Review Recent Advances in the Chemistry of Strecker Degradation and Amadori Rearrange

Food Sci. Technol. Res., 9 (1), 1–6, 2003 Review

Recent Advances in the Chemistry of Strecker Degradation and Amadori Rearrange- ment: Implications to Aroma and Color Formation

Varoujan A. YAYLAYAN

McGill University, Department of Food Science and Agricultural Chemistry 21,111 Lakeshore, Ste. Anne de Bellevue, Quebec, Canada, H9X 3V9

Received July 18, 2002; Accepted September 5, 2002

The importance of Strecker degradation lies in its ability to produce Strecker aldehydes and 2-aminocarbonyl compounds, both are critical intermediates in the generation of aromas during Maillard reaction, however, they can also be formed independently of the pathways established for Strecker degradation. Strecker aldehyde can be formed directly either from free amino acids or from Amadori products. Several pathways have been proposed in the literature for the mechanism of this transformation. On the other hand, Amadori or Heyns rearrangements of ammonia with reducing sugars can also generate 2-aminocarbonyl compounds without the formation of Strecker aldehyde. In addition, isomerization of the imine bond of the Schiff base formed between a reducing sugar and an amino acid, can initiate a transamination reaction and convert the amino acid into the corresponding -keto acid and the sugar into its -amino alcohol derivative. The reverse of this reaction, has been documented to produce Amadori products. The - keto acids can either decarboxylate to produce Strecker aldehydes or undergo Strecker degradation (as a -dicarbonyl compound) with amino acids to also produce Strecker aldehydes. This review will examine the role of Strecker degradation and Amadori rearrangement, under the light of recent ﬁndings, in controlling the balance among four critically important key intermediates: -dicarbonyl, -hydroxycarbonyl, 2-amino carbonyls and 2-(amino acid)-carbonyl compounds, during the Maillard reaction and hence control relative importance of aromagenic versus chromogenic pathways.

Keywords: Maillard reaction, Strecker degradation, Amadori rearrangement, aroma, browning, melanoidin, mechanisms

Although Strecker degradation (SD) has been delegated to a similar to the -hydroxy carbonyl compound (2) during Amadori “sub-reaction” category in the Maillard reaction scheme, how- rearrangement (AR), undergoes reductive amination and is con- ever, in its broader definition, it may play a more critical role, verted into 2-amino ketone (3), instead of its amino acid analog than what currently is assumed, in switching the Maillard reac- (4) as in the case of Amadori rearrangement process (see Fig. 1). tion towards the direction of aromagenic rather than chromoge- In addition, Amadori and Heyns rearrangements of free ammo- nic pathways. Similar to other sub-reactions occurring during the nia with -hydroxy carbonyl compounds also produce identical Maillard reaction, it has been discovered (Strecker, 1862) before intermediates to that of the SD of amino acids with correspond- the establishment of the mechanism of Amadori rearrangement ing -dicarbonyl compound, without, of course, the benefit of itself (Amadori, 1925) by Kuhn and Weygand (1937). Historical- formation of Strecker aldehyde (Fig. 1). Therefore, both process- ly, however, sugar amine reactions were investigated as early as es (SD and AR) serve the same purpose of reductively aminating 1866 by H. Schiff and later by E. Fischer (Wrodnigg & Eder, different sugar fragments (-dicarbonyls vs -hydroxy carbon- 2001) before the reaction was elevated to the status of indepen- yls) by the action of the amino acid (see Fig. 1). The question dent research field by Maillard (Maillard, 1912). Strecker degra- arises then as to the subsequent over-all effect on the future direc- dation is part of oxidative decarboxylation reactions of amino tion of the Maillard reaction induced by 2-amino ketones (3) acids that can be effected by variety of reagents and reaction con- generated by SD and their amino acid counterparts (4) generated ditions. It is particularly referred to as Strecker degradation, through Amadori rearrangement reaction. when -dicarbonyl compounds (1 in Fig. 1) act as oxidizing agents to effect decarboxylation of amino acids which is usually 1. Different pathways of formation of Strecker aldehyde followed by hydrolysis of the resulting imine to produce free Although the mechanism of Strecker degradation has been ammonia (if inorganic oxidizing agents are used) or a primary established for almost half a century ago (McCaldin, 1960), how- amine such as -keto amine and an aldehyde-referred to as ever, Strecker aldehyde and -amino carbonyl compounds can Strecker aldehyde. Although the amino acid itself undergoes oxi- also be formed independently of each other and by pathways dative decarboxylation, the -dicarbonyl compound however, other than that of Strecker degradation. Four different such pathways have been proposed for the formation of Strecker aldehyde E-mail: [email protected] (A, B, D & E in Fig. 2). Abbreviations: SD, Strecker degradation; AR, Amadori rearrangement; ARP, Amadori rearrangement product; SPME/GC-MS; Solid Phase Micro-extrac- From amino acids through oxidative decarboxylation and tion/Gas Chromatography-Mass spectrometry thermal reactions Mild oxidizing agents such as sodium hy- 2 V. A . YAYLAYAN

Fig. 2. Different pathways of formation of Strecker aldehyde. ARP Amadori rearrangement product; [O]oxidation.

Fig. 1. Comparison of Strecker degaradation (SD) and Amadori rearrangement (AR). of -dicrabonyl adduct of pathway C. ARP is therefore one oxidation level below the intermediates formed during Strecker degradation. pochlorite, N-bromosuccinimide, silver (II) picolinate, lead tetra From Amadori rearrangement products Formation of acetate, etc. (Barrett, 1985) can cause oxidative decarboxylation Strecker aldehyde directly from Amadori rearrangement product of amino acids at ambient temperatures followed by hydrolysis was ﬁrst suggested by Cremer et al. (2000) followed by Hof- of the resulting imine to give Strecker aldehyde (pathway A). mann and Schieberle (2000) in the same year. Both groups how- Amino acids alone (Yaylayan & Keyhani, 2001a) or in the pres- ever proposed different pathways (A and B) as shown in Fig. 3. ence of -hydroxy carbonyl compounds (Shu, 1998) can also Cremer et al. (2000) observed the formation of Strecker alde- undergo thermal deamination and produce Strecker aldehydes at hydes in a dry model system (aw 0.75) consisting of synthetic temperatures above 200˚C in the absence of oxidizing agents. ARP in the absence and presence of a different amino acid than Yaylayan and Keyhani (2001a) detected the formation of the imi- that of ARP. The model systems were incubated for 4 days at ne 5 formed between the Strecker aldehyde and the resulting 20˚C in a headspace vial and then analyzed by SPME/GC-MS amine from decarboxylation of the amino acid when pyrolyzed by heating for 60 min at 90˚C. The data indicated the formation alone at 250˚C for 20 s. On the other hand, Shu (1998) detected of Strecker aldehydes from both amino acids and more interest- the formation of tetramethylpyrazine and the Strecker aldehyde ingly, the molar ratio of these two aldehydes remained constant of amino acids heated at 200˚C for 7 min in the presence of 3- irrespective of the amount of added amino acid, after 1 : 1 ratio. hydroxy-2-butanone and proposed a decarbonylation mechanism To eliminate the possibility of Strecker aldehyde formation followed by deamination (pathway B in Fig. 2) to justify the for- through generation of a -dicarbonyl intermediates from degra- mation of Strecker aldehyde. dation product of the ARP (pathway C in Fig. 3), the authors per- From amino acids through -dicrabonyl assisted oxidative formed a control experiment using o-phenylenediamine as a decarboxylation Alloxan is the original “-dicarbonyl com- trapping agent for -dicrabonyl compounds, this model system pound” used by Strecker (1862) to effect decarboxylation/deami- also produced the Strecker aldehyde. To explain the results nation of amino acids and formation of the aldehyde named after obtained the authors proposed the reaction of free amino acid him. Similarly, the common ninhydrin reaction (McCaldin, (either added or released from ARP) with ARP to form the imine 1960) is also based on the ability of -dicarbonyl compounds to 8 (pathway A in Fig. 3) in equilibrium with the isomeric imine 9. deaminate amino acids. The mechanism of Strecker aldehyde Elimination of the free amino acid initiated by decarboxylation formation through -dicarbonyl-assisted oxidation is shown in can produce the corresponding Strecker aldehyde in addition to Fig. 2 (pathway C). Figure 2 also indicates that Amadori rear- structure 10 (Schiff base of 1-deoxyfructose with ammonia). On rangement products (ARP) in principle, could be oxidized into the other hand, Hofmann and Schieberle (2000) performed their -imino carbonyl compound (6 common intermediate with that experiments in aqueous solutions (buffered at pH 7.0, heated at of -dicrabonyl adduct in pathway C) and undergo Strecker deg- 100˚C for 2 h) of phenylalanine ARP under Argon and air atmo- radation as shown in Fig. 2 (pathway D). Alternatively, it can spheres. They also identiﬁed Strecker aldehyde in the mixtures, undergo oxidative decarboxylation similar to free amino acid however in much higher amounts under air and in the presence

(Pathway E) and produce the common intermediate 7 with that of CuSO4 than under Argon atmosphere. This prompted them to Recent Advances in the Chemistry of Strecker Degradation 3 propose a mechanism based on oxidation (catalyzed by metal) of 2. Formation of Amadori rearrangement product through the eneaminol moiety of the ARP into imino carbonyl (11) simi- transamination and strecker aldehyde through decarboxyla- lar to oxidation of enediols into -dicrabonyls (see Fig. 3 path- tion of -keto carboxylic acids way B). This intermediate can either get hydrolyzed into glu- In the absence of a catalyst, -keto carboxylic acids (19 in Fig. cosone or undergo a very similar decarboxylation reaction (after 5) acting as -dicrabonyl compounds, can also bring about oxi- ring closure) to that of intermediate 8 (Pathway A in Fig. 3), but dative decarboxylation of amino acids, being themselves convert- eliminating water instead of an free amino acid. Intermediate 12 ed into the corresponding amino acids. For example the odor of can undergo hydrolysis to produce the Strecker aldehyde. Both benzaldehyde could be detected above a boiling solution of phe- proposed mechanisms A & B are feasible but suffer from lack of nylglycine and pyruvic acid along with the formation of alanine direct evidence such as detection or isolation of the proposed side (Jones, 1979). The isomerization of the initial imine (20) into 21 products (intermediates 10 and 13 in Fig. 3). In a related study, followed by hydrolysis and decarboxylation can form the Streck- Hofmann et al. (2000) have demonstrated that during Strecker er aldehyde as shown in Fig. 5. However, formation of -keto degradation, the intermediate formed after decarboxylation and carboxylic acids during Maillard reaction has not been studied in hydration steps (structure 16 in Fig. 4) can also undergo air oxi- detail and therefore it is difﬁcult to ascertain their importance in dation and form intermediate 17 after an isomerization step. Fur- the formation of Strecker aldehydes during the Maillard reaction. ther hydrolysis of 18 can generate Strecker acid. Alternatively, Theoretically, -keto carboxylic acids could be formed from the structure 16 can lose a water molecule and generate Strecker isomerization of the Schiff base intermediate (23 in Fig. 5) and aldehyde as shown in Fig. 4. In addition, if the intermediate 14 is formation of imine 22. The latter can undergo hydrolysis and capable of ring closure (such as structure 11 in Fig. 3) then fur- generate an -keto carboxylic acid as shown in Fig. 5. Using 13C- ther oxidation and hence Strecker acid formation is prevented. labeled alanines, Yaylayan and Wnorowski (2002) provided evi- However, the generality of this pathway with other amino acids dence for the formation of pyruvic acid in alanine/glycolalde- apart from L-phenylalanine has to be demonstrated. hyde model system. On the other hand, the interaction of -keto

Fig. 3. Proposed mechanisms of Strecker aldehyde formation from Amadori rearrangement product. 4 V. A . YAYLAYAN carboxylic acids with -amino alcohols should generate imine 22 which after base catalyzed isomerization can form the Schiff base (23), the immediate precursor of Amadori product. Using 13C-labeled pyruvic acid, Yaylayan and Wnorowski (2002) provided evidence for the formation of alanine and glycolaldehyde in pyruvic acid/ethanolamine model system. Tressl et al. (1994) using labeled sugars confirmed the conversion of cystein into pyruvic acid through transamination reaction in a model glucose/ cysteine system. Similarly, Yaylayan et al. (2000) confirmed the formation of alanine during thermal decomposition of L-serine, through transamination reaction between pyruvic acid and ethanolamine generated in situ, using 13C-labeled L-serines. If the generality of this remarkable interconversion between reducing sugar/amino acid and their corresponding -keto carboxylic acid/-amino alcohol (Fig. 5) can be verified with different model systems, then mixtures consisting of -keto carboxylic acids and -amino alcohols can be used to mimic the Maillard reaction systems, since they can potentially generate an Amadori product, a reducing sugar and an amino acid as shown in Fig. 5. The Schiff base (23) is the common intermediate between the two pathways leading to Amadori rearrangement product.

3. Implications of Strecker degradation versus Amadori rearrangement in the generation of aroma and color During Strecker degradation (SD), -dicarbonyl compounds Fig. 5. Proposed mechanism of formation of Strecker aldehyde and Ama- dori rearrangement product through -keto carboxylic acids. (1), similar to the -hydroxy carbonyl compounds (2) during Amadori rearrangement (AR), undergo reductive amination and are converted into 2-amino ketones (3), instead of their amino acid counterparts (4) as in the case of Amadori rearrangement process (see Figs. 1 and 6). The significance of this transforma- secondary amino group in structure 4 prevents such amino-car- tion lies in the fact that the reactive primary amino group in bonyl type reactions to proceed to the extent of formation of sta- structure 3 allows dimerization and other reactions with different ble aromatic moieties, for example in the case of dimerization, it aldehydes or ketones to form neutral and stable N-containing het- leads to the formation of N,N¢-dialkyl-dihydropyrazines that are erocyclic aroma compounds such as pyrazines, pyrroles and unable to aromatize and eventually form, through a single elec- oxazoles (Kort, 1970; Yaylayan & Keyhani, 2001b), whereas the tron transfer process very unstable pyrazinium radical cations (Hofmann et al., 1999). These cations being unstable, they further undergo disproportionation (Hofmann et al., 1999) to regenerate dihydropyrazine and form doubly charged pyrazinium diquaternary salts (24 in Fig. 6) considered to be the precursors of colored melanoidins. According to Hofmann (1999), such intermediates, formed specifically from glycolaldehyde play an important role in the early melanoidization of Maillard mixtures compared with the longer chain analogs of -hydroxycarbonyl intermediates. Apart from this free radical based browning pathway, Hofmann (1998b) also identified an ionic pathway that leads to the formation of low molecular weight non-melanoidin type colored compounds at the later stages of Maillard reaction. This pathway is mainly initiated by the further interactions of furan moieties such as furan aldehydes and acetylformoin (2,4- dihydroxy-2,5-dimethyl-3(H)-furanone), the latter compound is described as a chemical switch activated in the presence of excess of either primary or secondary amino acids to direct the formation of amino acid specific chromophores (Hofmann, 1998c). Can aromagenic and chromogenic pathways be traced back to -aminocarbonyls and their amino acid counterparts as their principle initiators? According to the pathways described in Fig. 4. Proposed mechanism of Strecker acid formation (based on Hof- Fig. 6, it can be proposed that Strecker degradation reactions and mann et al., 2000). [O]oxidation. Amadori rearrangement of ammonia may direct the Maillard Recent Advances in the Chemistry of Strecker Degradation 5 reaction mainly towards aromagenic pathways through interme- furanoid species that lead to browning (Hofmann, 1998b). As to diate 3 and Amadori rearrangement of amino acids with - the variation in the differences of browning abilities (8 fold ver- hydroxycarbonyl compounds may lead the Maillard reaction sus 256) among the precursors, this could be related to their rela- mainly towards chromogenic pathways and melanoidization tive ability to be reduced or to be oxidized in the reaction system. through intermediate 4 and to chromogenic pathways through In the above experiment, the large difference (8 versus 256) in intermediate 1. Although there are no studies yet on the relative browning ability between glyoxal/glycoladehyde redox couple ability of structures such as 1, 3 and 4 in the generation of aroma and that of 2-oxopropanal/hydroxy-2-propanone could be related and browning, however, Hofmann (1999) investigated the role of to the much higher tendency of glyoxal to be reduced to glycola- the different -dicarbonyl and -hydroxycarbonyl intermediates ldehyde then the tendancy of hydroxy-2-propanone to be oxi- (such as 1 and 2 in Fig. 6) as browning precursors, in addition to Amadori products (4) of intact carbon chain. In this landmark study, Hoffman calculated relative browning activity of various carbohydrate precursors such as glyoxal, glycolaldehyde, pyruvaldehyde, hydroxyacetone, etc. in the presence of alanine, when refluxed in phosphate buffer (pH 7.0) for 15 min. Glycolalde- hyde (a C2 -hydroxycarbonyl), showed the highest browning activity among all the intermediates studied and compared with its dicarbonyl counterpart glyoxal, it showed 8 fold higher browning activity. On the other hand, 2-oxopropanal (a C3 - dicarbonyl) exhibited 256 fold higher browning activity than its -hydroxycarbonyl counterpart hydroxy-2-propanone. Further- more, 2,3-butanedione (a C4 -dicarbonyl) exhibited 8 fold higher browning activity than its -hydroxycarbonyl counterpart 2-hydroxy-3-butanone. This results are consistent with the proposed concept in Fig. 6. Furthermore, the results indicate that in addition to the nature of the chemical moieties involved as key precursors (1, 2, 3, and 4) their carbon chain lengths (C2, C3 or

C4) can also play a decisive role in promoting aldol condensations that lead to formation of furanoid species and hence induce browning through ionic pathway (see Fig. 6). In the case of glycolaldehyde/glyoxal couple, it is statistically less likely to generate a furan moiety through repeated aldol condensations, hence Fig. 7. Interconversion of key precursors of aroma and color during there is only one pathway of browning through ARP (free radi- Maillard reaction. [O]oxidation, [H]reduction. Double headed arrows cal). With C3 and C4 precursors, aldol condensation can genarate () indicate reversibility.

Fig. 6. Relationship of Strecker degradation (SR) and Amadori rearrangement (AR) to aromagenic and chromogenic pathways of Maillard reaction. [O]oxidation, [H]reduction. 6 V. A . YAYLAYAN dized to 2-oxopropanal. As a result both -dicrabonyl and - Biochemistry of amino acids.” ed. by G.C. Barrett, pp 355–375. hydroxycarbonyl species, due to their ease of interconversion Cremer, D.R., Vollenbroeker and Eichner, K. (2000). Investigation of the formation of Strecker aldehydes from the reaction of Amadori through redox reactions, can produce relatively similar browning rearrangement products with -amino acids in low moisture model (8 fold difference), for example through free radical pathway, as systems. Eur. Food Res. Technol., 211, 400–403. in the case of glyoxal/glycolaldehyde couple. On the other hand, Hofmann, T. (1998a). Studies on the relationship between molecular due to difficulty of redox interconversion between 2-oxopropanal weight and the color potency of fractions obtained by thermal treat- and hydroxy-2-propanone, a larger difference (256 fold) in ment of glucose/amino acid and glucose/protein solutions by using ultracentrifugation and color dilution technique. J. Agric. Food browning ability was detected, indicating the importance of ionic Chem., 46, 3891–3895. pathway when the carbon chain of the precursors exceeds C2. Hofmann, T. (1998b). Identifiaction of novel colored compounds con- According to Hofmann (1998a) in a typical Maillard reaction, taining pyrrole and pyrrolinone structures formed by Maillard reac- most of the browning is due to low molecular weight (1000 tions of pentoses and primary amino acids. J. Agric. Food Chem., 46, 3902–3911. amu) fraction of the product which constitiutes 78.4% by weight Hofmann, T. (1998c). Acetylformoin -A chemical switch in the forma- of the mixture in the case of glycine/glucose system. tion of colored Maillard reaction products from hexoses and pri- When the browning activity of the precursors produced in situ mary and secondary amino acids. J. Agric. Food Chem., 46, 3918– in a refluxing glucose/alanine and xylose/alanine mixtures at var- 3928. Hofmann, T. (1999). Quantitative studies on the role of browning pre- ious time intervals were also analyzed (through repeated derivati- cursors in the Maillard reaction of pentoses and hexoses with L-ala- zation at the end of each time interval) it was shown that relative nine. Eur. Food Res. Technol., 209, 113–121. browning activities of the intermediates changed with the time of Hofmann, T., Bors, W. and Stettmaier, K. (1999). Studies on radical the reaction, some decreased and some increased over a period of intermediates in the early stage of the nonenzymatic browning reaction of carbohydrates and amino acids. J. Agric. Food Chem., 47, 120 min (Hofmann, 1999). These observations indicate that 379–390. some chromophores are produced early (through pyrazinium Hofmann, T., Münch, P. and Schieberle, P. (2000). Quantitative model radical cation, formed mainly through glycolaldehyde) in the studies on the formation of aroma-active aldehydes and acids by course of the reaction and others later (further reactions of fura- Strecker-type reactions. J. Agric. Food Chem., 48, 434–440. Hofmann, T. and Schieberle, P. (2000). Formation of aroma-active noid species). In both model systems glyoxal contributed to the Strecker-aldehydes by a direct oxidative degradation of Amadori overall browning at the early phase of the reaction and pyruval- compounds. J. Agric. Food Chem., 48, 4301–4305. dehyde and 3-deoxyosons contributed towards the end of the Jones, J.H. (1979). Amino acids. In “Comprehensive Organic Chemis- reaction. The two types of chromophores (early and late) could try.” ed. by D.H.R. Barton and W.D. Ollis, Vol. 5, p 825. be therefore distinguished structurally. Kort, M.J. (1970). Reactions of free sugars with aqueous ammonia. Adv. Carbohydr. Chem., 25, 311–349. Kuhn, R. and Weygand, F. (1937). The Amadori rearrangement. Ber. Conclusion Dtsch. Chem. Ges., 70B, 769–772. In the context of Fig. 6, the contribution of Strecker degrada- Maillard, L.C. (1912). Acrion des acides amines sur les sucres; formation to the concentration of 2-amino carbonyl species (3) and its tion des melanoidines par voie methodique. Compt. Rend. Acad. Sci., 154, 66–68. role in depleting -dicarbonyl species (1), not to mention, gener- McCaldin, D.J. (1960). The chemistry of ninhydrin. Chem. Rev., 60, ation of Strecker aldehydes, verifies its known importance in aro- 39–51. magenesis during the Maillard reaction. On the other hand, Shu, C-K. (1998). Pyrazine formation from amino acids and reducing Amadori rearrangement plays a dual role during the reaction, it sugars—a pathway other than Strecker degradation. J. Agric. Food Chem., 46, 1129–1131. can contribute to aromagenesis through production of -dicarbo- Strecker, A. (1862). On a peculiar oxidation by alloxan. Justus Liebigs nyls both oxidatively and non-oxidatively (see Fig. 7) and Ann Chem., 123, 363–367. through formation of 2-amino carbonyls (3) in the presence of Tressl, R., Kersten, E., Nittka, C. and Rewicki, D. (1994). Formation ammonia. Simultaneously, it can contribute to the browning and of sulfur-containing flavor compounds from [13C]-labeled sugars, cysteine, and methionine. In “Sulfur compounds in Foods.” ed. by melanoidization of the reaction mixture through ionic and free C.J. Mussinan and M.E. Keelan, ACS symposium series 564, pp radical pathways. The balance among different key precursor 224–235. moities (1, 2, 3, and 4 in Fig. 7) which controls relative aromati- Wrodnigg, T.M. and Eder, B. (2001). The Amadori and Heyns rear- zation versus melanoidization can be easily disrupted through rangements: Landmarks in the history of carbohydrate chemistry or initiation of redox reactions (Fig. 7) that are affected by the unrecognized synthetic opportunities? Topics Curr. Chem., 215, 115–152. amount of dissolved oxygen, and by the amount and timing of Yaylayan, V.A. and Wnorowski, A. (2002). The role of -hydroxy- the release of reducing species produced by the reaction (reduc- amino acids in the Maillard reaction: Transamination route to Ama- tones), disproportionation and dehydration reactions and concen- dori products. In “Maillard Reaction in Food Chemistry and tration of metal ions. The role of Amadori rearrangement and Medical Sciences: Update for the Postgenomic Era.” ed. by S. Hori- uchi, N. Taniguchi, F. Hayase, T. Kurata, and T. Osawa. Interna- Strecker degradation is crucial in controlling this balance. tional Congress Series 1245. Elsevier Science. Amsterdam, The Knowledge of concentration changes of these species (1, 2, 3, Netherlands, pp. 195–200. and 4 in Fig. 7) over time, during the reaction is therefore critical Yaylayan, V.A. and Keyhani, A. (2001a). Carbohydrate and amino acid 13 in fully understanding and controlling melanoidization, aroma- degradation pathways in L-methionine/D-[ -C]glucose model systems. J. Agric. Food Chem., 49, 800–803. genesis and different browning pathways. Yaylayan, V.A. and Keyhani, A. (2001b). Elucidation of the mechanism of pyrrole formation during thermal degradation of 13C-la- References belled L-serines. Food Chem., 74, 1–9. Amadori, M. (1925). Products of condensation between glucose and p- Yaylayan, V.A., Keyhani, A. and Wnorowski, A. (2000). Formation of 13 phenetidine I. Atti Accad. Lincei, 2, 337–342. sugar-specific reactive intermediates from C-labeled L-serine. J. Barrett, G.C. (1985). Reactions of amino acids. In “Chemistry and Agric. Food Chem., 48, 636–641. Special Article May 2005: 158–165

Formation and Human Risk of Carcinogenic Heterocyclic Amines Formed from Natural Precursors in Meat Mark G. Knize and James S. Felton, PhD

A group of heterocyclic amines that are mutagens and during cooking and that the formation process was tem- rodent carcinogens form when meat is cooked to perature dependent. A large range of foods were ana- medium and well-done states. The precursors of these lyzed, and it was determined that cooked muscle meats compounds are natural meat components: creatinine, were the major sources of extractable mutagenic activity amino acids, and sugars. Defined model systems of in bacterial tests.6 dry-heated precursors mimic the amounts and propor- The precursors responsible for the mutagens were tions of heterocyclic amines found in meat. Results identified when the chemical structures of the first com- from model systems and cooking experiments suggest pounds from cooked fish7,8 and beef 9,10 were deter- ways to reduce their formation and, thus, reduce mined. These meat-derived mutagens were heterocyclic human intake. Human cancer epidemiology studies amines having an amino-imidazo structure, suggesting related to the consumption of well-done meat products that creatine or creatinine was involved in the reactions. are listed and compared in this review. Early work in adding creatine to meat before cook- Key words: heterocyclic amine, PhIP, IFP, cooked meat, ing showed that it increased mutagenic activity.11 Exper- epidemiology iments relating creatine levels in fish, which varied over © 2005 International Life Sciences Institute a range of 2.5-fold, showed mutagenic activity after doi: 10.1301/nr.2005.may.158–165 cooking to be only approximately correlated.12 A later study of cooked meat from 17 animal species also showed that creatine or creatinine levels do not explain INTRODUCTION differences in mutagenic activity. These results suggested that other components were also responsible for Diet has been one factor associated for many years with the mutagen levels in cooked meats. Other work showed 1,2 differing cancer rates worldwide. A biologically plau- free amino acids to be involved in the formation of sible factor in this association was the discovery in the mutagenic activity,13 but not amino acids from pro- 1970s of mutagenic activity, as detected by bacterial test teins.11 3,4 systems, in meats cooked for human food. Finding Analysis of the specific mutagenic compounds bacterial mutagens in meats paralleled the well-known formed during cooking shows that amino acids are im- presence of mutagens in the smoke from cigarettes at that portant, and changes in them can affect the amount and 5 time. types of mutagens found in the cooked meat. Knowledge The original discovery of mutagenic substances in of the formation conditions suggests ways to cook meat cooked meats was followed by demonstrations in many that greatly inhibit the formation and, thus, the human laboratories worldwide that the mutagens were formed intake of carcinogenic heterocyclic amines.

Mr. Knize and Dr. Felton are with the Biosciences FORMATION IN MEATS Directorate of the University of California, Lawrence Livermore National Laboratory, Livermore, California. Figure 1 shows the structures of creatine and seven of the This work was performed under the auspices of amino-imidazo heterocyclic amine mutagens pyrosyn- the US Department of Energy by the University of thesized from creatine and other small molecules, such as California, Lawrence Livermore National Laboratory amino acids and glucose. It is easy to see that the under contract no. W-7405-Eng-48 and was sup- N-methyl-amino-imidazo moiety could form intact from ported by NCI grant no. CA55861. Corresponding author: Dr. James S. Felton, Law- creatine, but the source of other rings are derived from rence Livermore National Laboratory 7000 East Ave- other small molecules, and their sources are not apparent nue, Livermore, CA 94550; Phone: 926-422-5656; from the reactants. These chemicals were isolated by Fax: 925-422-2282; E-mail: [email protected]. following their mutagenic activity in Salmonella-based

158 Nutrition Reviewsா, Vol. 63, No. 5 amounts of heterocyclic amine compounds identified per gram of cooked meat is shown for each of the four pan temperatures, indicating a direct correspondence between increased cooking temperature and heterocyclic amine content.18 Heat flow simulations to understand heterocyclic amine formation during the pan-frying of beef patties were done by Tran et al.19 These simulations accurately modeled the experimental temperature increases, meat cooking times, heterocyclic amine spatial distribution, and total amount of heterocyclic amines produced. Studies of the amounts of heterocyclic amines produced in foods as a result of regional cooking practices have been reported for Great Britain,20 Sweden,21,22 Switzerland,23 Spain,24 Japan,25 and the United States.26,27 In most cases, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) tend to be the Figure 1. Structures of heterocyclic amine mutagens/carcino- most mass-abundant heterocyclic amines. Their concen- gens and creatine. Heterocyclic amines are pyrosynthesized from creatine, amino acids, and sugars. trations in cooked meats typically range from nearly undetectable levels (typically 0.1 ng/g) to tens of nanograms per gram for MeIQx, and up to a few hundreds of mutation tests during extraction and chromatographic nanograms per gram for PhIP, depending on the cooking purification.7-9,14-16 A breakthrough in the analysis of method. heterocyclic amines in meats and model systems was made with the development of solid-phase extraction methods, enabling the extraction of the compounds, PAN RESIDUES AND FOOD FLAVORS followed by analysis for specific heterocyclic amines by high-performance liquid chromatography (HPLC), at a Another source of heterocyclic amines are pan residues reasonable cost.17 from meat and process flavors. Pan residues are some- The temperature dependence of the formation of times consumed after being made into gravy, and can be these compounds in beef patties cooked to 70°C that are a source of heterocyclic amines equivalent to or greater near the US Department of Agriculture Food Safety and than that of the meat itself.21,22,28 Process flavors are Inspection Services-recommended internal temperature commercially produced flavors derived from heated mix- of 71.1°C is shown in Figure 2. The sum of the mass tures of proteins, fats, and carbohydrates. These are

Figure 2. Formation of heterocyclic amines in beef patties after cooking to an internal temperature of 70°C at different frying pan temperatures. Error bars are the standard error of four or ﬁve replicate cooking experiments.

Nutrition Reviewsா, Vol. 63, No. 5 159 added to foods in amounts up to a few percent by weight Table 1. Concentration (mg/g meat wet weight) of to improve the food’s taste and color, and they can also Free Amino Acids, Creatine, and Glucose in Three be used as a base for soups. Because of their chemical Kinds of Meats (from Pais et al.28) complexity, specific sample preparation methods for het- Beef Chicken Breast Codfish erocyclic amine analysis have been developed for pro- L-Alanine 0.14 0.21 0.12 29-31 cess flavors. Although most have undetectable levels L-Arginine 1.07 1.19 0.03 of heterocyclic amines, some samples contain as much as L-Aspartic acid 0.02 0.13 0.01 20 ng of heterocyclic amines per gram of solid or liquid L-Glutamic acid 0.09 0.23 0.02 32-34 flavor. However, since process flavors are consumed L-Glycine 0.06 0.08 0.05 as only a tiny percentage of the human diet, the bulk of L-Histidine 0.14 0.18 0.03 heterocyclic amine exposure is from well-cooked meats. L-Isoleucine 0.05 0.08 0.02 L-Leucine 0.07 0.13 0.02 L-Lysine 0.07 0.14 0.18 MODEL SYSTEMS L-Methionine 0.06 0.08 0.04 L-Phenylalanine 0.05 0.08 0.01 Model systems to understand the formation of the muta- L-Proline 0.10 0.10 0.14 genic/carcinogenic heterocyclic amines were developed L-Serine 0.05 0.12 0.02 to help identify the foods and cooking conditions favor- L-Threonine 0.28 1.63 0.69 ing their formation in an effort to develop strategies to L-Tyrosine 0.06 0.10 0.03 reduce human intake. Defined model systems composed L-Valine 0.06 0.10 0.04 of creatine or creatinine, amino acids, and sugars have Creatine 6.33 3.54 7.06 been a good model for the trace-level formation of these Glucose 7.03 0.47 0.21 heterocyclic amines. Ja¨gerstad et al.11 developed a system for heating components in diethylene glycol, and this work was followed by many studies investigating heterocyclic amine precursors35-37 and kinetics38 in a the chicken model system, and PhIP is also known to be sealed-tube aqueous model. Reaction intermediates were formed from tyrosine and isoleucine, which are also identified that led to the formation of PhIP.39 It was highest in chicken.37 IFP is known to form from glutamic shown that 37°C is warm enough to produce PhIP from acid,44 and the IFP formation follows the glutamic acid a mixture of phenylalanine with creatinine and glucose content of the three mixtures shown. Results in Table 1 or MeIQx from glycine with creatinine and glucose in fit with the general findings of heterocyclic amines in aqueous buffers.40,41 No PhIP was found in a similar meats: that IQ and MeIQ are seldom detected in beef or model system kept at room temperature for two weeks.39 chicken; MeIQx is about equal in beef and chicken; large Another meat model system that produced heterocyclic amounts of PhIP can be formed in chicken that is amines was composed of boiled pork juice.42 overcooked, but PhIP levels similar to those seen in beef Simple dry-heating of heterocyclic amine precursors are measured in chicken cooked in most households also forms similar relative amounts and types of hetero- sampled.45 cyclic amines as are seen in cooked meats. Table 1 shows amino acid, creatine, and glucose content of beef, MODIFYING COOKING PRACTICES TO chicken breast, and codfish. When these components are REDUCE THE FORMATION OF combined and heated for 30 minutes at 225°C, a family HETEROCYCLIC AMINES of heterocyclic amines is formed. These vary with the mixture composition, as shown in Figure 3. Two of the As shown in Figure 2, the formation of heterocyclic compounds, 2-amino-3-methylimidaazo[4,5-f]quinoline amines is related to pan temperature when meat is (IQ) and 2-amino-3,4-dimethylimidaazo[4,5-f]quinoline cooked to the same final internal temperature. Surpris- (MeIQ), are most abundant in the codfish model, but the ingly, the time needed to reach the 70°C internal tem- codfish model produces the lowest amounts of MeIQx perature is about the same at 250°C (7 min) as at 160°C and 2-amino-(1,6-dimethylfuro[2,3-e])imidazole[4,5-b] (9 min).18 This is due to the limit of the slow heat transfer pyridine (IFP). The chicken model produced the largest through the meat, suggesting that simply using lower pan amount of PhIP, as is shown in studies of the cooked temperatures is a practical way to reduce heterocyclic chicken breast meat.43 amine formation without greatly increasing cooking The model systems in Table 1 show that arginine, time. glutamic acid, leucine, and phenylalanine are greatly Flipping pan-fried beef patties over every minute, as reduced in codfish compared with beef or chicken breast. opposed to turning the meat over once at 5 minutes, and Phenylalanine, a known precursor for PhIP, is highest in cooking at moderate pan temperatures until the target

160 Nutrition Reviewsா, Vol. 63, No. 5 component marinade to chicken breast meat before grilling can greatly decrease PhIP, although MeIQx is increased at the longest cooking time, probably due to sucrose in the marinade.50 Little change in heterocyclic amines was seen after marinating chicken in another study,51 possibly due to differences in marinating or cooking conditions. Conversely, a heterocyclic amine reducing effect was seen when sugar was mixed with ground meat formed into patties before frying.47 Exper- iments with eggs, bean cake, and pork show that boiling in sugar and soy sauce increases most of the heterocyclic amines.52 Microwave pretreatment was shown to reduce the amount of heterocyclic amines formed during the frying of ground beef.53 Beef patties received microwave pretreatment for various times before frying. Microwave pretreating for 2 minutes, then pouring off the resulting liquid and frying at either 200°C or 250°C for 6 minutes per side reduced heterocyclic amines. The liquid released by the microwave pretreatment contained creatine, cre- Figure 3. Heterocyclic amines formed in mixtures representing atinine, amino acids, glucose, water, and fat, and discard- the mass ratios of amino acid, creatine, and glucose found in ing these precursors resulted in lower amounts of het- beef, chicken breast, and codfish dry-heated at 225°C for 30 erocyclic amines. The sum of the heterocyclic amines min. present decreased 3-fold following microwaving and frying at 200°C or 9-fold following microwaving and frying at 250°C compared with controls (non-micro- internal temperature of 70°C is reached seems to be the wave-pretreated beef patties fried under identical condi- most effective way to reduce heterocyclic amine content tions). while also avoiding undercooking (defined as cooking to a final temperature below 70°C, the internal temperature needed to eliminate harmful bacteria).18 HUMAN RISK Minor changes in recipes for preparing different meat dishes may provide a way of reducing the amount Table 2 summarizes human studies that have investi- of heterocyclic amines formed. The addition of reaction gated the relationship between meat doneness and can- inhibitors or inert substances can change the concentra- cers at various sites. In rodents, the heterocyclic amines tion of precursors and show an inhibiting effect. Schemes are multisite carcinogens. The number of studies and for reducing mutagenic activity or the specific heterocy- number of human cancer sites with positive correlations clic amine by adding substances to ground meat have with meat doneness strongly suggest that these com- been reported. Food additives such as soy flour and pounds may be multisite carcinogens in humans as well. antioxidants46 or glucose or lactose47 were shown to Supporting these epidemiology studies is a study show- lower mutagenic activity. ing that women have an increased cancer risk with The heat and mass transport in meat during frying is increasing levels of PhIP-DNA adducts, and that the DNA adducts increase with a subject’s preference for very complex. Water is important for the transport of 85 water-soluble precursors for the formation of heterocy- well-done meat. clic amines within the food. The transport of precursors from the inner parts of the food to the surface can be CONCLUSION restricted by the addition of water-binding compounds such as salt, soy protein, or starch to minced meat, thus There is a general consensus that human exposure to reducing the formation of heterocyclic amines. Persson potent genotoxic heterocyclic amine carcinogens pro- et al.48 showed a significant effect with the addition of duced in meat during cooking is widespread. Under- sodium chloride/sodium tripolyphosphate. Enzyme treat- standing the parameters affecting their formation can ment with creatinase was used to reduce the available help us find ways to lessen exposure. The demonstrated creatine in meat.49 mutagenicity of these compounds in bacteria,3 cultured Heterocyclic amine formation can also be affected cells,86,87 and mice88 support the many studies of carci- by meat surface treatment. The application of a seven- nogenicity in mice89 and rats.90,91 Mechanistic data show

Nutrition Reviewsா, Vol. 63, No. 5 161 Table 2. Human Studies Investigating Cancer and Well-Done Meat Study Result* Cancer Site N (Age if given) Han et al., 200454 OR ϭ 2.38 Breast 635 Dai et al., 200255 OR ϭ 1.92 Breast 3015 (25–64) Zheng et al., 200256 OR ϭ 3.4 Breast 683 (postmenopause) Balbi et al., 200157 OR ϭ 2.66 (barbecued) Bladder 720 (40–89) OR ϭ NA (fried) Zheng et al., 200158 OR ϭ 2.0 Breast 488 (55–69) Delﬁno et al., 200059 NA Breast 394 (Ͼ39) Sinha et al., 200060 OR ϭ 1.9 Breast 930 (56–67) Zheng et al., 199861 OR ϭ 4.6 Breast 930 (55–69) Butler et al., 200362 OR ϭ 2.0 Colon 1658 (40–80) Kampman et al., 199963 OR ϭ 1.4 (men only) Colon 3402 (30–79) Sinha et al., 199964 OR ϭ 1.85/10 g meat Colon 374 Augustsson et al., 199965 NA Colon, rectum, bladder, kidney 1565 (56–80) Schiffman and Felton, 199066 OR ϭ 3.5 Colon 146 Barrett et al., 200367 OR ϭ 1.97 Colon/rectum 2164 (45–80) Tiemersma et al., 200468 NA Colon/rectum 864 Le Marchand et al., 200269 OR ϭ 8.8 Colon/rectum 1454 Nowell et al., 200270 OR ϭ 4.36 Colon/rectum 460 (20–88) Sinha et al., 200171 OR ϭ 1.29 Colon/rectum 374 Probst-Hensch et al., 199772 OR ϭ 2.2 Colon/rectum 976 (50–74) Gerhardsson de Verdier et al., 199173 RR ϭ 2.8 Colon/rectum 1064 (42–81) Gunter et al., 200574 NA Colon/rectum 565 (50–70) Navarro et al., 200475 OR ϭ 4.57 Colon/rectum 893 (23–80) Terry et al., 200376 NA Esophagus 1004 (Ͻ80) NA Gastric cardia 1077 (Ͻ80) OR ϭ 2.4 Esophagus-squamous cell 982 (Ͻ80) Bosetti et al., 200277 OR ϭ 1.89 Larynx 1824 (31–79) Sinha et al., 199878 OR ϭ 1.8 Lung 1216 (52–79) Zhang et al., 199979† OR ϭ 2.2 Non-Hodgkin’s lymphoma 88410 (48–74) Anderson et al., 200280 OR ϭ 2.19 Pancreus 867 (20–65ϩ) Norrish et al., 199981 Positive trend Prostate 787 Nowell et al., 200482 OR ϭ 8.27 Prostate 923 Murtaugh et al., 200483 OR ϭ 1.33 Rectum 2157 Ward et al., 199784 OR ϭ 2.4 Stomach 678 (ϳ67–82) OR ϭ 2.0 Esophagus 645 (ϳ67–82)

*OR ϭ odds ratio; RR ϭ relative risk; NA ϭ no association. †Prospective study; all other studies were case-control.

that, even at low doses, heterocyclic amines form DNA REFERENCES adducts in rodents, primates, and humans.92-94 The ma- 1. Doll R, Peto R. The causes of cancer: quantitative jority of epidemiological studies generally support the estimates of avoidable risks of cancer in the United hypothesis that degree of doneness increases risk for States today. J Natl Cancer Inst. 1981;66:1191- human cancers associated with meats. 1308. The goal of understanding and reducing cancer in- 2. Holmes S. Food, nutrition and the prevention of cidence is worthwhile. Here we describe conditions for cancer: A global perspective. Health Policy. 1998; both formation and reduction of carcinogenic heterocy- 45:169-171. clic amines in our diet. We also describe 30 human 3. Sugimura T, Nagao M, Kawachi T, et al. Mutagen- cancer epidemiology studies that relate consumption of carcinogens in food, with special reference to highly mutagenic pyrolytic products in broiled foods. In: well-done meat in our diet to various tumor sites. More Hiatt HH, Watson JD, Winsten JA, eds. Origins of than 80% of these studies show a positive correlation Human Cancer. New York: Cold Spring Harbor between cancer incidence and well-done meat consump- Press; 1977:1561-1577. tion. 4. Commoner B, Vithayathil AJ, Dolara P, Nair S,

162 Nutrition Reviewsா, Vol. 63, No. 5 Madyastha P, Cuca GC. Formation of mutagens in in pan-fried meat patties. Food Chem Toxicol. 2002; beef and beef extract during cooking. Science. 40:673-684. 1978;201:913-916. 20. Murray S, Lynch AM, Knize MG, Gooderham NJ. 5. Sugimura T, Kawachi T, Nagao M, et al. Mutagenic Quantification of the carcinogens 2-amino-3,8-di- principles in tryptophan and phenylalanine pyrolysis methylimidazo[4,5-f]quinoxaline, 2-amino-3,4,8-tri- products. Proc Jpn Acad. 1977;53:58-61. methylimidazo[4,5-f]quinoxaline, and 2-amino-1- 6. Bjeldanes LF, Morris MM, Felton JS, et al. Mutagens methyl-6-phenylimidazo[4,5-b]pyridine in food from the cooking of food: II. survey by Ames/Sal- using a combined assay based on gas chromatog- monella test of mutagen formation in the major raphy negative ion mass spectrometry. J Chrom protein-rich foods of the American diet. Food Chem (Biomedical Applications). 1993;616:211-219. Toxicol. 1982;20:357-363. 21. Johansson MAE, Ja¨ gerstad M. Occurrence of mu- 7. Kasai H, Yamaizumi Z, Nishimura S, et al. A potent tagenic/carcinogenic heterocyclic amines in meat mutagen in broiled fish. Part 1. 2-amino-3-methyl- and fish products, including pan residues, prepared 3H-imidazo[4,5-f]quinoline. J Chem Soc. 1981;1: under domestic conditions. Carcinogenesis. 1994; 2290-2293. 15:1511-1518. 8. Kasai H, Yamaizumi Z, Wakabayashi K, et al. Potent 22. Skog K, Augustsson K, Steineck G, Stenberg M, novel mutagens produced by broiling fish under Ja¨ gerstad M. Polar and non-polar heterocyclic normal conditions. Proc Jpn Acad. 1980;56:278- amines in cooked fish and meat products and their 283. corresponding residues. Food Chem Toxicol. 1997; 9. Kasai H, Yamaizumi Z, Shiomi T, et al. Structure of 35:555-565. a potent mutagen isolated from fried beef. Chem 23. Zimmerli B, Rhyn P, Zoller O, Schlatter J. Occur- Lett. 1981;485-488. rence of heterocyclic aromatic amines in the Swiss 10. Felton JS, Knize MG, Shen NH, et al. The isolation diet: analytical method, exposure estimation and and identification of a new mutagen from fried risk assessment. Food Addit Contam. 2001;18:533- ground beef: 2-amino-1-methyl-6-phenylimidazo- 551. [4,5-b]pyridine (PhIP). Carcinogenesis. 1986;7: 24. Busquets R, Bordas M, Toribio F, Puignou L, Gal- 1081-1086. ceran MT. Occurrence of heterocyclic amines in 11. Ja¨ gerstad M, Laser-Reuterswa¨ rd A, Oste R, et al. several home-cooked meat dishes of the Spanish Creatinine and Maillard reaction products as pre- diet. J Chromatogr B Analyt Technol Biomed Life cursors of mutagenic compounds formed in fried Sci. 2004;802:79-86. beef. In: Waller GR, Feather MS, eds. The Maillard 25. Wakabayashi K, Ushiyama H, Takahashi M, et al. Reaction in Foods and Nutrition. Washington, DC: Exposure to heterocyclic amines. Environ Health American Chemical Society; 1983:507-519. Perspect. 1993;99:129-134. 12. Marsh NL, Iwaoka WT, Mower HF. Formation of 26. Knize MG, Sinha R, Rothman N, et al. Heterocyclic mutagens during the frying of Hawiian fish: correla- amine content in fast-food meat products. Food tion with creatine and creatine content. Mutat Res. Chem Toxicol. 1995;33:545-551. 1990;242:181-186. 27. Knize MG, Sinha R, Brown ED, et al. Heterocyclic 13. O¨ vervik E, Kleman M, Berg I, Gustafsson J-Å. Influ- amine content in restaurant-cooked hamburgers, ence of creatine, amino acids and water on the steaks, and ribs. J Agric Food Chem. 1998;46:4648- formation of the mutagenic heterocyclic amines 4651. found in cooked meat. Carcinogenesis. 1989;10: 28. Pais P, Salmon CP, Knize MG, Felton JS. Formation 2293-2301. of mutagenic/carcinogenic heterocyclic amines in 14. Hayatsu H, Arimoto S, Wakabayashi K. Methods for dry-heated model systems, meats, and meat drip- separation and detection of heterocyclic amines. In: pings. J Agric Food Chem. 1999;47:1098-1108. Hayatsu H, ed. Mutagens in Food: Detection and 29. Gross GA, Gru¨ ter A, Heyland S. Optimization of the Prevention. Boca Raton, FL: CRC Press; 1991:101- sensitivity of high-performance liquid chromatogra- 112. phy in the detection of heterocyclic aromatic amine 15. Felton JS, Knize MG, Wood C, et al. Isolation and mutagens. Food Chem Toxicol. 1992;30:491-498. characterization of new mutagens from fried ground 30. Perfetti GA. Determination of heterocyclic aromatic beef. Carcinogenesis. 1984;5:95-102. amines in process flavors by a modified liquid chro- 16. Knize MG, Roper M, Shen NH, Felton JS. Proposed matographic method. J AOAC Int. 1996;79:813- structures for an amino-dimethylimidazofuropyri- 816. dine mutagen in cooked meats. Carcinogenesis. 31. Pais P, Moyano E, Puignou L, Galceran MT. Liquid 1990;11:2259-2262. chromatography-atmospheric-pressure chemical 17. Gross GA. Simple methods for quantifying muta- ionization mass spectrometry as a routine method genic heterocyclic amines in food products. Carci- for the analysis of mutagenic amines in beef ex- nogenesis. 1990;11:1597-1603. tracts. J Chromatogr A. 1997;778:207-218. 18. Salmon CP, Knize MG, Panteleakos FN, Wu RW, 32. Solyakov A, Skog K, Ja¨ gerstad M. Heterocyclic Nelson DO, Felton JS. Minimization of heterocyclic amines in process flavours, process flavour ingre- amines and thermal inactivation of Escherichia coli dients, bouillon concentrates and a pan residue. in fried ground beef. J Natl Cancer Inst. 2000;92: Food Chem Toxicol. 1999;37:1-11. 1773-1778. 33. Jackson LS, Hargraves WA, Stroup WH, Diachenko 19. Tran NL, Salmon CP, Knize MG, Colvin ME. Exper- GW. Heterocyclic amine content of selected beef imental and simulation studies of heat flow and flavors. Mutat Res. 1994;320:113-124. heterocyclic amine mutagen/carcinogen formation 34. Stavric B, Lau BP, Matula TI, Klassen R, Lewis D,

