INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Infonnation Company 300 North Zed) Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600
CAROTENE THERMAL DEGRADATION PRODUCTS AND THEIR EFFECTS ON THE OXIDATIVE STABILITY OF SOYBEAN OIL
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Donald Frank Steenson, M.S.
*****
The Ohio State University 1999
Dissertation Committee: Approved by Professor David B. Min, Advisor
Professor Steven J. Schwartz Advisor Professor Grady W. Chism III Food Science and Nutrition Professor Sheryl A. Barringer Graduate Program ÜMI Number: 9931679
UMI Microform 9931679 Copyright 1999, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized copying under Title 17, United States Code.
UMI 300 North Zeeb Road Ann Arbor, MI 48103 ABSTRACT
Soybean oil used in processed foods is susceptible to lipid oxidation. Carotenes are utilized as colorants in processed foods containing soybean oil. Though carotenes are easily degraded during thermal processing, little is known regarding the effects of thermally degraded carotenes on the oxidative stability of soybean oil.
Thermally degraded (3-carotene or lycopene solutions were added to soybean oil samples at concentrations of 0, 25,45, and 50 ppm, respectively. Each sample, as well as controls containing 5, 25, and 50 ppm o f unheated (3-carotene or lycopene, contained 3 ppm of chlorophyll to allow photosensitized singlet oxygen oxidation to occur. The vial containing each sample was sealed airtight and stored either at 25°C in a light box (1650 lumens) or in a dark oven at 60°C. The oxidative stability of each soybean oil sample was determined by measuring (every 4 hours for 24 hours) peroxide value and headspace oxygen depletion by thermal conductivity gas chromatography. Oxidative stability was further identified by first utilizing solid phase microextraction (SPME) fibers to adsorb volatile oxidation products in the headspace of the samples, then quantifying and comparing the volatiles’ total peak area from their respective gas chromatogram.
Soybean oil samples containing 50 ppm degraded (3-carotene displayed 11.5% higher peroxide values (under light) as well as higher headspace oxygen depletion values (in the dark) when compared with controls (p<0.01). Lycopene degradation products (50 ppm)
in soybean oil decreased peroxide values up to 10.5% under light, and significantly
decreased headspace oxygen depletion of samples in the dark (p<0.05). Over all
concentration ranges, headspace oxygen depletion values for samples stored under light
containing either P-carotene or lycopene degradation products did not differ significantly
from controls. After 30 days of storage in the dark at 60°C, samples containing 50 ppm
degraded p-carotene displayed a significantly higher (p<0.05) SPME-GC total volatile
peak area when compared with controls containing 50 ppm düLl-trans P-carotene. Under
similar conditions, samples containing 50 ppm degraded lycopene displayed a
significantly lower (p<0.05) SPME-GC total volatile peak area when compared with
controls containing only oil.
These results indicate that thermally degraded P-carotene can act as a prooxidant in
soybean oil exposed to elevated temperatures, which may cause a decrease in the
oxidative stability of thermally processed foods containing soybean oil. Thermally
degraded lycopene, however, may act as an antioxidant in soybean oil exposed to
elevated temperatures and therefore may actually increase the oxidative stability o f food systems containing soybean oil.
Ill Dedicated to my wife and family.
IV ACKNOWLEDGMENTS
I would like to thank my advisor. Dr. David B. Min, for his unequaled enthusiasm and dedication in helping me to achieve at the highest level possible in all my scholastic and research endeavors. To my dissertation committee. Dr. Grady W. Chism, Dr. Steven
J. Schwartz, and Dr. Sheryl A. Barringer: I would like to extend my sincere appreciation for the constructive criticism, comments, and suggestions afforded to me throughout my degree program. A special thanks goes out to Cathy Zirkle, A1 McRoberts, Carol Rogers,
Ed Zirkle, and Linda Burianek, members of the department support staff who provided a wide range of invaluable assistance and truly helped get ‘the necessary things’ done. I deeply appreciate the assistance with the GC-MS analyses that I received from Susan
Hatcher. I am of course grateful to all my friends and fellow colleagues in the department for their never-ending support and encouragement throughout the day-to-day nuances o f graduate school life, especially Sherri, Monica, Cindy, Kwok-Man, Steve,
Sarit, Joe, Julie, Minhthy, Mario, Keith, and Sinan. VITA
August 15,1970 Bom — Louisville, Kentucky
1992 B.S. Biology Bucknell University Lewisburg, PA
1994 M.S. Foods and Nutrition The Florida State University Tallahassee, FL
1994 - 1995 ...... Quality Control Chemist Mulberry Ethanol, L.P. Mulberry, FL
1995 - 1996 Laboratory Scientist P. E. LaMoreaux & Associates Lakeland, FL
1996 - 1997 ...... Graduate Research Assistant Dept, of Food Science and Technology The Ohio State University Columbus, OH
1997-1999 USD A Fellow Dept, of Food Science and Technology The Ohio State University Columbus, OH
VI PUBLICATIONS
1. Steenson, D. F., and Sathe, S. K. 1995. Characterization and digestibility of Basmati rice {Oryza sativa L. var. Dehraduni) storage proteins. Cereal Chemistry. 72 (3): 275- 280
2. Min, D. B., and Steenson, D. F. Crude fat analysis. In Food Analysis, 2"^* edition, Nielsen, S. S., ed. Gaithersburg: Aspen Publishers, Inc., 1998, pp. 201-215.
FIELDS OF STUDY
Major Field: Food Science and Technology
V ll TABLE OF CONTENTS
Page
Abstract...... ii
Dedication...... iv
Acknowledgments ...... v
Vita...... vi
List o f Tables...... xiii
List o f Figures ...... xv
Chapter 1 Introduction...... 1
Chapter 2 Literature Review...... 5 2.1 Vegetable oils ...... 5 2.1.1 Production and utilization...... 5 2.1.2 Soybean oil instability ...... 7 2.1.3 Free radical autoxidation...... 8 2.1.3.1 Prooxidants in oils ...... 9 2.1.3.2 Free radical scavengers...... 10 2.1.4 Photosensitized oxidation ...... 11 2.1.4.1 Type 1 and H processes ...... 12 2.1.4.2 Sensitizers in o ils ...... 14 2.2 Volatile soybean oil oxidation products ...... 15 2.2.1 Formation of volatile compounds ...... 17 2.2.2 Isolation o f volatile compounds ...... 19 2.2.2.1 Direct injection...... 19
viii 2.22.2 Static headspace analysis...... 20 2.2.2.S Dynamic headspace analysis ...... 21 2.2.2.4 Solid phase microextraction ...... 22 2.2.3 Correlation with flavor quality ...... 27 2.3 The chemistry of oxygen ...... 28 2.3.1 Nature of triplet and singlet oxygen ...... 28 2.3.1.1 Reaction rates...... 29 2.3.2 Singlet oxygen generation ...... 31 2.3.3 Singlet oxygen oxidation detection ...... 32 2.3.3.1 Headspace oxygen ...... 33 2.3.3.2 Peroxide value ...... 33 2.3.3.3 Gas chromatographic reactor...... 34 2.3.3.4 Photodiode detector ...... 34 2.4 Singlet oxygen-catalyzed oxidation ...... 35 2.4.1 Oxidative products in o il ...... 35 2.4.2 Singlet oxygen initiated biological damage...... 37 2.5 Quenching of singlet oxygen ...... 37 2.5.1 Effects of tocopherols...... 39 2.5.2 Effects of carotenoids ...... 41 2.5.3 Mechanisms and kinetics...... 45 2.6 Carotenoid chemistry ...... 49 2.6.1 Structure...... 50 2.6.1.1 Chromophore ...... 50 2.6.1.2 Carbon skeleton...... 52 2.6.2 Classification...... 53 2.6.2.1 Carotenes...... 53 2.6.2.2 Oxycarotenoids ...... 54 2.6.2.3 Apocarotenoids ...... 54 2.6.3 Stereochemistry ...... 54 2.7 Carotenoid distribution and function...... 58
IX 2.7.1 In nature...... 58 2.7.1.1 Plant contents...... 58 2.7.1.2 Animal contents...... 62 2.7.2 In processed foods ...... 63 2.7.3 As colorants and additives...... 64 2.8 Roles of carotenoids in human nutrition ...... 66 2.8.1 Metabolism ...... 67 2.8.1.1 Absorption ...... 68 2.8.1.2 Tissue distribution ...... 70 2.8.2 Vitamin A precursors ...... 72 2.8.3 Additional health effects ...... 74 2.8.3.1 Antioxidant protection ...... 75 2.8.3.2 Cancer prevention ...... 76 2.8.3.3 Cardiovascular disease...... 78 2.9 Carotenoid oxidation/degradation ...... 78 2.9.1 Effects of oxidizing agents ...... 78 2.9.2 Effects of thermal processes ...... 82 2.9.3 Effects of storage ...... 87 2.9.4 Effects of encapsulation in minimizing degradation ...... 88 2.10 Isolation and identification of carotenoids...... 89 2.10.1 Extraction procedures ...... 89 2.10.1.1 Solvents...... 90 2.10.1.2 Supercritical fluids...... 92 2.10.2 Quantitation...... 93 2.10.2.1 UV/Vis spectroscopy ...... 93 2.10.2.2 HPLC-PDA...... 94 2.10.3 Separation...... 94 2.10.3.1 HPLC...... 95 2.10.3.2 TLC...... 102 2.10.4 Structure elucidation...... 104 2.10.4.1 UV/Vis spectroscopy ...... 104 2.10.4.2 IR spectroscopy ...... 105 2.10.4.3 Mass spectrometry ...... 109 2.10.4.4 NMR spectroscopy ...... 112
Chapter 3 Materials and Methods...... 115 3.1 Materials...... 115 3.2 Lycopene extraction ...... 116 3.3 Thermal treatment of carotenes...... 116 3.3.1 B-carotene and lycopene ...... 116 3.3.2 Crude lycopene extract ...... 117 3.4 Sample preparation and storage for the chlorophyll-photosensitized singlet oxygen oxidation of soybean oil ...... 119 3.5 Sample preparation and storage for the thermally-induced oxidation of soybean o il ...... 123 3.6 Headspace oxygen analysis ...... 123 3.7 Solid phase microextraction (SPME) volatiles analysis ...... 124 3.8 Analysis o f carotene degradation products ...... 126 3.8.1 Separation ...... 126 3.9 Statistical analysis ...... 127
Chapter 4 Results and Discussion...... 128 4.1 Separation and identification of carotenes and their thermal degradation products...... 128 4.2 Effects o f degraded carotenes on chlorophyll-photosensitized singlet oxygen oxidation of soybean o il ...... 134 4.2.1 B-carotene...... 134 4.2.2 Lycopene ...... 142 4.2.3 Crude lycopene extract ...... 151 4.3 Effects o f degraded carotenes on thermally-induced oxidation of soybean oil ...... 158
XI 4.4 Separation and identification of soybean oil volatile oxidation products by SPME ...... 163 4.5 Effects of degraded carotenes on formation of volatile oxidation products in soybean oil ...... 172 Chapter 5 Conclusions...... 184
List of References...... 187
XU LIST OF TABLES
Page
1. Fatty acid composition and analytical constants of soybean oil ...... 6 2. Major volatile compounds responsible for off-flavors in oxidized soybean oil ...... 16 3. The carotenoid content of fresh/frozen vegetables (pg/lOOg ‘wet weight’) ...... 60 4. The contents o f lutein, lycopene, and P-carotene in 10 varieties of tomatoes (pg/lOOg) ...... 61 5. Carotenoid contents of several common tomato-based food products compared with whole tomatoes (pg/lOOg) ...... 64 6. Reported human tissue levels (nmol/g) of P-carotene and lycopene ...... 72 7. Relative vitamin A activity o f some carotenoids commonly foimd in vegetables .....73 8. UV/Vis spectroscopic data for several common carotenoids ...... 106 9. Tukey’s Studentized Range Test for the effects of P-carotene and thermally degraded P-carotene combinations on headspace oxygen depletion after 24 hours of light exposure ...... 135 10. Tukey’s Studentized Range Test for the effects of P-carotene and thermally degraded P-carotene combinations on peroxide formation after 24 hours of light exposure ...... 140 11. Tukey’s Studentized Range Test for the effects of lycopene and thermally degraded lycopene combinations on headspace oxygen depletion after 24 hours of light exposure ...... 145 12. Tukey’s Studentized Range Test for the effects of lycopene and thermally degraded lycopene combinations on peroxide formation after 24 hours of light exposure ...... 149
xui 13. Tukey’s Studentized Range Test for the effects o f crude lycopene extract and thermally degraded crude lycopene extract combinations on headspace oxygen depletion after 24 hours o f light exposure ...... 153 14. Tukey’s Studentized Range Test for the effects o f crude lycopene extract and thermally degraded crude lycopene extract combinations on peroxide formation after 24 hours of light exposure ...... 155 15. Volatile compounds detected in the headspace o f oxidized soybean oil by SPME-GC-MS ...... 167
XIV LIST OF FIGURES
Page
1. United States annual soybean oil production, 1988-1998 ...... 6 2. Excitation and deactivation of photosensitizers ...... 12 3. Sensitized oxidation reaction pathways for type 1 and 11 processes ...... 13 4. Main thermal decomposition products of linolenate hydroperoxides ...... 18 5. Solid phase microextraction device ...... 24 6. Molecular orbitals of triplet oxygen ...... 30 7. Molecular orbitals of singlet oxygen ...... 30 8. Generation of 'Oa by photochemical, chemical, and biological systems ...... 32 9. Singlet oxygen-catalyzed oxidation and degradation o f linoleate ...... 36 10. Scheme for the quenching of singlet oxygen and triplet sensitizer ...... 38 11. Singlet oxygen quenching rates (kq) as a function o f the length of the conjugated polyene system ...... 42 12. Steady state kinetics graph of triplet sensitizer quenching ...... 47 13. Plot of slope/intercept of Figure 10 regression line vs. quencher concentration (triplet sensitizer quenching only) ...... 48 14. Steady state kinetics graph of singlet oxygen quenching ...... 48 15. Plot of slope/intercept of the regression line from Figure 12 vs. quencher concentration (singlet oxygen quenching only) ...... 49 16. Isoprene, the basic unit of the carotenoid molecule...... 50 17. Lycopene, the prototypical carotenoid structure ...... 51 18. Numbering of the positions of the atoms in the carbon skeleton of lycopene 52 19. Structures of common carotenes ...... 53 20. Structures of typical oxygenated carotenoids (xanthophylls) ...... 55
XV 21. Structures of typical apocarotenoids ...... 56 22. Pathways involved in the conversion o f (3-carotene to retinoids ...... 68 23. Pathways and processes involved in the metabolism of carotenoids ...... 69 24. Possible reaction pathway for photosensitized oxidation of lycopene ...... 81 25. Nonvolatile compounds often formed during heating of (3-carotene ...... 83 26. Reaction sequence for the formation o f volatile compounds during heat treatment o f lycopene ...... 86 27. Sample procedure for isolation of carotenoids firom a natural source ...... 91 28. Main features o f a Cig bonded-phase silica material...... 96 29. Schematic diagram of a tjqjical HPLC system ...... 97 30. Typical DR. spectrum for lycopene ...... 107 31. Main fragmentations of the molecular ion of canthaxanthin by FAB MS ...... 110 32. Olefrnic section o f the IH-NMR spectrum of a typical carotenoid measured at 250 MHz (top), 400 MHz (middle), and 600 MHz (bottom)...... 113 33. Increased degradation of (3-carotene (decreased P-carotene peak size) with increased thermal processing at 90°C ...... 118 34. Increased degradation of lycopene (decreased lycopene peak size) with increased thermal processing at 90°C...... 118 35. Schematic of P-carotene or lycopene heating and fractionation into solutions with varying degrees of thermal degradation ...... 119 36. Addition of degraded carotene solutions to soybean oil to create the final concentrations in the tested sample solutions ...... 120 37. Addition of unheated carotene solutions to soybean oil to create the final concentrations in the tested control sample solutions ...... 121 38. Diagram of mirrored light box for sample storage under light ...... 122 39. Extent of thermal degradation of each collected P-carotene sample solution as seen by RP-HPLC ...... 129 40. Extent of thermal degradation of each collected lycopene sample solution as seen by RP-HPLC ...... 130
XVI 41. Extent of thermal degradation of each collected crude lycopene extract sample solution as seen by RP-HPLC ...... 132 42. Effects of thermally degraded (3-carotene on the headspace oxygen depletion of soybean oil containing 3 ppm chlorophyll under light storage at 25°C ...... 134 43. Effects of thermally degraded P-carotene on the peroxide value o f soybean oil containing 3 ppm chlorophyll under light storage at 25°C ...... 139 44. Correlation between headspace oxygen depletion and peroxide value for soybean oil containing degraded p-carotene under light storage at 25°C ...... 44 45. Effects of thermally degraded lycopene on the headspace oxygen depletion of soybean oil containing 3 ppm chlorophyll imder light storage at 25°C ...... 144 46. Effects of thermally degraded lycopene on the peroxide value o f soybean oil containing 3 ppm chlorophyll under light storage at 25°C ...... 148 47. Correlation between headspace oxygen depletion and peroxide value for soybean oil containing degraded lycopene under light storage at 25°C ...... 150 48. Effects of thermally degraded crude lycopene extract on the headspace oxygen depletion of soybean oil containing 3 ppm chlorophyll under light storage at 25°C ...... 152 49. Effects of thermally degraded crude lycopene extract on the peroxide value of soybean oil containing 3 ppm chlorophyll under light storage at 25°C ...... 154 50. Correlation between headspace oxygen depletion and peroxide value for soybean oil containing degraded crude lycopene extract under light storage at 25°C ...... 157 51. Effects of thermally degraded (3-carotene on the headspace oxygen depletion of soybean oil stored in the dark at 60°C ...... 159 52. Effects of thermally degraded lycopene on the headspace oxygen depletion of soybean oil stored in the dark at 60°C ...... 161 53. Effects of thermally degraded crude lycopene extract on the headspace oxygen depletion of soybean oil stored in the dark at 60°C ...... 162 54. Relative sensitivity of four different SPME fibers to hexanal in the headspace o f a soybean oil/acetone model system ...... 165
xvu 55. SPME-GC-MS chromatogram of volatiles in oxidized soybean oil ...... 166 56. SPME-GC mass spectrum of peak 17 (retention time 18.219 min.) (top) and 2-undecenal (bottom)...... 169 57. SPME-GC mass spectrum of peak 18 (retention time 18.681 min.) (top) and cyclododecane (bottom) ...... 170 58. SPME-GC mass spectrum of peak 19 (retention time 18.988 min.) (top) and dodecanal (bottom)...... 171 59. Percentage comparison of total GC peak area from volatiles formed in oil samples containing P-carotene or P-carotene degradation products after storage in a light box at 25°C ...... 173 60. Percentage comparison of total GC peak area from volatiles formed in oil samples containing P-carotene or P-carotene degradation products after storage in an oven at 60°C...... 177 61. Percentage comparison of total GC peak area from volatiles formed in oil samples containing lycopene or lycopene degradation products after storage in a light box at 25°C ...... 179 62. Percentage comparison of total GC peak area from volatiles formed in oil samples containing lycopene or lycopene degradation products after storage in an oven at 60°C...... 181
xvm CHAPTER 1
INTRODUCTION
In recent decades soybean oil production has seen dramatic growth as a result of its
widespread availability and relatively low cost. Today soybean oil accounts for over
75% of total vegetable oil consumption in the U. S., with total production exceeding its
nearest competitor, cottonseed oil, by nearly tenfold (Weiss, 1983). Food processors note
soybean oil’s superior physical properties, which include a wide liquid temperature
range, the ability to be hydrogenated selectively for blending with other oils, and
retention of pourable characteristics after partial hydrogenation (Pryde, 1980). As a
result, soybean is the predominant oil used in products such as mayonnaise, margarines,
prepared salad dressings, and salad or table oils.
A common drawback to the use of soybean oil in processed foods, however, is its
degree of unsaturation and resulting susceptibility to lipid oxidation. A reduction in
flavor quality and nutritional quality, as well as increased health risks such as heart
disease, are often associated with lipid oxidation in foods. Singlet oxygen CO 2) is a
reactive oxygen species that is often a promoter of lipid oxidation in food products
exposed to light (Foote et al., 1982). In these photosensitized oxidation processes, unsaturated fats must be exposed to light in the presence of oxygen and a sensitizer
1 before oxidation of the lipid can occur. Pigments and dyes such as chlorophyll, flavins,
acridine orange, erythrosine, methylene blue, and rose bengai have been found to be
effective photosensitizers (Faria, 1983).
Photosensitized oxidation can proceed by one of two different mechanisms: type I
and type II reactions. In a type I reaction, the excited triplet sensitizer interacts with a
substrate molecule directly, and the radicals that are formed subsequently undergo further
oxidation to form hydroperoxides. In a type II reaction, the triplet sensitizer first interacts
with oxygen to form singlet oxygen (^Oz). The singlet oxygen formed then reacts further
with various substrate molecules, while the triplet sensitizer interacts with other oxygen
molecules (Krinsky, 1979; Foote, 1982). As a result of its unique electron distribution,
singlet oxygen is very electrophilic and therefore extremely reactive with electron-rich
compounds containing double bonds such as linoleic acid (Frankel, 1985). The reaction
rate o f singlet oxygen with linoleic acid is 1,500 times faster than that of ground-state triplet oxygen and linoleic acid (Frankel, 1989).
Carotenoids are a group of lipid-soluble pigments created by the linking of five- carbon isoprene units together to form a polyene skeleton. They are classified into three main groups: the hydrocarbon carotenes, oxygenated xanthophylls, and smaller apocarotenoids (Britton et al., 1995). The carotenes 6-carotene and lycopene both possess polyene systems consisting of eleven conjugated double bonds, and have been shown to be excellent quenchers o f singlet oxygen (Matsushita and Terao, 1980; Packer et al., 1981; Di Mascio et al., 1989; Lee and Min, 1990; Jung and Min, 1991). The deactivation of * 0 2 by carotenoids occurs primarily by way o f physical quenching, a process o f transferring excitation energy firom *0% to the carotenoid resulting in formation of ground state (^0%) oxygen and triplet excited carotenoid. The energy is then dissipated
through rotational and vibrational interactions between the excited carotenoid and the
solvent to recover ground state carotenoid (Di Mascio et al., 1992; Stahl and Sies, 1993).
Carotenoids can therefore be effective at minimizing singlet oxygen-catalyzed lipid
oxidation in food products.
The extended polyene structure of carotenoid pigments also causes them to be very reactive in the presence o f light and/or heat. Because carotenoids such as P-carotene are widely used in industry as food colorants, thermal degradation (and resulting loss of color) during processing is of great concern to food manufacturers and nutritionists.
Although p-carotene and lycopene can be easily degraded by thermal processing (Marty and Berset, 1988; Handelman et al., 1991), very little information has been reported regarding the effectiveness o f thermally degraded carotenes as antioxidants. Therefore, my first objective was to determine on an individual basis the singlet oxygen quenching abilities of P-carotene and lycopene that have been exposed to thermal treatments and then added to a soybean oil model system. Peroxide values and headspace oxygen depletion of the oil samples were measured to determine the rate of oxygen uptake into the oil during light exposure, which can then be indirectly correlated with the rate of singlet oxygen quenching.
The second objective was to determine the ability of degraded carotenes (P-carotene and lycopene) to stop firee radical autooxidation initiated by triplet oxygen. To accomplish this, headspace oxygen depletion of oil samples stored in the dark containing degraded P-carotene or lycopene were measured and then compared with controls. Whether oxidized carotenes exhibit prooxidant or antioxidant properties in soybean oil
was determined by comparison with dl\-trans carotene (untreated) controls as well as controls containing only oil.
The third objective was to identify the extent of oxidation and/or hydrocarbon degradation of the carotenes after each thermal treatment by high-performance liquid chromatography (HPLC) methods, and to correlate the extent of degradation with antioxidant capacity.
The final objective was to compare the relative quantities of volatile lipid oxidation products formed in soybean oil containing degraded P-carotene or lycopene during storage. Solid phase microextraction (SPME) was utilized to collect the volatile compounds, which were then separated and identified using gas chromatography-mass spectrometry. The total peak area of volatile oxidation products present in the headspace of soybean oil samples containing degraded p-carotene or lycopene was compared to that of oil only controls.
