Cyclization of reactive carbonyls in foods upon processing: an origin for the formation of pyridine-derived flavours and heterocyclic aromatic amines

Department of Chemical Engineering Faculty of Chemistry University of Seville Department of Lipid Characterisation and Quality Instituto de la Grasa Spanish National Research Council Cyclization of reactive carbonyls in foods upon processing: An origin for the formation of pyridine-derived flavours and heterocyclic aromatic amines CRISTINA Mª LAVADO TENA Department of Chemical Engineering Faculty of Chemistry University of Seville Department of Lipid Characterisation and Quality Instituto de la Grasa Spanish National Research Council Cyclization of reactive carbonyls in foods upon processing: An origin for the formation of pyridine-derived flavours and heterocyclic aromatic amines Doctoral thesis for the degree of Philosophiae Doctor presented by Cristina Mª Lavado Tena Seville, 2021 Tutor: Professor Cecilio Carrera Sánchez Department of Chemical Engineering Faculty of Chemistry University of Seville Seville, Spain Advisors: Professor Francisco Javier Hidalgo García Instituto de la Grasa Spanish National Research Council (CSIC) Seville, Spain Professor Rosario Zamora Corchero Instituto de la Grasa Spanish National Research Council (CSIC) Seville, Spain Reviewers: Dr. Stephen Elmore Department of Food and Nutritional Sciences University of Reading Whiteknights, Reading, UK. Dr. Vural Gökmen Food Quality and Safety Research Group Department of Food Engineering University of Hacettepe Ankara, Turkey University of Seville Seville, 2021 Acknowledgements I am completely sure that I will not be able to express my complete gratitude in these acknowledgments. Firstly, I have to thank my research supervisors, Prof. Francisco J. Hidalgo and Prof. Rosario Zamora. Without their assistance and dedicated involvement in every step throughout the process, this thesis would have never been possible. I would like to thank you very much for your support, understanding, and above all their patience over these years. I am also very grateful to Prof. Cecilio Carrera for being the tutor of this thesis and for your patient support and for all of the opportunities I was given to start my research career. My sincere gratitude to Dr. Bruno De Meulenaer for giving me the opportunity to stay for a short period of time at Ghent University (Belgium). I am very grateful to Dr. Vural Gökmen and Dr. Stephen Elmore for being the reviews of this thesis. I am extremely thankful and indebted to Jose Luis Navarro for sharing expertise, and sincere and valuable guidance and encouragement extended to me. Thanks to all my colleagues at the Instituto de la Grasa, especially to Ainoa, Ana, Clemen, Elena, and Isa for sharing conversations and laughters, for your help and for being my great support to continue every day. To my Friends, Carmen, Claudia, Jeni, Natalia, Patri, and Sara, and their husbands above all Luis for being with me through thick and thin and understand and support me. Most importantly, none of this could have happened without my family and my husband. My parents, who teach me to fight for me goals, for their love, and for their unconditional support. My husband, who helps me to achieve this dream, who always looks on the bright side of everything. I am forever grateful. Acknowledgements Finally, I wish to pay special thanks to the Ministerio de Ciencia e Innovación of Spain for granting my fellowship (BES-2016-078018). The study was supported partly by the European Union (FEDER funds), and the Plan Nacional de I+D of the Ministerio de Economía, Industria y Competitividad of Spain (Project AGL2015-68186-R); this financial support is gratefully acknowledged. Contents I CONTENTS Contents VIII 4.2. Cyclization and oligomerization of the reactive carbonyls in presence of creatinine: Formation of heterocyclic aromatic amines .................................................................................................... 112 4.2.1. Reactive carbonyls and the formation of MeIQ .............................. 112 4.2.1.1. Formation of MeIQ in reaction mixtures of reactive carbonyls, creatinine, and amino compounds ........................... 112 4.2.1.2. Effect of reaction conditions on the formation of MeIQ in reaction mixtures of crotonaldehyde, creatinine, and glutamine ................................................................................... 114 4.2.2. Reactive carbonyls and the formation of IQ ................................... 120 4.2.2.1. Formation of IQ and MeIQ in mixtures of acrolein, crotonaldehyde, creatinine, and glutamine ............................... 120 4.2.2.2. Effect of reaction conditions on the formation of IQ and MeIQ in mixtures of acrolein, crotonaldehyde, creatinine, and glutamine ................................................................................... 122 4.2.3. Reactive carbonyls and the formation of MeIQx ............................ 128 4.2.3.1. Formation of MeIQx in reaction mixtures of reactive carbonyls and creatinine ........................................................... 128 4.2.3.2. Formation of MeIQx in reaction mixtures of acroleinproducing compounds and creatinine ....................................... 137 5. DISCUSSION ............................................................................................ 139 5.1. Formation of pyridines by cyclization/oligomerization of reactive carbonyls in the presence of ammonia and ammoniaproducing compounds .......................................................................... 141 Contents IX 5.1.1. Effect of ammonia and ammonia-producing compounds on pyridine formation ........................................................................... 141 5.1.2. Proposal of a reaction pathway for the formation of 2- alkylpyridines ................................................................................... 142 5.1.3. Proposal of a reaction pathway for the formation of 3- alkylpyridines ................................................................................... 144 5.1.4. Proposal of a reaction pathway for the formation of 2,5- dialkylpyridines ................................................................................ 145 5.1.5. Proposal of a reaction pathway for the formation of 2,6- dialkylpyridines ................................................................................ 147 5.1.6. Formation of pyridines in oxidized menhaden oil heated in presence of glutamine ...................................................................... 149 5.2. Formation of 3-hydroxypyridines in foods ........................................ 149 5.2.1. Formation of 3-hydroxypyridines as a consequence of lipid oxidation ........................................................................................... 149 5.2.1.1. Role of lipid-derived reactive carbonyls on 2-alkyl-3- hydroxypyridines formation ...................................................... 149 5.2.1.2. Role of ammonia and ammonia-producing compounds on 3- hydroxypyridines formation ...................................................... 150 5.2.1.3. 3-Hydroxypyridines formation pathway ................................... 151 5.2.1.4. Formation of 3-hydroxypyridines in oxidized oils heated in presence of glutamine ............................................................... 153 5.2.2. Formation of 3-hydroxypyridines as a consequence of Maillard reaction in foods ............................................................................... 154 Contents X 5.2.2.1. Proposal of a reaction pathway for 3-hydroxypyridines formation by ring expansion of 2-oxofurans ............................. 154 5.2.2.2. Formation of 3-hydroxypyridines in honeys as a consequence of heating ................................................................................... 155 5.3. Formation of heterocyclic aromatic amines by cyclization of reactive carbonyls in the presence of creatinine ................................ 157 5.3.1. Role of reactive carbonyls on the formation of heterocyclic aromatic amines ............................................................................... 158 5.3.2. Contribution of ammonia-producing compounds to the formation of heterocyclic aromatic amines with the structure of aminoimidazoazarenes ..................................................................... 161 5.3.3. Proposal of reaction pathways for the formation of heterocyclic aromatic amines with the structure of aminoimidazoazarenes ...... 163 5.4. Strategies for the mitigation of heterocyclic aromatic amines in foods ...................................................................................................... 170 6. CONCLUSIONS ....................................................................................... 171 7. REFERENCES .......................................................................................... 177 List of Figures XI LIST OF FIGURES List of Figures XII List of Figures XIII Figure 1. Pyridine formation through carbonyl-amine reaction in presence of reactive carbonyls ............................................................................... 6 Figure 2. Chemical structures of the main heterocyclic aromatic amine ............ 8 Figure 3. Proposed carbonyl pathway for heterocyclic aromatic amines formation. ......................................................................................... 14 Figure 4. Proposed free radical pathway for heterocyclic aromatic amines formation. ......................................................................................... 15 Figure 5. Proposed pathway for PhIP formation ............................................... 16 Figure 6. Structures of some carbohydrates. ..................................................... 19 Figure 7. A simplified scheme of Maillard reaction .......................................... 21 Figure 8. Carbohydrate-derived reactive carbonyls found in food. ................... 22 Figure 9. Formation of the most abundant carbohydrate-derived reactive dicarbonyls present in food .............................................................. 23 Figure 10. Structures of main fatty acids........................................................... 28 Figure 11. Main classes of lipid-derived reactive carbonyls found in food ...... 29 Figure 12. Lipid oxidation pathway. ................................................................. 30 Figure 13. Lipid oxidation products (LOPs) formation from linoleic acid ....... 33 Figure 14. Degradation of linoleic acid hydroperoxides to hydroxy-, epoxy- and oxo-fatty acids. .......................................................................... 34 Figure 15. Some compounds produced by β-scission of alkoxyl radicals. ........ 35 Figure 16. Structures of some tertiary lipid oxidation products from ω3 or ω6 polyunsaturated fatty acids. ........................................................ 36 Figure 17. General structure of amino acids. .................................................... 41 Figure 18. Strecker degradation of methionine and later conversion of methional into acrolein ..................................................................... 42 List of Figures XIV Figure 19. Conversion of amino acids into α-keto acids. .................................. 43 Figure 20. Thermal degradation of amino acids ................................................ 44 Figure 21. Contribution of o-quinones to degradation of amino acids .............. 46 Figure 22. Examples of advanced glycation end-products (AGEs). ................. 47 Figure 23. Non-enzymatic reactions between oxidised polyphenols and amino acids ....................................................................................... 49 Figure 24. Structure of chlorogenic acid and chlorogenic acid quinone ........... 50 Figure 25. Michael additions of glutathione (GSH) to p-quinone and oquinone ............................................................................................. 51 Figure 26. Structure of flavan-3-ols. ................................................................. 52 Figure 27. Structures of the carbonyl-phenol adducts ....................................... 53 Figure 28. Mechanism of pyrazine formation via Maillard reaction ................. 54 Figure 29. Syntheses of methyl 13-hydroperoxyoctadeca-9,11-dienoate (MeLOOH) and methyl 13-hydroperoxyoctadeca-9,11,15- trienoate (MeLnOOH). ..................................................................... 65 Figure 30. Synthesis of 4-hydroxy-2-nonenal (HNE). ...................................... 66 Figure 31. Synthesis of 4-hydroxy-2-hexenal (HHE). ...................................... 67 Figure 32. Syntheses of 4-oxo-2-nonenal and 4-oxo-2-hexenal. ....................... 68 Figure 33. Syntheses of 4,5-epoxy-2-alkenals................................................... 69 Figure 34. Linear retention indexes (LRI) of the trimethylsilyl derivatives of the hydroxyalkylpyridines. ............................................................... 78 Figure 35. Mass spectra of the trimethylsilyl derivatives of the studied hydroxyalkylpyridines. ..................................................................... 79 Figure 36. Effect of different lipid oxidation products on 3-hydroxy-2- methylpyridine formation. ................................................................ 93 List of Figures XV Figure 37. Effect of the chain length of the 2,4-alkadienal on the amount of 3-hydroxypyridines produced. ......................................................... 94 Figure 38. Effect of the type of ammonia-producing compound on 3- hydroxy-2-methylpyridine formation by cyclization of 2,4- hexadienal. ....................................................................................... 95 Figure 39. Effect of reaction pH on 3-hydroxy-2-methylpyridine formation by cyclization of 2,4-hexadienal in the presence of glutamine. ....... 96 Figure 40. Effect of aldehyde concentration on 3-hydroxy-2-methylpyridine formation by cyclization of 2,4-hexadienal in the presence of glutamine. ......................................................................................... 