Food Protein Analysis: Qualitative Effects on Processing

Home , Constance Kies, Edestin, Lowry protein assay

FOOD PROTEIN ANALYSIS Quantitative Effects on Processing

R. K. Owusu-Apenten The Pennsylvania State University University Park, Pennsylvania

Marcel Dekker, Inc. New York • Basel

Copyright © 2002 by Marcel Dekker, Inc. All Rights Reserved. ISBN: 0-8247-0684-6 This book is printed on acid-free paper. Headquarters Marcel Dekker, Inc. 270 Madison Avenue, New York, NY 10016 tel: 212-696-9000;fax: 212-685-4540

Eastern Hemisphere Distribution Marcel Dekker AG Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-261-8482;fax: 41-61-261-8896 World Wide Web http://www.dekker.com

The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special Sales/Professional Marketing at the headquarters address above. Copyright # 2002 by MarcelDekker, Inc. AllRightsReserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, micro®lming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit): 10987654321

PRINTED IN THE UNITED STATES OF AMERICA FOOD SCIENCE AND TECHNOLOGY

A Series of Monographs, Textbooks, and Reference Books

EDITORIAL BOARD

Senior Editors Owen R. Fennema University of Wisconsin–Madison Y.H. Hui Science Technology System Marcus Karel Rutgers University (emeritus) Pieter Walstra Wageningen University John R. Whitaker University of California–Davis

Additives P. Michael Davidson University of Tennessee–Knoxville Dairy science James L. Steele University of Wisconsin–Madison Flavor chemistry and sensory analysis John H. Thorngate III University of California–Davis Food engineering Daryl B. Lund University of Wisconsin–Madison Food proteins/food chemistry Rickey Y. Yada University of Guelph Health and disease Seppo Salminen University of Turku, Finland Nutrition and nutraceuticals Mark Dreher Mead Johnson Nutritionals Phase transition/food microstructure Richard W. Hartel University of Wisconsin– Madison Processing and preservation Gustavo V. Barbosa-Cánovas Washington State University–Pullman Safety and toxicology Sanford Miller University of Texas–Austin

1. Flavor Research: Principles and Techniques, R. Teranishi, I. Hornstein, P. Is- senberg, and E. L. Wick 2. Principles of Enzymology for the Food Sciences, John R. Whitaker 3. Low-Temperature Preservation of Foods and Living Matter, Owen R. Fenne- ma, William D. Powrie, and Elmer H. Marth 4. Principles of Food Science Part I: Food Chemistry, edited by Owen R. Fennema Part II: Physical Methods of Food Preservation, Marcus Karel, Owen R. Fenne- ma, and Daryl B. Lund 5. Food Emulsions, edited by Stig E. Friberg 6. Nutritional and Safety Aspects of Food Processing, edited by Steven R. Tan- nenbaum 7. Flavor Research: Recent Advances, edited by R. Teranishi, Robert A. Flath, and Hiroshi Sugisawa 8. Computer-Aided Techniques in Food Technology, edited by Israel Saguy 9. Handbook of Tropical Foods, edited by Harvey T. Chan 10. Antimicrobials in Foods, edited by Alfred Larry Branen and P. Michael Davidson 11. Food Constituents and Food Residues: Their Chromatographic Determination, edited by James F. Lawrence 12. Aspartame: Physiology and Biochemistry, edited by Lewis D. Stegink and L. J. Filer, Jr. 13. Handbook of Vitamins: Nutritional, Biochemical, and Clinical Aspects, edited by Lawrence J. Machlin 14. Starch Conversion Technology, edited by G. M. A. van Beynum and J. A. Roels 15. Food Chemistry: Second Edition, Revised and Expanded, edited by Owen R. Fennema 16. Sensory Evaluation of Food: Statistical Methods and Procedures, Michael O'Mahony 17. Alternative Sweeteners, edited by Lyn O'Brien Nabors and Robert C. Gelardi 18. Citrus Fruits and Their Products: Analysis and Technology, S. V. Ting and Russell L. Rouseff 19. Engineering Properties of Foods, edited by M. A. Rao and S. S. H. Rizvi 20. Umami: A Basic Taste, edited by Yojiro Kawamura and Morley R. Kare 21. Food Biotechnology, edited by Dietrich Knorr 22. Food Texture: Instrumental and Sensory Measurement, edited by Howard R. Moskowitz 23. Seafoods and Fish Oils in Human Health and Disease, John E. Kinsella 24. Postharvest Physiology of Vegetables, edited by J. Weichmann 25. Handbook of Dietary Fiber: An Applied Approach, Mark L. Dreher 26. Food Toxicology, Parts A and B, Jose M. Concon 27. Modern Carbohydrate Chemistry, Roger W. Binkley 28. Trace Minerals in Foods, edited by Kenneth T. Smith 29. Protein Quality and the Effects of Processing, edited by R. Dixon Phillips and John W. Finley 30. Adulteration of Fruit Juice Beverages, edited by Steven Nagy, John A. Atta- way, and Martha E. Rhodes 31. Foodborne Bacterial Pathogens, edited by Michael P. Doyle 32. Legumes: Chemistry, Technology, and Human Nutrition, edited by Ruth H. Matthews 33. Industrialization of Indigenous Fermented Foods, edited by Keith H. Steinkraus 34. International Food Regulation Handbook: Policy · Science · Law, edited by Roger D. Middlekauff and Philippe Shubik 35. Food Additives, edited by A. Larry Branen, P. Michael Davidson, and Seppo Salminen 36. Safety of Irradiated Foods, J. F. Diehl 37. Omega-3 Fatty Acids in Health and Disease, edited by Robert S. Lees and Marcus Karel 38. Food Emulsions: Second Edition, Revised and Expanded, edited by Kåre Lar- sson and Stig E. Friberg 39. Seafood: Effects of Technology on Nutrition, George M. Pigott and Barbee W. Tucker 40. Handbook of Vitamins: Second Edition, Revised and Expanded, edited by Lawrence J. Machlin 41. Handbook of Cereal Science and Technology, Klaus J. Lorenz and Karel Kulp 42. Food Processing Operations and Scale-Up, Kenneth J. Valentas, Leon Levine, and J. Peter Clark 43. Fish Quality Control by Computer Vision, edited by L. F. Pau and R. Olafsson 44. Volatile Compounds in Foods and Beverages, edited by Henk Maarse 45. Instrumental Methods for Quality Assurance in Foods, edited by Daniel Y. C. Fung and Richard F. Matthews 46. Listeria, Listeriosis, and Food Safety, Elliot T. Ryser and Elmer H. Marth 47. Acesulfame-K, edited by D. G. Mayer and F. H. Kemper 48. Alternative Sweeteners: Second Edition, Revised and Expanded, edited by Lyn O'Brien Nabors and Robert C. Gelardi 49. Food Extrusion Science and Technology, edited by Jozef L. Kokini, Chi-Tang Ho, and Mukund V. Karwe 50. Surimi Technology, edited by Tyre C. Lanier and Chong M. Lee 51. Handbook of Food Engineering, edited by Dennis R. Heldman and Daryl B. Lund 52. Food Analysis by HPLC, edited by Leo M. L. Nollet 53. Fatty Acids in Foods and Their Health Implications, edited by Ching Kuang Chow 54. Clostridium botulinum: Ecology and Control in Foods, edited by Andreas H. W. Hauschild and Karen L. Dodds 55. Cereals in Breadmaking: A Molecular Colloidal Approach, Ann-Charlotte Elias- son and Kåre Larsson 56. Low-Calorie Foods Handbook, edited by Aaron M. Altschul 57. Antimicrobials in Foods: Second Edition, Revised and Expanded, edited by P. Michael Davidson and Alfred Larry Branen 58. Lactic Acid Bacteria, edited by Seppo Salminen and Atte von Wright 59. Rice Science and Technology, edited by Wayne E. Marshall and James I. Wadsworth 60. Food Biosensor Analysis, edited by Gabriele Wagner and George G. Guilbault 61. Principles of Enzymology for the Food Sciences: Second Edition, John R. Whitaker 62. Carbohydrate Polyesters as Fat Substitutes, edited by Casimir C. Akoh and Barry G. Swanson 63. Engineering Properties of Foods: Second Edition, Revised and Expanded, edited by M. A. Rao and S. S. H. Rizvi 64. Handbook of Brewing, edited by William A. Hardwick 65. Analyzing Food for Nutrition Labeling and Hazardous Contaminants, edited by Ike J. Jeon and William G. Ikins 66. Ingredient Interactions: Effects on Food Quality, edited by Anilkumar G. Gaonkar 67. Food Polysaccharides and Their Applications, edited by Alistair M. Stephen 68. Safety of Irradiated Foods: Second Edition, Revised and Expanded, J. F. Diehl 69. Nutrition Labeling Handbook, edited by Ralph Shapiro 70. Handbook of Fruit Science and Technology: Production, Composition, Storage, and Processing, edited by D. K. Salunkhe and S. S. Kadam 71. Food Antioxidants: Technological, Toxicological, and Health Perspectives, edited by D. L. Madhavi, S. S. Deshpande, and D. K. Salunkhe 72. Freezing Effects on Food Quality, edited by Lester E. Jeremiah 73. Handbook of Indigenous Fermented Foods: Second Edition, Revised and Ex- panded, edited by Keith H. Steinkraus 74. Carbohydrates in Food, edited by Ann-Charlotte Eliasson 75. Baked Goods Freshness: Technology, Evaluation, and Inhibition of Staling, edited by Ronald E. Hebeda and Henry F. Zobel 76. Food Chemistry: Third Edition, edited by Owen R. Fennema 77. Handbook of Food Analysis: Volumes 1 and 2, edited by Leo M. L. Nollet 78. Computerized Control Systems in the Food Industry, edited by Gauri S. Mittal 79. Techniques for Analyzing Food Aroma, edited by Ray Marsili 80. Food Proteins and Their Applications, edited by Srinivasan Damodaran and Alain Paraf 81. Food Emulsions: Third Edition, Revised and Expanded, edited by Stig E. Fri- berg and Kåre Larsson 82. Nonthermal Preservation of Foods, Gustavo V. Barbosa-Cánovas, Usha R. Pothakamury, Enrique Palou, and Barry G. Swanson 83. Milk and Dairy Product Technology, Edgar Spreer 84. Applied Dairy Microbiology, edited by Elmer H. Marth and James L. Steele 85. Lactic Acid Bacteria: Microbiology and Functional Aspects, Second Edition, Revised and Expanded, edited by Seppo Salminen and Atte von Wright 86. Handbook of Vegetable Science and Technology: Production, Composition, Storage, and Processing, edited by D. K. Salunkhe and S. S. Kadam 87. Polysaccharide Association Structures in Food, edited by Reginald H. Walter 88. Food Lipids: Chemistry, Nutrition, and Biotechnology, edited by Casimir C. Akoh and David B. Min 89. Spice Science and Technology, Kenji Hirasa and Mitsuo Takemasa 90. Dairy Technology: Principles of Milk Properties and Processes, P. Walstra, T. J. Geurts, A. Noomen, A. Jellema, and M. A. J. S. van Boekel 91. Coloring of Food, Drugs, and Cosmetics, Gisbert Otterstätter 92. Listeria, Listeriosis, and Food Safety: Second Edition, Revised and Expanded, edited by Elliot T. Ryser and Elmer H. Marth 93. Complex Carbohydrates in Foods, edited by Susan Sungsoo Cho, Leon Prosky, and Mark Dreher 94. Handbook of Food Preservation, edited by M. Shafiur Rahman 95. International Food Safety Handbook: Science, International Regulation, and Control, edited by Kees van der Heijden, Maged Younes, Lawrence Fishbein, and Sanford Miller 96. Fatty Acids in Foods and Their Health Implications: Second Edition, Revised and Expanded, edited by Ching Kuang Chow 97. Seafood Enzymes: Utilization and Influence on Postharvest Seafood Quality, edited by Norman F. Haard and Benjamin K. Simpson 98. Safe Handling of Foods, edited by Jeffrey M. Farber and Ewen C. D. Todd 99. Handbook of Cereal Science and Technology: Second Edition, Revised and Expanded, edited by Karel Kulp and Joseph G. Ponte, Jr. 100. Food Analysis by HPLC: Second Edition, Revised and Expanded, edited by Leo M. L. Nollet 101. Surimi and Surimi Seafood, edited by Jae W. Park 102. Drug Residues in Foods: Pharmacology, Food Safety, and Analysis, Nickos A. Botsoglou and Dimitrios J. Fletouris 103. Seafood and Freshwater Toxins: Pharmacology, Physiology, and Detection, edited by Luis M. Botana 104. Handbook of Nutrition and Diet, Babasaheb B. Desai 105. Nondestructive Food Evaluation: Techniques to Analyze Properties and Quality, edited by Sundaram Gunasekaran 106. Green Tea: Health Benefits and Applications, Yukihiko Hara 107. Food Processing Operations Modeling: Design and Analysis, edited by Joseph Irudayaraj 108. Wine Microbiology: Science and Technology, Claudio Delfini and Joseph V. Formica 109. Handbook of Microwave Technology for Food Applications, edited by Ashim K. Datta and Ramaswamy C. Anantheswaran 110. Applied Dairy Microbiology: Second Edition, Revised and Expanded, edited by Elmer H. Marth and James L. Steele 111. Transport Properties of Foods, George D. Saravacos and Zacharias B. Maroulis 112. Alternative Sweeteners: Third Edition, Revised and Expanded, edited by Lyn O’Brien Nabors 113. Handbook of Dietary Fiber, edited by Susan Sungsoo Cho and Mark L. Dreher 114. Control of Foodborne Microorganisms, edited by Vijay K. Juneja and John N. Sofos 115. Flavor, Fragrance, and Odor Analysis, edited by Ray Marsili 116. Food Additives: Second Edition, Revised and Expanded, edited by A. Larry Branen, P. Michael Davidson, Seppo Salminen, and John H. Thorngate, III 117. Food Lipids: Chemistry, Nutrition, and Biotechnology: Second Edition, Revised and Expanded, edited by Casimir C. Akoh and David B. Min 118. Food Protein Analysis: Quantitative Effects on Processing, R. K. Owusu- Apenten 119. Handbook of Food Toxicology, S. S. Deshpande 120. Food Plant Sanitation, edited by Y. H. Hui, Bernard L. Bruinsma, J. Richard Gorham, Wai-Kit Nip, Phillip S. Tong, and Phil Ventresca 121. Physical Chemistry of Foods, Pieter Walstra 122. Handbook of Food Enzymology, edited by John R. Whitaker, Alphons G. J. Voragen, and Dominic W. S. Wong 123. Postharvest Physiology and Pathology of Vegetables: Second Edition, Revised and Expanded, edited by Jerry A. Bartz and Jeffrey K. Brecht 124. Characterization of Cereals and Flours: Properties, Analysis, and Applications, edited by Gönül Kaletunç and Kenneth J. Breslauer 125. International Handbook of Foodborne Pathogens, edited by Marianne D. Miliotis and Jeffrey W. Bier

Additional Volumes in Preparation

Handbook of Dough Fermentations, edited by Karel Kulp and Klaus Lorenz

Extraction Optimization in Food Engineering, edited by Constantina Tzia and George Liadakis

Physical Principles of Food Preservation: Second Edition, Revised and Expanded, Marcus Karel and Daryl B. Lund

Handbook of Vegetable Preservation and Processing, edited by Y. H. Hui, Sue Ghazala, Dee M. Graham, K. D. Murrell, and Wai-Kit Nip

Food Process Design, Zacharias B. Maroulis and George D. Saravacos For Mum and Dad, Elizabeth, James, Richard, Candida, and A®a

Preface

There is no book dealing with food protein analysis exclusively, that is, with the analysis of proteins in the food system. This books attempts to ®ll this niche. Protein analysis comes in two forms: 1) Quantitative analysis, and 2) fractionation and characterization. The ®rst activity is described here. This publication provides a reference for planning, performing and interpreting assays for food proteins. Many approved methods derive from the late-19th century, but they have undergone rigorous testing and modernization. This book does not focus on reviewing the latest research methods for protein analysis. With the exceptions of Chapters 6 and 7, each of the 14 self- contained chapters describes one protein assayÐprinciples, practices, and expected results. This book describes the effect of food processing on protein assay results with the emphasis on how to analyze proteins in real foods. A number of ``Methods'' sections provide instructions for speci®c tests. Sample pretreatment and clean-up procedures are described. General pretreatment strategies help in the avoidance of interference. More speci®c clean-up methods apply to particular protein assays and are described along with these. Example results, performance characteristics, case reports, and practical problems and solutions related to a wide range of foods are detailed in numerous ®gures, tables, and references.

v vi Preface

Food protein analysis is a hugely important activity performed by thousands worldwide. The book should appeal to professionals interested in food proteins and anyone working in the food system formerly called the food chain. This includes researchers and workers in agricultural production, food processing, and wholesale and/or retail marketing. It provides information for the grain or dairy farmer, extension worker, agricultural scientist, food scientists and technologists, or college professor. Some techniques described in this book were ®rst used by clinicians, nutritionists, and veterinary scientists. The book may also be of interest to those in small businesses, private or government laboratories, research institutes, colleges, and universities. It will be useful to undergraduate, postgraduate, or postdoctoral students. Sections dealing with mechanisms assume graduate level chemistry and/or analytical biochemistry. Any shortcomings of this project are wholly my responsibility. I thank all those colleagues worldwide whose research is reported here. My thanks to Anna Dolezal, Mr. DeSouza and Professor Arthur Finch for teaching me to think for myself. I am grateful to my past students: Drs. Yetunde Folawiyo, Despina Galani, Michael Anaydiegwu, Kiattisak Duangmal, Pitaya Adulyatham, Kwanele Mdluli, Halima Omar and Sripaarna Banerjee for raising my awareness of protein assay issues and for reading parts of the manuscript. Thanks to Dr. Bob Roberts (The Pennsylvania State University) for his advice on combustion methods. I am grateful to Dr. S. Khokhar and Marcel Dekker, Inc., for their commitment. I am also grateful to my family for their support.

R. K. Owusu-Apenten Contents

Preface v

Part 1. Fundamental Techniques Chapter 1. Kjeldahl Method, Quantitative Amino Acid Analysis and Combustion Analysis 1 1. Introduction to Food Protein Analyses 1 2. Kjeldahl Analysis 7 3. Colorimetric Analysis of Kjeldahl Nitrogen 18 4. Quantitative Amino Acid Analysis 25 5. Combustion Nitrogen Analyzers 29 References 38

Part 2. Copper Binding Methods Chapter 2. The Alkaline Copper Reagent: Biuret Assay 47 1. Introduction 47 2. The Alkaline Copper Reagent Protein Assay 48 3. Chemistry of the Alkaline Copper Reagent Protein Assay 50 4. Interference Compounds 53

vii viii Contents

5. Sample Pretreatment and Avoiding Interferences 55 6. The Micro-Biuret or Ultraviolet Biuret Protein Analysis 56 7. Applications of the ACR Solution for Food Protein Analysis 57 References 64 Chapter 3. The Lowry Method 69 1. Introduction 69 2. The Lowry Protein Assay 70 3. Chemistry of the Lowry Assay 73 4. Calibration Features 77 5. Interference Compounds 80 6. Sample Pretreatment, Avoiding Interferences, and Ensuring Accuracy 86 7. Applications of Lowry Assays to Food Protein Analysis 87 References 93 Chapter 4. The Bicinchoninic Acid Protein Assay 99 1. Introduction 99 2. The BCA Protein Assay 103 3. Chemistry of the BCA Protein Assay 105 4. Calibration Features 109 5. Interference Compounds 110 6. Sample Pretreatment, Avoiding Interference, Ensuring Accuracy 112 7. Automated BCA Protein Assays 113 8. Applications of the BCA Assay to Food Protein Analysis 116 References 121

Part 3. Dye Binding Methods Chapter 5. The Udy Method 125 1. Introduction 125 2. The Udy Method 127 3. Solid-Phase Dye-Binding Assays 131 4. The Chemistry of Dye-Binding Protein Assays 133 5. Interference Compounds and Their Avoidance 147 6. Applications of Dye-Binding Assays for Food Protein Analysis 147 References 160 Chapter 6. The Bradford MethodÐPrinciples 169 1. Introduction 169 Contents ix

2. Theory of the Bradford Assay 171 3. Effect of Protein-Dye Binding Parameters on the Bradford Assay 183 4. Linearization Plots for the Bradford Assays 184 5. Assay Sensitivity and the Maximum Number of Dye Binding Sites 184 6. Solid-Phase Dye-Binding Assays 185 7. Interference Compounds and Sample Pretreatment 186 References 191 Chapter 7. Bradford AssayÐApplications 195 1. Introduction 195 2. Coomassie Brilliant Blue Dye-Binding Assays 195 3. Performance Characteristics of CBBG Dye-Binding Assays 201 4. Applications to Food Protein Analysis 204 References 218

Part 4. Immunological Methods for Protein Speciation Chapter 8. Immunological Assay: General Principles and the Agar Diffusion Assay 221 1. Introduction 221 2. Immunological Methods 225 3. Speciation of Proteins by Agar Gel Double Immunodiffusion Assay 230 References 241 Chapter 9. Speciation of Meat Proteins by Enzyme-Linked Immunosorbent Assay 247 1. Introduction 247 2. Raw Meat Speciation by Indirect ELISA 252 3. Raw Meat Speciation by Sandwich ELISA 255 4. Muscle Protein Antigens for ELISA 257 5. Cooked Meat Analysis by ELISA 260 6. Monoclonal Antibodies for Meat Speciation 265 7. Fish and Seafood Identi®cation by ELISA 268 8. Performance Characteristics for Different ELISA Formats 270 9. Meat Testing for Transmissible Spongiform Encephalopathy Agents 271 References 274 x Contents

Chapter 10. Speciation of Soya Protein by Enzyme-Linked Immunoassay 281 1. Introduction 281 2. Sample Pretreatment and Analysis of Soy Protein 281 3. Structure, Denaturation, and Renaturation of Soybean Proteins 285 4. Solvent-Extractable Soybean Protein 289 5. Thermostable Antigens for Soybean Protein Analysis 289 6. Other Nonmeat Proteins 292 References 292 Chapter 11. Determination of Trace Protein Allergens in Foods 297 1. Introduction 297 2. Soya Bean Protein Allergens 301 3. Peanuts 305 4. Wheat and Related Cereals 312 References 329

Part 5. Protein Nutrient Value Chapter 12. Biological and Chemical Tests for Protein Nutrient Value 341 1. Introduction 341 2. Human and Other In Vivo Assays for Protein Nutrient Value 346 3. Small Animal Bioassays for Protein Nutrient Value 348 4. In Vitro Methods for Assessing Protein Nutrient Value 354 5. Protein Digestibility 366 References 374 Chapter 13. Effect of Processing on Protein Nutrient Value 381 1. Introduction 381 2. Milk and Milk Powders 381 3. Infant Formulas 384 4. Feedstuffs and Concentrates for Livestock 386 5. Legumes and Oilseeds 393 6. Cereal and Cereal Products 398 7. Improving Cereal Protein Quality by Screening 401 References 402 Chapter 14. Protein Digestibility±Corrected Amino Acid Scores 411 1. Introduction 411 2. Protein Digestibility 411 3. Protein Denaturation 414 4. Chemical Deterioration of Protein Ingredients 416 Contents xi

5. Matrix Effects on the Rate of Deterioration of Protein Ingredients 420 6. Protein Digestibility±Corrected Amino Acid Scores (PDCAAS) 427 References 440

Index 447

Kjeldahl Method, Quantitative Amino Acid Analysis and Combustion Analysis

1. INTRODUCTION TO FOOD PROTEIN ANALYSES

Protein analysis is a subject of enormous economic and social interest. The market value of the major agricultural commodities (cereal grains, legumes, ¯our, oilseeds, milk, livestock feeds) is determined partly by their protein content. Protein quantitative analysis is necessary for quality control and is a prerequisite for accurate food labeling. Proteins from different sources have varying aesthetic appeal to the consumer. Compliance with religious dietary restrictions means excluding certain protein (sources) from the diet. The variety of protein consumed is also extremely important in relation to food allergy. Detecting undeclared protein additives and substitutions is a growing problem. Proteins show differing nutritional quality or ability to support dietary needs. In summary, protein analysis has legal, nutritional, health, safety, and economic implications for the food industry (1). The estimated global food production total for 1988 was 4 billion metric tons. Allowing an average of 10% protein in foodstuffs yields 400 million metric tons of protein annually (2). Nonetheless, sensitivity is a major consideration for protein analysts. Some immunological methods can detect nanomole (10À9 mole) amounts of protein. Other important considerations when choosing a method for food protein analysis include

1 2 Chapter 1

TABLE 1 Approximate Chronology for Methods for Food Protein Analysis

Date Technique

1831 Dumasa 1843 Nessler's reagenta 1849 Biuret method 1859 Alkali-phenol reagent or Bethelot's methoda 1883 Kjeldahla 1927 Folin-Ciocalteau 1944 Dye bindinga 1951 Lowry 1960 Direct alkaline distillation 1960 Near-infrared re¯ectance (NIR)a 1971 Modi®ed Berthelot reaction 1975 Modi®ed Lowry method (Peterson) 1976 Bradford method (Coomassie Blue binding method) 1985 Bicinchoninic acid (BCA) method aTechniques for which semiautomated or fully automated apparatus has been manufactured. high sample throughput, simplicity, and low capital costs. Some of the most signi®cant methods (Dumas, Kjeldahl, and biuret assays) date from the late 1800s (Table 1). Techniques for food protein analysis are described in this book. I will focus on the techniques that feature most often in the food science literature. Infrared analysis of food proteins is not discussed here.

1.1. Characteristics of Food Protein Assays Techniques for food protein analysis need to be robust. This means one of several things. Foremost is compatibility with fresh produce (cereals, fruits, vegetables, meat, milk) and processed foods. Samples in various physical states (powders, slurries, dilute liquids, emulsions, gels, pastes) should be analyzable. A robust assay will also deal effectively with foods from either animal or plant sources. Such techniques are unaffected by the presence of dyes or pigments that absorb infrared, visible, or ultraviolet light. A robust protein assay needs mimimal sample pretreatment, which increases error and decrease analytical precision. Sample cleanup also increases the time per analysis (reduces sample throughput) and adds to costs. In the worst-case scenario, pretreatment can be too invasive, thereby invalidating results. In summary, a robust protein assay is simple, quick, sensitive, and reliable. It is also compatible with a diverse range of foods. The economic imperative Kjeldahl Method 3 leads to a preference for techniques requiring low capital expenditure and minimum training. Laboratories handling more than 8000 analyses per year tend to select techniques on the basis of their speed and ease of operation. A high sample throughput is usually achieved by automation or continuous ¯ow analysis (CFA). A rough ``time line'' for some food protein assays is given in Table 1. Common descriptive terms for protein analysis are de®ned in Table 2. Kjeldahl analysis gives accurate protein readings no matter what the physical state of the sample. This technique has approved status and is the reference method adopted by many national and international organiza- tions. However, the use of hazardous and potentially toxic chemicals in Kjeldahl analysis is creating concern. The Dumas combustion method is comparatively quicker, cheaper, easier to perform, safer, and more environment friendly;it is now considered on equal terms with Kjeldahl analysis in the United States, Canada, and Western Europe. Dye binding is another robust test for proteins (3,4). The biuret method is widely used,

TABLE 2 Some Important Calibration Indices and a Brief Explanation of Their Meaning

Calibration feature Explanation

Linear dynamic range Range over which signal is proportional to analyte concentration Sensitivity Slope of the calibration graph;analytical response per unit change in protein concentration. cf. parameters a, a0 in Eqs. (1)±(4) Accuracy Degree of agreement of results with a true value Precision, repeatability, or Agreement between repeated measurements reproducibility taken with a single sample or with different paired samples Speci®city Ability to discriminate between protein and interfering substance. Ratio of sensitivity for the analyte and interference Reliability A composite parameter combining speci®city, accuracy, precision, and sensitivity Lower limit of detection (LLD) Minimal protein concentration detectable above background noise Sample throughput (time per Numbers of samples analyzed per unit time, analysis) speed of analysis 4 Chapter 1 especially for cereal proteins (5). Procedures involving copper-based reagents (Lowry and bicinchoninic acid assays) continue to be important. Finally, a range of empirical (viscosity, refractive index, speci®c gravity) measurements are used for protein quantitation within industry.

1.2. Calibration and Statistical Principles The two common forms of calibration are (a) method calibration and (b) sample calibrations. With method calibration a set of food samples are analyzed using a new test method and a reference method that has been validated by a committee of the Association of Of®cial Analytical Chemists (AOAC). A calibration graph is then drawn by plotting results from the reference method (% Kjeldahl protein) on the Y-axis and the test results on the X-axis. The Xi and Yi observations are usually related by an equation for a straight line:

Yi aXi b 1 where a is the gradient and b is the intercept for the calibration graph. For each Xi result we can determine the calculated % Kjeldahl protein value (Ycalc) via Eq. (2).

Ycalc aXi b 2

Values for Yi and Ycalc can be compared in order to evaluate the test method (see later). Some investigators choose to plot the Kjeldahl results on the X- axis. Therefore, rather than Eq. (1) we get

* 0 * 0 Yi a X b 3 where Xi*is% Kjeldahl protein and Yi* is the test result. To compare Eq. 0 0 (1) and Eq. (3), notice that a 1/a and b Yi À (Xi / a). For sample calibration, the assay technique is assumed to be valid. We analyze a set of (standard) samples containing known amounts of protein.

In Eq. (1), Xi now represents a range of known protein concentrations and Yi are the corresponding instrument responses. Calibration factors (a, b, etc.) can be determined from simple algebra or statistical analysis of paired

(Xi, Yi) results. From the principles of least-squares analysis, P X À X = Y À Y a i m i m 4 P 2 Xi À Xm Kjeldahl Method 5 and

b Ym À aXm 5 where Xm and Ym are the mean values for all Xi and Yi observations. Agreement between the reference and test results is measured by the correlation coef®cient (R); R&1 shows excellent agreement. When Yi and Ycalc observations are poorly correlated, R & 0. The squared correlation coef®cient (R2) can be calculated from Eq. (6). Most handheld calculators can perform this operation automatically. "# Y À Y 2 R2 1 À i calc 6 P 2 Yi À Ym

Precision is another measure of the (dis)agreement between Yi and Ycalc values. This can be expressed as the standard deviation (SD) or coef®cient of variation (CV). High-precision methods produce low values for the SD and CV. P Y À Y 2 SD 2 i calc 7 n À 2

CV SD=Ym6100 8

We can also measure precision (commonly called error) from n-replicate (Yi) measurements on a single test sample. Thereafter, the numerator in Eq. (7) 2 becomes (Yi À Ym) , which is the square of the differences between individual observations and the mean for all observations. A low CV implies good agreement between successive test results.

1.3. Assay Performance Calibration parameters can provide a great deal of other information about assay performance (Table 2). The linear dynamic range is the concentration range over which a linear relationship exists between the instrumental response and protein concentration. Sensitivity is the slope of the calibration graph, and the lower limit of detection (LLD) is smallest quantity of sample that triggers an instrumental response above the background noise. The LLD is dependent on the instrument baseline quality and assay sensitivity. It is common to refer to ``sensitivity'' when we mean the LLD.We differentiate between sensitivity and LLD via the following exercise.

Measure the instrument baseline noise by recording the output (Yo) and 6 Chapter 1 the standard deviation (SDo) using a sample blank. The smallest instrumental response that can be distinguished from ``random noise'' in

95% of all cases is Yo+2:326SDo. Now substitute for Yi (Yo 2.326 SDo) and Xi ( LLD) in Eq. (1), leading to the following expression: Y 2:326SD Àb LLD o o 9 a

Usually Yo and b are both set to zero when the analyst sets the instrument baseline response to zero. Consequently, Eq. (9) becomes

LLD 2:326SDo=a 10 This relation shows that LLD decreases with increasing assay sensitivity and with increasing baseline quality (see decrease in the value for SDo). In order to ensure high sensitivity, it is important to obtain a stable instrumental baseline.

1.4. Calibrating Protein Assays The Kjeldahl method is used for calibrating other protein assays. Duda and Szot (6) evaluated six methods for analyzing porcine plasma protein during its manufacture. The techniques are simple and therefore of wider interest (Table 3). The protein content of porcine plasma was 5.58% (w/v). All techniques showed a good correlation with Kjeldahl results (R 0.905± 0.952). The precision for density and Kjeldahl assays was the same (CV 10.8%). The sensitivity of the former method was better. With appropriate calibration, density or viscosity measurements could be suitable for the routine analysis during the manufacture of plasma proteins.

TABLE 3 Some Simple Methods for Evaluating Porcine Plasma Protein

Method Instrument

Densitometry Standard picnometer Refractometry Laboratory refractometer Modi®ed refractometry Laboratory refractometer UV absorbance (215/225 nm) UV spectrophotometer UV absorbance (241 nm) UV spectrophotometer UV absorbance (280 nm) UV spectrophotometer Kjeldahl Method 7

Williams et al. (7) calibrated beer protein analyses using quantitative sodium dodecyl sulfate polyacrylamide gel electrophoresis (QSDS-PAGE). A range of test methods were investigated including biuret, bicinchoninic acid (BCA), Bradford, Kjeldahl, Lowry, and pyrogallol-red molybdate (PRM) assays. QSDS-PAGE revealed that beer has between 0.5 and 1mgmLÀ1 protein. Only the Bradford and PRM assays gave accurate results (Fig. 1). The main sources of error were low-molecular-weight interferences. Beer contains plant pigments, starch, sugars, alcohol, and natural dyes of barley origin. Both Kjeldahl and combustion analyses were subject to interferences by nonprotein nitrogenous (NPN) compounds. Dialysis did not improve accuracy for BCA, Lowry, and biuret assays, which were affected by high-molecular-weight Cu- reducing agents such as pectin and starch. Calibration issues are discussed in two articles by Pomeranz and co- workers (8,9). They considered the reliability of several test methods (biuret, dye binding, infrared re¯ectance, alkaline distillation method) for analyzing proteins in hard red winter wheat varieties from the American Great Plains. The test methods were highly correlated with the Kjeldahl assay (R 0.976± 0.992). The order of precision was Kjeldahl > biuret > dye binding > infrared analysis. Pomeranz and More (9) also considered the reliability of four ``rapid'' methods for barley or malt protein analysis.* A summary of assay performance statistics is given in Table 4. For barley samples, the precision and sensitivity of analysis were highest for the Kjeldahl and infrared analyses. The use of Kjeldhal analysis to calibrate protein assays for dairy products was discussed by Luithi-Pent and Puhan (10) and also Lynch and Barbano (11).

2. KJELDAHL ANALYSIS

Johan Kjeldahl was born on August 16, 1849 in the town of Jaegerpris in Denmark. In 1876 he was employed by the Carlsberg brewery to develop an improved assay for grain protein. The Kjeldahl method was published in 1883. The original technique has been extensively modi®ed. Key steps for the assay are (a) sample digestion, (b) neutralization, (c) distillation and trapping of ammonia, and (d) titration with standard acid. An exhaustive

* For the purposes of calibration, 44 samples of barley and 49 samples of malt were analyzed with biuret, dye binding, infrared, alkaline distillation, and Kjeldahl tests. Such results were the basis for deriving calibration relations between Kjeldahl and each test method. Then a further 76 samples of barley and 72 samples of malt were analyzed using only the rapid test methods.

Each Xi result gave rise to a corresponding Kjeldahl protein (Ycalc) value. 8 Chapter 1

FIGURE 1 Apparent protein concentrations in stout beer as determined by seven methods. (Data from Ref. 7.)

TABLE 4 Barley Protein Analysis Using a Range of Techniques

Regression line Analysis time and correlation Standard error of Test method (min) coef®cienta analysisb

Biuret 10 Y 0.857 cP 0.336 and 0.2336 1.942 R 0.972 Dye bindingc 15 Ð Ð Infrared 0.5±1.0 Y 1.060 cP 0.838±1.980 À1.03 R 0.96 Alkaline 90 Y 1.070 cP 2.383 and 1.540 distillation À 0.670 acP, crude protein determined by Kjeldhal method (N 6 6.25); Y, response from the test method. b P 2 Assay standard error calculated from Ycalc À Yi = n. cNo information given. Source: Ref. 8. Kjeldahl Method 9 account of the Kjeldahl method can be found in the monograph by Bradstreet (12). The book is divided into ®ve chapters: 1, introduction to the Kjeldahl method;2, the Kjeldahl digestion;3, digestion procedure (for fertilizers, leather, cereals, foods and proteins, coal and fuels);4, the distillation and detection of ammonia. Chapter 5 is an extensive bibliography. A standardized Kjeldahl procedure appears in the Interna- tional Standard ISO-1871 (13). Further descriptions are given by Gaspar (14) and Osborne (15). Initially, only sulfuric acid was used for sample digestion. Then solid potassium permanganate was added to facilitate sample oxidation. Mercuric oxide was introduced as a catalyst in 1885. During the acid digestion phase, the food sample is heated with concentrated sulfuric acid, which causes dehydration and charring. Above a sample decomposition temperature, carbon, sulfur, hydrogen, and nitrogen are converted to carbon dioxide, sulfur dioxide, water, and ammonium sulfate [Eq. (11)].

NH2 CH2pCOOH q 1H2SO4? p 1CO2 q SO2

4p H2ONH4HSO4 11

Digestion is complete when the mixture turns clear (light green color), usually after 20±30 minutes of heating. A further (after-boiling) period of heating is necessary to ensure quantitative recovery of nitrogen. Data from McKenzie and Wallace (cited in Ref. 14) show that adding X (mg) of potassium sulfate per mL of sulfuric acid increases its boiling point according to the relation Y (8C) 55.8X 331.2. A maximum boiling point elevation of 1308C is achivable by adding 2 mg (potassium sulfate) per mL acid. A high boiling point reduces the sample digestion time. Sample digestion can also be facilitated by using a catalyst;the order of effectiveness for metal oxide catalysts is Hg > Se > Te > Ti > Mo > Fe > Cu > V >W > Ag (16). A proprietary brand of Kjeldahl catalyst (Kjeltabs from Foss Electric Ltd.) comes as tablets. Each tablet contains 0.25 g of mercuric oxide and 5 g of potassium sulfate. A working selenium catalyst can be formulated with potassium sulfate (32 g), mercuric sulfate (5 g), and selenium powder (1 g). Chemical oxidants (hydrogen peroxide, perchloric acid, or chromic acid) can be added to the sulfuric acid to speed up sample digestion. Ammonium sulfate is ®rst neutralized with alkali to form ammonia. This is then distilled and trapped using 4% boric acid. Ammonium borate is then titrated with standard acid in the presence of a suitable indicator. Low- cost Quick-®t glassware is readily available for distillation and titration. Sophisticated semiautomatic distillation systems are also available. The 10 Chapter 1 processes of neutralization, distillation, and titrimetric analyses are summarized as follows.

distill 2À NH4HSO4 2OH À? NH3 2H2O SO4 12

À NH3 H3BO3 excess?NH4H2BO3?NH4 H2BO3 13

À H2BO3 HCl? titration?H3BO3 14 Suitable titration indicators include methyl orange, methyl red, Congo red, and Tashiro indicator (a 1:1 mixture of 0.2% methyl red solution and 0.1% methylene blue).

2.1. Nitrogen-to-Protein Conversion Factors

The Kjeldahl technique measures sample nitrogen (SN) as ammonia. The value for SN is later converted to crude protein (cP) by multiplying by a Kjeldahl factor, FK.

cP %SN F K 15

À1 The units for SN are g-N 100 g (g-nitrogen released per 100 g of sample). À1 The Fk (g-protein g N) is the amount of protein that produces a gram of nitrogen. Fk is also called the nitrogen-to-protein conversion factor. AOAC- recommended values for FK for meat and other food are summarized by Benedict (17). Frequently, FK is given a default value of 6.25 or 5.7 for animal and plant proteins, which are assumed to have an average N content of 16% and 17.5%, respectively. In fact, most proteins deviate signi®cantly from these averages (18). FK is also affected by the presence of NPN (e.g., adenine, ammonia, choline, betaine, guanidine, nucleic acid, urea, free amino acids). Soya beans have 3±10% NPN, which increases to about 30% for immature seeds. The amount of NPN also changes with growth conditions as well as with geographic factors. There is generally no correlation between NPN and protein content (19). No single FK value applies to all food types. Ideally, FK should be determined for each individual food type (Table 5).

FK can be calculated from amino acid data (18,20±24). Table 5 lists the 20 naturally occurring amino acids along with their formula weight, number of nitrogen atoms, percent nitrogen, and the value for FK. For arginine, FK is 3.11 100=32:15. An idealized protein having all 20 amino acids in equal numbers has a nitrogen content of 14.73%. The FK value is therefore 6.79 (100 g/14.73). Evaluating FK from amino acid data (for TABLE 5 Determination of the Kjeldahl Factor for Milk Protein Using 11 Amino Acid Composition Dataa Method Kjeldahl

12 345678910 Amino Formula % N Amino AA composition AA-N AA Mole fraction

Acid weight (bi) N atoms Acid FK (mg/g N) (mg/g N) (moles/g N) (Xi) biX Arginine 174.2 4 32.15 3.11 234.0 75.2 0.001343 0.027082 4.71761 Histidine 155.2 3 27.06 3.70 188.0 50.9 0.001211 0.024422 3.79022 Asparagine 132.1 2 21.20 4.72 292 61.9 0.00221 0.044564 5.88694 Glutamine 146.1 2 19.16 5.22 783 150.1 0.005359 0.108048 15.7859 Lysine 146.2 2 19.15 5.22 487.0 93.3 0.003331 0.067157 9.81828 Glycine 75.1 1 18.64 5.36 128.0 23.9 0.001704 0.034362 2.58058 Alanine 89.1 1 15.71 6.36 219.0 34.4 0.002458 0.049553 4.4152 Tryptophan 204.2 2 13.71 7.29 90.0 12.3 0.000441 0.008886 1.81447 Serine 105.1 1 13.32 7.51 338.0 45.0 0.003216 0.064837 6.81433 Proline 115.1 1 12.16 8.22 571.0 69.5 0.004961 0.100016 11.5118 Valine 117.1 1 11.96 8.36 428.0 51.2 0.003655 0.073687 8.6288 Threonine 119.1 1 11.75 8.51 278.0 32.7 0.002334 0.047059 5.60469 Cystine 121.1 1 11.56 8.65 47.0 5.4 0.000388 0.007825 0.94756 Isoleucine 131.2 1 10.67 9.37 290.0 30.9 0.00221 0.044563 5.84662 Leucine 131.2 1 10.67 9.37 600.0 64.0 0.004573 0.092199 12.0964 Aspartic acid 133.1 1 10.52 9.51 214.0 22.5 0.001608 0.032415 4.3144 Glutamic acid 147.1 1 9.52 10.51 574.0 54.6 0.003902 0.078669 11.5723 Methionine 149.2 1 9.38 10.66 148.0 13.9 0.000992 0.019999 2.98379 Phenylalanine 165.2 1 8.47 11.80 341.0 28.9 0.002064 0.041615 6.87481 Tyrosine 181.2 1 7.73 12.94 297.0 22.9 0.001639 0.033045 5.98774 Sum 5669.7 943.5 0.04960 1.0 132.0 Average 14.73

FK(1) 6.79 FK 6.01 a FK(1) is determined from the average nitrogen content for all amino acids, i.e., 14.73%.FK(2) value of whole protein or sum of amino acid nitrogen divided by sum of weight of nitrogen. FK 5669.7/943.5. Columns 8±10 contain data for calculating total protein content from quantitative amino acid analysis (Sec. 4). 12 Chapter 1 skimmed milk) can be achieved in the following steps: (a) express each amino acid as mg per gram of total nitrogen (see column 6 of Table 5) and (b) calculate the mass of nitrogen derived from each amino acid (column 7 in

Table 5). FK is the weight of amino acids divided by the weight of amino acid nitrogen (AA-N).

total weight of AA 5696:7 F 6:01 16 K total weight of AA À N 943:5

Some typical values for FK are listed in Table 6 for a range of foods. The use of FK values for quantitative amino acid analysis is discussed in Sec. 4.

TABLE 6 Nitrogen-Protein Conversion (Fk) Factors for Selected Food Protein Sources

Food product Fk Food product Fk Dairy products and egg Roots and tuber Casein 6.15 Carrot 5.80 Milk 6.02 Beet 5.27 Cheese 6.13 Potato 5.18 Egg 5.73 Potato protein 5.94 Egg white solids 5.96 Meat and ®sh products Fruit Beef 5.72 Tomato 6.26 Chicken 5.82 Banana 5.32 Fish 5.82 Apple 5.72 Leafy vegetables Microbial and fungal Lettuce 5.14 Yeast 5.78 Cabbage 5.30 Mushrooms 5.61 Cereals and legumes Wheat 5.71±5.75 Buckwheat 5.53 Rice 5.61±5.64 Oats 5.50 Corn 5.72 Millet 5.68 Sorghum 5.93 Mustard seed 5.40 Field pea 5.40 Rapeseed meal 5.53 Dry bean 5.44 Sun¯ower meal 5.36 Soya bean 5.69 Flax meal 5.41

Source: Data from Refs. 18 and 24. Kjeldahl Method 13

2.2. Macro- and Micro-Kjeldahl Analysis Kjeldahl digestion methods are discussed in this section. Illustrative examples are given to establish a pattern of work. Individual results are described in later parts of the chapter.

A. Grain and Cereals Kaul and Sharma (25) analyzed a range of legume and cereal grains by micro-Kjeldahl analysis. About 200 mg of each sample was weighed into several 75-mL Kjeldahl digestion tubes. Concentrated sulfuric acid (3 mL), hydrogen peroxide (1.5 mL), and one Kjeltab tablet were added. The tubes were heated using a Tectator digestion block at 3748C for exactly 25 minutes and then allowed to cool. The contents of each tube were diluted to 75 mL and any ammonia produced quanti®ed by colorimetric analysis (Sec. 3).

B. Potatoes Mohyuddin and Mazza (26) analyzed proteins from 14 potato cultivars. Potato tubers were peeled, sliced, diced, and dried in a vacuum oven at 708C and 48.8 mm Hg pressure. Each sample was milled and sieved through a 40- mesh sieve. Potato ¯our (100 mg) was added to each 100-mL Kjeldahl ¯ask, followed by concentrated sulfuric acid (3 mL), hydrogen peroxide (30% solution; 1.5 mL), and commercial catalyst (500 mg; 10:0.7 w/w ratio of potassium sulfate and mercuric oxide). Heating for 45 minutes digested the samples. After cooling to room temperature, the contents of Kjeldahl ¯asks were diluted to 75 mL and then subjected to colorimetric analysis to determine ammonia.

C. Dried Milk Powder Venter et al. (27) described a semimicro-Kjeldahl analysis for low-fat, medium-fat, and high-fat dried milk. About 200±250 mg of sample was mixed with 2.1 g of selenium catalyst and digested by heating with 10 mL of concentrated sulfuric acid. The digest was cooled and diluted to 100 mL with distilled water. Then 60 mL of 45% (w/v) NaOH solution was added and the liberated ammonia was distilled into 20 mL of 4% (w/v) boric acid solution. Titration was with 0.02 M HCl to an end point of pH 4.8. The results agreed well with the macro-Kjeldahl method [International IDF Standard (1962) No. 20]. 14 Chapter 1

D. Beer Protein Concentrated sulfuric acid (2 mL) was added to 50 mL of beer (bitter, lager, or stout) and the mixture was heated until nearly dry (7). Kjeldahl catalyst (10 g) and more sulfuric acid (20 mL) were added, followed by further heating for 25 minutes. After cooling for 2 hours, water (250 mL) was added and the Kjeldahl ¯ask was connected to a condenser with one end immersed in a 2% boric acid solution (200 mL). Bromocresol green was used as indicator. Sodium hydroxide (10 M, 70 mL) was added, followed by heating until the distillate tested neutral. The borate solution was later titrated with 0.1 N HCl. The nitrogen content in beer was calculated from the relation

%N 14Va=Wd 17 where Va (mL) is the volume of HCl required to neutralize ammonia and Wd (g) is the dry weight of beer. Table 7 summarizes characteristics of the Kjeldahl method used in the brewing and allied industries (28).

TABLE 7 Macro-Kjeldahl Procedures Currently in Use in the Brewing and Allied Industries

Parameter Commenta

Instrumentationb Kjel-Tec 1030 (Auto), Kjel-Tec 1007/ 1002*; Kjel-Foss 16120y Sample weight 0:92+0:23 g (barley or malt) range 0.5±1.5 g, 14+8:4 mL (beer) range 5±25 mL c Catalyst (weight range) K2SO4 CuSO4 TiO2;3.5±15.8 g Volume of conc. H2SO4 16+4:3 mL, range 10±25 mL Digestion temperatured 411Æ148C, range 380±4308C Digestion timee 75 min or 9.6 min End-point detection Mainly colorimetric aAverage values unless otherwise stated. bLocal suppliers *Tectator Ltd. and Perstop Ltd., both of Bristol, UK, { Foss Electric (UK) Ltd., The Chantry, Bishopthorpe, York, UK. cA large amount of a single catalyst or a smaller quantity of a combination catalysts was used. HgO was used in 2 of the 25 laboratories. dDigestion temperatures were not reported for seven laboratories using a manual Kjeldahl technique. e Digestion time plus after-boiling time. The digestion time is 9.6 min when H2O2 is used as a prooxidant. Source: Summarized from Ref. 28. Kjeldahl Method 15

2.3. Automated Kjeldahl Analysis Kjeldahl analysis has undergone three forms of automation. The Kjel- Foss1 instrument mechanizes the entire micro-Kjeldahl procedure (digestion, neutralization, distillation, and titration). The Kjel-Tec1 technique uses a digestion block in conjunction with apparatus for automated distillation and titrimetric analysis. The ®nal form of automation is the Technicon AutoAnalyzer Instrument1, which uses continuous ¯ow analysis (CFA).

A. The Kjel-Foss Instrument The Kjel-Foss instrument (N. Foss Electric Ltd., Hillerùd, Denmark) performs the entire Kjeldahl procedure automatically (29±31). Automation reduces the analysis time from 3 hours to 6 minutes. The ®rst analysis is completed in 12 minutes and succeeding analyses every 3 minutes. The sample throughput is 120±160 analyses per day. The Kjel-Foss instrument requires a reliable supply of electricity and tap water for installation and adequate drains and ventilation. A fume cupboard is not essential. Accuracy, precision, and economics of the Kjel-Foss method were compared with those of the manual Kjeldahl method, neutron activation analysis, proton activation analysis, combustion analysis, and the Kjel-Tech method (32). Results of the Kjel-Foss and manual Kjeldahl methods were highly correlated. Fish meal was analyzed using the Kjel-Foss instrument by Bjarno (33). Seven collaborators compared the ef®ciency of antimony versus mercuric oxide as catalyst. After modifying the Kjel-Foss procedure slightly with higher acid settings, the differences in recovery and repeatability of the two procedures were < +1%. Using mercuric oxide catalyst poses environ- mental concerns if the ef¯uent from the Kjel-Foss instrument is to be disposed of through the sewers. McGill (34) compared the Kjel-Foss method with an improved AOAC Kjeldahl method for meat and meat products. Over 80 analyses were performed with low (25%) fat, high (40%) fat, and dry sausage (50% fat). As shown in Fig. 2 and Table 8, the two techniques were highly correlated (Y 0.9904X 0.1797; R2 0.9896). There was no systematic error in the Kjel-Foss technique over the range of protein concentrations examined by McGill. This work validated the Kjel-Foss instrument for meat product analysis. Suhre et al. (35) also evaluated the Kjel-Foss instrument for meat analysis using the AOAC Kjeldahl method as reference. Twenty-three different laboratories analyzed six meat products having 10±30% crude protein. Eight laboratories used the automated Kjel-Foss instrument, ®ve used the of®cial AOAC method, and eleven used a block digester with steam 16 Chapter 1

FIGURE 2 Calibration graph for the Kjel-Foss automated method for protein determinations. (Data from Table 1 of Ref. 34.)

distillation. Recommendations from this work led to the block digester± steam distillation method being awarded ``®rst action'' status.

B. Automated Kjeldahl Continuous Flow Analysis The Technicon AutoAnalyzer has two reaction modules (36). The ®rst module digests water-dispersible samples. The digest is then pumped to a second module (AutoAnalyzer Sampler II). Colorigenic reagents are added in quick succession before the ¯ow stream passes to a delay coil to allow color formation. Ammonia is detected using Berthelot's reaction or the ninhydrin assay (Sec. 3). The AutoAnalyzer was applied for protein determinations in plant material (37), feedstuffs (38), grain ¯our (39), instant breakfasts, meat analogues (40), meat products (41), and over 40 assorted canned and processed foods (42,43). In general AutoAnalyzer results agreed with micro-Kjeldahl analysis. The AutoAnalyzer digestion unit is heated in two stages at 380±4008C and 300±3208C. To achieve ef®cient digestion, the ratio of acid to sample is Kjeldahl Method 17

TABLE 8 A Comparison of the Automated Kjel-Foss and Approved Kjeldahl Method for Protein Analysis in Sausage Samples

Sausage type Kjel-Foss (% protein) Kjeldahl (% protein)

Low-fat sausages Bologna 14.64 14.54 Polish 14.42 14.43 Krakow 16.76 16.60 Liver 16.64 16.74 Hungarian 16.01 16.02 Medium-fat sausages Breakfast 14.40 14.52 Italian 14.77 14.80 Pizza 17.42 17.33 Pork 14.09 14.21 Dry sausages Pepperoni 14.33 14.41 Summer 18.87 18.86

Source: Ref. 34.

higher than normal for batch digestion. A superheated layer of acid forms, which facilitates sample digestion (44). Later tests showed that the recovery of nitrogen from refractory materials (arginine, creatine, or nicotinic acid) was only 70%. Davidson, et al. (45) concluded that the AutoAnalyzer digestion module was not reliable if an accuracy of 1% was desired for Kjeldahl analysis. Over 70 different animal feeds (corn grain, wheat, barley, rice, alfalfa, mixed feeds, feed concentrates) were analyzed using the AutoAnalyzer digestion module. The recovery of nitrogen was 88±90% (46). In contrast, using a Technicon block digestor followed by AutoAnalyzer Sampler II led to 100% recovery of nitrogen from cattle supplement, swine ration, pig starter, and poultry ration (47). Suitable catalysts include copper sulfate and oxides of mercury, selenium, or titanium. Ammonia was detected using alkaline phenol reagent (Sec. 3.1). Quantitative recovery of nitrogen was also demonstrated by Kaul and Sharma (25), who used a Tectator1 heating block to digest 23 assorted strains of rice and 15 other cereal-legume mixtures. The electrically heated block digests 40 samples per hour under controlled temperature conditions. Samples were then transferred to the AutoAnalyzer Sampler II for ammonia detection using the alkaline phenol reagent. 18 Chapter 1

3. COLORIMETRIC ANALYSIS OF KJELDAHL NITROGEN

Colorimetric analysis simpli®es Kjeldahl analysis and increases the sample throughput. Other bene®ts include increased sensitivity and a greater potential for automation. Reagents for colorimetric Kjeldahl-N analyses include (a) alkali-phenol reagent (APR), also called indophenol reagent;(b) ninhydrin (indanetrione hydrate) reagent;(c) Nessler's reagent;and (d) acetylacetone formaldehyde reagent. These colorimetric techniques are reviewed next.

3.1. Alkali-Phenol Reagent (Indophenol) Method The alkali-phenol reagent is frequently used for the Technicon AutoAna- lyzer. Under alkaline conditions, ammonia, sodium hypochlorite, and phenol react to form a blue product. Berthelot ®rst reported this reaction in 1859. The principles of the APR assay have been reviewed (48±51) although the underlying reactions remain uncertain. Ammonia probably reacts with hypochlorite to form chloramine (NH2Cl). This reacts with phenol to form N-chloro-p-hydroxybenzoquinone monoimine or quinochloroamine (I).

À NH3 OCl ?NH2Cl

À PhO NH2Cl? I

PhOÀ I? II (I) Quinochloroamine and (II) Indophenol Alternatively, ammonia may ®rst react with phenol to form p- aminophenol, which is then oxidized by hypochlorite to form (I). Irrespective of how (I) forms, it reacts with 1 mole of phenol to form indophenol (II). The yellow compound ionizes under strongly alkaline 4 À1 À1 conditions (pKa 13.4) giving a blue anion De 1 À 2610 L mol cm . Substituted phenols (o-chlorophenol, m-cresol, and guaiacol) undergo the indophenol reaction. However, para-substituted phenols and some meta- derivatives do not react. Color intensity and the rate of color formation increase in the presence of manganese (2) ions or acetone. Sodium nitroprusside (10±40 mg LÀ1) is another catalyst. A simple APR assay suitable for detecting 3 ppm ammonia is described in Ref. 48 (Fig. 3). Indophenol formation is pH and temperature dependent. The linear dynamic range for ammonia was 0.3±3 ppm with a sensitivity of 0.3284 (absorbance units/1 ppm NH3). The assay precision (for 1 ppm NH3) was +3%. Although performed with boric acid as the background medium, the simple APR assay is probably not suitable for Kjeldahl Method 19

FIGURE 3 Calibration graph for ammonia determination using the alkali-phenol reagent assay. (Drawn from results in Ref. 48.)

Kjeldahl-N determination. Copper, zinc, and iron salts were found to act as interferences. Tetlow and Wilson (49) added ethylenediaminetetraacetic acid (EDTA) to APR to reduce metal ion interference. Temperature control was also improved using a thermostated water bath. An outline protocol is described below.

Method 1 Analysis of ammonia using the APR assay (48,49).

Reagents 1. Phenol (crystalline, 85% pure) 2. Sodium hydroxide solution (5 N) 3. Sodium hypochlorite solution (or commercial bleach) 4. Ammonium chloride solid 5. EDTA (6% w/v) 6. Acetone 20 Chapter 1

Procedure Preparation of alkali-phenol reagent. Place 62.5 g of solid phenol in a 500-mL beaker and add 135 mL of sodium hydroxide (5 N) slowly with stirring. Caution: Use an ice bath to avoid excessive heat buildup. Add 12 mL of acetone and make up the volume to 500 mL with deionized water. Sodium hypochlorite (1% w/v available chloride). Prepare by diluting commercially available bleach.

Ammonium chloride standards (1000 ppm NH3 and 100 ppm NH3). Dissolve 314.1 mg of solid NH4Cl in 100 mL of water and then dilute 10-fold. Prepare a working standard solution (0.5 ppm NH3) daily. The APR assay sequence. Place 1 mL of sample (or standard) in a test tube. Add EDTA solution (100 mL) with gentle shaking. Next, add APR (1 mL) and hypochlorite (0.5 mL) in quick succession, mixing after each addition. Finally, add 2.5 mL of water and incubate at

258C for 60 minutes. Take A625 readings for samples. Prepare a reagent blank as described next. Reagent blank (reverse addition method). First, mix hypochlorite (0.5 mL) and APR (1 mL) solutions and allow to react for 5±10 minutes. Next add EDTA (100 mL) followed by 3.4 mL of water (or the designated blank solution).

When reverse addition is used, hypochlorite reacts with phenol ®rst.

Traces of NH3 present in the blank are not detected (48,49). Reverse addition is useful where ammonia-free water is not available for sample preparation. After optimization, the linear dynamic range for ammonia analysis was 50± 500 ppb. Assay sensitivity was 200% greater than the results shown in Fig. 2.

Color formation with 50±800 ppb NH3 was virtually complete after 15 minutes at 14±308C. Temperature variations had little effect on the reaction. Thermostating at 258C for 60 minutes improved the precision. Addition of acetone to APR increased the response to ammonia by 10- fold. The color yield with 500 ppb ammonia declined by 2.65%,4.8%, and 6.8% for 4.5-hour-, 1-day- or 5-day-old APR. Addition of EDTA prevented interference from 100 ppb copper. The intervals between addition of various reagents must not exceed 1 minute to ensure optimum precision. Comparing results for normal and reverse addition provides a means for detecting very small amounts of NH3 in samplesÐsuch as water. Otherwise, ammonia-free water is needed for preparing reagents and blanks. The calibration curve for the APR assay is described by the equation A625 0.7120X, where X is the concentration of nitrogen (ppm). The analytical sensitivity was 0.7120 (absorbance units) for 1 ppm ammonia. The SD for the reagent blank was +0:0005 ppm. These values lead to an expected LLD for ammonia of 1.6 Kjeldahl Method 21

TABLE 9 The Comparative Costs of Manual APR Assay and Other Techniquesa

Capital Running cost Analysis Cost per Technique costb per yearc per year analysisd

Manual APR 6000 (1) 5200 (1) 8000 0.72 (1) method Micro-Kjeldahl 12000 (2) 7000 (1.3) 2000 4.1 (5.4) AutoAnalyzer 81,000 (13.5) 17,000 (3.3) 32,000 0.78 (1) APR test aCosts are given in deutsche marks. At current exchange rate 2.8 DM $1.4 £1. All methods employ a digestion unit costing DM6000. bCapital costs for the micro-Kjeldahl method include the cost of a distillation unit and an autotitrator. cThe running cost includes DM5000 for miscellaneous chemicals. dCalculated for a 10-year period as capital cost/10 running cost)/no. of samples. Ratios of costs are given in parentheses for each column.

ppb. Assuming a default FK of 6.25, the LLD for protein is 10 ppb. The APR assay is widely used in conjunction with the AutoAnalyzer.The composition of the APR used in CFA is pretty much the same as described earlier (45,52). Kaul and Sharma (25) describe a rare attempt to deploy a manual Kjeldahl-APR assay for protein analysis. They used a Tectator heating block for micro-Kjeldahl digestion of grain. Sample nitrogen was then analyzed by the APR assay. The analytical performance was similar to results obtained with the AutoAnalyzer-APR assay or the conventional micro-Kjeldahl analysis. From Table 9, the capital cost for the manual Kjeldahl-APR assay was 2 times lower than for the micro-Kjeldahl and 13.5 times lower than for the AutoAnalyzer method. Running costs were also lowest for the manual APR assay. For laboratories handling 40 or more analyses per day, it may be worth investing in an automated technique. The manual Kjeldahl-APR analysis was advantageous for small laboratories lacking the wherewithal to purchase an AutoAnalyzer. Mohyuddin and Mazza (53) used the manual Kjeldahl-APR assay to analyze potatoes (see Sec. II.B.2). The mean protein content for 14 potato cultivars was 10.65 (+1.23)% by the manual APR assay and 10.53 (+1.13)% using the AutoAnalyzer.

3.2. Nessler's Reagent Ammonia reacts with alkaline potassium iodomercurate II (Nessler's reagent) to form a colloidal complex (lmax 430±460). The linear range 22 Chapter 1 for analysis extends to 75 mg (ammonia) mlÀ1. A possible reaction scheme for Nesslerization is

2K2HgI4 3KOH NH3?OHg2NH2I 2H2O 7KI 18 Hach et al. (54) developed a commercial Nesslerization reagent for use in Kjeldahl analysis. A sulfuric acid±digested sample (0.4 mL) is diluted with 24.6 mL of 0.01% (w/w) polyvinyl alcohol (PVA) solution. One ml of Nessler's reagent is added and the sample is agitated mechanically before absorbance measurements are recorded at 430 nm. As the product of Nesslerization is colloidal in nature, spectrophotometric analysis is sensitive to the degree of sample agitation. PVA acts as a colloidal stabilizer and improves the precision of the Nessler method.

3.3. Acetylacetone-Formaldehyde Reagent The acetylacetone-formaldehyde assay is based on the Hantzsch reaction for the synthesis of pyridine (55). Prediluted digest is reacted with a mixture of acetyltacetone and formaldehyde in the presence of sodium acetate. The yellow product (3,5-diacetyl-1,4-dihydrolutidine) is measured at 410 nm De 1:46103LmolÀ1cmÀ1. The color-forming reaction is shown in Eq. (19).

Acetylacetone-formaldehyde reagent was used for the analysis of medicinal agents such as paracetamol, sulfanilamide, and chloropramide. The potential for colorimetric Kjeldahl analysis is obvious. Kjeldahl Method 23

Method 2 Determination of ammonia using acetylacetone-formaldehyde reagent (55).

Reagents Acetylacetone-formaldehyde reagent. Place 15 mL of formaldehyde (37% w/v) and acetylacetone (7.8 mL) into a 100-mL ¯ask. Make up to 100 mL with distilled water. Sodium acetate (2M). Dissolve sodium acetate (82 g) in 1 L of distilled water.

Procedure Add prediluted Kjeldahl digest (< 2 mL;25±100 mg N) to a 25-mL conical ¯ask followed by sodium acetate solution (3 mL) and acetylacetone-formaldehyde reagent (3 mL). Incubate the mixture at 97.88C for 15 minutes and cool to room temperature.

Bring the total volume to 25 mL and record A412 values using a 1-cm cuvette. The linear dynamic range for the preceding assay was 0.5±6.0 mgN (per ®nal reaction mixture). The calibration graph was described by À2 À3 2 A412 9:8610 X 4:2610 R 0:9999, where X is the amount of nitrogen present in the ®nal (25-mL) reaction mixture. The new method shows levels of accuracy and precision equal to those of the micro-Kjeldahl method.

3.4. Ninhydrin (Indanetrione Hydrate) Assay Ninhydrin* reacts with amino acids (Fig. 4) in two stages: (a) the amino acid is oxidized to aldehyde and ammonia while ninhydrin is converted to hydrindantin and (b) hydrindantin and 1 mole of ninhydrin react with ammonia to form Ruhemann's purple (56,57). For ammonia determination an added reducing agent is necessary to convert ninhydrin to hydrindantin. Ninhydrin solution is available commercially. Results from the ninhydrin- Kjeldahl assay agree closely with those from Kjeldahl analysis (56±58). The linear dynamic range for colorimetric Kjeldahl assay depends on the extent of sample dilution just before ninhydrin analysis. A 2-mL standard ammonium sulfate solution containing 5.6 mg ( or 2.8 ppm) reacted with

2 mL of ninhydrin solution yields an A570 reading of 0.805. Interference was noted for concentrations of selenium above 86 mgmLÀ1 (prediluted digest). No interferences were observed with Fe, Zn, Pb, Cu, Ca, Ba, Al, Mg, Co, or

* Ninhydrin is often encountered in detective novels as a reagent for ®ngerprint analysis. 24 Chapter 1

(a)

(b)

FIGURE 4 Reaction scheme between amino acids and ninhydrin reagent. Kjeldahl Method 25

FIGURE 5 Colorimetric ninhydrin analysis of Kjeldahl nitrogen calibrated against the conventional macro-Kjedahl analysis. (Drawn from data in Ref. 58.)

Ni at concentrations of 50 mgmLÀ1 or from Hg at 30 mgmLÀ1. The overall impression is that the ninhydrin assay is resistant to metal ions. Quinn et al. (58) analyzed rapeseed ¯our, rapeseed concentrate, soybean concentrate, and bovine serum albumin using the ninhydrin assay in conjunction with an AutoAnalyzer (Fig. 5). The precision of analysis was 1.40±1.76%. A manual Kjeldahl-ninhydrin assay has not been reported recently. This seems a pity. Compared with other colorimetric methods, the ninhydrin assay is more resistant to interferences from metal catalysts. The color is also formed at a more easily buffered pH between 4.9 and 5.4.

4. QUANTITATIVE AMINO ACID ANALYSIS

The following steps are involved in quantitative amino acid analysis: (a) hydrolyze a sample of food using concentrated hydrochloric acid, (b) determine the amino acid pro®le, (c) calculate the concentration of each amino acid in the sample, and (d) calculate the weight of each amino acid. Quantitative amino acid analysis is reportedly one of the most reliable methods for protein quantitation (59±61). 26 Chapter 1

4.1. Principles of Quantitative Amino Acid Analysis

Crude protein (cP) is expressed by Eq. (20), where Ci (mole) is the amount of À1 each amino acid in the sample and bi (g mole ), is the formula weight for each amino acid.

X20 cP Cibi 20 i1 However, amino acid pro®les are reported in terms of mole fraction of each amino acid (Xi):

Xi Ci=Cnet 21 where Cnet is the net concentration of amino acids found in the sample. Substituting Ci CnetXi in Eq. (20),

X20 cP Cnet biXi 22 i1

P À1 The term (biXi) is called the mean residue weight, W ( g mole ). This is the average formula weight for all amino acids in the sample adjusted for their frequency.* X À1 W g mole biXi 23 and

cP WCnet or cP FCnet 24 In Eq. (24), F is the mean residue weight. Usually, W (g moleÀ1) is adjusted to take into account two routine errors in amino acid analysis: (a) many colorimetric reagents for amino acids do not react with proline and (b) tryptophan is destroyed during acid hydrolysis of proteins. Proline and tryptophan are usually determined by separate experiments. After correcting for such errors, one gets the conversion factor, F (g moleÀ1):

P18 P18 biCi biXi F i1 or F i1 25 Cnet À CPro À CTrp 1 À XPro À XTrp where XPro and XTrp are the mole fractions of proline and tryptophan. For a

* Horstmann called this parameter the weight equivalent (WE). Kjeldahl Method 27 range of meat products, F (g moleÀ1) was 10±20% larger than W [Eq. (23)]. Calculations of F (g moleÀ1) appear in last three columns of Table 5. Typical values for F (g moleÀ1) are given in Table 10. Values for F range from 100 to 125 g moleÀ1 for most proteins.Most protein sources are now routinely analyzed for amino acid scores during nutritional evaluations (Chapter 14). This yields all the information necessary for total protein estimation. Zarkadas and co-workers in Canada are strong exponents of quantitative amino acid analysis (Table 11).

4.2. Quantitation of Speci®c Proteins Meat collagen was determined by measuring 5-hydroxylysine (5OH-Lys). This amino acid is found in collagen and no other meat protein. The concentration of collagen (Pj) was calculated from a modi®ed Eq. (24) (64,65);

1000Wj Pj Ci 26 njMi

TABLE 10 Corrected Mean Residue Weights (F) for Selected Food Protein Sources

Protein source F (g moleÀ1)

Barley ¯our 129.84 Soya bean ¯our 119.82 Pea ¯our 118.85 Fish meal 112.70 Beef sausage 107.06±109.01 Pig skin (rind) 94.02 Bone meal 104.21 Soya bean protein Flour 114.43 Isolate 114.48 Concentrate 115.69 Wheat Flour 113.13±116.00 Gluten 108.4±108.52 Egg white solids 118.43 Potato protein 108.52 Milk (nonfat) powder 112.98

Source: Data calculated or taken from references in Table 11. 28 Chapter 1

TABLE 11 Protein Determination by Quantitative Amino Acid Analysis

Sample/comments Reference

NASAÐSkylab meals Heidelbaugh et al. (59) Mushroom total protein Weaver et al. (62), Braaksma and Schaap (63) Porcine muscle and connective tissue Zarkadas et al. (64), Zarkadas et al. (65) proteins (myosin, actin, elastin, and collagen Actin, myosin, and collagen in Karatzas and Zarkadas (66) composite meat products;mixed meat sausages, bologna, frankfurters, sausages, hamburgers Additives and ingredients for meat Zarkadas et al. (67) products including soybean, wheat products, potato protein Porcine skin (rind) Nguyen et al. (68) Chicken meat and connective tissue Karatzas and Zarkadas (69) Apple ¯ower buds Khanizadeh et al. (70) Bone isolates Zarkadas et al. (71) New soybean cultivars Zarkadas et al. (72)

where Ci is the concentration of 5OH-Lys in the meat hydrolysate, Wj is mean residue weight from the amino acid pro®le for collagen (averaged for the different collagen types), nj is the number of 5OH-Lys residues per 1000 residues in collagen, and Mi is the formula weight for 5OH-Lys. Typically, À1 Wj 91:01 g mole , Ni 10 residues, Mj 145:18, and consequently

Amount of collagen 62:755OH-Lys 27

Similarly, the concentration of Nt-methylhistidine and 4-hydroxyproline was the basis for assessing the amount of myo®brillar protein and connective tissue (collagen and elastin) in meat. These methods are satisfactory for meat and meat products. They may have questionable validity for composite foods. Plant foods may contain 4-hydroxyproline± rich glycoproteins (extensins, lectins, salt-extractable glycoproteins). For example, alfalfa protein and potato protein contain signi®cant levels of 4OH-Pro. Kjeldahl Method 29

4.3. Examples and Relation to Kjeldahl Method Quantitative amino acid analysis is arguably one of the most accurate methods for food protein quantitation. One source of error is the high concentration of free amino acids in some foods. The protein content in nine strains of Agaricus was 28% (+3.4)% by quantitative amino acid analysis (62). The results were poorly correlated R 0:4 with Kjeldahl results (22.4% protein per dry weight basis). A more recent analysis of freeze-dried mushroom powder led to estimates of 7.0% protein by dry weight (63). In the later study, mushroom powder was extracted with 0.5 M NaOH and precipitating with trichloroacetic acid (TCA) before analysis. This removed large amounts of TCA-soluble NPN associated with mushrooms. Of the total NaOH-soluble nitrogen extracted from mushrooms, 20% was protein, 60% was urea or ammonia, and 20% was free amino acids. Food samples are now routinely extracted with organic solvents to remove NPN before quantitative amino acid analysis (Table 11). Advocates for quantitative amino acid analysis point to its compatibility with plant proteins. There is no interference from phenols, tannins, and lignin. By contrast, the Kjeldahl method is unsuitable for plant tissues regardless of the conversion factor used (73).

5. COMBUSTION NITROGEN ANALYZERS

The Dumas assay predates Kjeldahl analysis by 50 years (Table 1). The former technique was invented by Jean Baptiste Dumas. Early applications include the analysis of plant materials (74,75), meat (76), casein, whole powdered milk, soybeans, and maize ¯our (77). The ®rst-generation instruments for the Dumas method were not user friendly. The volume of nitrogen gas produced by combustion was determined with a manometer. The advent of easy-to-use and highly accurate combustion nitrogen analyzers (CNAs) rekindled interest in the Dumas method. CNAs from various manufacturers work on the same principle. The sample is dropped into a 950±10508C furnace, purged free of atmospheric gas, and ®lled with pure (99%) oxygen. Complete sample combustion leads to CO2, water, SO2,NO2, and N2. The product gases are cooled and a portion is passed through tubing packed with hot lead chromate, copper, sodium hydroxide (solid), or phosphorus pentoxide to remove SO2,O2, CO2, and water, respectively. The NO2 is then reduced to N2 and measured with a thermal conductivity detector (TCD). Sample protein content is calculated by taking into account the mass of sample injected, the 30 Chapter 1 proportion of the combustion gases analyzed, and the nitrogen-protein conversion factor (FK). The calculations are now automated.

5.1. Collaborative Studies and Approved Status for CNAs CNAs were calibrated with the Kjeldahl method. Interlaboratory studies appearing after 1987 are listed in Table 12. Such trials led to CNAs receiving approved status from the AOAC (Association of Of®cial Analytical Chemists), AOCS (American Oil Chemists' Society), ASBC (American Society of Brewing Chemists), AFI (American Feed Industry), BRF- International (Brewing Research Foundation-International), IOB (Institute of Brewing), and EBC (European Brewing Convention). The Canadian Grain Commission and U.S. Department of Agricul- ture (USDA) Federal Grain Inspection Services (FGIS) approved CNAs in

TABLE 12 Food Protein Analysis Using the Dumas Combustion Method

Samplea Reference

Animal feedstuffs, fertilizers Sweeney and Rexroad (78), Schmitter and Rhihs (79), Sweeney (80), Sachen and Thiex (81), Tate (82) Beer ASBC (83), Johnson and Johansson (84,85) Brewing grainsÐbarley, malt, rice ASBC (86), Buckee (28,87), Krotz et al. (88), Johansson (89), Angelino et al. (90) Cereal grainsÐwheat, barley, corn, Bicsak (91), Bicsak (92), Williams et al. sorghum (93) Dairy productsÐskimmed, powdered Wiles and Gray (94), Wiles et al. (95), milk etc. chocolate milkshake, Simonne et al. (96,97) cheeses, etc. FruitÐguava, peaches, plum Simonne et al. (96,97), Huang et al. (98) Infant food Bellemonte et al. (99) Meat and meat products, ®sh (raw, ®sh King-Brink and Sebranek (100), in oil, tuna) Simonne et al. (96,97), Buschmann and Westphal (101) Oilseeds (soybean, canola, sun¯ower, Bicsak (91), Duan and DeClercq (102), corn) Berner and Brown (103) Potatoes Young et al. (104) VegetablesÐcabbages, broccoli, Simonne et al. (96,97) ketchup, tomato aApproximate sample classi®cation;classes contain the other foodstuffs. Kjeldahl Method 31

TABLE 13 Advantages of the Dumas Method

1. Greater ease of operation 2. Higher operator safety owing to the nonrequirement for harzadous chemicals 3. The absence of wetchemistry 4. Reduced time of analysis 5. Higher performance characteristics (greatar accuracy, repeatablility) 6. Absence of waste disposal concerns (Table 14) 7. Simple instrument installation without a requirement for specialized ventilation 8. Low cost per analysis

1994 and 1996, respectively (91±93). Trials for the combustion method usually follow guidelines described by Youden and Sleiner (105): 1. The number of laboratories ranges from 7 to 12. Studies involving as few as three laboratories have been reported. 2. All studies compare CNAs with Kjeldahl analysis. 3. Interlaboratory studies focus on a single food group. Therefore, CNAs tend to receive approval for one food group at a time (Table 12). 4. Trials usually test a ``generic combustion method'' and are independent of the choice of instruments. Minimum performance guidelines for CNA instruments include (a) a furnace temperature of 9508C, (b) a separation system for trapping CO2 and water, (c) a thermal conductivity detector for nitrogen, (d) suf®cient accuracy to produce results within 0.15% of the mean (% nitrogen) results for 10 successive measurements using a standard compound, and (e) suf®cient precision to produce a relative standard deviation of 0.01%.The LECO FP-428 analyzer was used by about 80% of the laboratories involved in collaborative trials. The Foss-Heraue Macro-N analyzer, Carlo Erba NA-5000, and Perkin Elmer PE2410 also feature. The LECO FP-2000 combustion analyzer appears in the latest trials.

5.2. Advantages of the Combustion Method The modern CNA has advantages over the Kjeldahl method (Table 13). There is greater speed of analysis and greater operator safety stemming from the nonuse of aggressive chemicals. The estimated cost for analysis is $0.37± $0.50 per sample with the LECO FP-2000 protein analyzer (LECO Corporation, Saint Joseph, MI) compared with $1.0 per test for the Kjeldahl method (106±109). 32 Chapter 1

TABLE 14 A Comparison of Materials Reqirement for the Kjeldhal and Dumas Methods (74,84)

Requirement Kjeldahl Dumas

Chemical Conc. H2SO4,40% NaOH, Air, oxygen, helium, requirements K2SO4,TiO2/CuSO4 (or copper, turnings, EDTA, HgO), H3BO3 KH, nitrogen catalyst, Mg phthalate, methyl red, perchlorate, sodium phenolphthalein, pumice, hydroxide, alumina oxide water pellets Other suppliesa Kjeldahl and Erlenmeyer Tinfoil squares, brushes, tin ¯asks, burettes, acid, alkali capsules, combustion, and water dispensers, reduction and absorption stirring equipment, tubes, cotton wool, steel large containers for wool, particle ®lters, tubing acid, etc. Ancillary equipment Ductwork for corrosive Ductwork for warm airb fumes, acid-resistant fans, fume washer, fans, etc. Disposal of chemical Collected, professionally Nontoxic, wastebin or sink disposed disposal Time per analysis 120 min (24 samples) 3 min Degree of hazardc 62 Accuracy 70±98 100 Precision (CV %) 1.2 0.7 aDoes not include main equipment (Kjeldahl digester and distillation apparatus, or CNA instrument bOptional, but advisable for large-scale testing. cArbitrary scale of 1±10, with 10 being extremely hazardous and 1 completely safe. There may be a risk of burns when maintaining the combustion instrument.

A more detailed discussion of the relative costs of protein analysis by Kjeldahl or combustion analysis has to consider factors such as number of analyses per year, capital costs for instrumentation, depreciation, maintenance costs, and savings of labor, chemicals, and other consumables costs (106). It has been suggested that the combustion method provides cost savings of about 30% with a payback period within 2 years. For research institutes, universities, and small-scale laboratories, the safety of modern CNAs probably outweighs cost considerations. Further comparisons of the CNAs and Kjeldahl analysis are summarized in Table 14. Kjeldahl Method 33

5.3. Combustion Analysis of Feeds, Cereal Grains, and Oilseeds Combustion analysis ®rst received AOAC approval for feeds in 1968. The classical instruments (Coleman model 29A nitrogen analyzer) used a manometer for the volumetric assay of nitrogen (110). Comprehensive testing using modern TCD-based CNAs appeared in 1987 (78). A nine- laboratory collaborative trial to determine nitrogen in feeds was successfully completed in 1989 (80). The AOAC approved CNAs for animal feed testing in 1990. The small sample sizes (150±500 mg) used with modern CNAs raised concerns about sampling. Extensive grinding and mixing are essential to ensure sample homogeneity and representative sampling. Sweeney and Rexroad (78) analyzed 14 different animal feeds using the LECO FP-228 instrument with a prescribed sample size of <150 mg. Estimates of feed nitrogen agreed closely with results from Kjeldahl analysis. The precision of analysis was signi®cantly lower (0.013±0.052%) for the combustion method as compared with Kjeldahl analysis (0.006±0.035%). Schmitter and Rihs (79) increased the sample size for the LECO-F228 analyzer from 150 mg to 1 g by palletizing before loading into the instrument port. Adding a few drops of polyethylene (2% w/w in ethyl acetate solvent) prevented ¯aking of the pellets. Table 15 shows nitrogen and protein data for feeds determined using the CNA (78,79). Results have been averaged for samples of sizes 0.15±1.0 g. Protein values are calculated as %N 6 6.25. The results agree favorably with Kjeldahl analysis (Fig. 6). There appeared to be signi®cant positive bias for feedstuffs having < 2% nitrogen (Fig. 7). The bias was ascribed to plant- derived materials containing high levels of nitrate. The Kjeldahl method achieves low recoveries of nitrogen from refractory compounds (75) with N22OorN22N bonds (nitrite; nitrate; oximes; azo-, nitro-, nitroso- compounds). Sachen and Thiex (81) found that CNAs showed a 1.38% (protein) bias for hay samples. They attributed such results to ``atmospheric error'' arising from air being trapped in the interstices of the (¯uffy) hay samples. Compressing samples to remove trapped air led to agreement between the Kjeldahl and CNA results (81). The LECO FP-2000 nitrogen analyzer was not subject to an atmospheric blank because of improvements in instrument design and ef®cient purging of atmospheric gases before sample combustion. Sachen and Thiex examined a range of pelleting equipment and procedures for eliminating the atmospheric blank for the LECO FP-428 instrument. They proposed that powdered cellulose could be analyzed to check for an atmospheric error (81). Not all investigators agree about the nature of the atmospheric error. 34 Chapter 1

TABLE 15 Nitrogen and Protein in Feedstuffs Determined by Combustion Method

Sample N (g kgÀ1) Protein (%)

Straw 0.620 3.88 Corn silage 1.175 7.34 Porc soup 1.185 7.41 Hay 1.280 8.00 Corn grain 1.375 8.59 Barley 1.515 9.47 Grass silage 1.835 11.47 Oats 1.760 11.00 Triticale 1.835 11.47 Wheat 1.985 12.41 Dried grass 2.240 14.00 Cow premix 2.755 17.22 Alfafa pellets 2.770 17.31 Hog feed 3.390 21.19 Broiler ®nisher 3.420 21.38 Milk powder 4.365 27.28 Bone meal 4.325 27.03 Rapeseed 5.710 35.69 Protein conc. 6.045 37.78 Soybean meal 6.485 40.53 Cattle conc. 6.690 20.91 Yeast 6.895 43.09 Peanut meal 8.450 52.81 Meat meal 9.300 58.13 Fish meal 9.960 62.25 Gluten 11.730 73.31 Feather meal 13.610 85.06 Soy protein conc. 14.020 87.63

Bicsak (91) described a collaborative study to extend AOAC-approved status to cereal grains and oilseeds. Seven laboratories analyzed 15 matched pairs of samples (soybean, canola, sun¯ower, wheat, barley, corn, sorghum) having protein levels of 8±13%, 17±23%, or 35±40%. Six of the seven collaborators used the LECO FP-428 instrument. With 210 samples the average protein reading was 28.26% by combustion analysis and 28.01% by Kjeldahl analysis. Repeatability and reproducibility statistics were comparable. A recommendation to extend the AOAC combustion method to cereal grains and oilseeds was approved. The following non-English publications Kjeldahl Method 35

FIGURE 6 Comparison of protein results for feeds determined using the combustion method and the Kjeldahl method. Micro-CNA and macro-CNA refer to the use of 150-mg and 1-g sample sizes with the combustion nitrogen analyzer. List of feedstuffs is given in Table 15. (Data derived from Ref. 78.)

FIGURE 7 Residuals from Fig. 6 showing no systematic differences in results. 36 Chapter 1 describe collaborative tests leading to approved status for combustion analysis of cereal and cereal products including wheat, wheat bran, pasta, sorghum, and maize (111,112).

5.4. Barley, Malt, and Beer The combustion method was subjected to an 11-member interlaboratory trial for brewing grain (rice, barley, malt, spent grain) analysis. Agreement was reached in 1992 to include the Dumas method in the methods of analysis for brewing grains (86). The trial results showed that most commercial CNAs had a linear range of 0±9.5% nitrogen and an LLD of 0.0321%. The repeatability CV ranged from 0.013 to 0.055% compared with a reproducibility of 0.042±0.067%. Further collaborative studies to evaluate CNAs for barley, malt, and beer analysis were reported by the UK Institute of Brewing in 1996 (28). Fifteen of the 25 participating laboratories employed the LECO FP-428 instrument. Another ®ve laboratories used the Foss Heraeus Macro-N apparatus. The range of Kjeldahl techniques used is shown in Table 7. All laboratories examined eight samples each of barley, malt, and beer. CNAs gave slightly higher values for total nitrogen as compared with the Kjeldahl method. Essentially identical results were obtained when results from the Kjel-Foss instrument were omitted. The reproducibility was 0.03±0.07% for barley and 0.036±0.065% for malt analysis. These values were independent of total nitrogen over the range 1.17±1.71% (w/w) (barley), 1.45±2.03% (w/ w) (malt), and 268±1020 mg LÀ1 (beer). Based on such results, the IOB Analysis Committee (UK) approved combustion analysis for use alongside Kjeldahl analysis. However, the precision for beer analysis was low, perhaps because the trial participants lacked expertise with liquid samples. CNAs were judged unsuitable for beer protein determination. CNAs were approved for beer analysis in Europe in 1999 (84,85). In the collaborative trial organized by the IOB and EBC, ®ve samples of beer and malt were analyzed in duplicate by 18 laboratories from the brewing and allied industries. The collaborators used the following CNAs: LECO FP-428 instrument (11 laboratories), LECO FP-2000 (3 laboratories), and Macro-N analyzer (3 laboratories). Glycine, Tris, or EDTA was used as the calibrant. Samples of beer were found to contain 362±1159 mg (nitrogen) LÀ1 and malt samples had 0.534±0.706% nitrogen. Repeatability and reproducibility statistics for beer analysis were deemed satisfactory, leading to method approval. Kjeldahl Method 37

5.5. Milk and Related Dairy Products Despite the recent widespread use of CNAs for food protein analysis, few applications in the dairy ®eld have been published. Wiles et al. (95) described an 11-member interlaboratory study from New Zealand. They compared milk protein analysis by CNAs and the Kjeldahl method. Samples included ultra-heat-treated (UHT) whole milk, infant formulas, whole milk powder, skimmed milk powder, whey protein concentrate, casein, and sodium caseinate. Nine of the eleven laboratories employed the LECO FP- 428 instrument. Results for CNAs agreed closely with Kjeldahl ®ndings. There was no systematic bias associated with the former results. Indeed, no evidence was found for a systematic difference between CNA and Kjeldahl results reported between 1968 and 1997 (95).

5.6. Baby Foods and Infant Formulas Bellemonte et al. (99) analyzed ®ve categories of baby foods using the Carlo Erba model 1500 nitrogen analyzer. This instrument uses a high furnace temperature (18008C) combined with an oxygen-rich atmosphere to achieve complete sample combustion. Nitrogen is quanti®ed using a TCD. The analysis time for this instrument is reportedly 3 minutes. All sample types were analyzed successfully. Results obtained by the Kjeldahl method were 1±4% lower than those obtained with CNAs (Table 16). Results from both techniques compared favorably with protein values declared by food manufacturers. Compared with the Kjeldahl method, CNAs are convenient for baby food analysis. The sample throughput and safety considerations favor the Dumas method as described in Sec. 5.2.

5.7. Meat King-Brink and Sebranek (100) described a 12-laboratory trial to evaluate CNAs for meat product analysis. Participants in the trial used the LECO FP-428 instrument (9 laboratories), the Foss Heraeus Macro-N analyzer (2 laboratories), or the Perkin Elmer PE2410 analyzer. In all, 15 pairs of meat products, purchased from 30 different manufacturers, were analyzed. All participants used CNAs satisfactorily judging from (a) the low number of data outliers and (b) the high precision of results for standard compounds. The CNA results were slightly higher than Kjeldahl ®gures: 15.75% versus 15.59% (w/w). Estimates for repeatability and reproducibility were comparable. A recommendation that the Dumas method should be adopted as a reference test for meat proteins was approved by the AOAC. A 14- laboratory trial for analysis of meat and meat products was reported in 38 Chapter 1

TABLE 16 Analysis of Protein in Five Categories of Baby Foods (97)

Declared Samplea protein Kjeldahlb Dumasb

1. Formula milk (7) 15.55 14.94 15.36 2. Cereal-based products (6) 10.28 10.12 10.35 Cream of rice, semolina honey, wheat ¯our milk oats, milk soup cereal fruit, milk soup cereal apples 3. Biscuits (2) 9.50 9.40 14.70 4. Lyophilized products (2) 54.20 52.15 53.20 Veal, ham, and eggs 5. Homogenized products (6) 9.52 9.62 9.72 Beef, beef ham, chicken, veal brain, turkey, chicken carrot potatoes aNumbers in parentheses represent number of different foods in the category analyzed. Protein values are averaged for each food category b Protein values were calculated with a conversion factor (Fk) 6.38 for milk formula or 6.25 for all other categories of baby food.

Germany. This trial, too, concluded that the performance of the Dumas method was comparable to that of the Kjeldahl assay but that the former method was quicker and more environment friendly (101).

REFERENCES

1. WH Tallent. USA current developments in protein food regulationsÐ labeling. J Oil Chem Soc 56(3):239, 1979. 2. DK Salunkhe, SS Deshpande. Foods of Plant Origin. Production, Technology and Human Nutrition. New York: Van Nostrand Reinhold, 1991. 3. H Frankel-Conrat, M Cooper. The use of dyes for the determination of acid and basic groups in proteins. J Biol Chem 154:239±340, 1944. 4. DC Udy. Estimation of protein in wheat and ¯our by ion-binding. Cereal Chem 33:190±197, 1956. 5. AJ Pinckney. Wheat protein and the biuret reaction. Cereal Chem 26:423±439, 1949. 6. Z Duda, M Szot. A comparison of several methods of protein determination in pig blood plasma. Proceedings of the European Meeting of Meat Research Workers 32, Vol II 9, 1986, pp 447±450. Kjeldahl Method 39

7. KM Williams, P Fox, T Marshall. A comparison of protein assays for the determination of the protein concentration of beer. J Inst Brew 101:365±369, 1995. 8. Y Pomeranz, RB More, FS Lai. Reliability of ®ve methods for protein determination in barley and malt. J Am Soc Brew Chem 35:86±93, 1975. 9. Y Pomeranz, RB More. Reliability of several methods for protein determination in wheat. Bakers Digest February:44 ±58, 1975. 10. QQ Luthi-Peng, Z Puhan. The 4th derivative UV spectroscopic method for the rapid determination of protein and casein in milk. Milchwissenschaft 54:74± 77, 1999. 11. JM Lynch, DM Barbano. Kjeldahl nitrogen analysis as a reference methood for protein determination in dairy products. J Assoc Off Anal Chem Int 82:1389±1398, 1999. 12. RB Bradstreet. The Kjeldahl Method for Organic Nitrogen. London: Academic Press, 1965. 13. International Standard ISO-1871 (1975). Agricultural food productsÐgeneral directions for the determination of nitrogen by the Kjeldahl method. 14. L Gaspar. General laboratory methods. In I Kerese, ed. Methods of Protein Analysis. Chichester, UK: Ellis Horwood, 1984, pp 30±86. 15. BG Osborne. The determination of protein in cereals. In BJF Hudson, ed. Developments in Food ProteinsÐ4. London: Elsevier Applied Science, 1986, pp 247±290. 16. RA Osborne, JB Wilkie. A study of the Kjeldahl method IV. Metallic catalysts and metallic interferences. J Assoc Off Anal Chem 18:604±609, 1935. 17. RC Benedict. Determination of nitrogen and protein content of meat and meat products. J Assoc Off Anal Chem 70:69±76, 1987. 18. R Tkachuk. Nitrogen-to-protein conversion factors for cereal and oilseed meals. Cereal Chem 46:419±423, 1969. 19. OA Krober, SJ Gibbons. Nonprotein nitrogen in soybeans. J Agric Food Chem 10:57±58, 1962. 20. R Tkachuk. Note on the nitrogen-to-protein conversion factor for wheat ¯our. Cereal Chem 43:223, 1966. 21. JAD Ewart. Amino acid analyses of cereal ¯our proteins. J Sci Food Agric 18:558, 1967. 22. S Boisen, S Bech-Andersen, B Eggum. A critical view on the conversion factor 6.25 from total nitrogen to protein. Acta Agric Scand 37:299±304, 1987. 23. MAJS van Boekel, B Ribadeau-Dumas. Addendum to the evaluation of the Kjeldahl factor for conversion of the nitrogen content of milk and milk products to protein content. Neth Milk Dairy J 41:281±284,1987. 24. FW Sosulski, GI Ima®don. Amino acid composition and nitrogen-to-protein conversion factors for animal and plant foods. J Agric Food Chem 38:1351± 1356, 1990. 25. AK Kaul, TR Sharma. Rapid determination of nitrogen content in grain-meal samples with alkaline-phenol reaction, manually and with an AutoAnalyzer. Z Anal Chem 280:133±138, 1976. 40 Chapter 1

26. M Mohyuddin, G Mazza. Determination of potato protein by alkali-phenol, dye-binding and other methods. Am Potato J 55:621±626, 1978. 27. BG Venter, HP Sheepers, J Floor, JH Snyman. Semi-micro Kjeldhal procedures for protein determination of dairy products. S Afr J Dairy Technol 17(3):107±111, 1985. 28. GK Buckee. Determination of total nitrogen in barley, malt and beer by Kjeldahl procedures and the Dumas combustion methodÐcollaborative trial. J. Inst Brew 100:54±64, 1994. 29. MJ Brennan. Automated Kjeldahl yields poetry. Food Eng 46(5):102±103, 1974. 30. JE Trevis. Seven automated instruments. Cereal Sci Today 19(5):182±189, 1974. 31. R Oberrieth, NH Mermelstein. Instrument automates, accelerates nitrogen determinations. Food Technol 28(6):40±41, 43, 1974. 32. PC Williams, KH Norris, RL Johnsen, K Standing, R Fricioni, D MacAffrey, R Mercier. Comparison of physicochemical methods for measuring total nitrogen in wheat. Cereal Foods World 23(9):544±547, 1978. 33. OC Bjarno. Kjel-Foss automatic analysis using an antimony-based catalyst: collaborative study. J Assoc Off Anal Chem 63:657±663, 1980. 34. DL McGill. Comparison of automated method and improved AOAC Kjeldahl method for determination of protein in meat and meat products. J Assoc Off Anal Chem 64:29±31, 1980. 35. FB Suhre, PA Corrao, A Glover, AJ Melanoski. Comparison of three methods for determination of crude protein in meat: collaborative study. J Assoc Off Anal Chem 65:1339±1345, 1982. 36. JF Marten, G Catanzaro. Fundamental studies in automatic nitrogen digestion. Analyst 91:42±47, 1966. 37. JA Varley. Automatic methods for the determination of nitrogen, phosphorus and potassium in plant material. Analyst 91:119±126, 1966. 38. LL Wall, CW Gehrke. An automated total protein nitrogen method. J Assoc Off Anal Chem 58:1221±1226, 1975. 39. DE Uhl, EB Lancaster. Automation of nitrogen analysis of grain and grain products. Anal Chem 43:990, 1971. 40. KR Vincent, WF Shipe. The effect of calibration procedures on accuracy and precision of automated Kjeldahl nitrogen analysis in some formulated foods. J Food Sci 41:157±162, 1976. 41. WM Gantenbein. Collaborative study of the automated determination of nitrogen in meat products. J Assoc Off Anal Chem 56:31±35, 1973. 42. HG Lento, CE Daugherty. Automated determination of protein nitrogen in foods. Food Prod Dev 5:86, 1971. 43. HG Lento, CE Daugherty. The automated protein-nitrogen analysis of foods. Adv Auto Anal Technicon Int Congr 2:75±80, 1970. 44. JF Marten, G Catanzaro. Fundamental studies in automatic nitrogen digestion. Analyst 91:42±47, 1966. Kjeldahl Method 41

45. J Davidson, J Mathieson, AW Boyne. The use of automation in determining nitrogen by the Kjeldahl method, with ®nal calculation by computer. Analyst 95:181±193, 1970. 46. CW Gehrke, LL Wall Sr, JS Absheer. Automated nitrogen method for feeds. J Assoc Off Anal Chem 56:1096±1105, 1973. 47. LL Wall, CW Gehrke. Feeds: an automated total protein nitrogen method. J Assoc Off Anal Chem 58:1221±1226, 1975. 48. WT Bolleter, CJ Bushman, PW Tidwell. Spectrophotometric determination of ammonia as indophenol. Anal Chem 33:592±594, 1961. 49. JA Tettlow, AL Wilson. An absorptiometric method for determining ammonia in boiler feed-water. Analyst 89:453±465, 1964. 50. CW Gherke, FE Kaiser, JP Ussary. Automated spectrophotometric method for nitrogen in fertilizers. J Assoc Anal Chem 51:200±211, 1968. 51. PL Searle. The Berthelot or indophenol reaction and its use in the analytical chemistry of nitrogen. Analyst 109:549±568, 1984. 52. JE McNeal, A Karasz, E Gorge Jr. Automation of methods for meat and meat products. 1. Determination of protein. J Assoc Anal Chem 53:907±910, 1970. 53. G Mohyuddin, G Mazza. Determination of potato protein by alkali-phenol, dye-binding and other methods. Am Potato J 55:621±626, 1978. 54. CC Hach, SV Brayton, AB Kopelove. A powerful Kjeldahl nitrogen method using peroxymonosulfuric acid. J Agric Food Chem 33:1117±1123, 1985. 55. MB Devani, CJ Shishoo, SA Shah, BN Suhagia. Spectrophotometric method for micro-determination of nitrogen in Kjeldahl digest. J Assoc Off Anal Chem 72:953±956, 1989. 56. S Jacobs. The determination of nitrogen in organic compounds by the indanetrione hydrate method. Analyst 85:257±264, 1960. 57. S Jacobs. The effect of temperature on the Kjeldahl digestion process. Analyst 89:489±494, 1964. 58. JR Quinn, JGA Boisvert, I Wood. Semi-automated ninhydrin assay of Kjeldahl nitrogen. Anal Biochem 58:609±614, 1974. 59. ND Heidelbaugh, CS Huber, JF Bednarczyk, MC Smith, PC Rambaut, HO Wheeler. Comparison of three methods for calculating protein content of foods. J Agric Food Chem 23:611±612, 1975. 60. H-J Horstmann. A precise method for the quantitation of proteins taking into account their amino acid composition. Anal Biochem 96:130±138, 1979. 61. GL Peterson. Determination of total protein. Methods Enzymol 91:95±119, 1983. 62. JC Weaver, M Kroger, LR Kneebone. Comparative protein studies (Kjeldahl, dye binding, amino acid analysis) of nine strains of Agaricus bisporus (Lange) imback mushrooms. J Food Sci 42:364±366, 1977. 63. A Braaksma, DJ Schaap. Protein analysis of the common mushroom Agaricus bisporus. Postharvest Biol Technol 7:119±127, 1996. 64. CG Zarkadas, EA Meighen, GC Zarkadas, CN Karatzas, AD Khalili, JA Rochemont, M Berthelet. Determination of the myo®brillar and connective 42 Chapter 1

tissue proteins in the bovine diaphragm. J Agric Food Chem 36:1105±1108, 1988. 65. CG Zarkadas, N Karatzas, AD Khalili, S Khanizadeh, G Morin. Quantitative determination of the myo®brillar proteins and connective tissue content in selected porcine skeletal muscles. J Agric Food Chem 36:1131±1146, 1988. 66. CN Karatzas, CG Zakardas. Determination of the myo®brillar and connective tissue protein contents and amino acid composition of selected composite meat products. J Agric Food Chem 36:1109±1121, 1988. 67. CG Zarkadas, NJ Drouliscos, CN Karatzas. Comparison of the total protein, nitrogen and amino acid composition of selected additives and ingredients used in composite meat products. J Agric Food Chem 36:1121±1131, 1988. 68. Q Nguyen, MA Fanous, LH Kamma, AD Khalili, PH Schuepp, CG Zarkadas. Comparison of the amino acid composition of two commercial porcine skins (rind). J Agric Food Chem 34:565±572, 1989. 69. CN Karatzas, CG Zarkadas. Comparison of the amino acid composition of the intracellular and extracellular matrix protein fractions isolated from avian skeletal muscles. Poult Sci 68:811±824, 1989. 70. S Khanizadeh, D Buszar, CG Zarkadas. Comparison of three methods for calculating protein content in developing apple ¯ower buds. J Assoc Off Anal Chem 75:734±737, 1992. 71. CG Zarkadas, Y Ziran, GC Zarkadas, A Minero-Amador. Assessment of the protein quality of beefstock bone isolates for use as an ingredient in meat and poultry products. J Agric Food Chem 43:77±83, 1995. 72. CG Zarkadas, HD Voldeng, YI Ziran, Keijin-Shang, PL Pattsion, Compar- ison of the protein quality of ®ve new northern adapted natto soybean cultivars by amino acid analysis. J Agric Food Chem 45:2013±2019, 1997. 73. S Khanizadeh, D Buszard, CG Zarkadas. Misuse of the Kjeldahl method for estimating protein content in plant tissue. HortScience 30:1341±1342, 1995. 74. R Fiedler, G Proksch, A Koepf. The determination of total nitrogen in plant materials with an automatic nitrogen analyzer. Anal Chim Acta 63:435±443, 1973. 75. DW Nelson, LL Sommers. Total nitrogen analysis of soil and plant tissue. J Assoc Off Anal Chem 63:770±778, 1980. 76. YS Lee, CE Damon, JP Crisler. A micro method for the determination of protein and screening of added water in meat and meat products. Mikrochim Acta 4:477±483, 1972. 77. TL Lunder. Determination of total nitrogen in foodstuffs according to Dumas, by means of the Micro-Rapid N automatic analyzer. Lab Pract 23(4):170±172, 1974. 78. RA Sweeney, PR Rexroad. Comparison of LECO FP-228 ``Nitrogen Determinator'' with AOAC copper catalyst Kjeldahl method for crude protein. J Assoc Off Anal Chem 70:1028±1030, 1987. 79. BM Schmitter, R Rihs. Evaluation of a macrocombustion method for total nitrogen determination in feedstuffs. J Agric Food Chem 37:992±994, 1989. Kjeldahl Method 43

80. RA Sweeney. Generic combustion method for determination of crude protein in feeds: collaborative study. J Assoc Off Anal Chem 72:770±774, 1989. 81. RW Sachen, NJ Thiex. Effect of sample introduction and atmospheric blank on determination of nitrogen (crude protein) by combustion. J Assoc Off Anal Chem Int 80:14±19, 1997. 82. DF Tate. Determination of nitrogen in fertilizer by combustion: collaborative study. J Assoc Off Anal Chem 77:829±839, 1994. 83. American Society of Brewing Chemists. Report of Sub-Committee on Nitrogen in wort and beer by combustion method. J Am Soc Brew Chem 51(4):183±185, 1993. 84. BA Johnson, CG Johansson. Determination of total soluble nitrogen content of malt and beer by the Dumas combustion method: collaborative trial. J Inst Brew 105:360±364, 1999. 85. BA Johnson, CG Johansson. Determination of total soluble nitrogen of malt and beer by the Dumas combustion method: collaborative trial. Monatsschr Brauwiss 53(3/4):50±53, 2000. 86. American Society of Brewing Chemists. Report of Sub-Committee on Total Nitrogen in brewing grains by combustion method. J Am Soc Brew Chem 50:147±148, 1992. 87. GK Buckee. Review of methods for the measurement of total nitrogen. Ferment 8:357±361, 1995. 88. L Krotz, L Ragalia, F Adreolini. Nitrogen/protein determination in beer, wort and malt by combustion method as the alternative to Kjeldahl method. Current status and future trends. Proceedings of EURO FOOD CHEM VIII, Vol 3, Vienna, September 18±20, 1995, p 752. 89. CG Johansson. Determination of total nitrogen in barley and malt by combustion method. Collaborative trial. Monatsschr Brauwiss 49(11/12):326± 328, 1996. 90. SAGF Angelino, HPM van Laarhoven, JJM van Westerop, BM Broekhuijse, HCM Mocking. Total nitrogen content in single kernel malting barley samples. J Inst Brew 103(1):41±46, 1997. 91. RC Bicsak. Comparison of Kjeldahl method for determination of crude protein in cereal grains and oil seeds with generic combustion method: collaborative study. J Assoc Off Anal Chem 76:780±786, 1993. 92. R Bicsak. FGIS method of measuring protein. Bull Assoc Oper Millers August:6603±6606, 1995. 93. P Williams, D Sobering, J Antoniszyn. Protein testing methods at the Canadian Grain Commission. Proceedings of the Wheat Protein Symposium, Saskatoon, Saskatchewan, March 9 and 10, 1998 [conference paper online], October 19, 1998. World Web Web citation. Available from:http:// www.cgc.ca/Pubs/confpaper/Williams/ProteinOct98/protein1-e.htm. 94. PG Wiles, IK Gray. A collaborative trial for the establishment of a skim milk powder reference protein standard. Aust J Dairy Technol 51:17±21, 1996. 44 Chapter 1

95. PG Wiles, IK Gray, RC Kissling. Routine analysis of proteins by Kjeldahl and Dumas methods: review and interlaboratory study using dairy products. J Assoc Off Anal Chem Int 81:620±632, 1998. 96. EH Simonne, AH Simonne, RR Eitenmeiller, HA Mills, CP Creswell. Could the Dumas method replace the Kjeldahl digestion for nitrogen and crude protein determinations in food? IFT Annual Meeting, 1995, p 241. 97. EH Simonne, AH Simonne, RR Eitenmeiller, HA Mills, CP Cresman III. Could the Dumas method replace the Kjeldahl digestion for nitrogen and crude protein determinations in foods? J Sci Food Agric 73:39±45, 1997. 98. CJ Huang, R McDonald, RS Lyon, E Elkins. Comparison of Kjeldahl method for crude protein determination in fruit and vegetable products with combustion method. IFT Annual Meeting, 1995, p 198. 99. G Bellemonte, A Costantini, S Giammorioli. Comparison of modi®ed automatic Dumas method and the traditional Kjeldahl method for nitrogen determination in infant food. J Assoc Off Anal Chem 70:227±229, 1987. 100. M King-Brink, JG Sebranek. Combustion method for determination of crude protein in meat and meat products: collaborative study. J Assoc Off Anal Chem 76:787±793, 1993. 101. R Buschmann, K Westphal. Determination of the nitrogen content of meat and meat products with the Dumas method. Report from the Paragraph 35 LMBG working group `Fleischerzeugnisse' of the BgVV. Fleischwritschaft 81:82±84, 2001. 102. JK Duan, DR DeClercq. Comparison of combustion and Kjeldahl methods for determination of nitrogen in oilseeds. J Am Oil Chem Soc 71:1069±1072, 1994. 103. DL Berner, J Brown. Protein nitrogen combustion method collaborative study. I. Comparison with Smalley total Kjeldahl nitrogen and combustion results. J Am Oil Chem Soc 71:1291±1293, 1994. 104. MW Young, DKL Mackerron, HV Davies. Calibration of near infrared re¯ectance spectroscopy to estimate nitrogen concentration in potato tissues. Potato Res 40:215±220, 1997. 105. WJ Youden, EH Sliener. Statistical Manual of the AOAC. Arlington, VA:AOAC, 1975. 106. Anonymous. Protein: the Dumas technique. Food Rev 21(4):39, 1994. 107. PR Wilson, A new instrumental method in food protein analysis. Food Tech Int Eur 181±183, 1994. 108. JRN Taylor. Out with the old and in with the older. Food Ind 48(5):35, 37, 1995. 109. R Marsili. Don't waitÐautomate: revamping labs yields speedy results. Food Prod Design 7(3):71±72, 74, 79±80, 83±84, 1997. Kjeldahl Method 45

110. ME Ebeling. The Dumas method for nitrogen in feeds. J Assoc Off Anal Chem 51:766±770, 1968. 111. R Winker, S Botterbrodt, E Rabe, MG Linhauer. Nitrogen/protein determination in wheat and wheat products (¯our and meal) with the Dumas method. Getreide Mehl Brot 54(2):86±91, 2000. 112. MG D'Egidio, C Cecchini, G Novembre. Comparison of Dumas and Kjeldahl methods for determination of crude protein in cereal. Tec Molitoria 50:1189± 1195.

The Alkaline Copper Reagent: Biuret Assay

1. INTRODUCTION

Copper (Cu2) ions react with proteins to form a blue complex at high pH. The intensity of the color is proportional to the amount of protein present. The oldest copper binding technique for protein analysis is called the biuret method. The name, although quite deeply entrenched in the literature, is unfortunate. The assay has little to do with biuret. A more accurate name is the alkaline copper reagent (ACR) test. Older literature features almost exclusively the ``biuret reagent.'' Depending on the context, both these names will be used. First we will discuss the history and developments of the ACR assay. Modern and (where relevant) old protocols involving ACR are described. Applications in food protein analysis are then considered.

Berthelot reported that biuret (NH2CONHCONH2) forms a blue complex with copper ions under alkaline conditions. Proteins were reported to undergo a similar reaction in 1873. Early developments of the biuret method are described by Hiller (1) and also by Robinson and Hodgen (2). The two publications are some of the earliest accounts of the ACR assay. Blood plasma was diluted with 25 volumes of 0.9% saline and adjusted to 10% (w/v) TCA. The protein precipitate formed was recovered by ®ltration and redissolved in 3% sodium hydroxide solution (5 mL). Then copper

47 48 Chapter 2 sulfate solution (20% w/w; 2.5 mL) was added. After 1±2 hours the brown precipitate (Cu2 hydroxide) produced was removed by centrifugation. A linear relationship was found between DA560 readings and protein À1 concentration (0±0.2 mg mL ). The assay sensitivity was 6 DA560 units per (mg mLÀ1) protein. The formation of copper (hydro)oxide precipitate was a nuisance. Samples had to be centrifuged before DA560 readings could be taken.

2. THE ALKALINE COPPER REAGENT PROTEIN ASSAY

Mehl (3) added ethylene glycol to ACR to arrest the formation of Cu2 hydroxide. Ethylene glycol (100 mL), 60% (w/v) sodium hydroxide (40 mL), and 4% (w/v) copper sulfate were mixed and the solution was diluted to 400 mL with water. The mixture was heated to precipitate copper hydroxide and then ®ltered. Extra sodium hydroxide was added to bring the ®nal concentration to 10±11% (w/w). The resulting ACR was stable for several months. Dissolved egg albumin ( 3mgmLÀ1) was analyzed by mixing with an equal volume of ACR and taking DA550 readings after 90 minutes. The blue color, once formed, was stable for about 20 hours. Sols (4) used glycerin as a stabilizer for ACR. The amount of glycerin required was 100 times lower than ethylene glycol. ACR solution was prepared by mixing glycerin (2 mL), copper sulfate (5% w/v; 80 mL), and sodium hydroxide (20% w/v; 400 mL) and making up to 1 L with water. To perform analysis, 1 mL of test solution (< 1±2% w/v protein) was mixed with an equal volume of ACR solution. Colorimetric readings were recorded at two wavelengths, accounting for the unpopularity of the procedure. Weichselbaum (5) produced a modi®ed ACR by adding potassium sodium tartrate as the stabilizer for Cu2. Gornall et al. (6) evaluated several modi®ed ACR solutions, ®nding Mehl's copper reagent dif®cult to reproduce. Weichelsbaum also considered the effect of changing copper sulfate and sodium hydroxide concentrations and the reaction temperature on color formation. This work revealed some tolerance for variations in regent composition and experimental conditions. The ACR developed by Weichselbaum was employed by Gornall et al. for protein analysis in clinical samples. This variant ACR became more popular after Layne's account of it appeared in Methods in Enzymology in 1957 (7). ACR is now widely used for food protein analysis. A summary of this popular method is given next. The Alkaline Copper Reagent: Biuret Assay 49

Method 1 Protein analysis using alkaline copper reagent (5,6) Reagents 1. Copper sulfate CuSO4 Á 5H2O 2. Rochelle salt (potassium sodium tartrate) 3. Sodium hydroxide (10% w/w solution) Preparation of ACR. Dissolve copper sulfate (1.5 g) and potassium sodium tartrate (6 g) in 500 mL of water. Add 300 mL of sodium hydroxide (10% w/w) solution with constant mixing. Adjust the total volume to 1 L with distilled water. This solution should keep inde®nitely. Discard if it shows signs of forming a black or reddish precipitate. The ®nal concentrations of copper sulfate and sodium hydroxide are 0.015% and 3% w/v, respectively. Procedure Add 8 mL of ACR solution to 2 mL of protein solution. Mix and allow

to react for 30 minutes. Record DA540 using a 1-cm cuvette. Dilute plasma (protein) in 0.9% saline before analysis. To separate the albumin and globulin, dilute 0.5 mL of plasma with 9.5 mL of sodium sulfate solution (22.6% w/v). Remove 2 mL of the solution for biuret analysis of total plasma protein. Add 3 mL of ether to the remaining solution, mix for 30 seconds, and centrifuge. Remove 2 mL of the supernatant and analyze to ®nd the plasma albumin concentration. The difference (total protein À plasma albumin) is the plasma globulin concentration. The linear range for protein analysis with Method 1 was 1±10% (w/v) protein. The results were strongly correlated with Kjeldahl nitrogen.

TABLE 1 Effect of Different Proteins on the Color Yield from the ACR Assay

Protein Relative color intensity (%)

Serum albumin 100 Serum globulin 98.8 Egg albumin 98.6 Casein 88.1 Gelatin 74.2 Zein 92.1

Source: Ref. 6. 50 Chapter 2

Different proteins gave slightly different color yields (Table 1). Casein, zein, and gelatin produce low amounts of color compared with serum albumin. The biuret method cannot be used in the presence of ammonium salts. The ACR formulation is one of the most popular modi®cations of the biuret reagent and is the standard method for clinical samples (8). The ACR formulation is also widely used for the analysis of animal proteins.

3. CHEMISTRY OF THE ALKALINE COPPER REAGENT PROTEIN ASSAY

Strickland et al. (9) titrated proteins with increasing amounts of ACR. After each addition, the absorption spectrum was recorded between 425 and 850 nm. Sparingly soluble proteins were dissolved in 8 M urea, which had no deleterious effects on color formation. By plotting the maximum absorbance change versus copper concentration, the maximum amount of protein required to bind all Cu2 was found.

Table 2 shows that lmax for ACR-protein mixtures is 535±590 nm. The extinction coef®cient for the Cu2-protein complex (De,MÀ1 cmÀ1 with respect to protein concentration) increased with the molecular weight and the number of peptide groups per protein. The amount of Cu2 bound to a ®xed weight of protein is shown in Table 2. Note that gliadin and gelatin exhibit low color yields per unit weight of protein. This characteristic is partly related to the high proline content in these proteins. Other amino acid side chains also alter the speci®c color yield from protein-Cu2 binding. Cysteine binds copper strongly, forming Cu-mercaptide. Therefore cysteine- rich proteins show a reduced color yield during ACR analysis. High concentrations of glutamate and aspartate compete with the peptide backbone for Cu2 binding. Cu2 binding with biuret, glycinamide, and glycine oligopeptides was reviewed by Sigel and Martin (10). Cu2 binding re¯ects the acid-base properties of the peptide bond. Delocalization of the peptide nitrogen lone- pair electron leads to a planar geometry with partial double-bond character. The resulting canonical structures have a high electron density at the peptide oxygen. The amide hydrogen acts as a weak acid, becoming ionized only under very alkaline conditions (Fig. 1). Binding of Cu2 to the peptide group occurs via oxygen at neutral pH. In a highly alkaline medium (pH 13), complexation involves the nitrogen group. Under alkaline conditions, 2 moles of biuret bind with each Cu2 atom, forming a six- member ring (Fig. 2). Cu2 forms a tetradentate complex. The amide nitrogen becomes deprotonated in the process. Peptide-metal ion binding at high pH values occurs in competition with metal ion hydrolysis. Cu2 h laieCpe egn:Bue sa 51 Assay Biuret Reagent: Copper Alkaline The

TABLE 2 Selected Parameters for the ACR Analysis

Cu atoms Test Molecular Á"a Peptides "c (mole "d b À1 sample weight max per M [P] (g/Cu) per Cu per M ligand) per kg Biuret 103 515 27.8 206 4.0 13.9 0.5 134.95 Triglycine 189 590 90.3 189 3.0 90.3 1.0 477.78 Gliadin 27,000 555 143.4 700 6.0 5,531.1 38.6 204.86 b-Lacto- 36,800 544 151.4 620 6.0 8,986.3 59.4 244.19 globulin Zein 45,000 535 150.0 638 6.0 10,579.9 70.5 235.11 Gelatin 75,000 554 144.2 670 6.3 16,141.8 111.9 215.22 Serum 180,000 545 154.0 586 5.0 47,303.8 307.2 262.80 globulin Edestin 310,000 550 163.6 635 6.0 79,867.7 488.2 257.64 aExtinction coef®cient with respect to Cu2 (MÀ1 cmÀ1 ). bGrams of protein bound to 1 mole of Cu2 . cExtinction coef®cient with respect to protein (MÀ1 cmÀ1 ). dExtinction coef®cient with respect to protein (kgÀ1 cmÀ1 ). Source: Adapted from Ref. 9. 52 Chapter 2

FIGURE 1 Resonance hybrid forms of the peptide bond. Structures (1) and (2) predominate under physiological conditions at a ratio of 60:40. At neutral pH the electron density is highest at the oxygen atom. At the pH of the biuret reaction (pH 14), deprotonation of the amide nitrogen leads to signi®cant amounts of (3) with high electron density at the nitrogen atom. binding with the OHÀ ion leads to an insoluble precipitate. A model for the 2:1 biuret- Cu2 complex at high pH is shown in Fig. 2. Three kinds of Cu2 complexes are formed with polypeptides. At low pH Cu2 interacts exclusively with side-chain residues forming ``type S'' complexes. At intermediate pH values there is binding to both side chains

2 FIGURE 2 Schematic diagram of the Cu -(biuret)2 complex. The Alkaline Copper Reagent: Biuret Assay 53 and deprotonated amide nitrogen, forming type SP complexes. Lastly, when pH >10±12, the Cu2 interacts mainly with four peptide nitrogens to form a biuret-like (type B) complex. Cu2 binding does not involve consecutive amino acid residues. Interactions leading to type B complexes are incompatible with regular (a-helical or b-sheet) forms of polypeptide secondary structure. Interestingly, some (type B) polypeptide complexes with Cu 2 have rudimentary enzymatic (peroxidase and catalase) activity (10). The interactions just described can be usefully contrasted with Cu2 binding to native proteins, where binding occurs at the N-terminal region.

