Elucidation of selected Maillard reaction pathways in alanine and phenylalanine model systems through isotope labelling and pyrolysis-GC/MS based techniques
FONG LAM CHU
Department of Food Science and Agricultural Chemistry McGill University, Montreal
August 2009
Thesis submitted to the Faculty of Graduate and Post-Doctoral studies in partial fulfillment of the requirement of the degree of Doctor in Philosophy
© Fong Lam Chu, 2009
Library and Archives Bibliothèque et Canada Archives Canada
Published Heritage Direction du Branch Patrimoine de l’édition
395 Wellington Street 395, rue Wellington Ottawa ON K1A 0N4 Ottawa ON K1A 0N4 Canada Canada
Your file Votre référence ISBN:978-0-494-61905-6 Our file Notre référence ISBN: 978-0-494-61905-6
NOTICE: AVIS:
The author has granted a non- L’auteur a accordé une licence non exclusive exclusive license allowing Library and permettant à la Bibliothèque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par télécommunication ou par l’Internet, prêter, telecommunication or on the Internet, distribuer et vendre des thèses partout dans le loan, distribute and sell theses monde, à des fins commerciales ou autres, sur worldwide, for commercial or non- support microforme, papier, électronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats. . The author retains copyright L’auteur conserve la propriété du droit d’auteur ownership and moral rights in this et des droits moraux qui protège cette thèse. Ni thesis. Neither the thesis nor la thèse ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent être imprimés ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author’s permission.
In compliance with the Canadian Conformément à la loi canadienne sur la Privacy Act some supporting forms protection de la vie privée, quelques may have been removed from this formulaires secondaires ont été enlevés de thesis. cette thèse.
While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n’y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.
Suggested Short Title:
Elucidation of selected Maillard reaction pathways
II Abstract
Alanine and phenylalanine based model systems were utilized in this thesis to elucidate selected Maillard reaction pathways through isotope labelling, Fourier Transform
Infrared Spectroscopy (FTIR) and Pyrolysis-Gas Chromatography/Mass Spectrometry
(Py-GC/MS) based techniques. The formation of glycosylamines from reducing sugars and amino acids is a well-known process in the initial phase of the Maillard reaction.
They play a critical role in both the initiation and propagation stages, however, little attention has been paid so far on the ability of these imines to undergo isomerization and thus contribute to the diversity of Maillard reaction products. In this study, imine isomerization through 5-oxazolidinone formation was explored in phenylalanine and alanine sugar models systems. Spectroscopic evidence was provided for its formation by taking advantage of the strong carbonyl absorption band centered at 1784 cm-1 in the phenylalanine/glyceraldehyde and at 1778 cm-1 in phenylalanine/glycolaldehyde model system. The importance of 5-oxazolidinone formation lies in its ability to decarboxylate to azomethine ylide and subsequently form two isomeric imines, each capable of producing distinct Maillard reaction products. Evidence for the formation of such ylides was also provided through their ability to undergo 1,3-dipolar cycloaddition with dipolarophiles. Regarding the role of oxygen in the Maillard reaction, it was found that molecular oxygen can influence carbon-carbon bond cleavage through the formation and degradation of 1,2-dioxetane moieties generated from enol structures abundantly formed in the Maillard reaction from their corresponding ketones and aldehydes such as phenylacetaldehyde the Strecker aldehyde of phenylalanine and subsequently can be oxidized into benzaldehyde. Furthermore, the α-dicarbonyl compounds generated during
III the Maillard reaction play a significant role as precursors of important flavour-active heterocyclic compounds. The origin of many such α-dicarbonyl compounds still remains unknown. Using glucose and glyoxal with labelled [13C]- and [15N]-alanine, the mechanism of formation of 1,2-butanedione and 3,4-hexanedione were confirmed and proposed to proceed through amino acid-assisted chain elongation pathway. In addition, the role of α-dicarbonyl compounds in the formation of various heterocyclic compounds such as pyrazines, pyrazinones and imidazolidinones were also illustrated. The thesis further demonstrates the utility of pyrolysis-GC/MS as a powerful analytical tool especially if used in conjunction with isotope labelling techniques.
IV Résumé
Cette thèse comporte une étude approfondie des routes réactionnelles de la réaction de
Maillard dans des systèmes modèles à base d’alanine et de phénylalanine à l’aide de techniques basées sur les principes d’incorporation d’isotopes lourds avec la pyrolyse couplée à la chromatographie phase gazeuse et la spectrométrie de masse (Py-CG/SM) et ainsi que la spectroscopie infrarouge à transformée de Fourier (IR-TF). La formation des glycosylamines par des sucres réducteurs et des acides aminés est un processus bien connu dans la phase initiale de la réaction de Maillard. Ceux-ci jouent un rôle critique dans les étapes de déclenchement et de propagation. Cependant, peu d’attention est orientée vers la capacité de ces imines à subir l'isomérisation et de contribuer à la diversité des produits de la réaction de Maillard. Dans cette étude, l'isomérisation d'imine par la formation du 5-oxazolidinone fut explorée dans des systèmes modèles de phénylalanine/sucre et alanine/sucre. Les preuves spectroscopiques pour la formation du
5-oxazolidinone furent obtenues par la bande intense d'absorption carbonylique centrée à
1784 cm-1 dans le système modèle phénylalanine/glycéraldéhyde et à 1778 cm-1 dans le phénylalanine/glycolaldéhyde. L'importance de la formation du 5-oxazolidinone résulte dans sa capacité à se décarboxyler formant ainsi un ylide d'azomethine ayant l’habileté de produire deux imines isomériques, chacune capable de fabriquer des produits distincts de
Maillard. De plus, la formation de tels ylides fut également démontrée par la réaction de leur groupement 1,3-dipolaire avec des dipolarophiles par cycloaddition. Parallèlement, une étude sur le rôle de l'oxygène dans la réaction Maillard, nous a permis de constater que l'oxygène moléculaire peut influencer la rupture des liens carbone-carbone par la formation et la dégradation du 1,2-dioxetane. Ceci dit, le 1,2-dioxetane est formé par les
V structures d’énol produits par l’entremise de la réaction de Maillard, à partir de leurs cétones et aldéhydes correspondants tels que le phénylacétaldéhyde, l'aldéhyde de
Strecker de la phénylalanine. Il peut aussi être oxydé, produisant ainsi l'aldéhyde benzoïque. En outre, les α-dicarbonyles produits durant la réaction de Maillard jouent un rôle significatif en tant que précurseurs de composés hétérocycliques à effet aromatique important. L’origine de plusieurs de ces composés α-dicarbonyles demeure à ce jour inconnue. En faisant réagir le glucose ou le glyoxal avec de l’alanine isotopiquement marquée ( [13C]- et [15N]- alanine), les mécanismes de la formation du 1,2-butanedione et le 3,4-hexanedione furent confirmés, procédant par l’allongement de chaîne assisté par des acides aminés. De plus, le rôle des α-dicarbonyles dans la formation de divers composés hétérocycliques tels que des pyrazines, des pyrazinones et des imidazolidinones fut aussi étudié. Cette thèse démontre l'utilité de la pyrolyse-CG/SM en tant qu’outil analytique puissant, particulièrement lorsque utilisée en combinaison avec des techniques d'isotopes marqués.
VI Statement from the thesis office
In accordance with the regulations of the Faculty of Graduate and Postdoctoral studies of McGill University, the following statement from the guidelines for the Thesis preparation is included:
Candidates have the option of including, as part of the thesis, the text of one or more papers submitted, or to be submitted, for publication, or the clearly-duplicated text of one or more published papers. These texts must conform to the "Guidelines for Thesis Preparation" and must be bound together as an integral part of the thesis.
The thesis must be more than a collection of manuscripts. All components must be integrated into a cohesive unit with a logical progression from one chapter to the next. In order to ensure that the thesis has continuity, connecting texts that provide logical bridges preceding and following each manuscript are mandatory.
The thesis must conform to all other requirements of the "Guidelines for Thesis Preparation" in addition to the manuscripts.
In general, when co-authored papers are included in a thesis, the candidate must has made a substantial contribution for all papers included in the thesis. In addition, the candidate is required to make an explicit statement in the thesis as to who contributed to such work and to what extent. This statement should appear in a single section entitled "Contributions of Authors" as a preface to the thesis.
When previously published copyright material is presented in a thesis, the candidate must include signed waivers from the publishers and submit these to the Graduate and Postdoctoral Studies Office with the final deposition.
VII Contribution of Authors
This thesis is presented in manuscript format and consists of eight chapters. Chapter 1 is a brief introduction to the pyrolysis-GC/MS based techniques used in the thesis followed by the rational, research objectives and the significance of the proposed research. Chapter
2 is the literature review on the Maillard reaction and the analytical approaches. Chapters
3 to 7 constitute the main body of the thesis. Chapters 3, 4 and 5 are based on published manuscripts. Chapter 7 is submitted for publication and content of Chapter 6 will be submitted for publication. Chapter 8 is a brief summary of contribution to knowledge.
Connecting paragraphs provide logical bridges between different chapters. This dissertation is in accordance with the guidelines for thesis preparation as published by the faculty of Graduate Studies and Research of McGill University.
The present author was responsible for the concepts, experimental designs and manuscript preparation in all the published and submitted papers. The thesis supervisor,
Dr. Varoujan A. Yaylayan had a direct advisory input into the work as it progressed and as the manuscript co-author critically edited the dissertation prior to its submission. The manuscript which is not a part of this thesis and entitled “Oxidative pyrolysis and post- pyrolytic derivatization techniques for the total analysis of Maillard model systems: investigation of control parameters of Maillard reaction pathways” was co-authored with
Drs. Yaylayan, Haffenden and Wnorowski. Dr. Yaylayan was the first author and the present author’s contribution was to conduct the substantial part of the experimental work and to offer technical assistance.
VIII Publications
Yaylayan, V.A., Haffenden, L., Chu, F.L., and Wnorowski, A. 2005 Oxidative pyrolysis and postpyrolytic derivatization techniques for the total analysis of maillard model systems: investigation of control parameters of maillard reaction pathways. Annals of the New York Academy of Sciences, 1043, pp. 41-54.
Chu, F.L. and Yaylayan, V.A. 2008a. Model Studies on the Oxygen-Induced Formation of Benzaldehyde from Phenylacetaldehyde Using Pyrolysis GC-MS and FTIR. Journal of Agricultural and Food Chemistry, 56(22), pp. 10697-10704.
Chu, F.L. and Yaylayan, V.A. 2008b Post-schiff base chemistry of the Maillard reaction. Annals of the New York Academy of Sciences, 1126, pp.30-37.
Chu, F.L. and Yaylayan, V.A. 2009 FTIR monitoring of 5-oxazolidinone formation and decomposition in a glycolaldehyde-phenylalanine model system by isotope labelling techniques. Carbohydrate Research, 344(2), pp. 229-236.
Chu, F.L. and Yaylayan, V.A. (Submitted) Isotope labelling studies on the origin of 3,4- hexanedione and 1,2-butanedione in alanine/glucose model system. Journal of Agricultural and Food Chemistry
Chu, F.L. and Yaylayan, V.A. (in preparation) Identification and mechanism of formation of (2R) and (2S)-2,5-dimethyl-3-ethyl-4H-imidazolidinone in alanine/glucose model system using isotope labelling technique. Food Chemistry
Chu, F.L. and Yaylayan, V.A. (in preparation) Alkyl-substituted pyrazines, pyrazinones, pyridines and pyrroles formation in alanine/glucose model system. Food Chemistry
IX Conference Presentations
Yaylayan, V. A., Haffenden, L., Chu, F.L., Wnorowski, A. Technique of Oxidative Pyrolysis and Post-Pyrolytic Derivatization for the Total Analysis of Model Systems: Investigation of Control parameters of Maillard Reaction Pathways. Presented at the 8th International Symposium on the Maillard Reaction, Charleston, SC, USA, August 29- September 1st, 2004
Chu, F.L. and Yaylayan, V. Total analysis of l-alanine/glucose model system using oxidative pyrolysis and post-pyrolytic derivatization techniques. Presented at the CIFST/AAFC joint meeting, Montreal, Canada, May 28-30th, 2006
Yaylayan, V. and Chu, F.L. Post-Schiff base chemistry of the Maillard reaction. Presented at the 9th International Symposium on the Maillard Reaction, Munich, Germany, Sept-1-5th, 2007.
Chu, F. L. and Yaylayan, V. Isotope labeling and Pyrolysis-GC/MS based techniques for the analysis of Maillard reaction. Presented at IFT annual meeting, New Orleans, USA, June 28-July 1st, 2008.
Yaylayan V., Perez, C.P.; Chu F.L. Mechanism of furan, HMF and acrylamide formation: Recent findings. Presented at the COST Action 927 meeting held in Smolenice, Slovakia, Oct 2-4th, 2008.
X Acknowledgements
I want to sincerely thank Dr.Yaylayan for his patience and guidance. His wealth of knowledge has helped me to overcome many of the challenges that I encountered. He motivated me to go beyond my limits.
I want to thank all my labmates, Andre, Anja, Carolina, Luke and Plamen. I learned different perspectives of thinking from each one of them and without them, it is impossible to survive the countless number of hours in the lab. I want to thank Carolina again who kept me company from the start and who also proof-read or actually rewrote my French abstract.
I want to thank all the staff and friends in the department, they all had been very friendly and helpful.
I want to give thanks to my brothers and sisters at Montreal Chinese Alliance Church for their prayers with all the ups and downs.
I want to deeply thank all my family members for their continuing support and encouragement. I especially want to thank my wife Hoi Ying, who was there with me throughout my graduate studies.
I also want to praise the Lord who provided the wisdom to complete this thesis. May God be glorified.
XI Table of Contents
Abstract...... III Résumé...... V Statement from the thesis office ...... VII Contribution of Authors...... VIII Publications...... IX Conference Presentations...... X Acknowledgements...... XI Table of Contents...... XII List of Figures ...... XV List of Tables ...... XIX List of Abbreviations ...... XXI
Chapter 1: Introduction ...... 1 1.1 General introduction ...... 2 1.2 Isotope labelling, post-pyrolytic derivatization and oxidative pyrolysis techniques 3 1.3 Rationale and research objectives...... 4 1.4 Significance of the proposed research ...... 6
Chapter 2: Literature Review...... 8 2.1 General introduction ...... 9 2.2 Complexity and challenges of studying Maillard reaction ...... 9 2.3 Analytical techniques to study the Maillard reaction with emphasis on isotope labelling approach...... 11 2.4 Isotope labelling, post-pyrolytic derivatization and oxidative pyrolysis techniques ...... 15 2.5 Reaction pathways and reactive intermediates involved in the generation of Maillard products ...... 18 2.5.1 Reactive intermediates ...... 19 2.5.2 Amino acid degradation and interaction products ...... 22
Chapter 3: Post-Schiff base chemistry of the Maillard reactions: Mechanism of imine isomerization...... 32 3.1 Introduction...... 33 3.2 Material and methods...... 35 3.2.1 Pyrolysis GC/MS analysis ...... 35 3.2.2 Generation and FTIR monitoring of 5-oxazolidinone intermediate ...... 36 3.2.3 Browning measurement by UV/VIS...... 36 3.3 Results and discussion ...... 36 3.3.1 Isomerization of imines through transamination ...... 37 3.3.2 Isomerization of imines through 5-oxazolidinone and azomethine ylide formation...... 39 3.3.3 Evidence of imine isomerizations in phenylalanine/glyceraldehyde model system ...... 43
XII Chapter 4: FTIR monitoring of 5-oxazolidinone formation and decomposition in glycolaldehyde/phenylalanine model system by isotope labelling techniques...... 48 4.1. Introduction...... 49 4.2 Experimental...... 50 4.2.1 Extraction of 5-oxazolidinone “toluene extract” and FTIR analysis ...... 50 4.2.2 FTIR monitoring of 5-oxazolidinone and imine formation ...... 51 4.3 Results and discussion ...... 51 4.3.1 FTIR spectroscopy and isotope labelling studies...... 53 4.3.2 Infrared band assignments ...... 53 4.3.3 FTIR monitoring of 5-oxazolidinone formation and decomposition in phenylalanine/glycolaldehyde model system ...... 60 4.3.4 Implications of formation of an 5-oxazolidinone intermediate in the mechanism of the Amadori rearrangement...... 64
Chapter 5: Model Studies on the Oxygen Induced Formation of Benzaldehyde from Phenylacetaldehyde using Pyrolysis GC-MS and FTIR...... 66 5.1 Introduction...... 67 5.2 Material and Method...... 68 5.2.1 Pyrolysis GC/MS analysis ...... 68 5.2.2 Oxidative or wet Py-GC/MS analysis...... 69 5.2.3 FTIR monitoring of oxidation reactions ...... 70 5.3 Results and discussion ...... 70 5.3.1 Proposed mechanism of benzaldehyde formation from phenylacetaldehyde.. 80 5.3.2 Proposed mechanism of benzaldehyde formation from other precursors...... 84 5.3.2.1 Phenylalanine and phenylpyruvic acid...... 84 5.3.2.2 1-phenyl-1,2-ethandiol / phenylglyoxal /phenylglyoxylic acid ...... 84 5.3.2.3 Phenethylamine...... 85
Chapter 6: Formation of selected volatile, non-volatile and reactive intermediates generated in alanine/glucose model ...... 87 6.1 Introduction...... 88 6.2 Experimental Procedures ...... 90 6.2.1 Materials and Reagents ...... 90 6.2.2 Pyrolysis-Gas Chromatography-Mass Spectrometry (Py-GC/MS) ...... 90 6.2.3 Trapping of α-dicarbonyl intermediates with o-phenylenediamine...... 91 6.2.4 Spiking experiments with α-dicarbonyl compounds ...... 91 6.2.5 Ion Chromatography ...... 92 6.2.6 Post-Pyrolytic derivatization of non-volatiles...... 92 6.3 Results & Discussion ...... 92 6.3.1 Formation of α-dicarbonyls in alanine/glucose model system...... 93 6.3.2 Acetaldehyde formation...... 96 6.3.3 Volatiles generated from alanine/glucose Model system ...... 98 6.3.3.1 Pyrazine formation...... 98 6.3.3.2 Alkyl-substituted pyrazines formation...... 100 6.3.3.3 Pyrazinones formation...... 107 6.3.3.4 Imidazolidinones formation ...... 113
XIII 6.3.3.5 Pyrroles formation...... 117 6.3.3.6 Pyridines formation ...... 119 6.3.4 Non volatile products from alanine/glucose ...... 123 6.5 Conclusion ...... 125
Chapter 7: Isotope labelling studies on the origin of 3,4-hexanedione and 1,2-butanedione in alanine/glucose model system ...... 127 7.1 Introduction...... 128 7.2 Experimental Procedures ...... 129 7.2.1 Materials and Reagents ...... 129 7.2.2 Pyrolysis-Gas Chromatography-Mass Spectrometry (Py-GC/MS) ...... 129 7.2.3 Oxidative Py-GC/MS...... 130 7.2.4 Identification of pyrazines ...... 130 7.3 Results and discussion ...... 130 7.3.1 Chain elongation reactions of α-dicarbonyl compounds ...... 136 7.3.2 How alanine is involved in the chain elongation process of glyoxal?...... 141
Chapter 8: General conclusion and contribution to knowledge...... 143 8.1 General conclusions ...... 144 8.2 Contribution to knowledge ...... 145
Reference ...... 148
XIV List of Figures
Figure 2.1 Isotopically labelled precursors: D-glucose and glycine……………………..13
Figure 2.2 Mechanism of carboxylmethyllysine (CML) formation from glucose/lysine via β−dicarbonyl cleavage……………………………………………………………….14
Figure 2.3a Trapping step in oxidative pyrolysis..…………………..………………….17
Figure 2.3b Desorbing step in oxidative pyrolysis.……………..………………………17
Figure 2.4 Mechanism of formation of acetol, 2,3-butanedione, 1-deoxyglucosone and 3- deoxyglucosone, glyceraldehyde, glyceric acid, glycoaldehyde, glyoxal, pyruvaldehyde, from glucose……………………………………………………………………………...21
Figure 2.5 Mechanism of formation of (a) 2,3-butanedione (b) 2,3-pentanedione from alanine (c) and from glucose ……………………...……………….…………………..22
Figure 2.6 Mechanism of formation for phenethylamine and phenylacetaldehyde through Strecker degradation of phenylalanine…………………………………………...………23
Figure 2.7 Mechanism of Strecker degradation…………………………………….…...24
Figure 2.8 Mechanisms formations for 2,3-diethyl-5-methyl-pyrazine…………….…...25
Figure 2.9 Proposed mechanism of acetaldehyde formation from Strecker degradation of alanine…………………………………………………………………………...……….27
Figure 2.10 Proposed mechanism of formation of acetaldehyde containing C-1/C-2 atoms from D-Glucose……………………………………………………………………29
Figure 2.11 Proposed mechanism of formation of acetaldehyde containing C-5/C-6 from D-Glucose………………………………………………………………..………………30
Figure 3.1 Chemical transformations of Schiff bases…………………………………...34
Figure 3.2 Proposed mechanisms of imine isomerizations in the Maillard reaction (a) base catalyzed transamination, (b) oxazolidinone pathway, (c) decarboxylative transamination……………………………………………………………..……………..38
Figure 3.3 FTIR spectrum of phenylalanine/glyceraldehyde and [13C-]- phenylalanine/glyceraldehyde acquired after heating for 10 min in toluene at 110oC………………………………………………………………………….………….40
XV Figure 3.4 UV-VIS spectrum of phenylalanine/glyceraldehyde solution (A) acquired after heating for 8 min in DMSO at 80oC and (B) after additional storage at room temperature for 6h………………………………………………………………………..42
Figure 3.5 Isomerizations of the Schiff base formed between phenylalanine and glyceraldehyde through transamination and 5-oxazolidinone formation………………..44
Figure 3.6 Proposed mechanism of formation of 3-phenylpyridine and its mass spectrum as compared with authentic NIST library spectrum in head-to-tail fashion……………..45
Figure 4.1 Summary of chemical transformations of the Schiff base formed between an amino acid and glycolaldehyde………………………………………………………..…52
Figure 4.2 Band assignments of the intermediate compounds formed in the reaction between glycolaldehyde and phenylalanine……………………………………………...54
Figure 4.3 Absorption of the carbonyl and the imine regions (1900-1500 cm-1) of the toluene extracts of phenylalanine/glycolaldehyde and [13C-1]-phenylalanine /glycolaldehyde reaction mixtures acquired at (A) 35oC and (B) 80oC……………….....56
Figure 4.4 Infrared spectra (1900 – 1450 cm-1 region) of phenylalanine·HCl and [13C-1]- phenylalanine·HCl …………………………………………………………………...….57
Figure 4.5 Absorption of the carbonyl and the imine regions (1800-1540 cm-1) of the toluene extracts of phenylalanine/glycolaldehyde reaction mixtures acquired at 80oC over a period of 10 min showing the intensity of the band at 1778 cm-1 acquired at 35oC as a reference…………………………………………………………………………….……57
Figure 4.6 Second-derivative spectra (Savitsky Golay polynomial 2, points 15) of toluene extracts of phenylalanine/glycolaldehyde, [13C-1]-phenylalanine/glycolaldehyde and [15N]-phenylalanine/glycolaldehyde reaction mixtures heated at 80oC for 10 min…...…59
Figure 4.7 Absorption of the carbonyl and the imine regions (1800-1550 cm-1) of the phenethylamine/glycolaldehyde and phenethylamine·HCl /glycolaldehyde reaction mixtures heated at 70oC for 2 min……………………………………………………….59
Figure 4.8 Absorption of the carbonyl and the imine regions (1800-1550 cm-1) of the ethanolamine/phenylacetaldehyde and ethanolamine·HCl/phenylacetaldehyde reaction mixtures heated at 35oC for 2 min……………………………………………………….60
Figure 4.9 Time-dependent spectra of glycolaldehyde/phenylalanine mixture heated at 35oC for 55 min showing the initial 15 min.……………………………………………..63
Figure 4.10 Time-dependent spectra of glycolaldehyde/phenylalanine mixture heated at 35oC for 55 min showing the final 40 min……………………………………….………63
XVI Figure 5.1 Pyrograms generated at 250oC using temperature program B from (a) phenylalanine/glucose (1:1) under non-oxidative (b) phenylalanine/glucose (2:1) under non-oxidative (c) phenylalanine/glucose (1:1) under oxidative (d) phenylalanine/glucose (2:1) under oxidative conditions…………………………………………………………72
Figure 5.2 Pyrogram of phenylacetaldehyde (1mg) generated at 175oC under (a) wet/oxidative (b) dry/oxidative and (c) dry/non-oxidative conditions (d) pyrogram generated through oxidative pyrolysis of 1-phenyl-1,2-ethanediol at 175oC. using temperature program A…………………………………………………………………..76
Figure 5.3 Proposed mechanisms of formation of benzaldehyde from phenylalanine. Pathway A = Oxidative decarboxylation; pathway B = thermal decarboxylation, pathway C = Strecker or imine isomerization pathway………………………………….………...78
Figure 5.4 Proposed formation pathways of benzaldehyde precursors from phenylalanine…………………………………………………………………………….79
Figure 5.5 Proposed mechanism of free radical initiated oxidation of phenylacetaldehyde from its enol form………………………………………………………………………..81
Figure 5.6 FTIR spectra of phenylacetaldehyde, benzaldehyde acquired at 35oC, and phenylacetaldehyde acquired at 100oC.………………………………………………….82
Figure 5.7 Time-dependant spectra of phenylacetaldehyde oxidation catalyzed by 1,1’- azobis-(cyclohexane-carbonitrile) acquired at 100oC over 1h period scanned at 2 min intervals…………………………………………………………………………….…….83
Figure 6.1 Examples of α-dicarbonyls detected in alanine/glucose or alanine/glyoxal glyoxal (A), pyruvaldehyde (B) 2,3-butanedione (C), 1,2-butanedione (D), 2,3- pentanedione (E), 3,4-hexanedione (F) and 2,3-hexanedione (G)………………………93
Figure 6.2 Detection of dicarbonyl reactive intermediates by o-phenylenediamine as quinoxaline derivatives…………………………………………………………………..94
Figure 6.3 Mechanism formation for ethylamine and acetaldehyde from alanine……...97
Figure 6.4 General pathways of pyrazine formation from α-dicarbonyls……………..100
Figure 6.5 Some examples of pyrazine formation from their corresponding α-dicarbonyls ………………………………………………..…………………………………………103
Figure 6.6 Mechanism of formation for pyrazinone from alanine and α-dicarbonyls...110
Figure 6.7 Incorporation of alanine atoms in pyrazinone structures excluding the dicarbonyl moiety based on labelling experiments…………………………………..…112
XVII Figure 6.8 Proposed mechanism of formation of (2R) and (2S)-2,5-dimethyl-3-ethyl-4H- imidazolidinone in alanine/glucose model system………………….………………….115
Figure 6.9 Mass spectra and the proposed mass spectral fragments of 2,5-dimethyl,3- ethyl-4H-imidazolidinone found at retention times of 19.98 min (A) and 20.14 min (B) in alanine/glucose model system…………………………………………………………..116
Figure 6.10 Proposed mechanism of formations for pyrrole and N-ethyl-pyrrole from alanine/glycoaldehyde model…………………………………………….……………..118
Figure 6.11 Proposed mechanism of formations for N-alkyl-pyrrole from acetaldehyde and 2-amino-acetaldehyde……………………...………………………………..……..119
Figure 6.12 Mass spectra and the proposed mass spectral fragments of pyridines found at time 16.17 min, 16.48 min and 16.61 min………………………………………...……121
Figure 6.13 Proposed mechanism of pyridine formation from glyoxal/alanine……….122
Figure 6.14 Proposed mechanisms for the formation of selected non-volatiles from alanine/glucose ……….……………………………………………………………...…124
Figure 7.1 Ethyl-substituted pyrazines identified in glucose/alanine model system and percent incorporation of alanine C-2’ and C-3’ atoms…………………...……….……133
Figure 7.2 The α-dicarbonyl precursors required for oxidative and non-oxidative pyrazine formation in glucose/alanine model system………………………………….134
Figure 7.3 Proposed oxidative and non-oxidative mechanisms of pyrazine formation..137
Figure 7.4 Chromatogram generated at 210oC by the pyrolysis of glyoxal/alanine model system…………………………………………………………………………………..139
Figure 7.5 Proposed mechanisms of formation of 1,2-butanedione and 3,4-hexandione. …………………………………………………………………………………………139
Figure 7.6 The α-dicarbonyl precursors required for oxidative and non-oxidative pyrazine formation in glyoxal/alanine model system………………………….……….140
Figure 7.7 Proposed bi-cyclic transition state for the simultaneous aldol addition and decarboxylation reaction leading to chain elongation of simple α-dicarbonyl compounds……………………………………………………………………………...142
XVIII List of Tables
Table 3.1 Effect of addition of carbonyl compounds on the decarboxylation efficiency of phenylalanine at 250oC………………………………………………………………...... 41
Table 3.2 Effect of temperature on the decarboxylation efficiency of phenylalanine…..42
Table 5.1 Occurrence of benzaldehyde and its precursors in different model systems.…71
Table 5.2 Relative amounts of phenylacetaldehyde, phenethylamine and benzaldehyde formed in glucose/phenylalanine model systems at 250oC……………...…………….…74
Table 5.3. Amounts of benzaldehyde formed from different precursors between 175- 250oC normalized relative to phenylethanediol system………………………………….74
Table 5.4 Ability of phenylacetaldehyde to generate benzaldehyde relative to the listed precursors under optimum conditions……………………………………………………77
Table 6.1 Quinoxalines detected during pyrolysis of alanine/glucose model system…...95
Table 6.2 Isotope label incorporation in 2-ethyl-3-methyl-quinoxaline………….……..96
Table 6.3 Pyrazines detected during pyrolysis of alanine/glucose model system……....99
Table 6.4 Pyrazines detected during pyrolysis of alanine/glucose model system...…...102
Table 6.5 Mass spectrometric data and retention times of pyrazines compared to Baltes et al (1987a)……………………………………………………………………………….104
Table 6.6 Increase/decrease in intensity of the peaks associated with pyrazines when spiked with selected α-dicarbonyls……………………………………………………..106
Table 6.7 Possible pyrazinones detected during pyrolysis1 of alanine/glucose model system…..………………………………………………………………………………109
Table 6.8 Increase/decrease in intensity of the peaks associated with pyrazinones when spiked with selected α-dicarbonyls……………………………………………………..111
Table 6.9a The number of labelled atoms in imidazolidinones………………………..114
Table 6.9b Incorporation of labels in the mass spectral fragments of 2,5-dimethyl-3- ethyl-4H-imidazolidinone(tR= 19.98min)……………………………………………….114
Table 6.9c Incorporation of labels in the mass spectral fragments of 2,5-dimethyl-3-ethyl-
4H-imidazolidinone(tR= 20.14min)……………………………………………………..114
XIX Table 6.10 The number of alanine atoms incorporated in pyrroles during pyrolysis of alanine/glyoxal model system………………………………………………………….119
Table 6.11 The number of alanine atoms incorporated in pyridine during pyrolysis of alanine/glyoxal model…………………………………………………………………..122
Table 7.1 Pyrazine and ethyl-substituted pyrazines detected during pyrolysis of glyoxal/alanine model system…………………………………………………………..131
Table 7.2 The number of isotopic atoms incorporated in pyrazines generated from glyoxal/alanine model systems…………………………………………………………131
XX List of Abbreviations
AMDIS American Mass Spectral Deconvolution and Identification System ATR Attenuated Total Reflectance CAMOLA Carbon Module Labelling CE Capillary Electrophoresis CHE Chain elongation CML Carboxymethyllysine CuCl2 Copper (II) chloride DAB 1,2-diaminobenzene or o-phenylenediamine 1-DG 1-deoxyglucosone 3-DG 3-deoxyglucosone DMSO Dimethyl sulfoxide DTGS Deuterated triglycine sulphate EFCS Electronic Flow Controller EMV Electron Multiplier Voltage FTIR Fourier Transform Infrared Spectroscopy GC Gas Chromatrography GC-ITMS Gas Chromatography-Ion Trap Mass Spectrometry GC×GC–TOFMS Two-dimensional Gas chromatography -Time-of Flight Mass Spectrometry GC/MS Gas Chromatography -Mass Spectrometry GC-MS/MS Gas Chromatography -Tandem Mass Spectrometry GC–TOFMS Gas Chromatography-Time-of-Flight Mass Spectrometry HCl Hydrogen chloride He Helium HMDS Hexamethyl-disilazane HPLC High Performance Liquid Chromatography LC Liquid Chromatography MS Mass Spectrometry NIST National Institute of Standards and Technology NMR Nuclear Magnetic Resonance PhIP 2-amino-1-methyl-6-phenylimidazo-[4,5-b]-pyridine Py-GC/MS Pyrolysis-Gas Chromatrography/Mass Spectrometry RA Retroaldisation SD Strecker Degradation SPT Sample Pre-concentration Trap SR Strecker reaction TMS Trimethyl-silyl TMSDEA Trimethyl-silyldiethylamine UV Ultra-Violet UV-VIS Ultra-Violet-Visible
XXI Chapter 1: Introduction
1 1.1 General introduction
Maillard reaction is a term reserved for the reaction of reducing sugars with the amino- containing moieties, such as amino acids, peptides or proteins (Hodge, 1953). It is a form of non-enzymatic browning first described by Louis-Camille Maillard in 1912 (Maillard,
1912). During thermal processing, Maillard reaction is associated with the formation of aroma, texture and colour (Rizzi, 1997; Arnoldi et al., 1997; Hofmann, 1998b; Frank &
Hofmann, 2000; Gerrard et al., 2002; Yaylayan, 2003; Cerny, 2008), which are the key factors responsible for the consumer acceptance of many food products. The Maillard reaction is known to occur in heated, dried or stored foods (Duckham et al., 2002; Fay &
Brevard, 2005; Messia et al., 2005) and it is important to understand the mechanistic pathways underlying the changes in physical and chemical properties of food both during processing and storage. The reactive intermediates formed during this reaction (Weenen,
1998) can further react with other food components such as lipid oxidation products
(Hidalgo et al., 1999; Adams et al., 2005) and increase the diversity of the products. As a result, the Maillard reaction can be considered as a complex network of interactions among different precursors (Fayle et al., 2001) that proceeds via a complex series of steps and can be affected by various environmental factors such as temperature and time of heating, type of reactants, ratio of reactants and the presence or absence of oxygen
(Bemis-Young et al., 1993; Fayle et al., 2001; Tehrani et al., 2002; Chu & Yaylayan,
2008a; Chu & Yaylayan, 2008b). All the above factors can influence the specific course of the reaction and the final physical and chemical properties of the food product. The initial phase of the Maillard reaction is a well known process, which involves the formation of glycosylamine or the Schiff base. However, the subsequent reactions after
2 the Schiff base formation are in need of further investigation mainly due to the discovery of the fact that a previously unknown degradation product of asparagine, the acrylamide a potential food toxicant, can arise directly from the degradation of the corresponding glycosylamine (Stadler et al., 2002) or the Schiff base intermediate prior to the occurrence of Amadori rearrangement step (chapters 3 and 4 are dedicated specifically to study the post-Schiff base chemistry of the Maillard reaction). Such detailed understanding of the mechanism of Maillard reaction may provide the technical knowledge needed to control different pathways leading to the formation of desirable or undesirable products. The food industry today is facing the challenge of understanding the parameters that control the Maillard reaction and to successfully manipulate the conditions for the selective generation of aromas and colours while at the same time minimizing the formation of toxic and off-flavour components. However, generation of reliable mechanistic information can only be achieved through isotope labelling techniques (Tressl et al., 1993; Yaylayan & Keyhani, 2000; Schieberle, 2005) and requires the ability to perform analysis of Maillard reaction mixtures under various reaction conditions.