Nutrition Reviewsா, Vol. 63, No. 5 163 Downie RH. Mutagenic heterocyclic aromatic trient contents of pan fried meat from various animal amines (HAAs) in ‘processed food flavour’ samples. species. Acta Vet Scand. 1993;34:363-370. Food Chem Toxicol. 1997;35:185-197. 50. Salmon CP, Knize MG, Felton JS. Effects of mari- 35. Negishi C, Tsuda M, Wakabayashi K, Sato S, Sug- nating on heterocyclic amine carcinogen formation imura T, Ja¨ gerstad M. Formation of MeIQx and in grilled chicken. Food Chem Toxicol. 1997;35: another mutagen by heating a mixture of creatinine, 433-441. glucose and glycine. Mutat Res. 1984;130:378-379. 51. Tikkanen LM, Latva-Kala KJ, Heinio¨ R-L. Effect of 36. Skog K, Ja¨ gerstad M. Incorporation of carbon commercial marinades on the mutagenic activity, atoms from glucose into the food mutagens sensory quality and amount of heterocyclic amines MeIQx and 4,8-DiMeIQx using 14C-labelled glu- in chicken grilled under different conditions. Food cose in a model system. Carcinogenesis. 1993;14: Chem Toxicol. 1996;34:725-730. 2027-2031. 52. Lan CM, Chen BH. Effects of soy sauce and sugar 37. Johansson MA, Fay LB, Gross GA, Olsson K, Ja¨g- on the formation of heterocyclic amines in mari- erstad M. Influence of amino acids on the formation nated foods. Food Chem Toxicol. 2002;40:989- of mutagenic/carcinogenic heterocyclic amines in a 1000. model system. Carcinogenesis. 1995;16:2553- 53. Felton JS, Fultz E, Dolbeare FA, Knize MG. Effect of 2560. microwave pretreatment on heterocyclic aromatic 38. Arvidsson P, van Boekel MAJS, Skog K, Ja¨ gerstad amine mutagens/carcinogens in fried beef patties. M. Kinetics of formation of polar heterocyclic Food Chem Toxicol. 1994;32:897-903. amines in a meat model system. J Food Sci. 1997; 54. Han DF, Zhou X, Hu MB, et al. Sulfotransferase 1A1 62:911-916. (SULT1A1) polymorphism and breast cancer risk in 39. Zochling S, Murkovic M. Formation of the hetero- Chinese women. Toxicol Lett. 2004;150:167-177. cyclic aromatic amine PhIP: identification of precur- 55. Dai Q, Shu XO, Jin F, Gao YT, Ruan ZX, Zheng W. sors and intermediates. Food Chem. 2002;79:125- Consumption of animal foods, cooking methods, 134. and risk of breast cancer. Cancer Epidemiol Bio- 40. Kinae N, Kujirai K, Kajimoto C, Furugori M, Masuda markers Prev. 2002;11:801-808. S, Shimoi K. Formation of mutagenic and carcino- 56. Zheng W, Wen WQ, Gustafson DR, Gross M, Cer- genic heterocyclic amines in model systems without han JR, Folsom AR. GSTM1 and GSTT1 polymor- heating. In: The Maillard Reaction in Food Chemistry phisms and postmenopausal breast cancer risk. and Science. Hillsborough, NJ: Excerpta Medica; Breast Cancer Res Treat. 2002;74:9-16. 2002. 57. Balbi JC, Larrinaga MT, De Stefani E, et al. Foods 41. Manabe S, Kurihara N, Wada O, Tohyama K, Ara- and risk of bladder cancer: a case-control study in maki T. Formation of PhIP in a mixture of creatinine, Uruguay. Eur J Canc Prev. 2001;10:453-458. phenylalanine and sugar or aldehyde by aqueous 58. Zheng W, Xie D, Cerhan J, Sellers T, Wen W, Fol- heating. Carcinogenesis. 1992;13:827-30. som A. Sulfotransferase 1A1 polymorphism, endog- 42. Lee H, Lin MY, Chan SC. Formation and identifica- enous estrogen exposure, well-done meat intake, tion of carcinogenic heterocyclic aromatic amines in and breast cancer risk. Cancer Epidemiol Biomark- boiled pork juice. Mutat Res. 1994;308:77-88. ers Prev. 2001;10:89-94. 43. Sinha R, Rothman N, Brown ED, et al. High concen- 59. Delfino RJ, Sinha R, Smith C, et al. Breast cancer, trations of the carcinogen 2-amino-1-methyl-6-imi- heterocyclic aromatic amines from meat and N- dazo[4,5-b]pyridine (PhIP) occur in chicken but are acetyltransferase 2 genotype. Carcinogenesis. dependent on the cooking method. Cancer Res. 2000;21:607-615. 1995;55:4516-4519. 60. Sinha R, Gustafson DR, Kulldorff M, Wen WQ, Cer- 44. Pais P, Tanga MJ, Salmon CP, Knize MG. Forma- han JR, Zheng W. 2-amino-1-methyl-6-phenylimi- tion of the mutagen IFP in model systems and dazo[4,5-b]pyridine, a carcinogen in high-tempera- detection in restaurant meats. J. Agric Food Chem. ture-cooked meat, and breast cancer risk. J Natl 2000;48:1721-1726. Cancer Inst. 2000;92:1352-1354. 45. Knize MG, Kulp KS, Salmon CP, Keating GA, Felton 61. Zheng W, Gustafson DR, Sinha R, et al. Well-done JS. Factors affecting human heterocyclic amine in- meat intake and the risk of breast cancer. J Natl take and the metabolism of PhIP. Mutat Res. 2002; Cancer Inst. 1998;90:1724-1729. 506:153-162. 62. Butler LM, Sinha R, Millikan RC, et al. Heterocyclic 46. Wang YY, Vuolo LL, Spingarn NE, Weisburger JH. amines, meat intake, and association with colon Formation of mutagens in cooked foods: V. the cancer in a population-based study. Am J Epide- mutagen reducing effect of soy protein concen- miol. 2003;157:434-445. trates and antioxidants during frying of beef. Cancer 63. Kampman E, Slattery ML, Bigler J, et al. Meat con- Lett. 1982;16:179-189. sumption, genetic susceptibility, and colon cancer 47. Skog K, Ja¨ gerstad M, Laser-Reuterswa¨ rd A. Inhib- risk: A United States multicenter case-control study. itory effect of carbohydrates on the formation of Cancer Epidemiol Biomarkers Prev. 1999;8:15-24. mutagens in fried beef patties. Food Chem Toxicol. 64. Sinha R, Chow WH, Kulldorff M, et al. Well-done, 1992;30:681-688. grilled red meat increases the risk of colorectal 48. Persson E, Sjoholm I, Skog K. Effect of high water- adenomas. Cancer Res. 1999;59:4320-4324. holding capacity on the formation of heterocyclic 65. Augustsson K, Skog K, Jagerstad M, Dickman PW, amines in fried beefburgers. J Agric Food Chem. Steineck G. Dietary heterocyclic amines and cancer 2003;51:4472-4477. of the colon, rectum, bladder, and kidney: a popu- 49. Vikse R, Joner PE. Mutagenicity, creatine and nu- lation-based study. Lancet. 1999;353:703-707.

164 Nutrition Reviewsா, Vol. 63, No. 5 66. Schiffman MH, Van Tassell R, Andrews AW. Epide- of cooked meat and risk of prostate cancer. J Natl miologic studies of fecal mutagenicity, cooked meat Cancer Inst. 1999;91:2038-2044. ingestion, and risk of colon cancer. Prog Clin Biol 82. Nowell S, Ratnasinghe DL, Ambrosone CB, et al. Res. 1990;340E:205-214. Association of SULT1A1 phenotype and genotype 67. Barrett JH, Smith G, Waxman R, et al. Investigation with prostate cancer risk in African-Americans and of interaction between N-acetyltransferase 2 and Caucasians. Cancer Epidemiol Biomarkers Prev. heterocyclic amines as potential risk factors for 2004;13:270-276. colorectal cancer. Carcinogenesis. 2003;24:275- 83. Murtaugh MA, Ma KN, Sweeney C, Caan BJ, Slat- 282. tery ML. Meat consumption patterns and prepara- 68. Tiemersma EW, Voskuil DW, Bunschoten A, et al. tion, genetic variants of metabolic enzymes, and Risk of colorectal adenomas in relation to meat their association with rectal cancer in men and consumption, meat preparation, and genetic sus- women. J Nutr. 2004;134:776-784. ceptibility in a Dutch population. Cancer Causes 84. Ward MH, Sinha R, Heineman EF, et al. Risk of Control. 2004;15:225-236. adenocarcinoma of the stomach and esophagus 69. Le Marchand L, Hankin JH, Pierce LM, et al. Well- with meat cooking method and doneness prefer- done red meat, metabolic phenotypes and colorec- ence. Int J Cancer. 1997;71:14-19. tal cancer in Hawaii. Mutat Res. 2002;506:205-214. 85. Zhu JJ, Chang P, Bondy ML, et al. Detection of 70. Nowell S, Coles B, Sinha R, et al. Analysis of total 2-amino-1-methyl-6-phenylimidazo[4,5-b]-pyri- meat intake and exposure to individual heterocyclic dine-DNA adducts in normal breast tissues and risk amines in a case-control study of colorectal cancer: of breast cancer. Cancer Epidemiol Biomarkers contribution of metabolic variation to risk. Mutat Prev. 2003;12:830-837. Res. 2002;506:175-185. 86. Thompson LH, Carrano AV, Salazar E, Felton JS, 71. Sinha R, Kulldorff M, Chow WH, Denobile J, Roth- Hatch FT. Comparative genotoxic effects of the man N. Dietary intake of heterocyclic amines, meat- cooked-food-related mutagens Trp-P-2 and IQ in derived mutagenic activity, and risk of colorectal bacteria and cultured mammalian cells. Mutat Res. adenomas. Cancer Epidemiol Biomarkers Prev. 1983;117:243-257. 2001;10:559-562. 87. Wu RW, Wu EM, Thompson LH, Felton JS. Identi- 72. Probst-Hensch NM, Sinha R, Longnecker MP, et al. ﬁcation of aprt gene mutations induced in repair- Meat preparation and colorectal adenomas in a deﬁcient and P450-expressing CHO cells by the large sigmoidoscopy-based case-control study in food-related mutagen/carcinogen, PhIP. Carcino- California (United States). Cancer Causes Control. genesis. 1995;16:1207-1213. 1997;8:175-183. 88. Lynch AM, Gooderham NJ, Davies DS, Boobis AR. 73. Gerhardsson de Verdier M, Hagman U, Peters RK, Genetic analysis of PHIP intestinal mutations in Steineck G, O¨ vervik E. Meat, cooking methods and MutaMouse. Mutagenesis. 1998;13:601-605. colorectal cancer: a case-referent study in Stock- 89. Ohgaki H, Takayama S, Sugimura T. Carcinogenici- holm. Int J Cancer. 1991;49:520-525. ties of heterocyclic amines in cooked food. Mutat 74. Gunter MJ, Probst-Hensch NM, Cortessis VK, Kull- Res. 1991;259:399-410. dorff M, Haile RW, Sinha R. Meat intake, cooking- 90. Ito N, Hasegawa R, Sano M, et al. A new colon and related mutagens and risk of colorectal adenoma in mammary carinogen in cooked food, 2-amino-1- a sigmoidoscopy-based case-control study. Carci- methyl-6-phenylimidazo[4,5-b]pyridine (PhIP). Car- nogenesis. 2005;26:637-642. cinogenesis. 1991;12:1503-1506. 75. Navarro A, Munoz SE, Lantieri MJ, et al. Meat cook- 91. Shirai T, Sano M, Tamano S, et al. The prostate: A ing habits and risk of colorectal cancer in Cordoba, target for carcinogenicity of 2-amino-1-methyl-6- Argentina. Nutrition. 2004;20:873-877. imidazo[4,5-b]pyridine (PhIP) derived from cooked 76. Terry PD, Lagergren J, Wolk A, Steineck G, Nyren foods. Cancer Res. 1997;57:195-198. O. Dietary intake of heterocyclic amines and can- 92. Schut HAJ, Putman KL, Randerath K, 32P-postla- cers of the esophagus and gastric cardia. Cancer beling analysis of DNA adducts in liver, small and Epidemiol Biomarkers Prev. 2003;12:940-944. large intestine of male Fischer-344 rats after intra- 77. Bosetti C, Negri E, Franceschi S, et al. Diet and peritoneal administration of 2-amino-3-methylimi- ovarian cancer risk: a case-control study in Italy. Int dazo[4,5-f]quinoline (IQ). In: King CM, Romano LJ, J Cancer. 2001;93:911-915. Schniltzle, eds. Carcinogenic and Mutagenic Re- 78. Sinha R, Kulldorff M, Curtin J, Brown CC, Alavanja sponses to Aromatic Amines and Nitroarenes. New MC, Swanson CA. Fried, well-done red meat and York: Elsevier; 1988:265-269. risk of lung cancer in women (United States). Can- 93. Snyderwine EG, Schut HAJ, Sugimura T, Nagao M, cer Causes and Control. 1998;9:621-630. Adamson RH. DNA adduct levels of 2-amino-1- 79. Zhang S, Hunter DJ, Rosner BA, et al. Dietary fat methyl-6-phenylimidazo-[4,5-b]pyridine (PhIP) in and protein in relation to risk of non-Hodgkin’s tissues of cynomolgus monkeys after single or mul- lymphoma among women. J Natl Cancer Inst. 1999; tiple dosing. Carcinogenesis. 1994;15:2757-2761. 91:1751-1758. 94. Dingley KH, Curtis KD, Nowell S, Felton JS, Lang 80. Anderson KE, Sinha R, Kulldorff M, et al. Meat NP, Turteltaub KW. DNA and protein adduct forma- intake and cooking techniques: associations with tion in the colon and blood of humans after expo- pancreatic cancer. Mut Res. 2002;506:225-231. sure to a dietary-relevant dose of 2-amino-1-methyl- 81. Norrish AE, Ferguson LR, Knize MG, Felton JS, 6-phenylimidazo[4,5-b]pyridine. Cancer Epidemiol Sharpe SJ, Jackson RT. Heterocyclic amine content Biomarkers Prev. 1999;8:507-512.

Nutrition Reviewsா, Vol. 63, No. 5 165 Food Chemistry 132 (2012) 1316–1323

Contents lists available at SciVerse ScienceDirect

Food Chemistry

journal homepage: www.elsevier.com/locate/foodchem

Mechanism of formation of sulphur aroma compounds from L-ascorbic acid and L-cysteine during the Maillard reaction ⇑ Ai-Nong Yu , Zhi-Wei Tan, Fa-Song Wang

School of Chemistry & Environmental Engineering, Hubei University for Nationalities, Enshi, Hubei 445000, China article info abstract

Article history: The sulphur aroma compounds produced from a phosphate-buffered solution (pH 8) of L-cysteine and L-, 13 13 Received 16 April 2011 L-[1- C] or L-[4- C] ascorbic acid, heated at 140 ± 2 °C for 2 h, were examined by headspace SPME in Received in revised form 3 November 2011 combination with GC–MS. MS data indicated that C-1 of L-ascorbic acid was not involved in the formation Accepted 25 November 2011 of sulphur aroma compounds. The sulphur aroma compounds formed by reaction of L-ascorbic acid with Available online 3 December 2011 L-cysteine mainly contained thiophenes, thiazoles and sulphur-containing alicyclic compounds. Among these compounds, 1-butanethiol, diethyl disulphide, 5-ethyl-2-methylthiazole, cis and trans-3,5- Keywords: dimethyl-1,2,4-trithiolane, thieno[2,3-b]thiophene, thieno[3,2-b]thiophene, cis and trans-3,5-diethyl- Maillard reaction 1,2,4-trithiolane, 1,2,5,6-tetrathiocane, 2-ethylthieno[2,3-b]thiophene, 2,4,6-trimethyl-1,3,5-trithiane Sulphur compound Ascorbic acid and cyclic octaatomic sulphur (S8) were formed solely by L-cysteine degradation, and the rest by reaction Cysteine of L-ascorbic acid degradation products, such as hydroxybutanedione, butanedione, acetaldehyde, acetol, Headspace-SPME pyruvaldehyde and formaldehyde with L-cysteine or its degradation products, such as H2S and NH3.A new reaction pathway from L-ascorbic acid via its degradation products was proposed. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction in the processes of non-enzymatic browning, and a series of researches on the behaviour of ascorbic acid in the presence of Sulphur aroma compounds constitute the most powerful aroma amino acids, via the Maillard reaction, is reported in the literature. compounds and often play, although at trace levels, a dominant ASA is a common ingredient of the human diet, occurring espe- role in the flavour of cooked meats and roasted coffee (Cerny, cially in fruit, vegetables, herbs and meat (liver), and is frequently 2008). The aroma of cooked meat is provided by a complex mixture used as a food additive, as an antioxidant and as a flour improver in of volatile compounds produced during the cooking (Mottram, bakeries (Adams & De Kimpe, 2009). So, it is important to investi- 1998). Among these volatiles, sulphur aroma compounds are con- gate formation of aroma compounds from ASA and Cys during the sidered to be particularly important. Sulphur aroma compounds Maillard reaction. are among the key aroma compounds of meat flavour. Sulphur- However, there is a lack of research findings on the formation of containing heterocyclic aroma compounds are known to play an aroma compounds from ASA and Cys during the Maillard reaction. important role in contributing meaty flavour to roasted and cooked As far as we know, there are only two papers related to formation meats. During cooking, a major route to sulphur aroma compounds of aroma compounds in the model reactions of ASA with Cys is the Maillard reaction. L-Cysteine (Cys) is an important precursor (Adams & De Kimpe, 2009; Yu & Zhang, 2010b). Adams and De for the formation of sulphur compounds and has been extensively Kimpe (2009) reported formation of furan derivatives and thio- used in the manufacturing of reaction flavours. The Maillard model phenes produced by heating a model reaction of ASA with Cys system involving ribose and Cys has been used widely to study under dry-roasting conditions in the presence of K2CO3, but data generation of meaty flavours (Hofmann & Schieberle, 1995; were not shown. Another paper was published by our laboratory Mottram & Whitfield, 1995a, 1995b; Werkhoff, et al., 1990). Over (Yu & Zhang, 2010b). We reported mainly the effect of pH on the 180 compounds have been identified from these reaction systems, formation of aroma compounds from ASA and Cys during the Mail- and the key odorants elicit an overall roasty, meat-like odour lard reaction and discovered that the reaction between ASA and (Hofmann & Schieberle, 1995). As mentioned in our previous paper Cys led mainly to the formation of alicyclic sulphur compounds, (Yu & Zhang, 2010a), after reducing carbohydrates, L-ascorbic acid thiophenes, thiazoles and pyrazines, most of which contain sul- (ASA) appears to be the most widely-studied carbonyl component phur. Many of these volatiles have meaty flavour. But, the mechanism of formation of sulphur aroma compounds from ASA and Cys ⇑ Corresponding author. Tel.: +86 0718 8431586; fax: +86 0718 8437832. during the Maillard reaction has not been elucidated. The objective E-mail address: [email protected] (A.-N. Yu). of this study was to elucidate the formation chemical pathways for

13 13 Table 1 and L-[4- C] ascorbic acid (99 atom-% C) were from Omicron Bio- Model reactions. chemicals, Inc. (South Bend, IN). Cys was from Shanghai Yuanju Biolog-

13 13 No. L- L-Ascorbic L-[1- C] Ascorbic L-[4- C] Ascorbic ical Technology Co., Ltd. (Shanghai, China). C5–C22 n-alkanes were from Cysteine acid acid acid Pure Chemical Analysis Co., Ltd. Na2HPO4,NaH2PO4 and NaOH were of A 1.5a analytical grade. Authentic samples (thieno[3,2-b]thiophene, 2-meth- B 0.8 0.8 yltetrahydrothiophen-3-one, 2-acetylthiophene, 2-acetyl-3-methyl- C 0.8 0.8 thiophene, 4,5-dimethylthiazole, 2,4,5-trimethylthiazole and 2- D 0.8 0.8 acetylthiazole), for use as GC reference compounds, were from J&K a Amount (mmol). Chemical Ltd. (Beijing, China). Double-distilled water was used in all experiments.

Table 2 2.2. Degradation of Cys Degradation products of L-cysteine at pH8.

Compounds LRI Areas Â 106 Cys (1.5 mmol, Table 1) was dissolved in 15 ml of phosphate buffer (0.2 M, pH 8). The mixtures were then sealed in 48-ml Syn- Hydrogen sulphide <500 76.4 Ò Ethanethiol 513 7.7 thware pressure glass vials (Beijing Synthware Glass, Inc., China) Thiophene 661 54.1 and heated while stirring at 140 ± 2 °C for 2 h in an oil bath. After 1-Butanethiol 710 18.5 heating, the reaction mixtures were quickly cooled to room tem- 2-Methylthiophene 761 4.4 perature and then adjusted to neutral pH 7 before SPME analysis, Diethyl disulphide 915 16.8 and the resulting products were analysed using headspace- 2,4,5-Trimethylthiazole 990 3.5 5-Ethyl-2-methylthiazole 1002 15.9 SPME–GC–MS. 3,5-Dimethyl-1,2,4-trithiolane (cis or trans) 1132 964.5 3,5-Dimethyl-1,2,4-trithiolane(cis or trans) 1140 1009.3 2.3. Model reaction of Cys with ASA Thieno[2,3-b]thiophene 1192 9.9

Thieno[3,2-b]thiophene 1196 167.6 13 4,6-Dimethyl-1,2,3-trithiane (cis or trans) 1231 12.2 The experimentation scheme is shown in Table 1. ASA, L-[1- C] 13 4,6-Dimethyl-1,2,3-trithiane (cis or trans) 1242 419.4 ascorbic acid and L-[4- C] ascorbic acid were dissolved in phos- 3,5-Diethyl-1,2,4-trithiolane(cis or trans) 1333 1013.6 phate buffer (0.2 M, pH 8; 10 ml of buffer/mmol of ascorbic acid), 3,5-Diethyl-1,2,4-trithiolane(cis or trans) 1341 1048.7 and the pH of the solutions was adjusted to 8.0 using NaOH with 1,2,5,6-Tetrathiocane (C4H8S4) 1387 1149.5 2-Ethylthieno[2,3-b]thiophene 1403 82.2 pH meter (Shanghai Precision & Scientific Instrument Co., Ltd.). 2,4,6-Trimethyl-1,3,5-trithiane 1446 64.9 Cys were added to the solutions. The mixtures were then sealed Ò Cyclic octaatomic sulphur (S8) >1800 5.0 in 48 ml Synthware pressure glass vials (Beijing Synthware Glass, Inc., China) and heated while stirring at 140 ± 2 °C for 2 h in an oil sulphur aroma compounds formed from ASA and Cys during the bath. The reaction mixtures were immediately stopped by cooling 13 13 Maillard reaction. L-[1- C] Ascorbic acid and L-[4- C] ascorbic under a stream of cold water and then adjusted to neutral pH 7 acid were used to elucidate the origin of the carbons in sulphur before SPME analysis. aroma compounds. In this study, the pH 8.0 was set according to our previous research (Yu & Zhang, 2010b), and the reaction tem- 2.4. Headspace-SPME–GC–MS perature and time were set according to usual Maillard reaction conditions. The sulphur aroma compounds were analysed by head- The sample analysis conditions by headspace-SPME–GC–MS space-SPME–GC–MS, a effective technology to analyse aromatic were as previously reported (Liu, Shi, & Yu, 2009). The assayed compounds. fibre was CAR/PDMS (75 lm thickness; Supelco, Bellefonte, PA, USA) as previously reported (Yu & Zhang, 2010b). Before the SPME fibre was inserted into the vial, the sample was equilibrated for 2. Materials and methods 15 min at 40 °C. The extraction time was 50 min at 40 °C. Analyses were performed using an Agilent 6890N gas chro- 2.1. Reagents matograph coupled to a Agilent 5975i mass selective detector (Agi- lent, Santa Clara, CA). Aroma compounds were separated using a ASA (analytical grade, P99.7%) was from Sinopharm Chemical Re- DB-5 capillary column (30 m Â 0.25 mm(i.d) Â 0.25 lm). The 13 13 agent Co., Ltd. (Beijing, China). L-[1- C] Ascorbic acid (99 atom-% C) SPME fibre was desorbed and maintained in the injection port at

OH OH OH OH O O O O HOH2CHC O Feather, 1993 Rizzi, 2005 Barham, et al., 2010 or

HO OH OH OH O OH O OH OH

L-[1,4-13C]Ascorbic acid Aldopentose 1-Deoxypentosone Acetol Acetol

Rizzi, 2005 Rizzi, 2005 O Rizzi, 2005 OH O O O

Acetaldehyde O O O O Butadione Hydroxybutadione Pyruvaldehyde Pyruvaldehyde

13 13 Fig. 1. Reaction pathways for the degradation of L-[1,4- C] ascorbic acid (d = C). Table 3 1318 13 13 MS data and odour evaluation results of sulphur aroma compounds produced from L-, L-[1- C] or L-[4- C] ascorbic acid and L-cysteine, respectively.

a 13 13 No. Compounds LRI Identiﬁcation L-Ascorbic acid L-[1- C] Ascorbic acid L-[4- C] Ascorbic acid Odour description 1 Thiophene 665 MS,LRI 84 (100,+ M), 58 (52), 45 (27), 39 (14), 57 84 (100, M +), 58 (50), 45 (28), 39 (13), 57 84 (100), 58 (70), 85 (65), 45 (41), 59 (17), 39 (17), Garlic (10), 83 (6), 69 (6), 85 (6), 50 (5) (10), 69 (6), 83 (6), 50 (5), 85 (5) 57 (13), 40 (9), 83 (7), 86 (7) 2 2,5-Dihydrothiophene 752 MS 85 (100), 86 (52, M+), 45 (20), 87 (7), 39 (6), 85 (100), 86 (53, M +), 45 (21), 87 (7), 39 (7), 86 (100), 87 (48, M +), 45 (18), 85 (12), 88 (6), 40 Cabbage 71 (6), 53 (5), 57 (5) 71 (5) (6), 72 (5), 54 (5) 3 2-Methylthiophene 762 MS,LRI 97 (100), 98 (55,+ M), 45 (9), 99 (8), 39 (6), 53 97 (100), 98 (56, M +), 45 (9), 99 (8), 53 (6), 39 98 (100), 99 (56), 100 (9), 45 (8), 97 (7), 54 (5), 40 Mildly (6), 58 (4), 59 (3) (6), 58 (4), 59 (3) (4), 58 (4), 39 (3) sulphurous 42- 985 MS,LRI,Co- 60 (100), 116 (67, M +), 59 (26), 45 (19), 88 60 (100), 116 (69, M +), 59 (24), 45 (18), 88 60 (100), 117 (74), 59 (24), 45 (17), 61 (13), 88 Meat-like Methyltetrahydrothiophen- GCb (10), 58 (8), 61 (5) (11), 58 (7), 61 (6) (10), 58 (9), 116 (7), 89 (2) 3-one 5 2-Acetylthiophene 1083 MS,LRI,Co- 111 (100), 126 (44, M+), 39 (15), 83 (10), 43 111 (100), 126 (45, M +), 39 (14), 83 (10), 43 113 (100), 112 (99), 128 (81), 40 (22), 84 (13), 85 Sulphurous, GC (9), 112 (6), 45 (6), 113 (5), 57 (5) (8), 112 (6), 45 (5), 113 (5), 57 (5) (12), 43 (10), 44 (9), 114 (9), 46 (7) Meaty 6 2-Acetyl-3- 1147 MS,LRI,Co- 125 (100), 140 (51, M+), 97 (16), 53 (15), 45 125 (100), 140 (50, M +), 97 (16), 53 (13), 45 126 (100), 127 (91), 142 (79, M +), 98 (18), 45 (17), Vegetables, methylthiophene GC (11), 43 (8), 126 (8), 127 (6) (11), 43 (9), 126 (8) 54 (16), 99 (13), 128 (11), 44 (10), 43 (10) Sour 7 3-(Vinylthio)thiophene 1303 MS 142 (100,+), M 141 (81), 97 (48), 45 (15), 143 142 (100, M +), 141 (80), 97 (46), 143 (15), 45 143 (100, M +), 142 (82), 98 (46), 144 (16), 45 (14), Garlic, (14), 69 (11) (14), 69 (11) 70 (11) Onion + + 8 4,5-Dimethylthiazole 926 MS,LRI,Co- 113 (100, M), 71 (63), 85 (28), 86 (28), 45 - 114 (100, M ), 72 (65), 86 (39), 45 (28), 42 (22), 87 Earthy 1316–1323 (2012) 132 Chemistry Food / al. et Yu A.-N. GC (27), 41 (20) (22) 9 2,4,5-Trimethylthiazole 989 MS,LRI,Co- 127 (100,+ M), 71 (76), 86 (67), 85 (28), 59 127 (100, M +), 71 (72), 86 (61), 85 (25), 59 128 (100), 72 (75), 86 (64), 87 (63), 127 (63), 71 Frozen GC (19), 45 (14), 58 (11) (22), 45 (12) (54), 59 (30), 85 (18), 45 (16), 58 (14) meat 10 2-Acetylthiazole 1012 MS,LRI,Co- 43 (100), 127 (71, M+), 99 (61), 112 (43), 58 43 (100), 127 (77, M +), 99 (70), 112 (49), 58 128 (100, M +), 43 (89), 100 (89), 44 (76), 58 (57), Roasted, GC (40), 57 (24), 85 (19), 84 (15), 45 (10) (40), 57 (24), 85 (22), 84 (16), 45 (9) 113 (38), 57 (34), 112 (29), 85 (26), 86 (18) Meaty 11 4,6-Dimethyl-1,2,3- 1235 MS 166 (100,+ M), 60 (44), 59 (41), 102 (39), 45 166 (100, M +), 102 (38), 60 (36), 69 (34), 59 167 (100, M +), 60 (48), 103 (41), 59 (39), 70 (38), Garlic trithiane (cis or trans) (38), 69 (38), 101 (28), 64 (27) , 92 (27) (33), 101 (29), 45 (28), 92 (25), 64 (24) 102 (31), 64 (30), 92 (28) 12 4,6-Dimethyl-1,2,3- 1244 MS 166 (100,+ M), 60 (56), 59 (51), 45 (39), 102 166 (100, M +), 60 (50), 59 (45), 102 (37), 92 167 (100, M +), 60 (61), 59 (50), 103 (40), 70 (38), Garlic trithiane (cis or trans) (39), 69 (37), 92 (36), 64 (33), 101 (28) (34), 69 (33), 45 (32), 64 (30), 101 (28) 92 (37), 64 (36), 102 (30), 45 (27) 13 1-Butanethiol 712 MS,LRI 56 (100), 41 (80), 90 (79,+ M), 44 (56), 32 (41), 47 (39), 39 (24), 45 (24), 61 (22), 55 (21) Onion 14 Diethyl disulphide 916 MS,LRI 122 (100, M+), 66 (70), 94 (54), 107 (16), 59 (16), 45 (13), 67 (12), 60 (10) Sulphury, Cabbage 15 5-Ethyl-2-methylthiazole 1003 MS,LRI 127 (100, M+), 112 (84), 71 (67), 85 (47), 94 (42), 86 (36), 126 (29), 45 (26) Chicken broth 16 3,5-Dimethyl-1,2,4- 1134 MS,LRI 152 (100, M+), 59 (72), 92 (55), 88 (48), 60 (45), 64 (39), 45 (31), 55 (22), 58 (21), 154 (14) Sulphury, trithiolane(cis or trans) Onion 17 3,5-Dimethyl-1,2,4- 1142 MS,LRI 152 (100, M+), 59 (65), 92 (54), 88 (47), 60 (39), 64 (38), 45 (27), 55 (22), 58 (18), 154 (13) Sulphury, trithiolane(cis or trans) Onion 18 Thieno[2,3-b]thiophene 1194 MS,LRI 140 (100, M+), 96 (22), 69 (12), 142 (10), 141 (9), 95 (8), 45 (8), 70 (7), 63 (6) Smoky 19 Thieno[3,2-b]thiophene 1199 MS,LRI,Co- 140 (100, M+), 96 (21), 69 (14), 70 (10), 142 (9), 141 (8), 45 (8), 95 (7), 71 (5) Bacon GC 20 3,5-Diethyl-1,2,4- 1335 MS,LRI 180 (100, M+), 55 (58), 45 (44), 87 (41), 115 (37), 116 (36), 59 (36), 60 (32) Garlic trithiolane(cis or trans) 21 3,5-Diethyl-1,2,4- 1343 MS,LRI 180 (100, M+), 55 (53), 45 (45), 87 (38), 115 (36), 116 (36), 59 (33), 60 (28) Garlic trithiolane(cis or trans) 22 1,2,5,6-Tetrathiocane 1388 MS 59 (100), 60 (48), 184 (46, M +), 124 (34), 45 (18), 64 (16), 58 (11), 119 (11), 61 (10) Onion (C4H8S4) 23 2-Ethylthieno[2,3- 1406 MS 153 (100), 168 (47, M+), 167 (10), 69 (9), 154 (9), 155 (9), 45 (7), 169 (6), 109 (4) Roasted b]thiophene meat 24 2,4,6-Trimethyl-1,3,5- 1448 MS 180 (100, M+), 115 (92), 55 (84), 45 (70), 59 (55), 60 (54), 92 (50), 87 (50) Garlic trithiane 25 Cyclic octaatomic sulphur >1800 MS 64 (100), 256 (36, M +), 160 (31), 128 (30), 192 (22), 96 (17), 32 (15), 258 (13) Sulphurous (S8)

a LRI calculated for a DB-5 capillary column; mean values. b Co-injection with authentic sample. A.-N. Yu et al. / Food Chemistry 132 (2012) 1316–1323 1319

O O

S S S S S S S

1a 1b 2 3a 3b 4a 4b S

S S S S S S

5a O 5b O 5c O 6aO 6b O 7

N N N N N N N

S S S S S S S

8a 8b 9a 9b 9c 10a O 10b O

13 13 Fig. 2. Isotope-labelled sulphur aroma compounds formed from L-[4- C] ascorbic acid and L-cysteine (d = C). Compound nos. correspond to Table 3. the oven temperature (250 °C) and for the time (4.0 min) suggested result. At frequency of less than 4, detections were considered as by the manufacturer. The injection port was in split mode and split noises. ratio was 30:1. The temperature programme was isothermal for 5 min at 40 °C, raised to 260 °C at a rate of 5 °C minÀ1 and then raised to 280 °C at a rate of 15 °C minÀ1 and held for 1 min. C5– 3. Results and discussion C22 n-alkanes were run under the same chromatographic conditions as the samples to calculate the linear retention indices (LRI) 3.1. General of detected compounds. The transfer line to the mass spectrometer was maintained at 280 °C. The mass spectra were obtained using a The thermal degradation of Cys at 140 ± 2 °C and pH = 8 for 2 h mass selective detector by electronic impact at 70 eV, a multiplier gave a light yellow liquid, which had meat-like aroma and sulphur voltage of 1753 V, and collecting data at a rate of 1 scan sÀ1 over smell. The SPME-GC–MS analysis identified the volatile products the m/z range of 30–400 u.m.a. listed in Table 2, which were all sulphur-containing compounds. Aroma compounds were identified by comparing their mass Therefore, Cys itself can degrade to form certain sulphur-contain- spectra with those contained in the Nist05 and Wiley275 libraries ing aroma compounds. Zhang, Chien, and Ho (1988) investigated and by comparison of their LRIs with the National Institute of Stan- the volatile compounds obtained from thermal degradation of dards and Technology (NIST) 2009 Gas Chromatography Library cysteine in water at 180 °C for 1 h. Thiophene, 5-ethyl-2-methyl- (http://webbook.nist.gov/chemistry), as well as, whenever possi- thiazole, cis- and trans-3,5-dimethyl-1,2,4-trithiolane and 2,4,6-tri- ble, Co-GC injection with authentic samples available in our methyl-1,3,5-trithiane have also been found, but some compounds laboratories. When no published LRI information and authentic were not detected in the present work. This could be attributed to samples were available, the identification was established by com- the different reaction conditions of, e.g., pH, reaction temperature paring their fragmentation patterns of mass spectra with published and reaction time. 13 data (http://webbook.nist.gov/chemistry). Analysis of each tested The degradation of ASA has been widely reported. The 1,4- Cla- 13 condition was repeated twice and MS spectra data corroborated belled ASA used in this study can degrade and form 3- Clabelledaldo- with each other. pentose according to the pathway described by Feather (1993).Inour current work, the Maillard reaction takes place in phosphate buffer solution. The aldopentose formed by ASA degradation was reported 2.5. Odour analysis by GC–O to form intermediates, such as 1-deoxypentosone, hydroxybutanedione, butanedione and acetaldehyde, under the catalysis of phosphate GC–O analysis was carried out on an Agilent 6890N gas chro- (Rizzi, 2005). Besides, 1-deoxypentosone can form acetol, pyruvalde- matograph (Agilent, Santa Clara, CA) coupled to a Sniffer 9000 hyde (Barham et al., 2010), or degrade further to generate formalde- olfactometer (Brechbühler Scientific Analytical Solutions INC, Swit- hyde (Cerny & Davidek, 2004). These intermediates of ASA zerland). Aroma compounds were separated using an HP-5MS cap- degradation (hydroxybutanedione, butanedione, acetaldehyde, acetol, illary column (30 m Â 0.25 mm(i.d) Â 0.25 lm). The SPME fibre pyruvaldehyde, formaldehyde) may react with Cys or its degradation was desorbed and maintained in the injection port at the oven products to generate a variety of aroma compounds. The degradation temperature 270 °C and for a period of 4.0 min. The injection port pathway of 1,4-13C labelled ASA is shown in Fig. 1. was in splitless mode. The temperature programme was isother- Table 3 shows the mass spectral data and odour evaluation results mal for 5 min at 40 °C, raised to 210 °C at a rate of 5 °C minÀ1 of the major sulphur aroma compounds formed by reaction of Cys with À1 13 13 and then raised to 280 °C at a rate of 25 °C min . By smelling ASA, L-[1- C] ascorbic acid and L-[4- C] ascorbic acid, respectively, and recording the odour descriptions, three trained assessors were under the same conditions. 1-Butanethiol, diethyl disulphide, 2,4,6-tri- selected for the GC–O experiment. Retention times of the odour methyl-1,3,5-trithiane, thiophene, 2,5-dihydrothiophene and 2-meth- responses were converted into LRI values, using the retention ylthiophene were not reported in our previous research (Yu & Zhang, times of a series of n-alkanes (C5–C22). Analysis of each tested 2010b). In Table 3, the compounds (Nos. 13–25) have very similar mass condition was repeated twice and, in total, six assessments were spectral data (data shown only in Cys/ASA system), no isotopic signal, carried out. At a frequency of not less than 4, the odour found in and were found in the Cys degradation products (Table 2). Therefore, six assessments of the extract was treated as the odour evaluation they were formed by Cys degradation. Besides, in Table 3, all sulphur 1320 A.-N. Yu et al. / Food Chemistry 132 (2012) 1316–1323

OH OH OH A

-2H H2S -2H2O OH S S S OH H+ O OH O H 2 1b O H Red. O

H+ + H2S O O O Formaldehyde Pyruvaldehyde OH O O H

S OH H2S -H2O OH S S OH H+ H OH OH S S S

-2H2O -2H

S S S 7

O O B

Red. H2S H+ -2H2O S HO S H O O OH O O O S H+ 3b H O OH O O O OH O O Red. H2S -H2O Acetaldehyde Pyruvaldehyde OH S S OH H+ H 4b

O O O O C O H O

Red. H2S -H2O Red. S -H O S OH O OH HO SH HO 2 O O O O O O O 5b Pyruvaldehyde

OH OH O O O OH O OH O

H -H2O Red. H2S S -H2O S O OH OHHO OHHS O 6a O Hydroxybutadione Hydroxyacetone O O

13 13 Fig. 3. Proposed formation pathway for thiophenes from L-[4- C] ascorbic acid and L-Cysteine (d = C). Compound nos. correspond to Fig. 2.

13 compounds formed by Cys with ASA and L-[1- C] ascorbic acid had lar- aroma compounds, besides Cys degradation products, is discussed gely the same mass spectra. Moreover, the sulphur-containing com- as follows. 13 pounds formed by L-[1- C] ascorbic acid and Cys had no isotopic signal. Therefore, the C-1 of ASA was not involved in the formation of 3.2. Formation of thiophenes sulphur aroma compounds. According to the aforementioned degradation pathway of 1,4-13CASA(Fig. 1), the decarboxylated intermediate The formation mechanism of seven thiophenes (compounds Nos. 13 of ASA may have been involved in the formation of sulphur aroma 1–7) from L-[4- C] ascorbic acid and Cys, during the Maillard reac- compounds. tion, were studied in this work. Some of the thiophenes formed by L- The mass spectral data of compounds (Nos. 1–10) formed by [4-13C] ascorbic acid and Cys were a mixture of isotope-labelled and 13 13 reaction of Cys with ASA, L-[1- C] ascorbic acid and L-[4- C] unlabelled compounds, such as compounds 1a, 1b and 3a, 3b. These ascorbic acid, under the same conditions, were compared and ana- compounds may have formed, in part, by Cys degradation and, in 13 lysed (Table 3). The most likely structure of the isotope-labelled part, by reaction of L-[4- C] ascorbic acid with Cys. This point is fur- 13 sulphur aroma compounds formed by L-[4- C] ascorbic acid and ther confirmed as these thiophene compounds were found in the Cys is shown in Fig. 2. The formation mechanism of other sulphur Cys degradation products (Table 2). Compounds 4a and 5a had no A.-N. Yu et al. / Food Chemistry 132 (2012) 1316–1323 1321 isotope label but were not found in Cys degradation products, and 1999). The formation mechanisms of compounds 3b and 4b are shown their formation mechanism needs further study. in Fig. 3B. Among the isotope-labelled thiophenes, compounds 1b, 2 and 7 The formations of other thiophenes have similar mechanism to may have been formed by (1) aldol condensation of formaldehyde the aforementioned thiophenes shown in Fig. 3C. The a-diketones, with the pyruvaldehyde formed by ASA degradation (Fig. 1), or (2) a-hydroxy ketones (e.g., pyruvaldehyde, acetol, hydroxybutanedi- reduction of two carbonyls or one carbonyl of the condensation one) derived from ASA degradation and the acetaldehyde, derived product. A possible pathway for the formation of compounds 1b, 2 from Cys degradation, condense with each other, then react with is the reaction of the two carbonyl reduction products with hydro- the H2S released from Cys degradation and undergo dehydration gen sulphide, which was released by Cys degradation, followed by and reduction to generate other thiophenes. Thiophene, 2-methyl- dehydration, to yield compound No. 2, and subsequent dehydroge- thiophene, 2-acetylthiophene, 2-methyltetrahydrothiophen-3-one nation to get compound No. 1b. The one carbonyl reduction product and 2-acetyl-3-methylthiophene have been identified in thermal reacts with hydrogen sulphide to yield a sulphuration product, fol- reaction of ribose and cysteine (Chen, Xing, Chin, & Ho, 2000). lowed by dehydration to get tetrahydrothiophen-3-one, which was also identified in the cysteine/glucose system (Tai & Ho, 1997) and 3.3. Formation of thiazoles the cysteine/ASA system (Yu & Zhang, 2010b). However, it was not identified in this work and perhaps it was in trace amount and The formation mechanisms of three thiazoles (compound Nos. 13 was depleted in the following reaction. In addition, hydrogen sul- 8–10) from L-[4- C] ascorbic acid and Cys during the Maillard phide can react with acetaldehyde, which comes from Cys or ASA reaction were studied in this work. The possible formation mecha- degradation, to yield hemi-mercaptal. The hemi-mercaptal is a nisms of 4,5-dimethylthiazole (No. 8a) and 2,4,5-trimethylthiazole strong nucleophile for its mercapto group adds to tetrahydrothio- (No. 9b) are outlined in Fig. 4A (The formation of compounds 8b phen-3-one. The added product is dehydrated and dehydrogenated and 9c has a similar mechanism to 8a and 9b, respectively). In gen- to get compound No. 7. The proposed formation mechanism is eral, ASA degrades to give butanedione, which can react with H2S shown in Fig. 3A. to form thiols. The formaldehyde and acetaldehyde react with Compounds 3b and 4b may have been formed by aldol condensa- NH3, formed by Cys degradation (Sohn & Ho, 1995), to form the tion between pyruvaldehyde (from the degradation of ASA) and acet- imine (Schiff bases), which eventually react with the thiols to gen- aldehyde (from degradation of Cys or ASA) following tautomerisation. erate the thiazoles. The ion with m/z = 127 was found in the 2,4,5- 13 The other essential steps in the formation pathways of compound 3b trimethylthiazole formed by L-[4- C] ascorbic acid and Cys, arereactionwithhydrogensulphide,ring closure, reduction and dehy- whereas the ion with m/z = 126 was not observed in the 2,4,5-tri- dration. Compound 4b was formed through reduction, hydrogen sul- methylthiazole formed by ASA and Cys. The ion with m/z 127 is the phide addition and dehydration ring closure. Compound 4b was molecular ion of unlabelled 2,4,5-trimethylthiazole and has high identified in the 4-hydroxy-5-methyl-3(2H)-furanone/cysteine sys- abundance. Therefore, there was a certain amount of unlabelled tem and has similar formation pathways (Whitfield & Mottram, 2,4,5-trimethylthiazole (No. 9a), which came solely from Cys deg-

NH3 HCHO CH2=NH A O O O OH HN H2N H2S CH2=NH CH CH SH 2 2 HO S HO S O O H Butadione OH

NH N N N -H2O -H2O CH 2 HO HO HO S S S S 8a

O O

H2S O NH N O SH 2 N OH -H2O Butadione S O S S NH OH OH NH3 9b H H Acetaldehyde

O B O H SH SH O S NH N O CH2 CH2 -CO2 -4H Pyruvaldehyde CH2 CH NH2 CH N S S -H2O CH2 N COOH O O COOH 10a

13 13 Fig. 4. Proposed formation pathway for thiazoles from L-[4- C] ascorbic acid and L-cysteine (d = C). Compound nos. correspond to Fig. 2. 1322 A.-N. Yu et al. / Food Chemistry 132 (2012) 1316–1323

OH OH OH OH OH O Red. H2S H2SO -H2O, Red. H S S S S O O OH OH OH SH SH S S Hydroxyacetone 11 and 12

13 13 Fig. 5. Proposed formation pathway for 4,6-dimethyl-1,2,3-trithiane from L-[4- C] ascorbic acid and L-cysteine (d = C). Compound nos. correspond to Fig. 2. radation because 2,4,5-trimethylthiazole was found in the Cys deg- reaction of ASA degradation products, such as hydroxybutanedione, radation products (Table 2). 2,4,5-Trimethylthiazole was identiﬁed butanedione, acetaldehyde, acetol, pyruvaldehyde and formaldehyde, by Xi and Ho (2005) in the carbonyls/ammonium sulphide system. with Cys or its degradation products, such as H2SandNH3. The C-1 According to the isotopic label position in the 2-acetylthiazole of ASA was not involved in the formation of sulphur aroma compounds. 13 formed by L-[4- C] ascorbic acid and Cys, the 2-acetylthiazole (No. 10a) may have formed according to the pathway described Acknowledgements by Mulders (1973), as outlined in Fig. 4B. 3-13C Labelled pyruvalde- 13 hyde, originating from L-[4- C] ascorbic acid (Fig. 1), reacts well The authors thank the National Natural Science Foundation of with Cys, followed by a decarboxylation. The ring closure is China (20876036) for the ﬁnancial support of this investigation effected by nucleophilic attack of the thiol group on the double and the Flavour Chemistry group of the Beijing Technology and bond. Dehydrogenation of the thiazolidine compound produces Business University for participation and assistance during the the 2-acetylthiazole (No. 10a) in the suggested scheme. The forma- GC-O analysis. tion of compound 10b, formed by 1-13C labelled pyruvaldehyde, has a mechanism similar to 10a. Appendix A. Supplementary data