4. INTERFERENCE COMPOUNDS

ACR dissolves extraneous plant colors and dyes with a high absorbance at 530±550 nm (11).* To overcome this problem during ACR analysis, Jennings (11) diluted samples with high backgroud readings 100-fold using standard Lowry reagent A (see Chapter 4). The samples were then analyzed by treating with Lowry reagent B (11). Pinckney (12) reduced the interference from phopholipid by adding carbon tetrachloride to ground meals before analysis. Johnson and Craney (13) noted that isopropanol reduced the interfering effects from plant dyes. Some of the substances capable of interfering with the ACR method are listed in Table 3. Possible interferences during wheat protein analysis using ACR were investigated by Mitsuda and Mitsunaga (14). They extracted whole wheat ¯our or bran with 0.5% (w/v) sodium hydroxide. The alkali-soluble extracts were identi®ed as starch, glucose, pigment, and lipids (Table 4). The effects of each component were evaluated by adding known amounts to model solutions consisting of 0.5 or 5 mg mLÀ1 ovalbumin and then assaying the mixture by the ACR method. Glucose (<5mgmLÀ1) reduced the color yield for the ACR assay, probably by competing with the peptide group for Cu2 binding. The interference was less pronounced when colorimetric readings were taken after 30 minutes rather than 2.5 hours. The Cu2 reaction with glucose takes place more slowly than the Cu2 binding to protein. The different rates of reaction offer an opportunity for improving analytical performance (see later). Lipid and starch suspensions cause light scattering and an apparent increase in absorbance readings. Starch is also hydrolyzed

* Examples of strongly colored grain include barely and oats, which have blue or black chaff (pericarp, aleurone layer, and outer endosperm). 54 Chapter 2

TABLE 3 Some Interfering and Noninterfering Compounds for ACR Protein Assay

Reactive compounds

Colors Nonpeptide compounds Anthocyanins and other ¯avonol DNA Lycopenes (carotenoids and related) Glucose PigmentsÐchlorophyll and related Glucoseamine Starch Amino acids (histidine and cysteine) Biopolymers Buffer components Lignin and low molecular-weight Ethanolamine, ethylenediamine, Tris derivatives

Unreactive compounds

Amino acids Amides Gly, Tyr, Arg, Met, Phe, Asp, Glu, Acetamide, sulfanilic acid, Iso, Lys dimethylformamide, urea Organic acids N-methylacetamide Formic, acetic, lactic Amines Ammonia, ethylamine Others Creatine, betaine, EDTA

Source: Compiled from various sources.

to glucose during high-temperature ACR assays. For small amounts of sample, protein determinations can be performed after defatting by Soxhlet extraction or by stirring with cold acetone or hexane.

TABLE 4 Quantities of Alkali-Soluble Components Extracted from Wheat (mg gÀ1)a

Component Whole wheat Bran

Glucose 9.0 8.2 Starch 670.0 120.0 Protein 130.0 140.0 Pigment 0.0 7.4 Lipid 121.0 32.5 aA 1-g sample was extracted with 0.5% NaOH (100 mL) (14). The Alkaline Copper Reagent: Biuret Assay 55

TABLE 5 Effect of Hydrogen Peroxide Addition on ACR Assay of Colored Grain

Regression Sample/method equationa R SE (%)

Barleyb Y 13:3x 7:73 0.54 1.24 Wheat bran Y 40:8x 3:60 0.79 1.18 hydrogen peroxideb Barleyc Y 49:3x À 0:48 0.99 0.21 Wheat branc Y 58:5x À 0:14 0.98 0.28 a Y % Kjeldahl protein, x A550. bAnalysis by the method of Johnson and Craney (13). c Analysis after pretreatment with CCl4 and mixture made 1% with repeat to H2O2 for 30 minutes before A550 readings.

5. SAMPLE PRETREATMENT AND AVOIDING INTERFERENCES

Pinckney (12) successfully analyzed full-fat ¯our after precipitating lipid with carbon tetrachloride. The presence of 10% (v/v) tetrachloromethane or dichloroethane also reduced the solubilization of wheat starch in 0.5% (w/v) sodium hydroxide by 90%. These and other organic solvents (CF3CCl3, cyclohexane) probably exert their effect by complexing with starch. Chloroform, 1,2-dibromomethane, n-hexane, n-octane, and n-decane had no effect. Strategies for dealing with extraneous plant colors were also investigated by Mitsuda and Mitsunaga (15). Barley meal and wheat bran protein were analyzed by the method of Johnson and Craney (13). Colorimetric measurements were calibrated with the Kjeldahl method. All analytical parameters (sensitivity, linearity, standard error of analysis) were altered by the strong background color associated with barley grain and wheat bran. Adding hydrogen peroxide (1% v/v ®nal concentration) about

30 minutes before A550 readings improved assay sensitivity and LLD (Table 5). Jennings (11) and Williams (16) suggested that strongly colored grains cannot be analyzed using the ACR assay. They found ACR results were in poor agreement with Kjeldahl results for barley and wheat bran. Hydrogen peroxide provides a key ingredient needed for the Fenton reaction.

2 À 1 H2O2 Cu ? HO OH Cu 1 56 Chapter 2

TABLE 6 Sensitivity of the UV Biuret Method for a Range of Proteins

a Protein ÁA263 Insulin 5.2 Bovine serum globulin 6.7 Bovine serum albumin 5.3 Histone 6.3 Ribonuclease 5.9 Ovalbumin 5.0

a Protein ÁA310 Gelatin 1.56 Histone 1.71 Casein (Hammersten) 1.74 Ovalbumin 1.98 Trypsin 2.17 Bovine serum albumin 2.21 Lysozyme 2.20 aValue for 1 mg mLÀ1 protein solution.

The highly reactive hydroxyl radical HO catalyzes the oxidative destruction of extraneous plant colors.

6. THE MICRO-BIURET OR ULTRAVIOLET BIURET PROTEIN ANALYSIS

The UV ACR assay is performed at 255±320 nm. The technique is 10- to 15- fold more sensitive than the conventional ACR protein analysis. The use of UV absorption measurements in conjunction with the ACR method dates back to 1957. The method became popular owing to its rediscovery by Ellman (17). Itzhaki and Gill (18) used a similar approach with minor modi®cations. More recently, Kanaya and Hiromi (19) used a stopped-¯ow version of the UV biuret assay. The sensitivity of the UV biuret assay to gelatin was lower than observed with other proteins (17) (Table 6). Similar sensitivity differences are observed at visible wavelengths (see Table 1). The dipeptide L-ProGly had a high molar extinction coef®cient e 176 MÀ1 cmÀ1 when analyzed by the UV biuret method. In contrast, L-GlyPro had zero extinction. The biuret-positive dipeptide has a ``normal'' peptide nitrogen atom, whereas the peptide bond in L-GlyPro is formed from a secondary amine nitrogen The Alkaline Copper Reagent: Biuret Assay 57 without an ionizable hydrogen. In consequence, the complexation of Cu2 is severely weekened (18).

7. APPLICATIONS OF THE ACR SOLUTION FOR FOOD PROTEIN ANALYSIS 7.1. Cereal Proteins High-protein wheat ¯our is suitable for bread making. Low-protein soft wheat ¯our is used for manufacture of biscuits and cookies or for animal feed. Barley for brewing should have a low protein content in order to minimize haze formation. Understandably, the commercial value of grain depends on its protein content. Grain suppliers and purchasers are interested in rapid and accurate methods for protein determination. Protein analysis is also important in connection with plant-breeding programs. Pinckney (12) was the ®rst to analyze wheat proteins using ACR. Flour protein was extracted (peptization) before analysis. Jennings (11) and also Williams (16) later developed a one-step assay by adding ACR directly to cereal ¯our. Johnson and Craney (13) and Craney (20) eliminated the need to prepare ACR by adding solid copper carbonate directly to ¯our suspended in alcoholic alkaline solvent. Noll et al. (21) produced an ACR solution containing 50% (w/w) isopropanol to reduce the interference from plant dyes. Furthermore, the time taken for analysis was reduced from about 60 minutes to 5 minutes (22,23). Most of the early applications of the ACR centered on wheat grain and ¯our analysis (12±23). A smaller number of reports deal with the analysis of proteins from rice (24), barley and malt (25,26), corn (27), and sorghum and peal millet (28). Developments of the ACR analysis should be seen against a background of a number of perceived limitations of the manual Kjeldahl analysis. The method is considered slow and costly. There is a relatively high capital cost and a need for highly skilled staff (Chapter 1). By comparison, the ACR method is simple, fast, and affordable. As it is a colorimetric technique, there are nominal instrumentation costs. The requirement for personnel training is also minimal. Such considerations provided some of the impetus for the ®rst applications of the ACR for cereal protein analysis.

A. Wheat Proteins Pinckney (12) described a method for extracting wheat protein from ¯our using a 1:10 volume mixture of carbon tetrachloride and dilute (0.05 N) potassium hydroxide. The mixture was centrifuged and the aqueous phase analyzed using ACR stabilized with 0.32% glycerol. About 100 samples of 58 Chapter 2 hard red winter wheat, 36 samples of hard spring wheat, and 28 samples of hard white wheat were analyzed by Pinckney (12). The A550 readings were highly correlated with Kjeldahl protein (R 0.925±0.976). The standard error for the ACR assay was 0.30±0.32 with a reproducibility of 0.1%. Sources of error include suboptimal peptization by dilute potassium hydroxide. The rate of color formation was also undesirably slow, requiring 20±40 hours. Jennings (11) developed a one-step assay by adding ACR directly to ¯our. Protein extraction and color development occurred simultaneously as the ¯our was agitated with ACR. This innovation, by eliminating a separate peptization stage, led to a signi®cant increase in the speed of ACR analysis. Jennings also found that K-Na tartrate was a better stabilizer for Cu2 than glycerol. Samples of the former seemed to contain fewer impurities and the likelihood of copper oxide formation was reduced. The Cu2-tartrate complex (lmax 675 nm) also interfered less with A550 readings compared 2 with the Cu glycerol chelate (lmax 630 nm). During the one-step assay the protein extraction ef®ciency was 84±97%. With barley varieties having a black or blue aleurone layer, there was interference from extraneous plant colors. Unidenti®ed chromogenic species with lmax 500±800 nm were dissolved at high pH. The interference was especially acute for the analysis of whole-meal wheat ¯our or unhulled oats. The extraneous color was associated with oat bran (29). A one-step ACR method was employed to assess protein levels in 45 varieties of brown and milled rice (24). There was a high correlation between

A550 readings and crude protein levels (N 6 5.95).* The calibration equation for milled rice was cP 15:48A550 À 0:063, where cP is the % Kjeldahl protein. The linear range extended to A550 1.0 with a regression coef®cient of 0.964 and a standard error of 0.46. A further ACR analysis of 42 brown rice samples led to the regression equation cP 16:04A550 À 0:233 (R 0.981) with a standard error of analysis of 0.30. Other results for ground rice samples are summarized in Table 7. Further improvements of the ACR assay were introduced by Johnson and Craney (13). First, they modi®ed the ACR assay for use with strongly colored cereals such as barley, oats, and grain sorghum. Next, solid copper carbonate was added directly to ¯our suspended in alkaline-isopropanol solution, thereby doing away with a need for a prepared ACR solution. These changes led to a method that was signi®cantly faster than any previous ACR assay (Method 2).

* Ground rice (1 g) and 2 mL of carbon tetrachloride were added to several (25 Â 150 mm) test tubes along with ACR (40 mL). Samples were shaken with a mechanical shaker for 90 minutes.

Aliquots (15 mL) from each sample were centrifuged and A550 readings were recoreded. The Alkaline Copper Reagent: Biuret Assay 59

TABLE 7 Protein Content of Ground Rice Determined Using the Biuret Assay

Milled rice Brown rice Method Mean (%) Range (%) Mean (%) Range (%)

Micro- 7.50 5.74±11.69 8.25 6.15±11.74 Kjeldahl ACR or 8.18 5.95±12.50 8.85 6.30±12.45 biuret

Source: Data from Ref. 24.

Method 2 Rapid analysis of wheat proteins (13).

Reagents 1. Copper carbonate (solid). 2. Potassium hydroxide pellets 3. Isopropanol Alkaline isopropanol solvent. Add potassium hydroxide (5.61 g) to isopropanol (600 mL). Make up to 1 L with water.

Procedure Place 1.00 (+ 0.001) g of ¯our in a 250-mL Erlenmeyer (conical) ¯ask and add solid copper carbonate (1.0 + 0.1 g). Suspend the mixture in alkaline isopropanol solution (50 mL) and shake vigorously using a mechanical shaker for 15 minutes. Allow the sample to stand for a further 15 minutes and ®lter through a glass ®ber ®lter with vacuum

suction. Take A550 readings against a reagent blank. About 391 cereal samples were analyzed, including grain sorghum (47 samples), corn (48 samples), oats (40 samples), barley (44 samples), wheat (165 samples), and hard and soft wheat ¯our (47 samples). The ACR assay results are highly correlated with crude protein levels determined by Kjeldahl analysis (Table 8). The former technique could be applied to a wide range of cereal grains. Using multiple sample shakers, a throughput of 0.2 minÀ1 could be achieved. Compared with the preprepared ACR solution, solid copper carbonate is more stable for prolonged storage. The isopropanol reduced interference by extraneous plant dyes and also reduced the solubility of starch in the alkaline reaction medium. By eliminating a peptization stage, the time for the ACR analysis was reduced from 35±40 60 Chapter 2

TABLE 8 Analysis of Cereal Proteins Using the Method of Johnson and Craney

Regression Sensitivityb a Number of equation A550 per Analytical Cereal samples (Yi ) R (%) error (%)

Grain 47 13:36Xi 0.98 0.0748 0.19 sorghum 2:64 Corn 48 11:486Xi 0.95 0.0871 0.14 4:0 Oats 40 15:93Xi 0.99 0.0627 0.17 1:84 Barley 44 20:36Xi 0.97 0.0491 0.23 1:33 Wheat 165 16:07Xi 0.99 0.0622 0.18 (whole 1:47 meal) Wheat four 47 16:68Xi 0.99 0.0599 0.14 (re®ned) 0:26 a Regression equation with Yi % Kjeldahl protein and Xi absorbance at 550 nm. bSpeci®c color change for 1% increase in sample protein content. Source: Summarized from Ref. 13.

minutes to about 10 minutes. An example of the one-step biuret procedure for cereal protein analysis is described by Strong and Duate (30). On the downside, repeated measurement of solid copper carbonate for ACR analysis is unappealing. Availability of preweighed copper carbonate tablets could remove some of the objections to ``solid'' reagents. The pharmaceutical industry has been dispensing medicinal tablets for many years. Kjeldahl catalysts are also available in tablet form.

B. Sorghum and Millet Deosthale and Visweswara-Rao (28) at the National Institute of Nutrition, Hyderabad (India) employed the Johnson and Craney method for sorghum and peal millet protein analysis. The results were compared with the micro-

Kjeldahl method. For sorghum protein A550 0.04131 cP 3.854 (R 0.943), where cP represents Kjeldahl protein (%N 6 6.25). The calibration graph for pearl millet was expressed by A550 0.07135 cP 1.488 (R 0.9413). Comparing with data in Table 8, note that the current graphs were drawn with % Kjeldahl protein on the ordinate axis, and sensitivity for sorghum protein is lower here. Analysis of sorghum The Alkaline Copper Reagent: Biuret Assay 61 protein using the biuret and Kjeldahl methods is also described by Belavady et al. (31).

7.2. Meat Proteins The procedure of Torten and Whitaker (32) is a good example of the application of the ACR method for meat sample analysis. The method is summarized next. Method 3 Analysis of meat proteins using the ACR (32). Reagents 1. Prepare the ACR solution as described in Method 1. Procedure Meat comminution and alkaline digestion. Grind raw meat samples by passing through a Horbart meat grinder to produce a ®ne paste. Place comminuted meat samples (0.9±1.2 g) in a 50 mL Erlenmeyer ¯ask. Add 20 mL of sodium hydroxide (0.5 N) and heat the suspension for 10 minutes over a boiling-water bath. Allow to cool and ®lter using Whatman No. 3 paper to remove fat. Shake 15 mL of ®ltrate with an equal volume of petroleum ether to remove remaining fat. Centrifuge using a solvent-resistant centrifuge tube at * 4000 rpm and collect the clari®ed extract for protein analysis. Add 4 mL of ACR solution to a ®xed volume (0.4±1 mL) of meat extract. Bring the ®nal volume to 5 mL with distilled water, mix, and

allow to stand for 30 minutes. Using a colorimeter take, A550 readings against an appropriate reagent blank. Torten and Whitaker (32) added varying amounts of fat to comminuted lean beef, chicken, cod, or pork to provide a range of protein values. For a range of 9.9±24.2% protein the A550 readings were linearly related to Kjeldahl results (R > 0.98). The optimal time for meat protein dissolution using alkali was 10 minutes. Longer heating times led to peptide bond hydrolysis. Defatting with petroleum ether was ef®cient. Other defatting solvents include carbon tetrachloride and diethyl ether. Table 9 shows the range of protein values for raw meat samples as determined by the ACR analysis. A very high correlation exists between the Kjeldahl and ACR methods. Therefore, a comparative calibration exercise may be adopted by analyzing standard protein samples, e.g., serum albumin. Further examples of meat protein analysis using the ACR assay are reported by Lasztity et al. (33) and Reichardt and co-workers (34). Brooks and others 62 Chapter 2

TABLE 9 Meat Protein Analysis Using the Biuret Assay

% Protein Sample (n)a ACRb / biuret Kjeldahl % Moisture % Fatc

Beef (10) 24.2 Æ 0.3 24.3 Æ 0.3 60.3 Æ 0.11 20.3 Pork (11) 23.1 Æ 0.3 22.4 Æ 0.3 55.2 Æ 0.18 26.7 Chicken (9) 25.2 Æ 0.3 24.6 Æ 0.2 58.8 Æ 0.13 21.2 Cod (8) 18.8 Æ 0.2 19.2 Æ 0.2 64.5 Æ 0.07 20.3 aMeat sample and number replicate analysis.bACR alkaline copper reagent or biuret method. cApproximate fat content estimated as difference between protein and moisture. Source: Ref. 32.

(35) showed that the ACR method is probably the most accurate of colorimetric method for assaying whole-body protein concentration in test- animal carcasses during feeding trials. Silgjnic and Samardzija (36) considered the best ways to dissolve meat samples for biuret analysis. Beef, pork, chicken meat, frankfurters, sausage, and ®sh did not dissolve fully at high pH, leading to losses during the subsequent analysis. Dissolving samples at lower pH using concentrated urea solutions provided a better alternative. Protease action affects the results of the ACR assay for meat protein. Errors may arise for meat samples with high amounts of endogenous proteases. Horse muscle is thought to contain unusually high levels of catheptic activity. Autolysis leading to loss of some peptide bonds may affect results obtained using ACR analysis. Turgut (37) has shown that differently treated ®sh muscle gives different results with ACR analysis, probably as a result of ®sh muscle autolysis. Prusa and Bowers (38) used the biuret assay to determine the solubility of turkey muscle protein under the in¯uence of nonmeat ingredients (sodium nitrite, sodium chloride, and phosphate salts).

7.3. Meat Process End-point Temperatures For safety reasons, processed meat should be heated to certain minimum end-point temperatures (EPTs). With adequate heating, agents responsible for viral diseases such as foot-and-mouth disease, foul pest, Newcastle disease, and African swine fever are inactivated. Potential bacteriological hazards associated with underheated meat include Escherichia coli and Salmonella. Different classes of meat have different prescribed EPTs. A protein coagulation test has been suggested as the means for establishing The Alkaline Copper Reagent: Biuret Assay 63 whether meat has undergone adequate heat treatment. Heating leads to a reduction in the muscle proteins extracted with 0.9% saline. A reduction in the ratio of extractable biuret-positive compounds (EBPRs) can be used to monitor heat treatment. The extractable biuret-positive compound ratio is de®ned by the empirical relation test sample protein solubility EBPR 2 reference sample protein solubility Heating meat to the USDA-prescribed EPT produced an EBPR value of 1.1 + 0.012 (39±41).

7.4. Dairy Proteins A collaborative trial of the biuret method for assessing protein solubility was undertaken by Morr et al. (42). The tests of sodium chloride solubility at pH 3 or 7 were compared with Kjeldahl analysis. At pH 7 the range of solubility values determined by Kjeldhal analysis was 86.9±94.2% (whey protein), 70±99.9% (sodium caseinate), 17.3±19.7% (soy protein isolate), or 94.2±99.2% (egg white protein). Biuret results were correlated with Kjeldahl values but showed signi®cantly lower precision. The biuret assay was used for the measurement of insolubilized protein. Casein from milk was precipitated using acetate buffer. The precipitate was recovered by centrifugation and biuret reagent was added directly to the precipitate.

After 30 minutes, A540 readings were recorded. For 10 milk samples the mean casein content was 2.6±2.7% (w/v) (43). The method probably works because casein dissolves in the high-pH biuret reagent. Analyses of other dairy proteins including cheese have also been described (44).

7.5. Yeast Proteins and Fermentation Monitoring Analysis of yeast protein raises dif®culties owing to high levels of NPN, mainly nucleic acid. Yeast cells are also surrounded by a tough cell wall composed of glucans, mannan, and chitin as well as protein. The ACR method was successfully applied for yeast protein analysis by Ihl and Tagle from the University of Chile, Santiago (45). The method described in detail next should be applicable to a wider range of single-cell proteins. 64 Chapter 2

Method 4 Analysis of yeast protein by the ACR method. Reagents 1. ACR (Method 1) 2. Aqueous toluene solution (water containing 10 ppm toluene) 3. Sodium hydroxide solid 4. Torula yeast Procedure Yeast protein solubilization. Lyse dried yeast (3 g) by suspending in 100 mL of aqueous toluene solution. Shake or stir the suspension of cells in a water bath at 508C for 6 hours. Add sodium hydroxide to bring to 0.5 M and heat at 75±808C for 30 minutes. Freeze the sample at À 208C overnight, thaw, and reheat for a further 30 minutes. Protein analysis. To 1 mL of yeast protein extract add 4 mL of ACR

solution and allow 30 minutes for color formation. Record A540 readings. Yeast protein reacted with the ACR solution forming a purple-violet color with lmax 540 nm. The ACR assay results were strongly correlated with Kjeldahl results (%N 6 6.25). About 12% of nitrogen in dried yeast was NPN. Protein monitoring during fermenter operation was addressed by Nielsen et al. (46). They set up a laboratory scale fermenter and examined various ¯ow analyses for monitoring sugars, lactic acid, biomass, and protein in the feed stream. Growth media components (yeast extract and peptone) contain high levels of NPN and glucose. Despite such dif®culties, the ACR assay was successfully used to monitor the utilization of feed protein during the fermentation by lactic acid bacteria.

REFERENCES

1. A Hiller. Determination of albumin and globulin in urine. Proc Soc Exp Biol Med 24:385±386, 1927. 2. HW Robinson, CH Hodgen. The biuret reaction in the determination of serum proteins. 1. A study of the conditions necessary for the production of a stable color which bears a quantitative relationship to the protein concentration. J Biol Chem 135:707±724, 1940. 3. JWC Mehl. The biuret reaction of proteins in the presence of ethylene glycol. J Biol Chem 157:173±180, 1945. 4. A Sols. An improved biuret reaction of proteins and the two-standard colorimetry. Nature 160:89, 1947. The Alkaline Copper Reagent: Biuret Assay 65

5. TE Weichselbaum. An accurate and rapid method for the determination of proteins in small amounts of blood serum and albumin. Am J Clin Pathol 10(Tech Suppl):40±49, 1946. 6. AG Gornall, CJ Bardawill, MM David. Determination of serum proteins by means of the biuret reaction. J Biol Chem 177:751±766, 1949. 7. E Layne. Spectrophotometric and turbidimetric methods for measuring proteins. Methods Enzymol 3:447±454, 1957. 8. GH Grant, JF Kachmar. The proteins of the body ¯uids. In NW Tiettz, ed. Fundamentals of Clinical Chemistry. London:WB Saunders, pp 298± 376. 9. RD Strickland, ML Freeman, FT Gurule. Copper binding by proteins in alkaline solution. Anal Chem 961:545±552, 1961. 10. H Sigel, RB Martin. Coodinating properties of the amide bond. Stability and structure of metal ion complexes of peptide related ligands. Chem Rev 82:385± 426, 1982. 11. AC Jennings. Determination of the nitrogen content of cereal grain by colorimetric methods. Cereal Chem 38:467±479, 1961. 12. AJ Pinckney. Wheat protein and the biuret reaction. Cereal Chem 26:423±443, 1949. 13. RM Johnson, CE Craney. Rapid biuret method for protein content in grains. Cereal Chem 48:276±282, 1971. 14. H Mitsuda, T Mitsunaga. Evaluation and elimination of the interference by starch in the biuret determination of wheat protein. Agric Biol Chem 38:1649± 1655, 1974. 15. H Mitsuda, T Mitsunaga. A convenient method for rapid determination of cereal proteins: a device to eliminate the effect of color substances on the biuret procedure. Agric Biol Chem 38:2265±2266, 1974. 16. PC Williams. The determination of proteins in whole wheatmeal and ¯our by the biuret method. J Sci Food Agric 12:58±61, 1961. 17. GL Ellman. The biuret reaction: changes in the ultraviolet absorption spectra and its application to the determination of peptide bonds. Anal Biochem 3:40± 48, 1962. 18. RF Itzhaki, DM Gill. A micro-biuret method for estimating proteins. Anal Biochem 9:401±410, 1964. 19. K-I Kanaya, K Hiromi. Determination of low concentrations of protein by the biuret method using the ``stopped-¯ow time difference analysis'' technique. Agric Biol Chem 51:1885±1892, 1987. 20. CE Craney. A quick biuret method for protein in wheat. Cereal Chem 49:496± 497, 1972. 21. JS Noll, DH Simmonds, W Bushuk. A modi®ed biuret reagent for determination of protein. Cereal Chem 51:600±616, 1974. 22. WT Greenway, RM Johnson. Five-minute biuret method for protein content of wheat. Baker's Dig 48(2):38±39, 72, 1974. 23. DH Simmonds, JA Ronalds. Rapid protein determination in cereal grains using the biuret reaction. Baker's Digest 49(4):36±40, 51, 1975. 66 Chapter 2

24. LC Parial, LW Rooney, BD Webb. Use of dye-binding and biuret techniques for estimating protein in brown and milled rice. Cereal Chem 47:38±43, 1970. 25. M Gullord. Studies on the biuret method for determination of protein in cereals. Meld Nor Landbrukshoegsk 51(13):6 pp 1972. 26. Y Pomeranz, RB Moore, FS Lai. Reliability of ®ve methods for protein determination in barley and malt. J Am Soc Brew Chem 35(2):86±93, 1977. 27. PS Misra, R Barba-Ho, ET Mertz, DV Glover. Studies on corn proteins. V. Reduced color response of opakue-2 corn protein to the biuret reagent, and its use for the rapid identi®cation of opaque-2 corn. Cereal Chem 50:184±190, 1973. 28. YG Doesthale, K Visveswara-Rao. Application of rapid biuret technique for protein estimation in sorghum and pearl millet. Indian J Nutr Diet 14(3):65± 69, 1977. 29. PC Williams. The determination of proteins in whole wheatmeal and ¯our by the biuret method. J Sci Food Agric 12:58±61, 1961. 30. FC Strong, AMA Duate. A room temperature, rapid method for the determination of protein in wheat and other grains by the biuret reaction. Cereal Chem 69:659±664,1992. 31. B Belavady, MM Subramanya, KNRK Murthy, S Ramachandra, AN Bagali, SA Hosamani. Appropriate sample for determination of protein in sorghum (Sorghum bicolor (L) Moench) raised in agricultural trials. J Sci Food Agric 37:207±210, 1986. 32. J Torten, JR Whitaker. Evaluation of the biuret and dye-binding methods for protein determination in meats. J Food Sci 29:168±1174, 1964. 33. R Lasztity, D Torley, F Orsi. Contribution to the protein determination in meat products. Proceedings of the European Meeting of Meat Research Workers 24:L1:1±L1:6, 1978. 34. W Reichardt, J Mueller, S Mueller, B Eckhert. Beef and pork. Direct determination of connective-tissue-protein-free pure protein content [Rind und Schweine¯eisch. Zur direkten Bestimmung des bindegewebseiweissfreien Reineiweissgehaltes (Rein-BEFFE)]. Fleischwirtschaft 74:1327±1329, 1994. 35. SPJ Brooks, BJ Lampi, G Sarwar, HG Botting. A comparison of methods for determining total body protein. Anal Biochem 226:26±30, 1995. 36. D Smiljanic, S Samardzija. Biuret method for determination of proteins in food. I. Conditions of dissolving the sample. Technol Mesa 38:153±157, 1997. 37. H Turgut. Drawbacks in the use of the biuret method for determination of the same protein in differently treated ®sh samples. Food Chem 4:161±165, 1979. 38. KJ Prusa, JA Bowers. Protein extraction from frozen, thawed turkey muscle with sodium nitrite, sodium chloride, and selected phosphate salts. J Food Sci 49:709±713, 720, 1984. 39. WE Townend, JE Thomson, JR Hutchins. ``Coagulation test'' for cooked meat temperature: effect of sample preparation methods. J Food Sci 50:1179± 1180, 1186, 1985. The Alkaline Copper Reagent: Biuret Assay 67

40. CE Davis, AJ Bracewell, JB Andersen, JO Reagan. Time temperature heating effects on biuret-positive water-extractable porcine and bovine muscle proteins. J Food Prot 48:215±220, 1985. 41. CE Davis, BG Lyon, JO Reagan, WE Townsend. Effect of heating on water soluble biuret-positive compounds of canned cured pork picnic shoulder. J Food Prot 50:681±684, 1987. 42. CV Morr, B German, JE Kinsella, JM Regenstein, JP van Burent, A Kilara, BA Lewis, ME Mangino. A collaborative study to develop a standardized food protein solubility procedure. J Food Sci 50:1715±1718, 1985. 43. T Cheng, YB Sun, J Li. Determination of casein content in milk by biuret method. China Dairy Ind 28:33±35, 2000. 44. W Reichardt, B Eckert, Determination of protein in milk, cheese and meat by means of the biuret reaction [Zur Bestimmung des Proteingehaltes von Milch, Kaese und Fleisch mit Hilfe der Biuret-Reaktion]. Nahrung 35:731±738, 1991. 45. M Ihl, MA Tagle. Estimation of protein in yeast. J Sci Food Agric 25:461± 464, 1974. 46. J Nielsen, K Nikolajsen, S Benthin, J Villadsen. Application of ¯ow-injection analysis in the on-line monitoring of sugars, lactic acid, protein and biomas during lactic acid fermentation. Anal Chim Acta 237:165±175, 1990.

The Lowry Method

1. INTRODUCTION

The protein assay of Lowry et al. (1) is simply called the Lowry assay. The prototype assay has a number of drawbacks. The standard curve is nonlinear. The technique uses unstable reagents, which are generally prepared daily. The classical method is also subject to a variety of interference compounds. Nevertheless, the Lowry assay remains highly important. Peterson's modi®cation of the Lowry assay (2) is robust, sensitive, and impervious to most interferences. This chapter contains descriptions of the Lowry assay, the underlying principles (Sec. 3), calibration features (Sec. 4), interference compounds, and common sample pretreatment strategies for ensuring accurate results (Sec. 5 and 6). Applications of the Lowry method to food protein analysis are reviewed in Sec. 7. The design of the Lowry protein assay can be traced to the investigations of Wu (3,4). These were concerned with the use of tungstate- molybdate reagent for protein analysis. Folin and Ciocalteu (5) also determined tyrosine and tryptophan in protein hydrolysates using tungstate-molybdate. Herriot (6) achieved a 3- to 15-fold increase in sensitivity of the tungstate-molybdate assay in the presence of copper (Cu2) ions.

69 70 Chapter 3

2. THE LOWRY PROTEIN ASSAY

A mixture of Cu2 sulfate and sodium-potassium tartrate reacts with proteins. Then Folin-Ciocalteu reagent is added. A blue-purplish color forms that is measured at 750 nm (A750). The Lowry method is reviewed by Lane (7) and Peterson (8,9). The following procedure is based on descriptions from Refs. 1 and 7. Preparation of the Folin-Ciocalteu reagent (4) is described in Method 2.

Method 1 Analysis of soluble proteins using the Lowry method.

Reagents

1. Sodium carbonate (Na2CO3) 2. Sodium hydroxide (0.1 M) 3. Copper sulfate CuSO4 Á 5H2O 4. Sodium potassium tartrate 5. Folin-Ciocalteu reagent 6. Bovine serum albumin

Procedure Prepare stock solutions for reagents A, B, and E as described below. Mix reagents A and B in a volume ratio of 50:1 to produce reagent C, which is a copper sulfate solution stabilized with tartrate. Reagent A (2% sodium carbonate in 0.1 N sodium hydroxide). Dissolve sodium carbonate (2 g) in 100 mL of 0.1 M sodium hydroxide. Reagent B (0.5% copper sulfate in 1.0% sodium-potassium tartrate). Add 0.5 g of copper sulfate and 1.0 g of tartrate to 100 mL of distilled water.* Reagent C (alkaline copper tartrate). Mix 50 mL of reagent A and 1 mL of reagent B. Prepare daily. Reagent E (Folin-Ciocalteu reagent). Dilute the commercial (2 N) reagent 1:1 with distilled water. Standard protein assay. Mix 0.2 mL of protein sample (5±100 mg){ with 1 mL of reagent C. Incubate at room temperature for 10 minutes. Now add 0.1 mL of reagent E and mix immediately. After 30

* Copper sulfate and tartrate salts have a low solubility in alkali. Dissolve these separately in a small volume of water before adding to reagent A. { The prescribed reagent volumes lead to a 6.5-fold dilution of protein standards. The concentration of protein should be 20±500 mgmLÀ1, leading to a ``within cuvette'' concentration range of 15.4±77.0 mgmLÀ1. The Lowry Method 71

minutes take A750 readings or else A500 readings for strongly concentrated solutions. The Folin-Ciocalteu reagent is now widely available. The commercial reagent is stable for months at room temperature. There seems little justi®cation for preparing reagent E in house. In the absence of a commercial supplier, the Folin-Ciocalteu regent may be prepared as follows. Method 2 Preparation of the Folin-Ciocalteu reagent Reagents 1. Sodium tungstate Na2Wo4 ? 2H20 100 g 2. Sodium molybdate Na2MO4 ? 2H2O 25 g 3. Water 700 mL 4. Phosphoric acid (85% w/w) 50 mL 5. Hydrochloric acid (conc.) 100 mL Procedure Add the preceding compounds and some antibumping granules to a 1.5-L ¯ask. Fit a condenser and re¯ux gently for 10 hours in a fume cupboard. Switch off the gas or electric heater. Remove the condenser and add (a) 150 g of lithium sulfate and (b) 50 mL of water followed by (c) a few drops of liquid bromine. Bring the mixture to a boil for 15 minutes to remove excess bromine. Allow to cool, make up the volume to 1L with distilled water, and ®lter. The ®nished reagent, which should have no green tint, should be stored in an amber bottle. Reference 1 is one of the most often cited papers in analytical biochemistry.* It is one of the scienti®c papers for which the description classic is probably deserved. Much of what is known about the Lowry method was anticipated in the original publication 50 years ago. A number of investigators have modi®ed the Lowry assay to improve its performance characteristics. Bensadoun and Weinstein (10) reacted protein samples with sodium deoxycholate and trichloroacetic acid (TCA) before performing the Lowry assay. This pretreatment effectively frees protein samples from TCA- soluble compounds, thereby improving accuracy. When 5 and 50 mgof protein were precipitated with TCA, the recovery ranged from 7 to 91%. Pretreatment with DOC led to the quantitative recovery of protein no matter its initial concentration. With samples containing high concentrations of the interferences, two cycles of precipitation were used. The initial

* As of June 2001 Ref. 1 had 65,535 citations in the Science Citations index. 72 Chapter 3 protein-DOC precipitate was resuspended in distilled water and reprecipi- tated for a second time. Peterson (2) used DOC-TCA precipitation to develop an assay, now available in kit form (Sigma-Aldrich Ltd.), that rendered the classical Lowry method obsolete.* Method 3 Peterson's modi®cation of the Lowry protein assay (2,8,9) Reagents 1. Copper sulfate 2. Sodium potassium tartrate 3. Sodium carbonate 4. Sodium dodecyl sulfate (SDS) stock solution (5% w/v) 5. Sodium deoxycholate solution (0.15% w/v) 6. Sodium hydroxide stock solution (0.8 M) 7. Trichloroacetic acid (72% w/v) 8. Folin-Ciocalteu reagent (2 N) 9. Bovine serum albumin (BSA) (0.5 mg mLÀ1) as protein standard. Add 1 mgmLÀ1 sodium azide as preservative and store frozen as small aliquots. Procedurey Copper-tartrate-carbonate (CTC) stock solution. Dissolve copper sulfate (0.1 g) as well as sodium potassium tartrate (0.2 g) each in about 10 mL of distilled water. Add both solutions to sodium carbonate (10 g in 50 mL of distilled water) and make up to a ®nal volume of 100 mL. Preparation of reagent A. Mix CTC, SDS, and sodium hydroxide stock solutions in a volume ration of 1:2:1. Reagent A is stable at room temperature for 2±3 weeks. Refrigeration at 48C will double the useful life. SDS precipitates in the refrigerator. Redissolve by holding the reagent bottle under a running hot-water tap. Preparation of reagent B. Dilute commercial 2 N Folin-Ciocalteu reagent 1:1 with distilled water. This is stable for many months at room temperature. Protein analysis. Add suf®cient water to bring the protein sample (5± 100 mg) volume to 1 mL. Now, add 0.1 mL of DOC solution, followed 10 minutes later with 0.1 mL of TCA solution. Centrifuge the mixture at 10,000 g (mark 12 on a microcentrifuge) for 10 minutes. Decant the supernatant and place the upturned Eppendorf

* Peterson's modi®cation of the Lowry assay had received 6130 citations as of June 2001. { It is convenient to employ (1.5-mL) microcentrifuge (Eppendorf) tubes for this procedure. The Lowry Method 73

tubes over tissue paper to drain. The precipitate of DOC-protein may just be seen as a gray plaque on the wall of the microcentrifuge tube. Add reagent A (1 mL) and mix gently to redissolve the protein precipitate. After 10 minutes add reagent B and allow to stand for

30 minutes. Take A750 readings using a 1-cm (1.7 mL capacity) disposable plastic cuvette. The distinctive features of Peterson's method can be summarized as follows: 1. The increased concentration of copper tartrate increases the stability of the Lowry reagents from 24 hours to 2 weeks. 2. SDS (1% w/v) reduces interferences from nonionic detergents and dissolves membrane proteins. 3. Protein precipitation with DOC-TCA eliminates nonprotein interferences. 4. The precipitation step concentrates samples such that &1m gof protein can be detected accurately.

3. CHEMISTRY OF THE LOWRY ASSAY 3.1. Reactions of the Lowry Protein Assay The following account is based on Refs. 11±14. The Lowry assay involves two reactions. First, a protein-Cu2 complex forms. Six peptide bonds surround a central Cu2 atom. The high-pH solvent (OHÀ&0:1M;pH 13) induces protein denaturation. Loss of native structure precedes binding with Cu2 to form a type B protein-Cu2 complex (Chapter 2). Protein denaturation at high pH also exposes tyrosine and tryptophan residues, which then ionize. The second stage of the Lowry assay is a redox reaction with Folin- Ciocalteu reagent via two pathways. First, Mo6/W6 reacts directly with amino acid side chains (histidine, cysteine, asparagine, tyrosine, tryptophan). A high pH is not required for these reactions. Second, Cu2 mediates the dehydrogenation of the polypeptide via metal ion±catalyzed oxidation. The electrons are transferred to Mo6/W6, leading to a color change. Features of the Mo6/W6 reaction with Cu2 and protein are summarized in Table 1.

3.2. Metal Ion±Catalyzed Oxidation of Proteins The Cu2 catalyzes the oxidative degradation of polypeptides. The process involves the formation of Cu3 and Cu4 ions. It is possible to produce 74 Chapter 3

TABLE 1 Some Important Reactions for the Lowry Assay

6 6 4À 2 1. Mo /W reacts with reducing agents via a one-electron [e.g., Fe CN6 ,Fe , Sn2] or two-electron transfer (ascorbic acid and peptides). Reactions with tryptophan and tyrosine may involve four-electrons per residue. 2. Reduction of Mo6/W6 proceeds rapidly under acidic conditions. Adjusting the pH from 1 to 10 leads to deprotonation followed by a slow structural rearrangement and 1.7-fold increase in color. 3. Cu2 is not required for the reaction of Mo6/W6 with nonpeptide reductants. 4. The color yield is 3200 (+100) MÀ1 cmÀ1 per electron transferred to Mo6/W6. 5. Peptides without oxidizable side chains react with Mo6/W6 only if they can form a tetradentate peptide-Cu2 complex. 6. Each tetradentate Cu2 complex transfers approximately two reducing equivalents to Mo6/W6. With well-de®ned peptides there is a correlation between the color yield from the biuret and Lowry methods. 7. Color yield increases for polypeptides with oxidizable side chains. 8. Color yield decreases with the number of side chains with Cu2 complexing ability (glutamate, aspartate). 9. Cu1 is not involved in color formation.

Source: Summarized from Ref. 14.

Cu3 by chemical or electrochemical oxidation.

Cu2?Cu3 e; E À0:63V vs: NHE* 1

Metal ion±catalyzed oxidation (MCO) of tetraglycine-Cu2 has been investigated. One proposal is that in the presence of a strong oxidant 2À 2 (Qx), e.g., sodium chloroiridate NaIrCl6 , tetraglycine-Cu (designated as RH22Cu2) is degraded in three stages: (a) RH22Cu2 is oxidized to RH22Cu3 [Eq. (2)], (b) RH22Cu3 rearranges to (IIa) or (IIb), which is a dehydropeptide-Cu1 species [Eq. (3)], (c) compound IIb is hydrolyzed to diglycinamide, glycoxyglycine, and Cu1 [Eq. (4)]. It has been suggested that the structure for IIb is probably an iminopeptide (Fig. 1).

Qxn RH22Cu2 IÀk?1 Qx nÀ1 RH22Cu3 II 2

RH22Cu3 II /ÀkÀ2? ? R22Cu2 IIaH 3 /ÀÀ?R22Cu1 IIbH

* NHE, normal hydrogen electrode. The Lowry Method 75

FIGURE 1 Suggested mechanism of Cu2-mediated dehydrogenation of peptides leading to an iminopeptide or a dehydropeptide. (Adapted from Ref. 14.)

R22Cu1 IIbH O À?Cu1 2 4 diglycinamide glycoxyglycine

With molecular oxygen as oxidant, reaction (2) would lead to the superoxide radical [Eq. (5)], which probably remains protein bound as a ternary complex (III).

RH22Cu2 IO Àk?1 RH22Cu3 IIO 2 2 5 1 /ÀÀ?O222R22Cu IIIH Intramolecular oxidation of (III) then generates Cu2-hydroperoxide [Eq. (6)].

1 À 2 O222R22Cu IIIÀ?RO2 22Cu IV 6 76 Chapter 3

Formation of (IV) occurs via two-electron reduction of oxygen to produce a protein-bound hydroperoxide anion. Breakdown of the hydroperoxide (Fig. 1) accounts for the carbonyl compounds detectable using 2,4-dinitrophenylhydrazine (DNPH). A superoxide radical may be involved in the initial oxidation of 2 3 RH22Cu to RH22Cu [Eq. (2)] provided that O2 is released from the protein in Eq. (5). In summary, RH22Cu2 can be oxidized by strong 2 6 6 oxidants (NaIrCl6, hydrogen peroxide, and presumably Mo /W ). The formation of RH22Cu3 can also be initiated by atmospheric oxygen and 2 3 superoxide species O2 and RO222Cu . The RH22Cu may also form via the disproportionation of 2 moles of RH22Cu2 to produce RH22Cu3 and RH22Cu1. 2 2À Fragmentation of tetraalanine-Cu by IrCl6 produces alanylalanine amide (HAla2NH2) and pyruvyl alanine (PyrAlaOH). These were identi®ed by reacting with ninhydrin or DNHP and by analysis with high-voltage paper electrophoresis. Peptide fragments are also formed during the analysis of polyalanine by the Lowry method. In one study, the blue molybdate- tungstate complex was removed by adsorption with cross-linked polyacrylamide (Bio-gel P). The eluted (colorless) product reacted with DNPH. Hydrazone derivatives formed were analyzed by proton nuclear magnetic resonance (NMR) and by gel ®ltration on Sephadex G-25. In this manner, the products of polyalanine MCO were identi®ed as Ala6 or Ala8 peptide fragments. Tests showed that DNHP reacts with a-iminoisobutyric acid but not with dehydropeptide analogues such as N-acyldehydroalanine methyl ester. Therefore, MCO leads to an imino peptide and the dehydropeptide (Fig. 1). The initial oxidation involves electron abstraction from the lone- pair electrons on the peptide nitrogen rather than from the methylene group. According to Livitski et al. (12), the compound II is further oxidized as shown in Eq. (7). Hydrolysis of the dehydrogenated Cu222R leads to Cu2 [Eq. (8)] and not to Cu1 as shown in Eq. (4).

R22Cu2 IIQx n Àk?1 Qx nÀ1 R22Cu3?R22Cu2 7

2 2 R22Cu H2O À?Cu diglycinamide glycoxyglycine 8

Evidence for Eq. (8) comes from the deployment of 2.20-biquinoline (2,20DQ), which should form a purple complex with Cu1. Addition of 2,20DQ failed to detect the presence of Cu1. Therefore, the copper ion probably cycles between oxidation states 2 ,3, and 4 under strongly oxidizing conditions. Failure to detect free Cu1 shows that the sequence of reactions during the Lowry assay is probably Eq. (2)?, Eq. (3)?, Eq. (7)? The Lowry Method 77

Eq. (8). The net reaction is shown in Eq. (9).

2 n Polypeptide À Cu Qx H2O À?Cu2 Qx nÀ2 hexapeptide fragments 9

MCO reactions involving proteins are discussed further in Chapter 5.

3.3. Kinetics of the Lowry Protein Assay The duration of the Lowry assay is 40 minutes. Proteins react with Cu2 (reagent C) in 10 minutes. A further 30 minutes is needed for the reaction with reagent E before A750 readings are recorded. The speed of the Lowry assay depends on one or more of the reactions described in Sec. 3.2. 1. The time course for the alkali denaturation of proteins can be prolonged for stable proteins (e.g., lysozyme). 2. The oxidation of RH22Cu2 by atmospheric oxygen [Eq. (1)] À4 À1 occurs with the rate constant (k1)of5.56 10 (s ). The time for 99.9% completion (5/k1) is 151 minutes. 2 À 6 3. Oxidation of RH22Cu by IrCl6 (and presumably by Mo / W6) occurs with a rate close to the diffusion-controlled limit. Reduction of Mo6/W6 by simple reductants is nearly instanta- neous, being delayed only by the reagent mixing time. 4. Reduced tungstate-molybdate undergoes slow deprotonation and structural changes at high pH. At 258C this process takes about 30 minutes. 5. Rearrangement of RH22Cu3 to form a radical species [Eq. (2)] is 2À À3 À1 a slow process. With IrCl6 as oxidant, k2 is 1.66 6 10 (s )with a99.9% completion time of 36 minutes. Clearly, it is not possible to identify a single rate-limiting reaction for the Lowry assay of proteins. A high reaction rate is essential for automated protein analysis. Anderson and Marshall (15) and also Huang et al. (16) adapted the Lowry method for continuous ¯ow analysis. The sample throughput was 30 per hour.

4. CALIBRATION FEATURES

Calibration graphs for the Lowry assay are usually curved. A linear response is obtained for simple compounds (Sn2, tyrosine, etc.) (14). One explanation for the nonlinear calibration graphs for proteins is that Mo6/ 78 Chapter 3

W6 is degraded at high pH with a half-life of about 8 seconds (1). Assuming this explanation is correct, the color yield during protein analysis re¯ects a balance between Mo6/W6 decomposition by alkali and its reduction to form a blue complex. Inorganic reductants react rapidly with Mo6/W6 before its degradation. Nonlinearity was also ascribed to the declining copper/protein ratio as the concentration of protein is increased. Attempts to improve the linearity by increasing the concentration of copper led to higher readings for the sample blank.

Readings of A750 may be converted to protein concentration [P] using a nonlinear graph. This practice emphasizes data points adjacent to the unknown value. By contrast, a linear graph gives equal weighting to all the experimental points. In chemical analysis, ``linearity is next to cleanliness.'' Most analysts feel a sense of relief where calibration data conform to a linear function;

DA750 FP 10 where F is the slope of the calibration graph. The line described by Eq. (10) passes through the origin and hence

PDA750=F 11 Stauffer (17) ®tted Eq. (12) or its logarithmic form [Eq. (13)] to data from the Lowry assay.

F A700 oP 12

log A700 F logPw 13 where F and o are constants and w log o. For a protein concentration À1 range of 4±400 mgmL a plot of log A700 versus logP was linear. The unknown protein concentration can be determined from Eq. (14).

P10 YÀw=F 14

Notice that Y log A750 is the value for the unknown sample. Coakley and Jones (18) used reagent volumes three times higher than in the standard method.* Their calibration results were described by a hyperbolic curve.

* A 0.6-mL portion of protein standard solutions was reacted with 3.0 mL of Lowry reagent C.

After standing for 10 minutes, 0.3 mL of Folin-Ciocalteu regent was added. The A750 readings were taken 30 minutes later. The Lowry Method 79 P A 15 750 wP F

Equation (15) applies for within-cuvette BSA concentrations of 0.01± À1 0.77 mg mL and for A750 values of 0.18±2.2.* The constants F and w were determined by calibration. The protein concentration can be found using Eq. (16).

FA P 750 16 1 À wA750

For routine use, Eq. (16) can be linearized using a double-reciprocal transformation.

1 F w 17 A750 P

Equation (17) allows a two-point calibration: (a) analyze the unknown sample with two protein standards and (b) determine the constants F and w from Eqs. (18) and (19). y À y F 1 2 18 x1 À x2 where y1 1/A750 (1) and y2 1/A750 (2). Likewise, x1 1/ [P]1 and x2 [P]2. À1 The concentrations for the protein standards should be [P]1 0.1 mg mL À1 and [P]2 5±10 mg mL BSA. The intercept of Eq. (17) (w) is found from averages for y1, y2 and x1, x2.

w y À Fx where

y y x x y 1 2 and x 1 2 19 2 2

The previous treatment applies for a protein concentration between 0.1 and 10 mg mLÀ1. With protein concentrations below 0.2 mg mLÀ1, a straightforward calibration graph can be used with little loss of accuracy.

* Allowing for the 6.5 times dilution during the Lowry assay, a protein stock solution of 0.065± 5mgmLÀ1 leads to a ``within-cuvette'' concentration of 0.01±0.77 mg mLÀ1. Investigators are liable to cite either of these protein concentrations in their work. 80 Chapter 3

TABLE 2 Some Potential Interfering Compounds for the Lowry Assay

Biochemical classi®cation Food additives

1. Amine derivatives Acids and acidulants 2. Amino acids Amino acids 3. Buffers Colors 4. Chelating agents Dyes 5. Detergents (e.g., Triton X-100, Sweeteners Tween) 6. Drugs Arti®cial antioxidants (BHA, BHT, etc.) 7. Hexosamines Starch 8. Lipids and fatty acids Polysaccharides 9. Miscellaneous compounds Sulfur dioxide, sul®tes 10. Cryoprotectants Uric acid 11. Polyvinylpyrrolidone Fe2 12. Nucleic acids 13. Organic solvents 14. Phenols and polyphenols 15. Polysaccharides 16. Reducing agents 17. Salts 18. Sugars

5. INTERFERENCE COMPOUNDS

Interference compounds for the Lowry assay are either chelators or reducing agents. Chelators diminish color formation by sequestering Cu2. Reducing agents react with the Folin reagent to create extra color. Some interfering compounds act as both chelators and reducing agents. Dyes having an absorbance maximum near 600±750 nm are also interfering compounds. Peterson classi®ed interfering compounds into more than 13 classes comprising over 180 chemicals (Table 2). Many of these compounds are encountered in raw and processed foods.* Ascorbic acid and other reducing compounds are found in foods of plant origin (fruits, juices, pastes, concentrates). Wines and certain beverages have high concentrations of phenols and related compounds. Carbohydrates (simple sugarsÐsucrose, glucose, and fructose) and polysaccharides (pectin or starch) occur in foods including jams, preserves, and

* However, the analyst may not be aware of the presence of interfering compounds in a given sample. The Lowry Method 81 wheat products. Synthetic reductants, acidulants, and dispersants are just some of the other additives associated with foods. Low-molecular-weight interfering substances can be removed by dialysis. Another approach is to recover the protein from the surrounding solvent medium by TCA precipitation (Method 3). Where possible, sample pretreatment should be avoided for routine and high-throughput analysis. The effects of selected interferences on the Lowry assay are discussed in the following.

5.1. Buffers, Chelating Agents, and Detergents Peters and Fouts (19) showed that BICINE [N,N-bis- (2-hydroxethyl) glycine] and HEPES [N-(2-hydroxyethylpiperazine-N-2-ethanesulfonic acid)] produce color in proportion to their concentration. Ethylene diamine tetraacetic acid (EDTA), tris-hydroxythethylaminomethane (Tris), TRI- CINE, citrate, and Triton X-100 also produced similar effects (20). These interfering compounds increase the absorbance value for the reagent blank. However, they reduce the color yield from proteins. These effects are attributed to complex formation with copper. Interference by buffer components can easily be corrected using the appropriate buffer solution as the reagent blank. Sodium phosphate buffer had no effect on the Lowry assay.

5.2. Carbohydrates Reducing sugars (mainly hexoses) react with Cu2 and the Folin reagent. Nonreducing sugars (sucrose, oligosaccharides, polysaccharides) form complexes with Cu2. Tagatose, sucrose, and inulin interfered with the Lowry assay after exposure to hot alkali or acid (21). These solvents are routinely used to dissolve proteins. As well as reducing Cu2 to Cu1, sugars reduce tungstate-molybdate. The effects of simple sugars are signi®cant at concentrations above 1 mM. The order of effectiveness is fructose > sorbose > xylose > rhamnose > mannose > glucose (22). Toldra (23) considered protein analysis in connection with the production of high-fructose syrup. Effective control of this process requires the determination of a-amylase and glucoamylase speci®c activity (hence protein concentration) in samples containing up to 30±40% sugar. A solution of 1% (w/v) glucose or maltose produced an A750 response equivalent to that of 18.8 or 16.6 mgmLÀ1 BSA. Protein analysis is also necessary to determine the activity of fungal pectinases produced in submerged culture using high concentrations of pectin as inducer (24). Concentration of 0.1±1.0% (w/v) pectin interfered with both the classical and modi®ed Lowry assays. Calibration graphs showed a decrease in the 82 Chapter 3 slope and increased intercept. Assay sensitivity decreased while the LLD increased with increasing pectin concentration (Chapter 1). The color with protein-free samples increased linearly with the amount of added pectin. These results imply complex formation between pectin and Cu2. The Bradford method (Chapters 6 and 7) coped better with samples containing < 0.5% (w/v) pectin. In our experience, pectin forms a gelatinous mass during the Lowry assay. To avoid this problem, samples are exposed 0.1 M

CaCl2 and then centrifuged before protein analysis. Berg (25) warned that ``anyone attempting to determine protein contamination in polysaccharide preparations by the Lowry procedure could be misled by positive results which may be attributed to the carbohydrate.''

5.3. Chlorophyll Leaves, stems, and green fruit contain chlorophyll. Addition of chlorophyll À1 À1 (100 mgmL ) to BSA (300 mgmL ) increased A750 values by 400% (26). For accurate analysis, leaf protein must be separated from chlorophyll.*

5.4. Cryoprotectants, Sucrose, Glycerol, and Polyhydroxyalcohols Ethylene glycol, polyvinylpyrrolidone (PVP), and dimethyl sulfoxide (DMSO) are used to protect proteins against freeze damage. Glycerol, probably acting in a manner similar to sucrose, inhibits color formation. In the absence of copper, PVP reacts with Mo6/W6, leading to a blue coloration (27). Glycerol (2±5%) interferes with the Lowry procedure (28) but DMSO has no effect. Many foods and condiments contain very high concentrations of sucrose. Sucrose is also used for gradient ultracentrifugation. Sucrose by itself gives a positive response with the Lowry assay (29). However, protein analysis in the presence of sucrose leads to a reduced color yield. The interference becomes progressively worse with 0±65% sucrose. Effects can be reduced to tolerable levels (< 5% analytical error) by (a) diluting samples to < 10% sucrose or (b) doubling the concentration of copper in Lowry reagent B. Quadrupling the copper concentration led to an unstable reagent with a tendency to form colloidal copper (30).

* Leaf protein can be extracted with 0.1 N alkali. Next, precipitate the protein with 10% (w/v) TCA. A second approach is to treat freeze-dried leaf powder with acetone. This decolorized powder can be stored before analysis (this work is discussed in detail later). The Lowry Method 83

Sucrose (and other polyhydroxy compounds) appear to complex with Cu2 when monitored by ultraviolet absorbance spectrophotometry. In the absence of copper, sucrose or glycerol does not interfere with the tyrosine reaction with Folin-Ciocalteu reagent. Hydrolysis of sucrose by heat, alkali, or invertase generates fructose and glucose, which affect the Lowry assay as reducing sugars (31). Ficoll (synthetic polymer of sucrose) yields an intense blue color when analyzed by the Lowry procedure (32) (Fig. 2). Concentrations of 2±36% (w/v) ®coll are used for density gradient centrifugation of cell fractions. Because of its high average molecular mass (*400 kDa), it is not removed from samples by dialysis. Addition of the Folin-Ciocalteu reagent to ®coll leads to formation of color with a time course of over 18 hours at room temperature. Ficoll had no effect on the yield of color when added to protein samples after the Folin-Ciocalteu reagent. In the presence of ®xed concentrations (0.21±1.88%) of ®coll, the calibration graph of standard BSA solutions had progressively lower slopes while the intercept increased. Ficoll forms a UV-detectable complex with copper.

FIGURE 2 Effect of increasing ®coll concentrations on the Lowry assay. Boxed legend shows Ficoll concentrations. (Drawn from data from Ref. 32.) 84 Chapter 3

5.5. Lipids

Lipids increase sample turbidity and A750 values. The products of lipid autoxidation also react directly with the Folin reagent. Heating lipids with alkali (a common method of protein solubilization) leads to Lowry-positive products. The problem has been examined with arachidonic acid, phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol (33). Samples may be delipidated by shaking with chloroform or petroleum ether and then redissolved with 0.5 M NaOH by storing at 378C overnight or by heating at 1008C for 30 minutes. Solvent extraction will not be wholly successful if the protein and proteolipid complexes dissolve within an organic solvent phase. Emulsion formation can also lead to apparent loss of protein from the sample. Lees and Paxman (34) modi®ed the Lowry assay for proteolipids and lipoproteins. Their method is fast and does not require heating.* Markwell et al. (35) added 1% SDS directly to Lowry reagent C to dissolve membranes and other lipid. The concentration of copper sulfate was also increased from 0.5 to 4% (w/w) to reduce interferences from sucrose and EDTA. The modi®ed method gave results identical to those obtained with lipoprotein samples delipidated using petroleum ether. However, the new method allowed higher rates of analysis and greater convenience. There was no interference from 40±200 mM sucrose. SDS also features in Method 3, which is therefore able to deal with membrane lipids routinely. According to Kirazov et al. (36), treating membrane-containing fractions with 1 M NaOH or 0.5% SDS reduced the stability of A750 readings and did not reduce the interference from lipids.

5.6. Sulfhydryl Agents and Other Reducing Compounds The effects of various SH compounds are also related to their standard electrode potential (E8) measured against a normal hydrogen electrode (NHE) or a calomel electrode. At concentrations below 1 M, the redox potential is described by the Nernst equation: RT E E ln C 20 nF where n is the number of electrons transferred, E the observed redox potential corresponding to an activity of C (mol LÀ1), F the Faraday

* Treat protein samples (20±70 mg) with 0.5 mL of alkaline SDS solution (5% SDS, 0.5 M NaOH), vortex, and let stand at room temperature for 3 hours. Add 2.5 mL of Lowry reagent C followed

10 minutes later with 250 mL of Folin (1 N) reagent. Take A695 reading after 45 minutes. The Lowry Method 85 constant, and R the gas constant. For dilute solutions, C becomes equal to concentration. Therefore, the effect of reducing compounds on the Lowry assay is related to the ``intrinsic'' redox properties (E8 and n) and the concentration of reducing agent. Dithiothreitol (DTT), 2-mercaptoethanol (2ME), reduced glutathione (GSH), and oxidized gluthathione (GSSG) produce color with the Lowry assay (37). Reducing compounds react directly with the Folin-Ciocalteu reagent. Table 3 shows the degree of color formation with some SH compounds. As the underlying reaction is a redox process, calibration graphs for SH compounds will be nonlinear. For each SH compound De (MÀ1 cmÀ1) was determined from the steepest slope in Fig. 3. Values for De determined in this manner (Table 3, top half) agree with results from Ref. 14. DTT is one of the most effective interfering SH compounds. Samples with > 0.2 mM DTT have high A750 high background absorbances.

TABLE 3 The Color Yield from the Lowry Assay of Some Reducing Compounds

À1 À1 a b c Reducing agent Á" (M cm ) n E1/2(mV) Dithiothreitol 7625 (Æ176) 2 2-Mecaptoethanol 3900 1 Glutathione 3333 1 (reduced) Cysteine 3400 1 0.398 Glutathione 1870 (Æ760) 0.5 (oxidized) Cystine 1702 (Æ822) 0.5

Cysteine 3150 1 Fe2 3150 1 Sn2 6100 2 Ascorbic acid 6700 2 Phenol 12400 4 Tyrosine 12800 2 (4) 0.398 Indole 13000 2 (4) 0.253 Tryptophan 13200 2 (2) 0.205 aThe De (MÀ1 cmÀ1) values in the top half of Table 3 are calculated from the maximum slope in Fig. 3. Data in the bottom half of Table 3 from Ref. 14. bn electrons transferred to Mo6/W6 from one molecule of reducing compound. Values for n were determined by colorimetry,c Half-wave electrode potentials were determined by cyclic voltammetry, using standard calomel electrode (i.e., 0.242 V vs. NHE), from Ref. 78. Note that for dilute solutions E1/2 E8 À Eref. 86 Chapter 3

FIGURE 3 The Lowry assay of selected SH compounds. Concentrations of interference compounds were analyzed in the absence of protein. (Numerical data from Ref. 37.)

Reduction of the Mo6/W6 produces an absorbance change of 3200 units (MÀ1 cmÀ1) per mole of electrons transferred (see Table 1).

6. SAMPLE PRETREATMENT, AVOIDING INTERFERENCES, AND ENSURING ACCURACY

Dialysis provides a simple test for the presence of low-molecular-weight interferences. A sample is judged free of interferences when all attempts to remove such compounds meet with no success.

6.1. General Strategies A number of general strategies are available for dealing with interferences: 1. Use an appropriate reagent blank. Where the identity and concentration of interfering compound are known, a reagent blank can be prepared. Absorbance readings for the blank are then subtracted from the results for the sample. The Lowry Method 87

2. Prepare a reagent blank by analyzing a deproteinized sample. 3. Mask the interfering component using chemical additives. Excess copper sulfate will compensate for the effect of chelators. 4. Remove the interference. Lipids can be removed by defatting with an organic solvent. Pectin is removed by treating the sample with calcium chloride followed by centrifugation. Dialyze to remove low-molecular-weight interferences. 5. Precipitate the protein from the sample matrix before analysis (Method 3).

6.2. Destruction of SH Compounds The effect of SH compounds on the Lowry assay can be annulled by (a) oxidation with hydrogen peroxide (38), (b) carboxymethylation with iodoacetate (39), or (c) treatment with chloramine-T or CAT (40). The oxidation of SH groups by hydrogen peroxide in the presence of Cu2 involves the hydroxyl radical OHÐsee Chapter 2. Carboxymethylation recti®es analytical error due to 150 mM SH compounds. However, another popular SH-blocking agent, N-ethylmaleimide (NEM), was not an effective substitute for iodoacetate. NEM produced coloration with the Lowry assay. Interferences due to SH compounds were greatly reduced by oxidation with CAT. For instance, samples containing 20 mM Cys, 2ME, GSH, or

DTT gave A750 readings above 2. When treated with CAT, the same samples gave absorbances of 0.113, 0.032, 0.530, and 0.308. The unusually high reading from GSH was attributed to the peptide nature of this SH compound. Pretreatment with CAT and TCA also reduced inferences by

KCN, ascorbic acid, NaHSO3,Na2SO3, FeSO4, GSSG, and Cys-Cys. However, the sensitivity of the Lowry assay was reduced by about 50%. This effect is probably due to the oxidation of amino acid side chains (tyrosine, tryptophan, cysteine) by CAT, rendering them unavailable to react with molybdate-tungstate. Another consequence of the CAT procedure is that the resulting calibration curves are linear. Higuchi and Yoshida (41) found that using CAT under neutral conditions maintains analytical sensitivity at the level obtained with the classical Lowry procedure.

7. APPLICATIONS OF LOWRY ASSAYS TO FOOD PROTEIN ANALYSIS

Table 4 lists examples of food protein analysis using Lowry procedure. Selected cases are described in detail next. 88 Chapter 3

TABLE 4 Some Food Commodities and Processes Analyzed by the Lowry Procedure

Commodity Reference

Beer/brewing Hii and Herwig (42), Williams et al. (43) Cereal or legume ¯our Padhye and Salunke (44), Sathe and Salunkhe (45), Sebecic et al. (46), Paredes-Lopez et al. (47), Seguchi (48,49) Leaf proteins Eze and Dumbroff (25) Milk proteins Huang et al. (50), Kroening et al. (51) Various condiments Ogunbunmi and Bassir (52) Potato protein Hoff (53), Vigue and Li (54) Tomato seed proteins Latlief and Knorr (55) Yeast, microbes, fermentation Vananuvat and Kinsella (56), Gierhart and Potter (57), Abramov et al. (58), Kovar et al. (59), Oliveira et al. (60) Emulsions Galluzzo and Regenstein (61), Tran and Einerson (62) Proteolysis Juffs (63), Ory and Sekul (64), Kwan et al. (65), Siddiqui et al. (66), Ramana-Murthy et al. (67)

7.1. Cereal, Legume, and Flours Sathe and Salunkhe (45) extracted proteins from the great northern bean (Phaseolus vulgaris L) before assaying by the Lowry method.* Results were standardized against the Kjeldahl method (%N 6 6.25). Whole bean ¯our had 26.1% (w/w) protein (per dry weight) or 22.17% (w/w) crude protein. About 21.8% of the total protein was water soluble (albumin). Another 73.4% could be extracted with dilute salt solutions (globulin). Table 5 shows virtually quantitative recovery of bean protein using solutions of sodium carbonate (0.5%), potassium sulfate (5%), SDS (5%), or sodium hydroxide (0.02 N or 1 N). Sodium phosphate buffer was not an ef®cient extractant for bean ¯our protein. Urea and SDS, also commonly used as protein- solubilizing agents, did not perform so well. The extent of protein recovery by extraction is usually uncertain. Depending on the solvent, different protein fractions (albumin, globulins, prolamins, etc.) are extracted to different extents. The direct addition of Lowry reagent C to ¯our samples should be investigated. This approach

* Suspend whole bean ¯our (1 g) in 25 mL of solvent with thorough mixing. Let stand for 12 hours and then centrifuge (5000g) for 45 minutes. Analyze for protein content using Method 1. The Lowry Method 89

TABLE 5 Analysis of Bean Flour Protein Extracts Using the Lowry Assay

Solventa pH Total protein (%)

Sodium chloride 6.2 61.9 Sodium sulfate 6.3 65.3 Sodium acetate 6.5 52.9 sodium carbonate 10.0 93.4 Sodium dihydrogen phosphate 6.75 55.2 Sodium phosphate 5.0 38.3 Potassium chloride 6.7 57.4 Potassium sulfate 5.9 75.9 (100) Urea 7.2 69.8 (100) SDS 7.1 60.2 (100) HCl 1.5 75.6 Dimethylformamide 6.0 41.6 aAll salts were used at concentrations of 0.5% (w/v) aqueous solution. Values in parentheses are % extraction using 5% (w/v) solutions. Source: Adapted from Ref. 45. worked well for biuret analysis of ¯our (Chapter 2). After centrifugation, Lowry analysis could then be completed by adding Folin-Ciocalteu reagent. Sebecic (46) considered the Lowry procedure as a ``new'' method for wheat protein determination. Forty-®ve varieties of Yugoslav wheat were analyzed.* The Lowry and Kjeldahl methods were highly correlated. The sensitivity of these methods was 100-fold higher than that of the biuret assay. Sebecic concluded that the Lowry procedure was a good substitute for the Kjeldahl or Kjel-Foss method. It was certainly faster, simpler, less expensive, and easier to perform than the Kjeldahl method. Seguchi (49) analyzed starch granule surface proteins after extraction with 1% SDS overnight. Although present in very small amounts (0.06% w/w), the speci®cally bound protein may have important effects on the functional properties of wheat ¯our. Treating wheat ¯our at high temperatures (e.g., 1008C for 40 hours or 608C for 24 days) or storing at room temperature for long periods (130 days) caused greater than a 300% increase in wheat starch granule surface proteins.

* Wheat ¯our (0.4 g) was suspended in 50 mL of distilled water and heated at 958C for 2 minutes to gelatinize starch. The mixture was cooled and brought to a total volume of 100 mL. To the prediluted samples (with 4±17 mgmLÀ1 protein) were added 10 mL of Lowry reagent C. Folin-

Ciocalteu reagent (1 mL) was added after 30 minutes and A540 readings recorded 30 minutes later. 90 Chapter 3

7.2. Leaf Proteins and Protein from Other Chlorophyllous Tissue Adding 100 mgmLÀ1 chlorophyll to model BSA solutions (300 mgmLÀ1) increased the Lowry assay response by 400%. The interference from chlorophyll can be avoided by (a) precipitating the solubilized plant proteins with TCA, (b) decoloring the homogenized leaf extract by acetone treatment, or (c) extracting the freeze-dried leaf powder with acetone (25). The last strategy is most attractive.* Attempts to decolorize the leaf homogenate led to signi®cant losses of protein. The preceding method has general applicability. Most chlorophyllous tissue could be treated in this fashion: green fruits, stems, shoots, and green aquatic plants. Algae are another potentially rich source (30±70% dry weight) of crude protein (68).

7.3. Emulsions and Nondairy Creamers Protein stabilized emulsions are centrifuged to separate the cream (oil-rich) and aqueous phases. The aqueous layer is then analyzed for protein content. Tran and Einerson (62) used the Lowry assay to assess the stability of nondairy creamers. Emulsion samples were diluted (0.2% w/w) with distilled water to which salts had been added to simulate hardened water.{ The aqueous phase protein concentration, following emulsi®cation, gives an indication of protein emulsifying capacity. Galluzzo and Regenstein (61) de®ned emulsifying capacity as the amount of oil emulsi®ed per mg of protein. The order of emulsifying capacity for muscle proteins was myosin > actomyosin ( ATP added) > actomyosin > actin.

7.4. Monitoring Proteolysis and Digestion Anson's (69) method for monitoring proteolysis is straightforward. Adjust the sample to 10% ®nal concentration of TCA and centrifuge or ®lter. Take

A280 measurements for the TCA-soluble supernatant and a quartz cuvette. The assay sensitivity is increased 10-fold by reacting with Folin-Ciocalteu

* Shake freeze-dried or ground leaf tissue (50 mg) with 15 mL of acetone (62). Filter and store the decolorized powder for analysis. Extract protein from the decolorized leaf powder and analyze as usual. { Emulsions were centrifuged and to the aqueous phase (0.5 mL) was added 0.5 mL of SDS (10% w/w) as a dispersant. Lowry reagent C (5 mL) was added followed, 10 minutes later, with Folin-Ciocalteu reagent (1 mL). Absorbance readings were taken some 30 minutes later. Turbid samples were clari®ed by one or more freeze-thaw cycles before spectrometric measurements. The Lowry Method 91 reagent.* Hull (70) used this approach to monitor proteolysis in milk. The results were expressed as tyrosine equivalents by referring to a calibration graph for tyrosine. Juffs (63) found that tyrosine equivalents were signi®cantly correlated with milk bacterial count, age of the cow, period of lactation, and milk yield. There were variations in tyrosine equivalence for normal milk, and it was not possible to set a value for proteolyzed milk samples. The concentration of phosphotyrosine residues can also affect milk tyrosine values. Phosphotyrosine does not react with the Folin-Ciocalteu reagent (71). The total tyrosine value was determined after treating samples with added alkaline phosphates. It is therefore suggested that the Folic- Ciocalteu phenol reagent be used as basis for assaying phosphotyrosine phosphatase. Finally, milk proteolysis was monitored using the Lowry,

¯uorescamine, or trinitrobenzene sulfonate method or by A280 readings (65). The Lowry and Anson methods were highly correlated (R 0.99). Although 10-times more sensitive, the Lowry assay showed less agreement with techniques that monitor free amino groups. Ory and Sekul (64) recommended that milk proteolysis should be routinely measured with quantitative SDS±polyacrylamide gel electropho (SDS-PAGE) as well as spectrophotometry. Gut ¯uid is a challenging milieu for protein analysis. Crossman et al. (72) compared the modi®ed Lowry assay and other techniques (quantitative amino acid analysis,{ BCA, and Bradford method) for monitoring protein levels in the gut ¯uid from the marine herbivorous ®sh Kyphosus sydneyanus. The Lowry and BCA assays gave similar results. All spectrophotometric methods showed a low correlation with quantitative amino acid analysis. The results were also dependent on the mode of sample collection and pretreatment. In was important to freeze ®sh digesta immediately after collection. The sample could be thawed and gut ¯uid removed closer to the time for protein analysis. The Lowry assay was also applied in the study of dietary protein effects on sul®de production by bacteria in the human large intestines (73).

7.5. Yeast and Other Single-Cell Proteins Single-cell protein (SCP) is produced by growing microorganisms on a range of substrates including hydrocarbons and sweet whey (68,74,75). Growth

* Lowry reagent C is not required. { Quantitative amino acid analysis is described in Chapter 1. Descriptions of the BCA and Bradford methods appear in Chapters 4 and 5. 92 Chapter 3 media contain a range of interferences for protein assay. Two examples will illustrate the protein analysis issues. Vananuvat and Kinsella (56) grew Saccharomyces fragilis on spray-dried crude lactose (2% w/w) supplemented with ammonium sulfate (0.5%), peptone (0.5%), yeast extract (0.5%), and urea (0.03%) as nitrogen source. Microbial cell numbers and protein and nucleic acid concentrations were monitored during the fermentation. Protein concentrations from the Lowry method were always lower than Kjeldahl protein (56). Sample pretreatment to remove low-molecular-weight substances is recommended. Given the complexity of growth media, quantitative SDS-PAGE is probably a good idea for protein quantitation. Orban et al. (76) examined the production of yeast autolysates from Kluyveromyces fragilis grown on pasteurized whey (2%) supplemented with ammonium sulfate (0.4% w/v), potassium dihydrogen phosphate (0.2% w/v), and yeast extract. The cells were grown for about 8 hours in a batch fermenter and harvested by centrifugation. Autolysis was initiated by suspending freeze-dried cells (10% w/v) in dilute saline (5% w/v NaCl) solution. Sample results are given in Table 6. Determinations of SCP are usually performed after washing the wet cells free from growth medium. By washing the cells thoroughly, interferences from culture medium components (sugars, peptides, and proteins) can be avoided. However, Lowry and Kjeldahl results were consistently different because of interferences by the high levels of ribonucleic acid found in yeast cells.

TABLE 6 Determination of Protein and Other Components in Kluyveromyces fragilis cells before and after Autolysis

Dried yeast Autolysatea Parameter (% dw) (% dw)

Kjeldhal protein (N 6 6.25) 49.7 37.3 Lowry protein 40.4 19.7 TCA-soluble nitrogen 2.0b 5.8b Amino nitrogen 2.0 3.9 RNA 7.7 9.0 Carbohydrate 41.0 17.0 Moisture 2.2 2.92 a Optimal conditions of autolysisÐ10% yeast cells are suspended in 5% sodium chloride and incubated at 548C for 8 hours. bDetermined from Kjeldhal nitrogen analysis of material soluble in 20% TCA. Source: Data from Ref. 76. The Lowry Method 93

TABLE 7 Improved Lowry Methods for Collagen Analysis

Sensitivity

(ÁA562/ Linear Color Sample Method gÀ1) LLDa (g) range (g) stabilityb volume (L)

Modi®ed 0.0033 4 4±100 0.85 200 Enhanced 0.0085 3 3±80 0.85 200 aLLD, lower limit of detection. bColor stability % ®nal absorbance change in 10 minutes.

7.6. Collagen

À1 The sensitivity of the Lowry assay to collagen is low (0.1 A750 mL mg ) because of the low content of aromatic (oxidizable) amino acids (*1%). Copper binding is also inhibited by the high (hydroxyl) proline content (> 25%) and the triple-helix structure of collagen, which is apparently stable in alkaline media. Komsa-Penkova et al. (77) described two procedures for improving the Lowry assay sensitivity for collagen. In the ``standard modi®cation'' protocol, samples of collagen (200 mL) are heated with 180 mL of Lowry reagent A (see Method 1) at 508C for 20 minutes. Reagent B is added and the mixture is allowed to react at room temperature for 10 minutes. Finally, Folin-Ciocalteu reagent (600 mL) is added and the mixture heated at 508C for another 10 minutes before taking absorbance readings at 562 nm. With the ``enhanced protocol,'' collagen is reacted with Lowry reagent C at 508C for 20 minutes, cooled to room temperature, and then Folin-Ciocalteu reagent is added. Absorbance readings are taken 10 minutes later. Table 7 shows the performance characteristics of the proposed Lowry methods for collagen. The relative response to type I, type II, and type IV collagen was the same and 50% lower than the response obtained with type V collagen. The modi®ed methods were 10- to 20-fold more sensitivity than Method 1 for collagen analysis. The sensitivity increase is due to the denaturation of collagen by thermal treatment. Gelatin can be analyzed with the normal Lowry method at room temperature.

REFERENCES

1. OH Lowry, NJ Rosebrough, AL Farr, RJ Randall. Protein measurement with the Folin phenol reagent. J Biol Chem 193:265±273, 1951. 94 Chapter 3

2. GL Peterson. A simpli®cation of the protein assay method of Lowry et al. which is more generally applicable. Anal Biochem 83:346±356, 1977. 3. H Wu. Contribution to the chemistry of phosphomolybdic acid, phospho- tungstic acids and allied substances. J Biol Chem 43:189±220, 1920. 4. H Wu. A new colorimetric method for the determination of plasma proteins. J Biol Chem 51:33±39, 1922. 5. O Folin, V Ciocalteu. On tyrosine and tryptophan determination in proteins. J Biol Chem 73:627±650, 1927. 6. RM Herriot. Reactions of Folin [phenol] reagent with proteins and biuret compounds in the presence of cupric ion. Proc Soc Expt Biol Med 46:642±644, 1941. 7. E Lane. Spectrophotometric and turbidimetric methods for measuring proteins. Methods Enzymol 13:447±453, 1957. 8. GL Peterson. Review of the Folin phenol method of Lowry, Rosebrough, Farr and Randall. Anal Biochem 100:201±220, 1979. 9. GL Peterson. Determination of total protein. Methods Enzymol 91:95±119, 1983. 10. A Bensadoun, D Weinstein. Assay of proteins in the presence of interfering materials. Anal Biochem 70:241±250, 1976. 11. S-C Chou, A Goldstein. Chromogenic groupings in the Lowry assay protein determination. Biochem J 75:109±115, 1960. 12. A Livitski, M Anbar, A Berger. Speci®c oxidation of peptides via their copper complexes. Biochemistry 6:3757±3767, 1967. 13. JL Kurtz, GL Burce, DW Margerum. Trivalent copper catalysis of the autooxidation of copper (II) tetraglycine. Inorgan Chem 17:2455±2460, 1978. 14. G Legler, CM Muller-Plantz, M Mentges-Hettkamp, G P¯ieger, E Julich. On the chemical basis of the Lowry protein determination. Anal Biochem 150:278± 287, 1985. 15. ME Anderson, RT Marshall. An automated continuous protein analyzer: modi®cation of Lowry method. J Food Sci 40:728±731, 1975. 16. YW Huang, RT Marshall, ME Anderson, C Charoen. An automated modi®ed Lowry method for protein analysis of milk. J Food Sci 41:1219±1221, 1976. 17. CE Stauffer. A linear standard curve for the Folin Lowry determination of protein. Anal Biochem 69:646±648, 1975. 18. WT Coakley, CJ James. A simple linear transform for the Folin-Lowry protein calibration curve to 1.0 mg/ml. Anal Biochem 85:90±97, 1978. 19. MA Peters, JR Fouts. Interference by buffers and other chemicals with the Lowry protein determination. Anal Biochem 30:299±301, 1969. 20. TH Ji. Interference by detergents, chelating agents and buffers with the Lowry protein determination. Anal Biochem 52:517±521, 1973. 21. J Bonitati, WB Elliott, PG Miles. Interference by carbohydrates and other substances in the estimation of protein with the Folin-Ciocalteu reagent. Anal Biochem 31:399±404, 1969. The Lowry Method 95

22. J O'Sullivan, GE Mathieson. Interference by monosaccharides with the estimation of tyrosine and proteins using the Folin-Ciocalteu phenol reagent. Anal Biochem 42:540±543, 1970. 23. F Toldra. Effect of glucose and maltose on the Lowry assay. Nahrung 33:795± 796, 1989. 24. I Alkorta, MJ Llama, JL Serra. Interference by pectin in protein determination. Lebensm Wiss Technol 27(1):39±41, 1994. 25. DH Berg. Hexoseamine interference with the determination of protein by the Lowry procedure. Anal Biochem 42:505±508, 1971. 26. JMO Eze, ED Dumbroff. A comparison of the Bradford and Lowry methods for analysis of protein in chlorophyllous tissue. Can J Bot 60:1046±1049, 1982. 27. GW Pace, MC Archer, SR Tannenbaum. The effect of cryoprotective agents on the Lowry protein assay. Anal Biochem 60:649±652, 1974. 28. MK Zishka, JS Nishimura. Effect of glycerol on Lowry and biuret methods of protein determination. Anal Biochem 34:291±297, 1970. 29. B Gerhadt, H Beevers. In¯uence of sucrose on protein determination by the Lowry procedure. Anal Biochem 24:337±352, 1968. 30. H Schuel, R Schuel. Automated determination of protein in the presence of sucrose. Anal Biochem 20:86±93, 1967. 31. HL Rosenthal, WA Sobieszczanska. In¯uence of reducing sugars on protein determination by the Lowry procedure. Anal Biochem 34:591±598, 1970. 32. C-H Lo, H Stelson. Interference by polysucrose in protein determination by the Lowry method. Anal Biochem 45:331±336, 1972. 33. J Eichberg, LC Mokrasch. Interference by oxidized lipids in the determination of protein by the Lowry procedure. Anal Biochem 30:386±390, 1969. 34. MB Lees, S Paxman. Modi®cation of the Lowry procedure for the analysis of proteolipid protein. Anal Biochem 47:184±192, 1972. 35. AAK Markwell, SM Hass, LL Bieber, NE Tolbert. A modi®cation of the Lowry procedure to simplify protein determination in membrane and lipoprotein samples. Anal Biochem 87:206±210, 1978. 36. LP Kirazov, LG Venkov, EP Kirazov. Comparison of the Lowry and the Bradford protein assays for protein estimation of membrane-containing fractions. Anal Biochem 208:44±48, 1993. 37. CG Vallejo, R Lagunas. Interferences by sulfhydryl, disul®de reagents and potassium ions on protein determination by Lowry's method. Anal Biochem 36:207±212, 1970. 38. PJ Geiger, SP Bessman. Protein determination by the Lowry method in the presence of SH reagents. Anal Biochem 49:467±473, 1972. 39. E Ross, G Schatz. Assay of protein in the presence of high concentrations of sulfhydryl compounds. Anal Biochem 54:304±306, 1973. 40. M Higuchi, F Yoshida. Lowry determination of protein in the presence of sulfhydryl compounds or other reducing agents. Anal Biochem 77:542±547, 1977. 41. M Higuchi, F Yoshida. An improved Lowry procedure by using chloramine-T under the neutral or alkaline conditions. Agric Biol Chem 42:75±77, 1978. 96 Chapter 3

42. V Hii, WC Herwig. Determination of high molecular weight proteins in beer using Coomassie Blue. J Am Soc Brew Chem 40(2):46±50, 1982. 43. KM Williams, P Fox, T Marshall. A comparison of protein assays for the determination of the protein concentration of beer. J Inst Brew 101:365±369, 1995. 44. VW Padhye, DK Salunke. Biochemical studies on black gram (Phaseolus mungo): I. Solubilization and electrophoretic characterization of the proteins. J Food Biochem 1:111±129, 1977. 45. SK Sathe, DK Salunkhe. Solubilization and electrophoretic characterization of the Great Northern bean (Phaseolus vulgaris L.) proteins. J Food Sci 46(1):82± 87, 1981. 46. B Sebecic. A new possibility of wheat protein content determination. Nahrung 31:817±823, 1987. 47. O Paredes-Lopez, LF Gueverra, ML Schevenim-Pinedo, R Montes-Rivera. Comparison of procedures to determine protein content of developing bean seeds (Phaseolus vulgaris). Plant Foods Hum Nutr 39(2):137±148, 1989. 48. M Seguchi. Study of wheat starch granule surface proteins from chlorinated wheat ¯ours. Cereal Chem 67:258±260, 1990. 49. M Seguchi. Effect of wheat ¯our aging on starch-granule surface proteins. Cereal Chem 70:362±364, 1993. 50. Y W Huang, RT Mashall, ME Anderson, C Charoen. Automated modi®ed Lowry method for protein analysis of milk. J Food Sci 41:1219±1221, 1976. 51. TA Kroening, P Mukerji, RG Hards. Analysis of beta-casein and its phosphoforms in human milk. Nutr Res 18:1175±1186, 1998. 52. EM Ogunbunmi, O Bassir. Proteins and amino acid contents of some Nigerian food condiments. Nutr Rep Int 22:497±502, 1980. 53. JE Hoff. A simple method for the approximate determination of soluble protein in potato tubers. Potato Res 18:428±432, 1975. 54. J Vigue, PH Li. Correlation between methods to determine the protein content of potato tubers. HortScience 10:625±627, 1995. 55. SJ Latlief, D Knorr. Tomato seed protein concentrates: effects of methods of recovery upon yield and compositional characteristics. J Food Sci 48:1583± 1586, 1983. 56. P Vananuvat, JE Kinsella. Production of yeast protein from crude lactose by Saccharomyces fragilis. Batch culture studies. J Food Sci 40:336±341, 1975. 57. DL Gierhart, NN Potter. Effects of ribonucleic acid removal methods on composition and functional properties of Candida utilis. J Food Sci 43:1705± 1713, 1978. 58. ShA Abramov, DA Efendieva, STS Kotenko. Effect of the growth medium on the protein content of the yeast Saccharomyces cerevisiae. Appl Biochem Microbiol 30:225±227, 1994. 59. L Kovar, V Benda, B Hodrova, M Marounek. Fermentation of glucose, xylose, cellulose and waste paper by the rumen anaerobic fungus Orpinomyces joyonii. A J. Anim Feed Sci 9:727±735, 2000. The Lowry Method 97

60. MA Oliveira, C Rodrigues, EM dos Reis, J Nozaki. Production of fungal protein by solid substrate fermentation of cactus Cereus peruvianus and Opuntia ®cus indica. Quim Nova 24:307±310, 2001. 61. SJ Galluzzo, JM Regenstein. Emulsion capacity and timed emulsi®cation of chicken breast muscle myosin. J Food Sci 43:1757±1760, 1978. 62. KM Tran, MA Einerson. A rapid method for the evaluation of emulsion stability of non-dairy creamers. J Food Sci 52:1109±1110, 1987. 63. HS Juffs. Proteolysis detection in milk. I. Interpretation of tyrosine value data for raw milk supplies in relation to natural variation, bacterial counts and other factors. J Dairy Res 40:371±381, 1973. 64. RL Ory, AA Sekul. Spectrophotometric assay curves as anomalous indicators of proteolysis of oilseed proteins. J Food Biochem 1:67±74, 1977. 65. KKH Kwan, S Nakai, BJ Skura. Comparison of four methods for determining protease activity in milk. J Food Sci 48:1418±1421, 32, 1983. 66. SF Siddiqui, MK Pasha, F Ahmad, M Ahmad. Digestibility of some non- conventional seed proteins. J Oil Technol Assoc India 26(2):49±51, 1994. 67. MV Ramana-Murthy, Sriram-Padmanabhan, M Ramakrishna, BK Lonsane. Comparison of nine different caseinolytic assays for estimation of proteinase activity and further improvement of the best method. Food Biotechnol 11:1±23, 1997. 68. RK Robinson, DF Toerien. The algae: a source of protein. In BJF Hudson, ed. Developments of Food Proteins, Vol 1. Barking-Essex NJ: Applied Science Publishers, 1984, pp 289±325. 69. ML Anson. Estimation of pepsin, trypsin, papain and cathepsin with hemoglobin. J Gen Physiol 22:79±89, 1938. 70. ME Hull. Colorimetric determination of partial hydrolysis of the proteins in milk. J Dairy Sci 30:881, 1947. 71. BMK Gmeiner, CC Seelos. Measurement of phosphotyrosine phosphatase activity using the Folin-Ciocalteu phenol reaction. Biochem Mol Biol Int 36:943±948, 1995. 72. DJ Crossman, KD Clements, GJS Cooper. Determination of protein for studies of marine herbivory: a comparison of methods. J Exp Mar Biol Ecol 244:45±65, 2000. 73. EA Magee, CJ Richardson, R Hughes, JH Cummings. Contribution of dietary protein to sul®de production in the large intestine: an in vitro and a controlled feeding study in humans. Am J Clin Nutr 72:1488±1494, 2000. 74. CL Cooney, CK Rha, SR Tannembaum. Single-cell protein: engineering, economics and utilization in foods. In CO Chichester, ed. Advances in Food Research 26. New York: Academic Press, 1980, pp 1±52. 75. M Guzman-Juarez. Yeast protein. In BJF Hudson, ed. Developments in Food ProteinsÐ3. London: Elsevier Applied Science, 1982, pp 263±291. 76. E Orban, GB Qualia, I Casini, M Moresi. Effect of temperature and yeast concentration on the autolysis of Kluyveromyces fragilis grown on lactose based media. J Food Eng 21:245±261, 1994. 98 Chapter 3

77. R Komsa-Penkova, R Spirova, B Bechev. Modi®cation of Lowry's method for collagen concentration measurement. J Biochem Biophys Methods 32:33±43, 1996. 4

The Bicinchoninic Acid Protein Assay

1. INTRODUCTION

The bicinchoninic acid (BCA) protein assay was developed by Smith et al. (1). The plan was to substitute BCA for the Folin-Ciocalteu reagent in the Lowry assay. The advantages of the BCA method include decreased sensitivity to interferences, a need for one working reagent, and color stability. The BCA protein assay has the same sensitivity as the Lowry method. Reagents for the BCA assay are available commercially. As yet, the BCA assay does not feature greatly in the food science literature. As of June 2001, there were 16 references to bicinchoninic acid or bicinchoninate in the Food Science and Technology abstracts. General citations in the Science Citation Index number over 8500. The BCA assay is without a doubt more popular than indicated by the low number of formal citations. The characteristics of the BCA protein assay are reviewed in this chapter. Section 1 is an account of the history of BCA reagent and its use for chemical analysis. In Sec. 2, the basic BCA procedure is presented, followed, in Sec. 3, by a discussion of the chemistry underlying color formation. In Secs. 4±6 are descriptions of calibration features, interfering compounds, and sample pretreatment strategies for avoiding error. After a discussion of

99 100 Chapter 4 automated formats (Sec. 7), we turn to application of the BCA assay to food protein analysis in Sec. 8.

1.1. Determination of Copper Using BCA BCA is the trivial name for 2,20-diquinolyl-4,40-dicarboxylic acid. Early applications include the analysis of Cu2 in metal alloys and blood sugars. Hoste (2) evaluated 2,20-diquinolyl (2,20DQ) and nine related heterocyclic compounds for Cu2 analysis after reducing Cu2 to Cu1 with hydroxylamine hydrochloride. He found that 2,20DQ forms a 2:1 complex 1 with Cu (Fig. 1) having maximum absorbance (lmax) at 540 nm and a À1 À1 1 molar extinction coef®cient (De540) of 5490 M cm . To determine Cu , the working solution of 2,20DQ (0.02% w/v in ethanol, 10 mL) was added to

3 mL of sample. The mixture was diluted with 25 mL of ethanol and A540 measurements were recorded. The Cu1 was quantitatively determined in the presence of Cu2, with which there is no reaction. KerteÂ sz (3) employed 2,20DQ (0.05% w/v in glacial acetic acid) for the analysis of copper at the active site of the enzyme tyrosinase. To calibrate the assay, 2 mL of 2,20DQwas added to enzyme-free samples containing

FIGURE 1 Diagram of the 2:1 complex formed between 2,20-diquinolyl (R H) or 2,20-bicinchoninic acid (R COOH) and Cu1 The Bicinchoninic Acid Protein Assay 101

< 11.5 mgofCu1 in 2 mL of phosphate buffer (0.05 M, pH 6.8). First, Cu2 was reacted with hydroxylamine hydrochloride to produce Cu1. Upon the addition of 2,20DQ, a purple complex formed in 5±10 minutes at room temperature. There was no reaction with Cu2, although preincubating Cu2 with BSA or conalbumin led to reactivity with 2,20-DQ. Initial studies using 2,20DQsuggested that tyrosinase had a Cu 1 atom at the active site. Fesenfeld (4) also investigated the reaction of tyrosinase with 2,20DQ. He proposed that in the absence of substrate mushroom tyrosinase had an active site Cu2 species and that this might easily be reduced to Cu1 by a protein SH group. Colorimetric analysis of Cu1using BCA was studied by Gershuns et al. (5). They showed that Cu1 forms a complex with BCA and that this was insoluble at pH < 4. At high pH the red-violet complex had a lmax value of 560 nm. The linear range for Cu1 analysis was 1±100 mgmLÀ1. The analytical precision was 1%. There were no interferences from common 3 1 metal ions when present at 6±12 6 10 -fold mole excess. The Cu (BCA)2 complex was reportedly stable for a few hours at pH 4±12, although the optimal pH for Cu1 analysis was pH 6. As with previous investigations, Cu2 was analyzed after reduction using hydroxylamine hydrochloride. Nakano (6) also employed BCA for the analysis of Cu2. In a long- running study, the characteristics of several 4,40-substituted DQ derivatives were investigated. As shown in Table 1, BCA was the second most colorigenic derivative. Values in Table 1 are with water or isoamyl alcohol as solvent. The ratio of BCA to Cu1 found in the complex was con®rmed as 1 2:1. The Cu (BCA)2 complex had a lmax value of 565 nm and De565 equal

TABLE 1 Properties of Cu1 Complexes with 2,20DQ derivativesa

0 À1 À1 4,4 Substituent (R) max (nm) Á" (M cm ) 22COOHb 560±565 7920±8000 22CONHEt 565 3100

22CONHCHMe2 566 7100 22CONMe2 560 6130 22CONEt2 561 6200 22CON (CHMe2)2 556 28100 22CONBt2 561 5150 22COOEt 576 6600 22COOBt 576 6020 aValues are with water or isoamyl alcohol as solvent. bR COOH for bicinchoninic acid (BCA). Source: Compiled from Ref. 6. 102 Chapter 4 to 8000 MÀ1 cmÀ1. The color produced from BCA and Cu1 is reportedly stable for 48 hours at temperatures < 658C and at pH 3 to 13. The linear range for Cu1 analysis was 0.02±20 mgmLÀ1. Reports of the analysis of Cu2 using BCA were produced by Musta®n et al. (7) and Tikhonov (8). Noskova (9) used BCA for the photometric determination of copper in blood serum. The dissociation 1 À11 constant (Kd) for the Cu (BCA)2 complex was estimated as 1 6 10 Mat 208C by Buhl et al. (10). Capitan et al. (11) developed a solid-phase analysis of copper in natural water. The microdetermination of copper involved the reduction of Cu2 to Cu1, reaction with BCA, and adsorption of the resulting complex with a dextran cation exchange resin. Absorbance measurements were recorded directly using the resin phase. Copper was determined at concentrations of 1±20 6 10À9 gLÀ1. The relative standard deviation for analysis was 1.7%. The typical sample size was 2 L. Brenner and Harris (12) determined serum copper levels using BCA. Blood plasma (0.75 mL) was deproteinized by treating with 0.25 mL of TCA (30% w/w). After microcentrifugation, 0.5 mL of deproteinized serum was added to 0.1 mL of ascorbic acid (35±2 mg% w/w) to reduce Cu2 to Cu1. Then 0.4 mL of buffered BCA reagent was added and absorbance readings were measured. For calibration, standard amounts of copper were analyzed using 0.1 M sodium phosphate buffer as solvent. 1 A second lmax value was reported for the Cu (BCA)2 complex at 354.5 nm (13). At this wavelength the sensitivity toward copper is six to 4 À1 À1 seven times greater (De354.5 4.6 6 10 M cm ) than the sensitivity 3 À1 À1 562 nm (De562 7.7 6 10 M cm ). For a system containing 50 mM BCA, the linear dynamic range for copper detection was 2±25 mM. The upper limit of detection is consistent with the 2:1 stoicheometry for BCA 1 1 binding to Cu . The pH stability characteristics of the Cu (BCA)2 1 complex can be seen from Table 2. These results show that Cu (BCA)2 is

1 TABLE 2 The pH Stability of the Cu (BCA)2 Complex

4 À1 À1 Solution pH Á"354.5 /10 (M cm ) 4.5 3.95 (Æ0.1) 7.0 4.58 (Æ0.1) 9.5 4.43 (Æ0.2) 10.5 4.73 (Æ0.2) 12.5 3.88 (Æ0.2)

Source: Adapted from Ref. 12. The Bicinchoninic Acid Protein Assay 103 stable at pH 3±13;a similar conclusion is supported by absorbance measurements at 560±565 nm. At the time of writing, no protein assays have been conducted at 354.5 nm.

1.2. Analysis of Sugars Using BCA Between 1971 and 1973, researchers af®liated with Pierce-Warner Chemical Company used BCA for sugar analysis. An early iteration of the BCA reagent developed by Grindler (14) and Mopper and Grindler (15) was prepared from two stock solutions. Reagent A comprised BCA (170 g) and sodium carbonate (27 g) dissolved in 500 mL of distilled water. Reagent B contained copper sulfate (1 g) and aspartic acid (2.55 g) dissolved in 500 mL of distilled water. A working BCA solution was prepared by mixing equal volumes of reagents A and B. The analysis of sugars was performed in a ¯ow system using a reaction coil at 808C. The reaction time was 25 minutes and absorbance readings were recorded at 562 nm. McFeeters (13) described a manual BCA assay for sugars. Reagent A comprised 0.085% (w/v) BCA in 1.0 M phosphate buffer (pH 8.5). Reagent B contained 25 g of aspartic acid and 33.4 g of sodium carbonate predissolved in 500 mL of water to which copper sulfate (1.34%, 500 mL) was then added. A working BCA reagent was prepared from 23 parts of reagent A and 1 part of reagent B. In a typical analysis, 3 mL of working BCA reagent was added to a 1-mL solution of sugar. The mixture was heated in a boiling-water bath for 10 minutes. After allowing the samples to cool, absorbance readings were taken at 560 nm. The following sugars were assayed: galacturonic acid, glucuronic acid, glucose, galactose, fructose, mannose, xylose, maltose, melibiose, and cellobiose. The average color yield was 1.9 (+1.0) DA560 per mmole. The utmost color yield was obtained with fructose (10.76 DA560 per mmole). For a within-cuvette fructose concentra- À9 tion of 120 6 10 mole, DA560 was 1.29.

2. THE BCA PROTEIN ASSAY

As described earlier, the BCA protein assay was developed by a 10-member team (1) from the Biochemical Research Division, Pierce Chemical Company, Rockford, Illinois.* Their objective was to ®nd an alternative to the Folin-Ciocalteu reagent for detecting Cu1, which is formed when Cu2 is reduced by proteins. Actually, the BCA assay may be a totally new

* In Europe, Pierce and its sister company HyClone trade under the name Perbio Science. 104 Chapter 4 concept in protein analysis. The underlying reactions are signi®cantly different from those for the Lowry assay. The distinctiveness of the BCA- protein reaction probably accounts for the greater resistance to certain interferences. BCA reagent is prepared from two stock reagents. Reagent A and B are then mixed in a ratio of 50:1 to produce the working BCA solution.

Method 1 The bicinchoninic acid protein assay (1).

Reagents 1. Reagent A (alkaline BCA reagent). Add BCA (10 g), sodium carbonate (20 g), sodium tartrate (1.6 g), sodium hydroxide (4 g), and sodium hydrogen carbonate (9.5 g) to 500 mL of distilled water. Adjust to pH 11.25 using sodium hydroxide or solid sodium carbonate. Make up to a volume of 1 L. Reagent A is apparently stable inde®nitely at room temperature. 2. Reagent B (4% copper sulfate). Dissolve 4 g of copper sulfate CuSO4 Á 5H2O in 100 mL of deionized water. Reagent B is, reportedly, stable inde®nitely at room temperature. 3. Working BCA solution. Mix 50 volumes of reagent A with 1 volume of reagent B. Prepare the working BCA solution daily.*

Procedure Add 100 mL of sample (20±120 mg protein) to exactly 2 mL of BCA working reagent. As long as the 1:20 volume ratio is maintained, other sample and BCA reagent volumes can be used. Prepare a reagent blank by replacing 100 mL of sample with the same volume of distilled water. Incubate the mixture at 378C (30 minutes) or at room temperature (2 hours). The assay time may be reduced to 10±15 minutes by

performing the reaction at 608C. Record A562 against a reagent blank. The BCA assay was characterized with respect to the linear dynamic range, assay temperature, reaction pH, effect of interferences, and reagent stability. Performance characteristics were also compared with those from the Lowry assay. At room temperature, DA562 readings increased slowly over 21 hours. Heating the reaction mixture at 378C for 60 minutes

* This is a precautionary measure. The working BCA reagent was stable for over 7 days. A calibration graph produced using the 7-day-old BCA working reagent was virtually identical to a graph produced with a freshly prepared reagent. The Bicinchoninic Acid Protein Assay 105 produced a similar amount of color. Performing the assay at 608C increased the assay sensitivity four- and ®vefold (1). The effect of pH on the BCA assay was evaluated by altering the pH of reagent A with sodium hydroxide or solid sodium carbonate. The DA562 readings showed a bell-shaped response over the pH range 10.2±12.0. The optimal color yield occurred at pH 11.25. The working BCA reagent had suf®cient buffer capacity to provide accurate results for protein samples containing 0.1 N sodium 1 hydroxide or 0.1 N hydrochloric acid. Because Cu (BCA)2 is stable at pH 3±13 (Table 2), the pH range of Smith's assay is possibly dictated by the effect of pH on the protein-copper reaction. Protein-protein variations in the color yield follow a pattern similar to those for the Lowry assay (Chapter 3). The order of increasing color formation is gelatin < avidin < BSA < immunoglobin G < chymotrypsin < insulin < ribonuclease. The color yield from gelatin was only 50% of the value for avidin. A high concentration of (hydroxyl) proline residues probably reduces Cu2 binding to collagen. There is a further loss of color yield assay owing to the low concentrations of tyrosine and tryptophan in the gelatin chain (see later). Protein-protein variations in assay results are lower at 608C. For samples incubated at room temperature or 378C, DA562 readings increased slightly (0.25% per minute) after the prescribed assay time. For increased precision, the absorbance changes for all samples should be taken within a short time of each other.

3. CHEMISTRY OF THE BCA PROTEIN ASSAY

Cu2 reacts with the protein to produce to Cu1, which then binds to BCA. The reaction is different from those taking place for the Lowry assay (Chapter 3). In the latter case, the concentration of Cu1 remains very low due to the action of Mo6/W6. This strong oxidant converts Cu1 to Cu2 and Cu3.

3.1. Reactions of the BCA Protein Assay Two reactions occur between proteins and Cu2 during the BCA assay. The ®rst is a temperature-insensitive reaction between Cu2 and oxidizable protein side chains (tyrosine, tryptophan, cysteine). The second reaction is a temperature-sensitive process involving the binding of Cu2 to the peptide backbone. Once bound, Cu2 undergoes reduction to Cu1. Assays performed at high temperatures assist the peptide pathway. Because the peptide backbone is the same for different proteins, a high-temperature BCA assay lessens protein-protein variations in results. Wiechelman et al. 106 Chapter 4

TABLE 3 Some Features of the BCA Protein Interactions

1. Four amino acids (cysteine, tryptophan, tyrosine, and phenylalanine) react with BCA working reagent. 2. Biuret and dipeptides (unable to form a tetradentate Cu2 complex) do not react with the BCA reagent. 3. Di- or tripeptides containing tyrosine or tryptophan react with the BCA reagent. 4. There is no correlation between the redox potential and the color yield from the BCA assay. 5. The colors from amino acids and peptide backbone are not additive. 6. Oxidizable side chains do not react completely in the BCA assay at 378C. 7. The BCA assay is susceptible to interfering compounds that reduce Cu2 to Cu1.

(16) assessed the reactivity of various model compounds with BCA and Cu2. They also determined the redox potential for some of these compounds. Their ®ndings are summarized in Table 3. Clearly, the two- tier reaction scheme proposed by Smith et al. (1) is correct. However, oxidizable side chains do not react easily at either 37 or 608C. The following half-reactions probably lead to the reduction of 2 moles of Cu2.

2Cys?Cys À Cys 2e 2H 1

Tyr?semiquinone radical?L À dopa 2e H 2

TABLE 4 Determination of Reducing Compounds Using BCA

Compound Á" (MÀ1 cmÀ1)a Moles per Cu1b

Fructose 43,000 0.2 Glucose 6,120 0.8 Indole 3,621 2.2 Tryptophan 1,589 5.0 Tyrosine 1,172 6.8 Cysteine 1,357 5.8 Acetol 10,000 0.8 Ascorbic acid 3,030 0.26 2,4-Dinitrophenyl 16,400 0.5 hydrazine Hydroxylamine 3,230 2.5 aValues for the apparent extinction coef®cient are from Ref. 16 and 22. bMoles of each compound necessary to reduce 1 mole of Cu2 to Cu1. Value is calculated as De (MÀ1 cmÀ1) divided by 8000 (MÀ1 cmÀ1). The Bicinchoninic Acid Protein Assay 107

The majority of the half-reactions leading to the reduction of Cu2 are unknown. The Á" values for reactions with BCA are given in Table 4. Dividing Á" by 8000 (MÀ1 cmÀ1) yields the probable number of moles of 1 each compound needed to form 1 mole of Cu (BCA)2 complex. Clearly, acetol undergoes a one-electron reaction with Cu2. One mole of ascorbic acid apparently yields four electrons. 2,4-Dinitrophenylhydrazine contri- butes two electrons per mole. The reactions with protein side chains are dif®cult to describe in quantitative terms. Between 5 and 6.8 moles of tyrosine, cysteine, or tryptophan appear to be involved in the reduction of one Cu2.

3.2. Metal Ion±Catalyzed Oxidation of Proteins and Polypeptides Metal ion±catalyzed oxidation (MCO) reactions require a transition metal ion (Cu2,Zn2,Fe3,orNi2), hydrogen peroxide, and a reducing agent (cysteine, glutathione, or 2-mecaptoethanol, ascorbic acid). The reducing agent transform Cu2 to Cu1, which then reduces hydrogen peroxide to form the highly reactive hydroxyl radical OH. The literature concerning MCO of protein was reviewed by Levine et al. (17) and also Paci®ci and Davies (18). Possible initiation reactions for MCO involving ascorbic acid are shown next.

2Cu2 ascorbic acid?2Cu1 DHA* 3

1 À 2 H2O2 Cu ? OH OH Cu 4 Hydrogen peroxide can also be produced from a two-electron reduction of oxygen.

O2? O2?H2O2 5

. . Attack by O2 or OH can initiate a free radical reaction leading to covalent modi®cations of amino acid side chains. Depending on the reaction conditions, MCO can result in subtle changes in protein tertiary and quaternary structure, aggregation, or fragmentation into small polypeptides. Madurawe et al. (19) and also Bush and Lumpkin (20) showed that lactate dehydrogenase (LDH) is inactivated by MCO in the presence of added ascorbic acid, Cu2, and hydrogen peroxide. Inactivation of LDH by MCO also occurred without added ascorbic acid. Under such

* DHA=dehydro ascorbic acid. 108 Chapter 4 circumstances, the reducing power is probably derived from oxidizable protein side chains (tyrosine, tryptophan, cysteine). The MCO process is site speci®c. Protein structural modi®cations seem to occur close to the site for Cu2/Cu1 binding. The degradation of hydrogen peroxide via Fenton-type reactions leads to a protein-bound .OH that attacks adjacent groups. BCA inhibits MCO by forming a complex with Cu1. Other MCO inhibitors include catalase, thiourea, and ethylenediaminetetraacetate. These inhibitors remove hydrogen peroxide, quench free radicals, or chelate Cu2, respectively. Inhibitors did not wholly eliminate MCO when LDH was exposed to Cu2 alone in the absence of reducing compound. Superoxide dismutase, mannitol, isopropanol, and sodium formate were also ineffective MCO inhibitors under such circumstances. The failure of radical scavenges to eliminate MCO may be partly explained by the localized nature of these reactions. Protein-bound free radicals may be inaccessible to dissolved inhibitors. Lactate dehydrogenase modi®ed by MCO showed increased susceptibility to proteolysis due to changes in the enzyme quaternary and tertiary structure. The structural changes induced by MCO are subtle. Protease susceptibility was increased even when the residual LDH activity after MCO was > 95%. There was no fragmentation and cross-linking after LDH activity was reduced to < 50% by MCO. One view is that MCO alters the charge on histidine residues. The consequent changes in enzyme charge generate conformational changes including subunit dissociation. Due to the inhibitory effect of BCA on the progress of MCO reactions, we expect that proteins will not be degraded extensively during the BCA protein assay.

3.3. A Note on Protein-BCA Interactions with Copper

1 À6 1 Cu -protein binding (Kd *10 M) is signi®cantly weaker than Cu À11 1 binding to BCA (Kd 10 M). Consequently, Cu will be sequestered as 1 2 the Cu (BCA)2 complex. On the other hand, Cu will mainly complex with tartrate and protein. As BCA does not bind Cu2, the formation of a 1 2 1 Cu (BCA)2 complex will alter the Cu /Cu redox equilibrium in favor of Cu1.

Cu2 e?Cu1; E 0:153 6

In the presence of BCA, the apparent redox potential (E8*) for the Cu2/Cu1 redox couple can be estimated from Eq. (7) (21).

* E E 0:059 log 1=Kd 7 The Bicinchoninic Acid Protein Assay 109

1 From the Kd value given for Cu (BCA)2 the E8* calculated from Eq. (7) is approximately 0.802. The BCA increases the oxidizing power of the Cu2/ Cu1 system. By comparison, Cu2 is a relatively mild oxidizing agent (E8 0.153). This suggests that BCA is not a passive ligand for Cu2 but also plays an integral role in the protein detection mechanism.

4. CALIBRATION FEATURES

The linear dynamic range for the BCA protein assay extends to 40 mgof protein. The standard graph for 20±120 mg of BSA is curvilinear. The linear range is approximately the same for gelatin, avidin, immunoglobulin G, chymotrypsinogen, insulin, ribonuclease, and soybean trypsin inhibitor (1). The limit of linearity is probably an intrinsic feature of the BCA assay. À1 Depending on the protein, the maximum DA560 1±2 units 120 mg of protein. The calibration graph for the BCA protein assay is nonlinear just like the standard curve for the Lowry assay (Chapter 3). The two assays have roughly equal sensitivities. Protein-protein variations in color yield are also similar. The Lowry assay for inorganic reductants leads to straight- line graphs. Nonlinearity was ascribed to a slow reaction with proteins that allows time for the degradation of Mo6/W6 at high pH. With the BCA assay, the instability explanation for nonlinearty is less tenable. In contrast to Mo6/W6, BCA is stable in alkaline media. What is clear is that the reduction of Cu2 to Cu1 becomes less quantitative at high protein concentrations. Some possible causes for nonlinearity are listed in Table 5. A hyperbolic standard curve can be readily explained in terms of saturation phenomenon. Protein binding to a ®xed concentration of Cu2 is according to Eq. (8)

P Cu2?PCu2 8

2 2 Therefore, Kd PCu =PCu and total number of sites 2 2 2 2 Cu 0 Cu PCu . Therefore, the fraction of Cu bound is ÂÃ b PCu2 ÂÃÂÃ 9 2 2 2 Cu 0 Cu PCu where b is the concentration of the PCu2 complex. Substitution for [PCu2] 2 ( [P][Cu ]/Kd) and a brief rearrangement of Eq. (9) give the hyperbolic 110 Chapter 4

TABLE 5 Plausible Explanations for Nonlinear BCA Protein Assay and Their Rebuttal

Explanations 1. Incomplete reduction of Cu2 by proteins. 2. Cu1 is bound to the protein and unavailable to BCA, 3. Cu1 is lost as the insoluble hydroxide, 4. Cu2 concentration is limiting in the BCA assay.

Rebuttals for points 1±3 1. High temperature increases sensitivity or extent of reaction. There is no effect on linear range for analysis. 2. The binding af®nity of BCA for Cu1 is 10,000 times greater than the af®nity of proteins for Cu1. 3. Formation of insoluble copper precipitates is curtailed by tartrate or BCA.

function.

Cu2 P b 0 10 P Kd The absorbance change from the BCA reaction is proportional to the amount of bound copper ions;therefore,

A562 nm eb1 11 As with the Lowry assay (19), the calibration graph for the BSA-protein assay can be linearized using a range of transformations (Chapter 3).

5. INTERFERENCE COMPOUNDS

Smith et al. (1) evaluated 41 potential interferences for the BCA assay. These are either chelators or reducing agents. A further group of miscellaneous compounds disrupt the assay by virtue of their strong color. Glucose (100 mM) caused a large increase in the reagent blank. Sucrose (10%) had little effect on the BCA assay. EDTA, guanidine hydrochloride, sorbitol, and ammonium sulfate produced signi®cant color losses compared with control samples. A list of additional interfering compounds includes biogenic amines, pharmaceutical agents, and Benedict-positive compounds (Table 6). The Benedict test is a method for determining easily oxidizable compounds in biological samples. Upon heating with the test sample, Cu2 The Bicinchoninic Acid Protein Assay 111

TABLE 6 List of Potential Interferences for the BCA Protein Assaya

Interference Reference

Biogenic amines. Slocum and Duepree (23) Dopamine, norepinephrine, epinephrine, tyrosine, serotonin (5-HT), tryptophan Buffers (À) Kaushal and Barnes (24), Lleu and Ada, Ampso, Bes, Bicine, Bistris, Caps, Rebel (25) Epps, Hepes, Hepps, Mes, Mops, Pipes, Tes, Tricine Benedict-positive compounds Acetol, aminophenol, ascorbic acid, 2,3- Chen et al. (22,26) butanedione, glucose, glyoxal, 2,4- dinitrophenylhydrazine, pyruvic aldehyde Drugs Chlorpromazine, caffeine (À), carbachol Marshall and Williams (27) (À), chloramphenicol (À), codeine phosphate (À), lidocaine (À), penicillin G, paracetamol Lipids Phosphatidylcholine, phosphatidyletha- Kesller and Fenestil (28) nolamine, phosphatidylserine, phosphatidylinositol, cardiolipin, sphingomyelin Phenolsb Gallic acid, tannic acid, pyrogallic acid, Kamath and Pattabiraman (29) pyrocatechol a(À) With the exception of Tricine, buffers showed no response with BCA reagent. b Effects have been carefully documented in food systems (see Sec. 5.2).

is reduced, forming a brown precipitate of Cu1 (hydr)oxide. Chen et al. (22) used BCA to detect Cu1 formed in the Benedict procedure. Biogenic amines interfered with the BCA assay of proteins from brain or adrenal medulla (23). Catecholamines were highly reactive with the BCA reagent. In protein-free samples, there was a linear response for 1± 100 6 10À9 mole of biogenic amine. Very small quantities (1 6 10À9 mole) of biogenic amine affected protein results. Indeed BCA was suggested as a highly sensitive reagent for assaying biogenic amines. 112 Chapter 4

Zhang and Halling (30) showed that samples containing high concentrations of NaOH (1 M) gave unexpectedly low readings with the BCA assay. The interference was higher for the microassay format, where sample volume constitutes up to 50% of the total assay volume. The normal BCA response was restored by neutralizing samples. To analyze highly alkaline samples, the BCA reagent was adjusted to pH 10.27 in order to improve its buffer capacity.

6. SAMPLE PRETREATMENT, AVOIDING INTERFERENCE, ENSURING ACCURACY

The TCA-DOC precipitation method (31) was adapted for the BCA assay by Brown et al. (32). Protein pellets were prepared as described in Method 3 of Chapter 3 and resuspended in 50 mL of alkaline SDS solution (5% w/v SDS dissolved in 0.1 N NaOH). Then 1 mL of a standard BCA working solution was added, followed by sample incubation at 378C for 30 minutes. Interferences by glucose, DTT, 2ME, and ammonium sulfate were eliminated after the TCA-DOC procedure. Protein was recovered quantitatively from media containing various proprietary ampholytes or polybuffer. Shahabi and Dyers (33) proposed a two-point kinetic assay for dealing with interferences. Their approach exploits differences in the kinetics of the BCA reaction with proteins and interfering compounds. Many interferences (ascorbic acid, cysteine, uric acid ) react with BCA according to ®rst-order kinetics. By contrast, the reaction with proteins is usually zero order. This means that interfering compounds produce a ®xed amount of background color rapidly. In contrast, the absorbance change in the presence of proteins increases linearly with time. Using the rate of color formation as the index of protein concentration eliminated the effect of interferences. Solid-phase assays are another means for avoiding interferences. Gates (34) adsorbed protein samples (50 mL) on 1 6 2-cm strips of cationized nylon membrane (Zeta-Probe, Biorad Ltd.). The bound protein was rinsed free of interferences and then air dried. Each membrane was placed in a 1.2 6 10-cm test tube and 2 mL of BCA working solution was added. The reaction was incubated at 608C for 30 minutes and the concentration of nylon-immobilized protein determined from DA562 measurements. The sensitivity of the solid-phase protein assay was 78 (+5)±93 (+3.4)% of the sensitivity obtained with the conventional solution assay. The limits of detection (*10 mg protein) and the precision of analysis were the same as for the conventional BCA assay. Accurate results were obtained in the presence of glucose (20 mM), Tris-HCl buffer (200 mM), or DTT (0.1% w/v). Drying the protein solution facilitated binding to the nylon support. The Bicinchoninic Acid Protein Assay 113

7. AUTOMATED BCA PROTEIN ASSAYS

Automated BCA protein assays were developed using microwell plate readers or ¯ow injection analysis. Microwell plates are widely used for enzyme-linked immunoassay (EIA) and other colorimetric assays. The advantages of microwell plates assays include (a) a requirement for a small sample size (5±10 mL), (b) reduced reagent requirements (50±200 mL), (c) compatibility with multichannel pipettes and dispensers, (d) automated measurement using microwell plate spectrophotometers, and (e) the ability to download results to a computer for enhanced data processing. Flow injection analysis and continuous ¯ow analyzers allow rapid and more precise sample analysis compared with conventional batch analysis.