1.2 Isotope labelling, post-pyrolytic derivatization and oxidative pyrolysis techniques
One such analytical approach that can provide the ability to perform isotope labeling studies using minimum amounts of reactants with the flexibility of changing reaction parameters such as temperature and to perform derivatization and reactions under oxidative conditions is the pyrolysis technique in tandem with GC/MS. Pyrolysis can be considered a thermal extraction method suitable for gas phase study of the Maillard reaction. It is considered as a form of heating that induces chemical reactions and
3 decomposition of organic materials and which break apart large complex molecules into smaller and more analytically useful fragments for acquiring structural information.
Pyrolysis involves a rapid and controlled heating of the sample to a predetermined temperature ranging from 150-1000oC. Pyrolysis is usually performed in the absence of oxygen because capillary GC columns are sensitive to oxygen. To study the effect of oxygen and mimic the cooking system, a special set-up is required for conventional
GC/MS system. Yaylayan et al. (2005) introduced a new sample introduction method known as “oxidative pyrolysis” by using a gas stream switching valve together with a sample pre-concentration trap (SPT) coupled to a Gas Chromatography / Mass
Spectrometer (GC/MS). This approach also enabled the use of isotopically enriched starting materials to provide structural information for mechanistic studies. In addition, a derivatization technique was also introduced in this study for the investigation of α- dicarbonyl intermediates and other non-volatile components. For the reactive α- dicarbonyls intermediates, o-phenylenediamine was used to trap them as their corresponding quinoxaline derivatives. As for the other non-volatiles, trimethyl- silyldiethylamine (TMSDEA) derivatizing agent was used to increase the volatility for their subsequent detection by GC/MS. The main advantage of this novel technique lies in the fact that it requires only a small quantity of reactant material (~1mg) to analyze both volatiles and non-volatiles, which is especially important when expensive isotope enriched starting materials are used.
1.3 Rationale and research objectives
A simple model system for the Maillard reaction consisting of glucose and ammonia can form more than 15 compounds. A slightly more complicated system of glycine and
4 xylose for example can generate over 100 components when heated above 100oC and the structure and mechanism of formation of some of these products are still unknown.
Despite increased attention to the Maillard reaction in recent years, it still represents a continuing challenge to the researchers worldwide due to the diverse pathways and complex steps involved. A total analysis of Maillard reaction is difficult to achieve because of the inherent differences in the sample preparation protocols between volatile and non-volatile samples, requirement for different separation media and detection methods employed for the analysis of polar versus non-polar analytes. In addition, the lack of commercially available standards also hinders progress towards identification of complex mixtures. Different analytical approaches to achieve such a total analysis of
Maillard reaction mixtures are reviewed in Chapter 2. The application of isotope labelling and Py-GC/MS based techniques in solving specific problems in Maillard reaction chemistry was demonstrated in chapters 3 through 7. This approach not only provides structural and mechanistic information for the volatile components of the Maillard reaction but also for the non-volatiles and the reactive intermediates generated under oxidative and non-oxidative conditions.
The overall research objective of this thesis is to elucidate selected Maillard reaction pathways using Py-GC/MS based techniques and two model systems; (i) phenylalanine/sugar (Chapters 3, 4 and 5) and (ii) alanine/sugar (Chapters 6 and 7). The rationale behind the use of phenylalanine/sugar model system was to facilitate the detection and identification of relatively stable, easily detectable and commercially available decarboxylated amino acid moiety (phenethylamine) and its Strecker aldehyde
5 (phenylacetaldehyde) on the other hand, alanine/glucose represents a simple model system in which the amino acid possesses the smallest side chain possible (only one carbon atom) needed for the generation of substituted heterocyclic compounds and requiring only three 13C-labeleld analogues in addition to 15N atom for complete tracing of amino acid carbon atoms.
The specific objectives of the thesis were: (1) To investigate the post-Schiff base chemistry of Maillard reaction specifically the formation of 5-oxazolidinone intermediate and its ability to undergo decarboxylation. This objective was explored in chapters 3 and
4. In Chapter 4, isotope labelling studies were extended to include Fourier-transform infrared spectroscopy (FTIR) technique. (2) To investigate the role of oxygen and water in the Maillard reaction, specifically in the mechanism of generation of benzaldehyde; a potent aroma chemical that enzymatically forms in many fruits and potentially can develop as an off-flavour in processed food. The application of oxidative-pyrolysis technique was demonstrated in Chapter 5 in monitoring the conversion phenylacetaldehyde to benzaldehyde. (3) To investigate the origin and the mechanism of formation of complex α-dicarbonyls and their role in the generation of imidazolidinones, pyrazines and pyrazinones. These objectives were explored in chapters 6 and 7 using alanine/glucose as a model system together with the use of 13C and 15N isotope labelled precursors.
1.4 Significance of the proposed research
In addition to the known thermal decarboxylation of amino acids, identification of an alternative decarboxylation mechanism catalyzed by reducing sugars through the
6 formation of 5-oxazolidinone intermediate not only constitutes an important pathway of low energy degradation of amino acid derivatives but also provides a pathway of diversity through generation of isomeric and decarboxylated Schiff bases each capable of producing different Maillard reaction products. Furthermore, understanding the role of oxygen in the Maillard reaction constitutes an important element of control since performing the reaction under aerobic or anaerobic conditions is relatively easy. Finally, the formation and distribution of many aroma-active heterocyclic compounds is dependent on the amount and the type of α-dicarbonyl compounds in a given model system. Understanding the role of amino acid in their formation can dramatically enhance our ability to generate desired aroma effects in many processed foods.
7 Chapter 2: Literature Review
8 2.1 General introduction
The term Maillard reaction is reserved for the interaction of reducing sugars with the amino-containing moieties, such as amino acids, peptides or proteins (Hodge, 1953). It is a form of non-enzymatic browning associated with the formation of aroma, colour and texture (Arnoldi et al., 1997; Hofmann, 1998c; Frank & Hofmann, 2000; Gerrard et al.,
2002; Yaylayan, 2003; Cerny, 2008). Maillard reaction occurs during thermal processing such as baking, boiling and roasting of different types of food products (Saittagaroon et al., 1984; Baltes & Bochmann, 1987a; Oruna-Concha et al., 2002; Duckham et al., 2002;
Bianchi et al., 2008) and also occurs during drying and storage (Ferrer et al., 2003; Fay &
Brevard, 2005; Messia et al., 2005). Maillard reaction is responsible for the formation of distinct flavours and aromas in different foods and beverages, for example, chocolates, coffees, bakery products (Fayle and Gerrard 2002). During the reaction, off-flavours and toxic components can also be formed at the same time. Therefore, it is important to understand the mechanistic pathways of their formation. This may provide the technical knowledge needed to control the different pathways leading to the formation of desirable or undesirable products. The food industry today is facing the challenge of understanding the parameters that control the Maillard reaction, to successfully control the conditions for the selective generation of aromas and colours while at the same time minimizing the formation of toxic and off-flavour components.
2.2 Complexity and challenges of studying Maillard reaction
The current understanding of the Maillard reaction mechanism is mainly based on the analysis of the so called model systems using various sugars and amino acids as simple models representing a complex food matrix (Yaylayan, 1997; Fayle et al., 2001). Single
9 amino acid/sugar model systems had been widely used in the study of the Maillard reaction because a direct correlation can be established between the products and the specific amino acid such as alanine, asparagine, cysteine, glutamine, glycine, isoleucine, lysine, phenylalanine, proline, serine or threonine (Tressl et al., 1985; Baltes & Mevissen,
1988; Tressl et al., 1995; Ames et al., 1996; Tai & Ho, 1997; Chen et al., 1997; Yaylayan et al., 1998; Chen & Ho, 1999; Ames et al., 1999; Yaylayan & Keyhani, 2000; Frank &
Hofmann, 2000; Stadler et al., 2002; Mottram et al., 2002; Yaylayan & Haffenden,
2003a; Ehling & Shibamoto, 2005; Martins & van Boekel, 2005; Schieberle, 2005).
Furthermore, model systems containing peptides were also used to study the glycation process (Oh et al., 1991; Roscic et al., 2001; Kojic-Prodic et al., 2004; Horvat & Jakas,
2004; Lu et al., 2005; Hao et al., 2007). In these studies the chemical reactions were assumed to mimic the complex chemistry occurring in the foodstuff during processing
(Salter et al., 1989; Ames, 1998). However, despite the recent advances in the analytical separation methodologies and the increased attention paid to the Maillard reaction, studying this complex system still represents a difficult challenge (Ames, 2005; Fay &
Brevard, 2005; Hao et al., 2007). Fifty four components were identified in a simple model system such as glycine/glucose (Ames et al., 2001) and up to seven hundred compounds were reported in roasted coffee (Clifford, 1985; Baltes & Bochmann, 1987a).
Some of the products formed during this reaction are reactive intermediates (Weenen,
1998) and they can further react with other components and increase the diversity of the products (Yaylayan et al., 2003a). Furthermore, environmental factors such as the temperature, the presence of water or oxygen, the type and ratio of reactants will also determine the specific course of the reaction (Bemis-Young et al., 1993; Tehrani et al.,
10 2002). Thus, even though many compounds generated in the Maillard reaction were identified, the mechanisms of formation of many of these products still remain unknown.
Nevertheless, these components can be classified as volatiles, non-volatile, reactive intermediates and polymeric materials. Nursten (1981) had further classified the volatiles components into sugar fragmentation products, amino acid degradation products and interaction products. The diverse pathways of Maillard reaction products entail the availability of appropriate analytical procedures that allow their detailed identification however, such a total analysis of Maillard reaction systems is difficult to achieve due to the inherent differences in the sample preparation protocols, separation media and detection limits between analytical methodologies.
2.3 Analytical techniques to study the Maillard reaction with emphasis on isotope labelling approach
As mentioned above, the current understanding of the Maillard reaction mechanism is mainly based on the analysis of model-systems presumed to mimic the cooking process.
The extraction method used is considered the most important part of the experimental design, followed by separation and detection conditions (Fayle and Gerrard, 2002).
Certainly, some beverages and liquids products can be analyzed directly without extraction such as soya sauce, beer and fruit juices but the majority of foods are not suitable for direct analysis (Fayle and Gerrard, 2002). Typical food extraction procedures include homogenization with an appropriate solvent, centrifugation or distillation to remove the remaining solid. Some other current technology includes: solid phase micro extraction (Fadel & Farouk, 2002; Cerny & Davidek, 2003; Adams et al., 2008; Lojzova et al., 2009) and pyrolysis (Yaylayan et al., 2005; Paine et al., 2007) both are capable of providing a rapid and reliable method to extract especially the volatile Maillard reaction
11 products. As for the separation methods, capillary electrophoresis (CE), gas chromatography (GC) and liquid chromatography (LC) are some of the analytical tools that are currently being used to study Maillard reaction (Yaylayan & Keyhani, 1996;
Bailey et al., 1996; Ames et al., 1997; Monti et al., 1998; Royle et al., 1998; Wnorowski
& Yaylayan, 1999; de Sa et al., 2001; Fayle et al., 2001; Lojzova et al., 2009). Gas chromatography is used mainly to separate volatile compounds whereas capillary electrophoresis and liquid chromatography (LC) or High Performance liquid chromatography (HPLC) are used to separate the non-volatiles. In fact, capillary electrophoresis has been shown to resolve more components than reverse phase HPLC
(Ames et al., 1997; Royle et al., 1998). Gas chromatography can also be used for the analysis of non-volatiles through chemical derivatization (Yaylayan et al., 2005;
Haffenden & Yaylayan, 2008). These analytical tools can be equipped with different detectors to suit the particular research interests. For example, ultra-violet (UV) detector is one of the most commonly used for high performance liquid chromatography (HPLC) based separations and it can be used also with the capillary electrophoresis (CE), however, the drawback of UV is the fact that the specific wavelength chosen will influence the number of products detected (Fayle and Gerrard, 2002). For analyzing all eluting compounds, a complete UV visible spectrum can be recorded by using diode array detector on HPLC instruments (Bailey et al., 1996; Ames et al., 1998; Fayle &
Gerrard, 2002; Davidek et al., 2003). Gas chromatography / Olfactometry has been used to study the volatile components of Maillard reaction for sensory evaluations purposes
(Aaslyng et al., 1998; Kato, 2005; Bravo et al., 2008). Nuclear magnetic resonance
(NMR) and various specifications of Mass spectrometry (MS) are suitable techniques to
12 obtain structural information for mechanistic studies (Hofmann, 1998a; March, 2000;
Kojic-Prodic et al., 2004; Fay & Brevard, 2005). Mass spectrometry detectors are usually coupled to gas chromatography or liquid chromatography for the analysis of complex mixtures. For example, gas chromatography / ion trap mass spectrometry (GC–ITMS), gas chromatography / time-of-flight mass spectrometry (GC–TOFMS) and comprehensive two-dimensional gas chromatography / time-of-flight mass spectrometry
(GC×GC–TOFMS) were compared to study heterocyclic compounds in potato chips, and found that GC×GC–TOFMS offered the lower limit of detections for the target alkyl- pyrazine analyses (Lojzova et al., 2009). Nevertheless, reliable mechanistic information can only be achieved through isotope labelling techniques (Tressl et al., 1993; Yaylayan
& Keyhani, 2000; Schieberle, 2005). With the use of isotopically labelled precursors, the origin of carbon and nitrogen atoms in an unknown compound may be determined. The use of a complete set of labelled precursors ([13C-1]-, [13C-2], [13C-3]-, [13C-4]-, [13C-5]-,
[13C-6]-glucoses and [13C-1]-, [13C-2]- and [15N]-glycine - Figure 2.1) can allow easy determination of their eventual fate in specific target compounds.
c1 O c2 c'1
HOHc3 O c'2 HOH OH c4 NH HHO 2 c5 HOH
c6 N15 OH Figure 2.1 Isotopically labelled precursors: D-glucose and glycine
Schieberle (2005) had proposed a carbon module labelling (CAMOLA) technique, which uses a defined ratio of both unlabelled and labelled precursor mixtures, in order to
13 identify or clarify different mechanistic pathways based on mass spectrometric data and statistical rules. Kasper and Schieberle (2005) investigated the formation pathway of carboxymethyllysine (CML) by reacting hippuryllysine and a mixture of (1:1) unlabelled/13C-labelled glucose at various positions using LC/MS and propose the formation of carboxymethyllysine via a β−dicarbonyl cleavage of 2,4-dioxo intermediate
(see Figure 2.2) based on the changes in the distribution of different isotopomers. The
CAMOLA technique is a cost-effective and reliable approach which reduces the amount of expensive isotope labelled precursors needed by half and still provides mechanistic information.
O 1 1 H2N-R HN-R N-R 2 OH HOH 2 1 1 OH -H O HOH 2 2 -H2O H N-R 2 HHO + 2 HHO HHO HOH HOH HOH HOH HOH HOH HOH HOH
OH OH OH OH
CH3 NH-R NH-R N-R O OH 1 1 O 1 OH HOH NH-R 2 2 2 + OH +H2O 1 O OH 2 O O OH OH HOH HOH HOH
carboxymethyllysine OH OH OH Figure 2.2 Mechanism of carboxylmethyllysine (CML) formation from glucose/lysine via β−dicarbonyl cleavage. R = Lysine residue (Adapted from Kasper and Schieberle 2005)
14 2.4 Isotope labelling, post-pyrolytic derivatization and oxidative pyrolysis techniques
Different analytical procedures are reported in the literature for the identification of the
Maillard reaction products and attempts are still underway to develop methods to reduce the time of analysis and to increase the specificity. The total analysis of a reaction system is still difficult to achieve due to differences in analytical strategies between volatiles and non volatiles. Nevertheless, an example of a published method that combines extraction, separation and detection into a single process using pyrolysis coupled with gas chromatography/mass spectrometry (Py-GC/MS) and able to analyze volatiles and non volatiles both under oxidative and non-oxidative conditions is reported (Yaylayan et al.,
2005). This method had been used extensively in various studies on the Maillard reaction using cryogenic cooling to help trap volatiles component and silylation to help analyze the non-volatile residue remaining (Yaylayan et al., 2005; Haffenden & Yaylayan, 2008).
Pyrolysis is considered as a sampling and a thermal extraction method mainly used to analyze polymers. It is a form of heating that chemically decomposes organic materials in the absence of oxygen and at high temperatures. This involves a rapid and controlled heating of the sample to a predetermined temperature ranging from 150-1000oC.
Pyrolysis can break large complex molecules into smaller and more analytically useful fragments for acquiring structural information for polymeric materials. Pyrolysis is usually performed in the absence of air because capillary GC columns are sensitive to oxygen. Huyghues-Despointes et al (1994) introduced the concept of using Py-GC/MS system as a small scale chemical reactor to perform mechanistic studies with precursors.
Unlike the carbon module labelling (CAMOLA) technique introduced by Schieberle
(2005), the Py-GC/MS based technique allows utilization of labelled precursors without
15 the need for dilution with unlabelled material due to minimal requirements for the sample
(fraction of a milligram). Later Yaylayan et al. (2005) introduced a modification that allowed “oxidative pyrolysis” to be performed by using a gas stream switch-valve together with a sample pre-concentration trap (SPT) which were the two key components necessary to perform oxidative analyses. Oxidative pyrolysis involves two separate stages, in the first stage or the “trapping step” the volatiles are generated under air and are trapped in a Tenax sorbent while in the second stage the switch valve directs helium into the interface and the volatiles are thermally desorbed (Yaylayan et al., 2005). In the pyrolysis interface, the installed switch valve is used to direct either air or helium as the carrier gas and the volatiles generated during the pyrolysis are released and collected onto the sample pre-concentration trap (SPT), which contains a Tenax column and can absorb organic volatiles and vents the carrier gas (He or air) away from the GC column (see
Figure 2.3a). In the second stage or the desorption step, the volatiles trapped in the SPT can subsequently be “thermally desorbed” into the gas chromatography column for separation, by applying a reversed flow of helium (See Figure 2.3b).
16 Figure 2.3a Trapping step in oxidative pyrolysis (adapted from Yaylayan et al. 2005)
Figure 2.3b Desorbing step in oxidative pyrolysis (adapted from Yaylayan et al. 2005)
17 2.5 Reaction pathways and reactive intermediates involved in the generation of Maillard products
The initial step in the Maillard reaction between reducing sugars and amino acids is the formation of the Schiff base intermediate. Under acidic condition this intermediate can rearrange into 1-amino-1-deoxy-2-ketose, known as the Amadori rearrangement product if the sugar involved is an aldose and if a ketose sugar is used, the reaction product is known as “Heyns’ rearrangement product” (Ledl & Schleicher, 1990; Yaylayan, 1997).
Schiff bases play a critical role both in the initiation and propagation stages of the
Maillard reaction and their chemical modification can give rise to various processes such as transamination and oxazolidinone formation (Tsuge et al., 1987; Horvat et al., 1998;
Horvat et al., 1999; Aurelio et al., 2003; Kojic-Prodic et al., 2004; Perez Locas &
Yaylayan, 2008). Davidek et al (1990) suggested that isomerization of Schiff bases can generate the corresponding 2-keto acids. Yaylayan and Wnorowski (2002) used 13C- labelled precursor to confirm the imine isomerization formed in alanine/glycolaldehyde and pyruvic acid/aminoethanol model systems. Recently, Yaylayan et al. (2003b; 2004) have suggested a mechanism of decarboxylation of amino acids under dry heating conditions, in which the Schiff base undergoes intramolecular cyclization by the action of the carboxylate anion to form a 5-oxazolidinone intermediate capable of losing CO2 and forming a relatively stable azomethine ylide. Interestingly, this ylide can undergo 1,2- prototropic shift and generate two isomeric imines, providing a second mechanistic pathway of imine isomerization in addition to transamination. For example, degradation of the corresponding asparagine glycosylamine will form acrylamide, which is a potential food toxicant (Stadler et al., 2002). In general, the origin of different Maillard reaction
18 products can be classified into (1) sugar fragmentation products, (2) amino acid degradation products and (3) interaction products (Nursten, 1981).
2.5.1 Reactive intermediates
It is well accepted that degradation of sugars initiated by amino acids during the Maillard reaction can generate a large variety of reactive carbonyl or α-dicarbonyl compounds (See
Figure 2.4) such as acetol, 2,3-butanedione, 1-deoxyglucosone and 3-deoxyglucosone, glyceraldehyde, glyceric acid, glycoaldehyde, glyoxal and pyruvaldehyde (Weenen, 1998;
Yaylayan & Keyhani, 1999; Yaylayan & Keyhani, 2001b; Fay & Brevard, 2005). Glyoxal and glycoaldehyde for example can be generated through retroaldolisation in the C2/C3 position of glucose (Yaylayan & Keyhani, 1999) whereas 2,3-butanedione can be formed by the C-2/C-4 cleavage of glucose backbone (Huyghues-Despointes & Yaylayan, 1996).
The 1,2- and 2,3-enolization of Amadori product can lead to the formation of 3- deoxyglucosone and 1-deoxylglucosone (Weenen, 1998; Yaylayan & Keyhani, 1999; Fay
& Brevard, 2005). Both deoxyglucosones can undergo C-3/C-3 retroaldolisation to form pyrvualdehyde and glyceraldehyde based on isotope labelling data (Yaylayan & Keyhani,
1999). In addition, the 1-deoxyglucosone can isomerize into 2,4-dioxa-3,5,6- trihydroxyhexane which can undergo β-cleavage to form acetol and glyceric acid
(Weenen, 1998). In fact, longer α-dicarbonyl compounds such as 2,3-pentanedione, 2,3- hexanedione and 3,4-hexanedione were also detected in the Maillard model systems as free dicarbonyl compounds (Salter et al., 1989; Zheng et al., 1997; Whitfield & Mottram,
1999; Ames et al., 2001) and also in bread crust (Chiavaro et al., 2007). These longer chain α-dicarbonyl compounds including 2,3-butanedione, 2,3-pentanedione, 2,3- hexanedione and 3,4-hexanedione all are known to possess very strong characteristic
19 buttery odors (Abdel-Mageed & Fadel, 1995; Mosciano, 2005; Mosciano, 2007). In addition, they play a critical role in the generation of other heterocyclic aroma compounds during thermal processing particularly in the formation of pyrazines (Rizzi, 1972). The mechanism of formation of these longer alkyl chain α-dicarbonyl compounds is a less known and a more complex process due to their multiple origins and the involvement of both sugar and amino acid carbon atoms. For example, isotope labelling studies have shown that 2,3-butanedione can be generated from glucose via C-2/C-4 cleavage as mentioned above and it can also be formed through aldol condensation of pyrvualdehyde with glycine (see Figure 2.5a). Similarly, 2,3-pentanedione can be generated either through participation of alanine carbon atoms (see Figure 2.5b) or totally from sugar carbon atoms (Figure 2.5c) according to Yaylayan and Keyhani (1999). Weenen (1998) however, had proposed another mechanism for 2.3-pentanedione formation from glucose carbon atoms via aldol condensation of 2,3-butanedione and formaldehyde.
20
CH 3 CH3 O O OH H β-cleavage OH O Acetol OH HOH O
OH OH OH R R Glyceric acid O N NH O CH3 HOHH NR HOH 2 O 2,3-enolization O RA C3/C3 O HHO HHO HHO O CH3 HOH Pyruvaldehyde HOH HOH HOH HOH O HOH HOH HOH OH OH OH OH OH OH Glyceraldehyde Glucose Schiff base Amadori Compound 1-deoxglucosone
RA C2/C3 1,2-enolization
O O O O O O O OH RA C3/C3 CH HH 3 Glyoxal Glycoaldehyde Pyruvaldehyde HOH HOH O OH O OH HOH HOH OH OH HOH HOH 3-deoxglucosone Glyceraldehyde OH OH CH3 O
O
CH3 2,3-butanedione
Figure 2.4 Mechanism of formation of acetol, 2,3-butanedione, 1-deoxyglucosone and 3- deoxyglucosone, glyceraldehyde, glyceric acid, glycoaldehyde, glyoxal, pyruvaldehyde, from glucose (Adapted from Weenen, 1998; Yaylayan & Keyhani, 1999) RA=Retroaldolisation
21 CH 3CH CH3 2NH OH 3 NH2 3CH -CO2 - NH2 O +
A O O O O O O
2,3-butanedione
Pyrvualdehyde Glycine
2NH
3CH CH3 2NH OH 3CH CH3 3CH -CO2 - NH2 O + CH O O O B O 3 O O 2,3-pentanedione
Pyrvualdehyde Alanine
CH 3 3CH OH -H O CH 2 3 OH OH + CH OH C 3 CH3 CH O O 3 O O Acetol Acetaldehyde Figure 2.5 Mechanism of formation of (a) 2,3-butanedione (b) 2,3-pentanedione from alanine and (c) 2,3-pentanedione from glucose
2.5.2 Amino acid degradation and interaction products
Amino acids can also form specific degradation products such as reactive amines or aldehydes. For example, phenylalanine can generate phenethylamine and phenylacetaldehyde in the presence of a α-dicarbonyl source (Granvogl et al., 2006).
Both phenylacetaldehyde and phenethylamine possess very strong odor properties, in which phenylacetaldehyde is described as sweet, floral honey and rosy aroma (Mosciano,
2005), whereas phenethylamine has a vanilla and fishy odor (Merck Index). In fact, both phenethylamine and phenylacetaldehyde were found in chocolate and tomato (Schweitzer et al., 1975; Baker et al., 1987; Schnermann & Schieberle, 1997; Tieman et al., 2006).