97 Figure 41. Effect of amino acid concentration, on 3-hydroxy-2- methylpyridine formation by cyclization of 2,4-hexadienal in the presence of glutamine....................................................................... 98 Figure 42. Effect of time and temperature on the formation of 3-hydroxy-2- methylpyridine ................................................................................. 99 Figure 43. Arrhenius plot obtained for the formation of 3-hydroxy-2- methylpyridine. ................................................................................ 99 Figure 44. Effect of pH on the formation of 2-hydroxymethyl-5- hydroxypyridine and on the disappearance of 5- hydroxymethylfurfural (HMF) ....................................................... 102 Figure 45. Effect of HMF concentration on the formation of 2- hydroxymethyl-5-hydroxypyridine. ............................................... 103 Figure 46. Effect of the concentration of the amino compound on the formation of 2-hydroxymethyl-5-hydroxypyridine and on the disappearance of HMF ................................................................... 104 Figure 47. Effect of time and temperature on the formation of 2- hydroxymethyl-5-hydroxypyridine ................................................ 105 List of Figures XVI Figure 48. Arrhenius plot obtained for the formation of 2-hydroxymethyl-5- hydroxypyridine. ............................................................................ 105 Figure 49. Effect of time and temperature on the disappearance of HMF. ..... 106 Figure 50. Arrhenius plot obtained for the disappearance of HMF. ................ 106 Figure 51. Structures of 2-oxofurans assayed and the corresponding 3- hydroxypyridines. ........................................................................... 107 Figure 52. Formation of 3-hydroxypyridines from 2-oxofurans. .................... 108 Figure 53. Effect of heating time on the formation of 2-hydroxymethyl-5- hydroxypyridine and HMF in honey heated at A, 60, and B, 100 °C. ................................................................................................... 109 Figure 54. Effect of heating time on the formation of 2-hydroxymethyl-5- hydroxypyridine, HMF, 3-hydroxypyridine, and 3-hydroxy-2- methylpyridine in sugarcane honey heated at A, 60, and B, 100 °C. ................................................................................................... 111 Figure 55. Effect of reaction pH on MeIQ formation by reaction of crotonaldehyde, creatinine, and glutamine. .................................... 115 Figure 56. Effect of the concentration of crotonaldehyde on the MeIQ produced by reaction of crotonaldehyde, creatinine, and glutamine. ....................................................................................... 116 Figure 57. Effect of the concentration of creatinine on the MeIQ produced by reaction of crotonaldehyde, creatinine, and glutamine. ............. 116 Figure 58. Effect of the concentration of glutamine on the MeIQ produced by reaction of crotonaldehyde, creatinine, and glutamine. ............. 117 Figure 59. Effect of water concentration on MeIQ formation by reaction of crotonaldehyde, creatinine, and glutamine. .................................... 118 Figure 60. Effect of oxygen content on MeIQ formation by reaction of crotonaldehyde, creatinine, and glutamine. .................................... 118 List of Figures XVII Figure 61. Effect of time and temperature on MeIQ formation by reaction of crotonaldehyde, creatinine, and glutamine. .................................... 119 Figure 62. Arrhenius plot obtained for the formation of MeIQ....................... 120 Figure 63. Effect of the reaction time of the first incubation (at 25 °C) on the formation of IQ and MeIQ ............................................................. 121 Figure 64. Effect of reaction pH on the formation of IQ and MeIQ .............. 122 Figure 65. Effect of acrolein concentration on the formation of IQ and MeIQ ............................................................................................. 123 Figure 66. Effect of creatinine concentration on the formation of IQ and MeIQ .............................................................................................. 124 Figure 67. Effect of water added on the formation of IQ and MeIQ ............... 125 Figure 68. Time-course of IQ formation at different temperatures. ................ 125 Figure 69. Time-course of MeIQ formation at different temperatures. ........... 126 Figure 70. Arrhenius plot obtained for the formation of IQ. ........................... 127 Figure 71. Arrhenius plot obtained for the formation of MeIQ....................... 127 Figure 72. Effect of pH on MeIQx formation. ................................................ 129 Figure 73. Effect of acrolein concentration on MeIQx formation ................... 129 Figure 74. Effect of creatinine concentration on MeIQx formation. ............... 130 Figure 75. Effect of methylglyoxal addition on MeIQx formation. ................ 131 Figure 76. Effect of addition of: A, ammonium chloride; B, formaldehyde; C, ammonium chloride; and D, formamide, in mixtures of acrolein/creatinine and acrolein/creatinine/methylglyoxal............. 132 Figure 77. Effect of addition of water on MeIQx formation in mixtures of acrolein/creatinine and acrolein/creatinine/methylglyoxal............. 133 List of Tables XXIV List of Original Publications XXV LIST OF ORIGINAL PUBLICATIONS List of Original Publications XXVI List of Original Publications XXVII Zamora, R., Lavado-Tena, C. M., & Hidalgo, F. J. (2019). Oligomerization of reactive carbonyls in the presence of ammonia-producing compounds: A route for the production of pyridines in foods. Food Chemistry, 304, 125284. Zamora, R., Lavado-Tena, C. M., & Hidalgo, F. J. (2020). Reactive carbonyls and the formation of the heterocyclic aromatic amine 2-amino-3,4- dimethylimidazo(4,5-f)quinoline (MeIQ). Food Chemistry, 324, 126898. Hidalgo, F. J., Lavado-Tena, C. M., & Zamora, R. (2020). Conversion of 5- hydroxymethylfurfural into 6-(hydroxymethyl)pyridin-3-ol: A pathway for the formation of pyridin-3-ols in honey and model systems. Journal of Agricultural and Food Chemistry, 68, 5448-5454. Hidalgo, F. J., Lavado-Tena, C. M., & Zamora, R. (2020). Formation of 3- hydroxypyridines by lipid oxidation products in the presence of ammonia and ammonia-producing compounds. Food Chemistry, 328, 127100. Zamora, R., Lavado-Tena, C. M., & Hidalgo, F. J. (2020). Identification of precursors and formation pathway for the heterocyclic aromatic amine 2-amino- 3-methylimidazo(4,5-f)quinoline (IQ). Journal of Agricultural and Food Chemistry, 68, 7474-7481. Hidalgo, F. J., Lavado-Tena, C. M., & Zamora, R. (2021). Identification of acrolein as the reactive carbonyl responsible for the formation of 2-amino-3,8- dimethylimidazo[4,5-f]quinoxaline (MeIQx). Food Chemistry, 343, 128478. List of Original Publications XXVIII Introduction 1 1. INTRODUCTION Introduction 2 Introduction 3 1.1. Food processing and its consequences Foods can be consumed in a raw state, but, most frequently, they are processed both to improve their quality and to destroy deleterious components. Food processing can be carried out in many ways. Thus, milling, cooling/freezing, smoking, heating, canning, fermentation, drying, or extrusion cooking, among others, are common industrial processes used in food manufacturing (Augustin et al., 2016; Stoica et al., 2013). These processes have benefits, such as destruction of food-borne microbes and toxins, improving bioavailability of nutrients, extending shelf life, or enhancing sensory characteristics and functional properties (Boekel et al., 2010; Nawrot et al., 1999). On the other hand, food processing can also have detrimental consequences for the involved foods (Table 1). Among them, nutrient degradation (Elmore et al., 2010), production of deleterious sensory active compounds (Starowicz & Zieliński, 2019), or production of compounds with mutagenic or carcinogenic properties (Jaeger et al., 2010) have been described. Table 1. Detrimental effects of thermal processing. Food constituents Effect of heating Examples of produced compounds Carbohydrates Decomposition Carbonyl compounds Formation of carbonyl-amine adducts, including polymers Advanced glycation end products (AGEs) Lipids Oxidation Lipid hydroperoxides Hydroperoxide decomposition Carbonyl compounds Formation of carbonyl-amine adducts, including polymers Advanced lipoxidation end products (ALEs) Amino acids/ Proteins/ Creatine Decomposition Carbonyl compounds Formation of carbonyl-amine adducts AGEs/ALEs Reaction products with creatine Heterocyclic aromatic amines (HAAs) Phenolics Oxidation Quinones Introduction 4 This Ph.D. thesis will study the generation of two families of products formed as a consequence of food processing: flavours with the structure of pyridine and heterocyclic aromatic amines (HAAs), in an attempt to find out the chemical basis of the formation of both kinds of compounds. This knowledge should allow to promote their formation when they are desirable and to understand how their formation can be inhibited when they act as food toxicants. Thus, pyridines are potent odorants that contribute to the sensory properties of some foods (Maga, 1981). On the other hand, the formation of heterocyclic aromatic amines (HAAs) has been related to the claimed mutagenicity and carcinogenicity for processed meats (International Agency for Research on Cancer, 2018). 1.2. Pyridines The pyridines are a large group of N-heterocyclic compounds. The simplest pyridine derivative has the same name of the group, pyridine. Its structure is related to benzene but one methine group (=CH-) is replaced by a nitrogen atom. Other pyridines are more complex because of the presence of substituents in the aromatic ring. 1.2.1. Pyridines in foods The presence of alkylpyridines in foods has been investigated by diverse authors (Ferretti & Flanagan, 1971; MacLeod & Coppock, 1976; Watanabe & Sato, 1971; Yajima et al., 1979). The importance of heterocyclic compounds in food flavours is well-known. However, relatively little attention has been given to pyridine derivatives, even though many of them have been found in a large number of foods, specially alkylpyridines (Suyama & Adachi, 1980). In addition, hydroxypyridines also contribute to the sensory properties of some foods. However, contrarily to these alkylpyridines, hydroxypyridines are less frequently found in foods (Elhalis et al., 2021). Introduction 5 The food systems in which pyridines have been found include food systems containing high fat and low water contents, and submitted to roasting or frying. Therefore, pyridines have been hypothesized to be produced as a consequence of the interaction between primary amines in proteins or amino acids with alkadienals (Kim et al., 1996). Foods in which pyridines have been found include meat (beef, chicken, and lamb), fish, vegetables (beans, asparagus, potatoes, and tomatoes), cereals (rice and corn), bread, nut products (almond, peanuts, and pecan), soya bean, milk and dairy products (cheese), fruits, spices and condiments (pepper), honey, coffee, and alcoholic beverages (Bicas & Rodriguez-Amaya, 2021; Maga, 1981). 1.2.2. Role of pyridines in food flavours Pyridines are considered to contribute to food flavour to a lower extent than other heterocyclic compounds such as pyrazines. However, they provide foods with bitter, astringent, and roasted notes, among others. The flavour properties of some selected pyridines are collected in Table 2. Table 2. Flavour properties produced by pyridine derivatives. Pyridine Flavour properties 2-Methylpyridine Astringent and hazelnut 3-Methylpyridine Green, earthy, and hazelnut 2-Ethylpyridine Buttery, green, and caramel 2,5-Dimethylpyridine Roasted, green, and earthy 2,6-Dimethylpyridine Green 5-Ethyl-2-methylpyridine Fatty and green Elaborated from Maga, (1981) Introduction 12 Table 3 (cont.) Food Cooking method HAAs Reference Chicken burger Pan fried IQx (Haskaraca et al., 2016) Hamburger patties Fried IQ, MeIQ, MeIQx (Aeenehvand et al., 2016) - MeIQx Beer - PhIP (S Manabe et al., 1992) Wine - IQ, MeIQ, MeIQx, 4,7,8-TriMeIQx (Richling et al., 1997) Soup cubes - IQ, MeIQx, 4,8- DiMeIQx (Krach & Sontag, 2000) Soft cheese - MeIQx (Gonzalo- Lumbreras et al., 2010) Elaborated from Barzegar et al.(2019). Abbreviation: HAAs, heterocyclic aromatic amines. 1.4.3. Estimation of aminoimidazoazarene intake in the general population An estimate of population-average dietary intake of thermic heterocyclic aromatic amines is complex because the daily intake of them is different in each study. Thus, there are differences in study designs, and, also, on portion size, population, cooking conditions, intake frequency, etc. According to the obtained results in the different studies, heterocyclic aromatic amine intakes vary from < 2 to > 25 ng·kg -1 of body weight per day (Augustsson et al., 1997; Keating & Bogen, 2001). According to the study of Bogen & Keating (2001) (Table 4), heterocyclic aromatic amine intake depends on age, sex, socioeconomic factors, raceethnicity, type of food eaten, and cooking method. Thus, in decreased order of consumption, PhIP represents over 66% of the total heterocyclic aromatic amine considered, while MeIQx and AαC are 15%, and DiMeIQx and IQ each 2% of this total input. Estimated heterocyclic aromatic amine consumption takes into account race-ethnicity, age group, and sex, and it is about 1.5- to 2-fold greater Introduction 13 for African Americans and Asian/Pacific Islanders than for whites. In general, female vs. male heterocyclic aromatic amine specific intakes show small differences for each heterocyclic aromatic amine within each age and racialethnic group. An exception is that estimated intake by male African American children is over 30-50% greater than that of females (Bogen & Keating, 2001). Table 4. Estimated mean heterocyclic aromatic amine (HAA) intakes by raceethnicity, age group, and sex parameters. Race / ethnicity Age Sex Estimated HAA intake (ng/kg/day) PhIP MeIQx DiMeQx IQ White Child M/F 6.10/6.50 1.20/1.20 0.23/0.24 0.16/0.14 Young M/F 5.50/5.60 1.10/0.96 0.20/0.19 0.12/0.13 Older M/F 5.60/5.30 1.10/1.00 0.18/0.17 0.20/0.20 African American Child M/F 12.00/8.90 1.80/1.50 0.51/0.33 0.24/0.11 Young M/F 8.30/7.30 1.30/1.20 0.35/0.27 0.09/0.12 Older M/F 8.10/7.40 1.60/1.20 0.30/0.23 0.16/0.23 Asian / Pacific Islander Child M/F 10.00/9.20 1.90/1.90 0.27/0.22 0.29/0.90 Young M/F 5.80/5.90 1.00/1.10 0.22/0.16 0.33/0.20 Older M/F 8.20/9.60 1.70/1.90 0.20/0.21 0.69/1.20 Elaborated from Bogen & Keating, (2001). Abbreviations: M, masculine. F, feminine. Child <16 years, young (16–29 years), and older (>30 years). 