This research helped to determine how the addition of carotenoids to food products for coloration and/or antioxidant purposes pre-processing might affect the long-term oxidative stability o f the lipids in the product. CHAPTER 2
LITERATURE REVIEW
2.1 Vegetable oils
2.1.1 Production and utilization
The production of vegetable oil in the United States has traditionally come from a variety of sources, including soybean, cottonseed, com, palm, sunflower, and peanut plants. In recent decades soybean oil production has seen dramatic growth as a result of its widespread availability and relatively low cost (Figure 1). Today soybean oil accounts for over 75% o f total vegetable oil consumption in the U. S., with total production exceeding its nearest competitor, cottonseed oil, by nearly tenfold (Weiss, 1983).
In addition to its availability and low cost, soybean oil possesses other desirable properties that help make it the dominant oil in the U. S. market. It has a high linoleic acid content and a low saturated fatty acid content (Table 1), thus improving its nutritional value relative to the more saturated oils such as coconut or palm oil. Food processors note soybean oil’s superior physical properties, which include a wide liquid temperature range, the ability to be hydrogenated selectively for blending with other oils, and retention o f pourable characteristics after partial hydrogenation (Pryde, 1980).
Due to the many advantages the use of soybean oil provides over other vegetable oils, it 16,500 15,500 c 14,500 o 13,500 12,500 11,500 1988 1990 1992 1994 1996 1998 Year
Figure 1: United States annual soybean oü production, 1988-1998.
% Myristic acid 0.1 Palmitic acid 10.5 Stearic acid 3.2 Oleic acid 2Z3 Linoleic acid 54.5 Linolenic acid 8.3 Arachidonic acid 0.2 Eicosenoic acid 0.9 Iodine value 120 -141 Melting point -23° to -20°
Table 1: Fatty acid composition and analytical constants of soybean oil. is utilized in a wide variety of products in the U. S. After refining, bleaching, and
deodorization steps, a high quality food oil, or refined oil, is ready for incorporation into
various foodstuffs. Soybean oil is the major oil used in mayonnaise, prepared salad
dressings, and salad or table oils. Virtually all mayonnaise, imitation mayonnaise, and
prepared salad dressings produced commercially in the U. S. are prepared exclusively
with soybean oil. Salad and cooking oils often use a blend of partially hydrogenated
soybean oü and other oils such as cottonseed and/or com in their production. Most
margarines consist of partially hydrogenated soybean oil alone, or blended with minor
fractions of other oils such as com or cottonseed. Soybean oil is excellent in the
preparation of shortenings used in the production of bread, cakes, cookies, pie cmsts,
icings, and fillings. Its use has expanded to the point where it can even be found on the
labels of soup, pudding, waffle, and macaroni and cheese mixes, prepared spaghetti
sauces, pizzas, and popcorn (Brekke, 1980).
2.1.2 Soybean oil instability
In addition to improving nutritional quality, the large amount of unsaturated fatty
acids commonly found in soybean oil unfortunately also increases its susceptibility to
oxidation. Oxidative processes can easily occur in soybean oil due to the low energy of
activation necessary to cause double bond migration and hydroperoxide formation on the
individual fatty acids. When double bond migration does occur, essential fatty acids such as linoleic acid are often lost, lowering the oil’s nutritional quality. The high degree of unsaturation in soybean oü has been linked to increased oxidation of solubilized vitamin
A and other compounds present in the oil, as well (Budowski and Bondi, 1960).
7 2.1.3 Free radical autoxidation
Free radical autoxidation is a process that can destroy the unsaturated lipid components (such as soybean oil) in foods, causing the development of undesirable
flavors and odors. In some cases, toxic degradation products are created which can affect the overall safety of the food product (Frankel, 1989). The free radical autoxidation process involves the reaction o f unsaturated oils with oxygen to form hydroperoxides in three basic steps: initiation, propagation, and termination (Burton and Ingold, 1984). The initiation step (1) can take place by metal catalysis, ultraviolet irradiation, and/or thermal or photodecomposition of peroxides/hydroperoxides to form free radicals (R-)-
RH ------^ R- (1)
During the propagation steps (2 & 3) the free radical R- reacts with oxygen to form hydroperoxides (ROOM).
R- + O2 ------ROO- (2)
ROO + RH ------> ROOM (3)
Interactions between ROO- and R- can cause the formation o f nonradical products through the termination step (4, 5, & 6).
R- + R------RR (4)
R 0 0 -+ R------^ ROOR (5)
ROO- -f ROO------^ ROOR -i- O2 (6)
Hydrogen abstraction from an imsaturated fatty acid is selective for the most weakly boimd hydrogen, due to the slow rate of propagation step (3). As a result, the rate of autoxidation thus depends on the number of double bonds present. The relative rates of autoxidation of oleate:linoIeate:linolenate has been determined to be approximately
1:12:25 on the basis of peroxide development (Frankel, 1985).
2.1.3.1 Prooxidants in oils
Compounds known as prooxidants bave been shown to have the ability to quicken the
rate of autoxidation of unsaturated fatty acids. Transition metals having two or more
valence states with a significant oxidation-reduction potential between them have been
found to increase the rate of lipid oxidation by decreasing the length of the induction
period, and increasing the maximum rate o f oxidation. Metals of importance as catalysts
of fatty acid autoxidation include copper, iron, manganese, and nickel (Love, 1985).
Andersson and Lingnert (1998) reported that addition of only 0.07 ppm copper to rapeseed oil resulted in a doubling of the amount of bexanal formed in the headspace after 35 days of storage in air. Addition of 70 ppm copper to similar rapeseed oil samples exposed to air caused the headspace bexanal concentration to increase 70-fold. Jung and
Min (1992) added various concentrations o f oxidized a-, y-, and S-tocopherols to soybean oil, and found that in the dark all three tocopherols acted as prooxidants, with a- tocopherol exhibiting the greatest prooxidant effect. Prooxidant activity increased as oxidized tocopherol concentrations increased in the oil. In another study, thermally oxidized triglycerides at levels of up to 2% were found to exhibit prooxidant effects on the oxidative stabilities of refined, bleached, deodorized, and purified soybean oils (Yoon et al., 1988). Terao (1989) used the intense prooxidant effects of 2,2’-azoèw(2,4- dimethylvaleronitrile) (AMVN) to produce methyl linoleate hydroperoxides for antioxidant testing. Haila et al. (1996) examined the effects o f lycopene, lutein, annatto,
9 and y-tocopherol on autoxidized triglycerides. Surprisingly, results showed that lutein
and lycopene displayed prooxidant behavior when added in the dark to a model system of
oxidized rapeseed oil. Henry et al. (1998) also found that both P-carotene and lycopene
acted as prooxidants when added to safflower seed oil at concentrations >500 ppm.
2.1.3.2 Free radical scavengers
Lipid oxidation can be considered a “chain” reaction, initiated and propagated by the
formation of free radicals. Due to the undesirable results of free radical autoxidation in
lipids, i.e., oxidative rancidity, off-flavor formation, loss of essential fatty acids, etc.,
many free radical “scavengers” have been found that can either delay or eliminate
oxidative processes altogether. Scavengers usually work by removing or deactivating
free radicals before they can propagate the autoxidative process that leads to degradation
of lipids. The antioxidant action of the scavenger can be described by equation (7),
R- -t- AH ------^ RH + A- (7)
where R- is the free radical and AH indicates the antioxidant compound. Terao (1989)
investigated the free radical scavenging activity of the carotenoids P-carotene,
canthaxanthin, astaxanthin, and zeaxanthin on oxidized methyl linoleate by measuring the production of methyl linoleate hydroperoxides. Canthaxanthin and astaxanthin retarded hydroperoxide formation more efficiently than P-carotene and zeaxanthin, indicating the superior peroxyl-radical trapping abilities of the former compounds. Lutein, a common xanthophyll, was cited in another study as having excellent peroxy-radical scavenging ability when added to an oxidized methyl linoleic acid solution (Chopra et al., 1993).
10 Haila et al. (1996) examined the free radical quenching abilities of lycopene, lutein, annatto, and y-tocopherol on autoxidized rapeseed oil triglycerides, and found that annatto and y-tocopherol effectively inhibited hydroperoxide formation. The combination of lutein with y-tocopherol was found to be the most effective at minimizing lipid autoxidation.
2.1.4 Photosensitized oxidation
Most packaged edible oils in supermarkets are exposed to substantial light during the extent of their shelf lives, and are thus subject to photosensitized oxidation as well as autoxidation. The mechanism of photosensitized oxidation as a pathway for the production o f hydroperoxides in oils differs from that of free-radical autoxidation. In photosensitized oxidation, unsaturated fats must be exposed to light in the presence of oxygen and a sensitizer before oxidation of the lipid can occur. Photosensitized oxidation of unsaturated fats begins with the absorption of light by photosensitizers. Two systems of electronically excited states exist for photosensitizers: the singlet (^Sens*) and the triplet (^Sens*) (Foote, 1968; Foote, 1976). The singlet state of a sensitizer arises when upon absorption of visible or near UV light energy, an electron is raised to a higher energy level. The excited singlet sensitizer can then revert back to the ground state by emitting fluorescent light, or it can undergo intersystem crossing (ISC), a process by which the spin of the excited electron is reversed, and a change in the state of the molecule from singlet (*Sens*) to triplet (^Sens*) results (Figure 2). The excited triplet sensitizer slowly decays to the ground state by emitting phosphorescent light. Because
II Excited state Sens* ISC k=l-20xl07sec hv Sens*
k=l-3xl0 /sec Ground state Sens triplet-triplet annihilation
Figure 2; Excitation and deactivation of photosensitizers
the lifetime of ^Sens* is approximately 20,000x the lifetime of ^Sens*, photosensitized oxidations usually proceed by way o f ^Sens*. Therefore, the most effective sensitizers are those which offer the longest triplet state lifetime (Foote, 1968). Pigments and dyes such as chlorophyll, flavins, acridine orange, erythrosine, methylene blue, and rose bengal have been found to be effective photosensitizers (Faria, 1983). Unfortunately, soybean oil has been found to contain lipid oxidation-promoting sensitizers such as chlorophyll and pheophytin at levels as high as 15 ppb and 100 ppb, respectively, even after industrial refining and distillation o f the oil (Usuki et al., 1984; Frankel, 1989).
2.1.4.1 Type 1 and 11 processes
Photosensitized oxidation can proceed by one of two different mechanisms: type 1 and type II reactions. In a type 1 reaction, the excited triplet sensitizer interacts with a substrate molecule directly, usually transferring a hydrogen atom or electron in the
12 process. The radicals that are formed subsequently undergo further oxidation to form hydroperoxides (Figure 3). Type I processes are favored in conditions with high substrate reactivity and concentration, and low oxygen concentration (Foote, 1976).
In a type II reaction, the triplet sensitizer first interacts with oxygen to form an excited electronic state of oxygen known as singlet oxygen ('O 2). The singlet oxygen formed then reacts further with various substrate molecules, while the triplet sensitizer is free to interact with other oxygen molecules (Krinsky, 1979; Foote, 1982). Electron transfer directly to oxygen to form a superoxide anion also occurs in type II reactions, though these reactions make up less than 1% of the total sensitizer-oxygen collisions that occur in type II processes. Both the solubility and concentration o f oxygen present in the system largely decides the rate of type II reactions. Whether substrate or triplet oxygen, respectively, react with the excited triplet sensitizer is the major determinant of whether a type I or type II reaction will occur.
hv ISC Sens -► ^Sens* -► Sens*
Type I Type n + RH
R- + -SensH
+ RH
ROOH 02 •- + Sens •+ ROOH
Figure 3: Sensitized oxidation reaction pathways for type I and II processes.
13 2.1.4.2 Sensitizers in oils
Many compounds are known to act as efficient photosensitizers in vegetable oils by
generating singlet oxygen ('O 2) through transfer of energy in the presence of light and
atmospheric (triplet) oxygen (^Oi). Chlorophyll, which is often naturally present in
vegetable oils (Usuki et al., 1984; Tan et al., 1997), is perhaps the most commonly known
and researched photosensitizer. Fakourelis et al. (1987) determined that chlorophyll in
purified virgin olive oil acted as a photosensitizer for singlet oxygen formation under
light. At a concentration of 3.3 x 10'^ M, chlorophyll b was able to act as a
photosensitizer in the oxidation of soybean oil (Jung and Min, 1991). Kiritsakis and
Dugan (1985) determined that at a concentration of 6 ppm, chlorophyll functioned as a
photosensitizer in the rapid oxidation of bleached olive oil. In addition, chlorophyll b
was found to be twice as efficient as chlorophyll A at sensitizing oxidative reactions. Lee
and Min (1988) found that a dilute solution of soybean oil in methylene chloride
containing 4 ppm chlorophyll was effectively oxidized due to the photosensitizing
properties of the chlorophyll. In a similar study, 4.4 x 10'^ M chlorophyll initiated
photosensitized oxidation in soybean oil containing various carotenoids (Lee and Min,
1990).
The decomposition products of chlorophyll, including pheophytin, pheophorbide,
chlorophyllin, and protoporphyrin, have also been found to exhibit photosensitizer properties (Rahmani and Csallany, 1998). Endo et al. (1984) evaluated the photosensitizer activities of pheophorbide and pheophytin by subjecting safflower oil samples containing each of the chlorophyll derivatives to photooxidation at 0°C. Results showed that pheophytin and pheophorbide exhibited stronger prooxidant activities than
14 either chlorophyll A or B. Pheophytin was also found to be a more effective sensitizer
than chlorophyll A in the photooxidation of olive oil (Kiritsakis and Dugan, 1985) as well
as methyl linoleate (Usuki et al., 1984).
Yamauchi and Matsushita (1977) found the artificial color methylene blue to be an
effective sensitizer in the photooxidation of methyl linoleate. Both methylene blue and
chlorophyll were utilized by Faria and Mukai (1983) to act as sensitizers in the
photooxidation of linoleic acid and safflower oil.
Additional compounds have been shown to exhibit sensitizer properties. Seely and
Meyer (1971) utilized hypericin, a natural photosensitizer with a strong resistance to photooxidative destruction, to examine the products of oxidized (3-carotene. An acetone solution containing 2 x 10'^ M hypericin effectively catalyzed the oxidation o f (3-carotene into several epoxide compounds. Soybean oil itself was reported to act as a sensitizer in the photooxidation of 4,7-undecadiene (Clements et al., 1973).
2.2 Volatile soybean oil oxidation products
The oxidation of soybean oil, whether through autoxidative or photosensitized processes, usually results in the formation of lipid hydroperoxides. These unstable primary oxidation products often degrade to form a variety o f volatile, short-chain secondary oxidation products (Ullrich and Grosch, 1988). Many of these volatile compounds negatively affect the overall quality of soybean oil by producing imdesirable off-flavors that have been described as grassy, painty, or fishy in nature (Table 2). As a result, much effort has been made to both individually identify these volatile compounds and to determine possible mechanisms for how they might form.
15 Relative Threshold Major volatiles % value (ppm)
t, r-2,4-decadienal 33.7 0.10
t, c-2,4-decadienal 17.9 0.02
t, c-2,4-heptadienal 11.1 0.04
2-heptenal 5.6 0.2
t, r-2,4-heptadienal 4.5 0.1
n-hexanal 4.5 0.08
n-pentane 3.1 340
n-butanal 1.5 0.025
2-pentenal 1.2 l.O
l-octen-3-ol 0.9 0.0075
2-pentyl furan 0.8 2.0
n-pentanal 0.7 0.07
2-hexenal 0.7 0.6
n-nonanal 0.7 0.2
n-heptanal 0.6 0.055
l-penten-3-ol 0.5 4.2
2-octenal 0.5 0.15
Table 2: Major volatile compounds responsible for off-flavors in oxidized soybean oil.
16 2.2.1 Formation of volatile compounds
There are several catalysts that help initiate soybean oil oxidation that can result in
the formation of different volatile compounds, including light, increased temperatures,
and the presence of metal ions in the oil. Lee et al. (1995) reported that soybean oil
exposed to a 1937 lux fluorescent light source at room temperature favored the formation
of the volatile compoimds 2-heptenal and l-octen-3-ol when compared with samples
stored in the dark. Light exposure tests were conducted by Robards et al. (1988) on com
chips flried in various vegetable oils and stored in commercial polypropylene film
packaging. Results showed that exposure to light increased the formation of propanal,
pentanal, and hexanal, with propanal exhibiting the most pronounced increase.
Robards et al. (1988) also examined the effects of elevated temperatures on the
formation of volatile compoimds firom vegetable oil in com chips. While the control
chips stored at -20°C showed no changes in their volatiles profile after one year, identical
samples stored at 60°C in the dark showed significantly increased concentrations of
pentanal and hexanal after only 15 days. Snyder et al. (1985) found that polyunsaturated
oils such as soybean and safflower were particularly susceptible to oxidation when stored
at 60°C in the dark. The major volatile compounds apparently formed from linoleic acid
oxidation included pentane, hexanal, and 2-heptanal. The main volatile compounds formed by the thermally induced oxidative degradation of methyl linolenate hydroperoxides were described by Frankel et al. (1991) as 2,4,7-decatrienal, 2,4- heptadienal, and propanal (Figure 4).
Andersson and Lingnert (1998) examined the effects o f copper on the formation o f hexanal in rapeseed oil. The addition of 70 ppm copper to the rapeseed oil sample
17 exposed to air caused the hexanal concentration to increase 70-foId after 35 days of
storage compared to the sample without copper. The addition of only 0.7 ppm copper to rapeseed oil doubled hexanal concentrations.
OOH
2,4,7-decatrienal
OOH
\/ ------\ / i
3-hexenal OOH
2,4-heptadienal
HOO
16
propanal
Figure 4: Main thermal decomposition products of linolenate hydroperoxides.
18 2.2.2 Isolation o f volatile compounds
A major challenge in deterrnining which volatile compounds are responsible for the
off-flavors produced by the oxidation of soybean and other vegetable oils has been in the
attempt to isolate the volatiles from the original sample matrix. Even after isolation,
concentration o f the volatile compounds by adsorption processes is often necessary
before proper qualitative analysis by gas chromatography can occur. As a result, several
different methods have been developed in the attempt to properly identify and quantify
volatile compounds from a variety of sample matrixes.
2.2.2.1 Direct injection
One of the first methods developed for volatile compound isolation was direct
injection of the lipid sample into a gas chromatograph (GC). In this procedure, the glass
injection port liner is packed with glass wool, on top of which the oil sample is added
directly. The glass liner is then lowered into the heated injection port and sealed into the
GC carrier gas system, ready for analysis. Using the direct injection method, Snyder et al. (1988) isolated and identified over 20 different volatile soybean oil oxidation products, with pentane, as well as hepta- and decadienal isomers, being found in the largest quantities. The relatively higher concentration of carbonyl compounds was attributed to possible thermal decomposition of volatile precursors in the injector. Dupuy et al. (1985) reported acetone, pentane, 2,4-decadienal, 2-heptenal, and 2,3-octanedione as the major volatile oxidation products of soybean oil. The use of an external inlet device in the GC was shown to improve the direct injection procedure by decreasing the degradation o f volatile preciursors in the injector port.
19 2.2.2.2 Static headspace analysis
Static headspace (SHS) analysis is the simplest method for isolation of volatile compounds. SHS analyses take advantage o f the high volatility of typical oxidized compounds by relying on the equilibrium distribution of analytes between two coexisting phases (liquid and gaseous). Once the volatile compounds have equilibrated between both phases in a closed system such as a sealed vial, an aliquot of headspace is extracted from the vial with an airtight syringe. The syringe headspace gas is then immediately injected directly into a GC for analysis.
Though SHS analysis is the simplest, most inexpensive method for volatile analysis, it lacks the sensitivity of other methods due to dilution by various headspace gases
(Zhang et al., 1994; Field et al., 1996; Jelen et al., 1998). In addition, SHS sampling is generally only effective for compounds with low boiling points and/or high vapor pressures such as ethanol or ethyl acetate (Snyder et al., 1985; Chin et al., 1996). By focusing his studies only on similar low-molecular weight volatile compounds, Frankel
(1993) was able to effectively investigate by SHS analysis the thermal decomposition of various oxidized oils. Propanal was found to best characterize the oxidative degradation of omega-3 fatty acids, while pentane and hexanal characterized the decomposition of omega-6 fatty acids typical of soybean oil. Robards et al. (1988) used SHS to isolate and quantify the high vapor pressure compounds propanal, pentanal, and hexanal formed during the oxidative deterioration of vegetable oil in com chips. Snyder et al. (1988) revealed that for investigations involving isolation of volatiles from oxidized soybean oil, the SHS method favored a higher relative proportion of acrolein, propanal, and pentanal because of their higher vapor pressures. In a previous study, hexanal and pentanal were
20 among the major volatile compounds isolated from soybean oil stored for eight days at
60°C (Snyder et al., 1985).
2.2.23 Dynamic headspace analysis
The lack of sensitivity often associated with SHS analysis can be overcome by
performing dynamic headspace (DHS) analysis, also commonly known as piurge-and-trap
analysis. During DHS analysis volatile compounds are continuously purged out of the
sample while being concentrated onto a porous polymer adsorbent trap. The next and
final step in the analysis involves thermal desorption of the volatiles immediately prior to
GC separation. A DHS procedure was developed by Lee et al. (1995) for isolating the
volatile compoimds from oxidized soybean oil and trapping them on the adsorbent 2,6-
diphenyl-/7-phenylene oxide (Tenax TA®) that allowed only minimal decomposition of
hydroperoxides (50°C for 30 minutes; 75 mL/min He flow rate). Selke and Frankel
(1987) reported the principle components in the DHS-GC chromatogram profiles of
oxidized soybean oil as pentane, hexenal, 2-heptenal, and 2,4-decadienal from linoleate
hydroperoxides, and 2/3-hexanal and 2,4-heptadienal from linolenate hydroperoxides.
Andersson and Lingnert (1998) utilized DHS analysis to examine the effects of oxygen
and copper concentration on the rate and extent of lipid oxidation in rapeseed oU.
Morales et al. (1997) analyzed the flavor and off-flavor components o f virgin olive oü
using DHS techniques with Tenax TA® as the adsorbent. In this case, thermaUy desorbed volatiles from the Tenax TA® were condensed on a cold trap injector cooled to -110°C before being flash heated and injected into the capillary GC system.
21 Unfortunately, problems with DHS methods have also been discussed in the literature. DHS analysis has been described as expensive, time-consuming, labor intensive, and prone to methodological problems (Zhang et al., 1994; Song et al., 1997).
Park (1993) observed that during the thermal desorption step in DHS analysis in which trapped volatiles are released from the adsorbent and backflushed by hydrogen carrier gas, structural alterations occurred to 2-alkenals and 2,4-aIkadienais. Recommendations included replacing hydrogen with an inert gas such as helium for DHS-GC analyses involving quantification of volatile imsaturated compounds formed during lipid oxidation. Lower-boiling compounds may be lost during the DHS purge cycle while other less volatile compounds such as hexanal, heptadienal, and decadienal are concentrated on the trap, in effect altering the relative ratios of the volatiles observed on
GC chromatograms (Snyder et al., 1988). Park and Goins (1992) encountered frequent problems with volatile (nonanal and decanal) carryover from one sample to the next when utilizing DHS analysis. The requirement that blank water samples be run up to four times between each sample analysis to ensure complete elimination of nonanal and decanal carryover was reported as a serious drawback for the use of DHS analyses in attempts to isolate volatile compounds.
2.2.2.4 Solid phase microextraction
A relatively new technique for the isolation o f volatile compounds from sample matrixes is known as solid phase microextraction (SPME). First described by Arthur and
Pawliszyn (1990), SPME is rapidly gaining popularity as the technique of choice in the isolation and identification of volatile compounds because of its ability to integrate
22 sampling, extraction, concentration, and sample introduction in a single step. Both the
cost and time involved with sample analysis is reduced, while at the same time efficiency
and selectivity are increased (Zhang et al., 1994).
Headspace SPME (HS-SPME) utilizes a fused silica fiber coated with a polymeric
organic liquid that is connected to stainless steel tubing. The tubing acts as a syringe
needle, increasing the mechanical strength of the fiber assembly and allowing repeated
sampling without physical damage to the fiber (Figure 5). During HS-SPME, the fiber is
withdrawn into the protective syringe needle, which then punctures the septum of a
sealed vial. The fiber is pushed out of the syringe needle by a plunger, exposing the fiber
coating to the sample headspace containing volatile compounds. After a predetermined
exposure time to the headspace, the fiber is withdrawn back into the protective syringe needle, the needle is inserted into the injection port of a GC, and the fiber is pushed out of the needle. Analytes from the fiber coating are then thermally desorbed in the GC injector port and quantitatively analyzed (Zhang and Pawliszyn, 1993; Zhang et al.,
1994).