3.4. Formation of 4,6-dimethyl-1,2,3-trithiane Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.foodchem.2011.11.111. The 4,6-dimethyl-1,2,3-trithiane has cis and trans isomers (No.11, 12), which have identical mass spectra. According to the 13 References mass spectra, the 4,6-dimethyl-1,2,3-trithiane formed by L-[4- C] ascorbic acid and Cys has the isotopic label on C-5. The ion with Adams, A., & De Kimpe, N. (2009). Formation of pyrazines from ascorbic acid and m/z = 166, which is the molecular ion of unlabelled 4,6-dimethylamino acids under dry-roasting conditions. Food Chemistry, 115, 1417–1423. 1,2,3-trithiane, was not found in 4,6-dimethyl-1,2,3-trithiane Barham, P., Skibsted, L. H., Bredie, W. L., Frøst, M. B., Møller, P., Risbo, J., et al. (2010). 13 Molecular gastronomy: a new emerging scientific discipline. Chemical Reviews, formed from the Cys/L-[4- C] ascorbic acid system. However, 110, 2313–2365. among the Cys degradation products, 4,6-dimethyl-1,2,3-trithiane Cerny, C. (2008). The aroma side of the Maillard reaction. Annals of the New York was detected. The reason needs to be further studied. Considering Academy of Sciences, 1126, 66–71. Cerny, C., & Davidek, T. (2004). a-Mercaptoketone formation during the Maillard the formation mechanism of 4,7-dimethyl-1,2,3,5,6-pentathiepane, reaction of cysteine and [1-C-13]ribose. Journal of Agricultural and Food 4,6-dimethyl-1,2,3,5-tetrathiane and 3,5,7-trimethyl-l,2,4,6-tetrat- Chemistry, 52, 958–961. hiepane (Zhang et al., 1988), the possible formation mechanism of Chen, Y., Xing, J., Chin, C.-K., & Ho, C.-T. (2000). Effect of urea on volatile generation from Maillard reaction of cysteine and ribose. Journal of Agricultural and Food 4,6-dimethyl-1,2,3-trithiane is outlined in Fig. 5. The formation Chemistry, 48, 3512–3516. steps in the pathways of 4,6-dimethyl-1,2,3-trithiane are aldol con- Feather, M. S. (1993). Dicarbonyl sugar derivatives and their role in the Maillard densation between acetol (from the degradation of ASA) and acetal- reaction. In T. H. Parliment, M. J. Morello, & R. J. McGorrin (Eds.), Thermally Generated Flavors: Maillard, Microwave, and Extrusion Processes, ACS Symposium dehyde (from degradation of Cys or ASA), carbonyl reduction, Series 543 (pp. 127–141). Washington, DC: American Chemical Society. reaction with hydrogen sulphide, oxidative ring closure, dehydra- Hofmann, T., & Schieberle, P. (1995). Evaluation of the key odorants in a thermally tion and reduction. treated solution of ribose and cysteine by aroma extract dilution techniques. Journal of Agricultural and Food Chemistry, 43, 2187–2194. Liu, Y. X., Shi, Y. F., & Yu, A. N. (2009). Extraction and identification of volatile 4. Conclusions generation from Maillard reaction of ascorbic acid and cysteine. Journal of Hubei University for Nationalities (Natural Science Edition), 27, 241–247. Mottram, D. S. (1998). Flavour formation in meat and meat products: a review. Food This work studies the formation mechanism of sulphur aroma com- Chemistry, 62, 415–424. pounds from the Maillard reaction between ASA and Cys. The degrada- Mottram, D. S., & Whitfield, F. B. (1995a). Volatile compounds from the reaction of cysteine, ribose, and phospholipid in lowmoisture systems. Journal of tion products of Cys were first explored to determine their role in the Agricultural and Food Chemistry, 43, 984–988. Maillard reaction and ascertain which aroma compounds resulted Mottram, D. S., & Whitfield, F. B. (1995b). Maillard-lipid interaction in nonaqueous from Cys degradation, either partly or completely. The study on the systems: volatiles from the reaction of cysteine and ribose with 13 phosphatidylcholine. Journal of Agricultural and Food Chemistry, 43, 1302–1306. L ASA degradation focused on the degradation pathway of -[1- C] Mulders, E. J. (1973). Volatile components from the non-enzymic browning reaction 13 and L-[4- C] ascorbic acid and the role of degradation products in Mail- of the cysteine/cystine-ribose system. Zeitschrift für Lebensmitteluntersuchung lard reaction, based on previous reports. It was found that the sulphur- und -Forschung A, 152, 193–201. containing aroma compounds formed by reaction of ASA with Cys Rizzi, G. P. (2005). Role of phosphate and carboxylate ions in Maillard browning. In D. K. Weerasinghe & M. K. Sucan (Eds.), Process and Reaction Flavors: Recent mainly contained thiophenes, thiazoles and sulphur-containing alicy- Developments, ACS Symposium Series 905 (pp. 157–168). Washington, DC: clic compounds. Among these compounds, 1-butanethiol, diethyl American Chemical Society. disulphide, 5-ethyl-2-methylthiazole, cis and trans-3,5-dimethyl- Sohn, M., & Ho, C. T. (1995). Ammonia generation during thermal degradation of amino acids. Journal of Agricultural and Food Chemistry, 43, 3001–3003. 1,2,4-trithiolane, thieno [2,3-b]thiophene, thieno[3,2-b]thiophene, cis Tai, C. Y., & Ho, C. T. (1997). Influence of cysteine oxidation on thermal formation of and trans-3, Maillard aromas. Journal of Agricultural and Food Chemistry, 45, 3586–3589. 5-diethyl-1,2,4-trithiolane, 1,2,5,6-tetrathiocane, 2-ethylthieno[2,3- Werkhoff, P., Brüning, J., Emberger, R., Güntert, M., Köpsel, M., Kuhn, W., et al. (1990). Isolation and characterization of volatile sulfur-containing meat flavor b]thiophene, 2,4,6-trimethyl-1,3,5-trithiane and cyclic octaatomic sul- components in model systems. Journal of Agricultural and Food Chemistry, 38, phur (S8) were formed by Cys degradation, and the rest were formed by 777–791. A.-N. Yu et al. / Food Chemistry 132 (2012) 1316–1323 1323

Whitfield, F. B., & Mottram, D. S. (1999). Investigation of the reaction between 4- Yu, A. N., & Zhang, A. D. (2010a). The effect of pH on the formation of aroma a hydroxy-5-methyl-3(2H)-furanone and cysteine or hydrogen sulfide at pH 4.5. compounds produced by heating a model system containing L-ascorbic acid Journal of Agricultural and Food Chemistry, 47, 1626–1634. with L-threoine/L-serine. Food Chemistry, 119, 214–219. Xi, J., & Ho, C.-T. (2005). Formation of flavor compounds by the reactions of Yu, A. N., & Zhang, A. D. (2010b). Aroma compounds generated from thermal carbonyls and ammonium sulfide under low temperature. In D. K. Weerasinghe reaction of L-ascorbic acid with L-cysteine. Food Chemistry, 121, 1060–1065. & M. K. Sucan (Eds.), Process and Reaction Flavors: Recent Developments. ACS Zhang, Y., Chien, M., & Ho, C.-T. (1988). Comparison of the volatile compounds Symposium Series 905 (pp. 105–116). Washington, DC: American Chemical obtained from thermal degradation of cysteine and glutathione in water. Journal Society. of Agricultural and Food Chemistry, 36, 992–996. aey1t gL(eRvladBrrn 93 Bartowsky 1993; Bertrand and Revel (de approxi- mg/L malolactic is 5 and wines to alcoholic Bordeaux 1 these of in of product mately concentration most a its in is fermentations; DI incorporation vic- wines, its its out In and products. mark products, groups possible methyl of inal number the reduces symmetry exists ageing. reaction wine Maillard results during of These occur version can 2010). mild and a others 1912) that (Maillard and fact the Loscos with Tom- 2003; agree 2003; others others and and Ferreira inaga Silva 2000; others 2000; and 1999, others Marchand and (Cutzach proposed For been have pathway 2008). type Maillard- Guntz-Dubini classical the and furans, compounds, (Cerny heterocyclic and oxygenated studied pathways been generation have chemical some origins even identified been and have quantified them of and some hete- wines, 6-member In or compounds. rocyclic are 5- reaction nitrogen-containing Maillard or and the oxygen-, fried, from products (baked, sulfur-, odor products the food of processed Most many roasted). of colors and vor ute erdcinwtotpriso sprohibited is permission without reproduction Further 10.1111/j.1750-3841.2011.02261.x doi: 5/3/2011 Accepted 1/27/2011, Submitted 20110119 MS Introduction St 2,4,5-Trimethyloxazole and 2-Methylthiazolidine, 2-Methyl-3-thiazoline, 2-Methylthiazole, of Origins Mixed for Mechanisms Proposed Conditions: under Wine-Like Diacetyl with Reaction Cysteine The hmnd esteC 00,3 8 ilnv ’ro ee,Fac.Direct France. (E-mail:[email protected]). cedex, Marchand D’Ornon ISVV—210, author Villenave to 882 Bordeaux, inquiries 33 50008, de Leysotte—CS Bordeaux—Univ. de Chemin de polytechnique INRA—Institut C pai acad onAm,adGle eRevel de Gilles and Almy, John Marchand, ´ ephanie ictl(I a enapplrratn nmdlsses Its systems. model in reactant popular a been has (DI) Diacetyl fla- characteristic the for responsible is reaction Maillard The 01Isiueo odTechnologists Food of Institute 2011 2 ecin between reactions Keywords: Abstract: eeain h eeaino eeoylc.Te losgettepeec fuepoe dru compounds. odorous unexplored of presence the suggest heterocyclic on also parameters They physicochemical heterocyclics. and Like of compounds the compounds. other generation heterocyclic for of the to impact appropriate generations dicarbonyls the paths from anticipate reaction they leading pathways, mechanisms of chemical the terms the on in all in light run interpreted shed pathways are was These results conditions. reaction to The cysteine mild methyl-C(2) 2-methylthiazolidine. parallel the only the but for groups, provided a 2,4,5-trimethyloxazole methyl fragment cysteine and question, with and this 2-methyl-3-thiazoline 2-methylthiazole, marked DI supplied this of both thus formation reaction, explore DI the To cysteine of in and cysteine. carbons degrees DI fragment C(4) varying from the methyl-C(2) and in or C(1) that this demonstrated the DI of With distribution from product 3,4-hexanedione. origin by supplied The replaced be C(2). was DI at could which attached it groups clear; methyl not having was heterocycles member 5 1–3-N,O and ◦ ,ehnlwtr1%vv H35,pout ftedaey D)rato ihcsen nld ubro 1,3-N,S of number a include cysteine with reaction (DI) diacetyl the of products 3.5), pH v/v, 12% ethanol/water C, ato h buuto ie”cnb asdb h rsneo dru eeoylspoue ychemical by produced heterocycles odorous of presence the by caused be can wines” of “bouquet the of part A ytie ictl ao,mcaim thiazole mechanism, flavor, diacetyl, cysteine, S aioaisand acids -amino R α dcroy opud.Udrwn gigpyi-hmclcniin (20 conditions physic-chemical ageing wine Under compounds. -dicarbonyl . S Œnologie— USC n tes20) mn h oepoietpout r thia- groups are methyl having products oxazoles and prominent thiazolidines, more thiazolines, the (Marchandzoles, Among 2-acetylthiazole 2002). concerns others It and studied. been has ditions identified been have 2000). few others a and only daunting (Marchand which a of produces compounds, be- DI of reaction or number the methylglyoxal is and roasted cysteine origin and that Their nutty tween bouquet. on wine thiazoles the of thiazoles of impact component of dean olfactory set suggest to levels compared a tection quantities Then, The was wines. reaction conditions. in quantified Maillard-like ageing was of wine possibility under the studied precursors, as teine and (Starkenmann aroma a 2008). have and flavor others products, food other on the These influence with significant odors. proportions in intrinsic acceptable atoms intense in sulfur very odors, of with inclusion products the Maillard for many responsible is It and Yu 2010). 2008; Guntz-Dubini 1998; Zhang and others Cerny and 2007; others Horiuchi and Methven 1998; others Schieberle and and (Huang reactions Hofmann model functionality many 1998; of diverse subject Its the it 2001). made has others and (Pripis-Nicolau mg/L Jenny carbo- 2009). of 1999; classical Glomb step Keyhani and 1st the and (Yaylayan the of In degradation products thermal the 2005). hydrates of others one is and it reaction, Flamini Maillard 2004; Henschke and nyoeptwyo haoe eeainudrwn-iecon- wine-like under generation thiazoles of pathway one Only and systems, model Using 6 to 1 approximately at wines and musts in present is Cysteine o.7,N.6 2011 6, Nr. 76, Vol. α dcroy opud n cys- and compounds -dicarbonyl r ora fFo Science Food of Journal C861 ±

C: Food Chemistry C C, ◦ ◦ mfilm μ C/min to 220 ◦ L) were carried out in μ C and then at a rate of 5 C, the initial step lasting 1 min, at a rate of ◦ ◦ ´ I 0.2 mm column containing BP-21, 0.25- n-alkanes was mixed from samples purchased from All- C, respectively. The oven temperature was programed ◦ Cinthedark. 23 ◦ 2 -C 7 ± C/min to 200 An HP Model 5890 Series II gas chromatograph (Agilent tech- A stock wine model solution was prepared by dissolving 4 g of For the standard reactions, cysteine hydrate hydrochloride, Aliquots (5 mL) were withdrawn, spiked with internal standard The data for the 3,4-hexanedione and cysteine reaction were All the experiments were done in triplicate. The data were recorded and analyzed using anThe Agilent Wiley HP registry of mass spectral data (7th Ed, Chichester, ◦ thickness (Supelco, Lyon, France). Carrierconstant gas flow was rate helium 1 and mL/min. Injections the (2 splitless mode with inlet and transfer line temperatures at 250 nologies, Massy, France) was(HP coupled 5972; electronic with impact, a 70with eV; mass a eMV, 2.7 50 spectrometer kV) m and was fitted tartaric acid in 2003.5 mL by of addition distilled of water 1-Madded and 120 aqueous adjusting mL NaOH. the of Then, reagent-grade pH tothe ethanol to and this volume distilled to mix, water 1 were to L. raise 350 mg (2 mM)100 was mL of added the to wine172 model a mg solution. 100-mL Then, (2 brown to mM) this glass of88 mix, bottle DI mg was or added and (2 228 mM) mgwas of (2 added acetaldehyde. mM) until Aqueous of the 1-M 3,4-hexanedionewas final or or closed mixture 0.1-M with was NaOH a at20 PFTE pH screw 3.5, cap. then The the reactions bottle were stored at (St. Quentin Fallavier, France). We purchased 2-ethylthiazole from TCI Europe (Zwijndrecht, Belgium). Dichloromethane, pesticide grade, was purchasedEthanol, from liquid chromatography Carlo grade Erba was(Darmstadt, purchased (Val from Germany). Merck de Cysteine hydrate Reuil,chased hydrochloride from France). was Alpha pur- Aesar (Strasbourg,of France). A C standard mixture Gas chromatography-mass spectrometry (MS) analysis Reactions and sample preparation from 40 to 250 3 (tetrahydropyran 1 mg/L), and1-mL then dichloromethane were during extracted 1 3traction. min times Traces with of with water a were vortex, removed usingaccording for disodium each sulfate to ex- salt, (Vestling andunder a others gentle 1990) stream oference and compounds nitrogen then to were dissolved approximately concentrated directly 0.2 in mL. dichloromethane. Ref- taken after 2 h; after this,surement larger of peaks began the to interfere 2-methylFor with and mea- the 2-ethyl comparative products setthe reported of reaction below. time products chosen, fromIntegrations 1 DI of d, the and total gavepeaks cysteine, more representing ion and the accurate current products integrations. (Mebazaa tabulatedmanually, below and since were others carried most 2009) out werechromatography not after background among subtractionfirm the their was major identities. used products. to Ion con- and the final stepand lasting the 20 solvent min. delay was TheSCAN 5 mode. m/z min. range Detection was was 50 performed in to the 400; MS Chemstation with thedatabase. NIST/EPA/MSDC v.2 Mass Spectral England) provided additional spectraltion information. index Linear (LRI) reten- on BP21 values were calculated using retention tech France (Templemars, France). and 280 C for 2 h gave mostly 2- ◦ Vol. 76, Nr. 6, 2011 r C] (Yaylayan and Haffenden 2003), 13 C-2]-labeled glycine and DI produced 4,5- 13 Journal of Food Science Many of these methylated heterocyclic are found in wine (Marc- The complex group of chemical pathways taken by carbohy- Acetaldehyde derived from cysteine in the Maillard reaction Reagents, solvents, and reference samples were used with- This work does not mimic wine ageing but simplifies it in order C862 Materials and Methods Materials Cysteine reaction with diacetyl under wine-like conditions. . . at the 2-position. While carbonseither 4 from and DI, if 5 both are havedecarboxylation), assumed methyl if to groups, they originate or are from unsubstituted, cysteine acould (after methyl be group supplied at C(2) by either DI or cysteine. hand and others 2000; Almy andrelatively de volatile, Revel 2007). they Because are theytial are considered compounds in among determining the theThat quality more is and influen- one character reason of why wines. thiseration work is pathways centered of on the methylthiazoles studiesmain of under rationale gen- mild for seeking conditions. mechanismsbyproducts The is and to to anticipate understand on howrameters possible the and wine components physicochemical can pa- 5-member influence heterocyclic the generation. 3-N,S and 1–3-N,O drates and aminothe acids Maillard to reaction the hasreactants vast often to number been 2 model explored products compounds. by formedderived a Typically, simplifying carbonyl by from compound the carbohydrateamino degradation acid. is This treated simplificationproducts, with many gives of a a which morefoods single have limited and distinct number beverages. aromas Identification of or2 of tastes goals: these found products to in can predictreactants achieve the and products to of revealthe reaction parallel broader sequences reactions set or with of mechanisms similar wine. Maillard within products in complex mixtures such as (Vernin and Metzger 1981;cas Vernin and and Yaylayan 2004) othersC(2) has 1992; fragment been in Perez postulated the Lo- asand thermal others the degradation 1985). source Cysteine of alone of cysteine at the alone 180 (Shu methylthiazolidine and measurable amounts of 2-ethylthiazolidine, 2-methylthiazole, 2-ethylthiazole, 2-methyl-2-thiazolineother among heterocyclic products (Umano andthermal others 1995). study, cysteine In another alone gavethiazoline, 2-methylthiazole, 2-methyl-2- andexperiment, 2-methylthiazolidine. [ In a related pyrolysis out furthertrimethyloxazole, tetramethylpyrazine, purification. 3,4-hexanedione, tartaric acid, and Sodium sodium bicarbonate were purchased bicarbonate, from Sigma-Aldrich DI, 2,4,5- dimethyloxazole with 15% [ indicating that C(2) was onlyOn partially the supplied other by hand, there the is amino alsooriginates acid. evidence that from the DI. methyl-C(2) unit Pilotymonia and and Baltes (1979) found treated 2,4,5-trimethyloxazolemonia DI (cysteine in with produces am- its am- reactionDI with together DI). with hydrogen And, sulfide and severalthiazoles ammonia authors yielded and showed trimethyl that thiazolines (Takkenothers and 1990). others Thus, 1976;tual Vestling in source and of the themethylthiazoles, cysteine methyl-C(2) and their unit and reduced in counterparts DI the are 2-methyloxazoles, unclear. reaction, 2- theto examine ac- the role of DIof associated with odorous cysteine products in by the formation chemicalmechanistic means. interpretations In particular, of itlike these proposes conditions; reactions they under haveand mild implications character wine- during in ageing. terms of wine quality C: Food Chemistry bandfo netato ytiead34hxndoereaction 3,4-hexanedione and mixafter2h. cysteine of extract an from obtained mixture alkane (1965). C7-C23 Kovats a to of according injection independent an times from retention taken using calculated were indexes Kovats The mixture. eeec opud r itdi al .Fr4,5-diethyl-2- For of 1. condensation Table simple in 2,4,5-triethyloxazole, and listed methyloxazole are compounds reference opud rfrne Uao19][oas16][reference] 1965] [Kovats 1995] [Umano [reference] Compounds -tytizldn ytaieadpoaa H 98 558(0) 02) 61) 11) 117(14) 61(13); 56(18); 70(25); 88(100); 1515 ammonia, dimer, 1998] Mercaptoacetaldehyde [Ho propanal and Cysteamine 2-methyl-3-thiazoline 2-ethylthiazolidine 1 standard, Table Internal (2) 3,4-hexanedione; tetr (1) (14) acid; mix. propionic (13) 2-propionylthiazole. reaction propionin; (16) (12) 3,4-hexanedione 2-acetylthiazole; acid; and (15) acetic (11) cysteine 2-ethyl-3-thiazoline; 2,4,5-tri (10) 2-h (7) 2-methyl-3-thiazoline; 2-ethylthiazole; a (9) (6) 2-methyl-3-thiazoline; 2-methyl-2-thiazoline; of (5) extract 4,5-diethyl-2-methyloxazole; (4) an 2-methylthiazole; (3) from Tetrahydrothiopyran; chromatogram ion 1–Total Figure compounds Detected Discussion and Results C a of injection independent an from taken times . . conditions. wine-like under diacetyl with reaction Cysteine ,,-rehlxzl D NH HD, 2,4,5-triethyloxazole -ty--haoieMrataeadhd ie,ammonia, dimer, Mercaptoacetaldehyde 2-ethyl-3-thiazoline -ty--haoieMrataeadhd ie,ammonia, dimer, Mercaptoacetaldehyde 2-ethyl-3-thiazoline thiazoline 2-acetyl-2-methyl-3- methyloxazole 4,5-diethyl-2- neapeo hoaormi rsne nFgr .I was It 1. Figure in presented is chromatogram of example An hoaorpi n Sdt o neednl prepared independently for data MS and Chromatographic – eeto nee n assetao neednl yteie eeec compounds. reference synthesized independently of spectra mass and indexes Retention ecpoctleyedmr ammonia, dimer, Mercaptoacetaldehyde D NH HD, 1997] 1997] [Elmore n ctleye[sne 1958] [Asinger acetaldehyde and n rpnl[sne 1958] [Asinger propanal and n rpnl[sne 1958] [Asinger propanal and n rpnl[sne 1958] [Asinger propanal and 4 4 H n rpnl[Elmore propanal and OH, dimer acetaldehyde and OH, yteie rm R oasIdxsmz(eaieaudne%) abundance (relative m/z Indexes Kovats LRI from: Synthesized 7 -C 23 alkane 3212 3(-5 0) 5(,2) 62) 83(25); 56(25); 25); 153(M, 100); 138(M-15, 42(24); 83(30); 30); 139(M, 100); 124(M-15, 1325 1279 1322 1276 3115 0(,10;8(-5 5;5(3;68(36); 59(53); 95); 86(M-15, 100); 101(M, 1354 1351 3115 0(,10;8(-5 5;5(3;68(36); 59(53); 95); 86(M-15, 100); 101(M, 1354 1351 4514 6M2,10;15M 3;5(8;54(21); 59(38); 53); 115(M, 100); 86(M-29, 1441 1435 881010;5(0;114) 32) 4 (M,1) 143 43(25); 101(40); 59(90); 100(100); 1878 ,-eaein,NH 3,4-hexanedione, / 5famn,poal C probably fragment, 55 m/z a the ion of fragmentation a alkyl-C(2) present m/z both the for prod- spectra form Mass to product. desired oxazole acetaldehyde 3,4-hexanedione with the that competing indicating of was formed, 2,4,5- yields also 4,5-diethyl-2-methyloxazole, was of low triethyloxazole formation the gave Mottram in and and reactions ucts, (Elmore Both preparations 1997). unambiguous their assured opuds hti twstesple fteaklC2 nt the unit, alkyl-C(2) the of supplier the was it if that was so DI the compound which “marked” in This in reaction 3,4-hexanedione. parallel unit by a methyl-C(2) substituted out the of carried we origins products, the these reveal To data. LRI published MS with and/or agreement satisfactory by made were products = -5a h aein h aeinudrosfrhrls of loss further undergoes ion base The ion. base the as M-15 o.7,N.6 2011 6, Nr. 76, Vol. 4 H n ihraeadhd rpropanal or acetaldehyde either and OH, NS 1992] [NIST Emr 1997] [Elmore 124(5) 96(6); 41(18); 69(20); 55(23); 1997] [Elmore 68(5) 43(15); 110(3); 125(8); 57(12); 82(12); 41(15); 69(20); 55(24); Emr 1997] [Elmore 45(16) 42(18); 58(18); 74(22); 55(27); Emr 1997] [Elmore 45(16) 42(18); 58(18); 74(22); 55(27); Emr 1997] [Elmore 45(10) 58(11); 41(15); 82(15); 68(15); Aigr1958] [Asinger 2 H r 5 ora fFo Science Food of Journal N dnictoso l other all of Identifications CN. tyoaoe (8) ethyloxazole; aethylpyrazine; α -dicarbonyl C863

C: Food Chemistry 002 70 10 80 . . . . 0 < 4008 1 002 0 70 1 . . . . Ratio of 2-ethyl/ 0 2-methyl abundances < 90 100 100100 100 53 Intensity in Intensity in + + + 88 5886 1 0 Ion CH] m/z 2 h 1 d M–15 0 = Ion the 2-ethyl the 2-methyl m/z Fragment spectrum (%) spectrum (%) S N S N S H N N O Properties of the ions chosen for integration of 2-ethyl or Comparison of peak areas of 2-ethyl compared with 2- – S N S N S H N N O Confirmatory data were obtained by integration of mass chro- For the 2,4,5-triethyloxazole and 2-methyl-4,5-diethyloxazole Four main points emerge from the product data for the Ion chromatograms were used to locate each product exactly. 5-diethyloxazole 2-alkylthiazolidine2-alkyl-4, 88 [C3H6NS] M-15 100 100 2-alkylthiazole 58 [S-CH 2-alkyl-3-thiazoline 86 [C3H4NS] matograms also shown in Table 4.to The both ions ethyl chosen and were common methylucts, substituent and as free determined ofmain interfering by ions. prod- examination As thesumed in equal peak the response factors shapes TIC forTable of integrations, the 3 ethyl gives the several and the methyl ion ions substituent. chosen integrations for as- each alkylatedhomologues, product. the ions chosen wereresenting 124 and methyl 138, loss. respectively, rep- Sinceis the loss insignificant, of one ethylmethyl can from from either conclude 2-methyl-4,5-diethyloxazole product is thatthe from the 5-ethyl the predominant group. 4-ethyl As loss or could a be of corrected result, to the reflect2-ethyl ethyl/methyl the product peak 3 as possible area compared methyl ratios methyl losses to product. from the the Depending 2 on possible theethyl groups losses tendency at from of the the positions fragmentation of of 2- 4) oxazole, the a calculated maximum ratios correction (Table would33% reduce and the ethyl/methyl the ratio ratio by shown would be adjusted fromreaction 0.7 to 0.5. between 3,4-hexanedione and cysteine: (1) more Table 3– 2-methyl products. Table 4 methyl heterocycles produced3,4-hexanedione and over cysteine time (thetract by ion peak chromatograms). areas the are reaction obtained from between ex- in Table 4. Product ratiosincluded for because other the 2-alkyl areas heterocyclicboth) for were were not either not the large 2-methyl enough or to provide 2-ethyl reliable (or ratios. Then, for the 2-h sample,ternating integrations between the were peaks measured for 5all each times ethyl cases, and al- the methyl standard product. deviation In Overlap was from less other than peaks 15% prevented of comparisons1-day each data. for average. much of the Comparison of DI and 3,4-hexanedione reactions with cysteine DI + ∗ HD Cysteine + 0.01 nd < Relative area Relative area Cysteine Vol. 76, Nr. 6, 2011 r Relative peak areas of products of the reaction between – Journal of Food Science not detected. = Table 2 provides partial lists of products identified from the For the 3,4-hexanedione reaction with cysteine, peak ar- 2,4,5-trimethyl-4,5-diethyl-2-methyl-2,4,5-triethyl-2-acetyl-2-propionyl-2-ethyl-2-propionyl-2-acetyl-2-methyl- 4.62-ethyl-2-propionyl- nd2-acetyl-2-methyl-2-ethyl-2-propionyl- 3.52-acetyl-2-methyl- nd 362-(1-hydroxypropyl)-2,4,5-triethyl- 0.68 0.5 1.12-(1-hydroxyethyl)-2,4,5-trimethyl- nd 70 nd tetramethyl- 296tetraethyl- nd nd nd 36Acetic acid nd nd Propionic acid 10 ndAcetoin nd Propioin nd 3 8 nd 10 nd 122 94 36 20 nd nd 36 nd 12 5 nd 2-methyl-2-ethyl-2-methyl-2-ethyl-2-methyl-2-ethyl-2-methyl-2-ethyl- 1.0 1.5 0.03 0.03 1.0 7.3 nd 0.2 0.04 0.6 nd 0.88 nd 0.005 nd ∗ 2,4,5-Alkyloxazole 2-carboxylthiazole 2-Alkyl-2-carboxyl-3-thiazoline 2-Alkyl-2-carboxylthiazolidine 2-Alkyl-2-carboxyl-3-thiazoline 2-(1-hydroxyalkyl)-2,4,5-Alkyloxazoline Alkylpyrazines Carbonyls degradation products 2-Alkylthiazole 2-Alkyl-2-thiazoline 2-Alkyl-3-thiazoline 2-Alkylthiazolidine C864 Table 2 cysteine and eitherlike 3,4-hexanedione (HD) model or solutions),reaction. diacetyl after (DI) (in 2 wine h for HD reaction and 1 d for DI Cysteine reaction with diacetyl under wine-like conditions. . . product would have an ethyl, notin a these methyl group products. at the 2-position cysteine reaction with DIconditions and modeling with wine ageing. 3,4-hexanedione These under listsfor are mild each organized 2-methyl so that orand 2-ethyl product cysteine set reaction, of thecysteine the reaction corresponding 3,4-hexanedione appears product alongside. Peak of areasmethylthiazole are the in referenced each to DI 2- reaction. and Inreaction 3,4-hexanedione mix, and tetraethylpyrazine cysteine ischromatogram, one of yet the tetramethylpyrazine largestdemonstrates was peaks that in not there the was detected. nocrossover. significant As This 3,4-hexanedione expected, to DI tetramethylpyrazinepeaks is of one the DI of and the cysteine largest reaction products. eas’ measurements were carriedthiazoles, out 3-thiazolines, for thiazolidines,Ratios 2-methyl- and of and 4,5-dimethyloxazoles. peak areas, 2-ethyl- obtained from2-ethyl extract compared ion with chromatograms, 2-methyl of heterocyclic producedwere over monitored. time These ratiosor reveal the dicarbonyls contribution or both of to cysteine 2-alkyl residues. The results are shown C: Food Chemistry en nspligtemty-()ui f2mtytizl,2- 2-methylthiazole, in of role DI’s unit while 2,4,5-trimethyloxazole methyl-C(2) and methyl-3-thiazoline, the supplying in teine compar- 4,5-dialkyloxazole in and products. thiazolidine, different 3-thiazoline, are thiazole, ratios than the ratios 2-ethyl/2-methyl important ing product the less are that Other details fact 2-methylthiazole. the the over but affected, 1 be to similarly 2-ethylthiazole would 8 the for about preference a be require rate would to a 1 to with to combined 2-ethylthiazole 0.2 1 of to the ratio 1.5 and Thus, about of 3,4-hexanedione 1956). ratio area of Gettler 2-methylthiazole place and in (Fitzpatrick acetone DI and 3-pentanone and aebe rsn.Had present. been have rmD ihMdu neetdMedium Medium Undetected Dominant Medium Medium Trimethyloxazole 2Methylthiazolidine High Low 2-Methyl-3-thiazoline 2-Methylthiazole cysteine From DI From of group 2-methyl to Contribution 5 Table of employment the and mon, is 2,4,5-triethyloxazole. 2-methyl-4,5-diethyloxazole of of that amounts twice approximately the was (4) 2-ethylthiazolidine detected; but not produced (3) was 2-ethyl-3-thiazoline; 2-methylthiazolidine than 2- produced more was (2) methyl-3-thiazoline 2-methylthiazole; than produced was 2-ethylthiazole . . conditions. wine-like under diacetyl with reaction Cysteine on rmltrtr aawt oprberaet:hydroxy- reagents: comparable with (pKa(NH data lamine literature from found cysteine. was far by reaction the contributed but 3,4-hexanedione essentially Thus, detected the found. in not is found is 2-methylthiazolidine 2-methylthiazole products 2-ethylthiazolidine than 2-methyl fact, 2-ethylthiazole the In amounts more out. for significant ruled responsible that is possibility are 3,4- of the 3,4-hexanedione which reasoning, intermediate of in same common propionaldehyde). path the and a a acetaldehyde By as out form (such intermediates rules to of pair this modified and is uniform, hexanedione not are tios of purity isotopic the to M-1 cysteine. subject labeled the addition, the in as and, detected significant unlabelled is been of ion have absence not the would but dominated, 2-methylthiazolidine have would ion parent Iadcsen o h C(2)CH the for between competition cysteine a and showing DI generally result a produced have n eylwcnrbto fD ofr h aefragment same peaks the the form of to spectra the DI mass representing of the However, contribution 2-methylthiazolidine. low in very a and euti hc %of from 1% indistinguishable which of essentially in 3.5% been (about result have 103 a would m/z for ion) intensity base the the of intensity, 4.4% this be But to product. only unlabeled calculated the 104 represented of m/z have abundance ion, natural in would parent the produced 2-methylthiazolidine the that example, for for as intensity DI, accuracy labeled the Using with 4. Table ratios the measured have nterato fD ihcsen,D optswt cys- with competes DI cysteine, with DI of reaction the In h s fiooe otaeteoiiso rdcsi eycom- very is products of origins the trace to isotopes of use The og ae34hxndoeD ai faot02cnbe can 0.2 about of ratio 3,4-hexanedione/DI rate rough A ra- product 2-ethyl/2-methyl homologous the for areas Peak – otiuino Io ytiet -ehlrsde f5mmee heterocycles. 5-membered of residues 2-methyl to cysteine or DI of Contribution 2 13 ) -aee n naee itrswudnot would mixtures unlabeled and C-labeled = .)i lc fcsen (pKa(NH cysteine of place in 6.0) 13 13 -aee ytiebe sd h 104 the used, been cysteine C-labeled -aee -ehlhaoiiemight 2-methylthiazolidine C-labeled 13 -aee Io ytiewould cysteine or DI C-labeled 3 rgeti 2-methylthiazole in fragment S N 2 ) = 8.3) N S otoso hsCH this 2-methyl-3- of 2-methylthiazole, proportions increasing series among supplies the cysteine uniform 2-methylthiazolidine; in not thiazoline, and is products competition This cysteine these undetectable. with is compared 2-methylthiazolidine DI in unit this supplying ino ,-eaein ihcsen ie neetbeamounts undetectable gives cysteine with 3,4-hexanedione reac- of the tion since 2-methylthiazolidine of formation the exclude must to oxidations irreversible in successive shown are rings. also aromatic steps form 2-methylthiazole, final The form 2. and to Figure 2-methyl-3-thiazoline 2-methyl- steps oxidize form similar can to further then undergo acetoin can intermediate and DI water molecule This 3-thiazoline. of 2nd (2) loss a intermediate. concerted of form undergo orientation to water the allowing of can which or ion in to DI hydroxide regenerated tautomerize add 2, and either 2-methylthiazolidine Figure can form in (1) eventually displayed intermediate immonium-enol are the pathways and to combined (Griffith 1-pyrroline. by used to The proline decarboxylation oxidative been catalyzed its of have and is 1989) decarboxylation ions Hammond analogous it Immonium the reaction. that Strecker explain suggests a conditions via ring DI mild and under by occurs decarboxylation condensation followed That 1991). acid Shibamoto simple and (Yeo carboxylic decarboxylation reduction. involves 2-methyl-5-thiazolidine to nor route closure oxidation simplest neither The requires methylthiazolidine in 2-methylthiazolidine primarily 2-methyl-3- in oxidized). all found in least at (the is less contribution no unit with with and acetyl oxidized) thiazoline DI’s most the 2-methylthiazole. (the of hand, 2-methylthiazole formation other the the in DI) On to mi- (compared a role plays acetaldehyde nor 2-methyl-3-thiazoline; to of in cysteine formation DI with the with combines competes acetaldehyde DI) 2-methylthiazolidine; single produce not a of (and modifications Acetaldehyde from take products not product. these and pathways of different formation of by the of place that units suggests contributions cysteine methyl-C(2) and differing the DI from that formed 2- fact methylthazolidine—are and acetaldehyde the 2-methyl-3-thiazoline, products—2-methylthiazole, of And 3 contribution the DI. and significant from 3,4-hexanedione any between formed 3-ethylthiazolidine out reaction that rules the fact cysteine cysteine in The from undetected products. methyl-C(2) was formed heterocyclic the product the becoming some of in and compete acetaldehyde) DI (probably from unit acetyl rpsto fmcaitcpathways mechanistic of Proposition al 5. Table PathsinwhichDIisthesourceofthemethyl-C(2)fragment 2- to cysteine and acetaldehyde from path The the that suggests this view, of point mechanistic a From o.7,N.6 2011 6, Nr. 76, Vol. 3 C2 nt hs eut r yteie in synthesized are results These unit. -C(2) N H S r ora fFo Science Food of Journal O N C865

C: Food Chemistry H+ O S N 2MT 1.0 S H N 2MD S 0.005 +N - AcOH OH O H O 2 OH H O - DI 2 H - acetoin S O Not favored +N H OH O H S S N +N H O S O +N HO O H+ ~ H+ ~ H+ OH S +N OH O 2 S H +N i S +N O - DI S H N OH O O O 2 2 2 S N OH H H OH - H+ S H N - CO 2MD NOT found HO S N H O carbonyl group brought by DIaction (a of feature acetaldehyde) absent and in theFigure the 3 normal parallel suggests Strecker a re- path pathway to is3-thiazoline explain followed. and the formation 2-methylthiazole of but 2-methyl- not 2-methylthiazolidine. OH OH O H 2 HO S N H OH - H+ O N S O 2 HS +N O OH H O S + +N H+ O 2 OH OH O H 2 ii +DI + H - H + - H - acetoin S N - H N O HS O H OH OH O S N CO fragment. Decarboxyla- S N 3 2M3T S N H Vol. 76, Nr. 6, 2011 S O r +N H - acetoin COOH and a CH O O 2 O 2 2 2 H N S N - H O - H HS 2MT oxidation DI -acetoin + 0.88 S O N H+ HO 2MD3T O - AcOH N H 2 HS H DI + S N S S N +N 2M2T HO Journal of Food Science O OH H O cysteine DI + Figure 3–Formation of 2-methyl-3-thiazolinesource (2M3T) of and methyl-C(2). 2-methylthiazole (2MT) but not 2-methylthiazolidine (2MD) using cysteine and DI asC866 the Figure 2–Formation of 2-methylthiazolidinemethyl-C(2). (2MD), 2-methyl-3-thiazoline (2M3T), and 2-methylthiazole (2MT) using acetaldehyde as the supply of Cysteine reaction with diacetyl under wine-like conditions. . . of 3-ethylthiazolidine. The initial Strecker imine willclosure undergo and ring loss of CO tion is likely toMoehlenkamp occur 1983). 1st The as decarboxylation is this assisted step by is the relatively extra fast (Hanna and C: Food Chemistry ehlgop nC4 n ()ipyta hyaederived are 2- both they reaction, that cysteine imply and 3,4-hexanedione C(5) the and In DI. C(4) from on groups path. methyl Strecker the via cysteine from hand, compared other formed negligible acetaldehyde the be the On would to data. acetaldehyde the this of with 2-methylthiazolidine, amounts consistent small form not to is route cysteine this with but react separate can formed way that acetaldehyde this Any those problematic. in fragment, are methyl-C(2) DI the from acetaldehyde furnish to 3. DI Figure in ing proposed is acid car- acetic the of of loss hydration by followed of to bonyl sequence 2-acetyl-2-methyl-3-thiazoline a from 2-methyl-3-thiazoline, group, form acetyl the the For of 2-acetyl-2-methyl-3-thiazoline. shown to loss eno- was leads the that as with step compete route lization not this does and digression by this group Apparently However, formed above. acetyl 2-methylthiazolidine. not the of is of formation 2-methylthiazolidine eventual carbonyl the the with involves from It arrow. formation dashed 2-acetyl- a hydrate from with formed indicated ion is immonium possible 2-methylthiazolidine initial A the above. from given 2- digression properties satisfactory to chromatographic with and successful, leads was spectral 2-methylthiazole products 2-acetyl-2-methyl- the for that among search way A 3-thiazoline with 2. same product Figure the in this methyl-3-thiazoline of in a Oxidation affords 2-acetyl-2-methylthiazolidine, 2). DI (Table form found to product decarboxylation closure major by ring followed is then cysteine and and DI of methyl-C(2). of Condensation source the as acetaldehyde or cysteine using (TMO) 2,4,5-trimethyloxazole to pathways 4–Proposed Figure . . conditions. wine-like under diacetyl with reaction Cysteine ihrgr otefraino ,,-rmtyoaoe the 2,4,5-trimethyloxazole, of formation the to regard With us- 2-methyl-3-thiazoline form could that paths considering In From acetaldehyde and 3-amino-2-bu F rom cysteine:rom H HO HO 2 N HS O momooxime: acetaldehydeFrom anddiacetyl O N dehyde acetal- O + HN O HS HO N HO N tanone fromtheStrecker reaction: O N HO O N H HS N O O N H DI H HS +N O +N O O DI OH H u nycsen upisti rgett 2-methylthiazolidine. to fragment cysteine. this and supplies DI cysteine both only from But formed is 2,4,5- are products and trimethyloxazole results 2-methyl-3-thiazoline, 2-methylthiazole, methyl-C(2) These The of cysteine. 2-ethylthiazoline and fragment 2-methylthiazoline. DI between of reaction of the to amounts applied presence detectable the no despite 2-methyl- and than 2-ethyl-3-thiazoline 3-thiazoline, 2-ethylthiazole less more 2-methylthiazole, produced than cysteine and 3,4-hexanedione of gu ufrproducts. sulfur ogous n noiaini eesr o ,,-rmtyoaoe Both 2,4,5-trimethyloxazole. H for for opportunity necessary ample provide is paths oxidation an ammonia 2,4,5-trimethyl-3-oxazoline,and used, formed, and is is product 3-amino-2-butanone DI observed If another from 2003). Haffenden formed and (Yaylayan monooxime) 4. (DI Figure 3-imino- or 2-butanone DI, in of amine shown Strecker the 3-amino-2-butanone, are with either cysteine source from derived acetaldehyde methyl-C(2) include possibilities the These as us- cysteine 2,4,5-trimethyloxazole 2-ethyl consistent ing to obtain their routes Possible to nor ratios. sufﬁciently ethyl/methyl quantiﬁed they correspond- be neither could produced counterparts but also 2,4,5-trimethyltiazole, cysteine and and 2,4,5-trimethyl-3-thiazoline both and DI from 2,4,5-trimethyl-3-oxazoline ing cysteine. originates fragment and methyl-C(2) DI the that formed, indicating are 2,4,5-triethyloxazole and methyl-4,5-diethyloxazole Conclusion fe n1%ehnli ae ufrda H35 mixture a 3.5, pH at buffered water in ethanol 12% in h 2 After O HS N OH - H +N 2 O +N S O O OH o.7,N.6 2011 6, Nr. 76, Vol. H +N 2 O O O HO HO O N - H H OH 2 r O ora fFo Science Food of Journal H - acetoin 2 O xhnet fodanal- afford to exchange S O N O -DI TMO O N C867

C: Food Chemistry C- ). J 13 -dicarbonyl ´ equences bi- α Salmo salar ´ erocycles. Bull Soc Chim Belg ´ et ˆ omes: les h ´ erale des acides amines sur les sucres: ses conc ´ en ´ eaction g ´ e de Biologie 72:599–601. ´ et 2]-labelled glycine and alanine model systems. Food Chem 81:403–9. glucose/L-alanine maillard model systems. J Agric Food Chem 47(8):3280–4. different pH conditions. J Agric Food Chem 39:370–3. acid with L-cysteine. Food Chem 121:1060–5. quantification in must and wine. Anal Chim Acta 660:158–63. compounds from reactions between cysteineChem. and 48:4890–5. carbonyl compounds in wine. J Agric Food of cysteine in aroma production in wine. J AgricB, Food Camel Chem V. 2009. 50:6160–4. CharacterisationChem of 115:1326–36. volatile compounds in Tunisian fenugreek seeds. Food amino acids onAgric the Food volatile Chem and 55:1427–36. nonvolatile components of cookedpolyphenols in salmon a ( wine-model medium: impactActa. of 660:102–9. oxygen, iron, and sulfur dioxide. Anal Chim furan — a food toxicant. J Agric Food ChemLebensm. 52: Unters 6830–6. Forsch 168:374–80. method for the measurement ofapplications. free J amino Sci acids Food Agric including cysteine 81:731–8. in musts anddegradation wines: of first cystine in water. J Agric Food Chemodorant 33:438–42. of the typical aroma of oxidative aged portodour wine. precursors J in Agric the Food flavour Chem. and 51:4356–63. fragrance industry. Flavnitrogen Frag compounds J in 23:369–81. food flavors. ACS symposium series 26:114–21. of aged champagne wines. J Agric Food Chem 51:1016–20. the headspace of a43:2212–8. heated d-glucose/l-cysteine maillard model system. J Agric Food Chem 90:553–83. Kinetics and thermal-degradationGC-MC/specma data-bank of identification230:15–29. the of fructose volatile methionine aroma compounds. amadori Carbohydrperformance intermediates liquid Res chromatography – and mass spectrometry. Anal Chem 62:2391–4. ologiques. Soci oil heated with cysteine and trimethylamine oxide. J Agricstructurally Food related Chem thiazines 46:5232–7. inChem 46:664–7. a cysteamine/2,3-butanedione model system. J Agriccompounds. J Food Agric Food Chem 57:8591–7. index system. Adv Chromatogr 1:229–47. wines supplemented with grapewine flavour precursors ageing. from Food different Chem varietals 120:205–16. during accelerated Yaylayan VA, Keyhani A. 1999.Yeo HCH, Origin Shibamoto T. 1991. of Microwave-induced volatiles of 2,3-pentanedione the Maillard model and system under 2,3-butanedioneYu AN, in Zhang AD. D- 2010. Aroma compounds generated from thermal reaction of L-ascorbic Marchand S, de Revel G. 2009. A HPLC fluorescence-based methodMarchand for glutathione S, derivatives de Revel G, Bertrand A. 2000.Marchand Approaches S, to de Revel wine G, Vercauteren J, aroma: Bertrand release A. 2002. ofMebazaa Possible mechanism R, aroma for Mahmoudi involvement A, Fouchet M, Santos MD, Kamissoko F, Nafti A,Methven Cheikh L, RB, Tsoukka Rega M, Oruna-Concha MJ, Parker JK, Mottram DS. 2007. Influence of sulfur Nikolantonaki M, Chichuc I, Teissedre PL, Darriet P. 2010. Reactivity ofPerez volatile Locas thiols C, with Yaylayan VA. 2004. Origin and mechanistic pathways ofPiloty formation V, of the Baltes parent W. 1979. VolatilePripis-Nicolau products L, in de the Revel G, reaction Marchand of S, Anocibar amino A, acids Bertrand with A. 2001. diacetyl. AutomatedShu Z HPLC CK, Hagedorn ML, Mookherjee BD, Ho CT.Silva Ferreira 1985. AC, Barbe Volatile J-C, components Bertrand of A. 2003. the 3-Hydroxy-4,5-dimethyl-2(5H)-furanone: a thermal key Starkenmann C, Troccaz M, Howell K. 2008. The role ofTakken HK, cysteine van and der cysteine-S Linde conjugates LM, as de Valois PJ, DortTominaga T, HM, Guimbertau Boelens G, M. Dubourdieu 1976. D. 2003. Phenolic, Role sulfur of and certain volatileUmano thiols in K, the Hagi bouquet Y, Nakahara K, Shyoji A, Shibamoto T. 1995.Vernin Volatile G, chemicals formed Metzger in J. 1981. La chimie des ar Vernin G, Metzger J, Boniface C, Murello MH, Siouffi A,Vestling Larice M, Murphy JL, C, Parkanyi Fenselau C. C. 1990. 1992. RecognitionYaylayan VA, of Haffenden trypsin LJW. autolysis 2003. products Mechanism by of high- imidazole and oxazole formation in [ Horiuchi M, Umano K, Shibamoto T. 1998. AnalysisHuang of TC, volatile compounds Su formed YM, from Ho fish CT. 1998. Mechanistic studies on theJenny formation G, of Glomb thiazolidine MA. and 2009. Degradation ofKovats E. glucose: reinvestigation 1965. of Gas reactive chromatographic characterization ofLoscos N, organic Hernandez-Orte P, substances Cacho in J, Ferreira the V. 2010. retention Evolution of the aroma composition of Maillard LC. 1912. R Vol. 76, Nr. 6, 2011 r -dicarbonyls to form some lipid-Maillard interaction products α -hydroxyketones or α Journal of Food Science A number of questions are raised by the results in this work. The primary contribution of this work is to propose new path- J. A. gratefully acknowledges support form a Fulbright diacetyl and cysteine. Proceedings of the 9thSeptember international 1–5, symposium on 2007. the Munchen, Maillard reaction. Germany. and beyond. Int J Food Microbiol 96:235–52. Maillard reaction of Thiamine, Cysteine, and Xylose. J Agricvolatile Food compounds Chem during 56:10679–82. the aging46. of sweet fortified wines. J Agric Foodof Chem some 47:2837– volatile compounds inprocess. J white Agric fortified Food wines Chem (Vins 48:2340–5. Doux Naturels) duringIdentification the of aging wine aldehydes. In: Maarseresearch. H. Pays-Bas: and Elsevier. van p der Heij 353–61. DG editors. Trends in flavour des, and found in cooked beef. J Agric Food Chem 45:3595–602. of formation of several oximes. J Am Chem Socprincipal 78:530–6. carbonyl compoundsmicroextraction involved in and malolactic positive fermentation40:1558–64. ion of chemical wine by ionization solid-phase GC/MS analysis.of J amino-acids with Mass carbonyl-compounds. J Spectrom Dairy Sci 72:604–13. anisms of the oxidation ofand N-benzyliminodiacetic acidic acid sulfate with media. cerium(IV) J in Organic acidic Chem perchlorate 48:826–32. precursor systems in the formation3-furanthiol. of J the Agric intense food Food odorants Chem 2-furfurylthiol 46:235–41. and 2-methyl- C868 References Almy J, de Revel G. 2007. Approaches to wine aroma: C1 transfer during theBartowsky reaction EJ, between Henschke PA. 2004. The buttery’ attribute ofCerny wine C, diacetyl Guntz-Dubini desirability, R. spoilage 2008. Identification of 5-Hydroxy-3-mercapto-2-pentanone in the Cutzach I, Chatonnet P, Dubourdieu D. 1999. Study of theCutzach formation I, Chatonnet mechanisms P, Dubourdieu of D. some 2000. Influence of storage conditions on the formation de Revel G, Bertrand A. 1993. Dicarbonyl compounds and their reductionElmore products JS, Mottram in DS. wine. 1997. Investigation of the reaction between ammonium sulfide, aldehy- Acknowledgment Cysteine reaction with diacetyl under wine-like conditions. . . These findings are interpretedsistent in with terms the of relatively pathways mild that conditions are employed. con- Because this studyneed relates to be to examined. wine, Amongto these, a modify polyphenols has number mercaptans been equilibriumothers of shown 2010). in new In wines addition, variables (Nikolantonakiinfluenced the by role and the of reduced/oxidized DI glutathione asbe couple present oxidant known in is to musts likely and to wines be (Marchand and deways Revel in 2009). the generation ofwine-like odorous physicochemical heterocyclic compounds conditions. under These pathwaysto permit anticipate us the presence ofcompounds several in odorous wines (probably and nutty tothe like) propose strategies complex to wine quantify matrix. them in Aquitaine ResearchCommission and Award the Council throughScholars. for the the International Exchange Franco of American Fitzpatrick FW, Gettler JD. 1956. Kinetics of oxime formation;Flamini temperature R, coefficients of Dalla rate Vedova A, Panighel A, Perchiazzi N, Ongarato S. 2005. Monitoring ofGriffith the R, Hammond EG. 1989. Generation of Swiss cheese flavorHanna components SB, by Moehlenkamp the ME. reaction 1983. Metal ion oxidative decarboxylations. Kinetics and mech- Hofmann T, Schieberle P. 1998. Quantitative model studies on the effectiveness of different C: Food Chemistry View Online / Journal Homepage / Table of Contents for this issue Chem Soc Rev Dynamic Article Links

Cite this: Chem. Soc. Rev., 2012, 41, 4140–4149

www.rsc.org/csr TUTORIAL REVIEW

Flavour chemistry of methylglyoxal and glyoxal

Yu Wanga and Chi-Tang Ho*b

Received 30th January 2012 DOI: 10.1039/c2cs35025d

Methylglyoxal (MGO) and glyoxal (GO), known as reactive carbonyl species, can be generated endogenously and exogenously (human body and food system). They are attracting increased attention because of their relationship with diabetes and flavour generation. In this review, their characteristics relating to flavour chemistry are discussed. MGO and GO can be detected in food systems by GC and HPLC after derivatization. MGO and GO formed in the Maillard reaction play important roles as precursors of aroma and colour compounds, especially in Strecker degradation, a major flavour generation reaction. When combined with amino acids they undergo Schiff base formation, decarboxylation and a-aminoketone condensation leading to heterocyclic aroma compounds such as pyrazines, pyrroles and pyridines. They attack amine groups in amino acids, peptides and proteins to form advanced glycation end products (AGEs) and cause carbonyl stress followed by oxidative stress and tissue damage. Therefore, many studies about scavengers of MGO and GO are seen. The influence of these scavengers on flavour generation is also discussed.