1.5. Formation of aminoimidazoazarenes The heterocyclic aromatic amines with the structure of aminoimidazoazarene have an imidazole ring fused to either a pyridine or a pyrazine ring. Because of their similar structures (Figure 2), they are supposed to be produced by means of analogous reaction pathways, although their formation mechanisms are still unclear. Preliminary studies suggested that heterocyclic aromatic amines were produced as a result of complex reactions that involved creati(ni)ne, free amino Introduction 14 acids, and carbohydrates at high temperatures through Maillard reaction. Two hypothetical pathways were considered according to the Maillard reaction intermediates: A carbonyl pathway and a free radical pathway. The first proposed pathway was a consequence of the ability of Maillard reaction to produce pyrazines, and, to a lower extent, also pyridines. These heterocyclic compounds, which are produced as a consequence of the reaction between amino acids and reducing sugars, can later react with creati(ni)ne, and aldehydes, to produce the different heterocyclic aromatic amines (Figure 3) (Milic et al., 1993). Figure 3. Proposed carbonyl pathway for heterocyclic aromatic amines formation (Zamora & Hidalgo, 2020). Milic et al. (1993) firstly studied the pyrazine or pyridine formation by reaction of D-(+)-glucose and amino acids to produce 2,5-dimethylpyrazine or 2- methylpyridine. Then, the later reaction of these pyrazine or pyridine with creati(ni)ne and acetaldehyde produced MeIQx and MeIQ, respectively. However, employed reaction conditions are not representative of reactions occurring in real food systems. Some other studies have also pointed out that radicals, especially pyridine and pyrazine radicals might also be involved in the formation of aminoimidazoazarenes (Figure 4). Two different pathways for this free radical formation are proposed. One involves pyridine radical and other pyrazine radical, which would react with creati(ni)ne and the corresponding aldehyde. The Introduction 15 reaction would finish with the formation of IQ-type and IQx-type mutagens respectively. Nonetheless, this mechanism remains relatively controversial because it has missing steps (Murkovic, 2004). Figure 4. Proposed free radical pathway for heterocyclic aromatic amines formation (Zamora & Hidalgo, 2020). At present, only PhIP formation pathway has been clarified. As described below, it does not follow any of the two previously proposed formation mechanisms. Proposal of a reaction pathway for PhIP has been a long process. It was initiated by Shioya et al. (1987). These authors found that PhIP was produced in mixtures of creatinine, phenylalanine, and glucose. Later, Manabe et al. (1992) described that the reaction could also be produced in the presence of aldehydes. Some years later, Murkovic et al. (1999) demonstrated that phenylacetaldehyde was an intermediate in its formation. In fact, this group isolated the aldol condensation product between phenylacetaldehyde and creati(ni)ne (Zöchling & Murkovic, 2002). The formation pathway was completed some years ago by Zamora et al. (2014). The complete reaction pathway is shown in Figure 5. The reaction is initiated by the degradation of phenylalanine to produce phenylacetaldehyde. This degradation, also known as Strecker degradation, occurs when amino acids are heated in the presence of reactive carbonyls (Zamora et al., 2007). Once this aldehyde has been produced, it suffers the addition of creati(ni)ne to form the corresponding aldol product, which is later Introduction 16 dehydrated. The pyridine ring of PhIP is completed by successive additions of ammonia and formaldehyde. Ammonia and formaldehyde are produced by degradation of phenylalanine, phenylacetaldehyde, and/or creati(ni)ne. Additionally, formaldehyde and ammonia can react to produce formamide, which might also be an intermediate to close the pyridine ring (Zamora et al., 2014). CO2H NH2O N N O NH2 CH3 N N O NH2 CH3 HCHO NH3 N N N NH2 CH3 Figure 5. Proposed pathway for PhIP formation (Zamora et al., 2014). Analogously to PhIP, other heterocyclic aromatic amines with the structure of aminoimidazoazarene might also be produced similarly. Thus, reactive carbonyls other than phenylacetaldehyde would be produced and they would react with creati(ni)ne to produce the main skeleton of the different heterocyclic aromatic amines. Full structures might be later completed by means of ammonia and/or short chain aldehydes. As will be discussed below, this hypothesis, which has not been investigated so far, will be the main objective of this thesis. This hypothesis implies both that the reactive carbonyls required for the formation of the different heterocyclic aromatic amines should be produced in foods and that these reactive carbonyls should have a tendency to oligomerize and to produce cyclic/heterocyclic structures in the presence of creati(ni)ne. This tendency would also be the origin of the pyridines found in foods. Therefore, if this Introduction 17 hypothesis is confirmed, the formation of these kinds of compounds (some of which are desirable, but other are well-known food toxicants) can be understood and their formation potentially controlled based on this knowledge. The antecedents for the formation of reactive carbonyls in foods and their tendency to produce cyclic derivatives will be discussed in the next subsections. 1.6. Reactive carbonyls in foods: Formation of the food carbonylome Reactive carbonyls are continuously produced in foods and the most reactive of these compounds rapidly disappear. This group of compounds is therefore in a continuous change. The whole group of carbonyl compounds present in food at a certain time has been named “food carbonylome” (Zamora & Hidalgo, 2020). Its origin is quite diverse and, frequently, the same reactive carbonyl can have different precursors. Thus, the carbonyl compounds present in foods are produced from carbohydrates, lipids, amino acids, and phenolic compounds, among other sources. 1.6.1. Reactive carbonyls produced from carbohydrates 1.6.1.1. Carbohydrates In nature, carbohydrates are abundant components which are present in plant and animal tissues as well as in microorganisms. They are important nutrients that serve as energy source, fuel, and building materials. In addition, these compounds are often used as food additives, including anticaking agents, bulking agents, emulsifiers, gelling agents, humectants, stabilizers, sweeteners, or thickeners (Soga & Serwe, 2000; Voragen, 1998). Some examples of carbohydrates present in foods and foods products are described in Table 5 (Eggleston et al., 2018). Introduction 18 Table 5. Carbohydrates present in foods. Food Carbohydrates Monosaccharide Disaccharide Polysaccharide Fruit Apple Glucose, fructose and mannose Sucrose Starch and cellulose Grape Glucose, fructose and mannose Sucrose Cellulose Strawberry Glucose, fructose and mannose Sucrose Cellulose Vegetables Carrot Glucose and fructose Sucrose Starch and cellulose Onion Glucose and fructose Sucrose Cellulose Peanuts - Sucrose Cellulose Potato - - Starch and cellulose Others Sugar cane Glucose and fructose Sucrose - Honey Glucose and fructose Sucrose - Meat Glucose - Glycogen Milk - Lactose - Carbohydrates (Figure 6) can be classified into different groups. Thus, they are classified: • Into monosaccharides, oligosaccharides, and polysaccharides, according to their chemical structure. Monosaccharides are polyhydroxy-aldehydes (aldoses) or -ketones (ketoses), generally with an unbranched C-chain. When these monosaccharides polymerize in a number under 10 carbohydrate units, oligosaccharides are produced. The main oligosaccharides are the disaccharides, such as maltose or lactose. Finally, when the number of carbohydrate units are higher than 10, polysaccharides are produced. Examples are starch, cellulose, and pectin (Belitz et al., 2009). Introduction 19 • Into simple and complex carbohydrates according to the facility of absorption and digestion (Wong, 2018). Figure 6. Structures of some carbohydrates. 1.6.1.2. Maillard reaction Maillard reaction, also known as non-enzymatic browning, is one of the main chemical reactions that take place in foods. This reaction occurs as a consequence of food cooking, preservation, and processing, and many carbohydrates-derived reactive carbonyls are produced, which will be later involved in carbonyl-amine reactions. The carbohydrates that take part in Maillard reaction are reducing sugars, such as D-glucose, D-fructose, maltose, or lactose (Figure 6). The first references to this reaction were described by Louis-Camille Maillard in 1912. He observed that, when mixtures of amino acids and sugars were heated, browning was produced. However, it would not be until the 1950s, when the first stable products of the Maillard reaction in foods (glycosylamines or Amadori products) were identified. In 1953, John E. Hodge pointed out Amadori rearrangement with a key role in the reaction and divided the reaction into three parts: First, the formation of the Amadori products, followed by their Introduction 20 degradation, as a second step, and, then, the formation of melanoidins (Hellwig & Henle, 2014). A simplified scheme of Maillard reaction is shown in Figure 7. The pathway is divided into 3 main steps (Amaya-Farfan & Rodriguez-Amaya, 2021): • Initial stage begins with a sugar-amine condensation. Amino acids or amines, which are nucleophilic compounds, react with the carbonyl function of reducing carbohydrates. Then, imines (N-substituted glycosylamine)/Schiff bases are produced. Lately, these imines undergo a process named Amadori rearrangement to produce the corresponding Amadori (1-amino-1-deoxyketose) or Heyns (2-amino-2-deoxyaldose) compounds when the initial reducing sugar is either an aldose or a ketose, respectively. • Intermediate stage occurs when these Amadori compounds are degraded into 1-, 3-, and 4-deoxyosone compounds by means of different reactions including enolization, deamination, dehydration, cyclization, retroaldolization, isomerization, and fragmentation. These reactions occur under severe heating or prolonged storage. In addition to deoxyosones, other carbonyl compounds, furan derivatives, and other intermediates are also produced. • Final stage comprises aldol condensations, carbonyl-amine reactions, and the formation of heterocyclic derivatives with an atom of nitrogen. In addition, polymerisation reactions are also produced and the formation of coloured products with high molecular mass (melanoidins) is a common consequence of Maillard reaction. These pigments are responsible for the colour development observed in foodstuff during thermal processes (e.g. bread, biscuit, and meat). Polymerisation reactions also contribute to the hardening of cooked and stored food (Belitz et al., 2009; Rannou et al., 2016). Introduction 21 Figure 7. A simplified scheme of Maillard reaction (adapted from Purlis, 2010; Zamora & Hidalgo, 2005). 1.6.1.3. Formation of reactive carbonyls from carbohydrates Maillard reaction produces many compounds, including the formation of aldehydes, alcohols, thiols, sulphides, acids, lactones, and a large variety of heterocyclic compounds. The formed products depend on the intermediate that is degraded. Thus, 3-hydroxy-5-hydroxymethyl-2-methyl-(5H)-furan-4-one, 4- hydroxy-2,5-dimethyl-3(2H)-furanone (furaneol), or acetylformoin are produced by degradation of 3-deoxyosones, furfural is produced as a consequence of 1- deoxyosone degradation, and 2-hydroxyacetylfuran is produced by degradation of 4-deoxyosones (Zheng et al., 2019). Others secondary products of Maillard reaction are reductones, such as acetylformoin and aldehydes from the reactions between deoxyosones and amino acids. This last reaction is usually known as Introduction 28 (between 8 and 12 carbon atoms), and short chain (between 4 and 6 carbon atoms) fatty acids. The most abundant fatty acids in most organisms have 16-18 carbon atoms. Based on the presence or not of double bonds, fatty acids can be unsaturated or saturated, respectively. In addition, according to the number of double bonds, fatty acids can be monounsaturated (one double bond) or polyunsaturated (several double bonds). Double bonds can occur in two forms cis and trans, but most fatty acids found in natural fats have cis configuration. In polyunsaturated fatty acids, also named PUFAs, double bonds are usually separated by one methylene group. Depending on the position of the first carboncarbon double bond counted from the methyl end, fatty acids can be classified into families: ω3 (linolenic type), ω6 (linoleic type), ω7 (palmitoleic acid type), and ω9 (oleic acid type). Linolenic and linoleic acids are considered essential fatty acids for human beings and they cannot be produced within the human body. Chemical structures for some of these fatty acids are given in Figure 10 (Finley & DeMan, 2018; Wong, 2018). Figure 10. Structures of main fatty acids. 