Originally, most SPME applications involved analysis o f aqueous samples for the existence of possible contaminants. Arthur and Pawliszyn (1990) examined the effectiveness of SPME analysis in isolating the common groundwater contaminants
1,1,1 -trichloroethane, trichloroethene, and perchloroethylene. SPME analyses were conducted to determine whether the soil fumigant methyl isothiocyanate (Gandini and
Riguzzi, 1997) or the pesticide procymidone (Urruty et al., 1997) were present in wine samples.
23 plunger
barrel Z-slot
hub-viewing window
adjustable depth gauge
needle guide septum piercmg needle fiber attachment needle coated SPME fused silica fiber
Figure 5: Solid phase microextraction device.
SPME methods have also been used to isolate flavor volatiles firom a wide variety of sample matrixes (Yang and Peppard, 1994), including orange (Steffen and Pawliszyn,
1996; Jia et al., 1998) and apple (Matich et al., 1996; Song et al., 1997) flavor volatiles.
Aroma profiles of commercial vodkas (Ng et al., 1996) and black and white truffles
(Pelusio et al., 1995) have even been determined by SPME analyses. Bicchi et al. (1997)
24 differentiated various roasted coffees by their chromatographic flavor profiles. SPME-
GC patterns for characteristic volatile compounds were distinctly different among various
Swiss cheese samples in a study conducted by Chin et al. (1996). Clark and Bunch
(1997) performed both qualitative and quantitative analyses of flavor additives to tobacco samples. The affinity of SPME fibers for volatiles crucial to proper beer aroma, including the essential oils in hops (Field et al., 1996) and higher alcohols and esters
(Jelen et al., 1998) have been determined. Pyrazines, important constituents in a wide variety of food flavors, were analyzed by HS-SPME-GC. Results showed SPME to be very effective in the isolation, concentration, and analysis of pyrazines in model systems
(Ibanez and Bernhard, 1996). Pan et al. (1995) reported that SPME-GC could isolate semivolatile fatty acids (Ce-Cto) directly firom aqueous samples. The addition of a derivatizing agent in the SPME fiber further increased the analysis sensitivity.
Specific detection ranges for SPME volatile analyses have been determined firom different sample matrixes. Arthur et al. (1992) reported limit o f detection ranges firom
0.3 to 3 pg/L for the BTEX compounds (benzene, toluene, ethylbenzene, and xylene) using a 56 pm polydimethylsiloxane (PDMS) coating and FID detector. Using SPME-
GC-MS analysis, 31 typical tobacco flavors were detected, identified, and quantitated firom tobacco matrixes in concentrations as low as 10 ng/g to 6 pg/g (Clark and Bunch,
1997). Ethyl esters of Cg to Gig fatty acids present ia commercial vodkas were detected by SPME at pg/L levels (Ng et al., 1996), while Pan et al. (1995) reported detection of Cs to Cio fatty acids at levels lower than 1 ppb. Zhang and Pawliszyn (1993) described the
25 detection limits of HS-SPME at the ppt level when ion trap MS was used in conjunction with GC analysis.
The type of stationary phase present on the silica fiber used for SPME analysis greatly affects the affinity for the adsorption of each specific volatile compound to the fiber. Chin et al. (1996) found that adsorption of cheese volatiles on a polyacrylate- coated fiber was 3 to 20 times greater than that with a polydimethylsiloxane (PDMS) fiber. Several studied indicated that the polar stationary phase polyacrylate was selective for the more polar analytes, while PDMS stationary phases had a higher affinity for the non-polar hydrocarbons (Pan et al., 1995; Steffen and Pawliszyn, 1996; Clark and Bunch,
1997; Jelen et al., 1998).
As was previously mentioned, SPME analysis for the isolation of volatile compounds has many advantages over more traditional techniques. SPME completely eliminates consumption of solvents and corresponding solvent disposal costs, while greatly reducing extraction time (Arthur and Pawliszyn, 1990; Yang and Peppard, 1994; Pan et al., 1995).
Urruty et al. (1997) reported that SPME-GC-MS analysis was slightly more sensitive than enzyme-linked immunosorbent assays (ELISA) for the detection o f procymidone residues in wine.
Unfortunately, SPME analysis also has its drawbacks. The primary difficulties
Arthur and Pawliszyn (1990) encountered were poor precision due to difficulty in achieving exact partition equilibrium, and inconsistent positioning of the SPME fiber in the injector port of the GC. Jelen et al. (1998) revealed similar problems with SPME repeatability as determined by peak area comparisons. Yang and Peppard (1994) found the reproducibility of SPME analyses lacking due to its sensitivity to experimental
26 conditions, which can in turn influence the amount of volatiles adsorbed onto the SPME
fiber. Quantification of higher molecular weight apple volatile compounds by SPME was
hindered by the slow transport o f analytes into the gaseous phase, resulting in longer
equilibrium times (Matich et al., 1996). Pelusio et al. (1995) also reported that HS-SPME
techniques were not suited for quantitative analyses due to the fiber coating’s preferential
adsorption of the more polar, volatile compounds found in truffle aroma.
2.2.3 Correlation with flavor quality
Fresh soybean oil has a pleasant, nutty flavor that is considered acceptable to most
consumers (Kao et al., 1998). When soybean oil undergoes considerable lipid oxidation,
however, some o f the volatile oxidation products that are created drastically alter the
flavor profile of soybean oil to what has been described as a grassy, beany, painty flavor.
Photooxidized oils may develop distinctly different flavors described as sour, metallic, or buttery (Warner et al., 1989). The actual flavors imparted by lipid oxidation in oils are difficult to assess because of the wide variations in both their sensory impact and the actual methods used for their determination (Frankel, 1991). Indeed, many of the methods (SHS, DHS, and SPME) used to isolate and identify volatile oxidation products from oils have been developed with the ultimate goal of accurately correlating the quantitative and qualitative off-flavor profiles with decreases in sensory oil flavor quality scores.
Lee et al. (1995) attempted to correlate the amount of volatile compounds formed with the flavor scores for oxidized soybean oil. Results suggested that the major volatile oxidation products that were measured by GC accounted for less than half of the flavor o f
27 the soybean oil samples. A synergistic interaction of volatiles that gave them a much
stronger flavor than would be expected was offered as an explanation for the data.
Robards et al. (1988) showed a good correlation between oxidized com chip flavor scores
and GC volatile peak data. Using GC methods to determine com chip rancidity was
suggested as a sensitive, relatively precise altemative to traditional sensory evaluation
methods. Wamer et al. (1988) also reported good correlation between GC volatile
analyses and sensory flavor evaluations o f cmde soybean oil stored for various time
intervals.
2.3 The chemistry o f oxygen
Oxygen is the most abundant element on earth, occurring as free 0% gas and
combined in the form of water and oxides of many kinds. Molecular oxygen is unique in
being a paramagnetic molecule, even though it has an even number of electrons. Of the
six electrons in oxygen’s outer shell, two are assigned to the 2S orbital, while the other
four electrons are assigned to the three 2P orbitals.
2.3.1 Nature of triplet and singlet oxygen
Molecular oxygen exists in more than one form due to differences in the arrangement of the electrons in the highest occupied pair of molecular orbitals, the n* orbitals. The outermost pair of electrons (ti*) in the common ground state form of molecular oxygen are located on different orbitals and spin in parallel (Figure 6). Spin multiplicity, which is used to describe electron arrangements in molecular orbitals, can be defined by the
28 simple equation 2S + 1, where S is the total spin quantum number. Because ground state oxygen contains two %* electrons with parallel spins, its spin multiplicity is 3 [2(1) +1].
For this reason, ground state oxygen is also known as ‘triplet’ oxygen (^Oa). Triplet oxygen follows Hund’s Rule, which states that the most stable arrangement of electrons is that with the maximum number of unpaired electrons, all with the same spin direction.
A much more rare type of oxygen known as singlet oxygen (^0%) has its two outermost electrons spinning oppositely one-another, resulting in a spin multiplicity of 1
[2(0) + 1]. The specific electron configuration of singlet oxygen can be seen in Figure 7.
Singlet oxygen exists in two states: the higher energy state, and the lower energy state, *A (Foote, 1968). The ^2) state oxygen contains two electrons with opposite spins in different orbitals, and exists 37.5 kcal/mole higher than ground state triplet oxygen
(Foote, 1968). The state is very unstable and only exists for picoseconds before decaying to the state. The *A oxygen state, however, exists for several microseconds at 22.4 kcal/mole above ground state triplet oxygen. It is responsible for most singlet oxygen reactions involving lipid oxidation, and is therefore usually the specific state generally described as singlet oxygen ('Oz) in the literature.
2.3.1.1 Reaction rates
The presence of a completely empty tc* molecular orbital in the electron configuration of singlet oxygen makes it a very electrophilic compound that actively seeks electrons to fill the vacant orbital. As a result, singlet oxygen (’Oz) is a very reactive species and reacts with electron-rich compounds such as linoleic acid at least
29 Molecular orbital
Atomic. orbital Q Atomic...... orbital.... 00(S) (S)00 2Px 2Py 2Pz ''0''— ^ .p. ^ 2Px 2Py 2Pz \ W /
2S W 2S
Energy IS IS
Figure 6: Molecular orbitals of triplet oxygen.
Molecular orbital
Atomic orbital o Atomic orbital
2Px 2Py 2Pz X; M 2Px 2Py 2Pz
2S W 2S
Energy IS IS
Figure 7: Molecular orbitals of singlet oxygen.
30 1,500 times faster than ground state triplet oxygen CO 2 ) to form hydroperoxides (Frankel,
1989). Because of its relatively long lifetime (several microseconds) in the state, singlet oxygen has time to diffiise through condensed phases, and is often as a result able to initiate lipid oxidation. Singlet oxygen can also initiate lipid oxidation by accepting an electron from a donor molecule, resulting in the formation of a reactive superoxide ion
(O2- ) and substrate radical cation (Saito et al., 1983).
2.3.2 Singlet oxygen generation
Singlet oxygen can be produced by both chemical and biological means (Figure 8).
Several methods have been described in the literature for the generation of singlet molecular oxygen (^0%). Di Mascio and Sies (1989) generated singlet oxygen by two different methods; the thermal decomposition of the water-soluble endoperoxide of 3,3’-
( 1,4-naphthylidene) dipropionate (NDPO 2) to NDP, as well as by a hypochlorite/H202 mixture. Approximately one-half of the oxygen liberated in the NDPO 2 —> NDP reaction was found to be in the singlet state. The use o f the thermodissociable endoperoxide of
3,3 ’ -( 1,4-naphthy lidene) dipropionate (NDPO 2 ) to generate singlet oxygen for use in determining carotenoid (Di Mascio et al., 1989) and tocopherol (Kaiser et al., 1990) quenching ability, respectively, has also been performed. Foote (1968) has also reported the high-yield production of singlet oxygen by the chemiluminescent reaction between
NaOCl and H2O2 .
Carlsson et al. (1976) determined that the exposure of oxygen at low pressure to a microwave discharge in the presence of mercury vapor provided a very reliable route for the generation of singlet oxygen (^ 0 2 ). In addition, it is noteworthy that singlet oxygen
31 RCOO- + RCOO • Enzymes RC=0 O2 + Sens*
Endoperoxides p^ductX ^^^^ H2 O2 + OCl
Products + C1-
Ozonides ------^ H 2 O 2 + O 2 OH- + OH
OH- + O2 -
H2 O2 + HO2-
O2- + O2- 0 2 - + Y+
Figure 8: Generation of *02 by photochemical, chemical, and biological systems.
can be generated not only by standard photoexcitation methods, but also via chemiexcitation or “photochemistry in the dark” (Cilento and Adam, 1988).
2.3.3 Singlet oxygen oxidation detection
Researchers have developed methods to identify and quantify the creation of singlet oxygen (* 0 2 ) and its oxidative effects in a model system. Headspace oxygen, peroxide value, and gas chromatographic reactor methods indirectly measure singlet oxygen production by the determination of oxygen uptake into a model system, while germanium photodiode detectors have been used to quantitatively measure * 0 2 emission.
32 2.3.3.1 Headspace oxygen
By exposing lipid samples containing a photosensitizer to light in an airtight sealed
serum bottle, rates o f singlet oxygen oxidation in the oil can be determined by measuring
the headspace oxygen content in the vial. Oxygen disappearance in the headspace of the
sample vial can then be measured directly by a gas chromatograph equipped with a
thermal conductivity detector. Fakourelis et al. (1987) measured the rate of singlet
oxygen oxidation in virgin olive oil samples that contained the sensitizer chlorophyll and
were exposed to a light intensity of 4,000 lux. Rates of singlet oxygen oxidation of
soybean oil exposed to 4 ppm chlorophyll sensitizer at 10°C were determined by the
measurement of headspace oxygen disappearance (Lee and Min, 1988). Lee and Min
(1990) determined the singlet oxygen quenching mechanisms of carotenoids in the
chlorophyll-sensitized photooxidation of soybean oil by measuring headspace oxygen
depletion using a thermal conductivity gas chromatograph.
2.3.3.2 Peroxide value
Singlet oxygen oxidation detection can be performed through the use o f peroxide
value measurements, specifically when testing conditions include the use o f a photosensitizer under lighted conditions. Peroxide value testing determines all substances (peroxides or other similar products of fat oxidation), in milliequivalents of peroxide per 1000 grams of sample, which oxidize potassium iodide under specific conditions (AOCS, 1980). Endo et al. (1984) used peroxide value measurements in conjunction with UV absorbance at 234 nm to determine the ability o f chlorophyll derivatives to initiate photosensitized singlet oxygen oxidation of methyl linoleate. Two
33 other studies involved the measurement of peroxide values to determine the effects of
carotenoids in inhibiting the chlorophyll-photosensitized oxidation of soybean oil (Lee
and Min, 1990; Jung and Min, 1991). Fakourelis et al. (1987) correlated peroxide values
with the rate o f singlet oxygen oxidation in virgin olive oil samples that contained the
sensitizer chlorophyll and were stored under lighted conditions. Kiritsakis (1985)
suggested that because the ratio of peroxide value to conjugated dienoic acid content
increased when chlorophyll was present in an olive oil model system, a singlet oxygen
oxidizing effect was present. As a result, singlet oxygen oxidation could be measured by
determining extent of peroxide value formation.
2.3.3.3 Gas chromatographic reactor
A gas chromatographic reactor (OCR) can be used to detect the singlet oxygen (^0%)
oxidizing activities of different light sources on various polyunsaturated lipid sources
(Faria and Mukai, 1983). The main parts o f the OCR include the reactor (oxidation
chamber), the condenser or trap, the thermal conductivity and flame ionization detectors,
and an optional humidifier. By using the OCR Faria and Mukai (1983) determined that
the rate of singlet oxygen oxidation in both safflower oil and pure linoleic acid was
dependent on the wavelength of the light used. The rate of oxidation increased in the
following order: control (dark) < fluorescent < incandescent < UV (black light).
2.3.3.4 Photodiode detector
The potential for singlet oxygen (^0%) oxidation in a system can also be determined quantitatively by the direct measurement of 'Oz generation and emission through the use
34 o f a germanium diode photodetector. Through the use of a liquid nitrogen-cooled
germanium photodiode detector, Di Mascio et al. (1989) was able to quantitate both
singlet oxygen production and quenching by various carotenoids. Di Mascio and Sies
(1989) measured both monomol and dimol light emissions due to the chemiluminescent
transition of singlet oxygen to the triplet ground state. Singlet oxygen detection was
performed by monitoring infrared photoemission at 1270 nm.
2.4 Singlet oxygen catalyzed oxidation
Exposure to light in the presence of oxygen and a photosensitizer can lead to the
formation of singlet oxygen and subsequent oxidation of many compounds. Compounds that are most susceptible to oxidation by singlet oxygen are those that are highly unsaturated in nature, including many vegetable oils.
2.4.1 Oxidative products in oil
When reacting with unsaturated fats, singlet oxygen reacts directly with double bonds by the “ene” reaction, causing oxygen to be inserted at either carbon of the double bond
(Frankel, 1985; Frankel, 1991). The double bond, in turn, is shifted one carbon to the left or the right, resulting in the formation of a hydroperoxide in the trans configuration
(Figure 9). Singlet oxygen oxidation of linoleate produces a mixture of conjugated (9 and 13) and nonconjugated (10 and 12) hydroperoxides, while linolenate also produces both conjugated (9, 12, 13, 16) and nonconjugated (10 and 15) hydroperoxides. Singlet oxygen photooxidation o f the unsaturated diene 4-cfr,7-cfr-undecadiene was shown to yield 4-hydroperoxy-5-/ra«^,7-cfr-imdecadiene and 5-hydroperoxy-3-frar?iS’,7-cfr-
35 H H H H , I I J* ... h v /sensitizer/ 0% y ____
H
+
H breakdown [Hoducts H H OOH H = \ + I I ^ Y >R H H aldehyde
Figure 9: Singlet oxygen-catalyzed oxidation and degradation of linoleate.
undecadiene as initial products (Clements et al., 1973). Upon photosensitized oxidation at 0°C, methyl oleate produced a 50-50% mixture o f 9- and 10-hydroperoxides, linoleate a mixture of 66% conjugated and 34% unconjugated hydroperoxides, and linolenate a mixture of 75% conjugated and 25% unconjugated hydroperoxides (Frankel et al., 1979).
Matsushita and Terao (1980) utilized mass spectrometry to accurately determine the isomeric compositions of monohydroperoxides produced by singlet oxygen oxidation of methyl oleate and methyl linoleate. Equal amounts of the 9- and 10- isomers were found to be present in methyl oleate, while a slight predominance of the 9- and 16- isomers were foimd in the isomeric composition of methyl linoleate.
Hydroperoxides formed through singlet oxygen oxidation in oils can further degrade through metal catalysts and/or thermal treatments to form both volatile and nonvolatile
36 secondary products. Decomposition o f linoleate hydroperoxides in air at 37°C for a week
produced many nonvolatile monomeric compounds, including di- and tri-oxygenated
esters (Frankel et al., 1985). Volatile compounds such as octane, methyl octanoate, 2-
decenal, and 2-undecenal were identified by gas chromatography from the
photosensitized degradation of methyl oleate, while oxidized products of methyl linoleate
included 2-heptenal, methyl 9-oxononanoate, and lO-oxo-8-decenoate (Frankel et al.,
1985).
2.4.2 Singlet oxygen initiated biological damage
In addition to catalyzing the oxidation of the polyunsaturated lipids found in foods such as vegetable oils, singlet oxygen-catalyzed oxidation can also cause damage in biological systems. Cerutti (1985) found convincing evidence that cellular prooxidant states with increased concentrations of singlet oxygen ('O 2) can promote initiated cells to begin neoplastic growth, resulting in chromosomal damage from DNA structure alterations. Di Mascio et al. (1989) determined that singlet oxygen-inflicted damage of plasmid and bacteriophage DNA included loss of biological activity (as measured by transforming capacity in E. coli) and DNA single-strand breakage.
2.5 Quenching o f singlet oxygen
Because o f the oxidative damage singlet oxygen can cause in both food and biological systems, singlet oxygen quenching is a subject of much practical interest. The phrase “quenching of singlet oxygen” includes both “chemical” quenching, in which singlet oxygen (^Oz) reacts with a quencher Q, to give a product, QO 2 (1), and “physical”
37 quenching, in which the interaction of quencher and product leads only to reduction of
to the ground state with no oxygen consumption or product formation ( 2 ).
'O2 + Q > QO2 (1) Kr
‘O2 + Q ------> "O2 + Q (2 ) Icq Figure 10 reveals a general scheme that holds for photooxidation of a substrate (A) which is known to react only with singlet oxygen (^Oz) at a rate constant kr in the presence of a quencher (Q). In this scheme, physical quenching of ^Oz occurs at a rate kq, while chemical quenching occurs at a rate kox-Q- The decay o f singlet Oz in the surrounding solvent occurs at a rate kj. Chemical, or “charge transfer” quenching, involves interactions between the very electrophilic singlet oxygen molecule and an electron donor to give a charge-transfer complex. Because singlet oxygen acts as the electron-accepting component, the most easily oxidized compounds are usually the best quenchers.
Compounds likely to quench by the charge transfer mechanism include tocopherols, amines, phenols, metal complexes, and other electron-rich compounds (Foote, 1979).
Sen ► 'Sen* Sen* > AOz
kq
Sen Sen
Figure 10: Scheme for the quenching of singlet oxygen and triplet sensitizer.
38 A second quenching mechanism is known as physical, or “energy-transfer”
quenching. The mechanism involves formation of triplet quencher and ground state
oxygen (^Oi) through an energy transfer from a triplet sensitizer to triplet oxygen. The
efficiency of the triplet quencher depends on its ability to be very near or below the
energy of singlet O 2 in the state, which is 22 kcal/mole. Carotenoid compounds such
as p-carotene (Foote and Denny, 1968; Foote, 1970; Foote, 1976; Krinsky, 1979;
Krinsky, 1989; Stratton and Liebler, 1997) and lycopene (Foote, 1970; Krinsky, 1979;
Krinsky, 1989; Di Mascio et al., 1992), metal complexes (Foote, 1982), bilirubin (Stocker
et al., 1987), ascorbyl palmitate (Lee et al., 1997), and other compounds with very
extensive conjugated systems have triplet energies below 2 2 kcal/mole, and therefore can
be effective singlet oxygen physical quenchers.
2.5.1 Effects of tocopherols
Vitamin E, or a-tocopherol, can quench singlet oxygen at a high rate in all solvents,
but it is also converted into oxidized products in polar solvents (Foote, 1982; Liebler et
al., 1990). D-a-tocopherol was found to be an effective quencher of singlet oxygen (^0%) molecules at a rate of k = 2.5 x 10* moF^ s'^ in pyridine. The quenching process proved to be mostly “physical” in nature, with a-tocopherol deactivating about 1 2 0 ^0 % molecules before being destroyed (Fahrenholtz et al., 1974). Kaiser et al. (1990) discovered that the relative singlet oxygen physical quenching efficiencies of the tocopherol homologs decreased in the following order: a > p > y > ô-tocopherol.
Chemical reactivity of the tocopherols with ^0% was much lower, accounting for only 0.1-
39 1.5% of physical quenching. Singlet oxygen quenching rates for a-, y-, and 5-tocopherols
in ethanol were estimated by Yamauchi and Matsushita (1977) to be 2.6 x 10* M'^ s'\ 1.8
X 10* M'^ s'% and 1 .0 x 10* M*' s'\ respectively. The most effective quencher among
the three tocopherols was a-tocopherol. Burton and Ingold (1981) also found a-
tocopherol to be the most reactive antioxidant of the four tocopherol homologs a-, P~, y-,
and Ô-. The antioxidant activity of all tocopherols depends on the free hydroxyl group in
position 6 on the chromane ring; substitution of this group with an ester or ether group
eliminates the antioxidant activity (Sies and Stahl, 1995).
The effects o f tocopherols in an unsaturated oil system are not always beneficial,
however. The singlet oxygen quenching effect of a-tocopherol was found in one study to
be offset by the oxidation of the vitamin itself into hydroperoxidic products. The authors
therefore concluded that natural phenols such as a-tocopherol cannot prevent the buildup
of peroxide species in oils exposed to prolonged light (Carlsson et al., 1976). In another
study, the antioxidant ability of tocopherol was found to be only one-third that of various
anthocyanin compounds (Wang et al., 1997). Cilliard et al. (1980) demonstrated that
though tocopherols were effective singlet oxygen quenchers at low concentrations, at
concentrations higher than 5 x 10'^ M, a-tocopherol exhibited a pronounced prooxidant
effect in oxidized linoleic acid. Frankel et al. (1959) also foimd that natural soybean oil
exposed to light had a higher rate o f oxidation than soybean oils that had been ‘stripped’ to reduce their tocopherol content, again indicating the prooxidant effect of tocopherols present at elevated levels in vegetable oil.
40 2.5.2 Effects of carotenoids
The deactivation of ^0% by carotenoids occurs primarily by way of physical
quenching, a process of transferring excitation energy from ^ 0 % to the carotenoid
resulting in formation of ground state (^Oz) oxygen and triplet excited carotenoid (3).
The energy is then dissipated through rotational and vibrational interactions between ^C*
and the solvent to recover ground state carotenoid (4) (Di Mascio et al., 1992; Stahl and
Sies, 1993; Edge et al., 1997).
'O2 + c ------^ ^02 + ^C* (3)
^C* ------> C + thermal energy (4)
The existence of these quenching reactions has been proven in laser photolysis experiments in which triplet anthracene is produced and quenched by oxygen, yielding
^ 0 2 which then sensitizes absorption due to ^C* (Farmilo and Wilkinson, 1973). These tests have established beyond doubt that most quenching of ^ 0 2 by P-carotene (>99.9%) is in fact due to electronic energy transfer, resulting in production of ^C*, dissipation o f thermal energy into surrounding media, and a return of the carotenoid back to its ground state (Sies and Stahl, 1995). Because carotenoids can be “recycled” in this way, one molecule of P-carotene is capable of quenching as many as 250 to 1000 molecules of * 0 2
(Frankel, 1985; Liebler, 1993). As a result, carotenoids such as P-carotene and lycopene are among the most effective natural * 0 2 quenchers known (Di Mascio et al., 1992).