1. Introduction leading to heterocyclic aroma compounds such as pyrazines, pyrroles and pyridines. In other words, MGO and GO, that Reactive carbonyl species (RCS) such as glyoxal (GO) and contribute to flavour generation in food, can induce diabetes methylglyoxal (MGO), that can be generated endogenously and diabetes complications. How to evaluate their paradoxical and exogenously, are attracting increased attention because of roles becomes very interesting and challenging. Therefore, in Downloaded by FAC DE QUIMICA on 16 September 2012 their relationship with diabetes and flavour generation. In vivo this tutorial review, flavour chemistry of MGO and GO is Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D MGO is primarily formed during glycolysis in cells and discussed to provide an overview of controlling their formation generated from the metabolism of ketone body degradation in food. of threonine, and by the fragmentation of triosephosphates.1 In vitro, particularly in the Maillard reaction, RCS can be generated from Schiff’s base and Amadori compounds. 2. Chemical properties of methylglyoxal and A carbonyl group of RCS can attack an amine group in glyoxal amino acids, peptides or proteins to form advanced glycation end products (AGEs) and cause carbonyl stress, followed by MGO, a yellow hygroscopic liquid, is known as 2-oxopropanal, oxidative stress and tissue damage.2 Increasing evidence in pyruvaldehyde, or 2-ketopropionaldehyde. It is present in three both clinical and pre-clinical studies shows that MGO is rapidly equilibrium forms in aqueous solution; monohydrate is the associated with hyperglycemia in both Type I and Type II highest (71%) followed by dihydrate (28%), and the anhy- 4 diabetes and diabetes-related complications such as nephropathy, drated form is only about 1%. Depending on temperature Alzheimer’s disease and cataracts.3 Therefore, RCS are inducing and water content, MGO can change from a less reactive factors of diabetes or its complications. However, RCS that noncarbonyl form to more reactive carbonyl and dicarbonyl 5 are formed in Maillard reaction play important roles as forms. GO, which is also a yellow coloured liquid, is the precursors of aroma and colour compounds, especially in Strecker smallest dialdehyde. degradation, a major flavour generation reaction. MGO and GO, in combination with amino acids, undergo Schiff’s base 3. Analytical methods for quantification of MGO formation, decarboxylation and a-aminoketone condensation, and GO A derivatization process is needed, prior to chromatographic a Chair of Food Chemistry and Molecular Sensory Science, analysis, to quantify MGO or GO. Several derivatization agents Lise-Meitner-Straße 34, Technische Universita¨tMu¨nchen, listed in Table 1 and their adduct products shown in Fig. 1, D-85354 Freising, Germany b Department of Food Science, Rutgers University, New Brunswick, including diamino derivatives of benzene and naphthalene, react NJ 08901, USA. E-mail: [email protected] with MGO to form quinoxalines. Quinoxalines have been analysed

4140 Chem. Soc. Rev., 2012, 41, 4140–4149 This journal is c The Royal Society of Chemistry 2012 View Online

Table 1 Some examples of derivatization methods for methylglyoxal (MGO) analysis

Derivatization reagent Derivatives products Detector HPLC 6-Hydroxy-2,4,5-triaminopyrimidine Pteridine UV Cysteamine 2-Acetylthiazolidine UV Meso-stilbenediamine 2,3-Diphenyl-5-methyl-2,3-dihydropyrazine UV GC 1,2-Diaminobenzene Quinoxaline MS/SIM, NPD O-(2,3,4,5,6-pentaﬂuorobenzyl)- Oxime MS/SIM, ECD, NPD, FPD hydroxylamine-hydrochloride (PFBHA) a Source: Ref. 1. Note: ECD, electron-capture detector; NPD, nitrogen phosphorus detector; FPD, ﬂame photometric detector.

4. Generation of MGO and GO in foods MGO can be exogenously generated from sugar autoxidation, Maillard reaction, as well as degradation of lipid and microbial fermentation. In sugar autoxidation, MGO is formed from fragmentation of sugar by retro-aldol condensation in which oxygen plays an important role. This process mainly occurs in food containing a lot of carbohydrates, especially monosaccharides, from which the amount of MGO is higher than that from disaccharides.6 And the amount of MGO produced from glucose is higher than that from fructose. Honey with a high content of glucose and fructose forms MGO through sugar degradation during the heating processes in manufacturing and storage. The concentrations of GO and MGO in honey are in the range of 0.3–1.3 mg kg1 and 0.8–33 mg kg1, respectively.7 The most frequently used sweetener in foods or beverages is high fructose corn syrup which contains 90, 55 or 42% fructose. Levels of GO and MGO in commercial beverages which contain high amount of HFCS were 15.8–104.6 and 23.5–139.5 (mg per 100 ml), respectively.8 Coﬀee, as one of the popular drinks, was also studied for its MGO and GO levels. Four types of black coﬀee (espresso, bold, mild, and a

Downloaded by FAC DE QUIMICA on 16 September 2012 decaﬀeinated mild roast) were tested. Espresso was shown to

Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D contain the highest level of MGO at 230.9 mM, followed by 9 Fig. 1 Derivatization agents and products for methylglyoxal (MGO). bold coffee. The roasting process influences the content of MGO and GO in coffee beans because of the occurrence of and quantified with high-performance liquid chromatography Maillard reaction. When green beans were roasted at 210 1Cfor (HPLC) and can be monitored by an UV detector at 300–360 nm, 20 min, the glyoxal content increased in the first 6 min, peaked at by a fluorescent detector at 300–360 nm with excitation 13.07 0.39 mg per 100 g, then slowly decreased to 1.93 wavelengths and at 380–450 nm, with emission wavelengths, 0.05 mg per 100 g at 20 min. Methylglyoxal showed similar or by a mass detector (MS). Other agents, such as 6-hydroxy- behavior with its peak concentration (21.19 0.42 mg per 100 g) 10 2,4,5-triaminopyrimidine that forms a pteridine derivative, at 10 min and subsequently declined. Accumulation of MGO and a cysteamine forming 2-acetylthiazolidine, have also been in lipids is caused by lipid degradation during processing and analysed by HPLC. A reverse phase HPLC column is often storage. The amount of MGO formed in fish oils (tuna, cod applied. In the GC method, oxime derivatives of MGO with liver and salmon oils) heated at 60 1C for 7 days ranged from 1 O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine (PFBHA) can 2.03–0.13 to 2.89–0.11 mg kg , whereas among vegetable oils be detected by a flame ionization detector, a MS/SIM detector, (soybean, olive and corn oil) under these conditions, only olive 1 11 an electron-capture detector, or a flame photometric detector. oil yielded MGO (0.61–0.03 mg kg ). During fermentation, 1,2-Diaminobenzene derivatives of MGO can be analysed microorganisms release MGO into food products most commonly using a flame ionization detector, a MS/SIM, or a specific into alcoholic drinks and dairy products. Levels of MGO in 12 nitrogen/phosphorus detector. For biological samples, additional brandy, vinegar and wine were 1.9,35and10ppm,respectively. procedures might be needed to separate MGO from proteins. More than 90% of MGO was demonstrated to be bound to 5. Relationship between MGO and GO formation proteins, and perchloric acid was needed as a deproteinization and the Maillard reaction agent. Another benefit of the use of perchloric acid is to keep the samples in low pH, which prevents degradation of Maillard reaction plays an important role in MGO and GO dihydroxyacetone phosphate and glyceraldehyde 3-phosphate formation in vitro and in vivo. In fact, scavenging RCS may to MGO. terminate Maillard reaction, and suppressing Maillard reaction

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 4140–4149 4141 View Online

may reduce the level of RCS. Maillard reaction is the major route for the generation of ﬂavour and colour. Therefore, it is important to study the relationship between ﬂavour generation and RCS formation.

5.1. Chemistry of Maillard reaction Maillard reaction, which can be generally defined as the chemical interaction involving carbohydrates and amino compounds, is responsible for the generation of roasted, toasted and caramel-like aromas, as well as for the development of brown colour in foods. Maillard reaction also has both nutritional and toxicological effects on processed food.13–15 Many of the antinutritional aspects of the Maillard reaction, such as effects on the availability of essential Fig. 3 Strecker degradation. amino acids, on enzyme activity, as well as on the absorption/ utilization of minerals, have been extensively studied.13,14 Strecker degradation, which is considered a significant source of Maillard reaction occurs in three stages.13 The initial stage flavour compounds, is associated with the intermediate stage of (Fig. 2) is the condensation of the carbonyl group of a reducing Maillard reaction. If Maillard reaction can be seen as the sugar with an amino compound to form the Schiff base which degradation of sugar catalysed by amino compounds, from then cyclizes to the N-substituted aldosylamine. These Schiff another point of view, Strecker degradation can be taken as the bases can rearrange to form RCS such as 1-deoxyglycosone or degradation of amino acids initiated by RCS. 3-deoxyglycosone through amino-deoxyaldose or ketose by In Strecker degradation, dicarbonyl compounds such as Amadori or Heyns rearrangements. The Amadori rearrangement MGO and GO react with amino acids to produce carbon dioxide product is not stable, so the intermediate stages (Fig. 2) include and aldehydes with one less carbon atom and a-aminoketones enolization, deamination, dehydration, cyclization, retroaldoliza- that are key precursors of heterocyclic flavour compounds such tion, isomerization and fragmentation to carbonyl compounds as pyrazines, oxazoles and thiazoles (Fig. 3).16 (small molecular RCS), furan derivatives and other intermediates. The final stage is the reaction of theseRCSandfuranderivatives 5.3. Proposed mechanism of methylglyoxal formation to form flavour and colour compounds. In the intermediate stage, an amine group is released from the reaction, this means RCS, including glyoxal, methylglyoxal and 3-deoxyglucosone, amino acids play important roles in catalysing sugar activation have been demonstrated to be formed in early glycation and fragmentation. On the other hand, sugar can be degraded into in vitro from the degradation of glucose and Schiff base carbonyl compounds at high temperature through enolization, adduct. The reactions were influenced by the concentration 17 dehydration and fragmentation. of phosphate buffer and availability of trace metal ions. Downloaded by FAC DE QUIMICA on 16 September 2012 Glucosone can be formed from monosaccharide autoxidation.18

Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D 5.2. Strecker degradation Glyoxal can be generated in the degradation of glucose by Generally, ﬂavour compounds can be categorised into two groups: retro-aldol condensation, which is activated by deprotonation cyclization/condensation products and fragmentation products. of the 2- or 3-hydroxy groups. Hydrogen peroxide, which is formed in autoxidation of glycoaldehyde to glyoxal, and glucose to glucosone, can also stimulate glyoxal formation by hydroxyl radical-mediated acetal proton abstraction from glucopyranose and a-elimination reactions.18 Deprotonation of carbon-2 of glucose and re-distribution of the electron density between carbon-1 and carbon-2 or carbon-2 and carbon-3 in glucose lead to dehydration, forming the 1,2-enol or 2,3-enol and thereby 1-deoxyglucosone (1-DG) or 3-deoxyglucosone (3-DG), respectively.19 Methylglyoxal may be formed by fragmentation of 3-DG (Fig. 4). The formation of glyoxal, methylglyoxal and 3-DG from glucose is dependent on phosphate buﬀer and availability of trace metal ions.20 This may be due to phosphate 2 dianion HPO4 and metal ions catalysing the deprotonation of glucose and the autoxidation of glycoaldehyde and hydroxyl radical formation implicated in glyoxal formation.18 Moreover, trace metal ion phosphate complexes may be related to the activation of glucose for 3-DG formation.20 The formation of fructosamine residues is the major pathway of early glycation by glucose, but not the only one. Originally, RCS were considered to be formed from fructosamine only. However, some studies show that 3-DG, MGO and GO were Fig. 2 Initiate and intermediate stages of the Maillard reaction.13 generated throughout the whole reaction and changes in their

4142 Chem. Soc. Rev., 2012, 41, 4140–4149 This journal is c The Royal Society of Chemistry 2012 View Online

only, among all of which MGO and GO play important roles as intermediates. Carbohydrates can transform into glycosones at the beginning of Maillard reaction, and then these glycosones either cyclize into flavour compounds or break into a-dicarbonyls such as MGO and GO, then follow a recombination of these intermediates. One of these is 2,5-dimethyl-4-hydroxy-3(2H)- furanone (DMHF), which is generally considered the product of cyclization of intact glycosones and/or the recombination of MGO and GO. Aromas formed from carbohydrates and amino acids can be divided into amino acid-specific and amino acid-non-specific pathways. For the amino acid-non-specific pathway, a-dicarbonyl (MGO or GO) reacts with most types of amino acids forming a-aminoketone via Strecker degradation which leads to the formation of alkylpyrazines, oxazoles and 20 Fig. 4 Oxidative formation of methylglyoxal (MGO) from glucose. oxazolines. In the amino acid-specific pathway, amino acids such as cysteine and proline, a-aminoketone or a-dicarbonyl are involved in the generation of thiazoles, thiazolines, pyrrolines, and pyridines. Although some flavour compounds (e.g., methional, phenylacetaldehyde) are generated from amino acids only, a-dicarbonyl compounds are still involved in their formation pathways, particularly through Strecker degradation. In addition, peptides can also react with a-dicarbonyl compounds to generate some peptide-specific aromas (e.g., pyrazinones).

6.1. Formation of 2,5-dimethyl-4-hydroxy-3(2H)-furanone (DMHF) from carbohydrate or MGO 2,5-Dimethyl-4-hydroxy-3(2H)-furanone (DMHF, known as furaneol), with an intense caramel-like aroma, was originally discovered as a key ﬂavour component of strawberry in 1965.23 DMHF has been found to be an important odour-active Fig. 5 Formation of methylglyoxal (MGO) and glyoxal (GO) in compound in various natural and processed foods, such as 21 Maillard reaction. pineapple, raspberry, tomato and grape, as well as in roasted

Downloaded by FAC DE QUIMICA on 16 September 2012 24 21 coffee, bread crust, roasted almond and soy sauce. In fruits, Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D concentrations did not follow that of fructosamine. In other DMHF is usually formed by biosynthesis. Because of its words, RCS may be formed from other sources, such as from widespread occurrence, DMHF became a major reactant in glucose degradation and from the Schiff base (Fig. 5). The generating other flavour compounds. At low pH, DMHF has formation mechanism of RCS is similar to glucose degradation been shown to react with cysteine or hydrogen sulfide in except for the presence of aldimine that could be hydrolyzed to generating meat-like aroma compounds.25 Some roast aroma MGO, GO and 3-DG. The presence of the aldimine group 20 compounds such as alkylpyrazines can also be generated accelerates the formation of RCS. The formation of 3-DG through the decomposition of DMHF with phenylalanine.26 and fructosamine is parallel reaction pathways in which the The formation pathways of DMHF have been studied in model deprotonation of carbon-2 of the Schiff’s base is a critical point. experiments of thermal degradation of 6-deoxysugars, hexoses and Furthermore, in the Namiki pathway, glyoxal can be formed pentoses in the presence or absence of amino acids.27–29 Generally, directly from the Schiff’s base through erythritol, while methyl- 21 DMHF can be formed through 2,3-enolization of 6-deoxysugars, glyoxal is from 3-DG though retroaldolisation. Formation of hexoses and pentoses leading to 1-deoxyosones as intermediates.30 a superoxide radical, which is a key in GO generation, can be 22 Compared to hexoses and pentoses, 6-deoxysugars, such as detected in the early stage of glycation. MGO and 3-DG rhamnose, are more effective in forming DMHF through generation, however, does not involve oxidation. The formation 2,3-dioxo-4,5-dihydroxyhexane that is not easily formed from of glyoxal, methylglyoxal and 3-DG in glucose–amine reaction is hexoses and cannot be generated from pentoses.27 Schieberle also dependent on phosphate buffer concentration and availability 28 31 20 in 1992, and later Hofmann and Schieberle in 2001 showed of trace metal ions. that DMHF can be formed from hexose via acetylformoin (2,4-dihydroxy-2,5-dimethyl-3(H)-furanone) reduction, which 6. Importance of MGO and GO as intermediates may proceed either by disproportionation reaction or a in flavour generation Strecker reaction with amino acids. Only one study has shown the formation of DMHF from pentoses. Blank and Fay Flavour generated from Maillard reaction can follow three indicated that elongation of pentoses by Strecker aldehyde paths, (i) from carbohydrates only, (ii) from a combination of of glycine was an alternative pathway of DMHF formation in carbohydrates and amino acids, and (iii) from amino acids which acetylformoin was also proposed as an intermediate.29

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 4140–4149 4143 View Online

Fig. 6 Reaction pathway leading methylglyoxal to 2,5-dimethyl-4- hydroxy-3(2H)-furanone via Cannizzaro reaction.

Hexose or pentose can be cleaved into MGO and 1-hydroxy- Fig. 8 Proposed pathway of oxazoles and oxazolines formation. 2-propanone to generate DMHF. This reaction has been proved to be a major pathway when glucose is reacted with 6.2.2. Formation of oxazoles and oxazolines via Strecker proline in an aqueous solution.32 MGO, depending on the pH, degradation (non-specific amino acids). Oxazoles and oxazolines differently affected DMHF generation in the presence or absence are two closely related heterocyclic flavour compounds containing of amino acids. The DMHF level increased as pH increased when nitrogen and oxygen atoms. Oxazoles and oxazolines occur in cysteine reacted with MGO, whereas the trend was reversed in the various processed foods, such as roasted and ground coffee, presence of glycine.24 When MGO was heated alone at 120 1C, roasted cocoa, heated soy sauce, baked potatoes and roasted 34,35 the formation of DMHF was observed, and the MGO level was peanuts. significantly increased as the pH of the reaction increased. MGO Formation pathways of oxazoles and oxazolines were first may transform into 1-hydroxy-2-propanone and pyruvic acid proposed between a dicarbonyl compound and an amino acid through the Cannizzaro reaction (Fig. 6), and subsequently lead undergoing decarboxylation (Fig. 8). to DMHF by reacting MGO with 1-hydroxy-2-propanone.32 DMHF formation from MGO was pH-dependent because the Cannizzaro reaction is a base preferential reaction. 6.2.3. Formation of 2-acetyltetrahydropyridine and 2-acetylpyrroline via Strecker degradation (specific amino acids). Flavour generation reactions occur most often at high temperatures 6.2. Formation of aroma compounds from carbohydrates and especially Strecker degradation in which carbonyl groups react amino acids with amine groups through nucleophilic attack. However, some flavour compounds such as popcorn like flavour 2-acetyltetra- 6.2.1. Formation of alkylpyrazines via Strecker degradation hydropyridine and roasted aroma 2-acetylpyrroline can be (non-specific amino acids). Alkylpyrazines, which are nitrogen- formed from MGO in the presence of a specific amino acid containing heterocyclic compounds, occur in a wide range of proline, but without forming a-aminoketones. For example, raw and processed foods with potent and characteristic aroma. 2-acetyltetrahydropyridine, which is a popcorn like flavour in The most plausible formation mechanism of pyrazines is the Downloaded by FAC DE QUIMICA on 16 September 2012 rice and bread crust, can be formed from Strecker degradation condensation of two a-aminoketones. Dicarbonyls from sugar 36

Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D of proline and MGO through decarboxylation and dehydration degradation at the early stage of Maillard reaction react with (Fig. 9). The degradation product from the reaction between amino acids via Strecker degradation to form a-aminoketones proline and MGO can continually react with MGO to form leading to pyrazine formation through oxidation of dihydro- 2-acetylpyrroline (Fig. 10). On the other hand, 2-acetyltetra- pyrazines. If the alkyl groups in a-aminoketones are diﬀerent, hydropyridine can be biosynthesized through lysine and MGO.37 the isomers of pyrazines are observed33 (Fig. 7). The gene(lat) encoding the L-lysine e-aminotransferase (LAT) in Streptomyces clavuligerus was cloned and expressed in Escherichia coli. Lysine was found to be transformed to 1-piperideine-6-carboxylic acid. And 2-acetyltetrahydropyridine was characterized from the reaction mixture of 1-piperideine-6- carboxylic acid and MGO (Fig. 11).

Fig. 9 Hypothetical mechanism for the formation of 2-acetyltetra- Fig. 7 Proposed pathways of alkylpyrazines formation.33 hydropyridine.36

4144 Chem. Soc. Rev., 2012, 41, 4140–4149 This journal is c The Royal Society of Chemistry 2012 View Online

Fig. 10 Hypothetical mechanism for the formation of 2-acetylpyrroline.36

Fig. 12 Formation pathways of thiazoles and thiazolines.41

Fig. 11 Biosynthesis of 2-acetyltetrahydropyridine.37

6.2.4. Formation of thiazoles and thiazolines via Strecker degradation (cysteine specific). Other aroma compounds formed in the presence of cysteine via Strecker degradation are thiazoles and thiazolines. They are closely related to oxazoles Fig. 13 Formation of phenylacetaldehyde and phenylacetic acid and oxazolines, except a sulfur atom replaces the oxygen atom in from methylglyoxal (MGO) and phenylalanine.42 position 1 and a nitrogen atom occupies position 3. The first thiazole and thiazoline isolated from foods were 4-methyl-5- 6.4. Formation of pyrazinones from peptides and alpha- vinylthiazole and 2-acetyl-2-thiazoline, respectively.38,39 Later, dicarbonyls various thiazoles and thiazolines were identified in many food

Downloaded by FAC DE QUIMICA on 16 September 2012 Besides different amino acids, peptides are also important systems, especially in thermally treated foods such as baked 40 35 precursors of flavour generation with RCS. The major volatile Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D potatoes and roasted peanuts. Generally, 2-alkylthiazoles compounds from Maillard reaction of glycine, diglycine and possess green, vegetable-like aroma. Increasing substitutions in triglycine with glucose are alkylpyrazines. Triglycine was not positions 4 and 5 adds more nutty, roasted and sometimes meaty stable and degraded to cyclic diglycine and glycine, whereas notes. Because they contain sulfur thiazoles and thiazolines that diglycine had a higher stability than triglycine toward hydrolytic are formed mostly in the presence of cysteine. The most accepted cleavage of the peptide bond.44 Pyrazinone is a peptide-specific formation mechanism of these is that a-dicarbonyls, such as reaction product, and could generate from diglycine, triglycine 2,3-butanedione, react with hydrogen sulfide, ammonia and and tetraglycine with glucose.45 These compounds are formed acetaldehyde to form either 3-hydroxy-3-mercapto-2-butanone from the decarboxylation of 2-(30-alkyl-20-oxopyrazin-10-yl) or 3-mercapto-2-butanone, leading to 2,4,5-trimethythiazole and alkanoic acid at high temperatures, such as 180 1C(Fig.14), 2,4,5-trimethyl-3-thiazoline through condensation, cyclization or the condensation of a-dicarbonyl with isoasparagine.46 and condensation with ethylideneamine41 (Fig. 12). MGO and Model systems of Gly-Leu and Leu-Gly with MGO were also GO should undergo similar reaction. studied. It was found that a peptide sequence did not affect the type of aromas, but only changed the amount of pyrazinones 6.3. Formation of aroma compounds from amino acids via slightly.47 Strecker degradation Aromas, such as 3-methylbutanal (malty) or phenylacetaldehyde 7. Trapping of MGO and GO by phenolic (honey-like), that are derived from leucine and phenylalanine, compounds respectively, in the presence of a-dicarbonyl, generally are Strecker aldehydes formed in Strecker degradation. Besides MGO and GO are extremely reactive and readily modify aldehydes, some odour-active acids were also identified in lysine, arginine, and cysteine residues on proteins in vivo to the presence of RCS. In Strecker degradation, for example, form advanced glycation end products (AGEs) that are linked phenylalanine reacts with MGO or GO, preferentially leading to hyperglycemia and diabetes complications. As the first step to phenylacetic acid generation42 (Fig. 13), whereas reacting of AGEs formation, proteins in the tissues are modified by with 3-DG favors the formation of phenylacetaldehyde.43 reducing sugars through the reaction between a free amino

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 4140–4149 4145 View Online

Fig. 15 Adducts of methylglyoxal (MGO) and epigallocathin-3- gallate (EGCG).

Fig. 14 Mechanism of the formation of alkylpyrazinone from a tea polyphenolic compound was 3 to 1. Theaflavins, the dipeptide and methylglyoxal (MGO).45 main black tea components, were more reactive toward MGO than other polyphenols tested. Theaflavins showed high levels group of proteins and a carbonyl group of the sugars, leading to of MGO reduction with respect to the control sample, which the formation of fructosamines via a Schiff base by Amadori suggested that theaflavins would be excellent candidates in the rearrangement. Then, both Schiff base and Amadori product treatment of MGO scavenging in future in vivo studies. The undergo a further series of reactions through dicarbonyl inter- decreased amounts of MGO in TF1 (theaflavin), TF2 (theaflavin- mediates (e.g., GO and MGO), to form AGEs. The amount of 3- and -30-gallate), and TF3 (theaflavin-3,30-digallate) were AGEs is dependent on the concentration of dicarbonyl inter- 63.1, 60.1, and 66.7%, respectively. All tested theaflavins mediates, reactivity and concentration of amino acid residues, decreased MGO by about 66%, which implies that one theaflavin the half-life of protein, and the activity of glyoxalase system. molecule can trap two MGO molecules. In addition, the efficacy Accumulation of MGO and GO in cells may cause carbonyl of EGCG trapping to GO was much lower than that of MGO, stress that in its first step induces the formation of hydrogen because GO is much easier to polymerise in aqueous solution. peroxide. The hydrogen peroxide may then increase oxidative Trapping efficacy could be slowed down by the transformation stress and, therefore, tissue damage. Moreover, MGO can activate from polymers to free GO. The major adduct between EGCG Downloaded by FAC DE QUIMICA on 16 September 2012 NF-kB and induce the associated gene that is responsible for and MGO has been identified (Fig. 15). It is concluded that the 2 Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D inflammation and proliferation. Therefore, studies on the reaction between EGCG and MGO dominantly occurs at the prevention of accumulation of dicarbonyl compounds become C6- and C8-position in the A ring of EGCG. Slightly alkaline very urgent. Many synthetic chemicals have been reported to pH facilitates the addition of MGO at these two positions inhibit formation of AGEs. More recently, researchers have to form mono- and di-MGO adducts. Isomers with R- and documented the scavenging effect of MGO and GO by natural S-configuration could also be seen at each position.50 phenolic compounds, such as flavanols, chalcones, stilbenes, Chalcones, represented by phloretin and phloridzin, have isoflavones and phenolic acids. been studied for their trapping efficacy of MGO and GO.51 An inhibition of protein glycation on different stages was GO still showed a lower attachment than MGO because of studied by using different dietary flavonoids. Luteolin, rutin, easier polymerization. Phloretin had a much higher trapping epigallocathin-3-gallate (EGCG), and quercetin demonstrated rate than did phloridzin suggesting that glycosylation of a significant inhibitory effects on MGO mediated AGEs formation hydroxyl group at position 2 significantly slows down the by 82.2, 77.7, 69.1 and 65.3%, respectively, while catechin, formation of adducts. A similar phenomenon was observed in epicatechin (EC), epicatechin gallate (ECG), epigallocatechin EGCG. Positions 3 and 5 of the A ring in phloretin and (EGC), kaempferol, and naringenin showed a lower inhibitory phloridzin were active sites to bind with MGO to form mono- effect (13–54%).48 and di-MGO adducts (Fig. 16). Stilbenes were compared for Among all flavonoids, molecules with catechin-like structure theirefficacyoftrappingMGO.52 Among 2,3,5,40-tetrahydroxy- show the strongest effect on direct trapping of MGO. In tea stilbene-2-O-b-D-glucoside (THSG), pterostilbene and resveratrol, polyphenol adduct investigation, the mole ratio of MGO to each THSG was the most effective trapping agent. Positions 4 and 6 of specific polyphenol was set at three, and the % reduction of the A ring in stilbenes were the major active sites for trapping MGO was compared with that of the control sample at 0 1Con MGO (Fig. 17). Genistein, one of the isoflavones, was reported to an ice/salt bath for 1 h.49 After 1 h at 37 1C incubation, MGO trap MGO under neutral and alkaline conditions in vitro.53 was very stable, only decreasing by 5.8%. All tea polyphenolic Mono- and di-MGO adducts were identified. Mono-MGO compounds scavenge MGO. Most tea catechins decreased MGO adducted at position 8 on the A ring of genistein, and di-MGO by about 33%, which indicates that one catechin molecule reacts conjugated at both positions 6 and 8 on the A ring of genistein with one MGO molecule, since the initial mole ratio of MGO to (Fig. 18). Based on all the results obtained here, it is concluded

4146 Chem. Soc. Rev., 2012, 41, 4140–4149 This journal is c The Royal Society of Chemistry 2012 View Online

Fig. 19 Carbon electron charges on epicatechin.55

aminoguanidine.54 Compounds consisting of a single benzene Fig. 16 Adducts of methylglyoxal (MGO) and phloridzin.51 ring with the addition of at least one hydroxyl group were studied for MGO trapping. Compounds with one hydroxyl group on the benzene ring cannot react with MGO. Benzenetriols showed relatively higher trapping capability. The decreasing rate of MGO trapping for pyrogallol, 1,2,4-trihydroxybenzene, and 1,3,5-trihydroxybenzene were 55%, 49% and 64%, respectively. Steric hindrance and carbon electron charges on the benzene ring were reported as the inﬂuential factors. A carbon electron charge of 0.24 was the minimum value for high reactivity using a computational chemistry calculation.55 The carbon electron charges of epicatechin are shown in Fig. 19. The carbon electron charge on position 6 showed the lowest number of 0.26, followed by the one on position 8 (0.25). These results explain why positions 6 and 8 were the active sites for ﬂavanols.

8. Effect of polyphenolic compounds on Maillard flavour generations Some polyphenolic compounds can efficiently trap MGO and GO Downloaded by FAC DE QUIMICA on 16 September 2012 Fig. 17 Adducts of methylglyoxal (MGO) and stilbenes.52 usually generated from Maillard reaction. Therefore, scavenging

Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D RCS may influence Maillard reaction, while suppressing Maillard that phenolic compounds having the same A ring structure reaction may reduce the level of RCS. Moreover, Maillard (EGCG, phloretin, chalcone, stilbene and genistein) can efficiently reaction is the major route for the generation of flavour and trap MGO or GO to form mono- or di-MGO adducts even with colour. Thus, it is important to study the effect of polyphenolic different C rings. The mechanism is that the slightly alkaline pH compounds on flavour generation by trapping RCS. can increase the nucleophilicity of the unsubstituted carbons at Epicatechin was first studied for its influence on flavour 56 the A ring and facilitate the addition of MGO at these two generation in an aqueous model with glucose and glycine. positions to form mono- or di-MGO adducts. This observation Addition of epicatechin (EC) in the reaction decreased the also applies to polymers of flavonoids such as proanthocyanidins. generation of 2,3-butanedione, acetol, pyrazine, methylpyrazine, Peng et al. (2008) demonstrated that proanthocyanidin 2,3,5-trimethylpyrazine and cyclotene, because these are volatile B2 could effectively scavenge MGO, in a similar manner to compounds formed from C2/C3 or C3/C3 sugar fragments, and epicatechin can directly react with C2, C3 and C4 sugar fragments. Through NMR analysis, glyoxal, hydroxyacetone, and erythrose were identified as the sugar fragments. Methylglyoxal could also bind with EC, but EC-MGO underwent conformation/ constitutional exchange.57 One of the isomers consisted of a covalent binding between C1 of MGO and either the C6 or C8 position of the EC A ring, in accordance with a previous study.50 Besides EC, other flavan-3-ols, such as ECG and EGCG, showed the same phenomenon.58 Generation of phenolic-C2, C3, C4 and C6 fragment adducts statistically fit the reduction of glyoxal, glycolaldehyde, MGO, hydroxyacetone, diacetyl, acetoin, and 3-deoxyglucosone. Under a roast condition, addition of EC can significantly reduce the generation of hydroxyacetone, Fig. 18 Adducts of methylglyoxal (MGO) and genistein.53 2-methylpyrazine, 2,3,5-trimethylpyrazine, furfural, 2-acetylfuran,

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 4140–4149 4147 View Online

5-methylfurfural, 2(5H)-furanone, 2-acetylpyrrole and furfuryl 18 P. J. Thornalley, S. P. Wolff, J. Crabbe and A. Stern, Biochim. alcohol. However, it only reacted with C5 and C6 sugar Biophys. Acta, 1984, 797, 276–287. 59 19 P. J. Thornalley, Environ. Health Perspect., 1985, 64, 297–307. fragments. EC can quench 3-deoxy-2-hexosulose. Phloretin 20 H. Nursten, The Maillard Reaction: Chemistry, Biochemistry and and phloridzin were also studied for their effect on flavour Implication, The Royal Society of Chemistry, Cambridge, UK, generation in a model reaction with lysine and glucose.60 The 2005. amount of lysine could significantly affect the generation of 21 T. Hayashi and M. Namiki, Agric. Biol. Chem., 1980, 44, 2575–2580. flavour compounds. Furfural, 1,2,3,4-tetrahydroquinoline 22 C. J. Mullarkey, D. Edelstein and M. Brownlee, Biochem. Biophys. and eight pyrazines were identified in the reactant ratio of Res. Commun., 1990, 173, 932–939. 1 : 3 (glucose : lysine), while other additional compounds, 23 B.Willhalm,M.StollandA.F.Thomas,Chem. Ind. (London, U. K.), such as 2,3-dihydro-1-methyl-1H-indole, N-ethyl-2,3-xylidine 1965, 38, 1629–1630. 24 Y. Wang and C. T. Ho, J. Agric. Food Chem., 2008, 56, and 3-ethyl-2,5-dimethylpyrazine, were only found in lower 7405–7409. amounts of lysine. Both phloretin and phloridzin inhibited 25 Y. Zheng, S. Brown, W. O. Leding, C. Mussinan and C. T. Ho, the formation of volatile compounds, but phloretin showed J. Agric. Food Chem., 1997, 45, 894–897. 26 K. K. Jutta and B. Werner, Z. Lebensm.-Unters. Forsch., 1990, 190, a higher activity. Moreover, phloretin and phloridzin were 14–16. reported to show lower inhibitory effect with a higher content 27 T. Hofmann and P. Schieberle, J. Agric. Food Chem., 1997, 45, of lysine, suggesting the competition of these phenolic compounds 898–906. and amino groups for RCS. 28 P. Schieberle, Flavor Precursors: Thermal and Enzymatic Conversions, ed.R.Teranishi,G.R.TakeokaandM.Gu¨ntert, ACS Symposium Series 490, Washington, DC, 1992, pp. 164–175. 29 I. Blank and L. B. Fay, J. Agric. Food Chem., 1996, 44, 531–536. 9. Conclusion 30 J. E. Hodge, F. D. Mills and B. E. Fisher, Cereal Sci. Today, 1972, 17, 34–40. MGO and GO, which are intermediates in the Maillard 31 T. Hofmann and P. Schieberle, Flavour 2000, Perception Release reaction, are involved in flavour and colour generation for Evaluation Formation Acceptance Nutrition/Health, Proceedings of the 6th Wartburg Aroma Symposium. Eisenach, ed. M. Rothe, foods and drinks. Phenolic compounds having the same A ring Eisenach, Germany, 2001, pp. 311–322. structure (EGCG, phloretin, chalcone, stilbene and genistein) 32 P. Schieberle, Ann. N. Y. Acad. Sci., 2005, 1043, 236–248. can efficiently trap MGO or GO to form mono- or di-MGO/ 33 G. Vernin, The chemistry of heterocyclic flavoring and aroma GO adducts even with different C rings. Because of the compounds, Ellis Horwood, Chichester, UK, 1982. 34 O. G. Vitzthum, P. Werkhoff and P. Hubert, J. Food Sci., 1975, 40, trapping efficacy of phenolic compounds, adding them into 911–916. Maillard reaction would change the formation of MGO and 35 C. T. Ho, Q. Z. Jin, M. H. Lee and S. S. Chang, J. Agric. Food GO, and change the flavour properties. Therefore, RCS trapping Chem., 1983, 31, 1384–1386. agents provide an alternative way to influence flavour generation 36 P. Schieberle, J. Agric. Food Chem., 1995, 43, 2442–2448. 37 M. L. Wu, J. H. Chen, C. T. Ho and T. C. Huang, J. Agric. Food and affect the quality of food. Chem., 2007, 55, 1767–1772. 38 M. Stoll, P. Dietrich, E. Sundte and M. Winter, Helv. Chim. Acta, 1967, 50, 2065. Downloaded by FAC DE QUIMICA on 16 September 2012 Notes and references 39 C. H. Tonsbeek, H. Copier and A. J. Placken, J. Agric. Food

Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D Chem., 1971, 19, 1014–1016. 1 I. Nemet, L. Varga-Defterdarovic and Z. Turk, Mol. Nutr. Food 40 E. C. Coleman, C. T. Ho and S. S. Chang, J. Agric. Food Chem., Res., 2006, 50, 1105–1117. 1981, 29, 42–48. 2 L. Wu and B. H. Juurlink, Hypertension, 2002, 39, 809–814. 41 H. J. Takken, L. M. van der Linde, P. J. de Valois, H. M. van Dort 3H.Ju¨ergens, W. Haass, T. R. Castan˜ eda, A. Schu¨ermann, and M. Boelens, Phenolic, Sulfur, and Nitrogen Compounds in C. Koebnick, F. Dombrowski, B. Otto, A. R. Nawrocki, Food Favors, ed. G. Charalambous and I. Katz, ACS Symposium P. E. Scherer, J. Spranger, M. Ristow, H. Joost, P. J. Havel and Series 26, American Chemical Society, Washington, DC, 1976, M. H. Tschoep, Obes. Res., 2005, 13, 1146–1156. pp. 114–121. 4 D. J. Creighton, M. Migliorini, T. Pourmotabbed and 42 T. Hofmann and P. Schieberle, J. Agric. Food Chem., 2000, 48, M. K. Guha, Biochemistry, 1988, 27, 7376–7384. 4301–4305. 5 I. Nemet, D. Vikic-Topic and L. Varga-Defterdarovic, Bioorg. 43 T. Hofmann, P. Munch and P. Schieberle, J. Agric. Food Chem., Chem., 2004, 32, 560–570. 2000, 48, 434–440. 6 A. Hollnagel and L. W. Kroh, Z. Lebensm.-Unters. Forsch., 1998, 44 C. Y. Lu, Z. Hao, R. Payne and C. T. Ho, J. Agric. Food Chem., 207, 50–54. 2005, 53, 6443–6447. 7 E. Marceau and V. A. Yaylayan, J. Agric. Food Chem., 2009, 57, 45 Y. C. Oh, C. K. Shu and C. T. Ho, J. Agric. Food Chem., 1991, 39, 10837–10844. 1553–1554. 8 C. Y. Lo, S. Li, Y. Wang, D. Tan, M. Pan, S. Sang and C. T. Ho, 46 C. K. Shu and B. M. Lawrence, J. Agric. Food Chem., 1995, 43, Food Chem., 2008, 107, 1099–1105. 779–781. 9 J. Wang and T. Chang, J. Food Sci., 2010, 75, 167–171. 47 Y. C. Oh, C. K. Shu and C. T. Ho, J. Agric. Food Chem., 1992, 40, 10 M. Daglia, A. Papetti, C. Aceti, B. Sordelli, V. Spini and 118–121. G. Gazzani, J. Agric. Food Chem., 2007, 55, 8877–8882. 48 C. H. Wu and G. C. Yen, J. Agric. Food Chem., 2005, 53, 11 K. Fujioka and T. Shibamoto, Lipids, 2004, 39, 481–486. 3167–3173. 12 J. A. Rodrigues, A. A. Barros and P. G. Rodrigues, J. Agric. Food 49 C. Y. Lo, S. Li, D. Tan, S. Sang, M. H. Pan and C. T. Ho, Chem., 1999, 47, 3219–3222. Mol. Nutr. Food Res., 2006, 50, 1118–1128. 13 J. E. Hodge, J. Agric. Food Chem., 1953, 1, 928–943. 50 S. Sang, X. Shao, N. Bai, C. Y. Lo, C. S. Yang and C. T. Ho, 14 T. Hayashi and T. J. Shibamoto, J. Agric. Food Chem., 1985, 33, Chem. Res. Toxicol., 2007, 20, 1862–1870. 1090–1093. 51 X. Shao, N. Bai, K. He, C. T. Ho, C. S. Yang and S. Sang, 15 D. S. Mottram, B. L. Wedzicha and A. T. Dodson, Nature, 2002, Chem. Res. Toxicol., 2008, 10, 2042–2050. 419, 448–449. 52 L. Lv, X. Shao, L. Wang, D. Huang, C. T. Ho and S. Sang, 16 G. P. Rizzi, J. Agric. Food Chem., 1972, 20, 1081–1085. J. Agric. Food Chem., 2010, 58, 2239–2245. 17 P. J. Thornalley, A. Langborg and S. Harjit, Biochem. J., 1999, 53 L. Lv, X. Shao, H. Chen, C. T. Ho and S. Sang, Chem. Res. 344, 109–116. Toxicol., 2011, 24, 579–586.