The phenethylamine was also found in cheese (Baker et al., 1987), whereas, phenylacetaldehyde was found in buckwheat, cocoa beans, cooked beef and wheat flour
(Van Praag et al., 1968; Kato, 2005; Janes et al., 2008). Phenylalanine can undergo
22 Strecker reaction to generate phenylacetaldehyde (Mottram & Edwards, 1983) or undergo decarboxylation to form phenethylamine (Granvogl et al., 2006). In general, phenylalanine in the presence of a carbonyl or α-dicarbonyl source can form glycosylamine which can decarboxylate to form imine 1A and imine 1B (Granvogl et al.,
2006) both imines can hydrolyse and form corresponding phenylacetaldehyde, α-amino carbonyl, phenethylamine and the original α-dicarbonyl compound (see Figure 2.6).
OH O OH O R1 -H2O R O 1 + N NH2 O R2 O R Glycosylamine Phenylalanine 2
-CO2
R1 R1 N N OH O R2 R2 Imine 1B Imine 1A
+H2O +H2O
O R O R1 1
+ O + NH2 NH R O R2 2 2
Phenethylamine Phenylacetaldehyde
Figure 2.6 Mechanism of formation for phenethylamine and phenylacetaldehyde through Strecker degradation of phenylalanine (adapted from Granvogl et al., 2006)
23
Interaction products of the Maillard reaction retain moieties from the sugar and the amino acid components such as in the formation of pyrazines. They are considered to be important aroma-active components (Maga & Sizer, 1973; Shu, 1999; Adams et al.,
2008). Pyrazines impart baked, roasted, smoky and nutty types of aromas to the cooked foods and they are detected in cooked beef, cheese, roasted coffee and potato chips
(Mussinan et al., 1973; Baltes & Bochmann, 1987b; Qian & Reineccius, 2002; Lojzova et al., 2009). Pyrazines are the result of the Strecker degradation (Figure 2.7), a reaction between amino acids and α-dicarbonyl compounds at high temperatures (Rizzi, 1972).
Various mechanisms have been proposed for their formation based on different reaction models. For example, Bemis-Young et al (Bemis-Young et al., 1993) proposed the formation of 2-3-diethyl-5-methyl-pyrazine in glycine model system to proceed through the reaction of 1-hydroxy-2-butanal with acetaldehyde to form 2-hydroxyl-3-amino-4- hexanone, further reaction with 2-aminopropanal can form 2-3-diethyl-5-methyl-pyrazine
(Pathway A in Figure 2.8). Amrani-Hemaimi et al (1995) on the other hand, proposed an alternative pathway for this pyrazine in [13C-3]-alanine/hexose model system based on the
Strecker degradation of 1,2-butanedione and pyruvaldehyde to form 2-ethyl-5-methyl- dihydropyrazine, followed by the addition of acetaldehyde to form 2-3-diethyl-5-methyl- pyrazine (Pathway B in Figure 2.8). Low et al (2007) proposed the same pathway for this pyrazine identified in potato.
OO R1 R1 R 1 O R N 3 R N Strecker Aldehyde R O 3 1 R2 R3 -CO2 +H2O O R3 + NH2 -H O 2NH OH 2 OR2 O HRO 2 amino-acid H O R2 α−amino-ketone Figure 2.7 Mechanism of Strecker degradation with amino acid with α-dicarbonyl
24
3CH 3CH O OH OH O OH OH -H2O OH + CH OH 3 O OH O CH CH 3 3 CH CH 3 2-hydroxyl-butanal 3
3CH 3CH
2NH O O O OH AMADORI OH SD + Glycine O OH 3CH NH2 3 OCH CH3 CH3 Pyruvaldehyde 2-amino-propanal 2-hydroxy-3-amino- hexanone
Pathway A -3H2O
CH3
N CH3
O CH3 N oxidation -H2O CH3 CH3 H H Pathway B N CH3 N CH3 Pathway C
N N H H CH3 CH3 -2H2O -2H2O CH3
NH2 NH2 O CH3 O CH3 + + O O 2NH 2NH CH 3 CH3 1-amino-2-butanal 1-amino-2-propanal 3-amino-4-hexanal 1-amino-2-propanal Figure 2.8 Mechanisms formations for 2,3-diethyl-5-methyl-pyrazine Pathway A adapted from Bemis-Young et al. (1993); Pathway B adapted from Amrani- Hemaimi et al. (1995); Pathway C adapted from Rizzi (1972), SD=Strecker degradation
25 In fact, most accepted mechanistic pathways of alkyl-substituted pyrazine formation involves the condensation of two α-amino-ketones formed from their corresponding α- dicarbonyls through the Strecker reaction. This condensation results in the formation of dihydropyrazine which needs an oxidation step to form pyrazine (Rizzi, 1972; Adams et al., 2008). In this case, the Strecker degradation of 3,4-hexanedione together with pyrvuladehyde would form the two corresponding α-amino-ketones needed to form 2,3- diethyl-5-methyl-pyrazine (Pathway C in Figure 2.8). However, the pathway C has not been considered since the formation of 3,4-hexanedione from glycine/glucose was verified only recently in glycine/glucose model system (Ames et al., 2001).
Pyruvaldehyde on the other hand was detected as a glucose degradation product by many groups (Yaylayan & Keyhani, 2001b; Chen et al., 2005).
In alanine model systems, acetaldehyde can be formed by Strecker degradation of alanine and dicarbonyl sources (Rizzi, 2008). Alanine in the presence of α-dicarbonyl compound can form a hemiaminal (see Figure 2.9), which then undergoes dehydration and forms an imine 2A which is also capable of isomerization into imine 2B (Rizzi, 2008). Imine 2B can decarboxylate to form the azallylic ion 3 which is resonance stabilized; the hydrolysis of azallylic ions 3A and 3B can generate acetaldehyde and α-amino-ketone; whereas the hydrolysis of azallylic ion 3C will form ethylamine and the original α-dicarbonyl will be serving as a catalyst (Rizzi, 2008). Acetaldehyde can also be generated from glucose in glycine or other amino acid containing model systems. Paine et al (2008) investigated the mechanism of acetaldehyde formation from D-glucose using pyrolysis and 13C isotope labelling technique and found out that all the six carbon atoms from glucose can be incorporated into the acetaldehyde moiety. Acetaldehyde can be derived
26
3CH R1 O 3CH O NH 2 NH O + O R2 OH OH R2 O OH R1 Alanine Hemiaminal
-H2O
H 3CH 3CH + N O N O O O - OH R1 R2 O R1 R2 Imine (2B) Imine (2A)
-CO2
CH H H CH H 3 3CH 3 + + - + N O- N O C N O - C R1 R2 R1 R2 R1 R2
Azallylic ion (3A) Azallylic ion (3B) Azallylic ion (3C)
+H2O +H2O
R2 H O R2 CH O + 2NH 3 + NH O CH3 O 2 R1 Acetaldehyde R1 Ethylamine
Figure 2.9 Proposed mechanism of acetaldehyde formation from Strecker degradation of alanine (adapted from Rizzi 2008)
27 from C-1/C-2, C-2/C-3, C-3/C-4, C-4/C-5 and C-5/C-6 although the deconvolution analyses showed that C-1/C-2 and C-5/C-6 had the largest contributions; whereas the unusual odd combinations of C-2/C-3 and C-4/C-5 had relatively lower contribution and thus considered to be less important, nevertheless, C-2/C-3 and C-4/C-5 are suggested to be formed through carbo-cyclization from glucose as well as from re-fragmentation
(Paine et al., 2008). According to Paine et al. (2008), acetaldehyde generated from C-1/C-
2 or C-5/C-6 glucose resulted from dehydration and cyclic-Grob fragmentation (see
Figure 2.10 and 2.11) in which different mechanistic pathways illustrated the precise location of the carbonyl group with respect to each carbon atom.
The complex pathways of the Maillard reaction are still under investigation and there are many unknown reaction schemes that are being identified, the latest of which being the formation of acrylamide, a process-induced toxicant.
28 c1 OH
c2 OH c3 Enolization HOH Enolization c4 HHO H c1 O c5 OH HOH c1 c2 HOH c6 c2 O c3 OH c3 HOH HOH c4 c4 HHO HHO c5 Hydride migration c5 HOH HOH
-H2O c6 OH c6 OH Glucose Fructose O OH O c1 H c1 c2 c2 OH H CH OH c1 H 2 H c1 CH3 c2 c3 c2 O c3 OH Acetaldehyde c4 c4 HHO HHO c5 RA (2,3) c5 HOH HOH c6 O c6 OH OH H c1 c2 H H c3 O H c4 H O H O c5 c1 c1 H c2 c2 HOH OH OH H H c6 c3 O c3 O OH H c4 O c4 HHO -H2O c5 c5 HOH HOH O c1 OH Cyclic Grob c6 c1 c6 c2 fragmentation OH OH 3CH c2 2CH H H Acetaldehyde
Figure 2.10 Proposed mechanism of formation of C-1/C-2 containing acetaldehyde from D-Glucose (adapted from Paine et al. 2008). RA=Retroaldolisation
29 Pathway A
c1 H O c2 HOH c3 HOHc5 HOc5 HOH -H2O c4 H OH Cyclic Grob c6 CH c6 CH c5 2 3 HOHfragmentation Acetaldehyde c6 OH
Glucose
Pathway B
H c1 O H c1 O H c1 O c2 c2 c2 HOH HOH HOH c3 c3 -H2O c3 HOH HOH HOH c4 c4 c4 H O HHO OH c5 O HOH OH H H c5 c5 c6 H c6 c6 H OH Glucose
RA (4,5)
H c5 3 OCH OH c5 H c6 c6 H H Acetaldehyde
Figure 2.11 Proposed mechanism of formation of C-5/C-6 containing acetaldehyde from D-Glucose (adapted from Paine et al. 2008). RA=Retroaldolisation
30 Connecting paragraph
In Chapter 2, the literature review provided the background needed to comprehend the complexity of the Maillard reaction and the analytical approaches needed for its analysis.
Chapter 3 uses phenylalanine/glyceraldehyde as a model system to investigate the post-
Schiff chemistry of the Maillard reaction. In this study, the evidence for the occurrence of imine isomerization through 5-oxazolidinone formation was provided by FTIR spectroscopy and Py-GC/MS data. Chapter 3 was published in the Annals of the New
York Academy of Sciences (Chu, F.L. and Yaylayan, V.A. 2008b. Post-schiff base chemistry of the Maillard reaction. Annals of the New York Academy of Sciences, 1126, pp. 30-37)
31 Chapter 3: Post-Schiff base chemistry of the Maillard reactions: Mechanism of imine isomerization
+ N OH H
OH
N N OH OH
OH OH
32 3.1 Introduction
The initial step in the interaction between reducing sugars with amino acids is the formation of the so-called Schiff base intermediate (see Figure 3.1). Schiff bases, or imines, play a critical role, not only in initiating the Maillard reaction, but also in its propagation in the later stages of the reaction because of their chemical conversion into various reactive moieties able to interact and form new imines that are capable of further transformations. Little attention has been paid so far on the ability of these imines to isomerize and thus contribute to the diversity of Maillard reaction products. The possibility of the initial imine formed between a sugar and an amino acid will be converted into its isomeric imines during the Maillard reaction was first proposed by
Høltermand (1966) and has been referred to as transamination reaction. Similar isomerization of the imine, formed between α-keto acids and amino acids has been observed by Herbst and Engel (1934) and its mechanism characterized by Cram and
Guthrie (1966) as base-catalyzed methylene-azomethine rearrangement. Davidek et al
(1990) also suggested the conversion of amino acids through an imine isomerization mechanism into 2-keto acids in the presence of glyoxal during Maillard reaction. The first systematic study to confirm the isomerization of imines formed in alanine/glycolaldehyde and pyruvic acid/aminoethanol model systems was carried out by Yaylayan and
Wnorowski (2002) using 13C-labelled precursors. More detailed studies were carried out using Fourier-transform infrared spectroscopy (FTIR) and gas chromatography –tandem mass spectrometry (GC-MS/MS) to monitor the formation of the imine in the pyruvic acid/aminopropanediol model system and its isomerization and subsequent rearrangement into Amadori product (Wnorowski & Yaylayan, 2003). Recently, Yaylayan et al. (2003b;
33 2004) have suggested a mechanism of decarboxylation of amino acids under dry conditions, in which the Schiff base undergoes intramolecular cyclization by the action of the carboxylate anion to form a 5-oxazolidinone intermediate capable of losing CO2 and forming a relatively stable azomethine ylide (Figure 3.2 pathway b). Interestingly, this ylide can undergo 1,2-prototropic shift and generate two isomeric imines, providing a second mechanistic pathway of imine isomerization in addition to transamination. In this study, imine isomerization mechanisms are explored in phenylalanine/glyceraldehyde model system, and spectroscopic evidence is provided for the formation of 5- oxazolidinone intermediate and subsequent generation of azomethine ylide.
Figure 3.1 Chemical transformations of Schiff bases
34 3.2 Material and methods
All reagents and chemicals were purchased from Aldrich Chemical company
(Milwaukee, WI) and used without further purification. The labelled [13C-1] and [15N] phenylalanine were purchased from Cambridge Isotope Laboratories (Andover, MA).
3.2.1 Pyrolysis GC/MS analysis
Pyrolysis GC/MS (Py-GC/MS) analyses were conducted using a Varian CP-3800 GC coupled with a Varian Saturn 2000 ion trap mass spectrometry detector (Varian, Walnut
Creek, USA). The pyrolysis unit included a CDS 1500 valved interface and a CDS
Pyroprobe 2000 unit (CDS Analytical, Oxford, USA), was installed onto the GC injection port. The GC is equipped with a sample pre-concentration trap (SPT) filled with Tenax
GR. About one milligram of sample mixture was packed inside a quartz tubes (0.3mm thickness), plugged with quartz wool, and inserted inside the coil probe and pyrolyzed for
20 seconds at a temperature range from 175oC to 250oC. The volatiles after pyrolysis were concentrated on the sample pre-concentration trap (SPT) at 50oC for 4 minutes and subsequently desorbed at 100oC to the GC column for separation. A DB-5MS capillary column (J&W Scientific, 50m x 0.2mm i.d; coating thickness, 0.33μm) was used under the following conditions: a pressure pulse of 70 psi was set for the first 4 minutes and later maintained with a constant flow of 1.5mL/min for the rest of the run, regulated by an
Electronic Flow Controller (EFC). The GC oven temperature was set at –5oC for 5 minutes, using CO2 as the cryogenic cooling source. The temperature was increased to
50oC at a rate of 50oC/min and then to 180oC at a rate of 5oC/min. The oven temperature was further increased to 280oC at a rate of 20oC/min and kept for 9.5 min. MS data were collected using the electron impact ionization mode under the following conditions: MS
35 transfer line temperature, 250 °C; MS manifold temperature, 50°C; MS ion trap temperature, 175°C; ionization voltage: 70eV; electron multiplier voltage: 1700V; scan range, 20-650 m/z. Compound identification was performed using AMDIS (v2.62) and the National Institute of Standards and Technology (NIST) Standard Reference Database
(v05).3.2.2 Generation and FTIR monitoring of 5-oxazolidinone intermediate
An equimolar mixture (10mg) of sugar and amino acid was heated in toluene (methanol and/or p-toluene-sulfonic acid could be added to help dissolve insoluble models) for 10 min or until most reactants dissolved at 115oC in an open vial (2mL). The solution was filtered immediately and the sample was applied to the ATR crystal and scanned after evaporation of the solvent. Infrared spectra were recorded on a Nicolet single bounce
ATR. A total of 64 scans at 4 cm-1 resolution were co-added. Processing of the FTIR data was performed using GRAMS/32 AI (ThermoGalactic, Waltham, MA).
3.2.3 Browning measurement by UV/VIS
An equimolar solution of reactants (0.03M each) in dimethyl sulfoxide was stirred in the presence and absence of dimethyl fumarate (0.03M) at 80oC for 30min. The solution was cooled and browning was measured by scanning between 400-600 nm using Evolution
300 scanning spectrophotometer from Thermo Electron Corporation (Madison, WI).
3.3 Results and discussion
Imines formed between amino acids and reducing sugars are generally referred to as
Schiff bases (Figure 3.1). The fate of this initial Schiff base is mainly dependent on the moisture content of the system, the pH and the temperature. Under intermediate moisture conditions and pH between 5-7, Schiff bases are converted into Amadori products and
Maillard reaction cascade is initiated. Under lower moisture conditions, Schiff bases tend to prevail and subsequently undergo several transformations (Figure 3.1) depending on
36 the pH. Under basic pH and at room temperature, transamination (Wnorowski &
Yaylayan, 2003) can generate isomeric imines (Figure 3.2). Such isomeric imines can also be formed under pyrolytic conditions (Yaylayan & Wnorowski, 2002; Wnorowski &
Yaylayan, 2003). However, under slightly acidic or neutral pH, Schiff bases can undergo intramolecular cyclization and produce decarboxylated isomeric imines through 5- oxazolidinone formation, as shown in Figure 3.2. Other reactions of Schiff bases are summarized in Figure 3.1.
3.3.1 Isomerization of imines through transamination
Transamination reactions usually involve a base-catalyzed tautomerization of the imine and formation of a delocalized 2-azaallyl anion (Figure 3.2) (Cram & Guthrie, 1966).
This process has been recognized as a prototype of the biochemical transamination between amino acids and pyridoxal (Bishop et al., 1997). Recently, these intermediates have been generated under mild conditions at room temperature through deprotonation of imines using potassium tert-butoxide (Bishop et al., 1997). A concerted mechanism was proposed by Cram and Guthrie (1966) for this isomerization, in which the base removes a proton from one α carbon synchronously with the donation of a proton to the other carbon by its conjugate acid. In the specific case of pyruvic acid and ethanolamine, the initial imine formation and its subsequent isomerization through a 1,2-prototropic shift occurs very rapidly at room temperature, and the reaction goes to completion within 12 minutes as indicated by FTIR analysis (Wnorowski & Yaylayan, 2003). After hydrolysis, the two isomeric imines in this case, can generate alanine and glycolaldehyde, indicating the transfer of an amino group between a donor amino compound and an acceptor keto acid
(transamination).
37
Figure 3.2 Proposed mechanisms of imine isomerizations in the Maillard reaction (a) base- catalyzed transamination, (b) oxazolidinone pathway, and (c) decarboxylative transamination RT = room temperature.
38
3.3.2 Isomerization of imines through 5-oxazolidinone and azomethine ylide formation
The transamination pathway of imine isomerization described above applies to Schiff bases formed between any amino group and any carbonyl moiety. However, the second mechanism that can lead to the formation of isomeric imines is specific to Schiff bases formed between amino acids and any carbonyl-containing compounds and is accompanied by loss of a CO2 from the amino acid. Figure 3.2 depicts the formation of azomethine ylide (Cainelli et al., 1996) after the loss of CO2 from 5-oxazolidinone intermediate. In dry systems, the Schiff bases of reducing sugars and amino acids are prone to undergo intramolecular cyclization to form either 5-oxazolidinone or glycosylamines, in contrast to high-moisture systems, where they tend to undergo
Amadori rearrangement; the more stable isomer. However, the formation of 5- oxazolidinone and subsequent generation of azomethine ylides have so far been verified only in model systems consisting of amino acids and simple aldehydes (Tsuge et al.,
1987; Cainelli et al., 1996). Indirect evidence for the involvement of Schiff bases in assisting the decarboxylation process was obtained earlier (Yaylayan et al., 2004) when
[13C-4]-aspartic acid was pyrolyzed alone and in the presence of glucose. Analysis of the data showed that exclusive decarboxylation of aspartic acid form C-1 when pyrolyzed in the presence of glucose, indicating a preference for the formation of 5-oxazolidinone as an intermediate for the decarboxylation step. However, to provide direct evidence for the formation of 5-oxazolidinone, the amino acid/sugar reactions were analyzed by FTIR to monitor the formation of a peak in the range between 1780 to 1810 cm-1, where 5- oxazolidinones are known to exhibit a strong absorption band (Tsuge et al., 1987).
Spectroscopic studies using glyceraldehyde/amino acid model systems in toluene heated
39 at 110oC clearly indicated the formation of an intense peak in the range of 1780-1810 cm-1, depending on the amino acid. Figure 3.3 shows the carbonyl absorption peak centered at
1784 cm-1 for the glyceraldehyde/phenylalanine model system. The identity of the peak was verified by observing the expected 40 cm-1 shift when [13C-1]-labelled phenylalanine was used. Furthermore, evidence for the formation of azomethine ylide was also provided using their specific ability to undergo 1,3-dipolar cycloadditions with dipolarophiles
(Cainelli et al., 1996).
Figure 3.3 FTIR spectrum of phenylalanine/glyceraldehyde (solid lines) and [13C-] phenylalanine/glyceraldehyde (dotted line) acquired after heating for 10 min in toluene at 110oC
40
The addition of dipolarophiles, such as dimethyl fumarate, to the heated phenylalanine/glyceraldehyde model system has lead to a significant drop in intensity of the Maillard browning (Figure 3.4), indicating importance of the azomethine ylide as a browning precursor in the Maillard reaction. In addition to direct spectroscopic evidence for oxazolidinone formation, decarboxylation efficiencies can also provide indirect evidence for the involvement of Schiff bases in the decarboxylation process. In fact, the efficiency of decarboxylation of phenylalanine, as measured by the amount of phenethylamine produced per mole of the precursor during Py-GC/MS analysis, increased significantly in the presence of carbonyl containing compounds (Table 3.1). In addition, phenylalanine can undergo decarboxylation in the absence of carbonyls. Efficiency of this independent decarboxylation process is a function of temperature, as shown in Table 3.2.
The addition of carbonyl compounds not only increases the efficiency of decarboxylation, but also allows decarboxylation to occur at lower temperatures (results not shown).
Table 3.1 Effect of addition of carbonyl compounds on the decarboxylation efficiency1 of phenylalanine at 250oC Model System Relative efficiency2 (Phenethylamine formation) Phenylalanine 1 Phenylalanine + pyruvic acid 19 Phenylalanine + glyceraldehyde 99 Phenylalanine + 2,3-butanedione 115 Phenylalanine + phenylacetaldehyde 303 Phenylalanine + phenylpyruvic acid 300 1 estimated from the chromatographic peak area of phenethylamine produced /mole of phenylalanine 2 based on three replicates with coefficient of variation < 15%
41 Table 3.2 Effect of temperature on the decarboxylation efficiency1 of phenylalanine Temperature (oC) Relative efficiency2 (Phenethylamine formation) 175 0 200 1 250 191 1 estimated from the chromatographic peak area of phenethylamine produced /mole of phenylalanine 2 based on three replicates with coefficient of variation < 15%
Figure 3.4 UV-VIS spectrum of phenylalanine/glyceraldehyde solution (A) acquired after heating for 8 min in DMSO at 80oC and (B) after additional storage at room temperature for 6h.
42
3.3.3 Evidence of imine isomerizations in phenylalanine/glyceraldehyde model system
To investigate the occurrence of imine isomerization in Maillard model systems, phenylalanine/glyceraldehyde model was chosen because of the availability of the hydrolysis products of the four possible imines (1a, 1b, 2a & 2b in Figure 3.5) that could be formed in this model system. When the initial imine (1a) undergoes isomerization through transamination, the resulting imine (1b) can be obtained through the interaction of readily available 1-aminopropanediol and phenylpyruvic acid. Alternatively, if the imine 1a undergoes isomerization through oxazolidinone pathway, the resulting imines,
2a and 2b, could also be obtained through the interaction of readily available precursors such as glyceraldehyde, phenethylamine, 1-aminopropanediol and phenylacetaldehyde.
The formation of oxazolidinone in this particular system has already been confirmed, as described above, using the absorption band centered at 1785 cm-1. The four model systems generating imines 1a, 1b, 2a and 2b shown in Figure 3.5 were analyzed using
Py-GC/MS, and the formation of selected products was used to indicate the occurrence of specific isomerization. For example, detection of phenylacetaldehyde and phenethylamine in the pyrolysis products of glyceraldehyde/phenylalanine model system may indicate occurrence of isomerization between 2a and 2b, as shown in Figure 3.5. However, phenylacetaldehyde can also be formed through Strecker reaction, and phenethylamine can also be formed through independent decarboxylation of the amino acid without reactants necessarily passing through 5-oxazolidinone intermediates. To eliminate this possibility, the model systems were also studied at 175oC, a temperature at which independent decarboxylation was not observed (see Table 3.2). In addition, to identify a unique product indicating formation of imine 2b, the precursors of the two imines, 2a and
43
Figure 3.5 Isomerizations of the Schiff base formed between phenylalanine and glyceraldehyde through transamination and 5-oxazolidinone formation.
44
2b were studied, and the products formed in both model systems were compared. The analysis of the data indicated a significant formation of 3-phenylpyridine in the model system of 1-aminopropanediol/phenylacetaldehyde and its complete absence in the glyceraldehyde/phenethylamine system. Figure 3.6 shows a possible mechanism of formation of 3-phenylpyridine starting from 2b through two dehydration steps and an electrocyclic ring closure step. This mechanism is consistent with the inability of the 2a to form this compound. The identity of this peak was further verified by observing an identical retention time (25.05 min) with the commercially obtained 3-phenylpyridine; furthermore, the incorporation of one nitrogen atom in the structure when [15N]- phenylalanine was used is consistent with the proposed structure. Similarly, detection of
Figure 3.6. Proposed mechanism of formation of 3-phenylpyridine and its mass spectrum as compared with authentic NIST library spectrum in head-to-tail fashion
45
3-phenylpyridine and phenethylamine in the pyrolysis products of 1- aminopropanediol/phenylpyruvic acid model system (Figure 3.5) can also indicate the occurrence of isomerization, not only between 1b and 1a, but also between 2a and 2b.
Pyrolysis of glyceraldehyde/phenylalanine and 1-aminopropanediol/phenylpyruvic acid model systems indicated the formation of both indicator compounds, phenethylamine and
3-phenylpyridine, confirming the formation of 1b, 1a, 2a & 2b. In addition, these results indicate that transamination of 2a to 2b is difficult to achieve in imines lacking a carboxylic acid moiety in their structure that is able to catalyze prototropic shifts
(Yaylayan & Wnorowski, 2002; Wnorowski & Yaylayan, 2003), as in the case of isomerization of 1b to 1a, further supporting the intermediacy of azomethine ylide in the two model systems. Similar imine isomerizations were also observed by Granvogl et al
(2006) in phenylalanine dicarbonyl mixtures. They proposed a decarboxylative transamination (pathway c in Figure 3.2) mechanism similar to the suggestion of Bishop et al. (1997) in explaining the transamination reaction between amino acids and α-keto acids.
46 Connecting paragraph
In Chapter 3 it was demonstrated that during the early phase of the Maillard reaction, the
Schiff bases formed, can undergo in addition to Amadori rearrangement, intramolecular cyclization to form 5-oxazolidinone which can decarboxylate into unstable azomethine ylide capable of formation of two isomeric imines. In Chapter 4, FTIR was used in conjunction with [13C-1]-, [15N]-isotope labelled phenylalanine and glycoaldehyde to provide further evidence of the above proposed mechanism of imine-isomerization. In addition, time-dependent analysis of the FTIR spectra clearly illustrated the formation and decomposition of the 5-oxazolidinone intermediate. Chapter 4 was published in
Carbohydrate Research. (Chu, F.L. and Yaylayan, V.A., 2009. FTIR monitoring of 5- oxazolidinone formation and decomposition in a glycolaldehyde-phenylalanine model system by isotope labelling techniques. Carbohydrate Research. 344 (2), pp. 229-236.)
47 Chapter 4: FTIR monitoring of 5-oxazolidinone formation and decomposition in glycolaldehyde/phenylalanine model system by isotope labelling techniques
Amadori
faster
OH O R O OH R O O R cyclization H N N + H H OH ONH2 H OH CH3
5-oxazolidinone Schiff base
CO2
Isomeric Imines
48 4.1. Introduction
Imines formed subsequent to carbonyl-amine reactions between amino acids and reducing sugars tend to undergo various chemical transformations depending on the nature of the carbonyl moiety. They can undergo Amadori rearrangement, the Strecker reaction, or cyclizations to generate N-containing heterocyclic compounds. Prior to these transformations, imines are also susceptible to isomerization reactions, further increasing the diversity of Maillard reaction products. Such isomerization reactions can proceed either through 5-oxazolidinone formation in dry systems to generate decarboxylated imine isomers or through transamination reactions (Figure 4.1). The importance of 5- oxazolidinone formation lies in its ability to decarboxylate and form a non-stabilized azomethine ylide. This type of N-protonated azomethine ylide is prone to undergo 1,2- prototropic shifts and form two isomeric imines (B & C in Figure 4.1), each is capable of producing distinct Maillard products. The formation of 5-oxazolidinone and subsequent generation of azomethine ylides have so far been verified only in model systems consisting of amino acids and simple aldehydes (Tsuge et al., 1987; Aurelio et al., 2003).
The chemistry of imine isomerization reactions through 5-oxazolidinone formation in amino acid carbohydrate mixtures was first explored by Chu and Yaylayan (2008b) using a dry phenylalanine/glyceraldehyde model system. In this study, spectroscopic evidence was provided for the formation of 5-oxazolidinone intermediate by the strong carbonyl absorption band centered at 1784 cm-1. Spectroscopic studies using glyceraldehydes and various amino acids in toluene heated at 110oC clearly indicated the formation of an intense peak in the range of 1780-1810 cm-1, depending on the amino acid. The glyceraldehyde/asparagine model system (Perez Locas & Yaylayan, 2008), for example, exhibited an absorption peak centered at 1778 cm-1. The identity of the peaks was verified
49 by observing the expected 40 cm-1 shift when [13C-1]-labelled amino acids were used.
Furthermore, evidence for the formation of the resulting azomethine ylide was also provided (Chu & Yaylayan, 2008b) using their specific ability to undergo 1,3-dipolar cycloadditions with dipolarophiles (Tsuge et al., 1987). The addition of dipolarophiles, such as dimethyl fumarate to the heated model systems has lead to a significant drop in intensity of the Maillard browning indicating the importance of the resulting imines to the generation of colour. In previous studies we have examined imine isomerization reactions as a consequence of 5-oxazolidinone formation in phenylalanine/glyceraldehyde model system using Py-GC/MS (Chu & Yaylayan, 2008b) and acrylamide generation through 5- oxazolidinone formation in asparagine/glucose model system (Perez Locas & Yaylayan,
2008). In this paper, we provide in-depth spectroscopic analysis of oxazolidinone formation/decomposition in the phenylalanine/glycolaldehyde model system.
4.2 Experimental
All reagents and chemicals were purchased from Aldrich Chemical company (Milwaukee,
WI) and were used without further purification. The labelled [13C-1]-phenylalanine and
[15N]-phenylalanine were purchased from Cambridge Isotope Laboratories (Andover, MA).