As can be seen in Figure 11, an increasing number of conjugated double bonds are associated with a higher efficiency of quenching ability against singlet oxygen (Foote et al., 1970; Tsuchiya et al., 1992). In addition, it has been suggested that energy transfer from singlet oxygen to carotenoids is exothermic for carotenes with 11 or more
41 l.OE+11
l.OE+10
CO l.OE+09
l.OE+08
l.OE+07
C 0 njugated C= C in P o lyene C hain
Figure 11 ; Singlet oxygen quenching rates (kq) as a function of the length of the conjugated polyene system.
conjugated double bonds and endothermie for those with fewer than 11 conjugated double bonds (Foote, 1979). To determine relative quenching ability, quenching rate constants for several different carotenoids and xanthophylls have been identified. Lee and Min (1990) determined that the total singlet oxygen quenching rate of five carotenoids increased as their conjugated double bond count increased. Lutein, zeaxanthin, lycopene, isozeaxanthin, and astaxanthin, which contain 1 0 , 1 1 , 1 1 , 1 1 , and
13 conjugated double bonds, respectively, had increasing ^0% quenching rates of 5.72 x
10^, 6.79 X 10^, 6.93 x 10^, 7.39 x 10^, and 9.79 x 10^ M’^ sec"\ respectively. Di Mascio et al. (1989) found that lycopene, with 11 conjugated double bonds, was the most efficient biological carotenoid singlet oxygen quencher, with a quenching rate constant of
42 31 X 10^ M'* sec'\ over twice that of P-carotene (14 x 10^ M'* sec'*). Matsushita and
Terao (1980) revealed a quenching rate constant for p-carotene that was higher than that
reported for lycopene, at 1.5 x 10*° M'* sec'*, while Packer et al. (1981) reported the P-
carotene rate at a much lower 1.5 x 10^ MT* sec'*.
Variation in the functional groups of carotenoids such as lycopene and P-carotene
may contribute to differences in their quenching rates, with the opening of the P-ionone
ring to an open chain in lycopene increasing its quenching ability over p-carotene (Di
Mascio et al., 1991). Foote et al. (1970) suggested that significant cis-trans isomerization
of the carotenoids may accompany singlet oxygen quenching. Though different
structurally, both a- and P-carotene have been determined to act as effective singlet
oxygen quenchers (Kiritsakis and Dugan, 1985).
Foote and Denny (1968) first reported that singlet oxygen could be effectively
quenched by low concentrations of P-carotene: they found that 95% of the
photooxidation of 0.1 M 2-methyl-2-pentene was inhibited by IC^ M P-carotene.
Kellogg and Fridovich (1975) found that at concentrations as low as 10'^ M P-carotene
still inhibited the peroxidation of linolenate in a model system.
Another reaction that has been used to measure the quenching activity of carotenoids
is the chlorophyll-sensitized photooxidation o f soybean oil, monitored as oxygen consumption. Using this method, the *0% quenching rate constants of P-apo- 8 ’-carotenal,
p-carotene, and canthaxanthin were found to be 3.06 x 10° M'* sec'*, 4.60 x 10° MT* sec'*, and 1.12 x 10*° M'* sec'*, respectively (Jung and Min, 1991). Lee and Min (1988) examined the effects of 0, 5, 10, and 20 ppm P-carotene on the oxidation of a soybean
43 oil/methylene chloride model system containing 4 ppm chlorophyll. (3-carotene reduced
the oxidation o f soybean oil at every concentration, and most efiectively at the highest
concentration o f 20 ppm. Warner and Frankel (1987), however, found that at levels ^ 0
ppm, P-carotene contributed to poor flavor and color in soybean oil, while 5 to 10 ppm p-
carotene reduced the photosensitized oxidation of the oil without decreasing oil quality.
The question of whether carotenoids such as P-carotene could, in addition to
quenching singlet oxygen, also act as chain-breaking antioxidants has also been posed.
Though P-carotene does not have the structural features commonly associated with chain-
breaking antioxidants. Burton (1989) claimed that given low oxygen partial pressures, P-
carotene had the potential to act as a lipid-soluble chain-breaking antioxidant. Burton
and Ingold (1984) found that at oxygen pressures of 150 torr or higher, however, P-
carotene and related compoimds could actually act as prooxidants in a methyl linoleate
model system. Heinonen et al. (1997) reported similar P-carotene prooxidant effects in a
10% oil-in-water emulsion, and suggested protecting P-carotene from oxidative
destruction by adding tocopherols to the fat emulsion. Stratton and Liebler (1997) used
an isotope dilution assay to distinguish Type 1 vs. Type II lipid peroxidation reactions in a
lipid bilayer model system. Results suggested that singlet oxygen quenching, rather than
radical scavenging reactions, is responsible for the photoprotective actions o f P-carotene
in a lipid system.
The ability o f synthetic carotenoids to quench singlet molecular oxygen (^0%) has also been determined. Devasagayam et al. (1992) synthesized a Czg-polyene-tetrone that exhibited a quenching rate constant for 'O 2 of kq = 16 x 1 0 ^ sec'\ which was higher
44 than the quenching rate constants for both (3-carotene (5 x 10^ M’’ sec'^) and lycopene (9
X 10^ M** sec’*). The presence of two oxalyl chromophores at each end of the polyene chain was credited with enhancing the *0% quenching ability of the Czg-polyene-tetrone.
2.5.3 Mechanisms and kinetics
In the prevention of singlet oxygen oxidation, an effective quencher can interact with either a triplet sensitizer or singlet oxygen (see Figure 10). In the former case, a reaction between the triplet excited sensitizer (^Sen*) and the quencher (Q) occurs at a rate kq to yield the ground state sensitizer (Sen). In the ground state, the sensitizer can no longer react with triplet oxygen (^O^) to form singlet oxygen.
In the latter case, a quencher can prevent the reaction of singlet oxygen and substrate
(A) two different ways. First and foremost, singlet oxygen (^Oz) and a quencher (Q) interact at rate kq in a physical exchange of energy from oxygen to quencher that results in the formation of grotmd state triplet oxygen (^Oz) and a triplet excited quencher. The second interaction involves a chemical reaction between *0 % and quencher that proceeds at a rate kox-q and results in the formation of an oxidized quencher molecule (QO 2).
Singlet oxygen can also revert to triplet oxygen through decay in the surrounding solvent.
The actual type of quenching occurring (singlet oxygen or triplet sensitizer) can be determined by examining the steady state kinetics of the quenching mechanisms (Foote,
1979). The rate of formation of oxidized substrate (AO 2) due to singlet oxygen oxidation can be expressed as:
drAp2l = K fknlO?!} X ______(krlA]}______(5) dt {kq[(^ + ko[02]} {kr[A] + kq[Q ] + kox.q[Q] + ky}
45 with the reciprocal of equation (5) being equation ( 6 ):
fdFAO^ir' = K-‘ fknr01+k»r0?1! % fkrlAl + kglOl + k ^ ro i + k^\ (6) {dt} {ko[02Î} ■ {k,[A]>
where AO2, A, and Q are the concentrations of the oxidized substrate, the substrate, and
the quencher, respectively, K is the rate of formation of excited triplet sensitizer, ko, kr,
kq, and kox-q are the reaction rate constants of triplet sensitizer with triplet oxygen (^ 0 %),
substrate with singlet oxygen (^ 0 2 ), physical quenching of ^ 0 2 , and chemical quenching
of 'O2, respectively, and kd is the decay rate of ‘O 2 in a specific solvent.
Under specific conditions of constant irradiation time and quencher (Q)
concentration, a linear relationship exists between the reciprocal of the formation rate of the oxidized substrate and the reciprocal of the concentration of the substrate. When
prevention of photosensitized oxidation occurs solely as a result of interactions between the quencher (Q) and excited triplet sensitizer (^Sen*), equation ( 6 ) is simplified to the
following:
fdrAO,!}-' = K-' d + k n r o i} X (l+ k d > (7) {dt} {ko[02]} {kr[Af}
Quenching mechanisms and rates can be observed by graphing the reciprocals of each
level of quencher used. A graph plotting the reciprocals of [AO 2] vs. [A] at various quencher [Q] levels reveals a straight line with a different intercept and slope for each quencher level, which is indicative o f excited triplet sensitizer (^Sen*) quenching (Figure
1 2 ). In this case, slope = K'^ {kd(kq[QJ + ko[0 2 ])}/kr ko[0 2 ] and intercept = K‘^ {kq[Q] + ko[0 2 ])}/ ko[0 2 ], with the latter dependent upon both quencher (Q) and oxygen concentrations. The ratio of the slopes of the plots in Figure 12 to their intercepts vs.
46 [Qi]
[Q2]
1/[A02]
[Q3]
1 /[ A ]
Figure 12: Steady state kinetics graph of triplet sensitizer quenching.
quencher concentration (Q) results in a horizontal line because it is independent of both quencher and oxygen concentrations (Figure 13). Again, a horizontal line plot of this kind is indicative of excited triplet sensitizer (^Sen*) quenching.
When prevention of photosensitized oxidation occurs solely as a result of interactions between the quencher (Q) and singlet oxygen (^Oz), equation ( 6 ) is simplified to the following:
IdjAOall'* = K-' (l+(kg^+ kg)r01 + kd> (7) {dt} { k r [ A ] }
The plots of the reciprocals of [AO 2] vs. [A] at different [Q ] in this case yield plots of straight lines with different slopes = K'* {(k ox-q + k<,)[Q] + k d }/k r and a constant intercept
K*‘ (Figure 1 4 ) . Slope/intercept = (kox-q + k q ) [ Q ] + k j /k r for these plots, which is independent of oxygen concentration. Plotting slope/intercept vs. quencher concentration
47 S/I
[Q]
Figure 13: Plot of slope/intercept of Figure 12 regression line vs. quencher concentration (triplet sensitizer quenching only).
[Ql]
[Q2]
1/[A02] [Q3]
1/[A]
Figure 14: Steady state kinetics graph of singlet oxygen quenching.
48 [Q] results in a line with intercept = kj/kr and a slope = kox-q + kq/kr (Figure 15). Slope
changes as quencher [Q] concentration changes along with a constant intercept value are
indicative of singlet oxygen quenching in the system. In this case the concentration of
quencher [Q] is usually kept low enough to ensure that it acts as a physical quencher of
singlet oxygen, and not a quencher of excited triplet sensitizer.
2.6 Carotenoid chemistry
Carotenoids derive their name from a representative of their group, P-carotene, which
was first isolated firom carrots {Daucus carotd) by Wackenroder in 1831. They are
among the most widespread and important classes of pigments in nature, with over 600
known naturally occurring compounds, each with similar chemical structures.
S/I
[Q]
Figure 15: Plot of slope/intercept of the regression line firom Figure 14 vs. quencher concentration (singlet oxygen quenching only).
49 2.6.1 Structure
Carotenoids are isoprenoid polyenes formed by joining eight C 5 isoprene units
(Figure 16). The isoprene units are linked in a head-to-tail manner except in the center of
the molecule, where a tail-to-tail linkage provides symmetry for the molecule. The
resulting compound contains two methyl groups near the center of the polyene chain that
are separated by six carbon atoms, while the other methyl groups are separated by only
five carbon atoms. This structural arrangement is illustrated in lycopene (Figure 17),
which consists of the prototypical C 40 isoprenoid skeleton firom which all other
carotenoids can be derived by modifications such as cyclization, substitution, elimination,
addition, and rearrangement (Stahl and Sies, 1996).
2.6.1.1 Chromophore
Carotenoids owe their color to the absorption of light by a feature o f their molecular
structure known as the ‘chromophore’. In most carotenoids the chromophore consists
entirely of a series o f conjugated carbon-carbon double bonds, often referred to as the
‘polyene chain’. Though it is possible to have up to 15 conjugated double bonds in the
CH CH HC
Figure 16: Isoprene, the basic unit of the carotenoid molecule
50 Figure 17: Lycopene, the prototypical carotenoid structure.
chromophore of a C 40 carotenoid, structures with 7 to 11 such bonds are far more
common. A chromophore of seven or more double bonds, present in carotenoids such as
p-carotene and lycopene, conveys the ability to absorb light in the visible region so that
colors such as yellow/orange and red are observed, respectively. Likewise, the chromophores of phytoene (three conjugated double bonds) and phytofluene (five conjugated double bonds) are not long enough to reflect light and provide color (Britton etal, 1995).
Other properties of carotenoids are determined specifically by the chromophore, as well. First, each carotenoid is characterized by a specific electronic absorption spectrum based upon absorption of light by its chromophore. As a result, absorption spectroscopy is an important technique in carotenoid analysis. Second, the polyene chain renders the molecule extremely susceptible to oxidative degradation and geometrical isomerization by light, heat, or acids. Finally, the length of a carotenoid’s chromophore, in terms of number o f conjugated double bonds, has been found to be closely linked to its ability to
51 quench, reactive compounds such as singlet oxygen in vitro (Foote et al., 1970; Miller et al., 1996).
2.6.1.2 Carbon skeleton
The carbon skeleton o f many carotenoids consists of a C 40 backbone, although compounds with C45 and even C50 structures have been identified (Weedon and Moss,
1995). It has long been recognized that some carotenoids known as apocarotenoids also have carbon skeletons with fewer than 40 carbon atoms, though most of these have been found to be degradation products of a larger C 40 compoimd.
For convenience, the positions of the atoms in the carbon skeleton of a carotenoid are often numbered in a conventional manner as shown by Figure 18. Usually only key positions on the molecule are numbered in this fashion to focus attention directly to the area o f interest.
16’ 13’ 17) 4 15’
3 20 ’ 2
Figure 18: Numbering o f the positions of the atoms in the carbon skeleton o f lycopene.
52 2.6.2 Classification
Carotenoids can be generally classified into three major groups: carotenes, oxygenated carotenoids (xanthophylls), and the apocarotenoids.
2.6.2.1 Carotenes
The carotenes are comprised solely of carbon and hydrogen (C 40H56), and may or may not include a cyclical ring structure at one or both ends of the hydrocarbon. Common carotenes include (3-carotene, lycopene, and a-carotene (Figure 19).
(3-carotene
lycopene
a-carotene
Figure 19: Structures of common carotenes.
53 2.Ô.2.2 Oxycarotenoids
A second major group of carotenoids are the oxygenated derivatives of the carotenes,
also known as the xanthophylls. The oxygen functions most commonly observed in this
group are hydroxy (monols, diols, and polyols), methoxy, carboxy, oxo, aldehyde, and
epoxy (5,6- and 5,8-epoxides). Xanthophylls containing triple bonds are also known
(Pfander, 1992). When present, hydroxyl substituents are usually found at carbon 3 in
the 8 - or P-ring of the compound. In most cyclic xanthophylls the 5,6 and 5’, 6 ’ double
bonds are very susceptible to epoxidation, while the unconjugated double bond in the e-
ring does not undergo epoxidation. Examples of xanthophylls include canthaxanthin,
zeaxanthin, astaxanthin, and P-cryptoxanthin (Figure 20).
2.6.23 Apocarotenoids
A carbon skeleton containing less than 40 carbon atoms defines the group o f carotenoids known as the apocarotenoids. Most of the molecules in this group occur as the result of degradation of one end o f a C 40 carotenoid. One notable exception is vitamin A (retinol), a C20 compound which is a metabolite of P-carotene but is not considered an apocarotenoid. With the degraded end of the molecule usually comprised of an aldehyde or ketone group, typical apocarotenoids can be seen in Figure 21.
2.6.3 Stereochemistry
The stereochemistry of the carotenoids includes geometrical isomerism about the carbon-carbon double bond, as well as absolute configuration of an asymmetric carbon.
54 canthaxanthin
zeaxanthîn
astaxanthin
P-cryptoxanthin
Figure 20: Structures of typical oxygenated carotenoids (xanthophylls).
55 P-apo-8’-carotenal
P-apo-13-carotenone
retinal
P-apo-14’-carotenal
P-apo-lO’-carotenal
Figure 21: Structures of typical apocarotenoids.
56 Each of the disubstituted, and trisubstituted acyclic double bonds in the carotenoid polyene chain can exist in two forms known as geometrical isomers. The C=C double bond in a carotenoid is designated either cis or trans, with a cis double bond implying a configuration with the highest-priority groups on the same side, and a trans configuration having the groups on opposite sides. More recently, these designations have been largely replaced by Z (zusammen = together) and E (entgegen = opposite) terminology.
Typically, as the number of double bonds in a molecule increases, the number of possible stereoisomers increases accordingly. In carotenoids, however, the number of stereoisomers is restricted because o f steric hindrance from methyl groups along the chain. For example, of the 1056 theoretically possible cis-trans isomers o f lycopene, only 72 are sterically unhindered cis isomers; of 272 possible P-carotene isomers, only 20 are in the unhindered cis form. Carotenoids can often undergo cis-trans isomerization when exposed to high temperatures (Pfander, 1992; O’Neil and Schwartz, 1992) and/or light (Humbeck, 1990). Most carotenoids are found naturally in the dl\-trans (E) form due to its higher stability.
A second major characteristic of carotenoid stereochemistry is their possession of at least one asymmetrically substituted carbon, also known as a chiral center. This results in a molecule that can exist in two distinct stereoisomeric forms that are mirror images of each other. These isomers are known as optical isomers or enantiomers. Optical isomers have identical physical properties with the exception that one rotates polarized light to the right (dextrorotary or ‘d’ isomer), while the other rotates the plane of polarized light an equal amount to the left (levorotary or T’ isomer). Often the absolute configuration of chiral carotenoids is determined by chiroptical methods such as circular dichroism.
57 2.7 Carotenoid distribution and function
2.7.1 In nature
Carotenoids are widely distributed in nature, providing the well-known yellow-orange
color of flowers (sunflower and marigold), the orange-red coloring o f fruits (tomato, orange), and the orange roots of carrots. The greatest production o f carotenoids, however, occurs in the photosynthetic tissues of plants and algae where the abundance of chlorophyll often masks their presence.
2.7.1.1 Plant contents
Total annual natural plant production of carotenoids has been estimated in excess of
100 million tons (Pfander, 1992; Britton et al., 1995). Most of this production is in the form of four major carotenoids: fucoxanthin in marine seaweeds and algae, and lutein, violaxanthin, and neoxanthin in green leaves (Pfander, 1992).
Carotenoids are responsible for the colors of many ftmits, including pineapple, lemons, paprika, and rose hips. Mangos have been foimd to contain from 20 to 125 pg/g total carotenoids (depending on cultivar), with 60 to 76% of the total consisting of 13- carotene (Godoy and Rodriguez-Amaya, 1987). Khachik et al. (1989) found apricots, cantaloupe, and pink grapefruit to be excellent sources o f (3-carotene, with pink grapefimt containing significant quantities (>3000 pg/lOOg) of lycopene.
Hart and Scott (1995) found that among fresh/frozen vegetables, good sources (>1000 pg/lOO g) of lutein were broccoli, butterhead lettuce, parsley, peas, peppers, spinach, and watercress; of lycopene: tomatoes; and of (3-carotene: broccoli, carrots, greens.
58 butterhead lettuce, mixed vegetables, parsley, spinach, and watercress (Table 3).
Khachik et al. (1995) found vegetables from Fiji Island such as amaranthus leaves and
drumstick leaves to contain high levels of both lutein and p-carotene. In an earlier study,
Khachik et al. (1992b) reported the predominant carotenoids in raw green vegetables
(broccoli, spinach, and green beans) to be neoxanthin, violaxanthin, lutein epoxide,
lutein, a-carotene, and p-carotene. In addition, a comprehensive list of carotenoids in
raw tomatoes included lutein, 5,6-dihydroxy-5,6-dihydrolycopene, lycopene 1,2-epoxide,
lycopene 5,6-epoxide, lycopene, neurosporene, y-carotene, P-carotene, phytofluene, and
phytoene. Ben-Aziz et al. (1973) listed a series of oxygenated carotenoids isolated from
tomatoes which included apo- 6 ' -lycopenal, apo- 8 ’-lycopenal, and lycoxanthin. In a
similar study, Britton and Goodwin (1969) isolated a series o f carotenoids from ripe
tomatoes, which included phytoene 1 ,2 -epoxide, phytofluene, (^-carotene, and lycopene.
Carotenoids found in tomato varieties can be seen in Table 4.
The established functions of carotenoids in plants can often be related to their ability to absorb light in the visible spectrum. In photosynthetic tissues, carotenoids usually have two well-defined functions: 1) to act as accessory pigments in photosynthetic processes, and 2 ) to protect ‘photosynthetic apparatus’ against potential damage from visible light such as photosensitized oxidation (Goodwin, 1980). In the former case, carotenoids absorb light at wavelengths different from the chlorophylls and then transfer the absorbed fight energy to the chlorophylls with extreme efficiency that can nearly approach 100% (Siefermann-Harms and Ninnemann, 1982). In the latter case, carotenoids have been shown to exert a photoprotective effect in photosynthetic plant
59 Lutein Zeaxanthîn Lycopene a-carotene (3-carotene
Frozen Green beans 494 70 299
Broccoli 1614 800
Green cabbage 80 ------51
Carrots 283 ----- 3610 10800
Cauliflower Trace
Cucumber 670 ------2 2 2
Greens 3046 ------1663
Leeks 161 69
Iceberg lettuce 1 1 0 74
Butterhead lettuce 1611 1603
Mixed frozen veg. 882 84 ------1045 3670
Parsley 5812 3505
Frozen peas 1633 360
Green pepper 660 235
Frozen sweetcom 522 437 60 45
Spinach 5869 — --- 3397
Spring onions 255 1 1 2
Tomato 78 2937 415
Watercress 10713 4777
Table 3: The carotenoid content of fresh/frozen vegetables (fig/lOOg ‘wet weight’)
60 trans- Total trans-^- Lutein lycopene lycopene carotene
Red varieties Cherry 1 0 1 2686 3780 473 ‘large’ 6 8 1915 2270 349 ‘salad’ 78 2158 2547 509 Flavourtop 48 4958 5653 428 Tigerella 191 1223 1582 1702 Ida FI Hybrid 103 1324 1711 964 Shirley FI 79 2079 2347 771 Craig 149 2948 3907 1093 Moneymaker 59 3475 4255 427 Allicanti 91 3659 4037 525 Beefsteak 89 2729 4833 883
Yellow varieties Sungold 204 390 528 2232 Gold Sunrise 107 2 1 2 1 93
Table 4: The contents of lutein, lycopene, and (3-carotene in 10 varieties of tomatoes (pg/lOOg).
cells. Sistrom et al. (1956) first observed this in comparing a wild purple photosynthetic bacterium with a blue-green mutant strain that lacked carotenoids. When exposed to air and light, the mutant strain stopped growing because its cells and chlorophyll had been killed and destroyed, respectively, while the wild strain continued to grow.
61 2.7.1.2 Animal contents
Though commonly thought of merely as plant pigments, carotenoids are also often
found in microorganisms and animals. Carotenoids are often responsible for yellow,
orange, or red colors in non-phototrophic bacteria, molds, and yeasts. Animals are
sometimes colored by carotenoids from their diet. For example, the feathers of flamingos
are colored pink-red by the presence of ketocarotenoids, while astaxanthin in the skin and
flesh of goldfish and salmon provide these fish with their unique colors. Protein-
astaxanthin complexes provide the bluish-green pigmentation for many marine
invertebrate animals. However, when the animal (e.g. lobster) is subsequently cooked,
the astaxanthin is liberated from the protein, changing the animal’s color to the orange-
red color of the carotenoid (van Breemen, 1996). Even the skin of humans is capable of
taking on a yellowish-orange hue following ingestion of large amounts of carotenoid
supplements and/or carotene-rich foods.
The carotenoid lycopene has been found to be present in human blood samples (0.5
pmo 1/liter plasma) as well as tissues such as adipose, adrenals, and testes(1 nmol/g wet
wt. to 20 nmol/g wet wt.) (Stahl and Sies, 1996). Khachik et al. (1992a) identified
eighteen different carotenoids (as well as vitamin A) from extracts of human plasma,
including several isomers of (3-carotene, lutein, and zeaxanthin. Stahl et al. (1993)
separated five geometrical isomers of P-carotene and seven isomers of lycopene in human serum and tissues. The most prevalent P-carotene isomer in human serum was found to be 13-cw-P-carotene. In testes tissue, however, considerable amounts of 9-cis- and traces of IS-c/s'-P-carotene were also detected. Human liver, adrenal gland, and
62 testes tissues contain significantly higher amounts of carotenoids such as P-carotene and
lycopene than kidney, ovary, and fat tissues. In liver, kidney, adrenal gland, ovary, and
fat tissues p-carotene is the major carotenoid present, whereas lycopene is the
predominant carotenoid in testes tissue (Stahl et al., 1992).