1.6.2.2. Lipid oxidation as a main pathway for the formation of lipid-derived reactive carbonyls Lipid-derived reactive carbonyls include ketones and aldehydes with diverse structures (Figure 11) such as alkanals (Wu & Wang, 2019), 2-alkenals (Bastos Introduction 29 et al., 2017), 2,4-alkadienals (Beltrán et al., 2011), 4-hydroxy-2-alkenals (Csallany et al., 2015), 4-oxo-2-alkenals (Tullberg et al., 2019), 4,5-epoxy-2- alkenals (Zamora et al., 2017), and malondialdehyde (Bertolín et al., 2019), among others (Elmore et al., 2005). They are produced as a consequence of a cascade of reactions initiated by the reaction of fatty acyl chains with oxygen: The lipid oxidation. Figure 11. Main classes of lipid-derived reactive carbonyls found in food (Zamora & Hidalgo, 2020). The reaction needs the presence of initiators. Among them, heat, free radicals, light, photosensitizing pigments, and metal ions play a major role. Depending of the initiators, lipid oxidation may be a photooxidation (induced by light), an enzyme reaction (initiated by lipoxygenase), or a free-radical reaction (autooxidation) ( Figure 12 ). The three of them follow a similar scheme with slight differences (Finley & DeMan, 2018). The autooxidation is the most general pathway and will be described in this section. The process of autoxidation is produced through sequential free radical chain reactions and, like any other free radical reaction, implies three stages: Initiation, propagation, and termination (Rodriguez-Amaya & Shahidi, 2021). Introduction 30 Figure 12. Lipid oxidation pathway. • Initiation The initiation step involves the homolytic breakdown of the carbon-hydrogen bond at the α-position in relation to the double bond in the unsaturated acyl group (usually the bis-allylic carbon atom). This reaction produces the corresponding alkyl radicals (R • ). The reaction is thermodynamically unfavourable and the presence of an initiator (light, heat, metal ions) is needed (Schaich et al., 2013). RH + Initiator  R • Introduction 31 Hydrogens with lower bond dissociation energies are firstly removed. PUFAs are particularly susceptible to hydrogen abstraction due to the presence of single or multiple methylene-interrupted double bonds (McClements & Decker, 2018). Once produced, the resulting alkyl radical (R • ) is stabilized by delocalization over the double bond(s) to form a conjugated diene (McClements & Decker, 2018). • Propagation: The propagation phase of oxidation occurs with the addition of molecular oxygen to the produced alkyl radical (R • ) to form the corresponding peroxyl radical (ROO • ). This radical is less reactive than the alkyl radical. Consequently, most free radicals are in the form of peroxyl radicals. At the same time, a hydrogen atom from another fatty acid molecule (RH) is transferred to the peroxyl radical to produce the corresponding hydroperoxide (ROOH) and an alkyl radical (R • ). This new alkyl radical starts again the same process. This process continues indefinitely until no hydrogen source is available or the chain is interrupted. Hydroperoxides are decomposed into alkoxyl radicals, peroxyl radicals, and hydroxyl radicals by metal ions, heat, and ultraviolet (UV) light. When hydroperoxide decompositions are catalysed by metal ions, the scissions are heterolytic, producing one radical and one ion. However, when hydroperoxide decompositions are catalysed by heat and UV light, homolytic scission is produced and two radicals, alkoxyl (RO • ) and hydroxyl (OH • ) radicals, are produced. Both radicals react much more rapidly than ROO • . This second part of the propagation step is known as branching (McClements & Decker, 2018; Schaich et al., 2013). A brief scheme reaction of the propagation step is: R • + O 2  ROO • ROO • + RH  ROOH + R • Branching: (Metal ion) ROOH  RO • + OH - (Light or heat) ROOH  RO • + OH • Introduction 32 • Termination: As usual in all free radical reactions, free radical chain is broken when the collision of free radicals or the reactions between radicals with other non-radical compounds provoke the formation of stable molecules, thus making the free radicals disappear (McClements & Decker, 2018). Some of these reactions are: ROO • + R •  ROOR R • + R •  RR ROO • + ROO •  ROOR + O 2 R • + RO •  ROR R • + H •  RH RO • + H •  ROH ROO • + H •  ROOH R • + OH •  ROH These reactions are responsible for the formation of some secondary oxidation products. In addition, other secondary oxidation products are produced as a consequence of the breakage of free radicals. This will be discussed in the next subsection. 1.6.2.3. Formation of lipid oxidation products As indicated above, there are many routes and also many starting materials. Therefore, the number of produced compounds is large. They are named as lipid oxidation products (LOPs) and their formation occurs sequentially. For this reason, some of them are called primary LOPs and other are named secondary LOPs. In addition, the existence of tertiary LOPs is also considered by some authors (Lu et al., 2014).  Primary LOPs. Lipid peroxidation generates hydroperoxides in the propagation and termination steps, which are considered as the primary LOPs. Despite these compounds have not a role on food flavour, they are unstable and they have deleterious effects on health. For most foods, the main unsaturated fatty acids present in them and susceptible to oxidation are: Linoleic (LH) and linolenic acids (LnH). Their autooxidation produces the corresponding hydroperoxides: 9- Introduction 33 LOOH and 13-LOOH from linoleic acid and 9-LnOOH and 13-LnOOH from linolenic acid (Rodriguez-Amaya & Shahidi, 2021; Wong, 2018). Figure 13 shows the hydroperoxides formed as a consequence of linoleic acid oxidation. Figure 13. Lipid oxidation products (LOPs) formation from linoleic acid R 1 = C 5 H 11 and R 2 = C 8 H 15 O 2 .  Secondary LOPs. Secondary LOPs are formed from hydroperoxides, which can be decomposed in the presence of metal ions or at high temperatures. Primary LOPs are nonvolatile compounds and they are odourless and tasteless but they are easily degraded into a high number of volatile and non-volatile compounds, such as carbonyls (e.g. ketones and aldehydes), alcohols, hydrocarbons (e.g. alkanes, alkenes), and furans, that play a major role in the flavour of food products (Bastos et al., 2017; McClements & Decker, 2018; Yin et al., 2020) (Figure 14). Introduction 34 In fact, their formation has been related to deterioration of quality and production of rancidity, harmful substances, and discoloration (Domínguez et al., 2019). Figure 14. Degradation of linoleic acid hydroperoxides to hydroxy-, epoxy- and oxo-fatty acids. Only segments of the structures are presented. R 1 = C 5 H 11 and R 2 = C 8 H 15 O 2 (Belitz et al., 2009). Theses secondary LOPs are classified into three main groups (Berdeaux et al., 2012): • Low molecular weight volatile compounds formed through hydroperoxide decomposition. The main route to produce these LOPs is the homolytic βscission of the alkoxyl radicals derived from allylic hydroperoxides. The alkoxyl radical undergoes C−C bond cleavage on either side of the alkoxyl group. The produced radical reacts then with either other radicals Introduction 35 or neutral molecules and either gains or losses one atom of hydrogen to form hydroxy- or keto-acids. Figure 15 shows the general scheme of the alkoxyl-radical β-scission during unsaturated fatty acid oxidation. Figure 15. Some compounds produced by β-scission of alkoxyl radicals. • Oxidized monomers of triglycerides are compounds with molecular weights similar to those of the non-oxidized triacylglycerols, but with an oxygenated function. This function is produced by transformation of the hydroperoxy group. The produced functional groups are epoxy, hydroxyl, or keto groups (Figure 14) (Belitz et al., 2009; Wong, 2018). • Dimers and oligomers of triacylglycerols formed by triacylglycerol radical polymerization reactions (Berdeaux et al., 2012). Introduction 36  Tertiary LOPs. Secondary lipid oxidation products can suffer further oxidations and new compounds with two oxygenated functions are produced. These compounds are analogous to the above described 1,2-dicarbonyl derivatives (section 1.6.1.3) and diverse combinations of functional groups have been described: Epoxyketo, epoxyhydroxy, and ketohydroxy (Lu et al., 2014). They can have the same chain length of the starting fatty acid or have been broken and being short-chain carbonyl compounds. Thus, for example, the decomposition of polyunsaturated fatty acids, produce 4,5(E)-epoxy-2(E)-heptenal or 4-oxo-2-hexenal when starting from ω3 fatty acyl chains and 4,5(E)-epoxy-2(E)-decenal or 4-oxo-2- nonenal when starting from ω6 fatty acyl chains (Figure 16). These products are common in food systems and play a major role in the colour and flavour changes produced in foods as a consequence of lipid oxidation (Hidalgo & Zamora, 2004). Figure 16. Structures of some tertiary lipid oxidation products from ω3 (left) or ω6 (right) polyunsaturated fatty acids. 1.6.2.4. Factors influencing lipid oxidation rates Lipid oxidation depends on many factors, including lipid unsaturation degree, temperature, presence of prooxidants, antioxidants, amount of oxygen, water activity, distribution of lipids in the food, chelators and sequestering agents, light, surface area, and pH (Finley & DeMan, 2018; Wong, 2018). Some of these factors are discussed below: Introduction 37 A. Temperature Lipid oxidation increases exponentially as a function of temperature (Frankel, 2014). In addition, increasing temperature produces the decomposition of hydroperoxides, which promotes the propagation phase. Temperature is strongly related to the formation and decomposition rates of primary LOPs. At 130–140ºC, decomposition prevails over formation rates. So, the most important products at elevated temperatures are secondary LPOs due to the higher stability of some of them at such temperatures (Dobarganes et al., 2016; Dobarganes & Márquez-Ruiz, 2007; Velasco & Dobarganes, 2002). Besides the rates, temperature also determines the distribution of LOPs. Formation rates of different groups of compounds when the food is processed at high temperature also depend on how that temperature has been applied. For example, roasting applied for a longer time at a lower temperature produces a higher increase in secondary lipid oxidation products than heating at a higher temperature for a shorter time (Broncano et al., 2009). On the other hand, although freezing slows down lipid oxidation, it does not stop the process. Extracellular ice crystal formation by temperature fluctuations promotes oxidation as a result of an increasing cell disruption, which release prooxidant compounds (Erickson, 2002). Moreover, oxygen solubility in liquids decreases when temperature increases and changes in partitioning of antioxidants between phases are also produced (Gordon, 2004). B. Antioxidants Antioxidants are compounds that protect lipids from oxidation. They delay or slow the rate of oxidation by inhibiting the action of catalysts or the formation of free radicals (Pereira & Abreu, 2018). Antioxidants naturally present in foods include tocopherols, ascorbic acid, carotenoids, peptides, and enzymes. In addition, they can also be added to the food system (Carocho & Ferreira, 2013). Introduction 44 or serine produces carbonyl compounds, such as acetaldehyde and/or formaldehyde. Figure 20. Thermal degradation of amino acids (Yaylayan et al., 2000). 1.6.4. Reactive carbonyls produced from phenolic compounds 1.6.4.1. Phenolic compounds Phenols are a large group of compounds characterized for having at least one aromatic ring with one or more hydroxyl groups attached. They are associated with the nutritional and sensory quality of fresh and processed plant foods, and they are found in many herbs, fruits, vegetables, grains and cereals, green and black teas, coffee beans, propolis, and red and white wines (Ho, 1992). Phenolics can be classified into three groups according to the number and arrangement of their phenolic subunits. These three groups are simple phenols or phenolic acids, hydroxycinnamic acid derivatives, and polyphenols, which possess at least two phenol subunits. This last group includes flavonoids and stilbenes, among others. Introduction 45 Simple phenols include monohydroxylic derivates such as p-cresol, o-, m-, and p-diphenols such as catechol, resorcinol, and hydroquinone, respectively, and triphenols such as gallic acid. Hydroxycinnamic acids include p-coumaric, caffeic, ferulic, and sinapic acids. Finally, flavonoids are the largest group of polyphenols in food, and there are an estimated number of 4,000 compounds of this kind known. They possess two aromatic rings, each containing at least one hydroxyl group, which are connected through a three-carbon “bridge” (Shahidi et al., 2019). 1.6.4.2. Quinones as reactive carbonyls and as a source of them Analogously to carbohydrates and lipid oxidation products, polyphenols are also able to produce the Strecker degradation of amino acids (Rizzi, 2008). In addition, they participate as antioxidants during the lipid oxidation process and contribute to the flavours of foods (Delgado et al., 2015; Ho, 1992; Shahidi & Ambigaipalan, 2015). Although there are a high number of polyphenols, only a few of them are able to produce the Strecker degradation of amino acids. The structural characteristics required for this reaction are the presence of, at least, two hydroxyl groups in ortho- or parapositions. However, the presence in the same molecule of other aromatic ring with two hydroxyl groups in metaposition either cancelled or limited the ability of these complex phenols to degrade amino acids (Delgado et al., 2015). The reason for the required structural characteristics is the need that these compounds are converted into quinones. Quinones are 1,2-dicarbonyl compounds and react with amino acids analogously to other 1,2-dicarbonyl compounds (Figure 21). Thus, they form the corresponding imine in the initial step, which after decarboxylation and hydrolysis produce the Strecker aldehyde at the same time that the phenolic compound is converted into a hydroxyphenylamine. Introduction 46 Quinones can also be produced by fermentative processes of foods as observed in tea, coffee, and cacao (Rizzi, 2008). Figure 21. Contribution of o-quinones to degradation of amino acids (Schieber, 2018). 1.7. Disappearance of reactive carbonyls in foods: Food carbonylome fate As discussed in the previous section, reactive carbonyls have different origins and are formed by different pathways. However, many of these compounds are highly reactive and they take part in different reactions, which produces their rapid clearance. Some of these reactions will be discussed in this section. 