4148 Chem. Soc. Rev., 2012, 41, 4140–4149 This journal is c The Royal Society of Chemistry 2012 View Online

54 X. Peng, J. Ma, J. Chao, Z. Sun, R. C. Chang, I. Tse, 57 V. M. Totlani and D. G. Peterson, J. Agric. Food Chem., 2006, 54, E. T. Li, F. Chen and M. Wang, J. Agric. Food Chem., 2010, 58, 7311–7318. 6692–6696. 58 Y. Noda and D. G. Peterson, J. Agric. Food Chem., 2007, 55, 3686–3691. 55 C. Y. Lo, W. T. Hsiao and X. Y. Chen, J. Food Sci., 2011, 76, 59 V. M. Totlani and D. G. Peterson, J. Agric. Food Chem., 2007, 55, H90–H96. 414–420. 56 V. M. Totlani and D. G. Peterson, J. Agric. Food Chem., 2005, 53, 60 J. Ma, X. Peng, K. M. Ng, C. M. Che and M. Wang, Food Funct., 4130–4135. 2012, 3, 178–186. Downloaded by FAC DE QUIMICA on 16 September 2012 Published on 16 April 2012 http://pubs.rsc.org | doi:10.1039/C2CS35025D

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 4140–4149 4149 732

Asian Australas. J. Anim. Sci. Vol. 26, No. 5 : 732-742 May 2013 http://dx.doi.org/10.5713/ajas.2012.12619

www.ajas.info pISSN 1011-2367 eISSN 1976-5517

Flavour Chemistry of Chicken Meat: A Review

Dinesh D. Jayasena, Dong Uk Ahn1, Ki Chang Nam2 and Cheorun Jo* Department of Animal Science and Biotechnology, Chungnam National University, Daejeon 305-764, Korea

ABSTRACT: Flavour comprises mainly of taste and aroma and is involved in consumers’ meat-buying behavior and preferences. Chicken meat flavour is supposed to be affected by a number of ante- and post-mortem factors, including breed, diet, post-mortem ageing, method of cooking, etc. Additionally, chicken meat is more susceptible to quality deterioration mainly due to lipid oxidation with resulting off-flavours. Therefore, the intent of this paper is to highlight the mechanisms and chemical compounds responsible for chicken meat flavour and off-flavour development to help producers in producing the most flavourful and consistent product possible. Chicken meat flavour is thermally derived and the Maillard reaction, thermal degradation of lipids, and interaction between these 2 reactions are mainly responsible for the generation of flavour and aroma compounds. The reaction of cysteine and sugar can lead to characteristic meat flavour specially for chicken and pork. Volatile compounds including 2-methyl-3-furanthiol, 2-furfurylthiol, methionol, 2,4,5-trimethyl-thiazole, nonanol, 2-trans-nonenal, and other compounds have been identified as important for the flavour of chicken. However 2-methyl-3-furanthiol is considered as the most vital chemical compound for chicken flavour development. In addition, a large number of heterocyclic compounds are formed when higher temperature and low moisture conditions are used during certain cooking methods of chicken meat such as roasting, grilling, frying or pressure cooking compared to boiled chicken meat. Major volatile compounds responsible for fried chicken are 3,5-dimethyl-1,2,4-trithiolanes, 2,4,6-trimethylperhydro-1,3,5-dithiazines, 3,5- diisobutyl-1,2,4-trithiolane, 3-methyl-5-butyl-1,2,4-trithiolane, 3-methyl-5-pentyl-1,2,4-trithiolane, 2,4-decadienal and trans-4,5-epoxy- trans-2-decenal. Alkylpyrazines were reported in the flavours of fried chicken and roasted chicken but not in chicken broth. The main reason for flavour deterioration and formation of undesirable “warmed over flavour” in chicken meat products are supposed to be the lack of -tocopherol in chicken meat. (Key Words: Flavour, Chicken Meat, Maillard Reaction, Lipid Oxidation, Heterocyclic Compounds)

INTRODUCTION are primarily contributed by the volatile compounds originated through heat induced complex reactions between Continued consumption of meat and meat products can non-volatile components of lean and fatty tissues during be ensured through a tasty, nutritious and safety meat cooking (Mottram, 1998). supply for the consumers (Joo and Kim, 2011). Appearance, Chicken, the cheapest commercially produced meat in a taste, aroma, and texture of meat can generally persuade a global context, is supposed to have an increase in consumer’s decision to purchase meat. Flavour comprises consumption by 34% by 2018 (Jung et al., 2011). Being a mainly of taste and aroma and involves in consumers’ meat- white meat, chicken meat is more superior to red meat due purchasing behavior and preferences even before the meat to several other reasons, including its health benefits, as it is eaten (Shahidi, 1989; Sitz et al., 2005). It is well known contains less fat and cholesterol, easy to handle portions and that raw meat has only a blood-like taste with little or no less religious barriers (Liu et al., 2012). Few fast-growing aroma. Aromatic notes and most of the characteristic commercial broiler strains play the vital role in producing flavours responsible for the development of meat flavour the required amount of chicken meat to the world population (Jaturasitha et al., 2008). This is further * Corresponding Author: Cheorun Jo. Tel: +82-42-821-5774, supported by the production of meat from indigenous Fax: +82-42-825-9754, E-mail: [email protected] chicken breeds which have been neglected over the years. 1 Department of Animal Science, Iowa State University, Ames, IA However, the meat from indigenous chicken breeds is 50011-3150, USA. considered as a delicacy because of its unique taste and 2 Department of Animal Science and Technology, Sunchon texture compare to commercial broilers. As a result the National University, Suncheon, 540-742, Republic of Korea. Submitted Nov. 6, 2012; Accepted Jan. 14, 2013; Revised Jan. 26, 2013 price of the indigenous chicken meat is 2 or 3 times higher

Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742 733 than that of commercial broilers (Wattanachant et al., 2004; Choe et al., 2010). Chicken meat flavour, however, relies on several production and processing factors including the breed/strain of chicken, diet of bird, presence of free amino acids and nucleotides, irradiation, high pressure treatment, cooking, antioxidants, pH and ageing. Therefore these ante- and post- mortem factors can influence the status of chicken meat flavour. Additionally, chicken meat is more susceptible for quality deterioration mainly due to lipid oxidation and resulting off-flavours because chicken meat contains higher levels of unsaturated fatty acids compared to red meat. This off-flavour development is supposed to be one of the main problems regarding the quality of the chicken meat. As a result, the producers and processors in chicken meat sector, and even consumers try to avoid off-flavour development through various prevention mechanisms. Understanding the chemistry of chicken meat flavour is therefore vital in order to produce the most flavourful and consistent product possible. The intent of this paper is to highlight different Figure 1. Major classes of volatile compounds produced during mechanisms and chemical compounds responsible for the cooking of chicken meat (Mottram, 1998). chicken meat flavour and off-flavour development in detail and to brief the main factors affecting chicken meat flavour. (Perez-Alvarez et al., 2010). With respect to chicken meat, many of their key flavour GENERAL CHEMISTRY OF and odour compounds together with the mechanisms for the CHICKEN MEAT FLAVOUR formation have been identified (Aliani and Farmer, 2005). According to Shi and Ho (1994), 16 primary odour Bloody, metallic, and salty taste is generally a unique components have been identified in chicken broth, of which characteristic of fresh uncooked meat. Its aroma resembles 14 are structurally identified. They further demonstrated blood serum (Wasserman, 1972; Joo and Kim, 2011). that 2-methyl-3-furanthiol, generated from the Maillard However, significant changes take place in the flavour of reaction and lipid oxidation, as the most vital chemical meat during cooking. The main reactions involved during compound responsible for the meaty flavour of chicken cooking that are responsible for flavour developemnt are broth. In addition, other volatile compounds originated from Maillard reaction, thermal degradation of lipids and above two reactions include 2-furfurylthiol, methionol, Maillard-lipid interactions (Brunton et al., 2002). Flavour 2,4,5-trimethylthiazole, nonanol, 2-trans-nonenal, 2-formyl- gets developed during cooking through complex reactions 5-methylthiophene, p-cresol, 2-trans-4-trans-nonadienal, between components found in raw meat combining with 2-trans-4-trans-decadienal, 2-undecenal, -ionone, - heat. More than 1,000 chemicals have so far been identified decalactone and -dodecalactone. These compounds are in the volatiles of different muscle foods (Shahidi et al., obviously the major sources of chicken flavour (Shi and Ho, 1994). Majority of the volatile compounds identified in 1994; Varavinit et al., 2000). Sasaki et al. (2007) showed cooked poultry meat, have been recognized in chicken that components responsible for umami characteristics (Brunton et al., 2002). However many of these have little contribute to the brothy taste of meat. With respect to the influence on flavour of meat and relatively few make a key primary odourants of the broth, 2-trans-4-trans-decadienal contribution to the odour and flavour of cooked meat and -dodecalactone predominated in chicken broth (Aliani and Farmer, 2005). Melton (1999) has named sweet, compare to that of beef (Table 1). sour, salty, bitter and the “umami” or savory taste as the basic tastes of meat. Hydrocarbons, aldehydes, ketones, Flavour precursors of chicken meat alcohols, furans, thriphenes, pyrrols, pyridines, pyrazines, The major flavour precursors found in meat including oxazols, thiazols, sulfurous compounds (Figure 1), and chicken meat can be divided into two main groups: water many others have been identified as the flavour and aroma soluble components and lipids (Mottram, 1998). Although compounds found in meat (Ho et al., 1994; MacLeod, 1994). the flavour of meat from different species upon heat Therefore, sulfurous and carbonyl compounds are processing is expected to be similar due to the similarities considered to be the principal contributors to meat flavour of the free amino acids and carbohydrates in their meat

734 Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742

Table 1. Flavour dilution factors of odourants identified in broths furanthiol are pentose sugars and cysteine/cystine/ from chicken and beef (adapted from Gasser and Grosch, 1990) glutathione or thiamine. As pentose sugars are mainly Flavour dilution derived from ribonucleotides, in particular inosine-5’- Odour Compounds factor description monophosphate (IMP), 2-methyl-3-furanthiol and its Chicken Beef disulfide can be formed from reaction of IMP or ribose with 2-Methyl-3-furanthiol 1,024 512 Meat-like, sweet cysteine or cystine or glutathione or by thiamine bis (2-Methyl-3-furyl) <16 2,048 Meat-like degradation (Figure 2; Shi and Ho, 1994; Melton, 1999). disulphide Therefore, the relative importance of each precursor for 2-furfurylthiol 512 512 Roasty flavour generation in cooked chicken meat is still unclear 2,5-dimethyl-3-furanthiol 256 <16 Meaty (Aliani and Farmer, 2005). 3-mercapto-2-pentanone 128 32 Sulphurous Methionol 128 512 Cooked potato Lipid derived volatiles in chicken flavour 2,4,5-trimethylthiazole 128 <16 Earthy The role of lipid-derived carbonyl compounds in poultry 2-formyl-5-methylthiophene 64 64 Sulphurous flavour has been assessed by many researchers over the Phenylacetaldehyde 16 64 Honey-like years. The lean meat contains intramuscular triglycerides 2-trans-4-trans-decadienal 2,048 <16 Fatty and structural phospholipids. Therefore, desirable as well as 2-trans-4-cis-decadienal 128 <16 Fatty, tallowy undesirable flavours can be resulted in meat due to lipid 2-undecenal 256 <16 Tallowy, sweet oxidation. Mild thermal oxidative changes of lipids lead to -dodecalactone 512 <16 Tallowy, fruity generation of desirable flavour compounds and aromas in -decalactone 64 <16 Peach-like cooked meats (Shahidi, 2002). Nonanol 64 <16 Tallowy, green Similar to other meats, the flavour development of 2-trans-nonenal 64 <16 Tallowy, fatty poultry meat is partly attributed to its lipids (Perez-Alvarez 2-trans-4-trans-nonadienal 64 <16 Fatty et al., 2010). Several hundred volatile compounds are -ionone 64 <16 Violet-like generated in cooked meat through the lipid degradation, p-cresol 64 <16 Phenolic primarily the oxidation of the fatty acid components of lipids. Such compounds includes aliphatic hydrocarbons, (Shahidi, 2002), it was challenged by the lipid-derived aldehydes, alcohols, ketones, esters, carboxylic acids, some species-specific notes, mainly from the intramuscular lipids aromatic hydrocarbons (Figure 1), and oxygenated (Perez-Alvarez et al., 2010). In other words, differences in heterocyclic compounds such as lactones and alkylfurans lipid-derived volatile components between species are (Mottram, 1998). Forty-one out of 193 total compounds mainly responsible for the species differences in flavour, reported in the flavour of roasted chicken are lipid-derived whereas the precursors supplied by lean tissues generate the aldehydes. Selected aldehydes identified in roasted and meaty flavour common to all cooked meats (Mottram, cooked chicken flavour are shown in Table 2 (Shi and Ho, 1998). Mottram (1998) reported free sugars, sugar phosphates, nucleotide bound sugars, free amino acids, peptides, nucleotides and other nitrogenous components, such as thiamine as the main water-soluble flavour precursors. The reaction of cysteine and sugar can lead to characteristic meat flavour specially for chicken and pork (Varavinit et al., 2000). This was further confirmed by a research where the quantities of carbohydrates and amino acids, in particular ribose and cysteine, are reduced during heating. The main carbohydrates with flavour-forming potential include ribose, ribose-5-phosphate, glucose and glucose-6-phosphate (Meinert et al., 2009). Ribose which is associated with the ribonucleotides in the muscle is highly involved in flavour producing reactions during heating of meat (Mottram, 1998). However, many flavour compounds may be formed by two or more possible mechanisms. The best example is 2- methyl-3-furanthiol which is responsible for the meaty flavour of chicken broth (Shi and Ho, 1994; Aliani and Figure 2. Mechanism for formation of 2-methyl-3-furanthiol and Farmer, 2005). Principal precursors for 2-methyl-3- its disulfide by interaction of pentose sugars with cysteine or glutathione or by degradation of thiamine (Melton, 1999).

Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742 735

1994). According to the Table 2, hexanal and 2,4-decadienal are the most abundant aldehydes identified in chicken flavour which are known to be the primary oxidation products of linoleic acid (Figure 3). However, 2,4- decadienal is considered as a more important odourant for chicken flavour compared to hexanal due to its much lower odour threshold (Shi and Ho, 1994). Several studies have confirmed that phospholipids are much more important in the development of aroma volatiles during the cooking of meat than the triglycerides (Mottram, 1998). This is attributed to the presence of much higher proportion of unsaturated fatty acids, including significant amounts of polyunsaturated fatty acids such as arachidonic

Table 2. Selected aldehydes identified in roasted and cooked chicken flavour (adapted from Shi and Ho, 1994)

Concentration (mg per kg) Aldehyde Roasted Cooked chicken Figure 3. Oxidation and degradation products of linoleic acid (Shi chicken meat and Ho, 1994). Butanal 0.133 Pentanal 0.319 acid (20:4) in phospholipids (Mottram and Edwards, 1983). Hexanal 1.804 25.6 Saturated and unsaturated aldehydes having green, fatty and Heptanal 0.212 2.1 tallowy aroma play a vital role in all cooked meat aroma Octanal 0.422 2.3 (Mottram, 1998). The higher levels of unsaturated fatty Nonanal 0.467 1.7 acids in chicken compared with red meat generate more Decanal 0.052 0.3 unsaturated volatile aldehydes which are vital in Undecanal 0.058 determination of specific aromas of chicken meat (Noleau Dodecanal 0.022 and Toulemonde, 1987). Further, the aroma of fat-fried food Tridecanal 0.151 is reported to be due to 2,4-decadienal (Mottram, 1998). Tetradecanal 0.125 0.2 Therefore aliphatic aldehydes contribute to the fatty Pentadecanal 0.383 flavours of cooked meat including chicken meat. Hexadecanal 19.788 1.4 A different composition of volatiles that is responsible Heptadecanal 0.276 0.1 for the desirable flavours gets formed due to quick reactions Octadecanal 2.664 taken place in cooked meat (Mottram, 1998) compare to trans-2-butenal tr those volatiles formed during long-term storage leading to cis-2-pentenal tr rancid off-flavours. In chicken meat, lack of -tocopherol is trans-2-pentenal 0.085 1.1 however considered as the main reason for meat flavour cis-2-hexenal tr deterioration and formation of undesirable “warmed over trans-2-hexenal 0.060 0.3 flavour (WOF)” in chicken meat products (Shahidi, 2002). trans-2-heptenal 0.104 1.2 Lipid-derived compounds in meat volatiles have greater cis-2-octenal 0.004 odour threshold values (Table 3) as opposed to sulfur- and trans-2-octenal 0.195 3.7 nitrogen-containing heterocyclic compounds which make trans-2-nonenal 0.084 them less aroma significant. Even at relatively low cis-2-decenal 0.003 concentrations, heterocyclic compounds possess a trans-2-decenal 0.139 1.0 significant effect on aroma due to their low odour threshold cis-2-undecenal 0.002 values (Mottram, 1998). trans-2-undecenal 0.139 0.4 trans-dodecenal 0.002 0.3 Volatiles from the Maillard reaction in chicken flavour trans,cis-2,4-nonadienal tr The Maillard reaction is one of the main chemical trans,trans-2,4-nonadienal tr 0.3 reactions that take place during cooking of meat including trans,cis-2,4-decadienal 0.051 1.0 chicken meat. This typically occurs between amino trans,trans-2,4-decadienal 0.137 5.2 compounds and reducing sugars and eventually results a trans,trans-2,4-undecadienal 0.001 0.2 large number of compounds responsible for the flavour of tr = trace.

736 Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742

Table 3. Odour thresholds of some volatiles identified in boiled with these compounds forming intermediates that further meat (adapted from Gasser and Grosch, 1990) involve in flavour-forming reactions. This eventually Compound Threshold (ng/L; air) produces many important classes of flavour compounds 2-methyl-3-furanthiol 0.0025-0.001 including furans, pyrazines, pyrroles, oxazoles, thiophenes, Bis (2-methyl-3-furyl)disulphide 0.0007-0.0028 thiazoles and other heterocyclic compounds (Figure 5; 2-furfurylthiol 0.0045-0.002 Melton, 1999). Shahidi (1989) and Mottram and Madruga 2,5-dimethyl-3-furanthiol 0.0035-0.014 (1994) reported that sulfur-compounds derived from ribose 3-mercapto-2-pentanone 0.045-0.18 and cysteine, and carbonyl compounds are the principal 2,4,5-trimethylthiazole 1.8-7.2 contributors to meat flavour. Ribose is considered as the 2-formyl-5-methylthiophene 1.75-7.4 most important flavour precursor in chicken (Meinert et al., 2,4-decadienal 0.04-0.16 2009). In addition, thiamine has also been proved as an important precursor that provides wide range of sulfur any meat (Mottram, 1994a). During the initial stages of this compounds (Aliani and Farmer, 2005). Due to the reaction, Amadori products are formed via glycosylamine as degradation process of nucleotides such as IMP, ribose is a result of condensation of the carbonyl group of a reducing formed and it is then involved in a number of secondary sugar with amino compounds. Various sugar dehydration reactions yielding a large number of volatile compounds and degradation products such as furfural and furanone (Kavitha and Modi, 2007). Hence, IMP is generally derivatives, hydroxyketones and dicarbonyl compounds are considered as the major nucleotide in muscle that imparts formed by rearranging and dehydration of the resulted flavour to the meat (Yamaguchi, 1991). In general, the IMP product via deoxyosones. Subsequently these compounds content in chicken meat is 75 to 122 mg/100 g (Maga, 1983) interact with other reactive components such as amines, but some differences among breeds exist (Jung et al., 2011). amino acids, aldehydes, hydrogen sulfide, and ammonia and Formation of 2-methyl-3-furanthiol in chicken broth via the aroma compounds are formed (Mottram, 1998; Calkins Maillard reaction involves interaction between ribose and and Hodgen, 2007). sulphur-containing amino acids (cysteine or cystine) or Strecker degradation of amino acids by Maillard peptide (glutathione). Glutathione liberates hydrogen reaction-derived dicarbonyl compounds is a vital associated sulphide rapidly during the initial stages of cooking whereas reaction, during which an aldehyde is formed due to the cysteine does upon prolonged heating. Another important decarboxylation and deamination of an amino acid and an odourant in chicken broth, 2-furfurylthiol, is formed from -aminoketone or amino alcohol is resulted from the the reaction between furfural and cysteine (Shi and Ho, dicarbonyl compound as well. Hydrogen sulfide, ammonia 1994). In addition, the Maillard reaction-derived volatiles and acetaldehyde are also formed by Strecker degradation are the major components in meat grilled under severe when cysteine is used as the amino acid (Figure 4; Mottram, conditions (Mottram, 1998). 1998). Maillard reaction-derived carbonyl compounds act Volatiles compounds from lipid-Maillard interactions Whitfield (1992) and Mottram (1994b) reported that the interaction of lipid with the Maillard reaction leads to formation of a number of volatiles that have been identified in meat (Figure 6). Volatile compounds originating from

Figure 4. Strecker degradation of amino acids and the formation of hydrogen sulfide, ammonia and acetaldehyde from cysteine Figure 5. Mechanism for the formation of thiazolines and (Mottram, 1998). thiazoles in the Maillard reaction (Mottram, 1998).

Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742 737

Table 5. Occurrence of some long-chain alkylthiazoles in meat (adapted from Melton, 1999) Beef Chicken Thiazole Beef heart Longissimus breast dorsi 5-Octyl-4-ethyl -1 tr2 5 5-Nonyl-ethyl 3 4  5-Decyl-4-ethyl - tr  2-Tridecyl-4,5-dimethyl  tr - 2-Tridecyl-4/5-ethyl  - - 2-Tetradecyl-4,5-dimethyl  - -

2-Pentadecyl tr tr  Figure 6. Volatile compounds of lipid-Maillard interaction 2-Pentadecyl-4-methyl    (Melton, 1999). 2-Pentadecyl-4/5-ethyl    1 Absent. 2 Trace. 3 Slight. 4 Moderate. 5 Abundant. lipid-Maillard interactions are given in Table 4 and 5. Several thiazoles with C to C n-alkyl substituents in the 4 8 trithiolanes, trithianes that have low odour thresholds with 2-position and some other alkylthiazoles with much longer sulfurous, onion-like and, sometimes, meaty aromas (Fors, 2-alkyl substituents (C to C ) have been reported in fried 13 15 1983). These compounds contributed to the overall flavour chicken and heated chicken, respectively (Tang et al., 1983; and aroma of boiled meat. Farmer and Mottram, 1994). Melton (1999) reported that 2- Thermal degradation of cysteine and glutathione results octyl-4,5-dimethylthiazole found in chicken meat is also a in two major volatile compounds of fried chicken. At frying product from interaction of Maillard reactions and lipids. temperature (180C), cysteine produced 3,5-dimethyl-1,2,4-

trithiolanes and 2,4,6-trimethylperhydro-1,3,5-dithiazines Other compounds contributing to roasted, fried and (thialdine) whereas only the former compound was boiled chicken flavour produced by glutathione at the same condition. These Heterocyclic compounds such as pyrazines, thiazoles volatiles were identified in cooked chicken as well (Shi and and oxazoles are usually considered to be responsible for Ho, 1994). In addition to 3,5-dimethyl-1,2,4-trithiolanes, the roast flavours in foods including meat. Melton (1999) three other alkyl-substituted trithiolanes, 3,5-diisobutyl- reported that a large number of heterocyclic compounds are 1,2,4-trithiolane, 3-methyl-5-butyl-1,2,4-trithiolane and 3- associated with roasted, grilled, fried or pressure cooked methyl-5-pentyl-1,2,4-trithiolane, were identified in fried meats, but not boiled meat, due to higher temperatures used chicken flavour. The formation of latter 2 volatiles involves in those cooking methods. Different alkyl pyrazines and two thermal and oxidative degradation of lipids, pentanal and classes of bicyclic compounds, 6,7-dihydro-5(H)- hexanal (Shi and Ho, 1994). Possible mechanisms for the cyclopentapyrazines and pyrrolopyrazines, were found in formation of dithiazines and 3,5-diisobutyl-1,2,4-trithiolane meat volatiles (Mottram, 1998). It was noticed that both are given in Figure 7. According to Shi and Ho (1994), classes of compounds increased greatly with the increasing flavour of deep-fat-fried foods such as fried chicken is severity of heat treatment. However, Mottram (1991) attributed to lipid-derived aldehyde, 2,4-decadienal. Being reported that boiled meat contained higher levels of sulfur- an oxidation product of 2,4-decadienal, trans-4,5-epoxy- containing heterocyclic compounds such as thiophenes, trans-2-decenal having a low odour threshold plays a key Table 4. Selected lipid-Maillard product in cooked meat (adapted role in the flavour of fried chicken. from Melton, 1999) Heterocyclic compounds, mainly pyrazines, pyridines, Compound Beef Chicken Turkey Lamb pyrroles and thiazoles, found in fried and roasted chicken 1-Heptanethiol D1 - - - are listed in Table 6. Alkylpyrazines have a roasted, nut-like 2-Pentylpyridine D, R2 D, R R R or toasted flavour notes and are present in the flavours of 2-Buthylthiophene R R - - fried chicken and roasted chicken but not in chicken broth. 2-Hexylthiophene R D R - This confirms that the generation of pyrazines requires high 2-Pentylthiapyran D ND3 - - temperature and low moisture. Shi and Ho (1994) reported 2-Alkyl-3-formyldihydro-thiophenes ND ND - - that fried chicken flavour was further intensified by 2-Propyl-3-formyldihydro-thiophene ND D - - 2-pentylpyridine (strong fatty and tallow-like odour), 2-Butyl-3-formyldihydro-thiophene ND D - - 2-isobutyl-3,5-diisopropylpyridine (roasted cocoa-like 2-Hexyl-3-formyldihydro-thiophene D D - - aroma), 2-pentyl-4-methyl-5-ethylthiazole (strong paprika 1 D = Detected. 2 R = Reported in literature. 3 ND = Not detected. pepper flavour), 2-heptyl-4,5-dimethylthiazole (strong spicy

738 Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742

Table 6. Heterocyclic compounds identified in fried and roasted chicken flavours (adapted from Shi and Ho, 1994) Fried Roasted Compound Chicken Chicken Pyrazines Pyrazine  2-Methylpyrazine   2,3-Dimethylpyrazine   2,5-Dimethylpyrazine  2,6-Dimethylpyrazine   Trimethylpyrazine   2-Isopropylpyrazine  2-Methyl-3-ethylpyrazine   2-Methyl-6(5)-ethylpyrazine   2-Butylpyrazine  2,3-Dimethyl-5-ethylpyrazine   2,5-Dimethyl-3-ethylpyrazine  2,6-Dimethyl-3-ethylpyrazine  2,6-Diethylpyrazine   2-Methyl-5,6-diethylpyrazine  2-Methyl-3,5-diethylpyrazine  2-Methyl-3-butylpyrazine  2-Methyl-5-vinylpyrazine  Figure 7. Possible mechanisms for the formation of (a) dithiazines 2-Methyl-6-vinylpyrazine  and (b) 3,5-diisobutyl-1,2,4-trithiolane (Shi and Ho, 1994). 2-Isopropenylpyrazine  6,7-Dihydro-5H-cyclopentapyrazine  flavour) and 2-octyl-4,5-dimethylthiazole (sweet fatty 2-Methyl-6,7-dihydro-5H-  aroma). cyclopentapyrazine Pyridines Pyridine  

2-Methylpyridine   OFF-FLAVOURS OF CHICKEN MEAT 3-Ethylpyridine   4-Ethylpyridine  Lipid oxidation has been considered as the primary 2-Methyl-5-ethylpyridine   cause of flavour deterioration and development of off- 2-Ethyl-3-methylpyridine  2-Butylpyridine flavour known as oxidized flavours or ‘‘warmed-over  2-Pentylpyridine  flavour (WOF)’’ in poultry meat (Lillard, 1987; Shi and Ho, 2-Isobutyl-3,5-dipropylpyridine  1994). Generally WOF is generated within 24 h of Pyrroles Pyrrole   refrigerated storage in precooked poultry (Graf and Panter, N-methylpyrrole  1991). WOF development is also related to rancid, and 2-Methylpyrrole  sulfur/rubber sensory notes and a parallel reduction in 2-Ethylpyrrole  N-acetylpyrrole  chicken “meaty” flavour can also be observed. A 2-Acetylpyrrole  remarkable off-flavour problem in mechanically deboned 2-Isobutylpyrrole  chicken meat was reported by Shi and Ho (1994) and it was N-isobutylpyrrole  attributed to rancidity of fat. It was further stated by the N-(2-butanoyl)pyrrole  same authors that haem pigments, being the catalyst of the N-furfurylpyrrole  Thiazoles Thiazole   above reaction, are largely responsible for the off-flavour 2-Methylthiazole   formation. However, recent studies indicated that raw 2,4,5-Trimethylthiazole  poultry meats are highly resistant to oxidative changes due 2-Methyl-4-ethylthiazole  to various antioxidants present in chicken meat (Min et al., 2-Methyl-5-ethylthiazole  2010). Meanwhile, it has also been reported that increasing 2,4-Dimethyl-5-ethylthiazole  2-Isopropyl-4,5-dimethylthiazole  cooking temperature is associated with increased roasted, 2,5-Dimethyl-4-butylthiazole  toasted, and bitter sensory notes (Perez-Alvarez et al., 2010). 2-Isopropyl-4-ethyl-5-methylthiazole  Studies conducted to determine the critical changes in odour 2-Butyl-4,5-dimethylthiazole  compounds of boiled chicken during refrigerated storage 2-Butyl-4-methyl-5-ethylthiazole  and reheating proved that refrigerated storage and reheating 2-Pentyl-4,5-dimethylthiazole  2-Hexyl-4,5-dimethylthiazole  of boiled chicken showing WOF due to loss of meaty, 2-Heptyl-4,5-dimethylthiazole  chicken-like and sweet odour notes, and the formation of 2-Heptyl-4-ethyl-5-methylthiazole  green, cardboard-like, and metallic off-odours by the 2-Octyl-4,5-dimethylthiazole  secondary by-products of lipid oxidation. These changes + = Present

Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742 739 were caused primarily by an increase in hexanal (sevenfold) male birds. Many other researchers reported no significant and six fold decrease in both (E,E)-2,4-decadienal and 2- relationship between the two parameters (Farmer, 1999). furfurylthiol (Kerler and Grosch, 1997). Rhee et al. (2005) In addition to these intramuscular compounds, the diet reported that “cardboard’’ flavour and ‘‘sour’’ flavour of the bird also plays a vital role towards the flavour of intensity increased with storage time for chicken breast, chicken meat (Fanatico et al., 2007; Perez-Alvarez et al., while ‘‘cooked chicken’’ intensity decreased and, 2010). Diet can either positively or negatively influence the “cardboard’’ intensity increased for chicken thigh. Results flavour of chicken meat. Corn-fed chicken and arachidonic further illustrated relative storage effects on intensity of the acid enriched diet fed chicken have produced tastier meat species specific natural meat flavours and “cardboard’’ (Lyon et al., 2004; Kiyohara et al., 2011; Takahashi et al., flavour, with d-3 and d-6 scores expressed relative to d-0 2012) while diets supplemented with fish meal have scores. However, chicken exhibited the slowest rate of negatively affected the flavour of chicken meat (Poste, decrease in species-specific natural flavour intensity and the 1990). Processing steps such as aging, cooking, irradiation slowest rate of increase in “cardboard’’ flavour intensity and high pressure treatment also affect the flavour of compare to pork and beef. chicken meat. Post-mortem aging results in many chemical flavour compounds including sugars, organic acids, FACTORS AFFECTING peptides, free amino acids (Yano et al., 1995; Spanier et al., CHICKEN MEAT FLAVOUR 2004) and thereby leads to increased flavour. Cooking plays a vital role in flavour development and it affects the Findings of researches conducted over years have acceptability and volatile flavour components of poultry shown that several pre- and post-mortem factors affect the meat (Sanudo et al., 2000). Cooking methods such as flavour of chicken meat. Breed/strain of the chicken, diet of roasting, grilling, frying, and pressure cooking generates the bird, presence of free amino acids and nucleotides, many pyrazines, pyridines, pyrroles and thiazoles compared irradiation, high pressure treatment, cooking, antioxidants, to boiling of chicken meat (Shi and Ho, 1994). Irradiation pH, sex, and ageing are considered as the main affects flavour and aroma of chicken meat primarily determinants (Farmer, 1999; Jayasena et al., 2013). through the production of free radicals. Aldehydes (hexanal, Different breeds/strains contain different levels of flavour pentanal, heptanal, octanal, and nonanal) and sulfur precursors such as IMP leading to various types and volatiles mainly dimethyl trisulfide generated during concentrations of volatile compounds. Increased palatability irradiation result in the associated off-odour (Patterson and of indigenous chicken compared to broilers is well Stevenson, 1995; Perez-Alvarez et al., 2010). However, documented by many authors. Superior flavour of Korean very little negative effect was expressed by electron beam native/farm chicken, Hinai-jidori chicken (Japan), irradiation on the flavour of preheated chicken breast meat Wenchang and Xianju (China) to broilers has been reported (Rababah, 2006). Effect of high pressure treatment on (Tang et al., 2009; Jung et al., 2011; Kiyohara et al., 2011). flavour of meat including chicken meat had been variable Additionally, chicken results in more unsaturated volatile over the years. Hayman et al. (2004) showed that the aldehydes as compared to beef or lamb because their sensory quality of various meat products was not affected muscle contains higher levels of polyunsaturated fatty acids by high pressure treatment. However, exposure of chicken in the triglycerides than later species (Calkins and Hodgen, meat to a pressure of 300 MPa lead to a better flavour and 2007). These compounds contribute to the specific aromas taste as opposed to a 450 MPa treatment (Kruk et al., 2011). of chicken (Noleau and Toulemonde, 1987; Mottram, 1991). According to Cheah and Ledward (1996), pressure This confirms the effect of lipids on the flavour of chicken treatments at around 300 MPa at room temperature initiated primarily via the differences in fatty acid composition and the changes which eventually lead to catalysis of lipid the resulting carbonyls (Perez-Alvarez et al., 2010). Further, oxidation in pressure processed meat. positive correlations between the flavour of chicken meat and the intramuscular contents of amino acids, including CONCLUSION glutamic acid and nucleotides, such as IMP were demonstrated by Kurihara (1987), Rikimaru and Takahashi The Maillard reaction, thermal degradation of lipids and (2010) and Takahashi et al. (2012). Effect of sex on chicken Maillard-lipid interactions, are the main pathways by which meat flavour was demonstrated by many researchers, a large number of flavour and aroma compounds although the results were not consistent. Meat from male responsible for chicken meat flavour are formed during birds received higher scores for flavour as opposed to that cooking. Volatile compounds generated from the Maillard from female birds (Ramaswamy and Richards, 1982; reaction and lipid oxidation including 2-methyl-3-furanthiol, Farmer, 1999). However, it was also shown that the breast 2-furfurylthiol, methionol, 2,4,5-trimethylthiazole, nonanol, and leg meat of female birds were preferred to those of 2-trans-nonenal, 2-formyl-5-methylthiophene, p-cresol, 2-

740 Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742 trans-4-trans-nonadienal, 2-trans-4-trans-decadienal, 2- Farmer, L. J. and D. S. Mottram. 1994. Lipid-Maillard interactions undecenal, -ionone, -decalactone and -dodecalactone are in the formation of volatile aroma compounds. In: Trends in obviously the major sources of chicken flavour. However 2- Flavour Research (Ed. H. Maarse and D. G. vander Heij). Elsevier, Amsterdam. pp. 313-326. methyl-3-furanthiol that is produced from the reaction Fors, S. 1983. Sensory properties of volatile Maillard reaction between ribose and cysteine or cystine, and from products and related compounds. In: The Maillard Reaction in degradation of thiamin is considered as the most important Foods and Nutrition (Ed. G. R. Waller and M. S. Feather). compound in chicken flavour. American Chemical Society, Washington. pp. 185-286. A large number of heterocyclic compounds are formed Gasser, U. and W. Grosch. 1990. Primary odorants of chicken during roasting, grilling, frying or pressure cooking of broth: A comparative study with meat broths from cow and ox. chicken meat due to higher temperature and low moisture Z. Lebensm. Unters. Forsch. 190:3-8. conditions used in those cooking methods. These Graf, E. and S. S. Panter. 1991. Inhibition of warmed-over flavour compounds are absent in boiled meat. Major volatile development by polyvalent cations. J. Food Sci. 56:1055-1058. Hayman, M. M., I. Baxter, P. J. O’Riordan and C. M. Stewart. compounds responsible for fried chicken are 3,5-dimethyl- 2004. Effects of high pressure processing on the safety, quality, 1,2,4-trithiolanes, 2,4,6-trimethylperhydro-1,3,5-dithiazines, and shelf life of ready-to-eat meats. J. Food Prot. 67:1709- 3,5-diisobutyl-1,2,4-trithiolane, 3-methyl-5-butyl-1,2,4- 1718. trithiolane, 3-methyl-5-pentyl-1,2,4-trithiolane, 2,4- Ho, C. T., Y. C. Oh and M. Bae-Lee. 1994. The flavour of pork. In: decadienal and trans-4,5-epoxy-trans-2-decenal. Flavour of Meat and Meat Products (Ed. F. Shahidi). Blackie Alkylpyrazines were reported in the flavours of fried Academic and Professional, London. pp. 38-51. chicken and roasted chicken but not in chicken broth. Jaturasitha, S., T. Srikanchai, M. Kreuzer and M. Wicke. 2008. Flavour of chicken meat is affected by breed/strain of Difference in carcass and meat characteristics between chicken chicken, diet of bird, presence of free amino acids and indigenous to northern Thailand (Black-boned and Thai native) and imported extensive breeds (Bresse and Rhode Island Red). nucleotides, irradiation, high pressure treatment, cooking, Poult. Sci. 87:160-169. antioxidants and ageing. Jayasena, D. D., D. U. Ahn, K. C. Nam and C. Jo. 2013. Factors affecting cooked chicken meat flavor: A review. Worlds Poult. ACKNOWLEDGEMENT Sci. J. (In press). Joo, S. T. and G. D. Kim. 2011. Meat quality traits and control This work was carried out with the support of technologies. In: Control of Meat Quality (Ed. S. T. Joo). “Cooperative Research Program for Agriculture Sciecne & Research Signpost, Kerala: pp. 1-20. Technology Development (Project No. PJ90701104)” Rural Jung, Y., H. J. Jeon, S. Jung, J. H. Choe, J. H. Lee, K. N. Heo, B. S. Development Administration, Korea. Kang and C. Jo. 2011. Comparison of quality traits of thigh meat from Korean native chickens and broilers. Korean J.

Food Sci. Anim. Resour. 31:684-692. REFERENCES Kavitha, S. and V. K. Modi. 2007. Effect of water activity and temperature on degradation of 5’-inosine monophosphate in a Aliani, M. and L. J. Farmer. 2005. Precursors of chicken flavour II: meat model system. LWT-Food Sci. Technol. 40:1280-1286. Identification of key flavour precursors using sensory methods. Kerler, J. and W. Grosch. 1997. Character impact odourants of J. Agric. Food Chem. 53:6455-6462. boiled chicken: changes during refrigerated storage and Brunton, N. P., D. A. Cronin and F. J. Monahan. 2002. Volatile reheating. Eur. Food Res. Technol. 205:232-238. components associated with freshly cooked and oxidized off- Kiyohara, R., S. Yamaguchi, K. Rikimaru and H. Takahashi. 2011. flavours in turkey breast meat. Flavour Fragr. J. 17:327-334. Supplemental arachidonic acid-enriched oil improves the taste Calkins, C. R. and J. M. Hodgen. 2007. A fresh look at meat of thigh meat of Hinai-jidori chickens. Poult. Sci. 90:1817- flavour. Meat Sci. 77:63-80. 1822. Cheah, P. B. and D. A. Ledward. 1996. High pressure effects on Kruk, Z. A., H. Yun, D. L. Rutley, E. J. Lee, Y. J. Kim and C. Jo. lipid oxidation in minced pork. Meat Sci. 43:123-134. 2011. The effect of high pressure on microbial population, Choe, J. H., K. Nam, S. Jung, B. Kim, H. J. Yun and C. Jo. 2010. meat quality and sensory characteristics of chicken breast fillet. Differences in the quality characteristics between commercial Food Control 22:6-12. Korean native chickens and broilers. Korean J. Food Sci. Anim. Kurihara, K. 1987. Recent progress in the taste receptor. In: Resour. 30:13-19. Umami: A Basic Taste (Ed. Y. Kawamura and M. R. Kare). Fanatico, A. C., P. B. Pillai, J. L. Emmert, E. E. Gbur, J. F. Marcel Dekker, New York. pp. 3-39. Meullenet and C. M. Owens. 2007. Sensory attributes of slow- Lillard, D. A. 1987. Oxidative deterioration in meat, poultry, and and fast-growing chicken genotypes raised indoors or with fish. In: Warmed-Over Flavour of Meat (Ed. A. J. St. Angelo outdoor access. Poult. Sci. 86:2441-2449. and M. E. Bailey). Academic Press, Orlando. pp. 41-67. Farmer, L. J. 1999. Poultry meat flavour. In: Poultry Meat Science: Liu, X. D., D. D. Jayasena, Y. Jung, S. Jung, B. S. Kang, K. N. Poultry Science Symposium Series Vol. 25 (Ed. R. I. Heo, J. H. Lee and C. Jo. 2012. Differential proteome analysis Richardson and G. C. Mead). CABI Publishing, Oxon. pp. of breast and thigh muscles between Korean native chickens 127-158. and commercial broilers. Asian Australas. J. Anim. Sci.

Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742 741

25:895-902. plant extracts and subjected to electron beam irradiation. Poult. Lyon, B. G., D. P. Smith, C. E. Lyon and E. M. Savage. 2004. Sci. 85:1107-1113. Effects of diet and feed withdrawal on the sensory descriptive Ramaswamy, H. S. and J. F. Richards. 1982. Flavour of poultry and instrumental profiles of broiler breast fillets. Poult. Sci. meat- a review. Can. Inst. Food Sci. Technol. J. 15:7-18. 83:275-281. Rhee, K. S., L. M. Anderson and A. R. Sams. 2005. Comparison of MacLeod, G. 1994. The flavour of beef. In: Flavour of Meat and flavour changes in cooked–refrigerated beef, pork and chicken Meat Products (Ed. F. Shahidi). Blackie Academic and meat patties. Meat Sci. 71:392-396. Professional, Glasgow. pp. 4-37. Rikimaru, K. and H. Takahashi. 2010. Evaluation of the meat from Maga, J. A. 1983. Flavour potentiator. Crit. Rev. Food Sci. Nutr. Hinai-jidori chickens and broilers:Analysis of general 18:231-312. biochemical components, free amino acids, inosine 5’- Meinert, L., A. Schäfer, C. Bjergegaard, M. D. Aaslyng and W. L. monophosphate, and fatty acids. J. Appl. Poult. Res. 19:327- Bredie. 2009. Comparison of glucose, glucose 6-phosphate, 333. ribose, and mannose as flavour precursors in pork; the effect of Sanudo, C., M. E. Enser, M. M. Campo, G. R. Nute, G. Maria, I. monosaccharide addition on flavour generation. Meat Sci. Sierra and J. D. Wood. 2000. Fatty acid composition and 81:419-425. sensory characteristics of lamb carcasses from Britain and Melton, S. L. 1999. Current status of meat flavour. In: Quality Spain. Meat Sci. 54:339-346. Attributes of Muscle Foods (Ed. Y. L. Xiong, C. T. Ho and F. Sasaki, K., M. Motoyama and M. Mitsumoto. 2007. Changes in Shahidi). Kluwer Academic/ Plenum Publishers, New York. pp. the amounts of water-soluble umami-related substances in 115-130. porcine longissimus and biceps femoris muscles during moist Min, B., J. C. Cordray and D. U. Ahn. 2010. Effect of NaCl, heat cooking. Meat Sci. 77:167-172. myoglobin, Fe(II), and Fe(III) on lipid oxidation of raw and Shahidi, F. 1989. Flavour of cooked meats. In: Flavour Chemistry: cooked chicken breast and beef loin. J. Agric. Food Chem. Trends and Developments (Ed. R. Teranishi, R. G. Buttery and 58:600-605. F. Shahidi). American Chemical Society, Washington. pp. 188- Mottram, D. S. 1991. Meat. In: Volatile Compounds in Foods and 201. Beverages (Ed. H. Maarse). Marcel Dekker, New York. pp. Shahidi, F. 1994. Flavour of meat and meat products-an overview. 107-177. In: Flavour of Meat and Meat Products (Ed. F. Shahidi). Mottram, D. S. 1994a. Flavour compounds formed during the Blackie Academic and Professional, Glasgow. pp. 1-3. Maillard reaction. In Thermally Generated Flavours. Maillard, Shahidi, F. 2002. Lipid derived flavours in meat products. In: Meat Microwave, and Extrusion Processes (Ed. T. H. Parliament, M. Processing: Improving Quality (Ed. J. Kerry, J. Kerry and D. J. Morello and R. J. McGorrin). American Chemical Society, Ledward). Woodhead Publishing Ltd, Cambridge. pp. 105-121. Washington. pp. 104-126. Shi, H. and C. T. Ho. 1994. The flavour of poultry meat. In: Mottram, D. S. 1994b. Some aspects of the chemistry of meat Flavour of Meat and Meat Products (Ed. F. Shahidi). Blackie flavour. In: The Flavour of Meat and Meat Products (Ed. F. Academic and Professional, Glasgow. pp. 52-69. Shahidi). Chapman and Hall, London. pp. 210-230. Sitz, B. M., C. R. Calkins, D. M. Feuz, W. J. Umberger and K. M. Mottram, D. S. 1998. Flavour formation in meat and meat Eskridge. 2005. Consumer sensory acceptance and value of products: a review. Food Chem. 62:415-424. domestic, Canadian, and Australian grass-fed beef steaks. J. Mottram, D. S. and R. A. Edwards. 1983. The role of triglycerides Anim. Sci. 83:2863-2868. and phospholipids in the aroma of cooked beef. J. Sci. Food Spanier A. M., M. Flores, F. Toldrá, M. C. Aristoy, K. L. Bett, P. Agric. 34:517-522. Bystricky and J. M. Bland. 2004. Meat flavor: contribution of Mottram, D. S. and M. S. Madruga. 1994. Important sulfur proteins and peptides to the flavor of beef. Adv. Exp. Med. containing aroma volatiles in meat. In: Sulfur Compounds in Biol. 542:33-49. Foods (Ed. C. J. Mussinan and M. E. Keelan). American Spanier, A. M., M. Flores, K. W. Mcmillin and T. D. Bidner. 1997. Chemical Society, Washington. pp. 180-187. The effect of postmortem aging on meat flavour quality. Noleau, I. and B. Toulemonde. 1987. Volatile components of Correlation of treatment, sensory, instrumental, and chemical roasted chicken fat. LWT-Food Sci. Technol. 20:37-41. descriptors. Food Chem. 59:531-538. Patterson, R. L. and M. H. Stevenson. 1995. Irradiation-induced Takahashi, H., K. Rikimaru, R. Kiyohara and S. Yamaguchi. 2012. off-odor in chicken and its possible control. Br. Poult. Sci. Effect of arachidonic acid-enriched oil diet supplementation on 36:425-441. the taste of broiler meat. Asian Australas. J. Anim. Sci. Perez-Alvarez, J. A., E. Sendra-Nadal, E. J. Sanchez-Zapata and M. 25:845-851. Viuda-Martos. 2010. Poultry flavour: General aspects and Tang, H., Y. Z. Gong, C. X. Wu, J. Jiang, Y. Wang and K. Li. 2009. applications. In: Handbook of Poultry Science and Technology Variation of meat quality traits among five genotypes of Volume 2: Secondary Processing (Ed. I. Guerrero-Legarreta chicken. Poult. Sci. 88:2212-2218. and Y. H. Hui). John Wiley and Sons Inc, New Jersey. pp. 339- Tang, J., Q. Z. Jin, G. H. Shen, C. T. Ho and S. S. Chang. 1983. 357. Isolation and identification of volatile compounds from fried Poste, L. M. 1990. A sensory perspective of effect of feeds on chicken. J. Agric. Food Chem. 31:1287-1292. flavor in meats: Poultry meats. J. Anim. Sci. 68:4414-4420. Varavinit, S., S. Shobsngob, M. Bhidyachakorawat and M. Rababah, T., N. S. Hettiarachchy, R. Horax, M. J. Cho, B. Davis Suphantharika. 2000. Production of meat-like flavour. Science and J. Dickson. 2006. Thiobarbituric acid reactive substances Asia. 26:219-224. and volatile compounds in chicken breast meat infused with Wasserman, A. E. 1972. Thermally produced flavour components

742 Jayasena et al. (2013) Asian Australas. J. Anim. Sci. 26:732-742

in the aroma of meat and poultry. J. Agric. Food Chem. Yamaguchi, S. 1991. Roles and efficacy of sensory evaluation in 20:737-741. studies of taste. J. Japan Soc. Food Sci. Technol. 38:972-978. Wattanachant, S., S. Benjakul and D. A. Ledward. 2004. Yano, T., N. Kataho, M. Watanabe, T. Nakamura and Y. Asano. Composition, color, and texture of Thai indigenous and broiler 1995. Evaluation of beef aging by determination of chicken muscles. Poult. Sci. 83:123-128. hypoxanthine and xanthine contents: application of a xanthine Whitfield, F. B. 1992. Volatiles from interactions of Maillard sensor. Food Chem. 52:439-445. reactions and lipids. Crit. Rev. Food Sci. Nutr. 31:1-58.

Food & Function

View Article Online PAPER View Journal | View Issue

Formation and distribution of 2,4-dihydroxy-2,5- dimethyl-3(2H)-thiophenone, a pigment, an aroma and Food Funct. Cite this: , 2013, 4, 1076 a biologically active compound formed by the Maillard reaction, in foods and beverages

Rina Furusawa, Chiaki Goto, Miki Satoh, Yuri Nomi and Masatsune Murata*

We recently identiﬁed 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) from soy sauce as a low- molecular-weight pigment formed by the Maillard reaction. DHDMT has also been reported as an aroma compound in a model system and a biologically active compound of heated garlic. To utilize these functions eﬃciently, we here examined how DHDMT was formed during fermentation of soy Received 6th December 2012 sauce and in model systems. Although DHDMT was formed from cysteine and glucose, it was formed Accepted 10th March 2013 more from cystine and fructose in the model system. We also showed that this compound exists in DOI: 10.1039/c3fo30367e various kinds of soy sauce and miso as well as in some brown foods and beverages such as roasted www.rsc.org/foodfunction bread and beer.

Introduction sauce is mainly attributable to melanoidins which are formed by the Maillard reaction and are brown high-molecular-weight Soy sauce is made from soybean, wheat, NaCl and koji heterogeneous polymers. Flazine, a b-carboline derivative, has (Aspergillus oryzae or Aspergillus soyae). Macromolecules of the been reported as a low-molecular-weight pigment by Kihara, raw materials are decomposed during fermentation to such low- who described the relationship between its oxidation and the molecular-weight compounds as amino acids and sugars by the browning of soy sauce.7,8 However, its contribution to the color action of enzymes of koji in the presence of salt. Soy sauce is rich of soy sauce is unclear. in salt, glutamic acid, and other amino acids, and is one of the Based on these backgrounds, our group tried to identify a major traditional seasonings used in Japan. The color of soy low-molecular-weight pigment in soy sauce and recently iden- Published on 28 March 2013. Downloaded by FAC DE QUIMICA 12/11/2013 23:47:20. sauce is an important factor of its quality and is formed during tied 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) the production process involving heating, fermentation and as the major lipophilic and low-molecular-weight pigment in pasteurization by the Maillard reaction. The color of soy sauce soy sauce.9 Although its contribution to the total color of soy has aroused Japanese researchers' interests. In 1926, Kurono sauce was very low, this nding was considered to be mean- and Katsume partly puried soy sauce pigments by the precip- ingful, because the color contribution of components in foods itation method and showed for the rst time that these is cumulative. If color results from similar but not identical pigments contained nitrogen and were similar to melanin.1 pigments or chromophores, the contribution of each They called these pigments soyamelanic acid and soyamelanin. compound should be low. DHDMT has been reported as an Kato et al. have reported the presence of 3-deoxyglucosone in aroma compound formed by the Maillard reaction (Fig. 1) from soy sauce2 and Kato also showed that 3-deoxyglucosone was an cysteine and furanone10 and glucose.11 This compound was also important intermediate compound in the Maillard reaction.3 isolated from heated garlic as an antioxidant and anti-inam- These results indicate that the color of soy sauce is formed by matory substance,12,13 although the authors used a different the Maillard reaction. Hashiba separated and characterized the name, thiacremonone. Recently, the preventive effect against color of soy sauce by gel ltration chromatography.4 Motai et al. hepatocarcinogenesis and the anti-obesity effect of this examined the relationship between the oxidation of soy sauce compound have been reported.14,15 In other words, DHDMT and the change in color.5 Hayase et al. showed the presence of formed by the Maillard reaction is a pigment, an aroma pyrraline, imidazolone, pentosidine, and lysyl-pyrropyridine in compound, and a physiologically functional component in soy sauce.6 These studies have claried that the color of soy foods. However, we do not know the conditions for the formation of DHDMT and its distribution in foods. To utilize these functions of DHDMT efficiently, we have to Department of Nutrition and Food Science, Ochanomizu University, 2-1-1 Otsuka, know the conditions of its formation and its distribution in Bunkyo-ku, Tokyo, +112-8610 Japan. E-mail: [email protected]; Fax:  +81-3-5978-5755 foods. The aim of this present study is rst to explain how

1076 | Food Funct., 2013, 4, 1076–1081 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper Food & Function

Fig. 1 Plausible formation scheme of 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) from hexose and cysteine by the Maillard reaction.10,11 1-DG, 1- deoxyglucosone; 3-DG, 3-deoxyglucosone; DMHF, 2,5-dimethyl-4-hydroxy-3(2H)-furanone.

DHDMT was formed during the production of soy sauce, second to clarify the conditions for its formation in model systems, and third to show its distribution in foods and beverages.

Materials and methods Published on 28 March 2013. Downloaded by FAC DE QUIMICA 12/11/2013 23:47:20. Materials Soybean, wheat, soy sauce, miso (soy bean paste), brown rice, nampla (sh-sauce), mentsuyu (a soy sauce-based dip for Japanese noodles), beer, bread, onion, garlic, roasted coﬀee beans, cocoa, and chocolate were purchased at a local market (Tokyo, Japan).