2.7.2 In processed foods
The carotenoid content of foods has been proven to change both quantitatively and
qualitatively after exposure to industrial processing involving heat treatments (Khachik,
1998). Stewed tomatoes along with processed tomato paste were determined to have
lower epoxide concentrations than raw tomatoes (Khachik et al., 1992b). Tan et al.
(1988) also examined the carotenoid content of tomato paste, finding the four dominant
carotenoids to be (in increasing quantities) phytofluene, P-carotene, phytoene, and
lycopene. Several tomato-based food products revealed lycopene as the most abundant
carotenoid, with concentrations ranging firom 0.3 mg/lOOg in vegetable beef soup to 55
mg/lOOg in tomato paste (Table 5). The concentration of P-carotene ranged from 0.23
mg/lOOg in tomato soup to 1.51 mg/lOOg in vegetable beef soup (Tonucci et al., 1995).
Khachik et al. (1992b) revealed that cooked (by microwaving, steaming, and boiling) green vegetables did not have significantly changed carotenoid profiles from their raw counterparts. However, many of the epoxycarotenoids were destroyed in long-term boiling processes. The carotenoid contents of processed carrots have also been examined, and were observed to contain up to 15% less “total effective carotenes” when compared with raw carrots (Ogunlesi and Lee, 1979).
63 Lutein Lycopene Phytoene y-carotene P-carotene
Tomato soup 90 10920 1720 1950 230
Vegetable beef soup 1 1 0 310 350 1510
Minestrone soup 150 1480 280 920
Vegetarian veg. soup 160 1930 600 1500
Tomato juice 60 10770 1900 1740 270
Vegetable juice 80 9660 1710 830
Catsup 17230 3390 3030 590
Spaghetti sauce 160 15990 2770 3020 440
Tomato paste 340 55450 8360 9980 1270
Tomato puree 90 16670 2400 2940 410
Tomato sauce 17980 2950 3170 450
Whole tomatoes 80 P270 1860 1500 230
Table 5: Carotenoid contents of several common tomato-based food products compared with whole tomatoes (pg/lOOg).
2.7.3 As colorants and additives
Carotenoids are mainly produced for use as colorants in foodstuffs as well as for pigmentation of animal products by administration in the feed. The first synthesis of 13- carotene was reported in 1950 by Inhofifen and coworkers. The Inhoffen synthesis was
64 soon developed into an industrial process that facilitated the first commercial production
of P-carotene in 1954. Since then, improvements in carotenoid synthesis have lead to a
total annual sale value of synthetic carotenoids in 1995 of approximately $300 million,
which is expected to pass the $500 million mark in less than five years. Current
worldwide production capacity of P, P-carotene alone is approaching 500 tons per year
(Britton et al., 1995). Hoffinann-La Roche AG and BASF AG, the two major industrial
producers of carotenoids, today produce six different compounds: P,P-carotene,
canthaxanthin, astaxanthin, P-apo-8 ' -carotenal, P-apo- 8 ' -carotenoic acid ethyl ester, and
citranaxanthin. Recent market prices per kilogram for stable dispersible powders
containing 5-10% active carotenoid were $600 for p-carotene, $900 for the P-apo-8 ’- carotenoids, $1300 for canthaxanthin, and $2500 for astaxanthin (Khachik et al., 1992b).
Commercial synthetic carotenoids are used both as pigments for food (egg yolks, chickens, farm-raised salmon) and for coloration of food products such as margarines and cheeses. Once the carotenoids are transformed into formulations suitable for industrial use, they can be utilized in several different applications. A microcrystalline dispersion o f p-carotene in edible fat is commonly used in the manufacture of margarines. Powders containing different carotenoids microdispersed in a hydrophilic protective colloid are used in aqueous applications such as fruit juices (Britton et al., 1995).
Natural plant pigments have also been used as food colorants. Annatto (bixin, a C 25 diapocarotenoid), paprika extracts (containing casanthin and capsorubin), alfalfa and tagetes extracts (containing xanthophylls such as lutein), tomato extracts (containing lycopene), and carrot extracts (containing a- and p-carotene) are among the natural
65 extracts being used in industry. P-carotene normally produces yellow to orange colors,
while the apocarotenoids and lycopene produce a reddish hue. Carotenoids from
marigold petals can be added to poultry feed supplements, resulting in enhanced color
quality of the poultry meat and eggs (Emenhiser et al., 1996). Kearsley and Rodriguez
(1981) noted that when added as a final action in processing, P-carotene could be used to
color certain foodstuffs such as boiled sweets.
The main drawback to the use of natural carotenoids as colorants in food is their lack
of stability. Nielsen et al. (1996) found canthaxanthin to be a more stable colorant than
P-carotene and therefore more attractive both as a coloring agent and as an antioxidant.
Autoxidation appeared to be the primary mode o f degradation o f both a- and P-carotenes
found in natural encapsulated carrot powders (Wagner and Warthesen, 1995). Chen et al.
(1995) reported that canning o f carrot juice significantly reduced its brightness, with the
overall color changing from a dark orange to a yellowish hue. Regardless, the possible
upcoming government restrictions involving FD&C synthetic colors in food products (as
well as the increasing health-consciousness of consumers) should allow natural
carotenoid colors to play a larger role as food colorants and additives in the future.
2.8 Roles of carotenoids in human nutrition
Carotenoids have long been recognized for their importance as vitamin A precursors, but more recently strong evidence has surfaced that they may be protective against certain types of cancer regardless of provitamin A status. In addition, biological functions of carotenoids such as chemical quenching of free radicals and singlet oxygen
66 have also been demonstrated in the laboratory, proving that carotenoids do indeed serve
multiple roles in human nutrition.
2.8.1 Metabolism
The metabolic fate of carotenoids from both food and supplemental sources has recently received increased attention due to their potential use in the prevention of chronic disease and vitamin A deficiency. The conversion of P-carotene (and other carotenoids that contain an unsubstituted P-ring) into retinol in the intestine is well documented. The primary pathway of all-tram-p-carotene metabolism to retinol (vitamin
A) in the intestine is first through central cleavage of the molecule by p-carotene-15,15’- dioxygenase to form retinal (Figure 22). The retinal is then presumably bound to cellular retinol-binding protein type II (CRBP II) and subsequently converted to retinol by a microsomal reductase (van Vliet, 1996). The 9-cis isomer of p-carotene is also converted to retinal by the enzyme P-carotene-15,15’-dioxygenase, though less efficiently than the
3l\-trans form.
In human intestinal tissue vitamin A formation has also been shown to occur through eccentric cleavage, resulting in P-apo-carotenal byproducts which can be converted through chain-shortening processes to retinal (Figure 22). In the liver carotenoids may also be metabolized to compoimds other than vitamin A, with these retinoid-like metabolites possibly affecting growth regulation and other cellular activities (Rock et al.,
1996). Van Vliet et al. (1996) found that increasing lutein consumption caused reduced retinal formation from P-carotene in the liver, while lycopene had no such effect.
67 eccentric all-frarts-P-carotene cleavage central (S ’) à cleavage (15-15’)
P-apo-8’-carotenal chain shortening retinal retinal reductase OH retinol
Figure 22; Pathways involved in the conversion of p-carotene to retinoids.
2.8.1.1 Absorption
An average intake of 6 mg/day of the five major carotenoids has been observed in adults in the United States (Rock et al., 1996). However, only a fraction of the total intake is utilized by the body due to the relatively low efficiency of carotenoid absorption
(10-30%). With increasing intakes of carotenoids, percent absorption is reduced even further. Dietary fiber such as pectin has also been shown to inhibit P-carotene absorption.
68 Carotenoid availability from food sources depends partially on their release from the physical matrix in which they are ingested. Heating of plant foods before ingestion has been shown to dramatically increase the bioavailability of carotenoid pigments (Gartner et al., 1997), resulting in higher plasma carotenoid concentrations (Rock et al., 1998).
Dissolution o f released carotenoids into a bulk lipid phase is also crucial for efficient intestinal utilization (Figure 23). For this reason, consumption of dietary fat along with carotenoids can significantly increase their absorption into the human body.
Food Bioodstti
LPL
omueoM /lympiy Livor />
cfiylOflUcron / / ^ nutobolHoo , — n-VLtX.1 P'
LFL
VLOL
: X X •vil A mombfrnmw * i
Figure 23: Pathways and processes involved in the metabolism of carotenoids.
69 Once carotenoids have been solubilized in bulk lipid droplets in the stomach or
intestine, bile salts and pancreatic lipases help to capture the carotenoids in micelles.
Duodenal mucosal cells absorb micelles containing the carotenoids by a passive diffusion mechanism similar to that o f cholesterol and triglyceride lipolysis products. The carotenoids are then incorporated into chylomicrons and released into the lymphatics system (Stahl and Sies, 1996).
2.8 .1.2 Tissue distribution
In order to reach the tissues of the body, carotenoids must first be transported by chylomicrons from the intestinal mucosa to the bloodstream via the lymphatic system.
The more non-polar carotenoids such as p-carotene and lycopene reside in the hydrophobic center of the chylomicron, while carotenoids with polar functional groups such as lutein and zeaxanthin are found closer to the outer surface. This differential orientation within the chylomicron molecule may affect the carotenoids’ transfer to lipoproteins during circulation, as well as their uptake by extrahepatic tissues during hydrolysis of chylomicron triglycerides.
In the bloodstream chylomicrons are broken down by lipolytic action due to lipoprotein lipase, which gives rise to chylomicron remnants. The liver clears the remnants from the blood and then can resecrete the carotenoids within lipoproteins for transport to other tissues. The distribution of P-carotene, a-carotene, and lycopene among the very low-density lipoproteins (VLDL), low-density lipoproteins (LDL), and high-density lipoproteins (HDL) are similar for all three carotenoids, with 58-73% in
LDL, 17-26% in HDL, and 10-16% in VLDL. The more polar dihydroxy carotenoids
70 lutein and zeaxanthin, however, are found predominantly in HDL (53%), with
significantly lower proportions in LDL (31%) and VLDL (16%), respectively (Parker,
1996). Van Vliet (1996) actually determined that hepatic resecretion of carotenoids such
as P-carotene occurs first by VLDL, upon which the carotenoids are then transferred to
LDL by delipidation. In short, the hydrocarbon carotenoids are transported primarily in
LDL, whereas the slightly polar carotenoids tend to be transported primarily in the more
hydrophilic HDL lipoprotein fraction (Stahl and Sies, 1996). Significant differences have
also been discovered when comparing distribution of the different isomeric forms of
carotenoids in the human body. Though the amount of sl\-trans P-carotene greatly
exceeds that of cw-P-carotene in the plasma, much increased proportions of cw-P-
carotene has been observed in peripheral tissues.
After transport throughout the body in the lipoproteins, carotenoids are primarily stored in adipose tissue in humans, although they have also been foimd in liver, lung, corpus luteum, and adrenal tissues (Rock et al., 1996). Stahl and Sies (1996) reported that the highest levels of P-carotene and lycopene have been found in liver, adrenal, and testes tissues, with lesser amoimts in lung and kidney tissues (Table 6). In addition, lycopene concentrations exceeded those of P-carotene in every tissue except the ovary.
Prolonged over-consumption of vegetable juices has the capability of increasing carotenoid levels so high that the skin and liver can gain a yellow-orange discoloration
(Stahl and Sies, 1996).
71 Tissue P-carotene lycopene
liver 1.82-4.41 1.28-5.72 kidney 0.31-0.55 0.15-0.62 adrenal 5.6-9.39 1.9-21.60 testes 2.68-4.36 4.34-21.36 ovary 0.45-0.97 0.25 - 0.28 adipose 0.38 0 .2 - 1.3 lung 0.12-0.35 0.22 - 0.57 colon 0.17 0.31 breast 0.71 0.78 skin 0.27 0.42
Table 6 : Reported human tissue levels (nmol/g) of (3-carotene and lycopene.
2.8.2 Vitamin A precursors
Perhaps the most important physiological function o f carotenoids is to act as vitamin
A precursors in animals. Most animal species are capable o f enzymatically converting carotenoids into the vitamin A compound retinol (see Figure 22). The provitamin A activity of a carotenoid depends on the presence of at least one unsubstituted P-ring. As a result, p-carotene, with its two P-rings, is the carotenoid with the highest provitamin A activity. Other carotenoids containing one P-ring and exhibiting (to a lesser extent) provitamin A activity include: a-carotene, y-carotene, 5,6- and 5,8-monepoxides of p- carotene, cryptoxanthin, and the P-apocarotenais (Table 7). Other factors contributing to
72 Carotenoid Activity (%)
all-tranj-P-carotene 1 0 0
9-c/j-P-carotene 38
13 -cw-p-carotene 53
all-/ra«j-a-carotene 53
9 -c/5-a-carotene 13
13-cfj-a-carotene 16 all-/ra«j-cryptoxanthin 57
9-c/j-cryptoxanthin 27
15-c/j-cryptoxanthin 42
P-carotene 5,6-epoxide 2 1
P-carotene 5,8-epoxide (mutachrome) 50
y-carotene 42-50
P-zeacarotene 20-40
Table 7: Relative vitamin A activity of some carotenoids commonly found in vegetables.
provitamin A activity are state o f isomerization {cis vs. tram), gastrointestinal stability, and digestibility (O’Neil and Schwartz, 1992).
‘Retinol equivalent’ was introduced in 1974 by the NAS-Recommended Dietary
Allowances Panel to describe vitamin A content in terms of international unit activity, with 1 lU = 0.3 |ag retinol. Therefore by definition:
retinol equivalent = 1 pg retinol
73 retinol equivalent = 6 qg p-carotene = 12 qg other provitamin A = 3.33 lU vitamin A activity from retinol = 10 lU vitamin A activity from p-carotene
The recommended daily allowance for vitamin A is 1000 retinol equivalents. However, it is difficult to measure total carotenoid content (and therefore vitamin A content) in foods due to the sensitive nature of carotenoids to oxygen and solvents common in present analytical techniques. There are several specific nutritional roles for vitamin A in humans (Semba, 1998). The involvement of retinal, the vitamin A aldehyde, as the chromophore of the visual pigments in the eye, is crucial to human vision. In underdeveloped locations around the world where vitamin A deficiencies frequently occur, xerophthalmia, blindness, and premature death are all too common, especially among children. Vitamin A is also crucial in maintaining growth and reproductive efficiency, as well as maintenance of epithelial tissues and prevention of their keratinization. Recently it has even been shown to improve nonheme iron absorption from fortified rice, wheat, and com flours (Garcia-Casal, 1998).
2.8.3 Additional health effects
In addition to serving as precursors of vitamin A, other important biological functions have been shown to exist for carotenoids, including antioxidant activity (singlet oxygen quenching), and prevention o f diseases such as cancer and cardiovascular disease.
74 2.8.3.1 Antioxidant protection
A biological antioxidant can be defined as “compounds that protect biological systems against the potentially harmful effects of processes or reactions that can cause excessive oxidations” (Palozza and Krinsky, 1992). The antioxidant activities of carotenoids have been demonstrated both in vitro and in vivo in several studies. Stahl et al. (1998) found that lycopene and lutein acted synergistically to increase the antioxidant effects of carotenoid mixtures containing tocopherols and carotene (a and P) in multilamellar liposomes. Martin et al. (1996) reported that carotenoid-loaded cells were partially or completely protected against oxidant-induced changes in lipid peroxidation, demonstrating that P-carotene and lutein (or their metabolites) protect HepG2 human liver cells in vitro against oxidant-induced damage independent of provitamin A activity.
However, Bast et al. (1996) found that P-carotene administered via the diet did not significantly influence liver microsomal lipid peroxidation in rats. P-carotene was shown to act synergistically with a-tocopherol as an effective radical-trapping antioxidant in liver microsome membranes (Palozza and Krinsky, 1992). In a similar study, P-carotene supplementation increased the induction period and decreased PC-OOH production in plasma during AAPH-induced lipid peroxidation. The findings suggested that dietary p-carotene may play an important role in the overall antioxidant defense system of plasma (Meydani et al., 1994).
In vivo studies have shown similar results. Lepage et al. (1996) concluded that p- carotene deficiencies in children with cystic fibrosis led to excessive lipid peroxidation.
Subsequent p-carotene supplementation reduced serum malonaldehyde concentrations,
75 indicating the carotenoid acted as an antioxidant in vivo. Ribaya-Mercado et al. (1995)
found that when human skin was subjected to UV irradiation in vivo, skin lycopene and
P-carotene concentrations were lowered in a manner consistent with the consumption of
free radicals through quenching processes. The researchers concluded that the
antioxidant actions of lycopene and p-carotene might be an important defense mechanism
against the adverse effects of UV irradiation on the skin.
2.8.3.2 Cancer prevention
Studies have consistently sho-wn that individuals with the highest intakes of carotenoid-rich fruits and vegetables have the lowest risks for cancers such as lung, oral cavity, stomach, and esophagus (Krinsky, 1994). Pool-Zobel et al. (1997) determined that men on a diet supplemented with tomato, carrot, or spinach products showed significantly lower endogenous levels of strand breaks in lymphocyte DNA. High serum levels o f carotenoids have also been correlated with decreased risks for certain cancers
(Peto et al., 1981; Bendich, 1993; van Poppel, 1996). P-carotene has been shown to be particularly effective against smoking-related cervical intraepithélial neoplasia and cervical cancer (Charleux, 1996). Garewal (1995) found evidence o f a chemopreventative role for the carotenoid P-carotene against oral cavity cancer, the sixth most frequent cancer in the world. In another study, P-carotene supplemented elderly men had significantly greater natural killer cell activity than elderly men receiving placebos. Because natural killer cells are crucial in the body’s fight against tumor
76 growth, the increased activity can be correlated with decreased cancer risks (Santos et al.,
1996).
Lycopene has shown perhaps the most anti-cancer potential of all the carotenoids.
Giovannuci et al. (1995) foimd that among the carotenoids p-carotene, a-carotene, lutein, lycopene, and P-cryptoxanthin, only lycopene intake was related to a lower risk of non stage A1 prostate cancer. O f the four vegetables or Bruits that were found in the study to be significantly associated with lower prostate cancer risk, three (tomato sauce, tomato juice, and pizza) happened to be primary sources of lycopene. Dorgan et al. (1998) determined that the risk of developing breast cancer significantly decreased as serum concentrations of lycopene and lutein/zeaxanthin increased in women. Nagasawa et al.
(1995) examined the effects of chronic ingestion o f lycopene on the development of spontaneous mammary tumors in SHN virgin mice. The lycopene treatment significantly suppressed the mammary tumor development compared with the non-lycopene-fed control group. Sharoni et al. (1995) stated that lycopene inhibited tumor gro wth both in vitro and in vivo. In vitro treatments of lycopene inhibited the growth of human skin fibroblasts, while lycopene treated rats developed fewer and significantly smaller mammary tumors in vivo when compared to the control or P-carotene treated rats.
Lycopene delivered in cell culture medium from stock solutions in tetrahydrofuran more strongly inhibited proliferation of endometrial, mammary, and lung human cancer cells than did either a- or P-carotene (Levy et al., 1995).
77 2.8.3.3 Cardiovascular disease
Carotenoids have been linked to reduced risk for the development of cardiovascular disease. Krinsky (1994) reported on findings showing a distinct inverse association between serum carotene levels and ischemic heart disease. A study of U.S. health professionals showed a similar association between dietary intake of (3-carotene and heart disease risk. Men with the highest dietary (3-carotene intakes had a 29% decrease in heart disease risk. Further analysis, however, determined that heart disease risk was reduced
70% in current smokers with high (3-carotene intakes and 40% in former smokers, with lifelong nonsmokers actually having no significant correlation between (3-carotene intake and risk of heart disease. In a parallel study o f women nurses, those in the highest one- fifth of the population in terms of (3-carotene intake showed a 22% reduction in heart disease risk (Charleux, 1996).
2.9 Carotenoid oxidation/degradation
Due to their conjugated polyene backbone, most carotenoids are fairly unstable molecules and as a result are very sensitive to light, oxygen, and elevated temperatures.
The presence of these factors can cause oxidative degradation of carotenoids, resulting in destruction of the parent compound and formation of a variety of oxidized by-products.
2.9.1 Effects of oxidizing agents
El-Tinay and Chichester (1970) used the radical initiator azo-bis-isobutyronitrile
(AIBN) to accelerate the formation P-carotene oxidation products. AIBN is capable of
78 thermal breakdown to a radical species (R-) that can rapidly react with oxygen to form
peroxyl radicals (ROO-)- The resulting initial oxidized product was reported as P-
carotene-5,6- and 5,8-epoxides, with subsequent decomposition to other products such as
p-carotene-5,6-5’,6’-diepoxide derivatives. Handelman et al. (1991) disputed the former
claim, suggesting rather that the AIBN-initiated radical attack on P-carotene occurred at
multiple sites on the molecule, creating a ‘series’ of apo-carotenals of different sizes such
as retinal, P-apo-14’-carotenal, P-apo-12’-carotenal, and P-apo-10’-carotenal.
Yamauchi et al. (1993) utilized a similar radical initiator, 2,2’-azobis(2,4-
dimethylvaleronitrile) (AMVN), to allow alkylperoxyl radical oxidation of P-carotene.
The major resulting oxidized P-carotene products included 12-formyl-l l-nor-P,p~ carotene, 15’-fbrmy 1-15-nor-P,P-carotene, 5,6-epoxy-5,6-dihydro-P,P-carotene, and 19- oxomethyl-lO-nor-p,P-carotene. Using perphthalate as a chemical oxidizing agent, Seely and Meyer (1971) reported P-carotene-5,6-monoepoxide as the principle product of oxidation. Copper stearate was also effective as a catalyst in the degradation of lycopene to smaller oxidized products (Cole and Kapur, 1957). Khachik et al. (1998) prepared the oxidative metabolites lycopene 1,2-epoxide and lycopene 5,6-epoxide by oxidizing all-
/ra«5 -lycopene with m-chloroperbenzoic acid (MCPBA), followed by acid hydrolysis.
Micro-Cel C, a common chromatographic adsorbent, was found to react with various carotenoids to yield hydroxides and epoxides when exposed in the presence of a nonpolar solvent, a-carotene was converted to 4-hydroxy-a-carotene, while p-apo -8 ’ -carotenal underwent hydroxylation at the allylic 4-position of the P-ring. Lycopene was
79 completely oxidized by the Micro-Cel C, leaving lycopene 5,6-epoxide, 6 ’-apoIycopenal,
and Iycopene-5,6-diol as degradation products (Ritacco et al., 1984; Ritacco et al., 1984).
Photosensitizers can also be effective initiators of carotenoid oxidation. Seely and
Meyer (1971) utilized hypericin, a po^verful photodynamic agent, in the photosensitized
oxidation o f (3-carotene to yield products such as mutatochrome and aurochrome. Lutein
and zeaxanthin were found to undergo the photooxidative process more slowly than 13-
carotene. By irradiating lycopene in an 0% atmosphere with the presence of methylene
blue as a photosensitizer, Ukai et al. (1994) identified several oxidized products of
lycopene, including 2-methyl-2-hepten-6-one and apo- 6 ’-lycopenal (Figure 24).
Holman (1949) first investigated the oxidation of (3-carotene in imsaturated oils,
reporting that in the medium of an oxidizing imsaturated lipid, an intermediate product of
fat oxidation stimulated (3-carotene oxidation. Similar results were found in a later study
that determined the addition of unsaturated oil resulted in a shorter induction period for
the autooxidation of (3-carotene and vitamin A in a paraffin solution. The prooxidant effect of the oil increased with an increasing iodine value and degree of imsaturation
(Budowski and Bond, 1960). Camevale et al. (1979) disputed these previous findings by stating that increased imsaturation o f oil offers protection against autooxidation of carotenoids, because a higher degree o f unsaturation in oil provides a substrate diversion away firom the (3-carotene molecule, resulting in a lower rate of carotenoid oxidation.
Oxygen has been used to initiate carotenoid oxidation. By passing a slow current of pure oxygen through a solution of lycopene in hexane. Cole and Kapur (1957) claimed the oxidative products of lycopene to be acetone, methylheptenone, and laevulinic aldehyde.