1.7.1. Carbonyl–amine reactions An important fate of reactive carbonyls is carbonyl-amine reactions. These reactions are produced with the different reactive carbonyls described in the previous section. However, these reactions have been traditionally considered different depending on the origin of the carbonyl compounds. Thus, carbonyl-amine reactions produced as a consequence of carbohydrate/amino acids reactions have been named Maillard reaction. In this reaction, when carbohydrate-derived reactive carbonyls react with amines, amino acids, aminophospholipids, and proteins, advanced Maillard reaction products are Introduction 47 produced. These products are also known as advanced glycation end-products (AGEs) (Uribarri et al., 2010). The structure of the produced AGEs (Figure 22) depends on the nature of the amino moiety. Thus, when the amine group is attached to a protein, the reaction can result in cross-linking or unchangeable modifications of the protein, which may lead to functional alteration. For example, the reaction of a pentose with lysine (Lys) and arginine (Arg) residues of proteins produces a cross-linker, named pentosidine. Others examples of cross-linkers are those resulting from the reaction between two lysine side-chains and two molecules of 1,2-dicarbonyl compounds, such as glyoxal (GO), methylglyoxal (MGO), or 3-deoxyglucosone (3-DG). The formed AGEs are glyoxal-derived lysine dimer (GOLD), methylglyoxal-derived lysine dimer (MOLD), and 3-deoxyglucosone-derived lysine dimer (DOLD), respectively. The crosslinking between arginine and lysine results in GODIC (imidazolium cross-link derived from glyoxal and lysinearginine), MODIC (imidazolium cross-link derived from methylglyoxal and lysine-arginine), and DODIC (imidazolium cross-link derived from 3- deoxyglucosone and lysine-arginine) (Akıllıoğlu & Gökmen, 2019). Figure 22. Examples of advanced glycation end-products (AGEs). Other AGEs do not produce cross-linking. The most common non-cross- linking AGEs are pyrraline and N ε -carboxymethyllysine (CML), which is used as a marker of the AGEs formed in food. CML can be formed through different pathways. Thus, CML is produced by condensation of glucose with the ε-amino group of lysine, by reaction of GO directly with the ε-amino group in lysine, and through the oxidation of fructoselysine (Poulsen et al., 2013; Yu et al., 2017). Introduction 48 Analogously to GO, methylglyoxal forms N ε -carboxyethyllysine (CEL). CEL and CML are the better-studied AGEs and together with pentoside and pyrraline are usually used as indicators of the nutritional quality of foodstuffs. The high consumption of sugars, processed foods, and of fats or proteins have resulted in an increased exposure to AGEs. As discussed above, carbohydrates are not the only source of reactive carbonyls in foods. Lipid oxidation is also a major source of reactive carbonyls, which compete with carbohydrate-derived reactive carbonyls for amino compounds. Therefore, a major cause for lipid-derived reactive carbonyls disappearance is also the production of carbonyl-amine reactions (Hidalgo & Zamora, 2000). Lipid-derived reactive carbonyls react with amines, amino acids, peptides, proteins, and aminophospholipids, and produce advanced lipoxidation end products (ALEs) (Hidalgo & Zamora, 2017). Analogously to AGEs, ALEs can be classified into two groups: Low and high molecular weight products. Low molecular weight ALEs include ε-N-pyrrolylnorleucine. This compound has been found in many fresh food products and its content is directly correlated with lipid and iron contents (Zamora et al., 1999). Other low molecular weight ALEs are heterocyclic derivatives, such as pyrroles and pyridines. Some of these low molecular weight ALEs are stable. However, others are unstable and polymerize spontaneously to produce high molecular weight ALEs, which are responsible for the browning developed by fat-rich foods upon processing and storage (Zamora & Hidalgo, 2005). Foods with the highest levels of AGEs/ALEs are red meats and poultry. In addition, higher-fat and aged cheeses contain more AGEs/ALEs than lower-fat cheeses. It is remarkable the large amounts of AGEs/ALEs in these foods in spite of being uncooked foods. This is likely due to pasteurization and/or holding times at room temperatures (e.g., as in curing or aging processes). High-fat spreads also have high contents of AGEs/ALEs. However, grains, legumes, Introduction 49 breads, vegetables, fruits, and milk have lower contents of AGEs/ALEs (Uribarri et al., 2010). Analogously to reactive carbonyls from carbohydrates and lipids, quinones also react with amino acids and proteins through carbonyl-amine reactions. Polyphenols are oxidized by polyphenol oxidases (PPOs). Polyphenol oxidases catalyse the oxidation of the colourless phenols to coloured quinones at the expense of O 2 . These quinonic compounds are associated reversibly or irreversibly with amino acids. Quinones react with the free nucleophilic functional groups of proteins or amino acids such as sulfhydryl, amine, amide, indole and imidazole substituents (Figure 23). Figure 23. Non-enzymatic reactions between oxidised polyphenols and amino acids (adapted from Yaylayan, 2003). It results in imine formation in 1,4-Michael addition via nitrogen or sulphur, and in Strecker degradation forming aldehydes. Thus, primary products are Schiff bases, N-quinonyl derivatives, S-quinonyl derivatives, and simple aldehydes evolving from the Strecker degradation, which have been described in subsection 1.5.4.2 (Yaylayan, 2003). Introduction 50 The formation and activity of quinone-amino acids conjugates influence the colour, taste, and aroma of foods. For example, chlorogenic acid (Figure 24) is the major polyphenol in foods derived from plants and is a good substrate for polyphenol oxidase. Chlorogenic acid quinone (Figure 24), which is the oxidative product of chlorogenic acid by polyphenol oxidase is an important intermediate compound in enzymic browning. In addition to the reaction of chlorogenic acid quinone with amino acids, H 2 O 2 might also play an important role in the formation of the brown colour by enzymic browning (Bittner, 2006). Figure 24. Structure of chlorogenic acid and chlorogenic acid quinone (Bittner, 2006). Quinones can also be involved in reactions with proteins. They can react as electrophilic compounds, or undergo redox cycling and cause oxide reductive stress. Thus, quinones can react as Michael acceptors to alkylate proteins. In addition, quinones also react with sulphur nucleophiles, such as glutathione (GSH) or cysteine residues on proteins, or with nucleophilic amino groups of proteins (Bolton & Dunlap, 2017; Klopčič & Dolenc, 2019). p-Quinones react via 1,4-reductive addition reactions regenerating the hydroquinone with covalent attachment to the cysteine residue. Unsubstituted oquinones generally undergo 1,6-reductive addition reactions with thiol nucleophiles due to extended conjugation, although 1,4-reductive addition is often observed as a minor product (Figure 25). Reaction with nitrogen nucleophiles such as lysine, histidine, N-terminal amino acids, and purine and pyrimidine bases on DNA are much slower compared to that of sulphur Introduction 51 nucleophilic additions, although they occur to some extent. Generally, reductive 1,4-Michael additions predominate over Schiff base formation (Bolton & Dunlap, 2017). Figure 25. Michael additions of glutathione (GSH) to p-quinone and o-quinone (Bolton & Dunlap, 2017). 1.7.2. Carbonyl-phenol reactions Analogously to amino compounds, phenolic compounds are also able to trap reactive carbonyls. The carbonyl scavenging ability of phenolic compounds has been long studied. Thus, studies in model systems observed the influence of phenolic compounds on the formation of food flavours during Maillard reaction (Totlani & Peterson, 2005). Moreover, recent studies have shown that phenolic compounds can also influence the formation of advanced glycation end-products (AGEs) and of advanced lipoxidation-end products (ALEs) by trapping or scavenging reactive carbonyls (Hidalgo, et al., 2017b; Hidalgo, et al., 2018b). Most studies have been carried out with glyoxal (GO) and methylglyoxal (MGO). When these compounds are in the presence of phenolic compounds, they are trapped and the formation of AGEs and ALEs is inhibited (Zhu et al., 2020). The trapping ability of phenolic compounds is a consequence of their structural characteristics (Hidalgo, et al., 2017a; Zamora et al., 2017). Introduction 52 Trapping sites of MGO molecules are known in the case of flavonoids (Lund & Ray, 2017). The requirements for the trapping ability of phenolic compounds are briefly summarized (Figure 26): • The positions with the highest electronic densities are the carbons at the α-position to the carbons with the hydroxyl group in the phenolic ring. Meta configuration of the phenol is more reactive than ortho or para configurations. • The hydroxyl group at C-5 on the A-ring enhances trapping efficiency. • The double bond between C-2 and C-3 on the C ring could facilitate the trapping efficiency. • The number of hydroxyl groups on the B-ring does not significantly influence the trapping efficiency. Figure 26. Structure of flavan-3-ols (Lund & Ray, 2017). In addition to the ability of phenolic compounds to scavenge carbohydratederived reactive carbonyls, they can also form carbonyl-phenol adducts with lipid-derived reactive carbonyls, such as alkanals (Hidalgo, et al., 2017a), 2- alkenals (Hidalgo & Zamora, 2014), 2,4-alkadienals (Hidalgo & Zamora, 2018), 4-hydroxy-2-alkenals (Hidalgo & Zamora, 2019), 4-oxo-2-alkenals (Hidalgo, et al., 2018a), and 4,5-epoxy-2-alkenals (Zamora et al., 2017). These carbonyltrapping reactions have been shown to occur in foods under common cooking conditions (Zamora et al., 2016). The structures of the identified carbonyl-phenol adducts produced with the different reactive carbonyls are shown in Figure 27: Introduction 53 Carbohydrate-derived carbonyl-phenol adducts Lipid-derived carbonyl-phenol adducts Figure 27. Structures of the carbonyl-phenol adducts (Zamora & Hidalgo, 2020). 1.7.3. Cyclizations of reactive carbonyls in the presence of amino compounds After reacting with amine groups, reactive carbonyls are modified and they usually evolve into heterocyclic derivatives. This happens, for example, in the Strecker degradation of amino acids (Figure 28). As discussed previously, Strecker degradation (I) produces Strecker aldehydes and α-aminoketones. The formed α-aminoketones are unstable and dimerize spontaneously to produce pyrazines (II). As observed in the Figure 28, the Aims of study 60 Materials and Methods 61 3. MATERIALS AND METHODS Materials and Methods 62 Materials and Methods 63 3.1. Materials 3.1.1. Commercial products Many different products were employed in these studies. Some of them were purchased from commercial sources and others were prepared in the laboratory. Commercial products had always the highest available grade, usually analytical grade. Distributors included Sigma-Aldrich (St. Louis, MO, USA), Alfa Aesar (Haverhill, MA), TCI (Tokyo, Japan), and Merck (Darmstadt, Germany). Different lipid derivatives were employed, including both unoxidized and oxidized lipids. The employed unoxidized lipids were linoleic acid, linolenic acid, and their methyl esters, and menhaden and linseed oils. In addition, many commercial reactive carbonyls produced in the lipid oxidation pathway were assayed. They included alkanals (ethanal, propanal, butanal, and hexanal), 2- alkenals (2-hexenal and 2-heptenal), 4-alkenals (4-pentenal), 2,4-alkadienals (2,4-hexadienal, 2,4-heptadienal, 2,4-octadienal, 2,4-nonadienal, and 2,4- decadienal), ketones (3,5-heptadien-2-one and 6-methyl-5-hepten-2-one), and dialdehydes (glyoxal, malondialdehyde, and fumaraldehyde). Among carbohydrates, monosaccharides (glycolaldehyde, arabinose, rhamnose, ribose, glucose, fructose, and galactose), disaccharides (sucrose and lactose), and reactive carbonyls derived from them [furfural, 5-hydroxymethylfurfural (HMF), and 2-acetylfuran] were also employed. Finally, as amino compounds, ammonia, ammonium chloride, glutamine, urea, and creatinine, were used. Commercial products also included standards. The followed pyridines were used: 2-Methylpyridine, 3-methylpyridine, 4-methylpyridine, 2-ethylpyridine, 3- ethylpyridine, 2-pentylpyridine, 2,5-dimethylpyridine, 2,6-dimethylpyridine, 5- ethyl-2-methylpyridine, 3-acetylpyridine, 3-hydroxy-2-methylpyridine, 3- hydroxypyridine, and 2-hydroxymethyl-5-hydroxypyridine. In addition, Materials and Methods 64 standards of PhIP, IQ, MeIQ, and MeIQx (obtained from Toronto Research Chemicals, North York, ON, Canada) were also employed. Multi-floral honey and sugarcane honey or sugarcane syrup (a traditional plant syrup, namely ‘miel de caña’) were purchased from a local supermarket. All solvents used in these studies (hexane, ethyl acetate, acetone, acetonitrile, chloroform, tetrahydrofuran, and toluene) were analytical grade and were purchased from Merck (Darmstadt, Germany). Citric acid (99.5%), boric acid, sodium hydroxide (99%), and the sodium phosphate used to prepare the different buffers, were also obtained from Merck. Silica 60 (thickness of 0.063-0.2 mm) for column chromatography was purchased from Macherey Nagel (Gmbh, Düren, Germany). 3.1.2. Syntheses of lipid hydroperoxides Two lipids hydroperoxides and their methyl esters were employed in these studies. All of them were synthesised by following a previously described procedure (Zamora & Hidalgo, 2008). 13-Hydroperoxy-9,11-octadecadienoic acid (13-LOOH) and methyl 13-hydroperoxyoctadeca-9,11-dienoate (MeLOOH) were prepared by oxidation of linoleic acid with soya lipoxygenase and later esterification with diazomethane. Analogously, 13-hydroperoxy-9,11,15- octadecatrienoic acid (13-LnOOH) and methyl 13-hydroperoxyoctadeca-9,11,15- trienoate (MeLnOOH) were obtained from linolenic acid. The followed procedure is described briefly. A solution of the corresponding fatty acid (2.1 mmol) in absolute ethanol (1 mL) was added to 20 mL water and sonicated until permanent emulsion. Then, the emulsion was treated with 2 N NaOH until the solution was cleared and taken to 200 mL with 50 mM sodium borate, pH 9. Sodium borate buffer (500 mL of 50 mM), pH 9, was oxygenated for 15 min at 0 °C by bubbling in a slow stream of oxygen. After this time, 292 units of lipoxygenase (obtained from sigma), Materials and Methods 65 dissolved in 10 mL of 50 mM sodium borate buffer, pH 9, and the fatty acid solution was added. The lipoxygenation was carried out for 10 min at 0 °C in the presence of oxygen. The reaction was stopped by lowering the pH to 3 and extracting the hydroperoxide with diethyl ether (5 × 100 mL). The organic layers were combined, dried over anhydrous sodium sulphate and evaporated. 