Preparation of soy sauce Soy sauce was prepared on the laboratory scale. The preparation scheme is shown in Fig. 2. Soy beans (Glycine max cv. Toyo- masari; 2.5 L) were soaked in about 5 L of tap water and then gently boiled for 6 h. Wheat (a blend of Triticum aestivam; 2.5 L) was parched on a pan for 5 min and then cracked with a wooden hammer. Spores of Aspergillus oryzae (about 2 g, Kojiya Mizaemon Co., Toyohashi, Japan) were added to a mixture of the boiled soy beans and the parched cracked wheat. These Fig. 2 Preparation scheme of soy sauce. materials were mixed by hand and incubated at 30 C for 3 days to form koji, a fungal mixture. The koji was then mixed with centrifuged at 12 000 g for 30 min, and the obtained super- about 7.5 L of brine containing 3.3 kg NaCl, which was fer- natant was used for measurements of pH, DHDMT, and L-glu- mented for 7 months at room temperature with occasional tamic acid. L-Glutamic acid concentrations were measured stirring. The mash samples fermented for 0–7 months were using a kit (TC L-glutamic acid; Roche Diagnostic Japan, Tokyo).

This journal is ª The Royal Society of Chemistry 2013 Food Funct., 2013, 4, 1076–1081 | 1077 View Article Online Food & Function Paper

For measurement of absorbance, the supernatant was added to the same volume of 20% triuoroacetic acid solution. Aer removing the formed precipitate by centrifugation at 12 000 g for 20 min, the absorbance of the supernatant was measured. The mean of three measurements was obtained.

Analysis of 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) For liquid samples (10–20 mL), DHDMT was extracted with the same volume of ethyl acetate three times. For solid samples (10–20 g), DHDMT was rst extracted with MeOH (50–100 mL) three times. For cocoa and chocolate, samples were washed with hexane (35 mL 3 times) before the methanol extraction. Aer the methanol extract was concentrated in vacuo, DHDMT was re-extracted with ethyl acetate. Aer the ethyl acetate layer was concentrated in vacuo, DHDMT was dissolved in 1–2 mL of MeOH and analyzed with HPLC.HPLCconditionswereasfollows:system,Hewlett Packard series 1100 with a photodiode-array detector (Palo Alto, CA); column, TSKgel ODS 100V (4.6 mm i.d. 250 mm, Tosoh, Tokyo); eluent, solution A (water/MeOH ¼ 98 : 2) and solution B (MeOH), with a linear gradient from solution A to a mixture of solutions A and B (30 : 70) for 30 min; detection, 250–500 nm. The concentration was quantied at 400 nm with an external standard method using isolated DHDMT9 as Fig. 3 Time course of fermentation of soy sauce. The concentrations of a standard. DHDMT was eluted at 14.8 min with an absorp- L-glutamic acid (L-Glu) and 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone tion maximum at 360 nm. Each sample was analysed at least (DHDMT), pH and absorbance at 400 nm of soy sauce during fermentation were in triplicate. measured.

Model solution

A solution (4.0 mL) containing 50 mM L-cysteine, 0.75 M fruc- increased during fermentation. With a delay of increase in the ff – tose or glucose, and 0.5 M acetate bu er (pH 4.5 7.5) was put color, DHDMT was gradually formed aer 4 months of – – into a test tube with a cap and then heated for 1 5 h at 60 fermentation and accumulated (Fig. 3B). 120 C. The acetate buffer was replaced with 0.05–0.4 M acetate Published on 28 March 2013. Downloaded by FAC DE QUIMICA 12/11/2013 23:47:20. buffer, 0.5 M phosphate buffer (pH 5.5–8.0), and 0.5 M Tris–HCl buffer (pH 6.5–10.5). In an experiment, L-cysteine (50 mM) was Formation of DHDMT in model systems replaced with L-cysteine (1.56–400 mM) and cystine (1.56– The conditions for DHDMT formation from hexose and 200 mM). Each experiment was done in triplicate. cysteine were then examined in model systems. As more DHDMT was formed from fructose than from glucose in a Statistical analysis preliminary experiment at pH 5, fructose was used here in the ff The data obtained in the experiments on DHDMT contents of model system. Fig. 4A shows the e ectofreactiontimeand soy sauce and miso were analysed by a Student's t-test with temperature on its formation. Higher temperature and pro- Statcel2 soware (OMS Publishing, Tokorozawa, Japan) on longed heating promoted DHDMT formation. Fig. 4B shows ff ff Excel 2010 (Microso, Redmond, WA). the e ect of the concentration of the acetate bu er on the DHDMT formation. As the concentration of buffer was raised, Results more DHDMT was formed. Fig. 4C shows the effect of pH and buffer on the formation. The optimal pH for the DHDMT H Formation of 2,4-dihydroxy-2,5-dimethyl-3(2 )-thiophenone formation was around 7. More DHDMT was formed in the (DHDMT) during soy sauce production acetate buffer than in the Tris–HCl buffer at pH 7. DHDMT First, we examined in which step of soy sauce preparation was not detected in phosphate buffer. These results indicate DHDMT was formed. Mash, a fungal mixture (koji) with brine, that the composition and concentration of buffers inuenced was fermented for 7 months at room temperature. As shown in the DHDMT formation, although most DHDMT was formed Fig. 3, the pH value dropped a little due to lactic acid fermen- around a neutral pH region. Fig. 5 shows the difference of tation, and the concentration of L-glutamic acid was gradually glucose and fructose in DHDMT formation. A little more increased by the action of koji enzymes. The brown color of soy DHDMT was formed at pH 4.5 from fructose than from sauce, shown by the absorbance at 400 nm, was gradually glucose. As the pH of the reaction solution was made neutral

1078 | Food Funct., 2013, 4, 1076–1081 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper Food & Function

Fig. 5 Comparison of glucose and fructose for the 2,4-dihydroxy-2,5-dimethyl- 3(2H)-thiophenone (DHDMT) formation. Cysteine (50 mM) and hexose (0.75 M) were dissolved in 0.5 M acetate buﬀer (pH 4.5–7.0) and heated at 120 Cfor3h (n ¼ 3). Published on 28 March 2013. Downloaded by FAC DE QUIMICA 12/11/2013 23:47:20.

Fig. 4 Effect of temperature and time (A), acetate buffer concentration (B), and pH and buffer (C) on the formation of 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) in model systems. Cysteine (0.05 M) and fructose (0.75 M) were dissolved in the buffer (A, 0.5 M acetate buffer (pH 5); B, 0.05–0.5 M acetate buffer (pH 5); C, 0.5 M acetate buffer (pH 4.5–7.5), 0.5 M Tris–HCl buffer (pH 6.5– 10.0) and 0.5 M phosphate buffer (pH 5.5–8.0)), and heated for 6 h (A) and 3 h (B and C) (n ¼ 3).

from an acidic region, the formation of DHDMT from fructose was denitely raised, and the concentration of the formed

DHDMT at pH 7 from fructose was more than 10 times higher ff ff Fig. 6 E ect of concentrations of cysteine (A) and cystine (B) on the 2,4-dihy- than that from glucose. The e ect of cysteine and cystine on droxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) formation. Cysteine (1.56– the formation of DHDMT was then examined. As shown in 400 mM) or cystine (1.26–200 mM) were dissolved in 0.5 M acetate buffer (pH Fig. 6A, the concentration of cysteine inuenced the 5.0) containing fructose (0.75 M) and heated at 120 Cfor3h(n ¼ 3). production of DHDMT, and about 12.5 mM of cysteine was optimal for the formation of DHDMT in the presence of 0.75 M fructose. The addition of cystine in the fructose solu- of 25 mM of cystine (Fig. 6B). This concentration was tion dramatically increased the formation of DHDMT, and about 10 times higher than that in the 12.5 mM cysteine more than 300 mg L 1 of DHDMT was formed in the presence solution.

This journal is ª The Royal Society of Chemistry 2013 Food Funct., 2013, 4, 1076–1081 | 1079 View Article Online Food & Function Paper

Distribution of DMDTH in foods and beverages Table 3 Contents of 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) in various kinds of heated foodsa Tables 1–3 show the concentration of DHDMT in various kinds of foods and beverages. DHDMT was detected in all soy sauce Sample (heating DHDMT DHDMT (21 samples) and miso (9 samples) examined here. The contents conditions) (mg kg 1) Sample (retail) (mg kg 1) in soy sauce and miso ranged from 2 to 35 mg L 1 and 1 to 8 mg 1 Roasted wheat grain 5.3 Crust of bread 1.0 kg , respectively (Table 1). In general, colors of koikuchi, (10 min) saishikomi, and tamari soy sauce are darker than those of Roasted brown rice 0.67 Crumb of bread nd usukuchi and shiro soy sauce. The concentrations of DHDMT in (10 min) the former three groups of soy sauce (16 brands) were signi- Roasted wheat our nd Roasted coﬀee nd cantly higher than those of the latter two groups of soy sauce (10 min) beans Heated garlic 2.1 Chocolate nd (4 brands) (p < 0.01). The color of mamemiso is generally darker (150 C, 1 h) than other types of miso. Dense-colored soy sauce and miso Heated onion 7.6 Cocoa nd seem to contain higher amounts of DHDMT than pale colored (150 C, 1.5 h) ones. Mentsuyu, which is a soy sauce-based dip for Japanese Heated onion nd Kinako (roasted nd  noodles, contained DHDMT (Table 2). DHDMT was also (100 C, 30 min) soybean our) Boiled soy beans (6 h) nd detected in all kinds of beer examined here (8 samples, Table 2). a 1 DHDMT was detected in roasted wheat grains, a crust of bread, nd, not detected (<0.1 mg kg ). burnt garlic, burnt onion, etc., while it was not detected in boiled soy beans, a crumb of bread, roasted coﬀee beans, chocolate, etc. (Table 3). Discussion

DHDMT was rst isolated as a Maillard avor in a model Table 1 Contents of 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) system.10,11 This compound was then isolated as a biologically in various kinds of soy sauce and miso active compound from heated garlic. DHDMT showed anti- 12 13 14 oxidative, anti-inammatory, cancer-preventive, and anti- DHDMT (mg L 1) obesity activities.15 We then isolated this compound as a Sample n Mean (Min.–max.) Maillard yellow pigment from soy sauce.9 This nding might be important as a compound has the characteristics of an aroma Soy sauce (type) compound, a pigment and a functional component together. Koikuchi (standard) 9 18.1 11.4–35.4 Saishikomi (twice-fermented) 3 13.6 12.1–16.1 We can recognize this functional food component by the senses Tamari (soybean only) 4 16.3 6.9–24.5 of smell and sight. This characteristic is formed by the Maillard Nama (non-pasteurized) 1 8.1 reaction. The detection limits of DHDMT as a avor and a color9 – Usukuchi (pale-colored) 3 6.5 3.4 10.6 were about 10 mgmL 1 and 25 mgmL 1, respectively, and were Shiro (very pale-colored) 1 1.6

Published on 28 March 2013. Downloaded by FAC DE QUIMICA 12/11/2013 23:47:20. at a similar level. The content of DHDMT in foods ranged from – m 1 Miso (type) 0 35 gg . This indicates that this compound contributes the Kome (rice and soybean) 4 2.2 2.0–5.2 sensory quality to some extent, although the contribution is low. Mugi (barley and soybean) 3 2.7 1.2–4.9 We point out that the strength of color is cumulative and that a – Mame (soybean) 2 4.4 3.9 7.6 chromophore of melanoidin, a heterogeneous high-molecular- weight compound, is little known. It will be also necessary to evaluate the functional potencies from the aspect of concentration. Table 2 Contents of 2,4-dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT) a in various beverages or liquid foods We examined in which step DHDMT was formed in soy sauce production as we isolated it from soy sauce. Although DHDMT DHDMT (mg L 1) was not detected until 3 months of fermentation, it was detected aer 4 months of fermentation and gradually accumulated. Sample n Mean (Min.–max.) This indicates that it takes a long time to form DHDMT at room 11 Beer temperature. Tressl et al. showed that DHDMT was formed Pale-yellow (regular) 4 0.58 0.50–0.77 from glucose and cysteine by the Maillard reaction by using 13C Intermediate-color 1 0.82 labelled compounds. DHDMT in soy sauce seemed to be formed – Black 2 0.49 0.48 0.49 by the Maillard reaction between glucose and cysteine which Beer-type liquor 1 0.25 were produced during fermentation from macromolecules of Cola 2 nd raw materials by the action of koji enzymes. Nampla (sh-sauce) 1 nd In the model system, higher temperature and prolonged Mentsuyu (a soy sauce-based 1 2.9 heating promoted DHDMT formation. This supports that it dip for Japanese noodles) took a long time for the formation of DHDMT in soy sauce a nd, not detected (<0.1 mg L 1). production. DHDMT was detected in all samples of soy sauce

1080 | Food Funct., 2013, 4, 1076–1081 This journal is ª The Royal Society of Chemistry 2013 View Article Online Paper Food & Function

and miso (soybean paste), the production of which takes several Acknowledgements months at room temperature, and it was also detected in various foods or beverages which are highly heated or processed This study was partly supported by a Grant-in-Aid for Scientic with high temperature. The existence of DHDMT suggests the Research (B) (no. 22300257) from the Japan Society for the processing using prolonged heating or long time maturation. Promotion of Science. More DHDMT was formed from fructose than from glucose, and the optimal pH for its formation was around 7. These observations seem to have relevance to the fact that Notes and references formation of 1-deoxyglucosone from hexose is promoted at neutral pH than at acidic pH16 and that 1-deoxyglucose is 1 K. Kurono and H. Katsume, Nippon Nogeikagaku Kaishi, 1926, – more easily formed from fructose than from glucose. But we 3, 594 613. could not detect DHDMT in phosphate buffer. As 1-deoxy- 2 H. Kato, Y. Yamada, K. Izaka and Y. Kakurai, Nippon – glucoson and 3-deoxyglucosone are formed from an Amadori Nogeikagaku Kaishi, 1961, 35, 412 415. – compound in phosphate buffer at pH 8.2 and 9.0,17 the 3 H. Kato, Bull. Agric. Chem. Soc. Jpn., 1960, 24,1 12. – phosphate ion might inuence the formation and stability of 4 H. Hashiba, Nippon Nogeikagaku Kaishi, 1971, 45,29 35. intermediate compounds aer the formation of 1- 5 H. Motai, S. Inoue and Y. Nishizawa, Nippon Nogeikagaku – deoxyglucosone. Kaishi, 1972, 46, 631 637. We here showed that more DHDMT was formed from cystine 6 F. Hayase, Y. Takahashi, S. Sasaki, S. Shizuuchi and – than from cysteine in the model system. It seemed that cystine H. Watanabe, Int. Congr. Ser., 2002, 1245, 217 221. – was reduced to cysteine during the Maillard reaction, because 7 K. Kihara, Shoyu no Kenkyu to Gijutsu, 1985, 11,53 59. the Maillard reaction forms various kinds of reductants,18 and 8 S. Nakatsuka, B. Feng, T. Goto and K. Kihara, Tetrahedron – then the formed cysteine reacted with intermediate compounds Lett., 1986, 27, 3399 3402. as shown in Fig. 1 to form DHDMT. This seems to be important 9 M. Satoh, Y. Nomi, S. Yamada, M. Takenaka, H. Ono and – for the formation of DHDMT in foods, because S–S bond M. Murata, Biosci., Biotechnol., Biochem., 2011, 75, 1240 formation between cysteine residues in proteins is ubiquitous. 1244. In the model system containing 12.5 mM cysteine and 0.75 M 10 C.-K. Shu, M. L. Hagedorn, B. D. Mookherjee and C.-T. Ho, J. – fructose, the concentration of DHDMT reached more than Agric. Food Chem., 1985, 33, 638 641. 30 mg L 1 (0.5 M acetate buffer (pH 5.0), 120 C, 3 h). On the 11 R. Tressl, E. Kersten, C. Nittka and D. Rewicki, ACS Symp. – other hand, in the model system containing 50 mM cystine and Ser., 1994, 564, 224 235. 0.75 M fructose, the concentration of DHDMT reached more 12 I. G. Hwang, K. S. Woo, D. J. Kim, J. T. Hong, B. Y. Hwang – than 300 mg L 1. These data will be useful to raise the content and Y. R. Lee, Food Sci. Biotechnol., 2007, 16, 963 966. of DHDMT in processed foods. 13 J. O. Ban, J. H. Oh, T. M. Kim, D. J. Kim, H.-S. Jeong, S. B. Han DHDMT was not detected in coffee and chocolate. It is well- and J. T. Hong, Arthritis Res. Ther., 2009, 11, R145. known that these foods contain acrylamide.19 Since cysteine 14 T. M. Kim, H. S. Lee, T. J. Shim, H. Y. Kim, K. S. WooH, S. Jeong and D. J. Kim, Food Sci. Biotechnol., 2012, 21,

Published on 28 March 2013. Downloaded by FAC DE QUIMICA 12/11/2013 23:47:20. reacts with various intermediate compounds of the Maillard – reaction, the existence of cysteine inuences the reaction 1277 1284. products of the Maillard reaction. It has been reported that 15 J. O. Ban, D. H. Lee, E. J. Kim, J. W. Kang, M. S. Kim, cysteine represses the formation of acrylamide.20,21 The detec- M. C. Cho, H. S. Jeong, J. W. Kim, Y. Yang, J. T. Hong and – tion of DHDMT indicates the reaction between cysteine and the D. Y. Yoon, Phytother. Res., 2012, 26, 1265 1271. C6 intermediate compounds shown in Fig. 1. The formation of 16 N. A. M. Eskin, Biochemistry of Foods, Acdemic Press, San – DHDMT might imply the repression of the formation of acryl- Diego, 2nd edn, 1990, pp. 239 296. amide owing to the decrease in intermediate compounds for 17 J. Hirsch, V. V. Mossine and M. S. Feather, Carbohydr. Res., – acrylamide. It will be necessary to chemically investigate the 1995, 273, 171 177. relationship between formations of acrylamide and DHDMT in 18 H. Ukeda, T. Shimamura, T. Hosokawa, Y. Goto and – the future. M. Sawamura, Food Sci. Technol. Int., Tokyo, 1998, 4, 258 263. Conclusions 19 O. Pardo, V. Yusu, C. Coscola, N. Leon´ and A. Pastor, Food Addit. Contam., 2007, 24, 663–672. 2,4-Dihydroxy-2,5-dimethyl-3(2H)-thiophenone (DHDMT), 20 F. Mestdagh, J. Maertens, T. Cucu, K. Delporte, C. Van contributing color, aroma, and functional property to foods, Peteghem and B. De Meulenaer, Food Chem., 2008, 107, was formed more from cystine than cysteine in the model 26–31. system, and was detected in various brown foods and beverages 21 G. Koutsidis, S. P. J. Simon, Y. H. Thong, Y. Haldoupis, in which the Maillard reaction occurred at high temperature or J. W. Majica-Lazaro and D. S. Motram, J. Agric. Food Chem., during long time maturation. 2009, 57, 9011–9015.

This journal is ª The Royal Society of Chemistry 2013 Food Funct., 2013, 4, 1076–1081 | 1081

Chemistry and Physics of Lipids 165 (2012) 662–681

Contents lists available at SciVerse ScienceDirect

Chemistry and Physics of Lipids

j ournal homepage: www.elsevier.com/locate/chemphyslip

Review

Chemical alterations taken place during deep-fat frying based on certain reaction products: A review

∗

Qing Zhang, Ahmed S.M. Saleh, Jing Chen, Qun Shen

National Engineering and Technology Research Center for Fruits and Vegetables, College of Food Science and Nutritional Engineering,

China Agricultural University, Beijing 100083, China

a r t i c l e i n f o a b s t r a c t

◦

Article history: Deep-fat frying at 180 C or above is one of the most common food processing methods used for preparing

Received 17 April 2012

of human kind foods worldwide. However, a serial of complex reactions such as oxidation, hydrolysis,

Received in revised form 9 June 2012

isomerization, and polymerization take place during the deep-fat frying course and inﬂuence quality

Accepted 5 July 2012

attributes of the final product such as flavor, texture, shelf life and nutrient composition. The influence of

Available online 16 July 2012

these reactions results from a number of their products including volatile compounds, hydrolysis prod-

ucts, oxidized triacylglycerol monomers, cyclic compounds, trans conﬁguration compounds, polymers,

Keywords:

sterol derivatives, nitrogen- and sulphur-containing heterocyclic compounds, acrylamide, etc. which are

Deep-fat frying

Triacylglycerol present in both frying oil and the fried food. In addition, these reactions are interacted and inﬂuenced

by various impact factors such as frying oil type, frying conditions (time, temperature, fryer, etc.) and

Chemical alteration

Reaction product fried material type. Based on the published literatures, three main organic chemical reaction mecha-

Formation mechanism nisms namely hemolytic, heterolytic and concerted reaction were identiﬁed and supposed to elucidate

the complex chemical alterations during deep-fat frying. However, well understanding the mechanisms

of these reactions and their products under different conditions helps to control the deep-fat frying pro-

cessing; therefore, producing healthy fried foods. By means of comprehensively consulting the papers

which previously studied on the chemical changes occurred during deep-fat frying process, the major

Contents

1. Introduction ...... 663

2. Major reaction products ...... 663

2.1. TAG degradation products ...... 665

2.1.1. Oxidized volatile products...... 665

2.1.2. Hydrolysis products ...... 666

2.2. Oxidized TAG monomers ...... 667

2.3. Nonpolar TAG derivatives ...... 667

2.3.1. Cyclic fatty acid monomers...... 667

2.3.2. Trans isomers ...... 669

Abbreviations: AA, acrylamide; Ag-HPLC, silver ion-high performance liquid chromatography; APCI, atmospheric pressure chemical ionization; ATR, attenuated total

reﬂection; BHA, butylated hydroxyanisole; BHT, butylated hydroxytoluene; C=C, carbon-carbon double bond; CFAMs, cyclic fatty acid monomers; CLA, conjugated linoleic

acid; DAD, diode array detection; DAG, diacylglycerol; DPPH, 1,1-diphenyl-2-picrylhydrazyl; ECD, electron capture detector; EPR, electron spin resonance; ESI, electrospray

ionization; GC×GC, comprehensive two-dimensional gas chromatography; GC-MI-FTIR, gas chromatography-matrix isolation-Fourier transform infrared spectroscopy;

GLC, gas-liquid chromatography; GPC, gel permeation chromatography; HAAs, heterocyclic aromatic amines; HPLC, high performance liquid chromatography; HPSEC,

high performance size exclusive chromatography; HPTLC-UV-FLD, high performance thin layer chromatography-ultraviolet-ﬂuorescence detection; ITMS, ion trap mass

spectrometry; MAG, monoacylglycerol; NIR-PLS, near infrared spectroscopy-partial least squares regression; NMR, nuclear magnetic resonance; RI, refractive index; RPLC-

TSP-MS, reverse-phase high-performance liquid chromatography–thermospray-mass spectrometry; SP(M)E, solid-phase (micro-) extraction; TAGs, triacylglycerols; TD-GC-

MS, thermal desorption-gas chromatography-mass spectrometry; TOFMS, time-of-ﬂight mass spectrometry.

∗

Corresponding author. Tel.: +86 10 62737524; fax: +86 10 62737524.

E-mail address: [email protected] (Q. Shen).

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 663

2.4. TAG polymerized products ...... 670

2.5. Sterol derivatives ...... 671

2.6. Antioxidant alterations ...... 673

2.7. Products derived from interactions between frying oil and food material constituents ...... 674

2.7.1. Nitrogen- and sulphur-containing heterocyclic compounds ...... 674

2.7.2. Acrylamide ...... 675

3. Summary ...... 676

References ...... 676

1. Introduction The chemical reactions occurred during deep-fat frying roughly

involved in hydrolysis, oxidation, isomerization and polymeriza-

tion (Choe and Min, 2007; Velasco et al., 2009) resulted in the

It is well known that deep-fat frying is a prevalent and old food

◦ generation of free fatty acids, small molecular alcohol, aldehyde,

cooking method which can go back to 1600 BC. Although 180 C

ketone, acid, lactone and hydrocarbon (Pokorny,´ 1989), diglyceride

is usually recommended for frying foods, it is always higher than

◦ and monoglyceride, cyclic and epoxy compounds (Rojo and Perkins,

180 C in the practical deep-fat frying (Firestone, 1993). Fast food

1987), trans isomers (Martin et al., 1998a,b,c), TAG monomer, dim-

processing, palatable taste of fried food and considerable economic

mer, oligomer (Martin et al., 1998a,b,c). Besides, these reactions

beneﬁt make the deep-fat frying become one of the most pop-

interact and impact each other during the treatment at high-

ular food cooking methods used in household kitchen, fast-food

temperature. As a result, when the deep-fat frying excessively

restaurant and instant noodles industry. Furthermore, the sale of

proceeded, the phenomena of off-ﬂavor, foaming, color deep-

pre-cooked and ready-to-eat products which also refer to the deep-

ening and the increase of viscosity would appear in the frying

frying process has dramatically increased in the western world and

oil.

is rapidly expanding throughout the developing countries. In other

Polar compounds and TAG oligomers which produced during

words, fried food has become an industry chain in catering industry.

deep-fat frying course with content range of 20–27% and 10–16%

The fried food is endowed with attractive ﬂavor, golden pellicle and

respectively have been proposed to determinate the rejection of

crisp texture or mouth feel when it is ﬁred under the appropriate

used frying oil (Bastida and Sánchez-Muniz, 2002; Paul and Mittal,

conditions (Rossell, 2001; Warner, 2008).

1997). However, the measurement of polar compounds doesn’t

Under the established conditions of fried material’s natural

represent the whole content of reaction products formed during

properties and corresponding sample handling, frying can involve

the deep-fat frying and the reaction products are not detailed

all of the components to participate in a series of physical and chem-

in these indices. Some products were considered as potential

ical alterations. These changes not only include the decomposition

detrimental substances to the human body from the nutritional

reactions of the constituents such as the nutrients of raw material

aspect (Lamboni and Perkins, 1996; Saguy and Dana, 2003; Totani

and triacylglycerols (TAGs) of frying oil, but also include the inter-

et al., 2008). Therefore, it is quite necessary and useful to know

actions among these constituents (Chu and Luo, 1994; Dobarganes

more systematic precise information about the nature and quan-

et al., 2000a,b). Moreover, deep-fat frying is a complicated physic-

tity of the new compounds formed during the deep-fat frying

ochemical processes which is simultaneously inﬂuenced by many

course.

factors such as the nature of fried material and frying oil, time, tem-

Although the chemical alteration pathways occurred during

perature, intermittent or continuous heating, fresh oil complement,

deep-fat frying are complex and not well understood, some reac-

fryer model and use of Filters (Chatzilazarou et al., 2006; Kalogianni

tion pathways and the resulted reaction products have the same

et al., 2010; Rojo and Perkins, 1987). Therefore, many products are

characteristics and could be analyzed and classiﬁed as one class.

formed due to these complex substrates and chemical conditions.

Therefore, the major reaction products which have been previ-

Furthermore, frying with food and frying without food have a sig-

ously identiﬁed and reported to be elucidated by category and their

niﬁcant different chemical in reaction pathways (Barrera-Arellano

corresponding chemical reaction pathways are presented in this

et al., 1997; Houhoula et al., 2002).

work.

On the other hand, deep-fat frying is a process of drying and

cooking in hot oil at high-temperature with simultaneous heat and

mass transfer (Ahromrit and Nema, 2010; Dincer, 1996; Ni and 2. Major reaction products

Datta, 1999). Oil and fat is an excellent heat-transfer medium, thus

the material is quickly heated and cooked when it is immersed Analysis of substrate-known reaction products is not an easy

into the oil (Alvis et al., 2009). Mass transfer involves the loss of work to analysts due to the quantity and characteristic of the

moisture, oil and fat, carbohydrate, protein and vitamins and other products, not to mention the reactions involved in the deep-fat fry-

components from fried material and the oil uptake of material ing process. However, with a combination of developed analytical

from frying oil (Krokida et al., 2000; Sosa-Morales et al., 2006). instrument and chemometrics, the determination of the reaction

Therefore, both fried material and frying oil inﬂuence on each products with similar conﬁguration or trace quantity has become

other and collectively promote the occurrence of complex chemical possible and practical. Given that TAGs are the major kind of con-

reactions. stituents existing in frying oil, the probable emerging products and

Based on published data, there are few systemic and compre- their possible formation routes which have been investigated were

hensive studies to investigate the speciﬁc pathways of chemical reviewed as following details according to the alterations of TAGs.

reactions taken place during the deep-fat frying course. Among Moreover, being viewed as the major impact factor to the chemical

these reactions and products, some are expected to occur alterations, food material was also considered in term of the inter-

such as the desirable color and attractive ﬂavor resulted from actions between food material components and TAGs. The major

maillard reaction and some are undesired outcomes due to los- reaction products and their characteristics are shown in Table 1

ing of nutrients, producing of aldehyde, acrylamide and trans for direct and systematic understanding of the whole deep-fat fry-

conﬁguration-contained substances which are prejudicial to the ing course. The speciﬁc product types and their possible formation

quality of fried food and health of people (Boskou, 2003). mechanisms are elucidated as follows.

664 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

Table 1

The major reaction products formed during deep-fat frying, their characteristics and proposed formation mechanism and identifying methods according to the previously

reported studies.

Main products Characteristics Proposed formation mechanism Identifying methods

Oxidized decomposition Degradation product; volatility; alcohol, Oxidation TD-GC-MS (Fullana et al., 2004; Overton

compounds aldehyde, ketone, acid, lactone and Homolytic reaction (Free radical and Manura, 1995), SPME-GC-MS

hydrocarbon etc.; short-chain compounds reaction) (Mildner-Szkudlarz and Jelen,´ 2008), H

(saturate or unsaturate); depending on the NMR (Guillén and Uriarte, 2009), EPR

number and position of C C, extra oxygen (Roman et al., 2012)

etc.; molecule weight lower than that of

parent TAGs

Hydrolysis products Degradation product; polar; diacylglycerol Hydrolysis GC (Lee et al., 2002), GC and HPSEC

(DAG), monoacylglycerol (MAG), glycerol Heterolytic reaction (nucleophilic (Romero et al., 1998; Houhoula et al., 2003)

and free fatty acids; depending on the reaction)

presence of water; molecule weight lower

than that of parent TAGs

Oxidized TAG monomers Oxidized TAGs with keto, epoxy, hydroxyl, Oxidation GLC (Velasco et al., 2002), HPLC (Schulte,

aldehyde and epoxy groups (saturate or Epoxidation 2002), HPSEC (Caldwell et al., 2011),

unsaturate); extra oxygen-containing Free radical reaction GC-MS (Kamal-Eldin et al., 1997),

groups may simultaneously exist in one RPLC–TSP-MS (Yamane, 2002);

molecule; polar; core aldehydes; cis and RPLC-ESI-MS (Giuffrida et al., 2004a,b), H

trans conﬁguration; depending on the NMR (Aerts and Jacobs, 2004; Guillén and

number and position of C C, extra oxygen Ruiz, 2008)

etc.; molecule weight approximately

equals to that of parent TAGs

Cyclic fatty acid monomers Nonpolar; low concentration and latent Cyclization Ag-HPLC and GC-MS (Dobson et al., 1995),

biological hazard; ﬁve- or six-membered Intramolecular rearrangement GC-EI-MS (Berdeaux et al., 2007),

ring structures (saturate or unsaturate); catalyzed by free radicals GC-MI-FTIR and GC-EI-MS (Mossoba et al.,

monocyclic and bicyclic rings; cis and trans Concerted reaction 1996a,b)

conﬁgurations; depending on the number ([1,j]-prototropic migrations)

and position of C C; molecule weight

lower than that of parent TAGs

Trans isomers Nonpolar; trans fatty acids, conjugated Free radical reaction GC (Aro et al., 1998), GC-MS (Kandhro

linoleic acids; adverse effects on human (Addition-elimination mechanism) et al., 2008) and Ag-HPLC (Yurawecz et al.,

health; one more trans conﬁgurations may Heat-induced isomerization 1999; Eulitz et al., 1999), ATR-FTIR (Cho

simultaneously exist; depending on the Concerted reaction et al., 2011; Mossoba et al., 2007)

number and position of C C; molecule ([1,j]-sigmatropic rearrangements)

weight lower than that of parent TAGs

TAG polymerized products Dimmers, trimers and oligomers, etc.; with Oxidized polymerization and HPSEC (Dobarganes et al., 2000a,b;

the linkages of –C–C–, –C–O–C– and thermal polymerization Caponio et al., 2007), on-line LC-EI-MS

–C–O–O–C– among the TAG molecules; Free radical reaction (Byrdwell and Neff, 2004), ATR-FTIR

Nonpolar and polar; acyclic and cyclic Concerted reaction (Diels-Alder (Kuligowski et al., 2010a), On-line

polymers; depending on the number and reaction) GPC-FTIR (Kuligowski et al., 2010b),

position of C C, extra oxygen etc.; can be NIR-PLS (Kuligowski et al., 2012)

polymerized by both abovementioned

products and parent TAGs; molecule

weight higher than that of parent TAGs

Sterol derivatives Formed mainly result in special molecule Oxidation (Free radical reaction) GC-MS (Soupas et al., 2004); LC-APCI-MS

structure; Hydroxy, keto and epoxy Polymerization (Kemmo et al., 2008); SEC-APCI-MS

group-contained compounds; sterol (Rudzinska´ et al., 2010);

dimmers, trimers and oligomers also HPLC-RI/UV/APCI-MS (Saldanha et al.,

present; homologous 2006); SPE-HPSEC (Lampi et al., 2009)

Antioxidant alterations Rooted in natural and synthetic Free radical reaction HPLC (Rennick and Warner, 2006);

antioxidants or called free radical Esteriﬁcation HPLC-NMR (Verleyen et al., 2001a,b)

scavengers; accompany with many Dimerization

changes (i.e. the decrease of some

undesirable products) during the deep-fat

frying course; possess the quinine

structure

Heterocyclic compounds Nitrogen- and sulphur-containing Free radical reaction GC-MS (Horiuchi et al., 1998; Van Loon

heterocyclic compounds; volatility; latent Maillard reaction et al., 2005), Alternative GC–MS

mutagenicity and carcinogenicity; derived Electrocyclic and aromatization approaches: GC–ITMS, GC–TOFMS,

from interactions between frying oil and reaction GC × GC–TOFMS (Lojzova et al., 2009);

food material constituents or their reaction Nucleophilic reaction HPTLC-UV-FLD (Jautz et al., 2008)

products; formed from the interaction

between lipid oxidation and maillard

reaction

Acrylamide Possesses neurotoxicity, genetic toxicity Oxidation (Free radical reaction) GC-ECD (Zhang et al., 2006), GC-MS/MS

and carcinogenicity; depending on the Maillard reaction (Strecker (Hoenicke et al., 2004), HPLC-UV (Wang

oxidability of frying oil and the type of fried degradation) et al., 2008), LC–DAD (Gökmen et al.,

food; formed from the interaction between 2005), LC-MS/MS (Carrieri et al., 2010;

lipid oxidation and maillard reaction; Pedreschi et al., 2007; Viklund et al., 2007),

acrolein has been proposed as its precursor NIR-PLS (Pedreschi et al., 2010)

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 665

2.1. TAG degradation products condition of high temperature, lipid peroxidation becomes more

complex compared to the condition of ambient temperature, thus

Oil and fat is a mixture of TAGs which are composed of one the exhaustive elucidation of lipid peroxidation should be paid

glycerol and three groups of saturated or unsaturated fatty acids more comprehensively consideration in further studies.

with different carbon numbers. Not only the natures of fatty acid, Studies on volatile compounds have been greatly reported

but also the various combination positions of fatty acids to glyc- in the last century and mainly the odor constituents formed

erol molecule would impact the reaction activity of TAG. Therefore, during deep-fat frying with food material and without food mate-

the TAG degradation products mainly result from the breakages rial were investigated extensively (Brewer et al., 1999; Macku

occurred in the carbon-carbon double bond (C C) of aliphatic and Shibamoto, 1991a; Wu and Chen, 1992). Volatile products

chains and ester bond. These compounds have a smaller molecular formed from corn oil (Kawada et al., 1967), hydrogenated cotton-

weight compared with that of the parent TAG and almost possess of seed oil (Yasuda et al., 1968), trilinolein (Thompson et al., 1978),

volatility such as the decomposition compounds of lipid oxidation and triolein (May et al., 1983) during simulated deep-fat fry-

and TAG hydrolysis. ing were initially studied under well controlled frying conditions

and a total of 220 volatile compounds such as acids, hydrocar-

2.1.1. Oxidized volatile products bons, alcohols, aldehydes, ketones, esters, lactones and aromatic

It is well known that autoxidation is an important degradation compounds were detected (Chang et al., 1978). In addition, 140

reaction which is attributed to the rancidity of oil and fat. It is volatile constituents were detected and identiﬁed from used fry-

as well the major reaction occurred during frying along with the ing oils which were collected from large-scale commercial meat

increase of temperature. The mechanism of thermal oxidation is and poultry processing plant and the typical volatile constituents

principally similar with the autoxidation mechanism, the different were 1-pentanol, hexanal, furfuryl alcohol, (E)-2-heptenal, 5-

only in the reaction speed (Houhoula et al., 2003). During the fry- methylfurfural, 1-octen-3-ol, octanal, 2-pentylfuran, (E)-2-octenal,

ing treatment, except the absorbed parts by both fried material and nonanal, (E)-2-nonenal, hexadecanoic acid and pyrazines (Takeoka

frying oil, these compounds volatilize out of the frying system due et al., 1996). According to these studies, though there are many

to the high-temperature and their volatility. On the other hand, the degradation products, they could be classiﬁed into a number of

oxygen content decreases with the proceeding of high temperature categories based on similarity of structure or property.

treatment. In addition, when the state of oxygen-free occurs, many Volatile short-chain aldehyde compounds such as acrolein

other thermal reactions would take place. However, the reaction (2-propenal) have been detected in several cooking oils under

speed of oxidation increases during the frying treatment under the different conditions of frying oil type and frying temperature

condition of high temperature. and time (Katragadda et al., 2010; Osório and de Lourdes,

A classic peroxidation kinetic model of unsaturated fatty acids 2011). Unfortunately, these volatile aldehyde compounds are

which is regarded as three steps (i.e. initiation, propagation and not just small part of the degradation products but most of

termination) has been generally accepted to elucidate free radical them have been found to have certain potential threat to health

reaction pathways of lipid oxidation. The initiation step is catalyzed of consumer (Kamal-Eldin and Appelqvist, 1996; Seppanen and

by many factors such as light, heat and metal ions. Propagation Csallany, 2002), especially the genotoxic and cytotoxic properties

•

step involves the production of peroxyl radicals (ROO ) which then of ␣,␤-unsaturated aldehydes (Guillén and Goicoechea, 2008). The

abstract hydrogen from other organic substrates to form hydroper- genotoxic and cytotoxic oleﬁne aldehyde-contained compounds

•

oxides. Along with the decrease of lipid radicals (R ) during the such as 4-hydroxy-2-trans-hexenal, 4-hydroxy-2-trans-octenal,

• •

ROO ·formation course, an increase of R also occurs in the pro- 4-hydroxy-2-trans-nonenal and 4-hydroxy-2-trans-decenal are

cedure of formation of hydroperoxides. Therefore, the role of R produced and have been reported to be absorbed by the fried

during the propagation step is both reaction substrate and prod- material (Seppanen and Csallany, 2004; Seppanen and Csallany,

uct. In addition, the formed hydroperoxides are very unstable and 2006). The saturated and monounsaturated FAMEs such as methyl

◦

easy to give rise to ␤-scission homolytic cleavage of the O O, C C stearate and methyl oleate model systems heated at 185 C for 0

and C O around peroxide group to decompose into short-chain to 6 h did not generate any of the four ␣,␤-unsaturated 4-hydroxy-

compounds (Porter et al., 1995). The decomposition of hydroper- aldehydes, but 4-hydroxy-2-trans-hexenal existed in the products

oxides promotes the proceeding of propagation step and this step of both methyl linoleate and methyl linolenate model systems due

is a circulatory continuous reaction from reversibility point of view. to heat treatment (Han and Csallany, 2009; LaFond et al., 2011).

The ﬁnal step includes the combination of free radicals which pre- 4-oxo-trans-2-decenal, 4-oxo-trans-2-undecenal, 4-oxo-trans-2-

vent the proceeding of propagation step (Frankel, 1980; Schaich, nonenal, 4-hydroxy- trans-2-nonenal, 4-hydroxy-trans-2-hexenal,

2005). As shown in Fig. 1-1, these compounds include a homolo- and trans-4,5-epoxy-trans-2-decenal were detected in extra vir-

gous series of small molecular alcohols, aldehydes, ketones, acids, gin olive, sunﬂower and virgin linseed oils which were subjected

◦

lactones, etc. One result of thermal oxidation is the occurrence of to 195 C for prolonged periods of time in a discontinuous indus-

volatile compounds which is the main cause of the favorable odor trial fryer (Guillén and Uriarte, 2012). In combination with the

and off-ﬂavor of fried material and frying oil. instability of hydroperoxides, an important fragmentation of the

In a recent study, another alternate kinetic model has been pro- peroxide bond to produce two carbonyl fragments, which named

posed to explain the formation of aldehydes directly from peroxyl as Hock-cleavage, could be used to explain the formation mecha-

radicals through an independent pathway during thermal treat- nism of these oxidized aldehyde compounds (Schneider et al., 2005;

◦

ment of peanut oil at 180 C (Silvagni et al., 2010). As shown Schneider et al., 2001).

by the dashed arrows reactions in Fig. 1-1, two peroxyl radicals The remarkable products of short-chain compounds formed

combined to form two alkoxyl radicals and a molecular oxygen during deep-fat frying or just oil heating have also been inves-

by a bimolecular termination reaction. Actually, an intermediate tigated and identiﬁed. Short-chain fatty acids decomposed from

tetraoxide which is unstable and easily decomposes to the alkoxyl the carbon-carbon bond scission of alkoxy radical were detected in

radical and molecular oxygen at high temperature exist in this reac- frying oil without food materials and used to well indicate the all

tion. In addition, this proposed mechanism was veriﬁed by the alterations and early quality of used frying oil (Brühl and Matthäus,

results of reaction kinetics analysis and reaction products identiﬁ- 2008; Berdeaux et al., 1999a; Márquez-Ruiz and Dobarganes, 1996).

cation by means of NMR and EPR and had the same accordance with Fig. 1-2 shows the formation pathways of typical short-chain

the classical aldehydes formation from hydroperoxides. Under the glycerol-bound compounds from the 9-hydroperoxide of oleyl,

666 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

Catalyst

Initiation: R (CH2)6 CH2 CH CH R' + O2 R (CH2)6 CH CH CH R' + OOH

Catalyst

R (CH2)6 CH2 CH CH R' R (CH2)6 CH CH CH R' + H

Propagation: R (CH2)6 CH CH CH R' + O2 R (CH2)6 CH CH CH R'

OO +R (CH2)6 CH CH CH R'

R (CH2)6 CH CH CH R' + R (CH2)6 CH2 CH CH R'

2R (CH ) CH CH CH R' OO 2 6

R (CH2)6 CH CH CH R' + R (CH2)6 CH CH CH R' OOH

Scission, Rearrangement, etc.

R (CH2)6 CH CH CH R' Acids, Olefine acids Alcohols, Enol OOH Aldehydes, Olefine aldehydes Degradation Ketones, Ketenes compounds Lectones Furans

Etc.

R (CH2)6 CH CH CH R'

Termination : 2R (CH2)6 CH CH CH R'

R' CH CH (CH ) R

CH 2 6

R (CH2)6 CH CH CH R'

R (CH2)6 CH CH CH R' + R (CH2)6 CH CH CH R'

OO O R' CH CH CH (CH2)6 R

OOH HO O

R (CH ) CH 2 6 2 CH CH CH R' R (CH2)6 CH2 CH CH CH R'

O O

(CH ) R: R''-C R'' : Glyceridic backbone R 2 6 CH2 R (CH2)6 CH2 C H R': (CH ) -CH oleic 2 7 3 [H ] [O2] CH2-CH=CH-(CH2)5-CH 3 linoleic O

CH2-CH=CH-CH2-CH=CH-CH2-CH 3 linolenic R (CH 2 ) 6 CH 3 R (CH 2 ) 6 CH 2 C OH

Fig. 1. The possible mechanisms for formation of degradation compounds by the free radical reaction pathways of thermal oxidation (1) and the formation of typical

short-chain compounds from 9-hydroperoxide of major fatty acyl groups (2).

linoleyl and linolenyl groups (Velasco et al., 2004a,b). The same step and slow in the third step. The key phase is the water with

alterations would occur at C C in other aliphatic chain, conse- weak nucleophilicity to react with the protonated ester with strong

quently producing glycerides with more than one short carbon electrophilicity (Gillatt, 2001). However, it is hard to ﬁgure out the

chains. Monoacid TAGs (i.e. triolein and trilinolein) and FAME (i.e., breakage of speciﬁc ester bond position on the status of nonenzy-

methyl oleate and linoleate) were selected as model compounds matic hydrolysis. The breakage sequencing of ester bond at Sn-1,

◦

to undergo 180 C for different time periods and six compounds Sn-2 and Sn-3 might be inﬂuenced by the position and num-

which included methyl heptanoate, methyl octanoate, methyl 8- ber of C C, length of carbon chain, steric hindrance among the

oxo-octanoate, methyl 9-oxononanoate, dimethyl octanodiate and aliphatic chains, moister content, temperature or other unknown

dimethyl nonanodiate were identiﬁed (Berdeaux et al., 2002). In factors.

addition, the ratio between the compounds from 9-hydroperoxide When water and heat simultaneously exist, the decomposi-

and those from 8-hydroperoxide provided a good indication of the tion of ester linkage easily takes place. MAGs and DAGs initially

degree of unsaturation of the frying oil irrespective of the total increased and reached a plateau when potato chips fried in reﬁned

◦

alteration level (Velasco et al., 2005). cottonseed oil by heating from 155 to 195 C though it was not

signiﬁcant (p < 0.05) (Houhoula et al., 2003). The water would be

2.1.2. Hydrolysis products evaporated and the moisture content decrease along with the

Another source of decomposition products is the hydrolysis increase of temperature, which could slow down the proceeding

of TAG which results in the production of diacylglycerol (DAG), of hydrolysis. However, the formed glycerol volatilizes away above

◦

monoacylglycerol (MAG) and glycerol (Dobarganes et al., 2000a,b). 150 C which is in favor of the conduct of hydrolysis (Naz et al.,

Generally, TAG hydrolysis is a reversible reaction course which is 2005). Oil replenishment during deep-fat frying might minimize

promoted by heat or some catalysts and involves three steps. Out the producing of DAGs or MAGs and slow down the hydrolytic

of the feature of reversible reaction, the characteristic of this reac- changes (Romero et al., 1998). With the increase of frying time, the

tion is that the speed is slow in the ﬁrst step, fast in the second content of the formed fatty acids increased. The important result of

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 667

hydrolysis is the increase of the content of free fatty acids, which is happened during the deep-fat frying course, which has been con-

broadly used as an index for monitoring the quality of frying oil. ﬁrmed by detection of the resulted epoxy-TAGs using different

According to above-mentioned degradation compounds and instrumental approaches (Aerts and Jacobs, 2004). It is worth not-

their possible formation pathways, there are various impact factors. ing that the occurrence of cis and trans conﬁguration under the

Frying with food material which refers to the type and moisture formation of epoxy group, which increases the diversity of reaction

content of the material (Yu et al., 1993; Chyau and Mau, 1999; products.