4.2.1 Extraction of 5-oxazolidinone “toluene extract” and FTIR analysis
An equimolar mixture of glycolaldehyde (10 mg) and phenylalanine (6 mg) was heated in toluene (200 μL) for 1 min (the solution turns light yellow with the formation of a brown residue) at 110oC in an open vial (1 mL). Sample (1 μL) of the supernatant solution above the residue were repeatedly applied and evaporated onto an ATR crystal and immediately scanned at specified temperatures. FTIR spectra were recorded on a Bruker Alpha-P
FTIR spectrometer (Bruker Optic GmbH, Ettlingen, Germany) equipped with a deuterated triglycine sulphate (DTGS) detector, a temperature controlled single bounce
50 diamond attenuated total reflectance (ATR) crystal, and a pressure application device for solid samples. At a fixed temperature, the spectra were acquired every 60 s for 20 min or
40 min. A total of 32 scans at 4 cm-1 resolution were co-added. Processing of the FTIR data was performed using Bruker OPUS software.
4.2.2 FTIR monitoring of 5-oxazolidinone and imine formation
Glycolaldehyde dimer (10 mg) was heated in methanol (100 μL) for 5 min, and the solvent was evaporated to yield a colourless melt. The melt was applied onto the ATR crystal, and L-phenylalanine powder (6 mg) or phenethylamine or phenethylamine·HCl was spread over the melt. The infrared spectra were immediately recorded at indicated temperatures every 60 s over a 20 min period on a Bruker Alpha-P spectrometer (Bruker
Optic GmbH, Ettlingen, Germany) described above. Similarly, phenylacetaldehyde was applied onto the ATR crystal, followed by ethanolamine or ethanolamine·HCl, and immediately the infrared spectra were recorded at indicated temperatures.
4.3 Results and discussion
The initiation of the Maillard reaction is mainly attributed to the successful formation of a
Schiff base between the amino acid and the reducing sugar and its subsequent rearrangement into an Amadori product under appropriate moisture and pH conditions.
However, in low moisture systems, Schiff bases tend to prevail and subsequently undergo several transformations in addition to Amadori rearrangement. Under basic pH and at room temperature, transamination can generate its isomeric imine A shown in Figure 4.1
(Wnorowski & Yaylayan, 2003). Such isomeric imines can also be formed under pyrolytic conditions (Wnorowski & Yaylayan, 2003; Aurelio et al., 2003). However, under dry and slightly acidic or neutral pH, they can undergo
51
Figure 4.1 Summary of chemical transformations of the Schiff base formed between an amino acid and glycolaldehyde.
52
decarboxylation through 5-oxazolidinone intermediate to produce decarboxylated isomeric imines (B and C in Figure 4.1). The formation of an 5-oxazolidinone intermediate from Schiff bases, therefore, can provide an alternate pathway of degradation of sugars in addition to the Amadori pathway.
4.3.1 FTIR spectroscopy and isotope labelling studies
The 5-oxazolidinone intermediate can be conveniently studied using FTIR spectroscopy
(Tsuge et al., 1987; Aurelio et al., 2003) due to the strong and characteristic absorption band ~1800 cm-1. In previous studies, FTIR spectroscopy was employed to indicate 5- oxazolidinone formation in phenylalanine/glyceraldehyde (Chu & Yaylayan, 2008b) and in asparagine/glyceraldehyde model systems (Perez Locas & Yaylayan, 2008). Two sampling methods were developed to enhance the sensitivity of the detection. One method allows extraction of 5-oxazolidinone from the heated reaction mixtures using toluene or methanol as solvents. The second method allows for monitoring the formation/degradation of 5-oxazolidinone in excess molten glycolaldehyde used as solvent (see Experimental section).
4.3.2 Infrared band assignments
In the phenylalanine/glycolaldehyde model system, the band assignments of the different intermediates incorporating the carboxylic acid moiety depicted in Figure 4.2 were accomplished through the observation of a shift in the frequency of the absorption bands arising from C-1 of carboxylic acid moiety, when phenylalanine was replaced with [13C-
1]-phenylalanine. The specific bands of the imines 2a and 2b and their further reaction products were assigned through the use of glycolaldehyde/phenethylamine and ethanolamine/phenylacetaldehyde reaction systems respectively. Figure 4.3a shows the
FTIR spectra of the toluene extracts of [13C-1]-phenylalanine/glycolaldehyde and
53
Figure 4.2 Band assignments of the intermediate compounds formed in the reaction between glycolaldehyde and phenylalanine.
54
phenylalanine/glycolaldehyde systems acquired at 35oC. Figure 4.3a indicates a 42 cm-1 shift for the band centered at 1778 cm-1 and ~ 58 cm-1 shift for the band centered at 1628 cm-1. These bands, because of their shifts when [13C-1]-phenylalanine was used must be due to the C-1 atom of phenylalanine. The 1778 cm-1 band can be assigned to the oxazlidin-5-one (Tsuge et al., 1987; Aurelio et al., 2003; Perez Locas & Yaylayan, 2008;
Chu & Yaylayan, 2008b) and the band centered at 1628 cm-1 can be assigned to carboxylate moiety. It is interesting to note the absence of the free carboxylic acid absorption band that appears at 1732 cm-1 in the spectrum of phenylalanine hydrochloride salt and was confirmed by its shift to 1690 cm-1 when [13C-1]-phenylalanine·HCl was used (see Figure 4.4). However, when the temperature of the freshly prepared toluene extract was increased from 35oC to 80oC (see Figure 4.5) and monitored over a 10 min period, the oxzaolidin-5-one band centered at 1778 cm-1 sharply decreased, and a new band appeared and increased over time (see Figure 4.5). The new band was composed of two peaks, one centered around 1730 and the other around 1720 cm-1 (see Figure 4.5).
The 1730 cm-1 band was assigned to the carboxylic acid of the Amadori rearrangement product as confirmed by its shift to 1685 cm-1 when [13C-1]-phenylalanine was used (see
Figures 4.3b). The Amadori product should also exhibit a carbonyl absorption band that was logically assigned to the band at 1720 cm-1. Both peaks appeared and increased simultaneously with temperature. Contrary to the peak at 1730 cm-1, the peak at 1720 cm-
1 did not shift when [13C-1]-phenylalanine was used as shown in Figure 4.3, and it was confirmed by a second-derivative spectrum (Figure 4.6).
55
Figure 4.3 Absorption of the carbonyl and the imine regions (1900-1500 cm-1) of the toluene extracts of phenylalanine/glycolaldehyde (solid lines) and [13C-1]- phenylalanine/glycolaldehyde (dashed line) reaction mixtures acquired at (A) 35oC and (B) 80oC. An * indicates a 13C-1 atom.
56 Figure 4.4 Infrared spectra (1900 – 1450 cm-1 region) of phenylalanine·HCl (solid line) and [13C-1] phenylalanine·HCl (dashed line)
Figure 4.5 Absorption of the carbonyl and the imine regions (1800-1540 cm-1) of the toluene extracts of phenylalanine/glycolaldehyde reaction mixtures acquired at 80oC over a period of 10 min showing the intensity of the band at 1778 cm-1 acquired at 35oC as a reference.
57
Furthermore, to assign the bands of the two imines formed after decarboxylation of the 5- oxazolidinone intermediate, the glycolaldehyde/phenethylamine and glycolaldehyde/phenethylamine·HCl were analyzed by FTIR to assign the absorption band of imine 2a in Figure 4.2. Figure 4.7 shows the formation of a band centered around 1640 cm-1 when phenethylamine was used, as well as the appearance of two additional bands at 1666 and 1702 cm-1 when phenethylamine was replaced with phenethylamine·HCl. Under the strong basic conditions of phenethylamine, Amadori rearrangement was not encouraged and only initial imine was formed at 1640 cm-1 as shown in Figure 4.7. However, under the acidic conditions of phenethylamine·HCl, it is expected that the imine will rearrange into the Amadori product, Figure 4.7 shows the appearance of two bands, one at 1666 and the other at 1702 cm-1. The former was tentatively assigned as an enaminol band, and the latter as the aldehyde of the Amadori compound. Similar experiments were performed with ethanolamine/phenylacetaldehyde and ethanolamine·HCl/phenylacetaldehyde to identify the absorption band of imine 2b in
Figure 4.2. Figure 4.8 indicates the formation of a band centered at 1637 cm-1 with a shoulder at 1624 cm-1. The peak at 1637 cm-1 completely collapsed into 1624 cm-1 band when ethanolamine·HCl was used, indicating the acid catalyzed isomerization of the imine 2b into a conjugated system as shown in Figure 4.8. The results of the above band assignments studies are summarized in Figure 4.2.
58
Figure 4.6 Second-derivative spectra (Savitsky Golay polynomial 2, points 15) of toluene extracts of phenylalanine/glycolaldehyde, [13C-1]-phenylalanine/glycolaldehyde, and [15N]-phenylalanine/glycolaldehyde reaction mixtures heated at 80oC for 10 min.
Figure 4.7 Absorption of the carbonyl and the imine regions (1800-1550 cm-1) of the phenethylamine/glycolaldehyde (dashed line) and phenethylamine·HCl /glycolaldehyde (solid line) reaction mixtures heated at 70oC for 2 min.
59
Figure 4.8 Absorption of the carbonyl and the imine regions (1800-1550 cm-1) of the ethanolamine/phenylacetaldehyde (dash-dot) and ethanolamine·HCl/phenylacetaldehyde (solid line) reaction mixtures heated at 35oC for 2 min.
4.3.3 FTIR monitoring of 5-oxazolidinone formation and decomposition in phenylalanine/glycolaldehyde model system
The FTIR spectrum (see Figures 4.3a) of the toluene extract of a heated (1 min at 110oC) mixture of phenylalanine/glycolaldehyde indicated the presence of mainly 5- oxazolidinone intermediate in this extract. Figure 4.3a also shows the absorption band of carboxylate anion at 1628 cm-1 of the Schiff base 1a in Figure 4.2, indicating the formation of 5-oxazolidinone from precursor 1a. After 24 h of storage at room temperature, the toluene extract did not show any decomposition as opposed to the methanol extract (prepared in a similar fashion at 65oC) that showed complete disappearance of the 5-oxazolidinone band and formation of a dark-brown solution.
Monitoring of the toluene extract on the ATR crystal at 80oC indicated a fast
60 decomposition over a period of 10 min and its conversion into Amadori product as indicated by the concomitant appearance of carboxylic acid band centered at 1730 cm-1 and a carbonyl band centered at 1720 cm-1 (see Figure 4.5). Furthermore, examination of the second-derivative spectra of the toluene extracts generated from phenylalanine/glycolaldehyde, [13C-1]-phenylalanine/glycolaldehyde, and [15N]- phenylalanine/glycolaldehyde heated at 80oC for 10 min (see Figure 4.6) indicated the formation of imines (1620-1650 cm-1 region). Although the intensities of theses bands were weak, there is clear indication for the formation of imines 2a and 2b that absorb very close to each other (1640 vs. 1637 cm-1) and appear centered around 1640 cm-1 in the phenylalanine and [13C-1]-phenylalanine model systems, but, as expected, shifted to
1630 cm-1 in [15N]-phenylalanine system (Figure 4.6). The presence of a peak at 1624 cm-1 further confirms the formation of imine 2b (see Figure 4.8). There was no evidence for the presence of a peak at 1702 cm-1 (decarboxylated Amadori product), which was expected from the results of the model studies with glycolaldehyde/phenethylamine·HCl.
However, considering the fact that 1702 cm-1 peak in this model system quickly degraded when the sample was monitored over a 10 min period at 70oC, it is, therefore, not expected to be detected at 80oC. To monitor formation and decomposition of 5- oxazolidinone intermediate, excess glycolaldehyde was melted by heating in methanol and evaporating the solvent. The melt was applied onto the ATR crystal, followed by addition of powdered phenylalanine on the surface of the melt (see Experimental section). Even at 35oC, the rate of 5-oxazolidinone formation was very fast. By the time the first spectrum was acquired, the peak at 1778 cm-1 was already formed (see Figure
4.9). Within 15 min of the acquisition of the first spectrum, the intensity of the peak started to decrease, and it completely disappeared in the next 40 min with the
61 concomitant formation and increase of Amadori product as indicated by the appearance of carboxylic acid and aldehyde peaks at 1730 and 1720 cm-1, respectively (see Figure
4.10). Attempts to isolate the Amadori product from the mixture were not successful due to the instability of the product.
62
Figure 4.9 Time-dependent spectra of glycolaldehyde/phenylalanine mixture heated at 35oC for 55 min showing the initial 15 min.
Figure 4.10 Time-dependent spectra of glycolaldehyde/phenylalanine mixture heated at 35oC for 55 min showing the final 40 min.
63
4.3.4 Implications of formation of an 5-oxazolidinone intermediate in the mechanism of the Amadori rearrangement
The fact that the Amadori product appeared and increased in intensity only after the collapse and decrease of 5-oxazolidinone intermediate (see Figures 4.3, 4.5 and 4.10) indicates that this intermediate also undergoes Amadori rearrangement similar to the
Schiff base. Consequently, ring opening initiated by the nitrogen atom leads to decarboxylation of the 5-oxazolidinone and formation of the azomethine ylide, and ring opening through the proton shift of C-2 hydrogen atom of the sugar moiety leads to the formation of enaminol 1b, followed by the Amadori rearrangement as indicated in Figure
4.2. This observation may explain why the formation of the Amadori rearrangement product of phenethylamine with glycolaldehyde required a temperature of 70oC (see
Figure 4.7), whereas Amadori product formation with phenylalanine occurred even at
35oC (see Figure 4.9). Although it can be argued that the 5-oxazolidinone intermediate can revert to 1a and then undergo Amadori rearrangement, if this were the case, the peaks associated with the Amadori product would have appeared before the appearance of 5- oxazolidinone peak in Figure 4.9. In conclusion, the detection of an 5-oxazolidinone intermediate as an immediate precursor of the Amadori rearrangement product not only contributes to our detailed understanding of this important reaction, but also can explain why amino acids are more reactive than their amine counterparts in undergoing Amadori rearrangement process.
64 Connecting Paragraph
In Chapters 3 and 4, isotope labelling, pyrolysis-GC/MS and FTIR based techniques were used as powerful analytical tools for the elucidation of the detailed mechanism of the early phase of the Maillard reaction. The role of oxygen is another aspect of this reaction which remains to be elucidated in detail. In this chapter, oxidative and wet pyrolysis-
GC/MS based methodologies were used to study the effect of oxygen and water on the product distribution during the Maillard reaction. This study has revealed a distinct oxidation mechanism that can convert phenylacetaldehyde into benzaldehyde through interaction with molecular oxygen. Chapter 5 was published in Journal of Agriculture and
Food Chemistry. (Chu, F.L. and Yaylayan, V.A. 2008a. Model Studies on the Oxygen-
Induced Formation of Benzaldehyde from Phenylacetaldehyde Using Pyrolysis GC-MS and FTIR. Journal of Agricultural and Food Chemistry 56(22), pp.10697-10704)
65 Chapter 5: Model Studies on the Oxygen Induced Formation of Benzaldehyde from Phenylacetaldehyde using Pyrolysis GC-MS and FTIR
66 5.1 Introduction
Benzaldehyde is produced naturally in different fruits such as bitter almond, cherry, apricot and peaches through the action of β-glucosidases on cyanogenic glucosides followed by hydroxynitrile lyase (Chandra & Nair, 1993). Enzymatic transformation of phenylalanine into benzaldehyde using cultures of microorganisms, moulds and enzymes extracts are also reported. Porter and Bright (1987) used L-amino acid oxidase in the presence of horseradish peroxidase as an in vitro bi-enzymatic system to generate benzaldehyde from L-phenylalanine. Okrasa et al. (2004) used D-amino acid oxidase from
Trigonopsis variabilis and peroxidase from Coprinus cinereus to convert phenylalanine into benzaldehyde and proposed a mechanism starting from phenylalanine via the formation of phenylpyruvic acid and phenylacetaldehyde. Benzaldehyde has also been detected in numerous phenylalanine containing model systems where it can be produced thermally (Severin & Braütigam, 1970; Zamora et al., 2006; Granvogl et al., 2006). Due to its strong sensory properties, its generation under Maillard reaction conditions in food products may impart off flavour notes (Kato, 2005), especially in combination with phenylacetaldehyde, the Strecker aldehyde of phenylalanine. Recently, phenylacetaldehyde has been shown to be one of the off-aroma notes generated in aged beer (Bravo et al., 2008). On the other hand, Brinkman et al. (1972) identified benzaldehyde in the headspace of beef broth as one of the volatiles contributing to the characteristic aroma of beef. Severin and Braütigam (1970) reported formation of benzaldehyde through heating of phenylalanine, N-acetylphenylethylamine or N-glycyl- phenylalanine alone or with triglycerides between 190-240oC in the presence of air. The yield varied between 0.1 to 3.1%. Granvogl et al. (2006) observed the formation of benzaldehyde from a heated mixture of phenethylamine and pyruvaldehyde and proposed
67 a multistep mechanism based on imine isomerization, oxidation and water addition steps.
In the same study the authors made an interesting observation for the simultaneous generation of Strecker aldehyde and Strecker amine (decarboxylated amino acid) under
Strecker reaction conditions. Furthermore, Zamora et al. (2006) detected benzaldehyde in a model system heated at 60oC containing phenylalanine and 4,5-epoxy-2-alkenal. They attributed the formation of benzaldehyde to the generation of phenylpyruvic acid through transamination of phenylalanine in the reaction mixture. Due to the lack of systematic study on the thermal generation of benzaldehyde from phenylalanine, we provide evidence for its mechanism of formation using oxidative pyrolysis-GC/MS (Yaylayan et al., 2005) and FTIR spectroscopy.
5.2 Material and Method
All reagents and chemicals were purchased from Aldrich Chemical Company
(Milwaukee, WI) and used without further purification.
5.2.1 Pyrolysis GC/MS analysis
Py-GC/MS analyses were conducted using a Varian CP-3800 GC coupled with a Saturn
2000 ion trap mass spectrometry detector (Varian, Walnut Creek, USA). The pyrolysis unit included a CDS 1500 valved interface, and a CDS Pyroprobe 2000 unit (CDS
Analytical, Oxford, USA) was installed onto the GC injection port. The GC was equipped with a sample pre-concentration trap (SPT) filled with Tenax GR. About one milligram of sample mixture was packed inside a quartz tube (0.3mm thickness), plugged with quartz wool, and inserted inside the coil probe and pyrolyzed for 20 seconds at a temperature range from 175oC to 250oC. Liquid samples were mixed with silica gel in 1:10 ratio by weight. For the samples that were pyrolyzed in the presence of moisture, 3 μL of distilled water was added to the mix. The volatiles after pyrolysis were concentrated on the SPT at
68 50oC for 4 minutes and subsequently desorbed at 100oC to the GC column for separation.
A DB-5MS capillary column (J&W Scientific, 50m x 0.2mm i.d; coating thickness,
0.33μm) was used under the following conditions: a pressure pulse of 70 psi was set for the first 4 minutes and later maintained with a constant flow of 1.5mL/min for the rest of the run regulated by an Electronic Flow Controller (EFC). The GC oven temperature was
o set at –5 C for 5 minutes using CO2 as the cryogenic cooling source. Two temperature programs were used: (A) The temperature was increased to 100oC at a rate of 50oC/min and then to 180oC at a rate of 5oC/min. The oven temperature was further increased to
280oC at a rate of 20oC/min and kept for 9.5 min (benzaldehyde elutes at 12.5 min). (B)
The temperature was increased to 50oC at a rate of 50oC/min and then to 280oC at a rate of 8oC/min and kept for 5 min (benzaldehyde elutes at 15.5 min). MS data were collected using electron impact ionization mode under following conditions: MS transfer line temperature, 250°C; MS manifold temperature, 50°C; MS ion trap temperature, 175°C; ionization voltage, 70eV and EMV, 1750V; scan range, 20-650m/z. Compound identification was performed using AMDIS (v2.62) and NIST Standard Reference
Database (v05).
5.2.2 Oxidative or wet Py-GC/MS analysis
Pyrolysis under air or in the presence of moisture was achieved through modification of the above mentioned GC to allow gas stream switching and subsequent isolation of the pyrolysis chamber from the analytical stream. The pyrolysates generated under air or in the presence of moisture were initially collected onto the trap, which retained the organic volatiles and vented the carrier gas (air) and/or moisture. The trap was subsequently
69 flushed with helium and heated to desorb the collected volatiles. For details see Yaylayan et al. (2005).
5.2.3 FTIR monitoring of oxidation reactions
Pure samples such as benzaldehyde or phenylacetaldehyde were directly applied onto the
ATR crystal and immediately scanned at either 35 or 100oC. Infrared spectra were recorded on a Bruker Alpha-P spectrometer (Bruker Optic GmbH, Ettlingen, Germany) equipped with a deuterated triglycine sulphate (DTGS) detector, a temperature controlled single bounce diamond attenuated total reflectance (ATR) crystal and a pressure application device for solid samples. A phenylacetaldehyde oxidation mixture was prepared by dissolving 110 mg in two mL of acetonitrile, 25μL of water and 60mg of
1,1’-azobis-(cyclohexanecarbonitrile) 98% as a free radical initiator. After all the components had dissolved, 50 μL was applied onto the ATR crystal and scanned at 100oC for 1h after evaporation of the solvent. The spectra were acquired at specified temperature every 120 s for 60 min. A total of 32 scans at 4 cm-1 resolution were co-added. Processing of the FTIR data was performed using the Bruker OPUS software.
5.3 Results and discussion
Different precursors have been proposed in the literature as possible source of benzaldehyde in phenylalanine containing model systems. However, no systematic study has been performed on the mechanism of thermal generation of benzaldehyde. To assess the potential of benzaldehyde generation from phenylalanine and to identify the role of phenylalanine degradation products such as Strecker aldehyde and Strecker amine, various model systems (see Tables 5.1-5.3) were reacted under oxidative and non- oxidative conditions and in the presence and absence of moisture using oxidative Py-
GC/MS analysis, and the formation of benzaldehyde and its precursors
70
Table 5.1 Occurrence1of benzaldehyde and its precursors in different model systems Model system Phenyl Phenethyl Benzaldehyde Benzaldehyde acetaldehyde amine -imine adduct2 Phenylalanine, 200oC 0 0 0 + Phenylalanine, 200oC, Air 0 0 ++ + Phenylalanine, 250oC, He 0 +++ 0 + Phenylalanine, 250oC, Air 0 ++++ 0 ++
Phenylacetaldehyde, 175oC +++ 0 + 0 Phenylacetaldehyde, 175oC, Water +++ 0 + 0 Phenylacetaldehyde, 175oC, Air +++ 0 ++ 0 Phenylacetaldehyde, 175oC, Air and Water +++ 0 +++ 0 Phenylacetaldehyde, 250oC, He +++ 0 ++ + Phenylacetaldehyde, 250oC, Air +++ 0 +++ + Phenylacetaldehyde, 250oC, Air and Water +++ 0 +++ +
Glycolaldehyde/Phenylalanine (1:1), 175oC +++ 0 + 0 Glycolaldehyde/Phenylalanine (1:1), 250oC 0 ++++ 0 + Glycolaldehyde/Phenylalanine (2:1), 175oC +++ 0 + 0 Glycolaldehyde/Phenylalanine (2:1), 250oC +++ 0 + 0
Glycolaldehyde/Phenylacetaldehyde, 175oC +++ 0 + 0 Glycolaldehyde/Phenylacetaldehyde, 175oC, Air and Water +++ 0 +++ 0 Glycolaldehyde/Phenylacetaldehyde (1:1), 250oC ++++ 0 + 0
Glyoxal trimer/Phenylalanine (0.33-1), 175oC + 0 0 0 Glyoxal trimer/Phenylalanine (1:1), 175oC + 0 0 0
Glyceraldehyde/Phenylalanine (1:1), 200oC + 0 + 0 Glyceraldehyde/Phenylalanine (1:1), 250oC 0 ++++ 0 + Glyceraldehyde/Phenylalanine (2:1), 250oC +++ 0 + 0 1 based on chromatographic peak area per mole of phenylalanine. Values represent average of three replicates with percent standard deviation < 15% 2 imine adduct with phenethylamine 71
Figure 5.1 Pyrograms generated at 250oC using temperature program B from (a) phenylalanine/glucose (1:1) under non-oxidative (b) phenylalanine/glucose (2:1) under non-oxidative (c) phenylalanine/glucose (1:1) under oxidative (d) phenylalanine/glucose (2:1) under oxidative conditions. 1 = benzaldehyde; 2 = phenylacetaldehyde; 3 = phenethylamine
72
phenylacetaldehyde and phenethylamine were monitored. The data indicated that in sugar/phenylalanine model systems the reaction generated either phenylacetaldehyde or phenethylamine depending on the ratio of the starting sugar to phenylalanine (see Figure
5.1, Tables 5.1 and 5.2) and that benzaldehyde was present in all systems where phenylacetaldehyde was generated. When the sugar to phenylalanine ratio was ≥2:1, only phenylacetaldehyde was detected and not phenethylamine (except in the case of glucose).
When the sugar to phenylalanine ratio was ≤1:1, only phenethylamine was detected
(except in the case of glucose). Interestingly, phenylalanine alone also generated benzaldehyde under oxidative pyrolysis. It seems that excess starting aldehyde/sugar prevents trapping of phenylacetaldehyde as an imine adduct by providing other carbonyl functional groups to scavenge the reactive amines including the phenethylamine, thus allowing free phenylacetaldehyde to be detected. The ability of glucose to degrade into molar excess of smaller aldehydes and ketones can explain the above exception observed with glucose. On the other hand, phenethylamine was detected only under non-oxidative conditions when phenylalanine to sugar ratio was ≥2:1. Under these conditions neither phenylacetaldehyde nor benzaldehyde was observed. However, under oxidative condition both phenylacetaldehyde and benzaldehyde were detected whether the phenylalanine to glucose ratio was 1:1 or ≥2:1. Furthermore, benzaldehyde concentrations were much higher in the presence of air, which indicated oxygen was involved in the reaction (see
Tables 5.2 and 5.3). According to Table 5.2, in glucose/phenylalanine (1:1) model system there was a 23-fold increase in benzaldehyde formation in the presence of air. A similar increase was noted in the model system where glucose to phenylalanine ratio was
1:2; however, in this case benzaldehyde was trapped as phenethylamine adduct.
73 Table 5.2 Relative amounts1 of phenylacetaldehyde, phenethylamine and benzaldehyde formed in glucose/phenylalanine- model systems at 250oC. Model system2 Phenyl- Phenethyl- Benzaldehyde Benzaldehyde acetaldehyde amine -imine adduct2 Glucose/ Phenylalanine (1:1) 1 0 1 0 Glucose/ Phenylalanine (1:2) 0 45x 0 1 Glucose/ Phenylalanine (1:1), 1.6x 0 23x 0 Air Glucose/ Phenylalanine (1:2), 0 1 5x 38x Air 1 based on chromatographic peak area per mole of phenylalanine normalized relative to the lowest value in each column. Values represent average of three replicates with percent standard deviation < 15%, 2 Imine adduct with phenethylamine
Table 5.3. Amounts1 of benzaldehyde formed from different precursors between 175- 250oC normalized relative to phenylethanediol system Model system He Air Air/Water Phenylacetaldehyde, 175oC 122 223 5363 Phenylacetaldehyde, 250oC 512 3683 6593 Phenylethanediol, 175oC 1 2 1.3 Phenylethanediol, 250oC 32 nd 4 nd Phenethylamine/Pyruvaldehyde (1:1), 250oC 0 nd 67 Phenethylamine/Pyruvaldehyde (1:0.5), 250oC 0 nd 50 Phenylpyruvic acid, 175oC 245 143 210 Phenylglyoxal, 250oC 20 48 127 Phenylalanine, 200oC 0 2 nd Phenylalanine/Glucose5 (1:1), 250oC 2 64 nd 1 based on chromatographic peak area per mole of precursor. Values represent average of two to three replicates with percent standard deviation < 25%. 2 pre-formed in the bottle during storage due to oxidation 3 corrected for residual benzaldehyde in phenylacetaldehyde by subtracting area/mole value of He experiments before normalization 4 nd = not determined 5 based on area per mole of phenylalanine or glucose
74
Table 5.3 also indicates significant increases in benzaldehyde content when different precursors were pyrolyzed under air. In addition, water seemed to play an important role in enhancing benzaldehyde formation from phenylacetaldehyde. For example, at 175oC under wet oxidative pyrolysis, phenylacetaldehyde produced almost 24-fold excess benzaldehyde relative to dry oxidative pyrolysis. Figures 5.2, panels a and b, demonstrate the role of wet and dry oxidative pyrolysis of phenylacetaldehyde in the generation of benzaldehyde relative to dry non-oxidative pyrolysis (Figure 5.2c).
Detection of small amount of benzaldehyde in phenylacetaldehyde during non-oxidative pyrolysis indicates its facile oxidation into benzaldehyde during storage. Inspection of
Table 5.3 also reveals that there is more than one precursor of benzaldehyde and that phenylacetaldehyde is the most efficient precursor followed by phenylpyruvic acid.
Phenylalanine, phenethylamine, phenylglyoxal and phenylethanediol all can generate benzaldehyde, and all require oxidative conditions except phenylpyruvic acid, which does not require air, but the presence of water enhances its formation, indicating the possible existence of oxidative and non-oxidative pathways of benzaldehyde generation depending on the precursor. According to Table 5.3, phenylalanine and phenylethanediol have comparable abilities to generate benzaldehyde between 175 and 200oC under oxidative conditions. The phenylalanine/glucose model system, on the other hand, showed comparable ability to that of phenethylamine/pyruvaldehyde system. Table 5.4 summarizes the ability of phenylacetaldehyde to generate benzaldehyde relative to other precursors. According to this table phenylacetaldehyde has 2-3 fold higher ability to generate benzaldehyde relative to phenylpyruvic acid
75
Figure 5.2 Pyrogram of phenylacetaldehyde (1mg) generated at 175oC under (a) wet/oxidative (b) dry/oxidative and (c) dry/non-oxidative conditions, (d) pyrogram generated through oxidative pyrolysis of 1-phenyl-1,2-ethanediol at 175oC. Percentages represent areas of peak 1 (benzaldehyde) relative to peak 2 (phenylacetaldehyde). All above pyrograms generated using temperature program A.