13- LOOH and 13-LnOOH were obtained in that way. To obtain their methyl esters, diazomethane was employed. In this case, samples had to be purified by column chromatography on silica gel using hexane/diethyl ether (7:3) as the eluent. The separation was controlled by thin layer chromatography, 13-LOOH, 13-LnOOH, MeLOOH, and MeLnOOH (Figure 29) were obtained chromatographically pure. Additional confirmations of identity and purity were obtained by 1D and 2D NMR. 13 C NMR (CDCl 3 , 75.4 MHz) of MeLOOH: δ (ppm) 174.49 (C1), 133.96 (C9), 131.23 (C12), 130.11 (C11), 127.51 (C10), 86.85 (C13), 51.54 (OCH 3 ), 34.06 (C2), 32.49 (C14), 31.72 (C16), 29.35, 29.03, 29.00, 28.87 (C4-C7), 27.69 (C8), 24.98 (C15), 24.85 (C3), 22.51 (C17), and 14.04 (C18). 13 C NMR (CDCl 3 , 75.4 MHz) of MeLnOOH: δ (ppm) 174.53 (C1), 134.35 and 134.11 (C9 and C16), 130.37 (C12), 130.20 (C11), 127.50 (C10), 123.09 (C15), 86.21 (C13), 51.56 (OCH 3 ), 34.07 (C2), 30.58 (C14), 29.34, 29.03, 29.00, and 28.87 (C4-C7), 27.69 (C8), 24.85 (C3), 20.70 (C17), and 14.12 (C18) (Zamora & Hidalgo, 2008). Figure 29. Syntheses of methyl 13-hydroperoxyoctadeca-9,11-dienoate (MeLOOH) and methyl 13-hydroperoxyoctadeca-9,11,15-trienoate (MeLnOOH). Abbreviations: LOX, lipoxygenase. Materials and Methods 66 3.1.3. Syntheses of 4-hydroxy-2-alkenals Two 4-hydroxy-2-alkenals were employed in this study. They were 4- hydroxy-2-nonenal (HNE) and 4-hydroxy-2-hexenal (HHE). 4-Hydroxy-2- nonenal (Figure 30) was synthesised following the procedure of Gardner et al. (1992). Briefly, to a solution of 3-nonenol (35 mmol) in 14 mL dichloromethane, a 1.2 molar excess of 80-85% 3-chloroperoxybenzoic acid (54 mmol) was added. The solution was kept at room temperature for 1 h after which 2 mL of 10% NaHCO 3 was added with vigorous stirring for 45 min. The organic layer, containing 3,4-epoxynonanol, was oxidised with periodinane (38 mmol) in dichloromethane (150 mL). The oxidation mixture was stirred for 30 min at room temperature, and then, 550 mL of diethyl ether and 300 mL of 1.3 M NaOH were added with vigorous stirring for 10 min. The aqueous layer was removed, and the organic phase was washed with 300 mL of H 2 O. Evaporation of solvent afforded crude 4-hydroxy-2-nonenal, which was obtained chromatographically pure by column chromatography on silica gel using hexane:diethyl eter (2:1) as a solvent. The separation was controlled by thin layer chromatography. Its identity was confirmed by 1 H NMR. 1 H NMR (CDCl 3 , 300 MHz): δ 0.80 (br t, 3H), 1.4 (m, 8H), 2.58 (br s, 1H), 4.40 (m, 1H), 6.26 (ddd, J = 1.0, 7.0, 15.5, Hz, 1H), 6.80 (dd, J = 4.0, 15.5 Hz, 1H), 9.56 (d, J = 8.0 Hz, 1H). Figure 30. Synthesis of 4-hydroxy-2-nonenal (HNE). 4-Hydroxy-2-hexenal (Figure 31) was prepared from 3(Z)-hexenol according to Hidalgo & Zamora (2019) following a procedure similar to that of Gardner et al. (1992), but with significant modifications. Briefly, a solution of 3(Z)-hexenol (50 mmol) in 140 mL of dichloromethane was treated slowly with 3- Materials and Methods 67 chloroperoxybenzoic acid (9.3 g), and the obtained mixture was stirred for 1 h at room temperature. After this time, 10% potassium bicarbonate (140 mL) was added and the mixture was stirred for 2 min. The organic layer was then collected and dried over sodium sulphate. The obtained solution, which contained produced 3,4-epoxyhexanol, was slowly added to a solution of periodinane (16 g in 150 mL of dichloromethane), and the obtained mixture was stirred under dark for 30 min at room temperature. Then, 500 mL of diethyl ether and 230 mL of 1.3 M NaOH were added. The solution was vigorously stirred for 2 min, and the organic layer was collected, dried over sodium sulphate, and taken to dryness. The residue was fractionated by column chromatography on silica gel using mixtures of hexane and diethyl ether as the eluent. The separation was controlled by thin layer chromatography. Identity and purity of obtained HHE were confirmed by means of one-dimensional (1D) and two-dimensional (2D) nuclear magnetic resonance (NMR) and gas chromatography–mass spectrometry (GC– MS). 1 H NMR (CDCl 3 , 300 MHz): δ 1.02 (br t, 3H), 1.68 (m, 8H), 4.40 (m, 1H), 6.34 (ddd, J = 1.6, 7.9, 15.7 Hz, 1H), 6.81 (dd, J = 4.7, 15.7 Hz, 1H), 9.60 (d, J = 7.9 Hz, 1H). 13 C NMR (CDCl 3 ): δ 9.43 (CH 3 ), 29.47 (CH 2 CH 3 ), 72.27 (CHOH), 130.89 (CHCHO), 158.68 (CHCHOH). MS, m/z (%, ion structure): 114 (0.1, M + ), 96 (0.3, ethylfuran), 85 (60, M + – CH 3 CH 2 or CHO), 67 (6, 96 – CH 3 CH 2 ), 57 (100, 96 – C 3 H 3 ), and 55 (12, M + – CH 3 CH 2 CHOH). Figure 31. Synthesis of 4-hydroxy-2-hexenal (HHE). 3.1.4. Syntheses of 4-oxo-2-alkenals Two 4-oxo-2-alkenals were employed in this study: 4-Oxo-2-hexenal and 4- oxo-2-nonenal. Both compounds were prepared by ring opening of the corresponding 2-alkylfuran: 2-Ethylfuran and 2-pentylfuran, respectively (Figure 32) (Hidalgo et al., 2018a). Materials and Methods 68 Briefly, N-bromosuccinimide (10.7 mmol) and pyridine (10 mL) were added to a solution of 2-alkylfuran (16.6 mmol) in 11 mL of tetrahydrofuran/acetone/water (5:4:2) in an ice bath. The reaction mixture was stirred for 1 h at this temperature and, then, kept on stirring at room temperature for 2 h. After this time, the mixture was diluted with 10 mL of water and extracted three times with 50 mL of chloroform. The combined chloroformic extracts were washed successively with hydrochloric acid and water in order to remove pyridine, then dried over anhydrous sodium sulphate, and finally concentrated under vacuum. The residue was fractionated by column chromatography on silica gel 60 (230-400 mesh; Macherey-Nagel) using mixtures of hexane and ethyl acetate as eluent. The separation was controlled by thin layer chromatography. 4-Oxo-2-alkenals were obtained chromatographically pure. Confirmations of identity and purity were obtained by GC-MS by comparison with the retention times and mass spectra of authentic standard previously prepared in the laboratory. Figure 32. Syntheses of 4-oxo-2-nonenal and 4-oxo-2-hexenal. 3.1.5. Syntheses of 4,5-epoxy-2-alkenals Three 4,5-epoxy-2-alkenals (Figure 33) were employed in this study. 4,5- Epoxy-2-hexenal, 4,5-epoxy-2-heptenal, and 4,5-Epoxy-2-decenal were prepared by epoxidation of 2,4-hexadienal, 2,4-heptadienal, and 2,4-decadienal, respectively, with 3-chloroperoxybenzoic acid according to the procedure of Zamora et al. (2017). Materials and Methods 69 Briefly, 3-chloroperoxybenzoic acid (25 mmol) was dissolved in chloroform (175 mL), washed three times with 100 mL of buffer (0.2 M Na 2 HPO 4 adjusted to pH 7.5 with 0.1 M citric acid) followed by three 100 mL portions of water, and dried with anhydrous sodium sulphate. This solution was added slowly to a solution of 2,4-alkadienal (39.9 mmol) in chloroform (30 mL), which was stirred at room temperature overnight and, finally, washed three times with 100 mL of buffer (0.2 M Na 2 HPO 4 adjusted to pH 7.5 with 0.1 M citric acid), and, finally, followed by three extractions with 100 mL of water. The organic solution was dried over anhydrous sodium sulphate and concentrated under vacuum. The residue was fractionated by column chromatography on silica gel 60 (230-400 mesh; Macherey-Nagel) using mixtures of hexane and acetone as eluent. The separation was controlled by thin layer chromatography. 4,5-Epoxy-2-alkenals were obtained chromatographically pure. Confirmations of identity and purity were obtained by gas chromatography-mass spectrometry (GC/MS) comparison was carried out with retention times and mass spectra with authentic standards prepared previously in the laboratory. Figure 33. Syntheses of 4,5-epoxy-2-alkenals. Materials and Methods 76 3.3.8. Reaction of reactive carbonyls and creatinine. Formation of MeIQx The employed procedure was analogous to that described in sections 3.3.6 and 3.3.7. Briefly, mixtures of the reactive carbonyl(s) (50 µmol in 20 µmol of methanol), creatinine (50 µmol in 60 µL of water), 30 µL of 0.3 M sodium phosphate, pH 6.5, and 40 µL of water were singly homogenized with 0.063– 0.200 mm silica gel (300 mg). Samples were heated at 180 ºC in closed test tubes for 1 h. Heated samples were treated as described in section 3.3.6, and formed MeIQx was identified and determined by LC-MS/MS. 3.3.9. Reaction of acrolein-producing compounds and creatinine. Formation of MeIQx Model reactions were carried out analogously to section 3.3.8, but acrolein was replaced by acrolein-producing compounds. Assayed compounds were those that previous studies (Stevens & Maier, 2008) have shown to be acroleinproducing compounds. Studied compounds included carbohydrates (glucose, fructose, and ribose), fats (linseed oil, menhaden oil, and 13-hydroperoxide of linoleic acid), and amino acids (threonine and methionine). For linseed and menhaden oils the experimental procedure was slightly different to that employed for the other compounds. Thus, these oils were oxidized in a previous step to their reaction with creatinine to promote acrolein formation. The procedure employed for these oils was as follows: Mixtures of the oil (50 mg), 300 mg of silica gel, and 30 µL of 0.3 M of sodium phosphate, pH 6.5, were incubated for 4 days at 60 ºC in closed test tubes. After that time, 50 µmol of creatinine was added and mixtures were heated at 180 ºC for 1 h. After cooling, reaction mixtures were treated as described in section 3.3.6, and studied by LC-MS/MS. Materials and Methods 77 3.4. Identification and quantification of studied compounds 3.4.1. Identification and quantification of alkylpyridines Identification of most pyridines was carried out by comparison with retention times, mass spectra, and co-elution with authentic standards. When these standards were not available, tentative identifications were carried out on the basis of their mass spectra. Quantification of pyridines was carried out by preparing standard curves of these compounds. Seven concentration levels (0–10 µmol) for each pyridine was used and the procedure described in the section 3.3.1 was followed. Compound content was directly proportional to compound/internal standard area ratio (r 2 > 0.98, p < 0.001). RSD was always < 10%. When the commercial standards were not available, such as in the case of 2- propylpyridine or 2-butylpyridine, the concentration of the compound was estimated by using the calibration curve of the compound with the most similar chemical structure. Thus, the calibration curve of 2-ethylpyridine was used for 2- propylpyridine, and the calibration curve of 2-pentylpyridine was employed for 2-butylpyridine. 3.4.2. Identification and quantification of 3-hydroxypyridines Each compound was identified according to its mass spectrum, retention index, and by co-elution with authentic standards (when available). The M + and the (M + –methyl) ions of the 3-hydroxypyridine and the 3- hydroxy-2-alkylpyridines were employed for identification purposes. The second one, which was the base ion for most of studied pyridines, was used for quantification purposes. Materials and Methods 78 Linear retention indexes for the different hydroxypyridines increased linearly (r 2 = 0.994, p < 0.0001) as a function of the chain length of the alkyl substituent as observed in Figure 34. Figure 34. Linear retention indexes (LRI) of the trimethylsilyl derivatives of the hydroxyalkylpyridines. Mass spectra of the trimethylsilyl derivatives of the different studied hydroxypyridines are shown in Figure 35. Their mass spectra, relative intensities, and ion structures for the different ions were: 3-Hydroxypyridine, m/z (relative intensity, ion structure): 167 (32, M + ), 152 (100, M + –methyl), 73 (17, trimethylsilyl). 3-Hydroxy-2-methylpyridine, m/z (relative intensity, ion structure): 181 (40, M + ), 166 (100, M + –methyl), 73 (45, trimethylsilyl). 2-Ethyl-3-hydroxypyridine, m/z (relative intensity, ion structure): 195 (5, M + ), 180 (100, M + –methyl), 166 (23, M + –ethyl), 165 (22, M + –ethane), 73 (54, trimethylsilyl). 3-Hydroxy-2-propylpyridine, m/z (relative intensity, ion structure): 209 (5, M + ), 194 (100, M + –methyl), 166 (31, M + –propyl), 165 (26, M + –propane), 73 (46, trimethylsilyl). Materials and Methods 79 2-Butyl-3-hydroxypyridine, m/z (relative intensity, ion structure): 223 (8, M + ), 208 (100, M + –methyl), 180 (10, M + –propyl), 166 (40, M + –butyl), 165 (22, M + – butane), 73 (54, trimethylsilyl). 3-Hydroxy-2-pentylpyridine, m/z (relative intensity, ion structure): 237 (1, M + ), 222 (6, M + –methyl), 208 (14, M + –ethyl), 194 (23, M + –propyl), 181 (42, M + –butene), 180 (41, M + – butyl), 166 (100, M + –pentyl), 73 (63, trimethylsilyl). Figure 35. Mass spectra of the trimethylsilyl derivatives of the studied hydroxyalkylpyridines. Materials and Methods 80 In studies involving the formation of 2-hydroxymethyl-5-hydroxypyridine from HMF, the identification of both compounds was carried out by comparison of retention indexes and mass spectra, and by co-elution with authentic standards which were derivatizated analogously. The M + (m/z 269 and 198) and M + –1 (m/z 254 and 183) ions of 2-hydroxymethyl-5-hydroxypyridine and HMF, respectively, were employed for identification purposes. The second one was used for quantification purposes. Quantification of all 3-hydroxypyridines and HMF in model and food systems were carried out by preparing standard curves of all these compounds in the corresponding assayed systems and following the whole procedures described in sections 3.3.2, 3.3.3, 3.3.4, and 3.3.5 (without heating). Different concentration levels of each compound were used to prepare the standard curves. Compound contents were directly proportional to compound/internal standard area ratio (r 2 > 0.98, p < 0.001). The coefficients of variation at the different assayed concentrations were < 10%. When a standard of hydroxypyridine derivative was not available, the quantification of this compound was carried out by using the calibration curve of the available compound with the highest structural similarity. 