7.1. Microwell Plate Assays A microwell plate BCA assay was applied for the rapid analysis of protein fractions from sucrose gradient centrifugation by Redinbaugh and Turley (35). The assay involved a commercial BCA reagent kit from Pierce Chemical Company with BSA as the standard protein. Protein samples (10 mL) were placed in the 96-well microwell plate, to which BCA working reagent (200 mL) was added. The samples were incubated at 228C for about 14 hours (i.e., overnight) or at 608C for 2 hours. Absorbance readings were recorded at 570 nm. Lane et al. (36) also described a BCA microwell plate protein assay, as have Sorensen and Brodbeck (37). Typically, color formation took place at 378C for 30 minutes. Absorbance readings for 96 samples could be completed in 5 minutes. The performance of microwell plate-based BCA protein assays depends on a range of factors including sample size, volume of protein to BCA working reagent, incubation temperature, and time. With the standard assay (1:20 ratio of protein to BCA reagent), the linear dynamic range for analysis was between 1 and 12 mg of protein (compare with a linear range of 10±120 mg for the conventional assay). The calibration graph remains curvilinear (1). The proposal that microwell plate-based BCA assays are 10- fold more sensitive than the conventional assay is not supported by the available evidence when results are normalized for differences in the assay temperature or for protein±protein variations in the response. The maximum absorbance expected for a given amount of protein can be estimated for the BCA assay. Consider 10 mg of BSA in a 210-mL assay volume. First, six peptide bonds bind each Cu2 ion. Taking the molecular weights of BSA and amino acids as 66,000 and 110, then 100 moles of Cu2 ions will bind to 1 mole of BSA (see Chapter 2, Table 2). Consequently, 10 mg of protein (1.51 6 10À10 mole BSA) will bind with 15.1 6 10À9 mole of 114 Chapter 4

Cu2, which is equivalent to 72 6 10À6 MofCu2 in the 210-mL reaction volume. Assuming that the bound Cu2 is fully reduced to Cu1, the maximum absorbance change expected (A ecl) is 0.58. In practice, the maximum absorbance change obtained for 10 mg of BSA during a microwell plate assay was between 0.29 and 0.93 (Table 7). To account for the higher than expected absorbance change, note that the preceding calculation did not consider the role of oxidizable protein side chains. Groups such as tyrosine, tryptophan, and cysteine react directly with Cu2. It was also assumed that each sextet of peptide groups undergoes a one-electron redox reaction with bound Cu2. Table 4 shows that this assumption is probably unwarranted.

7.2. Flow Injection Analysis and Continuous Flow Analysis Using BCA Reagent The principles of ¯ow injection analysis (FIA) were described in the monograph by Ruzicka and Hansen (38) and also by Valcarcel and Luque de Castro (39). In FIA, a pump delivers reagents and analyte to a mixer (chamber) and then to a reaction coil. The products of the reaction then ¯ow to a detector and from there to a waste receptacle. Electrochemical and colorimetric detectors are popular. Davis and Radke (40) illustrated the use of BCA in a simple FIA system for proteins. They employed a peristaltic pump to impel BCA reagent (containing 0.5±2% BCA) through a ¯owline of Te¯on or polyethylene tubing [inside diameter (ID) 0.5 mm]. Protein samples were injected into the ¯ow stream via a septum using a (25- or 50-mL) Hamilton syringe. The mixture then passed through a 3-m length of Te¯on tubing incubated in a constant-temperature water bath. The color formed was monitored with a spectrophotometer ®tted with a ¯ow-through cuvette. Absorbance (peak height) measurements were recorded using a strip chart recorder. The BCA-FIA system's performance depends on variables such as sample volume (10±50 mL), length of the reaction coil, ¯ow rate (0.5 or 1 mL minÀ1), and reaction temperature (72±1008C). Note that the BCA assay is a kinetic method. The color-forming reactions do not go to completion. This is especially true for FIA when the sample residence time in the reaction coil is short. Precision can be ensured only by keeping the sample ¯ow rate (and hence residence time) constant. To improve the color yield (peak height) during FIA, sample ¯ow rate may be reduced, thereby increasing the sample residence time in the reaction coil. Another strategy is to use a longer reaction coil. Assay sensitivity can also be increased by using higher reaction temperatures and/or concentrations of BCA. h iicoii cdPoenAsy115 Assay Protein Acid Bicinchoninic The

a TABLE 7 Microwell Plate-Based BCA Protein Assays

Assay Linear range BSA/BCA volume Sensitivity ÁA/g % format (g) (L) (BSA) R Error

Redingbaugh and Turley (35) Microassay 0.2±2 10:200A 0.113 (Æ0.003) Ð Standard assay 1±10 10:200A 0.093 (Æ0.002) Ð Lane et al. (36) Microassay 0.5±2.5 10:50B 0.0547 0.9940 3.3 Standard assay 2±12 10:200C 0.0245 0.9967 3.5 aResults with bovine serum albumin as standard. Samples were incubated at (A) 228C overnight, (B) 608C for 30 minutes, or (C) 378C for 30 minutes. 116 Chapter 4

The FIA devised by Davis and Radke (40) employed a sample ¯ow rate of 0.5 mL per minute, a sample size of 10 mL and a reaction temperature of 808C. The reaction coil was 3 m in length. With such design features, the sample residence time was 4.7 minutes. The throughput achieved with this system was 60 samples per hour. The linear range for analysis was 1±10 mgof protein. The precision of analysis was surprisingly good (1%) in view of the crude system for sample injection. With BSA, ovalbumin, hemoglobin, beta- lactoglobulin, conalbumin, and myoglobin as standard proteins, the average sensitivity for analysis (DA560/mg) was 0.024 (+0.0013). Therefore, the sensitivity for FIA is comparable to that obtained for batch analysis (Table 7). At reaction temperatures above 808C, the ¯ow tubing became blocked by an unidenti®ed precipitate. A further example of protein FIA using BCA reagent was described by Wolfe and co-workers (41,42). Their setup was essentially as described by Davis and Radke (40) with the following modi®cations: (a) sample injection was via a Rheodyne2 sample valve or an autosampler device with a 20-mL sample loop, (b) samples were thoroughly degassed before use, (c) the carrier stream was phosphate buffer (0.1 M, pH 7.4) with 1% Triton X-100, and (d) protein analysis was performed at 808C with a ¯ow rate of 0.9 mL minÀ1. The linear range for analysis was 0±260 mgmLÀ1 (0±5.6 mg). The Technicon AutoAnalyzer was adapted for the BCA assay by Hawkes and Craig (43). The reagent, air, and sample ¯ow lines were assembled from 2.4-mm (ID) glass tubing. The sample stream was segmented with air bubbles before mixing with BCA reagent. It then passed to a reaction coil incubated at 558C at a ¯ow rate designed to produce a residence time of 4.2 minutes. Color formation (peak heights) was monitored at 570 nm. Calibration graphs for standard proteins were curvilinear and ®tted a polynomial equation. To save on the reagent cost, commercial BCA reagent was diluted 1:4 before use. The sensitivity of analysis decreased with decreasing concentrations of BCA reagent. Protein- protein variations in assay results were similar to those observed with the batch assay: gelatin < BSA < chymotrypsinogen < RNA < insulin. The within-day precision of analysis was 1.04±2.99% and day-to-day precision was 1.62±14%, depending on the nature and concentration of the protein analyzed.

8. APPLICATIONS OF THE BCA ASSAY TO FOOD PROTEIN ANALYSIS

There are only a small number of reports describing the use of the BCA assay for food protein analysis. No doubt, the numbers of applications will The Bicinchoninic Acid Protein Assay 117 increase in the near future.* Applications reported in the published literature are discussed next.

8.1. Solid-Phase Analysis of Cereal Proteins Chan and Wasserman (44) determined protein in corn meal ¯our. Commercial corn meal samples and/or zein (2±7 mg) were placed in microcentrifuge tubes. The BCA working reagent (1 mL) was added and the mixture was incubated with intermittent shaking at 378C for 30 minutes. Samples were cooled over an ice bath for 5 minutes and particulate material removed by centrifuging (13,000g;10 minutes). Thereafter, 0.2 mL of supernatant was diluted in 1 mL of BCA reagent A (see Method 1 for the reagent composition) and A562 readings were recorded. The BSA (50± 400 mg) was used as the standard protein while method calibration involved Kjeldahl analysis. Fig. 2 shows generally good agreement between Kjeldahl and BCA results. However, the former technique gave higher values for protein than the BCA assay. The discrepancy between Kjeldahl and BCA assays was

FIGURE 2 Correlation between Kjeldahl protein and BCA assay of corn meal protein and samples of zein. (Drawn from Ref. 44.)

* From personal experience, use of the commercially available reagent is underreported. 118 Chapter 4 ascribed to the presence of NPN. It is also likely that BSA is not a satisfactory standard for corn protein analysis.

8.2. Analysis of Forage Plant Leaf Proteins Some 23 forage plant (leaf) samples were analyzed using the BCA assay by Messman and Weiss (45). These included alfalfa (fresh, wilted, hay, silage, leaves), crown vetch (fresh, wilted, silage), spinach (fresh), perennial ryegrass (fresh), orchard grass (fresh, wilted), corn plant (silage), and peal millet (fresh). For sample pretreatment, leaves were lyophilized and ground using a 1-mm Wiley mill. The resulting powders were extracted with borate- phosphate buffer (0.1 M ionic strength) containing SDS (1% w/w). Sonicat- ing the suspension of leaf powder for up to 2 minutes facilitated protein extraction. In most cases, protein recovery was 85%. The protein extracts were analyzed with the BCA and Kjeldahl methods (N 6 6.25). The BCA method gave unreliable estimates of leaf proteins. There was poor agreement between BCA and Kjeldhal results. Leaf samples contain numerous interfering compounds (plant pigments, peptides, sulfhydryl compounds, and phenol derivatives) that can interfere with both the BCA and Kjeldahl methods. Attempts to circumvent interferences using the DOC-TCA procedure were not successful. The yield of leaf protein recovered by precipitation with DOC-TCA ranged from 40 to 80%. Protein recovery by cold-acetone precipitation was not signi®cantly higher.

8.3. Analysis of Seed Proteins in the Presence of Phenolic Compounds Salt-soluble proteins from soybean, tamarind, ragi, jack fruit, mango kernel, and sorghum were analyzed by Kamath and Pattabiraman (29). Whole meals were extracted with 0.3 M NaCl (buffered with 20 mM phosphate buffer, pH 6.9). Protein extracts were then analyzed using the BCA, Bradford, and Lowry assays. A range of pure proteins (BSA, casein, chymotrypsinogen, lysozyme, myoglobin, trypsin, zein) were also analyzed. The BCA, Bradford, and Lowry assays showed differences in protein- protein variations in color yield. Apparently, endogenous seed compounds affected the results. High responses were obtained with sorghum, mango kernel, and other samples known to have high concentrations of total phenol. The BCA reagent was more sensitive to phenolic substances than to protein. On a scale of 1.0 for BSA, the color yields from a range of phenolic compounds were pyrogallic acid (86), gallic acid (2.1), pyrocatechol (106.0), tannic acid (9.3), and phenol (0.8). The BCA response to protein and phenolic substances was additive. There was a linear response between A567 The Bicinchoninic Acid Protein Assay 119 values and the concentration of phenol. Most seed contained comparable amounts of protein and phenolic compounds. Therefore, the BCA response to these systems is likely to be due to the phenol. The BCA analysis of soybean protein was error free owing to the low concentrations of phenolic compounds in this seed.

8.4. Identi®cation of High-Lysine Cereals Using the BCA Assay

The A562 readings for ribonuclease, chymotrypsinogen, insulin, and BSA apparently showed a high degree of correlation with the number of lysine residues in each protein (R 0.99; P <.01). Therefore, the BCA assay response was determined for four high-lysine mutants of barley (Hordeum vulgare) and four normal-lysine variants. Whole meals were extracted with 4% SDS solution in 0.15 M Tris-HCl buffer (pH 7) at 608C for 20 minutes. The extracts were diluted three fold with distilled water before analysis using the BCA assay at 258C for 2 hours. There was a signi®cant correlation between BCA protein assay results and lysine content expressed per weight of barley meal (R 0.974, P <.001) (46). Available lysine was also determined using the TNBS (trinitrobenzenesulfonic acid) method of Kakade and Liener (47) (see Chapter 12). A signi®cant correlation was also observed between BCA results and sample protein content as determined by the Kjeldhal method or by UV absorbance. As a corollary, the sensitivity of the BCA assay for a range of synthetic polyamino acids increased in the order Asp & His < Glu < Arg Lys.

8.5. Determination of Plant Protein in the Presence of Reagents for Electrophoresis Extraction buffers for SDS-PAGE, two-dimensional (2D) electrophoresis, and isoelectric focusing of plant proteins are incompatible with many protein assays. The 2D electrophoresis samples may contain 9 M urea, 2,2ME, SDS, and ampholyte. Orr et al. (48) assayed such samples accurately using the BCA assay.

8.6. Analysis of Animal Carcass Total Protein Analysis of protein nutrient quality sometimes requires a quantitative determination of total carcass nitrogen (see Chapter 12). Brooks et al. (49) assayed whole-body nitrogen for male Sprague-Dawley rats using serveral methods including the BCA assay. Their approach should be applicable for 120 Chapter 4

TABLE 8 Effect of Extraction Solvent on the Apparent Carcass Protein Concentrationa

Assay method Extraction solvent [Protein] (g/100 g sample)

BCA Water 2.05 SDS-NaOH 12.03 GnHCl 10.43 Biuret Water 3.41 SDS-NaOH 14.70 GnHCl 12.11 Bradford Water 7.30 SDS-NaOH 102.00 GnHCl 6.10 aWhole-body protein concentration was 16.3% by quantitative amino acid analysis. other animal carcasses with roughly similar fat content. This work might be usefully compared with that of Toten and Whitaker (50) described in Chapter 2. Carcasses were passed through a Horbart meat grinder (model 4812) once. The skin (including hair) was cut into pieces with scissors and added to the meat. Homogenizing in a Waring blender and a Brinkman polytron homogenizer further reduced the meat particle size. Finally, 1 g of protein was solubilized by homogenizing with (a) water, (b) 5% SDS dissolved in 0.5 M sodium hydroxide, or (c) 6 M guanidine hydrochloride solution. Samples were analyzed by the BCA method, biuret assay, or Bradford assay. Protein extracts were also analyzed by three so-called absolute methods, i.e., quantitative amino acid analysis, protein hydrolysis followed by ninhydrin analysis, and Kjeldahl analysis. Quantitative amino acid analysis yields a value of 16.3 (+0.5) g protein/100 g sample (n 15 replicates). However, Kjeldahl results (N 6 6.25) were 34% higher than expected. Accurate results were obtained by using Fk 5.51. The ninhydrin assay results depended on the choice of amino acid standard. The BCA and biuret methods were not adversely affected by the presence of SDS or guanidine hydrochloride. The Bradford procedure was unusable in the presence of SDS (Chapters 6 and 7). These results are summarized in the following. Apparently, SDS-NaOH was the most ef®cient protein extraction solvent tested. Thereafter, the accuracy of results varied in the order biuret > BCA >> Bradford. The performance of the BCA assay may be better than indicated from results in Table 8. First, the BCA assay was not adversely affected by SDS-NaOH or GnHCl when serum albumin was The Bicinchoninic Acid Protein Assay 121 employed as the standard protein. Inadequate explanation was given for the ability of the BCA assay to detect only 75% of the carcass protein. Calibration graphs for the BCA are nonlinear. However, a hyperbolic function was not applied by the authors although such an equation was ®tted to results from the Bradford assay. There are good prospects that the BCA method can be adapted for animal protein analysis, perhaps with SDS- NaOH as the extraction solvent.

8.7. Analysis of Proteins from Freshwater Algae There were several sources of error during attempts to analyze proteins from freshwater algae. Meijer and Wijffels (51) noted that the ef®ciency of protein extraction from cells was variable. Attempts to facilitate extraction using chemical means led to interferences with the Bradford and BCA protein assays. Proteins could also undergo severe damage under harsh extraction conditions such as boiling with alkali. Such harsh treatments could lead to standard proteins being less representative of the sample. Boiling Chlorella cells with 1 M NaOH for 30 minutes led to a recovery of between 3% (Bradford assay) and 14% (BCA assay) of the crude protein. By comparison, extracting yeast cells under similar conditions produced protein recoveries between 76% (Bradford method) and 85% (BCA method). When BSA standard protein was exposed to boiling 1 M NaOH for 30 minutes, there was 32% (Bradford method) or 85% (BCA assay) of the response recorded for the untreated proteins. Apparently, chemical damage due to heating at high pH is only partly responsible for the poor assay results. Chlorella protein was ef®ciently extracted by sonicating 50-mL samples of fresh algae suspended in sodium phosphate buffer (25 mM) containing 1% SDS. After sonication for 0.5, 1, and 3 minutes, there was 36, 80, and 104% recovery of crude protein as determined by the BCA assay. To avoid foaming and a rise in sample temperature, the period of sonication was divided into six 30-second intervals. In the presence of SDS, the Bradford assay could not be used.

REFERENCES

1. PK Smith, RI Krohn, GT Hermanson, AK Mallia, FH Gartner, MD Provenzano, EK Fujimoto, NM Goeke, BJ Olson, DC Klenk. Measurement of protein using bicinchoninic acid. Anal Biochem 150:76±85, 1985. 2. J Hoste. On a new copper speci®c group. Anal Chim Acta 4:23±37, 1950. 3. D KerteÂ sz. State of copper in phenoloxidase (tyrosinase). Nature 180:506±507, 1957. 122 Chapter 4

4. G Felsenfeld. The determination of cuprous ion in copper proteins. Arch Biochem Biophys 87:247±251, 1960. 5. AL Gershuns, AA Verezubova, ZhA Tolstykh. Photocolorimetric determination of copper with 2,20-bicinchoninic acid. Izv Vyssh Uchebn Zaved Khim Khim Technol 4 (1):25±27, 1961. 6. S Nakano. 2,20-Biquinoline derivatives VI. Copper (I) chelates of the 4,40- substituted 2,20-biquinoline derivatives and the determination of copper by 2,20- bicinchoninic acid. Yakugaku Zasshi 82:486±491, 1962. 7. IS Musta®n, NS Frumina, VS Kovashova. Determination of copper in various samples with 2,20-bicinchoninic acid. Zavod Lab 29:782±785, 1963. 8. VN Tikhonov. Highly selective titration method for determining copper with 2,20-bicinchoninic acid. Zh Anal Khim 27:673±677, 1972. 9. NN Noskova. Photometric determination of copper in blood using 2,20- bicinchoninic acid. Mikroelem Sib 9:159±161, 1974. 10. F Buhl, Z Gregorowicz, B Piwowarska. Complex compound of 2,20- bicinchoninic acid with copper (I) ions. Pr Nauk Uniw Slask Katowicach 91:27±33, 1975. 11. F Capitan, JMR Navarro, LF Capitanvallvey, Solid-phase spectrophotometric microdetermination of copper. Anal Lett 24:201±1217, 1991. 12. AJ Brenner, ED Harris. A quantitative test for copper using bicinchoninic acid. Anal Biochem 226:80±84, 1995 [see Anal Biochem 230:360, 1995 for erratum]. 13. RF McFeeters. A manual method for reducing sugar determination with 2,20- bicinchoninic acid reagent. Anal Biochem 103:302±306, 1980. 14. EM Grindler. Automated determination of glucose via reductive formation of lavender Cu(I)-2,20-bicinchoninate chelate. Clin Chem 16:519, 536, 1970. 15. K Mopper, EM Grindler. A new noncorrosive dye reagent for automatic chromatography. Anal Biochem 56:440±442, 1973. 16. KJ Wiechelman, RD Braun, JD Fitzpatrick. Investigation of the bicinchoninic acid protein assay: identi®cation of the groups responsible for color formation. Anal Biochem 175:231±237, 1988. 17. RL Levine, D Garland, CN Oliver, A Amici, I Climent, A-G Lenz, B-W Ahn, S Shaltiel, ER Stadtman. Determination of carbonyl content in oxidatively modi®ed proteins. Methods Enzymol 186:464±478, 1990. 18. RE Paci®ci, KJA Davies. Protein degradation as an index of oxidative stress. Methods Enzymol 186:485±502, 1990. 19. RD Madurawe, Z Lin, PK Dryden, JA Lumpkin. Stability of lactate dehydrogenase in metal-catalyzed oxidation solutions containing chelated metals. Biotechnol Prog 13:179±184, 1997. 20. KD Bush, JA Lumpkin. Structural damage to lactate dehydrogenase during copper iminodiacetic acid metal af®nity chromatography. Biotechnol Prog 14:943±950, 1998. 21. DA Skoog, DM West. Fundamentals of Analytical Chemistry. 3rd ed. New York: Holt, Rinehart and Winston, 1976, p 305. 22. Q Chen, N Klemm, I Jeng. Quantitative Benedict test using bicinchoninic acid. Anal Biochem 182:54±57, 1989. The Bicinchoninic Acid Protein Assay 123

23. TL Slocum, JD Duepree. Interference of biogenic amines with the measurement of proteins using bicinchoninic acid. Anal Biochem 195:14±17, 1991. 24. V Kaushal, LD Barnes. Effect of zwitterionic buffers on measurement of small masses of protein with bicinchoninic acid. Anal Biochem 157:291±294, 1986. 25. PL Lleu, G Rebel. Interference of Good's buffers and other biological buffers with protein determination. Anal Biochem 192:215±218, 1991. 26. Q Chen, N Klemm, G Duncan, K Jeng. Sensitive Benedict test. Analyst 115:109±110, 1990. 27. T Marshall, KM Williams. Drug interference in the Bradford and 2,20- bicinchoninic acid protein assay. Anal Biochem 198:352±354, 1991. 28. RJ Kesller, DD Fanestil. Interference by lipids in the determination of protein using bicinchoninic acid. Anal Biochem 159:138±142, 1986. 29. P Kamath, N Pattabiraman. Phenols interfere in protein estimation by the bicinchoninic acid assay method. Biochem Arch 4:17±23, 1988. 30. J-X Zhang, PJ Halling. pH and buffering in the bicinchoninic acid (4,40- dicarboxy-2.20-biquinoline) protein assay. Anal Biochem 188:9±10, 1990. 31. A Bensadoun, D Weinstein. Assay of proteins in the presence of interfering materials. Anal Biochem 70:241±250, 1976. 32. RE Brown, KL Jarvis, KJ Hyland. Protein measurement using bicinchoninic acid: elimination of interfering substances. Anal Biochem 180:136±139, 1989. 33. ZK Shahabi, MS Dyers. Protein analysis with bicinchoninic acid. Ann Clin Lab Sci 18:235±239, 1988. 34. RE Gates. Elimination of interfering substances in the presence of detergent in the bicinchoninic acid protein assay. Anal Biochem 196:290±295, 1991. 35. MG Redinbaugh, RB Turley. Adaptation of the bicinchoninic acid assay for use with microtiter plates and sucrose gradient fractions. Anal Biochem 153:267±271, 1986. 36. RD Lane, D Federman, JL Flora, BL Beck. Computer-assisted determination of protein concentrations from dye-binding and bicinchoninic acid protein assays performed in microtiter plates. J Immunol Methods 92:261±270, 1986. 37. K Sorensen, U Brodbeck. A sensitive protein assay method using microtiter plates. Experientia 42:161±162, 1986. 38. J Ruzicka, EH Hansen. Flow Injection Analysis. 2nd ed. New York: Wiley, 1988. 39. M Valcarcel, MD Luque de Castro. Flow Injection Analysis: Principles and Applications. Chichester, England: Ellis Horwood, 1987. 40. LC Davis, GA Radke. Measurement of protein using ¯ow injection analysis with bicinchoninic acid. Anal Biochem 161:152±156, 1987. 41. AC Wolfe, DS Hage Automated determination of antibody oxidation using ¯ow injection analysis. Anal Biochem 219:26±31, 1994. 42. CAC Wolfe, MR Oates, DS Hage. Automated protein assay using ¯ow injection analysis. J Chem Educ 75:1025±1028, 1998. 43. WC Hawkes, KA Craig. Adaptation of the bicinchoninic acid protein assay to a continuous ¯ow analyzer. Lab Robot Autom 3:13±17, 1990. 124 Chapter 4

44. K-Y Chan, BP Wasserman. Rapid solid-phase determination of cereal proteins using bicinchoninic acid. Cereal Chem 70:27±28, 1993. 45. MA Messman, BP Weiss. Extraction of protein from forages and comparison of two methods to determine its concentration. J Agric Food Chem 41:1085± 1089, 1993. 46. H Ahokas, L Naskali. Bicinchoninic acid protein assay reveals high-lysine contents in barley seed protein extracts. J Cereal Sci 7:43±48, 1988. 47. ML Kakade, E Liener. Determination of available lysine in proteins. Anal Biochem 27:273±280, 1969. 48. A Orr, BA Wagnaer, CT Howard, OA Schwartz. Assay of plant proteins with bicinchoninic acid for high-resolution two-dimensional polyacrylamide gel electrophoresis. Plant Cell Rep 7:598±601, 1988. 49. SPJ Brooks, JJ Lampi, G Sarwar, HG Botting. A comparison of methods for determining total body protein. Anal Biochem 226:26±30, 1995. 50. J Torten, JR Whitaker. Evaluation of the biuret and dye-binding methods for protein determination in meats. J Food Sci 29:168±1174, 1964. 51. EA Meijer, RH Wijffels. Development of a fast, reproducible and effective method for the extraction and quanti®cation of protein of micro-algae. Biotechnol Tech 12:353±358, 1998. 5

The Udy Method

1. INTRODUCTION

Proteins react with certain organic dyes to produce insoluble complexes. The quantity of dye bound is proportional to (a) the dye-binding capacity (DBC)* and (b) the protein concentration. Farm-gate prices for milk (in Australia, Denmark, France, Netherlands, New Zealand, United States) is partly determined by its protein content. Dye-binding assays are widely used for milk protein determination. Amido Black 10B (C.I. 20470), Acid Orange 12 (C.I. 15970), and Orange-G (C.I. 16230) are the three main dyes used. Dye-binding assays are presented in Sections 1±3 of this chapter. Section 4 covers the chemistry of protein binding with azo dyes. Section 5 is a review of interference compounds and other sources of error. Section 6 covers applications of dye-binding assays in food protein analysis. Protein-dye interactions are of interest in relation to (a) dyeing of natural ®bers and textiles, (b) dyeing of tissue sections for microscopic observation, (c) the use of dyes to evaluate renal function, (d) application of dyes as pH indicators in protein-rich media, and (e) uptake of dyes by photographic materials. In 1925 Grollman (1) described phenol red

* DBC is the amount of dye bound by a gram of protein.

125 126 Chapter 5

(phenolsulfonphthalein) binding using the Freundlich adsorption isotherm:

1=n Db=cPKDf 1 where Db (moles) is the dye bound by cP (grams) of crude protein and Df is the equilibrium concentration of free dye. The empirical constants K and n were identical for serum albumin samples from healthy and diseased persons (2). Eq. (1) applied to solid adsorbents (charcoal, casein, granular gelatin) or soluble proteins (serum albumin). The Df was inversely related to serum albumin concentration. Robinson and Hogden (3) also showed that Df (measured as a decreases in A565) was inversely related to protein concentration (0±3 mg mLÀ1). Schmidt and co-workers (4±6) found that several proteins (gelatin, casein, ®brin, edestin) bound with acidic acid dyes (Biebrich Scarlet, Naphthylamine Brown,* Metanil Yellow, Tropaeolin O) at pH < 4.0 with maximum binding at pH 1.56±2.5. Contrary to earlier ®ndings, binding did not conform to the Fruendlich isotherm.{ The DBC was proportional to the concentration of basic amino acids in different proteins. Binding with basic dyes (Methylene Blue, Safranine-Y, and Induline Scarlet) depended on the number of acidic groups in the protein. The DBC was the same for soluble or granular gelatin. Fraenkel-Conrat and Cooper (7) con®rmed that Orange-G binds to basic amino acids. The basic dye Safranine-O bound with acidic amino acid side chains (carboxylic, phenolic, sulfhydryl groups). This study is recognizably the ®rst protein dye-binding assay.{ The number of basic (guanidyl, imidazole, amino) groups in several proteins is the same whether determined by titration or by Orange-G binding (Fig. 1).} The application of dye-binding assays to food proteins was developed and later commercialized by Udy (8,9). This subject has been reviewed by McGann (10), Lakin (11,12), Hartley (13), McGann (14), Lowe (15), and

* Naphthylamine Brown is also called Amido Black 10B (C.I. 20470), Amidoschwarz 10B, Naphthylamine Black 10BR, Aniline Blue Black, Naphthol Blue Black, Acid Black 1, or Pontacyl Blue Black SX.

{ From the Freundlich equation, log(Db/cP) log K (1/n)logDf. Thus, a plot of log(Db/cP) versus log Df should give a straight line graph with a slope (1/n) and an intercept log K. { About 5 mg of protein (egg albumin, beta-lactoglobulin, casein, ®brin, and zein) was placed in a 15-mL test tube, wetted with 1 mL of buffer followed by the addition 1±5 mL of dye solution (e.g. 0.1% w/v Orange-G in 0.2 M citrate±0.1 M phosphate buffer, pH 2.2). After shaking for 20±24 hours, the suspension was centrifuged to remove the insoluble protein-dye complex. The supernatant was diluted 100-fold with buffer and absorbance readings were taken. The free dye concentration was determined from a standard curve of absorbance readings plotted against known dye concentration. } Both collagen and casein appear to have a typical dye-binding characteristics. The Udy Method 127

McGann (16). The literature through to 1967 was discussed by Cole (17). Sherbon (18) reviewed dye-binding assays for milk proteins.

2. THE UDY METHOD

The Udy assay using Acid Orange 12 is described here (Method 1). Information in the public domain deals with milk and dairy products. Other applications developed by Udy Corporation during the 1960s were probably commercially sensitive and not published (Table 1). The Udy method is the basis of Udy Protein Systems2.

2.1. Protein Determination Using Acid Orange 12 Method 1 is equivalent to AOAC Method 967.12 for milk protein analysis (19). Instrument requirements include a spectrophotometer, short-path- length ¯ow cuvette, and automatic pipettes. For small-scale analysis, a degree of improvisation is possible. For large numbers of samples the required accessories include the Udy calorimeter, a 40-mL dye regent

FIGURE 1 Basic amino acid concentration in a range of proteins as measured by Orange G binding and by titration. Units of the Y-axis are 6 10À4 moles per gram protein. (Drawn using data from Ref. 7.) 128 Chapter 5

TABLE 1 A Range of Commodities Analyzed by the Udy Method

Alfafa Gaines burger Oats Soybean Barley Gram Oat groats Soybean hulls Beans Grass peas Peanut meal Soybean meals Bermuda grass Lentils Peas Sun¯ower meal Caseinate Linseed meal Pigeon peas Triticale Cheese (hard) Malted barley Rapeseed meal Urd beans Chickpeas Meat Rice Wheat Corn silage MilkÐ¯uid Rye Wheat germ Cottonseed meal MilkÐpowders Saf¯ower meal WheatÐgluten Cowpeas Mung beans Sesame meal WheyÐdelactose Fish meal Mustard meal Sorghum (milo) WheyÐfresh WheyÐpowdered

Source: Adapted from Udy Corporation advertising literature. dispenser, and a React-R-Shaker* for highly ef®cient mixing of powdered samples with dye solution.

Method 1 Acid Orange 12 dye binding (20±22).

Reagents{ 1. Acetic acid (glacial) 2. Acid Orange 12 3. Oxalic acid 4. Potassium dihydrogen phosphate

Procedure Puri®cation of Acid Orange 12. Dissolve 400 g of dye in 400 mL of boiling water. Add 400 mL of reagent-grade ethyl alcohol. Cool to room temperature and refrigerate at 0±58C overnight. Vacuum ®lter dye solids using a Buchner ¯ask-®ltration unit ®tted with a polypropylene ®lter. Wash with cold ethyl alcohol and dry the resulting solid in an oven at 1258C.

* Udy and React-R-Shaker are trademark terms for the Udy Corporation, 201 Rome Court, Fort Collins, CO 80524. Fax: 1-970-482-2067. Telephone: 1-970-482-2060. Internet address: http://www.udycorp.thomasregister.com { Other requirements include 2-oz polyethylene bottles or 125-mL conical ¯asks, automatic pipettes for dispersion of 40 mL of reagent, syringe pipettes (2±5 mL), sample mill, and ®ltration equipment or a low-speed centrifuge. The Udy Method 129

Phosphate buffer (0.05 M, pH 1.8±1.9). Dissolve potassium dihydrogen phosphate (3.4 g) and oxalic acid (2 g) in 100 mL of warm water. Add to 800 mL of water containing phosphoric acid (3.4 mL), acetic acid (60 mL), and propionic acid (1 mL) and dilute the mixture to 1L. Working Acid Orange 12 dye reagent (0.13% w/v). Dissolve 1.3 g of Acid Orange 12 in 100 mL of warm phosphate buffer. Allow to cool and dilute to 1 L with phosphate buffer. Reference dye solution (0.06% w/v). Dilute the working dye reagent with phosphate buffer. Prepare further dilutions and produce a

calibration curve of free dye concentration versus A480 readings. Performing a dye-binding assay. Place 1.5±2.4 mL of liquid sample (or 0.25±0.5 g of solid) in a 2-oz plastic polyethylene bottle. Add 40.44 g (40 mL) of working dye solution and shake vigorously for 30 seconds. Solid samples may be shaken for 5 minutes. Centrifuge at 3500 rpm for 30 minutes or ®lter to remove insoluble dye-protein complex. Dilute* the supernatant 100-fold with

phosphate buffer and record A480 readings versus an appropriate buffer blank. Determine the amount of free dye from the calibration graph. Calibration for protein determination. Determine dye binding for samples with known amounts of crude protein (%N 6 6.25).

Establish the regression equation relating Db versus crude protein.

Analyses of 73 whole milk or 34 spray-dried milk samples by Orange G binding led to a highly signi®cant correlation between crude protein (%N 6 6.38) and the amount of free dye (23);

V D À V D cP 100 1 2 f 2 kEm where k (226 g moleÀ1) is the dye equivalent weight,{ E (moles gÀ1) is the DBC expressed as equivalents of dye bound, and m is the weight of sample in grams. For a solid sample V1 & V2 (i.e., volume of sample & volume of

* Colorimetric measurements are possible without dilution when using a very short (0.3 mm) path-length ¯ow-through cuvette. { The equivalent weight (k) for an ionic species (g moleÀ1) is the molecular weight divided by the number of charges per molecule; k 226 (g moleÀ1) assuming that this dye has two positive charges. Note that the number of equivalents of dye bound is DBC/k. 130 Chapter 5 sample dye) and hence V D À D cP 100 1 f 3 kEm Rearranging Equation (2) leads to Eq. (4) or Eq. (5) for liquid or solid samples, respectively.

V1 cPEkm Df D À 4 V2 V2

cPEkm Df D À 5 V1

Therefore a graph of Df versus cP yields a straight line with a slope of À mkE. The parameter E was 0.792 (mEq gÀ1) for spray-dried milk and 0.805 (mEq gÀ1) for fresh milk. See Section 6 for further discussions of milk protein analysis.

2.2. Protein Determination Using Orange-G Fraenkel-Conrat and Cooper (7) employed Orange-G in their seminal study of 1944. Udy (8) also used Orange-G for protein determination and later changed to Acid Orange 12 in 1963 (22). The color change for Acid Orange 12 was apparently 100% greater than that obtained with Orange-G.* The former dye is also less hygroscopic and more easily puri®ed. These days, high-grade samples of Orange-G are readily available. The use of Orange-G is described further in Refs. 8,9,24,25, and 26.

2.3. Protein Determination Using Amido Black 10B Milk protein analysis using Amido Black 10B is important in both North America and Europe (27±30). Commercial instruments such as the ProMilk Mark II or ProMilk PMA (manufactured by Foss Food Technology Corp.) use Amido Black 10B. Several investigators reported dif®culties with Amido Black 10B staining of plant proteins;some Amido Black 10B samples may contain impurities with different af®nities for plant proteins (31). [Method 2 is adapted from Sherbon (32,33).]

* This view is incorrect (Section 3.2). Proteins bind equal amounts of Orange-G and Acid Orange 12. The molar extinction coef®cients for Orange-G and Acid Orange 12 are also similar. The Udy Method 131

Method 2 Protein analysis using Amido Black 10B dye binding* (32,33). Reagents 1. Amido Black 10B 2. Citric acid 3. Disodium hydrogen phosphate 4. Thymol blue (optional preservative) Procedure 0.05M Citrate±0.01 M phosphate buffer. Dissolve citric acid (52.6 g), sodium dihydrogen phosphate (3.3 g), and thymol blue (1 g) in 660 mL of water. Working dye solution (0.075% w/v). Dissolve Amido Black 10B (3 g) in 1 L of water by heating to 708C. Mix with citrate-phosphate buffer and add 3.33 kg of water. Reference dye solution. Dilute the working dye solution (1 volume) with 2.5 volumes of distilled water. Add 1 mL of sample to 20 mL of dye solution. Mix for 0.5±3 minutes. Filter to remove insoluble dye-protein complex. Dilute the super-

natant 100-fold with phosphate buffer. Record A620 readings. Determine the amount of free dye from a calibration graph of A620 plotted against reference dye solutions.{ For calibration analyze at least 10 samples of known protein concentration in duplicate. Methods 1 and 2 were readily scaled down by mixing 50±100 mLof sample with 1 mL of dye reagent in a polyethylene microcentrifuge tube. The protein-dye precipitate was then removed by microcentrifugation (13,000 rpm;5 minutes). Absorbance readings were recorded after diluting the dye- supernatant solution 100-fold. For strongly colored samples, the dry weight for the protein-dye complex was recorded. These micro-dye-binding assays led to signi®cant savings of reagent and improved convenience (34).

3. SOLID-PHASE DYE-BINDING ASSAYS

Protein is adsorbed on a ®lter support such as nitrocellulose, ®lter paper, or a glass ®ber ®lter. Sometimes, the adsorbed protein is ``®xed'' by treating

* Amido Black 10B dye binding is also called the ProMilk method. { Absorbance measurements can be taken without dilution if using the purpose-made 0.3 mm path length cuvette. Speci®c instructions are given for use with commercial instruments. 132 Chapter 5 with dilute TCA. Exposure to dye solution is followed by a destaining solution to remove excess dye. The dyed protein spot is then excised and placed in a test tube with elution solvent, which dissolves the protein-dye complex. Absorbance measurements are then recorded as before. The advantages of the solid-phase assay include (a) increased sensitivity and (b) improved resistance to interfering compounds. In the best cases, an LLD of 0.25 mg is attainable with a linear range extending to 200 mg of protein.

3.1. Nitrocellulose and Cellulose Acetate Membranes Kuno and Kihara (35) employed nitrocellulose membranes for solid-phase dye-binding assays.* The linear range for analysis was 10±50 mg protein. The ef®ciency of protein binding was >97%. Compared with the Lowry assay, there was increased resistance to interference from tyrosine or Tris. Heil and Zillig (36) analyzed protein (0.5±2.5 mg) using cellulose acetate membranes. Staining was with Amido Black 10B (0.24% w/w) dissolved in (10:45:45) acetic acid±methanol±water solvent. The same solvent was used for destaining. Protein spots were air dried, excised, and eluted with 0.5 mL of solvent (glacial acetic acid, formic acid, water, TCA). Schaffner and Weissmann (37) treated protein samples with 10% (w/w) TCA and recovered the resulting precipitate with a Millipore (HAWP 025CO) membrane ®lter. The rest of their methodology is essentially as already described. The linear dynamic range was 5±30 mg with a sensitivity À1 of 0.027 DA630 mg BSA. The LLD was 1.5 mg + 5% for a sample volume of 2 mL. The following compounds did not interfere: dextran (100 mg mLÀ1), polyethylene glycol (0.5 mg mLÀ1), glycogen (0.5 mg mLÀ1), RNA or DNA (30 mgmLÀ1), NaCl (>1 M), ammonium sulfate (2.5 M), magnesium chloride (>0.1 M), EDTA (>100 mM), 2-mecaptoethanol (>100 mM), SDS (1%), and sucrose (20%). The Schaffner-Weissmann method was modi®ed for protein analysis in the presence of 1000-fold excess lipid (38).BSA (20 mg) was accurately determined in the presence of 20 mg phospholipid (43% phosphatidylcholine, 30% phosphatidylethanolamine, 27% unidenti®ed lipids). The linear dynamic range (2±24 mg BSA) was unaffected by added phospholipid. Assay sensitivity was also unaffected by the presence of lipid. The A630 was 0.744 (+ 0.054) and 0.775 (+ 0.048) for 20 mg of BSA without and with 20 mg of

* Protein (5±50 mg) dissolved in 0.2 M magnesium chloride was ®ltered with a nitrocellulose membrane under suction. Each membrane was stained with 2 mL of Amido Black 10B (4 mgmLÀ1 in acetic acid±methanol±water (1:5:4) solvent). After destaining with 5 mL of acetic acid (1% w/w), membranes were eluted with 3.5 mL of 10 mM NaOH. The released dye was measured (A620). The Udy Method 133 lipid, respectively. There was compatibility with a great many buffer salts (200 mM): NaCl, KCl, Na phosphate, K phosphate, HEPES, Tris-HCl, MES, MOPS. However, protein-protein variations in the dye response were evident. The proceding technique was apparently more accurate than the modi®ed Lowry assay or biuret method. The speed for solid-phase analysis was increased by Nakamura and co-workers (39). They used micro®ltration apparatus* to apply multiple samples to nitrocellulose membranes. Stained protein spots were measured directly via a densitometer. The linear dynamic range for analysis was 1±10 mg. Sensitivity using Amido Black 10B staining was twofold higher than with Ponceau Red. For both dyes the order of increasing sensitivity was trypsin < lysozyme < cytochrome c < bovine serum albumin < human serum albumin < concanavalin A < histone II < human g-globulin. Assay results were unaffected at pH 3.6±9. There was no interference from a range of salts, sugars, amino acids, nucleotides, polyols or EDTA. Detergents (SDS, Triton X-100, Tweens) or high concentrations of denaturants reduced protein binding to the nitrocellulose membrane.

3.2. Whatman Paper and Glass Membrane Filters Protein samples (5±200 mg) were dried on to Whatman No. 42 paper followed by treatment with 7.5% (w/v) TCA (40). McKnight (41) spotted 100 mL of protein (0.5±5 mg) on glass ®ber ®lters (Whatman GF/C), dried the liquid using hot air, and then ®xed the protein with 20% (w/v) TCA. Esen (42) and Almand and Saleemuddin (43) dried protein solutions (5 mL; 1±4 mg mLÀ1) on Whatman No. 1 with no TCA ®xation step. The ®lter paper±bound protein was stained with Amido Black 10B or Coomassie Brilliant Blue G250 (see Chapter 7).

4. THE CHEMISTRY OF DYE-BINDING PROTEIN ASSAYS 4.1. Characteristics of Azo Dyes Azo dyes represent 60% of all known dye structures (44). The ®rst azo compounds were synthesized by Peter Gries in 1858 using building units designated A, D, E, M, and Z (Fig. 2). Mono-azo dyes were formed via electrophilic attack of a diazotized species (A) on a sulfonated amino-

* BIO-DOT apparatus;Bio-Rad Laboratories, Richmond, CA. Membrane-bound protein was stained using Amido Black 10B or Ponceau Red (0.1% w/v) dissolved in 7% acetic acid and destained with 7% (w/v) acetic acid. 134 Chapter 5

FIGURE 2 Building blocks for synthesis of azo dyes with a brief explanation of A, D, E, M, and Z notation. (Top) Diazotized (group A) compounds (left) or (right) a tetrazotized (group D) compound. (Bottom) Coupling agents with a capacity to react with one equivalent or two equivalents of a group A compound. M is an aromatic amine that can react with A. The product may be diazotized for a second round of coupling.

naphthol nucleus (E or Z). By altering the reaction pH, temperature, and concentration of reagents, diazo, triazo, or tetrakisazo dyes may be formed. The structure of azo dyes can written in shorthand: A?Z refers to a mono- azo dye produced by reacting a diazotized aromatic compound (A benzenediazonium chloride) with Z. Examples of A?Z dyes are Acid Orange 12, T-azo-R, and Acid Orange 1. A typical diazo dye is Amido Black 10B (1-amino-2p-nitrophenylazo-7-phenylazo-8-naphthol-3,6-disulfonic acid) with the formula A1?Z/A2. The A1 and A2 units are attached to a central (Z) unit, 1-amino-2-naphthol-3,6-disulfonic acid. These structures are shown in Figs 3±5. All azo dyes possess one or more azo (22N55N22) groups. The nitrogen-nitrogen double bond allows cis-trans isomerism. The naphthalene 2-hydroxyl group hydrogen bonds to the azo-group nitrogen, thereby stabilizing the trans isomer. The characteristics of some azo dyes are listed in Table 2. The absorptivity for Acid Orange 12 (w/v) is 26% higher than for Orange-G. The molar extinction coef®cients for the two dyes differ by only about 6%. The sulfonic acid group of azo dyes remains ionized at most The Udy Method 135

FIGURE 3 The structure of T-azo-R and Acid Red 1. accessible pH values. However, the exact acidity of benzenesulfonate or naphthylenesulfonate groups is uncertain (pKa * 0.7±1.5).

4.2. Protein Dye Binding

Azo dyes bind with the guanidino, imidazole, and the e-NH2 side chain of arginine, histidine, and lysine, respectively (4,5,7,45±48). Interactions with wool (composed of the protein keratin) occur via ionic bonding. Further 136 Chapter 5

FIGURE 4 The structure of Acid Orange 12 and Orange G. The Udy Method 137

FIGURE 5 The structure of Amido Black 10B. bonding is by van der Waals and hydrophobic interactions. These increase with the area of contact between the dye and protein (44). Nonionic interactions become more important at high dye/protein ratios. Formerly, the order of sensitivity for dye-binding assays was given as Amido Black 10B > Acid Orange 12 > Orange-G (21,22,26). Protein assays using Orange G were thought to be 100% less sensitive than assays using Acid Orange 12 because the two dyes bound 2 or 1 mole of arginine per mole of dye, respectively. However, more recent data (47) show that both Orange G and Acid Orange 12 form 1:1 mole complexes with protein basic amino acid residues. For Amido Black 10B the ratio of dye bound to basic amino acids is 1:0.5 (Table 3) (49,50). The small distance of separation between the sulfonate groups of Orange G may exclude binding to two sites.

TABLE 2 Characteristics of Some Acid Azo Dyes Used for Protein Assay

Dye AB 10B AO 12 OG

C.I. No. 20470 15790 16230 Molecular weight 616.50 350 452.38 Net charge 1 1 2 lmax 620 482 480 e (Abs mL/mg)a 81.5 (43,684) 59.0 (22,066) 46.9 (20,683)

AB10, Amido Black 10B;AO12, Acid Orange 12;OG, Orange G. aExtinction coef®cient, absorbance per unit concentration of dye bound (mg mLÀ1). Value in parentheses is molar extinction coef®cient. Source: Refs. 21 and 26. 138 Chapter 5

TABLE 3 Dye-Binding Capacity for a Range of Samples for Orange G (OG), Acid Orange 12 (AO 12), and Amido Black 10B (AB 10B)

DBC (mmoles gÀ1 cP)a

Sample OG AO12 AB 10B

BSA 1365.0 1775.7 743.5 HSA 1466.8 1780.0 834.4 HGG 1028.8 1354.3 582.8 k-Casein 692.5 857.1 274.4 Meat meal 630.5 905.7 339.3 Fish meal 708.0 920.0 336.0 Milk protein 800.9 1094.3 490.3 Soybean 984.5 1345.7 555.2 Average 959.6 1254.1 519.5 Ratiob 0.8 1.0 0.54 aDye-binding capacity (mmoles of dye bound per gram protein) from Ref. 50. bAverage dye: BAA ratio. BSA, bovine serum albumin;HSA, human serum albimin;HGG, human gamma globulin.

In summary, careful perusal of information in Table 3 shows that the sensitivities of dye-binding assays using Orange-G, Acid Orange 12, and Amido Black 10B are the same.

4.3. Soluble Protein Dye Complexes Protein-dye complexes can be studied by spectrophotometry or equilibrium dialysis (51). To avoid precipitate formation, the concentration of dye used (1±10 mM) is 350±1000 times below those used for the Udy assay. BSA binding with azosulfathiozole, Orange I, Orange II, methyl orange, and tetrazine yellow was investigated by Klotz et al. (52), Klotz (53), and Sheppard et al. (54). Pesavento and Profumo (55) examined T-azo-R binding to BSA. Other interesting reports describe protein binding to phenol red (3), bromophenol blue (56,57), thymol blue (58), and the reactive dye cibracron blue (59±61). Protein dye binding shifts the equilibrium between nonionized and

ionized dye forms. The extinction coef®cient for the bound dye (eb) increases while the wavelength for maximum absorption (lmax) shifts to lower values. The hyperchromic effect is explained by reference to the conjugation theory.

The lmax for dye molecules is determined by the energy required for p?p* electron transition. Protein binding alters the degree of conjugation The Udy Method 139 involving the p orbital and lowers the energy of the p* state. Lysine, arginine, or histidyl (auxochromic) groups donate electrons to the dye molecule, thereby increasing its conjugation extent. A further explanation centers on the transfer of free dye molecules from a polar low-viscosity solvent phase to a relatively nonpolar or restricted protein phase. Dye transfer to more nonpolar solvents and micelles leads to spectral changes resembling those observed during protein binding (62±64). Usually, a ®xed concentration of dye is exposed to increasing amounts of protein. Absorbance readings are recorded with a reference cuvette containing a dye solution of the same concentration as the sample cuvette (Table 4). Measuring the ``difference absorption'' (DA) is useful where a dye solution has a high background. The absorbance change for dye reagent depends on the total dye concentration (D), extinction coef®cient (ef), and the cuvette path length (1 cm) as described in Equation (6).

A1 ef D 6

Df, protein, and the protein-dye complex are in equilibrium. Db has its own extinction coef®cient (eb). The net absorption change is described by

TABLE 4 A Summary of Symbols Used in Describing Protein-Dye Binding

Symbol De®nition

A1 and A2 Absorbance for dye and dye protein DA ( A2 À A1) Difference absorbance a Fraction of dye bound D Total concentration of dye

Db Concentration of bound dye Df Concentration of free dye ef Extinction coef®cient for free dye eb Extinction coef®cient for bound dye De ( ef À eb) Extinction coef®cient difference for the free and bound dye eapp Apparent molar extinction change when a fraction of dye is bound

P, Pf Added, free concentration of protein Kd Conditional dissociation constant limax Wavelength for maximum absorbance liso Isobestic wavelength where De 0 n, ns Number of dye molecules bound per molecule protein; ns strong sites 140 Chapter 5

Equation (7).

A2 ef D À DbebDb 7

From Eqs (6) and (7) it can be seen that A2±A1 DA Db(eb À ef) and also

A À A D e À e 2 1 b b f or e a e À e e 8 D D app b f f and therefore

e À e a app f 9 eb À ef where a is the fraction of dye bound and eapp ( A2/D) is the apparent extinction coef®cient change when dye is bound.

The isobestic point (liso) is the wavelength at which bound and free dye molecules have equal absorptivity (ef eb). By running absorbance spectra with increasing dye or protein concentration, liso can be identi®ed as the wavelength at which there is no absorbance change (DA 0). The existence of an isobestic point is indication that the dye exists as two interconvertible forms (e.g., bound and free). No isobestic point will appear if ef = eb over the wavelength range examined. The corollary is that protein- dye binding will not generate an absorbance change if De 0.

4.4. Analysis of Protein Dye-Binding Reactions The protein dye-binding reaction is summarized by the following equation

Df nPf Db 10

Replacing Db with DA/De, we can de®ne the dissociation constant (Kd)as

D À DA=De nP À DA=De K 11 d DA=De

The concentration of dye species changes with pH and ionic strength.

Therefore, Kd is a conditional constant with a value that depends on the pH and ionic strength (52). Depending on the protein/dye ratio Eq. (11) takes on the two forms described in Cases 1 and 2. The Udy Method 141

Case 1, Low dye/protein ratio. With excess protein we have nP À (DA/De) & nP in Eq. (11). This approximation is also justi®ed if the number of binding sites is large;hence,

D À DA=DenP K d DA=De and

nDePD DA 12 Kd nP

For high protein concentrations (nP >10Kd) DA reaches a maximum (DAmax) where

DA De max 13 D

Equation (13) is the chief means by which De and also eb may be determined (65). First, invert all terms in Eq. (12). The resulting double-reciprocal relation [Eq. (14)]* allows the determination of DAmax by graphical means (see the following).

1 K 1 d 14 DA DeDnP DeD

Multiplying the former relation by DADeD gives Eq. (15). Other linearized forms result from multiplying Eq. (15) by 1/(DeKd) or 1/(DeKdD).

DAK DA DeD À d 15 nP

DA D DA À 16 nPDe Kd DeKd

DA n nDA À 17 PDDe Kd DeDKd

* The transformation is analogous to linearization of the Michaelis-Menten equation to give the Lineweaver-Burke double reciprocal plot, Eadie plot, Hanes plot, etc. 142 Chapter 5

Finally, Equation (17) may be restated as nD nGP G À 18 Kd Kd where G Db/P. To evaluate DAmax, De,andKd/n, proceed as follows: 1. Add varying concentrations of protein to a ®xed concentration of dye (D). 2. For each sample measure DA. 3. Using Equation (14), plot a graph of 1/DA (Y-axis) versus 1/P

(X-axis). From the X 0 intercept ®nd 1/DAmax ( 1/DeD). 4. Use the estimate for DAmax and ®nd De from Eq. (13). 5. From the slope and known values for De and D calculate Kd/n.It is not possible to determine Kd independently using Eqs (14)±(17). An alternative stratagem is to translate DA values to Db (e.g., Db DA/De) and Df ( D À DA/De). Thereafter, use Eq. (18) to evaluate all binding parameters. Note that Eqs (12)±(18) are valid only at high protein/dye ratios. Under these circumstances, only high-af®nity protein sites (strong sites) are occupied. Binding parameters therefore relate to strong sites. The number of strong binding sites (ns) is distinct from the total number of sites (n).

Case 2. High dye/protein ratio. With excess dye D À (DA/De) & D in Eq. (11);therefore, DenPD DA 19 Kd D Eq. (19) describes protein ligand binding when a small ®xed concentration of protein is exposed to varying concentrations of dye. As the concentration of dye increases (e.g., D > 10Kd), DA increases to a maximum value (DAmax) and Eq. (19) becomes*

DAmax nDeP 20 Using the De value determined before (see Case 1), ®nd the total number of binding sites (n) as follows: 1. Add varying concentrations of dye to a ®xed concentration of protein.

* A high dye concentration is de®ned in relation to Kd and not protein concentration. The Udy Method 143

2. For each mixture measure DA. 3. Plot a graph of 1/DA (Y-axis) versus 1/D (X-axis). The X 0

intercept yields 1/(nDeP) and the slope is Kd/nDeP. 4. Calculate the number of binding sites from Eq. (21).

n DAmax= DeP 21

Dividing the graph slope by the intercept gives Kd under conditions such that both strong and weak binding sites are ®lled. In summary, two different experimental designs and analyses for protein dye binding are possible. Case 1 employs a ®xed (low) concentration of dye and varying amounts of protein. The estimates of De, Kd, and n obtained are related to high-af®nity sites. Case 2 employs a ®xed (low) concentration of protein and varying (high) concentrations of dye. This study is useful mainly for determining the total number of binding sites.

Values of Kd are average parameters for both weak and strong dye-binding sites (57,58,65). The application of these relations to a study of BSA binding with T-azo-R is described next. A ®xed concentration of dye (D 10.8 mMin 5 6 10À3 M HCl solvent, pH 2.3) was titrated with increasing concentrations of BSA (see Case 1). Fig. 6 shows the pattern of binding of T-azo-R to BSA (55). At concentrations of dye above 100 mM, an insoluble protein-dye complex formed. Apparently T-azo-R binding to BSA could not be analyzed using the Scatchard plot (55). I have reanalyzed such data and others from Refs. 57, 58, and 65 using Eqs (14), (15), and (18) (Figs 6 and 7). Parameters for BSA binding with T-azo-R, bromophenol blue, and thymol blue are shown in Table 5. The original studies were not designed to measure the total number of (low- and high-af®nity) dye-binding sites. The proportion of basic amino acids (*110 per mole BSA) functioning as high-af®nity binding sites for T-azo-R did not exceed 50%. With bromophenol blue there was dye binding to a very small proportion of basic amino acids.

4.5. Solubility Relations for Protein Dye Complexes* The reaction between a charged dye molecule (DÀ) and protein (P ) produces a soluble, complex [PD]AQ that later forms an insoluble complex

* There is no strict adherence to the use of squared brackets to indicate concentration. Brackets are included only where their presence renders equations more readable. 144 Chapter 5

FIGURE 6 Analysis of protein-dye binding. A ®xed concentration T-azo-R dye (D 10.8 mM) was titrated with increasing concentrations of bovine serum albumin (P 0±6 mM). (Top graph) Difference absorbance

changes monitored at 510 nm (DA510) plotted versus total added protein concentration (P). (Lower graph) Determination of binding parameters using a double reciprocal plot of 1/P versus 1/DA. The Y-intercept

is 1/DAmax. Furthermore, DAmax/D De [see Eq. (13)]. The slope Kd/(nDeD). The Udy Method 145

FIGURE 7 Analysis of T-azo-R binding with bovine serum albumin. Same data as shown in Fig. 6. (Top graph) Determination of binding parameters in

accordance with Eq. (15). DA is plotted versus DA/P. The slope is Kd and intercept is DAmax. (Bottom graph) Determination of binding parameters according to Eq. (18). As P ? 0, then Db ? nP and Kd * D. Under such conditions it follows that G n (see Refs. 58 and 59). 146 Chapter 5

TABLE 5 Binding Parameters for Soluble Serum Albumin±Dye Complexes

Kd À1 À1 À1 Dye, equation n mM De (M cm ) n T-azo-R Eq. (14) 0.162 4709 Ð Eq. (15) 0.177 5024 Ð Eq. (18) 0.181 Ð 62.a Bromophenolblue Eq. (14) 3.11 87787 Eq. (15) 2.03 69250 Eq. (18) 2.23 6.0 Thymolblue Eq. (18) 69.0 1430 26 aThe intercept from Equation (18) n. See Refs. 57 and 58.

[PQ]S.

À À P D PDAQ; Kd P D =PDAQ 22

PDAQ PDS; K2 PDAQ=PDS 23 The net reaction is

À P D PSS 24 The overall process is comparable to isoelectric precipitation (4). Precipita- tion occurs when suf®cient numbers of dye molecules bind to neutralize all protein charges. In contrast, excess protein produces a ``colloidal protective effect'' that maintains the solubility of protein-dye complexes. From the

de®nition of solubility product (KS) we have

À KS KdK2 P D 25

The activity for [PQ]S is given a value of 1. To describe the effect of pH on protein-dye interactions, consider the ionization of dye-binding sites (pKa *12.5);

P H P ; Ka Pf H =P 26

The total protein concentration (P) [P ) [Pf] and after substituting for Pf,

Ka PÀP H =P 27 The Udy Method 147 and

P P 28 Ka=H 1 The concentration of P changes with the concentration of H in accordance with Equation (28) and consequently KS is given by Equation (29). PDÀ KS 29 Ka=H 1 To attain very low concentrations of soluble protein (in equilibrium with the protein-dye precipitate) requires a high dye concentrations at a low pH.

Strong protein-dye binding (small Kd) will also facilitate quantitative precipitation of protein from solution. At pH 4.84 the gelatin complex with À12 Amido Black 10B yields KS & 4 6 10 . Under higher acidity conditions the value KS was too low to determine (4). Refer to Skoog and West (66) for more information on KS-solubility relations.

5. INTERFERENCE COMPOUNDS AND THEIR AVOIDANCE

There was no interference from low-molecular-weight compounds, including amino acids and peptides. Lipids do not affect protein-dye binding. Ionic surfactants and benzoic acid derivatives (e.g., p-aminobenzoic acid) might interfere if present in high concentrations. Chaotropic agents such as urea are also likely to affect dye-binding results. Dye binding with nonprotein food components is possible. Calibration graphs for wheat ¯our protein had a nonzero intercept owing to dye binding with wheat b- glucan (67). Mass transfer or diffusion limitations may be important for solid or powdered food materials. Physical effects can be overcome by achieving high sample agitation, increasing the mixing time, and reducing sample particle size.

6. APPLICATIONS OF DYE-BINDING ASSAYS FOR FOOD PROTEIN ANALYSIS 6.1. Milk, Ice Cream, and Dairy Products Udy (9) analyzed whole milk and spray-dried milk samples by Orange-G binding. The milk samples and Kjeldahl protein values were supplied by Ashworth and co-workers at the Department of Dairy Science, Washington State University (Pullman, WA). Dye-binding studies at Ashworth's 148 Chapter 5 laboratory led to one of the ®rst Ph.D. dissertations in this area (68). Literature covering milk or dairy protein analysis using Orange-G and Acid Orange 12 is summarized in Table 6. Ashworth et al. (24) analyzed 354 milk samples from six breeds of cows. Milk powders were also analyzed. The average protein content for milks was 3.49 (+ 0.273)%. Some 95% of protein determinations were within + 0.67% of the crude protein content. NPN, proteose peptone, milk fat, and lactose caused little or no interference. Sample preservatives (hydrogen peroxide, formaldehyde, or mercuric hydrochloride) also did not affect the results. Adding mercuric chloride (1.35 mg %) to milk samples allowed room temperature storage before analysis. Antibiotics were not effective preservatives.

TABLE 6 Dye-Binding Assay of Milk and Dairy Protein

Dye, application Reference

Orange-G Milk (fresh, powder) Udy (9), Ashworth et al. (24), Dolby (25), Ashworth and Chaudry (26), and Conetta et al. (69), Park and King (70) Milk (fresh, evaporated, powdered), Ashworth (20) buttermilk, cheese, sherbet, cream, ice cream Ice cream, frozen desert Kroger et al. (71) Acid Orange 12 Milk (fresh, evaporated, powdered), Ashworth (20) buttermilk, cheese, sherbet, cream, ice cream Milk Sherbon (21) Chocolate milk drink, buttermilk Sherbon and Luke (72) Nonfat dry milk powder, ice cream, Sherbon and Luke (73) half-and-half Various dairy products Sherbon (74) Milk Conetta et al. (69), Lakin (75,76), Wilkinson and Richardson (77), Cheese Kristoffersen (78) Various dairy products, NFDM Sherbon and Fleming (79), Ice cream, ice milk, diet ice cream, Bruhn et al. (80) dietetic ice cream Other dyes Milk powder (delactosed)ÐRamazol Rawson and Mahoney (81,82) Blue R The Udy Method 149

The DBC for milk protein fractions was assessed by Ashworth and Chaudry (26). Milk protein fractions should have similar DBC values; otherwise, assays may be affected by variations in milk composition. Compared with whey proteins, caseins had a lower DBC (Table 7). Presumably the proportions of casein and whey protein remain fairly constant in different milk samples. The quantity of protein in several brands of milk drink (chocolate milk, two-ten, half-and-half, vitamin D milk, etc.) were determined by Ashworth (20). The Orange-G binding capacity for milk proteins was remarkably constant, notwithstanding processing into products such as ice cream and evaporated milk (Table 8). Fresh milk had a protein content of 3.5%, whereas evaporated milk contained 7% protein. Cheese manufacture had a signi®cant lowering effect on DBC, probably because of proteolysis. Notice that the values for the DBC are 50% lower than those reported in Table 3 for reasons discussed earlier.

6.2. Of®cial Approval of Dye-Binding Assays Collaborative studies led to dye-binding assays being granted AOAC approval for milk protein determination (21,72,83). The studies involved laboratories attached to the of®ces of the U.S. Federal Milk Market

TABLE 7 Apparent DBC for Different Milk Protein Fractionsa

Milk protein (fraction) AB 10B (mmoles gÀ1 cP) Orange G (mmoles gÀ1 cP)

Whole milk 566.6 389.4 Skim 561.7 396.0 NFDP 595.8 407.1 Whole casein 589.3 446.9 Paracasein 556.8 415.9 a-Casein 577.9 398.2 k-casein 592.5 420.4 b-casein 504.9 376.1 Whey protein 730.5 539.8 b-Lactoglobulin 763.0 557.5 Proteose peptone 584.4 278.8 Average 602.1 420.6 SD 76.1 76.2 aDBC was calculated from Ref. 26 using dye molecular weights from Table 2. Values are between 50 and 100% lower than expected from the basic amino acid content in milk proteins (see Table 3). 150 Chapter 5

TABLE 8 Dye-Binding Capacity and Protein Content for Different Milk Productsa

Product OG (mmoles gÀ1) AO12 (mmoles gÀ1) Protein (%)

Milk Fresh 393.5 1000.0 3.5 Evaporated 386.8 932.8 7.0 Powdered, nonfat 394.6 1000.0 36.0 Powdered, whole 366.9 963.6 26.0 Buttermilk 381.1 902.0 4.0 Cottage cheeses Creamed 394.8 966.4 15.0 Dry curd 394.8 966.4 18.0 Cheddar cheese Mild 306.8 801.1 24.0 Sharp 245.4 700.3 24.0 Cream 20±50% fat 394.8 1000.06 2.5 Ice cream Vanilla 394.8 1000.0 4.0 Chocolate 362.3 963.6 4.0 aA comparison with data from Table 3 shows that the values for OG are lower than expected. Source: DBC values calucalated from Ref. 21.

Administrators.* The extent of interlaboratory differences was considered practically insigni®cant. The agreement between dye-binding results and Kjeldahl results was excellent. Dye binding was granted of®cial ®rst action status for protein quantitation in fresh milk, dairy chocolate milk drink, cultured buttermilk, and half-and-half cream. A collaborative study to examine dye-binding assays for ice cream mix and nonfat dry milk powder is described by Sherbon and Luke (73). Five commercial samples of vanilla ice cream mix and 10 samples of nonfat dry milk were analyzed in powder form or after reconstitution. Seven laboratories performed dye-binding assays while three did Kjeldahl analysis. Tests using nonfat dry milk powder gave the same results as for the reconstituted material. The protein content for nonfat dry milk was 34.946% as determined by Kjeldahl analysis and 35.141% by dye binding. The average difference between Kjeldahl and dye-binding results was 0.195%.

* One lot of sterile, canned milk was analyzed by ®ve laboratories. Three laboratories carried out crude protein determination by the Kjeldahl method. A further 25 fresh milk samples were analyzed via 140 replicate determinations. The Udy Method 151

Ice cream mix had 3.852% protein or 3.968% protein by the Kjeldhal and dye-binding analysis, respectively. Consequently, dye binding received AOAC approval as a method for protein determination in ice cream and nonfat dry milk Analysis of ice cream mix and ice cream proteins using dye binding is further described by Kleyn (84) and by Bruhn (85). Further examples of these studies are listed in Table 9. Sherbon (32) compared the Pro-Milk and Udy methods for the analysis of milk proteins. The two techniques gave comparable results. Greater care was needed in calibrating the Pro-Milk instrument. A recommendation to grant of®cial status to the Pro-Milk method was deferred pending further work. A year later, a six-laboratory study of the Pro-Milk method led to AOAC approval (33). In addition to reports cited in Table 9 the Pro-Milk Analyzer is discussed in Refs. 86±91.

6.3. Wheat and Other Cereal Proteins Udy (8) found that wheat albumin, gluten, albumin gluten, and residue proteins bound constant amounts of Orange-G regardless of the variety of seed. These observations paved the way for a quantitative analysis of wheat

TABLE 9 Application of Amido Black 10B for Milk and Dairy Protein Analysisa

Amido Black 10B Reference

Cheese Kroger and Weaver (92) Condensed milk Lueck (93) Ice cream, frozen dessert Kroger et al. (71) Milk Dolby (25), Ashworth and Chaudry (26), Radcliffe (94), O'Connell (86), Conetta et al. (69), Sherbon (95), Kroger (96), Uzonyi (97), Patel et al. (98), Ng-Kwai-Hang and Hayes (99), van Reusel and Klijn (100) Whey protein, casein Roper and Dolby (101), McGann et al. (102), Renner and Ando (103), Reimerdes and Flegel (104) Milk (goat) Grappin et al. (105), Mabon and Brechany (106) Milk (mastitis) Waite and Smith (107) Skimmed milk powder Sanderson (108), O'Connell and McGann (109) aThe majority of investigators used the Pro-Milk Analyzer 152 Chapter 5 protein using dye binding (9). In all, 128 samples of whole wheat ¯our and 218 samples of re®ned wheat ¯our (from 58 known and 34 unknown wheat varieties) were examined for Orange-G binding capacity at pH 2.2. The samples were also analyzed for crude protein content by the Kjeldahl method. The correlation between amounts of dye bound and crude protein (%N 6 6.25) is described by Equation (30) (re®ned wheat ¯our) or Equation (31) (whole wheat ¯our).

cP 1:092Db À 4:62 30

cP 1:000Db À 5:53 31 The nonzero intercept indicates dye binding to nonprotein components, probably starch and/or wheat bran. Greenaway (110) at the U.S. Department of Agriculture (USDA;Beltsville, MD) reported a positive correlation between dye binding and Kjeldahl results for soft wheat (< 10% protein), hard winter wheat, hard spring wheat, and durum wheat; cP 0:8842X 1:7938 R 0:988 32 where X (% protein) is wheat protein content determined from dye binding and cP Kjeldahl protein (%N 6 6.25). Over 220 protein assays were performed. Methods were compared with respect to average protein content, correlation coef®cients, and standard error of estimates. The dye- binding assay gave reliable estimates for protein content for hard red spring wheat. For other classes of wheat, protein results were slightly lower than expected from Kjeldahl analysis. The mean difference between Kjeldahl and dye-binding tests was 0.5%. For wheat samples having less than 10% protein, dye-binding and Kjeldahl results differed by *1%. The reliability of the Udy method was compared with that of ®ve other techniques (Kjeldahl, alkaline distillation, biuret, Dumas, and infrared re¯ectance) for wheat protein determination by Pomeranz and More (111). Reliability encompasses speci®city, accuracy, precision, sensitivity, and the LLD (Chapter 1). Dye binding was the least precise method tested. Interestingly, no strong case is made for preferring any one method. Forty- ®ve varieties of rice produced in the 1966 season in the Philippines by the International Rice Research Institute (IRRI) were analyzed for protein (112). A typical set of results highlighting the performance of the dye- binding assay for rice protein analysis is given in Table 10. Barley and malt proteins were analyzed by Pomeranz et al. (113) using ®ve methods including the Udy method. About 120 samples each of barley and malt from all over the United States were analyzed. No details of the dye-binding assay were given other than a reference to Udy's 1971 paper The Udy Method 153

TABLE 10 Determination of Rice Protein Using Acid Orange 12 Dye-Binding Assay

Parameter Milled rice Brown rice

a Regression line Y 14.67À13.60A485 Y 14.78À14.12A485 Protein (%) 5.55±11.65 6.00±11.95

SY.X (%) + 0.48 + 0.26 R À 0.961 À 0.984

a Y crude protein content (%). Protein (%) is range for 45 samples. SY.X (%) standard deviation from the regression line. Data are from experiments using a laboratory shaker (112).

(22). The correlation coef®cient for dye-binding and Kjeldhal results was 0.974 (barley) or 0.984 (malt). The average mean squared error for analysis (with the Kjeldahl method as reference) was 0.897%. Using commercial apparatus, 200 protein determinations were completed daily. Baker and Hunt (114) evaluated the Pro-meter instrument (Foss America Inc) for dye-binding analysis of wheat protein. About 107 wheat samples were ground to pass 20-mesh screen and then analyzed according the instrument manufacturer's instructions. The graph of instrument response versus protein content was curvilinear for 50 samples of red wheat (hard red spring, hard red winter, durum wheat). By contrast, a linear calibration graph was obtained for 57 white wheat samples. The Pro-meter instrument was judged satisfactory despite some mechanical dif®culties. The ®lter system was periodically clogged, necessitating dismantling and cleaning of the measuring unit.

6.4. Legumes and Other Seed Proteins The Udy method is applicable to range of legumes including, chickpeas, cowpeas, gram, mungbeans, peas, and soybean (Table 1). Pomeranz (115) analyzed 24 soy ¯our samples using Orange-G and commercial apparatus from the Udy Corporation. The results were compared with the biuret and Kjeldahl methods. A highly signi®cant correlation was found between protein content assessed by dye binding (X, %) and crude protein (N 6 6.25%). For samples containing up to 80% protein, the regression equation was cP 1:003X À 4:559 R 0:989 33

The standard error of analysis was 1.8213%. Flour particle size had negligible effects on protein results. Mild heat did not affect dye-binding 154 Chapter 5 results although other studies show that severe heating reduces the DBC of soy proteins (116±120). Romo et al. (121) assessed seed protein extractability using the Udy method. The following DA482 changes were noted for the different seed protein solutions (10 mg mLÀ1): 1.363 (®eld bean), 1.197 (cowpea), 2.976 (rapeseed), 2.2454 (sesame seed), and 1.203 (cotton seed). Clearly, the assay sensitivity is different for different seed proteins. Medina et al. (122) proposed that a single calibration graph might be used for cereal, legume, and oilseed protein analysis. A composite graph would save time. Sesame ¯our, rapeseed meal, and rapeseed ¯our were analyzed by the standard Udy (shaker mixing) method. Fig. 8 shows a composite calibration graph for Acid Orange 12 binding to cereal, legume, and oilseed proteins. The least- squares equation for the composite graph is*

cP 0:2152Db 4:7333 R 0:981 34

FIGURE 8 A composite calibration graph relating dye binding (X-axis) and crude protein content for *28 samples of legumes, cereals, and oilseeds. (Drawn from Table IV in Ref. 123.)

* Actually, the regression equation reported in the literature was Y 0.245X 2.532 (R 0.995). In contrast, Fig. 1 was drawn using only 50% of the experimental data. The Udy Method 155

Apparently the average DBC for Acid Orange 12 is 465 (+ 17.11) mg gÀ1 (cP) or 1328.6 mmole gÀ1. The Udy method was further optimized for seed protein determination (122). Vacuum drying (558C) or atmospheric drying (1008C) had no effect on dye binding. Improving the degree of mixing, extending the shaking time from 30 to 150 minutes, and/or reducing the particle size to 40 mesh increased the perceived sample protein content. Table 11 summarizes results for sesame ¯our, rapeseed ¯our, and rapeseed meal. For all cases, a good correlation was obtained between dye binding and Kjeldahl results. Rapeseed protein was also determined by Goh and Clandinin (123). Twelve commercial meals and two laboratory samples were analyzed using the Udy method with Orange G dye reagent. The investigators also examined Acid Orange 12 as a dye reagent. Rapeseed meal had 30.1±44.8% (w/w) protein. For Orange-G the least-squares line relating dye binding and Kjeldahl results was

cP 0:49Db 3:91 R 0:93 35

For Acid Orange 12 dyes the corresponding equation is

cP 0:28Db 0:36 R 0:98 36

From such data it may be shown that the apparent DBC for Orange-G is 204 mg (dye) gÀ1 (cP) or 490 mmoles (dye) gÀ1 (cP). With Acid Orange 12 the DBC is 357 mg gÀ1 (cP) or 580 mmoles (dye) gÀ1 (cP). The calibration data were not affected by ¯our particle size (40 or 60 mesh). However, DBC was higher for a protein/dye ratio of 2:1 as compared with a ratio of 4:1. Values for the DBC were proportional to the net concentration of arginine, histidine, and lysine. The standard deviation for analysis was 1.3% (Orange G) or 0.80% (Acid Orange). From the higher DBC (per weight), precision,

TABLE 11 Protein Content in Sesame and Rapeseed Products Determined from Acid Orange 12 Dye Binding and Kjeldahl Analysis

Sample Protein (% w/w)a

Dye binding Kjeldahl

Sesame ¯our 59.4 (+ 0.524) 58.9 (+ 1.093) Rapeseed ¯our 59.4 (+ 1.743) 60 (+ 2.91) Rapeseed meal 36.1 (+ 0.595) 36 (+ 0.338) aValues are mean (+ SD). Source: Summarized from Ref. 122. 156 Chapter 5 and sensitivity of analysis, Goh and Clandinin (123) concluded that Acid Orange 12 was a more suitable dye reagent for rapeseed protein determination.

6.5. Fish, Meat, and Egg Products A. Animal Feedstuffs Bunyan (124) determined the protein content of feedstuffs containing animal protein. The procedure using Orange-G was essentially as given in Method 2. The dye-binding response was dependent on sample particle size. The extent of dye binding also increased with mixing time. With care, values of Db could be correlated with the protein content (Table 12). However, animal feeds were found to be highly heterogeneous owing to their different manufacturing and thermal histories. A number of meat meal samples had an unusually high content of gelatin. In one case, meat meal was positively identi®ed as feather meal (techniques for establishing protein authenticity are described in Chapters 9±11). It was concluded that dye-binding assays were not suited for animal feedstuffs. Differences in processing history, protein quality, and possible adulteration led to large variations in results.

TABLE 12 Analysis of Protein Content of Animal Feeds Using Orange G Dye Binding

DBC Sample (na) Regression line (mmole gÀ1 cP) % Error (CV)

Meat meal cP 0.278Db 30 796±925 6.4 (21)

Whale meat cP 0.216Db 30 842±770 2.0 meal (12)

Fish meal (8) cP 0.325Db 24 675 2.3 Soy bean (8) cP 0.217Db 28 1020 2.0 or Groundnut meals (6)

Miscellaneous cP 0.414Db 12 Ð foodsb an number of different feed samples. bIncluding casein, dried blood protein, egg, brewer's yeast, roller dried milk, and grass meal. Source: Summarized from Ref. 124. The Udy Method 157

TABLE 13 Protein Determination in Raw Meat Using Orange G Dye Binding

Sample Regression linea R DBC (mg gÀ1 cP)

Beef cP 0.301Db 8.18 0.90 DBC 209.2 À1.135cP Chicken breast cP 0.602Db À 2.50 0.94 DBC 265.5 À 3.721cP Pork loin cP 0.367Db 5.45 0.80 DBC 271.2 À 4.040cP Cod ®llet cP 0.632Db 3.00 0.95 DBC 246.9 À 3.397cP a Symbols cP and Db are as de®ned previously. Source: Summarized from Ref. 125.

B. Meat Proteins Raw beef, chicken, pork (loin), and cod ®llets were analyzed using Orange- G or Amido Black 10B* dye binding by Torten and Whitaker (125). Their procedure was described in Chapter 4. A signi®cant correlation was observed between crude protein values (Kjeldhal-N 6 6.25) and Db (Table 13). The DBC for raw meat proteins decreased linearly with increasing sample protein (see last column of Table 13). The amount of dye bound depended on the dye/protein ratio. In general, inadequate amounts of Orange-G were used in many early studies. Dye limitations and inadvertent changes in protein/dye ratio for different assays reduced the reliability of dye-binding assays. The regression equation (cP 0.301Db 8.18) for beef applies over a restricted range of protein content. The effect of a changing DBC is shown in the simulations reported in Fig. 9. One set of results are computed on the basis that the DBC is ®xed. Where DBC varies the dye/ protein ratio the simulated calibration graphs were nonlinear (Fig. 9). The curves are remarkably like actual calibration curves reported for ground pork and chicken (125). These samples showed a high dependence of DBC on protein content and large deviations from linearity. A linear equation did ®t the data but only over a highly restricted range of protein content. Ground chicken, pork loin, or cod ®llet having greater than 50% crude protein content should probably not be analyzed by Orange-G dye binding. It was on account of the dependence of DBC on protein content that Amido Black 10B was judged unsuitable for meat protein analysis.

* As Amido Black 10B was found to be unsuitable for raw meat analysis, the following discussion focuses on results obtained with Orange-G. 158 Chapter 5

FIGURE 9 The effect of a changing dye-binding capacity on calibration graphs for protein analysis in raw meat samples using Orange G binding. Shaded squares show normal response according to regression equations in Table 13. The open circles show simulated graphs for the assay response when DBC changes with sample protein content. The Udy Method 159

TABLE 14 Acid Orange 12 DBC of Egg and Meat Products

Meat product DBC (mg gÀ1 cP)a

Egg (whole) 410±440 Egg albumin (egg white) 390±410 Chicken meat 460±480 Chicken liver 360±390 Beef or pork (ground) 430±440 Beef liver 420±440 Proteose peptone 90±145 Gelatin 310±350 aRanges of values are given for analysis performed in the presence of excess of dye concentration of 0.4±0.6 mg mLÀ1. Source: Ref. 126.

C. Egg, Chicken, and Meat Protein Egg, chicken, and meat products were analyzed by Ashworth (126) (Table 14). As he was one of the ®rst investigators to apply dye-binding assays to foods, his approach merits attention. Reliable results were obtained provided that the free Acid Orange 12 concentration (after shaking with the protein sample) was kept within the range of 0.4±0.6 mg mLÀ1. To achieve this, the initial the dye/protein ratio was kept within a range of 0.64±0.92. Pork had the same DBC as beef, which was lower than the value of chicken. The DBC for mixtures of meat could be deduced from values for individual components. Dye binding was not affected by the presence of fat or by normal cooking (1608C, 40 minutes). It was concluded that dye binding is useful method for composition control in ground meats, eggs, and prepared mixes.

D. Sausage Protein Seperich and Price (127) determined protein in model sausage emulsions and muscle components (myo®brillar protein, sarcoplasmic protein, and stroma) from which they were produced. The approach was modi®ed from Ref. 128.* These studies con®rmed that protein dye binding was not affected by sausage emulsion fat content from 20 to 40%. The DBC was a function of

* Sausage emulsion samples (3.5 g) were homogenized with 51 mL of citrate (0.2 M)±phosphate (0.1 M) buffer (pH 5.5). Ten milliliters of the resulting homogenate was retained for Kjeldahl analysis. The remainder was shaken with 80 mL of Acid Orange 12 (0.56±3.64 mM;0.2± 1.27 mg mLÀ1) in a 250-mL centrifuge tube for 30 minutes and then centrifuged (5.680g;5 minutes). 160 Chapter 5 dye/protein ratio. At the highest dye concentration examined the DBC was of the order of 400 mg gÀ1 (cP), in line with values reported by other investigators. However, DBC decreased to about 33±34 mg gÀ1 (cP) at a dye concentration of 0.2 mg mLÀ1.

6.6. Mushrooms Nine strains of Agaricus bisporus (Lange) Imbach were analyzed by Weaver et al. (128) using dye binding, Kjeldahl, and quantitative amino acid analysis.*. The average protein content was 29.4 (+ 6.2)% by Kjeldahl analysis, 22.4 (+ 2.4)% by dye binding, and 28% (+ 3.4)% by amino acid analysis. Per wet weight basis, Agaricus had 2.6±2.8% protein. Quantitative amino acid analysis was more correlated with dye-binding analysis (R 0.74) than Kjeldahl analysis (R 0.4). Mushrooms are thought to contain high amounts of NPN, which could lead to errors in Kjeldahl analysis. Braaksma and Schaap (129) reported the protein content for Agaricus as 0.5% fresh weight or 7% per dry weight basis (Chapter 7).

REFERENCES

1. A Grollman. The combination of phenol red and proteins. J Biol Chem 64:141±160, 1925. 2. WW Smith, HW Smith. Protein binding of phenol red, Diodrast, and other substances in plasma. J Biol Chem 124:107±113, 1938. 3. HW Robinson, CG Hogden. The in¯uence of serum protein on the spectrophotometric absorption curve of phenol red in a phosphate buffer mixture. J Biol Chem 137:239±254, 1941. 4. LM Chapman, DM Greenberg, CLA Schmidt. Studies of the combination between certain acid dyes and proteins. J Biol Chem 72:707±729, 1927. 5. LMC Rawlings, CLA Schmidt. Studies on the combination of certain basic dyes and proteins. J Biol Chem 82:709±716, 1929. 6. LMC Rawlings, CLA Schmidt. The mode of combination between certain dyes and gelatin granules. J Biol Chem 83:271±284, 1930. 7. H Fraenkel-Conrat, M Cooper. The use of dyes for the determination of acid and basic groups in proteins. J Biol Chem 154:239±340, 1944. 8. D Udy. Dye-binding capacities of wheat ¯our protein fractions. Cereal Chem 31:389±395, 1954.

* Mushrooms were diced, freeze dried, and then oven dried to a constant weight. A 100-mg portion of mushroom powder was mixed with 20 mL of dye solution. BSA was used as the standard protein. Amido Black 10B was used in conjunction with the Pro-Milk Mk II instrument. The Udy Method 161

9. D Udy. Estimation of protein in wheat and ¯our by ion-binding. Cereal Chem 33:190±197, 1956. 10. TCA McGann. Dye-binding procedures could lead to better utilization of world protein resources. Farm Food Res 4(6):138±139, 1973. 11. AL Lakin. The estimation of protein and the evaluation of protein quality by dye-binding procedures. IFST Proc 6(2):80±83, 1973. 12. AL Lakin. The estimation of proteins by dye binding: principles and experimental parameters. J Sci Food Agric 26:549±550, 1975. 13. AW Hartley. Estimation of proteins by dye binding: applications to cereals and ¯ours. J Sci Food Agric 26:550±551, 1975. 14. TCA McGann. The estimation of proteins by dye binding: applications to dairy products. J Sci Food Agric 26:551±552, 1975. 15. RA Lowe. The estimation of proteins by dye binding. Applications to animal feed. J Sci Food Agric 26:552±553, 1975. 16. TCA McGann. Automated physico-chemical methods for the analysis of milkÐa review of major advances (1960±1978). Ir J Food Sci Technol 2:141± 155, 1978. 17. ER Cole. Alternative methods to the Kjeldahl estimation of protein nitrogen. Rev Pure Appl Chem 19:109±130, 1969. 18. JW Sherbon. Recent developments in determining protein content of dairy products by dye binding. J Dairy Sci 61:1274±1278, 1978. 19. AOAC International. AOAC Of®cial Method 967.12. Protein in milk. Dye binding method I. In AOAC International, ed. Of®cial Methods of Analysis of AOAC International. 16th ed. Vol II. Arlington, VA: Association of Of®cial Analytical Chemists, 1995, Chapter 33, p 14. 20. US Ashworth. Determination of protein in dairy products by dye binding. J Dairy Sci 49:133±137, 1966. 21. JW Sherbon. Rapid determination of protein in milk by dye binding. J Assoc Of®;Anal Chem 50:542±546, 1967. 22. DC Udy. Improved methods for estimating protein. J Am Oil Chem Soc 48:29A±33A, 1971. 23. DC Udy. A rapid method for estimating total protein in milk. Nature 178:314±315, 1956. 24. US Ashworth, R Seals, RE Erb. An improved procedure for the determination of milk proteins by dye binding. J Dairy Sci 43:614±623, 1960. 25. RM Dolby. Dye-binding methods for estimation of proteins in milk. J Dairy Res 28:43±55, 1961. 26. US Ashworth, MA Chaudry. Dye-binding capacity of milk proteins for Amido Black 10B and Orange-G. J Dairy Sci 45:952±957, 1962. 27. International Dairy Federation. MilkÐdetermination of protein content (routine dye-binding method using Amido Black). International IDF Standard 98:1980 (Provisional), 1980, 4 pp. 28. International Organization for Standardization. MilkÐdetermination of protein contentÐAmido Black dye-binding method (routine method). International Standard;ISO 5542±1984,1984, 5 pp. 162 Chapter 5

29. International Dairy Federation Milk. Determination of protein contentÐ Amido Black dye-binding method (routine method). International IDF Standard;98A:1985, 1985, 4 pp. 30. AOAC International AOAC Of®cial Method 975.17, Protein in milk. Dye binding method II. In AOAC International, ed. Of®cial Methods of Analysis of AOAC International. 16th ed. Vol II. Arlington, VA: Association of Of®cial Analytical Chemists, 1985, Chapter 33, p 15. 31. CM Wilson. Polyacrylamide gel electrophoresis of proteins: impurities in Amido Black used for staining. Anal Biochem 53:538±544, 1973. 32. JW Sherbon. Pro-Milk method for the determination of protein in milk by dye binding. J Assoc Of® Anal Chem 57:1338±1341, 1974. 33. JW Sherbon. Collaborative study of the Pro-Milk method for the determination of protein in milk. J Assoc Of® Anal Chem 58:770±772, 1975. 34. M A Al-Sketty, RKO Apenten. Modi®cations of protein dye binding assays. University of Leeds, 2001. 35. H Kuno, HK Kihara. Simple microassay of protein with membrane ®lter. Nature 215:974±975, 1967. 36. A Heil, W Zillig. Reconstitution of bacterial DNA-dependent RNA- polymerase from isolated subunits as a tool for the elucidation of the role of the subunits in transcription. FEBS Lett 11:165±168, 1970. 37. W Schaffner, C Weissmann. A rapid, sensitive, and speci®c method for the determination of protein in dilute solution. Anal Biochem 56:502±514, 1973. 38. RS Kaplan, PL Pedersen. Determination of microgram quantities of protein in the presence of milligram levels of lipid with Amido Black 10B. Anal Biochem 150:97±107, 1985. 39. K Nakamura, T Tanaka, A Kuwahara, K Takeo. Microassay for proteins on nitrocellulose ®lter using protein dye-staining procedure. Anal Biochem 148:311±319, 1985. 40. S Bramhall, N Noack, M Wu, JR Lowenberg. A simple colorimetric method for determination of protein. Anal Biochem 31:146±148, 1969. 41. GS McKnight. A colorimetric method for the determination of submicrogram quantities of protein. Anal Biochem 78:86±92, 1977. 42. A Esen. A simple method for quantitative, semiquantitative, and qualitative assay of protein. Anal Biochem 89:264±273, 1978. 43. H Ahmad, M Saleemuddin. A Coomassie Blue-binding assay for the microquantitation of immobilized proteins. Anal Biochem 148:533±541, 1985. 44. RLM Allen. Color Chemistry. London: Thomas Nelson & Sons, 1971. 45. L Sandler, FL Warren. Effect of ethyl chloroformate on the dye binding capacity of protein. Anal Biochem 46:1870±1872, 1974. 46. RF Hurrell, P Lerman, KJ Carpenter. Reactive lysine in foodstuffs as measured by a rapid dye-binding procedure. J Food Sci 44:1221±1227, 1979. 47. IM Perl, MP Szakacs, A Koevago, J Petroczy. Stoichiometric dye-binding procedure for the determination of the reactive lysine content of soya bean protein. Food Chem 16:163±174, 1985. The Udy Method 163

48. I MolnaÂ r-Perl, M PinteÂ r-SzakaÂ cs, D Medzihradszky. Dye-binding ``stoichiometry'' and selectivity of cresol red with various proteins. Food Chem 35:69± 80, 1990. 49. D Racusen. Stoichiometry of the Amido Black reaction with proteins. Anal Biochem 52:96±101, 1973. 50. I MolnaÂ r-Perl, M PinteÂ r-SzakaÂ cs, A KoÈ vago, I PetroÂ czy, UP KralovaÂ nszky, I MaÂ tyaÂ s. Dye-binding stoichiometry of AO12, AB 10B and OG with etalon proteins, feed and feedingstuffs and its application for reactive lysine determination. Food Chem 20:21±38, 1986. 51. WAB McBryde. Spectrophotometric determinations of equilibrium constants in solution. Talanta 21:979±1004, 1974. 52. IM Klotz, FM Walker, RB Pivan. The binding of organic ions by proteins. J Am Chem Soc 68:1487±1490, 1946. 53. IM Klotz. Spectrophotometric investigations of the interactions of proteins with organic anions. J Am Chem Soc 68:2299±2304, 1946. 54. SE Sheppard, RC Houck, C Dittmar. The sorption of soluble dyes by gelatin. J Phys Chem 46:158±176, 1942. 55. M Pesavento, A Profumo. Interaction of serum albumin with a sulphonated azo dye in acidic solution. Talanta 38:1099±1106, 1991. 56. KE Lind, U Kragh-Hansen, JV Moller. Protein binding to small molecules. V. Binding of bromophenol blue by chemical modi®cations of human serum albumin. Biochim Biophys Acta 371:451±461, 1974. 57. Y-J Wei, K Li, S-Y Tong. The interaction of bromophenol blue with proteins in acidic solution. Talanta 43:1±10, 1996. 58. Y-J Wei, K Li, S-Y Tong. Spectral study of interaction of thymol blue with protein in acidic solution. Anal Chim Acta 341:97±104, 1997. 59. AN Glazer. On the prevalence of ``nonspeci®c'' binding at the speci®c binding sites of globular proteins. Proc Natl Acad Sci USA 65:1057±1063, 1970. 60. AG Mayes, R Eisenthal, J Hubble. Binding isotherms for soluble immobilized af®nity ligands from spectral titration. Biotechnol Bioeng 40:1263±1270, 1992. 61. J Hubble, AG Mayes, R Eisenthal. Spectral analysis of interactions between proteins and dye ligands. Anal Chim Acta 279:167±177, 1993. 62. GS Hartley. The effect of long-chain salts on indicators: the valency-type of indicators and the protein error. Trans Faraday Soc 30:444±450, 1934. 63. L Michaelis, S Granick. Metachromasy of basic dyestuffs. J Am Chem Soc 67:1212±1219, 1945. 64. SE Sheppard, AL Geddes. Amphipathic character of proteins and certain lyophile colloids as indicated by absorption spectra of dyes. J Chem Phys 13:63±65, 1945. 65. RW Congdon, GW Muth, AG Splittgerber. The binding interaction of Coomassie Blue with proteins. Anal Biochem 213:407±413, 1993. 66. DA Skoog, DM West. Fundamentals of Analytical Chemistry. 3rd ed. New York: Holt, Rinehart & Winston, 1976. 67. PJ Wood, RG Fulcher. Interaction of some dyes with cereal beta-glucans. Cereal Chem 55:952±966, 1978. 164 Chapter 5

68. RG Seals. Some aspects of dye binding of milk and milk powder proteins. PhD thesis, Washington State University, 1960. 69. A Conetta, L Stooker, H Zehnder. An automated system for the determination of milkfat, protein and lactose in milk. Advances in Automated Analysis. Technicon International Congress, 1970, 2:81±85, 1971. 70. DL Park, RL King. Evaluation of automated dye-binding determination of protein in milk. J Assoc Of® Anal Chem 57:42±46, 1974. 71. M Kroger, EE Katz, JC Weaver. Determining protein content of ice cream and frozen desserts. J Dairy Sci 61:274±277, 1978. 72. JW Sherbon, HA Luke. Collaborative study of the dye binding method applied to chocolate milk drinks, cultured buttermilk, and half-and-half. J Assoc Of® Anal Chem 51:811±816, 1968. 73. JW Sherbon, HA Luke. Comparison of the dye binding and Kjeldahl methods for protein analysis of non-fat dry milk and ice cream. J Assoc Of® Anal Chem 52:138±142, 1969. 74. JW Sherbon. Dye binding method for protein content of dairy products. J Assoc Of® Anal Chem 53:862±864, 1970. 75. AL Lakin. The estimation of protein and the evaluation of protein quality by dye-binding procedures. ISFT Proc 6:80±83, 1973. 76. AL Lakin. Comparison of the amounts of dyes bound by milk proteins under the conditions employed in dye-binding procedures. XIX International Dairy Congress 1E:277±278, 1974. 77. RF Wilkinson, GH Richardson. Continuous ¯ow analysis of milk proteins using ultra-violet spectroscopy. J Dairy Sci 58:798, 1975. 78. T Kristoffersen, KH Koo, WL Slatter. Determination of casein by the dye method for estimation of cottage cheese curd yield. Cult Dairy Prod J 9:12±14, 1974. 79. JW Sherbon, R Fleming. Comparison of two formulations of Acid Orange 12 for the determination of protein in milk. J Assoc Of® Anal Chem 58:773±776, 1975. 80. JC Bruhn, S Pecore, AA Franke. Measuring protein in frozen dairy desserts by dye binding. J Food Prot 43:753±755, 1980. 81. N Rawson, RR Mahoney. Effect of processing and storage on the protein quality of spray-dried lactose-hydrolyzed milk powder. Lebensm Wiss Technol 16:313±316, 1983. 82. N Rawson, RR Mahoney. A modi®ed method for determination of reactive lysine in milk powder using Remazol Brilliant Blue R. Lebensm Wiss Technol 16:1±4, 1983. 83. AH Luke. Collaborative testing of the dye binding method for milk protein. J Assoc Anal Chem 50:560±564, 1967. 84. RH Kleyn. Frozen desserts under protein analysis. Dairy Ice Cream Field 159(7):44, 46, 1976. 85. JC Bruhn. Protein determinations in ice cream. Am Dairy Rev 40(2):34B± 34D, 1978. The Udy Method 165

86. JA O'Connell. Evaluation and modi®cation of the Pro-Milk Tester Mk II for protein estimation in milk. Lab Pract 19:1119±1120, 1970. 87. TCA McGann, JA O'Connell. Evaluation of the Pro-Milk automatic for rapid protein determination in milk. Dairy Ind 36:685±687, 1971. 88. TCA McGann, A Mathiassen, JA O'Conell. Applications of the Pro-Milk Mk II. IV. Monitoring the degree of denaturation of whey proteins in heat processing of milk, and the heat treatment classi®cation of milk powders. Lab Pract 21:865±871, 1972. 89. L Szijarto, DA Biggs, DM Irvine, DW Stanley. Mark II Pro-Milk Tester for estimation of protein percentage in plant milk supplies. J Dairy Sci 56:854± 857, 1973. 90. TCA McGann, JA O'Connell, R McFeely. Use of semi-automatic instruments for process and product control in the dairy-food industry. Routine assessment of total protein and heat treatment of milk powders by due binding. J Soc Dairy Technol 28:23±27, 1975. 91. R Grappin, VS Packard, RE Ginn, J Mellema. Precision of the Pro-Milk method in routine determination of protein in dairy testing laboratories. J Food Prot 43:52±53, 1980. 92. M Kroger, JC Weaver. Use of protein dye-binding values as indicators of the `chemical age' of conventionally made cheddar cheese and hydrolyzed-lactose cheddar cheese. J Food Sci 44:304±305, 1979. 93. H Lueck. The protein content of condensed milk as determined by the amino black dye-binding method. S Afri J Dairy Technol 5:77±79, 1973. 94. JC Radcliffe. Use of a recording spectrophotometer for Amido Black milk protein determinations. Aust J Dairy Technol 23:143, 1968. 95. JW Sherbon. Pro-Milk method for the determination of protein in milk by dye binding. J Assoc Of® Anal Chem 57:1338±1341, 1974. 96. M. Kroger. Techniques for milk protein testing. Food Prod Dev 6(7):68,77,1972. 97. M Uzonyi. Experiences with the Amido-Black 10B dye-binding method in the Hungarian dairy industry. Zesz Probl Postepow Nauk Roln 167:41±47, 1975. 98. MJ Patel, GK Patel, KC Patel, RD Patel. Protein determination in milk. Indian J Chem Educ 5:26, 1978. 99. KF Ng-Kwai-Hang, JF Hayes. Effects of potassium dichromate and sample storage time on fat and protein by Milko-Scan and on protein and casein by a modi®ed Pro-Milk Mk II method. J Dairy Sci 65:895±899, 1982. 100. A Reusel, CJ Klijn. Automated methods for routine analysis of raw milkÐthe dye-binding method for determination of the protein content of milk. Bull Int Dairy Fed 208:17±20, 1987. 101. J Roeper, RM Dolby. Estimation of the protein content of wheys by the Amido Black method. N Z J Dairy Sci Technol 6(2):65±68, 1971. 102. TCA McGann, A Mathassen, JA O'Connell. Rapid estimation of casein in milk and protein in whey. Ir Agric Creme Rev 26:17±22, 1973. 166 Chapter 5

103. E Renner, S Ando. Determination of the casein and whey protein contents of milk by Amido Black methods. XIX International Dairy Congress E:459±460, 1974. 104. EH Reimerdes, B Flegel. Casein micelles: heat-induced changes of the dye- binding capacity. XX International Dairy Congress E:2243±2244, 1978. 105. R Grappin, R Juenet, D Ale. Determination of the protein content of cows' and goats' milk by dye-binding and infrared methods. J Dairy Sci 62(Suppl 1):38±39, 1979. 106. RM Mabon, EY Brechany. The measurement of protein in fresh and stored goats' milk by a dye-binding technique. Lab Pract 31:26±27, 1982. 107. R Waite, GM Smith. Measurement of the protein content of milk from mastitic quarters by the Amido Black method. J Dairy Res 39:195±201, 1972. 108. WB Sanderson. Determination of undenatured whey protein nitrogen in skim milk powder by dye binding. N Z J Dairy Sci Technol 5:46±48, 1970. 109. JA O'Connell, TCA McGann. Rapid estimation of protein in skim milk powders. Ir Agric Cream Rev 25(110):17±19, 1972. 110. WT Greenaway. Comparisons of the Kjeldahl, dye binding, and biuret methods for wheat protein content. Cereal Chem 49:609±615, 1972. 111. Y Pomeranz, RB More. Reliability of several methods for protein determination in wheat. Bakers Dig 49:44±48, 58, 1975. 112. LC Parial, LW Rooney, BD Webb. Use of dye-binding and biuret techniques for estimating protein in brown and milled rice. Cereal Chem 47:38±43, 1970. 113. Y Pomeranz, RB Moore, FS Lai. Reliability of ®ve methods for protein determination in barley and malt. J Am Soc Brew Chem 35:86±93, 1977. 114. D Baker, WH Hunt. Pro-Meter evaluation. Cereal Foods World 20:246±247, 1975. 115. Y Pomeranz. Evaluation of factors affecting the determination of nitrogen in soya products by the biuret and Orange-G dye-binding methods. J Food Sci 30:307±311, 1965. 116. T Hymowitz, FI Collins, SJ Gibbons. A modi®ed dye-binding method for estimating soybean protein. Agron J 61:601±603, 1969. 117. L Sandler, FL Warren. Effect of ethyl chloroformate on the DBC of protein. Anal Chem 46:1870±1872, 1974. 118. IM Perl, MP Szakacs, A Koevago, J Petroczy. Stoichiometric dye-binding procedure for the determination of the reactive lysine content of soya bean protein. Food Chem 16:163±174, 1985. 119. I Molnar-Peal, M Pinter-Szakacs, D Medzihradszky. Dye-binding ``stoichiometry'' and selectivity of cresol red with various proteins. Food Chem 35:69± 80, 1990. 120. S Lin, AL Lakin. Thermal denaturation of soy proteins as related to their dye- binding characteristics and functionality. J Am Oil Chem Assoc 67:872±878, 1990. 121. CR Romo, AL Lakin, EF Rolfe. Properties of protein isolates prepared from ground seeds. I. Development and evaluation of a dye binding procedure for the measurement of protein solubility. J Food Technol 10:541±546, 1975. The Udy Method 167

122. MB Medina, DH Kleyn, WH Swallow. Protein estimation in sesame seed and rapeseed ¯ours and meals by a modi®ed Udy dye binding method. J Am Oil Chem Assoc 53:555±558, 1976. 123. YK Goh, DR Clandinin. The estimation of protein in rapeseed meal by a dye- binding method. Can J Anim Sci 58:97±103, 1978. 124. G Bunyan. Orange-G binding as a measure of protein content. J Sci Food Agric 10:425±430, 1959. 125. J Torten, JR Whitaker. Evaluation of the biuret and dye-binding methods for protein determination in meats. J Food Sci 25:168±174, 1964. 126. US Ashworth. Proteins in meat and egg products determined by dye binding. J Food Sci 36:509±510, 1971. 127. GJ Seperich, JF Price. Dye binding procedure for the estimation of protein content of meat components and sausage emulsions. J Food Sci 44:643±645, 1979. 128. JC Weaver, M Kroger, LR Kneebone. Comparative protein studies (Kjeldahl, dye binding, amino acid analysis) of nine strains of Agaricus bisporus (Lange) Imbach mushrooms. J Food Sci 42:364±366, 1977. 129. A Braaksma, DJ Schaap. Protein analysis of the common mushroom Agaricus bisporus. Postharvest Biol Technol 7:119±127, 1996.

The Bradford MethodÐPrinciples

1. INTRODUCTION

Proteins bind with Coomassie Brilliant Blue G250 (CBBG,* C.I. 42655) to produce a sparingly soluble complex (1). Protein-dye binding alters the absorption spectrum for CBBG. This is the basis of the assay developed by Bradford in 1976 (2). The Bradford assay has several advantages compared with the Lowry test: (a) four- to tenfold greater sensitivity, (b) tenfold greater speed, (c) decreased susceptibility to interferences, (d) requirement for a single reagent, and (e) lower cost. The Bradford assay is quicker than dye-protein precipitation (Udy assay) as no ®ltration step is required. Ready-to-use CBBG dye reagent is available from Bio-Rad Laboratory Ltd., Pierce Warriner Ltd., and the Sigma-Aldrich Chemical Company. Principles of the Bradford assay are described in this chapter. Applications for food protein analysis are discussed in Chapter 7. Coomassie Blue is the trade name for a group of dyes ®rst produced by Imperial Chemical Industries (ICI) Ltd. (UK). Weiler discovered CBBG in 1913. The 1971 edition of the Colour Index (3) lists 40 Coomassie dyes including Coomassie Blue FF (C.I. 42645), Coomassie Blue R (C.I. 42660), Coomassie Blue BL (C.I. 50315), Coomassie Brilliant Blue G (C.I. 42655), Coomassie Blue GL (C.I. 50320), and Coomassie Blue RL (C.I. 13390). The

* Abbreviations: CBBG, Coomassie Brilliant Blue G250;CBBR, Coomassie Brilliant Blue R250.

169 170 Chapter 6

FIGURE 1 The structure of Coomassie Brilliant Blue G250 (CBBG). Coomassie Brilliant Blue R250 (CBBR) lacks two methyl groups.

G and R labels refer to dyes having a greenish blue or reddish blue hue. Samples of dye with 2.5 times greater purity than the standard grade (19± 22% purity) are labeled ``250.'' CBBG is also known as C.I. Acid Blue 90, Xylene Brilliant Cyanine G, or Brilliant Blue G (4). CBBR* was ®rst used as a protein stain in 1963 by Fazekas de St. Groth et al. (5). Cellulose acetate electrophoresis support was soaked in sulfosalicylic acid to ®x protein bands and then transferred to the CBBR solution (0.25% w/v in water). Blue protein bands form against a clear background. Nonspeci®c staining increased if CBBR dye was prepared with methanol rather than water. From a densitometric analysis of polyacryla- minde gels, the blue color was proportional to protein amount (0±20 mg). In later developments, polyacrylamide gels were stained with CBBR dissolved with a 5:1:5 mixture of methanol, acetic acid, and water or 12.5% (w/v) TCA (6). Diezel et al. (1) introduced CBBG as a protein stain for electrophoresis. CBBG has two methyl groups more than CBBR (Fig. 1) and is therefore less soluble in 12% TCA. This lowers dye penetration into polyacrylamide gels and reduces background staining. The sensitivity of

* CBBR (CI 42660) is also known as Acid Blue 83, Coomassie Blue R, Xylene Brilliant Cyanine 6b, or Supranolcyanin 6B. Bradford MethodÐPrinciples 171

CBBG for protein is also signi®cantly greater than CBBR.* Reisner et al. (7) used perchloric acid (3.5%) as solvent for CBBG. The free dye exists as a colorless (leuco) molecule in perchloric acid solvent. Binding to protein leads to a blue protein-dye complex. There was virtually no background staining for polyacrylamide gels. Assay sensitivity was comparable to that obtained with Naphthylamine Black 10 (0.5% w/v) stain.{

2. THEORY OF THE BRADFORD ASSAY

Protein-protein variations in the sensitivity of the Bradford assay (8,9) led to interest in CBB-protein interactions. Coomassie Brilliant Blue binds with proteins by electrostatic interactions. The complex is also held together by van der Waals forces (5). The ®rst quantitative investigation of CBBR binding with proteins was reported by Tal et al. (10). Dye binding occurs only with polypeptides larger than about 3000 daltons. The number of dye molecules bound per molecule of protein (n) was strongly correlated with the total number of arginine, histidine, and lysine (Arg His Lys) residues. However, the average DBC was 100% greater than combined numbers of basic amino acid residues. Hydrophobic interactions may account for dye binding when charged protein sites were saturated. Rosenthal and Koussale (11) determined the critical micelle concentration (cmc) of nonionic surfactants with the Bradford reagent. Hydrophobic solubilization of CBBG within micelles led to marked increases in absorption at 620 nm.

2.1. Overview of CBBG Protein Binding Compton and Jones (12) assessed the effect of pH, protein, and surfactants on the absorption spectrum of CBBG. They concluded that CBBG exists in three ionized forms rather than two. The absorptivity of protein-CBBG complexes increased with protein molecular weight (13). Sign®cantly higher color yields were obtained for CBBG binding with polyarginine, polylysine, or polyhistidine. There was no dye binding with polyaspartic acid, polyglutamic acid, and polyproline. Lea et al. (14) recorded anomalous results with the Bradford assay for highly basic proteins. Chemical

* A polyacrylamide slab gel (10 6 15 cm) is soaked in 40 mL of 12.5% TCA for 5 minutes to ®x protein. Then 2.5 mL of CBBG solution (0.25% w/v dissolved in water) is added and the gel is incubated for 15±30 minutes. Transferring the gel to a 5% (w/v) acetic acid solution for 12 hours increases protein band intensity and reduces background staining. { This dye is the same as Amido Black 10B (Chapter 5). 172 Chapter 6 modi®cation to lower the numbers of positive charges increased (rather than decreased) the assay response for polylysine and histone. The expected reductions in assay sensitivity occurred when a more concentrated dye reagent solution was used (15). Protein-CGGB binding parameters were reported by Chial and Splittgerber (16), who also characterized the Bradford assay at pH 1 and pH 7. Congdon et al. (17) reported the dissociation constant for CBBG- protein binding. Although not involving CBBG, the study of Cibracron blue binding to lysozyme (18) is a ®ne example of contemporary methods for studying protein-dye binding. Atherton et al. (19) apply these principles to the CBBG system.

2.2. The Compton-Jones Scheme for CBBG Ionization

CBBG dye shows two lmax values at 470 and 650 nm at pH 0.8* (12). Adjusting the dye reagent to pH 1.2 produced the following changes: (a) decreased absorbance at 650 and 470 nm and (b) a new absorbance peak at 595 nm. There was no isobestic point, meaning that more than two interconverting dye species occurred over the pH range examined. In a different experiment, addition of 140 mg of BSA to CBBG dye reagent produced a difference spectra (dye protein versus dye) with lmax at 595 nm. Exposure of CBBG to excess SDS diminished the 470 peak and produced a new peak at 650 nm. Shareef and Shetty (20) identi®ed a fourth CBBG charged species. The 2 À purple bi-ionic form (CBBG , lmax 515 nm) appears at pH 11.5. With acidic conditions (pH 1.25) Coomassie Blue dye is a positively charged/ 1 cationic/red (CBBGH2) species (lmax 475 nm). This is in equilibrium 1 with other CBBG forms. (CBBGH2) is converted to the zero-charge/ 0 neutral/green (CBBGH) species (lmax 650 nm) at about pH 1.6. There- 1À after (CBBG) , which is the anionic blue form (lmax 595), is produced at pH 1.8±pH 7. The Compton-Jones scheme for CBBG ionization is summarized in Table 1. Crystal violet shows a similar three-state ionization as it changes color from violet to green to yellow at pH 8 to pH 2.4 and pH 0.6 (21);

1 2 3 R violet?RH green?RH2 yellow 1 Similar transformations occur with many other triphenylamine dyes (rosaniline, para-rosaniline, malachite green, aniline blue).

* Compton and Jones reported that the Bradford reagent had a pH of 0.8. Our own measurements found pH 1.25. Bradford MethodÐPrinciples 173

TABLE 1 The Compton-Jones Scheme Showing the Three-State Ionization of CBBG

Dye form Anion Neutral Cation

1À 0 1 Structure (CBBG) (CBBGH) (CBBGH2) Net charge À10 1 Color Blue Green Red/leuco lmax(nm) 595 650 470±475 ve charges 1 2 3 Àve charges 2 2 2 pH* 1.8±7 1.6 1.25

2.3. Spectrophotometric Analysis of CBBG±Protein Binding It is not easy to decide whether protein-bound CBBG is charged or not.

Compton and Jones (12) stated, ``based on the identical lmax for the dye- protein complex and the blue dye anion, . . . the bound dye species is in fact the dye anion. The (dye ionization) equilibrium shown above are forced to the left as the anion is bound by protein.'' CBBG binding to serum albumin leads to lmax 620 nm when the protein is present in excess (22). By contrast, lmax 595 nm with excess dye. A lmax value of 620 nm does not match the absorption maximum for any of the dye forms in solution (470, 595, or 650 nm;Table 2). Two possible models may be proposed for protein- CBBG binding.

A. Binding Scenario 1ÐExcess Protein Just prior to binding there is the anionic form (CBBG)1À in solution. Dye 1 binding is with the positively charged protein site (RNH3) . Protein-bound CBBG has a net charge of zero.

1À 1 0 CBBG RNH3 ? CBBG:RNH3 2 Electrostatic intereactions will neutralize charges on the dye and protein 0 molecule, forming (CBBG.RNH3) . The analogous reaction occurs between (CBBG)1À and the hydrogen ion:

CBBG1À H1? CBBGH0 3

0 Table 1 shows that lmax is 650 nm for the neutral dye species (CBBGH) in 174 Chapter 6

0 solution. For the protein-bound nuetral dye form (CBBG.RNH3) we ®nd lmax is 620 nm. To explain the hypsochromic shift from 650 to 620 nm, 0 consider the different dye environments for (CBBG.RNH3) and 0 0 (CBBGH) . The lmax shifts from 650 to 600±620 nm when (CBBGH) is internalized within the nonpolar environment of micelles formed by nonionic detergents. One implication is that charged protein sites for dye binding are (within) nonpolar environments. With scenario 1, the Bradford assay can be performed at 620 nm. Eq. (2) partly explains the correlation between DBC and total numbers of basic amino acids per molecule of protein. However, excess protein leads to the occupancy of strong binding sites (Chapter 5).

B. Binding Scenario 2ÐExcess Dye Just prior to binding, the free dye form is the blue (CBBG)1À species. Dye binding involves hydrophobic interactions with one of two types of neutral o 0 protein sites, R or (CBBGRNH3) . Protein-bound CBBG has a net charge of À1.

CBBG1À R0? CBBGR1À or

1À 0 1À CBBG CBBGRNH3 ? CBBG2RNH3 4 Dye binding shifts the dye ionization equilibrium toward (CBBG)1À and produces an absorbance increase at 595 nm. Equation (4) accounts for the 0 use of A595 readings for the Bradford assay. Dye binding with R does not account for the correlation between DBC and the number of positively charged basic amino acid residues. In contrast, the number of 0 (CBBGRNH3) sites is determined by the number of basic amino acids. The Bradford assay can be monitored at either 620 or 595 nm. The alternative dye-binding scenarios are not mutually exclusive. As described in Chapter 5, dye binding involves both nonpolar and ionic sites. CBBR binds to proteins with a dye basic amino acid ratio ranging from 1:1.5 to 1:2 (10). Compton and Jones (12) suggest that 60% of the Gibbs free energy change for protein-CBBG binding (*40 kJ molÀ1) is due to the nonpolar structure (benzene and methylene groups) of the dye. The remaining binding free energy (*30 kJ molÀ1) arises from protein interactions with the sulfonate groups of CBBG. The metachromatic properties of crystal violet provide relevant insights (23). In contemporary terms, metachromasia is a change in the dye absorption spectra due to changes in the dye environment. The lmax Bradford MethodÐPrinciples 175 may shift to shorter or longer wavelengths. Sometimes, the absorbance peak increases without lmax shifting. Crystal violet showed metachromatic behavior due to (a) increasing dye concentration, (b) addition of low concentrations of ammonium sulfate, or (c) binding to sites followed by optically signi®cant interactions between dye molecules. Low concentrations of ammonium sulfate induce the dimerization of crystal violet. The singly charged R (violet) form [see Eq. (1)] binds to phosphate groups from nucleic acids with a ratio of 1:1. For this reaction, lmax remains unchanged but the absorption peak increases. Crystal violet binding to agar sulfonate groups apparently shifts lmax from 585 to 510 nm as the peaks at these wavelengths decrease and increase, respectively. Each agar sulfonate group binds several dye molecules. The shift in lmax was ascribed to ``stacking interactions'' as excess dye molecules adsorb to dye molecules initially bound by electrostic interactions. Dye stacking is essentially a nonpolar process like those leading to dimerization. Examples of metachromatic phenomena were reported for dyes with nonionizing quaternary nitrogen groups;the process does not require dye ionization. Evidence from spectral measurements also supports two modes of 1À CBBG-protein binding. With excess protein lmax 620 and (CBBG) binds to positively charged sites. The formation of a neutral 1:1 protein-dye complex is accompanied by a metachromatic shift of lmax by *30 nm. By contrast, excess dye favors nonpolar interactions probably involving stacking. Eq. (4) shows two different hydrophobic binding sites;R 0 sites include the side chains of tyrosine, tryptophan, phenylalanine, leucine, and 0 1À isoleucine. (CBBGRNH3) is produced when (CBBG) binds to a charged protein group. The (CBBG)1À binding with nonpolar sites accounts for the 1À lmax value of 595 nm. Perhaps lmax is the same for (CBBGR) and CBBG1À because R0 sites are highly hydrated. Given the amphipathic nature of the CBBG molecule (Fig. 1), it is likely that the principal binding 0 sites (R and RNH4 ) are also amphipathic.

2.4. Quantitative Analysis of CBBG Binding to Proteins*

Protein-dye binding parameters (Kd, n, and De) were reported by Klotz (24,25), Klotz et al. (26), and also Aizawa (27). From the law of mass action, D À DA=De nP À DA=De K 5 d DA=De

The concentration of bound dye (Db)isDA/De while Df D À Db and

* All symbols are de®ned in Chapter 5, Sections 4.3 and 4.4. 176 Chapter 6 therefore

Df nP À Db Kd 6 Db

Db Kd Df nPDf 7 and

nPDf Db 8 Kd Df Eq. (8) describes protein-ligand binding with n independent sites. Linearization of this relation leads to Eqs (9)±(11) for extracting binding parameters (Kd and n). 1 K 1 d 9 Db Df nP nP

Multiplying this double-reciprocal equation with nPDb and rearrangement lead to D n D b À b 10 PDf Kd PKd

Substituting G Db/P, we obtain the equation for the Scatchard plot. G n G À 11 Df Kd Kd

The more familiar version of Eq. (11) (Db/Df nP/Kd À Db/Kd) is usable where the protein concentration is kept constant. Most dye-binding studies employ a constant concentration of dye while the protein concentration is varied. Two studies of CBBG-protein binding have been published. Congdon et al. (17) segregated protein binding sites for CBBG into ``strong'' and ``weak'' sites. To characterize strong binding sites, different amounts of BSA (0±41 mM) were added to a ®xed concentration (20 mM)* of CBBG. From a graph of 1/DA versus 1/P the x 0 yields a reciprocal for the maximum

* The concentration of CBBG was 0.0166% (w/v) or 200 mM. To a ®xed volume (100 mL) of dye reagent solution was added 0±2.66 mg of BSA. The ®nal volume of mixture was brought to

1 mL where needed with distilled water. Then DA620 was recorded for each protein concentration. Bradford MethodÐPrinciples 177 absorbance change (1/DAmax) at an in®nite protein concentration. From this, De DAmax/D. The absorbance measurements were also used to estimate the number of strong sites. First, values for Db ( DA/De) and Df ( D À Db) were determined for each protein concentration. Then results were analyzed using Scatchard and related equations. To determine the total number of binding sites, a ®xed concentration of CBBG (133 mM) was exposed to 0.5±100 mg of BSA (0.025±0.512 mM). The conditions (excess dye) are similar to those used for the normal Bradford assay. The number of binding sites was determined from the relation (16)

d DA nDe 12 dP

The left-hand side of Eq. (12) is supposedly the maximum gradient from the Bradford assay standard curve.* The results for CBBG binding to seven proteins are reported (Table 2). For BSA binding with T-azo-R, Eq. (11) gives a poor ®t to the results. The presence of two classes of binding sites (e.g., strong and weak sites) should lead to curvature in the Scatchard plot (28,29) although this has not been demonstrated. Because few graphs for CBBG-protein binding have been published, I have reexamined the data for BSA binding with T-azo-R

TABLE 2 Parameters for Coomassie Brilliant Blue G250 Binding to Selected Proteinsa

À1 À1 Protein Kd (mM) n(ns) eb(M cm ) BSA 18.6 105 (2.7) 48,000 Alcohol dehydrogenase 40.0 30 (7) 57,700 a-Lactalbumin 110.0 14 (2.4) 55,400 Glutamate dehydrogenase 8.9 ? (2.2) 57,900 Chymotrypsin 80.0 13 (2.3) 55,700 Ovalbumin 23.0 33 (1.9) 41,900 Carbonic anhydrase 35.0 28 (2.8) 51,900

a nS number of strong binding sites (in parentheses) and Kd are average values from the modi®ed Scatchard plot [Eq. (11)] and Hill plot [Eq. (15)]. Source: Based on results from Ref. 18.