76
Table 5.4 Ability1 of phenylacetaldehyde to generate benzaldehyde relative to the listed precursors under optimum conditions Model system He Air Air/Water Phenylethanediol, 175oC 268 Phenylethanediol, 250oC 20 Phenethylamine/Pyruvaldehyde, (1:1), 250oC 9.8 Phenylpyruvic acid, 175oC 2.1 3.7 2.5 Phenylglyoxal, 250oC 33 13.7 5.2 Phenylalanine/Glucose, (1:1), 250oC 330 10.2 1 Relative to air/water values of phenylacetaldehyde reported in Table 5.3
According to Figure 5.3 phenylalanine can generate phenylacetaldehyde through three pathways; oxidative decarboxylation (Yaylayan & Keyhani, 2001b) followed by hydrolysis (pathway A) or thermal decarboxylation (pathway B) or Strecker reaction
(pathway C). In addition, 1-phenyl-1,2-ethanediol, phenylglyoxal, and phenylglyoxylic acid theoretically can be generated from the oxidation and/or hydration of phenylacetaldehyde as proposed in Figure 5.3, and they are found to be also possible precursors of benzaldehyde as mentioned above (see Table 5.3). Furthermore, as reported by Granvogl et al. (2006), formation of benzaldehyde was also observed in the mixtures of phenethylamine and pyruvaldehyde, which was corroborated in this study (see Table
5.3). According to Figure 5.4, phenethylamine could be formed from phenylalanine either through thermal decarboxylation or through 5-oxazolidinone formation in the presence of sugar or carbonyl compounds (Chu & Yaylayan, 2008b). In addition, transamination of phenylalanine can also generate another important precursor of benzaldehyde the phenylpyruvic acid (Chu & Yaylayan, 2008b).
77
Figure 5.3 Proposed mechanisms of formation of benzaldehyde from phenylalanine. Pathway A = Oxidative decarboxylation; pathway B = thermal decarboxylation, pathway C = Strecker or imine isomerization pathway
78
Figure 5.4 Proposed formation pathways of benzaldehyde precursors from phenylalanine
79
5.3.1 Proposed mechanism of benzaldehyde formation from phenylacetaldehyde
Although phenylalanine can generate different precursors that are able to form benzaldehyde, phenylacetaldehyde, the Strecker aldehyde, is the most efficient and established precursor. In converting phenylacetaldehyde into benzaldehyde, an oxygen atom should be introduced at the benzylic carbon and the carbon-carbon bond should be cleaved. Free radical initiated oxidative cleavage of the carbon-carbon double bond of the enolized phenylacetaldehyde can serve as a suitable mechanism for this transformation
(see Figure 5.5). This mechanism is based on the findings of Tokunaga et al. (2005).
They identified molecular oxygen as a convenient oxidizing agent able to oxidize carbon- carbon double bond of enol ethers into alkyl formate and an aldehyde (or ketone). This
o oxidation was catalyzed either by Brønsted acids or preferably by CuCl2 at 40 C.
Previously, Kaneda et al. (1982) also reported on the ability of molecular oxygen to similarly oxidize a structurally related class of compounds, the enamines. We therefore propose that enols similar to enamines or enol ethers can undergo oxidation by a single electron transfer to molecular oxygen and formation of enol radical cation and a superoxide anion as shown in Figure 5.5. In the case of phenylacetaldehyde the enol radical cation is further stabilized through resonance by the formation of a benzylic radical in which the cation is also stabilized as oxonium ion (see Figure 5.5). The resulting superoxide radical can then react with the enol radical cation and generate the known 1,2-dioxetane ring structure capable of thermal cleavage into benzaldehyde and formic acid (Balci et al., 1989; Gollnick & Knutzen-Mies, 1991; Tokunaga et al., 2005).
The temperature dependence of benzaldehyde formation was also demonstrated here (see
Table 5.3). Supporting evidence for the proposed mechanism was provided through FTIR monitoring of the evolution of the benzaldehyde band centered at 1697 cm-1 from heated
80 phenylacetaldehyde in the presence and absence of 1,1’-azobis-(cyclohexanecarbonitrile), a free radical initiator on an ATR crystal. Figure 5.6 shows the aldehyde absorption bands of benzaldehyde and phenylacetaldehyde centered at 1697 cm-1 and 1721 cm-1, respectively. To initiate enolization, the spectrum of phenylacetaldehyde was acquired at
100oC and as shown in Figure 5.6, the enol band appeared at 1684 cm-1. Having assigned the characteristic bands needed to monitor the oxidation process, the phenylacetaldehyde was heated at 100oC on an ATR crystal over a period of 1h and scanned at 2 min intervals. As shown in Figure 5.7 in the presence of the free radical initiator, the enol band of phenylacetaldehyde at 1684 cm-1 formed and increased over a period of 18 min.
After this initial build-up of the band at 1684 cm-1, a new band centered at 1697 cm-1
(benzaldehyde) appeared and increased over time on the expense of the enol band of phenylacetaldehyde, indicating conversion of enol into benzaldehyde. In the absence of free radical initiator the characteristic bands were of lower intensities.
Figure 5.5 Proposed mechanism of free radical initiated oxidation of phenylacetaldehyde from its enol form.
81
Figure 5.6 FTIR spectra of phenylacetaldehyde (short dash), benzaldehyde (dashed) acquired at 35oC, and phenylacetaldehyde acquired at 100oC (solid line).
82 Figure 5.7 Time-dependant spectra of phenylacetaldehyde oxidation catalyzed by 1,1’- azobis-(cyclohexane-carbonitrile) acquired at 100oC over 1h period scanned at 2 min intervals. Dotted spectra represent the first 18 min and solid line spectra represent the remaining 42 min
83 5.3.2 Proposed mechanism of benzaldehyde formation from other precursors
5.3.2.1 Phenylalanine and phenylpyruvic acid
According to Table 5.3 pyrolyzing phenylalanine by itself at 200oC under oxidative condition also produces benzaldehyde. Figure 5.3 illustrates the proposed mechanism whereby phenylalanine undergoes oxidative decarboxylation to form phenethylimine
(pathway A in Figure 5.3) which can be hydrolyzed into phenylacetaldehyde followed by oxidation and formation of benzaldehyde as shown above; alternatively, it can isomerize into enamine and undergo oxidation (Kaneda et al., 1982) to benzaldehyde similar to phenylacetaldehyde as depicted in Figure 5.5. In the presence of carbonyl compounds phenylalanine can undergo transamination and generate phenylpyruvic acid (Zamora et al.,
2006; Chu & Yaylayan, 2008b). Although this intermediate theoretically can decarboxylate and form phenylacetaldehyde as a precursor of benzaldehyde, however, as shown in Table 5.3, the amount of benzaldehyde was not affected much when it was pyrolyzed under oxidative condition. A non-oxidative pathway based on water addition followed by retro-aldol cleavage is proposed in Figure 5.4.
5.3.2.2 1-phenyl-1,2-ethandiol / phenylglyoxal /phenylglyoxylic acid
In the presence of water, the enol form of phenylacetaldehyde can undergo Markovnikov addition of water to generate 1-phenyl-1,2-ethanediol followed by successive oxidations to generate first the enediol (2-hydroxy-1-phenylethanone and/or 2-hydroxyl-2-phenyl acetaldehyde), followed by phenylglyoxal and finally phenylglyoxylic acid formation (see
Figure 5.3). The decarboxylation of the latter hypothetically can yield benzaldehyde.
When commercially available 1-phenyl-1,2-ethanediol and phenylglyoxal were pyrolyzed under oxidative conditions (see Figure 5.2d), they both generated more benzaldehyde
84 compared to non-oxidative conditions, indicating thermal cleavage of gem-diols can also produce an aldehyde.
5.3.2.3 Phenethylamine
Granvogl et al. (2006) proposed a multistep mechanism for the formation of benzaldehyde from a heated mixture of phenethylamine and pyruvaldehyde based on imine isomerization, oxidation and water addition steps. According to Figure 5.4, the formation of phenethylamine Schiff base with any carbonyl compound followed by transamination can generate an imine that can either hydrolyze into phenylacetaldehyde or isomerize into enamine; both are capable to be oxidized into benzaldehyde as depicted in Figure 5.5. As shown in Table 5.3, non-oxidative conditions did not generate any benzaldehyde.
Different precursors generated from phenylalanine therefore can form benzaldehyde in
Maillard model systems; however, phenylacetaldehyde can be considered the major precursor under oxidative conditions, and phenylpyruvic acid can be considered the major precursor under non-oxidative conditions. The corresponding 4-hydroxybenzaldehyde identified in tyrosine model systems incubated under physiological conditions (Horvat &
Jakas, 2004) could also have a similar origin to that of benzaldehyde.
85 Connecting Paragraph
In Chapters 3, 4 and 5, phenylalanine/sugar was used to elucidate selected Maillard reaction pathways by Py-GC/MS based techniques. Phenylalanine/sugar model system was selected because the decarboxylated amino acid moiety of phenylalanine; the phenethylamine and the Strecker aldehyde phenylacetaldehyde, the two marker compounds used in the study, can be easily detected and identified in this model system relative to other amino acids. In Chapter 6, alanine/sugar model was selected as it represents a simple model system in which alanine requires only few 13C-enriched precursors to perform a complete labelling study on the formation of selected volatile, non- volatile and reactive intermediates. The content of this chapter will be submitted to the journal of “Food Chemistry”.
86 Chapter 6: Formation of selected volatile, non-volatile and reactive intermediates generated in alanine/glucose model
3CH N CH3
3CH N CH O 3 3CH N CH3
OH HOH O N CH3 CH 3 + HHO CH3 O OH 3CH O O HOH NH2 N HOH OH OH OH
3 NCH 3CH H CH N 3 H O N CH3
CH3
87 6.1 Introduction
Maillard reaction proceeds via a complex series of steps (Hwang et al., 1995; Keyhani &
Yaylayan, 1996a; Ames et al., 2001; Tehrani et al., 2002) in which various environmental factors such as the temperature, water, air and the type and ratio of reactants will determine the specific course of the reaction (Bemis-Young et al., 1993; Tehrani et al.,
2002; Chu & Yaylayan, 2008a; Chu & Yaylayan, 2008b). It is essential to study the mechanistic pathways responsible for the formation of various aroma, flavour, colour and polymeric compounds that ultimately changes the physical and chemical properties of food during processing. Mechanistic studies have shown for example that the formation of various alkyl-substituted pyrazines and pyrazinones requires a common substrate the α- dicarbonyls (Keyhani & Yaylayan, 1996a; Yaylayan & Haffenden, 2003b). In fact, the formation of pyrroles can also be explained by their involvement as well (Yaylayan &
Haffenden, 2003b). The simplest α-dicarbonyls such as glyoxal, pyruvaldehyde, 2,3- butanedione could be formed from retroaldisation and β-cleavage of the 1-deoxy or 3- deoxy-glucosone (Weenen, 1998; Yaylayan & Keyhani, 1999; Tehrani et al., 2002). The formation of longer chain α-dicarbonyls such as 1,2-butanedione, 2,3-pentanedione and
3,4-hexanedione can be explained by the chain elongation reactions (see Chapter 7). For the formation of pyrazines various mechanisms have been proposed and they were recently summarized by Adams et al. (2008). The most accepted pathway for alkyl substituted pyrazine formation is through Strecker degradation of α-dicarbonyls in the presence of amino acids to form various α-amino-ketones intermediates (Rizzi, 1972), the condensation of any two α-amino-ketones leads to the formation of dihydropyrazine, followed by oxidation to form pyrazine. These α-amino-ketones can also form pyrroles
88 through aldol condensation with acetaldehyde according to Knorr pyrrole synthesis
(Yaylayan & Haffenden, 2003b). Oh et al (1992) proposed two pathways for pyrazinones formations from a peptide specific model systems. The first pathway involved the reaction between the α-dicarbonyls from glucose degradation and the dipeptides, the second pathway involved the reaction of glucose degradation products together with free amino acid hydrolyzed from peptides (Oh et al., 1992). Keyhani and Yaylayan (1996a) also proposed two pathways for pyrazinones formation based on isotope labelling experiments from glycine/glucose model. The first pathway involved a multi-step reaction from three mol of glycine, the second pathway involved the interaction of a glycylglycine dipeptide together with an α-dicarbonyl moiety (Keyhani & Yaylayan, 1996a). As for pyrrole formation, Yaylayan and Haffenden (2003b) had proposed a mechanism based on the glycine and alanine model systems. Pyridines were detected in phenylalanine/glucose
(Keyhani & Yaylayan, 1996b) and phenylalanine/glyceraldehyde (Chu & Yaylayan,
2008b) Maillard model systems. The detection of phenyl-pyridine in both phenylalanine model systems depended on the imine formation between the aldehyde and the phenethylamine (from decarboxylated phenylalanine), followed by electrocyclic ring closure and dehydration or aromatization steps (Keyhani & Yaylayan, 1996b; Chu &
Yaylayan, 2008b). In addition to the α-dicarbonyls and heterocyclic volatile components, some non-volatiles lactones were also detected in a glycine/glucose model system
(Haffenden & Yaylayan, 2008). Haffenden and Yaylayan (2008) proposed that these lactones were generated through glucosones and their deoxy derivatives via three types of transformations such as oxidation, oxidative cleavage and benzylic acid rearrangement. In this chapter, the mechanisms of formation of selected volatiles and non-volatiles such as
89 pyrazine, pyrazinione, pyrrole and pyridine derivatives will be investigated using alanine/glucose based Maillard model systems.
6.2 Experimental Procedures
6.2.1 Materials and Reagents
All reagents and chemicals were purchased from Sigma-Aldrich Chemical Company
(Milwaukee, WI). The labelled [13C]- and [15N]-alanine, [13C]-glucoses were purchased from Cambridge Isotope Laboratories (Andover, MA).
6.2.2 Pyrolysis-Gas Chromatography-Mass Spectrometry (Py-GC/MS)
The Py-GC/MS analyses were conducted using a Varian CP-3800 Gas Chromatograph coupled to Saturn 2000 Ion Trap Mass Spectrometer (Varian, Walnut Creek, USA).
Pyrolysis unit included a CDS 1500 interface and a CDS Pyroprobe 2000 unit (CDS
Analytical, Oxford, USA) was installed onto the GC injection port. About 1mg of a sample mixture containing 1:2 molar ratio of glucose to alanine (or 1:1 molar ratio of glyoxal to alanine) in the presence of silica was packed inside a quartz tubes (0.3mm thickness), plugged with quartz wool, and inserted inside the coil probe and pyrolyzed for 20 seconds at a temperature of 250oC. Volatiles after pyrolysis were concentrated on the sample pre- concentration trap (SPT) at 50oC for 4 minutes and subsequently desorbed at 100oC to the
GC column for separation. A DB-5MS capillary column (J&W Scientific, 50m x 0.2mm i.d; coating thickness, 0.33μm) was used under the following conditions: a pressure pulse of 70 psi was set for the 4 minutes and later maintained with a constant flow of 1.5mL/min
o for rest of the run. GC oven temperature was set at –5 C for 5 minutes using CO2 as the cryogenic cooling source. Temperature was first increased to 50oC at a rate of 50oC/min, then to 180oC at a rate of 5oC/min and later to 280oC at a rate of 20oC/min and was kept
90 there for 9.5 min. MS data was collected using electron impact ionization mode with a scan range of 20-650 m/z. Compound identification was performed by using AMDIS
(v2.62) and NIST Standard Reference Database (v05).
6.2.3 Trapping of α-dicarbonyl intermediates with o-phenylenediamine
The reactive α-dicarbonyl intermediates generated during the alanine/glucose reaction were trapped as their corresponding quinoxaline derivatives using o-phenylenediamine.
The reagent was dissolved in methanol (in excess relative to alanine/glucose) and 1μl of the methanolic solution was added to the alanine/glucose mixture. Both in-situ derivatization (o-phenylenediamine methanolic solution was co-pyrolyzed with alanine/glucose) or post-pyrolytic derivatization (o-phenylenediamine methanolic solution was added to a previously pyrolyzed alanine/glucose sample and then re-pyrolyzed) techniques were used.
6.2.4 Spiking experiments with α-dicarbonyl compounds
The spiking experiments with commercially available α-dicarbonyl compounds were conducted by addition of catalytic amount (~1:10 w/wt) of α-dicarbonyl standards to the alanine/glucose (ratio of 2:1) model or (directly in the presence of alanine only) by using
Py-GC/MS as described in section 6.2.2. These experiments served to confirm the participation of a given α-dicarbonyl compound in the formation of pyrazines and pyrazinones by observing an increase in their peak areas. The peak areas were measured in the chromatograms of the alanine/glucose samples with and without spiking of the corresponding α-dicarbonyls.
91 6.2.5 Ion Chromatography
The presence of glucose in the pyrolytic residue was analyzed using a Metrohm MIC-8 modular IC system (Herisau, Switzerland) consisting of a pulsed amperometric detector
(E1 = 0.15 v, t1 = 400 ms, E2 = 0.75v, t2 = 200ms; E3 = -0.15v, t3 = 400ms), a pump and a sample injection unit connected to Metrosep Carb1-150 anion exchange column thermostated at 31oC. The mobile phase was 0.1N NaOH with a flow rate of 1 mL/min.
6.2.6 Post-Pyrolytic derivatization of non-volatiles
The analysis of non-volatiles was achieved through a post-pyrolytic derivatization technique (Yaylayan et al., 2005; Haffenden & Yaylayan, 2008). Several trimethylsilyl
(TMS) derivatizing agents were tested including trimethyl-silyldiethylamine (TMSDEA) and hexamethyl-disilazane (HMDS). Both of these agents provided satisfactory results and
TMSDEA was used in this study. The derivatization technique involved two stages of pyrolysis. The sample of alanine/glucose was first pyrolyzed to release the volatiles (see section 6.2.2). The non-volatile residue was subsequently derivatized by adding 1μl of
TMSDEA derivatizing agent. After 10 minutes, the sample was re-pyrolyzed at 250oC for
20 seconds to desorb the silylated (TMS) products. Compounds were identified using
AMDIS (v2.62) and NIST Standard Reference Databases (v05).
6.3 Results & Discussion
The analysis of alanine/glucose model system generated over 150 components at 250oC.
These components include pyrazines, pyrazinones, pyridines, imidazolidinones and pyrroles in addition to non-volatiles. Because the formation of α-dicarbonyl intermediates plays a critical role in the generation of both volatiles and non-volatiles, their formation in the alanine/glucose model system was studied through isotope labelling and trapping
92 experiments with o-phenylenediamine. In addition, their involvement in the formation of pyrazines and pyrazinones were verified through spiking experiments with appropriate α- dicarbonyl compounds.
6.3.1 Formation of α-dicarbonyls in alanine/glucose model system
The presence or formation of α-dicarbonyls in alanine/glucose model system (see Figure
6.1) was investigated either through trapping with o-phenylenediamine or through their involvement in pyrazine formation (see section 6.3.3 below).
O O O 3CH O 3CH O O 3CH O 3CH
O O O 3CH O 3CH O O O CH 3 CH3 CH3 3CH A B C D E F G
Figure 6.1: Examples of α-dicarbonyls detected in alanine/glucose or alanine/glyoxal model systesm. (A) glyoxal (B) pyruvaldehyde), (C) 2,3-butanedione (D) 1,2-butanedione (E), 2,3-pentanedione (F), 3,4-hexanedione (G) and 2,3-hexanedione.
Both in-situ derivatization and post-pyrolytic derivatization techniques were used to eliminate the formation of artefacts. In the post-pyrolytic derivatization technique the residue remaining after the initial pyrolysis of alanine/glucose model system was derivatized in the absence of glucose as confirmed by ion chromatography (see section
6.2.5) thereby eliminating the possibility of α-dicarbonyl formation from sugar by the action of the reagent. These studies have indicated that α-dicarbonyls formed in the model system can be trapped by o-phenylenediamine to form their corresponding quinoxaline derivatives (see Figure 6.2 and Table 6.1). The parent quinoxaline was only detected during in-situ derivatization with o-phenylenediamine, which indicated the formation of glyoxal. The 2-methyl-quinoxaline and 2,3-dimethylquinoxaline were also detected
93 during post-pyrolytic derivatization. From the GC chromatograms, 2-methyl-quinoxaline was the highest peak among the three quinoxaline derivatives in alanine/glucose model system when various concentrations of o-phenylenediamine were used, followed by the 2-
3-butandione and glyoxal. These quinoxaline derivatives were also detected when glucose alone was co-pyrolyzed with o-phenylenediamine, however in this case, glyoxal was the highest peak, indicating the difference in the pattern of α-dicarbonyl formation between glucose alone or in the presence of amino acids. Yaylayan and Keyhani (1999) based on a similar study, proposed that only 10% of 2,3-pentanedione could be generated from glucose carbon atoms and the rest 90% could be formed through the incorporation of C2’-
C3’ carbon atoms of alanine together with a C3 carbon units, either C1-C2-C3 or C4-C5-
C6 from D-glucose. The trapping experiments (Table 6.2) with o-phenylenediamine using labelled precursors indicated that 60% of 2,3-pentanedione was formed with the incorporation of C2’-C3’ carbon atoms from alanine, 30% from C1-C2-C3 and 10% from
C4-C5-C6 from D-glucose. The variation of label incorporation in 2,3 pentanedione between the present experiment compared with that of Yaylayan and Keyhani (1999) could be due to differences in the experimental conditions. For example, temperature of heating was 250oC (compared to 210oC) and the ratio of alanine/glucose was 2-1
(compared to 1:1). Furthermore, the pH could be altered in the presence of o- phenylenediamine, thus affecting the label incorporation pattern. Furthermore, 2,3- hexanedione and 3,4 hexanedione were also detected in this trapping experiment but the amount was very small and therefore it was very difficult to accurately determine the isotope label incorporation.
94 OR N R1 NH2 1 + NH N R2 2 O R2 Quinoxaline derivatives
Figure 6.2 Detection of dicarbonyl reactive intermediates by o-phenylenediamine as quinoxaline derivatives
Table 6.1 Quinoxalines detected during pyrolysis1 of alanine/glucose model system Compounds M.W. Structure α-dicarbonyl2 Reference quinoxaline 130 N A NIST, Spiking
N 2-methyl- 144 N CH3 B NIST, Spiking quinoxaline N 2,3-dimethyl- 158 N CH3 C NIST, Spiking quinoxaline N CH3 2-ethyl- 158 CH3 D Proposed quinoxaline N
N 3 2-ethyl,3-methyl- 172 CH3 E NIST, Spiking quinoxaline N
NCH3 2,3-diethyl- 186 CH3 F NIST, Spiking quinoxaline N
N
CH3 2-proply,3- 186 CH3 G Spiking methyl- quinoxaline N
NCH3 1 250oC for 20 s; 2 see Figure 6.1 3 see Table 6.2
95 Table 6.2 Isotope label incorporation in 2-ethyl-3-methyl-quinoxaline CH3 N
N CH3 M M+1 M+2 M+3 M+4 Total 172 173 174 175 176 C1 glu 70 30 0 0 0 100 C2 glu 70 30 0 0 0 100 C3 glu 60 40 0 0 0 100 C4 glu 90 10 0 0 0 100 C5 glu 90 10 0 0 0 100 C6 glu 90 10 0 0 0 100 C’1 ala 100 0 0 0 0 100 C’2 ala 40 60 0 0 0 100 C’3 ala 40 60 0 0 0 100 N15 100 0 0 0 0 100
6.3.2 Acetaldehyde formation
Paine et al (2008) showed that acetaldehyde could be formed from pyrolysis of glucose and is formed mainly from C-1/C-2 or C-5/C-6 atoms of glucose. Acetaldehyde could also be generated from Strecker degradation of alanine (Yaylayan & Wnorowski, 2001).
Acetaldehyde formation could also be explained based on the ability of 5-oxazolidinone formation from amino acid in the presence of carbonyls compound (Perez Locas &
Yaylayan, 2008; Chu & Yaylayan, 2008b; Chu & Yaylayan, 2009), which could decarboxylate and form two isomeric imines 1a and 1b (Figure 6.3). The hydrolysis of these two imines would form acetaldehyde and ethylamine. In fact, Rizzi (2008) had proposed a similar mechanism for acetaldehyde and ethylamine formation from alanine/pyruvaldehyde model system, without the step of 5-oxazolidinone formation.
Furthermore, acetaldehyde can also be formed from ethanolamine through deamination in alanine/glycoaldehyde model system (section 6.3.1.7 and Figure 6.10).
96 O
O 1' OH R1 O 3' 2' 1' O 3CH R2 N 3' + H O 3CH 2' NH2 R2 O R1 5-oxazolidinone
-CO2
3' 2' 3' 2' 3CH R2 Transamination 3CH R2 N N O O
R1 R1 1a 1b
R2 3' R 3' 2' 2' 2 NH CH 3CH + 2 3 + O O O NH2 O R acetaldehyde 1 ethylamine R1
Figure 6.3: Mechanism of formation for ethylamine and acetaldehyde from alanine
97 6.3.3 Volatiles generated from alanine/glucose Model system
6.3.3.1 Pyrazine formation
Pyrazines are common aroma compounds found in a variety of foodstuff, especially in cooked foods (Bondarovich et al., 1967; Walradt et al., 1971; Mussinan et al., 1973; Baltes
& Bochmann, 1987a). They impart the roasted, smoky and nutty type of aromas to cooked foods. Various mechanisms have been proposed for pyrazine formation based on different reaction models and they were recently summarized by Adams et al. (2008). The most accepted pathway for alkyl pyrazine formation is through Strecker degradation of α- dicarbonyls generated from reducing sugars in the presence of amino acids at high temperatures to form various α-amino-ketones (Rizzi, 1972). The condensation of two amino-ketones leads to the formation of dihydropyrazine, followed by oxidation to form pyrazine (Pathway A in Figure 6.4). An alternative pathway (Pathway B in Figure 6.4) involves the formation of dihydropyrazine from shorter chain α-dicarbonyls, such as glyoxal, pyruvaldehyde, 2,3-butanedione or 1,2-butanedione through Strecker degradation and condensation (Shibamoto & Bernhard, 1977; Maga, 1982; Amrani-Hemaimi et al.,
1995; Low et al., 2007) followed by the addition of an aldehyde such as formaldehyde or acetaldehyde and further dehydration in order to form the corresponding pyrazines. As mentioned above, acetaldehyde could be formed from glucose, or Strecker degradation of alanine or from the decarboxylation of 5-oxazolidinone followed by hydrolysis. Table 6.3 lists the pyrazines identified in alanine/glucose model system through the use of Py-
GC/MS together with the NIST library and the commercially available standards. The mechanism for their formations can be explained by both Pathway A and B in Figure 6.4.
98 Table 6.3 Pyrazines detected during pyrolysis1 of alanine/glucose model system Compounds M.W. Structure α-Dicarbonyls2 2-methyl-pyrazines 94 N CH3 A+B
N N CH 2,5(6)-dimethyl-pyrazine 108 3 3CH N CH3 B+B
3 NCH N 2,3,5-trimethyl-pyrazine 122 3CH N CH3 B+C
N CH3 2-ethyl-3,5-dimethyl 136 CH3 B+D CH N pyrazine and N 3
3-ethyl-2,5-dimethyl- NCH 3 NCH CH3 3 pyrazine (Costituente) CH 3 Tetramethyl-pyrazine 136 3CH N CH3 C+C
N CH 3CH 3 2-3-diethyl-5-methyl- 150 CH3 B+F pyrazine N CH3
N
CH3 1250oC for 20 s 2 see Figure 6.1
99 H R1 N R3 R1 NH2 O R3
+ R N R R O 2 4 1 O R3 R O H Strecker 2 2NH R4 + + R NH H R2 O O R 1 2 O R4 4 R N R + 1 4
R2 O 2NH R3 R2 N R3 Pathway A H [O]
R1 N R3 R1 N R4 +
R2 N R4 R2 N R3
H R NH O R 1 2 3 R1 N R3 + R O O R 1 3 R O NH H Strecker 2 2 R2 N H + H R O 2 O H R1 NH2 O H H + R1 N H
R2 O 2NH R3 Pathway B R2 N R3 H O
R4
R1 N R3 R1 N R4
R2 N R4 R2 N R3
Figure 6.4: General pathways of pyrazine formation from α-dicarbonyls
6.3.3.2 Alkyl-substituted pyrazines formation
Some of the longer chain multi-substituted pyrazines generated in the alanine/glucose
Maillard model system cannot be explained by pathway B (Figure 6.5) and hence they require the presence of pre-formed α-dicarbonyls. Examples of these pyrazines are shown in Table 6.4. Their structures were proposed based on their molecular weights, the spiking experiments with appropriate α-dicarbonyls and then further confirmed with the
100 use of 15N and 13C isotope labelled precursors. These pyrazines were also detected in other Maillard model systems and in roast coffee. For example, various pyrazines including ethyl-trimethylpyrazine (C+E) was detected in extruded mixtures of wheat flour at 100-120oC with addition of glucose and various amino acids including alanine, leucine, lysine, threonine and cysteine (Farouk et al., 2000). Farouk et al. (2000) reported that they detected low levels of ethyl trimethylpyrazine in threonine/glucose (0.06% of total pyrazine peak area) and alanine/glucose and (0.21% of total pyrazine peak area). The 2,5- diethyl-3,6-dimethyl- and 2,6-diethyl-3,5-dimethylpyrazines (E+E) were detected in 2,3- pentanedione and alanine reaction (Rizzi, 1972), in roast coffee (Baltes & Bochmann,
1987a) and in wine with the 2,3-pentanedione and cysteine as substrates (Pripis-Nicolau et al., 2000). Both 2,3-diethyl-5,6-dimethyl (C+F) and 2,3,5-triethyl,6-methylpyrazines
(E+F) were detected in roast coffee (Baltes & Bochmann, 1987a). Tetra-ethylpyrazine
(F+F) was detected in 3,4-hexanedione and alanine reaction mixture (Rizzi, 1972). One new tentatively identified pyrazine in this study was 2,6-dipropyl-2,5-dimethylpyrazines
(G+G), based on NIST library search and was confirmed with 2,3-hexanedione spiking experiments. Some of these pyrazines (Table 6.5) were further confirmed through their reported mass spectra published by Baltes and Bochmann (1987a). The next chapter
(Chapter 7) provides detailed formation pathways of some of the required α-dicarbonyls.
The type of alkyl-substituted pyrazines present in alanine/glucose model was depended on the initial α-dicarbonyl being generated. In the spiking experiments, it was confirmed that these pyrazines were formed from various α-dicarbonyls in the system. There was a significant increase in peak area by spiking the appropriate α-dicarbonyls as reactants to
101 the normal alanine/glucose (ratio of 2-1) mixtures. Table 6.6 shows the extent of increase in the peak areas of the effected pyrazines.