80 lycopene
O2, hv methylene blue
2 -methyl- 2 -hepten-6 -one
apo-6 ’-lycopenai
Figure 24: Possible reaction pathway for photosensitized oxidation of lycopene.
Ben-Aziz et al. (1973) later isolated a series of epoxides and apo-lycopenals from tomatoes, including 1,2-epoxy-l,2-dihydro-v|/,\|/-carotene and 5,6-epoxy-5,6-dihydro- v|/,vj/-carotene, which may have been early products of the oxidative degradation of lycopene due to tissue senescence and/or physical injury. Teixeira Neto et al. (1981)
81 examined the kinetics of P-carotene oxidation in a model system of “nonfat” dry milk,
AVTCEL microcrystalline cellulose, and crystalline P-carotene (simulating a dehydrated
food product). Carotenoid oxidation could be accurately predicted by colorimetric
assessment of the decoloration of P-carotene in the model system.
2.9.2 Effects of thermal processes
Carotenoid pigments possess an extended polyene structure that causes them to be
very reactive in the presence of light and/or heat (Minguez-Mosquera and Jaren-Galan,
1995). Because carotenoids such as p-carotene and annatto (bixin) are widely used in
industry as food colorants, thermal degradation (and resulting loss of color) during
processing is of great concern to food manufacturers and nutritionists. Common
degradation products from the heating of p-carotene can be seen in Figure 25.
Marty and Berset (1986) compared the degradation of all-/ra«j-P-carotene during two
thermal processes: heating in sealed glass tubes, and extrusion cooking. After heating
the all-frawi'-p-carotene in sealed glass tubes for 2 hours at 180°C, the main degradation
products were identified as P-carotene-5,6-epoxide, p-carotene-5,6,5’,6’-diepoxide, and
P-carotene-5,8-epoxide. Similar compoimds were found after extrusion cooking of all-
truMf-P-carotene, with P-carotene-5,6,5’,8’-diepoxide also being present. Marty and
Berset (1988) found that after extrusion cooking, only 8 % of the all-rra«j-p-carotene
remained, with the other 92% of the P-carotene degrading into one of six main groups: 1)
mono- or poly-cfj stereoisomers, 2) a diepoxide derivative, 3) five apo-carotenals, 4) a polyene ketone, 5) a dihydroxide derivative, or 6 ) a monohydroxide diepoxide derivative.
82 mutatochrome
5,6,5’,6’-diepoxy-P-carotene
aurochrome
luteochrome
Figure 25: Nonvolatile compounds often formed during heating of p-carotene.
The authors concluded that the resistance of all-trans-P-carotene to high temperatures depends largely on the processing conditions. Different thermal treatments resulted in all-/ra«5-P-carotene losses between 7.5% and 92%. Prolonged heating at 180°C caused
83 only limited breakdown o f the carotenoid molecule, but the presence o f other constituents such as starch and/or water combined with mechanical mixing favoring incorporation of additional oxygen lead to much higher losses of all-/ra«^-p-carotene (Marty and Berset,
1990).
Ouyang et al. (1980) identified the main decomposition products of P-carotene formed during a simulated commercial deodorization o f pahn oil to be P-13-apo- carotenone, P-15-apo-carotenal, and P-14’-apo-carotenal. When P-carotene was heated in glycerol at 210°C for 5 min., 15 min., 1 hour, and 4 hours (to simulate the time/temperature combinations seen in deep fat frying and edible oil deodorization), respectively, over seventy nonvolatile compounds were observed by ER/MS (Onyewu et al., 1986).
A study examining the effect o f microwave cooking on the stability o f carotenoid pigments in sweet potato leaves demonstrated that the epoxy-containing carotenoids were more susceptible to heat loss than other carotenoids. Two lutein dehydration products were identified in the sweet potato leaves after microwave processing: 3,4-didehydro- p,e-caroten-3’-ol, and 3,4-didehydro-P,P-caroten-3-ol (Chen and Chen, 1993). Khachik et al. (1992b) also found the epoxycarotenoids present in foods such as green vegetables and tomatoes to be more sensitive to the thermal processes involved in microwaving, boiling, steaming, and stewing than the hydrocarbon carotenoids such as neurosporene, a - and p-carotene, lycopene, phytofluene, and phytoene. Godoy and Rodriguez-Amaya
(1987) examined the effects of thermal processing on both mango slices and puree. The only significant change in the mango slices after heat treatments was an increase in the
84 measured luteoxanthin content. In the processed mango puree, processing at 80°C for 10 minutes resulted in a 13% decrease in P-carotene content, 33% decrease in violaxanthin, and an increase in the auroxanthin content due to a 5,6- to 5,8-epoxide transformation.
Cole and Kapur (1957) were among the first to examine the extent of lycopene breakdown as a result of exposure to elevated temperatures. The authors reported lycopene losses of 15% and 25% in 3 hours of thermal treatment at 65°C and 100°C, respectively. A heat treatment of 97°C in water produced several novel P-carotene thermal degradation products, including decanal, 4-ethylbenzaldehyde, and cetoisophorcne. The compound 5,6-epoxy-P-ionone was also shown to be an important reaction intermediate, acting as a precursor for various volatile compounds such as P- ionone and 2-hydroxy-2,6,6-trimethylcyclohexanone (Kanasawud and Crouzet, 1990).
Using a similar thermal process, Kanasawud and Crouzet (1990) determined the resulting degradation products of lycopene by GC/MS analysis. The main characterized products included 2 -methyl- 2 -hepten-6 -one and citral, as well as the previously uncharacterized compounds 5-hexen-2-one, hexane-2,5-dione, 6-methyl-3,5-heptadien-2-one, and geranyl acetate (Figure 26). AU-/raMS-lycopene was also found to partially isomerize to the cis- trans isomer as a result of the heat treatments. Henry et al. (1998) reported that in a safflower seed oil model system, the rates of thermal degradation for selected carotenoids between 75°C and 95°C were as follows: lycopene > all-p-aws'-P-carotene = cw-p- carotene > lutein.
85 aIl-/ra«5-P-carotene
Cs-Q C9 -C10 Cleavage Cleavage C7-C8 Cleavage
6-methyl-3,5-heptadien-2-one 2-methyI-2-hepten-6-one
CHO
geranial neral
Figure 26: Reaction sequence for the formation of volatile compounds during heat treatment of lycopene.
Chandler and Schwartz (1988) examined changes in the carotene content of sweet potatoes subjected to one of several different thermal processes. All-frnw-P-carotene was found to be more susceptible to isomerization reactions resulting in the formation of
CIS isomers such as 13- and I5-c/j-P-carotene than degradation reactions during most processing treatments. The extent of isomerization was largely related to the severity and
86 length of the individual heat treatment. Chen and Chen (1995) reported extensive
isomerization in carrot carotenoids during canning (121°C, 30 min.) and HTST heating
(120°C, 30 sec.) thermal processes. The formation of several cis isomers of P-carotene,
including 13-c/5-P-carotene, 13-c/j-lutein, and IS-c/^-a-carotene may have been
responsible for the carrot juice color change from orange to yellow during the intensive
heat treatments. Ogunlesi and Lee (1979) reported a substantial increase in the
concentration of cis isomers and a 25-35% decrease in dl\-trans isomers of P-carotene
after retorting processed carrots, resulting in a 15% decrease in the vitamin A value.
2.9.3 Effects of storage
In addition to changes incurred as a result of processing, carotenoids are also capable
of undergoing alterations in their composition and structure as a result of simple extended
storage. During storage of mango slices in lacquered or plain tin-plate cans, no
significant loss of p-carotene was observed after 10 months of storage. However, upon
extended storage of the slices, a 50% reduction in total p-carotene was observed after 14
months, with continued degradation resulting in a P-carotene loss of 84% after 24
months. Other carotenoids such as violaxanthin and luteoxanthin also decreased in
quantity during storage, while auroxanthin levels remained constant (Godoy and
Rodriguez-Amaya, 1987).
Kopas-Lane and Warthesen (1995) determined that light promoted carotenoid pigment losses in raw spinach, with 60% of the violaxanthin and 2 2 % of the lutein present in the spinach being degraded after only 8 days of storage. Storage of raw
87 spinach in the dark did not affect spinach carotenoid levels, except for an 18% loss of all-
/rawj-p-carotene. For raw carrots, however, neither hghted nor dark cold storage affected
the major carotenoids. In a similar study, degradation of lycopene in a vegetable juice
model system was about one-fifth that of a- and p-carotene after an 8 -day storage period
(Pesek and Warthesen, 1987). In addition, Wagner and Warthesen (1995) revealed that
degradation o f a- and p-carotene during storage at 37°C occurred at the same rate.
2.9.4 Effects o f encapsulation in minimizing degradation
Though the trend in the food industry is towards natural products as opposed to
synthetic additives, carotenoids are limited by application problems. For example,
creating acceptable water-soluble forms of carotenoids is extremely difficult due to the
hydrophobicity of pure carotene crystals. Encapsulation provides a method to transform
liquids such as solubilized carotenes into stable free-flowing powders that can be easily
incorporated into aqueous food systems.
Wagner and Warthesen (1995) determined that hydrolyzed starch of 36.5 DE was more effective than 25, 15, and 4 DE in improving carotene retention during storage, with encapsulated carotenes enjoying a predicted half-life of 450 days at 21 °C, compared with
2 days for the spray-dried carrot juice control. In another study, the stability of p- carotene encapsulated in 25 DE maltodextrin by spray drying, freeze drying, and drum drying was evaluated. After 15 weeks o f storage, drum drying gave the best p-carotene preservation o f all the encapsulation methods. Due to its smaller particle size and surface
88 carotenoid content, the spray-dried encapsulated p-carotene showed the fastest degradation, with 80% degraded after a seven-week storage period at 45°C.
2.10 Isolation and identification of carotenoids
Isolation of carotenoids from biological sources usually involves extraction, saponification, and separation processes. Once separated, identification of individual carotenoids is often based on a complex combination of spectrometric, chromatographic, and chemical tests. High performance liquid chromatography (HPLC), thin layer chromatography (TLC), Mass spectrometry (MS), Gas chromatography (GC), Nuclear magnetic resonance (NMR) spectroscopy, Ultraviolet/Visible (UV/Vis) spectroscopy, and circular dichroism (CD) are all routinely used for separation and/or structural elucidation of carotenoids and their oxidation/degradation products.
The analysis of carotenoids is further complicated due to their structural instability, tendency to stereomutate, photo- and thermolability, and propensity towards oxidation.
As a result, all analytical experiments must be carried out in dim light with inert (under nitrogen or vacuum) atmospheric conditions. In addition, solvents must be purified, environmental temperatures must be no higher than 40°C, and samples must be dried and stored at -20°C under nitrogen (Schiedt and Liaaen-Jensen, 1995).
2.10.1 Extraction procedures
Extraction of carotenoids firom biological materials must be done as rapidly as possible to minimize oxidative and/or enzymatic degradation. Often blanching of plant
89 tissue and addition of calcium carbonate (CaCOs) and antioxidants such as butylated
hydroxytoluene (BHT) is included before extraction to minimize enzymatic reactions,
acid hydrolysis, and oxidative degradation, respectively. To facilitate maximal extraction
yields material should be ground into small pieces. The lipophilic carotenoids require
organic solvents that are free of oxidizing compounds, acids, or halogens for efficient
extraction to take place (Figure 27).
Usually extractions are carried out in a blender so that grinding and extraction can
occur simultaneously. Hart and Scott (1995) prepared samples for carotenoid analysis by
freezing them in liquid nitrogen and then grinding them under liquid nitrogen with a
Waring blender. Homogenates can be suction filtered through a Buchner funnel lined with filter paper coated with Celite (Silveira, Jr., and Evans, 1995). The extraction process can be repeated several times until all visible pigment is extracted from the source material (Khachik et al., 1992b).
2.10.1.1 Solvents
Different solvents have been used with varying success in the extraction of carotenoids from biological sources. Hakala and Heinonen (1994) extracted 6 mg of lycopene from 10 grams o f tomato puree using both petroleum ether and acetone.
Recovery and purity o f the lycopene were both better in the sample extracted with petroleum ether compared to that extracted with acetone, due to a smaller portion of polar xanthophylls being extracted with the more nonpolar petroleum ether solvent.
Tetrahydrofuran (THF) was used as the solvent to extract carotenoids from several different raw and cooked vegetables for HPLC analyses. Complete extraction of the
90 WET BIOLOGICAL DEHYDRATED MATERIAL BIOLOGICAL MATERIAL
CRUDE EXTRACT
Extraction into ether/hexane Evaporation
DRY LIPID EXTRACT Partition hexane/aqueous 85% methanol
T i HYPOPHASIC EPIPHASIC NEUTRAL ACIDIC CAROTENOID CAROTENOID CAROTENOID CAROTENOID Partition
1 1 HYPOPHASIC EPIPHASIC CAROTENOID CAROTENOID
^ Chromatography CCC. TLC, HPLC) ^
INDIVIDUAL CAROTENOIDS Rechromatography Crystallization Recrystallization
PURE CAROTENOID Characterization
Figure 27: Sample procedure for isolation o f carotenoids from a natural source.
91 A wrist-action shaker containing a solvent mix of hexane-acetone-ethanol (50:25:25) was
utilized to extract carotenoids from tomato puree. Fifteen mLs of water was added to the
mix to allow improved separation into distinct polar (clear aqueous) and nonpolar
(lycopene-containing red) layers.
Craft and Soares, Jr. et al. (1992) reported the relative solubility, stability, and
absorptivity o f several carotenoids in eighteen different organic solvents. Results showed
that the solubility of both lutein and P-carotene was highest in tetrahydrofiiran (THF),
hexane exhibited the least solubility for lutein, and both methanol and acetonitrile
exhibited the least solubility for P-carotene. Cyclohexanone caused the most degradation
of carotenoid pigments after ten days o f storage, with only 37% of lutein and 32% of P-
carotene absorbance remaining, respectively.
2.10.1.2 Supercritical fluids
The increasing demand for natural p-carotene has resulted in a growing interest in rapid, cost-effective methods of carotenoid extraction from plant sources. Most extraction methods presently involve the use of organic solvents, which are generally undesirable due to exposure to potentially toxic compounds, as well as environmental concerns. Vega et al. (1996) reported a maximum 99.5% extraction of P-carotene from carrot pulp using supercritical carbon dioxide extraction with 1 0 % ethanol as a co solvent. Concentration of ethanol and temperature were determined to be the most important factors in determining extraction yield. High efficiency extractions of P-
92 carotene from natural sources such as carrot pulp were determined to be feasible by
supercritical CO2 + ethanol extraction methods.
2.10.2 Quantitation
Because carotenoids generally obey the Beer-Lambert law, their quantitative determination is often accomplished by spectrometric methods in which the absorbance of a known volume of carotenoid solution is read at the wavelength o f maximal absorption. Carotenoid content can therefore be determined in pg/g material by using the following equation:
pg carotenoid/g = A x V x 10^ A ‘’‘la. X 100 X G where V is the total volume (mL) containing G grams of sample, and A ”‘icm is the specific absorbance or extinction coefBcient. Specifically, the extinction coefficient is defined as the theoretical absorbance o f a 1% solution (w/v) in a 1 cm path-length cuvette. For colored carotenoids, extinction coefficient values are usually around 2500, so a solution with a carotenoid concentration of 1 pg/mL would give an absorbance (A) of approximately 0.25 (Davies, 1976).
2.10.2.1 UV/Vis Spectroscopy
The pursuit of accurate carotenoid quantitation is most commonly accomplished by using spectroscopy in the UV/Visible region of the spectrum. Kearsley and Rodriguez
(1981) utilized UV/Vis Spectroscopy to determine the content and stability of (3-carotene in solution after exposure to thermal treatments, light, and changes in pH.
93 2.10.2.2 HPLC-PDA
High performance liquid chrcmatography-photodiode array detection (HPLC-PDA) has also more recently been used to quantitate carotenoids from various sources. The use of a photodiode array detector to quantitate carotenoids in biological extracts depends upon calibration with authentic source substances, or standards. Analysis precision is greatly increased by the addition of internal standards early in the analysis. Peak height and peak area ratios of the compound o f interest vs. the internal standard can be utilized for quantitation. In the absence of standards only semiquantitative results can be compiled (De Leenheer and Nelis, 1992).
2.10.3 Separation
Chromatography is perhaps the most important single technique in the separation of carotenoid pigments. The separation process of chromatography is based upon two phases known as the stationary phase and the mobile phase. A mixture of compounds is added to the mobile phase that is subsequently carried through the chromatographic system. As the mobile phase passes through the stationary phase, each compound in the mixture reaches an equilibrium distribution at a specific point between the two phases, resulting in differential migration rates through the system (Pfander, 1995).
The compounds to be separated can interact with the two phases in two specific ways, partition and adsorption. Partitioning occurs if the sample mixture diffuses into the interior of a liquid stationary phase. The latter term is applicable if the sample mixture is attracted to the surface of a solid stationary phase. Many adsorbents are utilized in carotenoid chromatography, among them starch, CaCOs, MgCOs, AI 2O3, and silica gels.
94 However, these various absorbents can achieve separation o f compounds in different
ways. Materials such as CaCO] and CaCOs have affinities for double-bond systems, so
separation of carotenoids is determined by the number and type of double bonds in the molecule. Adsorbents such as alumina or silica, on the other hand, separate carotenoid groups of differing polarity such as hydrocarbons, monohydroxy carotenoids, and polyhydroxy carotenoids according to polarity, with the most polar carotenoids being most strongly adsorbed.
Silica and alumina are now the most widely used stationary phases for the separation of carotenoids, and are usually used in bonded-phase chromatography. Bonded phases are prepared by modifying the reactive groups on the surface of the stationary phase
(silica or alumina). To do this, the silanol groups (SiOH) are reacted with an alkylating agent such as dimethyloctadecylchlorosilane (ODS) to form a Cig bonded phase.
Trimethylchlorosilane can also be reacted with the silanol groups in a second alkylation process, known as endcapping, which minimizes the number of remaining silanol groups
(Figure 28).
2.10.3.1 HPLC
High-performance liquid chromatography (HPLC) is by far the method of choice for carotenoid analysis for many reasons. Its ability to distinctly separate many compounds in a relatively short time proves its efficiency. HPLC analyses can be highly sensitive, with the detection of small amounts of impurities, trace carotenoids, and/or geometrical isomers being possible. Finally, HPLC analyses with a photodiode array detector and computer-aided data processor can yield much information about the sample being
95 Si—OH Residual silanol group
CH3
Si1-—0—^i—-(CH2)i7CH3 Bonded ODS group
CH3
CH3
Si1—0 "—^ i—CH3 Endcapped silanol
CH3
Figure 28: Main features of a Cig bonded-phase silica material.
analyzed. High-performance liquid chromatography, which has also been referred to as
‘high-pressure liquid chromatography’, includes liquid chromatographic methods that used stationary phases of particle size not > 1 0 pm and pressures > 2 0 bar (300 psi).
An HPLC system is comprised of several important components that can be viewed schematically in Figure 29. The pump is responsible for producing the pressures necessary to force mobile phase through the stationary phase particles. A sintered metal frit is often placed after the solvent reservoir to prevent solid impurities from reaching the
96 I * _ : (— A 9
7 10 :
r-5-0:>-5SM' — — - __ ; ■III , 0 13
[irwri 14
Figure 29: Schematic diagram o f a typical HPLC system. 1. solvent reservoir, 2. sintered metal frit, 3. high-pressure pump, 4. pulse damper, 5. drain valve, 6 . pressure gauge, 7. pre-column, 8 . injection syringe, 9. injection valve, 10. column, 11. thermostatted oven (optional), 12. detector, 13. recorder/integrator/plotter, 14. fraction collector.
pump and damaging it. A pulse damper can lessen or even remove the pulsations that result from pump action. A pre-column is most commonly inserted between the sample injector and the separation column to protect the latter from impurities in the sample or mobile phase. The separation column itself is usually constructed from 316-grade stainless steel, which is resistant to the high HPLC pressures as well as most chemical corrosion. The detectors utilized in HPLC analyses are usually UV/Vis due to their sensitivity and ease in operation (Johnson and Stevenson, 1978). The development of photodiode-array (PDA) detectors has been very helpful in the analysis of carotenoids due to their rapid data acquisition and ability to store entire spectra for later comparison
(De Leenheer and Nelis, 1992; Pfander and Riesen, 1995). Most recently, electrochemical detection (BCD) has been shown to be a particularly useful alternative to
97 UV/Vis detection methods for LC analyses requiring extremely high (finol) sensitivity
(Ferruzzi et al., 1998).
Both UV/Vis and PDA detectors generate chromatograms that consist of a curved
peak for each separated compound from the sample. The chromatogram provides the
following information to the researcher: 1) the retention time of the compound(s), and 2 )
the area and/or height of the peak(s). Retention times of unknown compounds and
standards can be compared for tentative identification purposes. The area/height of a
peak can be used, with the help of a calibration curve, to estimate the relative amounts of
each compound present in the sample.
Carotenoid separations can be accomplished by either normal-phase or reverse-phase
HPLC. Normal-phase employs adsorptive phases such as silica and alumina, as well as
polar bonded phases such as alkylamine or alkylnitrile in combination with nonpolar
mobile phases. In this type of HPLC, polar sites on the carotenoid compete for
adsorptive sites on the stationary phase, resulting in the least polar carotenoids
(carotenes) eluting first, while the more polar oxygenated carotenoids (xanthophylls) are retained in the column longer. Reverse-phase HPLC includes nonpolar bonded phases
(Cg and Cig) and polymer phases in conjunction with polar mobile phases. During reverse-phase HPLC, xanthophylls are more induced to stay in the polar mobile phase and therefore elute first, while the carotenes partition preferentially into the stationary phase and elute later. Both normal-phase and reverse-phase HPLC can be used with the same mobile phase solvent compositions throughout the analysis (isocratic), or the solvent compositions can change during the analysis (gradient) (Craft, 1992).
98 The investigation of carotenoids by HPLC has been ongoing for almost 30 years, with a wide variety o f papers being published on the subject during that time. Handelman et al. (1991) utilized a mobile phase of 85% acetonitrile/15% methanol through a Cig column in a gradient elution process in order to separate the degradation products of 13- carotene. Ammonium acetate (0.01%) was added to the initial mobile phase to help increase HPLC recovery of carotenoids. Several apocarotenoids (autooxidized products of P-carotene) were recorded on chromatograms at a wavelength of 350 nm. In a similar study, epoxide products of P-carotene antioxidant reactions were separated by reverse- phase HPLC using a mobile phase of methanol-hexane (85:15 v/v) with a flow rate of 1.5 mL/min. Further resolution of epoxide products was successfully accomplished through the use of a cyano-column (Liebler and Kennedy, 1992). Marty and Berset (1990) separated thermally oxidized P-carotene compounds by using a HPLC elution system of n-hexane/diethyl ether (95:5 v/v) with a flow rate of 1 mL/min.
HPLC has been employed as an effective technique to separate and quantify various carotenoids present in human plasma samples. Eighteen different carotenoids, including vitamin A, were separated from extracts of human plasma by HPLC on reversed-phase
Cig silica-based nitrile-bonded columns (Khachik et al., 1992a). Epier et al. (1993) developed a HPLC method for quantitative measurement of the six major carotenoids foimd in human serum. The mobile phase consisted o f a mixture of acetonitrile, methanol, and ethyl acetate, each containing 0.05% triethylamine (TEA) to increase carotenoid recovery. In addition, ammonium acetate (0.05 M) was also added to the methanol to minimize carotenoid losses on the column. Stahl et al. (1993) separated five
99 geometrical isomers of P-carotene and seven of lycopene in human serum and tissues using improved reverse-phase HPLC methods. Mobile phases consisted of methanoI/acetonitrile/2-propanoI (54/44/2), or methanoI/acetonitrile/ 2 -propanoI/H2 0
(10/40/40/10) with a flow rate of 1.0 mL/min and detection at 460 nm.
The carotenoid content of human serum was first examined using a non-aqueous reverse-phase (NARP) HPLC method developed by Nelis and De Leenheer (1983).
NARP mobile phases used included mixtures of acetonitrile, dichloromethane, methanol, tetrahydrofuran (THF), and ethyl acetate. The NARP method showed superior sample solubility o f nonpolar carotenoid components when compared with conventional aqueous reverse-phase chromatography, mainly due to the fact that carotenoids typically are only sparingly soluble in partially aqueous solvents.