3.4.3. Identification and quantification of heterocyclic aromatic amines Heterocyclic aromatic amines (MeIQ, IQ, and MeIQx) were determined by LC-MS/MS in the samples prepared as described in sections 3.3.6, 3.3.7, 3.3.8, and 3.3.9, respectively. Three transitions were acquired for the identification of each compound and caffeine was employed as internal standard. To establish the appropriate MRM conditions, the mass spectrometric conditions of authentic standards were optimized by using infusion with a syringe pump. Precursor and product ions used for qualification and quantification purposes, and operating conditions for each compound are summarized in Table 6. The first transition was employed for quantification and the others were used to confirmation purposes. Materials and Methods 81 Quantitative determinations of heterocyclic aromatic amines were carried out by preparing calibration curves of authentic standards and following the whole procedures described in sections 3.3.6, 3.3.7, 3.3.8, and 3.3.9. For each curve, six concentration levels of the different heterocyclic aromatic amines were used. Heterocyclic aromatic amine content was directly proportional to the heterocyclic aromatic amine/internal standard area ratio (r 2 > 0.986, p < 0.0001). The coefficients of variation at the different concentrations were always < 15%. Table 6. MRM transitions employed for the detection of IQ, MeIQ, and MeIQx. HAAs Transition DP FP EP CEP CE CXP IQ 199.5 185.1 46 370 12 16 37 6 199.5 157.1 46 370 12 16 51 6 199.5 158.1 46 370 12 16 45 6 MeIQ 213.9 199.1 46 370 10 16 37 6 213.9 198.1 46 370 10 16 45 8 213.9 171.2 46 370 10 16 55 8 MeIQx 215.0 200.1 41 370 10 16 37 6 215.0 132.1 41 370 10 16 53 4 215.0 77.0 41 370 10 16 77 2 Caffeine (IS) 195.2 138.0 31 350 10.5 12 27 6 195.2 110.1 31 350 10.5 12 33 4 195.2 42.1 31 350 10.5 12 55 4 Abbreviations: DP, declustering potential; FP, focusing potential; EP, entrance potential; CEP, collision cell entrance potential; CE, collision energy; CXP, collision cell exit potential. 3.5. Statistical analyses All data given are mean ± SD values of, at least, three independent experiments. Analysis of variance was employed to compare group differences. When significant F values were obtained, group differences were evaluated by the Tukey test (Snedecor & Cochran, 1980). Statistical comparisons were carried out using Origin® v. 7.0 (OriginLab Corporation, Northampton, MA). The significance level is p < 0.05 unless otherwise indicated. Materials and Methods 82 Results 83 4. RESULTS Results 84 Results 85 This section is divided into two parts. The first part is dedicated to the role that cyclization and oligomerization reactions of reactive carbonyls have in the formation of food flavours, in particular pyridines. The second part is dedicated to cyclization reactions of reactive carbonyls in the presence of creatinine. 4.1. Cyclization and oligomerization of reactive carbonyls: A route for the production of pyridines in foods Pyridines are produced as a consequence of food processing and they contribute to the characteristic sensory properties of food products, although they are not major food flavours. This section describes the cyclizations and ringexpansions of reactive carbonyls (2-alkenals, 2,4-alkadienals, 4,5-epoxy-2- alkenals, HMF, furfural, and 2-acetylfuran) in the presence of ammoniaproducing compounds. 4.1.1. Cyclization and oligomerization of reactive carbonyls in the presence of ammonia-producing compounds 4.1.1.1. Formation of pyridines by oligomerization of crotonaldehyde in the presence of ammonia-producing compounds When crotonaldehyde was heated in presence of amino compounds, the formation of pyridines was observed (Table 7). As can be observed, four pyridines were produced: 2-Methyl, 3-methyl-, 2,5-dimethyl, and 5-ethyl-2- methyl-pyridine. The same four pyridines were always produced independently of the employed amino compound (as expected, pyridines were not produced when the amino compound was absent, data not show). In addition, the highest amounts of pyridines were produced when ammonia was added. This suggested that the contribution of amino compounds was related to their ability of producing ammonia. Results 92 4.1.2.1. Formation of 3-hydroxypyridines by lipid-derived reactive carbonyls in the presence of ammonia and ammonia-producing compounds A. Formation of 3-hydroxypyridines by cyclization of lipid-derived reactive carbonyls In an attempt to investigate the possibility that lipid-derived reactive carbonyls were also involved in the formation of 3-hydroxypyridines, different aldehydes were heated in the presence of glutamine and the formation of 3- hydroxypyridines was studied. The reactive carbonyls screened included alkanals (propanal and hexanal), 2-alkenals (2-hexenal and 2-heptenal), 2,4-alkadienals (2,4-hexadienal, 2,4-heptadienal, 2,4-octadienal, 2,4-nonadienal, and 2,4- decadienal), 4-hydroxy-2-alkenals (4-hydroxy-2-hexenal and 4-hydroxy-2- nonenal), 4-oxo-2-alkenals (4-oxo-2-hexenal and 4-oxo-2-nonenal), and 4,5- epoxy-2-alkenals (4,5-epoxy-2-hexenal, 4,5-epoxy-2-heptenal, and 4,5-epoxy-2- decenal). Obtained results showed that 4,5-epoxy-2-alkenals and 2,4-alkadienals were the lipid-reactive carbonyls responsible for the formation of 3- hydroxypyridines. As an example of the obtained results, Figure 36 shows the production of 3-hydroxy-2-methylpyridine in mixtures containing glutamine and different aldehydes with the same number of carbon atoms (hexanal, 2-hexenal, 2,4-hexadienal, 4-hydroxy-2-hexenal, 4-oxo-2-hexenal, and 4,5-epoxy-2- hexenal). As observed in the Figure 36, 2,4-hexadienal and 4,5-epoxy-2-hexenal produced 3-hydroxy-2-methylpyridine to a similar extent when the reaction was carried out under air. However, when the reaction was carried out under nitrogen, the amount of 3-hydroxy-2-methylpyridine produced by 2,4-hexadienal was considerably reduced. Nevertheless, the amount of 3-hydroxy-2-methylpyridine produced by 4,5-epoxy-2-hexenal remained unchanged. This suggested that 2,4- hexadienal had to be oxidized previously to be converted into 3-hydroxypyridine. Results 93 Figure 36. Effect of different lipid oxidation products on 3-hydroxy-2- methylpyridine formation. Abbreviations: HxA, hexanal; HxE, 2-hexenal; HxD, 2,4-hexadienal; HxD/N, 2,4-hexadienal under nitrogen; HHE, 4-hydroxy-2- hexenal; OHx, 4-oxo-2-hexenal; EHx, 4,5-epoxy-2-hexenal; and EHx/N, 4,5- epoxy-2-hexenal under nitrogen; LOP, lipid oxidation product. Means values with different letters are significantly different (p < 0.05) from three independent experiments. Because 2,4-alkadienals and 4,5-epoxy-2-alkenals produced 3-hydroxy-2- methylpyridine similarly, the rest of the studies described in this reaction was carried out with 2,4-alkadienals because they are commercial products. In addition to the kind of the lipid-derived reactive carbonyl, the effect of chain length on the formation of 3-hydroxypyridines was also studied and the results are shown in Figure 37. Five 2,4-alkadienals were studied comparatively: 2,4-Hexadienal, 2,4-heptadienal, 2,4-octadienal, 2,4-nonadienal, and 2,4- decadienal. The corresponding alkadienal with five carbons (2,4-pentadienal) could be neither obtained from commercial sources nor prepared in the laboratory. Instead of 2,4-pentadienal, 4-pentenal was employed. 4-Pentenal is also a lipid oxidation product and it has been shown to be produced as a consequence of lipid oxidation of pork meat (Park et al., 2007), for example. Results 94 All assayed reactive carbonyls produced 3-hydroxypyridines that differ in the alkyl chain length present at position 2. Thus, 4-pentenal produced 3- hydroxypyridine, 2,4-hexadienal produced 3-hydroxy-2-methylpyridine, 2,4- heptadienal produced 2-ethyl-3-hydroxypyridine, and so on. As observed in the Figure 37, the amount of 3-hydroxypyridines produced was different for each alkadienal. The highest yield was obtained with 2,4-hexadienal and this yield decreased linearly (r 2 = 0.99, p = 0.997) as a function of the alkadienal chain length. As observed, the amount of 3-hydroxypyridine produced by 4-pentenal was far from the regression line, which is likely a consequence of the different chemical structure of this aldehyde in relation to that of 2,4-alkadienals. Figure 37. Effect of the chain length of the 2,4-alkadienal on the amount of 3- hydroxypyridines produced. The compounds tested were 2,4-hexadienal (6 carbon atoms), 2,4-heptadienal (7 carbon atoms), 2,4-octadienal (8 carbon atoms), 2,4- nonadienal (9 carbon atoms), and 2,4-decadienal (10 carbon atoms). The compound with 5 carbon atoms tested was 4-pentenal because the corresponding 2,4-alkadienal was not available. Abbreviation: LOP, lipid oxidation product. Results 95 B. Effect of ammonia and ammonia-producing compounds on the formation of 3-hydroxy-2-methylpyridine by cyclization of 2,4-hexadienal. The effect of ammonia and ammonia-producing compounds on the formation of 3-hydroxy-2-methylpyridine was also studied. The obtained results are shown in Figure 38. As ammonia-producing compounds, ammonia, ammonium chloride, urea, glutamine, alanine, and creatinine were selected. All of them are well-known ammonia producers. Figure 38. Effect of the type of ammonia-producing compound on 3-hydroxy-2- methylpyridine formation by cyclization of 2,4-hexadienal. Abbreviations: AmmCl, ammonium chloride; Gln, glutamine; Ala, alanine; CRN, creatinine. Means values with different letters are significantly different (p < 0.05). As observed in Figure 38, all of them produced the formation of 3-hydroxy- 2-methylpyridine. However, this pyridine was produced to different extents depending on the employed ammonia-producing compounds. The compounds that most contributed to pyridine formation were ammonia and ammonium chloride. Following these two compounds, in a decreasing order, appeared glutamine, urea, alanine, and creatinine. Because glutamine was the ammoniaproducing compound employed in the study of the formation of alkylpyridines Results 96 described in section 4.1.1.1, this amino acid was also selected for the experiments carried out in this section. C. Effect of reaction conditions on the formation of 3-hydroxy-2-methylpyridine by cyclization of 2,4-hexadienal. This subsection describes the effect of pH, precursors, and reaction time and temperature, on the formation of 3-hydroxy-2-methylpyridine by cyclization of 2,4-hexadienal. The amount of 3-hydroxy-2-methylpyridine produced depended on the pH of the reaction (Figure 39). Thus, the amount of pyridine increased as a function of reaction pH from pH 3 until pH 7, and decreased afterwards. Figure 39. Effect of reaction pH on 3-hydroxy-2-methylpyridine formation by cyclization of 2,4-hexadienal in the presence of glutamine. The employed buffers were sodium citrate (○), sodium phosphate (△), and sodium borate (▽). As observed in the Figure 39, the reaction was favoured in the presence of phosphate buffer in relation to borate buffer. The maximum yield was obtained at pH 7-8. Results 97 The concentration of the reaction precursors (2,4-hexadienal and glutamine) also influenced amount of the 3-hydroxy-2-methylpyridine formed. Thus, the amount of 3-hydroxy-2-methylpyridine produced increased linearly (r 2 = 0.98, p < 0.0001) between 0 and 25 µmol of 2,4-hexadienal when this aldehyde was heated in the presence of 5 µmol of glutamine (Figure 40). Higher amounts of aldehyde did not produce further increases in the concentration of the 3-hydroxy- 2-methylpyridine. Figure 40. Effect of aldehyde concentration on 3-hydroxy-2-methylpyridine formation by cyclization of 2,4-hexadienal in the presence of glutamine. Analogously, the amount of 3-hydroxy-2-methylpyridine produced also depended on the amount of glutamine added (Figure 41). Thus, in the absence of glutamine, the pyridine was not produced. However, when a small amount of glutamine was added, the 3-hydroxy-2-methylpyridine was produced to a significant extent. The highest yield for the reaction was achieved when 20 µmol of 2,4-hexadienal were heated in the presence of 5 µmol glutamine. Higher amounts of glutamine decreased the amount of 3-hydroxy-2-methylpyridine produced. Thus, when 10 µmol of glutamine was added, the amount of the produced pyridine was reduced to almost one half, and further amounts of the Results 98 amino acid continued decreasing the amount of the produced pyridine. In fact, the amount of the produced pyridine decreased linearly (r 2 = 0.984, p = 0.0009) when the amount of added glutamine increased from 10 to 30 µmol. According to the obtained results, the optimal amount of precursors for 3-hydroxy-2- methylpyridine formation were 20 µmol of 2,4-hexadienal and 5 µmol of glutamine. Figure 41. Effect of amino acid concentration, on 3-hydroxy-2-methylpyridine formation by cyclization of 2,4-hexadienal in the presence of glutamine. Finally, the effect of reaction time and temperature were also studied. 3- Hydroxy-2-methylpyridine concentration increased linearly (r 2 > 0.98, p < 0.005) as a function of heating time between 100 and 180 ºC (Figure 42). In addition, reaction rates, which were calculated from the slopes of the obtained lines, increased as a function of the temperature. When these reaction rates were employed in an Arrhenius plot, a correlation was observed (r 2 = 0.988, p = 0.0055). From the slope of this line, an activation energy of 49.7 ± 4.5 kJ/mol was calculated for the formation of 3-hydroxy-2-methylpyridine from 2,4- hexadienal and glutamine (Figure 43). Results 99 Figure 42. Effect of time and temperature on the formation of 3-hydroxy-2- methylpyridine. The temperatures assayed were: 100 (□), 120 (○), 140 (△), 160 (▽), and 180 (◇) °C. Figure 43. Arrhenius plot obtained for the formation of 3-hydroxy-2- methylpyridine. Results 100 4.1.2.2. Formation of pyridines in oxidized oils heated in the presence of glutamine. Previous results showed that 3-hydroxypyridines were produced by lipidderived reactive carbonyls. To confirm that lipid oxidation was involved in the formation of this type of pyridines, two polyunsaturated oils (linseed and menhaden oils) were oxidized and heated in the presence of glutamine. Both oils are known to be a good source of reactive carbonyls as a consequence of oxidation (Dlugogorski et al., 2012). Thus, the oils were oxidized under air for 5 days at 60 °C. After this time, both oils were oxidized (oxidation of the oils was verified by the appearance of hexanal in the gas chromatogram of the incubated oils, data not shown). These oxidized oils were then mixed with glutamine and heated at 100 °C for 22 h. After this time, samples were derivatizated and the amount of 3- hydroxypyridines determined. Obtained data are shown in Table 9. Table 9. 