Table 6.4 Pyrazines detected during pyrolysis1 of alanine/glucose model system Compounds M.W. Structure Dicarbonyls2 Reference 2-ethyl-3,5,6-trimethyl 150 CH3 C+E (Farouk et pyrazines N CH3 al., 2000) (or ethyl-trimethyl- 3 NCH CH3 pyrazine) 2,5-diethyl-3,6- 164 CH3 E+E (Baltes & dimethyl- N CH3 Bochmann, pyrazines 1987a; 3 NCH Pripis- CH3 Nicolau et al., 2000)
2,6-diethyl-3,5-dimethyl 164 3CH N CH3 E+E (Baltes &
pyrazines CH3 Bochmann, N 1987a; CH3 Pripis- Nicolau et al., 2000)
2,3-diethyl-5,6 dimethyl 164 CH3 C+F (Baltes & pyrazines N CH3 Bochmann, 1987a) N CH3
CH3 2,3,5-triethyl,6-methyl 178 CH3 CH3 E+F (Baltes & pyrazines N Bochmann, (or triethyl-methyl- 1987a) N CH3 pyrazine) CH3 Tetra-ethyl-pyrazines 192 CH3 CH3 F+F (Rizzi, N 1972)
N
CH3 CH3 2,5-dipropyl-3,6- 192 CH3 G+G NIST
dimethyl-pyrazines 3CH N Library
N CH3
3CH 1 250oC for 20s 2 see Figure 6.1
102 O O 3CH O O 3CH O O 3CH O 3CH
O CH O CH O O O O O 3 3 CH CH3 CH3 3 3CH SD
CH O O O 3 O CH O O 3CH O 3 3CH
NH NH 2 NH NH2 2 3CH NH2 3CH NH2 2 NH2 CH CH3 CH3 3 3CH
A B C D E F G
CH H 3 CH H H 3 CH N CH H 3 3 3CH N 3CH N CH3 3CH N
N N CH3 N CH3 CH N H H H 3 CH CH H 3 3 CH3 CH B + C C+ F 3 E+E E+E
[O] Pathway A [O] CH3 CH3 3CH N CH3 3CH N 3CH N CH3 3CH N
N N CH3 N CH3 3CH N CH3 CH3 CH3 CH3 B + C C + F E+E E+E
-H2O -H2O O CH O 2 3CH O CH2 CH3 CH3 H H H H N N N CH3 3CH N
N N N N CH3 3CH H H H H CH CH 3 3 CH3 D'+D' A'+C' C'+D' D'+D'
CH3 NH 2 NH CH NH 2 3 2 NH2 Pathway B
CH O CH O O 3 3 O A' B' C' D'
SD CH3 O O 3CH O O
CH O CH O O 3 3 O Figure 6.5 Some examples of pyrazine formation from their corresponding α-dicarbonyls
103 Table 6.5 Mass spectrometric data and retention times of pyrazines compared to Baltes and Bochmann (1987a) Compound Retention MS Data time 2,3-diethyl-5,6- 20.350min Ref 13:164(100), 163(54), 149(87), 136(8), dimethyl-pyrazines 135(7),121(4),107(2),56(21), 54(13), 53 (12), 42 CH3 (4) N CH3 experimental result1: 165(68), 164(100), 163(85), N CH3 150(7), 149(29), 136(60), 135(11), 134(4), CH3 121(1), 108(3), 107(2),56(5), 54(1), 42 (11)
2,6-diethyl-3,5- 20.386min Ref 13:164(81), 163(100), 149(29), 136(13), dimethyl-pyrazines 135(5),121(13),108(4),107(4), 94(5), 80 (3), 3CH N CH3 67(21), 66(5), 56(6), 53(18), 41(14), 39(19)
CH3 N experimental result1: 164(63), 163(100), 149(21), CH3 136(6), 135(2), 121(9), 108(3), 107(3), 94(3), 80 (1), 67(18), 66(5), 56(8), 53(12), 41(23), 39(30)
2,5-diethyl-3,6- 20.426min Ref 13: 164(70), 163(53), 149(100), 136(9), dimethyl-pyrazines 121(7), 108(3), 94(4), 67(20), 66(4), 56(3), CH3 53(12), 41(12), 39(14) N CH3 experimental result2: 164(80), 163(66), 149(100), 3 NCH 136(4), 121(9), 108(2), 94(1), 67(17), 66(5), CH3 56(5), 53(13), 41(18), 39(20)
Triethyl-methyl- 21.5min Ref 13: 178(100), 177(57), 163(94), 149(11), pyrazine 135(17), 121(12), 94(5), 93(5), 81(5), 80(7), CH3 CH3 68(10), 67(22), 66(9), 56(10), 54(8),53(15) N experimental result3: 179(22), 178 (100), N CH3 177(48),164 (12), 163 (61), 150 (4), 149 (6), 148 CH3 (3), 135 (5), 68(2), 67(9), 66(4) 56(4), 53(7)
experimental result1,4: 179(17), 178 (88), 177(48), 164 (12), 163 (100), 150 (3), 149 (13), 148 (15), 135 (6), 122 (5) 67(10), 66(5) 56(8), 53(6)
Note: 1alanine spiked with 2,3-butanedione and 3,4-hexanedione (not devoluted because it is overloaded) 2alanine-glucose (not spiked and not devoluted because the intensity is too low) 3alanine-glucose spiked with 2,3-pentanedione and 3,4-hexanedione (intensity is high) 4alanine-glucose
104 Some of the pyrazines detected required a mixture of two α-dicarbonyls to enhance their peak areas, therefore, the limiting factor for their formation was the simultaneous availability of both α-dicarbonyls in the mixture. In the case of tetraethyl-pyrazines (F+F) and 2,5-dipropyl-3,6-dimethylpyrazines (G+G), the concentration was very low in the alanine/glucose mixture. This might be explained by the fact that the formation of 2,3- hexanedione and 3,4-hexanedione required multiple additions, which produced very little amount in the model system. In fact, their concentrations were increased only upon spiking the corresponding α-dicarbonyls, and the increase was not as significant as compared to other pyrazines. The explanation might be that 2,3-hexanedione and 3,4- hexanedione were readily reacted with other shorter chain α-dicarbonyls to form pyrazines. For example, 2,3-diethyl-5,6-dimethylpyrazines (C+F) and 2-propyl-3,5,6- trimethylpyrazines (G+G) were increased by 192 and 119 fold respectively when both corresponding α-dicarbonyls were spiked.
105 Table 6.6 Increase/decrease in intensity of the peaks associated with pyrazines when spiked with selected α-dicarbonyls CH CH3 CH3 3 CH3 CH3 CH3 N CH CH CH CH3 N CH N CH3 CH N CH N CH CH CH 3 3 3 3 3 3 3 3 3 3CH N CH3 N N N 3CH N NCH NCH CH N N CH CH N 3 N 3 3 3 N N CH N CH3 3 CH CH CH CH 3 CH N CH CH 3 CH CH CH 3 3 3 3 3 3 3 3 3CH Dicarbonyls1 B+F C+E C+G C+F E+E E+E D+F E+F F+F G+G R.T. 19.093 19.18 19.40 20.350 20.387 20.426 20.606 21.5 22.46 23.076 M.W. 150 150 150 164 164 164 164 178 192 192 alanine/glucose- 1.0 1.0 1.0 0.0 1.0 1.0 1.0 1.0 tiny
+ Pyruvaldehyde 1.4 1.2 check 0.0 0.0 check 1.1 1.4 0.0 + 2,3 butandione 0.5 0.8 0.0 1.0 0.0 0.4 0.5?? 0.5 0.0 + 2,3 pentanedione 1.9 3.7 0.0 0.0 31.7 33.4 0.0 19.1 0.0 + 2,3 hexanedione 1.0 1.0 12.1 0.0 1.2 check 30.1 check 0.0 + 3,4 hexanedione 2.7 0.2 0.0 4.0 0.0 0.4 0.4 19.8 1.0 + 1,2 cyclo hexanedione 0.2 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 + Pyruvaldehyde + 2,3 hexanedione 1.5 1.5 13.3 0.0 check check 18.2 0.0 0.0 + Pyruvaldehyde + 3,4 hexanedione 13.0 0.5 check 9.6 0.0 check 0.4 28.4 0.3 + 2,3 butanedione + 2,3 pentanedione 0.8 6.4 0.0 0.0 check 11.3 0.0 10.1 0.0 + 2,3 butanedione + 2,3 hexanedione 4.3 0.0 1.8 7.2 1.5 1.6 119.3 0.0 0.0 + 2,3 butanedione + 3,4 hexanedione 13.7 1.2 check 191.9 0.0 check 0.0 26.5 2.3 + 2,3 pentanedione + 3,4 hexanedione 20.5 1.9 0.0 14.5 16.5 8.2 0.0 194.3 1.7 1 see Figure 6.1
106 6.3.3.3 Pyrazinones formation
The role of the α-dicarbonyls identified above in alanine/glucose model system, was not limited only to the pyrazine formation, but also extended to the formation of pyrazinones.
Pyrazinones were found in Maillard models such as in glycine (excess) /glucose (Keyhani
& Yaylayan, 1996a; Yaylayan et al., 1997) and peptides/glucose (Oh et al., 1992). Table
6.7 illustrates some examples of the pyrazinone structures found in alanine/glucose model system. Similar α-dicarbonyl structures were involved in their formation as that of pyrazines such as glyoxal (A), pyruvaldehyde (B), 2,3-butanedione (C), 1,2-butanedione
(D), 2,3-pentanedione (E) and 3,4-hexanedione (F) (see Figure 6.1). Isotope labelling studies have revealed that all the pyrazinones have incorporated one carbon atom from
[13C-1]-alanine and two nitrogen atoms from [15N]-alanine in their structures. In the case of [13C-2] and [13C-3]-alanine label incorporation, this was varied depending on whether or not alanine was involved in the formation of more complex α-dicarbonyl structures from simple ones (Figure 6.7). This was also consistent with the label incorporation in similar α-dicarbonyl structures involved in the formation of corresponding pyrazines. As mentioned in 4.3.1.1, the extent of label incorporation of these α-dicarbonyls depended on the experimental conditions. A multi-step mechanism for the formation of pyrazinone was proposed in Figure 6.6. Alanine forms the imine (2a) with the first carbonyl group of the α-dicarbonyl. Then, a second molecule of alanine forms another imine (2b) with the second carbonyl group followed by transamination (2c), dehydration (2d) and decarboxylation steps. Another pathway similar to the pyrazine formation, involves the
Strecker degradation of the α-dicarbonyl to form α-amino-ketone (2e), which further reacts with another molecule of alanine and pyruvic acid to form imine (2b) and
107 subsequently to form the pyrazinone. Oh et al. (1992) used peptides, such as Gly-leu,
Leu-Gly and a mixture of Gly and Leu as their Maillard reaction model, heated with glucose. Therefore, their proposed pathway did not require the formation of the second imine. R1 and R2 in Figure 6.6 could be hydrogen, methyl or ethyl groups depending on the incorporated α-dicarbonyl structure. Figure 6.7 lists the tentatively identified pyrazinones. Table 6.8 shows the effect of spiking with α-dicarbonyls on the intensity of selected pyrazinones, using alanine/glucose (ratio of 2:1) mixture. In the case of pyvualdehyde and 2,3-butanedione, the peak areas of their corresponding pyrazionones were not increased probably due to the fact that these α-dicarbonyls were reacting with other components in the model system such as preferentially forming pyrazines. In another set of spiking experiments in which the two α-dicarbonyls were added to alanine without the presence of glucose, the peak area of 1-ethyl-3,5,6-trimethyl-2(1H)- pyrazinone was increased by 12.5 fold when 2,3-butanedione was added. In the case of pyvualdehyde, the peak areas for neither of the pyrazinones were increased as compared to the alanine/glucose model. This might be explained by the fact that significant amount of pyruvaldehyde generated from glucose, or pyrvualdehyde was involved in other reactions such as the formation of 2,3-pentanedione through chain elongation.
108 Table 6.7 Possible pyrazinones detected during pyrolysis1 of alanine/glucose model system Compounds M.W. Structure Spiking Dicarbonyls2 1-ethyl-3-methyl- 138 3CH NH Yes A 2(1H)-pyrazinone O HN
3CH 1-ethyl-3,5-dimethyl- 152 3CH NCH3 Yes B 2(1H)-pyrazinone O HN
3CH 1-ethyl-3,6-dimethyl- 152 3CH NH Yes B 2(1H)-pyrazinone O N CH3
3CH 1-ethyl-3,5,6- 166 3CH NCH3 Yes C trimethyl-2(1H)- O N CH pyrazinone 3 3CH 3CH N 1,5-diethyl-3-methyl- 166 CH3 No D 2(1H)-pyrazinone O HN
3CH 1,6-diethyl-3-methyl- 166 3CH NH No D 2(1H)-pyrazinone O N
CH3 3CH 3CH N 1,5-diethyl-3,6- 180 CH3 Yes E dimethyl-2(1H)- O N CH pyrazinone 3 3CH 1,6-diethyl-3,5- 180 3CH NCH3 Yes E dimethyl-2(1H)- O N pyrazinone CH3 3CH 3CH N 1,5,6-triethyl-3- 194 CH3 Yes F methyl-2(1H)- O N pyrazinone CH3 3CH 1 250oC for 20s 2 see Figure 6.1
109 OH OH OH 3' 3' 3' R1 O 1' Ala 1' Ala Transamination 1' 2' CH3 2' CH3 2' CH3 O O O
R1 N R1 N R1 N R2 O H
R2 O R2 N R2 N Ala 2a O O CH 2' CH 2' 3 1' 3 1' SD 3' OH 3' OH Ala R1 O 2b 2c
and pyruvic acid -H2O
R2 NH2 3' 2e CH 2' N 3 R1
1'
-CO2 O N R2 3' 1' CH 2' N O Pyrazines 3 R1 3CH 2' 1' 3' OH O N R 2 2d
3CH 2' 3'
Figure 6.6 Mechanism of formation for pyrazinone from alanine and α-dicarbonyls SD = Strecker reaction
110 Table 6.8 Increase/decrease in intensity of the peaks associated with pyrazinones when spiked with selected α-dicarbonyls 3CH NH 3CH NH 3CH NH3CH NCH3 3CH NCH3 3CH N CH N 3CH NCH3 CH 3 3 CH3 3CH N CH3 O N CH3 O N O HN O HN O N CH O HN O N CH O N 3 3 O N CH3 CH CH CH3 CH CH CH 3 CH CH 3 CH CH 3 3 3 3 3 3 3 3CH Dicarbonyls1 A B B C D D E E F Retention time. 21.228 22.390/22.722/23.735 25.018 23.528/23.803 25.70/25.80 26.40 M.W. 138 152 152 166 166 166 180 180 194 alanine/glucose 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
+ glyoxal 5.3 0.6 0.6 0.7 3.9 1.7 0.4 + Pyruvaldehyde 0.4 0.3 2.1 0.3 0.6 0.2 + 2,3 butandione 1.0 0.2 0.3 1.2 0.1 0.1 + 2,3 pentanedione 0.6 1.5 3.7 2.3 10.1 1.2 + 2,3 hexanedione 0.7 0.6 12.1 0.4 1.2 + 3,4 hexanedione 0.1 0.2 1.7 0.5 0.2 0.4 17.8 Alanine + pyruvaldehyde 1.0 0.6 0.9 alanine + 2,3 butandione 12.5 1 see Figure 6.1
111
3' CH 2' N 3 R1
1'
O N R2
3CH 2' 3'
3' 3' 3' CH 2' NH 3CH 2' NH 3CH 2' NCH3 3
1' 1' 1' O HN O HN O N CH3
CH 2' 3CH 2' 3CH 2' 3 3' 3' 3'
R =CH R =H M/Z=152 R =H R =CH M/Z=152 R1,R2=H M/Z=138 1 3, 2 1 , 2 3
3' 3' 3' 3CH 2' N 3CH 2' NH 3CH 2' NCH3 CH3 1' 1' 1 O HN O N O N CH3
CH3 3CH 2' 3CH 2' 3CH 2' 3' 3' 3'
R1=CH2CH3, R2=H M/Z=166 R1=H, R2=CH2CH3 M/Z=166 R1,R2=CH3 M/Z=166
3' 3' 3' CH 2' NCH 2' 3 3 3CH N 3CH 2' N CH3 CH3 1' 1' 1' O N O N CH3 O N
CH3 CH CH 2' 2' 3 3 3CH 3CH 2' 3' 3' 3' R R =CH CH M/Z=194 R1=CH3,R2=CH2CH3 M/Z=180 R1=CH3,R2=CH2CH3 M/Z=180 1, 2 2 3
Figure 6.7 Incorporation of alanine atoms in pyrazinone structures excluding the dicarbonyl moiety based on labelling experiments (see Figure 6.6)
112 6.3.3.4 Imidazolidinones formation
The formation of imidazolidinone moiety during Maillard reaction is only reported in peptide-sugar model systems (Horvat et al., 1998; Roscic et al., 2001; Roscic & Horvat,
2006). In this study, two new isomeric compounds having a molecular weight of 143 amu were identified and were tentatively assigned 2,5-dimethyl-3-ethyl-4H-imidazolidinone moiety. Based on the isotope labelling data, these isomers did not contain any sugar carbon atoms but they formed only from alanine atoms in the presence of glucose or in the presence of α-dicarbonyls with the exception of glyoxal. Table 6.9a shows that they both posses one carbon atom from [13C-1]-alanine, three carbon atoms from [13C-2]- and
[13C-3]-alanine and two nitrogen atoms from [15N]-alanine. In addition, Figure 6.9 shows their mass spectral fragmentation patterns and Tables 6.9b and 6.9c lists isotope label incorporation in the proposed mass spectral fragments consistent with the proposed structure. It is proposed that these two peaks represent two stereoisomers (2R)-and (2S)-
2,5-dimethyl-3-ethyl-4H-imidazolidinone formed during the reaction with acetaldehyde.
The mechanism of formation is proposed in Figure 6.8 which is based on the reaction of alanine with ethylamine generated from the decarboxylation of 5-oxazolidinone and followed by the hydrolysis of the imine (see Figure 6.3 and section 6.3.1.2). Alanine and ethylamine can form the amide, decarboxylated alanine dipeptide, which can also be formed through the decarboxylation of alanine dipeptide. This amide can further react with acetaldehyde and cyclize to form two stereoisomers of imidazolidinone moiety with the number of carbon and nitrogen atoms consistent with the labelling data. Interestingly, this decarboxylated alanine dipeptide could also react with α-dicarbonyls to form the corresponding pyrazinones discussed above.
113 Table 6.9a The number of labelled atoms in imidazolidinones Compounds M.W. Structure C1-ala C2-ala C3-ala N-ala (2R or 2S)-2,5-dimethyl-3- 143 3CH H 1 3 3 2 ethyl-4H-imidazolidinone N
O + N CH3 H CH3 (2R or 2S)-2,5-dimethyl-3- 143 3CH H 1 3 3 2 + H ethyl-4H-imidazolidinone N O N CH3
CH3
Table 6.9b Incorporation of labels in the mass spectral fragments of 1 2,5-dimethyl-3-ethyl-4H-imidazolidinone (tR= 19.98min) Models m/z m/z m/z m/z m/z m/z m/z 143 127 99 70 56 44 28 C1-ala-glyoxal 144 128 99 70 56 44 28 C2-ala-glyoxal 146 130 102 72 58 45 29 C3-ala-glyoxal 146 129 101 72 57 45 28 N-ala-glyoxal 145 129 101 71 57 45 29 1See Figures 6.8 & 6.9
Table 6.9c Incorporation of labels in the mass spectral fragments of 1 2,5-dimethyl-3-ethyl-4H-imidazolidinone (tR= 20.14min) Models m/z m/z m/z m/z m/z m/z m/z 143 127 99 70 56 44 28 C1-ala-glyoxal 144 128 99 70 56 44 28 C2-ala-glyoxal 146 130 102 72 58 45 29 C3-ala-glyoxal 146 129 101 72 57 45 28 N-ala-glyoxal 145 129 101 71 57 45 29 1See Figures 6.8 & 6.9
114
3' 3' CH 3CH 3 2' 2' 3' NO N 3CH 2' O 1' O 1' R1 NH 2 NH R1 R2 N O 1' R2 2' 2' NH 3' CH CH3 3' 3 3' 3CH 2' O R1 O Pyrazinone OH -CO2 Alanine dipeptide R2 O 3' 3' O 3CH 3' 3CH 3' 2' 2' CH CH3 N 3 2' 2' NH2 2' 3' NH 2NH O 1' 2 + O 1' CH3 CH3 N 2' O 1' 3' NH 2' OH 2' H CH Alanine Ethylamine CH 3' 3 3' 3
decarboxylated alanine dipeptide 3'
3CH H 2' N H 1' O 2' N CH3 3' 2' CH3 3'
mixture of R & S 2,5-dimethyl,3-ethyl-4H- imidazolidinone
Figure 6.8 Proposed mechanism of formation of (2R) and (2S)-2,5-dimethyl-3-ethyl-4H- imidazolidinone in alanine/glucose model system
115
Ion count
2' 2' A 100 3' 2' + H 2' + 3' N H N CH + + 3' 3 CH 1 2' N N 3 3' CH 2' O 3 N CH H 2' 3 75 CH 3' 3 CH 2' 3' 3 CH 3' 3 50
25
0
B 2' 3' 3' 100 2' + H + N CH3 + 3CH H HCHN N 2' 2' 2' 2' CH 3' N 75 1 2' N CH3 3' O + N CH3 2'
CH 2' H 50 3' 3 CH3 3'
25
0 m/z
Figure 6.9 Mass spectra and the proposed mass spectral fragments of 2,5-dimethyl,3-ethyl-4H-imidazolidinone found at retention times of 19.98 min (A) and 20.14 min (B) in alanine/glucose model system
116
6.3.3.5 Pyrroles formation
Pyrrole and N-ethyl-pyrrole were detected in the alanine/glucose and alanine/glyoxal model systems and were identified through NIST Standard Reference Database (v05) search. Table 6.10 lists the number of atoms from alanine incorporated into the structures from isotope labelling experiments using labelled alanine/glyoxal model system.
According to Table 6.10, the parent pyrrole incorporated two carbon atoms from alanine
(C-2 and C-3) and two from glyoxal whereas, N-ethylpyrrole incorporated four carbon atoms from alanine (2xC-2 and 2xC-3) and two from glyoxal. This pattern is consistent with the formation of the pyrrole moiety through the aldol condensation of acetaldehyde with either the 2-amino-acetaldehyde (Amadori rearrangement product of glycolaldehyde with ammonia as shown in Figure 6.10) to form pyrrole or with N-ethyl-2-amino- acetaldehyde (see Figure 6.10) to form N-ethyl-pyrrole as detailed in Figure 6.11.
Yaylayan and Keyhani (2001a) also reported such pyrrole formation pathway from the reaction of α-amino-carbonyl compounds and acetaldehyde. The formation of N-ethyl-2- amino-acetaldehyde (3c) can be rationalized as shown in Figure 6.10 through decarboxylation of 5-oxazolidinone intermediate formed between alanine and glycolaldehyde as demonstrated in chapters 3 and 4. The imine (3b) can undergo
Amadori rearrangement to form N-ethyl-2-amino-acetaldehyde (3c) and then react with acetaldehyde through aldol condensation to form N-ethyl-pyrrole as shown in Figure
6.11. The formation of ethanolamine was further confirmed due to its possible involvement in the formation of pyridine derivatives that were detected in alanine/glyoxal model system (see the following section 6.3.1.8). In alanine/glucose model system, the
117 carbon atom label incorporations from C2’-C3’alanine were not 100% in these two pyrroles, indicating acetaldehyde can be generated from multiple sources.
O OH 2' + CH3 O 1' 3' OH NH2 Amadori Rearrangement
O O NH3 O 1' 3' NH2 N 2' CH3 OH H
O 5-oxazolidinone 3'
CH3 2' -CO H 2 N N 3' 3' 3' NCH NCH3 3 + OH N 2' OH 2' 2' 3a 3b pyrazine pyrrole
3' -NH O O 3 NH 3' 2 + NH CH3 OH CH3 O CH3 2' 2' acetaldehyde ethanolamine 3c
O 3' CH 2' 3 pyridine 3' 2' CH3
N 3'
2' N-ethyl-pyrrole Figure 6.10 Proposed mechanism of formations for pyrrole and N-ethyl-pyrrole from alanine/glycoaldehyde model
118 O O OH NR NR CH3 -H2O O NHR aldol condensation NHR
R=H or R=CH2CH3
Figure 6.11 Proposed mechanism of formations for N-alkyl-pyrrole from acetaldehyde and 2-amino-acetaldehyde
Table 6.10 The number of alanine atoms incorporated in pyrroles during pyrolysis1 of alanine/glyoxal model system Compounds M.W. Structure C1-ala C2-ala C3-alaN-ala Pyrrole 67 H 0 1 1 1 N
N-Ethyl-pyrrole 95 CH3 0 2 2 1
N
6.3.3.6 Pyridines formation
Pyridines have been detected in simple model systems such as glycine/glucose and in over-heated and decaffeinated coffee (Hwang et al., 1995; Ames et al., 2001). As a group, pyridines exhibit mutagenic activity in the Ames assay (Sasaki et al., 1987). Furthermore,
2-amino-1-methyl-6-phenylimidazo-[4,5-b]-pyridine (PhIP), recognized as the main carcinogen in cooked meat is a pyridine derivative. It induces breast, colon and prostate tumor formations in rats (Lauber et al., 2004). Therefore, it is important to understand the general mechanistic pathway leading to pyridine ring. In the alanine/glyoxal model system three chromatographic peaks were observed at retention times of 16.17, 16.48 and 16.61 min having a common molecular weight of 121 amu. The isomers incorporated three carbons atoms from [13C-2]-alanine and three atoms from [13C-3]-alanine and two nitrogen atoms from [15N]-alanine (see Table 6.11). The NIST library search strongly indicated
119 structures related to alkyl-substituted pyridine but with no specific derivative matching accurately the unknown peaks. Based on the number of carbon atoms incorporated in the presumed pyridine structure and the molecular weight, three ethyl-substituted 2- methylpyridine derivatives were proposed as shown in Table 6.11. Their mass spectra and fragments consistent with the proposed structures and most importantly the incorporated labels are displayed in Figure 6.12. Furthermore, a plausible mechanism is shown in
Figure 6.13 for their formation again consistent with labelling data. The pyridines shown in Table 6.11 each have a total of eight carbon atoms, however, only six are originating from alanine and therefore two carbon atoms should arise from glycolaldehyde one of the two reactants. In the proposed scheme this two-carbon unit participates as ethanolamine in the formation of substituted pyridines. The ethanolamine generated from the interaction of glycoaldehyde with alanine (see section 6.3.1.7) can form the Schiff base (4a) in the presence of acetaldehyde. Dehydration followed by isomerization can lead to the formation of N-ethenyl-ethenamine (4b). Further reaction with another molecule of acetaldehyde followed by dehydration can eventually lead to the cyclised structure of 2- methy-1,4-dihydro-pyridine (4c). This key intermediate (4c) can react as enamine (C- nucleophile) with acetaldehyde at three possible positions indicated with arrows as a, b, c in the structure of 2-methyl-1,4-dihydro-pyridine (4c) to form corresponding 2-methyl-3- ethyl-pyridine (4d), 2-methyl-6-ethyl-pyridine (4e) and 2-methyl-5-ethyl-pyridine (4f) observed in the model system.
120
Ion count + CH + CH2 + H N 100 + + CH 3 NCH 2CH 2 + 3CH N A CH2 H N CH 75 2 3CH 2CH CH CH 2 50 2
25
0
100 B
75
50
25
0
100 C 75
50
25
0 m/z
Figure 6.12 Mass spectra and the proposed mass spectral fragments of pyridines found at time 16.17 min (A), 16.48 min (B) and 16.61 min (C).
121
3' 2' 2' CH 2' 3 OH 3' OH -H2O 3' CH N O + 2NH 3 3CH N CH2 4a (or 3a)
3' 2' isomerization 3' 3CH 2' 2' CH OH O 3' -H2O 2 3' CH NH CH N CH3 2 2 N CH 2' 2' 3' 3 CH 3' 2 2' 4b
3' 3' 3' H b CH 2' 3' 3 CH 2' N 3CH 2' N 2CH 3 N 2' 3' 3' 2' 2' a c 4c
3'' 3CH 2" O 3' CH 3' 3' 3 3' 3 2' NCH 3 2' NCH 3 NCH 2' 3' 2 3' 3CH CH3 3' 3' 3' 2' 2' 2' 2' 2'
4d 4e 4f 2-methy-3-ethyl-pyridine 2-methy-6-ethyl-pyridine 2-methy-5-ethyl-pyridine
Figure 6.13 Proposed mechanism of pyridine formation from glyoxal/alanine
Table 6.11 The number of alanine atoms incorporated in pyridine during pyrolysis1 of alanine/glyoxal model Compounds M.W. Structure C1-ala C2-ala C3-ala N15-ala 2-methyl-3-ethyl- 121 3 NCH 0 3 3 1 pyridine 3CH
2-methyl-6-ethyl- 121 CH3 0 3 3 1 pyridine 3 NCH
2-methyl-5-ethyl- 121 3 NCH 0 3 3 1 pyridine CH3
1250oC for 20s
122
6.3.4 Non volatile products from alanine/glucose
In addition to the various volatiles discussed above such as pyrazines and pyrroles, non- volatiles were also formed in the alanine/glucose model system and were detected using post-pyrolytic derivatization technique. In general, non-volatile compounds possessing hydroxyl, carboxyl or amino functional groups, require chemical derivatization for their increased volatility and for their subsequent detection by GC/MS. The trimethyl- silyldiethylamine (TMSDEA) derivatizing agent was used in this study. The samples of alanine/glucose were first pyrolyzed to release the volatiles and the non-volatile residue was subsequently derivatized by adding 1μl of TMSDEA derivatizing agent and re- pyrolyzed after 10 min at 250oC for 20 seconds (see section 6.2.2 and 6.2.6 for experimental conditions). The non-volatile residue from alanine/glucose generated over
30 trimethyl-silyl (TMS) derivatives, however, only four could be identified as TMS derivatives of 2,3,5-tris-O-(trimethylsilyl)-D-arabinoic acid (5A), ribo-hexonic-3-deoxy-
γ-lactone (5B), (cis)-dihydro-3,4-dihydroxy-2(3H)-furanone (5C) and (trans) -dihydro-
3,4-dihydroxy-2(3H)-furanone (5D) by using NIST Standard Reference Database (v05) search (see Figure 6.14). These non-volatiles were also detected previously in glycine/glucose model system and their structures were confirmed through isotope labelling analyses (Yaylayan et al., 2005; Haffenden & Yaylayan, 2008).
123
O
O O OH O OH OTMS O O O H OH H OH TMS HOH HOH OH HOH HOH TMSO OH OTMS
OH OH 5A
O O O O
OH H OH O OH O OH H OH OH HOH OH OH OH O OH OH HOH H H H OH OH H H H OH OH OH OH OH 3DG
O OH TMS
O OH
TMS O OTMS
CH3 OTMS O OH O O O O HOH OTMS HOH O OTMS TMSO HOH HOH 5C and 5D 5B OH OH
1DG
Figure 6.14 Proposed mechanisms for the formation of selected non-volatiles from alanine/glucose (adapted from Haffenden & Yaylayan (2008)
124 6.5 Conclusion
The research outlined in this chapter can contribute toward the development of a comprehensive analytical approach to the study of Maillard reaction using Py-GC/MS based techniques that allows the detection of volatiles, non-volatile and reactive intermediates. The main advantage of this technique lies in its ability to require only a small quantity of reactant material (~1mg) to analyze both volatiles and non-volatiles from the same sample. This is especially important in mechanistic studies using expensive isotope enriched starting materials.