Several HPLC methods have been developed to determine the carotenoid contents of firesh or processed fiaiits and vegetables both qualitatively and quantitatively. The major carotenoid constituents of extracts firom several raw and cooked green vegetables
(broccoli, green beans, and spinach) (EChachik et al., 1992b), as well as firesh tomatoes and tomato paste (Khachik et al., 1992b; Tonucci et al., 1995) have been separated by
HPLC on a Cig-reverse phase column. An isocratic mixture o f acetonitrile (85%), methanol (10%), dichloromethane (2.5%) and hexane (2.5%) was used for the mobile phase, with chromatographic analyses simultaneously being monitored at several different wavelengths (Khachik et al., 1992b). Tan (1988) also identified the carotenoids present in tomato paste by using isocratic reverse-phase HPLC with several different
NARP solutions. Carotenoids firom apricot, peach, cantaloupe, and pink grapefiruit extracts were also separated and quantitated on a Cig-reverse phase column (Khachik et
100 al., 1989). Kopas-Lane and Warthesen (1995) developed a reverse-phase gradient HPLC
method on a Cig column for the separation of xanthophylls, carotenes, and cis P-carotene
isomers from raw spinach and carrots. Initial conditions for the mobile phase included
90% acetonitrile/5% water/5% methanol, with a linear gradient increasing the methanol
content to 100% in 15 minutes. An HPLC solvent system of
acetonitrile/methanol/chloroform/hexane (75:12.5:7.5:7.5 v/v/v/v) pumped at a flow rate
of 1.0 mL/min was employed to determine the stability of carotenoids in sweet potato
leaves to microwave cooking. Eluate was monitored at 440 nm and comparing retention times of separated peaks with those o f the reference standards (Chen and Chen, 1993)
identified carotenoids.
More recently novel polymeric C 30 stationary phase columns have been shown to be effective in HPLC separation of carotenoid compounds and even geometric isomers
(Sander et al., 1994). Using such a column, Emenhiser et al. (1995) was able to obtain superior resolution of several isomers of the asymmetrical carotenoids lutein, a-carotene, and P-cryptoxanthin, as well as the symmetrical carotenoids zeaxanthin and P-carotene.
A mobile phase of methyl-terf.-butyl-ether (MTBE) in methanol was used isocratically to achieve separations on the polymeric C 30 stationary phase. The same C 30 column was later used by Emenhiser et al. (1996) to distinguish the different geometrical carotenoid isomers present in human serum, carrots, algae extract, and a poultry feed supplement.
Again a methanol-MTBE mobile phase was utilized for chromatographic analysis.
Addition o f triethylamine (0.1% v/v) to the mobile phase was found to increase the recovery o f p-carotene from the C 30 column from 52% to 88%.
101 Another proposal to increase HPLC resolution between carotenoid isomers was made
by Schmitz et al. (1995). The use of a calcium hydroxide stationary phase in HPLC
analyses was advocated for the specific purpose of increasing the resolution between the
various members o f the acyclic and cyclic geometric carotenes containing 5, 7, or 11
aliphatic double bonds.
Depending on how HPLC methods are used in the separation and identification of
carotenoids, occasionally unanticipated results have been known to occur. Piretti et al.
(1996) found that when using a normal-phase cyano-amino HPLC column, lycopene
samples and standards both exhibited diSering numbers of peaks depending on the
solvent used to prepare the sample for injection. Scott et al. (1992) reported that in
addition to reactions between carotenoids, injection solvents, and the mobile phase, metal
surfaces such as the stainless steel of metal frits in HPLC systems may be damaging to
carotenoid compounds. In addition, changes in ambient temperature during HPLC
analyses of carotenoids were reported to cause dramatic differences in the data collected.
A reduction of 1 minute in elution time for every 1°C rise in temperature was described,
with optimum resolution occurring at 20-22.5°C (Scott and Hart, 1993).
2.10.3.2 TLC
Thin-layer chromatography (TLC) first gained popularity in the 1950’s as a replacement for paper chromatography. Though more recently TLC has tended to be superseded by the more efficient and sensitive HPLC systems, TLC is still a simple and inexpensive method that is often utilized for pilot screenings of carotenoid mixtures of
102 unknown composition. The results of TLC screenings often direct the choice of conditions for subsequent preparative chromatography and/or HPLC (Schiedt, 1995).
TLC is generally applied on a micro- or semimicro scale, which permits rapid and sharp separation of the various compounds present in the sample as well as detection of substances at the trace level. Usually silica gels are used as the adsorbent; basic oxides or carbonates such as MgO are also commonly used. Tentative identification of a compound by TLC is based on comparison of its Rf value (the distance moved by the solute divided by the distance moved by the mobile front) with that of an authentic standard.
Carotenoid thermal degradation products have often been separated by TLC methods.
Kanasawud and Crouzet (1990) used aluminum oxide TLC plates of 1.5 mm thickness with an acetone-petroleum ether (4:95 v/v) elution solvent to separate the various non volatile thermal degradation products of P-carotene. In a similar study, alumina TLC plates separated five nonvolatile compounds produced by heat treatment of lycopene.
Separation by TLC was followed by determination of absorption spectra by UV/Vis spectroscopy (Kanasawud and Crouzet, 1990). The degradation products of P-carotene from heating in sealed glass tubes and extrusion cooking were fractionated on an alumina
TLC plate using 10% diethyl ether in n-hexane as a solvent. Each colored band was recovered in methylene chloride and rechromatographed twice to ensure purity (Marty and Berset, 1986). Onyewu et al. (1986) also utilized TLC plates with 10% diethyl ether in n-hexane to fractionate P-carotene products degraded by heating in glycerol.
Vegetable carotenoid pigment separation has also been achieved by using column chromatography in conjunction with TLC. Tomato carotenoids were separated on an
103 alumina column, then each fraction rechromatographed on TLC plates consisting of silica
gel and MgO-kieselguhr (1:1) (Ben-Aziz et al., 1973). Ritacco et al. (1984) reported the
use of a MgO TLC plate with an eluting solvent of 20% acetone in petroleum ether for the efficient separation of Micro-Cel C carotenoid artifacts.
2.10.4 Structure elucidation
In order to either identify a known naturally occurring carotenoid or elucidate the structure of a previously unknown carotenoid, the application o f a variety of physical and chemical methods is usually required. The assignment of stereochemistry in terms of not only chirality, but also geometrical configuration, usually is the source of most of the complexity in carotenoid analyses. As a result, structural elucidation involves various spectrometric methods as well as chemical derivatization (Britton et al., 1995).
Before a carotenoid can be characterized, chromatographic purification is necessary.
Chromatography also provides information regarding the polarity of the carotenoid in question so that tentative assignment as a hydrophobic carotene, monool, or more polar carotenoid can be made (Liaaen-Jensen, 1995). After chromatographic analyses are completed, spectrometric methods are usually applied next.
2.10.4.1 UV/Vis Spectroscopy
The UV/Vis spectrum of a carotenoid is usually examined first to provide information regarding the chromophore present. Specifically, examination o f the spectral fine structure and position of the X^ax of the main absorption band can help determine the specific number o f conjugated C=C double bonds present in an aliphatic, monocyclic, or
104 dicyclic carotenoid (Britton, 1995). Values for Xmax for some of the most commonly
encountered carotenoids can be seen in Table 8 .
Yamauchi et al. (1993) determined the main UV/Vis spectral peaks of several
oxidized products o f P-carotene in order to help characterize their structures. For
example, three peaks at 405, 426, and 452 nm helped to define one of the oxidized
products as 13,15'-epoxyvinyleno-13,15'-dihydro-P,P-carotene. In a similar study El-
Tinay and Chichester (1970) also utilized UV/Vis spectroscopy to help determine the
products formed upon oxidation of P-carotene. Lycopene epoxides present in firesh
tomatoes were partially characterized through the use of UV/Vis spectroscopy. Fractions
were dissolved in petroleum and values were recorded for comparison with known
standards (Ben-Aziz et al., 1973). Britton and Goodwin (1969) also identified a lycopene
oxidation product (phytoene 1 ,2 -oxide) in ripe tomato fiaiits by comparing Xmax values
with those of phytoene.
Jensen et al. (1982) examined the fine structure of UV/Vis chromatograms of
photoisomerized all-frnw-P-carotene in order to identify exactly where and when cis-
trans isomerization occurred. Findings showed a significant hypochromic shift and a reduction in vibrational fine structure in both the 9- and 15-C/5' isomers of p-carotene in hexane when compared with the dl\-trans isomer.
2.10.4.2 IR Spectroscopy
IR spectra are most useful in carotenoid structural elucidation for their ability to reveal the presence or absence of particular functional groups in the molecular structure.
105 Carotenoid ^max (nm) Solvent
Astaxanthin 480 A 485 B,C 478 EtOH 468 P
P-carotene 429 452 478 A 435 462 487 B 435 461 485 C 450 476 EtOH 425 450 477 H,P
Lutein 432 458 487 B 435 458 485 C 422 445 474 EtOH 421 445 474 P
Lycopene 448 474 505 A 455 487 522 B 458 484 518 C 446 472 503 EtOH 444 470 502 P
Phytoene 276 286 297 H,P
Zeaxanthin 430 452 479 A 440 463 491 B 433 462 493 C 424 449 476 P
Table 8 : UV/Vis spectroscopic data for several common carotenoids. Solvents: A, acetone; B, benzene; C, chloroform; EtOH, ethanol; H, hexane; P, light petroleum.
106 Infrared (IR) radiation emitted from an IR spectrophotometer enhances the vibration of atoms in a molecule so that stretching of the bonds and variation in specific bond angles occurs. Energy is absorbed (and an absorption band formed) when the frequency of the
IR radiation matches the frequency of a particular vibrational mode. Therefore, compounds with many different atoms and fimctional groups will give rise to many different absorption bands in the IR spectrum (Figure 30). It is the specific shape, size, and position of the absorption band that lends information as to the type of functional group that it represents. The positions of the absorption bands are typically indicated by the frequency in wavenumbers (the reciprocal value of the wavelength).
Unknown purified compounds can be identified by a direct comparison of its IR spectra with that of a standard (Onyewu et al., 1986). IR spectroscopy is advantageous as a method for structural elucidation due to its simplicity and rapidity at relatively low cost.
The recent advent of Fourier-transform (FTIR) instruments has greatly improved both the
V. i*j>nr •fl
Figure 30: Typical IR spectrum for lycopene.
107 sensitivity and accuracy of traditional IR spectroscopy. FTIR spectrometers can measure
an IR spectrum with 10 to lOOx the sensitivity and in 1/000^ the time required to obtain a
similar spectrum with a traditional instrument. The main disadvantage to IR
spectroscopy, however, is the limited information given, especially for the hydrocarbon
carotenes which lack any type of functional groups (Bernhard and Grosjean, 1995).
The methylene and methyl groups present in carotenoids and their derivatives
typically give rise to bands at 2950-2850 cm'\ while the aUcene groups present in their
polyene structure gives rise to weak bands between 1650 and 1550 cm'* and a strong
band between 990 and 960 cm'* (Yamauchi et al., 1993). Onyewu et al. (1986) determined that an unknown carotenoid compound with absorption bands only at the methyl C-H (2950-2850 cm'*) and polyene C=C (1650 cm'*, 990 cm'*) wavenumbers must be a hydrocarbon carotene with no oxygenated groups present.
The presence of a Z (cis) double bond in a carotenoid compound results in the appearance of strong absorption in the region around 750-800 cm'*. Mangoon and
Zechmeister (1957) utilized this information in showing the stepwise process in which prolycopene was converted into all-trnw-lycopene. As the cis double bonds in prolycopene were converted to trans, the absorption band at approximately 750-800 cm'* became smaller until it disappeared in the IR spectrum o f all-/rn«^-lycopene.
Alcohol groups often show a very distinct broad absorption band in the 3300-3400 cm-1 range due to the stretching o f the 0-H —-O bonds, while aldehyde and ketone groups have a C=0 stretching band near 1715 cm'*. Lu et al. (1995) declared that due to the absence of absorption bands in the OH and C=0 regions, an oxygenated carotenoid compound they were examining must have an ether linkage. In this study, the
108 information obtained from the IR spectra helped to identify the structure o f a novel
carotenoid with an epoxyiridane skeleton. Ouyang et al. (1980) reported that many of the
structures of p-carotene degradation products from palm oil deodorization contained
aldehyde end groups, due to the presence of strong absorption bands in the 1700 cm'^
region. Ukai et al. (1994) reported similar absorption bands in the region of 1678 cm'*
for an oxidized lycopene product containing an aldehyde group (apo- 6 ’-lyocopenal).
2.10.4.3 Mass spectrometry
The largest contribution mass spectrometry (MS) provides for structural elucidation of carotenoids is identification of the molecular mass of the molecule. In addition, characteristic fragmentation seen in a mass spectrum lends additional information as to the functional groups present in the compound, i.e., 18 mass units (water) from alcohols,
32 mass units (methanol) from methyl ethers, 80 mass units for 5,6- and 5,8-epoxides, etc
(Figure 31). To obtain a mass spectrum only a few pg of pure sample are required, though high precision instruments may require slightly more (Enzell and Back, 1995).
MS employs several different ionization techniques for compound identification. The first, and still most popular, technique is electron impact (El) mass spectrometry. In this method, electrons are accelerated and directed at the sample, with the resulting impact causing the generation of the positively charged molecular ion M*". The particular intensity of the ions produced by El mass spectrometry is highly dependent upon specific conditions such as temperature and ionizing voltage. Marty and Berset (1988) used El mass spectroscopy to help identify the degradation products of tra« 5 -P-carotene produced during extrusion cooking. The El mass spectrum presented fragments specific for
109 M-138 M-191 -H i -H M-56
M-203
Figure 31 : Main fragmentations of the molecular ion of canthaxanthin by FAB MS.
carotenoids with 1 or 2 cyclohexenyl rings, as well as two fragments characteristic of 5,6-
epoxide or 5,8-epoxide functions.
A comparison by Ouyang et al. (1980) of sample spectra with the El mass spectra of pure 9-c/j-retinal showed identical relative abundance and fragmentation patterns, allowing positive identification of the sample p-carotene degradation product as 9-cis- retinal. Khachik et al. (1992a) elucidated the structures of eighteen different carotenoids and their oxidation products in human serum samples with the help of El mass spectroscopy. El MS has been used by several authors to help determine the molecular structures of products formed during the oxidation/degradation of P-carotene
(Kanasawud and Crouzet, 1990; Handelman et al., 1991) and lycopene (Kanasawud and
Crouzet, 1990).
Chemical ionization (Cl) generates mass spectra through ionic reactions. In this method a reagent gas such as ammonia or methane is introduced under pressure to the ion source, and then the gas is bombarded with electrons to yield ions and neutral molecules. 110 The reagent gas molecules then react with the vaporized sample molecules, resulting in a less fragmented and simpler spectra compared with El mass spectrometry. Marty and
Berset (1988) utilized CIMS with NH 3 as the reagent gas to help identify fourteen different (3-carotene thermal degradation products. In a similar study Cl mass spectrometry was used to confirm the parent ion (and therefore the molecular weight) of several P-carotene thermal degradation products (Onyewu et al., 1986).
Fast atom bombardment (FAB) mass spectrometry is a ‘soft’ ionization technique normally used for structure elucidation of either nonvolatile or thermally labile compounds. In this technique high-energy atoms such as xenon or cesium bombard a solution of the sample dissolved in an involatile matrix such as glycerol. Volatilization and ionization of the sample compound occurs by way o f dissipating kinetic energy from the bombarding atom beam, resulting in the formation of even-electron protonated molecular ion(s). Delocalization of the charge in the carotenoid extended polyene system prevents the usual protonization of the carotenoid sample from occurring, however, and results in the formation of odd-electron molecular ions. Subjecting carotenoids to FAB mass spectrometry typically yields only the radical ion MT in the molecular-ion region of the spectrum. An advantage o f FAB MS over El or Cl techniques is the lack of sample vaporization prior to volatilization, which reduces the likelihood o f carotenoid degradation during MS analysis itself. Schmitz et al. (1992) described interfacing FAB mass spectroscopy with HPLC to provide a powerful tool in the analysis of carotenoids.
A mixture of 3-nitrobenzyl alcohol (NAB) and glycerol (80:20 v/v) was described as an effective FAB matrix for analysis of the more nonpolar carotenoids due to NAB’s ability to interact with and disperse the carotenoid throughout the matrix. Using FAB-LC/MS,
111 van Breemen et al. (1993) determined the limits of detection to be as low as 5 ng and 15
ng for lutein and a-carotene, respectively. Lu et al. (1995) used FAB-MS to identify a
new hydrogen peroxide oxidation product from lycopene. Van Breemen (1996) reported
that though FAB mass-spectroscopy linked with HPLC can produce highly sensitive and
unambiguous molecular weight confirmations for both oxygenated xanthophylls and
hydrocarbon carotenes, it is limited by NAB fouling of the FAB probe that results in
lower sample throughput and decreased sensitivity.
Matrix-assisted laser desorption ionization (MALDI) mass spectrometry is one more
type of MS which has been recently applied to the analysis of carotenoids. MALDI was
originally introduced as an extremely sensitive method for mass analysis of very large
molecules such as proteins up to 300 kD in size. While other MS methods have difficulty
analyzing carotenoids due to their lack of volatility and thermal lability (El and Cl) or
lack o f solubility and surface activity in common liquid matrices (FAB), MALDI mass
spectroscopy methods have the potential to overcome most of these drawbacks
(Kaufinann et al., 1996).
2.10.4.4 NMR Spectroscopy
Nuclear magnetic resonance (NMR) spectroscopy is commonly referred to as the most powerful technique for overall structural elucidation. The reason for this is simple: proton (^H) NMR is capable of identifying the structural surroundings o f every hydrogen atom in a carotenoid, while carbon-13 (^^C) NMR identifies the degree o f saturation for each carbon atom (sp^, sp^, sp) and its structural surroundings. Detailed and NMR analyses alone can identify a carotenoid structure as well as its stereochemistry and
112 geometry of carbon-carbon double bonds. In addition, a typical *H-NMR spectrum requires only 100-200 pg of sample. As the magnetic field strengths of NMR instruments have increased to as high as 750 MHz, the dispersion of chemical shifts in samples has increased to the point that interpretation of ^H-NMR spectra has been greatly simplified. In turn, the amount of structural information that can be deduced firom the spectra has greatly increased as well (Figure 32).
13,
-.4 r u 1" 409 HH2
»,
«e- (.4 «U am ChOj
1îl
TM TJ r.C « iE tl4 tLZ jrn !
Figure 32: Olefinic section of the ^H-NMR spectrum of a typical carotenoid measured at 250 MHz (top), 400 MHz (middle), and 600 MHz (bottom).
113 Mordi et al. (1991) identified several oxidized products of P-carotene (5,6-epoxy-P- carotene, retinal, p-ionone, p-apo-14’-carotenal) using ^H-NMR spectroscopy. In a similar study, ^H-NMR was performed at 270 MHz, the ^H-^H chemical shift-correlated
(COSY) NMR technique was employed to assign shifts and couplings, and ^^C-NMR was run at 70 MHz with proton decoupling in order to identify the products formed by the peroxyl radical oxidation o f P-carotene (Yamauchi et al., 1993).
A 400 MHz ^H-NMR and a 270 MHz ^^C-NMR spectrophotometer were utilized in an analysis of the photosensitized oxidation products of lycopene firom tomato puree
(Ukai et al., 1994). Lu et al. (1995) used *^C-NMR, ^H-NMR, and COSY two- dimensional NMR experiments to help identify lycopene-oxidized products. With the application of ^H-NMR spectroscopy at 270 MHz and the help of *^C-NMR spectroscopy, Englert et al. (1979) was first able to report the complete assignment of stereochemistry to the carotene prolycopene as 7,9,7’,9’-tetra-c«-\j/,\|/-carotene.
114 CHAPTERS
MATERIALS AND METHODS
3.1 Materials
(3-carotene, lycopene, and chlorophyll standards were obtained from Sigma Chemical
Company (St. Louis, MO). Lycopene (>98% pure) was generously donated by LycoRed
Natural Products Industries (Beer-Sheva, Israel). HPLC grade methanol, methyl-tert- butyl ether, acetonitrile, and chloroform were purchased from Fisher Scientific
(Pittsburgh, PA). Helium for both gas chromatography and HPLC sparging, compressed air for gas chromatography and HPLC autosampler function, hydrogen for gas chromatography, and nitrogen for gas chromatography, sample drying, and storage were obtained from the Chemical Store at The Ohio State University. Analytical reagent grade methanol, acetone, and hexane (Fisher Scientific, Pittsburgh, PA) were used for lycopene extraction. All solid phase microextraction (SPME) fibers [polydimethylsiloxane
(PDMS), 100 pm; polydimethylsiloxane/divinyIbenzene (PDMS/DVB), 65 pm; polydimethylsiloxane/carboxen (PDMS/carboxen), 75 pm; carbowax/divinylbenzene
(CW/DVB), 65 pm; polyacrylate, 85 pm] and fiber assembly holders used in soybean oil volatile analyses were purchased from Supelco, Inc. (Bellefonte, PA). Volatile standards
115 r,r-2,4-decadienal, 2-heptenal, r,/‘-2,4-heptadienal, n-hexanal, and n-pentane were procured from Aldrich (Milwaukee, WI). Refined, bleached, and deodorized (RED) soybean oil was obtained from Abitec Corp. (Columbus, OH).
3.2 Lycopene extraction
To obtain 85-90% pure crude lycopene extract for experimental procedures, lycopene was extracted from supermarket-bought (Kroger brand) canned tomato paste. Under dark conditions, 10 g of tomato paste were mixed with 25 mL of methanol and 4 g of Celite, with the mixture being tissumized for approximately 2-3 minutes. The homogenized sample was then filtered through #1 and #45 Whatman filter papers and the filtrant placed in a beaker with 50 mL of acetone/hexane (1:1). Again the mixture was tissumized for approximately 2-3 minutes and then filtered through the same #1 and #45 Whatman filter papers. The filtrate was poured into a separatory funnel, with the filtrant being discarded.
The filtrate was washed with 10 mL of distilled water 3 times in the fimnel, each time with removal of the bottom layer (waste) and retention of the top hexane layer. The top hexane layer was then transferred to 4 mL amber vials and dried under nitrogen gas until the lycopene extract was in a solid crystalline state. The vials were sealed, wrapped in aluminum foil to minimize light exposure, and stored at -4°C until needed.
3.3 Thermal treatment of carotenes
3.3.1 (3-carotene and lycopene
A solution of 500 ppm (3-carotene or lycopene solubilized in acetone was added to a glass vial that was sealed airtight with a Teflon-faced rubber septum and aluminum cap
116 (Supelco, Inc., Bellefonte, PA). A portion of the unheated solution was saved. The
re m a in in g portion in the glass vial was wrapped in aluminum foil to minimize light
exposure, then placed in a water bath set at 90°C to allow a specific amount of thermal
degradation of the carotene to occur. After the P-carotene or lycopene solution received
the proper length of thermal treatment (Figures 33 and 34, respectively), the sealed glass
vial was removed firom the water bath and allowed to cool to room temperature. Out of
the cooled glass vial one of three heated solutions containing either 250 ppm P-carotene +
250 ppm degraded products, 50 ppm p-carotene + 450 ppm degraded products, or 0 ppm
P-carotene + 500 ppm degraded products was collected (Figure 35), contents verified by
RP-HPLC, and stored in a fireezer at -4°C for further use, respectively.
3.3.2 Crude lycopene extract
A 40 mL solution o f 500 ppm lycopene solubilized in methyl-fgrf-butyl-ether
(MTBE) was added to a 100 mL glass vial that was sealed airtight with a Teflon-faced rubber septum and aluminum cap. A 13 mL portion of the unheated solution was saved.
The remaining portion in the glass vial was placed in a water bath at 90°C to allow thermal degradation of the carotene to occur. After receiving the proper length of thermal treatment, the sealed glass vial was removed from the water bath and allowed to cool to room temperature. Out of the cooled glass vial, a solution containing either 250 ppm lycopene + 250 ppm degraded products, 50 ppm lycopene + 450 ppm degraded products, or 0 ppm lycopene + 500 ppm degraded products was collected, contents verified with HPLC, then placed in a sealed glass vial and stored at -4°C.
117 2.00E+08 y = -504997% + 2E+08 1.50E-H)8 3 R^= 0.9469
.H l.OOE+08
s ft. 5.00E+07
O.OOE+00 100 150 200 250 Time (minutes at 90 deg Q
Figure 33: Increased degradation of P-carotene (decreased P-carotene peak size) with increased thermal processing at 90°C.
2.0E+08
1.6E+08 y = -208486% + 2E+08 P = 0.9983 S 1.2E+08 .1 ^ 8.0E+07 ft.2 4.0E+07
O.OE+OO 0 200 400 600 800 1000 Time (minutes at 90 deg C)
Figure 34: Increased degradation of lycopene (decreased lycopene peak size) with increased thermal processing at 90°C.