* Eq. (12) is not an appropriate expression for the assay sensitivity. The right-hand expression should nDeD/Kd (Section 3 of this chapter). The consequence of using Equation (12) to estimate n is described in Section 5. 178 Chapter 6

(see Chapter 5, Figs 5 and 6 and associated text). The data from Ref. 30 were replotted using the Scatchard equation [Eq. (11)] or modi®ed Scatchard relations [Eqs (13)±(15)]. 1 n 1 À 13 Df GKd Kd

1 K 1 d À 14 G nDf n

G À 1 n À 1 K À d 15 G n nDf Equations for the straight lines and binding parameters are reported in Table 3. The results show a poor ®t to the Scatchard plot;the regression coef®cient (R) for the graph shown in Fig. 2 was 0.7310. From the equation of the straight line Kd 11.6 mM and nS 106. Other equations led to more gratifying transformations of the data. Using Eqs (13)±(15), R 0.9941. In Table 3, two data entries are shown for each graph. In the ®rst case, values for Kd and nS were assessed assuming that all data conform to a straight line for a single class of binding sites. Figs 2±5 show deviations from linearity at the extremes. The second data entry in Table 3 is derived from results ®tted to the main linear phase in each graph. Eqs (13) and (15) emphasize data collected at high protein

TABLE 3 Parameters for T-Azo-R Binding to Bovine Serum Albumin Analyzed Using Scatchard and Modi®ed Scatchard Plots

Linearization equation Equation Kd (mM) n Eq. (11) Y 9.21 6 106 ± 8.62 6 104 X 11.6 106 Eq. (13) Y 1.06 6 107 X 164 6 105 6.1 64 (strong sites)a 3.8 43 Eq. (14) Y 9.26 6 10À8 X 164 6 10À2 5.6 61 (weak strong sites)a 19.61 76 Eq. (15) Y 0.984 ± 9.26 6 10À8 X 5.8 63 (strong sites)a 4.2 48 aDifferent equations emphasize data collected at a high protein/dye ratio (strong sites) or a low protein/dye ratio (weak strong binding sites). Second data entries are calculated using the major linear phase of each graph. Bradford MethodÐPrinciples 179

FIGURE 2 Scatchard plot for T-Azo-R binding with bovine serum albumin. Dye (10 mM) was titrated with 0±6 mM BSA. Study was performed at pH 2.3. Data from Ref. 31 plotted in accordance with Eq. (8).

concentrations and dominated by strong binding sites. Eq. (14) emphasizes both weak and strong dye binding sites. It appears there are 43±48 strong binding sites on the BSA molecule for T-azo-R (Table 3). By comparison, ns 26 for thymol blue and 6 for bromophenol blue (see Table 5, Chapter 5). The number of strong binding sites depends on the type of dye and also the reaction conditions. Counting the numbers of weak as well as strong binding sites using a single experiment may require the approach described by Rosenthal (28). Chial and Splittgerber (16) assessed protein-CBBG binding at pH 1 and pH 7. The total number of binding sites (n) was determined using Eq. (12). Detailed procedures are the same as reported by Congdon et al. (17). A summary of results is given in Table 4. The assay at pH 7 had decreased sensitivity. In a number of cases (BSA, a-lactalbumin, carbonic anhydrase) the sensitivity change from pH 1 to pH 7 could apparently be explained using Eq. (12). A 40-fold decrease in assay sensitivity for BSA at pH 7 compared with pH 1 could arise from 2-fold and 20-fold decreases in the values n and De, respectively. Eq. (12) predicts a sensitivity change of 2 6 20 ( 40-fold). The second footnote in this section suggests that this agreement is fortuitous. 180 Chapter 6

FIGURE 3 Modi®ed Scatchard plot for T-Azo-R binding with bovine serum albumin. Dye (10 mM) was titrated with 0±6 mM BSA. Study was performed at pH 2.3. Data from Ref. 31 plotted in accordance with Eq. (10).

FIGURE 4 Modi®ed Scatchard plot for T-Azo-R binding with bovine serum albumin. Dye (10 mM) was titrated with 0±6 mM BSA. Data from Ref. 31 plotted in accordance with Eq. (14). Bradford MethodÐPrinciples 181

FIGURE 5 Modi®ed Scatchard plot for T-Azo-R binding with bovine serum albumin. Dye (10 mM) was titrated with 0±6 mM BSA. Data from Ref. 31 plotted according to Eq. (15).

In Table 4 the molar absorptivity for protein-bound CBBG at 620 nm 3 (eb,620) was not affected by solvent pH. Hence eb,620 was 49 (+8.3) 6 10 MÀ1 cmÀ1 at pH 1 or 53 (+6.0) 6 103 MÀ1 cmÀ1 at pH 7.0. In contrast, the À1 À1 free dye had an extinction coef®cient (ef,620) of 43,700 M cm at pH 7.0 and 8800 MÀ1 cmÀ1 at pH 1.* Consequently, we have De 39.2 6 103 MÀ1 cmÀ1 at pH 1 and De 9.3 6 103 MÀ1 cmÀ1 at pH 7. The reaction time for protein-dye binding was about 60 minutes at pH 7 compared with about 5 minutes at the pH for the normal assay.

2.5. Identifying CBBG Binding Sites on Proteins

The relative color yield for CBBG binding to different poly-L-amino acids was poly-L-lysine (1), poly-L-histidine (4.2), poly-L-tryptophan (4.4), poly-L- tyrosine (4.7), and poly-L-arginine (36) (12). With increasing dye concentration there was a more similar CBBG response to different basic amino acids

* It is not certain whether these calculations were corrected for changes in the dye ionization with pH. The low ef,20nm value at pH 1 could be due to a sixfold lower concentration of blue (neutral) dye species at pH 1 compared with pH 7. 8 hpe 6 Chapter 182

TABLE 4 Effect of pH on Dye Binding Parameters and Sensitivity of the Bradford Assaya

Sensitivityb À1 4 À1 À1 Protein (DA620 mg )610 De620(M cm ) n Lys Arg (pH 1.0) Alcohol dehydrogenase 127.0 49,200 28 32 Bovine serumalbumin 193.0 39,200 100 86 Carbonic anhydrase 125.0 43,200 25 27 Chymotrypsin 101.0 47,200 16 17 a-Lactalbumin 143.0 44,200 14 13 b-Lactoglobulin 84.0 25,200 18 18 Ovalbumin 87.0 34,200 33 36 (pH 7.0) Alcohol dehydrogenase 2.6 11,300 2 Bovine serumalbumin 4.8 18,300 5 Carbonic anhydrase 0.9 7,300 1 Chymotrypsin 2.1 1,300 2 a-Lactalbumin 15.0 9,300 6 b-Lactoglobulin 3.0 12,300 12 Ovalbumin 0.9 8,300 2 a À1 À1 À1 À1 b ef (620) 43,700 M cm (pH 7) or 8800 M cm (pH 1). Sensitivity as determined from the maximum slope of the calibration graph. Source: Ref. 17. Bradford MethodÐPrinciples 183

(10); DA595 readings were 0.92, 1.2, and 1.5 for poly-L-lysine, poly-L- histidine, and poly-L-arginine, respectively. These studies agree on the importance of poly-L-arginine as a CBBG binding site. Moreno et al. (13) reported a relative color yield from different polyamino acids as poly-L- lysine (1), poly-L-tyrosine (1.9), poly-L-arginine (3), poly-L-histidine (>5.5). CBBG binding increased with the polypeptide molecular weight, but differences in color yield are less marked when absorbance changes are expressed per unit mass (microgram) of material analyzed. In summary, studies involving CBBG/R binding to poly-L-amino acids indicate that these dyes bind to basic groups. Nonpolar amino acid residues, notably tyrosine and tryptophan, are also important binding sites. Additional nonpolar sites for CBBG binding are created as the anionic dye molecule binds to positively charged protein sites. However, poly-L-amino acid results should be treated with caution. Proteins rarely feature the high density of sites associated with homopolymers. The correlation between CBBG binding and the number of basic amino acids occurs when ionic bonding predominates and the dye species is limiting (Chapter 5). The relations between protein-binding parameters (n, Kd, De) and characteristics of the Bradford assay are discussed in the next section.

3. EFFECTOF PROTEIN-DYE BINDING PARAMETERSON THE BRADFORD ASSAY

The plot of DA versus protein concentration leads to a hyperbolic calibration graph.* nDePD DA 16 Kd nP Eq. (16) is a result of the equilibrium between free dye, protein, and the bound dye. At low protein concentrations nP << Kd and

DA nDeDP=Kd 17 Therefore, a graph of DA versus P gives a straight line with a gradient equal to the sensitivity F, where

Sensitivity FnDeD=Kd 18

A linear calibration graph will be obtained when nP << Kd. For a linear relationship between DA and [P] it is necceary that Kd exceeds protein

* This relation is the same as Equation (12) of Chapter 5. 184 Chapter 6 concentration multiplied by the number of binding sites per molecule of protein. Eq. (18) also predicts that assay sensitivity will increase with (a) number of dye binding sites, (b) molar absorptivity change due to dye binding, (c) protein-dye binding af®nity (proportional to 1/Kd), and (d) increasing dye reagent concentration. Assay sensitivity will also be independent of the concentration of protein (provided nP << Kd). Another prediction from Eq. (18) is that protein-protein variations in assay results are not inevitable. Assume that De is the same for most proteins. Using BSA as standard, the relative assay sensitivity for different proteins can be stated as F/FBSA, where F n=K % d 19 FBSA n=Kd BSA Protein-protein variations in assay sensitivity are related to values for n and Kd. These parameters are not immutable. Encouraging nonspeci®c protein-dye binding may reduce protein-protein variations in sensitivity (see later).

4. LINEARIZATION PLOTS FOR THE BRADFORD ASSAYS

A double reciprocal plot (1/DA versus 1/[P]) or log(DA) versus log[P] will extend the linear range of a hyperbolic Bradford calibration graph (Chapters 3, Sections 3 and 4). Other linearization schemes have been proposed for the Bradford assay. Sedmak and Grossberg (31) increased the upper limit of linearity from 10 mg BSA to 50 mg by plotting A620/A456 versus protein amount. Bearden (32) obtained an upper limit of linearity of 40 mg

BSA by plotting the difference between A595 and A465 readings against the amount of BSA. Zor and Selinger (33) plotted A595/A450 versus protein concentration. These linearization schemes allow for (a) the hyperbolic protein-dye binding pro®le and (b) the overlap of the absorbance spectra for the free dye and bound dye.

5. ASSAY SENSITIVITY AND THE MAXIMUM NUMBER OF DYE BINDING SITES

Eq. (18), which properly describes the sensitivity of the Bradford assay, reduces to Eq. (12) only when [D] & Kd. Otherwise, the total number of binding sites calculated from Eq. (12) will be in error by the factor D/Kd. Bradford MethodÐPrinciples 185

This error appears in estimates for the total number of sites reported in Table 2 and Table 4. The CBBG concentration used in the studies was *113 mM and hence values for n need revising downward by 113/Kd. A further error appears with respect to pH 1 data in Tables 2 and 4. Allowing for a 0 shift from (CBBGH) to the (CBBGH2) species at low pH, ef,620 and hence De probably remain unchanged with pH (De620 is approximately 18,300). Eq. (12) then gives n 214 for BSA at pH 1. The corrected number of binding sites is 35 (i.e., 214 6 Kd/D). A further independent estimate comes from Fig. 1 of Chapter 7. A plot of DA/P versus D gives the regression line 10 Y 5.94 6 10 D. The gradient of this graph is n De/Kd [see Eq. (18)]. À6 À1 À1 Assuming that Kd 18.6 6 10 MandDe 18,300 M cm for BSA (see Tables 4 and 6), we obtain n 56. The preceding corrections lead to new estimates for the maximum number of CBBG binding sites for BSA (n 35±56). As BSA has approximately 117 Arg His Lys sites, only 30±48% of available cationic sites bind CBBG. In Table 4 the reported agreement between numbers of CBBG binding sites and (Arg Lys) residues is fortuitous. In any case, the n estimate from Equation (12) (Table 4) exceeds the total of Arg Lys residues. Hydrophobic interactions were invoked to explain excess dye binding. CBBR-protein complexes precipitate within the interstices of polyacrylamide gels. The number of dye binding sites exceeds the number of protein cationic sites (Lys Arg His) up to 150% (10).

However, Kd values for protein-CBBR binding are similar to those in Table 6.4: lysozyme (35.7 mM), cytochrome c (83.3 mM), RNAse (125 mM), trypsin (83.3 mM), pepsinogen (37 mM), pepsin (77 mM), and gramicidin S (111 mM). Between 1.3 and 3.0 CBBR molecules bind for each basic amino acid residue. Initially CBBR molecules bind to protein via 1:1 ionic interactions. There then follows the uptake of a second CBBR molecule via nonpolar interactions. This binding scheme accounts for (a) the correlation between DBC and numbers of basic residues and (b) the involvement of nonpolar interactions in protein-dye binding. Wilson (4) suggested the same idea.

6. SOLID-PHASE DYE-BINDING ASSAYS

To perform a solid-phase assay, the protein is ®rst bound to ®lter discs and stained with CBBG dye reagent. Destaining is carried out to remove nonspeci®cally bound dye and the stained protein spot is excised using a cork borer and immersed in a solubilization solvent. The amount of dye solubilized is measured from DA readings at 600±630 nm. A calibration graph can be produced as usual by analyzing standard concentrations of protein. Solid-phase dye-binding assays are described in Chapter 5. 186 Chapter 6

7. INTERFERENCE COMPOUNDS AND SAMPLE PRETREATMENT 7.1. Interferences and Compatible Solutes Compounds that interfere with the Bradford protein assay are listed in Table 5. These include food constituents such as chlorophyll, pectin, and ethanol and low-molecular-weight surface-active agents. Reducing compounds such as ascorbic acid have a fading effect on triphenylmethane colors (34). The action is slow and requires about 7 days. Compounds thought to be compatible with the Bradford method are listed in Table 6. Many buffer salts, EDTA, hydroxycinnamic acid derivatives, and low concentrations of ¯avanols apparently do not interfere with protein-CBBG interactions (Section 7.3). The preceding information is intended only as a rough guide. Food components not listed should be compared with their nearest listed relative. For example, pectin and gum arabic are interfering compounds. Expect interference from other structural polysaccharides (alginate, carrageenans) and plant gums (acacia, Tara gum, etc.). Tables 5 and Table 6 are not de®nitive guides: (a) potential interferences are usually tested in the absence of protein, (b) many interferences were tested using the prototype Bradford method (involving low dye concentrations), and (c) some interferences have

TABLE 5 A List of Interferences for the Bradford Assaya

Interfering substances

Ampholyte (pH 3±10.0;1 %) Potassium chloride (1M) Acetone Quercitin Apigenin rRNA Chlorophyll Rutin Chrysin SDS (1%) Detergents (various) Sodium acetate (2 M) Ethanol (95%) Sodium chloride (5 M) Fisetin Sodium hydroxide (2 M) Glycerol (99%) Sucrose (2 M) Gum arabic Tannic acid Hemosol (1%) Tris (2 M) Kaempferol Triton X-100 (0.1%) Magnesium chloride (1M) Tween 80 (>0.0001%) Myricetin Urea (>0.2 M) Pectin Vanadate Phenol (5%) aSummarized from Refs. 2, 13, 21, 32. Bradford MethodÐPrinciples 187 no effect or contrary effects at concentrations below those shown in Table 6-7. Interferences that compete with CBBG for protein binding will not be detected in the absence of protein. Some interferences have a different effect if the dye reagent/sample volume ratio is lower than used with the standard Bradford assay.

7.2. Detergents and Low-Molecular-Weight Surfactants

Nonionic detergent concentrations above the cmc produce a shift in the lmax value for CBBG from 650 nm to 600±620 nm. Thereafter A620 increases at the expense of A470 and A650 values. CBBG binding to nonionic detergents leads to an increase in A595 compared with A650. The reverse is observed for CBBG binding to anionic detergents (SDS). At concentrations below the cmc, SDS reduces the assay sensitivity by competing with dye molecules

TABLE 6 A List of Compounds Considered Compatible with the Bradford Method

Noninterfering substances

Adenosine (1 mM) Potassium chloride (1 M) Amino acids Potassium phosphate (pH 7;1 M) Ammonium sulfate (1 M) rRNA (0.25 mg/mL) Asparagine Rutin ATP (1 mM) Sinapic acid BES (2.5 M) Sodium acetate (0.6 M) Cacodylate-Tris (0.1 M) Sodium citrate (0.05 M) Caffeic acid Streptomycin sulfate (20%) CDGA (0.05 M) Thymidine (1 mM) Dethiothreitol (1 M) tRNA (0.4 mg/mL) DNA (1 mg/mL) Tyrosine (1 mM) EDTA (0.05 M) Valine Ethanol (95%) Glucuronic acid Ferulic acid Sitosterol Formic acid (1 M) Stigmasterol Glycine (0.1 M) Maltol HEPES (0.1 M) Chlorogenic acid Hexyl-b-D-glucopyranose (<10% w/w) Apigenin Maleic acid Phloretin MES (0.7 M) Chrystin MOPS (0.2 M) Fisetin PIPES (0.5 M)

Source: Summarized from Refs. 2, 13, 21, 32. 188 Chapter 6 for charged and noncharged protein sites. The spectral changes are due to the solubilization of neutral (CBBG)0 species inside detergent micelles (11). The positively charged dye species (CBBG)1 binds to SDS, forming a neutral complex, which is then solubilized within detergent micelles (35). To avoid interference, Zaman and Verwilghen (36) reduced the concentration of SDS in protein samples by precipitating with potassium salts. The SDS-depleted sample is then assayed using the Sedmak-Grossberg assay (Chapter 7, Section 2.5). Kapp and Vinogradov (37) removed SDS from protein samples using a column ®lled with the ion-exchange resin AG11A8 (Bio-Rad Laboratories). The SDS binding capacity was 1.7 mg/ mL (wet resin). Binding of the detergent was extremely ®rm and no simple techniques were found to regenerate the support. The recovery of protein from this support ranged from 52 to 90%. SDS may be removed from samples using charcoal cartridges made in house (38). Preextracting SDS from PAGE gels (with charcoal placed within a dialysis bag) allows gel staining using the conventional Bradford reagent. This can therefore serve as a dual-purpose reagent for protein analysis in solution as well as for PAGE. Alkyl-b-D-glucopyranoside detergents are compatible with the Brad- ford assay. Fanger (39) screened a number of commercially available detergents for solubilizing membrane proteins.* Low interferences were obtained with octyl-b-D-glucopyranose and related alkyl-b-D-glucopyrano- sides. There was tolerance for up to 10% (w/w) detergent in protein samples. Detergents with low cmc have more adverse effects on the Bradford assay (Fig. 6). Nonionic detergents showed a bell-shaped concentration response. Friedenauer and Berlet (40) found the optimum color yield for a Triton X-100 concentration of 0.008% in the ®nal assay mixture;the cmc for Triton X-100 is about 0.01% (w/w). Assay sensitivity increased by 11 to 128% with an average increase of 33% for 15 different proteins examined. Apparently, Triton X-100 facilitates protein-dye interactions via noncovalent bonding. The detergent had no effect on protein conformation. To enhance sensitivity the detergent should be added to the protein sample before dye.

7.3. Plant Secondary Metabolites Aromatic amino acids (tyrosine, phenylalanine, and tryptophan) from the shikimic acid pathway are not interferences. Hydroxycinnamic, caffeic acid,

* Detergent solutions (10% w/w;10 mL) were added to 4.9 mL of Bradford dye reagent and A595 and A650 measurements recorded. Bradford MethodÐPrinciples 189

FIGURE 6 The effect of detergents on the standard Bradford assay. A 100-mL portion of detergent solution (10% w/v) was added to 4.9 mL of Bradford dye reagent and absorbance measurements recorded 595 or 650 nm. (Drawn using data from Ref. 40.) ferulic acid, and sinapic acid are also compatible solutes. Flavanoids and related substances possessing a chalcone skeleton (anthocyanins, anthox- anthins, proanthocyanadins, and tannin) can interfere (41). Other potential interferences from plants include products of the mevalonate pathway (carotenoids, lycopenes, terpenes, etc.) and the molanic acid pathway (lipids, phospholipids, and lipid hydroperoxides). Some phenol-CBBG complexes have an absorption spectrum that overlaps that for the protein-CBBG complex. There is no evidence that protein precipitation with tannins is a source of error. Techniques for preparing plant proteins, to minimize interactions with phenolic compounds, are reviewed by Loomis and Battaile (42) and Loomis (43). Phenolic metabolites bind to proteins via (a) hydrogen bonding to the peptide oxygen, (b) oxidation to quinones followed by covalent reactions with the protein e-NH2 or SH group, (c) ionic interactions with phenolate or carboxylate groups, and (d) hydrophobic interactions between the phenolic benzene ring and aromatic amino acid residues. Some techniques for controlling protein-phenol interactions include (a) maintaining sample pH at just below neutral pH;(b) addition of phenolic adsorbents, e.g., polyvinylpryrrolidone (PVP), Polyclar AT, Amberlite XAD-2, XAD-4, or XAD-7;(c) use of antioxidants and phenol oxidase inhibitors including 190 Chapter 6 chelators, sul®te, and ascorbic acid;(d) addition of protective proteins such as BSA;and (e) use of inert atmospheres such as argon, carbon dioxide, or nitrogen (42,43). The Bradford assay is more resistant to interference by phenolics than the biuret, Lowry, and BCA assays. Indeed, Folin Ciocalteu and BCA reagents are used for the quantitative analysis of phenolics and other food antioxidants. These assays are therefore not suited to plant-derived samples including leafy vegetables, fruits, stem, and tubers. Processed foods such as beverages (tea, coffee, beer, wine, chocolate) are also not readily analyzed by the Lowry or BCA assays. The Bradford assay is widely applied to beer and wine protein analysis (Chapter 7, Sec. 4). Sample pretreatment involves size exclusion chromatography to remove low-molecular-weight interferences from beer or wine. Dialysis or treatment with phenolic adsorbents is also feasible.

7.4. Sugars, Glycation, and Glycosylation Products* Simple monosaccharides and disaccharides have no direct effect on the Bradford assay. At concentrations of 1 mM to 1 mM the following sugars produced no color with CBBG dye reagent: galactose, glucose, lactose, maltose, mannose, N-acetylglucosamine, and sucrose (44). However, protein glycation or glycosylation reduces the sensitivity of the Bradford assay. The formation of advanced glycation products (AGPs) is a feature of the pathology of diabetes. Incubating human serum albumin (HSA) with 1 M glucose for 12 weeks also leads to the glycation of lysine and arginine groups. So does heating dry mixtures of protein and sugar at oven temperatures overnight. The modi®ed proteins show improved foaming and emulsi®cation properties. The chemistry of protein-sugar reactions is similar to the early steps of the Maillard reaction (see Chapter 9). Brimer et al. (45) observed that nonenzymatic glycation interferes with the Bradford assay. HSA was underestimated by about 20% after glycation and exhaustive dialysis. Amino acid analysis showed a reduction of lysine and arginine residues of 45±54% and 24±30% due to glycation. There was no major error in the Bradford assay after HSA was incubated with 25 mM glucose. Relatively high levels of glycation were needed to cause interference. The Bradford assay gave 50% lower estimates for native glycosylated proteins (ovalbumin, ®brinogen, horseradish peroxidase, IgM,

* Glycation usually describes carbonyl-amine reactions between sugars and proteins in vitro as well as in vivo. Glycosylation is the attachment of sugar residues to proteins immediately after synthesis (posttranslational modi®cation) during protein secretion from cells. Many glycosylated proteins contain oligosaccharide units linked to asparagine, serine, and threonine. Bradford MethodÐPrinciples 191 and soybean trypsin inhibitor) (44). In some cases, the relative sensitivity of the Bradford assay was related to differences in protein hydrophobicity. The rate of color formation was also marginally reduced for some glycosylated versus nonglycosylated proteins. Rocher et al. (46) identi®ed six distinct g- and o-secalins by fractionation of rye storage proteins some of which were glycosylated. A 40-kDa rye protein was exceptional in having signi®cant absorbance in the visible region, low CBBG binding, and immunoreactivity. Other naturally glycosylated proteins [peanut conarachin, soybean conglycinin (47), lupin conglutin (48)] might show abnormally low reactivity with CBBG. The total protein content of rapeseed ¯our determined by the Bradford method did not agree with results of Kjeldahl method (49). The mechanism by which sugars interfere with the Bradford assay is uncertain. Sugars have no direct interactions with CBBG. Glycosylated proteins contain oligosaccharide units linked to asparagine (Asn-X-Ser or Asn-X-Thr), serine, or threonine residues that are not obviously involved in protein-CBBG binding (50). Perhaps steric hindrance and the generally lower hydrophobicity of glycosylated proteins may account for the reduction in CBBG binding.

7.5. Polysaccharides, Lipids and Nucleic Acids Interferences by nucleic acids, lipids, and polysaccharides have not been assessed in detail. Nondialyzable polysaccharides from cane sugar may bind CBBG (51). Alkorta et al. (52) reported results consistent with pectin binding with CBBG. Compared with either the Lowry or modi®ed Lowry assays, the Bradford method was resistant to interference from pectin at concentrations below 5 g LÀ1. Other acidic polysaccharides from terrestrial plants, marine plants, or bacteria are widely used in foods. The possible effects of gums and stabilizers on the Bradford assay have yet to be investigated. Interference by lipids is discussed in Chapter 7.

REFERENCES

1. W Diezel, G Kopperschlager, E Hoffman. An improved procedure for protein staining in polyacrylamide gels with a new type of Coomassie Brilliant Blue. Anal Biochem 48:617±620, 1972. 2. MM Bradford. A rapid and sensitive method for the quantitation of microgram quantities of protein utilising the principle of protein-dye binding. Anal Biochem 72:248±254, 1976. 192 Chapter 6

3. Colour Index. Vol 5. Bradford, W. Yorkshire, England: The Society of Dyers and Colourants, 1971, p 5400. 4. CM Wilson. Studies and critique of Amido Black 10B, Coomassie Blue R, and Fast Green FCF as stains for proteins after polyacrylamide gel electrophoresis. Anal Biochem 96:263±278, 1979. 5. S Fazekas de St. Groth, RG Webster, A Daytner. Two new staining procedures for quantitative estimation of proteins on electrophoretic strips. Biochim Biophys Acta 71:377±391, 1963. 6. A Chramback, RA Reisfeld, M Wyckhoff, J Zaccari. A procedure for rapi and sensititive staining of protein fractionated by polyacrylamide gel electrophoresis. Anal Biochem 20:150±154, 1967. 7. AH Reisner, P Nemes, C Pcholtz. The use of Coomassie Brilliant Blue G250 perchloric acid solution for staining in electrophoresis and isoelectric focusing on polyacrylamide gels. Anal Biochem 64:509±516, 1975. 8. J Pierce, CH Suelter. An evaluation of the Coomassie Brilliant Blue G250 dye binding method for quantitative protein determination. Anal Biochem 81:478± 480, 1977. 9. H Van Kley, SM Hale. Assay for protein by dye binding. Anal Biochem 81:485±487, 1977. 10. M Tal, A Silbersteine, E Nusser. Why does Coomassie Brilliant Blue R interact differently with different proteins: a partial answer. J Biol Chem 260:9976±9980, 1980. 11. KS Rosenthal, F Koussale. Critical micelle concentration determination of non-ionic detergents with Coomassie Brilliant Blue G-250. Anal Biochem 55:1115±1117, 1983. 12. SJ Compton, CG Jones. Mechanism of dye response and interference in the Bradford protein assay. Anal Biochem 151:369±374, 1985. 13. MR De Moreno, JF Smith, RV Smith. Mechanism studies of Coomassie Blue and silver staining. J Pharm Sci 75:907±911, 1986. 14. MA Lea, A Luke, C Martinson, O Velazquez. In¯uence of carbomoylation on some physical properties of basic polypeptides. Int J Peptide Prot Res 27:251± 260, 1986. 15. MA Lea, A Luke. Effect of carbomoylation with alkyl isocyanates on the assay of protein by dye binding. Int J Peptide Prot 29:561±567, 1987. 16. HJ Chial, AG Splittgerber. A comparison of the binding of Coomassie Brilliant Blue to proteins at low and neutral pH. Anal Biochem 213:362±369, 1993. 17. RW Congdon, WG Muth, AG Splittgerber. The binding interactions of Coomassie Blue with proteins. Anal Biochem 213:407±413, 1993. 18. AG Mayes, R Eisenthal, J Hubble. Binding isotherms for soluble immobilised af®nity ligands from spectral titration. Biotechnol Bioeng 40:1263±70, 1992. 19. BA Atherton, EL Cunningham, AG Splittgerber. A mathematical model for the description of the Coomassie Brilliant Blue protein assay. Anal Biochem 233:160±168, 1996. Bradford MethodÐPrinciples 193

20. MM Shareef, T Shetty. Effect of vanadate on different forms of Coomassie Brilliant Blue and protein assay. Anal Biochem 258:143±146, 1998. 21. EQ Adams, L Rosenstein. The colour and ionisation of crystal violet. J Am Chem Soc 36:1452±1473, 1914. 22. AG Splittgerber, J Sohl. Nonlinearity in protein assays by the Coomassie Blue dye-binding assay. Anal Biochem 179:198±201, 1989. 23. L Michaelis, S Granick. Metachromasia of basic dyes. J Am Chem Soc 67:1212±1219, 1945. 24. IM Klotz. Spectrophotometric investigations of the interactions of proteins with organic anions. J Am Chem Soc 68:2299±2304, 1946. 25. IM Klotz. The application of the law of mass action to binding by proteins. Interactions with calcium. Arch Biochem 9:109±117, 1946. 26. IM Klotz, FM Walker, RB Pivan. The binding of organic ions by proteins. J Am Chem Soc 68:1486±1490, 1946. 27. H Aizawa. Interactions of food colours with proteins. Part XII. Studies on the bindings of the food xanthene colours with trypsin. J Food Hyg Soc Jpn 12:81±85, 1971. 28. HE Rosenthal. A graphical method for the determination and presentation of binding parameters in a complex system. Anal Biochem 20:525±532, 1967. 29. JG Norby, P Ottolenghi, J Jensen. Scatchard plot: common misinterpretation of binding experiments. Anal Biochem 102:318±320, 1980. 30. M Pesavento, A Profumo. Interaction of serum albumin with a sulphonated azo dye in acidic solution. Talanta 38:1099±1106, 1991. 31. JJ Sedmak, SE Grossberg. A rapid, sensitive and versatile assay for protein using Coomassie Brilliant Blue G250. Anal Biochem 79:544±552, 1977. 32. JC Bearden Jr. Quantitation of submicrogram quantities of protein by an improved protein dye binding assay. Biochim Biophys Acta 533:525±529, 1978. 33. Z Zor, Z Selinger. Linearisation of the Bradford protein assay increases its sensitivity: theoretical and experimental studies. Anal Biochem 236:302±308, 1996. 34. DM Marmion. Handbook of US Colorants: Foods, Drugs, Cosmetics, and Medical Devices. New York: John Wiley & Sons, 1991. 35. GL Boccaccio, LA Quesada-Allue. Interference of sodium dodecyl sulfate in the Bradford assay for protein quantitation. An Assoc Quim Argent 77:79±88, 1989. 36. Z Zaman, RL Verwilghen. Quantitation of protein solubilized in sodium dodecyl sulfate mechaptoethanol-tris electrophoresis buffer. Anal Biochem 100:64±69, 1979. 37. OH Kapp, SN Vinogradov. Removal of sodium dodecyl sulfate from proteins. Anal Biochem 91:230±235, 1978. 38. RC Duhamel, E Meezan, K Brendel. A charcoal cartridge for the removal of anionic detergent and electrophoresis stains. J Biochem Biophys Methods 4:73±80, 1981. 194 Chapter 6

39. BO Fanger. Adaptation of the Bradford assay to membrane-bound proteins by solubilizing in glucopyranoside detergents. Anal Biochem 162:11±17, 1987. 40. S Friedenauer, HH Berlet. Sensitivity and variability of the Bradford protein assay in the presence of detergents. Anal Biochem 178:263±268, 1989. 41. JH Von Elbe, SJ Schwartz. Colorants. In O Fennema, ed. Food Chemistry. New York: Marcel Dekker, 1996, pp 651±722. 42. WD Loomis, J Battaile. Plant phenolic compounds and the isolation of plant enzymes. Phytochemistry 5:423±438, 1966. 43. WD Loomis. Overcoming problems of phenolics and quinones in the isolation of plant enzymes and organelles. Methods Enzymol 31:528±544, 1974. 44. M Fountoulakis, J-F Juranville, M Manneberg. Comparison of the Coomassie Brilliant Blue, bicinchoninic acid, and Lowry quantitation assays using nonglycosylated and glycosylated proteins. J Biochem Biophys Methods 24:265± 274, 1992. 45. CM Brimer, RP Murray-McIntosh, TJ Neale, PF Davis. Nonenzymatic glycation interferes with protein concentration determinations. Anal Biochem 224:461±463, 1995. 46. A Rocher, M Calero, F Soriano, E Mendez. Identi®cation of major rye secalins as celiac immunoreactive proteins. Biochim Biophys Acta 1295:13±22, 1996. 47. JR Books, CV More. Current aspects of soy protein fractionation and nomenclature. J Am Oil Chem Soc 62:1347±1354, 1985. 48. M Durante, E Cucchetti, P Cerletti. Changes in composition and subunits in the storage proteins of germinating lupin seeds. J Agric Food Chem 32:490± 493, 1984. 49. YL Folawiyo. PhD thesis, University of Leeds, 1996. 50. L Stryer. Biochemistry. 3rd ed. New York: WH Freeman, 1988, p 773. 51. MA Godshall. Interference of plant polysaccharides and tannin in the Coomassie Blue G250 test for protein. J Food Sci 48:1346±1347, 1983. 52. I Alkorta, MJ Llama, JL Serra. Interference by pectin in protein determination. Lebens Wiss Technol 27:39±41, 1994. 7

Bradford AssayÐApplications

1. INTRODUCTION

The principles of the Bradford assay (1) are discussed in Chapter 6. Applications of the Bradford assay for food protein analysis are discussed in this chapter. Section 2.2 introduces several modi®cations of the Bradford assay designed to increase its compatibility with buffer salts, reduce protein- protein variations in results, and enable analysis of detergent-containing samples. Protein precipitation using the TCA-DOC method is not suitable for the Bradford assay. Section 2.3 describes several protein precipitation methods not using detergents. The performance characteristics of the different assays are summarized in Section 3. Section 4 of this chapter is a review of applications of the Bradford assay for food protein analysis.

2. COOMASSIE BRILLIANTBLUE DYE-BINDING ASSAYS 2.1. The Bradford Assay The design of the Bradford assay owes much to classical dye-binding assays. There are similarities in the choice of buffer and the relative volumes for protein and dye reagent:

Method 1 CBBG dye binding assay for proteins (1).

195 196 Chapter 7

Reagents 1. Coomassie Brilliant Blue G250 (Sigma) 2. Ethanol (95% w/v) 3. Phosphoric acid (85% w/v) 4. Bovine serum albumin

Procedure Preparation of CBBG dye reagent. Dissolve CBBG (100 mg) in 50 mL of ethanol. Add 100 mL of phosphoric acid (85% w/v) and dilute to 1 L with distilled water. Filter through Whatman No. 1 paper and store in a stoppered bottle.* Standard protein assay. Pipette 100 mL of protein (100±1000 mgmLÀ1) into several 1.2 6 10 cm test tubes. Add 5 mL of dye reagent to each.

Mix and record A595 readings after about 5 minutes. Prepare a reagent blank with 100 mL of solvent and 5 mL of dye reagent. Microassay. Pipette 100 mL of protein standard (10±100 mgmLÀ1) into small test tubes.{ Add 1 mL of dye solution to each, mix, and record

A595 readings against an appropriate reagent blank. Calibration graph. Plot DA595 (i.e., absorbance corrected for the blank reading) versus amount of protein (10±100 mg). Determine the concentration of protein in 0.1mL of the unknown sample by referring to the calibration graph. A blue color develops within 2 minutes of mixing protein and dye solutions. The color remains virtually constant for up to 20 minutes and decreases slowly between 20 and 50 minutes. Assay reproducibility can be improved by measuring A595 readings within 5 and 60 minutes after mixing dye and protein. The calibration features of the Bradford assay are described in Chapter 6.

2.2. Modi®cations and Variants of the Bradford Assay The Sedmak-Grossberg (2) assay employs perchloric acid (3.5% w/w) as the solvent for CBBG. The sensitivity is higher than with Method 1 but the new method is subject to interferences by strongly buffered samples. The use of perchloric acid solvent for CBBG was ®rst suggested by Reisner et al. (3).

CBBG is red (lmax 465) when dissolved in 3.5% (w/v) perchloric acid solution. Addition to protein generates a blue protein-dye complex with

* The ®nal reagent composition is nominally 0.01% CBBG in 8.5% phosphoric acid and 4.5% ethanol. { Plastic microcentrifuge tubes are convenient vessels for microassay. Bradford AssayÐApplications 197 lmax 595 nm. Using a low concentration of mineral acid (nitric or sulfuric acid) as solvent leads to reagent instability with regard to small changes in sample ionic strength. If CBBG is dissolved with high concentrations of mineral acid, there is poor color change when the dye is added to protein. Weak organic acids (e.g., acetic, formic, and isobutyric acid) are not suitable solvents for CBBG because very high concentrations are needed to maintain the dye pH (2). The Sedmak-Grossberg reagent gives an immediate color reaction with proteins. The absorbance change is stable for 60±90 minutes. With 0.6 M HCl as solvent, the color was stable for 3±4 hours. Thereafter the protein-dye complex formed a precipitate. The main disadvantage of this assay is its susceptibility to pH variations. Samples containing 0.1 M phosphate buffer (pH 7.0) or Tris-HCl buffer (pH 8.0) are not usable. Gel ®ltration fractions with low buffer capacity (*10 mM) can be analyzed accurately. Bearden (4) modi®ed the Sedmak-Grossberg and Bradford assays to (a) increase the color stability, (b) increase assay sensitivity, and (c) reduce interference by buffer salts. The new dye reagent was prepared as described in Method 1 but without ethanol. Upon mixing equal volumes of protein and dye solutions (thereby avoiding the 10- to 50-fold dilution of samples by the dye reagent), a blue complex is formed in 1±2 minutes. The absorbance increases slowly over 60±90 minutes, then decreases from 90 to 180 minutes.

After 3 hours, A595 readings were within +8% of the maximum. The increase in color stability was attributed to the omission of ethanol from the reagent formulation. The protein-dye complex may be less soluble in perchloric acid plus organic solvent (ethanol or methanol). Plastic cuvettes are to be avoided as they may catalyze protein-dye precipitation. The Read and Northcote (5) assay reduced protein-protein variations in assay results by increasing the CBBG dye reagent concentration. The order of color intensity for Method 1 with different proteins is RNAse < ovalbumin < lysozyme < BSA cytochrome c (6);trypsin < chymotrypsin < pepsinogen < lysozyme < BSA < cytochrome c (7). More uniform responses were achieved by increasing the CBBG concentration. Table 1 lists a range of CBBG dyes from various suppliers (5,8). Commercial samples vary greatly in dye content because different users require different levels of purity. Only the highest purity CBBG (98% purity) samples should be used for protein analysis. The technical grades have 20±50% (w/w) dye matter. Variations in dye purity makes it impossible to compare the performance of Bradford reagents (prepared with CBBG) from different suppliers. Ideally, dye reagent formulations should be standardized as suggested by Sedmak and Grossberg (2). Reagents prepared from different batches of dye should be diluted to provide a CBBG content of 0.01% (w/v) 198 Chapter 7

TABLE 1 Composition of Coomassie Brilliant Blue G250 Samples Used for Protein Analysis

Dye purity (%), real dye content, and Dye (designation and supplier) comments

Coomassie Brilliant Blue G250, Sigma 36±60% purity (see Method 1). Bradford (Poole, UK or St. Louis, MO) reagent has 0.0036±0.006% (w/w) dye. Serva Coomassie Blue G, Serva *50% purity. Dye solution (0.01%) has

(Heidelberg, Germany) A550 0.6. Saturated solution 0.018% (w/v) dye. Serva Blue G, Serva (Heidelberg, 98% purity. Dye solution of 0.01% (w/v)

Germany) has A550 1.18. Saturated solution 0.01% (w/v) dyea Coomassie Blue G, Eastman (Rochester, 19±22% purity NY) a Dye purity is given as percent dry weight (% DW). This information has not been checked against current dye grades from suppliers. The formula weight of CBBB 854 Da. Saturating concentrations correspond to &117 mM. Source: Compiled from Refs. 5 and 8.

with A550 1.18 (Table 1). This is actually twofold greater than the CBBG concentration in Method 1. Stoscheck (9,10) reduced protein-protein variations in assay results by 15-fold. She achieved this by adding NaOH to the assay mixture. Alkali also increased assay sensitivity by 2.7-fold. Either 5±20 mL of 10 M NaOH or 50± 200 mL of 1 M NaOH should be added to 1 mL of Bradford reagent. The modest (0.02±0.12) pH change leads to no changes in dye ionization. NaOH is thought to produce a salting-out effect leading to enhanced protein-dye binding by hydrophobic interactions. The quantity of NaOH needed decreases with increasing CBBG dye concentration. At low dye concentrations about 200 mL of 1.0 M NaOH was needed for optimum effect. Dye reagent prepared from high-purity Serva Blue G dye required 50 mLof 1.0 M NaOH per mL of dye solution. The NaOH can be added to the protein sample or to the Bradford dye reagent. The effect of adding other salts (e.g., ammonium sulfate) on the Bradford assay should be examined. In summary, the Read-Northcote-Stoschek modi®cation can be implemented by simply using high-purity CBBG grade (such as Serva grade;Table 1) for routine analysis. The ®nal dye reagent should be alkalinized with 50 mL (of 1.0 M NaOH) per mL (dye reagent) before use. SDS, which is commonly added to electrophoresis buffers, lowers the sensitivity of the Bradford assay (Chapter 6). With the Zaman and Verwilghen (11) version of the Bradford assay, SDS is removed from Bradford AssayÐApplications 199 samples by precipitation with potassium phosphate. Boccaccio and Qesada- Allue (12) used potassium chloride as the precipitant for SDS. After centrifugation, the SDS-depleted samples are analyzed using Method 1. The linear range for analysis was 0±50 mg BSA with a sensitivity 10-fold lower than observed with SDS-free samples. The sensitivity for SDS-depleted samples reached 80% of the value for control samples when Boccaccio and Quesada-Allue (12) analyzed insect cuticular protein extracted with the aid of SDS. This approach should work for crustaceaÐcrabs, lobster, and related commodities. Proteins dissolved in electrophoresis buffers (normally containing 2% w/w SDS and 5% w/w 2-mercaptoethanol) were also successfully analyzed after simple dilution (13). Typically, 10 mL of sample was diluted to 100 mL with distilled water. Then 5 mL of Bradford dye reagent was added. Calibrations with BSA dissolved in electrophoresis sample buffer showed a linear dynamic range up to 100 mg BSA. Once more the assay sensitivity was 10-fold lower in the presence of SDS but comparable to the sensitivity of the Lowry assay.

2.3. Protein Precipitation for the Bradford Assay The DOC-TCA precipitation method (14,15) is not suitable for the Bradford assay. Normally 100 mL of DOC (5% w/w) solution is added to 1 mL of protein solution followed by 100 mL of TCA (72% w/w). Chang (16) found that several nonionic detergents (cholate, deoxycholate, Triton X- 100, Triton X-114, Tween 20, Nonidet P-40) formed dark precipitates with TCA. These interfered with the Bradford, Lowry, and BCA assays. Kirazov et al. (17) reported that >0.1% (w/w) sodium DOC formed a heavy precipitate with the Bradford dye reagent. Alternative methods for protein precipitation that are compatible with the Bradford assay are described here. Proteins can be precipitated with TCA at low temperature for the Bradford assay. With cultured animal cell protein the recovery by TCA precipitation was 98 + 0.17% (18).* To ensure a quantitative recovery of protein, the ®nal TCA concentration in the sample should be 10±20% (w/w).

* To 100 mL of protein solution add 25 mL of TCA (72% w/w). Incubate the sample at 08C on ice for 30 minutes and then centrifuge at 10,000g using a microcentrifuge. Discard the supernatant and remove excess TCA using absorbent paper. Redissolve the protein pellet with 50 mLof1M NaOH. For refractory samples incubate at 378C for 2±3 hours. Add 1 mL of Bradford reagent and record absorbance values at 595±620 nm. 200 Chapter 7

Protein precipitation should be performed at 08C. Bensadoun and Weinstein (14) obtained erratic protein recoveries of 7±91% by precipitating proteins using 6% (w/w) TCA at room temperature. Protein pellets from TCA precipitation dissolve readily with 0.1±1 M NaOH before analysis with the standard Bradford assay. Protein coprecipitation with calcium phosphate is also compatible with the Bradford assay (19). This mild sample pretreatment eliminates inferences from high concentrations of detergents and lipids.* Accurate results were obtained for protein samples containing 2 mg of cardiolipin, egg yolk phospholipid (mainly lecithin), or phosphatidylcholine. Other interferences can be annuled by this treatment, including potassium chloride (0.5 M), glycerol (30% v/v), Triton X-100 (3% w/w), SDS (0.33% w/w), CHAPS (2.5% w/v), and deoxycholate (0.075% w/v). With protein coprecipitation, interference effects were reduced by 40±120 times.

2.4. Puri®cation of Coomassie Brilliant Blue The purity of CBBG should be high in order to produce sensitive, accurate, and precise protein analyses. Impurities may affect the dye absorption characteristics as well as protein-dye binding. CBBG can be puri®ed by precipitation with sodium chloride (8){ or ammonium sulfate (20). The former method was ®rst applied to CBBR (8) and later adapted for CBBG pur®cation (21). Starting with Sigma-grade CBBG (Table 1), dye samples of 98% purity were obtained with a yield of 50%. The puri®cation fold was not reported for the ammonium sulfate method. Dye contaminants produced an absorption peak at 442 and 652 nm. Pure CBBG dissolved in 7.5% (w/w) acetic acid shows three major absorbance peaks at 590, 310, and 210 nm.

* Adjust the sample (100 mg protein) volume to 200 mL with distilled water, if necessary. Add 10 mL of potassium phosphate buffer (0.5 M pH 7.4), 10 mL of calcium chloride (0.25 M), and 1 mL of ethanol (80% v/v);mix after each addition. Centrifuge at 7000 g and remove the supernatant by aspiration. Wash the protein pellet by adding 100 mL of water and 1 mL of ethanol. Centrifuge and remove the supernatant as before. Repeat the washing step if necessary. Add CBCG dye concentrate (100 mL) and allow at least 5 minutes for the pellet to dissolve. Next add 400 mL of distilled water and record absorbance readings at 595 nm. { Dissolve CBBG or CBBR in a minimum volume of water-methanol (2:1) solvent. Add an equal volume of sodium chloride (5 M). Filter and dry CBBG precipitate at 808C. Alternatively, dissolve the CBBG solid with methanol. Dry the methanol solution to recover solid CBBG. Bradford AssayÐApplications 201

3. PERFORMANCE CHARACTERISTICS OF CBBG DYE-BINDING ASSAYS 3.1. Linearity, Sensitivity, and Precision A. Bradford Assay The calibration graph yields a hyperbolic pro®le over an extended range of (5±125 mg) protein. An approximately linear equation provides an adequate ®t for 0±50 mg BSA. Later investigators reported a narrower linear dynamic range of 0±10 mg of BSA (6,7). The assay reproducibility was 1.2% with a À1 sensitivity of 0.011 (DA595 mg ). The accuracy was satisfactory. Equal levels of color were obtained with HSA, BSA, hemoglobin, chymotrypsinogen, and cytochrome c (1). The Bradford microassay is reportedly less sensitive than the standard assay (1). Analysis of 5 mgmLÀ1 BSA gave an absorbance change of 0.1 (microassay) or 0.27 (standard assay). The results are due to different degrees of sample dilution. The volume of dye reagent is 5 mL for the standard assay and 1 mL for the microassay. To examine the effect of sample dilution, Spector (22) assayed ®xed volumes of BSA (0.1 mL;5±500 mgmLÀ1) with varying amounts of dye reagent (0.5, 1, 2, or 5.0 mL). The assay sensitivity and linear dynamic range can be seen from Table 2. With increasing volumes of dye reagent the assay sensitivity decreased while the linear dynamic range increased. These results remove a common misconception about so-called micro assays, namely that these have (a) increased sensitivity and (b) a more restricted linear dynamic range compared with the ``standard'' format. Where results are normalized for the total assay volume, sensitivity and linear dynamic range are independent of dye volume. Dilution effects can easily be addressed by ensuring that the same sample and reagent volumes are adopted for calibration and for

TABLE 2 Effect of Dye Reagent Volume on the Sensitivity of the Bradford Assay

Assay Sensitivityb Sensitivityc Linear range a À1 À1 volume (mL) DA595 mg DA595 (mL mg ) (mg) 5 0.011 0.058 5±50 2 0.028 0.059 2±20 1 0.057 0.063 1±10 0.5 0.099 0.059 0.5±5 a Volume of dye reagent added to 0.1 mL of BSA ( 0.5 mg mLÀ1). b Sensitivity determined from the slope (F) of the calibration graph. c Sensitivity calculated as F multiplied by ®nal assay volume. Source: Adapted from Ref. 22. 202 Chapter 7 sample analysis. Spector (22) suggested that 4- to 5-month-old samples of CBBG gave lower sensitivity. It is good practice to run several calibration samples alongside every assay as this adds minimally to the overall effort. Chiappelli et al. (18) compared the sensitivity of the Bradford and Lowry assays for proteins in crude cell-tissue extracts. The dye reagent was reportedly stable for 4±6 months when stored in a brown glass bottle at room temperature. The Bradford assay gave consistently lower protein readings compared with the Lowry assay. The ratio of protein values for the Lowry and Bradford assay was 1.58 (+0.09). It was proposed that results from the two assays could be interconverted using the factor 1.58. However, this conversion factor was not constant. Lowry assay results for TCA- precipitated proteins from the adrenal medulla were virtually identical to those from the Bradford assay if BSA was selected as standard protein. Using bovine g-globulin as standard, the ratio of results from the Lowry and Bradford assays was 2:1 (23).

B. Sedmak-Grossberg Modi®cation

A plot of A620 versus the amount of BSA (0±60 mg) produced a curvilinear graph. A linear response was obtained for 0±10 mg BSA. An extended linear graph was produced by plotting A620/A465 versus protein. The reproducibility of the standard and microassay formats was 4±5% for 10 replicate À1 measurements with 10 mg of BSA. The sensitivity was 0.044 (DA595 mg ). There was no color change with polypeptides smaller than 3000 Da. This so- called molecular weight selectivity is an important feature of CBBG dye- binding assays (Section 4). Many low-molecular-weight nonprotein constituents do not interfere with the Bradford assay (2).

C. Bearden's Modi®cation The linear dynamic range was 0±40 mg protein. For samples containing 0.02±1 mg of protein the assay sensitivity was 0.045 (DA mgÀ1), which is four times greater than that for the conventional Bradford method (4).

3.2. Effect of Coomassie Brilliant Blue Dye Concentration Increasing the CBBG dye reagent concentration led to the following effects: (a) a two- to four fold increase in assay sensitivity, (b) increased rate of color formation, and (c) reduced protein-protein variations in assay sensitivity. Optimum assay sensitivity was attained using Serva Blue G dye of 98% purity (Table 1). Addition of 50 mL of 1.0 M NaOH per mL dye reagent reduced protein-protein variations by 15-fold. Assay sensitivity increased by a further 2.7-fold (5). Bradford AssayÐApplications 203

Fig. 1 shows the linear relationship between assay sensitivity and

CBBG concentrations above a critical value, C*dye. With BSA as a standard protein C*dye was < 0.003%. With lysozyme a dye concentration (apparently) exceeding 0.01% was necessary for dye-protein binding. Concentrations of dye equal to C*dye might be necessary to act as nucleation sites for dye binding. Small quantities of dye may induce a conformational change in the protein before widespread binding occurs. Because C*dye differs for a range of proteins, it is not related to a ``dye-centered'' phenomenon such as micellation. Clearly, the sensitivity of the Bradford assay was underestimated in most early studies using low-purity CBBG. The maximum solubility for 98% pure CBBG, is 0.01% (w/v) or 117 mM.* The optimum dye reagent formulation described by LoÈ f¯er and Kunze (24) apparently had 0.017% CBBG (Serva grade) in 10.5% (w/w) phosphoric

FIGURE 1 Effect of CBBG concentration on the sensitivity of the Bradford assay. Bovine serum albumin (5 mg) was analyzed with dye Serva Coomassie Blue G (50% ``true'' CBBG dye content). Circles sensitivity, crosses apparent maximum numbers of dye-binding sites calculated from Eq. (11) of Chapter 6, i.e., n & sensitivity/De. Assume that De 44200 MÀ1 cmÀ1 for CBBG binding with BSA. (Based on results from Ref. 5.)

* Notice from Table 1 that Serva Coomassie Blue G has 50% dye purity. With classical studies using the dye grade the real dye reagent concentrations are one half the values published. 204 Chapter 7 acid and 2.5% (w/w) ethanol.* Making the dye reagent 0.008% (w/v) with respect to Triton X-100 increased assay sensitivity 25-fold compared with the standard Bradford assay. The effect of Triton X-100 and other detergents on the Bradford assay is described further in the next section.

3.3. Effect of Added Salts and Hydrophobic Interactions Nonspeci®c dye-binding interactions are expected to reduce protein-protein variations in assay results. Stoscheck (9) produced a 15-fold reduction in protein-protein variations by adding 50±200 mL of alkali (1 M NaOH) to 1 mL of Bradford reagent. The assay sensitivity for all proteins tested increased by an average of 2.7-fold. The added salt probably increased hydrophobic interactions between protein and dye. The sensitivity increase with Triton X-100 was also explained in terms of increasing hydrophobic interactions. Read and Northcote (5) reduced protein-protein variations by increasing the CBBG dye reagent concentration. Excess dye is thought to promote hydrophobic dye-protein interactions (Chapter 6).

4. APPLICATIONS TO FOOD PROTEIN ANALYSIS 4.1. Beer Adequate levels of protein ensure beer foam stability. However, too much protein leads to haze formation in cold beer. Rapid methods for beer protein analysis are necessary for improved quality assurance and for monitoring brewing processes. Beer is an aqueous extract from malted barley to which hops are added for ¯avor. The ®nished product contains protein, polypeptides, NPN, polyphenolic dyes, carbohydrates, and nucleic acids. The Kjeldahl method, although approved for beer protein analysis, is subject to interferences by NPN. Plant-derived substances interfere with the biuret, Lowry and BCA methods. Lewis et al. (25) were ®rst to use the Bradford assay for beer protein analysis (Table 3).{

* A dye concentration of 0.01% (w/w) appears to be saturating for a solvent comprising 8.5% phosphoric acid and 4.5% ethanol. The saturating concentration of CBBG appears to have doubled for a solvent with higher (12.5%) phosphoric acid and lower (2.5%) ethanol. { Beer and wort samples were subjected to ultra®ltration using a 10 kDa molecular size cutoff membrane to remove interferents. Pretreatment is also possible by size exclusion chromatography (SEC) with a Sephadex G50 (2.2 6 40 cm) column. Then 0.1±0.5 mL of pretreated beer or wort was added to 5 mL of Bradford reagent with mixing. A595 or A620 readings were recorded after 10±40 minutes. Bradford AssayÐApplications 205

TABLE 3 Analysis of Beer Proteins Using the Bradford Assay

Application and comments Reference

First reported analysis of beer proteins Lewis et al. (25)

Assay results (A595 or A620) agree with Hii and Herwig (26) micro-Kjeldahl analysis of alcohol- insoluble crude protein Proposal for organized collaborative American Society of Brewing Chemists study (27) Bradford assay results vs. beer foam Dale and Young (28) stability (head retention index) Collaborative study;of®cial approval American Society of Brewing Chemists deferred (29) Investigation of chillproo®ng using silica Kano and Kamimura (30) gel Analysis of foam polypeptides Onishi and Proudlove (31) Bradford and pyrogallol method for Williams et al. (32) beer protein

Their results were compared with micro-Kjeldahl and Lowry assays. Some potential interferences (reducing sugar, total carbohydrates, a-amino acids) were also measured. The Bradford assay detected high-molecular- weight (>4 kDa) polypeptides eluted within the column void volume during size exclusion chromatography. Lowry- and Kjeldahl-positive substances eluted after the void volume and were not detected. Wort was found to contain 660 mg LÀ1 or 4670 mg (protein) LÀ1 by the Bradford and Lowry methods, respectively. Simulated beer samples containing 20% (w/v) amino acids, phenols, nucleic acids, or other common beer constituents showed that these had no effect on the Bradford assay. Indeed, such preliminary studies (25) suggest that the Bradford assay can accurately monitor protein levels during brewing operations (chill proo®ng, proteolysis, mashing, lautering, and wort boiling). Hii and Herwig (26) compared the Bradford and Kjeldahl assay for beer protein. Sample pretreatment involved degassing beer by standing overnight and/or size exclusion chromatography with the Bio-Rad P2 (2562.5 cm) column.* To avoid interference from NPN, Kjeldahl analysis was performed with beer protein precipitated by 80% alcohol. Bradford

* Samples were eluted with 0.05 M NaCl (¯ow rate 84 mL hrÀ1). Fractions eluting from the column void volume (>2 kDa molecular size) were collected for analysis;0.4 mL of sample was added to 5 mL of Bradford reagent and the assay performed as described in Method 1. 206 Chapter 7 assay results were correlated (R 0.95) with crude protein content (N 6 6.25). For 29 beer samples with a range of 5±250 mg (protein) LÀ1 the regression equation was

À1 A595 0:001428 cP mg L À0:02857 1 The CV for analysis was 5.8%. Sensitivity toward beer proteins was lower than for lysozyme. Beer total nitrogen had 63±80% NPN. The protein standard should be chosen carefully. Compared with other protein assays, the Bradford technique had advantages of greater speed, higher resistance to interferences, and high molecular weight speci®city. Dale and Young (28) examined the correlation between Bradford assay results and beer foam stability (head retention index). Beer was pretreated by SEC using FPLC* with Sephadex G25, G50, and G57 columns. Fractions eluted in the void volume were collected for analysis. There was a signi®cant correlation (R 0.91) between foam stability and Bradford assay results. Foam stability was related to the quantity of polypeptides with sizes >5, >30, or >80 kDa. Beer contained polypeptides ranging from 2 to 100 kDa. A collaborative test for the Bradford assay organized by the American Society of Brewing Chemists was reported in 1987 (27,29).{ The major ®ndings of the ASBC study were: 1. The Bradford method is rapid, convenient, and highly sensitive for beer proteins. 2. The precision of analysis for within-laboratory errors is 2.6±4.5%. The between-laboratory errors range from 15.5 to 35.5%. 3. Degassing (14 hours vs. 1 hour standing) and the degree of mixing beer and dye reagent (3 vs. 5 seconds) have a signi®cant effect on the test results. 4. Temperature (208C versus 258C), color development time (10±55 minutes), or light exposure has no signi®cant effect on the Bradford assay. 5. Results with different dye reagents are unacceptably different (Fig. 2). The recommendation to grant the Bradford method ``approved'' status was deferred in 1987. Between-laboratory differences were considered too great. Variation arose from the use of different commercial samples of CBBG in different laboratories (Fig. 2). Ready-made dye reagents were

* FPLC=Fast protein liquid chromatography. { Two samples each of premium beer, light beer, and malt liquor were analyzed by 13 laboratories. Bradford AssayÐApplications 207

FIGURE 2 The effect of different commercial dye reagents on beer protein analysis by the Bradford assay. Boxed legend refers to different beer samples. (Drawn from results of Ref. 29.) Some data points were calculated. from Piece Ltd. or Bio-Rad Ltd. Other CBBG samples were purchased as solids and the dye reagent solutions were prepared in house. Within the remit of the test, it was not possible to determine which dye reagent gave accurate results. However, the ``true'' protein content of the beers was not determined. I discussed problems likely to arise from batch-to-batch differences in CBBG dye purity (Table 1). To improve assay repeatability (between-laboratory reproducibility), strictly de®ned dye formulations should be used. Williams et al. (32) analyzed protein levels for stout beer, bitter, and lager from British retail outlets. The CBBG reagent from Bio-Rad Ltd. was used. Beer was analyzed as is or after exhaustive dialysis to remove low- molecular-weight components. Results from this investigation are illustrated in Fig. 1 of Chapter 1. Reliable estimates for beer protein were obtained with the Bradford and the PMR* assays (Table 4). Both techniques had a CV of 1.7%. Kjeldahl, Dumas, biuret, Lowry, and BCA methods were subject to error due to the presence of dialyzable inteference compounds in beer.

* PMR=Pyrogallol-red molybdate. 208 Chapter 7

TABLE 4 Protein Levels in Some British Beers Determined Using the Bradford and PRM Methodsa

Protein (mg LÀ1)

Sample Bradford PRM

Stout 460 1050 430 800 Lager 200 630 160 500 Bitter 270 570 250 500 a Lower values show protein values after dialyzing samples. Source: Summarized from Ref. 32.

The effect of different protein assays on the study of beer foam stabilization was highlighted by Onishi and Proudlove (31). Beer foam proteins were fractionated by hydrophobic interaction chromatography, yielding ®ve fractions. In order of descending hydrophobicity, these were fraction 5 > 4 > 3 > 2 > 1. Each fraction was repeatedly ultra®ltered to remove solutes <3 kDa and then assessed for their foaming characteristics. Protein quantitation was via the BCA or Bradford assay. Estimates of protein content from the BCA method were 2.4±5.4 times greater than results from the Bradford assay (Table 5). The degree of foam stabilization by proteins from whole lager, foam protein isolate, or foam protein fraction

TABLE 5 Apparent Protein Levels in Lager Beer Fractions 1±5 Separated by Hydrophobic Interaction Chromatographya

À 1 Foam Protein (mg L ) fractions BCA Bradford

1 356 66 28432 3 181 114 45622 5 356 144 Total 1036 378 a Fractions 1±5 are listed in order of increasing hydrophobicity. Source: Values estimated from Ref. 35. Bradford AssayÐApplications 209

TABLE 6 Analysis of Wine Proteins by the Bradford Assay

Applications and comments Reference

First assay of wine protein Hsu and Heatherbell (33) Color reaction takes 60±90 minutes. Murphy et al. (34) Grape proteins, soluble solids, and Murphy et al. (35) maturation Phenol-CBBG and CBBG-protein Brenna and de Vecchi (36) spectra overlap. Pretreatment by SEC Sensitivity to wine protein is *34% of Waters et al. (37) response to BSA. Direct protein analysisÐpolyphenols Marchal et al. (38) and ethanol cause 30±90% error. Bradford assay underestimates wine Boyes et al. (39) protein by 10-fold. Pretreat samples with NaOH.

5 were equal when normalized for protein concentration. The BCA assay results were affected by reducing substances such as melanoidins.

4.2. Wine and Fruit Juice Wine protein is frequently analyzed using the Bradford assay (Table 6). Like beer, wine contains endogenous polyphenols and reducing compounds. The wide range of protein values reported for wine (1 mg LÀ1±1 g LÀ1) probably re¯ects inherent differences between types of wines, grape cultivars and maturity, and processing methods. Some variations are due to simple measurement error. The Bradford assay is more resistant to interfering compounds from wine (see Lowry, biuret, BCA, and Kjeldahl methods). However, claims that high concentrations of polyphenols (<400 mg LÀ1) do not interfere with dye binding were challenged (36). The choice of standard proteins for wine analysis is crucial. Research on wine proteins is mainly concerned with haze formation and identifying the protein fraction(s) responsible for wine instability. Hsu and Heatherbell (33)* in 1987 were probably the ®rst to analyze grape and

* Grapes were homogenized with liquid nitrogen using a Waring blender and then freeze dried. The resultant powder (3 g) was extracted for 2 hours with 30 mL of 0.1 M citrate, 0.2 M phosphate buffer (pH 5) containing Amberlite XAD-4 (6 g), PVP (3 g), and two protease inhibitors (diethyldithiocarbamate, 5 mM, and phenylmethylsulfonyl ¯uoride, 5 mM). The resultant mixture was ®ltered through muslin cloth and then centrifuged. The clear supernatant was analyzed by the Bradford assay. Then A595 readings were recorded after 15 minutes. 210 Chapter 7 wine proteins using the Bradford method. Treating grape juice or wine with ion-exchange resin (XAD-4) and PVP reduced total phenols by 44 and 57% (Table 7). However, protein-quinone interactions were already complete during the wine manufacturing process. Electrophoresis analysis achieved poor resolution of wine proteins despite treatment with Amberlite and PVP. There was also signi®cant adsorption of wine proteins to phenol adsorbents. In particular, it is thought that hydrophilic proteins adsorbed strongly on tannins, polyphenols, and phenol adsorbents.

The rate of A595 increase during wine protein analysis was examined by Murphy et al. (34). The A595 reached a maximum after 90 minutes for wine proteins and after 2±10 minutes with BSA. The incubation time for wine protein analysis using the Bradford assay has to be extended and strictly controlled. Dissolving BSA with protein-depleted wine reproduced the slow reaction with CBBG;addition of polyphenols to BSA in idealized buffer solutions did not. Wine protein values depend on the maturity of the grapes used by the manufacturer. GewuÈ rztraminer and white Riesling grapes were found to contain 50±120 mg (protein) LÀ1;such results are usually cited as mg L À1 equivalent to BSA (35). Interestingly, grape juice protein levels were highly correlated with acidity (R > 0.86). A highly signi®cant relationship also existed between wine soluble solids content (10±25% w/w) and wine proteins (25±105 mg LÀ1 eq. BSA). For wines having a solids content of 18±20% the protein value was 40 mg LÀ1 (BSA eq.). Adsorption of wine proteins with

TABLE 7 Wine Protein Analysis Using the Bradford Method: Apparent Protein and Phenol Levels in Grapes, Grape Juice, and GewuÈ rztraminer Wine

Protein Phenola Sample (mg LÀ1) eq. BSA (mg LÀ1)

Grapes 102 + 1.8 98 + 4.2 Db À34% À44% Grape juice 77 + 1.5 298 + 14 D À25% À49% Wine 29.4 + 0.5 124 + 6.3 D À29% À57% a Phenol levels were assayed using Folin Ciocalteu assay. b D percent change after treatment with Amberlite XAD-4, PVP, and enzyme inhibitors. Source: Data from Ref. 33. Bradford AssayÐApplications 211 bentonite (0.48 g LÀ1) or other ®ning agents could be measured by the Bradford assay. The Bradford technique is not wholly resistant to interference by low- molecular-weight wine constituents. The absorption spectra for CBBG± tannic acid or CBBG-proanthocyanidin mixtures overlap with the spectra for the CBBG-protein complex (36). The interfering compounds were readily removed by Sephadex G25 column chromatography. Fractions from the column void volume (1 mL) were mixed with an equal volume of

Bradford reagent and A595 values were recorded. The standard protein was wine protein isolate. Over the range 10±100 mg (protein) LÀ1, the assay response was described by

DA595 0:002213 cP À 0:00941 2

(R 0.998). Maximum color formation occurred in 10 minutes. No further

DA595 increases were observed after 30, 50, or 80 minutes of incubation. Interference compounds from pinot noir and chardonnay wines were assessed for their effects on the Bradford protein assay (38).* Some wine samples were analyzed without sample pretreatment. Low-molecular-weight nonprotein substances were removed by ultra®ltration with a 10- or 3-kDa molecular size cutoff membrane. Substances possessing a molecular size of <10 kDa accounted for 28±68% of the Bradford assay response. In one instance, NPN accounted for 93% of the apparent protein. The A595 values also increased linearly with ethanol levels. This effect led to an overestimate of proteins by 30±50% at normal alcohol levels in wine. Apparently, ethanol induced changes in the ionization equilibrium for CBBG dye. The impact of individual interference substances was further analyzed using simulated wine samples containing BSA.{ Endogenous phenols produced a level of error equivalent to about 100% of protein. Tannins (10 mg LÀ1) and pectin (100 mg LÀ1) had comparatively little effect on protein analysis in simulated wines. Wine protein may be severely underestimated using the standard Bradford assay. Boyes et al. (39) examined protein levels in gewuÈ rztraminer

* With so-called direct wine protein analysis, 0.2 mL of commercial CBBG dye reagent concentrate (Bio-Rad Ltd.) was added to 0.4 mL of wine and 0.4 mL of distilled water. The A595 readings were taken after 60 minutes. Wine samples were diluted as necessary to keep A595 readings below one. BSA was used for calibration. { Model wine solutions contain potassium hydrogen tartrate (0.3% w/v), alcohol (12% v/v), lysine (0.1% w/w), phenylalanine (100 mg LÀ1), sulfur dioxide (200 mg LÀ1) and an unspeci®ed amount of wine vitamins. Proteins and tannins were added as necessary. Add 50 mL of NaOH (1 M) to 10±100 mL of sample to adjust to pH > 11. After 5 minutes of incubation add 3 mL of Bradford dye reagent. For the control study omit NaOH. 212 Chapter 7 wine and sauvignon blanc wine. Kiwi fruit juice and corn protein extract were also analyzed using a new ``alkali'' Bradford assay.* Fruit juice and gewuÈ rztraminer wine protein estimates were *10 times higher with the new method (Table 8). The alkali Bradford assay is similar to the procedure of Stoschek (9,10). In both instances 50 mL of NaOH (1.0 M) is added to the sample before the Bradford reagent. The effect of pH adjustment may be to reduce protein H-bonding interactions with phenolic compounds.

4.3. Cereal Products Protein from wheat ¯our, dough, bread crumb, or gluten was estimated with the Bradford assay by Eynard et al. (40). Samples were preextracted with a range of solvents before analysis. The calibration graph (with 50±350 mg gluten as standard protein) was curved and described by a second-order 2 equation, A595 ax bx c, where x is the amount of protein. As an alternative, plotting A595/A460 versus protein gave a linear graph. Protein- protein variations in sensitivity followed the series serum albumin > water- soluble protein * acid-soluble protein > gluten * gliadin. Assay sensitivity matched the total number of CBBG-reactive amino acids (Arg Lys His Tyr Trp Phe). Water-extractable materials from ¯our, dough, and bread crumb had protein contents of 19.4%, 22.3%, and 6.9 g%, respectively. By comparison, the acetic acid±extractable material from ¯our, bread dough, and bread crumb has 57.7% protein, 58.2% protein, and 8.3% protein. Baking led to a large decrease in the solubility of bread crumb protein.

TABLE 8 Effect of NaOH Addition on Bradford Assay of Fruit Juice and Wine Protein

Protein (mg LÀ1)

Bradford assay Sample ( NaOH) Bradford assay

Pressed kiwi fruit juice 2000 Ð Heat-treated kiwi fruit juice 560±830 150±410 As above bentonite treated 120 0 GewuÈ rztraminer wine 75±250 9.7±33 ( bentonite treatment) Sauvignon blanc wine <50 <7

Source: Summarized from Ref. 39. Bradford AssayÐApplications 213

Changes in the breadmaking quality of wheat due to drying at high temperature were examined by CBBG dye-binding assay (41). The dried grain was milled and extracted with phosphoric acid. The amount of soluble protein assayed by the Bradford assay was correlated with bread baking quality. Excessive heating during drying resulted in a decrease in the soluble protein. Esen (42) analyzed corn protein by solid-phase CBBG dye binding. The calibration graph was linear for up to 10 mg mLÀ1 protein. The lower limit of detection (LLD) for zein was 0.1 mg mLÀ1 or 1 mg per spot. Results were highly correlated with micro-Kjeldahl analysis. The quantity of zein from corn grain was negatively correlated with lysine levels (R À0.896 and À 0.946). The Bradford assay was suggested as an indirect method for screening for high-quality maize. Caution is required so that corn is not ranked of higher quality by virtue of a low protein content.

4.4. Legumes Dhillon and Nainawatee (43) found that whole or defatted mungbeen ¯our had 24% or 24.9% protein (per dry weight, DW) using the Bradford assay and Kjeldahl (N 6 6.25) analysis.* As a form of sample pretreatment, legume protein was ®rst extracted using Tris-SDS-2ME buffer (0.05 M Tris- HCl buffer, pH 7.0, with 10 mM 2-mecaptoethanol and 2% SDS). The calibration graph had a linear range of 0±100 mg BSA.{ In agreement with Rubin and Warren (13), assay sensitivity was reduced 10-fold due to the presence of SDS (<0.05%). The maturation process for beans (Phaseolus vulgaris L. cv. ¯or de mayo) was monitored by the Bradford, Kjeldahl, and Lowry protein assays (44). Results from micro-Kjeldahl analysis of bean ¯our and the dialyzed ¯our homogenate are shown in Fig. 3. Fig. 4 shows the changes in PBS- soluble bean protein fractions. Broad beans had an apparent crude protein content of 40% DW. Dialysis removed approximate 45% of the nitrogen, which is therefore low-molecular-weight NPN. Bean protein levels (15% DW) decreased slightly with maturation. Results in Fig. 4 are for samples freed from NPN by dialysis. There were inherent differences between results from different assay methods. Using casein as a standard, the Bradford, micro-Kjeldahl, Lowry, and biuret

* Shake 100 mg of ¯our with 1 mL of Tris-SDS-2ME for 60 minutes. Centrifuge (5000g) and collect the supernatant. Wash the pellet with 2 6 2 mL of Tris-SDS-2ME buffer and combine with the supernatant. Dilute (5 mL) pooled extract with water to a ®nal volume of 40 mL. Determined protein content by adding 0.1 mL of extract to 5 mL of standard Bradford reagent. { Dissolve BSA standards in 0.05 M Tris-HCl buffer containing 0.05% (w/v) SDS. 214 Chapter 7

FIGURE 3 Determination of proteins in bean (Phaseolus vulgaris L., cv. ¯or de mayo) seeds. Samples were analyzed by the micro-Kjeldahl method. Left to right: bean ¯our;homogenized and dialyzed bean ¯our extract; homogenized, centrifuged and dialyzed extract.

methods all gave comparable results. By contrast, protein levels in soy protein isolate were estimated as 84 (+0.2)% by micro-Kjeldahl analysis, 87.8 (+2.1)% by Lowry assay, 78.3 (+0.7)% by the Bradford method, or 106 (+1.1)% by the biuret method. The differences in assay results are unexplained. Paredes-Lopez and co-workers (44) suggest that the Lowry, Bradford, and biuret methods are unsuitable for legume protein determination. It is essential to remove NPN from ¯our samples (e.g., by washing with 80% ethanol) before protein analysis. Improving bean protein extraction with SDS may also increase the accuracy (43).

4.5. Potatoes Potatoes contain high amounts of NPN. The amount and quality of protein also differ for different potato varieties and states of maturation. Rapid methods for protein analysis are necessary to assist in attempts to breed potato varieties having higher protein content. Criteria for selecting Bradford AssayÐApplications 215

FIGURE 4 Determination of proteins in bean (Phaseolus vulgaris L., cv. ¯or de mayo) seeds. Effect of assay method on PBS-soluble protein (dialyzed).

convenient assays for potato breeding include (45) (a) ef®cient extraction and solubilization of protein from potato tissue, (b) compatibility with different potato varieties and tubers at different stages of maturation, (c) lack of interferences from NPN, and (d) low expense and high throughput compared with the Kjeldahl method. Snyder and Desborough (45) determined potato tuber protein using the Bradford assay.* Bradford test results for potato tubers were strongly correlated with Kjeldahl results (R 0.93). The six hybrid potato varieties studied had 3±15.6% protein per DW in agreement with quantitative amino acid analysis or micro-Kjeldahl results. The cost per analysis was 100 times lower and the sample throughput 6 times higher for the Bradford assay

* Dice, freeze, and lyophilize fresh potato. Suspend the ground potato powder (15 g) in 2.5 mL of water and then add 2.5 mL of 1 M sodium hydroxide. Incubate at room temperature for 2.5 hours. Remove 0.4 mL of extract to another test tube and add 5 mL of Bradford reagent.

Record A595 readings as usual. 216 Chapter 7 compared with Kjeldahl analysis. A more extensive sample pretreatment is required for the Kjeldahl method. Potato powder was preextracted with 80% ethanol to remove NPN before Kjeldahl analysis. None of the 20 common amino acids (2 mg mLÀ1) interfered with the Bradford assay.

4.6. Mushrooms A wide range of protein values have been reported for mushrooms. Weaver et al. (46) found 19±39% protein (DW)À1. Pecora (47) explained some of the variations in mushroom protein levels in terms of differences in developmental stage, growth conditions, harvesting time, varietal differences, and different amounts of NPN. Also, two Kjeldahl factors (4.38 and 6.25) are available for mushroom protein analysis. The literature describing mushroom (Agaricus bisporus) protein analysis is summarized by Braaksma and Schaap (48). These investigators reexamined protein levels in mushroom using Bradford, Kjeldahl, and quantitative amino acid analysis. Just under 50% of mushroom protein was extractable with phosphate- buffered saline. The remainder dissolved with 0.5 M NaOH, which is also an ideal one-step extractant for mushroom protein. Kjeldahl analysis of whole mushroom indicated a nitrogen content of 63 mg N gÀ1 DW or 4.66 mg N gÀ1 fresh weight (FW). The nominal protein content (%N 6 6.25) was 39% DW or 2.92% FW. Analysis of TCA-insoluble material using the Bradford assay or quantitative amino acid analysis showed 7% protein (DW) or a nitrogen content of 12.8 mg N gÀ1 DW. Apparently, NPN accounted for 80% of the total nitrogen in whole mushrooms. The water-soluble nitrogen fraction from mushroom comprised 20% protein, 60% urea-ammonia, and 20% free amino acids. By comparison, the nitrogenous material extracted by 0.5 M NaOH contained 37% protein, 9% urea-ammonia, 15% free amino acids, and 20% unde®ned constituents, probably glucosamine from the cell wall. Compared with quantitative amino acid analysis, the Bradford assay underestimated mushroom protein by 10±14%. It is possible that Bradford assay results are more accurate than those from quantitative amino acid analysis. Bradford assay results (6.9% protein per DW) were not affected by NPN. The pretreatment procedure for mushroom protein determination starts with powdered mushrooms. Fresh mushrooms were frozen in liquid nitrogen, freeze dried, and ground into a ®ne powder. Protein was extracted with phosphate-buffered saline (50 mM sodium phosphate buffer, pH 7.0 0.5 M NaCl) followed by 0.5 M NaOH. Kjeldahl analysis can be performed on samples precipitated by 15% TCA (46±48). Bradford AssayÐApplications 217

4.7. Honey Bogdanov (49) determined protein levels in honey using the Bradford,* biuret, and Kjeldahl techniques. The average protein content for 12 different honey samples was 130 mg 100 gÀ1 (range 50±185 mg 100 gÀ1). Kjeldahl results were 30% lower than protein estimates obtained with the Bradford assay. Biuret analysis gave protein estimates 2.5-fold greater than expected, probably because of Cu2 binding by honey polypeptides. g-Globulin (1±20 mg) was the standard protein for the Bradford assay.

4.8. Whey Protein Concentrates Richard and Paquin (50) compared the Bradford and Kjeldahl assays for whey protein concentrate (WPC). It was thought desirable to control both the ionic strength and pH of the solvent. Thus, 1 g of WPC (33.8% protein) was dissolved in 100 mL of McIlvine buffer adjusted to various pH values. The solutions were centrifuged (40,000g) and then ®ltered using Whatman No. 1 paper. Solubility results from the Bradford and Kjeldahl assays were 3% different (*83% solubility) at pH 6±9. At pH 3 protein solubility determined via the Bradford assay was 7% lower than the value from Kjeldahl analysis (78%). Compared with Kjeldahl analysis, the Bradford assay was a simpler, faster, and a more affordable technique for monitoring the solubility of WPC.

4.9. Insoluble Proteins Gotham et al. (51) assayed BSA, beta-lactoglobulin, egg albumin, or lysozyme aggregated by heating at 1008C. Protein aggregates were recovered by microcentrifugation and resolubilized by heating for 2±4 minutes with urea-mercaptoethanol solvent (8 M urea 5% w/v 2-ME). Dissolved protein samples (50 mL) were added to 2.5 mL of the standard Bradford reagent and

A595 values were recorded. Assay sensitivity, precision, and accuracy were comparable to values obtained with soluble proteins. Guanidine hydrochloride was not a suitable substitute for urea as a diluent. Resolubilization strategies should be checked for particular proteins before use. The approach has potential for monitoring food protein adsorption and fouling of process equipment.

* Dissolve 0.5 g of honey with 100 mL of water. To 0.8 mL of diluted honey add 0.2 mL of commercial CBBG dye regent concentrate (Bio-Rad Ltd.). Use 0.8 mL of water as a reagent blank. Record A595 readings after 5±60 minutes. 218 Chapter 7

REFERENCES

1. MM Bradford. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72:248±254, 1976. 2. JJ Sedmak, SE Grossberg. A rapid, sensitive and versatile assay for protein using Coomassie Brilliant Blue G250. Anal Biochem 79:544±552, 1977. 3. AH Reisner, P Nemes, C Bucholtz. The use of Coomassie Brilliant Blue G250 perchloric acid solution for staining in electrophoresis and isolectric focussing on polyacrylamide gels. Anal Biochem 64:509±516, 1975. 4. JC Bearden Jr. Quantitation of submicrogram quantities of protein by an improved protein dye binding assay. Biochim Biophys Acta 533:525±529, 1978. 5. SM Read, DH Northcote. Minimization of variation in the response to different proteins of the Coomassie Blue G dye-binding assay for protein. Anal Biochem 116:53±64, 1981. 6. J Pierce, CH Suelter. An evaluation of the Coomassie Brilliant Blue G250 dye binding method for quantitative protein determination. Anal Biochem 81:478± 480, 1977. 7. H Van Kley, SM Hale. Assay for protein by dye binding. Anal Biochem 81:485±487, 1977. 8. CM Wilson. Studies and critique of Amido Black 10B, Coomassie Blue R, and Fast Green FCF as stains for proteins after polyacrylamide gel electrophoresis. Anal Biochem 96:263±278, 1979. 9. CM Stoscheck. Increased uniformity in the response of the Coomassie Blue G protein assay of different proteins. Anal Biochem 184:111±116, 1990. 10. CM Stoscheck. Quantitation of protein. Methods Enzymol 182:50±68, 1990. 11. Z Zaman, RL Verwilghen. Quantitation of protein solubilized in sodium dodecyl sulfate mercaptoethanol-tris electrophoresis buffer. Anal Biochem 100:64±69, 1979. 12. GL Boccaccio, LA Quesada-Allue. Interference of sodium dodecyl sulfate in the Bradford assay for protein quantitation. An Assoc Quim Argent 77:79±88, 1989. 13. RW Rubin, RW Warren. Quantitation of microgram amounts of protein in SDS-mercaptoethanol tris electrophoresis sample buffer. Anal Biochem 83:773±777, 1977. 14. A Bensadoun, D Weinstein. Assay of proteins in the presence of interfering materials. Anal Biochem 70:241±250, 1976. 15. GL Peterson. A simpli®cation of the protein assay method of Lowry et al. which is more generally applicable. Anal Biochem 83:346±356, 1977. 16. Y-C Chang. Ef®cient precipitation and accurate quantitation of detergent solubilized membrane proteins. Anal Biochem 205:22±26, 1992. 17. LP Kirazov, LG Venkov, EP Kirazov. Comparison of the Lowry and Bradford protein assays as applied to protein estimation of membrane- containing fractions. Anal Biochem 208:44±48, 1993. Bradford AssayÐApplications 219

18. F Chiappelli, A Vasil, DF Haggerty. The protein concentration of crude cell and tissue extracts as estimated by the method of dye binding: comparison with the Lowry method. Anal Biochem 94:160±165, 1979. 19. SV Pande, SR Murthy. A modi®ed micro-Bradford procedure for elimination of interferences from sodium dodecylsulfate, other detergents and lipids. Anal Biochem 220:424±426, 1994. 20. V Neuoff, R Stamm, H Eibl. Clear background and highly sensitive protein staining with Coomassie Blue dyes in polyacrylamide gels: a systematic analysis. Electrophoresis 6:427±448, 1985. 21. HJ Chial, AG Splittgerber. A comparison of the binding of Coomassie Brilliant Blue to proteins at low and neutral pH. Anal Biochem 213:362±369, 1993. 22. T Spector. Re®nement of the Coomassie Blue method of protein quantitation. Anal Biochem. 86:142±146, 1978. 23. HG Pollard, R Menard, HA Brandt, CJ Pazoles, CE Cruetz, A Ramu. Application of Bradford's protein assay to adrenal gland subcellular fractions. Anal Biochem 86:761±763, 1978. 24. B-M LoÈ f¯er, H Kunze. Re®nement of the Coomassie Brilliant Blue G assay for quantitative protein determination. Anal Biochem 177:100±102, 1989. 25. MJ Lewis, SC Krumland, DJ Muhleman. Dye-binding method for measurement of protein in wort and beer. J Am Soc Brew Chem 38:37±41, 1980. 26. V Hii, WC Herwig. Determination of high molecular weight proteins in beer using Coomassie Blue. J Am Soc Brew Chem 40:46±50, 1982. 27. EJ Knudson. Co-ordination of new and alternate methods of analysis. J Am Soc Brew Chem 45:99±100, 1987. 28. CJ Dale TW Young. Rapid methods for determining the high molecular weight polypeptide components of beer. J Inst Brew 93:465±467, 1987. 29. American Society of Brewing Chemists. High molecular weight protein in beer (Bradford method). J Am Soc Brew Chem 46:116±118, 1988. 30. Y Kano, M Kamimura. Simple methods for determination of the molecular weight distribution of beer proteins and their application to foam and haze studies. J Am Soc Brew Chem 51:21±28, 1993. 31. A Onishi, MO Proudlove. Isolation of beer foam polypeptides by hydrophobic interaction chromatography and their partial characterization. J Sci Food Agric 65:233±240, 1994. 32. KM Williams, P Fox, T Marshall. A comparison of protein assays for the determination of the protein concentration of beer. J Inst Brew 101:365±369, 1995. 33. JC Hsu, DA Heatherbell. Isolation and characterization of soluble proteins in grapes, grape juice and wine. Am J Enol Viticult 38:6±10, 1987. 34. JM Murphy, JR Powers, SE Spayd. Estimation of soluble protein concentration of white wines using Coomassie Brilliant Blue G-250. Am J Enol Viticult 40:189±193, 1989. 220 Chapter 7

35. JM Murphey, SE Spayd, JR Powers. Effect of grape maturation on soluble protein characteristics of Gerwurztraminer and White Riesling juice and wine. Am J Enol Viticult 40:199±207, 1989. 36. O Brenna, S de Vecchi. Evaluation of protein and phenolic content in must and wine. I. Assay of soluble proteins. Ital J Food Sci 2:269±273, 1990. 37. EJ Waters, W Wallace, PJ Williams. Heat haze characteristics of fractionated wine proteins. Am J Enol Viticult 42:123±127, 1991. 38. R Marchal, V Seguin, A Maujean. Quanti®cation of interferences in the direct measurement of proteins in wines from the Champagne region using the Bradford method. Am J Enol Viticult 48:303±309, 1997. 39. S Boyes, P Struebi, H Dawes. Measurement of protein content in fruit juices, wine and plant extracts in the presence of endogenous organic compounds. Lebensm Wiss Technol 30:778±785, 1997. 40. L Eynard, N Guerrieri, P Cerletti. Determination of wheat proteins in solution by dye binding in ¯our, dough, and bread crumb. Cereal Chem 71:434±438, 1994. 41. EA Tosi, ED Re, L Carbone, M Cuniberti. Breadmaking quality estimation by fast spectrophotometric method. Cereal Chem 77:699±701, 2000. 42. A Esen. Estimation of protein quality and quantity in corn (Zea mays L.) by assaying protein in two solubility fractions. J Agric Food Chem 28:529±532, 1980. 43. S Dhillon, HS Nainawatee. A rapid and sensitive method of protein estimation in legume grains. Int J Trop Agric 5:231±234, 1987. 44. O Paredes-Lopez, F Guevara-Lara, ML Schevenin-Pinedo, R Montes-Rivera. Comparison of procedures to determine protein content of developing bean seeds (Phaseolus vulgaris). Plant Foods Hum Nutr 39:137±148, 1989. 45. JC Snyder, SL Desborough. Rapid estimation of potato tuber total protein content with Coomassie Brilliant Blue G-250. Theor Appl Genet 52:135±139, 1978. 46. JC Weaver, M Kroger, LR Kneebone. Comparative protein studies (Kjeldahl, dye binding, amino acid analysis) of nine strains of Agaricus bisporus (Lange) Imbach mushrooms. J Food Sci 42:364±366, 1977. 47. RP Pecora. Determination of protein in edible mushroom (Boletus sp.). Int J Food Sci Technol 24:207±210, 1989. 48. A Braaksma, DJ Schaap. Protein analysis of the common mushroom Agaricus bisporus. Postharvest Biol Technol 7:119±127, 1996. 49. S Bogdanov. Determination of honey protein with Coomassie Brilliant Blue G. Mitt Geb Lebensm Hyg 72:411±417, 1981. 50. JP Richard, P Paquin. Use of a dye binding method for the determination of the protein content of dairy products. Milchwissenschaft 45:92±94, 1990. 51. SM Gotham, PJ Fryer, WR Perterson. The measurement of insoluble proteins using a modi®ed Bradford assay. Anal Biochem 173:353±358, 1988. 8

Immunological Assay: General Principles and the Agar Diffusion Assay

1. INTRODUCTION 1.1. Protein Adulteration, Authenticity, and Speciation Adulteration is any undeclared substitution or addition designed to enhance the economic value of a food product (1). The more general features of adulteration include (a) abstraction or omission of valuable constituents, (b) substitution by undeclared (usually cheaper or less safe) components, and (c) concealment of intrinsic low quality or damage (Fig. 1). The seriousness of each infringement is related to the position of a commodity in the food system, the origin of the foodstuff, the economic impact of the practice, consumer awareness of the problem, and safety and health considerations. Certain forms of adulteration may be unintentional but no less negligent. For instance, cross-contamination arises when poorly cleaned equipment is used for processing meats from two different species. Plant proteins can be legitimately added to meat products as extenders. They introduce new functional properties such as binding meat pieces, fat, and water. However, undeclared substitution of one protein by another (meat by nonmeat proteins) is usually undesirable. This and the following three chapters describe immunochemical tests for food proteins. The source of a protein affects its aesthetic appeal and economic value.

221 222 Chapter 8

FIGURE 1 Forms of food protein adulteration.

Protein speciation is a subject of legal, religious, and public health interest. Groups concerned with protein authenticity include consumer organiza- tions, the food industry (ingredient suppliers, food manufacturers), public analysts, enforcement agencies, government of®cials, and the legal profes- sion. Such interest groups work together to identify and advise on areas of food adulteration and encourage the use of new or existing methods of authentication. Protein adulteration is also associated with poor labeling. An accurately labeled food product is always authentic (2±5).

1.2. Implications of Adulteration Protein adulteration has health and safety consequences. Adulteration may expose sensitive individuals to allergens. To destroy meat-borne bacteria and viruses (foot and mouth, Newcastle disease, African swine fever, etc.), the United States Department of AgricultureÐFood Safety Inspection Services (USDA-FSIS) speci®cies minimum processing end-point temperatures (EPTs) for a range of imported meats. Processing at the EPT for beef Immunological Assay: General Principles 223 will lead to health risks if a product contains undeclared chicken or pork. The USDA-FSIS also periodically lists voluntary recall notices posted by companies due to accidental inclusion of ``undeclared allergens'' in various processed food products. In Europe, the use of meat and bone meal as feeds is subject to legislation. Religious dietary regimes (kosher, vegetarian, and halal practices) require accurate labeling information. Certain protein sources are unac- ceptable to people of the Jewish, Hindu, and Islamic religions. The geographic origin of meat can also be a signi®cant issue. Wildlife conservation requires the differentiation of farmed and free-living species (wildlife game, antelopes, buffalo, cattle venison). Species substitution may involve different classes of meat (poultry or beef), closely related meat (horse or kangaroo for beef or veal, chicken for turkey), or inadvertent cross-contamination (beef with pork or kangaroo). Adulteration undermines commerce, which is based on a rational relationship between pricing and quality. Undeclared substitution leads to purchasing decisions based on inaccurate product information. The practice erodes con®dence. The likelihood of meat adulteration has increased. According to Patterson and Jones (6), improvements in meat technology have made it easier to transport and utilize cheap undeclared meats that are dif®cult to identify when meat has been deboned and removed from the carcass. These products present few visual clues to their origin. Large frozen blocks and ¯aked or comminuted (minced) batches of meat are virtually impossible to identify by eye. The economic imperative can lead to the use of cheaper meats for certain processed foods. Blood (protein) may be added to mince meat to make it appear leaner. Mechanically recovered meat (MRM) from chicken may be used in place of beef or pork. High demand for poultry in India can lead to the adulteration of chicken meat products with beef and other meat types. Plant proteins may be added to processed meat products to increase their bulk. Readily available cow's milk can be added to ewe's milk for the manufacture of some specialty cheeses. The economic impact of adulteration may be estimated from a notional Protein Adulteration Price Index (PAPI) de®ned as

PAPI frequency of substitutions6substitution amount % 1 6price dividend

The potential ill gain from adulteration is partly related to (a) the frequency of species substitution, (b) the percent weight of adulterant added, and (c) the price dividend from substitution. Multiplying PAPI by the net value of 224 Chapter 8 trade gives the ®nancial cost to the consumer arising from undeclared substitutions. PAPI can be estimated only crudely for several reasons. First, information about the pattern of protein substitution worldwide is scarce. Wide variations will occur in the frequency and extent of undeclared substitutions in different countries and states and regions within any given country. Next, the incentive for adulteration is probably related to the net value of trade in that particular commodity. Most protein commodities could be subject to adulteration but higher value animal proteins are more likely targets. The estimated rate of undeclared substitutions for meat is about 16% in the state of Florida (7). For either raw or cooked meat, the weight of adulterant added was 1±5% (average 2.5%). The price dividend for the added meat can be estimated from the formula 1 À (Y/X), where Y is the real price of the adulterant and X is the advertised price. When the added protein has no intrinsic value, the price dividend is 1.0. Assuming a price dividend of 0.5, the PAPI estimate for the U.S. meat and poultry market is then 0.2%. The upper PAPI can be as high as 5%. Thus, the frequency of substitution is easily 50% in some sectors of the meat industry. A percent weight substitution of 10% also seems credible. For the worst-case scenario, we assume a maximum price dividend of 1, leading to PAPI value of 5%. The value of the U.S. meat and poultry trade is about US$120 billion according to FSIS ®gures from 1995. Consumer spending on meat and poultry products accounts for one third of the annual food budget. Meat purchases included raw beef, pork, lamb, chicken, turkey, and approximately 250,000 processed meat products, which are processed foods containing >2% poultry or >3% meat. Examples include ham, sausages, sources, soups, stews, pizzas, and frozen dinners. From the preceding PAPI and net trade values, we may suppose that the U.S. meat consumer is overcharged by between US$250 million and US$6 billion annually. The most ef®cient techniques for detecting protein adulteration are immunological. Other well-established approaches include quantitative sodium dodecyl sulfate polyacrylamide gel electrophoresis (QSDS-PAGE), isoelectric focusing (IEF), and capillary electrophoresis (8). High-pressure liquid chromatography (9) and fast protein liquid chromatography (10) have also been employed for the differentiation of milk from different species. Finally, mention must be made of non±protein-based methods. Detection of amino acids, sugars, or fats associated with particular species can be a clue to adulteration. Species-speci®c DNA testing using hybridization probes or polymerase chain reaction (PCR) is also increasing. The interested reader is referred to papers by Hunt et al. (11), Fairbrother et al. (12), Lenstra and Buntjer (13), and Wolf et al. (14);the subject is reviewed by Meyer and Candrian (15). Immunological Assay: General Principles 225

FIGURE 2 Methods for food protein immunoassay are divided into marker-free and marker-linked methods.

2. IMMUNOLOGICAL METHODS

Immunoassays are based on the use of antibodies (Fig. 2). The range of techniques includes marker-free techniques in solution or agar, e.g., solution-phase precipitation, agar gel immunodiffusion, immunoelectrophoresis, and counterelectrophoresis. The marker-linked techniques involve enzyme immunoassay (EIA) or radioimmunoassay (RIA). Application of immunoassays in food analysis and authenticity testing are reviewed by Samarajeewa et al. (16), Gazzaz et al. (17), Allen and Smith (18), Barai et al. (19), Lee and Morgan (20), Hernandez et al. (21), Taylor et al. (22), Smith (23), and Mandokhot and Kotwal (24).

2.1. Methods of Food Protein Immunoassay * With marker-free immunoassays, antibody-antigen binding is monitored directly. Excess antibody is added to a sample suspected to contain antigen.

* The term ``antiserum,'' although established in the literature, can lead to confusion. Descriptions such as rabbit antiserum for horse serum are not helpful. We will use antibody wherever possible. 226 Chapter 8

Antibody-antigen binding leads to a precipitate whose height and/or volume is proportional to the antigen concentration. Antibody-antigen binding can take place within an agar gel matrix (Section 3). Marker-free immunoassays are easy to perform and give qualitative (yes-no) answers in a relatively short time. They are, however, relatively insensitive. Marker-linked immunoassays provide more sensitive, although more expensive, analysis.

2.2. Principles of Immunoassay Antibody-antigen binding involves a reversible equilibrium reaction,

Ig P B 2 where, Ig is the free antibody concentration and P the free concentration of antigen. From Eq. (2), the association constant (Ka) for antibody-antigen binding is B B Ka 3 IgP IgT À BP where IgT is the total concentration of antibody and B the concentration of bound antibody. From this, one can readily write down the Scatchard equation: B B IgT À 4 PKa

A graph of [B] plotted versus [B]/[P] yields a straight line with a slope of 1/Ka and the intercept IgT. Dividing the value of the intercept by the known molar concentration of antibody gives the number of antigen binding sites per Ig molecule;ordinarily, the answer should be 2. The Scatchard plot is usually nonlinear, showing the presence of different populations of antibodies. Adsorbing Ig to a solid support will also obscure some antigen binding sites, leading to a range of Ig-antigen binding af®nity. We turn to the effect of antibody-antigen binding parameters on the immunoassay characteristics. The various terms in Eq. (3) can be expressed in terms of the fraction of antigen bound, X: X Ka 5 IgT À PX 1 À X where X B/PT and the term PT is total antigen concentration. Rearranging Immunological Assay: General Principles 227 this relation gives the quadratic relation

aX2 bX C 0 6

with a 1, b À[PT IgT À 1/Ka]/PT, and c IgT/PT. Theoretical X vs. PT graphs showing the fraction of antigen bound versus concentration were simulated from known values for Ka,IgT, and PT (Figs 3 and 4). The ®rst graph shows a binding curve for an immunoassay using two À8 À9 antibody concentrations (IgT)of10 and 10 M. The value of Ka is constant (1011 MÀ1). Fig. 3 shows that the analytically useful binding interactions occur over the antigen concentration where X 1. The concentration of antigen should not exceed the total concentration of antibody binding sites, otherwise the fraction bound decreases below 1. The linear dynamic range depends on the concentration of antibody. Simula- tions were also performed in which the concentration of antibody was kept 10 8 6 À1 constant while changing Ka from 10 to 10 and 10 M . Fig. 4 shows that decreasing antibody-antigen binding strength decreases the sensitivity (slope) of the calibration graph. We expect that the LLD for antigen will increase with decreasing Ig-antigen binding af®nity.

FIGURE 3 Simulated binding curves for food protein immunoassay. A plot of the fraction of antigen bound versus the total antigen concentration. 228 Chapter 8

FIGURE 4 Simulated binding curves for food protein immunoassay. Effect of

changing antigen binding af®nity constant. The Ka is given values of 1010,108,or106 MÀ1.

2.3. Background Immunology Antibodies are secreted by vertebrates in response to foreign matter or antigen. An antibody binds strongly to the antigen that triggered its formation. The antibodies are immunoglobins (Igs), which are proteins that dissolve in dilute salt solution and function in connection with immunity. The Igs are produced by blood cells called B lymphocytes. Each organism has at least 100 million different B lymphocytes. Each B lymphocyte has a distinct Ig protruding from its surface. One of the 100 million B lymphocytes is likely to recognize any antigen encountered in the environment. Binding of antigen to a cell-bound Ig leads to a proliferation of the recognizing B lymphocyte within the general population (an idea called the clonal selection theory). Antigen binding also triggers B-lymphocyte transformation into two new cell types. One cell type responsible for Ig production is called the plasma cell. The second cell type, B memory cells, primes the body to remember its encounter with the antigen. To induce Ig production, an antigen must possess a molecular weight >5000. Low-molecular-weight Immunological Assay: General Principles 229 compounds (haptens) act as antigens when covalently attached to carrier proteins. Large antigens possess multiple sites (epitopes) for Ig binding and are said to be polyvalent. Each epitope elicits Ig production by a speci®c B lymphocyte. The heterogeneous mixture of antibody produced by a population of B lymphocytes is described as a polyclonal antibody (pAb). A sample of pAb contains a mixture of Ig speci®c for the different epitopes on a single polyvalent antigen. Two proteins sharing a common epitope, perhaps because of sequence or structural homology, will be recognized by pAb, leading to cross-reactivity. The diversity of antibodies arises from their structure and large numbers of encoding genes. Antibodies are glycoproteins with a molecular mass >150 kDa. Each antibody comprises two heavy (55±75 kDa) polypeptide chains and two light (25 kDa) chains joined by (three) intermolecular disul®de bonds. Each heavy (H) or light (L) chain has variable (v) amino acid sequences at the C terminal and constant (c) regions near the N terminal. There are ®ve different H chains (g, a, m, d, and e) giving rise to ®ve classes of antibody; IgG, IgA, IgM, IgD, and IgE. The different classes of Ig occur at plasma concentrations of about 12, 3, 1, 0.1, and 0.001 mg mLÀ1 and differ in their stability and ease of handling. There are two groups of L chains designated k and l. The three-dimensional structures of the H chain comprises ®ve domains whereas each L chain is organized into two domains. The Hc and Lc regions are produced by two gene alleles and Lv and Hv regions are each encoded by about 300 genes. In all, an organism is able to produce at least 3002 or 9 6 104 types of antibody. Further diversity comes from a process of somatic recombination and the translation of messenger RNA (mRNA) transcripts using different reading frames. As a result of such mechanisms, vertebrates produce up to 1 6 108 different antibodies and their corresponding B lymphocytes before exposure to an antigen. The pAbs for commercial immunoassays are produced through the agency of live animals. B lymphocytes have not been successfully cultured in vitro. Such restrictions make the production of pAbs dif®cult to standardize. Immunizing a second animal must lead to variations in the crude reagent, which must then be puri®ed and standardized in some way. The technology for producing monoclonal antibodies (mAbs) was developed by Kohler and Milstein (25) of the University of Cambridge, U.K. Tumor cell lines, which survive inde®nitely when grown in cell culture, produce mAbs naturally. To produce mAbs of de®ned speci®city, spleen cells from immunized mice are fused with tumor cells. The resulting spleen cell±myeloma hybrid (hybridoma) possesses traits from both parental cell lines, i.e., immortality and the ability to synthesis speci®c mAbs derived from mice spleen cells. The hybridoma can be grown in PVC microwell 230 Chapter 8 plates and the cell supernatant examined for the mAb. It was possible to isolate a single hybridoma cell line (or clone) that produces the required Ig. Hybridomas can be grown inde®nitely and large amounts of mAbs produced for analytical use. Only a few of the available genes for the H and L regions are expressed owing to a process called allelic exclusion (25). Fusion is needed for the production of mAbs. The mere mixing of myeloma cells and spleen cells (polyethylene glycol as a fusion agent) will not lead to new Ig.

3. SPECIATION OF PROTEINS BY AGAR GEL DOUBLE IMMUNODIFFUSION ASSAY

Modern agar gel double immunodiffusion (AGID) assays were developed by Cutrufelli et al. (26,27). The technique is approved by the AOAC. The AGID assay is fast, requires minimal equipment, and can be performed by personnel with minimum training. The tests are based on the two-stage precipitin reaction. AGID comes in several forms (28): (a) linear diffusion, (b) double radial diffusion, (c) single radial immunodiffusion (SRID), (d) immunoelectrophoresis, and (e) counterimmunoelectrophoresis. With linear diffusion, antibody is added to molten agar and the mixture is allowed to set in a small tube. The sample antigen is placed on the column of gel and diffuses downward, forming an opaque precipitin phase. The distance of migration is a function of antigen concentration. For an SRID assay the antigen is placed in a circular well cut into the antibody- agar plate (29). The radius of diffusion, indicated by a precipitin phase, is related to the concentration of antigen. Double immunodiffusion assays involve two sets of circular wells cut into an agar gel. Into separate wells are placed antibody and analyte. They diffuse toward each other, combining to form an opaque precipitation band. With immunoelectrophoresis, the protein sample is placed within holes cut into the gel and forced to migrate under an electrical potential. Then antibody is placed along a rectangular trough parallel to the direction of electrophoresis. Each protein band reacts with the diffusing antibody, leading to a series of precipitin arcs (30).

3.1. Procedures for Meat Protein Identi®cation by AGID Assay AGID assays are an extension of the test tube precipitation test ®rst demonstrated by Uhlenauth in 1901 (cited in Refs. 5 and 31). Further work between 1902 and 1928 led to the production of less complex antigen (inoculation) mixtures designed to improve speci®city and reduce toxicity to Immunological Assay: General Principles 231 immunized animals. Gel-phase immunodiffusion assays were described by Oudin around 1946 and then by Ouchterlony (32). The last study led to a simple in vitro assay for diphtheria toxin and for toxin-producing bacteria. Heating diphtheria toxin led to its denaturation and failure to produce a precipitin reaction. Early serological tests for meat and other food proteins are described by Oswarld (33). To produce pAb for analytical applications, a host animal is immunized with antigen at intervals ranging from several weeks to months. Mice, rats, rabbits, sheep, goats, or horses are used for antibody production. Larger animals are preferred by large-scale, commercial, producers. After about 4 weeks, blood is taken from the immunized animal and centrifuged to produce antiserum, i.e., blood serum containing antibody. Further booster injections may be administered and antibody production monitored until a high plasma concentration (antibody titer) is attained. To avoid cross-reactivity, pAb may be puri®ed by immunoadsorption. When the titer of antibodies reaches a high level, larger samples of blood are removed, allowed to coagulate, and centrifuged to produce crude antibody. Preservative is normally added before storing at À 208C. Important variables for antibody production include the following:

1. Purity of the antigen. Puri®ed antigen is seldom required. However, more pure antigens have less toxic effects on the immunized animal. 2. Choice of host animal. A large animal such as a goat yields more serum than a mouse. To reduce cross-reactivity, pAb should be raised in a species similar to the species for which the antibody is intended to differentiate. Goats or sheep pAb is more able to differentiate between antigens from cattle, buffalo, and other bovine species (59). 3. Choice of adjuvant. Most investigators use Freund's complete adjuvant. This is an oil suspension of inactivated Mycobacterium tuberculosis. A muramyl peptide from the bacterial wall stimulates antibody production. Alternatively, antigen is coprecipitated or adsorbed on insoluble aluminum oxide, synthetic muramyl peptides, or carbohydrate (34). 4. Mode of immunization. Subcutaneous or intramuscular injection is common. Several investigators describe injections via the foot pads of rabbits. Intervals of injections are usually weekly. Larger animals may be injected at 30-day intervals. 5. Time and method for bleeding. Immunization schedules usually require trial bleedings at 4- to 7-day intervals after the injection of 232 Chapter 8

booster antigen. Blood may be collected from a vein in the ears or tail. 6. Puri®cation by immunoadsorption. To remove cross-reactivity with sheep, goat pAb may be exposed to sheep antigen and then centrifuged to remove any precipitate formed. The soluble pAb phase should not now react with sheep. Immunoabsorption can be performed with soluble sheep antigen or antigen immobilized on cyanogen bromide±activated support (35±39). AGID tests are performed according to the following steps: (a) extraction of antigen and immunization of host animal (see Method 1), (b) production of pAb (Method 1), (c) puri®cation of pAb by immunoadsorption (see Method 2), (d) preparation of agar gel plates (see Method 3), and (e) conduction of AGID test. Details of these steps are given next. The protocols are drawn from reference listed in Table 2 (Sec. 3.4). The inexperienced worker should seek expert advice regarding humane treatment and immunization of host animals. Commercial pAb should be used when available. The internet is a good source of addresses of companies currently supplying immunoassay reagents and ready-to-use kits. Method 1 Extraction of meat antigens and pAb production (40,59). Reagents 1. Phosphate-buffered saline (PBS;0.01 M phosphate buffer, pH 6.8±7.2 0.1 M KCl) 2. Muslin cloth 3. Freund's adjuvant Procedure Fresh meat antigen. Stir ground meat with 1 volume of PBS overnight at 48C. Filter through muslin cloth and centrifuge at 10,400g for 60 minutes. Filter the supernatant through Whatman No. 3 paper to remove ¯oating fat droplets. Freeze dry for prolonged storage. Immunization. Mix the antigen (5±10 mg mLÀ1 PBS) with an equal volume of Freund's complete adjuvant. Inject adult rabbits with 2 mL of antigen suspension at multiple sites. Reinject the animals with booster doses of antigen dissolved in Fruend's incomplete adjuvant at weekly intervals. With goats the ®rst injection can be 4 mL of antigen. The booster dose is injected every other week. Small amounts of blood should be removed in the intervening weeks and tested for pAb production. Treatment of blood and antibody production. Collect whole blood, allow to coagulate at room temperature for about 3 hours, and then Immunological Assay: General Principles 233

store at refrigerator temperatures overnight. Centrifuge at 2000g to remove blood cells. To the supernatant (antibody) add 100 ppm of methiolate as a preservative and store at À208C until needed. Immunization schedules are ¯exible, ranging from 3±4 weeks for rabbits to several months for goats, sheep, or horses. The pAbs for ORBIT and PROFIT were produced by injecting goats with 10 mL of antigen coprecipitated with alum. The same antigen was injected 21±30 days later (5 mL per hind leg), then at 3-weekly intervals. Blood samples were taken *7 days after each injection to monitor pAb production. The immunization process is continued until a desired concentration of antibody is attained. Method 2 Antibody puri®cation by immunoadsorption. 1. A simple immunoadsorption procedure (31,35). To 1 mL of crude pAb for species A that shows cross-reactivity with species Y, add 20 mg of antigen from species Y. Incubate the mixture at room temperature for 4 hours. Refrigerate at 48C overnight and centrifuge (2500g for 20 minutes) to remove pAb-antigen complexes. The supernatant phase should be pAb with improved speci®city for species A. 2. Immunoadsorption using ethylchrolorformate cross-linked antigen. The original method was described by Avrameas and Ternynck (36,37). Adjust a solution of antigen (e.g., 500 mg of whole serum protein in 10 mL of 0.2 M acetate buffer) to pH 5.0 with dilute HCl. Add ethylchloroformate (*3 mL) dropwise. Maintain the acidity at pH 4.5±5 using 1 M NaOH. Allow to react for about 10± 15 minutes, redisperse the resulting gel in a further 10 mL of 0.2 M acetate buffer, and allow to react for another 60 minutes. Recover the gel formed by ®ltration or by centrifuging gently. Wash with distilled water and keep refrigerated until use. Shake antibody for species A and with cross-linked antigen from species Y overnight. Recover the soluble serum phase, which should now be monospeci®c for species A. This method has been demonstrated by Kang'ethe et al. (59). Immunosorption can be performed with the dried antigen, cross-linked antigen, or antigen immobilized on CNBr-activated support. Method 3 Standard agar gel double immunodiffusion assay. Reagents 1. Agar (Oxoid Ltd.) 234 Chapter 8

2. PBS buffer 3. 6.5-mm-diameter ®lter discsÐoptional

Procedure Prepare 1% (w/v) molten agar in PBS by heating on a water bath. Pour about 4 mL of molten agar into a petri dish and allow to set at room temperature. Refrigerating for a few hours will harden the gel. Cut out several 6-mm-diameter holes or wells using a cork borer. Place one well in the center of the petri dish and then surround this with a quartet of wells each placed 6 mm from the central well. Remove the circular piece of gel using a Pasteur pipette attached to a vacuum line. Place a drop of molten agar into each well to act as a seal, thereby avoiding the migration of samples below the well. To perform a standard AGID assay, place pAb in the central well. In the surrounding wells, place one each of the test samples (*20 mL). Cover the petri dishes to avoid dehydration and allow to stand at room temperature for 24 hours.

3.2. Agar Gel Double Immunodiffusion Assay for Uncooked Meat The ®rst attempts to identify raw meat using the AGID assays were described by Warnecke and Saf¯e (38). They found that actomyosin was a poor antigen, with injections of 20±150 mg leading to only moderate pAb production in rabbits. Rabbit pAB for beef whole serum showed cross- reactivity for lamb extract. No reactivity was seen for horse meat or pork. The crude pAb was rendered monospeci®c for beef by immunoadsorption. An AGID assay with immunoadsorbed pAb allowed the detection of beef. Many of the methods described in this report are still in use at the present time. Further developments in AGID assays for meat speciation occurred in the laboratories of the U.S. Department of Agriculture, Beltsville, MD. Fugate and Penn (39) used AGID assays to identify meat from beef, horse, pig, and sheep. Of 12 meat samples examined, 11 were correctly identi®ed. They recommended that the AGID test should be subjected to collaborative testing. AGID assays using pAb for residual serum albumin from meat were produced by Hayden (40). Hers was probably the ®rst dissertation on this subject. At the Department of Food Science, University of Georgia (Athens, GA), Helm et al. (41) compared the AGID assay and a simple test tube precipitin test using (rabbit) pAb for beef, horse, lamb, and pork. The AGID test detected adulteration at the 2% level.The solution precipitin test was *three times faster and four times more sensitive. In Australia, Swart Immunological Assay: General Principles 235 and Wilks (42) differentiated beef, horse, kangaroo, and mutton by AGID assay. Species identi®cation ®eld tests (SIFTs) were developed in 1984 by Mageau et al. (43) at the USDA. Finally, Darwish et al. (44) also described an AGID assay for detecting beef adulteration with camel and pork. Some examples of the application of immunodiffusion assays include the detection of pork in beef mince meat (45), Alaska pollack surimi analysis in meat products (46), and the detection of various meat types (beef, pork, horse, poultry) in hamburger (47).

3.3. Species Identi®cation Field Tests (SIFTs) for Uncooked Meat SIFTs are designed for ®eld testing in abattoirs and meat inspection stations. The ®rst of these test kits is called ORBIT (43). The acronym stands for the Overnight Rapid Bovine Identi®cation Test. ORBIT was followed by PROFIT (Poultry Rapid Overnight Field Identi®cation Test) (48), SOFT (Serological Ovine Field Test) (49), PRIME (Porcine Rapid Identi®cation Method) (50), REST (Rapid Equine Serological Test) (51), DRIFT (Deer Rapid Identi®cation Field Test) (52), and MULTI-SIFT (Multispecies Identi®cation Field Test) (53). The last kit can simultaneously test for beef, poultry, pork, sheep, horse, and deer meat. After successful collaborative trials (26), ORBIT and PROFIT received approval from the AOAC. These methods can detect meat adulteration levels of about 10% or greater (27). SIFTs are not radically different from classical AGID assays. However, considerable design effort has gone into making SIFTs attractive and easy to use. Antibody and meat sample extract are adsorbed onto ®lter paper discs. Both are then freeze dried, providing stabilized reagent discs that can be stored for up to 12 months at refrigeration temperature. The antibody ®lter discs and all materials needed to perform SIFTs are available commercially. The tests are used by the USDA and meat inspection services in the United States. According to the team responsible for SIFTs, conventional methods for meat speciation (e.g., electrophoresis, chromatography) share many of the following disadvantages: (a) tests are usually performed within a formal laboratory, (b) relatively sophisticated equipment is needed, (c) high levels of staff expertise and training are necessary, (d) time delays arise due to off- site testing with an attendant need for transmitting samples and assay results to and from the analyst, and (e) relatively labile reagents are used. SIFTs were developed with the aim of avoiding such disadvantages. The assays are highly reliable, fast, easy to use, accurate, and sensitive. Little expertise or previous experience is needed for successful testing. Finally, SIFTs use 236 Chapter 8 stabilized reagents (antibody and gel plates) with a shelf life of up to 12 months. SIFTs use pAb speci®c for native serum albumin (40). The tests are therefore intended for raw meat speciation. A key feature is the use of lyophilized reagents that are more stable for storage. Commercial pAbs as well as those prepared in house proved usable. Assay results were affected by the distance of separation between ®lter discs, with a distance of 4±5 mm being optimal. The tests showed considerable temperature tolerance. Incubating agar gels plates at 25±378C had no ill effect. Higher temperatures led to gel dehydration. Lower temperatures prolonged the time needed for precipitin formation. Finally, SIFTs were suitable for the analysis of whole meats, ground meat, meat emulsions, and other formulated meat products, provided these were not heated. Table 1 summarizes some performance characteristics of SIFT assays. The MULT-SIFT assay allows simultaneous testing of an unknown meat sample (antigen) against six different reference pAbs (Table 1).