125 Connecting Paragraph
In Chapter 6, the mechanism of formation of selected heterocyclic compounds such as pyrazines, pyrazinones, pyridines, pyrroles and imidazolidinones generated in the alanine/sugar model system was elucidated through the use of isotope labelling technique. This study has identified α-dicarbonyl compounds as common precursors to imidazolidinones, pyrazines and pyrazinones. In chapter 7, the detailed mechanism of formation of two of the above α-dicarbonyl compounds, namely 1,2-butanedione and 3,4- hexanedione were investigated through the detection and identification of ethyl- substituted pyrazines in alanine/sugar model systems using isotope labelling and oxidative pyrolysis GC/MS based techniques. Chapter 7 was submitted to the Journal of
Agriculture and Food Chemistry.
126 Chapter 7: Isotope labelling studies on the origin of 3,4-hexanedione and 1,2- butanedione in alanine/glucose model system
CH3 O O O Alanine Alanine
O O O CH 3 CH3
127 7.1 Introduction
The origin of many reactive α-dicarbonyl compounds such as glyoxal, pyruvaldehyde and
2,3-butanedione formed in the Maillard model systems are well-established and can be attributed to the retro-aldol and elimination reactions of different glucosones and other sugar fragments. They play a critical role not only in the generation of different heterocyclic compounds during thermal processing of food but also in the formation of cross-links and advanced glycation end-products. However, the formation of longer alkyl chain containing α-dicarbonyls such as 2,3-pentanedione, 2,3-hexanedione and 3,4- hexanedione is a less known and a more complex process due to their multiple origin and the involvement of both sugar and amino acid carbon atoms. The 2,3-pentanedione for example can be generated in glucose/alanine model system (Yaylayan & Keyhani, 1999) either totally from sugar carbon atoms (10%) or through participation of amino acid carbon atoms (90%). These percentages can vary depending on the reaction conditions
(Cerny, 2008). Isotope labelling studies have revealed that a C3 carbon unit composed of either C1-C2-C3 or C4-C5-C6 (pyruvaldehyde or glyceraldehyde), from D-glucose and a
C2 unit either from amino acid or from glucose (C1-C2) are involved in a chain elongation reaction to generate 2,3-pentanedione in alanine/glucose model system. The
3,4-hexanedione and 1,2-butanedione on the other hand, although their involvement in the formation of pyrazines was predicted in model systems, to our knowledge the mechanism of their formation has not been reported. In this study the origin of 3,4- hexanedione and 1,2-butanedione was explored using isotope labelling techniques.
Similar to other α-dicarbonyl compounds, 3,4-hexanedione has very strong odor properties described as buttery, cooked and caramel (Mosciano, 2007). It has been
128 identified in different model systems as a free dicarbonyl compound (Salter et al., 1989;
Zheng et al., 1997; Whitfield & Mottram, 1999; Ames et al., 2001), in bread crust
(Chiavaro et al., 2007) or integrated into pyrazine moiety through Strecker reaction
(Rizzi, 1972).
7.2 Experimental Procedures
7.2.1 Materials and Reagents
L-Alanine (99%), D-glucose (99.5%), glyoxal trimer dihydrate (95%), acetal (99%) were purchased from Sigma-Aldrich Chemical Co (Oakville, ON, Canada). The labelled [13C-
1]alanine (98%), [13C-2]alanine (99%), [13C-3]alanine (99%), [15N]alanine (98%) were purchased from Cambridge Isotope Laboratories (Andover, MI).
7.2.2 Pyrolysis-Gas Chromatography-Mass Spectrometry (Py-GC/MS) A Varian CP-
3800 GC equipped with a sample pre-concentration trap (SPT) filled with Tenax GR was coupled to a Varian Saturn 2000 Mass Spectrometry detector (Varian, Walnut Creek,
USA). The pyrolysis unit included a valved interface (CDS 1500), which was installed onto the GC injection port and was also connected to a CDS Pyroprobe 2000 unit (CDS
Analytical, Oxford, USA). The sample separations were carried out on a DB-5MS column (5% diphenyl, 95% dimethyl-polysiloxane) with column dimension of 50m length x 0.2mm internal diameter x 33μm film thickness (J&W Scientific, ON). One milligram of sample mixture containing 1:1 or 1:3 molar ration of sugar to alanine in the presence of silica was packed inside a quartz tubes (0.3mm thickness), plugged with quartz wool, and inserted inside the coil probe and pyrolyzed at 250oC for 20 seconds.
The volatiles after pyrolysis were concentrated on the sample pre-concentration trap
(SPT), trapped at 50oC and subsequently directed towards the GC column for separation.
129 The GC column flow rate was regulated by an Electronic Flow Controller (EFC) and set at pressure pulse of 70 PSI for first 4 minutes and maintained with a constant flow of
1.5mL/minute for the rest of the run. The GC oven temperature was set at –5oC for first 5
o minutes using CO2 as the cryogenic cooling source and then increased to 50 C at a rate of
50oC/minute. Then, the oven temperature was again increased to 270o C at a rate of
8oC/minute and kept at 270oC for 5 minutes. The samples were detected by using an ion- trap mass spectrometer. The MS transfer-line temperature was set at 250oC, manifold temperature at 50oC and ion trap temperature at 175oC. The ionization voltage of 70eV was used and EMV was set at 1500V. The compounds were identified using NIST 05 mass spectral libraries.7.2.3 Oxidative Py-GC/MS
Pyrolysis under air was achieved through modification of the above-mentioned GC to allow gas stream switching and subsequent isolation of the pyrolysis chamber from the analytical stream. The pyrolysates generated under air or in the presence of moisture were initially collected onto the trap, which retained the organic volatiles and vented the carrier gas (air) and/or moisture. The trap was subsequently flushed with helium and heated to desorb the collected volatiles. For detailed description see (Yaylayan et al., 2005).
7.2.4 Identification of pyrazines
Pyrazines were identified by their retention times where possible and through NIST library matches in addition to the labelling data (see Tables 7.1 & 7.2).
7.3 Results and discussion
Identification of the origin and the structure of α-dicarbonyl compounds in model systems can be achieved through isotope labelling studies using known precursors such as sugars
130 Table 7.1 Pyrazine and ethyl-substituted pyrazines detected during pyrolysis of glyoxal/alanine model system Compounds M.W. Structure Retention Confirmation1 Abundance Time (min) Pyrazine2 80 N 10.78 NIST 28%
N 2,3 CH 2-ethyl-pyrazine 108 3 14.34 NIST & 46% N Retention time 14.42 N 2-6-diethyl- 136 CH3 CH3 17.61 NIST 18% pyrazine2,4 N
N 2-3-diethyl- 136 CH3 17.69 NIST & 6% pyrazine2 N Retention time
N 17.74
CH3 2,3,5-triethyl- 164 CH3 CH3 20.5 2% pyrazine2,3,5,6 N
N
CH3 Tetraethyl- 192 CH3 CH3 22.4 NIST 0.06% pyrazine4,7 N
N
CH3 CH3 1 Table 7.2 below also indicates the presence of two nitrogen atoms in all the pyrazines 2 (Kinlin et al., 1972) 3 (Fadel & Hegazy, 1993) 4 (Baltes & Bochmann, 1987b) 5 (Vitzthum et al., 1975) 6 (Mottram, 1985) 7 (Rizzi, 1972)
Table 7.2 The number of isotopic atoms incorporated1 in pyrazines generated from glyoxal/alanine model systems2 Compounds [13C-1’]ala [13C-2’]-ala [13C-3’]ala [15N]ala Pyrazine 0 0 0 2 2-ethyl-pyrazine 0 1 1 2 2,6-diethyl-pyrazine 0 2 2 2 2,3-diethyl-pyrazine 0 2 2 2 2,3,5-triethyl-pyrazine 0 3 3 2 Tetraethyl-pyrazine 0 4 4 2 1The number of atoms indicated were incorporated 100 ± 5% 2From four separate experiments using glyoxal/[15N]alanine, glyoxal/[13C-1’]-alanine glyoxal/[13C-2’]-alanine, glyoxal/[13C-3’]-alanine
131 and amino acids and analyzing for label incorporation in either free α-dicarbonyl compounds formed or in their derivatives such as quinoxalines and pyrazines. A derivatizing agent is needed for the formation of quinoxalines whereas pyrazines can be considered as “intrinsic derivatives” since their formation does not require the use of a reagent that may interfere with the normal course of the reaction. For example, the presence of 3,4-hexanedione can be inferred in alanine/glucose model system if either 2,3-diethyl-5-methylpyrazine or 2,3,5-triethyl-6- methylpyrazine is detected, similarly, the presence of 1,2-butanedione can be inferred if
2-ethyl-5,6-dimethylpyrazine is detected (see Figures 1 and 2). Although through
Strecker reaction the origin and the identity of α-dicarbonyl compounds can be traced back to the structure of pyrazines, however, the possibility exists that at the tetrahydropyrazine stage the enamine moiety can add the Strecker aldehyde or any other aldehyde formed in the system to its ring (see Figure 7.3) and following dehydration can generate a pyrazine through a non-oxidative pathway with one additional alkyl substituent on the pyrazine ring characteristic of the structure of the reacting aldehyde
(Amrani-Hemaimi et al., 1995; Adams et al., 2008; Adams & De Kimpe, 2009). In such cases only tetrahydropyrazine structure can reveal the identity of the α-dicarbonyl species formed in the model system. However, pyrazines formed through such non-oxidative pathway can also be used as derivatives of α-dicarbonyl compounds if the composition of the aldehydes is known in the model system or the added aldehyde can be distinguished through labelling techniques. In order for pyrazines therefore to serve as useful derivatives of α-dicarbonyl compounds and hence be used to trace their origin in model systems, their formation mechanism in a model system should be elucidated.
132
Figure 7.1 Ethyl-substituted pyrazines identified in glucose/alanine model system and percent incorporation of alanine C-2’ and C-3’ atoms.
133
Figure 7.2 The α-dicarbonyl precursors required for oxidative and non-oxidative pyrazine formation in glucose/alanine model system
134 In glucose/alanine model system, the reaction generated 2,3-diethyl-5-methylpyrazine,
2,3,5-triethyl-6-methylpyrazine and 2-ethyl-5,6-dimethylpyrazine (Figure 7.1), among others. Label incorporation studies have indicated that alanine played a major role in providing the two carbon atoms of the ethyl groups (see Figure 7.1). The 2-ethyl-5,6- dimethylpyrazine exhibited 60% incorporation of alanine carbon atoms and the 2,3- diethyl-5-methylpyrazine or 2,3,5-triethyl-6-methylpyrazine incorporated at least one ethyl group from alanine. The origin of the ethyl groups was confirmed through analysis of the presence or absence of label in the mass fragment generated by the loss of ethyl radical such as formation of ion at m/z 107 in 2-ethyl-5,6-dimethylpyrazine (see Figure
7.1).
The presence of the above mentioned pyrazines in the glucose/alanine model system may indicate the formation of 1,2-butanedione and 3,4-hexanedione in addition to other required α-dicarbonyls such as 2,3-butanedione, 2,3-pentanedione and pyruvaldehyde. As mentioned above, the structure of a given pyrazine may not necessarily allow deduction of the type of α-dicarbonyl compound needed to give rise to its formation due to the possibility of non-oxidative pathway as shown in Figure 7.3. However, one limitation of this pathway, is the fact that a given tetrahydropyrazine moiety can add only one aldehyde at a time to form a pyrazine with one additional alkyl group. For example, the tetrahydropyrazine I in Figure 7.3 can only form ethylpyrazine through non-oxidative pathway and not diethylpyrazine, as a result, the pyrazines listed in Figure 7.1 with multiple ethyl groups originating from alanine can only arise from preformed 1,2- butanedione and 3,4-hexanedione molecules irrespective of their method of formation
(oxidative vs non-oxidative) as illustrated in Figure 7.2. Furthermore, due to the
135 incorporation of a significant percentage of alanine carbon atoms into the pyrazines, the above α-dicarbonyls should arise through chain elongation processes (Yaylayan &
Keyhani, 1999) of smaller sugar-derived α-dicarbonyl compounds with the assistance of alanine.
7.3.1 Chain elongation reactions of α-dicarbonyl compounds
To justify the label incorporation patterns of the pyrazines shown in Figure 7.1, the four- carbon α-dicarbonyl compound the 1,2-butanedione should have two of its carbon atoms originating from alanine and the six-carbon α-dicarbonyl compound the 3,4-hexanedione should have four of its carbon atoms originating from alanine, thus pointing to glyoxal as the possible sugar-derived α-dicarbonyl precursor needed for the conversion into four- carbon and six-carbon analogues through chain elongation reactions involving alanine.
When alanine was reacted with glyoxal, in addition to the parent pyrazine many ethyl- substituted pyrazines (see Tables 1 and 2), pyrroles, pyridines and pyrazinones were also detected (see Figure 7.4), the major group being the pyrazine derivatives. Since glyoxal, the only carbohydrate source in the model system, can only generate the parent pyrazine through the Strecker reaction, all other substituted pyrazines in glyoxal alanine model system should therefore arise either through chain elongation reactions of glyoxal with alanine to form longer α-dicarbonyl compounds as was demonstrated before in the case of 2,3-pentanedione (Yaylayan & Keyhani, 1999) or through aldol condensation of acetaldehyde (the Strecker aldehyde) to glycolaldehyde to generate the same precursors
(see Figure 7.5).
136
Figure 7.3 Proposed oxidative and non-oxidative mechanisms of pyrazine formation. SR = Strecker reaction; [O] = oxidation.
Although in glyoxal/alanine model system, some glycolaldehyde may be formed, however the major contribution to the formation of substituted pyrazines should come from glyoxal the major constituent of the model system. In addition, spiking glyoxal/alanine model with glycolaldehyde (20 % relative to glyoxal content) did not increase the intensities of the pyrazine peaks, just the contrary they decreased indicating its preference to react with alanine to form Amadori intermediate rather than undergo aldol condensation with acetaldehyde. According to Table 7.2 all the carbon atoms of the ethyl groups of all the detected pyrazines originated 100% from C-2’ and C-3’ atoms of alanine. This pattern of isotopic substitution is consistent with both mechanisms proposed
137 in Figure 7.5. Although the addition of glycolaldehyde to the model system did not increase the intensities of pyrazines, however, to further confirm that aldol addition of acetaldehyde to glycolaldehyde does not contribute significantly to the formation of 1,2- butanedione, the [13C-3’]-alanine/glyoxal and [13C-2’]-alanine/glyoxal model systems were also analyzed in the presence of increasing concentrations of unlabelled acetaldehyde to estimate the percentage of acetaldehyde incorporation into the structure of pyrazines through aldol reaction. Adding a molar or less than a molar equivalent of acetaldehyde did not alter at all the isotopic distribution patterns of the pyrazines, however, higher concentrations caused 30-50% incorporation of mainly single acetaldehyde units into the pyrazine structures shown in Table 7.1. This observation indicates that aldol reaction of acetaldehyde does not contribute significantly to the formation of pyrazines but it points mainly to the occurrence of non-oxidative mechanism of pyrazine formation as illustrated in Figure 7.2 and only under unusually high concentrations of the aldehydes.
When alanine/glyoxal model system was pyrolyzed under air instead of helium, five-fold increase in the total intensities of the pyrazine peaks was observed indicating the importance of the oxidative pathway. Furthermore, as in the case of pyrazines identified in glucose/alanine system, the pyrazines detected in glyoxal/alanine model system, similarly require 1,2-butanedione and 3,4-hexanedione as preformed α-dicarbonyl compounds irrespective of their mechanism of formation as illustrated in Figure 7.6.
138
Figure 7.4 Chromatogram generated at 210oC by the pyrolysis of glyoxal/alanine model system.
Figure 7.5 Proposed mechanisms of formation of 1,2-butanedione and 3,4-hexandione. CHE = chain elongation, aldol = aldol condensation
139
Figure 7.6 The α-dicarbonyl precursors required for oxidative and non-oxidative pyrazine formation in glyoxal/alanine model system
140 7.3.2 How alanine is involved in the chain elongation process of glyoxal?
Similar to the mechanism proposed earlier (Yaylayan & Keyhani, 1999), alanine can interact with the aldehyde end of the α-dicarbonyl as a C-nucleophile through its α- carbon (Kaneko et al., 1974; Barton and Ollis, 1979) and undergoes aldol type addition with simultaneous decarboxylation reaction as shown in Figure 7.5. The resulting adduct produces 1,2-butanedione after a deamination step. In turn, the 1,2-butanedione undergoes similar alanine-assisted chain elongation to form 3,4-hexanedione, thus adding another set of two carbon atoms from alanine. Perhaps one of the reasons for the efficiency of this type of amino acid interactions with smaller and terminal α-dicarbonyl compounds such as glyoxal and pyruvaldehyde is the possibility and the ease of formation of a six-membered bicyclic transition state (see Figure 7.7) involving two hydrogen bonds to the trans configuration of the α-dicarbonyl moiety; one with the carboxylic acid hydrogen and the other with the hydrogen on the amino group (Figure
7.7). After simultaneous aldol addition and decarboxylation, the resulting 3-amino-2- hydroxycarbonyl compound can undergo deamination and generate the new α-dicarbonyl compound incorporating the carbon atoms of the amino acid. Chain elongation assisted by alanine requires multiple additions to glyoxal to generate longer chain dicarbonyls such as 3,4-hexanedione required for the formation of 2,3-diethyl, 2,3,5-triethyl and tetraethylpyrazines. Consequently increasing the ratio of alanine relative to glyoxal should enhance the peak areas associated with the above pyrazines.
141
Figure 7.7 Proposed bi-cyclic transition state for the simultaneous aldol addition and decarboxylation reaction leading to chain elongation of simple α-dicarbonyl compounds
142 Chapter 8: General conclusion and contribution to knowledge
OH O O OH R R O OH R O + O N N OH NH H OH 2 OH OH OH -CO2
R R R + + N N N OH OH H OH -CO2 OH OH OH
Strecker Degradation
O NH2 R OH R OH
NH O 2 OH OH
143 8.1 General conclusions
Alanine and phenylalanine based model systems were used to elucidate selected Maillard reaction pathways by isotope labelling and pyrolysis gas chromatography/mass spectrometry. Both mass spectrometry and FTIR data were used to propose suitable mechanisms for the formation of 5-oxazolidinone and its two decarboxylated isomeric imines in both alanine and phenylalanine model systems. The hydrolysis of these two imines in both model systems can generate various reactive precursors that so far remained unknown. In the alanine model system, the two imines hydrolysed into ethylamine and acetaldehyde and contributed to the formation of pyrazines, pyrazinones, pyrroles, pyridines or imidazolidinone depending on the initial sugar components. In phenylalanine model system, the two hydrolysed imines generated phenethylamine and phenylacetaldehyde that participated in the formation of benzaldehyde, 3-phenyl-pyridine and various other imine adducts. On the other hand, in the study on the role of oxygen and water in the Maillard reaction, it was found that molecular oxygen can influence carbon-carbon bond cleavage through the formation and degradation of 1,2-dioxetane moieties formed from enol structures abundant in Maillard reaction systems such as the enol form of phenylacetaldehyde generated from phenylalanine. To identify the precursors involved in its generation during Maillard reaction, various model systems containing phenylalanine, phenylpyruvic acid, phenethylamine, or phenylacetaldehyde were studied in the presence and absence of moisture using oxidative and non-oxidative
Py-GC/MS. Analysis of the data indicated that phenylacetaldehyde was the most effective precursor of benzaldehyde in which both air and water significantly enhanced the rate of its formation. Phenylpyruvic acid was the most efficient precursor under non-
144 oxidative conditions whereas phenethylamine needed the presence of a carbonyl compounds to generate benzaldehyde under oxidative conditions. Based on the FTIR results, a free radical initiated oxidative cleavage of the carbon-carbon double bond of the enolized phenylacetaldehyde was proposed as a possible major mechanism for benzaldehyde formation. Furthermore, the origin and formation mechanism of α- dicarbonyl reactive intermediates and their role in the generation of various heterocyclic compounds such as pyrazines and pyrazinones were investigated using alanine/glucose as a model system. The isotope labelling studies have indicated the occurrence of a chain elongation process of small sugar-derived α-dicarbonyl compounds into longer chain containing analogues assisted by the amino acid alanine. Based on the proposed mechanism, the glyoxal interaction with alanine through decarboxylative aldol addition reaction leads to the formation of 1,2-butanedione with the two terminal carbon atoms originating from C-2’ and C-3’ atoms of alanine. Similarly, interaction of 1,2- butanedione with a second molecule of alanine leads to the formation of 3,4-hexanedione with both terminal pairs of ethyl carbon atoms originating from C-2’ and C-3’ atoms of alanine. This thesis demonstrates for the first time the formation of 5-oxazolidinone in the early phase of the Maillard reaction and its importance as well as that of α-dicarbonyl compounds in the formation of various Maillard reaction products through the use of isotope labelling and pyrolysis gas chromatography/mass spectrometry based techniques.
8.2 Contribution to knowledge
Through its different chapters, the work carried out in this thesis provides for the first time:
145 1. Evidence for the formation and decomposition of 5-oxazolidinone intermediate in phenylalanine model system using spectroscopic data from Py-GC/MS and FTIR.
2. Evidence for the ability of 5-oxazolidinone intermediate to undergo decarboxylation and formation of two isomeric imines through azomethine ylide formation using both phenylalanine and alanine sugar model systems.
3. The application of oxidative-pyrolysis technique to study effect of oxygen and water in
Maillard reaction through the conversion of phenylacetaldehyde into benzaldehyde.
4. A detailed mechanistic study of benzaldehyde formation from various precursors such as phenylalanine, phenylpyruvic acid, phenethylamine and phenylacetaldehyde, indicating different pathways of its formation including oxidative decarboxylation, thermal decarboxylation and Strecker degradation.
5. Evidence for the role of molecular oxygen in cleaving carbon-carbon bonds through the formation and degradation of 1,2-dioxetane moieties generated from enol structures abundant in Maillard reaction systems such as the enol form of phenylacetaldehyde generated from phenylalanine.
6. The detailed analysis of mechanistic pathways of formations of heterocyclic compounds such as pyrazines, pyrazinones, pyrroles and pyridines in alanine/sugar model system.
7. Identification and elucidation of the mechanism of formation of (2R) and (2S)-2,5- dimethyl-3-ethyl-4H-imidazolidinone in alanine/sugar model system
8. A detailed mechanistic pathway for the generation of 1,2-butanedione and 3,4- hexanedione in alanine/sugar model system through identification of a chain elongation
146 process by which small sugar-derived α-dicarbonyl compounds can be converted into longer chain containing analogues assisted by the amino acid.
147 Reference
Aaslyng, M.D., Elmore, J.S., and Mottram, D.S. 1998. Comparison of the Aroma Characteristics of Acid-Hydrolyzed and Enzyme-Hydrolyzed Vegetable Proteins Produced from Soy. Journal of Agricultural and Food Chemistry, 46[12], pp. 5225-5231.
Abdel-Mageed, M.A. and Fadel, H.H.M. 1995. Effect of freeze storage on the volatiles of butter. Nahrung, 39[4], pp. 316-322.
Adams, A.N., Polizzi, V., van Boekel, M., and De Kimpe, N. 2008. Formation of Pyrazines and a Novel Pyrrole in Maillard Model Systems of 1,3-Dihydroxyacetone and 2-Oxopropanal. Journal of Agricultural and Food Chemistry, 56[6], pp. 2147-2153.
Adams, A., Borrelli, R.C., Fogliano, V., and De Kimpe, N. 2005. Thermal degradation studies of food melanoidins. Journal of Agricultural and Food Chemistry, 53[10], pp. 4136-4142.
Adams, A. and De Kimpe, N. 2009. Formation of pyrazines from ascorbic acid and amino acids under dry-roasting conditions. Food Chemistry, 115[4], pp. 1417-1423.
Ames, J.M. 2005. Application of semiquantitative proteomics techniques to the maillard reaction. Annals of the New York Academy of Sciences, 1043, pp. 225-235.
Ames, J.M. 1998. Applications of the Maillard reaction in the food industry. Food Chemistry, 62[4], pp. 431-439.
Ames, J.M., Arnoldi, A., Bates, L., and Negroni, M. 1997. Analysis of the Methanol- Extractable Nonvolatile Maillard Reaction Products of a Model Extrusion-Cooked Cereal Product. Journal of Agricultural and Food Chemistry, 45[4], pp. 1256-1263.
Ames, J.M., Bailey, R.G., and Mann, J. 1999. Analysis of Furanone, Pyranone, and New Heterocyclic Coloured Compounds from Sugar-Glycine Model Maillard Systems. Journal of Agricultural and Food Chemistry, 47[2], pp. 438-443.
Ames, J.M., Dafaye, A.B., Bailey, R.G., and Bates, L. 1998. Analysis of the non-volatile Maillard reaction products formed in an extrusion-cooked model food system. Food Chemistry, 61[4], pp. 521-524.
Ames, J.M., Defaye, A.B., and Bates, L. 1996. The effect of pH on the volatiles formed in an extruded starch-glucose-lysine model system. Food Chemistry, 58[4], pp. 323-327.
148 Ames, J.M., Guy, R.C.E., and Kipping, G.J. 2001. Effect of pH, Temperature, and Moisture on the Formation of Volatile Compounds in Glycine/Glucose Model Systems. Journal of Agricultural and Food Chemistry, 49[9], pp. 4315-4323.
Amrani-Hemaimi, M., Cerny, C., and Fay, L.B. 1995. Mechanisms of Formation of Alkylpyrazines in the Maillard Reaction. Journal of Agricultural and Food Chemistry, 43[11], pp. 2818-2822.
Arnoldi, A., Corain, E.A., Scaglioni, L., and Ames, J.M. 1997. New Coloured Compounds from the Maillard Reaction between Xylose and Lysine. Journal of Agricultural and Food Chemistry, 45[3], pp. 650-655.
Aurelio, L., Box, J.S., Brownlee, R.T.C., Hughes, A.B., and Sleebs, M.M. 2003. An Efficient Synthesis of N-Methyl Amino Acids by Way of Intermediate 5-Oxazolidinones. Journal of Organic Chemistry, 68[7], pp. 2652-2667.
Bailey, R.G., Ames, J.M., and Monti, S.M. 1996. An analysis of the non-volatile reaction products of aqueous Maillard model systems at pH 5, using reversed-phase HPLC with diode-array detection. Journal of the Science of Food and Agriculture, 72[1], pp. 97-103.
Baker, G.B., Wong, J.T.F., Coutts, R.T., and Pasutto, F.M. 1987. Simultaneous extraction and quantitation of several bioactive amines in cheese and chocolate. Journal of Chromatography, 392, pp. 317-331.
Balci, M., Taskesenligil, Y., and Harmandar, M. 1989. A novel dioxetane-hydroperoxide rearrangement: photooxygenation of 3-tert-butoxy- (and -methoxy)-6,7- benzobicyclo[3.2.1]octa-2,6-dienes. Tetrahedron Letters, 30[25], pp. 3339-3342.
Baltes, W. and Bochmann, G. 1987a. Model reactions on roast aroma formation. 1. Reaction of serine and threonine with sucrose under the conditions on coffee roasting and identification of new coffee aroma compounds. Journal of Agricultural and Food Chemistry, 35[3], pp. 340-346.
Baltes, W. and Bochmann, G. 1987b. Model reactions on roast aroma formation. IV. Mass spectrometric identification of pyrazines from the reaction of serine and threonine with sucrose under the conditions of coffee roasting. Zeitschrift für Lebensmitteluntersuchung und -Forschung, 184[6], pp. 485-493.
Baltes, W. and Mevissen, L. 1988. Model reactions on roast aroma formation. VI. Volatile reaction products from the reaction of phenylalanine with glucose during cooking and roasting. Zeitschrift für Lebensmitteluntersuchung und -Forschung, 187[3], p. 209-214.
149 Barton, D. H.; Ollis, W. D. (Eds.) 1979. Comprehensive Organic Chemistry: the synthesis and reactions of organic compounds. Pergamon Press: Oxford, New York, Vol. 2, p 825.
Bemis-Young, G.L., Huang, J., and Bernhard, R.A. 1993. Effect of pH on pyrazine formation in glucose-glycine model systems. Food Chemistry, 46[4], pp. 383-387.
Bianchi, F., Careri, M., Chiavaro, E., Musci, M., and Vittadini, E. 2008. Gas chromatographic-mass spectrometric characterisation of the Italian Protected Designation of Origin "Altamura" bread volatile profile. Food Chemistry, 110[3], pp. 787-793.
Bishop, J.C., Cross, S.D., and Waddell, T.G. 1997. Prebiotic transamination. Origins of Life and Evolution of the Biosphere, 27[4], pp. 319-324.
Bondarovich, H.A., Friedel, P., Krampl, V., Renner, J.A., Shephard, F.W., and Gianturco, M.A. 1967. Volatile constituents of coffee. Pyrazines and other compounds. Journal of Agricultural and Food Chemistry, 15[6], pp. 1093-1099.
Bravo, A., Herrera, J.C., Scherer, E., Ju-Nam, Y., Rubsam, H., Madrid, J., Zufall, C., and Rangel-Aldao, R. 2008. Formation of alpha-dicarbonyl compounds in beer during storage of Pilsner. Journal of Agricultural and Food Chemistry, 56[11], pp. 4134-4144.
Brinkman, H.W., Copier, H., De Leuw, J.J.M., and Tjan, S.B. 1972. Components contributing to beef flavour. Analysis of the headspace volatiles of beef broth. Journal of Agricultural and Food Chemistry, 20[2], pp. 177-181.
Cainelli, G., Giacomini, D., Trere, A., and Boyl, P.P. 1996. Efficient Transamination under Mild Conditions: Preparation of Primary Amine Derivatives from Carbonyl Compounds via Imine Isomerization with Catalytic Amounts of Potassium tert-Butoxide. Journal of Organic Chemistry, 61[15], pp. 5134-5139.
Cerny, C. 2008. The aroma side of the Maillard reaction. Annals of the New York Academy of Sciences, 1126, pp. 66-71.
Cerny, C. and Davidek, T. 2003. Formation of Aroma Compounds from Ribose and Cysteine during the Maillard Reaction. Journal of Agricultural and Food Chemistry, 51[9], pp. 2714-2721.
Chandra, A. and Nair, M.G. 1993. Quantification of benzaldehyde and its precursors in Montmorency cherry (Prunus cerasus L.) kernels. Phytochemical Analysis, 4[3], pp. 120- 123.