3-Hydroxypyridines produced by oil oxidation in the presence of glutamine. Oil 3-Hydroxypyridine Amount (nmol/g of oil) Linseed oil 3-hydroxypyridine 62 ± 11 nmol 3-hydroxy-2-methylpyridine 34 ± 4 nmol 2-ethyl-3-hydroxypyridine 32 ± 5 nmol Menhaden oil 3-hydroxypyridine 117 ± 11 nmol 3-hydroxy-2-methylpyridine 18 ± 1 nmol 2-ethyl-3-hydroxypyridine 13 ± 2 nmol Values are mean ± standard deviation (SD) for three independent experiments. As can be observed, menhaden oil produced mainly 3-hydroxypyridine. Although this pyridine was also the main hydroxypyridine produced by linseed oil, this oil also produced 3-hydroxy-2-methylpyridine and 2-ethyl-3- hydroxypyridine to a significant extent. Linseed oil produced these last pyridines to a higher extent (p < 0.05) than menhaden oil. Results 101 4.1.3. Conversion of 2-oxofurans into 3-hydroxypyridines Previous section has shown that 3-hydroxypyridines can be produced as a consequence of the lipid oxidation pathway. However, they can also be produced as a consequence of Maillard reaction. 2-Oxofurans are common products of Maillard reaction, in particular 5-hydroxymethylfurfural (HMF). However, when these compounds are heated in the presence of ammonia, they can suffer a ring expansion and the formation of 3-hydroxypyridines can be produced. 4.1.3.1. Conversion of HMF into 2-hydroxymethyl-5-hydroxypyridine When HMF was heated in the presence of ammonia or ammonium chloride, the formation of 2-hydroxymethyl-5-hydroxypyridine was observed. The reaction yield depended on reaction conditions, including pH, concentrations of reactants, and time and temperature. Figure 44 shows the effect of pH on both the formation of 2-hydroxymethyl- 5-hydroxypyridine and the disappearance of HMF. As observed in Figure 44, 2- hydroxymethyl-5-hydroxypyridine formation was parallel to HMF disappearance. Thus, 2-hydroxymethyl-5-hydroxypyridine formation increased when the pH increased from 3 to 7, with a maximum between pH 6–8 with sodium phosphate buffer. HMF disappearance followed a similar trend. However, this trend was not observed in the presence of borate buffer. When using this buffer, the amount of produced 2-hydroxymethyl-5-hydroxypyridine was lower and an effect in relation to pH was not observed. In spite of this lower production of the hydroxypyridine, HMF was decomposed, but its decomposition did not produce the hydroxypyridine as observed in the presence of citrate or phosphate buffers. Between pH 3 and pH 8 the formation of 2-hydroxymethyl-5- hydroxypyridine and the disappearance of HMF were correlated (r 2 = 0.9216, p = 0.002). The conversion yield of the furan into the hydroxypyridine depended on the pH and ranged from 0.4% at pH 3 to 1.3% at pH 7–8. Results 108 Figure 52. Formation of 3-hydroxypyridines: 2-Hydroxymethyl-5- hydroxypyridine (stripped bars), 3-hydroxypyridine (open bars), and 3-hydroxy-2- methylpyridine (crosshatched bars) from 2-oxofurans (HMF, furfural, and 2- acetylfuran, respectively). 4.1.3.3. Formation of 3-hydroxypyridines in honeys Previous results showed that conversion of 2-oxofurans into 3- hydroxypyridines occurs easily when these compounds are heated in presence of ammonia. To confirm that this reaction can also take place in foods, a food prone to suffer Maillard reaction, such as honey, was studied. In these studies, honeys were heated in the absence of additives. Thus, ammonia was not added because honeys contain compounds that can produce ammonia upon heating. Two different types of honey were assayed. One of them was a high quality honey, which had not suffered any previous thermal overheating. The other was a sugarcane honey, which is heated during its production process and develops Maillard reaction. Results 109 Both honeys were heated at 60 and 100 ºC and the formation of HMF and 3- hydroxypyridines was studied as a function of the heating time. Figure 53 shows the effect of heating on the high quality honey. Figure 53. Effect of heating time on the formation of 2-hydroxymethyl-5- hydroxypyridine (○) and HMF (△) in honey heated at A, 60, and B, 100 °C. Honey samples (1 g) were heated in closed test tubes for the indicated times and temperatures. Results 110 As expected, very small amounts of HMF and 2-hydroxymethyl-5- hydroxypyridine were detected in this honey at the initial step, due to the absence of heating during its production. However, when the honey was heated at 100ºC, HMF and 2-hydroxymethyl-5-hydroxypyridine were produced to a significant extent. In addition, trace amounts of 3-hydroxy-2-methylpyridine were also detected. On the other hand, formation of 3-hydroxypyridine was not observed. Significant differences were observed between the formation kinetics of HMF and 2-hydroxymethyl-5-hydroxypyridine. Thus, HMF increased exponentially as a function of heating time, as observed previously (Chua, 2018). However, 2- hydroxymethyl-5-hydroxypyridine concentration increased linearly (r 2 = 0.8836, p < 0.0002). Results were different when sugarcane honey was studied (Figure 54). As expected because of the treatment at which this honey had been submitted during its production processing, 2-oxofurans and 3-hydroxypyridines were already present in the commercial honey. In addition, 2-oxofurans and 3- hydroxypyridines concentrations increased linearly (r 2 > 0.8649, p < 0.0008) upon heating. In fact, there was a correlation between them at both 60 (r 2 > 0.9409, p = 0.005) and 100 ºC (r 2 > 0.9409, p < 0.0001). Surprisingly, 2- hydroxymethyl-5-hydroxypyridine/HMF ratio was higher at 60 ºC (~0.27) than at 100 ºC (~0.034). In addition to 2-hydroxymethyl-5-hydroxypyridine, small amounts of 3- hydroxypyridine and 3-hydroxy-2-methylpyridine were also present in sugarcane honey at the initial step and their concentrations increased linearly (r 2 > 0.81, p < 0.002) as a function of heating time and temperature. However, and analogously to the observed in the model systems and the honey, the concentrations of both pyridines were always lower than that of 2-hydroxymethyl-5-hydroxypyridine. Results 111 Figure 54. Effect of heating time on the formation of 2-hydroxymethyl-5- hydroxypyridine (○), HMF (△), 3-hydroxypyridine (□), and 3-hydroxy-2- methylpyridine (◇) in sugarcane honey heated at A, 60, and B, 100 °C. Sugarcane honey (1 g) was heated in closed test tubes for the indicated times and temperatures. Results 112 4.2. Cyclization and oligomerization of the reactive carbonyls in presence of creatinine: Formation of heterocyclic aromatic amines Section 4.1 has shown that reactive carbonyls are prone to suffer cyclizations and oligomerizations to produce heterocyclic derivatives, in particular pyridines, when heated in the presence of ammonia and ammonia-producing compounds. However, if this heating is produced in the presence of other nucleophiles, these compounds might also take part in the reaction and the formation of alternative compounds might be produced. As an example, when 2,4-hexadienal was heated in the presence of creatinine a much lower amount of 3-hydroxy-2-methylpyridine was obtained than when ammonia was employed (Figure 38). This might be a consequence of a low production of ammonia from creatinine, but also as a consequence of reaction between creatinine and the aldehyde. This would modify the aldehyde cyclizations described in section 4.1 and other products might be produced. This section describes the cyclization reactions of reactive carbonyls in the presence of creatinine. The formation of heterocyclic aromatic amines with the structure of aminoimidazoazarene is produced. 4.2.1. Reactive carbonyls and the formation of MeIQ This subsection is dedicated to the identification of precursors and reaction conditions that promote the formation of MeIQ. 4.2.1.1. Formation of MeIQ in reaction mixtures of reactive carbonyls, creatinine, and amino compounds When most reactive carbonyls were heated in the presence of creatinine and glutamine the formation of MeIQ was not detected in most assayed reaction mixtures (Table 10). However, when crotonaldehyde was present in the reaction mixtures, the formation of MeIQ was observed. Results 113 Table 10. MeIQ produced in mixtures of reactive carbonyls and amino compounds. Reactions of RCO with glutamine Reactions of Crot with amino compounds Reactive carbonyl MeIQ (nmol/mmol of RCO) Amino compound MeIQ (nmol/mmol of Crot) Carbohydrates n.d. a None 37 ± 7 a,b,c Fatty acids and esters n.d. a Alanine 64 ± 7 c,d Lipid hydroperoxides n.d. a Valine 34 ± 6 a,b,c Formaldehyde n.d. a Leucine 35 ± 6 a,b,c Acetaldehyde n.d. a Isoleucine 40 ± 9 a,b,c Propanal n.d. a Proline n.d. a Hexanal n.d. a Phenylalanine 24 ± 10 a,b,c Acrolein n.d. a Tryptophan 30 ± 5 a,b,c Crotonaldehyde 202 ± 22 b Methionine 48 ± 7 b,c 2-Pentenal n.d. a Glycine 40 ± 8 a,b,c 2-Octenal n.d. a Serine 109 ± 17 e 2,4-Heptadienal n.d. a Threonine 57 ± 6 b,c,d 2,4-Octadienal n.d. a Cysteine 99 ± 12 d,e 2,4-Decadienal n.d. a Tyrosine 41 ± 9 a,b,c 4-Hydroxy-2-hexenal n.d. a Asparagine 123 ± 20 e 4-Hydroxy-2-nonenal n.d. a Aspartic acid 99 ± 18 d,e 4-Oxo-2-hexenal n.d. a Glutamine 202 ± 22 f 4-Oxo-2-nonenal n.d. a Glutamic acid 108 ± 12 e 4,5-Epoxy-2-heptenal n.d. a Lysine 25 ± 8 a,b,c 4,5-Epoxy-2-decenal n.d. a Arginine 45 ± 2 b,c Glyoxal n.d. a Histidine 46 ± 5 b,c Malondialdehyde n.d. a Allylamine 18 ± 4 a,b Fumaraldehyde n.d. a Ammonia 191 ± 28 f Furfural n.d. a Ammonium chloride 195 ± 33 f Acet/Crot (1:1) $ 192 ± 21 b Acet/Crot (2:1) $ 140 ± 33 c Means in the same column with different letter are significantly (p < 0.05) different. Abbreviations: Acet, acetaldehyde; Crot, crotonaldehyde; RCO, reactive carbonyl; n.d., not detected. $ Data are given in nmol/mmol of crotonaldehyde. Results 114 Furthermore, when the same amount of acetaldehyde and crotonaldehyde were heated in the presence of creatinine and glutamine, the amount of MeIQ produced was not significantly different (p < 0.05) to that produced in absence of acetaldehyde. However, in the presence of a double amount of acetaldehyde, the formation of MeIQ was reduced. Table 10 also shows the MeIQ produced in mixtures of reactive carbonyls and amino compounds. Differently to the observed with reactive carbonyls, when different amino compounds were assayed, all of them produced MeIQ, including the control (only creatinine was added). However, significant (p < 0.05) differences among the assayed amino compounds were observed. The compounds that produced MeIQ to the highest extent were glutamine, ammonia, and ammonium chloride. In addition, the following compounds also produced more MeIQ than the control: Asparagine, serine, glutamic acid, aspartic acid, and cysteine. Because glutamine was used previously to induce cyclization of reactive carbonyls, this compound was also selected for studying the formation of heterocyclic aromatic amines. 4.2.1.2. Effect of reaction conditions on the formation of MeIQ in reaction mixtures of crotonaldehyde, creatinine, and glutamine Once identified the precursor of MeIQ, the effect of the concentration of the reactants and the reaction conditions on MeIQ formation was studied. The studied parameters were pH, reactant concentrations (amount of crotonaldehyde, creatinine and glutamine), water content, presence of oxygen, and time and temperature. The effect of reaction pH on MeIQ formation is shown in Figure 55. MeIQ was mainly produced at almost neutral pH and in the presence of phosphate buffer. When pH increased, a lineal decrease in the amount of MeIQ produced between pH 6.5 and pH 9 was observed (r 2 = 0.994, p = 0.003). The maximum Results 115 amount of MeIQ was obtained at pH close to 6.5. This was the pH selected for the other experiments. Figure 55. Effect of reaction pH on MeIQ formation by reaction of crotonaldehyde, creatinine, and glutamine. The employed buffers were sodium citrate (□), sodium phosphate (○), and sodium borate (△). MeIQ formation also depended on the concentration of the reactants. As shown in Figure 56, MeIQ was not produced in the absence of crotonaldehyde. However, even small amounts of crotonaldehyde induced MeIQ formation. In fact, the amount of MeIQ produced increased linearly (r 2 = 0.974, p < 0.0001) as a function of the amount of crotonaldehyde added in the range 0–80 µmol (Figure 56). Additional increases of crotonaldehyde concentration produced the decrease of the MeIQ formed. A similar behaviour was observed when different amounts of creatinine were added (Figure 57). In the absence of creatinine, MeIQ was not detected. In addition, MeIQ concentration increased linearly (r 2 = 0.996, p < 0.0001) as function of creatinine concentration in the range 0–20 µmol. Results 116 Figure 56. Effect of the concentration of crotonaldehyde on the MeIQ produced by reaction of crotonaldehyde, creatinine, and glutamine. Abbreviation: CRN, creatinine. Figure 57. Effect of the concentration of creatinine on the MeIQ produced by reaction of crotonaldehyde, creatinine, and glutamine. Abbreviation: CROT, crotonaldehyde. Results 117 The maximum amount of MeIQ obtained was produced when 20 µmol of creatinine were heated in the presence of 80 µmol of crotonaldehyde. Additional amounts of creatinine did not increase the amount of produced MeIQ. However, differently to that observed with the addition of crotonaldehyde, the amount of produced MeIQ did not decrease. Differently to crotonaldehyde and creatinine, MeIQ was produced in the absence of glutamine. In addition, only small amounts of glutamine were needed to achieve the maximum of MeIQ and additional amounts of glutamine decreased the MeIQ produced (Figure 58). In fact, the maximum amount of MeIQ was produced around 12 µmol. Figure 58. Effect of the concentration of glutamine on the MeIQ produced by reaction of crotonaldehyde, creatinine, and glutamine. Abbreviation: CROT, crotonaldehyde. MeIQ formation also depended on the amount of water added to the system (Figure 59). The maximum concentration of MeIQ was produced when 35–55 µL of water were added, and the MeIQ formed rapidly decreased when higher amounts of water were added.