Ramírez et al., 2004) and just heating frying oil (Fullana et al., Except for the above-mentioned short-chain aldehyde com-

2004) have the different composition of degradation products. Fur- pounds, the higher carbonyl compounds, called as core aldehydes

thermore, different frying oil types and their composition (Chyau (Kamal-Eldin et al., 1997), were also detected in deep-fat frying

and Mau, 2001; Özyurt et al., 2011), frying temperature and time, oils. Generally, these products originate from the oxidation of an

aerobic or anaerobic frying signiﬁcantly inﬂuence the chemical unsaturated fatty acid of a triglyceride and the subsequent cleavage

alterations resulting in production of degradation compounds. of the double bond. Used frying fats samples mixed with 2,4-

Degradation products mainly stem from ␤-scission cleavage of dinitrophenylhydrazine to detect the content of core carbonyl has

hydroperoxides and TAG hydrolysis according to previous studies. been investigated and the results showed that the method was

The deep-fat frying is indeed a complex physicochemical pro- helpful and could be used to well judge the quality of frying oil

cess which involves many substrates and reaction mechanisms (Schulte, 2002).

and is affected by various external contributors. The formed com- It is interesting that these oxidized groups were not only inde-

pounds with small molecular not only volatilize away, but also are pendently present in the fatty acyl chains, but also simultaneously

absorbed in the frying system for participating in or promoting exist in the same carbon chain or different carbon chains in the

further sophisticated reactions. glycerol backbone. On the other hand, the unsaturated degree of

the fatty acyl groups affect the ﬁnally products and their composi-

2.2. Oxidized TAG monomers tion. That’s to say, more than one oxygenated function group may

be present in the same fatty acyl chain and more than one oxi-

Structure of fatty acids impacts the chemical reactions hap- dized fatty acyl group may be present in one TAG molecule, such

pened in TAGs during deep-fat frying. The number and position of as the trilinolein and trilinolenin, which have higher unsaturated

C C are the most important structural factors. Oxidation occurred degree.

in the place of C C is the important and easy-to-happen reaction

during the deep-fat frying course. Oxidized TAG monomers involve 2.3. Nonpolar TAG derivatives

many classes of variations which are originated from hydroper-

oxides with extra oxygen. These compounds mainly include the Some non-altered TAGs which showed the nonpolar property

oxidized TAG monomers with keto, epoxy, hydroxyl and aldehyde can present after the deep-fat frying treatment, but it can not fully

groups supporting the modiﬁed ester acyl chains linked to the back- reﬂect the whole components of the nonpolar part. Many other

bone of glycerol (Giuffrida et al., 2004a,b; Guillén and Ruiz, 2008; nonpolar compounds could be formed during the high temperature

Kalogeropoulos et al., 2007; Marmesat et al., 2008). treatment. Nonpolar TAG derivatives mainly include the reaction

Oxo-fatty acids, monohydroxy-fatty acids and polyhydroxy- products without extra oxygen formed during the deep-fat frying

fatty acids were periodically analyzed and identiﬁed in the frying course. According to this deﬁnition, cyclic compounds and trans

fat of fast-food (mostly tallow), olive oil and safﬂower oil (Schwartz isomers which contain the compounds of carbon chains with cyclic

et al., 1994). Similar to the formation mechanism of short aliphatic and trans conﬁguration attached to the glycerol backbone or the

chain glycerides, the producing of oxidized TAG monomers with cyclic and trans monomers are the typical nonpolar TAG deriva-

keto and hydroxyl group also involves free radical reaction of 9- tives. It is worth noting that some extra oxygen-free TAG polymers

hydroperoxide, which are shown in Fig. 2-1 (Velasco et al., 2004a,b). and nitrogen-containing compounds such as heterocyclic amines

From the pathway, it could be observed that thermal-oxidation ini- which also belong to this kind of nonpolar products would be intro-

tiated the alterations. The hydroperoxide then changed into alkoxyl duced in the later sections.

radical to absorb another hydrogen atom to form oxidized TAG

monomers with hydroxyl group (route A) and react with other 2.3.1. Cyclic fatty acid monomers

changed alkoxyl radical to form oxidized TAG monomers with Cyclic fatty acid monomers (CFAMs) are a category of cyclization

keto and hydroxyl group on the basis of radical disproportiona- products intramolecularly or intermolecularly formed by alter-

tion (route B). In other words, the number and position of these ations occurred at C C in the aliphatic chains under condition

◦

two function groups in the carbon chains were determined by the of frying or reﬁning at 200 C or above (Sébédio et al., 1987).

number and position of C C in the original fatty acyl chains. Cyclization can occur in both the fatty acyl chains in TAG and the

Another common oxidized TAG monomers are epoxy-TAGs decomposed fatty acids, as long as the occurrence of C C. Apart

which are produced by epoxidation of C C in the fatty acyl chains. from previously discussed epoxy-TAGs, CFAM ring with only carbon

Olive oil and sunﬂower oil were heated and different monoepoxy atom also present in the deep-fat frying system. In spite of the low

products were identiﬁed. The results indicated that frying oil with concentration of these cyclic monomers present in the frying prod-

different unsaturated degrees led to produce monoepoxides with ucts (Romero et al., 2000a), the suspicious latent biological hazard

different amounts (Velasco et al., 2004a,b). The formation pathway to the health of consumer is a topic of worth exploring (Flickinger

of these products and some typical of them which were detected et al., 1997; Martin et al., 2000; Sébédio and Grandgirard, 1989).

from fatty acid methyl esters (FAMEs) heated by frying conditions is Several vegetable oils have been used to investigate the inﬂu-

shown in Fig. 2-2 (Berdeaux et al., 1999b; Giuffrida et al., 2004a,b; ence thereof on the formation amount of CFAMs and the results

Marmesat et al., 2008). From the proposed mechanism, a direct indicated that frying oil with high oleic acid had well frying effect

attack on the alkenyl in fatty acyl chains by a hydroperoxyl radical, and lower CFAMs yield (Romero et al., 2003, 2006). Fatty acid

resulting in the formation of an epoxy compound and an alkoxyl positional distribution in glycerol backbone has also been con-

radical. In addition, in conjunction with the typical epoxy com- sidered to impact the cyclization rate in model system and come

pounds, the position of epoxy group may present in any of the to the conclusion that the inﬂuence of TAG structure was hia-

original C C position in case of high unsaturated degree. Con- her than that of TAG composition on the formation tendency

version of C C into epoxy group is one of the most alterations of CFAMs, and the C18 polyunsaturated fatty acid in the Sn-2

668 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

OOH

1 R (CH2)6 CH2 CH CH CH R'

R (CH2)6 CH2 CH CH CH R'

A B O

+ RH + R (CH2)6 CH2 CH CH CH R'

OH R O R (CH2)6 CH2 CH CH CH R' R (CH2)6 CH2 C CH CH R' +

OH R (CH2)6 CH2 CH CH CH R'

ROO RO O

2 R (CH ) 2 7 CH CH CH2 R' R (CH2)7 CH CH CH2 R' O

CH HC

CH3OOC CH3 A: cis-9, 10-epoxystearate (CH2)7 O (CH2)7

CH OOC CH HC tran s 3 CH3 B: -9, 10-epoxystearate (CH2)7 O (CH2)7

CH HC

CH3OOC CH CH C: cis-9, 10-epoxyoleate (CH2)7 O CH2 (CH2)4 CH3

CH HC

CH3OOC CH CH D: trans-9, 10-epoxyoleate

(CH2)7 CH2 (CH2)4

O CH3

CH OOC CH HC 3 (CH ) CH CH E: cis-12, 13-epoxyoleate

2 7 CH2 O (CH2)4

CH3

CH OOC CH HC tran s 3 (CH ) CH CH F: -12, 13-epoxyoleate 2 7 CH2 (CH2)4 CH3

R: R'' -C R'' : Glyceridic backbone

R': (CH2)7-CH3 oleic

CH2-CH=CH-(CH2)5-CH3 linoleic

CH2-CH=CH-CH2-CH=CH-CH2-CH3 linolenic

Fig. 2. The possible formation pathways of oxidized TAG monomers with keto and hydroxyl group from 9-hydroperoxide of major fatty acyl chains (1) (Velasco et al., 2004a,b)

and epoxy-TAGs (2) (A-F are the typical epoxy compounds produced by fatty acid methyl esters.).

position of TAG made the oil easily undergo the cyclization reac- different cyclopentenyl, cyclopentyl, cyclohexenyl and cyclohexyl

tion upon heat treatment (Martin et al., 1998a,b,c). It is well known fatty acids would emerge in the frying products (Mossoba et al.,

that C C is the essential for cyclization; however, the degree of 1995). If the cyclic compounds have C C, C C may exist in both

cyclization, content and composition of the formed CFAMs dur- the ring’s interior and the alkyl chains. The existence of C C and its

ing the deep-fat frying course could be varied according to the stereomutation make these cyclic compounds have certain basic

unsaturation degree, position and conﬁguration of C C in differ- carbon skeleton, consequently producing the cis and trans conﬁg-

ent unsaturated aliphatic chains of frying oil (Christie et al., 1993; urations with equal extent (Christie and Dobson, 2000).

Dobson et al., 1997). After cyclization, one C C would be lost, so The formation mechanism of CFAMs could be explained by

most of the formed cyclic fatty acid monomers were mono-oleﬁnic intramolecular rearrangement catalyzed by free radical intermedi-

structure when sunﬂower oil undergone small-scale frying opera- ate, as shown in Fig. 3-1 (Christie and Dobson, 2000; Mossoba et al.,

tions (Dobson et al., 1996a) but diene is the main structure of the 1994). The ﬁrst step of this consecutive reaction is the combination

formed CFAMs when evening primrose oil (Dobson and Sébédio, of aliphatic chain with an allylic hydrogen radical, and sequen-

◦

1999) and linseed oil were heated at 275 C under nitrogen (Dobson tially comes about a ring closure by reaction of an ethylenic bond

et al., 1996b). In addition, the present of monocyclic, dicyclic and with a secondary carbon radical. Afterwards, the formed CFAM

even polycyclic monomers also constitute the cyclic compounds radical reacts with hydrogen radical to produce a stable cyclic struc-

(Chen and Chen, 2003; Mossoba et al., 1996a,b). Therefore, the type ture (A). Another pathway is that the formed CFAM radical loses a

and content of unsaturated fatty acids of frying oil could be reﬂected hydrogen radical to form a new C C structure, thereby producing

in the composition of the formed CFAMs. another kind of product (B). However, this kind of C C positional

According to previous studies, the formed CFAMs were all ﬁve- isomerization has been seldom reported in the investigated CFAMs.

or six-membered ring structures with a carboxyl-contained car- Thus, this radical catalysis of cyclization should be veriﬁed in future

bon chain and a hydrocarbon chain. That’s to say, a wide variety of studies.

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 669

1 R1 R2 R , RO

R1 R2

R R 1 R 2 A R1 R2 RH

R1 R2 R1 H R2 B

R1 R2

2 HOOC 9 10 H 14 HOOC 1014 B

10 HOOC H 9 HOOC 5 9 C 5

H 9 15 15 HOOC HOOC E

HOOC HOO C 4 H 10 4 F 9

9 H HOOC HOOC 9 10 12 G 12 H 9 HOOC H H HOOC 10 H

13 9

Fig. 3. The formation mechanism of cyclic fatty acid monomers (CFAMs) by free radical catalyzation (1, A is the typical formed CFAM and B has seldom been observed in

previous reports) and [1,6]- and [1,7]-prototropic migrations (2) from unsaturated fatty acid (Destaillats and Angers, 2005a).

Another convincible mechanism for interpreting the forma- products were detected. According to this mechanism, the for-

tion of CFAM is the thermally induced [1,6]- and [1,7]-prototropic mations of bicyclic fatty acid monomers, cyclopentenyl and

migrations cooperated with cycloaddition (Destaillats and Angers, cyclohexenyl CFAMs from linolein (Dobson et al., 1997) and

2005a; Spangler, 1976). Eight monocyclic saturated fatty acids linolenic (Dobson and Sébédio, 1999) were also reasonable and

which comprised four basic structures: cyclopentyl fatty acids with corresponded to the product analysis.

rings from C-5 to C-9 and C-10 to C-14, and cyclohexyl fatty acids

with rings from C-4 to C-9 and C-10 to C-15 of the original fatty acid 2.3.2. Trans isomers

chain, were identiﬁed from oleic acid and high-oleate sunﬂower oil In fact, some of the aforementioned cyclic monomers referred

during small-scale frying (Dobson et al., 1996a). According to Fig. 3- to cis/trans isomerization belonged to the trans isomers category.

2 and by observation of the linkage of carbon atoms, vinylic C-9 was Except the trans cyclic compounds, there were several other kinds

involved to form C and D under the induction of [1,6]-prototropic of trans isomers. It is well known that trans isomers of fatty acid

migration and vinylic C-10 was involved to form E and F under have many adverse effects on human health such as coronary heart

the induction of [1,7]-prototropic migration. In addition, the size of disease (Willett et al., 1993), sudden cardiac death (Kummerow,

the formed ring is feasible by the principal of Baeyer strain theory. 2009) and systemic inﬂammation (Mozaffarian, 2006). However,

On one hand, due to the discrepancies between bond angles and the source of trans isomers is very extensive in terms of both

◦

109.5 value of “normal” tetrahedrold and the certain steric hin- raw food materials and food products (Ledoux et al., 2007; Craig-

drance between C-9 and C-12 or C-13, the formation of cyclopropyl Schmidt, 2006). Hydrogenated vegetable oil was widely applied

(G) or cyclobutyl (H) is limited. In the second hand, the carbon on shortening or margarine, but the content of trans isomers was

band angle values of ﬁve-membered and six-membered rings are very considerable (Romero et al., 2000b). In addition, small amount

similar to tetrahedral angle value, hence no angle strain exist and of trans fatty acid also can be observed during edible oil reﬁning

their formations are possible, consequently only the four cyclic (Mezouari and Eichner, 2008; Schwarz, 2000).

670 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

During the deep-fat frying, all the breakage, shift and formation monomers is small under the condition of appropriate frying time

of C C involve the presence of trans conﬁguration. Therefore, it is and temperature. Therefore, there is no misgiving in terms of view

inevitable that the formation of trans fatty acid during vegetable oil of rational intake of fried food, but the understanding of the for-

heating or frying. Trans, trans-2,4-decadienal which related to the mation of these trans products would be beneﬁcial to control the

induction of low density lipoprotein oxidation was by-produced formation of undesirable products.

in fried potatoes (Boskou et al., 2006; Andrikopoulos et al., 2004).

Fortunately, an ordinary frying process in suitable time using un- 2.4. TAG polymerized products

hydrogenated edible oils has little impact on intake of trans fatty

acid from edible oils (Tsuzuki et al., 2010).With the increase of fry- TAG polymers, such as dimmers, trimers and oligomers (Steel

ing time, the amount of trans fatty acid increased but decreased et al., 2006; Tasioula-Margari et al., 1996) are categories of thermal-

when the frying system was added with butylated hydroxyanisole oxidized products whose molecule weight is higher than that of

(BHA) or phenolic extracts of dry rosemary (Gamel et al., 1999; the parent TAGs. Under the two main factors of oxygen and heat-

Tsuzuki et al., 2008). In addition, to select appropriate kind of frying, oxidized polymerization and thermal polymerization occur

ing oil (Martin et al., 1998a,b,c; Ribeiro et al., 2009) which also refers and lead to many complex polymers with the linkages of –C–C–,

to the fatty acid composition of frying oil is a good strategy to avoid –C–O–C– and –C–O–O–C– among the TAG molecules (Stevenson

the formation of trans fatty acid in conventional life. et al., 1984; Christopoulou and Perkins, 1989a). Therefore, oxygen

As a kind of isomerized products, trans conﬁguration-contained is present or not would make different kinds of polymeric products,

compounds were produced from cis-polyunsaturated fatty acids namely non-polar polymers without extra oxygen and polar poly-

during thermal treatment. Generally, lipid oxidation leads to the mers with extra oxygen. With the production and accumulation of

isomerization of cis-polyunsaturated structure to trans conﬁgura- polymers, many undesirable phenomena such as the easily deteri-

tion and very low content of trans fatty acid was produced (Liu oration, color deepening and viscosity increasing of the frying oil,

et al., 2007). The mechanism is illustrated by free radical chain occur and usually indicate the abandonment of the frying oil (Tseng

reaction of methyl oleate oxidation as shown in Fig. 4-1 (Porter, et al., 1996).

1986). Electron rearrangement and migration of C C were rec- Generally, the structure and content of polymeric products are

ommended to form the elaidic acid and other trans octadecenoic affected by frying condition and the nature of frying oil (mainly

acids when the triolein was isothermally heated at 280, 300, and the fatty acid composition) (Tompkins and Perkins, 2000; Takeoka

◦

325 C (Christy et al., 2009). Addition-elimination mechanism has et al., 1997; Choe and Min, 2007), which lead to complex reaction

also been used to interpret the isomerization induced by radical process. When oxygen is absent, dimerization or polymerization

species (Chatgilialoglu et al., 2006; Jiang et al., 1999). Therefore, is achieved by linkage of –C–C– to form dimers or polymers with-

the oxidation supplied free radicals to add with the C C to form a out extra oxygen atom. Two kinds of reactions named free radical

radical-adduct, which then underwent ␤-elimination of the radi- chain reaction and Diels-Alder reaction are mainly used to inter-

cal and formed the thermodynamically stable trans conﬁguration pret the formation mechanism of polymers. By the way, acyclic and

(Fig. 4-2). On the other hand, thermally induced isomerization of cyclic polymers can be produced in the light of different positional

C C in unsaturated fatty acids also exist during deep-fat frying carbons reacting with radicals (mainly allyl radicals). According to

(Christy, 2009a). Arrhenius plot was used to investigate the thermo- Fig. 6-1, on one hand, two allyl radicals bonded to form a dehy-

dynamics of the heat-induced isomerization of triolein (9-cis, 18:1) drodimer. On the other hand, an allyl radical was absorbed by an

and trielaidin (9-trans, 18:1) and the results indicated that heat- unsaturated molecule to form a dimeric radical, consequently com-

induced cis/trans isomerization of triolein and trielaidin occurred bined with a hydrogenous radical to produce a malposed acyclic

mainly through the formation of radical species (Tsuzuki, 2010). dimer. As described in Fig. 6-2, TAGs with polyunsaturated fatty

It is worth mentioning the formation of conjugated fatty acid chain lost a hydrogenous radical to form a radical with structure of

isomers, like conjugated linoleic acid (CLA) by linoleic acid during conjugated diene, which then combined with another unsaturated

◦

heat treatment at 180 and 220 C (Juanéda et al., 2003). Geometri- molecule to form a dimeric radical intermediate, consequently

cal isomerization of linoleic (9-cis, 12-cis, 18:2) acid into conjugated intramolecular addition took place and to form a cyclic dimer. The

C18:2 acids which were mainly composed of all-trans and cis/trans rest high-molecular polymers can be formed in the same manner.

CLA isomers has been investigated (Sébédio et al., 1988). 9t,12t fatty Diels-Alder reaction is an important combination reaction between

acid-contained TAG and its methyl ester also led to form conju- conjugated dioleﬁnic and oleﬁnic structures for formation of cyclo-

gated conﬁguration-contained products when they were heated hexene or cyclohexenyl-contained compounds, which also can be

◦

at 250 C (Christy, 2009b). Similar to the formation mechanism assigned to the cyclic polymers.

of above-mentioned CFAMs, [1,5]-sigmatropic rearrangements of However, extra oxygen is involved in participating in the sophis-

formation of CLA in heated oil with different levels of linoleic ticated chemical reactions during deep-fat frying and forming the

acid has been proposed and veriﬁed by the identiﬁed mono trans above-mentioned oxidized TAG monomers which have one or sev-

CLA isomers, such as 9-cis,11-trans, 9-trans,11-cis, 10-trans,12-cis, eral oxygen-contained groups. Under the effect of radical catalysis,

10-cis,12-trans, 8-trans,10-cis and 11-cis,13-trans octadecadienoic these oxygen-contained TAG monomers would be polymerized

acid (Destaillats and Angers, 2002). A mixture of 9-cis, 11-trans- by linkage of not only –C–C–, but also –C–O–C– and –C–O–O–C–

octadecadienoic acid and 10-trans, 12-cis-octadecadienoic acid (Christopoulou and Perkins, 1989b; Márquez-Ruiz et al., 1995), as

was also used to directly synthesize CLA under the similar per- shown in Fig. 7. Furthermore, one, two, or more additional oxygen

icyclic [1,5]-sigmatropic rearrangement mechanism (Destaillats atoms could be present as bridging oxygen in the formed dimmers,

◦

and Angers, 2003). Furthermore, when methyl linoleate was solely trimers or oligomers when the sample was heated at 190 C for 6 h

◦

heated at 200 and 220 C for a series of time period, the formation of (Byrdwell and Neff, 2004). This combination is consisted with the

9-trans,11-trans and 10-trans,12-trans CLA isomers was explained termination step of lipid oxidation. Allyl radical easily produced

by both a free-radical chain reaction mechanism (1) and [1,3]- by oxidation could combine with an alkoxy radical formed by scis-

sigmatropic rearrangement of intramolecular (2) under thermal soring of hydroperoxides to produce oxydimers. Meanwhile, two

oxidation condition, which can be seen in Fig. 5 (Destaillats and molecules of peroxy radicals bonded together and formed a peroxy

Angers, 2005b). dimer (Choe and Min, 2007).

From the proposed mechanism, the formation of trans isomers Polymers are an inevitable kind of products formed during

is unescapable. However, the yield of these isomers including cyclic deep-fat frying and can involve both the original polyunsaturated

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 671

OO OOH 1 RH R trans 1 11 R2 R1 11 R2 11-

O2 RH R R 9 R2 R 9 R2 cis R1 9 2 1 1 9- OO OOH OO OOH

O2 RH R R R R R tran s 1 11 2 1 11 2 1 11 R2 11-

11 8 R1 R2

OO OOH

O2 RH

cis R1 R R1 R2 R 8- 8 2 8 1 8 R2

O2 RH

10 R 10 R R 10 R trans

R1 R2 1 2 1 2 10- OO HO O R 8 RH trans 1 R2 R1 8 R2 8- OO OOH

R X H 2 2 X + R1 R2 H R2 X + R1

R1 R1= C7H15, R2= (CH2)6COO CH3

RH= methyl oleate

Fig. 4. The formation mechanism of trans conﬁguration induced by free radical chain reaction (1) and addition–elimination (2) of methyl oleate oxidation.

constituents of frying oil and the decomposed products of parent 2002). The known kinds of sterols are cholesterol (animal sterol),

TAGs in various polymerizations. Furthermore, the simultaneous stigmasterol, stigmastenol, ␤-sitosterol, campesterol, brassicast-

formation of non-polar and polar polymers and the formation ten- erol, avenasterol, etc., which broadly exist in various kinds of

dency of the two polymers are still needed to be investigated. food. As a kind of triterpenoid, 28 or 29 carbons and one or two

Therefore, rightly out of the structural complexity and lack of effec- C C (the ﬁrst one in sterol nucleus and the second one in the

tive analysis methods of the polymers, there have been few studies alkyl side chain) are present in the most phytosterols (Moreau

focused on the structure analysis and formation pathways of poly- et al., 2002). Out of the special molecule structure, many reac-

mers produced during deep-fat frying. tions could occur in sterols during the deep-fat frying course.

The most important one is oxidation which leads to many sterol

2.5. Sterol derivatives derivatives and the sterol oxidation products have been widely

investigated (Ryan et al., 2009; Soupas et al., 2007). In addition,

As minor components existing in the oils and fats, sterols, mainly the side-effect of these oxidation products has also been studied

the phytosterols, have many beneﬁts to human health (Ostlund, and some of these products have been considered to have cell

RH R

1 R1 R2 R1 R2

R , RO RH, RO H trans-10,tran-12 18:2

2 R1 R2 R1 H H RH R R

1 R2 R1 R2 trans-9,tr ans-11 18:2

2 R1 2

trans-10 ,tran-12 18:2 R = (CH ) COO H R2 R1 1 2 6

R2= C4H9 H H

R1 R2

trans-9 ,trans-11 18:2

Fig. 5. The formation of conjugated linoleic oil resulted from free radical chain reaction (1) and intramolecular [1,3]-sigmatropic rearrangement (2) during linoleic oil heat

treatment. (Destaillats and Angers, 2005b).

672 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

1 Acyclic polymers

R1-CH2-CH=CH-R2 H R -CH-CH=CH-R + R1-CH-CH=CH-R2 1 2

R1-CH-CH=CH-R2 R1-CH-CH =CH-R2 + R1-CH2-CH=CH-R2 dimers R1-CH-CH=CH-R2 + H R1-CH-CH=CH-R2

R1-CH2-CH-CH-R2 R1-CH2-CH-CH2-R2

+ R1-CH2-CH=CH-R2 R1-CH-CH=CH-R2 R1-CH-CH=CH-R2 R -CH -CH-CH-R + H 1 2 2 R1-CH2-CH-CH-R2 trimers

R1-CH2-CH-CH-R2 R1-CH2-CH-CH2-R2

+ R1-CH2-CH=CH-R2

polymers 2 Cycli c polymers H R1-CH=CH-CH2-CH=CH-R2 R1-CH-CH=CH-CH=CH-R2

+ R1-CH=CH-CH2-CH=CH-R2

R1-CH-CH-CH-CH=CH-R2 R1-CH-CH =CH-CH=CH-R2

R1-CH-CH-CH2-CH=CH-R2 R1-CH-CH-CH2-CH=CH-R2 + H

+ R1-CH=CH-CH2-CH=CH-R2 R1-CH-CH-CH-CH=CH-R2 R1-CH-CH2-CH-CH =CH-R2 R1-CH-CH-CH2-CH=CH-R2

R1-CH-CH-CH2-CH-CH-R2 R1-CH-CH-CH2-CH=CH-R2 R1-CH-CH-CH-CH=CH-R2 + H mon ocyclic diene dimers R1-CH-CH-CH2-CH=CH-R2 R -CH-CH-CH-CH=CH-R 1 2 + H R -CH-CH -CH -CH=CH-R

R1-CH-CH-CH2-CH-CH2-R2 1 2 2 2 bicyclic mon oene dimers R1-CH-CH-CH-CH=CH-R2

R : R'-C R' : Glyceridic backbone

1 R1-CH-CH-CH2-CH=CH-R2

R2: Aliphatic chain mon ocyclic triene trimers

Fig. 6. The formation mechanism of acyclic and cyclic polymers via radical chain reaction on the condition of lacking of extra oxygen during deep-fat frying (Choe and Min,

2007).

␤ cytotoxicity in vivo studies (Adcox et al., 2001; Koschutnig et al., 6␣-hydroxy-3-ketostigmasterol, 6 -hydroxy-3-ketostigmasterol,

␣ ␣

2009). 7-ketostigmasterol, 5 ,6 -epoxystigmasterol, 5␤,6␤-

Similar to the oxidation of TAGs, sterol hydroperoxide and epoxystigmasterol and stigmastanetriol. The oxidation products

its decomposition products would be formed due to the free of other sterols are similar to those of stigmasterol (García-

radical chain reaction. The main identiﬁed oxidation products Llatas and Rodríguez-Estrada, 2011). Further studies showed

are hydroxy, keto and epoxy group-contained compounds (Lampi that the amount of oxidized sterols decreased along with the

et al., 2002; Smith, 1996). Taking stigmasterols for an example increase of heating temperature, indicating that these oxidized

(Kemmo et al., 2005), 6-OOH-stigmasterol, 7-OOH-stigmasterol, sterols could be used as precursors of some other products,

25-OOH-stigmasterol were initially formed and their amount such as sterol dimmers, trimers, oligomers, etc. (Rudzinska´

◦

decreased with the proceeding of heating at 180 C. Free radical et al., 2009, 2010). Thermo-oxidation products of stigmasterol

chain reaction, rearrangement and epoxidation were proposed to have been identiﬁed as monomeric, dimeric and polymeric

interpret the formation of 6␤-hydroxy-stigmasterol, 7␤-hydroxy- products according to their polarity and proposed as a result

␣ stigmasterol, 7 -hydroxy-stigmasterol, 25-hydroxy-stigmasterol, of the dehydration and condensation when the stigmasterol

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 673

O R (CH2)6 CH CH CH R' R (CH2)6 CH CH CH R' + R (CH2)6 CH CH CH R' O

R (CH2)6 CH CH CH R' R (CH ) CH CH CH O O 2 6 R'

R (CH2)6 CH CH CH R' + R (CH2)6 CH CH CH R' O

O R (CH2)6 CH CH CH R'

R: R'' -C R'' : Glyceridic backbone

R': (CH2)7-CH3 oleic

CH2-CH=CH-(CH2)5-CH3 linoleic

CH2-CH=CH-CH2-CH=CH-CH2-CH3 linolenic

Fig. 7. The formation mechanism of acyclic and cyclic polymers via radical chain reaction on presence of extra oxygen during deep-fat frying. Structural formula A shows a

possible extra oxygen-contained trimer (Choe and Min, 2007).

◦

samples were heated at 180 C for different time (Lampi et al., 2011), trans fatty acids (Filip et al., 2011), CLA (Ko et al., 2010), sterol

2009). oxidation products (Kmiecik et al., 2011), acrylamide (Ou et al.,

The effect of concentration and structure of phytosterol on the 2010), nonvolatile compounds which contribute to the negative

performance of frying oil has been studied by adding exogenous odor (Neff et al., 2003), and anti-polymerization effects (Lampi and

phytosterol into frying oil whose indigenous tocopherols and phy- Kamal-Eldin, 1998) were observed under the presence of antiox-

tosterols were aforehand removed away. The added phytosterols idant during the deep-fat frying. On the other hand, under the

signiﬁcantly impacted the thermal and oxidative stability of frying treatment of high temperature, the antioxidation activity of them

oils at higher concentration (Winkler and Warner, 2008a). In addi- would be inﬂuenced, such as ␣- and ␦-tocopherol lost their effec-

◦

tion, the number and location of double bonds in the ring structure tiveness when the temperature was higher than 150 C (Réblová,

have been conﬁrmed to be more related to the anti-polymerization 2006). These alterations were found to be related to the degradation

effect than the presence of an ethylidene group in the side chain of of antioxidant or the interaction between antioxidant and other

phytosterol (Winkler and Warner, 2008b). These phenomena indi- components.

cated that phytosterols should be involved in interaction with the Free radical scavengers have been introduced to describe the

components of frying oil or the intermediate products formed dur- role of the antioxidant when it is added into the frying system

ing the thermal treatment, which should be conﬁrmed in further (Rossi et al., 2007; Yeo et al., 2011). 1,1-diphenyl-2-picrylhydrazyl

research. (DPPH) tests have been investigated and the results indicated that

Sterol derivatives in fried food also have been investigated. After the rosemary extract showed more antioxidative effect than that of

◦

14th frying at 180 C for 24.5 min in total, the amount of total ␣-, ␥-, ␦-tocopherols alone, their homologous mixture and BHT in

phytosterols decreased and the amount of total oxyphytosterols the TAGs isolated from rapeseed oil (Nogala-Kalucka et al., 2005).

(mainly the epoxy- and 7-hydroxyphytosterols) increased in both In view of the free radical reaction easily occurred during deep-fat

frying oil and French-fries (Rudzinska´ et al., 2005). According to course, a lot of free radicals are present in the frying system. The

previous studies of different thermally treatments, the total phytos- antioxidative effect is related to their ability to inhibit the formation

terol content decreased more rapidly in French-fries and ﬁsh ﬁllets of free radicals or quench the formed free radicals. From this point

than that of noodles, minced meats and readymade ﬁsh products of view, antioxidant works on the initial steps of the free radical

which may resulted from the initial high total phytosterol con- reaction in order to inhibit the formation of the above-mentioned

tent of the former products (Derewiaka and Obiedzinski,´ 2012). products.

However, the correlation between the phytosterol loss or formed Besides the description of antioxidant loss during frying, the

oxyphytosterols and the fatty acid proﬁle of chosen frying oil was speciﬁc chemical alterations occurred in antioxidants have been

not signiﬁcant (Tabee et al., 2008; Winkler et al., 2007). Due to reported for these years (Barrera-Arellano et al., 1999; Bruscatto

complex chemical environment of deep-fat frying system, more et al., 2009). However, its loss during the high-temperature treat-

derivatives of sterol could be produced and shouldl be isolated and ment is possibly related with their degradation for the reason

identiﬁed in further research. of their volatility (Marmesat et al., 2010). The ␣-tocopherol

oxidation products, ␣-tocopherolquinone, 4a,5-epoxy-␣-

␣

2.6. Antioxidant alterations tocopherolquinone, and 7,8-epoxy- -tocopherolquinone, have

been identiﬁed when the triolein mixed with ␣-tocopherol and

Due to the loss of natural antioxidants in frying oil during the heated at different temperatures for different times (Verleyen et al.,

reﬁning course, some natural and synthetic antioxidants such as 2001a,b). It was proposed that peroxyl radicals were captured by

␣

tocopherol, butylated hydroxytoluene (BHT), BHA, etc., are always -tocopherol to form the polar oxidation products (Verleyen et al.,

added into frying oil to prevent the oxidation of frying oil and 2001a,b). Tocopherolquinones increased along with the increase

improve the performance of deep-fat frying course. Speciﬁcally, of frying time when the heating treatment of sunﬂower and soy-

␣

the reduction of formation amount of total polar compounds bean oils mixed with -tocopherol (Rennick and Warner, 2006).

(Aladedunye and Przybylski, 2011), 4-hydroxynonenal (Gerde et al., Esteriﬁcation and dimerization were proposed to interpret the

674 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

formation of 2,4 -di-tert-butyl-5 -hydroxy-2 ,4-dimethoxy- 2.7.1. Nitrogen- and sulphur-containing heterocyclic compounds

diphenyl ether, 2,2 -dihydroxy-5,5 -dimethoxy-3,3 -di-tert- With the exception of previously discussed volatile decom-

butylbiphe, etc., when BHA are subjected to high temperature position compounds, nitrogen- or sulfur-containing heterocyclic

treatment, which had been investigated with other three synthetic compounds are other typical volatile products which contribute to

◦

antioxidants in heating of individual antioxidant at 185 C for the main odor of frying oil or fried material (Van Loon et al., 2005;

different heating time (Hamama and Nawar, 1991). Hartman et al., 1983; Kiatsrichart et al., 2003). Obviously, the source

of nitrogen or sulfur is protein which extensively exist in common

fried material.

2.7. Products derived from interactions between frying oil and A total of 130 volatile compounds were isolated and identi-

◦

food material constituents ﬁed from fried chicken at 185 C for 8 min (Tang et al., 1983).

Except for the little molecular compounds formed by degra-

Based on aforementioned detected products and their possi- dation of hydroperoxides, several heterocyclic compounds of

ble formation mechanisms, it can be observed that deep-fat frying pyridines, thiazoles, thiazolines, oxazoles, oxazolines, thiophenes,

without food material is already a complex physicochemical pro- pyrroles, furans, and thialdine were also present in the volatile

cess. However, the practical deep-fat frying is a more intricate part. Apart from lipid auto-oxidation and lipolysis, proteoly-

process which involves not only the components of frying oil, but sis was also the reason for the formation of ﬂavor compounds

also the components of food material, such as water, protein, lipid, (Jerkovic´ et al., 2007). 2,4-decadienal which is the well known

carbohydrate, inorganic salt in participating in the various alter- as a secondary lipid oxidation product was reacted with either

ations (Dobarganes et al., 2000a,b). In other words, it is considerably cysteine or glutathione and come to the following conclusions:

abstruse that a system refers to so many reaction substrates and both the two mixtures had complex reactions occurred under

drastic reaction condition. The following content introduced is on the simulated fried condition and the corresponding hetero-

the basis of studies reported previously. cyclic compounds were formed. The possible reaction pathway

Strictly, the hydrolysis products come from the reaction of 2-pentylpyridine production which involves electrocyclic and

between TAGs and food material constituents in that the water aromatization reactions is shown in Fig. 8-1 (Henderson and Nawar,

is present by food materials. However, they were discussed 1981; Zhang and Ho, 1989). Model systems with different two

in an indie unit due to their properties of degradation prod- kinds of amino acids under simulated deep-fat frying conditions

◦

ucts. Even though except for water, the conventional fried food of 180–183 C in corn oil were investigated and the results of

material is still a sophisticated composition matrix and each producing of volatile pyrazines were obtained (Chun and Ho,

constituent of food material could participate in the complex 1997).

reactions during deep-fat frying (Karel et al., 1975). Broadly speak- As a kind of minor byproduct of food processing, heterocyclic

ing, heat and mass transfer dominate the whole frying process aromatic amines (HAAs) have been recognized as an important kind

involving interrelated physicochemical alterations among vari- of potential mutagens and carcinogens for human health (Cheng

ous components of both frying oil and food material (Dana and et al., 2006; Jägerstad and Skog, 2005). The main precursors of these

Saguy, 2006; Mellema, 2003). Consequently, the delicious taste HAAs are creatinine, reducing sugars and amino acids which com-

and attractive color and aroma are obtained by suitable frying monly exist in the deep-fat frying system. The maillard reaction

condition. and strecker degradation were also proposed to be the formation

The typical and well-studied alterations come into contact with pathways of these HAAs, such as imidazo-quinolines and imidazo-

the changes of both frying oil and food material components are quinoxalines in the fried products (Jägerstad et al., 1998; Murkovic,

lipid oxidation and maillard reaction. Maillard reactions which are 2004). Along with the increase of frying temperature and frying

also called as non-enzymatic browning reactions between carbonyl time, the amount of HAAs increased. The inﬂuence of frying oil type

and amino group are a series of important chemical alterations on the formation of HAAs has also been studied. The results indi-

contributing to both color and odor of food in processing (Hodge, cated that the amount of HAAs was signiﬁcantly lower in sunﬂower

1953). On one hand, the formation of aldehydic compounds sup- seed oil and margarine than that in butter, margarine fat phase, liq-

plies the presence of carbonyl group in the frying system as a result uid margarine and rapeseed oil when the beefburgers were fried at

◦

of the thermal oxidation. On the other hand, the amino group was 165 and 200 C (Johansson et al., 1995) and extra virgin olive oil was

supplied by frying materials. effective to inhibit the formation of HAAs in the fried beefsteak in

These two kinds of chemical alterations run through the whole suitable adding amount (Lee et al., 2011). Fortunately, heating could

frying course and both inﬂuence other reactions. But most impor- be used to decrease the amount of imidazo-quinoxalines and 2-

tant of all, these two typical alterations had certain relation in amino-1-methyl-6-phenylimidazo[4,5-b]pyridine and the content

the mutual inﬂuence of intermediates and end-products. As a of unsaturated fatty acids in frying and their primary or secondary

kind of maillard reaction products, nonenzymatically browned pro- lipid oxidation products were related with the chemical degrada-

teins had been conﬁrmed to have good antioxidative effect and tion of HAAs (Randel et al., 2007).

well induce lipid oxidation (Mastrocola et al., 2000; Ahmad et al., Garlic slices were treated by frying, oil-cooking, microwave-

1998; Alaiz et al., 1997). The interaction relationships between frying, baking, and microwave-baking and the produced volatile

the lipid oxidation and maillard reaction has been reviewed and compounds were studied and identiﬁed. The identifying analy-

come to the conclusion that both reactions were so interrelated sis indicated that the sulfur-containing volatile compounds were

that they should be considered simultaneously to understand the possibly decomposed and/or rearranged products of alk(en)yl thio-

products of the lipid oxidation in the presence of proteins and vice sulﬁnates (Yu et al., 1993). Cysteine and corn oil were mixed and

◦

versa (Zamora and Francisco, 2005; Zamora and Hidalgo, 2011). heated at 180 C for 4 h and ﬁfty-four volatile compounds were

Thus, even under the condition of absence of reducing sugar, the identiﬁed. Being produced in greatest amount, 2-alkylthiophenes

␣ ␤

maillard reaction would also occur during deep-fat frying and was proposed to be formed from reaction of , -unsaturated

the main volatile nitrogen-containing heterocyclic compounds, fatty aldehydes and hydrogen sulﬁde, as explained in Fig. 8-2

such as pyrazines, pyridines, pyrroles, etc. could be produced. (Macku and Shibamoto, 1991b). As a nucleophile, hydrogen sulﬁde

␣ ␤

Another worth concerned nitrogen-containing compound is acry- attacked , -unsaturated fatty aldehydes and made the oxygen

lamide which has become a hot topic in the ﬁeld of baking atom replaced by sulfur atom. Afterward, the resulting sulphur-

food. containing radical rearranged its electrons through carbon atoms

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 675

O C H

CH CH CH CH CH CH CH CH (CH ) (CH ) 1 2 4 H2O 2 4 CH3 A + CH3 N CH NH2 B HO CH R R COOH O R CH (CH ) COOH 2 4 N (CH ) CH3 2 4 N CH3 CH R COOH H O N OH R: N (CH2)4 H N O O CH3 HS

H O

O 2 S S

2 + R

R-CH2-CH=CH-C + H2S R-CH2-CH=CH-C R-CH-C H=CH-C H H RH H

HC CH HC CH HC CH

HC C-R HC C-R C CH-R S H S H H S + OH OH HC CH

HC C-R R

S H H2O S

Fig. 8. The possible reaction pathways for the formation of 2-pentylpyridine from 2,4-decadienal (A) and glutathione (B) (Zhang and Ho, 1989) and 2-alkylthiophenes from

cysteine and corn oil on the condition of heating (Macku and Shibamoto, 1991b).

and reacted with a hydroxyl radical to produce a thiophene ring by mixed with asparagines to investigate the formation degree of

dehydration. AA under certain frying conditions. The results indicated that

␣ ␤ ␥ ␦

, , , -diunsaturated carbonyl compounds were the most reac-

tive compounds to participate in the synergism of AA (Zamora and 2.7.2. Acrylamide

Hidalgo, 2008).

It is well known that acrylamide (AA) possesses neurotoxicity,

As it often happens, conventional fried material such as potato

genetic toxicity and carcinogenicity (Friedman, 2003; Hagmar et al.,

and cereal food contains the reaction substrates of AA (Matthäus

2005; Keramat et al., 2011a,b) and it has been highly kept a watch-

et al., 2004). It is easy to form AA when these foods are subjected

ful eye on since it was observed in heated starch-rich foodstuffs

to high temperature process. It also could be said that the for-

with high amount (Zhang et al., 2005; Tareke et al., 2002). Dur-

mation of AA is related to the type of the fried food. During the

ing the food processing, AA is mainly produced by the reaction

process of potato chips, when the frying temperature and other

between some amino acids and reducing sugar on high temper-

factors were controlled, the yield of AA directly related to the fry-

ature or during the course of maillard reaction (Keramat et al.,

ing time (Romani et al., 2008). Secondly, the yield of AA increased

2011a,b; Stadler et al., 2002). As shown in Fig. 9 (Mottram et al.,

when the frying temperature was increased (Fiselier et al., 2006;

2002), an amino acid combined with a dicarbonyl compound by

Sanny et al., 2012). Furthermore, the effect of frying oil types on

losing a water and then subjected to intramolecular rearrangement

the formation of AA has been studied and no signiﬁcant differ-

of ion and strecker degradation to form an acrolein, a hydrogen

ences were observed among the vegetable oils (Mestdagh et al.,

nitride and other compounds or strecker aldehyde. At last, AA was

2005). But the AA concentration was much higher when palm olein

formed from the combination of acrylic acid and hydrogen nitride

was used as frying oil due to its high content of 6 to 8% of diglyc-

or ammonium originated from nitrogen containing compounds.

erides. This might be attributed to another formation pathway of

Except the reducing sugar source of carbonyl compound,

the AA precursor of acrolein which derived from electro rearrange-

another possible source is the lipid oxidation which occurs during

ment during the hydrolysis of MAGs (Gertz and Klostermann, 2002)

deep-fat frying. In addition, the more susceptible to oxidation of

or formed from glycerol when the frying temperature was higher

the frying oil, the more AA was formed in model systems (Capuano

than the smoke point of frying oil (Claeys et al., 2005). There-

et al., 2010). Obviously, some of the lipid oxidation products pro-

fore, control of the type of frying oil, frying temperature and time

mote or participate in the formation of AA. Some typical lipid

are effective to control the producing of AA during French fries

oxidation products such as 2,4-decadienal, 2-octenal, methyl 13-

process.

oxooctadeca-9,11-dienoate and 4,5-epoxy-2-decenal have been

676 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

Amino acid Dica rbonyl compoun d R

X-CH-NH H N 1 2 O O 2 + Amino C ketone

R1-C=C-R2 H N O OH 2 O R2 + C-CH -CHO

- H2O H 2 O ?

N X C Strecker X R1 X aldeh yde - CO H N

H 2 C=N R1 O 2 H ? C-CH=C H Acrylamide O OO R2 2 O HO R2 H R1-CO-CO-R2

+NH3 + CH3S H

NH3 +

+ CH2=CH-CHO CH2=CH-C OOH CH2=CH-COO NH4

Acrol ein

Fig. 9. The proposed formation mechanism of acrylamide between asparagines or methionine and dicarbonyl compounds from maillard reaction. (It is asparagines when X

means –CH2CONH2 and methionine when X means –CH2CH2SCH3.) (Mottram et al., 2002).

3. Summary water supplies a nucleophile and the hydrolysis is supposed to be an

ionic reaction. In addition, the proposed formation mechanism of

The beneﬁts of well understanding of the chemical alterations 2-alkylthiophenes is another application of nucleophilic reaction.

occurred during deep-fat frying are mainly embodied in analyzing Concerted reaction is an important reaction which just refers to

the beneﬁcial part (e.g. attractive favor, color and tasty crispness the transition state of chemical bond transformation or the forma-

resulted from maillard reaction and caramelization) and hazard tion and breakage of chemical bond in one step. The recommended

part (e.g. trans fatty acids and AA formed during frying) of the [1,j]-prototropic migrations and sigmatropic rearrangement well

products produced during the high-temperature process in order explain the formation of some CMFAs and trans isomers. More-

to improve the mouthfeel and safety of fried food, choose suitable over, the diene synthesis, i.e. Diels-Alder reaction also belongs to

frying conditions, determine the appropriate service life of the fry- concerted reaction. Therefore, the deep-fat frying products which

ing oil, etc. and reusing the used frying oil in the ﬁeld of energy and have been detected and identiﬁed so far are well interpreted by

chemical industry. At last, it is useful for both healthy prospective these three reactions.

of consumers and the economic prospective of fried food industry However, on the basis of the three main organic chemistry reac-

or the whole society. tions and the conditions of high temperature and various reaction

Though the reaction products formed during deep-fat frying are substrates, the possible reaction products are more complex than

of multiformity and complexity, they could be identiﬁed and clas- those of known at present. Take polymers as example, the pos-

siﬁed into several categories according to their similar nature. In sible produced structure is so multifarious that it is difﬁcult and

addition, the reaction substrates also have certain exploitable rules unnecessary to clearly understand their speciﬁc formation path-

such as the TAGs proﬁle of frying oil and the presence of water ways. Furthermore, being accompanied with the development of

in conventional food frying. As previously mentioned, the applied instrument and its raised detection precision, more and more trace

frying conditions which comprise frying temperature, time, con- products would be detected and identiﬁed. As a result, to con-

tinuous or incontinuous frying, fryer type, oil turnover rate or fresh clude and interpret the formation mechanism of reaction products

oil replenishment, etc. would impact the product types and for- become more difﬁcult. These obstacles should be paid more atten-

mation routes. Therefore comprehensively consideration of all of tion to and conquered in the future studies.

these possible factors is achievable to expound the product forma-

tion mechanism and forecast the possible products during deep-fat

frying. References

According to above-mentioned varieties of products, there are

Adcox, C., Boyd, L., Oehrl, L., Allen, J., Fenner, G., 2001. Comparative effects of phytos-

several overlaps among the varieties and even more mutual inﬂu-

terol oxides and cholesterol oxides in cultured macrophage-derived cell lines.

ence among them. From their corresponding possible formation Journal of Agricultural and Food Chemistry 49, 2090–2095.

Aerts, H.A.J., Jacobs, P.A., 2004. Epoxide yield determination of oils and fatty acid

mechanisms, homolytic reaction, also called as free radical reac-

methyl esters using H NMR. Journal of the American Oil Chemists’ Society 81,

tion, heterolytic reaction, also called as ionic reaction and concerted

841–846.

reaction have been successfully used to interpret their forma- Ahmad, I., Alaiz, M., Zamora, R., Hidalgo, F.J., 1998. Effect of oxidized lipid/amino acid

tion pathways. These three reactions dominate the whole typical reaction products on the antioxidative activity of common antioxidants. Journal

of Agricultural and Food Chemistry 46, 3768–3771.

reactions of organic chemistry, which also reﬂect the complex of

Ahromrit, A., Nema, P.K., 2010. Heat and mass transfer in deep-frying of pumpkin,

deep-fat frying course.

sweet potato and taro. Journal of Food Science and Technology 47, 632–637.

Free radicals are pervasive during the whole frying course in Aladedunye, F.A., Przybylski, R., 2011. Antioxidative properties of phenolic acids

and interaction with endogenous minor components during frying. European

spite of their instantaneous lifetime. The initiation step of ther-

Journal of Lipid Science and Technology 113, 1465–1473.

mally oxidation is the main source for the presence of free radicals.

Alaiz, M., Hidalgo, F.J., Zamora, R., 1997. Antioxidative activity of nonenzymati-

As a kind of reactive intermediate, free radicals easily attack other cally browned proteins produced in oxidized lipid/protein reactions. Journal of

Agricultural and Food Chemistry 45, 1365–1369.

chemicals and capture their electro to form a more stable sub-

Alvis, A., Velez, C., Rada-Mendoza, M., Villamiel, M., Villada, H.S., 2009. Heat transfer

stance. In the sight of this point of view, the interpreting to the

coefﬁcient during deep-fat frying. Food Control 20, 321–325.

formation of some volatile products, short-chain compounds, oxi- Andrikopoulos, N.K., Chiou, A., Mylona, A., Boskou, G., Dedoussis, G.V.Z., 2004. Mon-

itoring of 2,4-decadienal in oils and fats used for frying in restaurants in Athens,

dized TAG monomers, CFAMs, trans isomers polymers, heterocyclic

Greece. European Journal of Lipid Science and Technology 106, 671–679.

compounds, etc. is reasonable and acceptable. Heterolytic reaction

Aro, A., Kosmeijer-Schuil, T., Van De Bovenkamp, P., Hulshof, P., Zock, P., Katan, M.B.,

seems not so obviously like homolytic reaction, but the presence of 1998. Analysis of C18:1 cis and trans fatty acid isomers by the combination of

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 677

gas-liquid chromatography of 4,4-dimethyloxazoline derivatives and methyl Christy, A.A., 2009a. Thermally induced isomerization of trilinolein and trilinoe-

◦

esters. Journal of the American Oil Chemists’ Society 75, 977–985. laidin at 250 C: analysis of products by gas chromatography and infrared

Barrera-Arellano, D., Márquez-Ruiz, G., Dobarganes, M.C., 1997. A simple procedure spectroscopy. Lipids 44, 1105–1112.

to evaluate the performance of fats and oils at frying temperatures. Grasas y Christy, A.A., 2009b. Evidence in the formation of conjugated linoleic acids from ther-

Aceites 48, 231–235. mally induced 9t12t linoleic acid: a study by gas chromatography and infrared

Barrera-Arellano, D., Ruiz-Méndez, V., Márquez-Ruiz, G., Dobarganes, M.C., 1999. spectroscopy. Chemistry and Physics of Lipids 161, 86–94.

Loss of tocopherols and formation of degradation compounds in triacylglycerol Christy, A.A., Xu, Z., Harrington, P.B., 2009. Thermal degradation and isomerisation

model systems heated at high temperature. Journal of the Science of Food and kinetics of triolein studied by infrared spectrometry and GC–MS combined with

Agriculture 19, 1923–1928. chemometrics. Chemistry and Physics of Lipids 158, 22–31.