150 Chen, C.W., Lu, G., and Ho, C.T. 1997. Generation of Proline-Specific Maillard Compounds by the Reaction of 2-Deoxyglucose with Proline. Journal of Agricultural and Food Chemistry, 45[8], pp. 2996-2999.
Chen, J. and Ho, C.T. 1999. Comparison of Volatile Generation in Serine/Threonine/Glutamine-Ribose/Glucose/Fructose Model Systems. Journal of Agricultural and Food Chemistry, 47[2], pp. 643-647.
Chen, S.L., Jin, S.Y., and Chen, C.S. 2005. Relative reactivities of glucose and galactose in browning and pyruvaldehyde formation in sugar/glycine model systems. Food Chemistry, 92[4], pp. 597-605.
Chiavaro, E., Vittadini, E., Musci, M., Bianchi, F., and Curti, E. 2007. Shelf-life stability of artisanally and industrially produced durum wheat sourdough bread ("Altamura bread"). LWT-Food Science and Technology, 41[1], pp. 58-70.
Chu, F.L. and Yaylayan, V.A. 2008a. Model Studies on the Oxygen-Induced Formation of Benzaldehyde from Phenylacetaldehyde Using Pyrolysis GC-MS and FTIR. Journal of Agricultural and Food Chemistry, 56[22], pp. 10697-10704.
Chu, F.L. and Yaylayan, V.A. 2008b. Post-schiff base chemistry of the Maillard reaction. Annals of the New York Academy of Sciences, 1126, pp. 30-37.
Chu, F.L. and Yaylayan, V.A. 2009. FTIR monitoring of oxazolidin-5-one formation and decomposition in a glycolaldehyde-phenylalanine model system by isotope labeling techniques. Carbohydrate Research, 344[2], pp. 229-236.
Clifford, M.N. 1985. Coffee: Botany, Biochemistry, and Production of Beans and Beverage, Editors: Clifford, M.N. and Willson, K.C. Croom Helm, London, UK. pp. 457.
Cram, D.J. and Guthrie, R.D. 1966. Electrophilic substitution at saturated carbon. XXVII. Carbanions as intermediates in the base-catalyzed methylene-azomethine rearrangement. Journal of the American Chemical Society, 88[24], pp. 5760-5765.
Davidek, J., Velisek, J., Pokorny, J 1990. Chemical Changes During Food Processing. Elsevier. New York, NY. pp. 138
Davidek, T., Clety, N., Devaud, S., Robert, F., and Blank, I. 2003. Simultaneous Quantitative Analysis of Maillard Reaction Precursors and Products by High- Performance Anion Exchange Chromatography. Journal of Agricultural and Food Chemistry, 51[25], pp. 7259-7265.
151 de Sa, P.F.G., Treubig, J.M., Brown, P.R., and Dain, J.A. 2001. The use of capillary electrophoresis to monitor Maillard reaction products (MRP) by glyceraldehyde and the epsilon amino group of lysine. Food Chemistry, 72[3], pp. 379-384.
Duckham, S.C., Dodson, A.T., Bakker, J., and Ames, J.M. 2002. Effect of Cultivar and Storage Time on the Volatile Flavour Components of Baked Potato. Journal of Agricultural and Food Chemistry, 50[20], pp. 5640-5648.
Ehling, S. and Shibamoto, T. 2005. Correlation of Acrylamide Generation in Thermally Processed Model Systems of Asparagine and Glucose with Colour Formation, Amounts of Pyrazines Formed, and Antioxidative Properties of Extracts. Journal of Agricultural and Food Chemistry, 53[12], pp. 4813-4819.
Fadel, H.H.M. and Farouk, A. 2002. Caramelization of maltose solution in presence of alanine. Amino Acids, 22[2], pp. 199-213.
Fadel, H.H.M. and Hegazy, N.A. 1993. Improvement and stabilization of bread flavour. Nahrung, 37[4], pp. 386-394.
Farouk, A., Pudil, F., Janda, V., and Pokorny, J. 2000. Effect of amino acids on the composition and properties of extruded mixtures of wheat flour and glucose. Nahrung, 44[3], pp. 188-192.
Fay, L.B. and Brevard, H. 2005. Contribution of mass spectrometry to the study of the maillard reaction in food. Mass Spectrometry Reviews, 24[4], pp. 487-507.
Fayle, S.E. and Gerrard, J.A. 2002. The Maillard Reaction. Editor: Belton, P.S. Royal Society of Chemistry ,Cambrigde, U.K., pp. 120
Fayle, S.E., Healy, J.P., Brown, P.A., Reid, E.A., Gerrard, J.A., and Ames, J.M. 2001. Novel approaches to the analysis of the Maillard reaction of proteins. Electrophoresis, 22[8], pp. 1518-1525.
Ferrer, E., Alegria, A., Farre, R., Abellan, P., Romero, F., and Clemente, G. 2003. Evolution of available lysine and furosine contents in milk-based infant formulas throughout the shelf-life storage period. Journal of the Science of Food and Agriculture, 83[5], pp. 465-472.
Frank, O. and Hofmann, T. 2000. Characterization of Key Chromophores Formed by Nonenzymatic Browning of Hexoses and L-Alanine by Using the Colour Activity Concept. Journal of Agricultural and Food Chemistry, 48[12], pp. 6303-6311.
152 Gerrard, J.A., Brown, P.K., and Fayle, S.E. 2002. Maillard crosslinking of food proteins III: the effects of glutaraldehyde, formaldehyde and glyceraldehyde upon bread and croissants. Food Chemistry, 80[1], pp. 45-50.
Gollnick, K. and Knutzen-Mies, K. 1991. Dye-sensitized photooxygenation of 2,3- dihydrofurans: competing [2 + 2] cycloadditions and ene reactions of singlet oxygen with a rigid cyclic enol ether system. Journal of Organic Chemistry, 56[12], pp. 4017-4027.
Granvogl, M., Bugan, S., and Schieberle, P. 2006. Formation of Amines and Aldehydes from Parent Amino Acids during Thermal Processing of Cocoa and Model Systems: New Insights into Pathways of the Strecker Reaction. Journal of Agricultural and Food Chemistry, 54[5], pp. 1730-1739.
Haffenden, L.J.W. and Yaylayan, V.A. 2008. Nonvolatile oxidation products of glucose in Maillard model systems: Formation of saccharinic and aldonic acids and their corresponding lactones. Journal of Agricultural and Food Chemistry, 56[5], pp. 1638- 1643.
Hao, Z., Lu, C.y., Xiao, B., Weng, N., Parker, B., Knapp, M., and Ho, C.T. 2007. Separation of amino acids, peptides and corresponding Amadori compounds on a silica column at elevated temperature. Journal of Chromatography A, 1147[2], pp. 165-171.
Herbst, R.M. and Engel, L.L. 1934. A reaction between alpha -ketonic acids and alpha - amino acids. Journal of Biological Chemistry, 107, pp. 505-512.
Hidalgo, F.J., Alaiz, M., and Zamora, R. 1999. Effect of pH and temperature on comparative nonenzymatic browning of proteins produced by oxidized lipids and carbohydrates. Journal of Agricultural and Food Chemistry, 47[2], pp. 742-747.
Hodge, J.E. 1953. Browning reactions in model systems. Journal of Agricultural and Food Chemistry, 1, pp. 928-943.
Hofmann, T. 1998a. Application of Site Specific 13C Enrichment and 13C NMR Spectroscopy for the Elucidation of the Formation Pathway Leading to a Red 1H-Pyrrol- 3(2H)-one during the Maillard Reaction of Furan-2-carboxaldehyde and L-Alanine. Journal of Agricultural and Food Chemistry, 46[3], pp. 941-945.
Hofmann, T. 1998b. Characterization of the Chemical Structure of Novel Coloured Maillard Reaction Products from Furan-2-carboxaldehyde and Amino Acids. Journal of Agricultural and Food Chemistry, 46[3], pp. 932-940.
153 Hofmann, T. 1998c. Characterization of the most intense coloured compounds from Maillard reactions of pentoses by application of colour dilution analysis. Carbohydrate Research, 313[3-4], pp. 203-213.
Høltermand, Arvid. 1966. Browning reaction. Staerke, 18[10], pp. 319-328.
Horvat, S. and Jakas, A. 2004. Peptide and amino acid glycation: new insights into the Maillard reaction. Journal of Peptide Science, 10[3], pp. 119-137.
Horvat, S., Roscic, M., and Horvat, J. 1999. Synthesis of hexose-related imidazolidinones: novel glycation products in the Maillard reaction. Glycoconjugate Journal, 16[8], pp. 391-398.
Horvat, S., Varga-Defterdarovic, L., and Horvat, J. 1998. Synthesis of novel imidazolidinones from hexose-peptide adducts: model studies of the Maillard reaction with possible significance in protein glycation. Chemical Communications (Cambridge), [16], pp. 1663-1664.
Huyghues-Despointes, A. and Yaylayan, V.A. 1996. Retro-Aldol and Redox Reactions of Amadori Compounds: Mechanistic Studies with Variously Labeled D-[13C]Glucose. Journal of Agricultural and Food Chemistry, 44[3], pp. 672-681.
Huyghues-Despointes, A., Yaylayan, V.A., and Keyhani, A. 1994. Pyrolysis/GC/MS Analysis of 1-[(2'-carboxyl)pyrrolidinyl]-1-deoxy-D-fructose (Proline Amadori Compound). Journal of Agricultural and Food Chemistry, 42[11], pp. 2519-2524.
Hwang, H.I., Hartman, T.G., and Ho, C.T. 1995. Relative Reactivities of Amino Acids in the Formation of Pyridines, Pyrroles, and Oxazoles. Journal of Agricultural and Food Chemistry, 43[11], pp. 2917-2921.
Janes, D., Kantar, D., Kreft, S., and Prosen, H. 2008. Identification of buckwheat (Fagopyrum esculentum Moench) aroma compounds with GC-MS. Food Chemistry, 112[1], pp. 120-124.
Kaneda, K., Itoh, T., Kii, N., Jitsukawa, K., and Teranishi, S. 1982. Oxygenation of enamines using copper catalysts. Journal of Molecular Catalysis, 15[3], pp. 349-365.
Kaneko, T.; Izumi, Y.; Chibata I.; Itoh, T. (Eds.) 1974. Synthetic production and utilization of amino acids. Wiley: New York, p 196-198.
154 Kasper, M. and Schieberle, P. 2005. Labeling studies on the formation pathway of Ne- carboxymethyllysine in Maillard-type reactions. Annals of the New York Academy of Sciences, 1043, pp. 59-62.
Kato, Y. 2005. Influence of butter and/or vegetable oil on flavours of roux prepared from wheat flour and fat/oil. Food Science and Technology Research, 11[3], pp. 278-287.
Keyhani, A. and Yaylayan, V.A. 1996a. Elucidation of the Mechanism of Pyrazinone Formation in Glycine Model Systems Using Labeled Sugars and Amino Acids. Journal of Agricultural and Food Chemistry, 44[9], pp. 2511-2516.
Keyhani, A. and Yaylayan, V.A. 1996b. Pyrolysis/GC/MS Analysis of N-(1-Deoxy-D- fructos-1-yl)-L-phenyl- alanine: Identification of Novel Pyridine and Naphthalene Derivatives. Journal of Agricultural and Food Chemistry, 44[1], pp. 223-229.
Kinlin, T.E., Muralidhara, R., Pittet, A.O., Sanderson, A., and Walradt, J.P. 1972. Volatile components of roasted filberts. Journal of Agricultural and Food Chemistry, 20[5], pp. 1021-1028.
Kojic-Prodic, B., Peric, B., Roscic, M., Novak, P., and Horvat, S. 2004. Novel carbohydrate-induced modification of peptides: crystal structure and NMR analysis of ester-bridged bicyclic galactose-enkephalin adduct containing imidazolidinone moiety. Journal of Peptide Science, 10[1], pp. 47-55.
Lauber, S.N., Ali, S., and Gooderham, N.J. 2004. The cooked food derived carcinogen 2- amino-1-methyl-6-phenylimidazo[4,5-b] pyridine is a potent oestrogen: a mechanistic basis for its tissue-specific carcinogenicity. Carcinogenesis, 25[12], pp. 2509-2517.
Ledl, F. and Schleicher, E. 1990. The Maillard reaction in food and in the human body - new results in chemistry, biochemistry and medicine. Angewandte Chemie, International Edition, 29[6], pp. 565-594.
Lojzova, L., Riddellova, K., Hajslova, J., Zrostlikova, J., Schurek, J., and Cajka, T. 2009. Alternative GC-MS approaches in the analysis of substituted pyrazines and other volatile aromatic compounds formed during Maillard reaction in potato chips. Analytica Chimica Acta, 641[1-2], pp. 101-109.
Low, M.Y., Parker, J.K., and Mottram, D.S. 2007. echanisms of alkylpyrazine formation in a potato model system containing added glycine. Journal of Agricultural and Food Chemistry, 55[10], pp. 4087-4094.
155 Lu, C.y., Hao, Z., Payne, R., and Ho, C.T. 2005. ffects of Water Content on Volatile Generation and Peptide Degradation in the Maillard Reaction of Glycine, Diglycine, and Triglycine. Journal of Agricultural and Food Chemistry, 53[16], pp. 6443-6447.
Maga, J.A. 1982. Pyrazines in foods: an update. Critical Reviews in Food Science and Nutrition, 16[1], pp.1-48.
Maga, J.A. and Sizer, C.E. 1973. Pyrazines in foods. Review. Journal of Agricultural and Food Chemistry, 21[1], pp. 22-30.
Maillard, L.C. 1912. Action of Amino Acids on Sugars. Formation of Melanoidins in a Methodical Way. Comptes rendus hebdomadaires des séances de l'Académie des sciences, 154, pp. 66-68.
March, R.E. 2000. Quadrupole ion trap mass spectrometry. A view at the turn of the century. International Journal of Mass Spectrometry, 200, pp. 285-312.
Martins, S.I.F.S. and van Boekel, M.A.J.S. 2005. Kinetics of the glucose/glycine Maillard reaction pathways: influences of pH and reactant initial concentrations. Food Chemistry, 92[3], pp. 437-448.
Messia, M.C., Caboni, M.F., and Marconi, E. 2005. Storage Stability Assessment of Freeze-Dried Royal Jelly by Furosine Determination. Journal of Agricultural and Food Chemistry, 53[11], pp. 4440-4443.
Monti, S.M., Bailey, R.G., and Ames, J.M. 1998. The influence of pH on the non-volatile reaction products of aqueous Maillard model systems by HPLC with diode array detection. Food Chemistry, 62[3], pp. 369-375.
Mosciano, G. 2007. Organoleptic characteristics of flavour materials. Perfumer & Flavourist, 32[9], pp. 56-58.
Mosciano, G. 2005. Organoleptic characteristics of flavour materials. Perfumer & Flavourist, 30[6], pp. 38-44.
Mottram, D.S. and Edwards, R.A. 1983. The role of triglycerides and phospholipids in the aroma of cooked beef. Journal of the Science of Food and Agriculture, 34[5], pp. 517-522.
Mottram, D.S. 1985. The effect of cooking conditions on the formation of volatile heterocyclic compounds in pork. Journal of the Science of Food and Agriculture, 36[5], pp. 377-382.
156 Mottram, D.S., Wedzicha, B.L., and Dodson, A.T. 2002. Acrylamide is formed in the Maillard reaction. Nature, 419[6906], pp. 448-449. 3-10-
Mussinan, C.J., Wilson, R.A., and Katz, I. 1973. Isolation and identification of pyrazines present in pressure-cooked beef. Journal of Agricultural and Food Chemistry, 21[5], pp. 871-872.
Nursten, H.E. 1981. Recent developments in studies of the Maillard reaction. Food Chemistry, 6[3], pp. 263-277.
Oh, Y.C., Shu, C.K., and Ho, C.T. 1991. Some volatile compounds formed from thermal interaction of glucose with glycine, diglycine, triglycine, and tetraglycine. Journal of Agricultural and Food Chemistry, 39[9], pp. 1553-1554.
Oh, Y.C., Shu, C.K., and Ho, C.T. 1992. Formation of novel 2(1H)-pyrazinones as peptide-specific Maillard reaction products. Journal of Agricultural and Food Chemistry, 40[1], pp. 118-121.
Okrasa, K., Guibe-Jampel, E., Plenkiewicz, J., and Therisod, M. 2004. In vitro bi- enzymatic synthesis of benzaldehyde from phenylalanine: practical and mechanistic studies. Journal of Molecular Catalysis B: Enzymatic, 31[4-6], pp. 97-101.
Oruna-Concha, M.J., Bakker, J., and Ames, J.M. 2002. Comparison of the volatile components of two cultivars of potato cooked by boiling, conventional baking and microwave baking. Journal of the Science of Food and Agriculture, 82[9], pp. 1080- 1087.
Paine, J.B., Pithawalla, Y.B., and Naworal, J.D. 2008. Carbohydrate pyrolysis mechanisms from isotopic labeling: Part 2. The pyrolysis of d-glucose: General disconnective analysis and the formation of C1 and C2 carbonyl compounds by electrocyclic fragmentation mechanisms. Journal of Analytical and Applied Pyrolysis, 82[1], pp. 10-41.
Paine, J.B., Pithawalla, Y.B., Naworal, J.D., and Thomas, C.E. 2007. Carbohydrate pyrolysis mechanisms from isotopic labeling: Part 1. The pyrolysis of glycerin: Discovery of completing fragmentation mechanisms affording acetaldehyde and formaldehyde and the implications for carbohydrate pyrolysis. Journal of Analytical and Applied Pyrolysis, 80[2], pp. 297-311.
Perez Locas, C. and Yaylayan, V.A. 2008. Further insight into thermally and pH-induced generation of acrylamide from glucose/asparagine model systems. Journal of Agricultural and Food Chemistry, 56[15], pp. 6069-6074.
157 Porter, D.J. and Bright, H.J. 1987. The cyanogenic substrate for horseradish peroxidase is a conjugated enamine. Journal of Biological Chemistry, 262[19], pp. 9154-9159.
Pripis-Nicolau, L., De Revel, G., Bertrand, A., and Maujean, A. 2000. Formation of Flavour Components by the Reaction of Amino Acid and Carbonyl Compounds in Mild Conditions. Journal of Agricultural and Food Chemistry, 48[9], pp. 3761-3766.
Qian, M. and Reineccius, G. 2002. Identification of aroma compounds in Parmigiano- Reggiano cheese by gas chromatography/olfactometry. Journal of dairy science, 85[6], pp. 1362-1369.
Rizzi, G.P. 1997. Chemical structure of coloured Maillard reaction products. Food Reviews International, 13[1], pp. 1-28.
Rizzi, G.P. 1972. Mechanistic study of alkylpyrazine formation in model systems. Journal of Agricultural and Food Chemistry, 20[5], pp. 1081-1085.
Rizzi, G.P. 2008. The Strecker Degradation of Amino Acids: Newer Avenues for Flavour Formation. Food Reviews International, 24[4], pp. 416-435.
Roscic, M. and Horvat, S. 2006. Transformations of bioactive peptides in the presence of sugars-Characterization and stability studies of the adducts generated via the Maillard reaction. Bioorganic & medicinal chemistry, 14[14], pp. 4933-4943.
Roscic, M., Versluis, C., Kleinnijenhuis, A.J., Horvat, S., and Heck, A.J.R. 2001. The early glycation products of the Maillard reaction: mass spectrometric characterization of novel imidazolidinones derived from an opioid pentapeptide and glucose. Rapid Communications in Mass Spectrometry, 15[12], pp. 1022-1029.
Royle, L., Bailey, R.G., and Ames, J.M. 1998. Separation of Maillard reaction products from xylose-glycine and glucose-glycine model systems by capillary electrophoresis and comparison to reverse phase HPLC. Food Chemistry, 62[4], pp. 425-430.
Saittagaroon, S., Kawakishi, S., and Namiki, M. 1984. Aroma constituents of roasted coconut. Agricultural and Biological Chemistry, 48[9], pp. 2301-2307.
Salter, L.J., Mottram, D.S., and Whitfield, F.B. 1989. Volatile compounds produced in Maillard reactions involving glycine, ribose and phospholipid. Journal of the Science of Food and Agriculture, 46[2], pp. 227-242.
Sasaki, Y., Shibamoto, T., Wei, C.I., and Fernando, S. 1987. Biological and chemical studies on overheated brewed coffee. Food and Chemical Toxicology, 25[3], pp. 225-228.
158 Schieberle, P. 2005. The carbon module labeling (CAMOLA) technique: A useful tool for identifying transient intermediates in the formation of Maillard-type target molecules. Annals of the New York Academy of Sciences, 1043, pp. 236-248.
Schnermann, P. and Schieberle, P. 1997. Evaluation of Key Odorants in Milk Chocolate and Cocoa Mass by Aroma Extract Dilution Analyses. Journal of Agricultural and Food Chemistry, 45[3], pp. 867-872.
Schweitzer, J.W., Friedhoff, A.J., and Schwartz, R. 1975. Chocolate, beta - phenethylamine, and migraine reexamined. Nature, 257[5523], pp. 256.
Severin, T. and Braütigam, K.H. 1970. Formation of benzaldehyde from phenylalanine and n-acetylphenylethylamine by heating under and air blast. Zeitschrift für Lebensmitteluntersuchung und -Forschung, 144[5], pp. 321-323.
Shibamoto, T. and Bernhard, R.A. 1977. Investigation of pyrazine formation pathways in sugar-ammonia model systems. Journal of Agricultural and Food Chemistry, 25[3], pp. 609-614.
Shu, C.K. 1999. Pyrazine formation from serine and threonine. Journal of Agricultural and Food Chemistry, 47[10], pp. 4332-4335.
Stadler, R.H., Blank, I., Varga, N., Robert, F., Hau, J., Guy, P.A., Robert, M., and Riediker, S. 2002. Acrylamide from Maillard reaction products. Nature, 419[6906], pp. 449-450.
Tai, C.Y. and Ho, C.T. 1997. Influence of Cysteine Oxidation on Thermal Formation of Maillard Aromas. Journal of Agricultural and Food Chemistry, 45[9], pp. 3586-3589.
Tehrani, K.A., Kersiene, M., Adams, A., Venskutonis, R., and De Kimpe, N. 2002. Thermal Degradation Studies of Glucose/Glycine Melanoidins. Journal of Agricultural and Food Chemistry, 50[14], pp. 4062-4068.
Tieman, D., Taylor, M., Schauer, N., Fernie, A.R., Hanson, A.D., and Klee, H.J. 2006. Tomato aromatic amino acid decarboxylases participate in synthesis of the flavour volatiles 2-phenylethanol and 2-phenylacetaldehyde. Proceedings of the National Academy of Sciences of the United States of America, 103[21], pp. 8287-8292.
Tokunaga, M., Shirogane, Y., Aoyama, H., Obora, Y., and Tsuji, Y. 2005. Copper- catalyzed oxidative cleavage of carbon-carbon double bond of enol ethers with molecular oxygen. Journal of Organometallic Chemistry, 690[23], pp. 5378-5382.
159 Tressl, R., Helak, B., Kersten, E., and Rewicki, D. 1993. Formation of proline- and hydroxyproline-specific Maillard products from [1-13C]glucose. Journal of Agricultural and Food Chemistry, 41[4], pp. 547-553.
Tressl, R., Nittka, C., Kersten, E., and Rewicki, D. 1995. Formation of Isoleucine- Specific Maillard Products from [1-13C]-D-Glucose and [1-13C]-D-Fructose. Journal of Agricultural and Food Chemistry, 43[5], pp. 1163-1169.
Tressl, R., Rewicki, D., Helak, B., Kamperschroer, H., and Martin, N. 1985. Formation of 2,3-dihydro-1H-pyrrolizines as proline specific Maillard products. Journal of Agricultural and Food Chemistry, 33[5], pp. 919-923.
Tsuge, O., Kanemasa, S., Ohe, M., and Takenaka, S. 1987. Simple generation of nonstabilized azomethine ylides through decarboxylative condensation of a-amino acids with carbonyl compounds via 5-oxazolidinone intermediates. Bulletin of the Chemical Society of Japan, 60[11], pp. 4079-4089.
Van Praag, M., Stein, H.S., and Tibbetts, M.S. 1968. Steam volatile aroma constituents of roasted cocoa beans. Journal of Agricultural and Food Chemistry, 16[6], pp. 1005-1008.
Vitzthum, O.G., Werkhoff, P., and Hubert, P. 1975. Volatile components of roasted cocoa. Basic fraction. Journal of Food Science, 40[5], pp. 911-916.
Walradt, J.P., Pittet, A.O., Kinlin, T.E., Muralidhara, R., and Sanderson, A. 1971. Volatile components of roasted peanuts. Journal of Agricultural and Food Chemistry, 19[5], pp. 972-979.
Weenen, H. 1998. Reactive intermediates and carbohydrate fragmentation in Maillard chemistry. Food Chemistry, 62[4], pp. 393-401.
Whitfield, F.B. and Mottram, D.S. 1999. Investigation of the Reaction between 4- Hydroxy-5-methyl-3(2H)-furanone and Cysteine or Hydrogen Sulfide at pH 4.5. Journal of Agricultural and Food Chemistry, 47[4], pp. 1626-1634.
Wnorowski, A. and Yaylayan, V. 1999. Effect of pressure programming on the efficiency of sample transfer during pyrolysis-GC/MS analysis of glucose-glycine mixtures. Journal of Analytical and Applied Pyrolysis, 48[2], pp. 77-91.
Wnorowski, A. and Yaylayan, V.A. 2003. Monitoring carbonyl-amine reaction between pyruvic acid and a-amino alcohols by FTIR spectroscopy-A possible route to amadori products. Journal of Agricultural and Food Chemistry, 51[22], pp. 6537-6543.
160 Yaylayan, V.A. 1997. Classification of the Maillard reaction: a conceptual approach. Trends in Food Science & Technology, 8[1], pp. 13-18.
Yaylayan, V.A. 2003. Recent advances in the chemistry of Strecker degradation and Amadori rearrangement: Implications to aroma and colour formation. Food Science and Technology Research, 9[1], pp. 1-6.
Yaylayan, V.A. and Haffenden, L.J.W. 2003a. Mechanism of imidazole and oxazole formation in [13C-2]-labelled glycine and alanine model systems. Food Chemistry, 81[3], pp. 403-409.
Yaylayan, V.A. and Haffenden, L.J.W. 2003b. Mechanism of pyrazole formation in [13C-2] labeled glycine model systems: N-N bond formation during Maillard reaction. Food Research International, 36[6], pp. 571-577.
Yaylayan, V.A., Haffenden, L., Chu, F.L., and Wnorowski, A. 2005. Oxidative pyrolysis and postpyrolytic derivatization techniques for the total analysis of maillard model systems: investigation of control parameters of maillard reaction pathways. Annals of the New York Academy of Sciences, 1043, pp. 41-54.
Yaylayan, V.A. and Keyhani, A. 1996. Pyrolysis/GC/MS analysis of non-volatile flavoir precursors. Editors: Pickenhagen W., Spanier A.M. and Ho,C.T. In Contribution of Low and Non-volatile Materials to the Flavour of Food. Allured Publishing Co., Carol Stream, IL, pp.13-26.
Yaylayan, V.A. and Keyhani, A. 1999. Origin of 2,3-Pentanedione and 2,3-Butanedione in D-Glucose/L-Alanine Maillard Model Systems. Journal of Agricultural and Food Chemistry, 47[8], pp. 3280-3284.
Yaylayan, V.A. and Keyhani, A. 2000. Origin of Carbohydrate Degradation Products in L-Alanine/D-[13C]Glucose Model Systems. Journal of Agricultural and Food Chemistry, 48[6], pp. 2415-2419.
Yaylayan, V.A. and Keyhani, A. 2001a. Elucidation of the mechanism of pyrrole formation during thermal degradation of 13C-labeled L-serines. Food Chemistry, 74[1], pp. 1-9.
Yaylayan, V.A. and Keyhani, A. 2001b. Carbohydrate and Amino Acid Degradation Pathways in L-Methionine/D-[13C] Glucose Model Systems. Journal of Agricultural and Food Chemistry, 49[2], pp. 800-803.
161 Yaylayan, V.A., Keyhani, A., and Huygues-Despointes,A. 1998. Generation and the fate of C2, C3, and C4 reactive fragments formed in Maillard model systems of [13C]glucose and [13C]glycine or proline. Advance in Experimental Medicine and Biology, 434, pp. 237-244.
Yaylayan, V.A., Machiels, D., and Istasse, L. 2003a. Thermal Decomposition of Specifically Phosphorylated D-Glucoses and Their Role in the Control of the Maillard Reaction. Journal of Agricultural and Food Chemistry, 51[11], pp. 3358-3366.
Yaylayan, V.A., Matni, G., Pare, J.R.J., and Belanger, J.M.R. 1997. Microwave-Assisted Synthesis and Extraction of Selected Maillard Reaction Products. Journal of Agricultural and Food Chemistry, 45[1], pp. 149-152.
Yaylayan, V.A., Perez Locas, C., Wnorowski, A., and O'Brien, J. 2004. The role of creatine in the generation of N-methylacrylamide: A new toxicant in cooked meat. Journal of Agricultural and Food Chemistry, 52[17], pp. 5559-5565.
Yaylayan, V.A. and Wnorowski, A. 2001. Role of L-serine and L-threonine in the generation of sugar-specific reactive intermediates during the Maillard reaction. Editors: Spanier, A., Shahidi, F., Parliament, T., Mussinan, C., Ho, C.T. and Contis, E. In food Flavours and Chemistry Advances of the New Millennium, Cambridge, U.K. pp. 313- 317.
Yaylayan,V.A. and Wnorowski,A. 2002. The role of β-hydroxyamino acids in the Maillard reaction-transamination route to Amadori products. In Maillard Reaction in Food Chemistry and Medical Sciences: Update for the Postgenomic Era. Editors: Horiuchi S., Taniguchi N., Hayase F., Kurata T. & Osawa T. International Congress Series, 1245. Elsevier Science. Amsterdam, the Netherlands. pp. 195–200.
Yaylayan,V.A., Wnorowski,A., and Perez,L.C. 2003b. Why asparagine needs carbohydrates to generate acrylamide. Journal of Agricultural and Food Chemistry, 51[6], pp. 1753-1757.
Zamora,R., Navarro,J.L., Gallardo,E., and Hidalgo,F.J. 2006. Chemical Conversion of a- Amino Acids into a-Keto Acids by 4,5-Epoxy-2-decenal. Journal of Agricultural and Food Chemistry, 54[16], pp. 6101-6105.
Zheng,Y., Brown,S., Ledig,W.O., Mussinan,C., and Ho,C.T. 1997. Formation of Sulfur- Containing Flavour Compounds from Reactions of Furaneol and Cysteine, Glutathione, Hydrogen Sulfide, and Alanine/Hydrogen Sulfide. Journal of Agricultural and Food Chemistry, 45[3], pp. 894-897.
162