3.4. Thermostable Meat Antigens for Cooked Meat Analysis Ideal antigens for cooked meat analysis survive thermal treatment at 708C or autoclaving temperatures of 1208C for 15 minutes. Hayden (54) considered troponin as a heat-stable meat antigen (Table 2). Extensive research since 1977 has shown that thermostable meat antigens are usually troponin C, troponin I, or troponin T. Schweiger et al. (55) used puri®ed turkey troponin T for AGID assay. Thermostable antigens were initially developed from adrenal gland extracts (56,57) by Milgrom and Witebsky (58).* Rabbit pAb for boiling-resistant, ethanol-insoluble (BE) antigen from beef showed cross-reactivity with BE antigen from sheep. There was no reaction with heated adrenal extracts from pig, rat, guinea pigs, or humans. The method for BE antigen preparation has remained unchanged for nearly 40 years. Hayden (56,57) used BE antigen for detecting cooked beef sausage adulteration with horse, sheep, and pig meat. Radhakrishna et al. (62) showed that (buffalo, goat, oxen) muscle BE antigen had SDS-PAGE bands corresponding to troponin (C, I, and T) and tropomyosin. Bhilegaonkar et al. (67) found that the concentration of (buffalo, sheep, goat, pig) BE antigen was 21±130% higher in adrenal tissue

* Boiling-resistant ethanol-insoluble (BE) antigen is produced by boiling an aqueous extract from bovine adrenal glands or muscle tissue followed by centrifugation. The supernatant is autoclaved at 1208C for 30 minutes and centrifuged. Then three volumes of 95% ethanol are added to the soluble fraction. A whitish precipitate forms after *12±14 hours at 378C and is then dissolved in normal saline for immunization. Immunological Assay: General Principles 237

TABLE 1 Characteristics of Some Species Identi®cation Field Tests (SIFTs)a

Test acronym and Immunized speci®city host Performance

ORBIT, Rabbit, 70 samples. No false beef (39) goat, sheep positives/negatives, Speci®city: bison ( ), bovine ( ), water buffalo ( ), deer ( À ), elk ( À ), goat ( À ), horse ( À ). PROFIT, Goat 66 samples. No false poultry (40) / À results. Speci®city: chicken ( ), turkey ( ), goose ( ), quail ( ), partridge ( ), bovine ( À ), deer ( À ), horse ( À ), pig ( À ), sheep ( À ).LLD 3% or 5% in pork or beef PRIME, Goat 83 samples. No false / À pork (42) results. LLD 5% pork (in beef), 3% pork in lamb. SOFT, Calf 90 samples. No false / À sheep (41) results. LLD 3% mutton in beef. REST, Sheep 101 samples. No false / À horse (43) results. Speci®city: deer, ( À ), donkey ( ), mule ( ), beef ( À ), pork ( À ), sheep ( À ), chicken ( À ), turkey ( À ), kangaroo ( À ). LLD 3% horse DRIFT, Goat 100 samples. No false / À deer (44) results. Speci®city: mule deer ( ), elk ( ), moose ( ), reindeer ( ), beef ( À ), horse ( À ), sheep ( À ), chicken/turkey ( À ). MULTI-SIFT, Various No false / À results. six different Speci®city (as above) species (45) aLLD, Lower limit of detection or minimum % weight of adulterant detectable. All antiserum ®lter discs are stable for 4±5 months at room temperature and 12 months at 48C. ( ) positive test, ( À ) negative test.

than in muscle. SDS-PAGE analysis of adrenal BE antigens showed a single 35.5-kDa component (troponin T). Cow and buffalo muscle BE antigen had three components with molecular masses of 35.5 kDa (troponin T), 19 kDa (troponin I), and 16 kDa (troponin C). Sheep and goat muscle BE antigen had the 35.5- and 19-kDa components. Pork thermostable antigen showed three troponin bands and an extra 66-kDa protein. 238 Chapter 8

TABLE 2 Boiling-resistant Ethanol-Insoluble (BE) Protein as Thermostable Antigen for Agar Gel Double Diffusions Assay of Cooked Meat

Analysis/comments Reference

Troponin T, as chicken muscle antigen Hayden (54) BE antigen for sausage analysis Hayden (56,57) Troponin T antigen for turkey analysis Schweiger et al. (55) BE antigen for differentiating domestic Kang'ethe et al. (59) meat sources (beef, goat, and sheep) from 14 game species BE antigen for detecting chicken Sherikar et al. (60,61) adulteration by buffalo, sheep, goats, pig meat BE antigen identi®ed as troponin T; Radhakrishna et al. (62); isolation from buffalo, cattle, sheep, Bhilegaonkar et al. (67); goat, and pig adrenals and muscle Sherikar et al. (63) tissue BE antigen for detection of pork in beef, Reddy and Giridhar-Reddy (64); buffalo, sheep, goat, chicken meat Saisekhar and Reddy (65) End-point temperature determination, Levieux and Levieux (66) myoglobin

Saisekhar and Reddy (65) isolated troponin T from raw beef and buffalo meat. An AGID assay based on native troponin T antigen from buffalo cross-reacted with cattle, goat, and sheep meat. There was no reaction with chicken or pork. The (rabbit) pAb for buffalo was rendered monospeci®c by immunoadsorption with cattle, goat, and sheep antigen. The AGID assay using monospeci®c buffalo pAb could detect beef or mutton adulteration with 1% of buffalo meat. Interestingly, (rabbit) pAb for bovine troponin T was monospeci®c for beef without prior immunopur- i®cation. No cross-reactivity occurred with buffalo, goat, sheep, or chevon meat. The LLD was 10% beef addition to samples of buffalo, chevon, or mutton. These tests based on pAb for native troponin T did not detect cooked meat. In contrast, pAb for native troponin T from chicken or turkey was sensitive to (poultry meat ) antigen in fried sausages (54,55). Clearly, troponin T is not heat resistant. Heating may generate a denatured but soluble troponin T that functions as a BE antigen. Much evidence points to troponin T being the major BE antigen. However, other muscle proteins might also play this role. SDS-PAGE analysis of muscle proteins extracted by 0.6 M KCl shows bands for myosin (200 kDa), actinin, actin (42 kDa), and troponin T (37 kDa). In the 32±28 kDa region appeared bands for troponin I, tropomyosin, and myosin light Immunological Assay: General Principles 239 chain (68). After comminution, troponin T was susceptible to proteolysis, forming a 30-kDa fragment. Proteolysis also occurred for meats up to 7 days old (69). After heating meat to 808C, only bands for troponin T, actin, and the myosin light chains remained. As actin is highly conserved between species (70), antibodies for this protein will not differentiate meat from different mammals (62,67). The pAb for actin could form a general test for animal tissue.

3.5. Agar Gel Double Immunodiffusion Assays for Cooked Meat Myoglobin was not wholly successful as a heat-stable antigen (31). To produce an AGID test for cooked meat, (rabbit) pAb was prepared using pure myoglobin preheated to 908C for 15 minutes. Cooking meat samples to 708C led to loss of sensitivity to meat myoglobin. There was cross-reactivity between (rabbit) pAb for lamb and beef myoglobin. However (goat) pAb for porcine myoglobin showed no cross-reactivity with beef. Thus 3% pork could be detected in unheated meat samples. Mild heating raised the LLD (lower limit of detection) to 10% adulteration of beef with pork. Hayden (31) proposed that (rabbit) pAb for heated myoglobin might be suited for developing AGID tests for mildly heated meat from bovine, horse, seal, or whale. Actually, pAb for heat-denatured pure myoglobin did not recognize myoglobin heated within a meat matrix. Apparently, the conformation of denatured myoglobin depends on the heating matrix. The denaturation of heme proteins involves an in-series mechanism. First, there is a conformational change and dissociation of the heme group. This binds covalently to the apo protein and other proteins or else forms heme oligomers (71);it is possible that some heme degradation products are antigenic. The apo protein then undergoes incorrect refolding and/or aggregation. The presence of other meat proteins can easily affect the end point for myoglobin denaturation. The AGID assay using (rabbit) pAb for chicken troponin T could detect the presence of 1±5% chicken meat in beef sausages cooked to an internal temperature of 70±908C (54). Using (rabbit) pAb for turkey troponin T, turkey sausages heated to an internal temperature of 708C and cooked for 10±15 minutes were also successfully analyzed. All assays were also highly sensitive to turkey meat in raw meat balls (mixture of pork and beef) and extract from poultry fried sausages (55). Thermostable BE antigen was the basis for the AGID tests for differentiating meat of domesticated species (cattle, goats, sheep) from 14 240 Chapter 8 wildlife game species.* The tests developed by Kang'ethe et al. (59) from the University of Nairobi used (goat) pAb. The crude pAb was surprisingly monospeci®c in most cases. Cross-reactivity was observed for some closely related species: buffalo and cattle, bushbuck and cattle, Grant's gazelle and sheep, and Grant's gazelle and Thomson's gazelle. After immunoadsorbtion, each pAb was rendered monospeci®c for thermostable antigen, cooked meat extracts, or fresh meat. The domesticated species were clearly distinguished from the game species. Differentiating between Grant's and Thomson's gazelle, kongoni and topi, and kongoni and wildebeest remained problematic. Using goat as the host for pAb production probably accounts for low cross-reactivity. By comparison, (rabbit) pAb shows lower speci®city and a greater likelihood of cross-reactivity between closely related species. Chicken meat is in high demand in parts of Asia owing to religious restrictions related to the consumption of beef. Sherikar et al. (61,63) produced an AGID assay for chicken meat adulteration with beef, buffalo, goat, mutton, or pork. BE antigen was prepared from heart, kidney, liver, spleen, or lung tissue. Crude (rabbit) pAb for pork was monospeci®c. The other (rabbit) pAbs were puri®ed by immunoadsorption. Partially puri®ed pAbs reacted only with homologous antigen from raw tissue or tissue mildly heated at 708C. Fully cooked meat could not be detected unless further processed to BE antigen (by autoclaving, centrifugation, and ethanol precipitation). Apparently, components in the relatively complex cooked meat extract interfered with antibody-antigen binding and/or precipitin formation. AGID assays were performed on samples with BE antigen. Adulteration of chicken with 10% (w/w) beef, buffalo, goat, or sheep meat was detectable. The LLD for pork was 5% (w/w). Reddy and Giridhar-Reddy (64) also produced an AGID assay for cooked pork. Porcine muscle BE antigen and the corresponding (rabbit) pAb were prepared as usual. The crude (rabbit) pAb was monospeci®c for pork. Samples for analysis were extracted from raw pork or after heating at 1208C for 30 minutes. The AGID assay detected 10% (w/w) pork in cooked meat from buffalo, cattle, chicken, goat, and sheep. The LLD for pork in raw meat mixtures was 20% (w/w). The frequency of undeclared meat substitutions was referred to earlier. Hsieh et al. (7) examined 806 raw meat and 96 cooked meat samples (mostly

* The species include buffalo (Syncerus caffer), bushbuck (Tragelaphus scriptus), cattle (Bos indicus), eland (Taurotragus oryx), goat (Capra aegagrus hircus), Grant's gazelle (Gazella granti), impala (Aepyceros malampus), kongoni (Alcelaphus buselaphus cokii), oryx (Oryx spp.), sheep (Ovis ammonaires), Thomson's gazelle (Gazella thomsoni), topi (Damaliscis lunatus), waterbuck (Kobus spp.), and wildebeest (Connochaetes taurinus). Immunological Assay: General Principles 241 beef and ground veal). The AGID tests involved commercially available pAb speci®c for raw sheep, pork, beef, and horse meat. Enzyme-linked immunosorbent assay (ELISA) kits utilizing pAb for thermostable antigen were used for the analysis of cooked meats. About 16% of raw meat samples were adulterated with meat from another species. The frequency of adulteration increased to 23% for cooked meat. The most frequent adulterants for ground veal or beef were sheep (47%), pork (42%), or poultry (31%). There were no substitutions involving horse meat. Cross- contamination via (improperly cleaned) processing equipment was not signi®cant. Ground lamb and pork had adulteration frequencies of 66.7% and 53%, respectively. So far, immunological methods cannot distinguish beef from veal or mutton from lamb.

3.6. Agar Gel Double Immunodiffusion Assays for Other Food Proteins AGID assays have not proved popular for the analysis of nonmeat proteins. The analysis of soy protein in meat is reviewed by Llewellyn (5). Hammond et al. (72) found that cross-reactivity with other legume proteins was widespread. The reliability of AGID tests for vegetable proteins is poor because of the effect of heat, extrusion, texturization, and other forms of processing. More promising are methods based on ELISA (Chapter 10).

REFERENCES

1. KD Hargin. Authenticity issues in meat and meat products. Meat Sci 3(Suppl):S277±S289, 1996. 2. PB Hutt, PB Hutt II. A history of government regulation of adulteration and misbranding of food. Food Drug Cosmet Law J 39:2±73, 1984. 3. RS Singhal, RR Kulkarni, DV Rege. Handbook of Indices of Food Quality and Authenticity. Cambridge, UK: Woodhead Publishing, 1997. 4. PR Ashurst, MJ Dennis, eds. Analytical Methods of Food Authentication. London: Blacke Academic & Professional, 1988. 5. JW Llewellyn. Analysis of novel proteins in meat products. In: BJF Hudson, ed. Developments in Food Proteins. Barking, Essex, UK: Applied Science Publishers, 1982, pp 171±216. 6. RLS Patterson, SJ Jones. Species identi®cation of meat in raw, unheated meat products. In: BA Morris, MN Clifford, eds. Immunoassay in Food Analysis. Barking, Essex, UK: Elsevier Applied Science, 1985, pp 85±94. 7. Y-HP Hsieh, BB Woodward, S-H Ho. Detection of species substitution in raw and cooked meats using immunoassays. J Food Prot 58:555±559, 1995. 242 Chapter 8

8. L Kurth, FD Shaw. Identi®cation of the species of origin of meat by electrophoretic and immunological methods. Food Technol Aust 35:328±223, 1983. 9. C Romero, O Perez-Andujar, A Olmedo, S Jimenez. Detection of cow's milk in ewe's or goat's milk by HPLC. Chromatographia 42:181±184, 1996. 10. W Haasnoot, DP Venema, HL Elenbass. Determination of cow milk in the milk and cheese of ewes and goats by fast protein liquid chromatography. Milchwissenschaft 41:642±645, 986. 11. DJ Hunt, HC Parkes, ID Lumley. Identi®cation of the species of origin of raw and cooked meat products using oligonucleotide probes. Food Chem 60:437± 442, 1997. 12. KS Fairbrother, AJ Hopwood, AK Lockley, RG Bardsley. The actin multigene family and livestock speciation using the polymerase chain reaction. Anim Biotech 9:89±100, 1988. 13. JA Lenstra, JB Buntjer. On the origin of meat. Food Chem 64:1, 1999. 14. C Wolf, J Rentsch, P Huebner. PCR-RFLP analysis of mitochondrial DNA: a reliable method for species identi®cation. J Agric Food Chem 47:1350±1355, 1999. 15. R Meyer, U Candrian. PCR-based DNA analysis for the identi®cation and characterization of food components. Lebensm Wiss Technol 29:1±9, 1996. 16. U Samarajeewa, CH Wei, TS Huange, RR Marshall. Applications of immunoassay in the food industry. CRC Crit Rev Food Sci Nutr 29:403± 434, 1991. 17. SS Gazzaz, BA Rasco, FM Dong. Applications of immunochemical assays to food analysis. CRC Crit Rev Food Sci Nutr 32:197±229, 1992. 18. JC Allen, CJ Smith. Enzyme-linked immunoassay kits for routine food analysis. Trends Biotech 5:193±199, 1987. 19. BK Barai, RR Nayak, RS Singhal, PR Kulkarni. Approaches to the detection of meat adulteration. Trends Food Sci Technol 3:69±72, 1992. 20. HA Lee, MRA Morgan. Food immunoassays: application of polyclonal, monoclonal and recombinant antibodies. Trends Food Sci 4:129±134, 1993. 21. PE Hernandez, R Martin, T Garcia, P Morales, G Anguita, AI Haza, I Gonzales, B Sanz. Antibody-based analytical methods for meat species determination and detecting adulteration of milk. Food Agric Immunol 6:95± 104, 1994. 22. WJ Taylor, NP Patel, J Leighton-Jones. Antibody based methods for assessing seafood authenticity. Food Agric Immunol 6:305±213, 1994. 23. DM Smith. Immunoassay process control and speciation of meats. Food Technol 49(2):116±119, 1995. 24. UV Mandokhot, SK Kotwal. Enzyme linked immunosorbent assays in detection of species origin of meatsÐa critical appraisal. J Food Sci Technol India 34:369±380, 1997. 25. G Kohler, C Milstein. Continuous cultures of fused cells secreting antibody of prede®ned speci®city. Nature 256:495±497, 1975. Immunological Assay: General Principles 243

26. ME Cutrufelli, RP Mageau, B Schwab, RW Johnston. Detection of beef and poultry by serological ®eld screening tests (ORBIT and PROFIT): collaborative study. J Assoc Off Anal Chem 70:230±233, 1987. 27. Association of Of®cial Analytical Chemists. Beef and poultry adulteration of meat products. Species identi®cation. First action. J Assoc Off Anal Chem 70:389±390, 1987. 28. S Baudner. Analysis of plant proteins using immunological techniques based on the antigen-antibody precipitation. Ann Nutr Aliment 31:165±177, 1977. 29. G Mancini, AO Carbonara, JF Heremans. Immunochemical quantitation of antigens by single radial immunodiffusion. Immunochemistry 2:235±254, 1965. 30. P Grabar, CA Willliams Jr. MethodeÂ immunoeÂ lectrophreÂ tique d'analyse de meÂ langes de substances antigeÂ niques. Biochim Biophys Acta 17:67±74, 1955. 31. AR Hayden. Immunochemical detection of ovine, porcine and equine ¯esh in beef products with antisera to species myoglobin. J Food Sci 44:494±500, 1979. 32. O Ouchterlony. In vitro methods for testing toxin-producing capacity of diphtheria bacteria. Acta Pathol Microbiol Scand 25:186±191, 1948. 33. EJ Oswarld. Serological methods in the regulatory control of foods. J Assoc Agric Chem 36:107±111, 1953. 34. SL He¯e. The chemistry and biology of food allergens. Food Technol 50(3):86, 88±92, 1996. 35. MO Warnecke, RL Saf¯e. Serological identi®cation of animal proteins. 1. Mode of injection and protein extracts for antibody production. J Food Sci 33:131±135, 1968. 36. A Avrameas, T Ternynck. Biologically active water-insoluble protein polymers. I. Their use for isolation of antigens and antibodies. J Biol Chem 242:1651, 1967. 37. T Ternynck, S Avrameas. Polymerization and immobilization of proteins using ethylchloroformate and glutaraldehyde. Scand J Immunol Suppl 3:29± 35, 1976. 38. MO Warnecke, RL Saf¯e. Serological identi®cation of animal proteins. 1. Mode of injection for antibody production. J Food Sci 33:131±135, 1968. 39. HG Fugate, SR Penn. Immunodiffusion technique for the identi®cation of animal species. J Off Anal Chem 54:1152±1156, 1971. 40. AR Hayden. Determination of residual species serum albumin in adulterated ground beef. J Food Sci 43:476±478, 492, 1978. 41. MB Helm, MO Warnecke, RL Saf¯e. Gamma globulin isolated from rabbit antiserum for rapid detection of meat adulteration. J Food Sci 36:998±1000, 1971. 42. KS Swart, CR Wilks. An immunodiffusion method for the identi®cation of the species of origin of meat samples. Aust Vert J 59:21±22, 1982. 43. RP Mageau, ME Cutrufelli, B Schwab, RW Johnston. Development of an Overnight Rapid Bovine Identi®cation Test (ORBIT) for ®eld use. J Assoc Off Anal Chem 67:949±954, 1984. 244 Chapter 8

44. AM Darwish, MA Soliman, HA Aideia, TM Nouman. Evaluation of the agar-gel immunodiffusion technique for differentiation of meats. J Egypt Vert Med Assoc 51:739±745, 1991. 45. DR Martin, J Chan, JY Chiu. Quantitative evaluation of pork adulteration in raw ground beef by radial immunodiffusion and enzyme-linked immunosorbent assay. J Food Prot 61:1686±1690, 1998. 46. MS Dreyfuss, ME Cutrufelli, RP Mageau, AM McNamara. Agar-gel immunodiffusion test for rapid identi®cation of pollock surimi in raw meat products. J Food Sci 62:972±975, 1997. 47. ME Flores-Munguia, MC Bermudez-Almada, L Vazquez-Moreno. Detection of adulteration in processed traditional meat products. J Muscle Foods 11:319±325, 2000. 48. ME Cutrufelli, RP Mageau, B Schwab. Development of Poultry Rapid Overnight Field Identi®cation Test (PROFIT) J Assoc Off Anal Chem 69:483± 487, 1986. 49. ME Cutrufelli, RP Mageau, B Schwab, RW Johnston. Development of Serological Ovine Field Test (SOFT) by modi®ed agar-gel immunodiffusion. J Assoc Off Anal Chem 72:60±61, 1989. 50. ME Cutrufelli, RP Mageau, B Schwab, RW Johnston. Development of Porcine Rapid Identi®cation Method (PRIME) by modi®ed agar-gel immunodiffusion. J Assoc Off Anal Chem 71:444±445, 1988. 51. ME Cutrufelli, RP Mageau, B Schwab. Development of a Rapid Equine Serological Test (REST) by modi®ed agar-gel immunodiffusion. J Assoc Off Anal Chem 74:410±413, 1991. 52. ME Cutrufelli, RP Mageau, B Schwab. Development of a Deer Rabid Identi®cation Field Test (DRIFT) by modi®ed agar-gel immunodiffusion. J Assoc Off Anal Chem 75:74±76, 1992. 53. ME Cutrufelli, RP Mageau, B Schwab. Development of a multispecies identi®cation ®eld test by modi®ed agar-gel immunodiffusion. J Assoc Off Anal Chem Int 76:1022±1026, 1993. 54. AR Hayden. Detection of chicken ¯esh in beef sausages. J Food Sci 42:1189± 1192, 1977. 55. A Schweiger, S Baudner, HO Gunther. Isolation by free-¯ow electrophoresis and immunological detection of troponin T from turkey muscle: an application in food chemistry. Electrophoresis 4:158±163, 1983. 56. AR Hayden. Use of antisera to a heat-stable antigen from equine adrenals for detection of horse meat in cooked beef sausages. J Anim Sci 49(Suppl 1):221± 222, 1979. 57. AR Hayden. Use of antisera to heat-stable antigens of adrenals for species identi®cation in thoroughly cooked beef sausages. J Food Sci 46:1810±1813, 1819, 1981. 58. F Milgrom, E Witebsky. Immunological studies on adrenal glands. 1. Immunization with adrenals of foreign species. Immunology 5:46±66, 1962. 59. EK Kang'ethe, JM Gathuma, KJ Lindqvist. Identi®cation of the species origin of fresh, cooked and canned meat and meat products using antisera to Immunological Assay: General Principles 245

thermostable muscle antigens by Ouchterlony's double diffusion test. J Food Sci Agric 37:157±164, 1986. 60. AT Sherikar, JB Khot, BM Jayarao, SR Pillai. Differentiation of organs of meat animals and identi®cation of their ¯esh in chicken using anti-adrenal BE sera. Indian J Anim Sci 58:565±573, 1988. 61. AT Sherikar, JB Khot, BM Jayarao, SR Pillai. Use of species-speci®c antisera to adrenal heat-stable antigens for the identi®cation of raw and cooked meats by agar gel diffusion and counter immunoelectrophoresis techniques. J Sci Food Agric 44:63±73, 1988. 62. K Radhakrishna, DV Rao, TR Sharma. Characterization of the major component in thermostable muscle proteins. J Food Sci Technol India 26:32± 35, 1989. 63. AT Sherikar, UD Karkare, JB Khot, BM Jayarao, KN Bhilegaonkar. Studies on thermostable antigens, production of species-speci®c antiadrenal sera and comparison of immunological techniques in meat speciation. Meat Sci 33:121± 136, 1993. 64. PM Reddy, M Giridhar-Reddy. Immunochemical detection of pork using thermostable muscle antigens. J Food Sci Technol India 32:326±328, 1995. 65. Y Saisekhar, PM Reddy. Use of troponin for species identi®cation of cattle and buffalo meats. J Food Sci Technol India 32:68±70, 1995. 66. D Levieux, A Levieux. Immunochemical quanti®cation of myoglobin heat denaturation: comparative studies with monoclonal and polyclonal antibodies. Food Agric Immunol 8:111±120, 1996. 67. KN Bhilegaonkar, AT Sherikar, JB Khot, UD Karkare. Studies on characterization of thermostable antigens of adrenals and muscle tissues of meat animals. J Sci Food Agric 51:545±553, 1990. 68. C-S Cheng, FC Parrish. Heat induced changes in myo®brillar proteins of bovine logissimus muscle. J Food Sci 44:22±24, 1979. 69. S Ebachi, T Wakabayashi, F Ebashi. Troponin and its components. J Biochem 69:441±445, 1971. 70. L Stryer. Biochemistry, 3rd ed. New York: WH Freeman, 1988, pp 920±974. 71. JL Forsyth, RKO Apenten, DS Robinson. The thermostability of puri®ed isoperoxidases from Brassica oleracea Var. gemmifera. Food Chem 65:99±109, 1999. 72. JC Hammond, IC Cohen, B Flaherty. A critical assessment of Ouchterlony's immunodiffusion technique as a screening test for soya protein in meat products. J Assoc Public Anal 14:119±126, 1976.

Speciation of Meat Proteins by Enzyme-Linked Immunosorbent Assay

1. INTRODUCTION

Enzyme-linked immunosorbent assay (ELISA) was invented by Eva Engvall during her Ph.D. studies in Stockholm in 1971 (1). A similar assay was produced by Van Weemen and Shuurs (2). Developments between 1971 to 1981 are reviewed in reference 3. ELISA is an example of an enzyme immunoassay (EIA). These are immunological tests ending with enzymatic analysis. Enzyme multiplied immunoassay tests (EMITs) are performed in solution. ELISA employs immunoglobins attached to a solid surface. EIA is further categorized as competitive or noncompetitive (Table 1). Secs 2±5 of this chapter describe authenticity testing for raw and cooked meat by ELISA. The meat antigens detected with such tests are mostly residual serum proteins from the blood or else muscle proteins. Sec. 6 covers ELISA for meat protein using monoclonal antibodies (mAbs). Seafood speciation is discussed in Sec. 7, followed by a comparison of the different ELISA formats in Sec. 8. Attempts to detect meat and bone meal in animal feeds are described in Sec. 9, as are some new ELISA tests for the bovine spongiform encephalopathy (BSE) agent. During EMIT an enzyme-antigen conjugate (Enz-antigen) reacts with antibody (Ig). Steric hindrance from Ig binding reduces enzyme activity. The

247 248 Chapter 9

TABLE 1 A Range of Enzyme Immunoassays (EIAs)a

Solution-phase EIA Solid-phase EIA

EMIT ELISA Competitive Competitive Enz-antigen Enz-antigen, Enz-Ig Noncompetitive Direct, indirect, sandwich aEnz, enzyme;Ig, antibody. Other acronyms are de®ned in the general text. added sample (with known concentration of antigen) competes with the Enz- antigen for Ig. This process relieves steric inhibition of the Enz-antigen. EMITs are suitable for analyzing small molecular weight antigens. The measured enzyme activity is proportional to the added analyte concentration. To perform EMIT requires access to pure antigen and Enz-antigen conjugate. Competitive ELISA uses enzyme-labeled antibody (Enz-Ig or Enz- Ig0)* or Enz-Antigen conjugate as reagent:

1. Adsorb pure antigen on microwell plates (Fig. 1) 2. Block excess adsorption sites with milk protein. 3. In a separate step, preincubate Enz-Ig with the sample (containing an unknown concentration of antigen). 4. Transfer the mixture to the antigen-coated microwell plate. 5. Wash to remove excess reagent. 6. Assay the microwell plate for bound Enz-Ig using enzyme substrate.

The enzymatic activity is inversely related to the concentration of antigen in step 3. An alternative one-step competitive ELISA is performed as follows:

1. Coat microwell plates with antibody. 2. Block nonspeci®c sites (as described above). 3. Add Enz-antigen and sample and allow competition for the microwell-bound Ig. 4. Wash microwell plates and assay with enzyme substrate.

The enzyme activity is inversely related to sample antigen concentration.

* Enz-Ig enzyme linked to a primary antibody speci®c for a food protein. Eng-Ig0 enzyme conjugate with antibody (Ig0) speci®c for the primary antibody, e.g., goat antibody for (rabbit) Ig. Speciation of Meat Proteins 249

FIGURE 1 Competitive ELISA using enzyme-labeled antibody or enzyme-labeled antigen. (Top) Direct competitive ELISA, bound antigen (1) and free antigen (3) compete for Enz-Ig (2). (Bottom) Enz-Antigen (2) competes with antigens (3) for bound antibody (1).

Noncompetitive ELISA is performed using a direct, indirect, or sandwich format. For indirect ELISA the order of reagent addition is sample, ®rst antibody, Enz-Ig0 conjugate, and enzyme substrate (Fig. 2). After each step, wash the microwell plates to remove excess reagents. The ``indirect'' pre®x refers to the use of a second antibody (see Enz- Ig0) to visualize the bound antigen. With direct ELISA, the antigen is visualized (directly) using one antibody (Enz-Ig). Although requiring two antibodies, indirect ELISA is a more ¯exible and ultimately cheaper. The second antibody (Ig0) is produced by immunizing goat or sheep with the primary rabbit Ig. Commercially available Enz-Ig0 can be used for visualization wherever (rabbit) pAb is used for analysis. By contrast, a distinct Enz-Ig has to be synthesized for every new direct ELISA. The sandwich ELISA format may be implemented by binding ``capture'' 250 Chapter 9

FIGURE 2 Formats for noncompetitive ELISA. Meat antigen (1) adsorbed on the surface of a microwell plate, (2) species-speci®c rabbit antibody, (3) enzyme-antibody directed against the rabbit antibody. antibody to the microwell plate, followed by antigen. Thereafter one may adopt either the indirect or direct visualization strategy. Applications of ELISA for food analysis were ®rst discussed at a symposium held at the University of Surrey in September 1983 (4). The same theme was addressed by Hitchcock (5) at the second symposium on the application of immunoassays in veterinary and food analysis. Potential targets for immunoassay are listed in Table 2. Reviews dealing with ELISA for food analysis and authenticity testing include those by Allen and Smith (6), Samarajeewa et al. (7), Gazzaz et al. (8), Barai et al. (9), Lea and Morgan (10), Hernandez et al. (11), Taylor et al. (12), Smith (13), Simpkins and Harrison (14), Mandokhot and Kotwal (15), and Luethy (16). Hitchcock and co-workers (17)* were ®rst to use ELISA for food protein speciation. Compared with ELISA, other immunological assays have some of the following disadvantages: (a) a requirement for large

* They were af®liated with Unilever Research Laboratories, Colworth House, Bedford (UK). Speciation of Meat Proteins 251

TABLE 2 Potential Food Analytes (Antigens) for ELISA

Major groups Examples

Trace components/contaminants Mycotoxins Ochratoxin A1, a¯atoxins Bacterial toxins Clostridium, Staphylococcus, E. coli Hormones and anabolic agents Natural, synthetic reproductive or growth-affecting hormones Drugs and Antibiotics Antinutrients Solanine, trypsin inhibitor Vitamins Plant hormones Indoleacetic acid, abscisic acid Pesticide and residues Additives Low molecular weight Colors, ¯avors High molecular weight Gums, stabilizers, emulsi®ers Food protein speciation Meat and egg Beef, buffalo, camel, poultry Milk Cow, ewe, goat's milk (speciation), speci®c milk proteins Blood Serum protein speciation Bacterial and fungal Single-cell proteins Plant proteins Soya, wheat gluten, pea, potato

Source: Adapted from Refs. 4 and 5. quantities of antibody;(b) long assay times of between 18 and 24 hours, although 2±3 hours have been used for qualitative studies, and (c) high cost per analysis due to the cost of antibody. The advantages of ELISA compared with argar gel immunoassays include (18) (a) a 100- and 1000-fold lower requirement for antibody, (b) small sample size (1 mg of meat) per assay, (c) increased (10- and 100-fold higher) sensitivity, (d) more easily interpreted results, (e) screening for multiple adulterations possible, and (f) increased potential for partial automation. ELISA is a major technique for meat protein speciation (Table 3). Implementing ELISA involves the following stages: (a) production of antibody, (b) puri®cation of antibody by af®nity chromatography, (c) preparation of Enz-Ig conjugate, (d) extraction of meat antigen, and (e) ELISA. Depending on past experiences, one can perform all operations in house or use a commercially available ready-to-use kit. Some off-the-shelf components may be combined with reagents prepared in house. Essential equipment and accessories include a 96-well microwell plate reader, microwell plate washer, precision micropipettes, and microwell plates. A 252 Chapter 9

TABLE 3 Speciation of Raw Meat by ELISA with pAb

Analysis/comments Reference

ELISA Engvall and Perlman (1) Indirect ELISA, horse, beef Kang'ethe et al. (19) Indirect ELISA, beef, camel, horse, kangaroo, sheep Whittaker et al. (18,20) Sandwich ELISA, beef, sheep, horse, kangaroo, pig, Patterson et al. (21) camel, buffalo, and goat Sandwich ELISA, pork, beef Jones and Patterson (22) Sandwich ELISA, beef, buffalo Patterson and Spencer (23) Indirect ELISA, beef, pork, horse Jones and Patterson (24) Meat identi®cation ELISA kits (UK) Pelly and Tindle (25) Sandwich ELISA,a beef, pork Martin et al. (26,27) Sandwich ELISA,a chicken Martin et al. (27) Indirect/competitive ELISA, direct/competitive Ayob et al. (29) ELISA, pork, beef Indirect ELISA, hand or mechanically recovered Stevenson et al. (30) chickena meat Seafood (sardine, tuna, and crustacea) Taylor et al. (31) aStudies based on muscle protein antigen. Indirect ELISAs are noncompetitive methods unless stated.

plate reader enables colorimetric readings from microwell plates in situ. The 96 wells can be read within a space of 1.5 minutes. Several precision micropipettes are necessary for dispensing reagents;most essential are 50- mL and 100-mL pipettes. A continually adjustable (50±200 mL) multiwell pipette is convenient for rapid dispensing. Polystyrene microwell plates appear to be the solid phase of choice.

2. RAW MEATSPECIATION BY INDIRECTELISA

Noncompetitive indirect ELISA is probably the simplest EIA format. A reasonably comprehensive description of this technique is provided here to encourage those wishing to implement this assay for the ®rst time. This section also considers the limits and tolerances of this technique. Method 1 Analysis of uncooked meat by noncompetitive indirect ELISA. Reagents 1. Detection antibody (e.g., peroxidase-labeled antibody) Speciation of Meat Proteins 253

2. Antigen standard 3. Coating buffer (0.1 M sodium carbonate buffer, pH 9) 4. PBST (phosphate-buffered saline with 0.05% Tween 20)Ðwash and diluent buffer 5. Enzyme assay buffer (citrate-phosphate buffer, pH 4.2) 6. Enzyme substrate (variousÐsee the following) 7. Enzyme stopping solution (variousÐsee the following) Procedure 1. Coat microwell plates. Add 100 mL of meat extract (diluted in PBST) to microwell plates and incubate for 60 minutes at room temperature. Wash with PBST (100 mL) three times. 2. First antibody. Add 100 mL of rabbit antibody (diluted in PBST). Incubate for 30±60 minutes and wash wells with PBST (100 mL) three times. 3. Detection antibody. Enzyme conjugate. Add 100 mL of Enz- antibody conjugate. Incubate for 60 minutes. Wash with PBST (100 mL) three times. 4. Enzyme assay. Add 100 mL of enzyme substrate. Incubate for 30 minutes. Add stopping solution and record absorbency reading with the plate reader. Not counting the PBST washing steps, indirect ELISA involves four steps. Perform preliminary experiments to establish the optimum sample (antigen) dilution as well as the required concentrations of antibody and Enz-Ig conjugate. Incubation times ranging from 30 minutes to 3 hours have been employed. Horseradish peroxidase (HRP) is the most common enzyme label for ELISA. Suitable substrates for HRP are ABTS [2,20-azino-di(3- ethylbenzthiazoline sulfonate)] and hydrogen peroxide. Other HRP substrates are listed in Table 4. Alkaline phosphatase, urease, and glucose oxidase have also been used as labels. Kang'ethe et al. (19) developed indirect ELISA for horse meat. Polyclonal antibody (pAb) for horse serum albumin (HrSA) was raised by immunizing rabbits. Preliminary tests using AGID assay showed that (rabbit) pAb cross-reacted with BSA and sheep serum albumin (SSA). Therefore crude (rabbit) pAb for HrSA was puri®ed by column immunoadsorption. Antibody samples were diluted by about 100À1 and meat extracts diluted by 200À1 and 3200À1 before assay. Indirect ELISA was performed essentially as described in Method 1. Substitution of beef with 5± 80% horse meat produced the calibration response 1 DA À K %Horse ln 492 1 1 C K2 254 Chapter 9

TABLE 4 Enzymes and Substrates for ELISA

Substrate and assay Enzyme conditions Stopping solution

Horseradish peroxidase ABTS (2 mM) H2O2 30 mL of NaCN (37 mM) (2 mM), citrate- or 50 mL citric acid phosphate (0.1 M, pH (0.1 M) 4.2), DA 414 nm

Horseradish peroxidase o-Dinisidine (0.08%) 50 mLofH2SO4 (4 M) H2O2 (0.006%), citrate- phosphate buffer (0.1 M, pH 5); DA 620 nm

Horseradish peroxidase o-Diphenylenediamine 50 mLofH2SO4 (12.7 M) H2O2 in citrate- phosphate buffer (0.1 M, pH 5), DA 492 nm

Horseradish peroxidase o-Toluidine (2.5 mM) 50 mLofH2SO4 (12.7 M) H2O2 (2.5 mM), citrate- phosphate (0.1 M, pH 4.5), DA 620 nm 0 0 Horseradish peroxidase 3,3 ,5,5 -Tetramethylben- 100 mLofH2SO4 (3 M) zidine (1 mM) H2O2 (3 mM) in 0.2 M citrate buffer (pH 3.95), DA 450 nm Alkaline phosphatase p-Nitrophenyl phosphate 25 mL of NaOH (0.4 N (1 mg/mL) in NaOH) diethanolamine buffer, pH 9.8, DA 405 nm

Source: Compiled from multiple sources in Table 3.

where C, K1,andK2 are constants and A492 is the absorbance reading. A simple straight-line equation applied for 0±60% substitution of beef by horse meat. The assay precision was 2.3±8%. Whittaker et al. (18,20) employed indirect ELISA for the identi®cation of uncooked meat from cattle, camel, horse, kangaroo, and sheep. To improve speci®city, (rabbit) pAbs for serum proteins were puri®ed via af®nity chromatography. Antigen adsorption to microwell plates was optimum at pH 5±6. The working range for analysis was 10±80% (w/w) adulteration. Speciation of Meat Proteins 255

3. RAW MEATSPECIATIONBY SANDWICH ELISA

The ®rst step for sandwich ELISA is coating microwell plates with capture antibody. Then meat extract is added followed by either Enz-Ig conjugate or the combination of second Ig Enz-Ig0. Binding meat antigens to a ``bed'' of antibody introduces selectivity. Noncomplementary proteins do not bind to the ®rst antibody and can therefore be washed from the microwell plate. Further speci®city derives from the second antibody (Fig. 3). Patterson et al. (21) were the ®rst to develop a sandwich ELISA for meat speciation. Different ELISA tests were produced with speci®city for meat from buffalo (water buffalo),* camel, cattle, goat, horse, kangaroo, pig, or sheep. Capture pAbs were usually raised by immunizing sheep with whole serum protein from camel, cattle, goat, etc. Speci®c pAb for sheep was produced using cattle as host. The yield of pAb was greater if the host animal was phylogenetically different from the donor species. Sheep produced greater quantities of pAb when injected with kangaroo antigen as compared with beef antigen. Detection pAbs were raised using rabbits. Samples of crude pAb were puri®ed by af®nity chromatography using the complementary antigens immobilized on CNBr-activated Sepharose.

FIGURE 3 Sandwich ELISA format. (1) Microwell plate±bound ®rst (capture) antibody, (2) meat antigen, (3) second (usually rabbit) antibody, and (4) enzyme-conjugated goat antibody for rabbit IgG.

* The American plains bison is also buffalo. The buffalo referred to in this chapter is the water buffalo of Asia and Africa called simply buffalo in the literature. 256 Chapter 9

Tests with 8 pAbs versus 8 meat extracts (64 tests) showed three false positives. The (cow) pAb for sheep cross-reacted with goat meat. However (sheep) pAb for goat did not cross-react with sheep. The (sheep) pAb for beef cross-reacted with buffalo meat, but (sheep) pAb for buffalo did not react with beef. By choosing the host animal for antibody production carefully, speci®c pAb could be produced for sandwich-ELISA. The relatively high pAb speci®city was ascribed to the following factors: (a) choice of host species (sheep or goats produced more speci®c antibodies than rabbit or mice) and (b) choice of antigen. Using whole serum protein for immunization, rather than a single pure protein, introduces many antigenic determinants. The pAbs are produced that are more discriminat- ing between species. Meat samples (1 g) were extracted by 10 mL of solvent and diluted by a factor of 10À1±5000À1 before analysis. The LLD was 1% (w/w) kangaroo meat substitution for beef or 1% (w/w) substitution of goat meat for sheep. Cross-reactivity between closely related species (beef- buffalo, goat-sheep, and donkey-horse) was evident. Patterson and Spencer (23) also produced monospeci®c pAbs for buffalo, goat, or donkey using cattle, sheep, or horse as host, respectively. Each pAb was then puri®ed by immunoaf®nity chromatography. Thus, (sheep) pAb for goat was puri®ed with a column of Sepharose±goat serum protein. Bound pAb was eluted with ammonium thiocyanate (2.5 M, pH 7.0), desalted by gel ®ltration with Sephadex G25, and concentrated by ultra®ltration. The pAb sample containing 8 mg mLÀ1 protein was divided into two portions;half was covalently conjugated to HRP for detection and the other half was used for capture. Enzymatic detection was via o-toluidine±hydrogen peroxide (Table 4). From visual inspection the LLD was 0.1% (w/w) donkey meat added to horse meat or 0.1% (w/w) goat meat added to mutton. Beef adulteration with > 1.0% (w/w) buffalo meat was detectable. Jones and Patterson (22) showed that it was possible to detect 0.5±1% (w/w) adulteration of beef by pork using a sandwich ELISA. The linear range of analysis was 1±3%. The capture antibody was (rabbit) pAb for porcine serum albumin (PSA). The detector pAb was produced with a sheep host. Both (rabbit) pAb and (sheep) pAb for PSA were puri®ed by a two- stage af®nity procedure. The order of reagent addition was (rabbit) pAb for PSA, meat extract, (sheep) pAb for PSA, and HRP-pAb conjugate for sheep Ig. The (sheep) pAb for PSA was unstable when directly adsorbed on microwell plates.* Presumably, instability prevents the preparation of HRP

* Reagent stability is an important feature of sandwich ELISA. Once coated with capture antibody, microwell plates may be dried and stored at refrigerator temperatures for 6 months. Speciation of Meat Proteins 257 conjugate with (sheep) pAb for PSA. Also important for assay design is the low speci®city of (rabbit) pAb for PSA. Using (sheep) pAb for PSA to complete the ``sandwich'' improved assay sensitivity. For details of the sample pretreatment see footnote.* The LLD was 1% (w/w) pork in minced beef, beef sausage mix, or beef burger mix. For 1±10% (w/w) substitution, the assay response was described by the relation

A492 0:301 0:153 log %Pork 2 These results should be compared with the analysis of pork in commercial meat (ham, pork-soy sausages, pate) products (29) by indirect ELISA. Using af®nity-puri®ed (rabbit) pAb for pork, the linear range of analysis was 1±40% (w/w) substitution.

4. MUSCLE PROTEIN ANTIGENS FOR ELISA

The ELISAs described so far were speci®c for residual blood proteins within meat. These assays are unsatisfactory. Cross-contamination by blood from another species gives a positive ELISA test (26,30). Furthermore, the amount of blood lost during the conversion of muscle to meat is variable (32). Attempts to identify different cuts of meat using ELISA for blood serum proteins were not successful. There were large variations in the blood content in different samples (33). Martin and co-workers (26) developed a sandwich ELISA for (porcine) muscle protein. Extracts of pork diluted by between 20À1 and 20,480À1 gave absorbance readings of < 1.0. Substitution of beef with 1±50% (w/w) pork led to the calibration response

A492 0:268 0:114 ln %Pork 3 With experienced personnel, the assay precision was 2±3%. In another study, chicken muscle antigen was isolated using an af®nity column ®lled with Sepharose±protein A complex with (rabbit) pAb for chicken (28). Bound antigen was eluted with diethylamine buffer (0.05 M, pH 11.5). An

* The assay was calibrated using samples of 0±10 g of pork added to 1±2 kg of lean minced beef. The mixture was blended with 900 mL of distilled water for 2 minutes, ®ltered through Whatman No. 3 paper, and stored at À208C. Other meat products of known formulation were also analyzed. Ten-kilogram amounts of beef burger and beef sausage mixtures were prepared according standard recipes. Then 40-g samples were extracted with 360 mL of water and ®ltered. The resulting extracts were diluted with PBST for immunoassay. Samples were diluted by 50À1 and 250À1 for assay. 258 Chapter 9

SDS-PAGE analysis followed by Western transfer and immunostaining showed that the chicken-speci®c antigen was probably pyruvate kinase. The antigen was used for sandwich ELISA with both the capture and detector antibody being (rabbit) pAb. Substitution of beef with 1±10% (w/w) chicken meat gave the following response:

A492 0:621 0:161 ln %Chicken 4 The corresponding assay for pork adulteration with 1±10% (w/w) chicken meat led to the following performance.

A492 0:590 0:154 ln %Chicken 5 Stevenson et al. (30) developed an assay for mechanically recovered chicken meat and hand-deboned chicken using indirect ELISA for bone marrow antigen. Crude (rabbit) pAb was puri®ed by ammonium sulfate precipitation and af®nity chromatography using Sepharose-immobilized bone marrow antigen extracted with 7 M urea. The ELISA was visualized using af®nity-puri®ed (rabbit) pAb for chicken bone marrow, followed by commercial HRP-labeled (goat) pAb for rabbit IgG. HRP was assayed with the substrate 3,30,5,50-tetramethylbenzidine (TMB), which, unlike o-diphenylenediamine, is noncarcinogenic. The ®nal assay showed only slight selectivity toward MRM as compared with hand-deboned meat. There was no cross-reactivity with hand-deboned beef, mutton, or pork. Substitution of beef with MRM chicken could be detected at levels of 2±50%. The identity of muscle protein antigens has not been fully established. An SDS-PAGE analysis of soluble protein from horse meat revealed *20 proteins. Immunoaf®nity chromatography using immobilized (rabbit) pAb for horse muscle reduced the number of SDS-PAGE bands to nine. Three antigens (37, 70, and 96 kDa) increased in concentration after af®nity chromatography (34). SDS-PAGE also featured in the partial identi®cation of the bone marrow antigen from chicken (30). Proteins from bone marrow or muscle were separated by SDS-PAGE followed by Western blot transfer to a nitrocellulose membrane. The bound protein was visualized with species-speci®c (rabbit) pAb and commercially available HRP-labeled (goat) antibody for rabbit IgG. Immunostaining revealed three antigenic proteins with molecular sizes of 69, 45, and 96 kDa. The 69-kDa protein was tentatively identi®ed as chicken serum albumin. The identities of the 45- and 96-kDa proteins are not known. The case for adopting muscle protein antigens for ELISA is compelling. However, the extra effort involved in isolating muscle proteins may require further justi®cation. Fig. 4 shows results for sandwich ELISA using pAb speci®c for muscle soluble protein or blood serum albumin. These Speciation of Meat Proteins 259

FIGURE 4 Blood versus muscle protein antigen for meat speciation using sandwich ELISA. Adulteration of beef with pork was analyzed using (rabbit) pAb for porcine muscle protein (open circles) or pAb for porcine serum albumin (closed circles). 260 Chapter 9 results are comparable. Furthermore, it is not certain that the quantity of soluble muscle proteins is the same for different meat tissues from any single species. Myoglobin levels vary with tissue type and levels of exercise.

5. COOKED MEATANALYSIS BY ELISA 5.1. Boiling-Resistant Ethanol-Soluble (BE) Antigen Kang'ethe and Lindqvist (35) found that BE antigen was not wholly suitable for indirect ELISA. The antigen showed irregular adsorption to microwell plates because of the presence of extraneous proteins (probably gelatin). Samples of BE antigen gelled at 48C. Notwithstanding partial puri®cation by size exclusion chromatography, indirect ELISA using BE antigen yielded poor sensitivity (36). Tests involved (goat) pAb for partially puri®ed BE antigen from 4 domesticated species (cattle, camel, pig, and sheep) and 14 games species.* With a total of 324 tests (18 antibody 6 18 meat samples), no cross-reactivity was observed using pAbs for water buffalo, camel, horse, topi, and pig. The (goat) pAb for cattle BE antigen cross-reacted with virtually every species tested. Using the appropriate species-speci®c (goat) pAb between 1 and 10% (w/w) adulteration of beef (with buffalo), pork (with warthog), or goat (with impala) was detectable. The sensitivity of meat tests using BE antigen was improved 100-fold by adopting the sandwich ELISA format (37). The higher performance was attributed to BE antigen binding to capture antibody rather than to a ``bare'' microwell plate. Direct binding to surfaces alters antigen conformation and can reduce antibody binding (38,39). In all, six sandwich ELISA tests were developed using (goat) pAb for BE antigen from buffalo, bushpig, camel, cattle, horse, and pig. Each assay was tested with meat samples from 14 species.* The ELISA test for beef showed cross-reactivity for buffalo, horse, and bushbuck meat. All other assays were species speci®c. Adulteration of beef or pork with 1±20% (w/w) buffalo or camel meat was readily detected using the appropriate sandwich ELISA for these adulterants.

* Buffalo (Syncerus caffer), bushbuck (Tragelaphus scriptus), cattle (Bos indicus), eland (Taurotragus oryx), goat (Capra aegagrus hircus), Grant's gazelle (Gazella granti), impala (Aepyceros malampus), kongoni (Alcelaphus buselaphus cokei), oryx (Oryx spp.), sheep (Ovis ammonaires), Thomson's gazelle (Gazella thomsoni), topi (Damaliscus linatus), waterbuck (Kobus spp.), and wildebeest (Connochaetes taurinus). Speciation of Meat Proteins 261

5.2. Thermostable Muscle Protein Antigen Hsieh and co-workers produced thermostable muscle protein antigens for analysis of cooked poultry (40), pork (41,42), or red meat (43). Lean meat paste was blended with three volumes of water or 0.85% saline. The resulting meat slurry was heated at 1008C for 15 minutes, cooled to refrigeration temperatures, and shaken gently for 2 hours. The supernatant obtained after centrifugation (14,000g) was employed as thermostable antigen. SDS-PAGE and Western blot analyses revealed that thermostable antigens from chicken skeletal muscle had molecular masses of 22±25, 30±35, and 120 kDa. The proteins were not identi®ed but are likely to correspond to troponin C, troponin T, and myosin fragment. Analysis of the antigen from other animal sources showed bands with molecular sizes ranging from 14.5 to 26.5 kDa. Thermal-stable muscle protein antigens were the basis for developing mAb for ELISA (Sec. 6). Rencova and co-workers described indirect competitive ELISA tests for heat-processed meat from horse, kangaroo, poultry, or rat. The tests involved (rabbit) pAb for heat-stable muscle antigen. This antigen was prepared by heating meat sample extracts with PBS at 100±1208C for 30 minutes. The assay detected chicken or kangaroo meat within commercial meat products from the retail market in the Czech Republic (44).

5.3. Native Thermostable Antigen Native thermostable antigens (nTAs) are produced from raw tissue by ammonium sulfate fractionation and ion-exchange chromatography on carboxymethylcellulose (45). The isolation process, although technically straightforward, lasts several days. The yield of nTA was 25 mg per kg of meat. Sandwich ELISA for using nTA for chicken cross-reacted with turkey (Fig. 5). No cross-reactivity occurred with red meat (beef, deer, horse, kangaroo, or sheep). Both cooked and uncooked poultry meat could be analyzed. This is understandable because a genuinely heat-resistant meat antigen should be unaffected by thermal treatment. A sandwich ELISA using nTA for pork was also developed (Fig. 6). The nTA-based tests can detect chicken or pork in a wide range of cooked meat products: (a) frankfurters (horse, beef, pork, sheep, deer, chicken or turkey);(b) bologna (beef, pork, chicken, or turkey);(c) chopped, pressed, and sliced meats (beef, ham, chicken, turkey);(d) canned baby foods (beef, pork, lamb, chicken, or turkey);and (e) canned meat spreads (beef, ham, chicken). In every case, product varieties containing poultry or pork were correctly identi®ed. The LLDs for chicken and pork were 126 and 250 ppm, respectively. The sensitivity, ascribed to the biotin-streptavidin 262 Chapter 9

FIGURE 5 Speci®city of sandwich ELISA for chicken native thermostable antigen. (Drawn using results from Ref. 45.)

ampli®cation system, was more than adequate to detect meat adulterations of practical signi®cance. Assay performance was not affected by the nature of the meat matrix. The nTA was also the basis for sandwich ELISA tests for beef, deer, horse, and mutton (46). The nTAs were routinely puri®ed by immunoadsorption (47,48). Even then some cross-reactivity occurred with the following samples: beef/American bison, goat/sheep, donkey/horse, white- tail deer/mule deer±caribou. With cooked meat adulterants the LLD was 0.16% (w/w). Apparently the assays were less ef®cient for raw meat. A commercial ELISA kit utilizing nTA was used by Hsieh et al. (49) to survey cooked meat adulteration in Florida. Commercial kits, said to be based on the USDA protocol, are available for cooked beef, pork, poultry, sheep, horse, and deer meat (ELISA Technologies, Alachua, FL). In the United Kingdom similar kits are available form Cortecs Diagnostics Ltd (Newtech Square, Deeside Industrial Park, Flintshire, CH5 2NT, UK). Potential dif®culties arising from the analysis of cooked meat samples containing high amounts of gelatin using commercial ELISA kits were described (50). Speciation of Meat Proteins 263

FIGURE 6 Speci®city of sandwich ELISA for cooked pork based on antibody for native thermostable antigen. (Drawn using results from Ref. 45.)

The identity of the 50-kDa protein associated with nTA has not been established. Antibody for nTA did not react with a-acid glycoprotein (also called a-HS-glycoprotein) (45,46). The list of blood serum proteins includes immunoglobins (160 kDa), transferrin (85 kDa), albumin (66 kDa), Ig fragment (45 kDa), a1-antitrypsin (45 kDa), orosomucoid (44 kDa), GC globulin (51 kDa), a-HS-glycoprotein (49 kDa), g-globulin (25 kDa) and b2-microglobulin (11.8 kDa). The a-HS-glycoprotein is probably the most heat-stable serum protein. However, several serum proteins have molecular sizes around the region 45±50 kDa (51). The thermal stability characteristics of nTAs are unlike those of other muscle proteins (Table 5, Fig. 7). Levieux et al. (52) heated readily extractable muscle proteins and analyzed the residual soluble proteins by QSDS-PAGE. The order of thermal resistance was albumin > myoglobin lactate dehydrogenase (M4) > IgG > transferrin. All proteins were completely denatured by heating at > 808C for 30 minutes. I estimate that the half-life (t1/2) of chicken nTA is 213.6 or 52 minutes at 100 or 1258C, respectively. Turkey or beef nTAs were relatively less heat resistant by comparison with t1/2 of 51.3 or 103 minutes at 1008C. These t1/2 values were reduced to 35 minutes (turkey nTA) or 68 minutes (beef nTA) at 1208C. The heat deactivation mechanism for nTA was also different from 264 Chapter 9

TABLE 5 Thermal Inactivation Parameters for Some Soluble Muscle Proteins

Muscle DH# DS# protein (kJ molÀ1) (J molÀ1 KÀ1) Arrhenius equation

IgG 184.3 280 ln k 58.6 À 2.23 6 104 (1/T) LDH (M4) 340.7 740 ln k 113.0 À 4.05 6 104 (1/T) Myoglobin 455.7 1040 ln k 136.4 À 4.90 6 104 (1/T) Albumin 475.3 1130 ln k 150.6 À 5.3 6 104 (1/T) Chkn-nTA 86 À49 ln k 17.9 À 1.03 6 104 (1/T) Trk-nTA or 22 À201 ln k À1.25 À 2.67 6 103 pork-nTA (1/T)

Source: Based on data from Refs. 45, 46, and 60.

FIGURE 7 Simulated thermal inactivation pro®les for bovine muscle proteins and native-thermostable antigen from chicken (Chkn-nTA) or turkey (Trk-nTa). Samples are held at 20±958C for 30 minutes. (Calculated from data in Table 5.) Speciation of Meat Proteins 265 that for the other muscle proteins. The activation enthalpy change (DH#) and entropy change (DS#) for heat deactivation were large and positive for most of the muscle proteins. This is indicative of large conformational changes being the rate-limiting step during heat denaturation. For nTA we ®nd that DH# < 100 and DS# was negative. Low transition state parameters are characteristic of conformationally plastic proteins (53). Such proteins survive heat treatment owing to the ability to refold once the thermal stress is removed. A low DH# can also arise where bioactivity (antigenicity) resides in lower order (primary or secondary) protein structure. Simulated thermal inactivation pro®les for nTA and some other muscle proteins are compared in Fig. 7. I have assumed that (a) the Arrhenius equation (Table 5) applies over a temperature range of 20±1208C and (b) thermal deactivation kinetics are ®rst order (54). Caution is also warranted because the initial data (45,60) for modeling came from a limited temperature range of 54±668C for muscle proteins and 100±1208C for nTA.

5.4. End-Point Temperature Determination According to USDA guidelines, meat imported into the United States should be heated to a certain minimum end-point temperature (EPT) to ensure that it is free from pathogens and viruses. Compliance with the USDA guidelines on meat EPT can be assessed by ELISA (55±59). For example, Denise Smith and co-workers developed a sandwich ELISA for chicken or turkey skeletal muscle lactate dehydrogenase (LDH). Heart LDH was detected with 1000-fold lower af®nity. No cross-reactivity occurred with porcine or bovine skeletal muscle or heart muscle LDH. The limit of detection was 1 ng mLÀ1 or 1 ppb. Heating to an EPT of 68.3±728C inactivated muscle LDH. Didier and Annie Levieux (60) also developed immunological EPT indicators.

6. MONOCLONAL ANTIBODIES FOR MEAT SPECIATION

The production and use of pAbs for protein analysis has several disadvantages: (a) the process involves live animals and results in batch- to-batch variation in pAb, (b) there is a requirement for reagent standardization with respect to pAb concentration and af®nity for antigen, (c) there is limited pAb production from a single animal host, and (d) crude pAb requires purifying to lessen cross-reactivity. This process is technically demanding, slow, and expensive. Such disadvantages can be overcome with hybridoma technology for monoclonal antibody (mAb) production. 266 Chapter 9

Between 1989 and 1999 two research groups explored the use of mAbs for meat speciation. One group (from Spain) utilized af®nity-puri®ed (unheated) meat antigen for immunization. Hsieh's group at the Depart- ment of Nutrition and Food Science, Auburn University initially used (crude) thermostable meat antigens. The comparatively small number of publications describing ELISA with mAbs are summarized in Table 6.

6.1. Poultry (Chicken and Turkey) Af®nity-puri®ed chicken muscle antigen (28) (Section 4) was used for mAb production. Three hybridoma cell lines (designated AH4, BC9, and CF2) were identi®ed that produced mAb for chicken muscle antigen. SDS-PAGE and immunoblotting revealed that mAb-CF2 was speci®c for chicken muscle pyruvate kinase (58 kDa). Both mAb-AH4 and mAb-BC9 bound to a 47-kDa protein tentatively identi®ed as 3-phosphoglycerate kinase. Martin et al. (62) puri®ed mAb-BC9 by ion-exchange chromatography with a Mono-Q column for use as capture antibody. The detection antibody for their sandwich-ELISA was (rabbit) pAb for chicken muscle antigen. Visualization was by commercial HRP-labeled (goat) antibody for rabbit IgG. A sandwich ELISA using mAb-BC9 was selective for raw poultry (chicken and turkey) meat. No cross-reactivity was observed with beef, horse, pork, rabbit, or mutton. There was also no reaction with puri®ed proteins such as casein, gelatin, or soy protein. The sandwich ELISA test

TABLE 6 Monoclonal Antibody for the Speciation of Raw or Heated Meat Proteins by ELISA

Analysis, commentsa Reference

Chicken MartiÂ n et al. (61) Chicken Martin et al. (62) Porku,c Morales et al. (63), Chen et al. (41), Chen and Hsieh (42) Horse Garcia et al. (64) Chicken LDHc Abouzied et al. (55) Turkey LDHc Wang and Smith (57) Poultryc Sheu and Hsieh (40) Meat (beef, pork, etc.)c Hsieh et al. (43), Chen et al. (41) Central nervous system tissue O'Callaghan (65), Schmidt et al. (66) a c, Cooked or autoclaved meat, all other samples, and u, unheated samples. Speciation of Meat Proteins 267 responses were described by

A405 0:455 0:47 ln %Chicken 6

A405 0:404 0:46 ln %Chicken 7 for beef or pork adulteration by chicken, respectively. The linear dynamic range for analysis was 1±100% (w/w). The preceding assays are for raw meat samples only. Notice that the sandwich ELISAs using mAb are more sensitive than those with pAb [Eqs (4) and (5)].

6.2. Pork and Horse Meat A pork-speci®c mAb-DD9 did not react with beef, chicken, horse, casein, soy, gelatin, or BSA (63). The mAb-DD9 was prepared from unheated muscle protein antigen puri®ed by immunoaf®nity chromatography (26,27). Indirect ELISA utilizing mAb-DD9 detected beef or chicken adulteration by 1±100% (w/w) pork. The calibration response was described by

A405 0:0692 0:7989 ln %Pork 8

A405 0:0745 0:7612 ln %Pork 9 for beef and chicken samples, respectively. The LLD was 0.1% (w/w), which is below levels probably economically advantageous to the retailer. The assay sensitivity compares favourably with results expressed in Eq. (3). Heating meat samples to 658C for 30 minutes had no adverse on the assay reponse, but autoclaving samples at 1208C for 20 minutes led to loss of assay sensitivity. Chen and Hsieh (42) have recently described an mAb-based ELISA for detecting the presence of pork within cooked or processed meat products. The assay employs porcine thermostable antigen (Sec. 5.3). The LLD for pork was 0.5% (w/w) with intrassay and interassay precision of 5.8% and 7.9%. The highly accurate method was able to identify pork in 45 commercial processed meat samples. Sawaya and co-workers (67) also produced an ELISA test sensitive to cooked pork although they used pAb. Horse meat±speci®c mAb-DD3 (64) showed no cross-reactivity for beef, chicken, pork, soy proteins, casein, gelatin, or BSA. Addition of 0±50% (w/w) horse meat to beef led to the following ELISA response.

A405 0:4626 0:0314 %Horse 10 The LLD for horse meat was 2% (w/w). The Spanish group suggest that 268 Chapter 9 antigen puri®cation by af®nity chromatography may be necessary for mAb production.

6.3. Cooked Red Meat The mAb-2F8 had selectivity for beef, sheep, or lamb and reduced sensitivity for deer meat (43). There was no cross-reactivity with raw pork, beef, lamb, mutton, or deer. Chicken or turkey was not detected in either the raw or cooked state. Indirect ELISA using mAb-2F8 was speci®c for red meat. Adulteration of poultry (chicken and/or turkey) with 0.5±15% (w/w) beef, horse, sheep, or pork was easily detectable. The LLD for deer meat was * 5% (w/w) addition to poultry. Multiple adulterants produced a cumulative signal. Red meat is often more expensive than chicken. However, the preceding test will be useful for detecting the substitution of chicken meat with less valuable beef or pork trimmings (49). Chicken is also in higher demand in some parts of Asia where the consumption of beef is restricted by religious stricture.

7. FISH AND SEAFOOD IDENTIFICATION BY ELISA

Seafood adulteration may be growing in importance for three main reasons. First, there is a trend away from red meat toward the consumption of ®sh and seafood. Fish and seafood are perceived as healthy because of their higher content of unsaturated fatty acids. Second, world ®sh stocks continue to decline, increasing the economic incentive for adulteration. Third, improvements in processing technology have led to larger markets for comminuted ®sh products (68) and ®sh protein concentrates (69). These products are dif®cult to identify visually and offer signi®cant scope for adulteration. Speciation by immunological methods is seen as complementary to traditional methods such as isolectric focusing and SDS-PAGE (70). Suzuki (71) used an agar gel diffusion assay to identify tuna. Fish protein authentication using ELISA is discussed in this section.

7.1. Sardine and Tuna Taylor and Jones (72) and also Taylor et al. (73) described an indirect ELISA using (rabbit) pAb for soluble antigens from canned sardine, bonito, yellow®n tuna, or skipjack tuna. Crude (rabbit) pAb for canned ®sh antigen was nonspeci®c. High responses were obtained for most canned ®sh samples (Fig. 8). To improve the speci®city, pAb was puri®ed via (a) batch immunoadsorption, (b) immunoadsorption using antigens immobilized on Speciation of Meat Proteins 269

FIGURE 8 Noncompetitive indirect ELISA for the identi®cation of canned ®sh species. Graph shows the speci®city of crude (rabbit) antibody for different canned ®sh samples. (Based on data from Ref. 64.) magnetic beads, and (c) immunoaf®nity chromatography using the antigen immobilized on CNBr-activated Sepharose. Tuna species (albacore, yellow®n, and skipjack) were dif®cult to differentiate from each other or from bonito (a potential tuna substitute). Attempts to increase the speci®city of pAb for tuna species (by immunoadsorption) led to large reductions in the net concentration of free pAb. The indirect ELISA for canned ®sh was able to differentiate between ®sh and crustacea (prawn and scampi). The assay is potentially useful for detecting the adulteration of prawn, scampi, and other crustacean meat with cheaper ®sh (74).

7.2. Rock Shrimp SDS-PAGE analysis of seafood tissue extracts showed that protein M was unique to rock shrimp (Sicyonia brevirostris). The electrophoresis band corresponding to protein M was excised, homogenized with buffer, and centrifuged. The supernatant was dialyzed overnight and utilized as antigen for mAb production. Indirect ELISA using mAb-4H2±10D3 successfully 270 Chapter 9 identi®ed rock shrimp from three geographic locations in the United States. These samples were also correctly differentiated from 23 other seafood samples including white shrimp (from Colombia, Ecuador, Honduras, and Peru) and blue shrimp (from Ecuador). The speci®city for rock shrimp was attributed to the use of puri®ed antigen for mAb production. The test was more sensitive for heated samples probably because protein M was heated during isolation by SDS-PAGE. The antigen had molecular size of 17.7±18.5 kDa and made up *20% of the total water-soluble protein (75).

7.3. Red Snapper Adulteration of red snapper (Lutjanus campechanus) appears widespread. Approximately 14 out of 15 (91%) ®sh samples branded as red snapper by U.S. retailers were inaccurately labeled. Another survey involving 24 samples found 70% mislabeling. The high level of adulteration indicates genuine error on the part of retailers. Huang et al. (76) developed mAb-C1C1 and mAb-A1B1 for red snapper. The species-speci®c antigen (protein A, pI 5.93 ) was isolated by isoelectric focusing of red snapper muscle extract. Both mAb-C1C1 and mAb-A1B1 were applied for indirect ELISA. The detection system was (rabbit) pAb for mouse IgG conjugated to alkaline phosphatase. In tests involving 24 ®sh, seafood, and meat samples, indirect ELISA with mAb-C1C2 gave positive results for red snapper samples of vermilion snapper, lane snapper, mutton snapper, and yellowtail snapper. Using both mAb-C1C2± and mAb-A1B1±based ELISA tests simultaneously allowed a clear differentiation of red snapper samples from most other seafoods.

8. PERFORMANCE CHARACTERISTICS FOR DIFFERENT ELISA FORMATS

Factors affecting assay performance include (a) the ELISA format, (b) the type of enzyme label, (c) antibody characteristics, and (d) the nature of the antigen used. Calibration graphs for sandwich ELISA [Eqs (6) and (7)] or indirect ELISA [Eqs (8) and (9)] show that the indirect format is more sensitive. Using pAb for capture can lead to loss of assay response for sandwich ELISA. Multiple interactions involved in pAb-antigen binding may leave few epitopes for the detector antibody. Poor assay sensitivity also arises when a sandwich ELISA utilizes the same mAb for both capture and detection. Under such circumstances, highly speci®c epitopes become occupied by the capture mAb with few left for binding with the (same) mAb for detection. Comparing the sandwich ELISA tests for muscle antigen Speciation of Meat Proteins 271 shows that using mAb for capture enhances assay sensitivity by about an order of magnitude compared with the use of pAb. Finally, whether puri®ed antigens are necessary for mAb production is contentious. With mAb for unheated samples it seems necessary to use puri®ed antigen. However, thermostable antigens appear to induce speci®c mAb formation (see earlier).

9. MEAT TESTING FOR TRANSMISSIBLE SPONGIFORM ENCEPHALOPATHY AGENTS 9.1. Meat and Bone Meal Bovine spongiform encephalopathy (BSE) or ``mad cow disease'' is linked with feeding contaminated meat and bone meal to cattle. The ®rst recognized cases of BSE appeared in the United Kingdom in 1985±1986. New-variant Creutzfeldt-Jakob disease (nvCJD), thought to be a human form of BSE, was detected 10 years after the ®rst BSE cases (77). Transmission to other animals was demonstrated in laboratory studies with cats, pigs, goats, and sheep. The origin of BSE and other transmissible spongiform encephalopathies (TSEs) remains uncertain. One hypothesis highlights changes in meat rendering procedures in the late 1970s and early 1980s. The discontinuation of meat rendering processes involving hydro- carbon solvent±steam treatment may have led to a critical (even if slight) increase in the infectivity of BSE-contaminated feed. The theory, although compelling, is not necessarily accepted by all (78). The government of the United Kingdom banned the use of meat and bone meal in cattle feed in 1988. The incidence of BSE in the United Kingdom declined as a result of this ban, which was later extended to feed materials for all farm animals. However, large numbers of cattle remain affected with BSE. Infected cattle have also been discovered in the European Community countries including France and Belgium. Public health concerns about BSE seem likely to continue because the projected incubation period for nvCJD could extend from 4 to 30 years (79). Methods are being developed to monitor compliance with the legislation excluding meat and bone meal from feeds. Adequate heat treatment can also turn out safe meat and bone for use and/or disposal (80). The inactivation characteristics of the BSE agent were studied by Taylor and co-workers (79). BSE survives irradiation or boiling. Heating at 1218C (15 psi) leads to partial inactivation and thermal treatment at 1348C for 60 minutes produces complete inactivation. Meat-rendering plants in Germany heat treat samples at 1338C for 20 minutes at 3 bars of pressure. Klaus Hofmann used the Cortecs ELISA test kit for pork to test for bone and meat meal. Similar ELISA test kits exist for cooked beef or 272 Chapter 9 mutton. Samples of meat and bone meal heated to a rendering temperature of 1338C for 20 minutes were fully degraded and undetectable by ELISA (81,82). In the near future such tests might be used for routine monitoring of meat-rendering plants in Germany. Collaborative trials involving 21 laboratories from 12 countries (83) showed that these tests had suf®cient reliability as judged by a precision between 11 and 12%. Understerilized meat and bone meal were clearly distinguishable by a 10-fold greater assay signal.

9.2. Central Nervous System Tissue Concentrations of the BSE agent are higher in the central nervous system (CNS) tissue compared with peripheral nerves. A sandwich ELISA for CNS tissue was developed using mAb for GFAP (glial ®brillary acidic protein) (66). This protein is found only in astrocyte cells in the CNS.* The ELISA for GFAP had an approximate LLD of 1 ng (GFAP) with a linear range extending to 40 ng. The within-assay precision was 3.25±4%. While assay sensitivity remained constant for different matrices, the LLD increased in the order ground beef brain tissue > ground beef spinal cord tissue > puri®ed GFAP standards > brain > spinal cord. The ELISA gave the concentrations of GFAP in spinal cord tissue (55,000±220,000 ng mgÀ1 tissue), brain tissue (9000±55,000 ng mgÀ1 tissue), and cerebral cortex (17,000 ng mgÀ1 tissue). Neck muscle and ground beef were free of GFAP. The antigen is not very stable;therefore CNS tissue could be detected for only up to 8 days when samples were stored at 48C. CNS tissue was also analyzed using immunoblot analysis (84). The tests were directed at two antigens: GFAP and neuron-speci®c enolase. Immunoblot analysis did detect CNS tissue if samples were subjected to extreme temperature processing.

9.3. Direct Immunological Detection of the BSE Agent As of December 2000, the European Commission accepted ®ve direct tests for the BSE agent for further evaluation (85). The tests were produced by

* The details of the sandwich ELISA for GFAP were essentially as described elsewhere. (a) Coat microwell plates with a commercial pAb for GFAP (supplied by Dako Corporation, Carpentaria, CA) at 378C for 1 hour or at 48C overnight. (b) Block nonspeci®c microwell plate sites with PBST±powdered milk protein. (c) React pAb-coated microwells with GFAP standards or samples for 60 minutes. (d) Add diluted mAb for GFAP (supplied by Boehringer Mannheim, Indianapolis, IN). (e) Add enzyme-labeled antibody, i.e., alkaline phosphatase± labeled (rabbit) pAb for mouse IgG. Assay for enzyme activity. Speciation of Meat Proteins 273

ID Lalyated, Netherlands Imperial College of Science and Technology, UK Institute of Neurodegenerative Diseases, University of California, San Francisco Perkin Elmer Life Sciences, UK Prionics AG, Switzerland The BSE test produced by Prionics AG, Switzerland (Prionics AG, University of Zurich, 8057 Zurich, Switzerland) appears to be the favorite. Prionics-checkTM employs immunoblot analysis. Brain tissue extract is ®rst exposed to a protease solution followed by SDS-PAGE and then transferred to a nitrocellulose membrane by Western transfer. The membrane-bound proteins are detected immunologically using mAb speci®c for prion particles. The test is able to differentiate between the benign prion protein (PrPC) and the disease-causing PrPSc because the former is susceptible to protease attack but the latter is not. According to the advertising literature, Prionics-check is intended for (a) identi®cation of suspected BSE cases, (b) diagnostic testing in abattoirs and slaughterhouses, and (c) general monitoring for scrapie and BSE. Validation of the Prionics-Check tests has been documented (86). Prionics- check will be used for mandatory BSE testing in the European Union from 2001. Some important characteristics of the Prionic-checks include (a) high selectivity and speci®city, (b) ability to distinguish cattle with BSE from those with other neurological disease states, (c) detection of subclinical cases of BSE, (d) ease of use and availability in a kit form, (e) suitability for ®eld use, (f) high sample throughput (the time of analysis is reportedly 6±8 hours from tissue extraction to ®nal test results), and (g) current use for Swiss BSE surveillance for all sick and falling cattle. At this time, Prionics AG manufacture at least two prion-speci®c mAbs (6H4 and 34C9) as well as pAb (RO29). The antibodies are suitable for developing ELISA. The company also has available the full kit for BSE detection. The life science diagnostics company Bio-Rad Ltd. currently manufactures an ELISA test for BSE (PLATELIA2 BSE test). This test is commercially available in the United Kindom, France, Germany, Belgium, Luxembourg, Norway, Sweden, Switzerland, Italy, and Spain. At this time the test is not being sold in the United States (87). The list of companies now entering the BSE diagnostics market is growing rapidly as shown by the following list (88). 274 Chapter 9

Abbeymoy Ltd. Altegen Inc. Anonyx Inc. Bayer Caprion Pharmaceuti- Bio-Rad Inc. cals Inc. Boehringer-Ingelheim AG Centre Suisse Celcus Inc. d'EÂ lectronique et de Microtechnique SA Commissariat a l'EÂ nergie Enfer Scienti®c Ltd. Disease Sciences Ltd. Atomique Genesis Biventures IGEN International Microsens Biohase Ltd. Mary Jo Schmerr Paradigm Genetics Inc. Nen Life Science Products Inc. New York State Basic Proteome Sciences Ltd. Prion Developments Research Institute for Laboratory Neurological Disorders Prionics AG Q-One Biotech Ltd. V.I. Technologies Inc.

For further discussions of immunological tests for the BSE agent see Refs. 89 and 90.

REFERENCES

1. P Engvall, P Perlman. Enzyme-linked immunosorbent assay (ELISA). Quantitative assay of immunoglobin G. Immunochemistry 8:871±874, 1971. 2. BK Van Weemen, AHWH Shuurs. Immunoassay using antigen-enzyme conjugates. FEBS Lett 15:232±236, 1971. 3. E Ishikawa, T Kawai, K Miyai, eds. Enzyme Immunoassay. New York: Igaku-Shoin, 1981. 4. MN Clifford, The history of immunoassays in food analysis. In: BA Morris, MN Clifford, eds. Immunoassays in Food Analysis. Barking, Essex, UK: Elsevier Applied Science Publishers, 1985, pp 3±20. 5. CHS Hitchcock. Opportunities for developing food immunoassays. In: BA Morris, NM Clifford, R Jackman, eds. Immunoassays for Veterinary and Food AnalysisÐ1. Barking, Essex, UK: Elsevier Applied Science, 1988, pp 3±16. 6. JC Allen, CJ Smith. Enzyme-linked immunoassay kits for routine food analysis. Trends Biotechnol 5:193±199, 1987. 7. U Samarajeewa, CH Wei, TS Huange, RR Marshall. Applications of immunoassay in the food industry. CRC Crit Rev Food Sci Nutr 29:403± 434, 1991. 8. SS Gazzaz, BA Rasco, FM Dong. Applications of immunochemical assays to food analysis. CRC Crit Rev Food Sci Nutr 32:197±229, 1992. 9. BK Barai, RR Nayak, RS Singhal, PR Kulkarni. Approaches to the detection of meat adulteration. Trends Food Sci Technol 3:69±72, 1992. Speciation of Meat Proteins 275

10. HA Lee, MRA Morgan. Food immunoassays: application of polyclonal, monoclonal and recombinant antibodies. Trends Food Sci 4:129±134, 1993. 11. PE Hernandez, R Martin, T Garcia, P Morales, G Anguita, AI Haza, I Gonzales, B Sanz, Antibody-based analytical methods for meat species determination and detecting adulteration of milk. Food Agric Immunol 6:95± 104, 1994. 12. WJ Taylor, NP Patel, J Leighton-Jones. Antibody based methods for assessing seafood authenticity. Food Agric Immunol 6:305±314, 1994. 13. DM Smith. Immunoassay process control and speciation of meats. Food Technol 49(2):116±119, 1995. 14. W Simpkins, M Harrison. The state of the art in authenticity testing. Trends Food Sci 6:321±328, 1995. 15. UV Mandokhot, SK Kotwal. Enzyme linked immunosorbant assays in detection of species origin of meatsÐa critical appraisal. J Food Sci Technol India 34:369±380, 1997. 16. Luethy, J. Detection strategies for food authenticity and genetically modi®ed foods. Food Control 10:359±361, 1999. 17. CHS Hitchcock, FJ Bailey, AA Crimes, DAG Dean, PJ Davis. Determination of soya proteins in food using an enzyme-linked immunosorbent assay procedure. J Sci Food Agric 32:157±165, 1981. 18. RG Whittaker, TL Spencer, JW Copland, Enzyme-linked immunosorbent assay for meat species testing. Aust Vet J 59(4):125, 1982. 19. EK Kang'ethe, SJ Jones, RLS Patterson. Identi®cation of the species origin of fresh meat using an enzyme-linked immunosorbent assay procedure. Meat Sci 7:229±240, 1982. 20. RG Whittaker, TL Spencer, JW Copland. An enzyme-linked immunosorbent assay for species identi®cation of raw meat. J Sci Food Agric 34:1143±1148, 1983. 21. RM Patterson, RG Whittaker, TL Spencer. Improved species identi®cation of raw meat by double sandwich enzyme-linked immunosorbent assay. J Sci Food Agric 35:1018±1023, 1984. 22. SJ Jones, RLS Patterson. Double-antibody ELISA for detection of trace amounts of pig meat in raw meat mixtures. Meat Sci 15:1±13, 1985. 23. RM Patterson, TL Spencer. Differentiation of raw meat from phylogenically related species by enzyme-linked immunosorbent assay. Meat Sci 15:119±123, 1985. 24. SJ Jones, RLS Patterson. A modi®ed indirect ELISA procedure for raw meat speciation using crude anti-species antisera and stabilized immunoreagents. J Sci Food Agric 37:767±775, 1986. 25. J Pelly, RW Tindle. Meat species identi®cation: a semi-quantitative ELISA based test to con®rm the species of a given meat and to detect contamination of one meat species with another. Tech Ser Soc Appl Bacteriol 24:255±257, 1987. 276 Chapter 9

26. R Martin, JI Azcona, C Casas, PE HernaÂ ndez, B Sanz. Sandwich ELISA for detection of pig meat in raw beef using antisera to muscle soluble proteins. J Food Prot 51:790±794, 1988. 27. R Martin, JI Azcona, PE HernaÂ ndez, B Sanz. Parial puri®cation of porcine- speci®c soluble muscle proteins by immunoadsorption chromatography. Fleischwirtschaft 72:889±900, 1992. 28. R Martin, JI Azcona, J Tormo, PE HernaÂ ndez, B Sanz. Detection of chicken meat in raw mixtures by a sandwich enzymes immunoassay. Int J Food Sci Technol 23:303±310, 1988. 29. MK Ayob, AA Ragab, JC Allen, RS Farag, CJ Smith An improved, rapid, ELISA technique for detection of port in meat products. J Sci Food Agric 49:103±116, 1989. 30. A Stevenson, K Pickering, M Grif®n, Detection of chicken meat in raw meat mixtures by the double method of an enzyme immunoassay and an immunoblotting technique. Food Agric Immunol 6:297±304, 1994. 31. WJ Taylor, NP Patel, J Leighton-Jones. Antibody-based methods for assessing seafood authenticity. Food Agric Immunol 6:305±314, 1994 32. PD Warris. The residual blood content of meatÐa review. J Sci Food Agric 28:457±462, 1977. 33. NM Grif®ths, MJ Billington. Evaluation of an enzyme-linked immunosorbent assay for beef blood serum to determine indirectly the apparent beef content of beef joints and model mixtures. J Sci Food Agric 35:909±914, 1984. 34. R Martin, T Carcia, B Sanz, P Hernandez. Partial puri®cation of horse- speci®c soluble muscle proteins by immunoadsorbtion chromatography. J Sci Food Agric 58:447±449, 1992. 35. EK Kang'ethe, KJ Lindqvist. Thermostable muscle antigens suitable for use in enzyme immunoassays for identi®cation of meat from various species. J Sci Food Agric 39:179±184, 1987. 36. EK Kang'ethe, JM Gathuma. Species identi®cation of autoclaved meat samples using antisera to thermostable muscle antigens in an enzyme immunoassay. Meat Sci 19:265±270, 1987. 37. DN Gacheru, EK Kang'ethe, HFA Kaburia, FM Njeruh. Sandwich enzyme immunoassay for speciation of cooked meat and for detecting trace amounts of adulterants in phylogenically related species. East Afri Agric For J 59:205± 212, 1994. 38. SE Dierks, JE Buttler, HG Richerson. Altered recognition of surface- adsorbed compared to antigen-bound antibodies in the ELISA. Mol Immunol 23:403±411, 1986. 39. GC Varshney, W Mahana, AM Filloux, A Venien, A Paraf. Structure of native and heat-denatured ovalbumin as revealed by monoclonal antibodies: epitopic changes during heat treatment. J Food Sci 56:224±227, 233, 1991. 40. S-C Sheu, Y-H Hsieh. Production and partial characterization of monoclonal antibodies speci®c to cooked poultry. Meat Sci 50:315±326, 1998. Speciation of Meat Proteins 277

41. FC Chen, YHP Hsieh, RC Bridgman. Monoclonal antibodies to porcine thermal-stable muscle protein for detection of pork in raw and cooked meats. J Food Sci 63:201±205, 1998. 42. FC Chen, YHP Hsieh. Detection of pork in heat-processed meat products by monoclonal antibody-based ELISA. J Assoc Off Anal Chem Int 83:79±85, 2000. 43. YHP Hsieh, S-C Sheu, RC Bridgman. Development of a monoclonal antibody speci®c to cooked mammalian meats. J Food Prot 61:476±481, 1998. 44. E Rencova, I Svoboda, L Necidova. Identi®cation by ELISA of poultry, horse, kangaroo and rat muscle speci®c proteins in heat-processed meat samples. Vet Med (Prague) 45:353±356, 2000. 45. RG Berger, RP Mageau, B Schwab, RS Johnston. Detection of poultry and pork in cooked and canned meat foods by enzyme-linked immunosorbent assays. J Assoc Off Anal Chem 71:406±409, 1988. 46. CD Andrews, RG Berger, RP Mageau, B Schwab, RW Johnston. Detection of beef, sheep, deer, and horse meat in cooked meat products by enzyme- linked immunosorbent assay. J Assoc Off Anal Chem Int 75:572±576, 1992. 47. A Avrameas, T Ternynck. Biologically active water-insoluble protein polymers. I. Their use for isolation of antigens and antibodies. J Biol Chem 242:1651, 1967. 48. T Ternynck, S Avrameas. Polymerization and immobilization of proteins using ethylchloroformate and glutaraldehyde. Scand J Immunol Suppl 3:29± 35, 1976. 49. Y-HP Hsieh, BB Woodward, S-H Ho. Detection of species substitution in raw and cooked meats using immunoassays. J Food Prot 58:555±559, 1995. 50. K Hofmann, K Fishcher, E Mueller, W Babel. ELISA-test for cooked meat species identi®cation on gelatine and gelatine products. Nahrung 43:406±409, 1999. 51. GH Grant, JF Kachmar. The proteins of body ¯uids: plasma and serum proteins. In NW Tietz, ed. Fundamentals of Clinical Chemistry. London: WB Saunders, 1976, p 356. 52. D Levieux, A Levieux, A Venien. Immunochemical quanti®cation of heat denaturation of bovine meat soluble proteins. J Food Sci 60:678±684, 1995. 53. RKO Apenten, K Mahadevan. The heat resistance and conformational plasticity of Kunitz soybean trypsin inhibitor. J Food Biochem 23:209±224, 1999. 54. RKO Apenten. The effect of protein unfolding stability on their rates of irreversible denaturation. Food Hydrocolloids 12:1±8, 1998. 55. MM Abouzied, CH Wang, JJ Prestka, DM Smith. Lactate dehydrogenase as safe endpoint cooking indicator in poultry breast rolls: development of monoclonal antibodies and application to sandwich enzyme-linked immunosorbent assay (ELISA). J Food Prot 56:120±124, 1993. 56. CH Wang, JJ Pestka, AM Booren, DM Smith. Lactate dehydrogenase, serum protein, and immunoglobulin G content of uncured turkey thigh rolls as 278 Chapter 9

in¯uenced by endpoint cooking temperature. J Agric Food Chem 42:1829± 1833, 1994. 57. CH Wang, DM Smith. Lactate dehydrogenase monoclonal antibody immunoassay for detection of turkey meat in beef and pork. J Food Sci 60:253±256, 1995. 58. DM Smith, LD Desrocher, AM Booren, CH Wang, MM Abouzied, JJ Pestka, J Veeramuthu. Cooking temperature of turkey ham affects lactate dehydrogenase, serum albumin and immunoglobulin G as determined by ELISA. J Food Sci 61:209±212, 1996. 59. DS Smith, LD Desrocher. Immunoassays for determination of endpoint processing temperatures in poultry and beef products. J Muscle Foods 7:335± 344, 1996. 60. D Levieux, A Levieux. Immunochemical quanti®cation of myoglobin heat denaturation: comparative studies with monoclonal and pAb. Food Agric Immunol 8:111±120, 1996. 61. R MartiÂ n, RJ Wardale, SJ Jones, PE HernaÂ ndez, RLS Patterson. Production and characterization of monoclonal antibodies speci®c to chicken muscle soluble proteins. Meat Sci 25:199±207, 1989. 62. R MartiÂ n, RJ Wardale, SJ Jones, PE HernaÂ ndez, RLS Patterson. Monoclonal antibody sandwich ELISA for the potential detection of chicken meat in mixtures of raw beef and pork. Meat Sci 30:23±31, 1991. 63. P Morales, T GarciÂ a, I GonzaÂ lez, R MartiÂ n, B Sanz, PE HernaÂ ndez. Monoclonal antibody detection of porcine meat. J Food Prot 57:146±149, 1994. 64. T GarciÂ a, R MartiÂ n, P Morales, AI Haza, G Anguita, I GonzaÂ lez, B Sanz, PE HernaÂ ndez. Production of a horse-speci®c monoclonal antibody and detection of horse meat in raw meat mixtures by an indirect ELISA. J Sci Food Agric 66:411±415, 1994. 65. JP O'Callaghan. Quantitation of glial ®brillary acidic protein: comparison of slot-immunoassay with a novel sandwich ELISA. Neurotoxicol Teratol 13:275±281, 1991. 66. GR Schmidt, KL Hossner, RS Yemm, DH Gould, JP O'Callaghan. An enzyme-linked immunosorbent assay for glial ®brillary acidic protein as an indicator of the presence of brain and spinal cord in meat. J Food Prot 62:394±397, 1999. 67. WN Sawaya, MS Mameesh, E El-Rayes, A Husain, B Dashti. Detection of pork in processed meat by an enzyme-linked immunosorbent assay using antiswine antisera. J Food Sci 55:293±297, 1990. 68. V Venugopal. Mince from low cost ®sh species. Trends Food Sci Technol 3:2± 5, 1992. 69. SR Tannenbaum, BB Stillings, NS Scrimshaw, eds. The Economics, Market- ing, and Technology of Fish Protein Concentrate. Cambridge, MA: MIT Press, 1974. 70. W Sidwell. Fish speciation by immunochemical techniques. In: MRA Morgan, CJ Smith, PA Williams, eds. Food Safety and Quality Assurance. Applica- Speciation of Meat Proteins 279

tions of Immunoassay Systems. Barking, UK: Elsevier Science Publishers, 1992, pp 49±54. 71. A Suzuki. Recovery of species-speci®c antigenicity from heat denatured serum protein of tuna. Bull Jpn Soc Sci Fish [Nihon Suisan Gakkai Shi] 41:373, 1975. 72. WJ Taylor, JL Jones. An immunoassay for verifying the identity of canned sardines. Food Agric Immunol 4:169±175, 1992. 73. WJ Taylor, NP Patel, J Leighton-Jones. Antibody-based methods for assessing seafood authenticity. Food Agric Immunol 6:305±314, 1994. 74. WJ Taylor, JL Jones. An immunoassay for distinguishing between crustacean tailmeat and white ®sh. Food Agric Immunol 4:177±180, 1992. 75. H An, PA Klein, K-J Kao, MR Marshall, MWS Otwell, C We. Development of monoclonal antibody for rock shrimp identi®cation using enzyme-linked immunosorbent assay. J Agric Food Chem 38:2094±2101, 1990. 76. T-S Huang, MR Marshall, K-J Kao, WW Otwell, C Wei. Development of monoclonal antibodies for red snapper (Lutjanus campechansus) identi®cation using enzyme-linked immunsorbent assay. J Agric Food Chem 43:2301±2307, 1995. 77. RG Will, JW Ironside, M Zeidler, SN Coursens, K Estibeiro, A Alperovitch, S Poser, M Pocchiary, A Hofman, PG Smith. A new variant Creutzfeldt-Jakob disease in the UK. Lancet 347:921±925, 1996. 78. P Brown. On the origins of BSE. Lancet 352:252±253, 1998. 79. D Taylor. BSE: our future in the balance. Chem Ind (June):444±447, 1998. 80. P Brown. BSE: the ®nal resting place. Lancet 351:1146±1147, 1998. 81. K Hofmann. Proof of proper heating at meat-and-bone meals. Fleisch- wirtschaft 76:1037±1039, 1996. 82. K Hofmann, K Fischer, E Mueller, V Kemper. Experiments to demonstrate the effectiveness of heat treatments applied to canned meats and meat-and- bone meals. Fleischwirtschaft 76:920±923, 1996. 83. C von Holst, KO Honickel, W Unglaug, G Kramer, E Anklam. Determina- tion of an appropriate heat treatment of animal waste using the ELISA technique: results of a validation study. Meat Sci 54:1±7, 2000. 84. EH Luecker, E Eigenbrodt, S Wenisch, R Leier, M Buelte. Identi®cation of central nervous system tissue in retail meat products. J Food Prot 63:258±263, 2000. 85. European Commission. DG24. Directorate B, Unit B3. The evaluation of tests for the diagnosis of transmissible spongiform encephalopathy in bovines (8 July 1999). http:/europe.eu.int./comm/dg24/health/bse. 86. O Schaller, R Fatzer, M Stack, J Clark, W Cooley, K Bif®ger, S Egli, M Doherr, M Vandevelde, D Heim, O Oesch, M Moser. Validation of a Western immunoblotting procedure for bovine PrPSc detection and its use as a rapid surveillance method for the diagnosis of bovine spongiform encephalopathy (BSE). Acta Neuropathol 98:437±443, 1999. 87. http://www.fda.gov/oc/opacom/hottopics/bse.html#3anchor. 88. Prion Disease Diagnostics. http://www.mad-cow.org/00/feb01_last.html. 280 Chapter 9

89. LKJ van Keulen, JPM Langeveld, GJ Garssen, JG Jacobs, BEC Schreuder, MA Smits. Diagnosis of bovine spongiform encephalopathy: a review. Vet Q 22:197±200, 2000. 90. D Momocilovic, A Rasooly. Detection and analysis of animal materials in food and feed. J Food Prot 63:1602±1609, 2000. 10

Speciation of Soya Protein by Enzyme-Linked Immunoassay

1. INTRODUCTION

There are usually guidelines for adding plant and other nonmeat protein to meat products. Food technologists use such ingredients legitimately to enhance functional properties such as water holding, fat binding, and gelation. Nevertheless, levels of nonanimal proteins in meat should be monitored. Much research has appeared in connection with soybean protein, this being the most important nonmeat protein ingredient. This chapter describes immunological methods for detecting bulk quantities (>0.5% w/w) of soybean protein in meat and meat products. The topic is dominated by methods of sample pretreatment designed to ensure accurate results no matter the sample processing history.

2. SAMPLE PRETREATMENT AND ANALYSIS OF SOY PROTEIN

Hitchcock et al. (1) were ®rst to use ELISA for soy protein analysis. The assay was designed for a wide range of commercial soy samples including ¯our, protein isolates, and texturates. To correct for variable (heat) processing history, samples are pretreated with 8 M urea. Denatured soy

281 282 Chapter 10 protein is then renatured before analysis by competitive indirect ELISA. Details of this pretreatment regime are given later (Method 1). Develop- ments leading to the eventual commercialization and of®cial approval for soybean protein ELISA tests are summarized in Table 1. With indirect ELISA (1),* the ®nal absorbance measurement is inversely related to the concentration of soy protein. The assay was highly speci®c with negligible responses toward beef, milk, ®eld bean, or wheat proteins. Speci®city was for conglycinin, which is the 7S soybean globulin (Fig. 1). The precision for soy protein determination was 10.5%. Grif®ths et al. (3) found that a commercial pAb for native and/or heat-denatured soy protein was as effective as the laboratory-developed (rabbit) pAb for renatured soy protein (Fig. 2). Collaborative testing of a commercial ELISA kit for soya protein involved 13 laboratories from the United Kingdom (4,7). The kits were supplied by Biokits Ltd. (Newtech Square, Deeside Industrial Park, Deeside, Clwyd CH5 2NU, UK), who also organized a 1-day workshop to familiarize trial members with the test procedures. For the actual trial,

TABLE 1 Determination of Soybean Proteins by ELISA

Analysis and comment Reference

First ELISA for soya protein Hitchcock et al. (1), Grif®ths et al. (2) ingredients (¯our, isolates, concentrates, and extrudates) Commercial pAb for soy protein Grif®ths et al. (3) ELISA Commercial pAb for soy protein in Crimes et al. (4) ELISA, collaborative study Europe-wide collaborative study Olsman et al. (5) ELISA of soya protein by Ravestein and Driedonks (6) immunoblotting Collaborative study of ELISA kit Hall et al. (7) AOAC approval for ELISA for soy McNeal (8) proteins Detection of soy milk in bovine milk Hewedy and Smith (9)

* The following steps are involved: (a) Pretreat soy sample and standards by denaturation- renaturation protocol. (b) React soy samples or standards with (rabbit) pAb for soy protein. (c) Add mixture to a microwell plate precoated with bound (renatured) soy protein. (d) Incubate to allow antibody-antigen reaction. (e) Wash thoroughly with PBST. (f) Add enzyme-labeled pAb for rabbit IgG. (g) Wash thoroughly with PBST. (h) Add enzyme substrate. (i) Stop reaction after a ®xed incubation time and record absorbance readings. Speciation of Soya Protein 283

FIGURE 1 Competitive indirect ELISA for soya bean protein using commercially available (rabbit) polyclonal antibodies (1) or experimental (rabbit) polyclonal antibodies for renatured soya bean protein. (Drawn from Refs. 1 and 3.)

FIGURE 2 Effect of cooking temperatures on soy protein analysis in beefburgers using competitive indirect ELISA. (Drawn from Refs. 1 and 3.) 284 Chapter 10 participants analyzed duplicate samples of raw beefburger mix (three samples), raw sausages (two samples), cooked pate (three samples), and a trial sample. The average LLD was 0.7% (w/w) soy protein. The percent protein detected (accuracy) was 89 (+6.1)% (7). A soybean ELISA kit was also evaluated by Rittenburg et al. (10). The accuracy depended on the type of soy additive, type of food matrix, and subsequent processing. Protein recovery was 93 (+15.5), 75 (+7.5), 81 (+8.5), or 82 (+8.8)% for soy isolate, ¯our grit, concentrate, or texturate, respectively. For 72 retail meat products (including beef and/or pork sausages, bacon and ham loaf, and mince), the average recovery for soy protein was 91 (+12.3)% when added at levels of 1.2±1.6% (soy protein isolate), 2.4±5% (soy ¯our), or 4.8% (soy texturate). Autoclaving (1218C, 15 psi;20 minutes) produced a signi®cant decline in accuracy (Fig. 2). Overall, the performance of the commercial ELISA kit was deemed acceptable. ELISA results agreed with the soy protein levels declared by most manufacturers. Method 1 Denaturation-renaturation sample treatment for soy protein antigens (1,4,7). Samples or soy protein standards are dissolved with hot urea- mercaptoethanol* solution. This unfolds proteins and destroys S22S bonds. The sample is then transferred to a renaturation buffer. This treatment encourages protein refolding and restores soy proteins to a baseline conformation regardless of their previous thermal history. Reagents 1. Urea 2. Dithiothreitol (DTT) 3. Tris-HCl buffer (0.25 M, pH 8.6) Procedure Denaturation-extraction buffer (urea 10.6 M, DTT 18.8 mM in *25 mM Tris-HCl, pH 8.6). To a 250-mL volumetric ¯ask add 80 g of urea, 20 mL of Tris-HCl buffer (0.25 M, pH 8.6), and warm to dissolve. Add 30 mg of DTT to the hot solution, dissolve, and keep solution in a 1008C water bath. Add 60 mL of distilled water. Renaturation buffer (0.06 M NaCl with 7.5 mM L-cystine). Dissolve 1.8 g of L-cysteine with sodium hydroxide (1.0 M, 20 mL). Add the solution to 900 mL of 0.06 M NaCl solution. Adjust to pH 9 with 1 M HCl and make up to 1000 mL.

* Mercaptoethanol (2-ME) was later replaced with the less odorous dithiothreitol (DTT). Speciation of Soya Protein 285

Sample extraction. Homogenize 12 g of ®nely chopped (meat) sample with 48 mL of Tris-HCl buffer (0.05 M, pH 8.6) using an Untra- Turax homogeneizer. Add 2.5 g of meat homogenate or 40 mg of soy protein standard to a 50-mL ¯ask. Add 7.5 mL of urea-DTT solution, mix, and heat at 1008C for 60 minutes. Transfer the mixture to a 508C water bath. Renaturation. Add renaturation buffer (20 mL, prewarmed to 508C), mix, and then allow to cool to room temperature. Bring the ®nal volume to 100 mL and ®lter through Whatman No. I paper. Collect the ®rst 10 mL of ®ltrate for ELISA. Meat samples suspected of containing undeclared soy protein were extracted with acidic ethanol, followed by acetone, and then air dried to produce acetone powder. These samples and soy antigen (for immunization or precoating microwell plates) were dissolved using urea-DTT solvent and renatured as just described. In theory, the renaturation treatment transforms all soy protein to a baseline renatured state with reconstituted antigenic determinants. In practice, the recovery of antigenicity is found to be about 20% for 11S soy globulin and 70% for the 7S soy protein. Reasons for this are twofold. First, severe heat treatment leads to covalent modi®cation of protein side chains. The chemical changes (lysinoanaline formation, cross-link formation, deamidation, etc.) cannot be reversed by the renaturation procedure. Second, protein refolding is usually less than 100% ef®cient. Competing side reactions result in the formation of protein aggregates and misfolded structures with improperly aligned S22S bonds. The ranaturation procedure reverses protein sulfhydryl/-disul®de exchange and noncovalent interactions produced during food processing.

3. STRUCTURE, DENATURATION, AND RENATURATION OF SOYBEAN PROTEINS 3.1. Soy Protein Structure Soya beans contain between 40 and 50% protein by dry weight. The major protein groups are (a) storage proteins (70±80% total), (b) enzymes (notably lipoxygenase, lactate dehydrogenase), (c) protease inhibitors (notably the Bowman-Birk and the Kunitz inhibitors), and (d) other storage proteins (e.g., lectin). The foremost storage protein (*40% total protein) is 11S glycinin. It is a member of the leguminin family of proteins. The second storage protein (*30% total protein) is the 7S conglycinin. Glycinin (molecular mass of 350 kDa) is a hexamer of A-B subunits. The six acidic (A) and basic (B) subunit pairs are each joined by a disul®de 286 Chapter 10 bond. The A-B subunit is synthesized as a single polypeptide chain containing an intermolecular disul®de bond and a linker sequence. Posttranslational proteolysis removes the linker sequence, leaving the A and B subunits linked by a disul®de bond. Isoelectric focusing reveals ®ve A- subunit isoforms (A1,A2,A3,A4,A5). There are at least four B-chain isoforms (B1,B2,B3,B4). The structural characteristics of the glycinin were reviewed by Peng et al. (11), Nielsen (12), and Fukushima (13). Conglycinin (170 kDa) is a trimer. From the structural model proposed by Thanh and Shibasaki (14), conglycinin has six isomers. They are produced from the random combination of four subunits; a, a0, b, and g. Conglycinin contains about 5% carbohydrate but is virtually devoid of cysteinyl residues or S22S bonds (15).

3.2. Thermal Denaturation The effects of heating on soy proteins were extensively investigated by nonimmunological techniques (16±21). Turbidimetric measurements showed that heating glycinin at 1008C (0.5% w/v in potassium phosphate buffer, I 0.5) led to aggregation. The rate of aggregation increased (while the net aggregation decreased) in the presence of 2-ME as a source of free sulfhydryl groups. The kinetics of aggregation conformed to a reaction order of 1.2. The Arrhenius plot for the reaction was biphasic between 70 and 908C (16). Studies using turbidity and UV difference measurements do not differentiate between intermolecular and intra-intermolecular effects such as dissociation, unfolding, and aggregation. Precise thermal stability data were obtained with differential scanning calorimetry (DSC). Glycinin (7±10% w/v) had a denaturation temperature

(TD) of 85±948C, whereas the TD for conglycinin was 10±128C lower. The TD increased by a maximum of 308C with increasing salt (0±2 M NaCl) concentration for both glycinin and conglycinin (22±25). The ratio of calorimetric and Van't Hoff enthalpies was *1, implying that seed globulins denature via a one-step (all or nothing) process (24,25). By contrast, the kinetics of conglycinin and glycinin thermal denaturation is biphasic with separate dissociation and protein unfolding steps (26,27). Apparently, DSC studies do not register changes in soybean protein quaternary structure. The thermal dissociation of 11S glycinin leads to a 7S trimer and then to individual (3S) A-B subunits. Glycinin trimer has a sedimentation number equal to that of native 7S conglycinin and is therefore designated 7sÏ. Further heating produces disul®de bond lysis and the separation of the A and B subunits. The hydrophobic B subunit forms aggregates, leaving the A units in solution. From the many studies of glycinin denaturation comes the Speciation of Soya Protein 287 following commonly accepted scheme below.

11S 7s 3S A subunits B subunits ; B Subunits sol-aggregates 1 ; B Subunits solid-aggregates

Processes in Equation (1) will be modi®ed by soy protein interactions with muscle proteins, lipid, rusk, and other ingredients. Indeed, TD for soy protein depends on the moisture level, ionic strength, pH, and the presence of sulfhydryl compounds including other meat proteins. Within a defatted

¯our matrix containing about 60% (w/w) moisture, TD for glycinin was 908C. The TD value increased to 1608C for absolutely dry ¯our (28,29).

3.3. Effect of Heat Treatment on Glycinin Structure and Antigenicity Preheating soy protein for 60 minutes in a high-ionic-strength medium (0.035 M phosphate buffer, pH 7.6 with 0.4 M NaCl) at 30±808C did not impair pAb binding with glycinin* at room temperature (30,31). By contrast, the thermal treatment of glycinin dissolved in low-salt solvent (0.035 M phosphate buffer, pH 7.6, with 0.15 M NaCl) produced a total loss of antigenicity (32). Apparently, pAbs for native glycinin bind to the nondissociated protein [Eq. (1)]. The antibody recognizes discontinuous epitopes formed by the intact protein. High salt concentrations stabilize glycinin. Thermal treatment at 1008C (5 minutes) led to a total loss of glycinin antigenicity in the presence of 0±0.7 M salt. However, the residual antigenicity increased in the presence of 0.7±2.0 M NaCl. Glycinin retained 90% antigenicity at the highest salt concentrations tested (33). Exposure to extremes of pH (pH < 2andpH > 10) also diminished glycinin antigenicity. High acidity or basicity is known to cause the dissociation of glycinin (34). The order of anitgenicity for soy protein preparations (with a rabbit host) was 11S 4 A1a*A2 > A3

* The effect of heating on glycinin antigenicity was examined by agar gel double immunodiffusion (AGID) or single radial immunodiffusion (SRID). The tests involved crude (rabbit) pAb for native glycinin, puri®ed by immunoadsorption using ethylchloroformate cross-linked glycinin (see Chapter 8 for a description of this method). AGID assay for soy proteins showed speci®city for glycinin with no cross-reactivity for whey proteins or conglycinin. 288 Chapter 10

* B1 (35). No antibodies formed against A4,B3, and B4 glycinin subunits despite multiple injections. Gel immunodiffusion assays con®rmed that pAb for native glycinin did not react with isolated A or B glycinin subunits.

3.4. Effect of Heat Treatment on Conglycinin Structure and Antigenicity Immunoanalytical investigations reveal that b-conglycinin is more heat resistant than glycinin. Heating conglycinin (in 0.035 M K-phosphate buffer with 0±0.1 M NaCl) at 1008C for 5 minutes produced 50±75% retention of antigenicity (compared with zero for glycinin). Conglycinin subunits and soluble aggregates produced by thermal treatment or native subunits (a, a0, and b) isolated by ion-exchange chromatography are able to react with pAb for the native protein (36). These results are consistent with one or more of the following conclusions: (a) antigenic sites for conglycinin are continuous, i.e., involve a contiguous sequence of amino acids formed from lower order (18 and 28) protein structure;(b) thermal dissociation of conglycinin does not involve large changes in conformation of subunits;and (c) antigenic sites for conglycinin (subunits) are surface located and unaffected by the protein association-dissociation transition (37). Even though it has a lower TD, b-conglycinin is apparently less vulnerable to thermal processing than glycinin.

3.5. Renaturing Ef®ciency of Glycinin and Conglycinin The antigenicity of soy protein is dominated by the 7S protein (Fig. 1). This is the result of a more ef®cient renaturation of conglycinin compared with glycinin. After exposure to urea, extreme pH or high temperature, glycinin dissociates into its constituent subunits (37±39). German et al. (19) showed that heating then disrupts S22S bonds between A-B subunits. Disul®de bond cleavange is catalyzed by indigenous free sulfhydryl compounds associated with soy protein. Isolated glycinin B subunits then aggregate via SH/S22S exchange while the A chains remain soluble (20). On the other hand conglycinin does not aggregate extensively upon heating. This protein lacks S22S bonds and also has low levels of SH groups. Extensive heating eventually produces soluble conglycinin aggregates although SDS-PAGE analysis shows little irreversible damage. Conglycinin and glycinin refold with *80% and *70% ef®ciency after exposure to 8 M urea and dialysis, respectively. However, treatment with urea±2-ME (which disrupts A-B disul®de bonds) leads to a glycinin renaturing ef®ciency of *20%. Breaking the A-B disul®de linkage produces a marked reduction in glycinin renaturation ef®ciency and enhanced Speciation of Soya Protein 289 aggregation. A small proportion of glycinin molecules also form incorrectly refolded soluble monomers with altered antigenic properties. Reconnecting S22S bonds in a correct orientation is the limiting step for glycinin renaturation (39). According to Berkowitz and Webert (40), the renaturation procedure for soybean products, although long and laborious, is inescapable. They proposed using SDS-PAGE analysis and immunoblotting as a more rapid ELISA format. During immunoblotting, soybean protein is denatured, separated by SDS-PAGE, and transferred to a nitrocellulose membrane before immunoassay. There is no distinct sample renaturation step, which reduces the assay time.

4. SOLVENT-EXTRACTABLE SOYBEAN PROTEIN

There may be other more convenient methods for preparing soy protein for analysis. Medina (41) ultrasonicated soy protein standards (1±2 mg) with 10±20 mL of coating buffer (3.2 mM sodium carbonate, pH 9.8 0.1% thimerosol and 0.05% Tween 20). Samples of cooked sausage (1 g) were similarly homogenized with 10 mL of carbonate buffer and subjected to untrasonication and (1000-fold) dilution followed by ®ltration. Results using soy protein fractions showed pAb binding to glycinin A subunits. Pretreating commercial soy protein isolate with reducing agent increased antibody binding. Assay precision was improved by increasing the time for pAb coating on microwell plates. The linear dynamic range was 0±0.2 mg soy protein per well, or 0±5% (w/w) soy protein in cooked sausages. For laboratory-prepared sausages the amount of soy protein found (Y, %) was described by the relation Y %1:22 1:09X 2 where X is the amount of soy protein known to be present in samples. From the gradient in Equation (2) there was a quantitative recovery of soy protein antigen from the cooked sausage matrix.

5. THERMOSTABLE ANTIGENS FOR SOYBEAN PROTEIN ANALYSIS

ELISA for soy products would be improved by using thermostable antigens that are unaffected by cooking and other forms of processing. Possible 290 Chapter 10 thermostable antigens include protease inhibitors, peptide antigens (natural or synthetic), and conglycinin.

5.1. Protease Inhibitors The Bowman-Birk inhibitor is a possible thermostable antigen for ELISA of soybean additives (42,43). Kunitz soybean trypsin inhibitors (KSTIs), which rival the Bowman-Birk inhibitor in thermal resistance (44), are another potential thermostable antigen (45).

5.2. Peptide Antigens Peptide antigens with continuous epitopes are potentially useful thermostable antigens. Yasumoto, et al. (46) developed a noncompetitive indirect ELISA for soybean protein predigested with trypsin. The assay was highly speci®c with no cross-reactivity for pork, beef, egg, or azuki bean proteins. Soya protein (10.4±20.8% w/w) was quantitatively determined for pork sausages cooked at 808C for 20 minutes. The LLD was 0.4%. The pAbs used for this assay were raised by immunizing rabbits with intact glycinin. As a form of pretreatment, glycinin standards were autoclaved at 1208C for 180 minutes and then digested with trypsin for 24 hours. Microwell plates were coated with the peptide digest and blocked with BSA to avoid nonspeci®c binding. Food samples, e.g., cooked sausages, were pretreated by homogenization with acidic ethanol, followed by acetone precipitation and drying. The powders were suspended in buffer, autoclaved for 1208C for 180 minutes, and digested with trypsin for 24 hours at 378C. Proteolysis was terminated by boiling brie¯y. The supernatant was removed for indirect ELISA. Carter et al. (47) used glycinin peptides as antigens for pAb and mAb production. The absence of discrete protein bands during SDS-PAGE analysis con®rmed a complete hydrolysis of glycinin (1 mg mLÀ1)by treatment with 20 mg of subtilisin (in ammonium bicarbonate buffer, 50 mM, pH 8) overnight. The (rabbit) pAbs for glycinin peptides were nonspeci®c (Fig. 3). Avidity for intact glycinin was attributed to surface- located antigenic determinants. There was cross-reactivity with other seed globulins (b-conglycinin, pea 11S globulin), probably due to partial amino acid sequence homology. Three mAbs (designated mAbs IFRN024, IFRN025, and IFRN026) were also produced with interesting speci®city characteristics (Fig. 4). The mAb-IFRN024 was nonspeci®c and recognized an epitope that was susceptible to denaturation by SDS treatment. Both mAb-IFRN025 and mAb-IFRN026 were speci®c for both intact glycinin and glycinin peptides. The former mAb showed a 100-fold greater avidity Speciation of Soya Protein 291

FIGURE 3 Indirect ELISA for soya bean and other seed globulins using polyclonal antibodies for glycinin peptides. (Drawn from Ref. 47.)

FIGURE 4 Indirect ELISA for soybean and other seed globulins using mAbs for glycinin peptides. Results for mAb-IFRN0024 and mAb-IFRN0026 were multiplied by 10 before plotting. (Drawn from Ref. 47.) 292 Chapter 10 for these samples. The epitopes recognized by mAb-IFRN025 and mAb- IFRN026 were resistant to combined SDS and thermal treatment. One should expect that peptide antigens will be resistant to processing effects owing to their small size.

5.3. Conglycinin Structural changes for thermally treated conglycinin were monitored using mAb-IFRN089 (48). The antigenicity of b-conglycinin increased after preheating to 658C, which is the TD for this protein. Huang et al. (49) also monitored heating effects on glycinin using mAb-IFRN025.* Thermal treatment led to progressive ``denaturation'' at temperatures above 908C and an increase in protein antigenicity (49). The insoluble protein precipitate was not analyzed. In summary, the use of mAbs for continuous epitopes should lead to ELISAs that are suitable for processed foods. Continuous epitopes comprising consecutive amino acids (protein 18 structure) are more thermoresistant than conformational or discontinuous epitopes formed from higher order protein (48,38,or28) structure. Beyond these considerations, sample solubility may still be a limiting factor. Ef®cient strategies are needed for resolubilizing proteins that have undergone severe processing.

6. OTHER NONMEAT PROTEINS

Nonmeat protein additives from cereal sources, milk, and egg have been analyzed by ELISA. However, there is far greater interest in the detection of trace amounts of these proteins in relation to their ability to cause allergic reactions. The analysis of protein allergens is discussed in Chapter 11.

REFERENCES

1. CHS Hitchcock, FJ Bailey, AA Crimes, DAG Dean, PJ Davis. Determination of soya proteins in food using an enzyme-linked immunosorbent assay procedure. J Sci Food Agric 32:157±165, 1981. 2. NM Grif®ths, MJ Billington, W Grif®ths. A review of three modern techniques available for the determination of soya protein in meat products. J Assoc Public Anal 19:113±119, 1981.

* Glycinin (4 mg mLÀ1 in 0.35 mM K-phosphate buffer 0.1 M NaCl) was heated and then centrifuged to remove insoluble aggregates. The soluble fraction was analyzed by ELISA. Speciation of Soya Protein 293

3. NM Grif®ths, MM Billington, AA Crimes, CHS Hitchcock. An assessment of commercially available reagents for an enzyme-linked immunosorbent assay (ELISA) of soya protein in meat products. J Sci Food Agric 35:1255±1260, 1984. 4. AA Crimes, CHS Hitchcock, R Wood. Determination of soya protein in meat products by an enzyme-linked immunosorbent assay procedure: collaborative study. J Assoc Public Anal 22:59±78, 1984. 5. WJ Olsman, S Dobblelaere, CHS Hitchcock. The performance of an SDS- PAGE and an ELISA method for the quantitative analysis of soya protein in meat products: an international collaborative study. J Sci Food Agric 36:499± 507, 1985. 6. P Ravestein, RA Driedonks. Quantitative immunoassay for soya protein in raw and sterilized meat products. J Food Technol 21:19±32, 1986. 7. CC Hall, CHS Hitchcock, R Wood. Determination of soya protein in meat products by a commercial enzyme immunoassay procedure. Collaborative trial. J Assoc Public Anal 25:1±27, 1987. 8. JE McNeal. Semi-quantitative enzyme-linked immunosorbent assay of soy protein in meat products: summary of collaborative study. J Assoc Anal Chem 71:443, 1988. 9. MM Hewedy, CJ Smith. Modi®ed immunoassay for the detection of soy milk in pasteurized skimmed bovine milk. Food Hydrocolloids 3:485±490, 1990. 10. JH Rittenburg, L Adams, J Palmer, JC Allen. Improved enzyme-linked immunosorbent assay for determination of soy protein in meat products. J Assoc Off Anal Chem 70:583±587, 1987. 11. IC Peng, DW Quassa, WR Dayton, CE Allen. The physicochemical and functional properties of soybean 11S globulinÐa review. Cereal Chem 61:480±489, 1984. 12. NC Nielsen. The chemistry of legume storage proteins. Philos Trans R Soc Lond Ser B 304:287±296, 1984. 13. D Fukushima. Structures of plant storage proteins and their functions. Food Rev Int 7:353±381, 1991. 14. VH Thanh, K Shibasaki. Beta-conglycinin from soybean proteins. Isolation and immunological and physical properties of the monomeric forms. Biochim Biophys Acta 490:370±384, 1977. 15. MC Garcia, M Torre, ML Marina, F Laborda. Composition and characterization of soyabean and related products. CRC Rev Food Sci Nutr 37:361± 391, 1997. 16. WJ Wolf, T Tamura. Heat denaturation of soybean 11S protein. Cereal Chem 46:331±402, 1969. 17. K Hashizume, T Watanabe. In¯uence of heating temperature on conformational changes of soybean proteins. Agric Biol Chem 43:683±690, 1979. 18. VH Thanh, K Shibasaki. Major proteins of soybean seeds. Reversible and irreversible dissociation of b-conglycinin. J Agric Food Chem 27:805±809, 1979. 294 Chapter 10

19. B German, S Damodaran, JE Kinsella. Thermal dissociation and association behavior of soy proteins. J Agric Food Chem 30:807±811, 1982. 20. S Damodaran, JE Kinsella. Effect of conglycinin on the thermal aggregation of glycinin. J Agric Food Chem 30:812±817, 1982. 21. S Iwabuchi, H Watanabe, F Yamauchi. Observations on the dissociation of b- conglycinin into subunits by heat treatment. J Agric Food Chem 39:34±40, 1991. 22. AM Hermansson. Physico-chemical aspects of soy proteins structure formation. J Texture Stud 9:33±58, 1978. 23. S Damodaran. Refolding of thermally unfolded soy proteins during the cooling regime of the gelation process: effect on gelation. J Agric Food Chem 36:262±269, 1988. 24. AN Danilenko, EK Grozav, TM Bikbov, VY Grinberg, VB Tolstogozov. Studies of the stability of 11S globulin from soybeans by differential scanning microcalorimetry. Int J Biol Macromol 7:109±112, 1985. 25. VY Grinberg, AN Danilenko, TV Burova, VB Tolstoguzov. Conformational stability of 11S globulins from seeds. J Sci Food Agric 49:235±248, 1989. 26. S Iwabuchi, H Watanabe, F Yamauchi. Thermal denaturation of b- conglycinin. Kinetic resolution of reaction mechanisms. J Agric Food Chem 39:27±33, 1991. 27. K Watanabe. Kinetics of heat insolubilization of soybean 11S protein in phosphate buffer system. Agric Biol Chem 52:2095±2096, 1988. 28. DJ Sessa. Hydration effects on the thermal stability of proteins in cracked soybeans and defated soy ¯our. Lebensm Wiss Technol 25:365±370, 1992. 29. DJ Sessa. Thermal denaturation of glycinin as a function of hydration. J Am Oil Chem Soc 70:1279±1284, 1993. 30. N Catsimpoolas, EW Meyer. Immunochemical properties of the 11S component of soybean proteins. Arch Biochem Biophys 125:742±750, 1968. 31. N Catsimpoolas, TG Campbell, EW Meyer. Association-dissociation phenomena in glycinin. Arch Biochem Biophys 131:577±569, 1969. 32. N Catsimpoolas, J Kenney, EW Meyer. The effect of thermal denaturation on the antigenicity of glycinin. Biochim Biophys Acta 229:451±458, 1971. 33. S Iwabuchi, K Shibasaki. Immunochemical studies of the effect of ionic strength on thermal denaturation of soybean 11S globulin. Agric Biol Chem 45:285±293, 1981. 34. A Demonte, IZ Carlos, EJ Lourenco, JE Dutra de Oliveira. Effect of pH and temperature on the immunogenicity of glycinin (Glycine max L.). Plant Foods Hum Nutr 50:63±69, 1997. 35. M A Moreira, WC Mahoney, BA Larkins, NC Nielsen. Comparison of the antigenic properties of the glycinin polypeptides. Arch Biochem Biophys 210:643±646, 1981. 36. S Iwabuchi, K Shibasaki. Immunochemical studies of the effects of ionic strength on thermal denaturation of soybean 7S globulin. Agric Biol Chem 45:1365±1371, 1981. Speciation of Soya Protein 295

37. VH Thanh, K Shibasaki. b-Conglycinin from soybean proteins. Isolation and immunological and physicochemical properties of the monomeric forms. Biochim Biophys Acta 490:370±384, 1977. 38. RC Roberts, DR Briggs. Isolation and characterization of the 7S component of soybean globulins. Cereal Chem 42:71±85, 1965. 39. K Kitamura, T Takagi, K Shibasaki. Renaturation of soybean 11S globulin. Agric Biol Chem 41:833±840, 1977. 40. DB Berkowitz, DW Webert. Determination of soy in meat. J Assoc Off Anal Chem 70:85±89, 1987. 41. MB Medina. Extraction and quantitation of soy protein in sausages by ELISA. J Agric Food Chem 36:766±771, 1988. 42. D Brandon, AH Bates, M Friedman. Monoclonal antibody±based enzyme immunoassay of the Bowman-Birk proteinase inhibitor of soybeans. J Agric Food Chem 37:1192±1196, 1989. 43. H Frokiaer, L Horlyck, V Barkholt, H Sorensen, S Sorensen. Monoclonal antibodies against soybean and pea proteinase inhibitors: characterization and applications for immunoassays in food processing and plant breeding. Food Agric Immunol 6:63±72, 1994. 44. CM DiPietro, IE Liener. Heat inactivation of the Kunitz and Bowman-Birk soybean protease inhibitors. J Agric Food Chem 37:39±44, 1989. 45. RE Oste, DL Brandon, AH Bates, M Friedman. Effect of Maillard browning reactions in the Kunitz soybean trypsin inhibitor on its interaction with monoclonal antibodies. J Agric Food Chem 38:258±261, 1990. 46. K Yasumoto, M Sudo, T Suzuki. Quantitation of soya protein by enzyme linked immunosorbent assay of its characteristic peptide. J Sci Food Agric 50:377±389, 1990. 47. JM Carter, HA Lee, EN Mills, H Lambert, HW-S Chan, MRA Morgan. Characterization of polyclonal and monoclonal antibodies against glycinin (11S storage protein) from soya (Glycine max). J Sci Food Agric 58:75±82, 1992. 48. GW Plumb, N Lambert, EN Clare Mills, MJ Tatton, CCM D'Ursel, T Bogracheva, MRA Morgan. Characterization of monoclonal antibodies against b-conglycinin from soya bean (Glycine max) and their use as probes for thermal denaturation. J Sci Food Agric 67:511±520, 1995. 49. L Huang, EN Clare Mills, JM Carter, MRA Morgan. Analysis of thermal stability of soya globulins using monoclonal antibodies. Biochim Biophys Acta 1388:215±226, 1998.