118 water bath ® 90°C
40 mLs 500 ppm mcreasmg B-carotene or heating time lycopene in acetone
15 mLs 8 mLs 8 mLs 8 mLs unheated 250 ppm + 50 ppm + 0 ppm + 500 ppm 250 ppm degraded 450 ppm degraded 500 ppm degraded + 0 ppm
Figure 35: Schematic of P-carotene or lycopene heating and fractionation into solutions with varying degrees of thermal degradation.
3.4 Sample preparation and storage for the chlorophyll-photosensitized singlet oxygen oxidation of soybean oil
To study the effects of thermal processing on the singlet oxygen quenching abilities of degraded carotenes, a 2.5 mL solution containing thermally processed p-carotene or
119 lycopene (250 ppm carotene + 250 ppm degraded products, 50 ppm carotene + 450 ppm degraded products, or 0 ppm carotene + 500 ppm degraded products) was added in triplicate to 30 mL glass serum vials containing 22.5 mL of purified soybean oil and 3 ppm chlorophyll sensitizer. The resulting 25 mL soybean oil solutions then contained either 25 ppm carotene + 25 ppm degraded products, 5 ppm carotene + 45 ppm degraded products, or 0 ppm carotene + 50 ppm degraded products (Figure 36).
25 ppm 250 ppm 2.5 mL + 250 ppm 25 ppm
5 ppm + 50 ppm + 45 ppm 450 ppm
0 ppm 0 ppm + 2.5 mL + ► 500 ppm 50 ppm
22.5 mL soybean oil 25 mL soybean oil + 3 ppm chlorophyll + carotene
Figure 36: Addition of degraded carotene solutions to soybean oil to create the final concentrations in the tested sample solutions.
120 Similarly, 2.5, 1.25, or 0.25 mL of 500 ppm unheated p-carotene or lycopene were added to vials containing identical quantities of soybean oil and chlorophyll to create 25 mL control solutions containing either 50, 25, or 5 ppm of either lycopene or P-carotene.
Other controls included samples containing only oil, or oil and chlorophyll (Figure 37).
50 ppm
0.25 mL 500 ppm 25 ppm + 0 ppm
5 ppm
oil + oil only chlorophyll
Figure 37: Addition of unheated carotene solutions to soybean oil to create the final concentrations in the tested control sample solutions.
121 Each vial was then sealed airtight (for headspace oxygen content determination only)
with Teflon-faced rubber septa and aluminum caps and stored at 25°C in a mirrored light
box (70 cm X 50 cm x 60 cm) for 24 hours (Figure 38). The light source itself consisted
of a 100-watt fluorescent lamp (Lights of America, Inc., Walnut, CA) with an intensity of
1,650 lumens. The oxidative stability of the sample was determined by measuring both
peroxide value (AOCS, 1980) and the headspace oxygen content of the sample vial by
thermal conductivity gas chromatography every 4 hours during the aforementioned 24
hour period.
m
Figure 38: Diagram o f mirrored light box for sample storage imder light.
122 3.5 Sample preparation and storage for thermally-induced degradation of soybean oil
To study the effects of degraded carotenes on the oxidative stability of thermally-
treated soybean oil, 2.5 mL acetone solutions containing either 500 ppm P-carotene or
lycopene, or 0 ppm carotene + 500 ppm degraded P-carotene or lycopene were added in
triplicate to 30 mL glass serum vials containing 22.5 mL soybean oil, respectively. Each
vial (including controls containing 22.5 mL soybean oü and 2.5 mL acetone) was then
sealed airtight with Teflon-faced rubber septa and aluminum caps and then stored at 60°C
in the dark for 8 days. The oxidative stability of the sample was determined by
measuring the headspace oxygen content of the sample vial by thermal conductivity gas
chromatography after 8 days of storage at 60°C.
3.6 Headspace oxygen analysis
A Hewlett Packard 5890 Gas Chromatograph equipped with a thermal conductivity detector (TCD) and a stainless steel molecular sieve packed column (13x, 80/100, '
Alltech, Deerfield, IL) was utilized to determine the extent of headspace 0% depletion
(and corresponding soybean oil oxidation). High purity (99.995%) helium gas was used as both the carrier and auxiliary gas at a rate o f 30 mL/min. The injector, detector, and oven temperatures were set to 120°C, 150°C, and 40°C, respectively. Every 4 hours for
24 hours 100 pL o f headspace air from each sealed vial was injected into the GO manually using an airtight syringe. Ambient air was used as a reference o f O 2 content to correct for day to day chromatographic variability. A Hewlett Packard HP 3396A integrator (Avondale, PA) was used to record the oxygen peak in electronic counts.
123 3.7 Solid phase microextraction (SPME) volatiles analysis
Volatile compounds created as a result o f thermal (stored at 60°C in oven) or light- induced (stored in light box, see Figure 35) oxidation of soybean oil containing either 13- carotene or lycopene were isolated by solid phase microextraction (SPME), then separated and tentatively identified by gas chromatography. All soybean oil/carotene samples exposed to either oxidizing treatment were prepared by adding the following to
30 mL serum vials: 1) 22.0 mL RED soybean oil; 2) 2.5 mL of acetone containing either
500 ppm P-carotene/lycopene, 500 ppm degraded P-carotene/lycopene, or acetone only
(control); 3) 500 pL pure acetone (for thermally treated samples) or 500 pL 150 ppm chlorophyll in acetone (for samples placed in light box). Resulting soybean oil samples therefore contained 50 ppm P-carotene/lycopene, 50 ppm degraded P-carotene/lycopene, or just oil and acetone (controls), respectively. In addition, all samples placed in the light box contained 3 ppm chlorophyll.
Three different samples (50 ppm P-carotene/lycopene, 50 ppm degraded P- carotene/lycopene, oil only control), each in triplicate, received either the oven storage treatment or the light box storage treatment for 0, 5, 10,20, or 30 days. At the end of the storage time, SPME headspace analysis of each vial was conducted by penetration of the vial septum and adsorption of volatiles present with a fijsed silica fiber coated with a stationary phase of polydimethylsiloxane (PDMS) (100 pm film). The adsorption time was exactly 40 minutes, occurring at a fixed temperature of 60°C. The fiber was then retracted firom the vial and immediately inserted and desorbed for 2 minutes into the inlet liner of a Hewlett Packard 5890 Series II Gas Chromatograph equipped with an HP-5
124 crosslinked 5% phenyl-methyl silicone fused silica capillary column (30 m x 0.32 mm
ID, 0.25 pm film) and flame ionization detector (FID). Each GC run consisted of a ramped oven temperature profile as follows: 40°C (0 minutes) to 240°C at 8°C/min., then 240°C for 5 minutes (total run time 30 minutes). The carrier gas was nitrogen at a column flow rate of 4 mL/min. The FID detector was set at 300°C, while the injector
(and desorption temperature) was set at 250°C. The GC chromatogram total peak area from the oil only control sample was set at a value of 100%, and the peak areas of samples containing dX\-trans or degraded carotenes were compared at 0, 5, 10, 20, and 30 days as a percentage of the oil only control.
Gas chromatography-mass spectrometry (GC-MS) was utilized to positively identify volatile oxidation products of soybean oil that were adsorbed onto a solid phase microextraction (SPME) polydimethylsiloxane (PDMS) fused silica fiber and then desorbed onto a GC capillary column. Specifically, a Hewlett Packard 5890 Gas
Chromatograph equipped with an HP-5 crosslinked 5% phenyl-methyl silicone fused silica capillary column (30 m x 0.32 mm ID, 0.25 p.m film) as well as a computer monitor interface containing Hewlett Packard 59970 MS Chemstation Software was utilized for volatile separation. A Hewlett Packard 5970 Series Mass Selective Detector equipped with a Hewlett Packard 59822 B Ionization Gauge Controller was used to collect a mass spectrum for each volatile compoimd. Structural identification of the main volatile peaks was made with the help of Wiley Mass Spectrum Library computer software.
125 Each sample prepared for GC-MS analysis was held at 60°C for twelve hours in a
water bath while a PDMS SPME fiber was exposed to the sample’s headspace. The fiber
was then immediately inserted into the GC that was equipped with a special septum and
inlet liner for SPME fibers. The fiber was desorbed in the inlet liner for 4 minutes at
250°C, then retracted. Each GC-MS analysis consisted of an overall run time of 35
minutes, with a 25 minute ramping program in the middle. The oven was held at 40°C
for 5 minutes at the beginning of each nm, then ramped at a rate of 8°C/min. from 40°C
to 240°C, upon which the oven temperatiure was held at 240°C for the last 5 minutes of
the run. The injector was set at 250°C, while the detector temperature was 300°C.
Helium was used as the carrier gas through the capillary column at a flow rate of
approximately 1.5 mL/min.
3.8 Analysis of carotene degradation products
3.8.1 Separation
P-carotene and lycopene, along with their degradation products, were separated using
a Hewlett Packard Series 1050 HPLC system and Hewlett Packard 3396A integrator.
Isocratic reverse-phase liquid chromatography was conducted utilizing an Hewlett
Packard CDS Hypersil analytical column (5 pm, 200 x 4.6 mm) and a mobile phase
solvent system of methanol: methyl-Zert-butyl-ether (90:10). The flow rate for each run
was 1 mL/min. with a total run time of 20 minutes. A UV/Vis detector was set at 460 nm
for P-carotene, 472 for lycopene, and 350 nm for degraded product analyses to maximize peak sensitivity.
126 3.9 Statistical analysis
Two sample t-test and one-way Analysis of Variance (ANOVA) were used to determine the presence of differences (a=0.05) between two samples and groups of samples, respectively. Tukey’s Studentized Range Test determined which particular sample groups were different at a=0.05. All the aforementioned statistical analyses were calculated on the Minitab Version 10.1 for Windows statistical software package.
127 CHAPTER 4
RESULTS AND DISCUSSION
4.1 Separation and identification of carotenes and their thermal degradation products
Isocratic nonaqueous reverse-phase high performance liquid chromatography (RP-
HPLC) was utilized to isolate, quantify, and determine the extent of degradation in heated solutions of aS\rtrans P-carotene, all-tranj-lycopene, and crude lycopene extract, respectively (Nelis and De Leenheer, 1983). Though hexane is often used to solubilize carotenes for HPLC analysis, it did not effectively solubilize carotenes at concentrations higher than 500 ppm (Craft and Soares, 1992). Acetone, which more effectively solubilized the carotenes, was therefore utilized as the solvent in each experiment.
After solubilization, heating a solution of P-carotene in sealed vials at 90°C resulted in a fairly rapid linear degradation, or “bleaching”, of the dl\-trans peak (Figure 39). El-
Tinay and Chichester (1970) reported a similar linear relationship between the loss of P- carotene and time. The A,max for the dl\-trans p-carotene peak was found to be 460 nm, so the UV/Vis detector was set at this wavelength to quantify (in AU) the remaining all- trans P-carotene at each stage of thermal degradation (Stahl et al., 1993). As seen in
Figure 39, after approximately 3.5 hours of thermal treatment, the size of the dl\-trans P-
128 1.8 X 10* AU 8.5 X 10^ AU
I I 460 nm 2.0 X 10^ AU I
■ 1 2.0 X 10^ AU JJL I
350 nm
jw A ji.
500 ppm 250 ppm 50 ppm 0 ppm P-carotene + + + + 0 ppm 250 ppm 450 ppm 500 ppm degraded products
Figure 39: Extent of thermal degradation of each collected P-carotene sample solution as seen by RP-HPLC (arrows indicate all-fruw-p-carotene peak).
carotene peak at 460 nm was halved, from 1.8 x 10* AU to 8.5 x lO’ AU. A detector wavelength of 350 nm has been shown to effectively reveal chromatographic peaks o f carotene thermal degradation products, due to the presence of conjugated trienes in these compounds (Handelman et al., 1991). Such P-carotene degradation peaks were clearly visible at 350 nm on the chromatogram showing the 250 ppm P-carotene + 250 ppm degraded products solution. After 8 hours of thermal treatment at 90°C, only one-tenth of 1% o f the ail-trans p-carotene peak remained (effectively 0 ppm p-carotene + 500 ppm
129 degraded products). At the point in which all of the a!L\-trans ^-carotene had been degraded, many of the peaks representing oxidized products seen in the corresponding chromatogram run at 350 run were reduced in size as well, indicating extensive degradation of the carotene conjugated double bond system.
Continued thermal treatment of lycopene solubilized in acetone also resulted in linear reduction o f the dl\-trans peak size (Figure 40). Quantitation of lycopene was conducted at a wavelength o f472 nm, the reported A,max for the ail-trans isomer (Mangoon and
1.7 X 10* AU 7.0 X 10^ AU I I 472 nm 1.4 X 10^ AU I 0 AU L L I 350 nm
I
rJkJu\_ U i 500 ppm 250 ppm 50 ppm 0 ppm lycopene + + + + 0 ppm 250 ppm 450 ppm 500 ppm degraded products
Figure 40: Extent o f thermal degradation of each collected lycopene sample solution as seen by RP-HPLC (arrows indicate all-traw-lycopene peak).
130 Zechmeister, 1957). The small peaks visible at 472 nm that eluted before the a[\-trans
peak in the unheated lycopene solution were most likely oxidized products of lycopene
formed unintentionally during sample storage, handling, and chromatography (Emenhiser
et al., 1995). Piretti and Diamante (1996) hypothesized that anomalous peaks seen in
commercial standards o f all-/ran5 lycopene using RP-HPLC could also be geometrical
isomers (13-cfj) o f the predominant dX\-trans form.
As was the case with P-carotene, great care was taken to heat the lycopene solution
for just the right amount of time necessary to obtain the proper proportion of dl\-trans
lycopene to degraded products. In a manner similar to the results seen in Figure 39,
chromatograms o f the 250 ppm lycopene + 250 ppm degraded products solution viewed
at 472 nm and then 350 nm in Figure 40 revealed a 50% reduction in the size (in AU) of
the all-trans lycopene peak, and a marked increase in the size of the oxidized/degraded
peaks, respectively. After 14 hours of thermal treatment at 90°C, the elY-trans lycopene
peak was completely gone, which indicated thermal degradation of the carotene
equivalent to the desired 0 ppm lycopene + 500 ppm degraded products solution.
Lycopene appeared more resistant to thermal degradation when compared with (3-
carotene (14 hours at 90°C to completely degrade vs. 8 hours, respectively). This observation did not agree with the findings of Henry et al. (1998), who reported that lycopene was most susceptible to thermal degradation, with a rate constant approximately double that of p-carotene. Differences in the carotenes’ surrounding matrixes (acetone vs. safflower seed oil) may have accounted for the observed differences in the relative thermal degradation rates of lycopene and P-carotene (Onyewu et al., 1986). In addition.
131 the relatively high pressures that were present in the sealed vials during heating of the
carotene solutions have also been shown to specifically increase the degradation rate of
P-carotene (Marty and Berset, 1990).
A crude lycopene extract from tomato paste was the third solution to be exposed to
extended thermal degradation at 90°C. HPLC chromatograms set at both 472 and 350 nm were run at each stage of thermal degradation o f the crude lycopene extract solution, from unheated to completely degraded (Figure 41). The unheated crude lycopene
1.5 X 10“ AU 8.4 X 10^ AU 472 nm
1.9 X 10' AU
1.4 X 10^ AU
11. I
350 nm
r A_ 500 ppm 250 ppm 50 ppm 0 ppm lycopene + extract 0 ppm 250 ppm 450 ppm 500 ppm degraded products
Figure 41 : Extent of thermal degradation of each collected crude lycopene extract sample solution as seen by RP-HPLC (arrows indicate all-P-nw-lycopene peak).
132 extract solution revealed, in addition to the predominant all-tra«j lycopene peak at
approximately 14.3 minutes, a minor peak at approximately 17.5 minutes that was
identified (by spiking the solution with a standard) as dl\-trans P-carotene. Previous
studies have also shown red tomatoes to contain a small quantity of P-carotene in
addition to the predominant lycopene pigment (Hakala and Heinonen, 1994; Hart and
Scott, 1995). However, ai\-trans lycopene typically elutes before a!L\-trans P-carotene
only when acetonitrile is used in the mobile phase (Nelis and De Leenheer, 1983; Stahl et
al., 1992; Scott and Hart, 1993). The reason why in this case lycopene actually eluted
before p-carotene without the use o f acetonitrile in the mobile phase could not be
determined.
The crude lycopene extract solution degraded at a significantly slower rate than either
the p-carotene or lycopene solutions. While P-carotene and lycopene only required 3.5
and 8 hours to lose half of their peak area (in AU), respectively, the crude lycopene
extract solution required 24 hours to become half degraded. To become completely
degraded, the crude lycopene extract solution required 275 hours, or over eleven days!
Lipid-soluble antioxidants such as tocopherols that were present in the tomato paste may
have been extracted with the lycopene and helped to prevent thermal oxidation and
degradation of the lycopene present by acting as free radical scavengers (Burton and
Ingold, 1989; Palozza and Krinsky, 1992; Heinonen et al., 1997). After 275 hours of thermal treatment, less than 1/10 o f 1% of the dM-trans lycopene peak area remained, effectively creating the 0 ppm lycopene extract + 500 ppm degraded products solution.
133 4.2 Effects of degraded carotenes on chlorophyll-photosensitized singlet oxygen oxidation of soybean oil
4.2.1 P-carotene
To determine how degraded carotenes might affect the oxidative stability of a soybean oil model system exposed to light, proper controls must first be created to account for any extraneous variables. For this reason, in addition to the three samples containing 3 ppm chlorophyll and varying concentrations of degraded carotenes (5 ppm,
25 ppm, or 50 ppm), six control samples were also tested. Figure 42 shows seven of
^— B-car 50 ppm
#— B-car 25 ppm B-car 5 ppm 25+25
5 +45 0+ 50
oil+chloro (light)
12 Time (hours)
Figure 42: Effects of thermally degraded P-carotene on the headspace oxygen depletion of soybean oil containing 3 ppm chlorophyll under light storage at 25 °C.
134 the nine soybean cil/carotene/chlorophyll samples that were tested for headspace oxygen
depletion over a 24 hour period by injection of 100 pL headspace air into a gas
chromatograph equipped with a thermal conductivity (TC) detector.
The control sample containing 50 ppm unheated al\-trans (3-carotene (B-car 50 ppm)
had significantly less (p<0.05) headspace oxygen depletion after 24 hours than all other
samples, which indicated that there was less oxygen uptake and subsequent oxidation of
the oil (Figure 42, Table 9). This result was not surprising considering the numerous
studies citing p-carotene as an effective quencher of singlet oxygen (Foote and Dermy,
Depleted pmoles Sample* Grouping** Oi/mL headspace
50 ppm ail-trans 3.35 A
25 ppm all-trans 4.00 B
5 ppm al\-trans 4.21 B
25 ppm all-tram + 25 ppm degraded 4.06 B
5 ppm all-tram + 45 ppm degraded 4.06 B
0 ppm all-tram + 50 ppm degraded 4.56 C
oil + chlorophyll (in light) 4.24 BC
* Indicates final concentration of P-carotene in soybean oil. ** Means with different letters are significantly different (p<0.05).
Table 9: Tukey’s Studentized Range Test for the effects o f P-carotene and thermally degraded p-carotene combinations on headspace oxygen depletion (pmoles Oi/mL headspace) after 24 hours of light exposure.
135 1968; Edge et al., 1997), particularly in soybean oil model systems (Warner and Frankel,
1987; Jung and Min, 1991). The control sample containing 25 ppm unheated al\-trans p- carotene (B-car 25 ppm) was very close in headspace oxygen depletion after 24 hours to the sample containing 25 ppm 2l\-trans P-carotene + 25 ppm degraded products (25 + 25)
(4.00 vs. 4.06 pmoles Oz/mL, respectively). Similarly, the control sample containing only 5 ppm unheated dl\-trans P-carotene (B-car 5 ppm) was very close in headspace oxygen depletion after 24 hours to the sample containing 5 ppm d!i\-tram p-carotene + 45 ppm degraded products (5 +45) (4.21 vs. 4.06 pmoles Oz/mL, respectively) (Figure 42).
None of the headspace oxygen depletion values for the 5 and 25 ppm controls and their corresponding degraded mixture samples (5 + 45, 25 + 25) were significantly different
(Table 9). The only dissimilarity between the 5 and 25 ppm aï\-trans P-carotene controls and their corresponding 5 + 45 or 25 + 25 degraded mixtures was the presence of degraded carotenes in the non-control samples. Therefore, the degraded P-carotene compounds in this case did not act as either prooxidants or antioxidants in the soybean oil model system, but rather had no overall effect on the soybean oil oxidative stability under the light.
Of particular interest was the relationship in headspace oxygen depletion between the sample containing 50 ppm zil-trans p-carotene (0 + 50), and the control samples containing 50 ppm all-n*ara p-carotene (B-car 50 ppm) and oil and 3 ppm chlorophyll
[oil + chloro (light)], respectively. The 50 ppm degraded P-carotene sample showed a significantly higher (p<0.05) level of headspace oxygen depletion after 24 hours than all the other samples (Figure 42) at 4.56 pmoles Oz/mL, except for the oil and chlorophyll
136 only control at 4.24 fimoles Oz/mL (Table 9). Obviously the degradation of the P~
carotene polyene system from 50 ppm dll-trans p-carotene to 50 ppm degraded products
caused a dramatic change in the effects o f the carotene on soybean oil stability. With the
polyene system intact in the 50 ppm dl\-trans sample (B-car 50 ppm), headspace oxygen
depletion was significantly reduced. After the thermal destruction of the polyene system
in the 50 ppm degraded sample (0 + 50), however, headspace oxygen depletion was much
higher. This indicated that the loss of the extensive conjugated double bonds present in
dl\-trans P-carotene by heating eliminated its ability to quench singlet oxygen and
prevent photosensitized oxidation of the soybean oil. Similarly, Lee and Min (1990)
reported a decrease in the singlet oxygen quenching rate of five carotenoids as their
respective conjugated double bond coimts decreased. The 50 ppm degraded P-carotene
sample not only provided no antioxidant activity in the form of singlet oxygen quenching,
but in fact caused a slight, though not significant, prooxidant effect when compared with
the control containing only oil and chlorophyll (Figure 42).
In order to prove that oxidation of the soybean oil was indeed coming from singlet
oxygen oxidation, two control samples were prepared, stored for 24 hours, and examined
for headspace oxygen depletion (data not shown). The first control contained only
soybean oil and was stored in the light at 25°C, while the second control contained soybean oil as well as 3 ppm chlorophyll and was stored in the dark at 25°C. After 24 hours of storage in the light, the headspace oxygen depletion value for the sample containing only soybean oil and no added chlorophyll was only 0.74 pmoles Oz/mL, significantly less than the control in light storage containing soybean oil and 3 ppm
137 chlorophyll (4.24 p.moles Oz/mL). These results reiterated the importance of chlorophyll
as a photosensitizer that can excite ground-state triplet oxygen to the more reactive
singlet oxygen state (Clements et al., 1973; Usuki et al., 1984; Fakourelis et al., 1987), as
well as supported singlet oxygen as the main source of soybean oil oxidation under light
storage. After 24 hours of storage in the dark, the second control sample containing
soybean oil as well as 3 ppm chlorophyll experienced almost no headspace oxygen
depletion (0.004 p.moles Oz/mL). The importance of light as an exciter of sensitizers in
photosensitized oxidation reactions has been well documented in the literature (Clements
et al., 1973; Warner and Frankel, 1987; Rahmani and Csallany, 1998). The controls
showed that in the presence of light and a sensitizer like chlorophyll (two requisites for
singlet oxygen oxidation), oxidation o f soybean oil occurred, while samples lacking
either light exposure or chlorophyll were oxidized to a much lesser extent. The data
therefore strongly supported singlet oxygen as the main source of soybean oil oxidation
under light storage.
Peroxide values were also determined for soybean oil samples exposed to light
containing the same combinations of dl\-trans and degraded P-carotene to better determine the relationship between carotenes and oil oxidative stability. In a manner similar to the headspace oxygen depletion data for P-carotene, control samples containing
50 ppm dl\-trans P-carotene had the lowest peroxide values after 24 hours of storage at
25°C under the light (Figure 43). Their average value of 8.83 meq HzOz/Kg soybean oil after 24 hours o f light storage was significantly lower than those of either the 25 ppm
(11.33 meq HzOz/Kg) or 5 ppm (13.67 meq HzOz/Kg) dl\-trans p-carotene samples
138 15 -, . B-car 50 ppm . B-car 25 ppm 12.5 - . B-car 5 ppm .25+25 .5+45
= 10 - .0+50 o . oil+chloro (light) DX)