Determination of Trace Protein Allergens in Foods

1. INTRODUCTION

Food allergy is one of a number of adverse reactions to foods (Table 1). The nontoxic adverse reactions are mediated by the immune system (allergy). Non±immune system mediated adverse reaction is termed a food intolerance. Some allergic reactions to food, for example, anaphylactic shock, may be severe and even fatal. General symptoms of food allergy are reviewed by Anderson (1), Blades (2), Jones (3), and also Burks and Sampson (4). Allergenic foods are discussed by He¯e et al. (5) and also Taylor et al. (6). Food allergens are reviewed by Matsuda and Nakamura (7), Bush and He¯e (8), and He¯e (9). This chapter describes the analysis of trace amounts of protein allergens in food. Allergens are associated with eight or nine major food groups (Sec. 1.1). Soybean, peanut, and gluten allergies are described in Secs 2, 3, and 4. In each instance we also consider the structure of the protein allegens and the effect of processing on assay results.

1.1. Food Allergens and Labeling Regulations Articles by Campbell (10) and Amor (11) provide summaries of current food labeling regulations related to allergens. It is believed that food may be rendered safe by providing information related to allergen content.

297 298 Chapter 11

TABLE 1 Adverse Reactions to Foods

Class Compounds Adverse effects

Toxic reactions Natural food Range of effects at high components, food concentrations additives Nontoxic reactions Immune mediated IgE mediated Protein allergens Anaphylaxis, oral allergy syndrome, atopic dermatitis, gastroin- testinal effects, etc. Non-IgE mediated Protein allergens Celiac disease and related gastroenteropathy Non-immune mediated Lactase de®ciency, inborn Various effects errors of Food intolerance Metabolism, pharmaco- logical compounds Food aversion Psychological avoidance

Therefore, manufactures are not to consider allergens as incidental additives that are exempt from declaration. On the contrary, voluntary declaration of allergens is encouraged even where they are obviously parts of proprietary formulations. Of major concern is the presence of hidden allergens. As of 1996, manufacturers are required to take all necessary steps to prevent, remove, or otherwise arrange not to have inadvertent allergens in foods. Steps should be taken to avoid the transfer of allergens from one food product to another. Firms must make all reasonable efforts to provide adequate labeling describing the allergenic status of their product. According to the Food and Drug Administration (FDA), which is concerned with food regulations in the United States, food allergens are associated with the following eight commodities: (a) legumes (soybean, peanut), (b) milk, (c) eggs, (d) ®sh, (e) crustacean, (f) mollusks, (g) wheat, and (h) tree nuts. The United Nations Codex Alimentarius Commission also provide a list of the major sources of food allergens. The Codex list is similar to the that provided by the FDA although (allergenic food groups are) de®ned in a slightly broader manner: (a) peanuts, soybean, and products made from these;(b) milk and milk products (including lactose);(c) eggs and egg products;(d) ®sh and ®sh products;(e) crustacea and derived products;(f) tree nuts and nut products;(g) cereals containing gluten Determination of Trace Protein Allergens 299

(wheat, rye, barley, oats) and derived products;and (h) sul®te in concentrations of 10 mg kgÀ1 or greater. Future FDA and Codex Commis- sion policy may require the declaration of foods containing allergenic proteins produced through genetic engineering. The concern is that genetic modi®cation may result in the inadvertent transfer of allergenic proteins from one food source to another via genetic modi®cation (12). Food regulations and mandatory labeling requirements related to allergens are also discussed in articles by Nestle (13) and others (14,15).

1.2. Detecting Allergens and Allergies Some of the major protein allergens in foods have been isolated and partially characterized. Many are heat resistant and survive common food processing operations. They also tend to be resistant to digestion. The presence of antibodies in the circulation suggests that allergens pass through the digestive tract (16). Allergenic peptides and perhaps whole proteins may be absorbed intact, especially in infants. Some well-characterized food allergens are listed in Table 2. The detection of food allergens plays a vital role in the management of diets for susceptible subjects. Two groups of analyses are employed for food allergens: (a) in vivo tests using live patients, in which mild symptoms are induced in the subject during a skin prick test or the double-blind, placebo-controlled food challenge (DBPCFC) test, and (b) in vitro detection, which covers three types of tests involving isolated pAbs. A well-known procedure is the radioallergoadsorbent test (RAST) for circulating antibody. The food sample containing antigen is mixed with serum (collected from the allergic subject). The mixture is then added to a ®xed amount of antigen immobilized on CNBr-activated Sepharose. After washing the support, any bound pAb is determined by reacting with iodine-125-labeled (rabbit) antibody for human IgE. The amount of radioactivity bound to the solid phase, measured with a scintillation counter, is inversely related to the concentration of circulating pAb as well as to the food allergen content. RAST is essentially a competitive immunosorbent assay using iodine-125- labeled (rabbit) pAb for ``visualization'' (17,18). The second in vitro assay for food allergens involves immunoblotting (19,20);examples of these studies are listed in Table 3. The food sample is analyzed by SDS-PAGE, transferred to a nitrocellulose membrane, and exposed to serum from the allergic subject. The bound IgE is visualized using iodine-125-labeled (rabbit) pAb for human IgE followed by radiography (Table 3). The ®nal group of in vitro tests for food allergens involves classical ELISA. Included in this group are tests described in Chapter 10 for bulk food proteins. 300 Chapter 11

TABLE 2 Some Food Allergens Involved in IgE-Mediated Adverse Reactions

Food group Major allergen, comments

Milk aS1-casein (23 kDa, heat stable) b-lactoglobulin (18.4 kDa, heat stable) Egg white Ovalbumin (43 kDa) Ovamucoid (23-kDa glycoprotein trypsin inhibitor, heat stable) Soybean 30-kDa Gly m Bd 30 (oil body protein, papain analogue) 28-kDa allergen, KSTI-16 protein allergen protein in all Peanut Ara h 1. (65-kDa allergen, conarachin) Ara h 2. (17±21 kDa allergen) Castor bean 12-kDa allergen (amylase/trypsin inhibitor, heat stable) Cod®sh Parvalbumin or allergen M (12-kDa Ca2 binding protein, heat resistant) Shrimp and other Tropomyosin (70-kDA muscle protein), heat stable crustacea

Source: Refs. 7 and 8.

The merits and disadvantages of in vivo or in vitro tests for food allergens were described by Nordlee and Taylor (49) and also Taylor and Lehrer (50). One advantage of using human pAb for immunoblotting is that the results are related to functional allergens (49). The test can be done in

TABLE 3 Analysis of Food Allergens by SDS-PAGE and Immunoblotting with Human pAb

Allergen source References

Soybean Shibasaki et al. (21), Bucks et al. (22), Bush et al. (23), Herian et al. (24), Ogawa et al. (25), E-F Ebabiker et al. (26) Peanut Sachs et al. (27), Barnett et al. (28), Bucks et al. (29), Bucks et al. (30), Uhlemann et al. (31), He¯e et al. (32), de Jong et al. (33) Shrimp Hoffmann et al. (34), Nagpal et al. (35), Lehrer et al. (36), Reese et al. (37) Brazil nuts Hide (38), Morgan et al. (39), Nordlee et al. (12), Borja et al. (40) Almonds Bargman et al. (41), Hlywka et al. (42) Egg white Hoffman (43), Langeland (44), Leduc et al. (45) Milk proteins Ball et al. (46), Restani et al. (47), Rosendal and Barkholt (48) Determination of Trace Protein Allergens 301 the absence of the allergenic individuals. Immunoblotting is the default method for detecting allergens in food materials. The main disadvantages are that (a) care is needed in the selection of human donors for pAb and (b) a requirement for human pAb increases costs. Once a major allergen is identi®ed and isolated, less expensive sources of pAb or mAb may be developed.

2. SOYA BEAN PROTEIN ALLERGENS 2.1. Structure and Characteristics of Soya Bean Allergens Soybean allergy is common in children. The allergic reaction is induced by soy protein, with soybean oil being comparatively hypoallergenic (51). The identities of soybean protein allergens were established by SDS-PAGE and immunoblotting with serum pAbs from allergic patients. The same approach enabled the identi®cation of food allergens associated with peanut, shrimp, almonds, Brazil nut, and egg white (Table 3). Research leading to the identi®cation, characterization, and quantitative assays for the major soybean allergens is described in this section. Sixteen soybean allergens were identi®ed SDS-PAGE analysis, Western transfer to a nitrocellulose membrane, and immunoblotting using pAb pooled from 361 patients exhibiting atopic dermatitis (eczema). Protein bands were visualized using iodine-125-labeled (rabbit) pAb for human IgE and autoradiography (25). Approximately 20% of patients with atopic dermatitis showed IgE production for soy protein. The major soybean allergens were associated with the 7S protein fraction (7SF). Of the subset of 69 patients showing sensitivity to soybeans, two thirds produced pAb for a 30-kDa protein thereafter named Gly m Bd 30K. About 23% of patients had circulating pAb for a 70-kDa protein designated Gly m Bd 70 (Table 4). Only 1.4% and 2.9% of patients produced pAb for the A subunit of glycinin and the Kunitz soybean trypsin inhibitor, respectively. It has been suggested that the pattern of IgE speci®city can vary in different populations depending on (a) the age of the subjects, (b) history of sensitization to the allergen, (c) route of sensitization (airways vs. digestive tract), and (d) method of sample pretreatment for immunoblotting. It remains to determined whether Gly m Bd 30K is the major soy allergen in other clinical populations. The foremost soybean allergen (Gly m Bd 30K) seems identical to a 34-kDa oil body protein from soybean seeds. Ogawa et al. (52) found that native Gly m Bd 30K forms a 350-kDa oligomer that elutes ahead of glycinin during gel ®ltration column chromatography. Treatment with SDS and 2-ME produced a dissociated 30-kDa protein with a pI of 4.5. The 302 Chapter 11

TABLE 4 Major Soybean Allergens Detected by Immunoblotting with pAb from 69 Patients (average age 6 years)

Protein size Patients with (kDa) Identity antibody (%)

30 7SF, Gly m Bd 30K 65.0 68±70 b-Conglycinin a- 23.3 subunit 28 7SF, Gly m Bd 28K 23.3 63±67 7SF 18.8 52±55 7SF 14.5 47±50 7SF 10.1 43±45 7S globulin, b-unit 10.1 33±35 7SF 15.9 35 11S globulin, a-unit 1.4 20 2S, KSTI 2.9

Source: Ref. 25.

N-terminal 15-amino-acid sequences for Gly m Bd 30K and 34-kDa oil body protein were the same. The mAb for either protein cross-reacted with the other. Following established convention, Gly m Bd 30K, being the ®rst soybean allergen identi®ed, was designated Gly m 1. Kalinski et al. (53) independently characterized the 34-kDa oil body protein. Within intact cells, the protein was a vacuolar protein designated P34. Like other storage proteins, protein P34 undergoes post-translational glycosylation and proteolysis. During processing, P34 appears along the endoplasmic recticulum, Golgi bodies, and eventually within vacuoles or protein bodies. Protein P34 had partial sequence homology with cysteine proteases from the papain superfamily. It is not certain that P34 has proteolytic activity (54). Its association with soybean oil bodies was an experimental artifact produced by cell disruption. A survey of a large number of soybean strains shows that P34 is widespread. The possibility of eliminating P34 from soybean lines by breeding seems doubtful (55). The next important soybean allergen (Gly m Bd 70K) is the 70-kDa b-conglycinin a-subunit. Antibody speci®c for the a-subunit did not cross- react with the other two (a0 or b) subunits of b-conclycinin despite some sequence homology between these polypeptides (56). The third soybean allergen (Gly m Bd 28K) is a 26-kDa glycosylated protein with a pI of 6.1 (57). It is apparently unstable and present in very small quantities in defatted soy ¯our (*15 ppm). Perhaps for these reasons, Gly m Bd 28K was dif®cult to detect in processed foods containing soy protein. Finally, pAbs from some soybean-sensitive individuals recognized the A-chain residue Determination of Trace Protein Allergens 303 from glycinin. The epitope consisted of a 114-amino-acid residue fragment (58). The epitope for Gly m Bd 30K recognized by mouse mAb was also identi®ed by Hosoyama et al. (59).

2.2. Quantitative Analysis of Soybean Allergens Quantitative ELISA for Gly m Bd 30K was developed by Tsuji et al. (60) using mAb produced by conventional means. Eventually, two cell lines producing mAb (mAb-F5 and mAb-H6) were isolated. Immunoblot analysis showed that mAb-F5 and mAb-H6 were both speci®c for Gly m Bd 30K. Sandwich ELISA for Gly m Bd 30K employed mAb-H6 as the capture antibody and an HRP conjugate with mAb-F5 as detector. The linear range for analysis was 2±200, 5±500, or 10±500 ng protein for reduced/ carboxymethylated-allergen, SDS/2-ME±treated sample, or native soy protein, respectively. Ten soybean products and ®ve meat products containing soy protein isolate (SPI) were analyzed by a sandwich ELISA for Gly m Bd 30K (61). Results from this study are summarized in Table 5.* From such data it may be surmised that (a) high concentrations of Gly m Bd 30K are detectable within a range of processed soybean products, (b) fermented soy products (miso, soyu, and natto) are free of allergen, (c) Gly m Bd 30K can be accurately determined in cooked meat products known to contain soy protein isolate, (d) sandwich ELISA results were corroborated by immunoblot analysis using pAbs from soybean-sensitive patients, and (e) Gly m Bd 30K is heat resistant, accounting for its detection in cooked products. The thermal stability of soybean antigens is discussed in Chapter 10. The allergen was also resistant to digestion by chymotrypsin and trypsin. Levels of Gly m Bd 30K declined to negligible values in fermented soy products, probably as a result of digestion by microbial (acid) proteases. Treating soybean samples with added microbial proteases reduced the level of allergen, leading to hypoallergenic soybean products (62). Clearly, the preceding assay has potential uses for monitoring soybean allergen in processed foods. The ELISA test could also ®nd use as a quality control tool for hypoallergenic soybean products.

* As pretreatment each (5-g) sample was homogenized with Na- phosphate buffer (50 mM, pH 8.0 with 1% w/v SDS and 20 mM 2-ME). The extract was centrifuged and SDS was precipitated with KCl (1 M ®nal concentration). After centrifugation, the SDS-depleted sample was analyzed by sandwich ELISA. 304 Chapter 11

TABLE 5 Determination of the Major Allergen (Gly m Bd 30K) in Soybean-Containing Products using Sandwich ELISA

Sandwich ELISA results (mg allergen Immunoblot Product gÀ1 N) analysisa

Soybean 126 ve Soymilk 106 ve Tofu (kinugoshi) 89 ve Tofu (momen) 65 ve Kori-dofu 64 ve Kinako 29 ve Abura-age 59 ve Yuba 66 ve Miso npb À Shoyu np À Natto np À Meatballs 17 ve Beef croquettes 21 ve Fried chicken 9 ve Fish sausages np À Hamburger np À aSDS-PAGE/immonoblotting with IgE from soy-sensitive patients. ve positive, (À) negative results. bnp, no allergen present. Source: Adapted from Ref. 61.

2.3. Indirect Monitoring of Soybean Allergens by ELISA ELISA for bulk soy protein analysis (Chapter 10) may be useful for monitoring soybean allergens indirectly. Yeung and Collins (63) described a competitive ELISA using (rabbit) pAb for whole soybean extract. The assay, which was virtually identical to those used by Hitchcock and co- workers (64±67), was intended to detect bulk soy protein as opposed to trace allergen. The antigen for immunizing rabbits was prepared from soybean ¯our defatted with cold acetone.* The competitive ELISA for soy protein (63) had the following characteristics:

* Soy ¯our was extracted with cold Tris-HCl buffer (10 mM, pH 8, with 1% w/v SDS and 10 mM 2-ME) at 48C for 16 hours. Determination of Trace Protein Allergens 305

1. Speci®city. There was no cross-reactivity with 10 other legumes, 8 varieties of nuts, 11 common food ingredients, and 2 phytoestro- gens. 2. Acuity. The concentration of soy proteins that produced 50% À1 inhibition of pAb-antigen binding (IC50) was 35 ng mL and the linear range was 3±117 ng mLÀ1. For various real foods, the detection limit was 2 ppm. 3. Recovery. For processed tuna ®sh having 13.5±54 ppm of soy protein, the recovery was 77±95%. Similar recoveries (63±96%) were observed with hamburger matrix. 4. Precision. For model solutions and hamburger matrix containing soy protein, the interassay precision was 3.5±4.2% and 3.6±8%, respectively. The corresponding intra-assay precision 2.4±5.1% or 1.5±2.6%. Features such as assay speci®city and acuity were not routinely reported for ELISA designed to quantify soy protein in meat samples (Chapter 10).

3. PEANUTS 3.1. Peanut Allergy Peanut (groundnut) allergy occurs in about 1 in 200 of the general population in the United States. The incidence rate is probably similar in Western Europe. Emmett et al. (68) found a partial association between peanut allergy and allergy to tree nut. Peanut allergy also persists for life. Allergic subjects appear in all age groups and with no gender bias (69). About 25±28% of all cases of food allergy involve peanuts. The ®gures for tree nuts are Brazil nut (10.2%), almond (7.8%), and hazelnut (7.1%). The incidence of peanut allergy also appears related to the frequency of exposure (70). In a study of 868 children in Singapore, 27% of subjects showed sensitivity to bird's nest soup. The incidences of other food allergies were 24% (crustacean), 11% (eggs and cow milk), and 7% (traditional Chinese herbs). There was no recorded adverse reaction to peanuts or tree nuts. The low incidence of peanut allergy was attributed to the low per-capita consumption of peanuts and other nuts in Singapore (71). Symptoms of anaphylactic shock arise in approximately 6% of allergic reactions to peanuts. The onset of fatal and near-fatal anaphylaxis can be very rapid. The ®rst reaction may appear within 5±30 minutes of contact with the allergen (72). More frequent clinical effects include atopic dermatitis or eczema (40%), angioedema (37%), and asthma (14%);digestive symptoms occur more rarely (1.5%). The quantity of peanut necessary to produce an adverse reaction ranges from a few micrograms to 1 g (73). The 306 Chapter 11 anaphylactic response can arise from topical contact with peanut residue on another individual, by inhalation (74), or by exposure to peanuts in vegetarian ``beef '' burgers (75). Small amounts of peanut protein associated with peanut oil can also induce an allergic episode. Peanut protein was found in some infant formula prepared from peanut oil (76,77). Highly re®ned protein-free oil apparently poses little threat (51). Reported fatalities due to peanut allergy are of the order of 125 persons per year in the United States (84). Figures from Pumphrey et al. (78) suggest that there were 21 deaths from peanut and tree nut allergy in the United Kingdom during the 5- to 6-year period following 1992. Of 541 patients showing sensitivity to nuts, 90% had serum IgE speci®c for peanut. About 67% of patients showed serum IgE for another nut besides peanut, while 34% of patients had immunoglobins for all nuts tested. The threshold quantity necessary to induce allergy varies for different subjects. Exposure to peanut allergens in early infancy (<1 year old) or before birth may contribute to sensitization (70). It is not yet certain whether pregnant and lactating women should avoid peanuts and tree nuts in their diet (78±80). Avoidance of peanuts is the main strategy for managing the diet of sensitive individuals. Hidden allergens from peanut-derived ingredients are a major issue. Eating outside the home setting can also be problematic. Restaurateurs cannot provide strict assurances that their ingredients or recipes are free from peanut allergens. People thought to be highly sensitive to peanut protein are also advised to have rapid access to epinephrine (adrenaline) by injection. In summary, peanut and nut allergies present serious health threats. Reviews by Sampson (81), Hourihane (82), Burks et al. (83,84), and also Warner (85) deal with the clinical and public health aspects of peanut allergy. Some distinctive features of peanut allergy are (a) high incidence in the population, (b) speed and severity of the symptoms, (c) lack of an effective treatment, (d) widespread use of peanuts and peanut ingredients in the food chain, and (e) persistence of peanut sensitivity with age. The availability of tests for trace amounts of peanut will reduce the risk of exposure to peanut allergens.

3.2. Structure and Characteristics of Peanut Allergens Proteins from peanuts are predominately globulins (* 90% total). Approximately 10% of the storage proteins are albumins. Other classes of storage proteins (glutelins, prolamins, etc.) are not present. Johns and Jones (86) divided the peanut globulins into two groups, arachin and conarachin. The foremost peanut allergen (Ara h 1) is thought to be the 60±65 kDa subunit of conarachin. Allergin Ara h 2 corresponds to a 21-kDa arachin subunit (33). Therefore arachin and conarachin are legitimate targets for Determination of Trace Protein Allergens 307

ELISA tests designed to monitor peanut allergens. SDS-PAGE analysis of peanut protein treated with 2-ME revealed 14 bands with molecular masses of 10±70 kDa. Western transfer to nitrocellulose membranes and immunoblotting with human pAb led to visualization of 13 of the 14 polypeptides;a 26 kDa protein was not allergenic, probably due to denaturation by heat and SDS treatment. Six major allergens (i.e., recognized by antibody from 50% of patients) had molecular masses ranging from 17 to 44 kDa (Table 6). Fractionation of arachin and conarachin can be achieved by three strategies (87,88):

1. Cryoprecipitation. Extract crude peanut protein by stirring defatted ¯our with 0.2 M phosphate buffer (pH 7.9) for up to 4 hours. Dialyze against several changes of tap water at 48C. Arachin precipitates while conarachin remains soluble. This can be recovered by a range of methods, e.g., acid precipitation or freeze-drying. 2. Ammonium sulfate fractionation. Precipitate arachin and conarachin from solution by adjusting to 0±40% and 60±80% saturated ammonium sulfate, respectively. When this is preceded by cryoprecipitation, the samples of arachin and conarachin obtained are homogeneous. Chiou's procedure (89) for isolating conarachin is simple and effective. Add peanut (acetone) powder directly to 60% saturated ammonium sulfate solution. Centrifuge and then adjust the supernatant to 85% saturation by adding more ammonium sulfate. The precipitated protein is conarachin. 3. Ion-exchange chromatography. Fractionate crude peanut protein using low-pressure chromatography on a DEAE column. With FPLC, use a Mono-Q support. Three peaks (P1, P2, and P3) are

TABLE 6 Major Peanut Allergens

Molecular size (kDa) Patients with antibody (%)

26.0, 0 29.0, 28.0, 16.0 20 36 20±30 71.0, 60±64, 17.0 30.0±40.0 44.0, 40.0, 33.0 50±60 21.0, 20.0, 17±18 60±80

Source: Adapted from Ref. 33. 308 Chapter 11

eluted with a 0±0.6 M gradient of NaCl. Protein peaks P2 (conarachin) and P3 (arachin) normally elute at 0.23 M and 0.33 M NaCl.

Native arachin (11±12S globulin) has a molecular mass of 350 kDa (90). Treatment with 2-ME followed by SDS-PAGE generates seven polypeptides with molecular masses of 72.4, 60.3, 39.8, 33.1, 29.0, and 21 kDa.* Native conarachin is a 7S globulin with a molecular mass of 180 kDa. The quaternary structure is a trimer of 60-kDa subunits. SDS- PAGE analysis of 2-ME-treated conarachin yields eight bands (72.4, 39.8, 33.1, 26.9, 24.0, 21.9, 18.6, and 16 kDa). These results from Monteiro and Prakash (90) suggest that peanut 7S protein has a different architecture from soybean 7S protein. The a, a0, and b subunits of conglycinin appear to be devoid of S22S bonds. In contrast, peanut 7S protein is affected by 2-ME, indicating the presence of S22S bonded subunits (91).

3.3. Thermal Denaturation of the 11S Peanut Allergen (Arachin) Ara h 2 is either the A or B subunit of arachin. Investigations employing dry and moist heat treatment show that arachin is very heat resistant. Neucere (92,93) identi®ed 14 peanut antigenic species by immunoelectrophoresis using (rabbit) pAb. Moist heat treatment (40% w/w moisture, 1108C) or dry heat treatment (5% w/w moisture, 110±1508C) for 60 minutes reduced the antigenicity of nonarachin proteins monitored by immunoelectrophoresis, rocket immunoelectrophoresis, or AGID. To achieve the same effect for arachin required moist heat treatment at 1508C or dry heat treatment at 1758C. Heating also reduced protein solubility. SDS-PAGE analysis of peanut proteins heated at 1008C for 15±210 minutes showed little changes in the patterns for arachin (94). Dry and moist heating studies by Chiou (89) also con®rm that arachin is heat resistant. Shokraii and Esen (91) found that treating peanut protein with SDS and 2-ME had no effect on the immunological characteristics. This raises the possibility that some epitopes for peanut allergens are continuous. That is, antibody binding occurs with consecutive sequences of amino acids. Discontinuous epitopes constructed from higher order protein structure would be destroyed by denaturants.

* Only six from seven bands were prominent. Determination of Trace Protein Allergens 309

3.4. Thermal Denaturation of the 7S Peanut Allergen (Conarachin) The effect of thermal treatment on the conformation and immunological characteristics of Ara h 1 (b-subunit of conarachin) was examined by Koppelman et al. (95). The Ara h 1 was isolated by cation-exchange chromatography of whole peanut protein. Gel ®ltration and sedimentation velocity measurements showed that native Ara h 1 was a 180-kDa trimer. An SDS-PAGE analysis con®rmed the presence of 63-kDa subunits. The far-UV spectrum for native Ara h 1 was consistent with there being 31% a-helix, 36% b-sheets, and 33% aperiodic structure. The ¯uorescence spectrum for Ara h 1 indicated that tryptophan residues were solvent- exposed in the native protein. Differential scanning calorimetry thermo- grams showed a denaturation temperature (TD) equal to 878C with a small À1 transition enthalpy (DHm) of about 120 kJ mol for an unde®ned irreversible structure change. Upon heating samples of Ara h 1 for 15 minutes at 20, 50, 80, 90, 110, or 1408C, protein solubility decreased to 75% and 32% at the last two temperatures. SDS-PAGE analysis showed the same molecular weights for heated and unheated protein. Thermal treatment did not produce signi®cant changes in IgE binding avidity for the soluble protein fraction. Such results are consistent with one of the following explanations: (a) Ara h 1 is heat resistant, (b) the epitopes for IgE binding are in parts of the three- dimensional structure not affected by denaturation, or (c) the epitopes of Ara h 1 are of the continuous type (con®rmed in the following). The 3D structure of Ara h 1 was elucidated from computer modeling studies using the crystal structure of phaseolin as template (96). The locations of 23 linear/continuous epitopes for human IgE binding to this allergen were also established. For the native protein, antigenic sites were clustered in the outer exposed region of the 3D structure. Most epitopes of IgE binding comprised nine amino acids. There was no other common structural feature for the epitopes. Single amino acid substitution within epitopes abolished antibody binding.

3.5. Digestibility of Peanut Allergens Protein stability and/or resistance to digestion appears to be a feature of allergenic proteins. Compared with a random group of plant proteins, peanut allergen (Ara h 2) was more resistant to digestion by simulated gut ¯uid (97). The same was true for allergens from soybean (b-conglycinin b-subunit, KSTI, Gly m 1), egg (ovalbumin), or milk (b-lactoglobulin, casein, and BSA). Analysis by enzyme allergosorbent test (EAST) or 310 Chapter 11 immunoblot analysis showed that simulated gut digestion reduced the allergenicity of peanuts and hazelnuts by 50 and <10% (98). The digestion of soy proteins by rats proceeded within 3 days to immunoreactive peptides (99). Allergic reactions are liable to be invoked by contact with raw, cooked, or roasted peanuts. Peanut and other food allergens will also have a tendency to survive common food processing operations.

3.6. Determination of Peanut Allergens Peanut allergens were initially identi®ed by immunoblotting with human pAb (Table 3) or using the RAST assay (17,18). Both techniques use iodine- 125 as label. These approaches have limited general utility for several reasons: (a) there is limited availability of human antibody, (b) human antibody is dif®cult to obtain as a standard reagent, (c) there are possible biohazards associated with handling human body ¯uid as a reagent, and (d) RAST employs radionuclides for visualization (100). More convenient ELISAs for peanut allergen are now being developed by research groups in several countries including the United States (100), Netherlands (101), Canada (102), United Kingdom (103), and Germany (104). Keck- Gassenmeier et al. (105) evaluated several ELISA kits for peanut allergens before adapting one for semiquantitative analysis of confectionery products (see later). The ®rst of the new generation of tests for peanut allergens* was developed by He¯e et al. (100). Results from the sandwich ELISA were positively correlated with RAST results for a range of samples including cheesecake mix, chocolate peanut cookies, chocolate-nut granolar bars, confection±peanut brittle, and peanut butter. The high correlation (R 0.85) is to be expected because the mAb used for the ELISA test was speci®c for peanut allergens recognized by human IgE. The working range of the ELISA was 3.2±3162 mgmLÀ 1 of peanut extract. With ice cream, the LLD for peanut protein was 40 mgmLÀ 1. This level of acuity is higher than observed with the traditional skin test for peanut allergens. A competitive indirect ELISA for trace amounts of peanut protein was developed by Yeung and Collins (102). The IC50 was adopted as an index of speci®city. The linear range for analysis was 1±63 ng mL À 1 extract (R2 > 0.98) with an LLD equal to *400 ng g À 1 (food sample). Negligible responses were obtained for protein extracts from the following legumes and

* Whole peanut protein was used as antigen. An mAb and biotinylated (rabbit) pAb were used as capture antibody and for detection, respectively. Visualization was with an HRP-streptavidin conjugate assayed with OPD±hydrogen peroxide substrate. Determination of Trace Protein Allergens 311 nuts: soybean, green pea, chick pea, lupine, hazelnuts, Brazil nuts, pine nut, pecan, almonds, walnut, and cashew nut. Also, no interferences occurred with general foodstuffs such as corn, cocoa, chocolate, milk, sugar, soy lecithin, peanut lectin, and coconut. Peanut proteins were quantitatively determined in over 40 processed food items belonging to the following food groups: potato chips (crisps), cooking oil, pasta sources, plain cookies, ice cream, and milk chocolate bars. All products labeled as containing peanut protein were correctly identi®ed with no false-positive results. A sandwich ELISA using (rabbit) pAb for puri®ed Ara h 1 was described by Koppleman and co-workers (101). The assay characteristics can be summarized as follows: 1. Range. The linear dynamic range for peanut protein was 0.005±1 mgmLÀ 1. Fried peanut samples gave 45±80% of the response obtained from raw samples. 2. Speci®city. No cross-reactivity occurred with 40 food products including, 9 legumes, 9 nuts, 7 cereals, 4 oil seeds, and 9 animal proteins. In tests involving 31 processed goods (10 cookie brands, 6 chocolates, 3 cereals, 3 baby foods, and 8 sources and soups) all products labeled as containing peanuts were accurately identi®ed. 3. Recovery. For samples spiked with 0.0001±1.0% peanut protein, recoveries were typically 85±105%. Reproducible recoveries were obtained from high-fat foods. The competitive indirect ELISA described by Holzhauser and Vieths (104) employed commercial pAbs for unheated peanut protein. Before use, pAbs were puri®ed by immunoadsorption with soy protein, white bean, and marzipan. Precooking peanuts had little effect on the levels of Ara h 1 and its binding by both human and rabbit pAbs. The ®nal assay was successfully applied to over 30 commercial food products. Three commercial ELISA kits for peanut allergen were evaluated (105) at the Nestle Research Center (Lausanne, Switzerland). The kits were those from Cortecs Ltd. (UK), Prolab Ltd. (Canada), and the TNO (Nether- lands). The last two kits appear to be derivatives of the assays developed by Yeung and Collins (102) and Koppleman and co-workers (101). Detailed results concerning the preliminary evaluation were not reported. The Cortecs kit was further evaluated for a semiquantitative determination of trace amounts of peanuts in a range of confectionery products. The particular kit employs a sandwich ELISA format with antibody for concanavalin A for capture. Biotinylated antibody for concanavalin is used for detection. Bound antigen is visualized with streptavidin-labeled HRP assayed using tetramethylbenzidine. Recovery of peanut allergen from dark chocolate was only 2±3% when using the manufacturer's extraction 312 Chapter 11 buffer (TBST). Inclusion of 12.5% ®sh gelatin in the extraction solvent increased immunogen recovery to 65±97%. With the modi®ed sample extraction procedure, the Cortecs ELISA kit was successfully applied to a wide range of foods including milk chocolate, chocolate confectionery, ice cream, sherbet, ready-to-use refrigerated cookie dough, lemon curd, apple sauce, piccalilli, honey, jelly, and a frozen meal. The test kit originally intended for a qualitative assay of peanut protein could be adapted for a semiquantitative analysis (105). In conclusion, several ELISA assay formats are available for detecting trace amounts of peanuts and tree nut allergens (106). These highly sensitivity assays can detect as little as 2 mg (allergens) per kg of food (2 ppm). Making sure that allergen can be effectively solubilized from highly processed foods is a limiting factor. Present assays can readily provide a yes/ no indication for hidden peanut allergens. Self-operated diagnostic tests for peanut allergens would be useful for domestic use.

4. WHEATAND RELATEDCEREALS 4.1. Wheat Allergy Adverse reactions to cereals and cereal products include dermatitis herpetiformis, baker's asthma, and celiac disease.* The last condition is related to the consumption of the alcohol-soluble cereal proteins (prolamins) from wheat, barley, rye, or triticale. It is doubtful whether oat is a source of celiac-active protein (107,108). Rice, millet, sorghum, maize, and buckwheat are reportedly nontoxic to gluten-sensitive individuals (109,110). Baker's asthma is ascribed to inhalation of airborne cereal amylase inhibitor proteins (111). This section focuses on allergy to wheat gluten and related proteins. Various aspects of celiac disease have been reviewed by Goodwin and Rawcliffe (112), Baldo and Wrigley (113), Ciclitira and Ellis (114), Cornell (115), Brozzone and Asp (116), Silano and Vincenzi (117), Feighery (118) and Hadjivassiliou et al. (119). Marsh (120) wrote a monograph on this subject. Celiac disease appears in 1 in 300 of the general population in Western Europe except in Italy, where the frequency is nearer 1 in 4700. The incidence of nontropical sprue in the United States is about 1 in 250 (121). One of the highest occurrences of celiac disease was reported for the Seharawi tribe from western Algeria and parts of Morocco. Their staple diet

* Celiac disease is also known as gluten-sensitive enteropahathy, gluten-induced enteropathy, or nontropical sprue. Determination of Trace Protein Allergens 313 consists of couscous and bread. Catassi and co-workers (122) suggested that celiac disease has adaptive value in preventing intestinal adhesion and infection by pathogenic microorganisms such as Vibrio cholerae. The major lesion accompanying celiac disease is atrophy of microvilli from the small intestine, duodenum, and jejunum. Microscopic observation shows a ¯attening of the mucosal lining and reduced surface area. There is malabsorption of most nutrients. The intestinal absorptive layer, being compromised, may allow the uptake of a greater range of psychoactive peptides. Theories explaining the etiology of gluten-sensitive enteropathy are summarized by Cornell (115), who also describes in vitro methods for assaying celiac-reactive polypeptides or allergens. Bruzzone and Asp (116) provide an informative introduction to gluten-sensitive enteropathy starting from its discovery in 1950±1953 by Dickie and covering possible links between diet and the immune system. Progress in the study of gluten- sensitive enteropathy has suffered owing to the lack of reliable assays for the condition. Current methods for diagnosis for this condition involve 1. Use of organ culture. Intestinal biopsy spceimens when grown in cell-tissue culture show normal microvilli within 24 hours. Addition of gluten (peptides) to culture media reintroduces abnormalities in microvilli. 2. Detection of circulating antigliadin antibodies. Blood serum from sufferers from gluten-sensitive enteropathy is used as the basis of indirect ELISA. The symptoms of celiac disease in infants and young children include growth impairment, abnormal stool, abdominal distention, and psychological effects, which show up as generally poor temperament. Adults sometimes exhibit diarrhea, ¯atulence, bulky stools, anemia, fatigue, weight loss, bone weakness, and neurological disorders including depression and schizophrenia. Treatment for celiac disease involves the exclusion of gluten from the diet. United Kingdom regulations state that gluten-free foods should contain no more than 300 mg (gliadin) per 100 g foodstuff (i.e., 300 mg % or 0.3% w/w). European Union and WHO limits for gluten-free foods are between 10 and 1 mg %. Wheat gluten is an inexpensive protein and therefore widely used as a ``vegetable protein'' ingredient for meat products (123). Wheat ¯our is used for thickening soups and other processed foods. Gluten appears in confectionery products and is also used as a binder for pharmaceutical tablets. Beer contains barley protein. Rye bread is widely consumed in parts of Eastern Europe. Raw meat products such as sausages may contain wheat proteins derived from rusk. Hidden allergens need to be considered in attempts to reduce celiac-active proteins and peptides in the diet. There is 314 Chapter 11 generally inadequate labeling information with respect to products having wheat or barley protein. Levels of potentially active celiac proteins and peptides in exotic foods, ethnic cuisine, and restaurant or cafeteria foods are ill de®ned. A number of manufacturers specialize in the production of gluten-free products. Most developed countries have celiac disease societies that coordinate information useful for sufferers.

4.2. Structure and Characteristics of Dietary Wheat Allergens A. Classi®cation of Cereal Proteins Thomas Burr Osborne divided plant seed proteins into ®ve classes according to their solubility in water, 0.1 M sodium chloride, alcohol, or alkali. Sequentially extracting ¯our using this range of solvents recovers albumins (in water), globulins (with 0.1 M salt), prolamins (in 40±70% ethanol, and glutelins (with 0.1 M sodium hydroxide). A ®fth class of protein (called residue protein) can be extracted only with severe solvents. The glutelins are extractable in alcohol and 2-ME or DTT. About 70±90% of the storage proteins in cereals consist of nearly equal proportions of prolamin and glutelins.* An important exception is cereal oat, which contains 80% globulins. Rice has 1±5% prolamin and 80±90% glutelin. The allergens responsible for celiac disease are thought to be amino acid repeat sequences found in cereal storage proteins. Tatham et al. (124) devised a new classi®cation scheme for cereal proteins (Table 7). The new scheme acknowledges fundamental structural

TABLE 7 Cereal Prolamins Thought to Be Involved in Celiac Disease

Prolamin Wheat Rye Barley

Sulfur rich Monomeric g-Gliadin g-Secalin g-Hordein b-Gliadin a-Gliadin Polymeric LMW glutenin B hordein Sulfur poor o-Gliadin o-Secalin C hordein HMW HMW glutenin HMW secalin D hordein

* By contrast, legume seeds (soya bean, pea, kidney bean, and peanut) contain globulins as the major storage proteins. There is relatively little (0±10%) prolamin or glutelin present. Determination of Trace Protein Allergens 315 relationships between the different protein fractions. First, all cereal protein fractions (besides albumins) are soluble in 70% alcohol and are therefore classed as prolamins. Next, the storage proteins are grouped according to similarities in their amino acid and cDNA sequences into three classes: (a) sulfur-rich prolamins (a-, b-, g-gliadin), (b) sulfur-poor prolamins (o-gliadin), and (c) high-molecular-weight (HMW) prolamins. The sulfur- poor prolamins are monomers possessing only intramolecular disul®de bonds. This group includes o-gliadin (wheat), o-secalin (rye), and C hordein (barley). Some sulfur-rich prolamins form HMW polymers. By contrast, a-, b-, and g-gliadins represent monomeric sulfur-rich prolamins. The three classes of prolamins have related architectures.

B. Gluten Gluten is prepared by washing wheat ¯our with water to remove the starch. This leaves a viscoelastic protein network (gluten) that is responsible for the gas-retentive properties of wheat dough. Gluten is a network formed by interacting wheat prolamins (Table 7). Each of the four gliadins can be prepared on a large scale by ion-exchange chromatography (125).* A single chromatographic run involving about 54 g crude gliadin yields 1 g of o- gliadin and approximately 10 g each of g-, b-, and a-gliadin. Electrophoresis at low pH separates of gliadin into four subtypes. SDS-PAGE analysis shows the following order of increasing electrophoretic mobility: a-gliadin > b-gliadin >, g-gliadin > o-gliadin. Glutelin is highly polymerized by S22S bonds. The sulfur-rich prolamins possess N-terminal repetitive amino acid sequences containing high numbers of three amino acids: glutamine (Q), proline (P), and phenylalanine (F) (Figure 1). The C-terminal areas have variable sequences bearing a number of half cysteine residues normally involved in intramolecular S22S bonding. Also noteworthy is the rare occurrence of SH groups in o-gliadin. Regularities are also evident in the 28 structure of cereal proteins (126,127). Thus, a-gliadin, b-gliadin, and g-gliadin have 36±37% a-helix, 11±12% b-sheet, and 52±53% aperiodic structure plus b-turns. The precise quantity of b-turn or hairpin loop structure is uncertain because this does not produce a distinct circular

* Stir 1 kg of ¯our with water-saturated n-butanol to remove lipids. Next, extract with four volumes of 70% (v/v) aqueous ethanol. Dry the extract by rotary evaporation and redissolve the protein in 0.1 M acetic acid solvent. Dialyze against the same solvent and fractionate with a CMC-cellulose column (10 cm 6 22 cm, containing 1.7 kg support) equilibrated with sodium acetate buffer (5 mM 1 M dimethylformamide and adjusted to pH 3.5 with acetic acid). Elute the bound protein using stepwise changes in NaCl concentration gradients. 316 Chapter 11

FIGURE 1 The primary structure of cereal prolamins. (Adapted from Refs. 124 and 126.)

dichroism absorbance peak. However, numerical calculations using amino acid sequence data suggest that the repetitive sequences are b-turns. Celiac toxic peptides have been identi®ed at the N-terminal regions of a-gliadin and b-gliadin (117,128,129). The epitopes recognized by intestinal T cells appear to be two tetrapeptides, PSQQ and QQQP. The ®rst occurs in a-gliadin at amino acid residues 13±16, 50±53, and 213±216. QQQP occurs at residues 16±19, 33±36, and 188±191. Four out of six potentially toxic sequences appear within the ®rst 56 amino acids from the N-terminal. The epitope bound by mAb-WC2 is probably QQQP, which is apparently common to all celiac-active proteins (129). Tanabe et al. (130), using Determination of Trace Protein Allergens 317 antibody from celiac sufferers, identi®ed the sequence QQQPP as being celiac active. Bromelain digests proline-rich peptides leading to wheat ¯our that may be useful for producing hypoallergenic bread (131). Ginger protease has high speci®city for proline residues (132).

4.3. Thermal Denaturation of Wheat Allergens The thermal denaturation of gliadin was examined using 70% (v/v) ethanol as solvent. Heating g-gliadin led to a conformational change at 20±608Cas monitored by far-UV circular dichroism.The spectral changes showed an isobestic point implying that denaturation was a two-state (helix-coil type) transition. At elevated temperatures the proportion of a-helix declined by 7% and the structure of gliadin loosened to expose phenylalanyl residues to the solvent. Puri®ed a-, b-, and o-gliadin behaved similarly. Conformational changes produced by the relatively short (5-minute) thermal treatments were reversible (126,127). Heating wheat ¯our matrix revealed that o-gliadin is more heat resistant than the sulfur-rich gliadins.* The levels of a/b-gliadin decreased substantially after 20 minutes of heating. The g-gliadin was resistant for up to 50 minutes of heating at 1008C. There was no change in the quantity of o-gliadin recovered after 0±100 minutes of heating. This result is due to the virtual absence of half cystine residues in o-gliadin. This protein can not undergo sulfhydryl-disul®de exchange during heating (133).

4.4. Determination of Wheat Allergens Immunological assays for cereal proteins are generally concerned with the detection of celiac-active peptides. Some attempts to assess grain varietal differences by EIA were reported (134). Also of interest is the identi®cation of (bread making) quality-related polypeptides by immunoassay (135,136); literature in this area was reviewed by Skerritt et al. (137). Sections 4.4 through to 4.10 consider ELISA tests for trace (nanogram to microgram) quantities of cereal proteins. ELISAs for wheat allergens and bulk protein adulterants are not fundamentally different in their design. However, the former techniques possess greater acuity. SDS-PAGE±immunoblot analysis is another ELISA format.

* A suspension of wheat ¯our in water was heated at 1008C for 0, 10, 20, 30, 50, and 100 minutes. Flour proteins were then extracted with 1 M urea solution and analyzed by gradient gel electrophoresis. 318 Chapter 11

4.5. Sandwich ELISA Tests for Gluten Using pAb McKillop et al. (141) provide a useful introduction to ELISA of wheat gliadin. Sandwich ELISA and competitive indirect ELISA formats were implemented using (rabbit) pAb for unfractionated gliadin. The two- antibody format was selected for full evaluation. The assay characteristics were as follows: 1. Linear range. 0.05±6 mg (gliadin) mLÀ1. 2. Detection limit. A concentration of 23 ng (gliadin) mLÀ1 produced absorbances of 2 SD above the reagent blank. 3. Speci®city. Gliadin and gliadin-containing foods. Wheat ¯our had 5.7% (w/w) gliadin equivalent protein. The values for oats, rice, and corn ¯our were 240, 46, and 47 mg %, respectively. 4. Precision. The within-assay reproducibility for analysis ranged from 5 to 40%. Samples with > 330 ng (gluten) mLÀ1 were determined with below + 10% error. The preceding assay enabled the identi®cation of ``gluten-free'' ingredients. No tests were performed on heated samples. Table 8 summarizes other ELISAs for wheat proteins. Fritschy et al. (140) used (rabbit) pAb for total gliadin as the basis for a gluten-sensitive ELISA. Wheat ¯ours (20 different types) were found to contain 2.9±6.7% (w/w) gliadin equivalent proteins. The gliadin contents of many commercial gluten-free foods were insigni®cant. Rice contained 57.5 mg % (w/w) and sorghum 12.5 mg % (w/w) gliadin on a dry weight basis. Food samples were extracted with 70% (v/v) ethanol with a recovery of 77±107%. The LLD was 10 ng (gliadin) mLÀ1 sample extract. The assay was speci®c for cereal proteins. It was possible to assay rice, corn, oats, and barley protein when 1±2% BSA was included in samples to avoid nonspeci®c protein binding to microwell plates. Troncone et al. (145) puri®ed (rabbit) pAb by af®nity chromatography using Sepharose 4B±immobilized gliadin. They then implemented a conventional sandwich ELISA using (rabbit) pAb for capture. The linear dynamic range for gliadin analysis was 5±400 ng mLÀ1 with an LLD of 0.5 ng (3 ng for the competitive assay). There was cross-reactivity with maize and oat prolamins, although these were detected with 1000-fold lower sensitivity. The response for gliadin was 104-fold higher than obtained for rice prolamin. Rye and barley were not assayed. Interference from maize, rice, or oat prolamins could be avoided by using high protein dilution.*

* Equal amounts of prolamin from different cereals yield 1000- to 10,000±fold differences in assay sensitivity. The ELISA response for wheat protein (cf. maize, oats, and rice proteins) measures ``gluten''-like structure and functionality. ELISA results do not necessarily show the absolute amounts of prolamins present in the different cereals. Determination of Trace Protein Allergens 319

TABLE 8 Analysis of Gliadin and Other Prolamins by ELISA

ELISA format and antibody Antigen References

Sandwich, pAb a-Gliadin, gliadin Windemann et al. (138), Meier et al. (139), Fritschy et al. (140) Sandwich, Competitive, Gliadin McKillop et al. (141) pAb dot blotting, mAb o-Gliadin, gliadin Skerritt and Smith (133), Skerritt (142), Skerritt et al. (143), Freedman et al. (144) Sandwich, Competitive, Gliadin Troncone et al. (145) pAb Sandwich, mAb Gliadin Freedman et al. (146) Immunoblotting Gliadin Janssen et al. (147) Sandwich, pAb Gliadin Ayob et al. (148) Competitive, pAb Gliadin Friis (149), Chirdo et al. (150) Sandwich, mAb a-Gliadin, g-gliadin Mills et al. (151) Sandwich, mAb AOAC o-Gliadin, glutenin Hill and Skerritt (152), approved laboratory subunits Skerritt et al. (153), test Skerritt and Hill (154,155) Sandwich, mAb Home kit o-Gliadin, glutenin Skerritt and Hill (156) subunits Sandwich-ELISA Gliadin Hekkens and Twist de Graaf (157) Sandwich, mAb Gliadin peptides Ellis et al. (158), Denery- Competitive Papini (159), Ellis et al. (160), Nicolas et al. (161)

Gliadin was accurately determined in human milk. Eleven of 18 lactating mothers who had consumed gluten meals 2 hours prior to testing showed gliadin-reactive material in their breast milk. Gliadin was detectable in human plasma after heat treatment to denature circulating human antibodies. Thermal treatment (1218C, 5 minutes) had no adverse effect on assay sensitivity. There was apparently no immunoreactive material in blood plasma from celiac patients (145). By contrast, Lane et al. (162) found 320 Chapter 11 signi®cantly high levels of gluten in the serum of patients suffering from dermatitis herpetiformis and celiac disease (compared with normal controls). Different assay conditions used in these studies may explain the lack of agreement. It is perhaps signi®cant that Lane and co-workers calibrated their assay by adding gliadin to human plasma. The performance of a low-cost sandwich ELISA test for gluten was compared with two commercial gluten tests produced by Cortecs Ltd. (UK) and R-Biopharm (Germany) (163). Results for the in-house test were highly correlated (R 0.0.9967) with both commercial tests. The former was more sensitive than the Cortecs gluten test (see the following). The in-house test had the advantage of a 10-fold lower cost. The linear range for the in-house assay was 50±150 ng mLÀ1 compared to 20±80 ng mLÀ1 for the R-Biopharm test. The LLD values of the two tests were 0.04 and 0.1% mg %, respectively.

4.6. Competitive ELISA for Gluten Using pAb A competitive ELISA test for gliadin (149) had a working range of 10±250 ng mLÀ1 with an LLD of 1 ng mLÀ1. The highly accurate test had an interassay precision of 33%. There was speci®city for a-, b-, g-, and o- gliadin as well as barley, oats, and rye prolamins. No responses were obtained for maize, soya, or millet protein. Buckwheat gave a low response. The competitive ELISA test for gluten was not tested for heat-processed samples. Chirdo et al. (150) used (rabbit) pAb for competitive ELISA for gluten in a range of processed foods including cake ¯our, breakfast cereal, processed meat, soup, and chocolate. The working range for gliadin standards was 0.6 ng mLÀ1 to 10 mgmLÀ1. The LLD of 1 ng mLÀ1 (0.002 mg %) was the same as reported by Friss (149). The average within-assay precision was about 9%. It was claimed that this test was useful for thermally processed foods.

4.7. The AOAC-Approved ELISA Test for Gluten The AOAC-approved test for gluten employs mAbs for capture and detection (154). Skerritt and Smith (133) and Skerritt (142) found two mAbs (mAb-401/21 and mAb-304/13) with speci®city for o-gliadin, HMW and LMW glutenin subunits. Because o-gliadin is heat resistant (Section 4.3), the preceding tests may be suitable for the analysis of heat-processed foods (see later). Either mAb-401/21 or mAb-304/13 could be used for capture or detection, leading to similar results. Preliminary tests (153) showed speci®city for wheat (bread or durum) > rye*barley 4 maize > oats. No responses were observed for oat, maize, or rice protein at levels comparable Determination of Trace Protein Allergens 321 to those for wheat gluten. Sorghum was not tested. Assay performance depended on these factors (152,153,154): 1. Extraction solvent. The optimal solvent for gluten extraction from a range of food samples was 40 % (v/v) ethanol. Using 70% (v/v) ethanol, 1 M urea, or 1 mM HCl as solvent led to underestimation or overestimation of the gluten content. 2. Form of extraction. Homogenization of ¯our samples using an omnimixer or Ultraturrax mixer for about 30 seconds led to accurate analysis. Vortexing for 30 or 60 seconds duration (four times per hour) led to inaccuracy, probably due to shear-induced precipitation of gluten. 3. Choice of gluten standard. Normal gluten is a suitable standard. This is relatively soluble, stable in the freeze-dried form, and usable over a wide concentration range (dilutions of up to 10,000À1 were used for analysis). 4. Choice of solid phase. Polystyrene microwell plates were preferable to PVC plates, which produced high nonspeci®c binding. Back- ground adsorption of gluten was not ameliorated by a range of blocking solutions including PBST with 0.05±5% BSA, 0.65 M NaCl, or 2% Tween. 5. mAb characteristics. The nature and concentration of the capture and detector antibody and coating conditions (time and temperature of coating) affected the assay. Following optimization, the linear dynamic range of the AOAC- approved ELISA test was 0.0075±5 mg mLÀ1 (15 mg % to 10% w/w gluten in actual foods). The LLD was 0.1±0.2 mgmLÀ1, which is equivalent to 0.2±0.38-mg gliadin per 100 g of ¯our (0.2 ±0.38 mg %). A partial list of food types successfully analyzed includes starches (wheat, maize), cake ¯ours (plain, self-raisin, bread crumb, cookie), gluten-free bread mixes, processed meat (cooked beef, ham sausages, salami), baby foods (beef or chicken based with and without ¯our thickener), breakfast cereals, baked goods (bread crumb mix, sweet cookies, crisp bread), soups, confectionery (caramel, chocolate), and others (lentils, eggs, milk powder). 322 Chapter 11

4.8. Collaborative Testing of AOAC-Approved ELISA Test for Gluten Prior to AOAC approval, the previous ELISA was subjected to a collaborative trial by 15 laboratories* (155). Eighteen samples including ®ve prestudy samples (three wheat starches and two meat-gluten blends) with known amounts of added gluten were tested. Participating laboratories familiar with ELISA methodology were supplied with commercial versions of the Skerritt-Hill test. The linear dynamic range for gluten was 16 mg % to 11% (w/w) basis. Assay reproducibility was 24±33% with a repeatability of 19±22%. Gluten was accurately determined in a range of foods (164). The accuracy of the Skerritt-Hill ELISA test is indicated by its high precision. Gluten levels reported by collaborators were less than +12% different from expected values. The collaborative study led to AOAC approval for the Skerritt-Hill ELISA test for gluten. Potential limitations of the AOAC assay for gluten have been suggested (165). A minicollaborative trial using a limited number of gluten test kits was organized by the Celiac Society of Great Britain. Participating groups were major UK clinical laboratories with longstanding interest in celiac disease.{ Results from ®ve laboratories agreed with respect to four gluten-free foods and nine gluten-containing foods (spaghetti bolognese, egg and bacon breakfast, crispbread, malted drink, porridge, barley, plain ¯our, and wheat starches). By contrast, large disparities (®ve- to sixfold differences) were reported for gluten levels in 11 foods including egg and bacon, beer, low-alcohol lager, stout, gluten-free ¯our, gravy powder, corn¯our, and rice pudding. An mAb-based gluten test developed by Mills et al. (151) showed a linear range comparable to that of the AOAC test. They used pAbs from chicken{ for capture while mAb-IFRN 033

* The list of participants includes M. Billington (City of Birmighman Public Analysts, Birmingham, UK), A. Crimes, (Unilever, Bedford, UK), C. Cuncliffer, (Somerset County Council, Taunton, UK), M. Cutrufelli, (USDA, Betsville, MD), T. McKenny (Cerestar, Manchester, UK), M. Murlry, (Kraft Research Center, Glenview, IL), N. Paterl, (Campden, Food and Drink Research Association, Chipping Campden, UK), J. Rhodes, and D. Lord, (Lancashire County Lab., Preston, UK), B. Ritter, (ABC Research, Gainesville, FL), M. Scooter, (Food Science Division, MAFF, London, UK), M. Smith, (Avon County Scienti®c Services, Bristol, UK), C. Stanley, (Laboratory of Government Chemists, Teddington, UK), P. Sutton, and S. Cooper, (British Food Manufacturing Industries Research Association, Letherhead, JK), and B. Taylor and D. Ansell, (Greater Manchester Public Analyst, Manchester, UK). { Participating institutions were St James University Hospital (Leeds), St Bartholomew's Hospital (London), Western General Hospital (Edinburgh), Radcliff In®rmary (Oxford), and Bristol Royal In®rmary (Bristol). { The pAb is recovered from the eggs of immunized chicken. Determination of Trace Protein Allergens 323 functioned as the detector antibody. The mAb-IFRN 033 was speci®c for a- gliadin and g-gliadin, which together constitute *85% of gliadin. The linear range for gluten was 0.25±2.5 mgmLÀ1 with the LLD being 0.1 mgmLÀ1. The preceding tests may be compared with the 1987 sandwich ELISA test developed by Freedman et al. (146) from Guys and St Thomas's Hospital, London. They used (rabbit) pAb for capture and mAb for unfractionated gliadin for detection. The bound mAb was visualized with alkaline phosphatase±labeled (goat) pAb for murine IgG/IgM. The linear dynamic range for this assay was 10 ng mLÀ1 to 1 mgmLÀ1. Strong wheat ¯our contained 3.7% (w/w) gliadin-equivalent proteins. Several brands of ``gluten-free'' ¯our produced from wheat starch had 1.9±3.3 mg % of gliadin. The day-to-day precision of the assay was better than 5%.

4.9. A Cocktail mAb-Based Sandwich ELISA Test for Gluten Gluten tests should be equally sensitive to all celiac-active proteins from wheat, barley, or rye. However, proteins from celiac-negative cereals (maize, rice, millet, etc.) should not interfere. Single mAbs had different sensitivities for different celiac toxic prolamins (166). Combining different mAbs is one way to ensure cross-reactivity for a range of celiac-active prolamins. A cocktail antibody test for gluten was developed by mixing two capture mAbs (mAb-13B4 and mAb-Rye5) with speci®city for barley and rye (167). A third mAb for rye protein (mAb-Rye3) was chosen for detection. Results compared favorably with the commercial AOAC-ELISA test for gluten. The cocktail sandwich ELISA produced comparable calibration graphs for gliadin, hordein, and secalin with a linear range of 3±100 ng mLÀ1 and an LLD of 1.5 ng (gliadin) mLÀ1, 0.05 ng (hordein) mLÀ1, 0.15 ng (secalin) mLÀ1, or 12 ng (avenin) mLÀ1. The acuity toward barley and rye prolamin was higher than obtained for wheat. The response toward avenin was 10±100 times lower, which is as expected from the lower toxicity of oats. Compared with the AOAC test, the cocktail mAb test had an LLD for hordein or gliadin that was *25-fold and 4- to 10-fold lower. The mAb- Rye3 used for detection in the cocktail ELISA had low speci®city for promlamins from wheat, barley, and rye. This feature probably explains the relatively high response for barley and rye.

4.10. A Home Test for Gluten The Skerritt and Hill (156) home test kit for gluten performed well with ordinary citizens, who successfully identi®ed gluten-free foods with 82± 100% accuracy. The home test agreed closely with results from the AOAC- approved laboratory test kit. The home test kit can be usefully compared 324 Chapter 11 with the home pregnancy test kits, which are now commonplace. Similar self-use kits are possible for other food allergens. The home-gluten test consists of the following components: 1. Polystyrene test tubes (Nunc, Denmark) precoated with mAb-401/ 21 and blocked with 1% BSA. The antibody-coated test tubes are stable at 48C for 12 months or at 208C for 6 months. 2. Graduated tubes for sample extraction. 3. Enzyme-labeled mAb-402/21 for detection. 4. Substrate solution (TMB/hydrogen peroxide). The kit was ®eld tested in eastern Australia by 5 dieticians or food technologists and 47 ordinary citizens registered with the Celiac Society. Participants were 10- to 77-year-old urban and rural dwellers of varied educational background. The testers were given six food samples to grade as having ``low,'' ``borderline,'' or ``high'' gluten. Samples classed in the borderline (40 mg %)orhigh(>150 mg %) gluten categories were not acceptable as ``gluten-free'' foods, whereas those with <10 mg % were placed in the gluten-free category. For sample pretreatment, 0.5 g of ground or ®nely chopped foodstuff was shaken with 5 mL of dilute hydrochloric acid (2 mM) for 1 minute. There was 100% protein recovery from wheat starch, 62% recovery from wheat starch having 100 mg % gluten, and 18% recovery from wheat ¯our with 11% gluten. Home testing for gluten involved ®ve simple steps: (a) add 0.8 mL of reaction buffer (1% BSA solution in PBST) to the antibody-coated test tubes provided, (b) add one drop (30 mL) of food extract and mix gently for 30 1 seconds, (c) add three drops of HRP-mAb conjugate, (d) wait about 1 =2 minutes and rinse the tubes with running tap water, and (e) add enzyme substrate solution. A positive test gives a blue coloration within 2 minutes. Classify tests results as high, low, or borderline by comparing results with color produced by standard samples.

4.11. Gluten ELISA Tests Using Peptide Antigens As the benchmark technique, the AOAC-approved ELISA test has been subject to a range of criticisms, not all of which are justi®ed: 1. Low accuracy. The approved method measures o-gliadin. How- ever, the amount of o-gliadin fraction may be only weakly correlated with the total gliadin content (168). Determining total gluten by extrapolation from the o-gliadin may lead to error. 2. Low selectivity. There is some cross-reactivity with cereal proteins not proved to be celiac toxic (e.g., starch granule proteins). Determination of Trace Protein Allergens 325

3. Poor limit of detection. The LLD achieved by the AOAC- approved gluten test may be inadequate to deal with future, more stringent legislation regarding gluten content. 4. Unproven range of applicability. Although maize, rice, and oats were tested by the approved AOAC method, it may be desirable to test a wider range of materials. 5. Low reliability for certain foods. A commercial ELISA kit based on the approved method gave discordant results for 11 out of 24 foods measured in ®ve laboratories (Section 4.8).

Peptides are useful antigens for developing gluten ELISA tests. A sandwich ELISA for gluten was developed using mAb-WB8 with speci®city for peptide B3144 as the detector (158). B3144 consists of the celiac-toxic N- terminal residues 3±56 from a-gliadin (129). The capture antibody was (rabbit) pAb for unfractionated gliadin. The mAb-WB8 bound a-gliadin, b- gliadin, g-gliadin, o-gliadin, glutenin subunits, and unfractionated gliadin. There was also high reactivity toward rye and barley prolamins. The response for celiac-active proteins was 125±4000 times greater than that with celiac-negative cereals such as oats, maize, millet, or sorghum. Very low binding occurred with ovalbumin or BSA (Fig. 2).

FIGURE 2 Speci®city of two mAbs for celiac-active peptide B3144. Speci®city was tested by noncompetitive indirect ELISA using cereal proteins extracted with 50% (v/v) ethanol. The Y-axis shows mAb titer from ascite ¯uid. (Drawn from results in Ref. 129.) 326 Chapter 11

The working range for gliadin detection was 2 ng mLÀ1 to 1 mgmLÀ1. The LLD ( 1.5 SD above background absorbance) was 15 ng mLÀ1 (extract) or 0.03 mg % of gliadin in wheat ¯our. The corresponding LLD values for other cereals were 0.03 mg % (rye), 0.25 mg % (barley), and 0.5 mg % (oats). The LLD for gliadin is four orders of magnitude lower than the current UK upper limit (300 mg %) for gluten-free foods. The preceding assay was suitable for uncooked foods only. Prolamins were extracted with 20 volumes of 40±50% (w/w) ethanol before analysis. The peptide-based ELISA test for gluten underwent further re®nements. A 19-residue peptide fragment from a-gliadin N-terminal (residues 31±49) was used as antigen. By comparison with mAb-WB8, the new detector antibody (mAb-PN3) (160) was more selective for unfractionated gliadin and a-gliadin at the expense of o-gliadin, rye, and barley proteins (Fig. 3). The LLD for gliadin was 4 ng mLÀ1 of extract or 0.008 mg % (¯our). The assay repeatability was 8.7±9.3% with a reproducibility of 18.8±22.3%. Cooking reduced the apparent gluten levels by 70% when measured by the mAb-PN3-based ELISA. An attempt to increase sample solubility by

FIGURE 3 Sandwich ELISA tests for gluten developed with mAbs for peptide antigens. The capture antibody is (rabbit) pAb. The dectector antibody is either mAb-WB8 or mAb-PN3 with speci®city for a-gliadin N- terminal residues 3±56 or 31±49, respectively. (Drawn using results from Refs. 158 and 160.) Determination of Trace Protein Allergens 327 extracting with a reducing buffer was unsuccessful. Residual mercaptoethanol affected the stability of antibody during ELISA. The inability to assay gliadin in heat-processed samples is a serious shortcoming. The test using mAb-PN3 is also likely to underestimate celiac-toxic potential when a product contains ingredients from barley or rye. This is because the limit of detection for wheat protein was 250-fold lower than values for hordein or secalin. The latter proteins will remain ``hidden'' in foods exhibiting relatively low amounts of celiac±active proteins. Denery-Papini and co-workers (159) produced pAbs for a number of synthetic peptides. Each peptide matched a unique amino acid sequence found in one of the four classes of gliadins. Each 8- to 12-residue peptide was covalently coupled to a protein carrier for immunization. The (rabbit) pAbs for these antigens were evaluated by ELISA as well as SDS-PAGE and immunoblot analysis. A pAb speci®c for the N-terminal sequence from g-gliadin or o-gliadin bound only to the corresponding protein. A particular pAb speci®c for the N-terminal sequence of a/b-gliadin was used as the basis for a competitive ELISA test for gliadin (161). Gliadin levels were found to be 2333±5040 mg % using ¯our from 20 wheat cultivars. The results were highly correlated with gliadin determinations by reverse-phase HPLC (R 0.82) and Kjeldahl analysis (0.86). The linear range for analysis was 45±250 mgmLÀ1 (extract) with an LLD of *9 mgmLÀ1. Stirring wheat ¯ours with 70% (w/w) ethanol for 14 hours led to the solubilization of 48% to the total protein. The solubilized protein was gliadin with little contamination from glutenin. Results from ELISA tests for gluten are summarized in Table 9. This information was compiled from the limited number of studies involving real foods (140,141,144,146,149,150,154,160). The agreement between gluten values is reasonable. Results are routinely reported in milligrams of gliadin (equivalent protein) per 100 g dry food (mg %). Measuring gluten in this fashion is convenient because ELISA tests are usually calibrated using whole gliadin extract. The ideal assay for gluten should show equal responsivity toward prolamins from all celiac-active cereals. It is necessary to test samples of alcohol-soluble proteins from each cereal. The results in Table 9 do not include contributions from glutelin subunits, which (although celiac active) dissolve poorly in 40±70% (v/v) aqueous alcohol. Glutenin could be more effectively solubilized for immunoassay using solvents containing SDS or guanidine hydrochloride. Conroy and Esen (169) found that zein was ef®ciently adsorbed by polystyrene microwell plates from acetic acid (60% v/v) or 6 M urea (dissolved in carbonate coating buffer). Sample extraction with these solvents could improve the recovery of glutenin. 328 Chapter 11

TABLE 9 Apparent Gluten Levels in Range of Foods Determined by ELISA for Gliadin

Foodstuff or commodity Gliadin (mg %)

Gluten-negative foods Tropical cereals Millet, sorghum, rice, jawar Negative Legumes Soya ¯our, chickpea ¯our, buckwheat, quinoa Miscellaneous Potato ¯our, molasses, skimmed milk powder, sugar beet Gluten-free (wheat starches) Kinderm, Gluta®n, Wheatex, Nutregen 0.5±7.75 Wheat starch-based products Cooked foodsa Gluta®n GF ®ber loaf, Gluta®n GF white bread with 0.12±0.24 soya, Lopro®n white bread, low-protein chocolate chip cookies Uncooked products Gluta®n GF white mix, ®ber mix, Rite Diet low- 0.9±1.4 protein ¯our mix Wheat-free products Cooked products Chocolate digestive biscuits, custard creams, tea biscuits, 0.144±1.18 orange ¯avor cream wafers, mince pies Uncooked products Gluta®n pasta spirals, vermicelli, spaghetti, macaroni 0.07±6.7 Flours Buckwheatb 4 Sorghumb 13 Maizeb 46 Riceb 47±57 Oatsb 100±240 Barley 400 Rye 580 Wheat (self-raising, plain) 4000±8100 a Levels can be underestimated by 70% after heating. b Sometimes nil gliadin is found in these cereals.

A meaningful comparison of ELISA tests results for gluten is dif®cult because successive investigators aimed for innovation rather than standardization (170). Slight changes in assay conditions can produce substantial changes in assay performance. The following general observations seem Determination of Trace Protein Allergens 329 relevant. The most sensitive tests are ELISA using pAb. The LLD for such assays is about 0.5±1 ng mLÀ1. To achieve high acuity it is necessary to avoid nonspeci®c binding of gliadin to microwell plates. Exposing microwell plates to 3% skimmed milk powder or BSA for 60±120 minutes reduces nonspeci®c protein binding. Precoating with extraneous protein is better than adding the coating protein together with the sample. Coating of microwell plates at 48C overnight leads to better assay performance in contrast to coating at 378C for 90 minutes. Using mAb in place of pAb increases the cost of ELISA tests by an estimated 10-fold. On the other hand, mAbs exhibit increased speci®city and decreased likelihood of detecting celiac-negative cereals. Immunoaf®ntity-puri®ed pAbs show high selectivity for celiac- active cereals and are more frequently employed than mAbs.

REFERENCES

1. JA Anderson. Allergic reactions to foods. CRC Crit Rev Food Sci 36(S):S19± S38, 1996. 2. M Blades. Food allergy and food intolerance. Food Sci Tehnol Today 10:82± 86, 1996. 3. SM Jones. The role of food allergy and other allergic disease in atopic dermatitis. Clin Rev Allergy Immunol 17:293±321, 1999. 4. AW Burks, HA Sampson. Anaphylaxis and food allergy. Clin Rev Allergy Immunol 17:339±360, 1999. 5. SL He¯e, H Nordlee, SL Taylor. Allergenic foods. Crit Rev Food Sci Nutr 36(S):S69±S89, 1996. 6. SL Taylor, SL He¯e, A Munoz-Furlong. Food allergies and avoidance diets. Nutr Today 34:15±22, 1999. 7. T Matsuda, R Nakamura. Molecular structure and immunological properties of food allergens. Trends Food Sci 41:289±293, 1993. 8. RR Bush, LS He¯e. Food allergens. CRC Crit Rev Food Sci Nutr 36(S):S119± S163, 1996. 9. SL He¯e. The chemistry and biology of food allergens. Food Technol 50(3):86, 88±92, 1996. 10. EJ Campbell. AllergensÐgovernment standpoint. Manuf Confect 79(12):103± 105, 1999. 11. D Amor. Food Allergies. Food Ind News (May):10±11, 1998. 12. JA Nordlee, SL Taylor, JA Townsend, LA Thomas, RK Bush. Identi®cation of a Brazil-nut allergen in transgenic soybeans. N Engl J Med 334:688±692, 1996. 13. M Nestle. Allergies to transgenic foodsÐquestions of policy. N Engl J Med 334:726±728, 1996. 14. World Health Organization. Gluten-free foodsÐCanada. Int Dig Health Legisl 47(1):50, 1996. 330 Chapter 11

15. United Kingdom Ministry of Agriculture, Fisheries & Food (MAFF). Food Advisory Committee Annual Report 1998. London: MAFF Publications, 1999. 16. JD Astwood, JN Leach, RL Fuchs. Stability of food allergens to digestion in vitro. Nat Biotechnol 14:1269±1273, 1996. 17. JA Nordlee, SL Taylor, RT Jones, JG Yunginger. Allergenicity of various peanut products as determined by RAST inhibition. J Allegy Clin Immunol 68:376±382, 1981. 18. JW Yunginger, MB Gauerke, RT Jones, M-JE Dahlberg, SJ Ackerman. Use of radioimmunoassay to determine the nature, quantity and source of allergenic contamination of sun¯ower buffer. J Food Prot 46:625±628, 1983. 19. H Towbin, T Staehelin, J Gordon. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Natl Acad Sci USA 76:4350, 1979. 20. ER Tovey, BA Baldo. Characterization of allergens by protein blotting. Electrophoresis 8:452±463, 1987. 21. M Shibasaki, S Suzuki, S Tajima, H Nemoto, T Karoume. Allergenicity of major component proteins of soybean. Int Arch Allergy Appl Immunol 61:441±448, 1980. 22. AW Bucks Jr, JR Brooks, HA Sampson. Allergenicity of major component proteins of soybean determined by enzyme linked immunosorbent assay (ELISA) and immunoblotting in children with atopic dermatitis and positive soy challenges. J Allergy Clin Immunol 81:1135±1142, 1988. 23. RK Bush, D Schroeckenstein, S Meiner-Davis, J Balmes, D Rempel. Soybean ¯our asthma. Detection of allergens by immunoblotting. J Allergy Clin Immunol 82:251±255, 1988. 24. AM Herian, SL Taylor, RK Bush. Identi®cation of soybean allergens by immunoblotting with sera from soy-allergic adults. Int Arch Allergy Appl Immunol 92:193±198, 1990. 25. T Ogawa, N Bondo, H Tsugi, H Okajima, K Nishikawa, K Sasaoka. Investigations of the IgE-binding proteins in soybeans by immunoblotting with the sera of the soybean-sensitive patients with atopic dermatitis. J Nutr Sci Vitaminol 37:555±565, 1991. 26. E-F Ebabiker, A Hiroyuki, N Matsodomi, H Iwata, T Ogawa, N Bando, A Kato. Effect of polysaccharide conjugation or transglutaminase treatment on the allergenicity and functional properties of soy protein. J Agric Food Chem 46:866±871, 1998. 27. MI Sachs, RT Jones, JW Yunginger. Isolation and partial characterization of a major peanut allergen. J Allergy Clin Immunol 67:27±34, 1981. 28. D Barnett, BA Baldon, MEH Howden. Multiplicity of allergens in peanuts. J Allergy Clin Immunol 72:61±68, 1983. 29. AW Bucks, LW Williams, RM Helm, C Connaughton, G Cockrell, T O'Brien. Identi®cation of a major peanut allergen, Ara h 1, in patients with atopic dermatitis and positive peanut challenges. J Allergy Clin Immunol 88:172±179, 1991. Determination of Trace Protein Allergens 331

30. AW Bucks, LW Williams, C Cannaughton, G Cockrell, TJ O'Brien, RM Helm. Identi®cation and characterization of a second major peanut allergen, Ara h 2, with use of the sera of patients with atopic dermatitis and positive peanut challenge. J Allergy Clin Immunol 90:962±969, 1992. 31. L Uhlemann, W-M Becker, M Schlaak. Food allergy: identi®cation and characterization of peanut allergens with patients' sera and monoclonal antibodies. Z Ernaehrungswiss 32:139±151, 1993. 32. SL He¯e, JP Folgert, RK Bush, FS Chu. Monoclonal antibodies against selected peanut allergens: production and use as af®nity agents. Food Agric Immunol 6:197±208, 1994. 33. EC de Jong, M van Zijverden, S Spanhaak, SJ Koppelman, H Pellegrom, AH Pernninks. Identi®cation and partial characterization of multiple major allergens in peanut proteins. Clin Exp Allergy 28:743±751, 1998. 34. DR Hoffmann, ED Day, JS Miller. The major heat stable allergen of shrimps. Ann Allergy 47:17±22, 1981. 35. S Nagpal, L Rajappa, DD Metcalfe, PV Subba Rao. Isolation and characterization of heat stable allergens from shrimp (Penaeus indicus). J Allergy Clin Immunol 64:259, 1989. 36. SB Lehrer, MD Ibanez, ML McCants, CM Daul, JE Morgan. Characteriza- tion of water-soluble shrimp allergens released during boiling. J Allergy Clin Immunol 85:1000±1013, 1990. 37. G Reese, R Ayuso, SB Lehrer. Tropomyosin: an invertebrate pan-allergen. Int Arch Allergy Immunol 119:247±258, 1999. 38. DW Hide. Clinical curio: allergy to Brazil nut. Br Med J 287:900, 1983. 39. NL Morgan, CJ Albert, K Spears. The use of image analysis to study the allergens in foods. Tech Ser (Soc Appl Bacteriol) 24:153±161, 1987. 40. JM Borja, B Bartolome, E Gormez, PA Galindo, F Feo. Anaphylaxis from Brazil nut. Allergy 54:1007±1008, 1999. 41. TJ Bargman, JH Rupnow, SL Taylor. IgE-binding proteins in almonds (Prunus amygdalus), identi®cation by immunoblotting with sera from almond- allergic adults. J Food Sci 57:717±720, 1992. 42. JJ Hlywka, SL He¯e, SL Taylor. A sandwich enzyme-linked immunosorbent assay for the detection of almonds in foods. J Food Prot 63:252±257, 2000. 43. DR Hoffman. Immunochemical identi®cation of the allergens in hen's egg white. J Allergy Clin Immunol 71:481±486, 1983. 44. TA Langeland. A clinical and immunological study of allergy to hen's egg white. IV. Speci®c IgE antibodies to individual allergens in hen's egg white related to clinical immunological parameters in egg-allergic patients. Allergy 38:493±500, 1983. 45. V Leduc, C Demuelemester, B Polack, C Guizard, L Guern, G Peltre. Immunochemical detection of egg-white antigens and allergens in meat products. Allergy 54:464±472, 1999. 46. G Ball, JJ Shelton, BJ Walsh. A major continuous allergenic epitope of bovine beta-lactoglobulin recognized by human IgE binding. Clin Exp Allergy 24:758±764, 1994. 332 Chapter 11

47. P Restani, A Gaiaschi, A Plebani, B Beretta, G Cavagni, A Fiocchi, C Poiesi, T Velona, AG Ugazio, CL Galli. Cross-reactivity between milk proteins from different animal species. Clin Exp Allergy 29:997±1004, 1999. 48. A Rosendal, V Barkholt. Detection of potentially allergenic material in 12 hydrolyzed milk formulas. J Dairy Sci 83:2200±2210, 2000. 49. JA Nordlee, SL Taylor. Immunological analysis of food allergens and other food proteins. Food Technol 49(2):29±132, 1995. 50. SL Taylor, SB Lehrer. Principles and characteristics of food allergens. CRC Crit Rev Food Sci Nutr 36(S):S91±S118, 1996. 51. SL He¯e, SL Taylor. Allergenicity of edible oils. Food Technol 53(2):62±68, 70, 1999. 52. T Ogawa, H Tsuji, N Bando, K Kitamura, Y-L Zhu, H Hirano, K Nishikawa. Identi®cation of the soybean allergenic protein, Gly m Bd 30K, with soybean seed 34-kD oil-body-associated protein. Biosci Biotechnol Biochem 57:1030± 1033, 1993. 53. A Kalinski, JM Weisemann, BF Matthews, EM Herman. Molecular cloning of a protein associated with soybean seed oil bodies that is similar to thiol proteases of the papain family. J Biol Chem 265:13843±13848, 1990. 54. A Kalinski, DL Melroy, RS Dwivedi, EM Herman. A soybean vacuolar protein as a glycoprotein precursor during seed maturation. J Biol Chem 267:12068±12076, 1992. 55. RW Yaklich, RM Helm, G Cockrell, EM Herman. Analysis of the distribution of the major soybean seed allergens in a core collection of Glycine max accessions. Crop Sci 39:1444±14447, 1999. 56. T Ogawa, N Bando, H Tsuji, K Nishikawa, K Kitamura. a-Subunit of b- conglycinin, an allergenic protein recognized by IgE antibodies of soybean- sensitive patients with atopic dermatitis. Biosci Biotechnol Biochem 59:831± 833, 1995. 57. H Tsuji, N Bando, M Hiemori, R Yamanishi, M Kimoto, K Nishikawa, T Ogawa. Puri®cation and characterization of soybean allergen Gly m Bd 28K. Biosci Biotechnol Biochem 61:942±947, 1997. 58. MG Zeece, TA Beardslee, JP Markwell, G Sarath. Identi®cation of an IgE- binding region in soybean acidic glycinin G1. Food Agric Immunol 11:83±90, 1999. 59. H Hosoyama, A Obata, N Bando, H Tsuji, T Ogawa. Epitope analysis of soybean major allergen Gly m Bd 30K recognized by the mouse monoclonal antibody using overlapping peptides. Biosci Biotechnol Biochem 60:1181± 1182, 1996. 60. H Tsuki, H Bando, M Kimoto, N Okada, T Ogawa. Preparation and application of monoclonal antibodies for a sandwich enzyme-linked immunosorbent assay of the major soybean allergen, Gly m Bd 30 K. J Nutr Sci Vitaminol 39:389±397, 1993. 61. H Tsuji, N Okada, R Yamanishi, N Bando, M Kimoto, O Tadashi. Measurement of Gly m Bd 30K, a major soybean allergen in soybean product Determination of Trace Protein Allergens 333

by a sandwich enzyme-linked immunosorbent assay. Biosci Biotechnol Biochem 59:150±151, 1995. 62. R Yamanishi, H Tsuji, N Bando, Y Yamada, Y Nadaoka, Tao-Huang, K Nishikawa, S Emoto, T Ogawa. Reduction of the allergenicity of soybean by treatment with proteases. J Nutr Sci Vitaminol 42:581±587, 1996. 63. JM Yeung, PG Collins. Determination of soy proteins in food products by enzyme immunoassay. Food Technol Biotechnol 35:209±214, 1997. 64. CHS Hitchcock, FJ Bailey, AA Crimes, DAG Dean, PJ Davis. Determination of soya proteins in food using an enzyme-linked immunosorbent assay procedure. J Sci Food Agric 32:157±165, 1981. 65. NM Grif®ths, MJ Billington, W Grif®ths. A review of three modern techniques available for the determination of soya protein in meat products. J Assoc Public Anal 19:13±119, 1981. 66. NM Grif®ths, MM Billington, AA Crimes, CHS Hitchcock. An assessment of commercially available reagents for an enzyme-linked immunosorbent assay (ELISA) of soya protein in meat products. J Sci Food Agric 35:1255±1260, 1984. 67. AA Crimes, CHS Hitchcock, R Wood. Determination of soya protein in meat products by an enzyme-linked immunosorbent assay procedure: collaborative study. J Assoc Public Anal 22:59±78, 1984. 68. SE Emmett, FJ Angus, JS Fry, PN Lee. Perceived prevalence of peanut allergy in Great Britain and its association with other atopic conditions and with peanut allergy in other household members. Allergy 54:380±385, 1999. 69. RSH Pumphrey, PB Wilson, EB Faragher, SR Edwards. Speci®c immunoglobulin E to peanut, hazelnut and Brazil nut in 731 patients: similar patterns found at all ages. Clin Exp Allergy 29:256±259, 1999. 70. PW Ewan. Clinical study of peanut and nut allergy in 62 consecutive patients: new features and associations. Br Med J 312:1074±1078, 1996. 71. DLM Goh, YN Lau, FT Chew, LPC Shek, BW Lee. Pattern of food-induced anaphylaxis in children of an Asian community. Allergy 54:84±86, 1999. 72. HA Sampson, L Mendelson, JP Rosen. Fatal and near-fatal anaphylactic reactions to food in children and adolescents. N Engl J Med 327:380±384, 1992. 73. DA Moneret-Vautrin, F Rance, G Kanny, A Olsewski, JL Gueant, G Dutau, L Guerin. Food allergy to peanuts in FranceÐevaluation of 142 observations. Clin Exp Allergy 28:1113±1119, 1998. 74. RS Dawe, J Ferguson. Allergy to peanut. Lancet 348:1522±1523, 1996. 75. KL Donovan, J Peters. Vegetableburger allergy: all was nut as it appeared. Br Med J 300:1378, 1990. 76. DA Moneret-Vautrin, R Hatachet, G Kanny, Z Ait-Djafer. Allergenic peanut oil in milk formulas. Lancet 338:1149, 1991. 77. A Olszewki, L Pons, F Moutete, I Aimone-Gastin, G Kanny, DA Moneret- Vautrin, JL Gueant. Isolation and characterization of proteic allergens in re®ned peanut oil. Clin Exp Allergy 28:850±859, 1998. 334 Chapter 11

78. RSH Pumphrey, PB Wilson, AS Bansal. Peanut allergy. Lancet 352:741±742, 1998. 79. PW Ewan. Prevention of peanut allergy. Lancet 352:4±5, 1998. 80. DE Fox, G Lack. Peanut allergy. Lancet 352:741, 1998. 81. HA Sampson. Managing peanut allergy. Br Med J 312:1050±1051, 1996. 82. JOB Hourihane. Peanut allergyÐcurrent status and future challenges. Clin Exp Allergy 27:1240±1246, 1997. 83. W Burks, HA Sampson, GA Bannon. Peanut allergens. Allergy 53:725±730, 1998. 84. W Burks, GA Bannon, S Sicherer, HA Sampson. Peanut-induced anaphylactic reactions. Int Arch Allergy Immunol 119:165±172, 1999. 85. JO Warner. Peanut allergy: a major public health issue. Pediatr Allergy Immunol 10:14±20, 1999. 86. CO Johns, DB Jones. The proteins of the peanut, Arachis hypogaea. 1. The globulins arachin and conarachin. J Biol Chem 28:77±87, 1916. 87. NJ Neucere. Isolation of a-arachin, the major peanut globulin. Anal Biochem 27:15±24, 1969. 88. KJ Shetty, MSN Rao. Studies on groundnut proteins. V. Physico-chemical properties of arachin prepared by different methods. Anal Biochem 73:458± 470, 1976. 89. RYY Chiou. Effects of heat treatments on peanut arachin and conarachin. J Food Biochem 14:219±232, 1990. 90. PV Monteiro, V Prakash. Effect of proteases on arachin, conarachin 1, and conarachin II from peanut (Arachis hypogaea L). J Agric Food Chem 42:268± 273, 1994. 91. EH Shokraii, A Esen. Characterization of disul®de-linked forms of seed protein in peanuts (Arachis hypogaea L.). J Agric Food Chem 40:1491±1495, 1992. 92. NJ Neucere. Effect of heat on peanut proteins. I. Solubility properties and immunochemical-electrophoretic modi®cations. J Agric Food Chem 20:252± 255, 1972. 93. NJ Neucere. Antigenic and electrophoretic changes of a-arachin after heating in-vitro. J Agric Food Chem 22:146±148, 1974. 94. JP Cherry, KH McWatters, MR Holmes. Effect of moist heat on solubility and structural components of peanut proteins. J Food Sci 40:1199±1204, 1975. 95. SJ Koppelman, CAFM Bruijnzee-Koomen, M Hessing, HHU de Jongh. Heat- induced conformational changes of Ara h 1, a major peanut allergen, do not affect its allergenic properties. J Biol Chem 274:4770±4777, 1999. 96. DS Shin, CM Compadre, SJ Maleki, RA Kopper, H Sampson, SK Huang, AW Burks, GA Bannon. Biochemical and structural analysis of the IgE binding sites on Ara h1, an abundant and highly allergenic peanut protein. J Biol Chem 273:13753±13759, 1998. 97. JD Astwood, JN Leach, RL Fuchs. Stability of food allergens to digestion in vitro. Nat Biotechnol 14:1269±1273, 1996. Determination of Trace Protein Allergens 335

98. S Vieths, J Reindl, IJ Mueller, A Hoffmann, D Haustein. Digestibility of peanut and hazelnut allergens investigated by a simple in vitro procedure. Eur Food Res Technol 209:379±388, 1999. 99. MD Perez, ENC Mills, N Lambert, IT Johnson, MRA Morgan. The use of anti-soya globulin antisera in investigating soya digestion in vivo. J Sci Food Agric 80:513±521, 2000. 100. SL He¯e, RK Bush, JW Yunginger, FS Chu. A sandwich enzyme-linked immunosorbent assay (ELISA) for the quantitation of selected peanut proteins in foods. J Food Prot 57:419±423, 1994 101. SJ Koppleman, H Bleeker-Marcelis, G Duijn, M van. Hessing. Detecting peanut allergens. The development of an immunochemical assay for peanut proteins. World Ingredients (December):35±38, 1996. 102. JM Yeung, PG Collins. Enzyme immunoassay for determination of peanut proteins in food products. J Assoc Anal Chem 79:1411±1416, 1996. 103. EN Clare Mills, A Potts, GW Plumb, N Lambert, MRA Morgan. Development of a rapid dipstick immunoassay for the detection of peanut contamination of food. Food Agric Immunol 9:37±50, 1997. 104. T Holzhauser, S Vieths. Indirect competitive ELISA for determination of traces of peanut (Arachis hypogaea L.) protein in complex food matrices. J Agric Food Chem 47:603±611, 1999. 105. B Keck-Gassenmeier, S Benet, C Rosa, C Hischenhuber. Determination of peanut traces in food by a commercially available ELISA test. Food Agric Immunol 11:243±250, 1999. 106. WJ Hurst. Allergen test kits. Manuf Confect 80(8):43±47, 2000. 107. U Srinivasan, N Leonard, E Jones, DD Kasarda, DG Weir, C O'Farrelly, C Feighery. Absence of oats toxicity in adult celiac disease. Br Med J 313:1300± 1301, 1996. 108. T Thompson. Do oats belong in a gluten-free diet? J Am Diet Assoc 97:1413± 1416, 1997. 109. JH Skerritt. Molecular comparison of alcohol-soluble wheat and buckwheat proteins. Cereal Chem 63:365±369, 1996. 110. E Aubrecht, PA Biacs. Immunochemical analysis of buckwheat proteins, prolamins and their allergenic character. Acta Aliment 28:261±268, 1999. 111. J Franken, U Stephan, HE Meyer, W Konig. Identi®cation of alpha-amylase inhibitor as a major allergen of wheat-¯our. Int Arch Allergy Immunol 104:171±174, 1994. 112. BJF Goodwin, PM Rawcliffe. Food allergies associated with cereal products. Food Chem 11:321±338, 1983. 113. BA Baldo, CW Wrigley. Allergies to cereals. Adv Cereal Sci Technol 6:289± 356, 1984. 114. PJ Ciclitira, HJ Ellis. Investigation of cereal toxicity in celiac disease. Postgrad Med J 63:767±775, 1987. 115. HJ Cornell. Wheat proteins and celiac disease. Comments Agric Food Chem 1:289±313, 1988. 336 Chapter 11

116. CM Bruzzone, EH Asp. The cereal science and disease etiology of gluten- sensitive enteropathy. Cereal Foods World 44:109±114, 1999. 117. M Silano, M de Vincenzi. Bioactive antinutritional peptides derived from cereal prolamins: a review. Nahrung 43:175±184, 1999. 118. C Feighery. Celiac disease. Br Med J 319:236±239, 1999. 119. M Hadjivassiliou, RA Gruenewald, GAB Davies-Jones. Gluten sensitivity: a many-headed hydra. Br Med J 318:1710±1711, 1999. 120. MN Marsh. Celiac Disease. Oxford: Blackwell Scienti®c Publications, 1992. 121. GR Corazza, ML Adreani, F Biagi, F Biagi, G Corrao, S Pretolani, G Guilianeli, G Ghironzi, G Gasbarrini. The smaller size of the celiac iceberg in adults. Scand J Gastroenterol 32:917, 1997. 122. C Catassi, IM Raetsch, L Gando®, R Pratesi, E Fabiani, R El-Asmar, M Frijia, I Bearzi, L Vizzoni. Why is celiac disease endemic in the people of the Sahara? Lancet 354:647±648, 1999. 123. V Krishnakumar, I Gordon. The world wheat gluten market. Int Food Ingredients 4:41±42,45, 1995. 124. AS Tatham, PR Shewry, PS Belton. Structural studies of cereal prolamins, including wheat gluten. Adv Cereal Sci Technol 10:1±78, 1990. 125. AL Patey, DJ Evans. Large-scale preparation of gliadin proteins. J Sci Food Agric 24:1229±1233, 1973. 126. AS Thatham, BJ Mi¯in, PR Shewry. The beta-turn conformation in wheat gluten proteins: relationship to gluten proteins. Cereal Chem 62:405±442, 1985. 127. AS Tatham, PR Shewry. The conformation of wheat gluten proteins. The secondary structures and thermal stabilities of a-, b-, g-, and o-gliadins. J Cereal Sci 3:103±113, 1985. 128. JM Devery, V Bender, I Penttila, JH Skerritt. Identi®cation of reactive synthetic gliadin peptides speci®c for celiac disease. Int Arch Allergy Appl Immunol 95:356±362, 1991. 129. HJ Ellis, AP Doyle, H Sieser, RP Sturgess, PJ Ciclitira. Speci®city of monoclonal antibodies to domain I of a-gliadins. Scand J Gastroenterol 28:212±216, 1993. 130. S Tanabe, S Arai, Y Yanagihara, H Mita, K Takahashi, M Watanabe. A major wheat allergen has a Gln-Gln-Gln-Pro-Pro motif identi®ed as an IgE- binding epitope. Biochem Biophys Res Commun 219:290±293, 1996. 131. A Tanabe, S Arai, M Watanabe. Modi®cation of wheat ¯our with bromelain and baking hypoallergenic bread with added ingredients. Biosci Biotechnol Biochem 60:1269±1272, 1996. 132. KH Choi, RA Laursen. Amino-acid sequence and glycan structures of cysteine protease with proline speci®city from ginger rhizome Zingiber of®cinale. Eur J Biochem 267:1516±1526, 2000. 133. JH Skerritt, RA Smith. A sensitive monoclonal-antibody-based test for gluten detection: studies with cooked or processed foods. J Sci Food Agric 36:980± 986, 1985. Determination of Trace Protein Allergens 337

134. A Stevenson, PK McCarthy, M Grif®n. Polyclonal antisera against unheated and heated common wheat speci®c gamma and omega gliadins for the detection of adulteration of durum wheat and durum wheat products with common wheats. Food Agric Immunol 6:435±442, 1994. 135. JH Skerritt. A simple antibody-based test for dough strength. I. Development of method and choice of antibodies. Cereal Chem 68:467±474, 1991. 136. ECN Mills, GM Brett, S Holden, JA Kauffman, MJ Tatton, MRA Morgan. Production of monoclonal antibodies to gluten proteins and their use in developing tests for gluten quality. Food Agric Immunol 7:189±196, 1995. 137. JH Skerritt, JL Andrews, M Blundell, HL Beasley, F Bekes. Applications and limitations of immunochemical analysis of biopolymer quality in cereals. Food Agric Immunol 6:173±184, 1994. 138. H Windemann, F Fritschy, E Baumgartner. Enzyme-linked immunsorbent assay for wheat a-gliadin and whole gliadin. Biochim Biophys Acta 709:110± 121, 1982. 139. P Meier, H Windemann, E Baumgartner. Determination of the a-gliadin content of heated foods containing gluten or being ``gluten free''. Z Lebensm Unters Forsch 178:361±365, 1984. 140. F Fritschy, H Windemann, E Baumgartner. Quantitative determination of wheat gliadins in foods by enzyme-linked immunsorbent assay. Z Lebensm Unters Forsch 181:379±385, 1985. 141. DF McKillop, JP Gosling, FM Stevens, PF Fottrell. Enzyme immunoassay of gliadin in food. Biochem Soc Trans 13:468±487, 1985. 142. JH Skerritt. A sensitive monoclonal-antibody-based test for gluten detection: quantitative immunoassay. J Sci Food Agric 36:987±994, 1985. 143. JH Skerritt, JA Diment, CW Wrigley. A sensitive monoclonal-antibody-based test for gluten detection: choice of primary and secondary antibody. J Sci Food Agric 36:995±1003, 1985. 144. AR Freedman, G Galfre, HJ Ellis, PJ Ciclitira. Detection of wheat gliadin contamination of gluten-free foods by a monoclonal antibody immunobinding assay. Clin Chim Acta 166:323±328, 1986. 145. R Troncone, M Vitale, A Donatiello, E Farris, G Rossi, S Aurrichio. A sandwich enzyme-immunoassay for wheat gliadin. J Immunol Methods 92:21± 23, 1986. 146. AR Freedman, G Galfre, E Gal, HJ Ellis, PJ Ciclitira. Monoclonal antibody ELISA to quantitate wheat gliadin contamination of gluten-free foods. J Immunol Methods 98:123±127, 1987. 147. FW Janssen, G Voortman, JA De Barrij. Detection of wheat gluten, whey protein, casein, ovalbumin, and soy protein in heated meat products by electrophoresis, blotting, and immunoperoxidase staining. J Agric Food Chem 35:563±567, 1987. 148. M Ayob, J Rittenbunburg, JC Allen, CJ Smithe. Development of a rapid enzyme-linked immunosorbent assay (ELISA) for gliadin determination in food. Food Hydrocolloids 2:39±49, 1988. 338 Chapter 11

149. SU Friis. Enzyme-linked immunosorbent assay for quantitation of cereal proteins toxic in celiac disease. Clin Chim Acta 178:261±270, 1988. 150. FG Chirdo, MC Anon, A Fossati. Optimization of a competitive ELISA with pAb for quantitation of prolamins in foods. Food Agric Immunol 7:333±343, 1995. 151. EN Mills, CA Spinks, MRA Morgan. A two-site enzyme-linked immunosorbent assay for wheat gliadins. Food Agric Immunol 1:19±27, 1989. 152. AA Hill, JH Skerritt. Monoclonal antibody based two-site enzyme immunoassays for wheat gluten proteins. I. Kinetic characteristics and comparison with other ELISA formats. Food Agric Immunol 1:147±160, 1989. 153. JH Skerritt, KL Jenkins, AS Hill. Monoclonal antibody based two-site enzyme immunoassays for wheat gluten proteins. II. Speci®city analysis. Food Agric Immunol 1:161±171, 1989. 154. JH Skerritt, AS Hill. Monoclonal antibody sandwich enzyme immunoassays for determination of gluten in foods. J Agric Food Chem 38:1771±1778, 1990. 155. JH Skerritt, AS Hill. Enzyme immunoassay for determination of gluten in foods: collaborative study. J Assoc Off Anal Chem 74:257±264, 1991. 156. JH Skerritt, AS Hill. Self-management of dietary compliance in celiac disease by means of ELISA ``home test'' to detect gluten. Lancet 337:379±382, 1991. 157. WTJM Hekkes, M van Twist de Graaf. What is gluten-freeÐlevels and tolerances in the gluten-free diet. Nahrung 34:483±487, 1990. 158. HJ Ellis, AP Doyle, H Wieser, RP Sturgess, P Day, PJ Ciclitira. Measurement of gluten using a monoclonal antibody to a sequence peptide of a-gliadin from the celiac activating domain 1. J Biochem Biophys Methods 28:77±82, 1994. 159. S Denery-Papini, JP Brand, L Quillien, Y Popineau, MH van Regenmortel. Immunological differentiation of various gliadins and low Mr subunits of glutenin using anti-peptide antisera. J Cereal Sci 20:1±14, 1994. 160. HJ Ellis, S Rosen-Bronson, N O'Reilly, PJ Ciclitira. Measurement of gluten using a monoclonal antibody to a celiac toxic peptide of a-gliadins. Gut 43:190±195, 1998. 161. Y Nicolas, S Denery-Papini, JP Martinant, Y Popineau. Suitability of a competitive ELISA using anti-peptide antibodies for determination of the gliadin content of wheat ¯our: comparison with biochemical methods. Food Agric Immunol 12:53±65, 2000. 162. AT Lane, JC Huff, WL Weston. Detection of gluten in human sera by an enzyme immunoassay: comparison of dermatitis herpetiformis and celiac disease patients with normal controls. J Invest Dermatol 79:186±189, 1982. 163. J LeszczynÂ ska, J Masl-owska, A Owczarek, U Pytasz. Estimation of gluten content in some of non-gluten products. Chem Anal (Warsaw) 42:715±719, 1997. 164. JH Skerritt, AS Hill. How ``free'' is ``gluten free''? Relationship between Kjeldahl nitrogen values and gluten protein content for wheat starches. Cereal Chem 69:110±112, 1992. Determination of Trace Protein Allergens 339

165. CC Booth, MS Losowksy, JA Walker-Smith, JDW Whitney. Inter-laboratory variation in gluten detection in ELISA kit. Lancet 337:1094, 1991. 166. M Rumbo, FG Chirdo, CA Fossati, MC Anon. Analysis of anti-prolamin monoclonal antibody reactivity using prolamin fractions puri®ed by preparative electrophoresis. Food Agric Immunol 12(1):41±52, 2000. 167. L Sorrell, JA Lopez, I Valdes, P Alfonso, E Camafeita, B Acevedo, F Chirdo, J Gavilondo, E Mendez. An innovative sandwich ELISA system based on an antibody cocktail for gluten analysis. FEBS Lett 439:46±50, 1998. 168. H Wieser, W Seilmeier, H-D Belitz. Quantitative determination of gliadin subgroups from different wheat cultivars. J Cereal Sci 19:149±155, 1994. 169. JM Conroy, A Esen. An enzyme-linked immunosorbent assay for zein and other proteins using unconventional solvents for antigen adsorption. Anal Biochem 137:182±187, 1984. 170. S Denery-Papini, Y Nicolas, Y Popineau. Ef®ciency and limitations of immunochemical assays for the testing of gluten-free foods. J Cereal Sci 30:121±131, 1999.

Biological and Chemical Tests for Protein Nutrient Value

1. INTRODUCTION

The quality of dietary protein should meet the physiological needs of the consumer. Protein with a high nutrient value is digestible and provides a full complement of the essential amino acids. Protein nutrient value (PNV) raises issues beyond the normal terms of reference for chemical analysis. A great deal of progress was made in this ®eld recently and a consensus was reached about procedures for assessing protein quality.This chapter considers PNV and its evaluation. Sec. 1 describes some of the most important indicators of protein quality, factors affecting these indicators, and an outline of the literature. Human bioassays for protein quality are discussed in Sec. 2. Investigations using human subjects are slow and expensive. Alternative tests involve rats, chicks, and other small animals (Sec. 3). A number of chemical tests for protein quality are also available (Secs. 4 and 5) that are fast and inexpensive but have yet to undergo rigorous interlaboratory evaluation. The range of in vitro tests for protein quality includes measurements of reactive lysine, dye binding, and the rate of hydrolysis by proteolytic enzymes.

341 342 Chapter 12

1.1. Indicators for Protein Nutrient Value Indices of protein quality are listed in Table 1. The protein ef®ciency ratio (PER) and the net protein ratio (NPR) are determined noninvasively by measuring the body weight and the amount of protein consumed. Another group of noninvasive indices includes nitrogen balance (B), nitrogen balance index (NBI), and biological value (BV). They are derived from the amount of protein ingested (nitrogen intake, I) and excreted in the feces (fecal nitrogen, F) or urine (urinary nitrogen, U). Finally, the net protein utilization (NPU) and gross protein value (GPV) require the determination of total body or carcass nitrogen. The purpose of dietary protein is to provide amino acids for protein synthesis. This process occurs at a higher rate in young, actively growing, animals. Gender, physiological status, and natural differences between individuals are also important (Table 2). There is some recycling of amino acids as a consequence of the constant breakdown of old protein and its transformation into new protein;indeed, the amount of protein resynthe- sized may exceed the daily intake. Amino acid recycling is not 100%

TABLE 1 Indicators of Protein Nutrient Quality and Their De®nition

Quality index De®nitiona

weight gained Protein ef®ciency ratio (PER) PER protein consumed weight change protein groupnonprotein group Net protein ratio (NPR) NPR protein consumed body N protein groupÀnonprotein group Net protein utilization (NPU) NPU N consumed NPU test protein Gross protein value (GPV) GPV NPU casein

IÀ FÀF0 Total digestibility coef®cient (TDC) TDC I Nitrogen balance (B) B I À (F À U)

Nitrogen balance index (NBI) NBI (B À B0)/I Biological value (BV) BV 100(B À B0)/A response slope fed group Relative nutritive value (RNV) RNV response slope basal group (graphs with common intercept)

response slope fed group Relative protein value (RPV) RNV response slope basal group (graphs with no common intercept) a Letters with subscript 0 refer to parameters for a protein-free diet. For example, F0 fecal nitrogen concentration for rats fed a nonprotein diet, also called metabolic nitrogen. Tests for Protein Nutrient Value 343

TABLE 2 Factors Affecting Protein Quality

Consumer or host related, physiological Species, age, gender, reproductive status (pregnancy and lactation), general health and pathological states (trauma, stress, neoplasia) Sociological Economic, hygiene, and sanitation Protein related Amino acid composition, primary, secondary, tertiary, and quaternary structure, stability, interactions, digestibility Processing history Alkali, chemical, heat, or pressure treatment etc., storage, contamination by bacteria, presence of antinutritional factors Dietary Total protein, total calori®c intake, dietary ®ber, frequency of feeding

ef®cient. Over time there is a gradual net loss of body protein, which must be recovered from the diet. A maintenance diet provides protein at a level that is just suf®cient to avoid a net loss of body mass. Excess dietary amino acids are deaminated and then oxidized to supply energy. Such processes lead to a complicated relationship between protein (nitrogen) intake and utilization. The large number of variables affecting PNV makes it dif®cult to measure accurately.

1.2. Factors Affecting Protein Nutrient Value High-quality protein meets an organism's requirement for essential amino acids. These are not synthesized by the body and must be provided in the diet. The levels of essential amino acids affect how proteins are utilized for growth and maintenance. Protein quality is related to factors intrinsic to the protein, e.g., amino acid composition, protein source, presence of antinutritional factors, processing history, and storage conditions. Extrinsic factors that can affect protein quality include nonprotein dietary constituents, total energy intake, and contamination by bacteria or toxins. Consumer-related variables, socioeconomic factors, and hygiene are also important considerations. Protein quality also varies with the species of consumer (Table 2). 344 Chapter 12

High-quality protein is readily assimilated by the organism, leading to large differences between the nitrogen intake and output. Protein digestion leads to amino acids and small (3±10 residues) peptides. These are absorbed from the gut across the intestinal lining into the blood stream. From here they pass across cell membranes and into the cytoplasm. Unabsorbed protein and other products of hydrolysis are lost as fecal nitrogen. Absorbed but nonmetabolized peptides and amino acids are excreted via the urine. Some dietary protein is degraded by intestinal micro¯ora with no bene®t to the consumer. This is especially true for ruminants, although such losses also occur in the large intestines of monogastric animals. PNV is especially signi®cant for certain sections of the population. These include pregnant women, lactating women, infants and preschool children, and persons with dietary restrictions (vegetarians, patients on liquid feeds, weight watchers, and others on low-calorie diets). People may be on special diets by choice or as a part of medical treatment. Of special concern are populations with diets comprising mainly plant proteins or where the average total calori®c intake is low. In the United States and Canada, food manufacturers must provide food labeling information concerning PNV. The evaluation of PNV is important for monitoring food product formulation, new sources of protein, and for quality control. The animal feeds industry is also interested in PNV. A high protein content (%N 6 6.25) is not a reliable indicator of a food's ability to support growth. The quality of dried milk decreases during storage under high-humidity conditions at room temperature. The quality loss is due to reactions between casein and lactose. In model systems comprising casein and glucose, losses of amino acid side chains occurred after 5±30 days at 378C and 70% relative humidity. Kjeldahl protein levels remained unchanged. Fig. 1 shows sample results for casein or egg albumin stored with D-glucose. This diagram was drawn using data from Lea and Hannan (1) and Tanaka et al. (2). The patterns of amino acid losses are similar for both casein and egg albumin (mainly ovalbumin). Decreases in the level of lysine and arginine are most notable. Baldwin et al. (3) obtained comparable results by heating a mixture of casein, dextrose, and water (1:1:0.5) at 1218C (15 psi) for up to 260 minutes. Protein quality was assessed using the rat feeding trial, by dye binding, and by monitoring the concentration of essential amino acids. Heat treatment caused losses of essential amino acids within the ®rst 15±80 minutes. The pattern of amino acid destruction (arginine 94%, lysine 80%, tryptophan 42%, methionine 41%, histidine 32%) was not unlike results in Fig. 1. The declines in arginine and lysine were closely shadowed by losses in protein quality measured by bioassay. In conclusion, processing and storage can cause changes in protein quality, whereas the quantity of protein need not be affected. Tests for Protein Nutrient Value 345

FIGURE 1 The effect of 5 days of storage on residual amino acid levels for a casein- glucose mixture or an egg albumin±glucose mixture. Samples were stored dry at 378C, pH 6.3 at an RH of 68±70%. (Drawn from data from Refs. 1 and 2.)

The literature covering protein quality is extensive. Since 1973, procedures for the assessing PNV have been reviewed by Rao and Shurpalekar (4), Rosen®eld (5), Friedman (6), Pellet (7), Hopkins (8), Anderson (9), Friedman (10), Bender (11), Satterlee and Chang (12), Walker (13), Sarwar (14), and Sarwar and McDonough (15). Reviews have discussed proteins from various sources, including leaf (16), wheat (17), potato (18,19), sprouted cereal (20), dried beans (21,22), ®sh (23), various plant proteins (24), and single-cell protein (25,26). Friedman (27) surveyed the PNV of most of the major food protein groups. Conference proceedings on protein quality were edited by Claassens and Potgieter (28), Porter and Rolls (29), Friedman (30±32), Bodwell et al. (33), and also Phillips and Finley (34). The following books devote one or more chapters to the subject of protein quality: Milner (35), Bender et al. (36), White and Fletcher (37), Olson (38), Whitaker and Tannenbaum (39), Milner et al. (40), Rakosky (41), Liepa (42), and Damodaran and Paraf (43). The book edited by Finely and Hopkins (44) is another authoritative source of information on protein nutrient value. Speci®c chapters are cited from these sources where relevant. 346 Chapter 12

2. HUMAN AND OTHER IN VIVO ASSAYS FOR PROTEIN NUTRIENT VALUE

Protein quality may be measured with human subjects randomly selected from the population of demographic, ethnic, or geographic interest. To determine PNV for feedstuffs, farming livestock may be used (45). Precise and accurate measurements of PNV require that many of the variables listed in Table 2 are controlled. Careful experimental design and due regard to statistical methods (sampling, appropriate numbers of repetitions, proper control experiments) are necessary. The evaluation of PNV using humans is described by Scrimshaw et al. (46±48). The following account is based on this work. High protein quality is the same as the ability to support growth in humans. Indices of human growth such as body weight or height could be used to monitor PNV. In practice, inducing poor growth in humans is considered unethical. PNV is usually assessed by the nitrogen balance method (Table 1). Subjects are fed a diet containing a submaintenance level of protein [0.3 g kgÀ1 (body weight) dayÀ1]. The amount of nitrogen excreted is carefully monitored over the trial period. A nitrogen balance curve is constructed by plotting nitrogen balance versus the N intake. From such a graph, several indices of quality are determined. Fig. 2 shows two N balance curves produced by feeding diets containing varying amounts of egg or whole wheat protein to human subjects. The equation of the line relating N balance and N intake is

B 0:5 IÀ44 1 where B N balance and I N intake. The (dimensionless) gradient (0.5) is a straightforward measure of the ef®ciency of nitrogen utilization. The ``B 0'' intercept [88 mg N kgÀ1 (body weight) dayÀ1] is the maintenance protein-nitrogen requirement. This is the nitrogen intake necessary to maintain the body weight with no net gain or loss. The `Ì 0'' intercept shows the N balance for subjects subsisting on a protein-free diet. A negative intercept implies that subjects lost 44 mg N kgÀ1 (body weight) dayÀ1 on a protein-free diet. This value should be independent of the identity of the sample protein. In practice, different `Ì 0'' intercepts may occur when high N intake values are extrapolated to zero N intake. In Table 1, some PNV indicators (e.g., the relative nutritive value, RNV) are calculated assuming that all test proteins have the same `Ì 0'' intercept value;i.e., the organism exhibits the same maintenance nitrogen requirement (for different test samples). Actually, the `Ì 0'' intercept from Fig. 2 is simply a mathematical value. The real issue is Tests for Protein Nutrient Value 347

FIGURE 2 An N balance curve for egg or whole wheat protein. Young men were ®rst fed a protein-free diet (I 0) followed by a single diet (e.g., with 40 mg N/kg/day) or at multiple levels of nitrogen. The equation of the straight line is B 0.5(I) À 44 for egg protein and B 0.27(I) À 35 for whole wheat protein. Each data point is averaged from seven test subjects.*

whether the nitrogen balance curves are linear or not. For completely linear responses the ``I 0'' intercept is more likely to be independent of the type of protein sample. Protein quality can also be determined at a single level of N intake. PNV is then calculated using two experimental points [e.g., I 0 and I 40 g kgÀ1 (body weight) dayÀ1]. This approach is prone to error if the N balance plot is not a straight line. It is preferable to determine PNV by feeding several diets with different levels of N intake. According to the slope method, Eq. (1) parameters and associated PNV indices are determined from several experimental points. From Table 3, the slope method is more sensitive and shows clearly that the order of increasing quality is egg protein > soy isolate > whole wheat protein. The level of protein required for maintenance is also a sensitive index of PNV. A feeding trial using one level of nitrogen failed to identify egg protein as possessing superior quality.

* Where B Nitrogen balance and I Nitrogen intake as protein (see Table 1). 348 Chapter 12

TABLE 3 Estimation of Protein Nutrient Value from N Balance Studies with Young Adult Mena

Protein source, N utilization ef®ciency N required for levels of N (slope)* maintenance (intercept)**

Single test Egg 0.63 81 Soy isolate 0.54 94 Whole wheat 0.65 78 Slope method Egg 0.50 88 Soy isolate 0.43 107 Whole wheat 0.27 131 aProtein quality was determined as the (*) the dimensionless slope or (**) intercept [mg N kgÀ1 (body weight) dayÀ1] of the N balance curve.

3. SMALL ANIMAL BIOASSAYS FOR PROTEIN NUTRIENT VALUE

Bioassays using humans or large mammals are time consuming and expensive for three reasons: (a) larger animals grow slowly, (b) there is a requirement for special and more exacting accommodation, and (c) people (even students) may require remuneration. In contrast, small animals such as rats exhibit higher growth rates than adult humans and are cheaper to house. The use of model animals cuts cost and speeds up analysis. Protein bioassays using weanling rats have received of®cial approval in Canada and the United States. Many proposed chemical methods for assessing PNV are routinely calibrated using the rat bioassay. Nevertheless, the use of small animal models (as opposed to humans, livestock, etc.) represents a compromise. Rats and chicks have different amino acid requirements than people and farming livestock.

3.1. The Rat Bioassay Before 1994 there was a statutory requirement in the United States and Canada for food labeling information based on the PER. Protein quality is now measured by a new technique described in Chapter 14. The classic PER assay remains important. This is usually determined using 4-week-old rats. The PER technique was thoroughly discussed in the proceedings from the 38th Annual Meeting of the Institute of Food Technologist Dallas, Texas Tests for Protein Nutrient Value 349

(1978) titled Protein Quality Testing: Industry Problems, Needs, Approaches. Articles by the following authors provide a comprehensive overview of the topic: Rosen®eld (49), Staub (50), Hackler (51), Anderson (52), Burnette and Russoff (53), and Hsu et al. (54). The proceedings from the symposium Measuring Protein Quality for Human Nutrition (American Association of Cereal Chemist, New Orleans, 1976) likewise provides informative summaries by Steinke (55), Bodwell (56), and also Hackler (57). The following discussion is based on materials from these sources and a number of other general reviews (4±16). Osborne et al. (58) introduced PER as an index for measuring the growth-promoting effects of proteins. To determine the PER, cohorts of 10 rats were fed a diet containing 10% protein for 28 days. Their weights were then measured. PER is the ratio of weight gained (or lost) to the weight of protein consumed. The procedure was later standardized with respect to test animals (strain, gender, age) and nonprotein dietary factors by Chapman et al. (59). They adopted casein as a standard protein. A casein standard diet is assessed along with the test protein. Results are then adjusted such that the PER for casein is 2.5. The casein-corrected PER value (cPER) is calculated from the relation

2:5PER cPER SAMPLE 2 PERCASEIN

Values for cPER allow more reliable comparisons of assay results from different laboratories. Bender and Doell (60) incorporated weight loss for rats fed on a protein-free diet into the PER equation. The resulting index, called the net protein ratio (NPR), is a measure of the protein requirement for growth and maintenance. In contrast, PER measures only the protein requirement for growth. Dividing the NPR by the corresponding value for a standard protein (casein or lactalbumin) yields the relative NPR value (RNPR). PER, cPER, NPR, and RNPR represent increasingly robust indices for protein quality determined using the rat bioassay. In Europe, the preferred rat bioassay involves the measurement of carcass nitrogen. To determine net protein utilization (NPU), rats are fed for 10 days on a protein-free diet. Another group of rats is fed the same basal diet supplemented with 10% protein. Both groups of rats are sacri®ced and their carcasses are dried at 1058C or dissolved in concentrated sulfuric acid. Whole-body nitrogen is determined by Kjeldahl analysis or else estimated from the known body moisture content. Rats have a ®xed water/nitrogen ratio (61). The PER, NPR, and NPU assays are single-point assays. These tests involve a single level of N intake. A slope method or multiple-level tests 350 Chapter 12 involving diets with different levels of protein yield more reliable estimates of quality. PER measurements based on multiple levels of N intake were introduced by Hegsted and Chang (62) and Hegsted et al. (63). Feeding multiple levels of nitrogen led to two slope measurements for protein quality called the relative nutritive value (RNV) and the relative protein value (RPV).

3.2. Sources of Error During PER determination, the test sample should contribute a dietary protein level of about 10%. Provided that the sample and casein standard diets have similar proximal composition (moisture, protein, fat, crude ®ber, minerals), some leeway is permitted in the amounts of nonprotein dietary constituents. Values for PER change with the amount of dietary water, ®ber, and mineral mix as well as net food intake. The PER test is also more suited for simple foods or protein concentrates (milk powder, infant formula, wheat ¯our, soy ¯our). With complex foods (hamburgers, processed meat, sausages, bread, rice, and ready-to-eat meals), measuring PER requires a great deal of attention to detail. Samples having one or more of the following characteristics may pose dif®culty: (a) high moisture content, (b) high fat (greater fat content than protein content), (c) high moisture and fat content, and (d) low (<11%) or very high protein content. To obtain accurate values for cPER, the protein and casein standard diets should have a similar proximal composition. PER increases with moisture levels over the range 0±30%. Therefore, food samples should be lyophilized and controlled amounts of moisture reintroduced as necessary. Food components should also be ®nely ground to avoid dietary separation by rats, which can reject certain foods while favoring others. Rats have a preference for sweetened foods and may consume greater quantities compared with the unsweetened casein standard. Residual feed should be routinely tested to con®rm that unconsumed food is representative of the general diet. Other rat-related factors (sex, age, strain) should obviously be standardized. To ensure greater precision, only healthy weanling rats should be used. The transfer of rats to the test laboratory should be done carefully to avoid overly stressful conditions. A 2- to 7-day acclimation period in the new laboratory environment is recommended with longer acclimation periods resulting in higher PNV.

3.3. A Critique of Animal Testing According to Bodwell (56), the chief merits of animal testing are their convenience and low cost. According to this investigator, Tests for Protein Nutrient Value 351

1. Animal tests have few other advantages over human tests. 2. Rat bioassays have not been satisfactorily calibrated against the human tests. An equal number of studies show agreement or disagreement between human- and rat-based results. The number of studies comparing PNV results in rats and humans is inadequate. 3. Protein requirements for growth versus maintenance differ for rats and humans. 4. The PER assay does not measure maintenance requirements. 5. The PER scale is nonlinear and values are not proportional to quality. For example, a PER change from 1 to 2 does not mean a 100% improvement in quality. 6. Speci®c amino acid de®ciencies lead to different effects in rats and people. Finally, Bodwell concludes that the performances of various rat bioassays are no better than chemical methods. A number of methodological differences between human and small animal bioassays were also highlighted by Kies and co-workers (64);they routinely chose mice (rather than rat) for feeding trials owing to the smaller size of feeding rations required. Whereas growth measurements are employed for animal assays, nitrogen balance measurements are used for human assays. Small animal bioassays usually involve groups of genetically pure strains of mice. With human subjects, such homogeneity is not possible or desirable. Therefore, all experimental variables including control diets are usually applied to all human subjects during successive periods within the feeding trial.

3.4. AOAC Guidelines for the Rat Bioassay A collaborative study involving eight laboratories was conducted in order to establish the best practice for rat bioassays. Standard techniques for assessing true protein digestibility (TPD) and associated statistical procedures (65) were developed.* The principles apply to other PNV

* Participants McDonough, Steinke, Sarwar, Eggum, Bressani, Huth, Barbeau, Geraldine Mitchell, and Philips were af®liated with USDA-Beltsville, USDA-Washington, USDA- Philadelphia, Protein Technologies International (St Louis), Health and Welfare Canada (Ontario), National Institute of Animal Science (Tjele, Denmark), Institute of Nutrition of Central America and Panama (Guatemala City, Guatemala), Kraft General Foods Inc. (Glenview, IL), Virginia Polytechnic Institute and State University (Blacksburg, VA). 352 Chapter 12 indicators (Table 1). In vivo digestibility of the major food protein groups has been reviewed by Hopkins (66) and Sarwar (67). Method 1 An outline of AOAC guidelines for conducting a rat bioassay. True protein digestibility was determined for six samples (casein, canned macaroni and cheese, tuna ®sh in water, rolled oats, pinto beans, and heat-treated milk powder). Sample Pretreatment 1. Freeze dry food samples and (where necessary) grind to pass through a 20-mesh screen. 2. Determine the proximate composition (e.g., % nitrogen, fat, water, ®ber) for each sample. 3. Assess the nitrogen content for each sample by Kjeldhal analysis. 4. Formulate standard rat diets to provide equal levels of protein-N (*1.6% or 10% true protein). 5. From the known proximate composition, normalize the level of fat and ®ber in each diet. Table 4 shows the composition of some of the diets used in AOAC study (65). Notice that diet 1 is a nonprotein diet. Butylated hydroxyanisole (BHA) was added as antioxidant. Dietary constituents were mixed thoroughly using a mechanical mixer to avoid selection by rats. Male Sprague-Dawley rats (50±70 g) were acclimated in wire net cages (at 18± 268C and 40±70% relative humidity) for 2 days on normal rat food. Each test diet was fed (15 g per day, water available ad libitum) to two groups of four rats (i.e., eight rats per diet) for a total of 9 days. The ®rst 4 days was an adaptation period. The next 5 days constituted the balance period, during which the weight of food eaten was recorded. Rat fecal output was also collected, dried, and stored for nitrogen determination. Nitrogen intake was determined from the weight of feed consumed and the percent nitrogen content of each diet. TPD was calculated from Eq. (3).

I À F À F TPD P P O;P 3 IP where IP, FP,andFO,P are protein intake, fecal protein, and metabolic protein. Eq. (3) is a speci®c form of the digestibility equation (see Table 1) applied to dietary protein. Digestibility can also be de®ned in relation to amino acids (Chapter 14). Tests for Protein Nutrient Value 353

TABLE 4 Example Diets Used in a Collaborative Rat Assay for Protein Digestibilitya

Diet Component (g/100 g) 1 2345

Casein Ð 10.8 Ð Ð Ð Macaroni and Ð Ð 54.8 Ð Ð cheese Tuna ®sh Ð Ð Ð 11.0 Ð Rolled oats ÐÐÐÐ57.8 Vitamins 2.0 2.0 2.0 2.0 2.0 Minerals 3.5 3.5 3.5 3.5 3.5 Corn oil 10.0 10.0 1.5 9.7 6.6 Cornstarch 79.5 68.7 34.5 68.8 29.4 Cellulose/®ber 5.0 5.0 3.7 5.0 0.7 BHA 0.005 0.005 0.005 0.005 0.005 aThe plan conforms to the AOAC test diets for rat bioassays. Source: Adapted from Refs. 65 and 121.

3.5. The Chick Bioassay Choppe and Kratzer (68) used White Plymouth Rock chicks to evaluate PNV for meat and bone meal. Chicks were fed with normal starter mash rations until 5 days old. They were then divided into groups of 10 and fed a basal diet (Table 5) with or without a supplement of test sample. Suf®cient meat and bone meal were added to supply 24% of protein (Kjeldahl, %N 6 6.25). The chicks were weighed at 5, 10, and 15 days old. Moran et al. (69) used New Hampshire 6 Delaware chicks to evaluate the effect of

TABLE 5 An Example Basal Diet for Chick Feeding Trials

Basal diet Composition (%)

Cornstarch *75 Mineral mixture 2 Soybean oil (crude, 5 degummed) Cellulose 4 Choline chloride 12.5 Vitamin supplements 2.16

Source: Adapted from Ref. 68. 354 Chapter 12 heating on the quality of soybean and ®shmeal. Groups comprising ®ve male and female chicks are randomly selected and fed a basal diet with and without added test sample as a protein supplement. During feeding trials chicks should be kept under constant temperature and controlled lighting (usually 15 hours lighting per day). They are also fed on demand. Weight changes are recorded at weekly intervals for up to 4 weeks. Each measurement is taken in triplicate or quadruplicate and averaged over 30± 40 chicks. Various protein quality indicators (Table 1) are calculated from the weight changes.

4. IN VITRO METHODS FOR ASSESSING PROTEIN NUTRIENT VALUE 4.1. Amino Acid Scores The use of amino acid data to determine PNV was suggested in 1946 by Mitchell and Block (70). They are also credited with introducing the concept of protein chemical score. To determine the chemical score, divide the quantity of each amino acid from the test sample by the amount found in egg protein or another reference protein [Eq. (4)]. Repeat the calculation for each of the 9±10 essential amino acids (EAAs). amount of EAA sample Chemical score 4 amount of EAA egg

The chemical score is the smallest ratio found;this also identi®es the limiting amino acid. Oser (71) introduced the integrated chemical score, which is the geometric mean of the ``egg ratio'' for all essential amino acids. By the 1970s the egg protein standard had fallen into disuse. Protein quality is now determined in relation to the essential amino acid requirements for humans (Table 6). Amino acid scores (AASs) are estimated using the FAO/WHO/UNU (1985) amino acid requirement pattern for 2- to 5-year-old preschool children (Table 6). The calculation of AAS follows the same principles as shown in Eq. (4). First, determine the essential amino acid pro®le for the test sample. Next, divide the amount of each essential amino acid by the corresponding value from the FAO/WHO/UNU (1985) list. The smallest ratio (a) identi®es the limiting amino acid and (b) yields a numerical value for the AAS. It is a relatively simple matter to calculate the AAS value for a hypothetical mixture of foods. Some investigators have taken the sum of the chemical scores for the nine essential amino acids (EAA9) as the index for quality. Tests for Protein Nutrient Value 355

TABLE 6 Suggested Pattern for Essential Amino Acid Requirements

Suggested pattern of requirement (mg/g protein)

Infanta Infanta 2±5 10±12 Amino acids mean range years years Adult

Histidine 26 18±36 19 19 16 Isoleucine 46 47±53 28 28 13 Leucine 93 83±107 66 44 19 Lysine 66 53±76 58 44 16 Methionine cystine 42 29±60 25 22 17 Phenylalanine 72 68±118 63 22 19 tyrosine Threonine 43 40±45 34 28 9 Tryptophan 17 16±17 11 9 5 Valine 55 44±77 35 25 13 Total EAA 460 398±589 339 241 127 Total nonessential 540 411±602 661 759 873 AA Total 1000 1000 1000 1000 1000 aMean and range of values found in human milk. Source: From FAO/WHO/UNU (1985) [see Refs. 75, 76 of Chapter 14].

4.2. Sources of Error Errors in amino acid analysis (Chapter 1) will affect values of the AAS. Ideally, the hydrolysis of proteins using 6 M HCl should produce ef®cient release of all amino acids followed by accurate analysis by column chromatography. In practice, the sulfur amino acids and tryptophan are partially destroyed during acid hydrolysis. The release of valine and isoleucine is dependent on the hydrolysis time. In order to obtain data for all amino acids, multiple protein hydrolysis is necessary. The majority of amino acids are analyzed after protein hydrolysis in 6 N HCl at 1108C for 24 hours. The sulfur amino acids are ®rst oxidized by performic acid treatment to cysteic acid or methionine sulfone. Tryptophan determination is performed after alkaline hydrolysis. Many criticisms of amino acid analysis appear to have been addressed with the advent of more sophisticated equipment. The precision of amino acid analysis was thought to be questionable in the past. However, recent collaborative studies have shown a within-laboratory precision of about 3%. The interlaboratory precision for assaying labile amino acids was 16±17%. 356 Chapter 12

Another disadvantage of AAS is that it takes no account of protein digestibility. This suggests (incorrectly) that (a) all proteins are optimally digestible and (b) all amino acid residues are absorbed and are bioavialable to the same degree. Following recent improvements in amino acid analysis, protein digestibility corrected amino acid scores (PDCAASs) have become widely adopted as indicators for protein quality (Chapter 14).

4.3. Available or Reactive Lysine Lysine is the ®rst limiting essential amino acid in cereals. The most reactive essential amino acid also appears to be lysine (Fig. 1). Protein-bound lysine residues with low chemical reactivity tend to be nutritionally unavailable. Therefore available lysine is widely monitored as an index of protein quality (Table 7).* To measure lysine reactivity, the food sample is exposed to FDNB, TNBS, MIU, or EVS at high pH for periods ranging from 2 hours (FDNB) to 4 days (MIU). The chemically modi®ed protein is hydrolyzed by re¯uxing with 6 M HCl. The hydrolysate is analyzed by colorimetry to determine the concentration of reactive lysine. The FDNB (difference) or EVS (difference) methods involve conventional amino acid analysis. Chemically modi®ed

TABLE 7 Reagents for Measuring Protein-Bound Reactive Lysine

Reagent References

1-Fluoro-2,4-dinitrobenzene (FDNB) Sanger (72), Carpenter (73), Hurrell and Carpenter (74), Booth (75) 1-Fluoro-2,4-dinitrobenzene (FDNB) Booth (75), Roach et al. (76), Ostrowski (difference method) et al. (77), Couch (78) Trinitrobenzenesulfonic acid (TNBS) Kakade and Liener (79) o-Methylisourea (MIU) Hurrell and Carpenter (74) o-Phthalaldehyde (OPA) Gondo et al. (80) Ethyl vinyl sulfone (EVS) (difference Friedman and Finley (81) method) Differential dye-binding capacity Sandler and Warren (82)

* Carpenter suggests that ``reactive lysine'' is a more accurate description. Lysine chemical reactivity is not synonymous with nutritional availabilityÐsee Section 5.1. However, reactive lysine is not an effective search keyword in the food science literature. We use ``available lysine'' to imply chemically reactive protein lysine groups. Tests for Protein Nutrient Value 357 lysine is eluted with a different retention time, allowing unmodi®ed (so- called bound) lysine to be found.

4.4. The o-Phthalaldehyde Method The o-phthalaldehyde (OPA) method does not involve protein hydrolysis (80). The sample (* 25 mg in 0.1 M sodium phosphate buffer, pH 7) is treated with OPA reagent. Fluorescence intensity is measured after 2 minutes using excitation and emission wavelengths of 340 and 455 nm. The ¯uorescence emission (calibrated with quinine sulfate) is directly proportional to the concentration of reactive lysine. Corrections are needed for small peptides because OPA also reacts with the terminal amino groups. There is usually no interference from common reagents such as urea, guanidine chloride, imidazole, ethanol, tricine, sodium acetate, or 2- mecaptoethanol at concentrations of 100 mM. So far, the OPA method has been applied only to relatively pure protein samples.

4.5. Fluorodinitrobenzene Acid Reactive Lysine by Spectrophotometry FDNB introduced by Sanger (72) is the most widely used reagent for assaying available lysine. Lea and Hannan (1,83) used FDNB to monitor changes in reactive lysine concentration for a casein-glucose mixture stored in the dry state. They also followed changes in the reactive lysine content of dry milk powder. Later, Carpenter and co-workers (73,74) from the School of Agriculture, University of Cambridge (UK) extended the FDNB procedure for protein-bound lysine in foods. In Method 2, the sample is ®rst modi®ed with FDNB and then hydrolyzed by heating with 6 M HCl. The hydrolysate is extracted with diethyl ether and the absorbance reading is recorded for the aqueous phase (Tube A). A regent blank (Tube B) is prepared by neutralizing the hydrolysate, treating with an acylating agent, and then extracting with diethyl ether. The reactive lysine concentration is proportional to the absorbance of Tube A minus the absorbance for Tube B. Method 2 Determination of FDNB-reactive lysine (73,75). Reagent 1. 1-Fluoro-2,4-dinitrobenzene (2.5% w/w in ethanol). Melt solid FDNB by warming to about 408C and dispense using an automatic pipette. About 12 mL of FDNB solution is required per sample. 358 Chapter 12

2. Methoxycarbonyl chloride (MCC;also called methyl chloroformate). 3. Diethyl ether.

4. Sodium carbonate buffer (8% w/v;pH 8.5). Add NaHCO 3 (8% w/v) to a solution of 8% (w/w) Na2CO3 until this reaches pH 8.5. 5. Concentrated hydrochloric acid (8 M). Procedure STAGE 1. Production of FDNB-reactive lysine and acid hydrolyses of the modi®ed protein. Place 1.5 g of ground (50 mesh) sample in a ¯at-bottomed ¯ask and wet using 8 mL of sodium carbonate buffer. Add 12 mL of FDNB solution and shake at room temperature for 2 hours. Warm the mixture over a boiling-water bath until it no longer effervesces and all ethanol has been evaporated. Add 24 mL of 8 M HCl and re¯ux at near 1008C for 16 hours. Allow the sample to cool slightly and ®lter through Whatman No. 541 ®lter paper. Wash the ®ltrate thoroughly with water, collect the washings, and make up to ®nal volume of 200 mL. Remove 2 mL (62) samples and label as Tube A and Tube B. STAGE 2. Prepare sample (Tube A) for absorbency measurements. Extract the contents of Tube A with 5 mL (64) diethyl ether. Discard the ether phase and warm the aqueous sample over a boiling-water bath to remove trace amounts of ether. Make up the contents of Tube A to 10 mL using 1 M HCl and retain this sample (Tube A-aq) for absorbency readings. STAGE 3. Acylate the regent blank (Tube B) for absorbency measurements. Extract Tube B with 5 mL of ether once. Discard the ether phase and warm the aqueous phase to remove trace ether as before. Add one drop of phenolphthalein indicator and neutralize the sample using 0.1 N alkali until pink. Adjust the sample to pH 8.5 by adding 2 mL of sodium carbonate buffer. Acylate the sample with 45±50 mL of methoxycarbonyl chloride and shake vigorously. Using Tube B, follow guidelines from Stage 2. STAGE 4. Record absorbance measurements at 435 nm.

Record A435 readings for Tube A (aq) and Tube B (aq) using a 1-cm path length disposable cuvette. The absorbency difference for Tubes A and B is proportional to the reactive lysine concentration. Tests for Protein Nutrient Value 359

Food samples usually contain protein-bound lysine and free amino acids (lysine, histidine, arginine). One mole of FDNB reacts with protein- bound lysine via the epsilon amino group (e-NH2), forming e-DNP-Lys. Free lysine or the N-terminal lysine residue reacts with 2 moles of FDNP to form Lys(DNP)2. This species (and His-DNP, Arg-DNP, AA-DNP) lacks a formal charge and therefore partitions into the ether phase (see Stage 2).

The a-NH2 group e-DNP-Lys remains free and ionized (when released by acid hydrolysis of the DNP-modi®ed protein) and therefore extracts into the aqueous phase of Tube A. Finally, acylation and extraction by ether remove all DNP-AA derivatives including e-DNP-Lys (in an acylated form) from the water phase. Table 8 shows the distribution of different DNP±amino acid derivatives between the aqueous and ethereal fractions during reactive lysine analysis. DNP±amino acid derivatives (30 mM) were added at stage 1 of the FNDB assay. DNP-Arg remained in the aqueous phases in both Tubes A and B. The ®nal absorbency difference between Tube A (aq) and Tube B (aq) provides an excellent correction for the presence for free histidine which reacts with FDNP. The recovery of e-DNP-Lys from animal proteins was 90% during the measurement of reactive lysine. Therefore, results are normally multiplied by 1.09 as a correction factor for physical losses (73,74). With plant protein the recovery of e-DNP-Lys was 60±85% owing to one or more of the following factors: (a) plant proteins have a higher resistance to acid hydrolysis, (b) the degree of hydrolysis is further reduced by the low solubility of DNP-protein adducts compared with unmodi®ed proteins, (c) DNP-Lys adsorbs to plant residues not hydrolyzed by concentrated HCl, and (d) the NO2 group DNP-Lys may be reduced to 22NH2, leading to a reduced color yield. The correction factor of 1.09 was inapplicable for plant

TABLE 8 Recovery of DNP Derivatives of Three Basic Amino Acids

Final A435 for tube

Amino acid Tube A (aq)a Tube B (aq) Tube B (eth) A À B (aq) e-DNP-Lys 0.203 0.002 0.201 0.201 DNP-Arg 0.090 0.089 0.008 0.001 a-benzoyl His 0.002 0.007 0.138 À0.005 Me ester aTube A (ether phase) discarded. Source: Based on results from Ref. 73. 360 Chapter 12 proteins. The correction factor should be carefully determined during plant protein analysis (75).

4.6. Fluorodinitrobenzene Reactive Lysine by Chromatography To determine chemically available lysine by chromatography, hydrolyze FDNB-modi®ed protein with 6 M HCl (see Method 1). Then subject the hydrolysate to chromatographic analysis. The majority of reports involve low-pressure chromatography (Table 9). However, these are probably only of historic interest. They have now been superseded by reverse-phase high- pressure chromatography (RP-HPLC). Peterson and Warthesen (92) were ®rst to describe chromatographic analysis of e-DNP-Lys using RP-HPLC. Their system comprised a m-Bondapak C18 column coupled to a Waters model 600A pump, UK6 injector, and a 440 absorbance detector. Samples were separated by isocratic elution with 20:80 (v/v) acetonitrile±0.01 M acetate buffer (pH 4.0) and

TABLE 9 Chromatographic Analysis of Available Lysine

Protein source Column, comments References

Various including oilseeds Amberlite IR-120 Rao et al. (84), Frampton (cottonseed, sesame, (*1 6 6 cm) wash and Kuck (85), Kuck peanut, soybean), ®sh 3 N HCl, eluent 3N and Frampton (86), meal, meat meal, HCl Me-Et-ketone Hussein (87) feather Various including maize, Nylon powder, Amberlite Blom et al. (88) gluten, milk replacer, C.G. 12) polarograp- soybean hic detection Groundnut meal, arachin, Liquid-liquid chromato- Matheson (89,90) conarachin graphy with kieselguhr support Various including ®sh Amino acid analyzer, Roach et al. (76), Booth meal, meat meal, milk subtractive amino acid (75), Ostrowski et al. powder, wheat, maize, analysisÐSilcock met- (77), Bailey (91) soybean, sun¯ower hod meals

Casein, soybean, gluten, C18 HPLC reverse phase, Peterson and Warthesen zein, ®sh meal, e.g., m-Bondapak, Nov- (92), Rabasseda et al. albumin, lysozyme a-Pak, or Spherisorb (93) Castillo et al. (94) ODS-2 Hernandez et al. (95) Tests for Protein Nutrient Value 361 products were detected at 436 nm. Two peaks for 2,4-DNP and e-DNP-Lys were well separated with retention times of 10 and 12 minutes. Comparing Carpenter's technique (Method 1) with the HPLC method showed that the former gave signi®cantly higher lysine readings for high-carbohydrate samples. Rabasseda and co-workers (93) attempted to optimize the RP- HPLC method using product detection at 274 nm. The LLD was 4.5 6 10À9 gofe- DNP-Lys or 2 6 10À9 g for unmodi®ed lysine. The precision of analysis was 0.95% for ®sh meal and 1.66% for soybean meal. Castillo and co-workers (94) criticized the previous HPLC analysis for the arbitrary choice of assay conditions (e.g., conditions for sample hydrolysis, choice of mobile phase pH, choice of detector wavelength). The optimal RP-HPLC assay involved a

C18 column operated at 408C to avoid precipitation of DNP during analysis. Peak elution was with 22:77 (v/v) acetonitrile±0.01 M acetic acid (pH 5) solvent at a ¯ow rate of 2 mL minÀ1. DNP-protein was more ef®ciently hydrolyzed by treating solid samples with an excess of relatively dilute HCl acid (1 mL 6 M HCl per 1 mg sample) as compared with highly concentrated acid (1 mL of 10 M HCl per 1 mg sample). Apparently, e-DNP-Lys was degraded by 10 M HCl. Analytical precision varied with sample type;the CV ranged from 1.41 to 3.19% for enteral formula, milk, and lentil.

4.7. Trinitrobenzene Sulfonic Acid The analysis of free amino acids and peptides using TNBS (2,4,6- trinitrobenzene sulfonic acid) was introduced by Okuyama and co-workers (96,97). TNBS reacts with primary amines in a manner similar to FDNB. Habeeb (98) used TNBS to monitor protein-bound lysine under relatively mild conditions. The BSA reaction with formaldehyde could be followed using TNBS. Binding of SDS to lysine residues interfered with the TNBS reaction. Mokrasch (99) used TNBS for the coestimation of amines, amino acids, and proteins. Kakade and Liener (79) were ®rst to determine available lysine using TNBS. The trinitrophenyl (TNP) protein derivatives can be hydrolyzed by autoclaving with concentrated HCl for 60 minutes as compared with a minimum hydrolysis time of 4 hours for DNP-proteins. Analysis of trinitrophenylated products is by spectrophotometry (see Method 2) or RP-HPLC (100).

Method 3 Determination of reactive lysine using TNBS (79).

Reagents 1. 2,4,6-Trinitrobenzene sulfonic acid (0.1% w/w solution in water). 362 Chapter 12

2. Sodium carbonate buffer (4% w/v;pH 8.5). To a solution of 4 %

(w/w) Na2CO3 add NaHCO3 (4% w/v) until the correct pH is attained. 3. Concentrated HCl. 4. Diethyl ether. Procedure Add 1 mL of TNBS solution to 1 mg of sample (dispersed in 1 mL sodium carbonate buffer) and incubate at 408C for 2 hours. Add 3 mL of concentrated HCl and autoclave at 1208C (15 psi) for 60 minutes. Cool to room temperature and adjust to 10 mL with distilled water. Extract the hydrolysate with 2610 mL of diethyl ether.

Record the A346 reading for the aqueous phase using a 1-cm cuvette. Calculate the concentration of TNP-Lys; c A/e, where e 1.46 6 104 MÀ1 cmÀ1. The TNBS method gave quantitative readings for lysine in a wide selection of proteins including BSA (60 moleÀ1), ovalbumin (20 moleÀ1), b-lactoglobulin dimer (29 moleÀ1), bovine hemoglobin (47 moleÀ1), and insulin (1 moleÀ1). In all cases, the numbers of lysine residues detected equaled literature values. Examples of the application of the TNBS method for food samples include the analysis of cotton seed meal (101) and casein and rapeseed protein (102). High amounts of lactose interfere with the TNBS method. Overnight dialysis followed by freeze-drying is a simple strategy for reducing the error from lactose (103,104). Interferences also occur in the presence of high concentrations of other carbohydrates. Sample treatment with bleach may ameliorate this problem (105).

4.8. Differential Dye-Binding Capacity or Dye-Binding Lysine Sandler and Warren (82) measured DBC before and after acylating protein samples using ethyl chloroformate. The differential dye binding capacity (dDBC) is the change in dye-binding capacity produced when lysine residues are blocked by chemical modi®cation. Hurrell et al. (106) used propionic anhydride as an acylating agent and called their measurement the dye- binding lysine procedure. As described in Chapters 5±7, protein DBC is determined by the net concentration of basic amino acids. Values for the dDBC are directly proportional to lysine levels alone and therefore provide a sensitive indicator of protein quality. The dDBC procedure is summarized next. Tests for Protein Nutrient Value 363

Method 4 Determination of differential dye-binding capacity using propionic anhydride (106,107).

Reagents 1. Propionic anhydride 2. As listed for the Udy method (Method 1, Chapter 5) 3. Sodium acetate (0.2 M)

Procedure Dye-binding capacity. To 20±500 mg of ground sample (*20±90% w/w protein) in a 100-mL ¯ask, add three 6-mm glass beads and 2 mL of sodium acetate (0.2 M). Shake for 10 minutes to wet. Add 40 mL of dye solution and shake for a further 60 minutes. Filter the sample, dilute the supernatant as needed, and read its absorbency. Determine the DBC (mg gÀ1 protein) from a dye calibration graph. DBC for a chemically modi®ed sample. Weigh 20±500 mg of sample and wet with sodium acetate as above. Add 0.4 mL of propionic anhydride, mix, and allow to react for 10±15 minutes. Add 40 mL of dye solution and treat as before. Calculate the dDBC as the difference between DBC values for the modi®ed and unmodi®ed protein. Protein acylation with propionic anhydride is straightforward and quantitative. The procedure for determining dDBC is no more protracted than a conventional dye-binding assay. The lysine content of 24 different varieties of grain legumes was rapidly determined by the dDBC method. Hurrell et al. (106) found that the dDBC method had potential for monitoring plant breeding. The technique was successfully applied for monitoring processing damage. Values for dDBC (mmoles of dye bound per 16 g N) are numerically equal to the concentration of total lysine (mmoles per 16 g N) determined by amino acid analysis or using the FDNP technique (Fig. 3). The dDBC measurements are free from interference by starch, lipid, or free lysine (up to 30% w/v). Sandler and Warren (82) found that carbethoxylation of ®sh meal required 30±60 minutes, with 40 minutes being the least time needed to ensure complete reaction. The rate of dye binding was slow at 208C although equilibrium is reached in about 20 minutes at 408C. There was good agreement between values for DBC and the total basic amino acid concentration (TBAA) and also between dDBC and lysine values measured by amino acid analysis (Table 10). Pearl and co-workers (107) determined the dDBC values for soybean meal using Method 3 with minor modi®cations. The amount of Orange G dye bound to soybean protein 364 Chapter 12

FIGURE 3 Relationship between the total lysine and the differential dye-binding capacity (dDBC) for a range of commodities: bovine serum albumin, skim milk powder, ®sh meal, pea ¯our, soya ¯our, meat and bone meal, barley, peanut ¯our, corn gluten, as well as a variety of legumes (winged bean, pea, green gram, chickpea, lentil, cowpea, dry bean, soybean, runner bean, and pigeon pea). The equation of the line Y mX C, where m 1.0328 (+0.013), C 0.071 (+0.3810). R2 0.9967. For statistical testing, F 6287.38 with 21 degrees of freedom.

TABLE 10 Dye-Binding Capacity and Differential Dye-Binding Capacity in Protein Mealsa

TBAA DBC Lys* dDBC Protein source (mmoles) (mmoles) (mmoles) (mmoles)

Fish-a 79 (+8.6) 87 43 (+4.7) 44 (+4.8) Fish-b 74 (+8.1) 73.6 40 (+4.4) 38 (+3.6) Soya 57.7 59.4 24.1 36.4 Cashew 47.0 52.1 13.7 17.6 aTBAA, total basic amino acids, i.e., Lys His Arg concentration (mmole/120 mg of sample); DBC, dye-binding capacity (mmole dye bound/120 mg sample);Lys, results from amino acid analysis (mmole/120 mg sample);dDBC, differential dye-binding capacity. Samples were ®sh meal (®sh-a), defatted ®sh meal (®sh-b), soya meal (soya), or cashew nut meal (cashew). Tests for Protein Nutrient Value 365 was determined before and after acylation with propionic anhydride. Slight changes in protein-bound lysine due to thermal treatment were detected. These results (82,106,107) con®rm that the dDBC method is a highly affordable alternative to amino acid analysis when one wants to measure lysine levels.

4.9. Comparisons of Methods for Available Lysine One advantage of the TNBS method (compared with the FDNB technique) is the shorter time for protein hydrolysis. The solvent extraction steps for both techniques reduce interference. Hurrell and Carpenter (74) compared FDNB, TNBS, and other reagents for reactive lysine determination in model systems comprising ovalbumin, lactalbumin, and glucose (3:2:5 weight ratio).* Samples were stored at 378C for up to 30 days or autoclaved at 1218C for 15 minutes. The reacted protein-sugar mixtures underwent large changes in the ability to sustain growth in chicks or rats. The MIU and TNBS methods underestimate total lysine levels for the unheated (control) sample (Table 11). The ability to detect mild levels of damage (378C, 30 days storage) also varies for different techniques. Amino acid analysis and FDNP and TNBS methods did not accurately measure

TABLE 11 Total Lysine and FDNB-Reactive Lysine in Ovalbumin, Lactalbumin, and Glucose Mixture by Different Methods

Lysine (mg/g protein)

Methoda Control 378C, 30 d 1218C, 15 min

Total Lys 86.3 50.9 29.3

Total Lys (NaBH4) 81.1 12.2 18.7 FDNB (diff) 84.7 45.7 19.5 FDNB (dir) 81.6 19.6 12.2 TNBS 53.6b 35.9 11.2 MIU 71.9b 15.8 6.5 a Total lysine, amino acid analysis. Total Lys (NaBH4), amino acid analysis after sample reduction by sodium borohydride. Other abbreviations are de®ned in the text. bResults are signi®cantly different from the control. Source: Summarized from Ref. 74.

* Lactalbumin is spray dried whey protein. It has low solubility owing to heat damage by spray drying. 366 Chapter 12 mild heat damage. The errors were ascribed to two factors: (a) products of early Maillard reactions are not acid stable and some glycosylated lysine may be degraded to free lysine during protein hydrolysis in 6 M HCl, and (b) early Maillard reaction products may react with TNBS or FDNB. This may be because the nucleophilic character of the lysine derivatives formed during the early Maillard reaction is similar to that for unmodi®ed lysine. Both errors can be ameliorated by the addition of sodium borohydride

(74,101,108). The Schiff base formed between lysine e-NH2 and sugars is rendered acid stable by treatment with sodium borohydride. Without this precaution, available lysine assays involving acid hydrolysis at high temperatures will be subject to error. By comparison with Method 1, the chromatographic detection of e- DNP-Lys leads to some of the following improvements: 1. Reduction of interference from other yellow substances formed during protein chemical modi®cation (e.g., dinitrophenol and o- DNP tyrosine). Plants also contain brown humin and other dyes that can act as interferences. 2. Ability to measure free lysine and N-terminal lysine residues in one experiment. 3. Decreased error in the case of high-carbohydrate samples. The color yield for Method 1 can be impaired when e-DNP-Lys is reduced to the diaminophenol derivative. Alternatively, carbohydrate may react with FDNB to generate water-soluble derivatives that interfere with the analysis of e-DNP-Lys. 4. Absence of sample pretreatment. Protein hydrolysate can be subjected to chromatographic analysis directly without an ether extraction step. 5. High sample throughput.

5. PROTEIN DIGESTIBILITY

Digestion begins in the mouth. Food is macerated by chewing, mixed with saliva, and swallowed. Further particle size reduction arises from muscular contractions of the stomach. Within the gastric environment, protein is broken down by pepsin (or rennin in young animals). The stomach contents then pass to the small intestines and are neutralized by pancreatic juice. This contains the principal enzymes for protein digestionÐtrypsin, chymotrypsin, and elastase as inactive zymogens. Following zymogen activation, protein is degraded to low-molecular-weight peptides and then attacked by carboxypeptidase A and B. The products of protein digestion (amino acids, Tests for Protein Nutrient Value 367 dipeptides, tripeptides) are absorbed in the small intestines. Undigested protein functions as dietary ®ber or else undergoes microbiological degradation in the large intestines. Human nutritionist and veterinary scientists are concerned with protein digestibility in vivo. Guidelines for assessing digestibility in the rat were described earlier (see Method). More arti®cial tests for digestibility are described in this section.

5.1. De®nition of Terms Alternative de®nitions for digestibility are listed in Table 12. Apparent protein digestibility (APD) is the dietary protein intake (I) minus fecal protein (F) expressed as a percentage of the dietary protein intake.

Correcting for the metabolic or fecal nitrogen (F0) yields the true protein digestibility (TPD). The metabolic nitrogen is that which is excreted by a test subject fed on a protein-free diet. It is derived from cells sloughed off the intestinal wall and proteins comprising gastric enzymes and other secretions. Both APD and TPD are sometimes called nitrogen digestibility because of being calculated from intake and fecal nitrogen concentrations.* The formula for TPD is the same as that for (bio)availability (Table 12). Both indices can be further re®ned by taking account of urinary nitrogen (U or

U0). In practice, this correction is not often applied for tests using human

TABLE 12 Some Common Expressions for Digestibility

Expressed in terms of Term Equationa retention

Apparent protein (I À F)/I (R U)/I digestibility (APD)

True protein digestibility (I À (F À F0))/I (R U F0)/I (TPD)

Availability/bioavailability (I À (F À F0))/I or (R F0 U0)/I (I À (F À F0) À (U À U0)/I True adsorption I À (F À F0) Retention (R) I À (F U) Nitrogen balance (B) I À (F À U) aI, intake; F, total fecal excretion; R, retention; U, total urinary excretion. Subscript 0 refers to metabolic values determined with a protein-free diet.

* These relations also apply to dietary nitrogen, which is assumed to be proportional to crude protein (N 6 6.25). See Chapter 1 for a more detailed description of Kjeldahl analysis. 368 Chapter 12 subjects or rats. Dietary protein levels probably do not affect the amount of protein-N secreted via healthy kidneys and hence U À U0 is close to zero. For amino acids and other low-molecular-weight nutrients, ignoring urinary losses may lead to signi®cant error (109). The importance of ``digestibility'' can be seen by comparing relations in Tables 1 and 12;notice that parameters such as NBI and BV (Table 3) also feature digestibility. Digestibility can be measured for speci®c amino acids and is called amino acid availability (AAAv), amino acid digestibility, or simply availability. The three terms are used interchangeably. The AAAv (unlike TPD) allows for different adsorption rates for speci®c amino acids. I À F À F AAAv AA AA O;AA 5 IAA

In Equation (5), the term IAA is the intake concentration of a speci®ed amino acid, FAA is the fecal concentration, and FO,AA is the fecal concentration with a protein-free diet.* Protein digestibility is routinely measured using the rat assay (Method 1). Constance Kies (110) refers to ``protein bioavailability'' owing to the similarity between Eq. (3) and Eq. (5). TPD is determined by measuring nitrogen (crude protein), and AAAv is determined from the corresponding amino acid concentrations (111). Most nutritionists do not distinguish between protein digestibility and protein (bio)availability.

5.2. In Vitro Protein Digestibility and Protein Quality In vitro protein digestibility refers to the rate of proteolysis of different proteins by proteases in vitro. Tests of this type do not take account of the effect of absorption. The link between PNV and in vitro protein digestibility is shown by results from Hankes and co-workers (112). Autoclaving casein for 4 minutes or 20 hours produced no signi®cant changes in the amino acid composition compared with the unheated protein. However, in vitro protein digestibility was impaired during successive treatments with pepsin, pancreatin, and erepsin (a crude source of peptidase). Similar ®ndings were reported by Eldred and Rodney (113). Melnick and Oser (114) found that heating legume or egg proteins improved their biological value

* Measurement of fecal concentrations does not allow for amino acid degradation by colon bacteria. With farm animals, ileal digestibility is routinely determined by ®tting catheters and collecting digestion products (for nitrogen or amino acid determination) before they reach the colon. Tests for Protein Nutrient Value 369 probably through inactivation of protease inhibitors, which improves proteolysis.

5.3. Determination of in Vitro Protein Digestibility The food sample is exposed to one or more proteases and the degree of proteolysis is monitored. Single enzymes or combinations of enzymes can be used including pepsin, chymotrypsin, trypsin, papain, and peptidase. The degree of protein hydrolysis is generally measured as TCA-soluble products at 275±280 nm, by following the fall in system pH, or by monitoring the release of free amino groups using TNBS. Techniques for monitoring protein digestibility were reviewed by Swaisgood and Catignani (115,116) and by Boisen and Eggum (117). The AOAC-approved test for in vitro protein digestibility is based on the three-enzyme method of Hsu et al. (118) or the four-enzyme approach (119,120). A pH meter or pH-Stat instrument is used to monitor these multienzyme reactions (see Method 5).

Method 5 Determination of in vitro protein digestibility using three or four enzymes (118±120).

Reagents 1. Porcine pancreatic trypsin (14,190 units per mg) 2. Bovine pancreatic a-chymotrypsin (60 BAEE units per mg) 3. Porcine intestinal peptidase (40 units per g powder) 4. Pronase 5. Dilute (0.1 M) hydrochloric acid and sodium hydroxide Enzyme solution A. Dissolve trypsin (16 mg), chymotrypsin (31 mg), and peptidase (13 mg) in distilled water at 378C. Adjust to pH 8 using dilute HCl or NaOH and bring to a ®nal volume of 10 mL. Enzyme solution B. Dissolve 65 units of pronase in distilled water at 378C. Adjust to pH 8 using dilute HCl or NaOH and bring to a ®nal volume of 10 mL. Store the enzyme solution in ice.

Procedure* Determination of in vitro protein digestibility via a three-enzyme test. Add the food sample (62.5 mg or equivalent to 10 mg N) to about 6± 8 mL of distilled water and soak for 60 minutes at 378C. Adjust to pH 8 using dilute HCl or NaOH and bring to a ®nal volume of

* Pronase (enzyme solution B) is not needed in the three-enzyme method. 370 Chapter 12

10 mL. Add 1 mL of enzyme solution A, incubate for 10 minutes at 378C, and record the pH change at 378C. Determination of in vitro protein digestibility via a four-enzyme test. Proceed as above. Exactly 10 minutes after adding enzyme solution A, add 1 mL of enzyme solution B. Transfer the mixture to 558C for 9 minutes. Return the sample mixture to 378C and record the pH change. The net reaction time should be 20 minutes. The decline in pH during proteolysis follows a power law, DpH Btm, where t is reaction time and B and m are constants. Equation parameters were determined by regression analysis. Results from in vitro protein digestibility tests were highly correlated with the apparent protein digestibility (APD) measured using the rat bioassay:

APD %210:46 À 18:10 DpH 10 min 6

APD %234:84 À 22:56 DpH 20 min 7

Fig. 4 shows the relation between in vitro protein digestibility and in vivo results for over 20 processed and unprocessed food protein sources. The right-hand axis in Fig. 4 shows the residuals for pH (10 min) results, i.e., difference between in vitro and in vivo test results. These is no systematic trend in the residuals. In vitro protein digestibility estimates using the multienzyme method were not affected by the buffer capacity of normal food samples. In a collaborative test for the three-enzyme method, pH was monitored with a pH-Stat instrument (121).* For sample pretreatment fresh or canned foods were freeze-dried and ground into a powder. High- lipid foods were defatted by extracting with anhydrous cold ether. In vitro protein digestibility was examined essentially as described in Method 4 with minor modi®cations. Reaction progress was monitored from the volume of 0.1 N NaOH added to the reaction vessel by the pH-Stat set at pH 8. In vitro protein digestibility{ was calculated from

IVPD %79:28 40:74 V mL 8

* The test using a pH-Stat is AOAC approved. Participants in the interlaboratory tests were af®liated with the USDA (Beltsville) Health and Welfare Canada (Ottawa), Protein Technologies International (St. Louis, MO), CIVO-TNO Food Analysis Institute±AJ (Zeist, The Netherlands), Kraft Inc. (Glenview, IL), National Institute of Animal Science (Tjele, Denmark). { The term in vitro protein digestibility has been substituted for ``percent true digestibility'' in the original reference to avoid confusion with true protein digestibility (TPD). Tests for Protein Nutrient Value 371

FIGURE 4 Correlation between in vitro protein digestibility and in vivo apparent protein digestibility (APD) in the rat. Results for three-enzyme (ÁÁÁÁÁ circles) or four-enzyme (ÐÐ) test. Protein sources are casein, soy isolate, soy protein concentration, delactosed whey, corn, high-protein wheat bran ¯our, general wheat ¯our, nonfat dry milk, cottonseed meal, wheat protein concentrate, yeast protein concentrate, bean protein concentrate, soy concentrate, and full lactose whey. Triangles show residuals. (Drawn using results from Refs. 118±120.) where IVPD stands for in vitro protein digestibility and V (mL) is the volume of alkali added to the enzyme reaction over a 5-minute period. In vitro protein digestibility values were adjusted according to a scale with casein assigned an arbitrary value of 100. From 204 independent analyses (17 samples* 6 2 replicates 6 6 laboratories) the between-laboratory reproducibility was 0.83±5.0%. The within-laboratory repeatability ranged from 0.35 to 1.4%. The multienzyme in vitro protein digestibility method allows a precise and rapid assessment of digestibility. The correlation between TPD and in vitro protein digestibility implies that protein

* The 17 protein sources examined were casein, nonfat dried milk, nonfat dried milkÐheated, tuna ®sh, salami, canned sausage, chicken frankfurters, beef stew, macaroni and cheese, peanut butter, soy isolate, pea concentrate, chick peas, rolled oats, pinto beans, whole wheat cereal, and rice-wheat gluten cereal. 372 Chapter 12 bioavailability is largely determined by the rate of protein breakdown by digestive enzymes.

5.4. Sample Pretreatment to Inactivate Trypsin Inhibitors Suspend the sample (62.5 mg or 10 mg N) with 2.5 mL of water and 2.5 mL of 0.2 N NaOH. Then incubate at 308C for 30 minutes. The high pH conditions inactivate trypsin inhibitors. Thereafter, add 5 mL of 0.075 N HCl and adjust the mixture to pH 8 and a volume of 10 mL (121). Add the three-enzyme mixture (1 mL) and proceed as described in Method 5.

5.5. Degree of Protein Hydrolysis Adler-Nissen (122) measured in vitro protein digestibility from the degree of hydrolysis (DH). This is the number of peptide bonds hydrolyzed (h; milliequivalents per g)* divided by the total number of peptide bonds (htot). 100h DH % 9 hTOT

The htot is also equal to the net concentration of amino acids (mmoles) per gram of protein. The DH was determined using a pH-Stat instrument to measure the volume of alkali required to maintain a static pH value during proteolysis (V): VN DH % b 10 a MhTOT where Nb is the concentration of the alkali and M the mass of protein (Kjeldahl N 6 6.25). The term a (degree of dissociation of the ammonium groups produced by proteolysis) is dependent on the sample temperature and pH. During initial rate measurements with DH below 20%, the following relations apply:

10pHÀpK a 11 1 10pHÀpK

* One milliequivalent millimolar multiplied by the valence of the chemical species. Tests for Protein Nutrient Value 373 and 298 À T pK 7:8 2400 12 298:T Under standard conditions used for the approved test (pH 8, T 378C and pK 6.91) we may suppose that a & 0.91. For a wide range of proteins À1 the typical value for htot is about 8±9 meq g protein (122), that is, htot 1/F where F (&112 g moleÀ1) is the average formula weight for all 20 naturally occurring amino acids (see Chapter 1).

5.6. In Vitro Protein Digestibility with Dialysis During in vitro protein digestibility tests the concentration of products increases within an enclosed container. By contrast, digestion products are continually removed from the digestive tract. To simulate the absorption process, apparatus for protease digestion coupled with continuous dialysis was devised by Baldwin et al. (3). Mauron et al. (123) used a re®ned version of the dialysis apparatus to examine the PNV for condensed milk and milk powder. A digestion unit with continuous dialysis was also used by Gauthier et al. (124) to assess the effect of heat and alkali treatment on casein, soybean, and rapeseed protein. The device consists of a dialysis bag (containing enzyme, food sample, and a stirrer) suspended in a cylindrical vessel containing buffer. The cylinder is ®tted with a tap so that dialysate can be periodically withdrawn for analysis and replaced with an equal volume of buffer. Undigested protein and enzyme are retained within the dialysis bag. In later studies, dialysate was continuously removed, and compensatory amounts of buffer were added using a peristaltic pump. Using the digestion-dialysis method, in vitro protein digestibility is measured in terms of the ND or the ``nitrogen digestibility index'' (125). N dialysate ND 6100 13 N dialysateN dialysis bag where N(dialysate) and N(dialysis bag) are the amount of nitrogen remaining in the dialysate and within the dialysis bag after 24 hours. The main disadvantage of the dialyzed±in vitro protein digestibility tests is their slowness. The 1-kDa cutoff dialysis membrane impedes the release of products. Kennedy et al. (126) described an in vitro protein digestibility test employing pepsin followed by dialysis of reaction products. The system was monitored by Kjeldahl or automated amino acid analysis. The results were described as the pepsin digest dialysate (PDD) index. This and other 374 Chapter 12 attempts to correct amino acid scores for digestibility are discussed further in Chapter 14.

REFERENCES

1. CH Lea, RS Hannan. Studies of the reaction between proteins and reducing sugars in the ``dry state.'' III. Nature of the protein groups reacting. Biochim Biophys Acta 5:433±454, 1950. 2. M Tanaka, TC Lee, C Chichester. Effect of browning on chemical properties of egg albumin. Agric Biol Chem 39:863±866, 1975. 3. RR Baldwin, JR Lowry, R Thiessen Jr. Some effects of processing on the nutritive properties of proteins. Food Res 16:107±117, 1951. 4. MN Rao, KS Shurpalekar. Recent trends in protein methodology. World Rev Nutr Diet 17:150±163, 1973. 5. D Rosen®eld. Utilizable protein: quality and quantity concepts in assessing food. Food Prod Dev 7:57, 60, 62, 1973. 6. M Friedman Jr. Protein Nutritional Quality of Foods and Feeds. Part I. Assay MethodsÐBiological, Biochemical, and Chemical. New York: Marcel Dekker, 1975. 7. PL Pellet. Protein quality evaluation revisited. Food Technol 32 (5):60, 62, 64± 66, 68, 70±72, 74, 76, 78±79, 1978. 8. DT Hopkins. Updating protein quality measurement techniques. Cereal Foods World 23:539±543, 1978. 9. TR Anderson. Available lysineÐII. Chemical methods. S Afri Food Rev 7(6):113, 115, 119, 121, 123, 1980. 10. M Friedman. Chemically reactive and unreactive lysine as an index of browning. Diabetes 31(Suppl 3):5±14, 1982. 11. AE Bender. Nutritional value of proteins and its assessment. In: PF Fox, JJ Condon, eds. Food Proteins. Barking, Essex: Applied Science Publishers, 1982, pp 121±131. 12. LD Satterlee, KC Chang. Nutritional quality of deteriorated proteins. In JP Cherry, ed. Food protein deterioration. Mechanisms and functionality. Washington, DC: American Chemical Society, pp 409±431, 1982. 13. AN Walker. The estimation of protein quality. In BJF Hudson, ed. Developments of Food ProteinsÐ2. London: Elsevier Applied Science, 1983, pp 293±323. 14. G Sarwar. Digestibility of protein and bioavailability of amino acids in foods. Effects on protein quality assessment. World Rev Nutr Diet 54:26±70, 1987. 15. G Sarwar, FE McDonough. Evaluation of protein digestibilityÐcorrected amino acid score method for assessing protein quality of foods. J Assoc Off Anal Chem 73:347±356, 1990. 16. AA Woodham. The nutritional evaluation of leaf protein concentrates. In L Telek, HD Graham, eds. Leaf Protein Concentrates. Westport, CT: AVI Publishing Co, 1983, pp 415±133. Tests for Protein Nutrient Value 375

17. SB Vaghe®, DD Makdani, O Mickelsen. Lysine supplementation of wheat proteins. Am J Clin Nutr 27:1231±1246, 1974. 18. D Knorr. Protein quality of the potato and potato protein concentrates. Lebensin Wiss Technol 11:109±115, 1978. 19. JE Kinsella. Texturized proteins: fabrication, ¯avoring, and nutrition. CRC Crit Rev Food Sci Nutr 10:147±207, 1978. 20. K Lorenz. Cereal sprouts: composition, nutritive value, food applications. CRC Crit Rev Food Sci Nutr 13:353±385, 1980. 21. KC Change, LD Satterlee. Chemistry of dry bean proteins. J Food Process Preserv 6:203±225, 1982. 22. SK Sathe, SS Deshpande, DK Salunkh. Dry beans of Phaseolus. A review. I. Chemical composition: proteins. CRC Crit Rev Food Sci Nutr 20:1±46, 1984. 23. K Devadasan. Oxidation of residual lipids in ®sh protein concentrates and its effect on the nutritional quality of the protein. Fish Technol 18:71±77, 1981. 24. R Uauy, E Yanez. Plant foods for human protein nutrition;studies on soy, lupin and mixed vegetable sources. Plant Foods Hum Nutr 33:17±28, 1983. 25. VR Young, N Scrimshaw. Clinical studies on the nutritional value of SCP. In SR Tannenbaum, DIC Wang, eds. Single Cell Protein II. Cambridge, MA: MIT Press, 1975, pp 564±586. 26. HF Erbersdobler, AB Holstein, HW Niehoff. Protein quality of a bacterial single cell protein (SCP) (Methylomonas) and its improvement by various processes. Food Rev 11:75±77, 1984. 27. M Friedman. Nutritional value of proteins from different food sources. A review. J Agric Food Chem 44:6±26, 1996. 28. JW Claassens, HJ Potgieter. Proteins and Food Supply in the Republic of South Africa. Capetown, Republic of South Africa: AA Balkema, 1971. 29. JWG Porter, BA Rolls. Proteins in Human Nutrition. London: Academic Press, 1973. 30. M Friedman, ed. Nutritional Improvement of Food and Feed Proteins. New York: Plenum, 1978. 31. M Friedman, ed. Protein Nutritional Quality of Foods and Feeds. Part I. Assay MethodsÐBiological, Biochemical, and Chemical. New York: Marcel Dekker, 1975. 32. M Friedman, ed. Protein Nutritional Quality of Foods and Feeds. Part II. Quality FactorsÐPlant Breeding, Composition, Processing, and Antinutri- ents. New York: Marcel Dekker, 1975. 33. CE Bodwell, JS Adkins, DT Hopkins, eds. Protein Quality in Humans. Westport, CT: AVI Publishing Co, 1981. 34. RD Phillips, JW Finley, eds. Protein Quality and the Effects of Processing. New York: Marcel Dekker, 1989. 35. M Milner, ed. Protein-Enriched Cereal Foods for World Needs. St Paul, MN: American Association of Cereal Chemists, 1969. 36. AE Bender, R Kihberg, B Loefqvist, L Munck, eds. Evaluation of Novel Protein Products. Oxford, UK: Pergamon Press, 1970. 376 Chapter 12

37. PL White, DC Fletcher. Nutrients in Processed Foods. Proteins. Acton, MA: Publishing Sciences Group, 1974. 38. RE Olson, ed. Protein-Calorie Malnutrition. New York: Academic Press, 1975. 39. JR Whitaker, SR Tannenbaum. Food Proteins. Westport, CT: AVI Publish- ing Co, 1977. 40. M Milner, NS Scrimshaw, DIC Wang, eds. Protein Resources and Technology: Status and Research Needs. Westport, CT: AVI Publishing Co, 1978. 41. J Rakosky, ed. Protein Additives in Foodservice Preparations. New York: Van Nostrand±Reinhold, 1989. 42. GU Liepa, ed. Dietary Proteins. How They Alleviate Disease and Promote Better Health. Champaign, IL: American Oil Chemists Society, 1992. 43. S Damodaran, A Paraf, eds. Food Proteins and Their Applications. New York: Marcel Dekker, 1997. 44. JW Finely, DT Hopkins, eds. Digestibility and Amino Acid Availability in Cereals and Oilseeds. St. Paul, MN: American Association of Cereal Chemists, 1985. 45. J Wiseman, DJA Cole, eds. Feedstuff Evaluation. London: Butterworths, 1990. 46. VR Young, WM Rand, NS Scrimshaw. Measuring protein quality in humans: a review and proposed method. Cereal Chem 54:929±948, 1977. 47. VR Young, NS Scrimshaw, B Torun, F Vilteri.. Soybean protein in human nutrition: an overview. J Am Oil Chem Soc 56:110±120, 1979. 48. VR Young, NS Scrimshaw, DM Bier. Whole body protein and amino acid metabolism: relation to protein quality evaluation in human nutrition. J Agric Food Chem 29:440±447, 1981. 49. D Rosen®eld. Protein quality testing: introductory remarks. Food Technol 32(12):51±56, 1978. 50. HW Staub. Problems in evaluating the protein nutritive quality of complex foods. Food Technol 32(12):57±61, 1978. 51. L R Hackler. An overview of the AACC/ASTM collaborative study on protein quality evaluation. Food Technol 32(12):62±64, 1978. 52. RH Anderson. Protein quality testing: industry needs. Food Technol 32(12):65, 68, 1978. 53. MA Burdette III, II Russoff. GMA test protocol for protein quality assays. Food Technol 32(12):66±68, 1978. 54. HW Hsu, NE Sutton, MO Banjo, D Satterlee, JG Kendrick. The C-PER and T-PER assays for protein quality. Food Technol 32(12):69±73, 68, 1978. 55. FH Steinke. Protein ef®ciency ratio pitfalls and causes of variability: a review. Cereal Chem 54:949±957, 1977. 56. CE Bodwell. Application of animal data to human protein nutrition: a review. Cereal Chem 54:958±983, 1977. 57. LR Hackler. Methods for measuring protein quality: a review of procedures. Cereal Chem 54:984±995, 1977. Tests for Protein Nutrient Value 377

58. TB Osborne, LB Mendel, EL Ferry. A method of expressing numerically the growth promoting effect of protein. J Biol Chem 37:223±229, 1919. 59. DG Chapman, R Castillo, JA Campbell. Evaluation of protein in foods. I. A method for the determination of protein ef®ciency ratios. Can J Biochem Physiol 37:679, 1959. 60. AE Bender, BH Doell. Biological evaluation of proteins. A new aspect. Br J Nutr 11:140, 1957. 61. DS Miller, AE Bender. The determination of net utilization of proteins by a shortened method. Br J Nutr 9:382, 1955. 62. DM Hegsted, YO Chang. Protein utilization in growing rats. I. Relative growth index as a bioassay procedure J Nutr 85:159±168, 1965. 63. DN Hegsted, R Neff, J Worcester. Determination of the relative nutritive value of proteins. Factors affecting precision and validity. J Agric Food Chem. 16:190±195, 1968. 64. C Kies, HM Fox, PJ Mattern, VA Johnson, JW Schmidt. Comparative protein quality as measured by human and small animal bioassays of three lines of wheat. In: M Friedman, ed. Nutritional Improvement of Food and Feed Proteins. London: Plenum, 1978, pp 91±102. 65. FE McDonough, FH Steinke, G Sarwar, BO Eggum, R Bressani, PJ Huth, WE Barbeau, GV Mitchell, JG Phillips. In-vivo rat assay for true protein digestibility. Collaborative study. J Assoc Off Anal Chem 73:801±805, 1990. 66. DT Hopkins. Protein quality in humans: assessment and in vitro estimation. In: CE Bodwell, JS Adkins, DT Hopkins, eds. Westport, CT: AVI Publishing Co, 1981, pp 169±193. 67. G Sarwar. Digestibility of protein and bioavailability of amino acids in foods. Effects on protein quality assessment. World Rev Nutr Diet 54:26±70, 1987. 68. W Choppe, FH Kratzer. Methods for evaluating the feeding quality of meat- and-bone meals. Poultry Sci 42:642±646, 1963. 69. ET Moran Jr, LS Jensen, J McGinnis. Dye binding by soybean and ®sh meal as an index of quality. J Nutr 79:239±244, 1963. 70. HH Mitchell, RJ Block. Some relationships between the amino acid contents of proteins and their nutritive value for the rat. J Biol Chem 163:599, 1946. 71. BL Oser. Method for integrating essential amino acid content in the nutritional evaluation of protein. J Am Diet Assoc 27:396±402, 1951. 72. F Sanger. The free amino groups of insulin. Biochem J 39:507±515, 1945. 73. CK Carpenter. The estimation of the available lysine in animal-protein foods. Biochem J 77:605±610, 1960. 74. RF Hurrell, KJ Carpenter. Mechanism of heat damage in proteins 4. The reactive lysine content of heat-damaged material as measured in different ways. Br J Nutr 32:589±604, 1974. 75. VH Booth. Problems in the determination of FDNB-available lysine. J Sci Food Agric 22:658±666, 1971. 76. AG Roach, P Sanderson, DR Williams. Comparison of methods for the determination of available lysine value in animal and vegetable protein sources. J Sci Food Agric 18:274±278, 1967. 378 Chapter 12

77. H Ostrowski, AS Jones, A Cadenhead. Availability of lysine in protein concentrates and diets using Carpenter's method and a modi®ed Silcock method. J Sci Food Agric 21:103±107, 1970. 78. JR Couch. Collaborative study of the determination of available lysine in proteins and feeds. J Assoc Off Anal Chem 58:599±601, 1975. 79. ML Kakade, IE Liener. Determination of available lysine in proteins. Anal Biochem 27:273±280, 1969. 80. CC Gondo, HE Swaisgood, GL Catignani. A ¯uorimetric assay for available lysine in proteins. Anal Biochem 115:203±211, 1981. 81. M Friedman, JW Finley. Reactions of proteins with ethyl vinyl sulfone. Int J Peptide Prot Res 7:481±486, 1975. 82. L Sandler, FL Warren. Effect of ethyl chloroformate on the dye binding capacity of protein. Anal Chem 46:1870±1972, 1974. 83. CH Lea, RS Hannan. Studies of the reaction between proteins and reducing sugars in the ``dry state.'' II. Further observations on the formation of casein- glucose complex. Biochim Biophys Acta 4:518±534, 1950. 84. SR Rao, FL Carter, VL Frampton. Determination of available lysine in oilseed proteins. Anal Chem 35:1927±1930, 1963. 85. VL Frampton, JC Kuck. Sources of error in determination of available lysine in cottonseed and peanut meals. J Am Oil Chem Soc 50:304±306, 1973. 86. JC Kuck, VL Frampton. Improvement in determination of available lysine in cottonseed meals. J Am Oil Chem Soc 52:214, 1975. 87. LA Hussein. Comparison of methods for the determination of available lysine value in animal protein concentrations. J Sci Food Agric 25:117±120, 1974. 88. L Blom, P Hendricks, J Caris. Determination of available lysine in foods. Anal Biochem 21:382±400, 1967. 89. NA Matheson. Available lysine. 1. Determination of non±N-terminal lysine in protein. J Sci Food Agric 19:492±495, 1968. 90. NA Matheson. Available lysine. II. Determination of available lysine in feedingstuffs by dinitrophenylation. J Sci Food Agric 19:496±502, 1968. 91. CJ Bailey. Automated analysis of available lysine and tyrosine in foodstuffs. J Sci Food Agric 25:1007±1014, 1974. 92. WR Peterson, JJ Warthesen. Total and available lysine determinations using high pressure liquid chromatography. J Food Sci 44:994±997, 1979. 93. J Rabasseda, G Rauret, T Galceran. Liquid chromatographic determination of available lysine in soybean and ®sh meal. J Assoc Off Anal Chem 71:350± 353, 1988. 94. G Castillo, MA Sanz, MA Serrano, T Hernandez, A Hernandez. An isocratic high-performance liquid chromatographic method for determining the available lysine in foods. J Chromatogr Sci 35:423±429, 1997. 95. A Hernandez, MA Serano, MM Munoz, G Castillo. Liquid chromatographic determination of the total available free and intrachain lysine in various foods. J Chromatogr Sci 39:39±43, 2001. 96. T Okuyama, K Satake. The preparation and properties of 2,4,6-trinitrophenyl-amino acids and peptides. J Biochem (Tokyo) 47:454±466, 1960. Tests for Protein Nutrient Value 379

97. K Satake, T Okuyama, M Ohashi, T Shinoda. The spectrophotometric determination of amines, amino acids, and peptides with 2,4,6-trinitrobenzene-1-sulfonic acid. J Biochem (Tokyo) 47:654±660, 1960. 98. AFSA Habeeb. Determination of free amino groups in proteins by trinitrobenzenesulfonic acid. Anal Biochem 14:328±336, 1966. 99. LC Mokrasch. Use of 2,4,6-trinitrobenzenesulfonic acid for the coestimation of amines, amino acids, and protein mixtures. Anal Biochem 18:64±71, 1967. 100. RM Tomarelli, RJ Yuhas, A Fisher, JR Weaber. An HPLC method for the determination of reactive (available) lysine in milk and infant formulas. J Agric Food Chem 33:316±318, 1985. 101. JR Couch, MC Thomas. A comparison of chemical methods for the determination of available lysine in various proteins. J Agric Food Chem 24:943±946, 1976. 102. A Eklund. On the determination of available lysine in casein and rapeseed protein concentrates using 2,4,6-trinitrobenzenesulphonic acid (TNBS) as a reagent for free epsilon amino group of lysine. Anal Biochem 70:434±439, 1976. 103. LP Posati, VH Holsinger, ED DeVilbiss, MJ Pallansch. Factors affecting the determination of available lysine in whey with 2,4,6-trinitrobenzene sulfonic acid. J Dairy Sci 55:1660±1665, 1972. 104. R Greenberg, HJ Dower, JH Woychik. An improved trinitrobenzene sulfonic acid procedure for the determination of available lysine in nonfat dry milk. Abstr Papers Am Chem Soc 173:AGFD 72, 1977. 105. RJ Hall, K Henderson. An improvement in the determination of available lysine in carbohydrate-rich samples. Analyst 104:1097±1100, 1979. 106. RF Hurrell, P Lerman, KJ Carpenter. Reactive lysine in foodstuffs as measured by a rapid dye-binding procedure. J Food Sci 44:1221±1227, 1231, 1979. 107. IM Pearl, MP Szakacs, A Kovogo, J Petroczy. Stoichiometric dye-binding procedure for the determination of reactive lysine content of soya bean. Food Chem 16:163±174, 1985. 108. PA Finot, F Mottu, E Bujard. Biological availability of true Schiff's bases of lysineÐchemical evaluation of the Schiff's base between lysine and lactose in milk. Abstr Papers Am Chem Soc 172:ACFD 70, 1976. 109. JK Thompson, VR Fowler. The evaluation of minerals in the diets of farm animals. In: J Wiseman, DJ Cole, eds. Feedstuff Evaluation. London: Butterworths, 1990, pp 235±266. 110. C Kies. Bioavailability: a factor in protein quality. J Agric Food Chem 29:435±440, 1981. 111. W Wu, WP Williams, ME Kunkel, JC Acton, Y Huang, FB Wardlaw, LW Grimes. True N conversion factor for diet and excreta in evaluating protein quality. J Food Sci 60:854±857, 1995. 112. LV Hankes, WH Hiesen, LM Henderson, CA Elvehjem. Liberation of amino acids from raw and heated casein by acid and enzyme hydrolysis. J Biol Chem 176:467±476, 1948. 380 Chapter 12

113. NR Eldred, G Rodney. The effect of proteolytic enzymes on raw and heated casein. J Biol Chem 162:261±265, 1946. 114. D Melnick, BL Oser. The in¯uence of heat-processing on the function and nutritive properties of protein. Food Res 3(2):57±71, 1949. 115. HE Swaisgood, GL Catignani. In vitro measurement of effects of processing on protein nutritional quality. J Food Prot 45:1248±1256, 1982. 116. HE Swaisgood, GL Catignani. Protein digestibility: in vitro methods of assessment. Adv Food Nutr Res 35:185±236, 1982. 117. S Boisen, BO Eggum. Critical evaluation of in vitro methods for estimating digestibility in simple-stomach animals. Nutr Res Rev 4:141±162, 1991. 118. HW Hsu, DL Vavak, LD Satterlee, GA Miller. A multienzyme technique for estimating protein digestibility. J Food Sci 42:1269±1273, 1977. 119. LD Satterlee, HF Marshall, JM Temmyson. Measuring protein quality. J Am Oil Chem Soc 56:103±106, 1976. 120. LD Satterlee, JG Kendrick, HF Marshall, DK Jewell, RA Ali, MM Heckman, F Steinke, P Larson, RD Philips, G Sarwar, P Slump. In vitro assay for predicting protein ef®ciency ratio as measured by rat bioassay: collaborative study. J Assoc Off Anal Chem 65:798±806, 1982. 121. FE McDonough, G Sarwar, FH Steinke, P Slump, S Garcia, S Boisen. In vitro assay for protein digestibility: inter-laboratory study. J Assoc Off Anal Chem 73:622±625, 1990. 122. J Adler-Nissen. Enzymic Hydrolysis of Food Proteins. London: Elsevier Applied Science, 1986, chapters 2 and 6. 123. J Mauron, F Mottu, E Bujard, RH Egli. The availability of lysine, methionine and tryptophan in condensed milk and milk powder. In vitro digestion studies. Arch Biochem Biophys 59:433±451, 1955. 124. SF Gauthier, C Vachon, JD Jones, L Savoie. Assessment of protein digestibility by in vitro enzymatic hydrolysis with simultaneous dialysis. J Nutr 112:1718±1725, 1982. 125. L Savoie, SF Gauthier. Dialysis cell for the in vitro measurement of protein digestibility. J Food Sci 51:496±498, 1982. 126. JF Kennedy, RJ Noy, JA Stead, CA White. A new rapid enzyme digestion method for predicting in vitro protein quality (PDD index). Food Chem 32:277±295, 1989. 13

Effect of Processing on Protein Nutrient Value

1. INTRODUCTION

Chemical and biological techniques for determining protein nutrient value (PNV) are described in Chapter 12. The effect of food processing on PNV for speci®c food commodities is discussed here. Up until 1993±1994 the protein effeciency ratio (PER) assay was the reference method for assessing protein quality (1). It has been superseded by a new measure of protein quality called protein digestibility±corrected amino acid score, PDCAAS (Chapter 14). The PER remains important for comparative purposes and is still widely used. The present discussion is broadly arranged according to different food commodity groups.

2. MILK AND MILK POWDERS

Lea, Hannan and co-workers studied the effects of sterilization, drying, and storage on milk protein quality over 50 years ago (2±5). Their ®ndings are discussed throughout this section. Milk is subjected to various thermal treatments before drying. A range of temperature-time treatments (728C/ 15 sec, 958C/30 sec, 758C/30 min, or 808C/30 min) had no signi®cant effect on the available lysine (6). Pasteurization (728C for 15 seconds), domestic

381 382 Chapter 13 boiling (30 seconds), or UHT treatment (135±1508C for a few seconds) did not reduce the PER or digestibility determined using the rat bioassay (7). From the standpoint of protein quality, there is no nutritional reason for discouraging heat processing of milk for human consumption. The total lysine content of milk ranges from 6.2 to 11.1 g 100 gÀ1 (protein). Roller drying milk that had been preprocessed by UHT treatment produced 2.5± 6.4% reduction in available lysine (8). Scorching during roller drying produced up to 35% loss of lysine (6), whereas storing dried milk for 6 months produced 15.4% decline in the available lysine (9).

2.1. Spray-Dried Delactose Milk A high lactose content is undesirable in milk intended for adults. Intestinal lactase levels are reduced in so-called hypolactasic populations, leading to a reduced tolerance for dietary lactose. The acceptable oral dose of lactose is thereby limited to 5±10 g or 100±200 mL of milk. Hydrolysis of milk lactose using exogenous lactase increases the tolerance of hypolactasic subjects for milk. Such lactose-hydrolyzed milk has different processing requirements due to its higher content of reducing sugars. The effect of heat processing on the protein quality for delactase milk was investigated by Burvall et al. (10). Normal milk samples were prepackaged in 1-L paper cartons and injected with ®lter-sterilized lactase. Enzyme treatment reduced milk lactose levels by 99%. Normal milk and lactose-treated milk were UHT sterilized by steam injection (1428C, 2 seconds) and evaporated to 25% dry weight and/or spray dried (®nal water content 4%;air inlet and outlet temperatures of 185 8Cand 858C). The total digestibility coef®cient (TDC), biological value (BV), and net protein utilization (NPU) were then measured.* Spay drying reduced the BV for delactose milk by 45% compared with 10% for normal spray-dried milk powder (Fig. 1). The NPU was also signi®cantly lower for delactosed spray-dried milk. Pre-evaporation allowed more rapid drying and reduced the loss of quality. There was no visible browning. Protein quality losses were due to the formation of early Maillard reaction products. Hence supplementation with lysine improved the nutrient value of delactosed spray-dried milk.

2.2. Storage Stability and Protein Nutrient Value The effect of prolonged storage on the PNV of freeze-dried versus spray- dried lactose-hydrolyzed milk powder was examined by Rawson and

* Measures of protein quality are de®ned in Chapter 12. Effect of Processing on Protein Nutrient Value 383

FIGURE 1 Effect of processing on milk protein quality. Y-axis values are compared with freeze-dried normal milk powder as control. X-axis abbreviations: N-SD, spray-dried normal milk;LH SD, lactose-hydrolyzed and spray dried;LH EV SD, lactose hydrolyzed±evaporated±spray dried; LH SD ( Lys), lactose hydrolyzed and spray dried with 1% (w/w) lysine added. Insert shows TDC, total digestibility;BV, biological value; NPU, net protein utilization determined with a rat bioassay. (Data from Ref. 10.)

Mahoney (11,12). Freeze drying is a more expensive method for producing dry milk. The dried milk powders were stored at À208Cor308C for up to 120 days. As shown in Table 1, spray drying had a more adverse effect on the initial quality. However, PNV was the same for freeze-dried and spray- dried milk powder samples after prolonged storage;both lost 46±48 % of FDNP-reactive lysine after 120 days storage. The net protein ratio for both samples also declined by nearly equal amounts. Owing to the lower processing costs, spray drying may be advantageous for milk powders intended for prolonged storage. Although Rawson and Mahoney (11,12) found reactive lysine losses of 18% after spray drying delactose milk, others have reported higher (50%) losses (10,13).

2.3. Effect of Water Activity, pH, and Temperature Freeze-dried milk powder was equilibrated with saturated salt solutions of known water activity (AW 0.11±0.62). The moisture-equilibrated samples 384 Chapter 13

TABLE 1 Effect of Processing and Storage on Spray-Dried or Freeze-Dried Delactosed Milk

FDNB-Lys DBCb Total Lys Rat assay Samplea (mg gÀ1) (mg gÀ1) (mg gÀ1) NPR

Normal milk (fresh) 28 218 29.1 4.8 Freeze-dried 25 178 28.9 Ð Freeze-dried, 120 d, À208C 25 200 29.1 Ð Freeze-dried, 120 d, 308C 12 120 15.7 3.0 Spray-dried 23 195 23.4 Ð Spray-dried, 120 d, À208C 23 235 23.6 4.0 Spray-dried, 120 d, 308C 17 150 17.8 3.2 aProcessing and storage conditions. bDye-binding capacity determined using Remazo Brilliant Blue R as described in Ref. 11. were then stored for 1±12 months. Protein quality was assessed using the rat bioassays. Available lysine was assayed using the o-methylisourea method (13). The loss of available lysine was greater for lactose-hydrolyzed milk powder compared with normal milk powder. Chemical and biological assays of available lysine were highly correlated. There was a bell-shaped relationship between AW and quality loss. Maximum losses of PNV occurred at AW 0.55. These observations agree with results for the dry caseinate-glucose system (3). Lea and Hannan (3) investigated the reaction between dry sodium caseinate and glucose at temperatures of 0±708C and

AW < 0.95. The system was adjusted to pH 3±10 before drying. The decline in free amino groups after 65 days storage at 378C, followed a bell-shaped curve with a maximum at AW 0.6±0.8. At low AW the reaction rate may be limited by the restricted diffusion of reagents. At high AW reactant dilution or inhibition by water accounts for the decreased rate of deterioration. The loss of lysine increased steadily from pH 3 to pH 8 and leveled off at pH 8±

10. The effect of temperature (AW 0.75) on the kinetics of lysine loss conformed with the simple Arrhenius equation ln(k) 14446.7(1/T) 38.0 (r 0.9999), where k is the percent loss of lysine per second and T temperature in kelvin (Fig. 2). Alternative models involving nonlinear temperature dependence are described in Chapter 14.

3. INFANTFORMULAS

Mitchell and Grundel (14) compared several chemical methods (total lysine determination by amino acid analysis, differential dye-binding capacity, in Effect of Processing on Protein Nutrient Value 385

FIGURE 2 Arrhenius plot showing the effect of temperature on the rate of Lys loss for casein-glucose mixture at 70% RH. The temperature range studied was 0±708C. Reaction pH (not given) is probably pH 6.3 or near the natural pH for dried milk. (Drawn from data in Refs. 3±5.) vitro protein digestibility) and bioassays (TPD, PER) for evaluating protein quality for ®ve infant formulas.* These protein quality indices were evaluated for their sensitivity to ``humidity-related'' damage. For all ®ve infant formulas, in vitro protein digestibility was said to give a poor estimate of TPD. For example, the in vitro digestibility was was 88±89% for soy- based infant formulas compared with the TPD of 91±94%. For milk-based formulas, the in vitro protein digestibility was 85±88% compared with a preadjusted TPD of 85±88%;the latter TPD value was adjusted to 96±97 % to allow for lactose intolerance in rats. Apparently there is a digestibility de®cit of 11% when rats are fed diets with lactose levels comparable to the concentration of lactose in infant formulas. The dDBC values agreed with total lysine results. The rat bioassay showed no protein quality changes for 30-day-old infant formulas;however, there were large drops for in vitro

* The designation and composition of the ®ve commercial infant formulas were: milk-based I (skimmed milk, reduced minerals, whey, lactose, oleo, coconut, soy, oleic oils, and lecithin), milk-based II (skimmed milk, lactose, coconut and corn oils), milk-based III (skimmed milk, lactose, soy and coconut oils), soy-based I (soy protein isolate, corn syrup solids, coconut and corn oils), and soy-based II (soy protein isolate, soy oil, lecithin, sucrose, tapioca dextrin). 386 Chapter 13 protein digestibility and dDBC. Mitchell and Grundel (14) concluded that (a) agreement among the various assays is poor although different in vitro methods show qualitative similarities in some cases, (b) changes in dDBC and browning of the stored milk-based and soy-based formulas do not re¯ect changes in relative PER, (c) in vitro methods are not accurate predictors of protein quality for the rat, and (d) only certain humidity- related protein nutritional damage in infant formulas could be predicted by monitoring dDBC or in vitro protein digestibility (14). An interesting feature of this study is the assumption that lysine is the limiting amino acid for soy-based infant formulas (see later). Sarwar et al. (15) evaluated the impact of amino acid supplementation on protein quality for four commercial soy-based infant formulas. The samples were fed to 2-week-old weanling rats as the sole source of protein. Each diet contained net 8% protein, 20% fat, and adequate amounts of minerals and vitamins. The RPER and RNPR for casein plus methionine were assigned a value of 100. By comparison, the RPER and RNPR values for the infant formulas were 71±81 and 78±85, respectively. There was no improvement in RPER or RNPR values when diets were supplemented with either lysine (0.2%), methionine (0.2%), threonine (0.1%), or tryptophan (0.05%). Increased levels of amino acids appeared in the serum, indicating ef®cient adsorption. Protein quality improved following the supplementation with four essential amino acids (RPER or RNPR values 100). Apparently several amino acids are colimiting for soy-based formulas. Measuring available lysine is a useful way to monitor protein quality where it has been demonstated that lysine is the (only) limiting amino acid. Otherwise, focusing on a single amino acid will lead to errors of the type described in the preceding paragraph. Processing losses and shelf-life studies for infant formulas have also been reported by other investigators (16,17). Under optimal conditions, infant formulas can be stored for prolonged periods (16±20 weeks) without losses of available lysine. Poor storage conditions can lead to a 20±30% loss of available lysine.

4. FEEDSTUFFS AND CONCENTRATES FOR LIVESTOCK

Protein concentrates and dietary supplements for livestock are derived from both animal and plant sources (Table 2). The commercial value for feed concentrates depends on their total nitrogen content as determined by the Kjeldahl analysis. Differences in feed protein quality arise due to (a) intrinsic differences in the quality of raw materials and (b) effects of processing. Heat processing renders feed material more portable and palatable. It also improves the chemical and microbiological stability of Effect of Processing on Protein Nutrient Value 387

TABLE 2 Protein Quality Testing of Feedstuffs and Concentrates (1955±1960)

Feedstuffs Meat and bone meals Fish meal Whale meal Dried whale solubles Dried skimmed milk Soya bean meals Groundnut meal Cottonseed meal Sun¯ower seed meal Procedures (all samples) Gross protein value (GPV) Net protein utilization (NPU) Protein quality index (PQI) Nitrogen solubility (NS) Dye-binding capacity (Orange G) Reactive Lys (FDNB method) Total Lys (Lueconostoc mesenteriodes assay) Protein quality test by Tetrahymena pyriformis Streptococcus faecalis Streptococcus zymogens Total protein (Kjeldahl method) Procedures (selected samples) Biological value (BV) Net protein ratio (NPR) Sulfur amino acids Total free amino groups

feedstuffs. Thermal treatment eliminates antinutritional factors, bacteria, toxins, and viruses. Excessive heating can lead to a reduction in PNV. An optimal heat treatment ensures microbiological safety while minimizing the loss of PNV. Techniques for testing feed protein quality should provide results that agree with bioassay data. Proposed methods should also be quick, inexpensive, and reliable. The Agricultural Research Council (UK) organized a collaborative study of chemical methods for determining PNV of animal feeds (18). Participants were companies and research institutes involved in the production and utilization of protein concen- 388 Chapter 13 trates.* A range of chemical methods were compared with the chick and rat bioassays for protein quality. Techniques of interest had to be simple enough to be performed in the average analytical laboratory. They also had to be rapid, allowing (a) on-line monitoring of quality during feedstuff production and (b) quality assurance at storage or shipment depots.

4.1. Tests for Feed Protein Quality Table 2 lists some of the chemical methods for assessing feed protein quality examined in the ARC study. The tests were compared with a chick bioassay for gross protein value (GPV) and the rat bioassay for NPU. The test samples were protein feedstuffs commercially important for feeding nonruminant livestock. To foster cooperation from participants, the names of feed manufacturers and sources of feed materials were not recorded. About 130 samples covering seven classes of feedstuffs were investigated. Key ®ndings of this study were that (a) rankings of feed protein quality based on the GPV and NPU tests did not agree with each other except in the case of whale meat meal;(b) when all samples taken together there was no correlation between GPV and the crude protein, nitrogen solubility, or Orange G binding;and (c) no chemical method was positively correlated with feeding trials for all protein concentrates. However, available lysine showed a positive correlation with the GPV test for meat and bone meals (R 0.82), ®sh meal (R 0.96), and whale meat meal (0.94). Orange G binding was a useful index for quality, being correlated with available lysine. With oilseed protein concentrates, GPV was correlated with the nitrogen solubility index. No chemical test showed any correlation with NPU or GP. The presence of antinutritional factors (trypsin inhibitors in groundnuts and soybean, gossypol in cottonseed) may be partly to blame, as was bacterial contamination of feedstuffs. In applying PNV testing to practical situations, the ARC group referred to many issues summarized in Table 2 of Chapter 12. For example, protein meals are usually fed as compound rations containing different protein types. Livestock fed with these materials differed in terms of their variety, strain,

* The participants include J. Bibby & Sons Ltd., Liverpool (UK);Bovril Ltd., London (UK); Cros®eld & Calthrop Ltd., Liverpool (UK);Department of the Government Chemist, London (UK);Glaxo Laboratories Ltd., Greenford (UK);National Institute for Medical Research (UK);National Institute for Research in Dairying, Shin®eld (UK);Rowett Research Institute, Aberdeen (UK);Tory Research Station, Aberdeen (UK);Unilever Ltd., Shanbrook (UK); University of Cambridge School of Agriculture (UK);U.S. Department of AgricultureÐ Southern Utilization Research Division, New Orleans;Walton Oaks Experimental Station, Vitamins Ltd., Tadworth (UK). Effect of Processing on Protein Nutrient Value 389 and age;older pigs or chickens were less sensitive to low-quality protein. Chemical methods for assessing the quality of protein concentrates/feeds were also assessed by Bunyan and Price (19). The meat meal (26 batches), whale meat meal (16 batches), and ®sh meal (13 batches) samples were similar to those described for the ARC collaborative study. However, this study considered only animal-derived protein concentrates. Results for each class of protein concentrate were considered separately (Table 3). Orange G binding capacity was a useful quality indicator for most protein concentrates. FDNB-reactive lysine was not correlated with bioassay results for samples other than whale meat meal. Three meat meal samples with unusually high crude protein content (N 6 6.25) had somewhat anomalous properties. For a crude protein content > 79% the true protein value was only 21±28%. At least one of the samples appeared to be adulterated with feather meal judging from the low methionine content and high amounts of gelatin. The NPR value was 0.4±2.2 for most meat meals with a value of 1.6 signifying weight loss during the rat bioassay.

4.2. Orange G Binding Soybean and ®sh meal samples were heated in steel pans or autoclaved at 1208C (20). The DBC was positively correlated with PER determined using

TABLE 3 Summary Characteristics of Concentrates and Feedstuffs

Meat meal Whale meat meal Fish meal Sample and characteristics (n 26) (n 16) (n 13)

Crude protein (%) 40±90 25±95 60±73 NPU (%) 9±38 17±62 22±66 TDC (%) 70±90 75±88 74±90 BV (%) Ð 34±69 27±80 Total Lys (g 16 gÀ1 N) 5.4±7.8 Ð Ð Reactive Lys (g 16 gÀ1 N) 3.0±4.6 4.0±7.3 3.4±6.9 Methionine (g 16 gÀ1) 0.9±2.6 1.6±2.6 1.3±2.5 Correlations identi®ed a Dye binding (mg gÀ1)

1. Crude protein (Y1) Y1 0.278Db 30 Y1 0.261Db 30 0.325 Db 24 2. True protein (Y2) Y2 0.239Db 27 Y2 0.287D 17 Ð 3. NPU (Y3) Y3 3.86Db±0.43 Y3 6.37Db±19.8 Ð Reactive Lys (X1) None BV 7.52X1 4.0 None a Y1 crude protein, Y2 true protein, Y3 NPU, Db amount of dye bound (acid equivalents 4 of dye bound/10 grams of protein), X1 reactive Lys. 390 Chapter 13 a chick bioassay. Signi®cant decreases in DBC occurred within the ®rst 15 minutes of heat treatment. The quality of raw soybean meal was lower than expected from its DBC, probably due to the presence of protease inhibitors in the sample. The overall impression is that the measurement of DBC is a useful for monitoring heat process damage. Despite this, the DBC for heat- treated ®sh meal was high even though samples showed low levels of cysteine and low PNV;lysine does not appear to be the limiting essential amino acid in ®sh meal. The results from Choppe and Kratzer (21) also show that Orange G binding is positively correlated with protein quality results from the chick growth assay.

4.3. Acid Orange 12 Binding The thermal processing damage suffered by ®sh meal, groundnut meal, blood meal, freeze-dried beef steak, and lactalbumin was evaluated using Acid Orange 12 binding by Hurrell and Carpenter (22). The DBC was also determined for Remazol Brilliant Blue and Cresol Red. The results were compared FDNP-reactive lysine and total basic amino acids. With severely heated samples (1218C, 28 hours), Acid Orange 12 binding capacity agreed closely with the concentration of histidine arginine reactive lysine (or the HARL value). Proteins that were subjected to mild heat treatment (378C for 30 days) and/or experienced early Maillard reactions exhibited little change in DBC even after a 90% decline in available lysine. The early Maillard products were dye reactive. Carpenter and Opstvedt (23) collated results from a collaborative study of eight commercial ®sh (capelin) meals. Protein quality was evaluated from the measurement of Acid Orange 12 binding, chick bioassay and FDNB-reactive lysine. Interlaboratory variations in results were generally as large as differences between samples. Outlying results were identi®ed and those samples reanalyzed. A correlation matrix was established indicating a to strong relationship between the indices of quality. Much of the variation between samples was dependent on one sample. Omitting this sample led to the collapse of all correlation below signi®cant levels. The numbers of samples used in this trial should, perhaps, have been larger. The effect of commercial processing and laboratory heat treatment on 126 rapeseed meal samples was evaluated by Goh et al. (24±26). Heat damage was measured via Acid Orange 12 binding and as FDNB-reactive lysine. The DBC was 124.4 mg gÀ1 (sample) or 349±356 mg gÀ1 (cP) for unheated rapeseed meal. The lysine content for unheated Brassica seed protein is about 5.45 (+ 0.53) g per 100 gram (27). Autoclaving produced changes in the DBC at 45±120 minutes of heating. The concentration Effect of Processing on Protein Nutrient Value 391

FIGURE 3 Effect of autoclaving on rapeseed meal protein quality assessed from Acid Orange 12 binding, FDNB-reactive lysine, and total levels. (Drawn using data in Refs. 24±26.) of FDNB-reactive lysine declined steadily over the entire heating time (Fig. 3). The mismatch between DBC and available lysine values occurs because DBC is dependent on the total basic amino acid concentration. I have performed multiple regression analysis of DBC versus the concentration of lysine, histidine, and arginine yielding the following relations: DBC 26 +2:3 Lys 203 +11:8 1

DBC 1 +0:067 TBAA 2

DBC 17 +4:7 Lys 74 +37 His 5 +6:4 Arg 30 +80 3

These relations were generated from data in Table 5 of Ref. 26. The ranges of values used were 2.54±5.9 g (lysine) 100 gÀ1 protein, 2.43±2.74 g (histidine) 100 gÀ1 protein, and 2.98±5.65 g (arginine) 100 gÀ1 protein. The total basic amino acid range is therefore 7.95±14.3 g per 100 gÀ1 protein. The regression coef®cients for Eq. (1), Eq. (2), and Eq. (3) were 0.9201, 0.9533, and 0.9532, respectively. The nonzero intercept from Eq. (1) suggests that Acid Orange 12 binds to sites other than lysine. Eq. (2) con®rms that the DBC is strongly 392 Chapter 13 correlated with TBAA and that Acid Orange 12 binds no sites other than the three basic amino acids. Eq. (3) con®rms the previous results. The coef®cient for arginine, histidine, and lysine shows their relative contributions to DBC. Apparently, the order of dye binding to basic amino acids is histidine > lysine 4 arginine. Nevertheless, heating produced the greatest decline in lysine with histidine being the least heat susceptible. The extent of heat damage to rapeseed meal is also a function of the sample moisture content (Fig. 4). The DBC was reduced to a greater extent by heating rapeseed meal at 10±20% moisture as compared with 2% or 40% moisture. Reactions producing quality loss are limited by lack of reagent mobility in low-moisture systems and by dilution effects in high-moisture systems. Changes in available lysine show a more complex dependence on moisture levels. With prolonged heating, losses in available lysine were higher for samples having 2% moisture as compared with 40%. The assessment protein heat damage from DBC is described further by Hook (28,29), Peal et al. (30), Randall et al. (31), and also by Kratzer et al. (32). Protein samples investigated include wheat, soybean, defatted milk, whole egg, and whole blood proteins. In general, changes in DBC tended to

FIGURE 4 Effect of sample moisture content and autoclaving time on the Acid Orange 12 binding capacity for rapeseed meal. Effect of Processing on Protein Nutrient Value 393 lag behind PER. Quality loss was strongly correlated with decreases in available lysine. Lin and Lakin (33) reported disparities between the DBC and FDNB-reactive lysine for heated soy meal samples. Steaming soy meal at atmospheric pressure led to the progressive loss of the nitrogen solubility index (NSI) due to protein denaturation and insolubilization by covalent (sulfur-disul®de exchange) and noncovalent aggregation. Urease activity decreased because of enzyme inactivation after 60±80 minutes of heating. In vitro protein digestibility increased, probably due to the inactivation of soybean trypsin inhibitors. The level of unreactive lysine increased gradually from 0.14 g (lysine) gÀ1 protein and leveled off at 0.26 g (lysine) gÀ1 protein after heating of soybean meal for 120 minutes. Assuming an initial lysine content of 6.3 g per 100 g, then 95±98% of lysine residues remained available after heating. Steam treatment led to an increase in DBC, probably due to heat denaturation of soy protein.

4.4. Hot Water±Soluble Protein and Quality Meat and bone meal samples had high levels of gelatin measured as the amount of hot water±soluble protein. There was a considerable difference in the PNV for individual samples owing to their varied thermal history. For 20 different samples, Choppe and Kratzer (21) found a strong (negative) correlation between the amount of hot water±soluble bone meal gelatin and PNV. El (34) suggested a regression equation for predicting PER values for meat or ®sh based on the collagen content. Calculated PER values agreed closely with experimental values for sardine, lamb, bovine liver, chicken meat, or beef sausages. Collagen content may provide a rapid and inexpensive assay for estimating PNV for meat.

5. LEGUMES AND OILSEEDS

The structure and characteristics of legume proteins and effect of processing on their quality were reviewed by Vanderstoep (35), Chang and Satterlee (36), Sathe et al. (37), Friedman (38), and de Lumen and Uchimiya (39). Common processing operations for legumes include baking/roasting, dehulling, cooking, canning, extrusion cooking, fermentation, germination, and hydrothermal treatment. A list of some legumes for which the protein quality has been investigated over the last decade is given in Table 4. Most of these processes improve the quality of legume proteins through the reduction of protein antinutritional factors (trypsin inhibitors, hemagglutinins/lectins) and chemical antinutrients (oxalates, phytic acid, and tannins). 394 Chapter 13

TABLE 4 Determination of Protein Nutritional Value for Legumes

Commodity References

Acacia farnesiana, Cercidium Ortega-Nieblas et al. (40) microphyllium, Cercidium sonorae, Mimosa grahamii, Olneya tesota, Parkinsonia aculeata, and Prosopis juli¯ora. African locust bean Kapu et al. (41) Alfalfa protein concentrate Hernandez et al. (42) Bauhinia purpurea L Karuppanan et al. (43) Canavalia brasiliensis Oliveira et al. (44) Egyptian legumes;faba beans, lentils Youssef and Abdel-Gawad (45) (Lens culinaris), common beans (Phaseolus vulgaris), cowpea, and soybeans (Glycine max L.). Lupin varieties; Lupinus polyphyllus Aniszewski (46), Egana et al. (47) Lindl var., L. angustifolius, Lupinus albus cv. Multolupa Mung beans (Phaseolus aureus), black Khalil and Khan (48) gram, and wild beans (Vigna sublobata) Pea protein Wang et al. (49) Pigeonpea (Cajanus cajan) Singh et al. (50) Velvet bean (Mucuna pruriens L.) Permal et al. (51) Soybean varieties Vibha and Simlot (52) Sudanese legumes: lupin (Lupinus Ahmed and Nour (53) terminis), pigeon pea (Cajanus cajan), two types of cowpea (Vigna sinensis and Vigna unguiculata), bonavist bean (Dolichos lablab), faba bean (Vicia faba), and soybean

5.1. Available Lysine Reductions in legume available lysine levels in occur during domestic cooking with little nutritional consequence. Lysine is not normally the limiting essential amino acid in legumes;methionine is. Heating also produces bene®cial reductions in antinutritional factors. Available lysine levels from dhal (split pulses) of ®ve legumes (Lens esculenta, Phaseolus mungo, Lathyrus sativus, Cicer arietinum,andPisum sativum) was 5.55± 9.12 g 100 gÀ1 (54). Cooking each legume for 30 minutes produced a maximum of 25% reduction in available lysine (range 4.21±6.79 g 100 gÀ1). Effect of Processing on Protein Nutrient Value 395

Pressure cooking for 10 minutes produced a 35% reduction in available lysine (3.89±5.87 g 100 gÀ1). A similar study involving ®ve types of beans (white kidney beans, black eye beans, crab eye beans, butter beans, red kidney beans) found available lysine levels of 6.34 (+ 0.11) g 16 g NÀ1 in the raw beans (55). Pressure cooking (30 minutes at 15 lb inÀ2) produced a mean available lysine of 5.42 (+ 0.36) g 16 g NÀ1, a reduction of about 14.5% compared with the raw beans. The available lysine values were 5.48 (+ 0.48) g 16 g NÀ1 and 4.78 (+0.16) g 16 g NÀ1 after cooking for 2 and 8 hours. These changes did not produce net declines in PNV according to rat bioassays.

5.2. Steaming and Hydrothermal Treatment Subjecting whole rapeseed to steam treatment for 10 minutes inactivated the enzyme myrosinase, decomposed glucosinolates, and improved protein quality (56). The bene®ts of hydrothermal treatment are also observed with other legumes. Steam treating beans (Phaseolus vulgaris L) at 102, 119, and 1368C resulted in a loss of lectin and trypsin inhibitor activity. There was also a decline in the available lysine (57). Inactivation of trypsin inhibitor involved biphasic ®rst-order kinetics. Effects of steaming temperature on rate constants followed Arrhenius-type relations. Both total lysine and available lysine were reduced to differing extents. Steam treatment at 1198C for 5 or 10 minutes was proposed as a compromise for inactivating antinutritional factors while avoiding high degrees of protein damage as measured by total and available lysine.

5.3. Soaking, Dehulling, and Cooking Dehulling before cooking leads to greater improvements in protein quality as compared with cooking alone. The effects of dehulling, sprouting, and/or steam cooking on rice beans and mung beans were evaluated by Mehta, et al. (58). As well as the in vitro protein digestibility, they measured proximate composition, sugars, starch, and trace minerals for the raw and processed beans. The maximum protein content was 23.5% for rice beans and 26.5% for mung beans. After a combination of sprouting, dehulling, and cooking, in vitro protein digestibility increased by 35% for rice beans and 30.8% for mung beans. Soaking a range of legume [soybean, lupin, bean (Phaseolus vulgaris)] seeds in 0.5% sodium bicarbonate reduced the level of antinutritional factors (phytic acid, tannin, trypsin inhibitor, and hemagglutinin activity). Interestingly, protein extractability (in distilled water, NaCl, or sucrose solutions) was also reduced although in vitro digestibility and available lysine were improved (59). 396 Chapter 13

Seeds of perilla (Perilla frutescens Linn, Britton) have concentrations of essential amino acids above FAO/WHO/UNU recommendations for infants. Lysine was limiting. Longvah and Deosthale (60) found that dehulling and cooking perilla seeds further increased the net protein ratio (NPR), NPU, and TPD as determined by a rat bioassay although values were lower than for a casein-based control diet. Therefore, perilla seed represents a good source of protein for human and animal nutrition, particularly after dehulling and cooking. In conclusion, legume protein quality can be improved by a number of inexpensive processing methods. These studies show that weaning foods and food supplements may be produced from locally available foodstuffs.

5.4. Roasting and Irradiation Roasting and malting led to signi®cant improvements in the protein quality for a range of local legumes and cereals examined by Gupta and Sehgal (61), Dahiya and Kapoor (62), and Gahlawat and Sehgal (63). Plahar et al. (64) evaluated the effect of roasting (preceded by dehulling) on four varieties of cowpea (Vigna unguiculata). Dehulling reduced the tannin content by up to 98%. With cowpea having a highly pigmented coat, dehulling improved the protein quality. Roasting signi®cantly improved digestibility and more than doubled the PER. It was suggested that dehulled and roasted cowpea may be useful as a protein supplement in cereal-based weaning foods. The bene®cial effects of roasting have also been con®rmed with chickpeas (Cicer arietinum) and peanut (Arachis hypogaea), where cooking led to a decrease in the trypsin inhibitor activity and an increase in the true digestibility, relative nitrogen utilization, and PER values. Hira and Chopra (65) found that roasting chickpea or peanuts decreased the trypsin inhibitory activity, available lysine, and BV (P < 0.05). However, the true digestibility, relative nitrogen utilization, and PER values were all improved. Microwave heating presoaked legume seeds (faba beans, peas, chickpeas, soybeans, lentils, common beans) reduced levels of protein antinutritional factors (hemagglutinins and inhibitors) comparable with conventional cooking. In consequence, PER values were increased. Microwave processing of dried seeds had less effect. Antinutrients from common bean were destroyed by microwave heating (66). These effects are in agreement with more general ones showing that effects of both industrial and domestic microwave cooking on nutritional characteristics including proteins are comparable to those of conventional cooking (67,68). Effect of Processing on Protein Nutrient Value 397

5.5. Multiple ProcessingÐWeaning Foods Weaning foods are produced from mixtures of cereal and legumes (and occasionally milk). The multiple protein sources provide complementary supplies of essential amino acids cysteine/methionine and lysine. A recent emphasis is on the production of low-cost weaning foods using materials locally available in developing countries. Gupta and Sehgal (61), Dahiya and Kapoor (62), Gahlawat and Sehgal (63) from Haryana University (Hisar State, India) describe a number of weaning food formulations based on local cereals and legumes including wheat, pearl millet (bajra), Bengal gram, green gram (mung beans), groundnuts, peal millet, rice, kangini (Setaria italica), and sanwak (Echinochloa frumentacea). Formulations containing two or three components were generally subjected to a range of processing techniques including sprouting, roasting, and malting. For commodities such as bajra (peal millet), barley, green gram, amaranth grain (Amaranthus sp.), and jaggery, malting and/or roasting led to protein quality indices comparable to those for a commercial weaning food;PER 2.04±2.13, BV 79.56±80.68, NPU 66.75±67.86, NPR 2.13±2.76, and PRE (protein retention ef®ciency) 34.18±44.18. Dahiya and Kapoor (62) produced food supplements for preschool children using malted and/or roasted bajra, Bengal gram, green gram (mung beans), groundnuts, jaggery, or amaranth leaves. Bajra-based food supplements had quality indices (PER, food ef®ciency ratio, BV, NPU, NPR, and PER) signi®cantly higher than those of wheat-based supplements (P < 0.05). The authors suggest that the quality of their formulations was equal to that of Cerelac2, a commercial supplement. However, PNV was found to be lower (P < 0.05) than the value for casein (standard protein). Rats fed on bajra- based supplements showed an excellent growth pattern throughout the feeding trial. Santos et al. (69) prepared extruded weaning foods using a mixture of rice, mung bean, and milk (70:25:5). Protein quality was determined by a rat bioassay. The quality of the weaning food was signi®cantly improved if rice and mung bean were extruded ®rst before milk was added. The optimal PER was 2.25, which is comparable to the growth-promoting effect of casein. Extrusion cooking the complete mixture led to a PER value of 1.93. Supplementation with lysine increased the PER value to 2.10. Obviously, extrusion cooking destroyed some lysine. Fermentation and supplementation of a traditional Ghanaian cornmeal weaning food with soybean meal improved its protein quality (70). Studies also suggest that roasted cowpea, widely grown in the Sahel, may be suitable as a weaning food supplement (64,71). 398 Chapter 13

Mahgoub (72) produced ®ve weaning formulations comprising 8.5% skim milk powder and/or sorghum, groundnuts (peanuts), sesame seeds, and chickpeas (various concentrations) also along with 5% sugar and a vitamin and mineral mixture. The formulations were processed by a twin- roller drum dryer. Protein digestibility and available lysine PER, NPR, and NPU were determined for the different formulas. Formulation F3 (60% sorghum, 20% chickpeas, 5% sesame, 8.5% skim milk powder, 5% sugar, and 1.5% vitamins minerals) and F2 (55% sorghum, 15% chickpeas, 5% groundnuts, 10% sesame, 8.5% skim milk powder, 5% sugar, and 1.5% vitamins minerals) had compositions and properties comparable to those of Cerelac. In summary, cosupplementation using different protein sources is a key feature of weaning formulations. The choice and order of processing applied are also important, with some forms of processing (sprouting, fermentation, malting) sometimes able to compensate for others (autoclaving) (73).

6. CEREAL AND CEREAL PRODUCTS

Cereal products (bread, biscuits, cooked rice, noodles, pasta, etc.) are part of the staple diet in much of North and South America, North Africa, and Asia. Cereals also provide proteins indirectly when used as animal feedstuffs. We now consider the nutritional quality of cereal proteins. Bread, biscuit, and pasta making quality etc. are not discussed. The reader should refer to the following reviews for information on protein functional quality (74±76). The effect of extrusion cooking on protein quality was reviewed by Bjoerck and Asp (77), Cheftel (78), and Mercier (79). Salunkhe et al. (80) reviewed the nutritional quality of cereal proteins in general. Lorenz (81) considered the effect of sprouting on cereal protein quality. Dixon-Philip (82) discussed the consequences of milling, baking, extrusion, hydrothermal processing, and fermentation on the quality of cereal products. Examples of processing effects on PNV in cereal-based products are described here.

6.1. Amaranth Protein nutrient value for amaranth grain (Amaranthus cruentus) was determined using a nitrogen balance study with 12 adult men over three periods of 9 days each (83). There was no signi®cant difference in digestibility of popped and extruded amaranth. The nitrogen balance index (NBI) was 0.97, 0.98, and 0.96 for cheese, extruded amaranth, and popped Effect of Processing on Protein Nutrient Value 399 amaranth, respectively. These results indicate that amaranth is a good source of high-quality protein.

6.2. Maize Bressani et al. (84) examined the effect of processing maize into tortillas on their nutritional characteristics. Eleven ordinary maize cultivars and one variety of quality-protein maize (QPM) called `Nutricta' were processed into cooked maize or tortillas according to methods used in rural Guatemala. Protein quality was signi®cantly higher (P < 0.03) in tortillas than in raw maize. The QPM cultivar had superior PNV both as raw grain and as tortilla. Gupta (85) reported increased protein quality for maize after sprouting provided that radicals and plumules were removed from corn kernels. The true digestibility was unaffected by sprouting although BV, NPU, and utilizable protein increased. Gupta and Eggum (86) developed a process for transforming the by-product from corn oil production into a food-grade protein meal. Commercial oil cake was extracted with hexane and 80% ethanol and then sieved to remove undesirable materials. The defatted maize germ oil cake had 24.7% protein. Albumin, globulin, and zein decreased while glutelin and residue protein fraction increased. The meal protein had higher levels of lysine and tryptophan than whole maize grain. Protein digestibility and BV were improved as compared with the starting material.

6.3. Rice, Millet-Sorghum Eggum and Juliano (87) found that rice protein quality was not adversely affected by simple cooking or parboiling. Extrusion cooking led to adverse effects on protein quality. Lysine is the limiting essential amino acid in millet (88). The digestibility of all cultivars was high (Table 5). Autoclaving led to a 19±25% decrease in the digestibility and an overall increase in BV. Forti®cation with lysine led to greater increases in BV as compared with the effect of heating. Geervani (89) reviewed effects of processing on protein quality in millet sorghum and other cereals important for developing countries. Changes in amino acid composition and protein quality characteristics for millet (as well as barley, oats, wheat, rye, and maize) before and after boiling are discussed. Dry heat processing (e.g., as used for baking biscuits and bread), frying, fermentation (especially sorghum and millet products), and germination are also discussed. Pawar et al. (90) obtained a PER increase from 2.14 to 2.32 for pearl millet by soaking in 0.2 N HCl for 15 hours. Cooking for 20 minutes led to further improvements 400 Chapter 13

TABLE 5 Protein Quality in Millet Cultivars

Index Value

Protein (%) 10.78±17.3 Lysine (g 16 gÀ1 N) 0.995±1.39 True digestibility (%) 95±99.3 Biological value (%) 48.3±56.5

Range of samples: Italian millet (Setaria italica), French millet (Panicum miliaceum), barnyard millet (Echinochloa colona), Kodo millet (Paspalum scrobiculatum), and Little millet (P. miliare) were investigated. Source: Summarized from Ref. 88.

in protein quality. The TPD, BV, NPU, and utilizable protein were also increased for soaked and/or scari®ed pearl millet.

6.4. Storage and Insect Infestation Insect infestation produced a signi®cant decline in protein quality in wheat grain. Jood and Kapoor (91) examined the effect on wheat grains of 25, 50, and 75% infestation by mixed populations of Trogoderma granarium Everts and Rhizopertha dominica Fabricius. Changes in PNV were evaluated using the rat PER assay. With a diet containing insect-infested wheat grain (at 50 and 75% infestation) there was a decrease in food intake, body weight gain, PER, nitrogen absorption, BV, NPU, and dry matter digestibility. Virtually all protein nutrient quality parameters showed a negative association with infestation levels. Below 25% grain infestation, protein quality was not affected signi®cantly. The mechanism by which protein quality is reduced by insect infestation is uncertain. Infestation may elicit physiological changes associated with plant defense (92). Infested seed samples have increased levels of polyphenols, protease inhibitors, and other plant defense metabolites. Infestation of wheat, maize, and sorghum produced signi®cant decreases in amino acid scores for all essential amino acids. Levels of nonessential amino acids were also reduced. Large reductions were found in methionine, isoleucine, and lysine concentrations for infested wheat, maize, and sorghum grains, respectively. Insect infestation did not change the order of ®rst (lysine) and second (isoleucine) most limiting amino acids. Effect of Processing on Protein Nutrient Value 401

7. IMPROVING CEREAL PROTEIN QUALITY BY SCREENING

Much effort is being devoted to breeding high-lysine cereal varieties. Developments in this ®eld are discussed by Johnson et al. (93), Whitehouse (94), Bressani (95), Mertz (96), and also de Lumen and Uchimiya (97). The breeding programs require methods for the rapid identi®cation of high- lysine seeds. Suitable methods should enable the evaluation of hundreds of seed varieties quickly and cheaply. Many investigators used dye-binding assays to detect high-lysine cereal cultivars. A high correlation is reported between DBC and lysine concentration determined by amino acid analysis. The correlation between lysine content and DBC is signi®cantly improved by normalizing results for Kjeldahl protein (98±100). Example of breeding programs evaluated using dye binding are listed in Table 6. There is generally a negative correlation between lysine content and other grain quality indices. Grain yield, size, and crude protein content

TABLE 6 Identi®cation of High-Lysine Cereal Cultivars Using Dye Binding

Cereal References Country

Barley Doll et al. (98) Denmark Barley Bhatty and Wu (104), Canada LaBerge et al. (105), LaBerge et al. (106) Barley Lekes and Rozkosna Czechoslovakia (107) Barley Saastamoinen (108) Finland Barley Bansal et al. (101) India Barley, ®eld beans, and Gullord (99) Norway wheat Oats Young et al. (109) USA Peal millet Rabson et al. (110) USA Rice Kaul et al. (111,112) India Rice Juliano et al. (113) Philippines Rice and maize Le Thi Xuan et al. (114) Vietnam Rice Chutima et al. (115) Thailand Sorghum Jambunathan et al. (116) India Triticale Knoblauch (117) USA Wheat Mossberg (118) Sweden Wheat Sharma and Kaul (119) Germany Wheat Iqbal-Khan (120) Pakistan 402 Chapter 13 decreased as the lysine content increased (98,101). Chatterjee and Abrol (102) reported that high-lysine barley varieties were less resistant to damage. Increased lysine content is probably a consequence of the increased synthesis of water-soluble protein (albumin and globulins) at the expense of less soluble proteins (prolamins and glutelins). The soluble protein fractions have a high total basic amino acid content. This increases the avarage protein hydrophilicity, increases their DBC, and increases their electrophoretic mobility. Lawrence et al. (103) examined DBC for seven wheat protein fractions separated by polyacrylamide gel electrophoresis. PAGE lanes were stained with Amido Black 10B. Matching protein bands were excised and analyzed for their amino acid pro®le. Electro- phoretic mobility and dye-binding strength are positively correlated with the arginine and lysine content and negatively correlated with the glutamate/ glutamine levels. Total basic amino acid content varied with electrophoretic mobility from the top to the bottom of the gel: *2.5±5% (gliadins), 18±19% (albumins), and 28% (globulins). The genetic linkage between the high-lysine phenotype, impaired endosperm development, and decreased yield is not well understood. Horvatic et al. (121) noted that lysine, methionine, and tryptophan levels in nine cultivars of wheat were negatively correlated with total crude protein and with the gluten content. In high-lysine opaque (o2) maize the endosperm was ¯oury, chalky, and soft. Kernel hardness was also low, leading to reduced resistance to insect damage and low processability. Quality maize Ào2 hybrids had kernels with increased density, vitreosity, and increased grinding time. Paulis et al. (122) found a positive correlation between kernel hardness and the content of prolamins. Perhaps prolamins somehow enhance protein-starch interactions. Clore and Larkins (123) reported high levels of a protein (designated EF-1alpha) in high-lysine maize varieties. Grain lysine content was highly correlated with concentrations of protein EF-1alpha. Using immunocyto- chemistry and confocal microscopy, they showed that EF-1alpha was associated with the cytoskeletal network within the developing endosperm. The network of proteins they suggested might be necessary for the formation of protein bodies. It is feasible that increasing EF-1alpha and other lysine-rich proteins may have adverse effects on the cohesiveness of the starchy endosperm.

REFERENCES

1. MA Burnette III, II Rosof. GMA test protocol for protein quality assays. Food Technol 32(12):66±68, 1978. Effect of Processing on Protein Nutrient Value 403

2. KM Henry, SK Kon, CH Lea, JCD White. Deterioration on storage of dried milk. J Dairy Res 15:292±356, 1948. 3. CH Lea, RS Hannan. Studies of the reaction between proteins and reducing sugars in the ``dry'' state. I. The effect of activity of water, of pH and of temperature on primary reactions between casein and glucose. Biochim Biophys Acta 3:313±325, 1949. 4. CH Lea, RS Hannan. Studies of the reaction between proteins and reducing sugars in the ``dry'' state. II. Further observations on the formation of casein- glucose complex. Biochim Biophys Acta 4:518±531, 1950. 5. CH Lea, RS Hannan. Studies of the reaction between proteins and reducing sugars in the dry state. III. Nature of the protein groups reacting. Biochim Biophys Acta 5:433±454, 1950. 6. AMR van den Bruel, PJ Jenneskens, JJ Mol. Availability of lysine in skim milk powders processed under various conditions. Neth Milk Dairy J 26:19± 30, 1972. 7. M E®genia, B Povoa, T Moraes-Santos. Effect of heat treatment on the nutritional quality of milk proteins. Int Dairy J 7:609±612, 1997. 8. E Renner. Content of available lysine in heat-treated milk products. Kiel Milchwirtsch Forschungsber 35:313±314, 1983. 9. SN El, A Kavas. Available lysine in dried milk after processing. Int J Food Sci Nutr 48:109±111, 1997. 10. A Burvall, N-G Asp, A Dalhqvist, R Oste. Nutritional value of lactose- hydrolyzed milk: protein quality after some industrial processes. J Dairy Res 44:549±553, 1977. 11. N Rawson, RR Mahoney. Effect of processing and storage on the protein quality of spray-dried lactose-hydrolyzed milk powder. Lebensm Wiss Technol 16:313±316, 1983. 12. N Rawson, RR Mahoney. A modi®ed method for determination of reactive lysine in milk powder using Rezamol Brilliant Blue R. Lebensm Wiss Technol 16:1±4, 1983. 13. A Burvall, N-G Asp, A Bosson, CS Jose, A Dahlqvist. Storage of lactose- hydrolyzed dried milk: effect of water activity on the protein nutritional value. J Dairy Res 45:381±389, 1978. 14. GV Mitchell, E Grundel. Nutritional value for proteins in powdered infant formula. In vitro and in vivo methods. J Agric Food Chem 34:650±653, 1986. 15. G Sarwar, RW Pearce, HG Botting. Effect of amino acid supplementation on protein quality of soy-based infant formulas fed to rats. Plant Foods Hum Nutr 43:259±266, 1993. 16. R Nagendra, V Mahadevamma, V Baskaran, S Venkat-Rao. Shelf-life of spray-dried infant formula supplemented with lactulose. J Food Process Preserv 19:303±315, 1995. 17. E Ferrer, A Alegria, R Farre, P Abellan, F Romero. Effects of thermal processing and storage on available lysine and furfural compounds contents of infant formulas. J Agric Food Chem 48:1817±1822, 2000. 404 Chapter 13

18. AW Boyne, JJ Carpenter, AA Woodham. Progress report on an assessment of laboratory procedures suggested as indicators of protein quality in feedingstuffs. J Sci Food Agric 12:832±848, 1961. 19. J Bunyan, SA Price. Studies on protein concentrates for animal feeding. J Sci Food Agric 11:25±37, 1960. 20. ET Moran Jr, LS Jensen, J McGinnis. Dye binding by soybean and ®shmeal as an index of quality. J Nutr 79:239±244, 1963. 21. W Choppe, FH Kratzer. Methods for evaluating the feeding quality of meat- and-bone meals. Poultry Sci 42:642±646, 1963. 22. RF Hurrell, KJ Carpenter. The use of three dye-binding procedures for the assessment of heat damage to food proteins. Br J Nutr 33:101±115, 1975. 23. KJ Carpenter, J Opstvedt. Applications of chemical and biological assay procedures for lysine to ®shmeals. J Agric Food Chem 24:389±393, 1976. 24. YK Goh, DR Clandinin, AR Robblee. Application of the dye-binding technique for quantitative and qualitative estimation of rapeseed meal protein. Can J Anim Sci 59:181±188, 1979. 25. YK Goh, DR Clandinin, AR Robblee. The application of the dye-binding method for measuring protein denaturation of rapeseed meals caused by autoclave or oven heat treatments for varying periods of time. Can J Anim Sci 59:189±194, 1979. 26. YK Goh, AR Clandinin, AR Robblee. Protein quality evaluation of commercial and laboratory heat-damaged rapeseed meals by the dye-binding technique and by biological assay with chicks. Can J Anim Sci 59:195±201, 1979. 27. FW Sosulski. Rapeseed protein for food use. In: BJF Hudson, ed. Developments in Food ProteinsÐ2. London: Applied Science Publishers, 1983, pp 109±132. 28. SCW Hook. A test for heat-damaged wheat using the RAPID-meter. Milling Feed Fertilizer 162:20±21, 29, 1979. 29. SCW Hook. Dye-binding capacity as a sensitive index for the thermal denaturation of wheat protein. A test for heat-damaged wheat. J Sci Food Agric 31:67±81, 1980. 30. IM Peal, MP Szakacs, A Kowvago, J Petroczy. Stoichiometric dye-binding procedure for the determination of the reactive lysine content of soya bean protein. Food Chem 16:163±174, 1985. 31. PG Randall, L Stone, R Jacobs, AEJ McGill. Detection and quanti®cation of heat damage in wheat. Food Rev 15(Suppl 2):45±47, 1988. 32. FH Kratzer, S Bersch, P Vohra. Evaluation of heat-damage to protein by Coomassie Blue G dye binding. J Food Sci 55:805±807, 1990. 33. S Lin, AL Lakin. Thermal denaturation of soy proteins as related to their dye- binding characteristics and functionality. J Am Oil Chem Soc 67:872±878, 1990. 34. SN El. Evaluating protein quality of meats using collagen content. Food Chem 53:209±210, 1995. Effect of Processing on Protein Nutrient Value 405

35. J Vanderstoep. Effect of germination on the nutritive value of legumes. Food Technol 35(3):83±85, 1981. 36. KC Chang, LD Satterlee. Chemistry of bean proteins. J Food Process Preserv 6:203±225, 1982. 37. SK Sathe, SS Deshpande, DK Salunkhe. Dry beans of Phaseolus. A review. I. Chemical composition: proteins. CRC Crit Rev Food Sci Nutr 20:1±46, 1984. 38. M Friedman. Nutritional value of proteins from different food sources. A review. J Agric Food Chem 44:6±29, 1996. 39. BO de Lumen, H Uchimiya. Molecular strategies to enhance the nutritional quality of legume protein: an update. AgBiotech News Info 9(3):53N±58N, 1997. 40. M Orgega-Nieblas, L Vazquez-Moreno, MR Robles-Burgueno. Protein quality and antinutritional factors of wild legume seeds from the Sonoran desert. J Agric Food Chem 44:3130±3132, 1996. 41. MM Kapu, KM Shehu, RAI Ega, HO Akanya, GO Obodo, DJ Schaeffer. Protein quality of tamarind and African locust bean seed meals. Lebensm Wiss Technol 23:260±261, 1990. 42. T Hernandez, C Martinez, A Hernandez, G Urbano. Protein quality of alfalfa protein concentrates obtained by freezing. J Agric Food Chem 45:797±802, 1997. 43. V Karuppanan, S Perumal, J Karnam. Chemical composition, amino acid content and protein quality of the little-known legume Bauhinia purpurea L. J Sci Food Agric 73:279±286, 1997. 44. JTA Oliveira, IM Vasconcelos, MJL Gondim, BS. Cavada, RA Moreira, CF Santos, LIM Moreira. J Sci Food Agric 64:417±424, 1994. 45. MKE Youssef, AS Abdel-Gawad. Protein quality and trypsin inhibitors in some common Egyptian legume seeds. Assiut J Agric Sci 23:3±18, 1992. 46. T Aniszewski. Nutritive quality of the alkaloid-poor Washington lupin (Lupinus polyphyllus Lindl var. SF/TA) as a potential protein crop. J Sci Food Agric 61:409±421, 1993. 47. JI Egana, R Uauy, X Cassorla, G Berrera, E Yanez. Sweet lupin protein quality in young men. J Nutr 122:2341±2347, 1993. 48. IA Khalil, S Khan. Protein quality of Asian beans and their wild progenitor, Vigna sublobata (Roxb). Food Chem 52:327±330, 1995. 49. N Wang, PR Bhirud, RT Tyler. Extrusion texturization of air-classi®ed pea protein. J Food Sci 64:509±513, 1999. 50. U Singh, R Jambunathan, K Sexana, N Subrahmanyam. Nutritional quality evaluation of newly developed high-protein genotypes of pigeonpea (Cajanus cajan). J Sci Food Agric 50:201±209, 1990. 51. S Permal, V Karuppanan, J Karnam. Chemical composition and protein quality of the little-known legume, velvet bean (Mucuna pruriens L. DC.). J Agric Food Chem 44:2636±2641, 1996. 52. D Vibha, MM Simlot. Effect of trypsin inhibitor on protein quality of black- soybean and mothbean meals. J Food Sci Technol India 34:208±211, 1997. 406 Chapter 13

53. AH R Ahmed, AAAM Nour. Protein quality of common Sudanese leguminous seeds. Lebensm Wiss Technol 23:301±304, 1990. 54. S Datta, SC Datta. Available lysine in cooked pulses. Indian J Nutr Diet 15:128±130, 1978. 55. AE Bender, H Mohammadiha, K Almas. Digestibility of legumes and available lysine content. Qual Plant Plant Foods Hum Nutr 29:219±226, 1979. 56. H Kozlowska, M Piskula, D Rotkiewicz. Steaming whole rapeseeds to improve protein and oil quality. In: Proceedings of the World Conference on Oilseed Technology and Utilization. American Oil Chemists Society, Champaign, Ill pp 458±460, 1993. 57. TFB Van der Poel, J Blonk, DJ van Zuilichem, MG van Oort. Thermal inactivation of lectins and trypsin inhibitor activity during steam processing of dry beans (Phaseolus vulgaris) and effects on protein quality. J Sci Food Agric 53:215±228, 1990. 58. U Mehta, P Verma, I Singh. Effect of cooking, sprouting and dehulling on the nutritional quality and protein digestibility of rice bean (RBL-1) Vigna umbellata in comparison to mung bean (Phaseolus aureus). J Dairy Food Home Sci 12:165±172, 1993. 59. TA El-Adawy, EH Rahma, AA El-Bedawy, TY Sobihah. Effect of soaking process on nutritional quality and protein solubility of some legume seeds. Nahrung 44:339±343, 2000. 60. T Longvah, YG Deosthale. Effect of de-hulling, cooking and roasting on the protein quality of Perilla frutescens seed. Food Chem 63: 519±523, 1998. 61. C Gupta, S Sehgal. Protein quality of developed homemade weaning foods. Plant Foods Hum Nutr 42:239±246, 1992. 62. S Dahiya, AC Kapoor. Biological evaluation of protein quality of home- processed supplementary foods for pre-school children. Food Chem 48:183± 188, 1983. 63. P Gahlawat, S Sehgal. Protein quality of weaning foods based on locally available cereal and pulse combination. Plant Foods Hum Nutr 46:245±253, 1994. 64. WH Plahar, NT Annan, CA Nti. Cultivar and processing effects on the pasting characteristics, tannin content and protein quality and digestibility of cowpea (Vigna unguiculata). Plants Foods Hum Nutr 51:343±356, 1997. 65. CK Hira, N Chopra. Effects of roasting on protein quality of chickpea (Cicer arietinum) and peanut (Arachis hypogaea). J Food Sci Technol India 32:501± 503, 1995. 66. M Hernandez-Infante, V Saousa, I Motalvo, E Tena. Impact of microwave heating on hemagglutinins, trypsin inhibitors and protein quality of selected legume seeds. Plant Foods Hum Nutr 52:199±208, 1998. 67. K Lorenz. Microwave heating of foodsÐchanges in nutrient and chemical composition. CRC Crit Rev Food Sci Nutr 7:339±370, 1976. 68. GA Cross, DYC Fung. A review of the effects of microwave cooking on foods. J Environ Health 44(4):188±193, 1982. Effect of Processing on Protein Nutrient Value 407

69. RV Santos, MA Udarbe, CC Marcado, JM Gonzales. Effects of extrusion on the protein quality of a milk supplemented rice-mungbean weaning food. ASEAN Food J 8:61±65, 1993. 70. K Addo, S Lykins, C Cotton. Indigenous fermentation and soy forti®cation: effects on protein quality and carbohydrate digestibility of a traditional Ghanaian corn meal. Food Chem 57:377±380, 1996. 71. CA Nti, WA Plahar. Chemical and biological characteristics of a West African weaning food supplemented with cowpea (Vigna unguiculata). Plant Foods Hum Nutr 48(1):45±54, 1995. 72. SEO Mahgoub. Production and evaluation of weaning foods based on sorghum and legumes. Plant Foods Hum Nutr 54(1):29±42, 1999. 73. S Mbithi-Mwikya, W Ooghe, J van Camp, D Ngundi, A Huyghebaert. Amino acid pro®les after sprouting, autoclaving, and lactic acid fermentation of ®nger millet (Eleusine coracan) and kidney beans (Phaseolus vulgaris L.). J Agric Food Chem 48:3081±3085, 2000. 74. NE Pogna, P Tusa, G Boggini. Genetic and biochemical aspects of dough quality in wheat. Adv Food Sci 18:145±151, 1996. 75. CY Lui, KW Sheperd, AJ Rathjen. Improvement of durum wheat pasta making and bread making qualities. Cereal Chem 73:155±166, 1996. 76. SK Bhupendar, JD Scho®eld. Molecular and physico-chemical basis of bread makingÐproperties of wheat gluten proteins: a critical appraisal. J Food Sci Technol India 34:85±102, 1997. 77. I Bjoerck, NG Asp. The effects of extrusion cooking on nutritional valueÐa literature review. J Food Eng 2:281±308, 1983. 78. JC Cheftel. Nutritional effects of extrusion cooking. Food Chem 20:263±283, 1986. 79. C Mercier. Nutritional appraisal of extruded foods. Int J Food Sci Nutr 44(Suppl 1):S45±S53, 1993. 80. DK Salunkhe, SS Kadam, JK Chavan. Nutritional quality of proteins in grain sorghum. Qual Plant Plant Foods Hum Nutr 27:187±205, 1997. 81. K Lorenz. Cereal sprouts: composition, nutritive value, food applications. CRC Crit Rev Food Sci Nutr 14:353±385, 1980. 82. R Dixon-Philip. Nutritional quality of cereal and legume storage proteins. Food Technol 51:62, 64±66, 1997. 83. R Bressani, ECM de Martell, CM de Godinez. Protein quality evaluation of amaranth in adult humans. Plant Foods Hum Nutr 43:123±143, 1990. 84. R Bressani, V Benavides, E Acevedo, MA Ortiz. Changes in selected nutrient contents and in protein quality of common and quality-protein maize during rural tortilla preparation. Cereal Chem 67:515±518, 1994. 85. HO Gupta. Protein quality evaluation of sprouted maize. Plant Foods Hum Nutr 46:85±91, 1994. 86. HO Gupta, BO Eggum. Processing of maize germ oil cake into edible food grade meal and evaluation of its protein quality. Plant Foods Hum Nutr 52:1± 8, 1998. 408 Chapter 13

87. BO Eggum, BO Juliano. Properties and protein quality in growing rats of a low-glutelin content rice mutant. Cereal Chem 74:200±201, 1997. 88. P Geervani, BO Eggum. Nutrient composition and protein quality of minor millets. Plant Foods Hum Nutr 39:201±208, 1989. 89. P Geervani. The in¯uence of home processing on the quality of cereal and millet proteins. In: International Association of Cereal Chemistry, ed. Amino Acid Composition and Biological Value of Cereal Proteins. 1985, pp 495±519. 90. VD Pawar, MV Khandagale, NF Quadri. Assessment of protein quality of depigmented pearl millet. J Food Sci Technol India 27:109±110, 1990. 91. S Jood, AC Kapoor. Biological evaluation of protein quality of wheat as affected by insect infestation. Food Chem 45:169±174, 1992. 92. S Jood, AC Kapoor, R Singh. Amino acid composition and chemical evaluation of protein quality of cereals as affected by insect infestation. Plant Foods Hum Nutr 48:159±167, 1995. 93. VA Johnson, PJ Mattern, JW Schmidt. The breeding of wheat and maize with improved nutritional value. Proc Nutr Soc 29:20±31, 1970. 94. RNH Whitehouse. The prospects of breeding barley, wheat and oats to meet special requirements in human and animal nutrition. Proc Nutr Soc 29:31±39, 1970. 95. R Bressani. Protein quality of high-lysine maize for humans. Cereal Foods World 36:806±811, 1991. 96. ET Mertz, ed. Quality Protein Maize. St. Paul, MN: American Association of Cereal Chemists, 1995. 97. BO de Lumen, H Uchimiya. Molecular strategies to enhance the nutritional quality of legume protein: an update. AgBiotech News Info 9:53N±58N, 1997. 98. H Doll, B Koie, BO Eggum. Induced high lysine mutants in barley. Radiat Bot 14:73±80, 1974. 99. M Gullord. The dye binding capacity method (DBC-method) for determination of basic amino acids in protein. Meld Nor Landbrukshogsk 53, 21 pp. 100. TR Sharma, AK Kaul. Rationale of using dye-binding capacity (DBC) for the evaluation of protein content and quality in segregating lines of wheat. Z P¯anz 76:204±214, 1976. 101. HC Bansal, RR Singh, S Bhaskaran, IM Santha, BR Murty. Hybridization and selection for improving seed protein in barley. Theor Appl Genet 58:129± 136, 1980. 102. SR Chatterjee, YP Abrol. Protein quality evaluation of popped barley grains (sattu). J Food Sci Technol India 14:247±250, 1977. 103. JM Lawrence, SC Liu, DR Grant. Dye-binding capacity and amino acid content of wheat-protein gel-electrophoresis bands. Cereal Chem 47:110±117, 1970. 104. RS Bhatty, KK Wu. Lysine screening in barley with a modi®ed Udy dye- binding method. Can J Plant Sci 55:685±689, 1975. 105. De LaBerge, DR Metcalfe, R Tkachuk. Comparison of methods for lysine screening in barley. Can J Plant Sci 56:25±30, 1976. Effect of Processing on Protein Nutrient Value 409

106. DE LaBerge, AW MacGregor, DR Metcalfe. Screening for high-lysine cultivars in a barley breeding program. Can J Plant Sci 56:817±821, 1976. 107. J Lekes, A Rozkosna. A contribution to the breeding of productive spring barley varieties with higher biological value of the protein. Z P¯anzenzucht 74:199±210, 1975. 108. M Saastamonien. On the DBC protein content and on the amino acid contents in F5 lines of the barley line Hiproly. J Sci Agric Soc Finl 51:40±50, 1979. 109. VL Young, KD Gilchrist, DM Peterson. Protein the current emphasis in oat quality. Cereal Sci Today 18:409±411, 1973. 110. R Rabson, WW Habbam, H Axman. Potential for improving the protein content of pearl millet grain using induced mutation. In: Seed Protein Improvement in Cereals and Grain Legumes, Vol 11. Vienna: International Atomic Agency, 1979, pp 367±376. 111. AK Kaul, RD Dhar, P Raghaviah. The macro and micro dye-binding techniques of estimating the protein quality in food samples. J Food Sci Technol Mysore 7:11±16, 1970. 112. AK Kaul, RD Dhar, MS Swaminathan. Microscopic and other dye binding techniques of screening for proteins in cereals. Plant Foods Hum Nutr 2:113± 117, 1972. 113. BO Juliano, AA Antonio, BV Esmama, Effects of protein content on the distribution and properties of rice proteins. J Sci Food Agric 24:295±306, 1973. 114. Le Thi Xuan, KC Nguyen, HT Nguyen, VU Nguyen. Use of the dye binding method (DBC) for estimating protein and lysine content in rice and maize. Acta Agron Acad Sci Hung 25:391±394, 1976. 115. K Chutima, BR Jackson, S Duangratana, R Boonduang, N Kongseree, K Suwantaradon. Results of multi-location tests over several years for yield and seed protein content of indigenous. Thai rice varieties. In: Seed Protein Improvement in Cereals and Grain Legumes, Vol 11. Vienna: International Atomic Agency, 1979, pp 279±291. 116. R Jambunathan, NS Rao, S Gurtu. Rapid methods for estimating protein and lysine in sorghum (Sorghum bicolor (L.) Moench). Cereal Chem 60:192±194, 1983. 117. CJ Knoblauch. Early generation selection in triticale for factors contributing to protein quality. Dissertation Absracts International B;36 (1) 31: Order no. 75-14769. 118. R Mossberg. Practical problems concerning the marketing of cereals with improved protein value. Proc Nutr Soc 29:39±48, 1970. 119. TR Sharma, AK Kaul. Correction between some quality characters and protein determinants in bread wheat. Curr Sci 40:150±152, 1971. 120. M Iqbal-Khan. Lysine estimation with the modi®ed Udy-dye binding method in hexaploid wheat. Experientia 34:711±712, 1978. 121. M Horvatic, M Gruener, N Mulalic. Relationship between protein nutritive quality and technological property parameters of wheat ¯our. Nahrung 33:901±903, 1989. 410 Chapter 13

122. JW Paulis, AJ Peplinski, JA Bietz, TC Nelsen, RR Berquist. Relation of kernel hardness and lysine to alcohol-soluble protein composition in quality protein maize hybrids. J Agric Food Chem 41:2249±2253, 1993. 123. AM Clore, BA Larkins. Protein quality and its potential relationship to the cytoskeleton in maize endosperm. J Plant Physiol 152:630±635, 1998. 14

Protein Digestibility±Corrected Amino Acid Scores

1. INTRODUCTION

The protein digestibility±corrected amino acid score (PDCAAS) was adopted as the AOAC-approved index for protein quality in 1993±1994. The PDCAAS is the amino acid score (AAS) multiplied by protein digestibility. Alterations in either parameter changes protein nutrient value (PNV). PDCAAS is discussed in this chapter. In Sec. 2 considration is given to digestibility and its relation to protein structure. In Secs 3 and 4 is a review of protein denaturation and chemical deterioration during food processing and their effect on the PDCAAS. There is also increasing realization that moisture-temperature-time relations affect the food matrix and protein quality. Some possible links between Tg (glass transition temperature) and PNV are explored in Sec. 5. The determination of PDCAAS for a range of foods is discussed in Sec. 6.

2. PROTEIN DIGESTIBILITY

True protein digestibility (TPD) measured using a rat bioassay agrees with in vitro protein digestibility (Chapter 13). This results has two implications:

411 412 Chapter 14

1. Absorption across the intestinal membrane does not limit (bio)availability. Exceptions to this rule appear in Section 6.7, where availability-corrected amino acid scores (AvCAAS) are discussed. 2. Fundamental research on protein structure and susceptibility to proteolysis can be marshaled to help make sense of the PDCAAS.

2.1. Protein Conformation, Molten Globules, and Digestibility Binding interactions between a protease and its substrate follow the approach of two molecules of comparable sizes. An exposed region on the substrate protein then enters the protease active site. Linderstrom-Lang et al. (1) described enymatic digestion via a two-stage reaction. First, the native protein (N) conformation changes to an unfolded (U) state that exposes peptide bonds to the external solvent [Eq. (1)]. N U 1

Protein unfolding increases within the stomach. The low-pH, high-chloride environment transforms globular proteins into a molten globule state. This is an expanded, more ¯exible structure compared with the N state (2,3). At low pH, ®brous proteins (myosin, collagen) exhibit changes in protein- protein interactions followed by unfolding. Second, the U state is hydrolyzed by an enzymatic reaction that leads to the irreversibly modi®ed (I) state [Eq. (2)]. U ! I 2

The rate of proteolysis (v) is determined by the fraction of protein substrates unfolded (FU) and the rate constant for the U ! I reaction (ki):

kiPKeq v & kiPFU 3 1 Keq where P is the protein concentration. The preceding relation applies to the whole protein or to speci®c segments.*

* Protein unfolding is often described as a two-state (all or nothing) process. In this idealized model all peptide bonds are either inaccessible or fully accessible to the external solvent. In contrast to this global (un)folding process, a non±two-state unfolding event leads to the localized exposure of bonds. Corrected Amino Acid Scores 413

2.2. Protein Structure, Stability, and Digestibility In vitro digestibility is a sensitive probe for protein structure. Under nondenaturing conditions Keq 5 1 and FU&Keq [see Eq. (3)]. The value of Keq is ®nitely large or small but never ``zero.'' Proteases attack the U state readily and therefore slight changes in the N/U equilibrium [Eq. (1)] alter protease susceptibility. Proteolytic attack on the N state conformation occurs at surfaces, exposed peptide loops, beta-turns, and at random (aperiodic) sequences.* Digestion is more likely between protein domains. Oligomeric proteins are digested following dissociation into subunits. The rate of digestion is low within areas of regular secondary structure. Protein stability is inversely related to digestibility (4). Matthyssens et al. (5) monitored the thermal denaturation of lysozyme at pH 2 by measuring the rate of proteolysis by pepsin. With increasing temperatures, lysozyme unfolds and the rate of proteolysis increases. Imoto et al. (6,7) monitored the thermal unfolding transition for lysozyme from the kinetics of proteolysis with pronase. Church et al. (8) used immobilized pronase activity as a probe for normal or chemically modi®ed lysozyme, beta- lactoglobulin, and casein. Daniel et al. (9) established an inverse correlation between protein stability and susceptibility to proteolysis. Ueno and Harrington (10) monitored structural changes in myosin at 5±408C using papain, chymotrypsin, or trypsin. A relationship between protein digestibility, ¯exibility, surface hydrophobicity, and physical functional properties (foaming, emulsi®cation) was proposed by Kato et al. (11). Kato et al. (12) reported a similar relation for heat-denatured lysozyme and ovalbumin. The sites of autolysis for thermolysin had high segmental mobility as detected by temperature factors (B values) from X-ray crystallography (13). Unstable proteins for which the N/U equilibrium lies toward the U state are more susceptible to proteolytic degradation. Acid-stable proteins (e.g., beta- lactoglobulin) are resistant to unfolding and hence to protease attack. The relationship between protein stability and digestibility is altered within the stomach. The majority of food proteins become destabilized at low pH at 378C. Using the ¯uorescent dye anilino-naphthalene-8-sulfonic acid (ANS). Folawiyo and Apenten (14) showed that oilseed proteins undergo a conformational change at pH 1±2. The degree of ANS binding at low pH was 10-fold greater than obtained by heating alone. Preheating had little effect on ANS binding sites at low pH.

* Under nondenaturing conditions, the small proportion of the U state in equilibrium with the native state is attacked by proteases. It is a moot point whether the N state itself undergoes proteolytic attack. 414 Chapter 14

2.3. Enzyme Speci®city The rate of proteolysis is also determined by protease speci®city. Schechter and Berger (15) supposed that proteases have up to seven active sites (. . . S3, S2, S1, S10,S20,S30) with speci®city for six peptidyl sites (P3, P2, P1, P10,P20, P30). The peptide bond subjected to cleavage occurs between P1 and P10. Different proteases recognize different P1, P10 residues for bond hydrolysis. For trypsin, chymotrypsin, and pepsin, P1 lysine or arginine, an aromatic amino acid, or a hydrophobie residue, respectively. The effect of enzyme speci®city on the rate of proteolysis appears in the ki term of Eq. (3). In accordance with Michaelis-Menten formalism, ki Vmax=Km (ratio of the maximum velocity to the Michaelis constant).

3. PROTEIN DENATURATION

Changes in protein structure affect digestibility and PNV. The important variables are temperature, pressure, pH, ionic strength, presence of surfaces, and shear rate (Table 1). Changes in protein structure follow a two-stage process [Eq. (1)±Eq. (2)]. Unfolding is either a global process or is restricted to speci®c segments of the protein. Exposure to chemical denaturing agents (urea, guanidine hydrochloride) leads to a U state that is approximately a random coil. Extremes of pH, extremes of temperature, and/or high pressure produce the partially unfolded (molten globule) state (Table 1, row C). The N/U transition [Eq. (1)] is formally reversible. Removing the protein from stress restores the N state. Irreversible U ! I reactions occur after prolonged exposure to extreme conditions [Eq. (2)]. Besides Linderstrom- Lang and co-workers (1), Equations (1) and (2) were also developed by Eyring and Lumry (16), who applied them to describe protein denaturation. Proteolysis is an irreversible denaturation process where the U ! I reaction is peptide bond hydrolysis catalyzed by a protease. Nonenzymatic proteolysis occurs at low pH and at temperatures of about 1008C. Structural changes of practical signi®cance are irreversible. Charles Tanford's treatment of this subject is still unsurpassed after 30 years (17). Ahern and Klibanov (18,19) have also discussed denaturation at length. Kinsella (20) reviewed the chemistry of food protein reactions. On the relatively long time scale associated with aging and development, there is a constant leakage from the U state to the I state. The postharvest state is characterized by increased formation of the I state. Table 1 (row B) lists factors that are likely to support reversible or Corrected Amino Acid Scores 415

TABLE 1 Forms of Food Protein Denaturationa

A. Reversible denaturation A*. Irreversible denaturation

N UN! I B. Conditions for the N U transition B*. Conditions for N ! I transition Low protein concentration High protein concentration (<0.1% w/v) (>5% w/v) pH removed from pI Solvent pH&pI Presence of sulfhydryl reagents Extremes of pH (pH < 4, pH > 8) Transient exposure to denaturing Prolonged exposure to denaturant conditions Low intensity of denaturant High intensity of denaturant/severe process conditions Low ionic strength High ionic strengh C. Some N U processes C0. Some N ! I processes Transient unfolding, expansion Covalent aggregation Semimolten globule Noncovalent aggregaration Molten globule Peptide bond lysis Domain denatured Racemization Cross-link formation Carbonylamine reactions Deamidation D. Processing variables High temperatures (sterilization, cooking, drying) Low temperature (cold storage, freezing) Moisture control (dehydrated storage, drying) High pressure (sterilization, low-temperature gelation) High or low pH (solubilization, texturization) Procesing chemicals (alcohols, alkali, nitrate, reducing compounds, etc.) Exposure to interfaces (emul®cation, foaming, etc.) High shear treatment aNotes: A, Two forms of protein denaturation;B, conditions that facilitate reversible and irreversible denaturation;C, types of reversible and irreversible denaturation;D, forms of food processes leading to denaturation.

irreversible denaturation. Irreversible changes in protein conformation arise from covalent aggregation via sulfhydryl-disul®de exchange or sulfhydryl oxidation. Physical denaturation processes such as gelation will be reversed in the low-pH gastric environment. 416 Chapter 14

4. CHEMICAL DETERIORATION OF PROTEIN INGREDIENTS

Denaturation was initially de®ned only in operational terms by reference to the changing physical appearances of proteins. Protein precipitation, aggregation, gelation, or increases in viscosity were taken as signs of denaturation. These changes in appearance can be induced (by salting-out, exposure to organic solvents) without denaturation. After the publication of the 3D structure for myolglobin in 1963, denaturation was rede®ned in terms of changes in the 28,38,or48 structure. Alterations in protein 18 structure were not included in the de®nition of denaturation. It seemed unlikely that changes in 18 structure could occur under normal physiological conditions (atmospheric pressure, T 0±37 C). There is now much sympathy for the idea that proteins from archaebacteria evolved at temperatures of 80±1208C. Many organisms persist under ``extreme'' physiological conditions of temperature and pressure where changes in protein 18 structure are conceivable. The conventional de®nition of denaturation is also too restricting in view of recent advances in protein chemistry and technology. Proteins exhibit 18 structure changes during food processing. The range of deteriorative changes includes peptide bond hydrolysis, carbonyl-amine reactions, racemization to form D-amino acids, and formation of covalent cross-links. Deamidation of glutamine and asparagine transforms these residues into glutamate and aspartate, respectively. The SH-disul®de exchange produces cystine cross-links from cysteine. Loss of hydrogen sul®de and dehydration generate dehydroalanine. Heating in the presence of sugars leads to glycation. Most deteriorative processes are accelerated by water. Addition of water to intermediate moisture foods encourages deterioration by lowering the glass transition temperature (Tg). With dissolved proteins, deterioration follows protein unfolding to expose reactive amino acid residues to the external solvent.

4.1. Carbonyl-Amine Reactions Within a polypeptide all except three essential amino acids are protected from the Maillard reaction. Luise Maillard (21) discovered the reactions between amino acids and sugars. Nonenzymatic browning leads to the formation of melanoidins and ¯avor precursors. The reactive protein residues are lysine (e-NH2), methionine (22S22CH3), and tryptophan (indole group). With high-carbohydrate (plant) foods, lysine is the most reactive essential amino acid and the most limiting. Methionine or tryptophan is usually limiting for animal proteins. Corrected Amino Acid Scores 417

Carbonyl-amine reactions take place in four stages: (a) addition of the e-NH2 group of lysine to a carbonyl group of an aldose, (b) elimination of water to form a cationic Schiff base, (c) proton loss to form an enol, and (d) enol-keto rearrangement to form 1-amino-1-deoxy-2-ketose. Reactions (c) and (d) together constitute the Amadori rearrangement. For a keto-sugar the corresponding reaction is the Heyens reaction leading to 2-amino-2- deoxyaldose. The effect of carbonyl-amine reactions on protein nutrient value (PNV) is reviewed by Dworschalk (22), Labuza and Saltmarch (23), Friedman (24,25), Saltmarch and Labuza (26), Feeney and Whitaker (27), Hurrell (28), and Feather (29).

4.2. Controlling Factors for Carbonyl-Amine Reactions

The effect of pH, water activity (Aw), temperature, and different sugars on the rate of the carbonyl-amine reaction was studied by Labuza and Saltmarch (23,26). Maillard browning shows a bell-shaped pH dependence with a maximum at about pH 8. Extremes of pH lead to a decline in the rate of browning by (de)protonating the carbonyl and amino functions. The dependence on AW is also bell shaped. Browning decreases with decreasing moisture content due to the high medium viscosity and diffusion restrictions. With increasing AW the reaction rate increases until suf®cient water is available for monolayer coverage. With still higher moisture content the rate of reaction decreases due to the dilution of reactants. The temperature dependence of the Maillard reaction was a described with linear Arrhenius equation. The activation energy for lysine loss was about 80 kJ molÀ1 (Section 5.3). The relative rates of reaction with different sugars follow the order lactose > D-ribose > D-fructose > D-glucose;pentose > hexose > disaccharide.

4.3. Impact on Protein Quality The products of the carbonyl-amine reactions have two principal effects on PNV: (a) the bioavailability of lysine is reduced and (b) digestibility is reduced by the presence of protein-bound sugar residues. The Amadori compound (1-amino-1-deoxy-2-ketose) is biologically unavailable (30). After ingestion by rats, a signi®cant proportion is excreted in the urine. The nonabsorbed fraction appears in the feces and/or is degraded by intestinal bacteria. Peptide bonds in the vicinity of glycated lysine residues are not susceptible to protease attack. Maillard reaction products also inhibit the absorption of other amino acids. Suggested physiological effects include increased protein allergenicity, mutagenicity, and effects on the reproductive capacity of rats. Normal cooking does not affect PNV. 418 Chapter 14

Microwave cooking also had no adverse effects (31,32). Maillard reaction products form during baking that have a detrimental effect on protein quality. The effects of the Maillard reaction on available lysine is also discussed in Chapter 12 and 13.

4.4. Racemization and Cross-Linking Cysteine, serine, and phosphothreonine residues are converted to dehydroalanine at high pH. The process involves the elimination of hydrogen sul®de, water, or the phosphate group, respectively. Dehydroalanine then reacts with a range of protein groups forming cross-links (Fig. 1). Racemization also takes place under alkaline conditions (33). The two reactions proceed via the following common steps: (a) abstraction of a proton from an a-C atom by a base (B) produces a carbanion ion, (b) readdition of H to the a-C then generates the corresponding D-amino acid, (c) b-elimination from the carbanion ion produces dehydroalanine, and (d) addition of protein side chains (histidine, cysteine, lysine) to dehydroala- ninie forms cross-links. The subject is reviewed by Hurrell (28), Maga (34), Otterburn (35), and Swaisgood and Catignani (36).

4.5. Factors Controlling the Rate of Racemization and Cross-Linking Friedman and Liardon (37) found that the rate of racemization is sensitive to ( )or(À) inductive effects. For simple amino acids the rate of racemization was predicted using linear free energy relationships. Free amino acids are *10-fold less reactive than protein-bound amino acids. The presence of an adjacent carboxyl group destabilizes the intermediate carbanion ion. A high pH (>9) increases the rate of racemization by facilitating the initial proton abstraction. Proline residues (especially when next to aspartate or aparagine) are more reactive than other amino acid residues. Polypeptide sequences containing serine and cysteine (e.g., Ser/ Cys22X22Lys or Ser/Cys22X22X) are also highly reactive. Racemization and cross-linking can be observed by heating model proteins in alkaline solution (38,39). Similar reactions take place in foods when proteins are exposed to high pH and temperature. A traditional method for producing tortilla involves steeping corn in 1% lime solution at 808C for 20±45 minutes. The preparation of ®sh protein concentrates, texturization of soy protein, recovery of residual protein from bone, alkaline extraction of plant or yeast proteins, and lye peeling also exposes food proteins to high pH values. Cross-linking and racemization reduce in vitro protein digestibility (40,41) as well as PNV (42). Corrected Amino Acid Scores 419

FIGURE 1 A scheme for amino acid racemization, dehydroalanine formation, and protein cross-linking. Based on a scheme proposed by Masters and Friedman (33). B base. 420 Chapter 14

5. MATRIX EFFECTS ON THE RATE OF DETERIORATION OF PROTEIN INGREDIENTS

The glass transition temperature (Tg) is an important index for molecular mobility and diffusivity within food matrices (43). Studies of the relationship between Tg and the rate of deterioration of protein ingredients are just beginning. Matrix-based models are also currently being formulated and tested in relation to freeze-dried (pharmaceutical) proteins. These studies and the small number of investigations involving food protein ingredients are discussed here.

5.1. The Glass Transition Temperature Changes within highly concentrated, glass-forming, food systems were described using a polymer chemistry approach by Slade and Levine (44,45). Key elements of glassy state and its relation to food quality deterioration can be summarized as follows: 1. Foods and food materials can be treated as classic polymer systems.

2. The glass transition temperature (Tg) for a glass-rubber transition is a critical parameter that determines processability, quality, stability, and safety.

3. At temperatures below Tg the viscosity of a food matrix is extremely high such that physical and chemical changes are inhibited.

4. Above the Tg the viscosity of a food matrix decreases and various relaxation process (or chemical changes) become possible. 5. Food stability may be assured by storing at temperatures below

Tg. 6. Water is a ubiquitous plasticizer for both natural and fabricated food ingredients and products. The plasticizing effect of water

reduces Tg. 7. At temperatures above the Tg materials are in a disordered (rubbery or amorphous) state. 8. William-Landel-Ferry (WLF) kinetics apply at

Tg < T < Tg 1008C. The kinetics of food change does not conform to the linear Arrhenius equation. 9. Nonequilibrium glass/rubbery state transitions affect all time- dependent structural and mechanical properties in real-world foods;in contrast, thermodynamic concepts such as water activity are inapplicable. Corrected Amino Acid Scores 421

5.2. Models for Food Matrices Food matrices are described using the fringed-micelle model or the folder-chain lamella model. According to the fringed-micelle model, food materials are composed of crystalline and amorphous phases. Strong intermolecular associations between polymer chains account for regions of high order. Amorphous or disordered regions arise from the lack of strong polymer-polymer interactions. Heating leads to consecutive order- disorder transitions involving a glassy/rubber state (at Tg) followed by a crystalline/amorphous transition at the melting temperature (Tm). At 12 temperatures below the Tg the system possesses a viscosity of about 10 Pa s. Most physical and chemical processes (microbial growth, enzymatic activity, protein deterioration) are severely inhibited. For pure biopolymers Tg is generally about 150±2008C. As Tg 5 Tm a solid may undergo decomposition before the Tm is reached. According to the folded-chain lamella model (46), crystalline regions are formed by intramolecular associations involving a single polymer chain folded back on itself. Amorphous regions occur at interruption zones or regions of the polymer possessing defects, kinks, and other irregularities in the structure.

The plasticizing action of water leads to Tg decreasing with increasing moisture content. At high moisture levels (>20% w/w) Tg decreases to below the freezing point for water. Cooling produces a freeze-concentrated system as pure solvent water solidi®es into ice. The freeze-concentrated component then undergoes a liquid/glass phase transition at characteristic temperature 0 0 designated Tg . Solutes for which the Tg value is high exhibit greater preserving action (see later). The polymer science approach provides satisfactory explanations for a wide range of food-related phenomena including microbiological stability, enzymatic activity at low Aw, inhibition of collapse, and improved freeze drying, cooking, and frying processes. For low or intermediate moisture foods, the new approach leads to an integrated discussion of moisture, temperature, and time relations during the processing or storage. Speci®c process end points can be achieved using different time-temperature, moisture-temperature, and moisture-time combinations.

5.3. The Glass Transition and Protein Quality

The Tg for gluten was reported by several workers (47±52). The Tg has also been reported for beef proteins (53), caseinate (54), and gliadin (55±57).

Noel et al. (58) determined the Tg for fractionated wheat gluten proteins (a-, g-, and o-gliadins and HMW glutenin) using the Perkin-Elmer DSC2 422 Chapter 14 microcalorimeter ®tted with a liquid nitrogen cooling accessory (Fig. 2). Temperature scans from 270 and 370 K showed an exothermic peak at about 320 K (or 478C) for o-gliadin equilibrated with 12% water. For a moisture content < 20%, Tg decreased in accordance with the Gordon- Taylor equation,

w1Tg w2 Tg k 1 2 Tg 4 w1 w2k where T is the observed glass transition temperature, T is the glass g g1 transition temperature for the pure protein, T is the glass-transition g2 temperature for the pure plasticizer (À138 K for water), w1 and w2 are the weight fractions of protein and moisture, respectively, and k is a constant whose value increases with increasing plasticization. Nonlinear regression analysis lead to prediction of T for dry proteins based on the g1

FIGURE 2 Effect of moisture on the glass transition temperature for fractionated glutenin proteins. High-molecular-weight glutenin subunits (HMW) a-gliadin, g-gliadin, and o-gliadin. Pro®les were generated according the Gordon-Taylor relation, Eq. (4), using values for T and k given in Refs. g1 55 and 56. Corrected Amino Acid Scores 423 observed T values at different moisture levels. Values of T ranged from g g1 397 K (1248C) to 417 K (1448C) for the four gluten proteins. In the presence of 20% moisture, Tg decreased to between 270 K (À38C) and 280 K (78C).

Morales and Kokini (59) measured Tg for 7S and 11S soy globulin fractions using DSC and mechanical spectrometry. There were two Tg values due to cross-contaminating amounts of the 7S globulin in the sample of 11S globulins and vice versa. Single Tg values were obtained for highly puri®ed soy globulins. With 7S soy protein Tg was 387 K (1148C) to 206 K (678C) for moisture contents of 0 to 35%, respectively. For the puri®ed 11S fraction Tg ranged from 433 K (1608C) to 256 K (À178C) for moisture contents from 0 to 40%. Soy globulins behave as polymers that are highly plasticizable by water.

The effect of moisture on the Tg and caking properties of ®sh protein hydrolysate was examined by Aguiliera et al. (60). Increasing the relative vapor pressure from 0 to 0.64 reduced Tg from 352.1 K (79.18C) to 230.2 K (À42.88C). The plasticizing action of water was described by the Gordon- Taylor equation [Eq. (4)]. At a ®xed temperature of 198C (room temperature) collapse was initiated at a relative vapor pressure of 0.44, corresponding to the TÀTg value of 35.88C. Above a relative vapor pressure of 0.55 the following quality defects occurred: nonenzymic browning, collapse, shrinkage, and setting into a sticky, high-viscosity brown liquid. Using the WLF equation, the viscosity of the matrix at the onset of collapse was estimated as 105±107 (PaÁs). To preserve protein samples from deterioration it is necessary to employ storage temperatures below the Tg. However, the glassy-rubbery state transition takes place over a ®nite temperature interval. Peleg (61±63) modeled mechanical changes during the glass-rubber transition using Fermi's equation:

Y Y g 5 1 exp T À Tc=A

TC TC;0 exp Àkw 6

00 A A0 exp k w 7 where Y is an apparent stiffness parameter; Yg is the stiffness in the glassy state; T is temperature; TC is the in¯exion temperature, which is not necessarily coincident with the Tg;and A is a parameter that measures the slope of the Y-temperature graph. Both TC and A are dependent on the % 424 Chapter 14 moisture content (w). We have replicated these simulations (Fig. 3) with the following two results: (a) the in¯ection temperature decreases with increasing moisture, and (b) the gradient of each graph increases with increasing moisture. Therefore, sensitivity to temperature increases with increasing moisture. To preserve high-moisture foods requires a low storage temperature and improved temperature control. Drying allows higher storage temperatures and increased tolerance with respect to temperature variations. So far, only a few investigators have applied the polymer science approach to food protein deterioration. The physical-mechanical basis for glass-liquid transitions for proteins has not been extensively discussed.

According to Ferry (64), Tg is the temperature below which wriggling motions and conformational rearrangements within a polymer cease. Liquids and polymers have a constitutive volume (determined by van der Waals contacts) and free volume arising from packing irregularities or defects. During cooling the free volume changes in accordance with the

FIGURE 3 Effect of temperature and moisture on the apparent stiffness of gliadin near the glass transition temperature. The moisture content (%) is shown in the boxed legend. The in¯ection temperature for each transition is 138C, 38C, À148C, À288C, and À258C at a moisture content of 11.25%, 13%, 19.25%, 24.4%, and 27.2%. Simulations were performed using equation parameters from Peleg (56,61±63). Corrected Amino Acid Scores 425 thermal expansion coef®cient. The Tg is a narrow temperature region in which the thermal expansion coef®cient, heat capacity, adiabatic compres- sibility, and speci®c volume show discontinuity. What can undergo a glass transition in a protein solid? In answer to this self-posed question, Morozov and Gevorkian (65) suggested that the glass-liquid transition in globular proteins involves surface groups and packing defects. These structural components may or may not be hydrated. There are also aperiodic or random sequences in most proteins.

5.4. Studies on Freeze-Dried Proteins

The storage stability of pharmacologically active proteins is related to Tg. Freeze-dried high-value proteins undergo so-called moisture-induced deteriorative reactions. Most of these reactions have been identi®ed by accelerated testing (Table 1). Liu et al. (66) found that the storage stability of freeze-dried proteins was increased by adding excipients, mainly polyhydroxy alcohols. Costantino et al. (67) reported similar ®ndings. For proteins in the dry state additives act as platicizers that reduce Tg in accordance with the Gordon-Taylor equation. The destabilizing action of added excipients is inversely related to the Tg of the pure additive. Added excipients are stabilizing as compared with the equivalent weight of water. To achieve stabilization the pure additive should have T higher than water. g2 Excipients also increase the denaturation temperature (Tm) in direct proportion to their effect on Tg (68±70). Indeed, for some synthetic polymers the two transition temperatures are strongly correlated:

Tg K0:66 +0:04 Tm 8 The glass-liquid transition and the crystalline-amorphous transition appear to be subject to similar constraints (46).

5.5. Protein Conformational Stability, Tm, and Quality

Protein solutions have Tg values below room temperature although Tm remains high (30±908C). Regular 28 structure yields regions of order and crystallinity within the N state. The net conformational (dis)order determines the peak temperature (Tm) and enthalpy change (DH) measured by differential scanning calorimetry (DSC). Equation (8) suggests that a high Tm (compare Tg) will produce lower rates of protein deterioration in solution. Protein stability-function relations were discussed by Apenten and Berthalon (71) and Apenten (72) using the two-stage denaturation scheme. Reactive amino acid groups are buried in the N state and become accessible 426 Chapter 14 to the solvent after protein unfolding. High conformational stability therefore attenuates U ! I reactions [see Eqs (1)±(3)]. The magnitude of DE# provides useful information about the mechanism of deterioration. The N/U transition is a highly cooperative process with a large temperature coef®cient (Q10 > 2). Where the rate- limiting step for deterioration is a conformational change [Eq. (1)] the value of DE# should be 250±700 kJ molÀ1. In contrast, when the U ! I reaction determines the rate of deterioration then DE# is small (20±80 kJ molÀ1). The magnitude of DE# is not only determined by the rate-limiting step for protein deterioration because 1. Deterioration is a multistage reaction [Eqs (1)±(3)]. Therefore, DE# is the sum of values for several reactions. 2. Quality loss might not conform to the N ! U ! I scheme. Multiple U states (U1, U2, U . . . Ui) and/or U ! I reaction occur. 3. DE# is temperature dependent (71,72). Labuza and Saltmarch (73) investigated the effect of moisture and temperature on available lysine and nonenzymatic browning in whey powder. Labuza et al. (74) examined similar reactions for pasta. The loss of FDNB-lysine and nonenzymatic browning followed ®rst- and zero-order kinetics, respectively. The rates of lysine and PNV losses were two to three orders of magnitude greater than the rate of browning. The reaction temperature dependence ®tted a linear Arrhenius equation at 35±558C with # a squared regression (R2) coef®cient of *0.601. The DE for lysine loss and À1 browning ranged from 52 to 85 kJ mol depending on the prevailing AW. Adding data from 258C produced curved Arrhenius plots. I have reexamined these results using a higher order Arrhenius equation that allows for variations in the value for DE# with temperature. The simple Arrhenius equation is

DE# ln k ln k0 À 9 RT where k is a rate constant and DE# is the energy required to form an activated complex. A semilog arithmic plot of ln k versus 1/T leads to a straight-line graph having a slope equal to DE#=R and an intercept value of ln k0. A more comprehensive discussion of reaction rates is based on the transition-state theory. Here the activated complex from the Arrhenius model is replaced by a transition state, and ``activation energy'' becomes DG# (the Gibbs free energy change for producing the transition state). Comparing the Arrhenius and transition state reaction rate models leads to the realization that DE# is essentially equal to DH# (enthalpy change for Corrected Amino Acid Scores 427 producing a transition state) where DH# DE# À 2RT. Reactions taking place ``in water'' can be accompanied by signi®cant heat capacity change; # # DCp d DH =dT.* To allow for a temperature-dependent DH# value, the Arrhenius equation is expanded to a second-order polynomial.

ln k a bx cx2 10 where a, b and c are constants and x 1=T. From general thermodynamic # # # # principles DE Rd ln k=dx &DH and d DH =dT DCp and therefore 2c DH# b 11 T À1 and 2R DC# À c 12 p T À2 Fig. 4 and Table 2 show results for the deterioration of pasta (74). The loss of FNDB-lysine and browning reactions were adequately described by Equation (10) (R2 0:98 À 0:99). # Close to the Tg (1208C) for a dry protein the magnitude of DH is large, implying that quality loss may be limited by conformational transitions. The low DH# value near room temperature is a consequence of the temperature dependence of this parameter.

6. PROTEIN DIGESTIBILITY±CORRECTED AMINO ACID SCORES (PDCAAS)

There has long been dissatisfaction with the rat PER assay. In 1991 an FAO/WHO Expert Consultative Group (75) agreed that the PER test should be replaced by a new method with the following characteristics: (a) greater accuracy and reproducibility than the PER test, (b) shorter time for assay, (< 48 hours per test), (c) incorporation of digestibility, (d) wider applicability to samples without extensive prepretreatment, (e) simplicity

À1 À1 * The heat capacity (Cp;Jg K ) is the amount of heat required to raise the temperature of 1 g of material by 1 K. The reaction A B ! C D will lead to a heat capacity change (DCp) if the reactants and products interact differently with the solvent. 428 Chapter 14

FIGURE 4 Arrhenius plot for protein deterioration in pasta at different temperatures. Relative vapor pressure is 0.44. (Based on data from Ref. 74.)

TABLE 2 Activation Parameters for Protein Deterioration in Pasta;Analysis by a Nonlinear Arrhenius Equation

Parameter Lys loss (RVP 0.44) Browning (RVP 0.44) a 212.3 192.7 b À1.328 6 105 À1.22 6 105 c 1.968 6 107 1.825 6 107 R2 0.991 0.980 DH# (kJ molÀ1) 42.2a (272.2) 40.1 (252.6) DS# (J molÀ1KÀ1) À202.8 (419.1) À229.1 (264.5) DG# (kJ molÀ1) 104.79 (107.6) 110.7 (146.6) # À1 À1 DCp (J mol K ) 4395 2359 aThe ®rst and second values are for temperatures of 378C and 1208C. Corrected Amino Acid Scores 429 and suitability for routine use, and (f) low cost not exceeding $200 per test at 1991 prices. The panel of eminent nutritionists selected a new test involving protein digestibility-corrected amino acid score (PDCAAS). They further agreed that PDCAAS should be based on the FAO/WHO/UNU 1985 list showing the essential amino acid requirements for 2±5 yr. preschool children (76). The values are given in Chapter 12 (p. 355). In 1993 the FDA (USA) adopted the PDCAAS procedure for the routine evaluation of PNV and for food labeling purposes. Developments leading to the adoption of the PDCAAS test are reviewed by Sarwar and McDonough (77), Boutrif (78), Madi (79), Henley and Kuster (80), and also Kuntz (81). The basic principles for evaluating PDCAAS and examples of its use are discussed in Section 6.5. First, however, we review older methods for PNV evaluation and how the PDCAAS index evolved (Sections 6.1±6.4).

6.1. Protein Chemical Score Mendel (82) stated that a protein's quality is related to its minimum quantity of essential amino acids. In 1946 Mitchell and Block (83) de®ned the chemical score for protein quality by calculating the de®cit of essential amino acids as compared with essential amino acids from whole egg. Samples with equal amounts of crude protein (%N 6 6.25) were analyzed for their essential amino acid content. The results were each divided by the essential amino acid content for whole egg. The result having the largest de®cit compared with egg protein shows the limiting essential amino acid. EAA Chemical score SAMPLE 13 EAAWHOLE EGG The chemical score for 28 proteins was positively correlated BV determined using the rat bioassay (Table 3). Thermal processing produced signi®cant changes in BV without affecting the essential amino acid content of proteins (Fig. 5). Knorr (84) found that potato protein concentrates prepared in different ways possessed high chemical scores and may be useful for human nutrition. Based on their chemical scores, immature durum wheat has a higher nutrient value than mature wheat (85). Chemical scores for eight peanut cultivars were highly correlated with RNV (R 0:98) and PER (R 0:88) (86). Ruales and Nair (87) found that the limiting amino acids in quinoa ¯our were tyrosine and phenylalanine, yielding a chemical score of 0.86. The chemical score has two shortcomings with regard to protein quality determination. First, adopting a Kjeldahl factor of 6.25 for two protein sources with equal nitrogen will lead to error if the samples vary 430 Chapter 14

TABLE 3 Chemical Scores for a Range of Food Proteins Determined Using Whole Egg Standard

Chemical Limiting Chemical Digestibility score 6 Protein source EAA score BV (%) (%) Dig/100

Beef muscle Cys Met 0.71 76 100 71.0 Beef liver Ile 0.70 77 97 67.9 Egg albumin Lys 0.69 82 100 69.0 Cow's milk Cys Met 0.68 90 95 64.6 Lactalbumin Met 0.66 84 98 64.7 Beef kidney Cys Met 0.65 77 99 64.4 Beef heart Ile 0.65 74 100 65.0 Casein Cys Met 0.58 73 99 57.4 Sun¯ower seed Lys 0.53 65 94 49.8 Soybean (heated) Met 0.49 75 96 47.0 Rolled oats Lys 0.46 66 93 42.8 Yeast (average) Cys Met 0.45 69 93 41.9 White rice Lys 0.44 66 78 34.3 Corn germ Met 0.39 78 85 33.2 Sesame seed Lys 0.39 71 92 35.9 Wheat germ Ile 0.38 75 95 36.1 Whole wheat Lys 0.37 70 91 33.7 Cottonseed Lys 0.37 61 90 33.3 Whole corn Lys 0.28 60 94 26.3 White ¯our Met 0.28 52 100 28.0 Peanut Met 0.24 58 97 23.3 Pea Met 0.24 48 91 21.8 Gelatin Trp 0.0 25 95 0.0 Human milk Met 0.86 Blood serum Ile 0.44 Hemoglobin Ile 0.10 Flax seed Lys 0.35 Alfalfa Ile 0.45

Source: Adapted from Ref. 78.

greatly in the amount of nonprotein nitrogen (Chapter 1). Second, changes in protein characteristics unrelated to their essential amino acid content are not measured by the chemical score. Protease inhibitors will affect PNV although such effects are not re¯ected by changes in the chemical score. Corrected Amino Acid Scores 431

FIGURE 5 The chemical score as an index for protein quality. (Top) Chemical score is strongly correlated with the biological value. (Bottom) Showing the correlation between biological value and ``chemical score corrected for digestibility'' (open circles). A low correlation is observed between BV and digestibility (see closed symbols). 432 Chapter 14

6.2. Pepsin Digest Residue Index Sheffner et al. (88) introduced the pepsin digest residue (PDR) index in 1956. The PDR is determined by combining in vitro digestibility with the essential amino acid pattern. The food sample (containing 1 g of protein) was incubated with pepsin (25 mg) in 30 mL of sulfuric acid (0.1 N) solution for 24 hours.* Precipitating with sodium tungstate and 0.66 M sulfuric acid separated undigested protein from the products. The essential amino acid pro®les for the substrate and soluble digest (products) are determined by microbiological assay. Subtraction of these values gave the essential amino acid pattern for the nonhydrolysed protein residue. For both the digest and residue, each essential amino acid is expressed as a percentage of the total content of essential amino acids. The two columns of results were then divided by the corresponding values for egg protein. Next, the geometric mean was determined for the ``egg ratios'' and the resulting values were multiplied by a factor that takes into account the relative amounts of digest and residue formed by the action of pepsin on the sample and the egg protein. The PDR index was not easy to calculate. Sheffner et al. noted that any process that decreases pepsin digestibility will also lower protein nutritional value.

6.3. Pepsin Pancreatic Digestion Index Akeson and Stahmann (89) introduced the pepsin pancreatin digestion (PPD) index in 1964. Samples of proteins were digested with pepsin at low pH, neutralized, and then treated with pancreatin. The soluble products were concentrated, freeze dried, and subjected to amino acid analysis by ion- exchange chromatography. The PPD index was then calculated in the same way as the PDR index. The PPD index showed an excellent correlation (R 0.99) with BV determined using the rat bioassay. Kennedy et al. (90) modi®ed the PPD index by performing the pepsin digestion within a dialysis cell. The pepsin digest dialysate (PDD) index for a range of protein ingredients (soy ¯our, gelatin, gluten, casein, whole egg powder, cows' milk forti®ed with carbohydrates, proteins, and vitamins) was highly correlated with BV. Some suggested advantages of the PDD index compared with the PDR and PPD indices include (a) use of simpler apparatus, (b) requirement

* Some experiments were performed with multiple enzyme digestion using trypsin, pancreatin, and erepsin. However, the ®nal method used a single enzyme digestion by pepsin. It was suggested that the proportion of egg protein digested as chyme leaves the duodenum was about 30%. The in vitro study of Sheffner et al. employed conditions designed to ensure about 30% digestibility of their samples. Corrected Amino Acid Scores 433 for only one enzyme, (c) use of modern amino acid analysis instrumentation, (d) higher reproducibility, and (e) use of computerized calculations.

6.4. Computed-PER Index Satterlee et al. (91) introduced a computed-PER (c-PER) index in an attempt to correct AAS for digestibility in 1977. First the essential amino acid pattern for a food sample was expressed as a percentage of the FAO/ WHO pattern for humans (1973). Then each ``%FAO value'' was multiplied by protein digestibility determined with a rat bioassay or in vitro. The digestibility-corrected AASs were assigned a statistical weighting inversely related to the abundance of each essential amino acid. The sum of the weighted values was divided by the corresponding value for casein. Finally, the results were transformed by one of four mathematical functions to generate c-PER values. A collaborative study involving seven laboratories found that the c-PER gave the same rankings of protein quality as the PER assays. The c-PER test could be completed in 72 hours compared with 28 days required for the PER determination.

6.5. Available Amino Acid Score and PDCAAS The PDCAAS is considered an excellent index for protein quality. Particularly appealing is the ease of calculation. It combines protein digestibility and AAS in a simple manner. With 20-20 hindsight the PDCAAS could have been anticipated earlier. Chemical scores were ®rst criticized for not taking digestibility into account in 1946 (92). Satterlee et al. (91) pretty much calculated the PDCAAS value in 1977 and then proceeded to transform such data to give the c-PER. In 1984 Sarwar (93) introduced the term ``available amino acid score'' to describe protein scores* corrected for protein digestibility. His formula was identical to that used later for calculating PDCAAS (Table 4). Published lists for TPD may be used to calculate PDCAAS. Digestibility data are also available from the FDA. Manufacturers should note that processing can lead to signi®cant deviations in the TPD compared with published values. Where a food source contains several proteins (Pi)

* The protein score is calculated in the same way as the chemical score except that the reference essential amino acids is the pro®le for humans. 434 Chapter 14

TABLE 4 Stages for the Calculation of Protein Digestibility± Corrected Amino Acid Score (PDCAAS) for a Food Protein

Procedure

1. Determine sample nitrogen content 2. Calculate protein content (N 6 6.25 or speci®c conversion factor) 3. Analyze sample for essential amino acids 4. Determine the amino acid score ASS =; mg EAA per 1 g protein mg EAA per 1 g FAO=WHO=UNU

5. Determine digestibility 6. Calculate PDCAAS Lowest AAS 6 digestibility

the TPD is replaced by the weighted average (TPDav) value: P TPDi6Pi TPDav P 14 Pi All proteins having a PDCAAS in excess of 100% (or > 1) are assigned a value of 100%. A protein with a PDCAAS value above 100% does not provide further bene®t as excess amino acids are utilized for energy. According to Henley and Kuster (80), the PDCAAS method provides a measure of protein quality that is directly correlated with human requirements. The PDCAAS method also has considerable ¯exibility. Manufacturers and diet planners can provide larger quantities of lower quality dietary protein in order to meet the recommended daily requirements.

6.6. Applications Sarwar et al. (94) showed that a number of rat bioassays (PER, PER, RPER, NPR, or RNPR) ranked 20 food products in the same order of protein quality. The correlations between different rat assays were highly signi®cant (r 0.98±0.99). Chemical scores were also correlated with the results of rat bioassays. Correcting the chemical score for the digestibility improved the observed correlation. Carnovale et al. (95) compared PNV for wild and cultivated species of Vigna. The wild type had a signi®cantly higher protein content, trypsin inhibitory activity, and tannin content. Protein digestibility was lower although PNV assessed in terms of the PDCAAS was not signi®cantly different. Corrected Amino Acid Scores 435

A. Quinoa Quinoa (Chenopodium quinoa Wild) seed protein quality was evaluated using amino acid analysis and animal feeding trials. The ®rst limiting amino acids were tyrosine and phenylalanine with a chemical score of 0.86. Quinoa protein (14% w/w of the seed) had levels of lysine, methionine, and cysteine superior to those found in most other plant proteins. From animal feeding experiments NPU was 75.7, BV was 82.6, and digestibility was 91.7%. These results yield an estimated PDCAAS of 79%. This value is higher than the value for meat (87).

B. Rice The quality of protein associated with cooked milled rice and a typical rice- based menu for Filipino preschool children and adults was assessed by Eggum et al. (96). Digestibility, BV, and NPU were assessed with growing rats. Digestibility was 88.8% for the preschool child diet, compared with a BV of 90.0 and NPU of 79.9. For the adult diet digestibility was 87.3%,BV was 86.6, and NPU was 75.5. On its own, cooked rice had a digestibility of 90.0%, BV equal to 82.5, and an NPU value of 74.3. The availability of lysine (the limiting essential amino acid) was 95.4% for the preschool child diet, 95.7 for the adult diet, and 100.0 for rice. For whole diets the chemical scores were 1.00 for the preschool child diet, 0.92 for the adult diet, and 0.62 for rice. From such ®gures the PDCAAS may be estimated as 88.8% for the preschool diet, 80.4% for the adult diet, and 56.0% for cooked rice.

C. Maize The quality of protein from Canadian maize cultivars adapted to Northern latitudes (greater than 458 N) was assessed by Zarkadas et al. (97). The cultivars designated Dent CO251, Flint CO255, and Pioneer 3953 compared favorably with quality protein maize inbred (QPM-C13). Total protein levels for maize meal were 7.95% (QPM), 8.2% (Pioneer), 10.5% (Dent), and 11.79% (Flint). Compared with ordinary maize, QPM protein had double the amount of lysine and arginine, increased levels of tryptophan and cystein, and lower levels of leucine. QPM protein had a good balance of essential amino acids limited only in lysine. The PDCAAS was 67% (QPM), 28.5% (Pioneer), 31.0% (Dent), or 33.0% (Flint). PNV for a white ¯oury maize variety from the Indian Agricultural Program of Ontario (98) was also evaluated. Maize (IAPO-13) harvested over 1992, 1993, and 1994 seasons had an average protein content of 10.14 (+0.1)%. The limiting essential amino acid was lysine with an AAS of 0.414. The human digestibility of maize protein is reportedly 89%, leading to an estimated 436 Chapter 14

PDCAAS for IAPO-13 of 37±38%. It was concluded from such studies that Canadian QPM maize variety had higher quality than common maize and that breeding maize for high protein quality showed a great deal of promise.

D. Oats Hull-less or naked oats (Avena sativa var. nuda) from temperate zones in Asia were studied by a Canadian group for their potential use for both human and animal nutrition. Zarkadas et al. (99) determined the protein content for three high-yielding and rust-resistant naked oat cultivars. The protein levels were 13.67 (+ 0.60)%, 13.93 (+ 0.53)%, and 14.40 (+ 0.55)% for varieties AC Percy, AC Hill, and AC Lotta. All cultivars had a good balance of nine essential amino acids. Lysine was limiting, followed by threonine. Assuming a human protein digestibility of 86%, estimates for PDCAAS are 54.9% (AC Hill), 56.3% (AC Lotta), and 59.3% (AC Percy). The PDCAAS values for other protein sources are mechanically dehulled oats (62%), maize (29%), soybean (86%), and egg (95%). Zarkadas et al. (100) also evaluated the PNV of two newly released Canadian oat cultivars (Newman and AC Stewart). Oat wholemeal had a total protein content of 10.75 (+ 0.23)% and 11.92 (+ 0.06)% for strains Newman and AC Stewart, respectively. The corresponding total protein content for the oat groats (dehulled grains) was 13.27 (+ 0.24)% and 12.61 (+ 0.94)%. The limiting amino acid was lysine for oat wholemeal and lysine plus threonine for groats. Values for PDCAAS based on the FAO/WHO/ UNU pattern (2-year-old preschool children) were reportedly 58±62.3% (strain Newman) or 66.7±62.3% (strain AC Stewart). Dehulling had no easily predictable effect on the PNV.

E. Collagen Connective tissue protein (from skin, rind, tendon, or bone) is potentially useful for livestock feed supplements. The low PNV of collagen was inferred from feeding trials conducted in the 1960s. Meat meal protein quality is also inversely related to the hot water±soluble protein (collagen) content. Feather meal had a distinctly low PNV during an ARC collaborative study of animal foodstuffs reported in Chapter 13. A contemporary evaluation of PNV for collagen-containing meat by-products is reported by Zarkadas et al. (101). Three commercial batches of demineralized beef bone powder were subjected to quantitative amino acid analysis to determine total protein content and PDCAAS. The amino acid pro®les were consistent with the type I collagen being the main protein constituent. Glycine, proline, and hydroxyproline each constituted 20% of the total amino acids. Alanine (10% total amino acids) and basic amino acids (12.5% total) were also relatively Corrected Amino Acid Scores 437 abundant. Comparatively low amounts of cysteine (*0.2%), methionine (*1.24±1.33%) and threonine (1.17±2.24%) were found. The protein content of bone meal ranged from 18 to 18.6%. Taking the average digestibility of collagen as 90%, the PDCAAS value was reportedly 14.4± 16.4%;it is not clear how such ®gures were determined. Table 5 lists the essential amino acids pro®le for the three batches of demineralized bone powder. Allowing for rounding-up errors, threonine is the limiting amino acid for one sample with a chemical score of 0.344. Luecine and isoleucine are the next limiting amino acids with a score of 0.510±0.550. Following the normal rules, the PDCAAS is between 31.0% and 49% for bone meal products.

F. Beans Changes in the protein quality of red kidney bean (Phaseolus vulgaris L) due to autoclaving, domestic cooking, or canning were examined by Wu and co- workers (102) from Clemson University, South Carolina. A large number of quality indices were considered: available lysine, c-PER, mean chemical score for lysine, methionine-plus-cysteine levels, and the essential amino acids index. In vitro protein digestibility was determined with a multienzyme method. Cysteine plus methionine was limiting with a chemical score of 0.928. In vitro digestibility was 43.2% for uncooked kidney beans and 79±82% for heat-treated beans. The PDCAAS was 40% for raw beans compared with 70±75.4% for cooked beans. Other quality indices showed

TABLE 5 Estimation of the PNV for Bone Meal Products

FAO/WHO/UNUa Amino acids EAA for collagen (mg/1 g protein)

Histidine 11 19 Isoleucine 16 28 Leucine 35 66 Lysine 335 58 Methionine cystine 15 25 Phenylalanine tyrosine 23 63 Threonine 12±22 34 Tryptophan 11 11 Valine 27 35 Total EAA 184±195 339 Total nonessential AA 805±816 661 aFrom FAO/WHO/UNU (1985);[Refs. 75±81];see Table 6 of Chapter 12. 438 Chapter 14 sensitivity to heating when corrected for digestibility. For example, raw and home-cooked beans had available lysine contents of 6.21 and 6.19 g per 100 g protein, respectively. This yields corresponding in vitro protein digestibility±corrected available lysine scores 2.68 and 5.1 g per 100 g protein. After normalization with the FAO/WHO/UNU (1985) pattern, the protein digestibility±corrected available lysine score (PDCALS) was 40% for raw beans and 74% for cooked beans. Apparently, lysine as well as cysteine plus methionine may be limiting for red kidney beans.

6.7. Beyond Protein Digestibility±Corrected Amino Acid Scores Amino acid availability (AAAv) is a more precise measure of digestibility than TPD (Chapter 12). Calculations of PDCAAS using TPD assume that all essential amino acids are released from proteins with equal ease. The rates of absorption are also assumed to be equal. Batterham (103) has reviewed the signi®cance of AAAv. Kuiken and Lyman (104) showed that availability of essential amino acids from a single protein source can differ. The AAAv was determined for a range of protein sources (egg, liver extract, roast beef, cotton seed, peanut, and wheat ¯ours) using a rat feeding trial. Employing a microbiological assay for amino acids, they determined the dietary and fecal concentrations of individual essential amino acids. With roast beef, egg, or liver protein values for AAAv were the same for all essential amino acids. The amino acids from cottonseed meal protein fell into three groups: (a) arginine, histidine, and tryptophan with high availability (>90%); (b) isoleucine, phenylalanine, and threonine with intermediate availability (70±90%);and (c) lysine with low availability (65 %). Sarwar (105) con®rmed that AAAv was up to 25% lower than the TPD for legume protein. The AAAv and TPD were determined for 17 protein sources using male rats. Diets and feces were freeze dried, ground, and analyzed for amino acid pro®les and for crude protein. The AAS values for the different diets were expressed as percentages of human requirements speci®ed by the FAO/WHO (1985). Finally, these results were corrected for AAAv. The resultant index for protein quality was termed ``available amino acid score'';I have renamed this quantity availability-corrected amino acid score (AvCAAS).* Correcting the chemical scores for TPD led Sarwar to generate what is probably the ®rst list for PDCAAS. Both AvCAAS and the

* Sarwar used AAS as an abbreviation for the available amino acid score. I have changed this to AvCAAS (availability-corrected amino acid scores). AAS is used by many investigators to mean amino acid score. Some investigators use AASTPB to describe amino acid scores corrected for true protein digestibility. Corrected Amino Acid Scores 439 notional PDCAAS were highly correlated with the RNPR for 17 diets (R 0.92). However, values for PDCAAS were consistently higher than AvCAAS by between 2% and 8% (Fig. 6). Wu and co-workers (106,107) found much more signi®cant differences in PNV after correcting amino acid scores for TPD and AAAv. With raw kidney beans, cysteine plus methionine was limiting with a chemical score of 0.944. The TPD value was 15.7% for raw beans, increasing to 72±87% for heat-processed beans. By comparison, AAAv was negative (À18.6%) for raw bean protein and positive (39.8±68.0%) for heat-processed beans. Rats fed with raw beans had higher concentrations of cysteine plus methionine in their feces as compared with amounts initially present in their diets. The AAAv values for several other amino acids (alanine, proline, valine, luecine, and threonine) were also negative. The AvCAAS for raw kidney bean protein was therefore À17.6% compared with a PDCAAS of 13.9%. The relative merits of the PDCAAS and AvCAAS are discussed by Darragh et al. (108). Estimating TPD using a rat assay is a slow process, requiring about 8 days for completion. In view of the high correlation between TPD and in vitro protein digestibility, correcting AAS using in vitro protein digestibility is more cost effective. Rozan and co-workers (109) considered AAS for soybean, lypine, and rapeseed meal proteins using the FAO/WHO/UNU (1995) pattern for 2- to 5-year-old preschool children. The AAS values were then corrected using in vitro protein digestibility measured as the degree of

FIGURE 6 Protein nutrient value estimates based on amino acid scores corrected for true protein digestibility (PDCAAS) or amino acid availability (AAS). [Drawn from the data of Sarwar (93).] Y-axis shows PDCAAS minus AAS. 440 Chapter 14 hydrolysis (DH) or/and nitrogen digestibility index (ND). Degree of hydrolysis±corrected AAS (DHCAAS) or nitrogen digestibility±corrected AAS (NDCAAS) values were highly correlated with values of PDCAAS (p < 0.001).

REFERENCES

1. K Linderstrom-Lang, RD Hotchkiss, G Johansen. Peptide bonds in globular proteins. Nature 142:996, 1938. 2. OB Ptitsyn. How the molten globule state became. Trends Biochem Sci 20:376±379, 1995. 3. M Hirose. Molten globule state of proteins. Trends Food Sci 4:48±51, 1993. 4. JE Wilson. The use of monoclonal antibodies and limited proteolysis in elucidation structure-function relationships in proteins. Methods Biochem Anal 35:207±205, 1991. 5. GE Matthyssens, G Simons, L Kanarek. Study of the thermal-denaturation mechanism of hen egg-white lysozyme through proteolytic degradation. Eur J Biochem 26:449±454, 1972. 6. T Imoto, K-I Fakuda, K Yagishita. A study of the native-denatured (N D) transition in lysozyme. I. Detection of the transition by product analysis of protease digests. Biochim Biophys Acta 336:264±269, 1974. 7. T Imoto, K-I Fuakuda, K Yagishita. A study of the native-denatured (N D) transition in lysozyme. II. Kinetic analysis of protease digestion. J Biochem (Tokyo) 80:1313±1318, 1976. 8. FC Church, GL Catignani, HE Swaisgood. Use of immobilized Streptomyces griseus proteases (pronase) as a probe of structural transitions of lysozyme, b- lactoglobulin and casein. Enzyme Microbial Technol 4:317±321, 1982. 9. RM Daniel, CA Cowan, HW Morgan, MP Curran. A correlation between protein thermostability and resistance to proteolysis. Biochem J 207:641±644, 1982. 10. H Ueno, WF Harrington. An enzyme-probe method to detect structural changes in myosin rod. J Mol Biol 173:35±61, 1984. 11. A Kato, K Komatsu, K Fujimoto, K Kobayashi, Relationship between surface functional properties and ¯exibility of proteins detected by the protease susceptibility. J Agric Food Chem 33:931±934, 1985. 12. A Kato, K Fujimoto, N Matsudomi, K Kobayashi. Protein ¯exibility and functional properties of heat-denatured ovalbumin and lysozyme. Agric Biol Chem 50:417±420, 1986. 13. A Fontana, G Fassina, C Vita, D Dalzoppo, M Zamai, M Zambonin. Correlation between sites of limited proteolysis and segmental mobility in thermolysin. Biochemistry 25:1847±1850, 1986. 14. RKO Apenten, YL Folawiyo. The effect of pH on rapeseed globulin (cruciferin) binding to anilinonaphthalene-8-sulphonate. J Food Biochem 19:455±465, 1997. Corrected Amino Acid Scores 441

15. I Schechter, A Berger. On the size of the active site in proteases. I. Papain. Biochem Biophys Res Commun 27:157±162, 1967. 16. R Lumry, H Eyring. Conformation changes of proteins. J Phys Chem 58:110± 120, 1954. 17. C Tanford. Protein denaturation. Adv Protein Chem 23:121±282, 1968. 18. TJ Ahern, AM Klibanov. Mechanism of irreversible enzyme inactivation at 1008C. Science 228:1280±1284, 1985. 19. TJ Ahern, AM Klibanov. Analysis of processes causing thermal inactivation of enzymes. Methods Biochem Anal 33:91±127, 1988. 20. JE Kinsella. Texturized proteins: fabrication, ¯avoring, and nutrition. CRC Crit Rev Food Sci Nutr 10:147±207, 277, 1978. 21. LC Maillard. Action des acides amines sur les sucres: formation des meÂ lanoidines par voie meÂ thodique. C R Hebd Seances Acad Sci 154:66±68, 1912. 22. E Dworschak. Nonenzyme browning and its effect on protein nutrition. CRC Crit Rev Food Sci Nutr 13:1±40, 1980. 23. TP Labuza, M Saltmarch. The nonenzymatic browning reaction as affected by water in foods. In: LB Rockland, GF Stewart, eds. Water Activity: In¯uence on Food Quality. New York: Academic Press, 1981, pp. 605±650. 24. M Friedman. Lysinoalanine formation in soybean proteins: kinetics and mechanisms. In: J Cherry ed. Food Protein Deterioration. Mechanisms and Functionality. Washington, DC: American Chemical Society, 1982, pp. 231± 273. 25. M Friedman. Chemically reactive and unreactive lysine as an index of browning. Diabetes 31(Suppl 3):5±14, 1982. 26. M Saltmarch, TP Labuza. Nonenzymatic browning via the Maillard reaction in foods. Diabetes 31(Suppl 3):29±36, 1982. 27. RE Feeney, JR Whitaker. The Maillard reaction and its prevention. In: J Cherry ed. Food Protein Deterioration. Mechanisms and Functionality. Washington, DC: American Chemical Society, 1982, pp 201±229. 28. RF Hurrell. Reactions of food proteins during processing and storage and their nutritional consequences. In: BFJ Hudson, ed. Developments of Food ProteinsÐ3. London: Elsevier Applied Science, 1984, pp 213±244. 29. MS Feather. D-Glucose-L-alanine interactions: their role in Maillard polymer formation and Strecker degradation reactions. In: RD Philips, JW Finley, eds. Protein Quality and the Effects of Processing. New York: Marcel Dekker, 1989, pp 263±272. 30. PA Finot. Nonenzymic browning products: physiological effects and metabolic transit in relation to chemical structure: a review. Diabetes 31(Suppl 3):22±28, 1982. 31. K Lorenz. Microwave heating of foodsÐchanges in nutrient and chemical composition. CRC Crit Rev Food Sci Nutr 7:339±370, 1976. 32. GA Cross, DYC Fung. A review of the effects of microwave cooking on foods. J Environ Health 44:88±193, 1982. 442 Chapter 14

33. PM Masters, M Friedman. Racemization of amino acids in alkali-treated food proteins. J Agric Food Chem 27:507±511, 1976. 34. JA Maga. Lysinoalanine in foods. J Agric Food Chem 32:955±964, 1984. 35. MS Otterburn. Protein cross-linking. In: RD Philips, JW Finley, eds. Protein Quality and the Effects of Processing. New York: Marcel Dekker, 1989, pp 247±261. 36. HE Swaisgood, GL Catignani. Protein digestibility: in vitro methods of assessment. Adv Food Nutr Res 35:185±236, 1991. 37. M Friedman, R Liardon. Racemization kinetics of amino acid residues in alkali-treated soybean proteins. J Agric Food Chem 33:666±672, 1985. 38. PM Masters, M Friedman. Racemization of amino acids in alkali-treated food proteins. J Agric Food Chem 27:507±511, 1979. 39. M Friedman, PM Masters. Kinetics of racemization of amino acid residues in casein. J Food Sci 47:760±764, 1982. 40. R Hayashi, I Kameda. Racemization of amino acid residues during alkali treatment of proteins and its effects on peptic digestibility. Agric Biol Chem 44:891±895, 1980. 41. S-Y Chung, HE Swaisgood, GL Catignani. Effects of alkali treatment and treatment in the presence of fructose on digestibility of food proteins as determined by an immobilized digestive enzyme assay (IDEA). J Agric Food Chem 34:579±584, 1986. 42. AP De Groot, P Slump. Effect of severe alkali treatment of proteins on amino acid composition and nutritive value. J Nutr 98:45±56, 1969. 43. YJ Roos. Phase Transitions in Foods. New York: Academic Press, 1995. 44. L Slade, H Levine. Beyond water activityÐrecent advances based on an alternative approach to the assessment of food quality and safety. Crit Rev Food Sci Nutr 30:115±360, 1991. 45. L Slade, H Levine, J Ievolella, M Wang. The glassy state phenomenon in applications for the food industry: application of the food polymer science approach to structure-function relationships of sucrose in cookie and cracker systems. J Sci Food Agric 63:133±176, 1993. 46. JW Nicholson. The Chemistry of Polymers. 2nd edi. Cambridge: The Royal Society of Chemistry, 1997, pp 47±62. 47. GE Attenburrow, AP Davies, RM Goodband, SJ Ingman. The fracture behavior of starch and gluten in the glassy state. J Cereal Sci 16:1±12, 1992. 48. MT Kalichevsky, EM Jaroszkiewicz, JMV Blanshard. Glass transition of gluten. I. Gluten and gluten-sugar mixtures. Int J Biol Macromol 14:257±266, 1992. 49. MT Kalichevsky, EM Jaroszkiewicz, JMV Blanshard. Glass transition of gluten. II. The effect of lipids and emulsi®ers. Int J Biol Macromol 14:267± 273, 1992. 50. N Gontard, S Guilbert, JL Cuq. Water and glycerol as plasticizers affect mechanical and water vapor barrier properties of an edible wheat gluten ®lm. J Food Sci 58:206±211, 1993. Corrected Amino Acid Scores 443

51. G Sartor, GP Johari. Polymerization of a vegetable protein, wheat gluten, and the glass-softening transition of its dry and reacted state. J Phys Chem 100:19692±19701, 1996. 52. M Pouplin, A Redle, N Gontard. Glass transition of wheat gluten plasticized with water, glycerol, or sorbitol. J Agric Food Chem 47:538±543, 1999. 53. G Sartor, GP Johari. Structural relaxation of a vitri®ed high-protein food, beef, and the phase transformations of its water content. J Phys Chem 100:10450±10463, 1996. 54. M Le Mester, RB Duckworth. Effect of water content on the mobility of solute molecules and of protein side chains in caseinate preparations. Int J Food Sci Technol 23:457±466, 1988. 55. EM Graff, H Madeka, AM Cocero, JL Kokini. Determination of the effect of moisture on gliadin glass transition using mechanical spectrometry and differential scanning calorimetry. Biotechnol Prog 9:210±213, 1993. 56. M Peleg. Mathematical characterization and graphical presentation of the stiffness-temperature-moisture relationship of gliadin. Biotechnol Prog 10:652±654, 1994. 57. JL Kokini, AM Cocero, H Madeka, E de Graff. The development of state diagrams for cereal proteins. Trends Food Sci Technol 5:281±288, 1994. 58. TR Noel, R Parker, SG Ring, AS Tatham. The glass-transition behavior of wheat gluten proteins. Int J Biol Macromol 17:81±85, 1995. 59. A Morales, JL Kokini. Glass transition of soy globulins using differential scanning calorimetry and mechanical spectrometry. Biotechnol Prog 13:624± 629, 1997. 60. JM Aguiliera, G Levi, M Karel. The effect of water-content on the glass- transition and caking of ®sh±protein hydrolysates. Biotechnol Prog 9:651±654, 1993. 61. M Peleg. A model of mechanical changes in biomaterials at and around their glass transition. Biotechnol Prog 10:385±388, 1994. 62. M Peleg. Mathematical characterization and graphical presentation of the stiffness-temperature-moisture relationship of gliadin. Biotechnol Prog 10:652±654, 1994. 63. M Peleg. On modeling changes in food and biosolids at and around their glass transition temperature range. Crit Rev Food Sci Nutr 36:49±67, 1996. 64. JD Ferry. Viscoelastic Properties of Polymers. New York: John Wiley & Sons, 1980. 65. VN Morozov, SG Gevorkian. Low temperature glass-transition in proteins. Biopolymers 24:1785±1799, 1995. 66. WR Liu, R Langer, AM Klibanov. Moisture-induced aggregation of lyophilized proteins in the solid state. Biotechnol Bioeng 37:177±184, 1991. 67. HR Costantino, R Langer, AM Klibanov. Moisture-induced aggregation of lyophilized insulin. Pharm Res 1:21±29, 1994. 68. LN Bell, MJ Hageman, LM Muraoka. Thermally-induced denaturation of lyophilized bovine somatotropin and lysozyme as impacted by moisture and excipients. J Pharm Sci 84:707±712, 1995. 444 Chapter 14

69. BS Chang, RM Beauvais, AC Dong, JF Carpenter. Physical factors affecting the storage stability of freeze-dried interleukin-1 receptor antagonist: glass transition and protein conformation. Arch Biochem Biophys 331:249±258, 1996. 70. RG Strickley, BD Anderson. Solid-state stability of human insulin .1. Mechanism and the effect of water on the kinetics of degradation in lyophiles from pH 2±5 solutions. Pharm Res 13:1142±1153, 1994. 71. RKO Apenten, N Berthalon. Determination of enzyme global thermostability from equilibrium and kinetic analysis of heat inactivation. Food Chem 51:15± 20, 1994. 72. RKO Apenten. The effect of protein unfolding stability on their rates of irreversible denaturation. Food Hydrocolloids 12:1±8, 1998. 73. TP Labuza, M Saltmarch. Kinetics of browning and protein quality loss in whey powders during steady state and nonsteady state storage conditions. J Food Sci 47:92±96, 1981. 74. TP Labuza, K Bohnsack, MN Kim. Kinetics of protein quality change in egg noodles stored under constant and ¯uctuating temperatures. Cereal Chem 59:142±148, 1982. 75. FAO/WHO. Protein Quality Evaluation. FAO/WHO Nutrition Meetings Report Series 1991;51, Rome, Italy. 76. FAO/WHO/UNU. Energy and Protein Requirements. Report of a joint FAO/ WHO/UNU expert consultation, WHO Technical Report 1985;Series No. 724, Geneva, Switzerland. 77. G Sarwar, FE McDonough. Evaluation of protein digestibility±corrected amino acid score method for assessing protein quality of foods. J Assoc Off Anal Chem 73:347±356, 1993. 78. E Boutrif. Recent developments in protein quality evaluation. Food Nutr Agric 1:36±40, 1991. 79. RL Madi. Evolution of protein quality determination. Cereal Foods World 38:576±577, 1993. 80. EC Henley, JM Kuster. Protein quality evaluation by protein digestibility± corrected amino acid scoring. Food Technol 48:74±77, 1994. 81. LA Kuntz. Protein possibilities. Food Prod Design 7:76±78, 81, 85±86, 88, 90, 1997. 82. LB Mendel. Nutrition: The Chemistry of Life. New Haven, CT: Yale University Press, 1923 (cited by Ref. 76). 83. HH Mitchell, RJ Block. Some relations between the amino acid contents of proteins and their nutritive values for the rat. J Biol Chem 163:599±620, 1946. 84. D Knorr. Protein quality of the potato and potato protein concentrates. Lebensm Wiss Technol 11:109±115, 1978. 85. HR Takruri, MA Humeid, MAH Urari. Protein quality of parched immature durum wheat (frekeh). J Sci Food Agric 50:319±327, 1990. 86. PK Ghuman, SK Mann, CK Hira. Evaluation of protein quality of peanut (Arachis hypogaea) cultivars using Tetrahymena pyriformis. J Sci Food Agric 52:137±139, 1990. Corrected Amino Acid Scores 445

87. J Ruales, BM Nair. Nutritional quality of protein in quinoa (Chenopodium quinoa Wild) seeds. Plant Food Hum Nutr 42:1±11, 1992. 88. AL Sheffner, GA Eckfeldt, H Spector. The pepsin-digest-residue (PDR) amino acid index of net protein utilization. J Nutr 60:105±120, 1956. 89. WR Akeson, MA Stahmann. A pepsin pancreatin digest index of protein quality. J. Nutr 83:257±261, 1964. 90. JF Kennedy, RJ Noy, JA Stead, CA White. A new rapid enzyme digestion method for predicting in vitro protein quality (PDD index). Food Chem 32:277±295, 1989. 91. LD Satterlee, JG Kendrick, GA Miller. Rapid assays for estimating protein quality. Food Technol 31(6):78, 1977. 92. D Melnick, BL Oser, S Weiss. Rate of enzymic digestion of proteins as a factor in nutrition. Science 103:326, 1946. 93. G Sarwar. Available amino acid score for evaluating protein quality in foods. J Assoc Off Anal Chem 67:623±626, 1984. 94. G Sarwar, RW Pearce, HG Botting, D Brule. Relationship between amino acid scores and protein quality indices based on rat growth. Plant Foods Hum Nutr 39:33±44, 1989. 95. E Carnovale, E Lugaro, E Marconi. Protein quality and antinutritional factors in wild and cultivated species of Vigna spp. Plant Foods Hum Nutr 41:11±20, 1991. 96. BO Eggum, MIZ Cabrera, BB Juliano. Protein and lysine digestibility and protein quality of cooked Filipino rice diets and milled rice in growing rats. Plant Foods Hum Nutr 43:163±170, 1993. 97. CG Zarkadas, Y Ziran, RI Hamilton, PL Pattison, NGW Rose. Comparison between the protein quality of northern adapted cultivars of common maize and quality protein maize. J Agric Food Chem 43:84±93, 1995. 98. CG Zarkadas. Assessment of the protein quality of native white ¯oury maize designated IAPO-13, by amino acid analysis. J Agric Food Chem 45:1062± 1069, 1997. 99. CG Zarkadas, Y Ziran, VD Burrows. Protein quality of three new Canadian- developed naked oat cultivars using amino acid compositional data. J Agric Food Chem 43:415±421, 1995. 100. CG Zarkadas, Z Yu, VD Burrows. Assessment of the protein quality of two new Canadian-developed oat cultivars by amino acid analysis. J Agric Food Chem 43:422±428, 1995. 101. CG Zarkadas, Z Yu, GC Zarkadas, A Minero-Amador. Assessment of the protein quality of beefstock bone isolates for use as an ingredient in meat and poultry products. J Agric Food Chem 43:77±83, 1995. 102. W Wu, WP Williams, ME Kunkel, JC Acton, FB Wardlaw, Y Huang, LW Grimes. Thermal effects on in-vitro protein quality of kidney beans (Phaseolus vulgaris L.). J Food Sci 59:1187±1191, 1994. 103. ES Batterham. Availability and utilization of amino acids for growing pigs. Nutr Res Rev 5:1±18, 1992. 446 Chapter 14

104. KA Kuiken, CM Lyman. Availability of amino acids in some foods. J Nutr 36:359±368, 1948. 105. G Sarwar. Available amino acid score for evaluating protein quality of foods. J Assoc Anal Chem 67:623±626, 1984. 106. W Wu, WP Williams, ME Kunkel, JC Acton, Y Huang, FB Wardlaw, LW Grimes. True digestibility and digestibility-corrected amino acid score of red kidney beans (Phaseolus vulgaris L.). J Agric Food Chem 43:1295±1298, 1995. 107. W Wu, WP Williams, ME Kunkel, JC Acton, Y Huang, FB Wardlaw, LW Grimes. Amino acid availability and availability corrected amino acid score of red kidney beans (Phaseolus vulgaris L.). J Agric Food Chem 44:1296±1301, 1996. 108. AJ Darragh, G Schaarfsma, PJ Moughan. Impact of amino acid availability on the protein digestibility corrected amino acid score. Bull Int Dairy Fed 336(Dairy Foods Health):46±50, 1998. 109. P Rozan, R Lambhari, M Linder, C Villaume, J Fanni, M Parmentier, L Mejean. In-vivo and in-vitro digestibility of soybean, lupine and rapeseed meal proteins after various technological processes. J Agric Food Chem 45:1762± 1769, 1997. Index

AAAvÐamino acid availability, [Adulteration] 438±440 PCR analysis and, 224 AASÐamino acid score, 354±355, 433 US meat trade and, 224 Absorptivity, 171 Advanced glycation products, Bradford Accuracy, de®nition, 3 assay and, 190 Acetylacetone, Kjeldahl assay and, 22 Adulteration frequency, 240 Acid Orange 12, 125, 127±130, 136 Adverse reaction, food and, 298 feedstuffs protein and, 390 Agar gel double immunodiffusion assay, rice protein and, 153 230±246 Acid Red 1, 134±135 Agaricus, Udy assay, 160 Activation energy, browning, 427±428 Aggregated protein, Bradford assay, Actomysin, antigens and, 234 217 Acylation, 363 AGIDÐagar gel double Adjuvant, 231 immunodiffusion assay, 230±246 Adulteration, 221 Albumin, biuret assay, 48±49 commerce and, 223 Alkali treatment, digestibility and, 373 health and, 222 Alkaline copper reagent, 47 protein prices and, 223 Alkali-phenol reagent, Kjeldahl assay, economics and, 222 18 ELISA tests for, 255±257 Allergens, 299±300 Halal food and, 223 castor bean, 300 Kosher food and, 223 egg-white, 300

447 448 Index

[Allergens] Applications milk, 300 biuret assay, 57±67 peanuts, 305±312 Bradford assay, 195 shrimp, 300 Udy assay, 147 soybean, 301±305 Ara h 2, peanut allergens and, 308 wheat, 312±329 Arachin, 306 Allergy Arginine, dye binding and, 135, 139, infants and, 306 212 pregnancy and, 306 Arrhenius equation, 384, 426±427 Amaranth, 398±399 PNV and, 384 Amido Black 10B, 130±131 Assay performance, 5 Amino acid analysis (see Quantitative Association of of®cial Analytical amino acid analysis, QAAA) chemists (see AOAC) Amino acid availability (AAAv), 368, Atmospheric error, Dumas assay, 33 438 Atopic dermatitis, 305 Amino acid availability, protein quality Authenticity, proteins and, 222 and, 438±440 Autoanalyzer, 21, 16 Amino acid score, 354±355, 433 Autoclaving, effects on digestibility, 368 errors in, 355 Automated, Kjeldahl assay, 15±16 Ammonia determination, 18±23 Availability corrected amino acid score, Anaphylactic shock, 297, 305 438 Anilino-naphthalene-8-sulfonic acid, Available amino acid score (see AAS) 413 Available lysine, 356±366 ANS ¯uorescence, 413 chromatographic analysis, 360±361 Antigen comparison of assays, 365 actomysin, 234 cottonseed, 360 binding curves, 227, 261 interferences, 359, 362 boiling and ethanol resistant, 236, 239, lactalbumin, 365 240, 260 legumes, 394±395 cooked meat and, 236, 238 moisture and, 426 peptides as, 324±327 ovalbumin, 365 thermostable proteins as, 236, 261 reagents for, 356 troponin T as, 236, 238, 239 sodium borohydride and, 365±366 AOAC, Association of Of®cial temperature and, 426±427 Analytical Chemists AvCAAS (see Availability corrected digestibility, 369 amino acid score) Dumas assay, 30, 34,36 Azo dyes, 133±134,137 dye binding assay, 149, 151 gluten ELISA, 320±321, 324±325 B lymphocytes, 228±229 guidelines, rat bioassay and, Baby foods 351±353 Dumas assay, 38 immunodiffusion assay, 230 meat protein detection in, 262 Kjeldahl assay, 15 peanut allergens in, 311 factors and, 10 Baby formula, 37±38 Udy assay, 149 Bacteria, feedstuffs quality and, 388 Index 449

Baker's asthma, 312 b-Lactoglobulin, 149, 182, Barley digestibility, 309 biuret assay, 55 dye binding, 149 Dumas assay, 36 Binding constant, protein-CBBG, 177 dye binding assay, 153 Binding sites, dyes, 140 Kjeldahl analysis, 8 Binding, T-azo-R, 177±181 protein analysis, 7, 8 Bioassay, 387, 388 rapid assays for analysis, 7 chick, 354 Udy assay, 152 human, 341 BCA assay, 99±122 protein quality and, 346±354 algae protein, 121 rat 348±353 animal carcass protein, 119±121 Biological value (BV), 342 automated, 113 Bio-Rad LTD, 207 calibration features, 109 Biotin-streptavidin detection, 261±262 copper analysis, 100 2,20-Biqinoline, Lowry assay and, 76 ¯ow injection analysis, 114±116 Birds nest soup, allergy, 305 interferences, 110±111 Bitter, Bradford assay, 207 lysine and, 119 Biuret assay, 47 mechanisms, 105 barley analysis, 55 metal ion catalyzed oxidation, 107 casein determination, 49 method, 104 cereal proteins, 57 microwell plate format, 113±115 protein solubility determinations by, phenolic compounds effects, 118 63 reducing compounds and, 106 Biuret, structure, 47 sample pretreatment for, 112 Boiling and ethanol resistant antigen, serum copper, 102 236, 239, 240, 260 solid phase assay, 117 Bone meal, 271 sugars and, 103 Bovine spongiform encephalopathy (see TCA-DOC precipitation and, 112 BSE agent) BCA, derivatives, 101 Bradford assay BE antigen (boiling and ethanol resistant advantages, 169 antigen), 236, 239, 240, 260 beer protein, 204±205 Bean protein, Bradford assay, 213 beverages and, 190 Beans, PDCAAS value, 437 bread crumb, 212 Beef calibration features, 201 analysis, ORBIT, 237 carotenoids and, 189 croquettes, soy protein detection in, Chardonnay wine, 211 304 compatible solutes, 186, 187 protein, dye binding assay, 157 effect of SDS, 187 Beer interferences, 186 Bradford assay, 190, 204±206 legumes, 213 celiac disease and, 312 lipids and, 191 Dumas analysis, 30, 36 mechanisms, 172 Kjeldahl analysis, 8, 14 microassay format, 196 Lowry assay, 88 modi®ed, 196 450 Index

[Bradford assay] [Calibration] monosaccharides and, 190 Lowry assay, 72, 77±80 mungbean ¯our, 213 meat ELISA, 253 mushrooms, 216 pork ELISA, 257, 267 pinot noir, 211 soy ELISA, 284 polypeptide selectivity, 171 Udy assay, 154 polyphenols and, 209±210 Camel, ELISA, 252, 254, 255, 260 polysaccharides and, 186, 191, 199 Canned baby foods, ELISA and, 262 reagent pH, 172 Canned ®sh, ELISA, 268±269 sensitivity, 202±203 Canned foods solid phase, 185 digestibility, 370 standard method, 196 Kjeldahl analysis, 16 wavelength of, 173±175 Canned milk, 150 Bran protein, 55 Canned tuna, 268 Bread crumb, Bradford assay and, 212 Carbethoxylation, ®sh meal, 363 Bread making quality, 212±213 Carbohydrates Bread mix, gluten free, 321 available lysine determination and, Bread wheat, 57 359, 362, 366 Brest milk, 318 Lowry assay and, 81 Brewing grains, Dumas assay, 30, 33, 36 Carbonyl-amine reaction, (see also British beers, Bradford assay, 208 Maillard Reaction), 416±418 BSA binding, CBBG, 185 Caroteinoids, Bradford assay and, 189 BSE agent, 271±273 Casein bone meal and, 271 biuret assay, 49 immunological detection, 272 digestibility, 370 test kits for, 271±273 dye binding assay, 126 Buckwheat, celiacs and, 320 PER value, 349 Buffalo, antigen, 238 protein quality, 344 Buffer effects, Lowry assay, 81 Udy assay, 138, 149 Butter milk, dye binding assay, 148 Catalysts, Kjeldahl assay and, 9 BV (see Biological value) CBBC sources, Bradford assay and, 197±198 Cake ¯our, gluten free, 321 CBBG, bi-ionic form, 172 Calibration, 3±7 Celiac disease, 312±313 ammonia analysis and, 19, 20, 23 Celiac Society of Great Britain, 322 BCA assay and, 109±110, 113 Celiac-negative cereals, 323 biuret assay for rice, 58±59 Cereal products Bradford assay and, 201 Bradford assay, 212 de®nitions, 3 PNV, 398±401 Dumas assay and, 35 Cereal proteins, classi®cation, 314 dye binding assay, 129, 131, 154, 155 Cereals ELISA format and, 270 allergens, 312±329 gliadin ELISA, 323 biuret assay, 57 gluten ELISA, 323 Lowry assay, 88 horse meat ELISA, 267 Udy assay, 151 Index 451

Cerelac, 397 [Collagen] Chardonnay wine, Bradford assay, 211 quantitative amino acid analysis, Cheddar cheese, dye binding, 150 27±28 Cheese, 151 Colorimetric assay, ammonia, 18±23 Chemical deterioration, proteins, Colorimetric Kjeldahl analysis, 18±25 416±420 Combustion analysis, milk products, 37 Chemical score, 429±431 Combustion nitrogen analyzer (see egg protein and, 354 Dumas assay) Chemistry Commodities, Udy assay, 128 biuret assay, 50 Compatible solutes, Bradford assay, Lowry assay, 77 186, 187 Udy assay, 133±147 Compton-Jones scheme, CBBG, Chick bioassay, protein quality, 353±354 172±173 Chick pea, Udy assay, 153 ComputedÐPER index (see c-PER) Chicken Conarachin, 306 antigen for, 238 Condensed milk, 151 determination amino acid analysis, 27 Confectionary, gluten ELISA, 322 dye binding assay, 157, 159 Confectionary products, ELISA, 311 immunodiffusion assay, 240 Conformational stability, 425 Chlorophyll, Lowry assay and, 81±82 Conglycinin, 285±289, 292 Chocolate renaturing ef®ciency, 288 bars, ELISA, 310 Conversion factor, Kjeldahl assay, 10, gluten ELISA and, 320, 321 12 milk, dye binding assay, 148 Cooked beef, ELISA, 260 peanut allergens in, 310±311 Cooked meat Chromatographic determination, antigens for, 236, 238 available lysine, 360±361 immunoassay, 239, 247, 260, 268 Circular dichroism, 317 soy allergens and, 303 Citations off, Lowry assay, 71 Cooked poultry, ELISA, 261, 262 CNS tissue in feed, 272 Cooking oil, peanut allergens and, Cod ®llet, dye binding assay, 157 310 Codex Alimentarius Commission, Cooking, gluten analysis and, 326 allergens, 298 Coomassie Brilliant Blue G250, 169 Collaborative testing Copper analysis, BCA assay and, 101 Bradford assay, 205±206 Copper complex, Biuret assay, 50±52 Dumas assay, 30, 34, 36 Copper concentration, Lowry assay and, gluten ELISA, 322 82 ice cream analysis dye binding, 150 Copper hydroxide, biuret assay, 48 PNV, 390 Corn ¯our, gluten detection in, 322 soya protein ELISA, 282 Corn meal, BCA assay, 117 three-enzyme assay for digestibility, Corn protein, Bradford assay, 212, 213 370 Correlation coef®cient, de®nition, 5 Collagen Cortecs, 262, 271 Lowry assay, 93 gluten ELISA kit, 319 PDCAAS value, 436±437 peanut ELISA kit, 311 452 Index

[Cortecs] Detergent effect, Bradford assay, pork ELISA kit, 271 187±188 Costs Dialysis assay, digestibility, 373 Dumas assay, 32 Differential dye-binding capacity PNV tests, 429 (d DBC), 362±365 Cottage cheese, dye binding, 150 Differential scanning calorimetry (see Cottonseed, available lysine, 360 DSC) Cowpea, Udy assay, 154±155 Dif®culties, Udy assay, 130 c-PER, 433 Digestibility Creutzfeldt-Jacob disease, 271 de®nitions of, 367±368 Cross reactivity (see Speci®city) peanut allergens, 309±310 Cross-linking, 418±419 three enzyme method for, 369±372 Crystal violet, protein binding and, Dimethyl sulfoxide and Lowry assay, 82 172 Diphtheria toxin, 231 Crystalline-amorphous transition, 425 Disul®de bonding Cyoprotectants, Lowry assay and, 82 gluten and, 315 Cysteine, racemization, 418 in soybean protein, 285, 288 Donkey, ELISA, 256, 262 Dairy products Double immunodiffusion, 230 Dumas assay, 30 Dried milk, 344 dye binding, 147 Kjeldahl assay, 13 Dairy proteins, Biuret assay, 63 PNV for, 382±384 DBC (see Dye binding capacity) DRIFT, 235 d DBC, differential dye-binding capacity DSC, soy protein, 286 available lysine determination and, DSC2, Tg measurements, 422 362±365 Dumas assay, 29 legumes and, 364 Advantages of, 31, 36 soybean meal and, 363 atmospheric error and, 33 total lysine and, 364 beer protein, 30, 36 Deer analysis, DRIFT, 237 feedstuffs, 34 Defatting, meat, 61 instrumentation, 31±33 Degree of hydrolysis, 373 ketchup, 30 Dehulling, PNV and 395 malt, 30 Dehydrogenation, Lowry assay, 75 materials, 32 Denaturation, 414±419 milk products, 37 meat antigens, 263±265 oilseed protein, 33 peanut allergens, 308±309 semolina, 38 processing and, 415 Dye binding assay Denaturation temperature (TD), soy beef protein, 157 proteins, 286±287 calibration with Kjeldahl, 150, Denaturation-renaturation, soy 152±155 antigens, 284 chicken meat, 157, 159 Denaturing agents, 414 cod ®llet, 157 Dent CO251 (see also Maize), 435 Dye binding lysine, 362 Design, Lowry assay, 69 Dye equivalent weight, 129 Index 453

Dye purity, Bradford assay and, [ELISA] 197±198, 200 analysis, 256, 257 pyruvate kinase as antigen, 258 Effect of dye volume, Bradford, 201 raw meat, 252, 255 Egg albumin, 344 red snapper, 270 biuret assay, 48, 63 reviews, 250 Bradford assay, 217 rock shrimp, 269±270 Egg allergens, 305, 309 sardine, 268 Egg products, dye binding assay, 159 sea food, 268±270 Egg protein, 290, 292, 346, 347, 392 sheep, 256 Udy assay, 159 soy protein, 281±296 Egg ratio, 354 ELISA test kit Egg white allergens, 300 BSE, 273 Egg yolk, Bradford assay and, 201 deer meat, 262 Egyptian legumes, PNV, 394 gluten, 323 Electrophoresis (see also SDS-PAGE) meat, 262 CBBG, 170 allergens, 310 samples, Bradford assay, 198, 199 peanut, 311 ELISA pork, 262 a¯atoxins, 251 soybean, 283±284 bacterial toxins, 25 tuna, 268 boiling resistant antigen, 260 working range, 254, 320 BSE agent, 271±273 Emulsion proteins, Lowry assay, 90 buffalo, 256 End-point temperature, meat ELISA, meat, 260 265 cattle, 256 End-point temperatures, Biuret assay, commercial, 282 62 competitive, 248 Enzyme linked immunosorbent assay confectionary products, 311 (see ELISA) cooked beef, 262 Enzyme, protein digestibility and, 366 meat, 260±265 Equations, Udy assay, 141±142 pork, 262 Equilibrium, Udy assay, 138, 156 poultry, 262 Errors enzyme substrates, 254 chemical score, 355 format, 248±250, 255 PER determination, 350 game meat, 260 Essential amino acids, 355 gliadin, 317±329 Ethyl vinyl sulfone (EVS), 356 gluten, 317±329 Evaporated milk, protein analysis, horse meat, 252±256, 258, 260±263, 149 266±268 muscle antigen for, 257±260 FAO/WHO/UNU, 433 Ochratoxin, 251 Fatalities, peanut allergy, 306 performance characteristics, 270 FDA, 298 peroxidase assay and, 254 FDNB, 1-Fluoro-2,4-dinitrobenzene pork, 260 assay, 356 454 Index

[FDNB] Fruit juice, Bradford assay, 211 available lysine determination, 356±359 Game meat, ELISA, 260 correction factor, 359 Gel immunodiffusion assay, cooked Fecal nitrogen, 367, 368 meat, 239±241 Feed supplements, 386 Gelatin Feedstuffs and concentrates, 386±393 BCA assay, 105 Feedstuffs, 387 Biuret assay, 49, 56 Dumas assay, 30, 33, 34, 35 complexes with Amido Black 10B, 147 dye binding assay, 156, 390 feedstuffs and, 156 Kjeldahl analysis, 16 Lowry assay, 93 Orange G binding, 388±390 Gelatin granules, 126 PNV evaluation, 346, 387±393 Gel-®ltration, beer, 206 Udy assay, 156 Glass transition temperature (see also

Fermented soybean products, Tg), 420±425 allergenicity, 304 Gliadin, 314±317 Fertilizer, Dumas assay, 30 in human milk, 318 FIA, BCA assay using, 114 Tg value for, 424 Ficol, Lowry assay, 82, 83 Gluten, 315 Fish, allergens, 298 available lysine, 360 Fish and seafood identi®cation, 268±271 confectionary products and, 322 Fish gelatin, 311 ELISA, 317±329 Fish meal, 387, 388, 390 home test, 323±324 Acid Orange 12 binding, 390 preparation, 315 DBC, 389 standards for ELISA, 321

Kjeldahl analysis and, 15 Tg value for, 421 Udy assay, 138, 156 Gluten-free foods, 313, 318, 322, 323 Fish protein, Tg, 423 Gly m Bd 30k, as soybean allergen, Fish protein concentrate, 268 301±304 Fish sausages, soy protein detection in, Glycation, Bradford assay, 190 304 Glycinin, 285±289 Flint CO255 (see Maize), 435 Gordon-Taylor equation, 422±425 Flow injection analysis (see FIA) GPV (Gross protein value), 342 Flower bud protein, quantitative amino Grain and cereal, Kjeldahl assay, 13 acid analysis, 27 Grains, Dumas assay, 33 1-Fluoro-2,4-dinitrobenzene (see Grape juice, Bradford assay, 210 FDNB) Graphical analysis, Udy assay, 144±145 Folin-Ciocalteu reagent, 71 Gross protein value (see GPV) Food allergy, 297 Ground rice, biuret assay, 59 intolerance, 297 matrix, protein quality and, 420±425 Half and half milk, protein assay, 149 Foods, Dumas assay, 30 Hamburger, meat identi®cation and, 235 Freeze dried proteins, 425 Heat effects, soy antigens, 287 Fringe-micelle model, 421 Heat resistance, conglycinin, 288 Frozen dessert, 151 Histidine, dye binding and, 135, 139 Index 455

Home test, gluten, 323±324 [Infant formula] Honey Dumas assay, 37±38 Bradford assay, 217 peanut protein detection in, 306 Kjeldahl analysis, 217 PER, 350 Horse PNV bioassay, 385 antigen production using, 231 rat bioassay, 385 ELISA, 267 Infrared analysis, cereal proteins, 7 immunoassay, 234 Insect protein, Bradford assay, 199 immunodiffusion assay, 239 Insoluble protein, Bradford assay, 217 meat, 252±256, 258, 260±263, 266±268 Instant breakfasts, Kjeldahl assay meat protein, immunoassay, 252±256, analysis, 7 258, 260±263, 266±268 Interferences muscle antigen, identity, 258, 261 available lysine, 362, 363, 366 muscle protease, biuret assay and, 62 BCA assay list for, 111 protein assay, 235 biuret assay, 53±55 Horseradish peroxidase, 253±254 Bradford assay, 186±191 Human milk, gliadin detection in, 318 d DBC, 363 Hybridoma, 229 Lowry assay, 80 Hydrogen peroxide ninhydrin assay, 23 Biuret assay, 55 plant dyes as, 57, 58, 59 Kjeldahl assay and, 9 Udy assay, 147 Hypoallergenic bread, 316 Isobestic point, 140, 172 IVPD (see In vitro digestibility) Ice cream dye binding, 148 protein, 150±157 Kangaroo meat, immunoassay, 234, 235 Udy assay, 148, 150±151 Ketchup, Dumas assay, 30 Immunization schedule, 231, 233 Kinetics, 395 Immunoabsorption, 232 Lowry assay, 77 Immunoassay, principles, 226 proteolysis, 412±414 Immunoblotting, 299±300, 307 Kiwi fruit juice, Bradford assay, 211, Immunodiffusion assay 212 chicken meat, 240 Kjeldahl analysis, 1, 7±25, 344 meat, 230 barley, 8 Immunological assays, disadvantages, beer, 8, 14 250 catalysts, 9 Immunology, 228 colorimetric, acetylacetone and, 22 Imperial Chemical Industries (ICI), 169 comparison with dye binding assay, In vitro digestibility, 368±374 150, 152, 153 protein stability and, 413 honey, 217 protein structure and, 413 gliadin, 327 Incidental additives, 298 Kjeldahl factor and, 10, 12 Indanetrione assay, Kjeldahl assay, 23 nitrogen-protein conversion factor, Indophenol reagent, Kjeldahl assay, 18 10, 12 Infant formula, 384±386 PER values and, 349 amino acid supplementation, 386 PNV determination, 349 456 Index

[Kjeldahl analysis] [Lowry assay] reactions of, 9 cereals, 88 reliability, 7 chlorophyll and, 81±82 sausages, 17 citations of, 71 Kjel-foss, Instrument, 15 cryoprotectants and, 82 design, 69 Labeling, food allergens, 297 emulsion proteins, 90 Lactalbumin, PER value, 349 leaf protein, 90 Lactose, protein quality and, 382, legumes, 88 385 mechanisms of, 77 Larger, Bradford assay, 208 nondairy creamers, 90 Leaf protein, Lowry assay, 88, 90 sample pretreatments for, 86±87 LECO FP-2000, Dumas assay, 31 single cell protein, 91 LECO FP-228, Dumas assay, 33 Lysine LECO FP-428, Dumas assay, 31±32 BCA assay and, 119 Legumes bioassay and, 344 available lysine, 394±395 CBBG binding and, 171 baking and PNV of, 395 determination, 356±366 Bradford assay, 213 dye binding and, 135, 138 dehulling and PNV, 395 glycation and Bradford assay, 190 dye biding, 153±156 loss, Arrhenius plot 385 ELISA and, 312 Lowry assay, 88 Maillard reaction, 382, 416±418 PNV, 393±398 available lysine, 366 roasting and PNV, 396 pH and, 417 sprouting and PNV, 395 Maize steaming and hydrothermal available lysine, 360 treatment, 395 cultivars, 435 Lesions, celiac disease, 312±313 Dumas analysis, 29 Linear dynamic range, 5 PNV, 399 Linear free energy relations, amino Malt acids, 418 Dumas assay, 30, 36 Linearity, Bradford assay, 201 dye biding, 152 Lipid Materials, Dumas assay, 32 Bradford assay and, 191, 200 Matrix effects, protein deterioration interferences by, 84 and, 420 Lower limit of detection (LLD) Mean residue weight, tables of values, 27 de®nition, 3, 6 Meat, 247±281, 284, 303 ELISA, 265 analysis for soybean, 285 gluten ELISA, 320, 324 antigenic components, 232, 251 soy bean ELISA, 284, 290 BCA assay and, 119±121 Udy assay, 152 biuret assay, 61±63 Lowry assay Dumas analysis, 29 buffers and, 81 dye binding assay, 157 carbohydrates and, 81 gel immunodiffusion assay, 230 Index 457

[Meat] Moist heating, allergen stability and, quantitative amino acid analysis, 27 308 sample pretreatment for ELISA, 284 Moisture soy protein detection in, 303 food deterioration and, 416 Udy assay, 156±157 peanut allergens and, 308 Meat analogue, Kjeldahl analysis, 16 PER values and, 350 Meat and bone meal, PNV, 393 PNV and, 383±384 Meat and meat products, Kjeldahl rapeseed heat damage and, 392 analysis, 15 Moisture-temperature-time relations, Meat antigen, 231 411 Meat rendering plants, 272 Molten globule and digestibility, Meat speciation, 247±280 412±415 Meatballs, soy protein detection in, 304 Monoclonal antibodies Mechanism development for ELISA, 265±268 biuret assay, 50 gluten ELISA and, 325±327 Lowry assay, 73, 77 meat analysis and, 265±267 Udy assay, 133±147 pork ELISA and, 267 Melanoma, 229 soy bean ELISA and, 290±292 Metachromasia, Crystal violet, 174 turkey analysis, 266 Metal ion catalyzed oxidation, Lowry Monosaccharides, Bradford assay and, assay, 73 190 Michaelis-Menten kinetics, 414 Multiple processing, legumes, 397 Microassay, Bradford, 196 Mungbean, Bradford assay, 213 Microbiuret assay, 56 PNV, 394 Micro-Kjeldahl analysis, 21, 13 Muscle lactate dehydrogenase, ELISA, Microwave heating, 396 264, 265 Microwell plate, BCA assay using, 113 Mushrooms Milk, 282 Bradford assay, 216 allergens, 298, 300 Kjeldahl analysis, 160 ELISA, 282 quantitative amino acid analysis, 27 powder, 381 Udy assay, 160 powder, PER, 350 Myoglobin, thermostable antigen and, prices and protein content, 125 238, 239 products, Dumas assay, 37 protein, dye binding assay, 147, 148 Naphthylamine brown, 126 protein, ELISA plate coating, 248, NB, nitrogen balance, 342, 346±347, 367 272 NDI, nitrogen digestibility index, 373 protein, Udy assay, 147±151 Nessler's reagent, Kjeldahl method and, soymilk detection in, 281 21 Millet Net protein utilization (NPU), 342 PNV, 399±400 Ninhydrin reagent, Kjeldahl method protein, biuret assay, 60 and, 23±24 Mince, soy protein detection in, 284 Nitrogen balance (NB), 342, 346±347, Model wine solution, Bradford assay, 367 211 Nitrogen digestibility index (NDI), 373 458 Index

Nondairy creamers, Lowry assay, 90 Peanut allergy, 305±312 Nonanimal protein ingredients, 281 average fatalities, 306 Nonfat milk, 149 in infants, 306 NPU, net protein utilization, 342 in pregnancy, 306 Nucleic acid, Bradford assay and, 191 Peanut protein, in confectionary, 311 Nuts Peanuts, chemical score, 429 allergens and, 300±301 Pepsin digest dialysate index (PDD ELISA, 310±312 index), 432 digest residue index (PDR index), 432 Oats pancreatin digestion index (PPD Ac Hill, 436 index), 432 Ac Lotta, 436 Peptide antigens Ac Percy 436 gluten ELISA and, 324±327 biuret assay, 60 soybean ELISA and, 290 PDCAAS value, 436 Peptidyl sites, digestibility and, 414 Oilseeds, Dumas assay, 30, 33 PER (Protein ef®ciency ratio), 342, OPA (see o-Phthaldehyde) 348±351 o-Phthaldehyde, 356 AOAC guidelines, 351 Orange G complex foods and, 350 binding, PNV, feedstuffs, 389±390 rat acclimation period, 350 structure of, 136 sweetened foods and, 350 Udy assay using, 130 Perilla, 396 ORBIT, 235 Peterson's Lowry assay, 72 Ovalbumin, 182 Phenol red, 125 Phenol pAb, for detection of donkey muscle Bradford assay and, 188 protein, 256, 262 BCA assay and, 119 PAGE, 81, 402 Phospholipids, Lowry assay, 84 Pavalbumin, allegen and, 300 Phosphothreonine, racemization, PDCAAS, 411, 427, 429, 433±438 418 applications, 434±439 Pierce Warriner LTD, 207 beans, 437 Pinot Noir, Bradford assay, 211 bene®ts, 434 Pioneer 3953 (see Maize), 435 calculation, 434 Plant colors, Biuret assay and, 55 collagen, 436±437 Plant metabolites, Bradford assay, maize, 434 188±189 oats, 436 Plant proteins, 221 rice, 434 Amido Black 10B binding, 130 Vigna, 434 Plasticizer, water as, 420 PDR index (see Pepsin, digest residue PNV (protein nutrient value), 341±380, index) 411 Pea globulin, ELISA, 290, 291 amino acid losses and, 344±345 Peanut allergens, 300, 306±308 animal tests, 348±354 cooking oil and, 310 assessment by PDCAAS, 434 preparation, 307 bioassays, 346±354 Index 459

[PNV (protein nutrient value)] [Potato protein] consumer factors and, 343 Lowry assay, 88 factors affecting, 343 Poultry analysis, PROFIT, 237 food labeling and, 344 Poultry, ELISA for, 266±267, 268 human assays and, 346±348 PPPD index, 432 hygiene and, 343 Precision, protein assays and, 5 indicators for, 342 Preservatives, milk, 148 infant formulas and, 384 Prionics AG, 273 insect infestation and, 400 Procedure Kjeldahl analysis and, 349 biuret assay, 49 legumes, 393±398 Bradford assay, 195 literature, 345, 348±349 Lowry assay, 70 meat and bone meal, 393 PDCAAS and, 433±434 pregnant and lactating women and, soy protein antigen preparation, 284 344 Udy assay, 128 processing and, 381±402 Processed foods, 2 quality control and, 344 digestibility, 370±371 rat bioassay, 348 gluten detection in, 310, 320 soaking and, 395 interferences in, 80 Polyacrylamide gels, CBBG stain, 170 Kjeldahl analysis, 16 Polyamino acids, dye binding and, 171, PROFIT, 235 181, 183 Pro-meter instrument, Udy assay, 153 Polyclonal antibody production, 232 Propionic anhydride, acylation, 363 Polyethylene glycol, 132 Protein Polymer chemistry, 420 analysis, signi®cance, 2 Polymer science approach, 424 assay, characteristics, 2 Polymerase chain reaction, 224 binding, CBBG, 173, 175, 177 Polyols, 133 binding, dyes, 135 Polypeptide selectivity, Bradford assay concentrates, 386±393 and, 171 digestibility, 366, 411 Polyphenols, Bradford assay and, digestibility corrected amino acid 209±210 score, (see PDCAAS) Polysaccharide interference, Bradford ef®ciency ratio (see PER) assay, 186, 191, 199 haze, 209 Polyvalent antigen, 229 immunoassay, 225 Porcine rind protein, 27 M, shrimp antigen, 269 Pork molecular weights, 57 analysis, PRIME, 237 nutrient value (see PNV) dye binding assay, 157 precipitation, Bradford assay, 199 ELISA, 267±268 carbonyl-amine reaction and, thermostable antigen, 238 417±418 Potato protein factors in¯uencing, 343 Bradford assay, 214±216 milk, 381 chemical score, 429 segmental mobility, digestibility and, Kjeldahl analysis, 13 413 460 Index

[Protein] Reverse phase high pressure stain, CBBR, 170 chromatography (see RP-HPLC) Protein-protein-variation, Reviews BCA assay, 105 celiac disease, 313 Bradford assay, 197, 201, 212 dye binding, 126±127 Protein-quinone interactions, 210 peanut allergy, 306 Proteolysis, Lowry assay, 90±91 protein quality, 345 Puri®cation, Coomassie Brilliant Blue, Rice, 402, 435 200 beans, protein, 395 protein QPM (see Maize), 435 biuret assay, 59 QSDS-PAGE, 7 Udy assay, 153 Quantitative amino acid analysis, pork PDCAAS value, 435 rind, 27 PNV and, 399 Quantitative sodium dodecylsulfate RNV, 342 polyacrylamide gel Rock shrimp, 260±270 electrophoresis (see QSDS- RP-HPLC PAGE) available lysine, 361 gliadin analysis, 327 Racemization RPV, 342 amino acids and, 418±419 cysteine, 418 Salt effects, Bradford assay, 201 factors controlling, 418±419 Sample calibration, 4, 7 rate prediction, 418 Sample pelleting, Dumas assay, 33 Radcliff In®rmary, 322 Sample pretreatment Radioallergo absorbent test (see RAST) beer, Bradford assay, 205 Rapeseed cereals, 55 heat damage, 392 corn meal, BCA assay 117 meal, PNV, 390±392 Lowry assay and, 86±87 protein, Udy assay, 154±155 meat, 261 RAST, 299 Biuret analysis, 61 Rat bioassay, 411 ELISA, 285 Raw meat, ELISA, 252, 255 Plasma, 49 R-Biopharm, ELISA test kit, gluten, soy bean product, 284 320 Sardine, 268 Reactive lysine (see also Available Sausages lysine), 356 dye binding, 159 Red snapper, 270 gluten and, 313, 321 Red wheat, Udy assay, 153 Immunodiffusion assay, 239 Relative nutritive value (see RNV) Kjeldahl analysis, 17 Relative protein value (see RPV) soy protein, 284, 289, 290 Reliability Udy assay, 159 Kjeldahl assay, 7 Scenario, dye binding, 141±143 Udy assay, 152 Schecter and Berger scheme, 414 Renaturing buffer, 284 Screening, high-lysine cereals, 401±402 Index 461

SDS, Bradford assay, 187 Sorghum, 399±400 SDS-PAGE, 300, 307, 309 biuret assay, 60 antigen analysis, 237 Soup, gluten detection in, 322 peanut allergens, 309 Soy wheat allergen, 315±317 bean allergens, analysis, 303±305 Seal meat, immunodiffusion, 239 ELISA Secondary structure, Gliadin, 317 cooking and, 283 Seed globulins, ELISA and, 291 speci®city, 283, 291 Seharawi, celiac disease and, 312 ¯our, Udy assay, 153 Semolina, Dumas assay, 38 products, ELISA, 289 Sensitivity, 1, 5 protein, 285±286 BCA assay of copper, 102 in sausages, 289 biuret assay, 55, 56 thermal denaturation, 286±287 Bradford assay, 177, 179, 183±185, protein antigen, 284, 285 197, 201±204 Soybean colorimetric Kjeldahl assay, 18 ELISA, 281±296 dye binding, 131,132, 133 7S protein, 285±289 gluten ELISA, 329 Udy assay, 138 immunoassay, 227 Soybean meal, Orange G binding, 389 Lowry assay for wheat protein, 89 Soybean protein, available lysine, 360 meat ELISA, 260 Soymilk, 282 protein analysis, 1, 2, 5±6 Species identi®cation ®eld test (see Udy assay, 137 SIFTS), 235 Serine, racemization, 418 Speci®city Sesame ¯our, Udy assay, 155 de®nition, 3 Sheep antigen, 231 gluten, ELISA, 318, 325, 326 Sheep, ELISA, 249, 252, 255±256, 257, meat ELISA, 256, 263 260, 261, 262, 268 peanut ELISA, 311 Sheep serum albumin (SSA), 253 soya bean ELISA, 283 Shrimps, 269, 300, 301 Spray dried milk, 382±383 Single cell protein, Lowry assay, 91 dye binding, 129, 130,147 Skimmed milk PNV, 382 dye binding assay, 151 Udy assay, 129 Kjeldahl factor for 12 Sprouted cereals, protein quality, 345 Slope assay, protein quality, 347 St. Bartholomew's Hospital, 322 Small animal bioassay, 348 St. James University Hospital, 322 Soaking, PNV and, 395 Standard assay, Bradford, 196 Sodium borohydride, available lysine, Starch interferences, 55 365±366 Statistical principles, 4 Sodium dodecyl sulfate (see SDS) Steaming, PNV and, 395 Solid phase assay Stiffness parameter, 423 Bradford assay, 185 Stout, Bradford assay 207 Udy assay, 131±133, 143 Structure Solubility relations, Udy assay, 143 Acid Orange 12 dye, 136 Soluble complexes, Udy assay, 138±140 Acid Red 1, 135 462 Index

[Structure] [Thermostable antigen] Amido Black 10B, 137 soybean and, 289±292 bicinchoninic acid, 100 Three enzyme assay, in vitro biuret, 47 digestibility, 369±372 CBBG, 170 TNBS assay, 119, 356, 361±362, 365, CBBR, 170 369 copper-biuret complex, 47 Tomato protein, Dumas assay, 30 dehydropeptide, 75 Tomato seed, Lowry assay, 88 iminopeptide, 75 Tortillas, 399 Orange G, 136 Total digestibility co-ef®cient (see T-azo-R, 135 TDC) Sudanese legumes, PNV, 394 TPD (see True protein digestibility) Sulfhydryl compounds Trace allergens, 297 BCA assay and, 118 Trinitrobenzene Sulfonic acid, (see Lowry assay and, 84±85 TNBS) Sulfhydryl group, dye binding, 126 Troponin, meat antigens and, 236, Sulfhydryl-disul®de exchange, 317, 415 238±239 Symptoms Troponin T, ELISA antigens and, 261 celiac disease, 312±313 True protein digestibility, 367, 433 food allergy, 298 Trypsin inhibitors, protein digestibility, peanut allergy, 305 372 Tuna, 268 T-azo-R, 134 PER value for, 352 TCA-DOC precipitation soy protein in, 305 Bradford assay and, 118, 195, 199±200 Turkey sausages, 239 Lowry assay and, 72 Two-state denaturation, 412, 414±415, TDC (total digestibility coef®cient), 342 425 Tectator heating block, 21 Tyrosinase, copper detection in, 99 Test kits, species identi®cation and, 235±236 Udy assay, 125±126, 138 Tg (glass transition temperature), barley, 152 420±425 cereal proteins, 151 ®sh protein hydrolysate, 423 dif®culties, 130 gluten, 421 dye-protein solubility and, 143 moisture and, 422 equations for, 141±142 protein quality and, 420±425 ®sh meal, 138, 156 proteins and peptides, 421 mechanisms of, 133±137 Theory (see Mechanism) milk protein, 147±151 Thermal reliability, 152 conductivity detectors, 30 sausages, 159 denaturation, peanut allergen, various commodities, 128 308±309 UHT milk, protein quality and, 382 stability, meat antigens, 263±265 Ultra®ltration, 211 Thermostable antigen, 238 Uncooked meat, ELISA, 254±257 meat and, 261 immunoassay, 234 Index 463

Water [Wheat] ammonia determination in, 19±20 biuret assay, 53, 57±60 PER and, 350, 352 dye binding assay, 151 Water activity, 383±384 reliability of analysis, 7 Maillard reaction and, 383±384, 417 Udy assay, 151 Water holding, 281 Whey protein, Bradford assay, 217 Water supply, Kjel-Foss instrument and, Whole milk, dye binding, 149 15 William-Landel-Ferry kinetics, 420 Weaning foods, PNV, 397±398 Wine, Bradford assay, 190, 209±212 Western transfer, 273 Whale meal, 387 Yeast protein, Lowry assay, Wheat, 345 63±64, 91 allergens of, 314±317 allergy, 312±313 Zein, biuret assay, 49