Bastida, S., Sánchez-Muniz, F.J., 2002. Polar content vs. TAG oligomer content in the Chu, Y.H., Luo, S., 1994. Effects of sugar, salt and water on soybean oil quality during

frying-life assessment of monounsaturated and polyunsaturated oils used in deep frying. Journal of the American Oil Chemists’ Society 71, 897–900.

deep-frying. Journal of the American Oil Chemists’ Society 79, 447–451. Chun, H.K., Ho, C.T., 1997. Volatile nitrogen-containing compounds generated from

Berdeaux, O., Márquez-Ruiz, G., Dobarganes, M.C., 1999a. Selection of methylation maillard reactions under simulated deep-fat frying conditions. Journal of Food

procedures for quantitation of short-chain glycerol-bound compounds formed Lipids 4, 239–244.

during thermoxidation. Journal of Chromatography A 863, 171–181. Chyau, C.C., Mau, J.L., 1999. Release of volatile compounds from microwave heating

Berdeaux, O., Márquez-Ruiz, G., Dobarganes, M.C., 1999b. Characterization, quan- of garlic juice with 2, 4-decadienals. Food Chemistry 64, 531–535.

titation and evolution af monoepoxy compounds formed in model systems of Chyau, C.C., Mau, J.L., 2001. Effects of various oils on volatile compounds of deep-fried

fatty acids methyl esters and monoacid triglycerides heated at high temperature. shallot ﬂavouring. Food Chemistry 74, 41–46.

Grasas y Aceites 50, 53–59. Claeys, W.L., Vleeschouwer, K.D., Hendrickx, M.E., 2005. Quantifying the formation

Berdeaux, O., Velasco, J., Márquez-Ruiz, G., Dobarganes, M.C., 2002. Evolution of carcinogens during food processing: acrylamide. Trends in Food Science &

of short-chain glycerol-bound compounds during thermoxidation of FAME Technology 16, 181–193.

and monoacid TAG. Journal of the American Oil Chemists’ Society 79, Craig-Schmidt, M.C., 2006. World-wide consumption of trans fatty acids. Atheroscle-

279–285. rosis Supplements 7, 1–4.

Berdeaux, O., Fournier, V., Lambelet, P., Dionisi, F., Sébédio, J.L., Destaillats, F., 2007. Dana, D., Saguy, I.S., 2006. Review: mechanism of oil uptake during deep-fat frying

Isolation and structural analysis of the cyclic fatty acid monomers formed from and the surfactant effect-theory and myth. Advances in Colloid and Interface

eicosapentaenoic and docosahexaenoic acids during ﬁsh oil deodorization. Jour- Science 128, 267–272.

nal of Chromatography A 1138, 216–224. Derewiaka, D., Obiedzinski,´ M., 2012. Phytosterol oxides content in selected

Boskou, D., 2003. Frying fats. In: Sikorski, Z.E., Kołakowska, A. (Eds.), Chemical Func- thermally processed products. European Food Research and Technology 234,

tional Properties of Food Lipids. CRC Press, Boca Raton, Florida, pp. 320–338. 703–712.

Boskou, G., Salta, F.N., Chiou, A., Troullidou, E., Andrikopoulos, N.K., 2006. Content Destaillats, F., Angers, P., 2002. Evidence for [1,5] sigmatropic rearrangements of CLA

of trans,trans-2,4-decadienal in deep-fried and pan-fried potatoes. European in heated oils. Lipids 37, 435–438.

Journal of Lipid Science and Technology 108, 109–115. Destaillats, F., Angers, P., 2003. Directed sequential synthesis of conjugated linoleic

7,9 12,14

Brewer, M.S., Vega, J.D., Perkins, E.G., 1999. Volatile compounds and sensory char- acid isomers from to . European Journal of Lipid Science and Tech-

acteristics of frying fats. Journal of Food Lipids 6, 47–61. nology 105, 3–8.

Bruscatto, M.H., Zambiazi, R.C., Sganzerla, M., Pestana, V.R., Otero, D., Lima, R., Paiva, Destaillats, F., Angers, P., 2005a. On the mechanisms of cyclic and bicyclic fatty acid

F., 2009. Degradation of tocopherols in rice bran oil submitted to heating at monomer formation in heated edible oils. European Journal of Lipid Science and

different temperatures. Journal of Chromatographic Science 47, 762–765. Technology 107, 767–772.

Brühl, L., Matthäus, B., 2008. Short-chain fatty acids as marker for the degradation Destaillats, F., Angers, P., 2005b. Thermally induced formation of conjugated iso-

of frying fats and oils. Lipid Technology 20, 60–63. mers of linoleic acid. European Journal of Lipid Science and Technology 107,

Byrdwell, W.C., Neff, W.E., 2004. Electrospray ionisation MS of high M. W. TAG 167–172.

oligomers. Journal of the American Oil Chemists’ Society 81, 13–26. Dincer, I., 1996. Modelling for heat and mass transfer parameters in deep-frying of

Caldwell, J., Cooke, B., Greer, M., 2011. High performance liquid products. Heat and Mass Transfer 32, 109–113.

chromatography–size exclusion chromatography for rapid analysis of Dobarganes, M.C., Márquez-Ruiz, G., Velasco, J., 2000a. Interactions between fat and

total polar compounds in used frying oils. Journal of the American Oil Chemists’ food during deep-frying. European Journal of Lipid Science and Technology 102,

Society 88, 1669–1674. 521–528.

Caponio, F., Gomes, T., Pasqualone, A., Summo, C., 2007. Use of the high performance Dobarganes, M.C., Velasco, J., Dieffenbacher, A., 2000b. Determination of polar com-

size exclusion chromatography analysis for the measurement of the degree pounds, polymerized and oxidized triacylglycerols, and diacylglycerols in oils

of hydrolytic and oxidative degradation of the lipid fraction of biscuits. Food and fats – results of collaborative studies and the standardized method (Tech-

Chemistry 102, 232–236. nical Report). Pure and Applied Chemistry 72, 1563–1575.

Capuano, E., Oliviero, T., Ac¸ ar, Ö.C¸ ., Gökmen, V., Fogliano, V., 2010. Lipid oxida- Dobson, G., Christie, W.W., Brechany, E.Y., Sébédio, J.L., Le Quere, J.L., 1995. Silver ion

tion promotes acrylamide formation in fat-rich model systems. Food Research chromatography and gas chromatography-mass spectrometry in the structural

International 43, 1021–1026. analysis of cyclic dienoic acids formed in frying oils. Chemistry and Physics of

Carrieri, G., Anese, M., Quarta, B., De Bonis, M.V., Ruocco, G., 2010. Evaluation of Lipids 75, 171–182.

acrylamide formation in potatoes during deep-frying: the effect of operation Dobson, G., Christie, W.W., Sébédio, J.L., 1996a. Gas chromatographic properties of

and conﬁguration. Journal of Food Engineering 98, 141–149. cyclic dienoic fatty acids formed in heated linseed oil. Journal of Chromatogra-

Cheng, K.W., Chen, F., Wang, M.F., 2006. Heterocyclic amines: chemistry and health. phy A 723, 349–354.

Molecular Nutrition and Food Research 50, 1150–1170. Dobson, G., Christie, W.W., Sébédio, J.L., 1996b. Monocyclic saturated fatty acids

Chang, S.S., Peterson, R.J., Ho, C.T., 1978. Chemical-reactions involved in deep-fat formed from oleic acid in heated sunﬂower oils. Chemistry and Physics of Lipids

frying of foods. Journal of the American Oil Chemists’ Society 55, 718–727. 82, 101–110.

Chatgilialoglu, C., Ferreri, C., Lykakis, I.N., Wardman, P., 2006. Trans-fatty acids and Dobson, G., Christie, W.W., Sébédio, J.L., 1997. Saturated bicyclic fatty acids formed

radical stress: what are the real culprits? Bioorganic & Medicinal Chemistry 14, in heated sunﬂower oils. Chemistry and Physics of Lipids 87, 137–147.

6144–6148. Dobson, G., Sébédio, J.L., 1999. Monocyclic dienoic fatty acids formed from (-

Chatzilazarou, A., Gortzi, O., Lalas, S., Zoidis, E., Tsaknis, J., 2006. Physicochemical linolenic acid in heated evening primrose oil. Chemistry and Physics of Lipids 97,

changes of olive oil and selected vegetable oils during frying. Journal of Food 105–118.

Lipids 13, 27–35. Eulitz, K., Yurawecz, M.P., Sehat, N., Fritsche, J., Roach, J., Mossoba, M., Kramer, J.,

Chen, Y.C., Chen, B.H., 2003. Determination of polycyclic aromatic hydrocarbons in Adlof, R., Ku, Y., 1999. Preparation, separation, and conﬁrmation of the eight

fumes from fried chicken legs. Journal of Agricultural and Food Chemistry 51, geometrical cis/trans conjugated linoleic acid isomers 8,10- through 11,13-18:2.

4162–4167. Lipids 34, 873–877.

Cho, I.K., Kim, S., Khurana, H.K., Li, Q.X., Jun, S., 2011. Quantiﬁcation of trans fatty Filip, S., Hribar, J., Vidrih, R., 2011. Inﬂuence of natural antioxidants on the formation

acid content in French fries of local food service retailers using attenuated of trans-fatty-acid isomers during heat treatment of sunﬂower oil. European

total reﬂection – Fourier transform infrared spectroscopy. Food Chemistry 125, Journal of Lipid Science and Technology 113, 224–230.

1121–1125. Fiselier, K., Bazzocco, D., Gama-Baumgartner, F., Grob, K., 2006. Inﬂuence of the frying

Choe, E., Min, D.B., 2007. Chemistry of deep-fat frying oils. Journal of Food Science temperature on acrylamide formation in French fries. European Food Research

72, 77–86. and Technology 222, 414–419.

Christie, W.W., Brechany, E.Y., Sébédio, J.L., Le Quéré, J.L., 1993. Silver ion chromatog- Flickinger, B.D., McCusker, R.H., Perkins, E.G., 1997. The effects of cyclic fatty acid

raphy and gas chromatography-mass spectrometry in the structural analysis of monomers on cultured porcine endothelial cells. Lipids 32, 925–933.

cyclic monoenoic acids formed in frying oils. Chemistry and Physics of Lipids 66, Frankel, E.N., 1980. Lipid oxidation. Progress in Lipid Research 19, 1–22.

143–153. Friedman, M., 2003. Chemistry, biochemistry, and safety of acrylamide. A review.

Christie, W.W., Dobson, G., 2000. Formation of cyclic fatty acids during the frying Journal of AgrIcultural and Food Chemistry 51, 4504–4526.

process. European Journal of Lipid Science and Technology 102, 515–520. Firestone, D., 1993. Worldwide regulation of frying fats and oils. Information 4,

Christopoulou, C.N., Perkins, E.G., 1989a. Isolation and characterization of dinners 1366–1371.

formed in used soybean oil. Journal of the American Oil Chemists’ Society 66, Fullana, A., Carbonell-Barrachina, Á.A., Sidhu, S., 2004. Volatile aldehyde emissions

1360–1370. from heated cooking oils. Journal of the Science of Food and Agriculture 84,

Christopoulou, C.N., Perkins, E.G., 1989b. Dimer acids: synthesis and mass spec- 2015–2021.

trometry of the tetrahydroxy, dihydroxy, and diketo dinners of methyl stearate. Gamel, T.H., Kiritsakis, A., Petrakis, C., 1999. Effect of phenolic extracts on trans fatty

Journal of the American Oil Chemists’ Society 66, 1344–1352. acid formation during frying. Grasas y Aceites 50, 421–425.

678 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

García-Llatas, G., Rodríguez-Estrada, M.T., 2011. Current and new insights on phy- Kalogeropoulos, N., Salta, F.N., Chiou, A., Andrikopoulos, N.K., 2007. Formation and

tosterol oxides in plant sterol-enriched food. Chemistry and Physics of Lipids distribution of oxidized fatty acids during deep- and pan-frying of potatoes.

164, 607–624. European Journal of Lipid Science and Technology 109, 1111–1123.

Gerde, J.A., Hammond, E.G., White, P.J., 2011. Inﬂuence of polydimethylsiloxane on Kalogianni, E.P., Karastogiannidou, C., Karapantsios, T.D., 2010. Effect of potato pres-

the formation of 4-hydroxynonenal in soybean oil at frying temperature. Journal ence on the degradation of extra virgin olive oil during frying. International

of the American Oil Chemists’ Society 83, 1503–1510. Journal of Food Science and Technology 45, 765–775.

Gertz, C., Klostermann, S., 2002. Analysis of acrylamide and mechanisms of its forma- Kamal-Eldin, A., Márquez-Ruiz, G., Dobarganes, M.C., Appelqvist, L.A., 1997. Char-

tion in deep-fried products. European Journal of Lipid Science and Technology acterisation of aldehydic acids in used and unused frying oils. Journal of

104, 762–771. Chromatography A 776, 245–254.

Gillatt, P., 2001. Flavour and aroma development in frying and fried food. In: Rossell, Kamal-Eldin, A., Appelqvist, L.A., 1996. Aldehydic acids in frying oils: formation,

J.B. (Ed.), Frying: Improving Quality. Woodhead Publishing Limited, Cambridge, toxicological signiﬁcance and analysis. Grasas y Aceites 47, 342–348.

pp. 266–334. Kandhro, A., Sherazi, S., Mahesar, S.A., Bhanger, M.I., Tapur, M.Y., Rauf, A., 2008.

Giuffrida, F., Destaillats, F., Robert, F., Skibsted, L.H., Dionisi, F., 2004a. Formation GC-MS quantiﬁcation of fatty acid proﬁle including trans FA in the locally man-

and hydrolysis of triacylglycerol and sterols epoxides: role of unsaturated tria- ufactured margarines of Pakistan. Food Chemistry 109, 207–211.

cylglycerol peroxyl radicals. Free Radical Biology and Medicine 37, 104–114. Karel, M., Schaich, K., Roy, R.B., 1975. Interaction of peroxidizing methyl linoleate

Giuffrida, F., Destaillats, F., Skibsted, L.H., Dionisi, F., 2004b. Structural analysis of with some proteins and amino acids. Journal of Agricultural and Food Chemistry

hydroperoxy- and epoxy-triacylglycerols by liquid chromatography mass spec- 23, 159–163.

trometry. Chemistry and Physics of Lipids 131, 41–49. Katragadda, H.R., Fullana, A., Sidhu, S., Carbonell-Barrachina, Á.A., 2010. Emissions

Gökmen, V., S¸ enyuva, H.Z., Acar, J., Sarıoglu,˘ K., 2005. Determination of acrylamide of volatile aldehydes from heated cooking oils. Food Chemistry 120, 59–65.

in potato chips and crisps by high-performance liquid chromatography. Journal Kawada, T., Krishnamurthy, R., Mookherjee, B., Chang, S., 1967. Chemical reactions

of Chromatography A 1088, 193–199. involved in the deep fat frying of foods. II. Identiﬁcation of acidic volatile decom-

Guillén, M.D., Goicoechea, E., 2008. Toxic oxygenated ␣,␤-unsaturated aldehydes position products of corn oil. Journal of the American Oil Chemists’ Society 44,

and their study in foods: a review. Critical Reviews in Food Science and Nutrition 131–135.

48, 119–136. Kemmo, S., Ollilainen, V., Lampi, A.M., Piironen, V., 2008. Liquid chromatography

Guillén, M.D., Ruiz, A., 2008. Monitoring of heat-induced degradation of edible oils mass spectrometry for plant sterol oxide determination in complex mixtures.

by proton NMR. European Journal of Lipid Science and Technology 110, 52–60. European Food Research and Technology 226, 1325–1334.

Guillén, M.D., Uriarte, P.S., 2009. Contribution to further understanding of the evolu- Kemmo, S., Soupas, L., Lampi, A.M., Piironen, V., 2005. Formation and decom-

tion of sunﬂower oil submitted to frying temperature in a domestic fryer: study position of stigmasterol hydroperoxides and secondary oxidation products

by H nuclear magnetic resonance. Journal of Agricultural and Food Chemistry during thermo-oxidation. European Journal of Lipid Science and Technology 107,

57, 7790–7799. 805–814.

Guillén, M.D., Uriarte, P.S., 2012. Aldehydes contained in edible oils of a very dif- Keramat, J., LeBail, A., Prost, C., Jafari, M., 2011. Acrylamide in baking products: a

ferent nature after prolonged heating at frying temperature: Presence of toxic review article. Food Bioprocess Technology 4, 530–543.

oxygenated ␣, ␤ unsaturated aldehydes. Food Chemistry 131, 915–926. Keramat, J., LeBail, A., Prost, C., Soltanizadeh, N., 2011b. Acrylamide in foods: chem-

Hagmar, L., Wirfält, E., Paulsson, B., Törnqvist, M., 2005. Differences in hemoglobin istry and analysis. A review. Food Bioprocess Technology 4, 340–363.

adduct levels of acrylamide in the general population with respect to dietary Kiatsrichart, S., Susan Brewer, M., Cadwallader, K., Artz, W., 2003. Pan-frying stability

intake, smoking habits and gender. Mutation Research: Genetic Toxicology and of NuSun oil, a mid-oleic sunﬂower oil. Journal of the American Oil Chemists’

Environmental Mutagenesis 580, 157–165. Society 80, 479–483.

Hamama, A.A., Nawar, W.W., 1991. Thermal decomposition of some phenolic antiox- Kmiecik, D., Korczak, J., Rudzinska, M., Kobus-Cisowska, J., Gramza-Michalowska,

idants. Journal of Agricultural and Food Chemistry 39, 1063–1069. A., He˛s,´ M., 2011. ␤-Sitosterol and campesterol stabilisation by natural and

␣ ␤

Han, I.H., Csallany, A.S., 2009. Formation of toxic , -unsaturated 4-hydroxy- synthetic antioxidants during heating. Food Chemistry 128, 937–942.

aldehydes in thermally oxidized fatty acid methyl esters. Journal of the American Ko, S.N., Kim, C.J., Kim, C.T., Kim, Y., Kim, I.H., 2010. Effects of tocopherols and

Oil Chemists’ Society 86, 253–260. tocotrienols on the inhibition of autoxidation of conjugated linoleic acid. Euro-

Hartman, G.J., Jin, Q.Z., Collins, G.J., Lee, K.N., Ho, C.T., Chang, S.S., 1983. Nitrogen- pean Journal of Lipid Science and Technology 112, 496–501.

containing heterocyclic compounds identiﬁed in the volatile ﬂavor constituents Koschutnig, K., Heikkinen, S., Kemmo, S., Lampi, A.M., Piironen, V., Wagner, K.H.,

of roasted beef. Journal of AgrIcultural and Food Chemistry 31, 1030–1033. 2009. Cytotoxic and apoptotic effects of single and mixed oxides of ␤-sitosterol

Henderson, S., Nawar, W., 1981. Thermal interaction of linoleic acid and on HepG2-cells. Toxicology in Vitro 23, 755–762.

its esters with valine. Journal of the American Oil Chemists’ Society 58, Krokida, M.K., Oreopoulou, V., Maroulis, Z.B., 2000. Water loss and oil uptake as a

632–635. function of frying time. Journal of Food Engineering 44, 39–46.

Hoenicke, K., Gatermann, R., Harder, W., Hartig, L., 2004. Analysis of acrylamide in Kuligowski, J., Carrión, D., Quintás, G., Garrigues, S., De La Guardia, M., 2012. Direct

different foodstuffs using liquid chromatography–tandem mass spectrometry determination of polymerised triacylglycerides in deep-frying vegetable oil by

and gas chromatography–tandem mass spectrometry. Analytica Chimica Acta near infrared spectroscopy using Partial Least Squares regression. Food Chem-

520, 207–215. istry 131, 353–359.

Hodge, J.E., 1953. Dehydrated Foods, chemistry of browning reactions in model Kuligowski, J., Quintas, G., Garrigues, S., De La Guardia, M., 2010a. Monitoring of

systems. Journal of Agricultural and Food Chemistry 1, 928–943. polymerized triglycerides in deep-frying oil by on-line GPC-FTIR spectrometry

Horiuchi, M., Umano, K., Shibamoto, T., 1998. Analysis of volatile compounds formed using the science based calibration multivariate approach. Chromatographia 71,

from ﬁsh oil heated with cysteine and trimethylamine oxide. Journal of Agricul- 201–209.

tural and Food Chemistry 46, 5232–5237. Kuligowski, J., Quintás, G., Garrigues, S., De La Guardia, M., 2010b. Direct determi-

Houhoula, D.P., Oreopoulou, V., Tzia, C., 2003. The effect of process time and temper- nation of polymerized triglycerides in deep-frying olive oil by attenuated total

ature on the accumulation of polar compounds in cottonseed oil during deep-fat reﬂectance-Fourier transform infrared spectroscopy using partial least squares

frying. Journal of the Science of Food and Agriculture 83, 314–319. regression. Analytical and Bioanalytical Chemistry 397, 861–869.

Houhoula, D., Oreopoulou, V., Tzia, C., 2002. A kinetic study of oil deterioration during Kummerow, F.A., 2009. The negative effects of hydrogenated trans fats and what to

frying and a comparison with heating. Journal of the American Oil Chemists’ do about them. Atherosclerosis 205, 458–465.

Society 79, 133–137. LaFond, S.I., Jerrell, J.P., Cadwallader, K.R., Artz, W.E., 2011. Formation of 4-hydroxy-

Jautz, U., Gibis, M., Morlock, G.E., 2008. Quantiﬁcation of heterocyclic aromatic 2-(E)-nonenal in a corm-soy oil blend: a controlled heating study using a French

amines in fried meat by HPTLC/UV-FLD and HPLC/UV-FLD: a comparison of two fried potato model. Journal of the American Oil Chemists’ Society 88, 763–772.

methods. Journal of Agricultural and Food Chemistry 56, 4311–4319. Lamboni, C., Perkins, E.G., 1996. Effects of dietary heated fats on rat liver enzyme

Jägerstad, M., Skog, K., 2005. Genotoxicity of heat-processed foods. Muta- activity. Lipids 31, 955–962.

tion Research/Fundamental and Molecular Mechanisms of Mutagenesis 574, Lampi, A.M., Juntunen, L., Toivo, J., Piironen, V., 2002. Determination of thermo-

156–172. oxidation products of plant sterols. Journal of Chromatography B 777, 83–92.

Jägerstad, M., Skog, K., Arvidsson, P., Solyakov, A., 1998. Chemistry, formation and Lampi, A.M., Kamal-Eldin, A., 1998. Effect of ␣- and ␥-tocopherols on thermal

occurrence of genotoxic heterocyclic amines identified in model systems and polymerization of purified high-oleic sunflower triacylglycerols. Journal of the

cooked foods. Zeitschrift für Lebensmittel-Untersuchung und -Forschung A 207, American Oil Chemists’ Society 75, 1699–1703.

419–427. Lampi, A.M., Kemmo, S., Mäkelä, A., Heikkinen, S., Piironen, V., 2009. Distribu-

Jerkovic,´ I., Mastelic,´ J., Tartaglia, S., 2007. A study of volatile ﬂavour substances tion of monomeric, dimeric and polymeric products of stigmasterol during

in Dalmatian traditional smoked ham: Impact of dry-curing and frying. Food thermo-oxidation. European Journal of Lipid Science and Technology 111,

Chemistry 104, 1030–1039. 1027–1034.

Jiang, H., Kruger, N., Lahiri, D.R., Wang, D., Vatele, J.M., Balazy, M., 1999. Nitro- Ledoux, M., Juanéda, P., Sébédio, J.L., 2007. Trans fatty acids: deﬁnition and occur-

gen dioxide induces cis-trans-isomerization of arachidonic acid within cellular rence in foods. European Journal of Lipid Science and Technology 109, 891–900.

phospholipids. detection of trans-arachidonic acids in vivo. Journal of Biological Lee, J., Dong, A., Jung, K., Shin, H.S., 2011. Inﬂuence of extra virgin olive oil on

Chemistry 274, 16235–16241. the formation of heterocyclic amines in roasted beef steak. Food Science and

Johansson, M.A.E., Fredholm, L., Bjerne, I., Jägerstad, M., 1995. Inﬂuence of frying fat Biotechnology 20, 159–165.

on the formation of heterocyclic amines in fried beefburgers and pan residues. Lee, J., Lee, S., Lee, H., Park, K., Choe, E., 2002. Spinach (Spinacia oleracea) pow-

Food and Chemical Toxicology 33, 993–1004. der as a natural food-grade antioxidant in deep-fat-fried products. Journal of

Juanéda, P., Brac De La Pérrière, S., Sébédio, J.L., Grégoire, S., 2003. Inﬂuence of Agricultural and Food Chemistry 50, 5664–5669.

heat and reﬁning on formation of CLA isomers in sunﬂower oil. Journal of the Liu, W.H., Inbarai, B.S., Chen, B.H., 2007. Analysis and formation of trans fatty acids

American Oil Chemists’ Society 80, 937–940. in hydrogenated soybean oil during heating. Food Chemistry 104, 1740–1749.

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 679

Lojzova, L., Riddellova, K., Hajslova, J., Zrostlikova, J., Schurek, J., Cajka, T., 2009. Alter- Murkovic, M., 2004. Chemistry, formation and occurrence of genotoxic heterocyclic

native GC–MS approaches in the analysis of substituted pyrazines and other aromatic amines in fried products. European Journal of Lipid Science and Tech-

volatile aromatic compounds formed during Maillard reaction in potato chips. nology 106, 777–785.

Analytica Chimica Acta 641, 101–109. Naz, S., Siddiqi, R., Sheikh, H., Sayeed, S.A., 2005. Deterioration of olive, corn and

Macku, C., Shibamoto, T., 1991a. Headspace volatile compounds formed from heated soybean oils due to air, light, heat and deep-frying. Food Research International

corn oil and corn oil with glycine. Journal of Agricultural and Food Chemistry 38, 127–134.

39, 1265–1269. Neff, W.E., Warner, K., Eller, F., 2003. Effect of (-Tocopherol on formation of non-

Macku, C., Shibamoto, T., 1991b. Volatile sulfur-containing compounds generated volatile lipid degradation products during frying of potato chips in triolein.

from the thermal interaction of corn oil and cysteine. Journal of Agricultural and Journal of the American Oil Chemists’ Society 80, 801–806.

Food Chemistry 39, 1987–1989. Ni, H., Datta, A.K., 1999. Moisture, oil and energy transport during deep-fat frying of

Marmesat, S., Morales, A., Velasco, J., Dobarganes, M.C., 2010. Action and fate of food materials. Food and Bioproducts Processing 77, 194–204.

natural and synthetic antioxidants during frying. Grasas y Aceites 61, 333–340. Nogala-Kalucka, M., Korczak, J., Dratwia, M., Lampart-Szczapa, E., Siger, A.,

Marmesat, S., Velasco, J., Dobarganes, M.C., 2008. Quantitative determination of Buchowski, M., 2005. Changes in antioxidant activity and free radical scav-

epoxy acids, keto acids and hydroxy acids formed in fats and oils at frying enging potential of rosemary extract and tocopherols in isolated rapeseed oil

temperatures. Journal of Chromatography A 1211, 129–134. triacylglycerols during accelerated tests. Food Chemistry 93, 227–235.

Márquez-Ruiz, G., Dobarganes, M.C., 1996. Short-chain fatty acid formation during Osório, V.M., de Lourdes, C.Z., 2011. Determination of acrolein in french fries by solid-

thermoxidation and frying. Journal of the Science of Food and Agriculture 70, phase microextraction gas chromatography and mass spectrometry. Journal of

120–126. Chromatography A 1218, 3332–3336.

Márquez-Ruiz, G., Tasioulamargari, M., Dobarganes, M.C., 1995. Quantitation and Ostlund Jr., R.E., 2002. Phytosterols in human nutrition. Annual Review of Nutrition

distribution of altered fatty acids in frying fats. Journal of the American Oil 22, 533–549.

Chemists’ Society 72, 1171–1176. Ou, S., Shi, J., Huang, C., Zhang, G., Teng, J., Jiang, Y., Yang, B., 2010. Effect of antiox-

Martin, J.C., Dobarganes, M.C., Nour, M., Márquez-Ruiz, G., Christie, W.W., Lavil- idants on elimination and formation of acrylamide in model reaction systems.

lonnière, F., Sébédio, J.L., 1998a. Effect of fatty acid positional distribution and Journal of Hazardous Materials 182, 863–868.

triacylglycerol composition on lipid by-products formation during heat treat- Overton, S.V., Manura, J.J., 1995. Analysis of volatile organics in cooking oils by ther-

ment. I. Polymer formation. Journal of the American Oil Chemists’ Society 75, mal desorption gas chromatography-mass spectrometry. Journal of Agricultural

1065–1071. and Food Chemistry 43, 1314–1320.

Martin, J.C., Joffre, F., Siess, M.H., Vernevaut, M.F., Collenot, P., Genty, M., Sébédio, Özyurt, G., Kafkas, E., Etyemez, M., 2011. Effect of the type of frying oil on volatile

J.L., 2000. Cyclic fatty acid monomers from heated oil modify the activities of compounds of goatﬁsh (Upeneus pori) during cold storage. International Journal

lipid synthesizing and oxidizing enzymes in rat liver. Journal of Nutrition 130, of Food Science and Technology 46, 2598–2602.

1524–1530. Paul, S., Mittal, G.S., 1997. Regulating the use of degraded oil/fat in deep-fat/oil food

Martin, J.C., Lavillonnière, F., Nour, M., Sébédio, J.L., 1998b. Effect of fatty acid frying. Critical Reviews in Food Science and Nutrition 37, 635–662.

positional distribution and triacylglycerol composition on lipid by-products for- Pedreschi, F., Kaack, K., Granby, K., Troncoso, E., 2007. Acrylamide reduction

mation during heat treatment: III-Cyclic fatty acid monomers study. Journal of under different pre-treatments in French fries. Journal of Food Engineering 79,

the American Oil Chemists’ Society 75, 1691–1697. 1287–1294.

Martin, J.C., Nour, M., Lavillonnière, F., Sébédio, J.L., 1998c. Effect of fatty acid Pedreschi, F., Segtnan, V.H., Knutsen, S.H., 2010. On-line monitoring of fat, dry matter

positional distribution and triacylglycerol composition on lipid by-products for- and acrylamide contents in potato chips using near infrared interactance and

mation during heat treatment: II. Trans isomers. Journal of the American Oil visual reﬂectance imaging. Food Chemistry 121, 616–620.

Chemists’ Society 75, 1073–1078. Pokorny,´ J., 1989. Flavor chemistry of deep fat frying in oil. In: Min, D.B., Smouse,

Mastrocola, D., Munari, M., Cioroi, M., Lerici, C.R., 2000. Interaction between Maillard T.H., Zhang, S.S. (Eds.), Flavor Chemistry of Lipid Foods. American Oil Chemists

reaction products and lipid oxidation in starch-based model systems. Journal of Society, Champaign, pp. 113–154.

the Science of Food and Agriculture 80, 684–690. Porter, N.A., 1986. Mechanisms for the autoxidation of polyunsaturated lipids.

Matthäus, B., Haase, N.U., Vosmann, K., 2004. Factors affecting the concentration of Accounts of Chemical Research 19, 262–268.

acrylamide during deep-fat frying of potatoes. European Journal of Lipid Science Porter, N.A., Caldwell, S.E., Mills, K.A., 1995. Mechanisms of free radical oxidation of

and Technology 106, 793–801. unsaturated lipids. Lipids 30, 277.

May, W., Peterson, R., Chang, S., 1983. Chemical reactions involved in the Ramírez, M.R., Estévez, M., Morcuende, D., Cava, R., 2004. Effect of the type of frying

deep- fat frying of foods. IX. Identiﬁcation of the volatile decomposition culinary fat on volatile compounds isolated in fried pork loin chops by using

products of triolein. Journal of the American Oil Chemists’ Society 60, SPME-GC-MS. Journal of Agricultural and Food Chemistry 52, 7637–7643.

990–995. Randel, G., Balzer, M., Grupe, S., Drusch, S., Kaina, B., Platt, K.L., Schwarz, K., 2007.

Mellema, M., 2003. Mechanism and reduction of fat uptake in deep-fat fried foods. Degradation of heterocyclic aromatic amines in oil under storage and frying

Trends in Food Science and Technology 14, 364–373. conditions and reduction of their mutagenic potential. Food and Chemical Tox-

Mestdagh, F.J., De Meulenaer, B., Van Poucke, C., Detavernier, C., Cromphout, C., Van icology 45, 2245–2253.

Peteghem, C., 2005. Inﬂuence of oil type on the amounts of acrylamide generated Rennick, K.A., Warner, K., 2006. Effect of elevated temperature on development

in a model system and in french fries. Journal of Agricultural and Food Chemistry of tocopherolquinones in oils. Journal of Agricultural and Food Chemistry 54,

53, 6170–6174. 2188–2192.

Mezouari, S., Eichner, K., 2008. Trans fatty acid content and positional distribu- Réblová, Z., 2006. The effect of temperature on the antioxidant activity of toco-

tion in reﬁned and heated rice bran oil. European Journal of Lipid Science and pherols. European Journal of Lipid Science and Technology 108, 858–863.

Technology 110, 1058–1061. Ribeiro, A., Grimaldi, R., Gioielli, L.A., Goncalves, L., 2009. Zero trans fats from soybean

Mildner-Szkudlarz, S., Jelen,´ H.H., 2008. The potential of different techniques for oil and fully hydrogenated soybean oil: Physico-chemical properties and food

volatile compounds analysis coupled with PCA for the detection of the adulter- applications. Food Research International 42, 401–410.

ation of olive oil with hazelnut oil. Food Chemistry 110, 751–761. Rojo, J.A., Perkins, E.G., 1987. Cyclic fatty acid monomer formation in frying fats.

Mossoba, M.M., Milosevic, V., Milosevic, M., Kramer, J.K.G., Azizian, H., 2007. Deter- I. Determination and structural study. Journal of the American Oil Chemists’

mination of total trans fats and oils by infrared spectroscopy for regulatory Society 64, 414–421.

compliance. Analytical and Bioanalytical Chemistry 389, 87–92. Roman, O., Courtois, F., Maillard, M.N., Riquet, A.M., 2012. Kinetic study

Mossoba, M.M., Yurawecz, M.P., Roach, J.A.G., McDonald, R.E., Flickinger, B.D., of hydroperoxide degradation in edible oils using electron spin res-

Perkins, E.G., 1996a. Analysis of cyclic fatty acid monomer 2-alkenyl-4,4- onance spectroscopy. Journal of the American Oil Chemists’ Society,

dimethyloxazoline derivatives by gas chromatography-matrix isolation -fourier http://dx.doi.org/10.1007/s11746-012-2048-4.

transform infrared spectroscopy. Journal of Agricultural and Food Chemistry 44, Romani, S., Bacchiocca, M., Rocculi, P., Rosa, M.D., 2008. Effect of frying time on

3193–3196. acrylamide content and quality aspects of French fries. European Food Research

Mossoba, M.M., Yurawecz, P.M., Roach, J.A.G., McDonald, R.E., Perkins, E.G., 1996b. and Technology 226, 555–560.

Conﬁrmatory mass-spectral data for cyclic fatty acid monomers. Journal of the Romero, A., Cuesta, C., Sánchez-Muniz, F.J., 1998. Effect of oil replenishment during

American Oil Chemists’ Society 73, 1317–1321. deep-fat frying of frozen foods in sunﬂower oil and high-oleic acid sunﬂower

Mossoba, M.M., Yurawecz, P.M., Roach, J.A.G., Lin, H.S., McDonald, R.E., Flickinger, oil. Journal of the American Oil Chemists’ Society 75, 161–167.

B.D., Perkins, E.G., 1994. Rapid determination of double bond conﬁguration and Romero, A., Cuesta, C., Sánchez-Muniz, F.J., 2000a. Cyclic fatty acid monomers and

position along the hydrocarbon chain in cyclic fatty acid monomers. Lipids 29, thermoxidative alteration compounds formed during frying of frozen foods

893–896. in extra virgin olive oil. Journal of the American Oil Chemists’ Society 77,

Mossoba, M.M., Yurawecz, P.M., Roach, J.A.G., Lin, H.S., McDonald, R.E., Flickinger, 1169–1175.

B.D., Perkins, E.G., 1995. Elucidation of cyclic fatty acid monomer structures. Romero, A., Cuesta, C., Sánchez-Muniz, F.J., 2000b. Trans fatty acid production in

Cyclic and bicyclic ring sizes and double bond position and conﬁguration. Journal deep fat frying of frozen foods with different oils and frying modalities. Nutrition

of the American Oil Chemists’ Society 72, 721–727. Research 20, 599–608.

Moreau, R.A., Whitaker, B.D., Hicks, K.B., 2002. Phytosterols, phytostanols, and their Romero, A., Cuesta, C., Sánchez-Muniz, F.J., 2003. Cyclic fatty acid monomers in high

conjugates in foods: Structural diversity, quantitative analysis, and health- oleic acid sunﬂower oil and extra virgin olive oil used in repeated frying of fresh

promoting uses. Progress in Lipid Research 41, 457–500. potatoes. Journal of the American Oil Chemists’ Society 80, 437–442.

Mottram, D.S., Wedzicha, B.L., Dodson, A.T., 2002. Acrylamide is formed in the mail- Romero, A., Cuesta, C., Sánchez-Muniz, F.J., 2006. Cyclic fatty acid monomer for-

lard reaction. Nature 419, 448–449. mation in domestic frying of frozen foods in sunﬂower oil and high oleic acid

Mozaffarian, D., 2006. Trans fatty acids – effects on systemic inﬂammation and sunﬂower oil without oil replenishment. Food and Chemical Toxicology 44,

endothelial function. Atherosclerosis Supplements 7, 29–32. 1674–1681.

680 Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681

Rossell, J.B., 2001. Frying: Improving Quality. Woodhead Publishing Limited, Cam- Takeoka, G.R., Full, G.H., Dao, L.T., 1997. Effect of heating on the characteristics and

bridge, pp. 1–355. chemical composition of selected frying oils and fats. Journal of Agricultural and

Rossi, M., Alamprese, C., Ratti, S., 2007. Tocopherols and tocotrienols as free radical- Food Chemistry 45, 3244–3249.

scavengers in reﬁned vegetable oils and their stability during deep-fat frying. Takeoka, G.R., Perrino Jr., C., Buttery, R., 1996. Volatile constituents of used frying

Food Chemistry 102, 812–817. oils. Journal of Agricultural and Food Chemistry 44, 654–660.

Rudzinska,´ M., Korczak, J., Wasowicz, E., 2005. Changes in phytosterols and their Tang, J., Jin, Q.Z., Shen, G.H., Ho, C.T., Chang, S.S., 1983. Isolation and identiﬁca-

oxidation products during frying of French fries in rapeseed oil. Polish Journal tion of volatile compounds from fried chicken. Journal of Agricultural and Food

of Food and Nutrition Sciences 14, 381–387. Chemistry 31, 1287–1292.

Rudzinska,´ M., Przybylski, R., Wasowicz,˛ E., 2009. Products formed during thermo- Tareke, E., Rydberg, P., Karlsson, P., Eriksson, S., Tornqvist, M., 2002. Analysis of acry-

oxidative degradation of phytosterols. Journal of the American Oil Chemists’ lamide, a carcinogen formed in heated foodstuffs. Journal of Agricultural and

Society 86, 651–662. Food Chemistry 50, 4998–5006.

Rudzinska,´ M., Przybylski, R., Zhao, Y.Y., Curtis, J.M., 2010. Sitosterol thermo- Tasioula-Margari, M., Marquez-Ruíz, G., Dobarganes, M.C., 1996. Fractionation of

oxidative degradation leads to the formation of dimers, trimers and oligomers: oligomeric triacylglycerides and the relation to rejection limits for used frying

a study using combined size exclusion chromatography/mass spectrometry. oils. Journal of the American Oil Chemists’ Society 73, 1579–1584.

Lipids 45, 549–558. Thompson, J., May, W., Paulose, M., Peterson, R., Chang, S., 1978. Chemical reactions

Ryan, E., McCarthy, F.O., Maguire, A.R., O’Brien, N.M., 2009. Phytosterol oxidation involved in the deep-fat frying of foods. VII. Identiﬁcation of volatile decompo-

products: their formation, occurrence, and biological effects. Food Reviews sition products of trilinolein. Journal of the American Oil Chemists’ Society 55,

International 25, 157–174. 897–901.

Saguy, I.S., Dana, D., 2003. Integrated approach to deep fat frying: engineering, nutri- Tompkins, C., Perkins, E.G., 2000. Frying performance of low-linolenic acid soybean

tion, health and consumer aspects. Journal of Food Engineering 56, 143–152. oil. Journal of the American Oil Chemists’ Society 77, 223–229.

Saldanha, T., Sawaya, A.C.H.F., Eberlin, M.N., Bragagnolo, N., 2006. HPLC separation Totani, N., Burenjargal, M., Yawata, M., Ojiri, Y., 2008. Chemical properties and cyto-

and determination of 12 cholesterol oxidation products in ﬁsh: comparative toxicity of thermally oxidized oil. Journal of Oleo Science 57, 153–160.

study of RI, UV, and APCI-MS detectors. Journal of Agricultural and Food Chem- Tseng, Y.C., Moreira, R., Sun, X., 1996. Total frying-use time effects on soybean-oil

istry 54, 4107–4113. deterioration and on tortilla chip quality. International Journal of Food Science

Sanny, M., Jinap, S., Bakker, E.J., van Boekel, M.A.J.S., Luning, P.A., 2012. Possible and Technology 31, 287–294.

causes of variation in acrylamide concentration in French fries prepared in food Tsuzuki, W., 2010. cis-trans isomerization of carbon double bonds in monounsatu-

service establishments: an observational study. Food Chemistry 132, 134–143. rated triacylglycerols via generation of free radicals. Chemistry and Physics of

Schaich, K.M., 2005. Lipid oxidation theoretical aspects. In: Bailey’s Industrial Oil Lipids 163, 741–745.

and Fat Products. John Wiley & Sons Inc, New York, pp. 269–355. Tsuzuki, W., Matsuoka, A., Ushida, K., 2010. Formation of trans fatty acids in

Schneider, C., Boeglin, W., Yin, H., Stec, D., Hachey, D., Porter, N., Brash, A., 2005. edible oils during the frying and heating process. Food Chemistry 123,

Synthesis of dihydroperoxides of linoleic and linolenic acids and studies on their 976–982.

transformation to 4-hydroperoxynonenal. Lipids 40, 1155–1162. Tsuzuki, W., Nagata, R., Yunoki, R., Nakajima, M., Nagata, T., 2008. cis/trans-

Schneider, C., Tallman, K.A., Porter, N.A., Brash, A.R., 2001. Two distinct pathways of isomerisation of triolein, trilinolein and trilinolenin induced by heat treatment.

formation of 4-hydroxynonenal, Mechanisms of non-enzymatic transformation Food Chemistry 108, 75–80.

of the 9- and 13-hydroperoxides of linoleic acid to 4-hydroxyalkenals. Journal Van Loon, W.A.M., Linssen, J.P.H., Legger, A., Posthumus, M., Voragen, A.G.J., 2005.

of Biological Chemistry 276, 20831–20838. Identiﬁcation and olfactometry of French fries ﬂavour extracted at mouth con-

Schulte, E., 2002. Determination of higher carbonyl compounds in used frying fats by ditions. Food Chemistry 90, 417–425.

HPLC of DNPH derivatives. Analytical and Bioanalytical Chemistry 372, 644–648. Velasco, J., Berdeaux, O., Márquez-Ruiz, G., Dobarganes, M.C., 2002. Sensi-

Schwarz, W., 2000. Formation of trans polyalkenoic fatty acids during vegetable oil tive and accurate quantitation of monoepoxy fatty acids in thermoxidized

reﬁning. European Journal of Lipid Science and Technology 102, 648–649. oils by gas–liquid chromatography. Journal of Chromatography A 982,

Schwartz, D.P., Rady, A.H., Castaneda,˜ S., 1994. The formation of oxo and hydroxy- 145–152.

fatty acids in heated fats and oils. Journal of the American Oil Chemists’ Society Velasco, J., Marmesat, S., Berdeaux, O., Márquez-Ruiz, G., Dobarganes, M.C.,

71, 441–444. 2004a. Formation and evolution of monoepoxy fatty acids in thermoxidized

Sébédio, J.L., Grandgirard, A., 1989. Cyclic fatty acids: Natural sources, formation olive and sunﬂ ower oils and quantitation in used frying oils from restau-

durin heat treatment, synthesis and biological properties. Progress in Lipid rants and fried food outlets. Journal of Agricultural and Food Chemistry 52,

Research 28, 303–336. 4438–4443.

Sébédio, J.L., Grandgirard, A., Prevost, J., 1988. Linoleic acid isomers in heat treated Velasco, J., Marmesat, S., Berdeaux, O., Márquez-Ruiz, G., Dobarganes, M.C.,

sunﬂower oils. Journal of the American Oil Chemists’ Society 65, 362–366. 2005. Quantitation of short-chain glycerol-bound compounds in thermoxi-

Sébédio, J.L., Prevost, J., Grandgirard, A., 1987. Heat treatment of vegetable oils I. dized and used frying oils, A monitoring study during thermoxidation of

Isolation of the cyclic fatty acid monomers from heated sunﬂower and linseed olive and sunﬂower oils. Journal of Agricultural and Food Chemistry 53,

oils. Journal of the American Oil Chemists’ Society 64, 1026–1032. 4006–4011.

Seppanen, C.M., Csallany, A.S., 2002. Formation of 4-hydroxynonenal, a toxic alde- Velasco, J., Marmesat, S., Dobarganes, M.C., 2009. Chemistry of frying. In: Sahin, S.,

hyde, in soybean oil at frying temperature. Journal of the American Oil Chemists’ Sumnu, S.G. (Eds.), Advances in Deep-fat Frying of Foods. CRC press, Boca Raton,

Society 79, 1033–1038. Florida, pp. 33–56.

Seppanen, C.M., Csallany, A.S., 2004. Incorporation of the toxic aldehyde 4-hydroxy- Velasco, J., Marmesat, S., Márquez-Ruiz, G., Dobarganes, M.C., 2004b. Format ion of

2- trans-nonenal into food fried in thermally oxidized soybean oil. Journal of the short-chain glycerol-bound oxidation products and oxidized monomeric tria-

American Oil Chemists’ Society 81, 1137–1141. cylglycerols during deep-frying and occurrence in used frying fats. European

Seppanen, C.M., Csallany, A.S., 2006. The effect of intermittent and continuous Journal of Lipid Science and Technology 106, 728–735.

heating of soybean oil at frying temperature on the formation of 4-hydroxy- Verleyen, T., Kamal-Eldin, A., Dobarganes, M.C., Verhe, R., Dewettinck, K., Huyghe-

2-trans-nonenal and other ␣-, ␤-unsaturated hydroxyaldehydes. Journal of the baert, A., 2001a. Modeling of ␣-tocopherol loss and oxidation products

American Oil Chemists’ Society 83, 121–127. formed during thermoxidation in triolein and tripalmitin mixtures. Lipids 36,

Silvagni, A., Franco, L., Bagno, A., Rastrelli, F., 2010. Thermoinduced lipid oxidation of 719–726.

a culinary oil: a kinetic study of the oxidation products by magnetic resonance Verleyen, T., Verhe, R., Huyghebaert, A., Dewettinck, K., Greyt, W.D., 2001b. Identiﬁ-

␣

spectroscopies. Journal of Physical Chemistry A 114, 10059–10065. cation of -tocopherol oxidation products in triolein at elevated temperatures.

Smith, L.L., 1996. Review of progress in sterol oxidations: 1987–1995. Lipids 31, Journal of Agricultural and Food Chemistry 49, 1508–1511.

453–487. Viklund, G., Mendoza, F., Sjöholm, I., Skog, K., 2007. An experimental set-up for study-

Sosa-Morales, M.E., Orzuna-Espiritu, R., Velez-Ruiz, J.F., 2006. Mass, thermal and ing acrylamide formation in potato crisps. LWT – Food Science and Technology

quality aspects of deep-fat frying of pork meat. Journal of Food Engineering 77, 40, 1066–1071.

731–738. Wang, H., Lee, A.W.M., Shuang, S., Choi, M.M.F., 2008. SPE/HPLC/UV studies on

Soupas, L., Huikko, L., Lampi, A.M., Piironen, V., 2007. Pan-frying may induce phy- acrylamide in deep-fried ﬂour-based indigenous Chinese foods. Microchemical

tosterol oxidation. Food Chemistry 101, 286–297. Journal 89, 90–97.

Soupas, L., Juntunen, L., Säyn äjoki, S., Lampi, A.M., Piironen, V., 2004. GC-MS method Warner, K., 2008. Chemistry of frying oils. In: Akoh, C.C., Min, D.B. (Eds.), Food Lipids:

for characterization and quantiﬁcation of sitostanol oxidation products. Journal Chemistry, Nutrition, and Biotechnology. CRC Press, Boca Raton, Florida, pp.

of the American Oil Chemists’ Society 81, 135–141. 189–202.

Spangler, C.W., 1976. Thermal [1,j] sigmatropic rearrangements. Chemical Reviews Willett, W.C., Stampfer, M.J., Manson, J.E., Colditz, G.A., Speizer, F.E., Rosner, B.A.,

76, 187–217. Sampson, L.A., Hennekens, C.H., 1993. Intake of trans fatty acids and risk of

Stadler, R.H., Blank, I., Varga, N., Robert, F., Hau, J., Guy, P.A., Robert, M.C., Riediker, coronary heart disease among women. Lancet 341, 581–585.

S., 2002. Acrylamide from maillard reaction products. Nature 419, 449–450. Winkler, J.K., Warner, K., 2008a. The effect of phytosterol concentration on oxidative

Steel, C.J., Dobarganes, M.C., Barrera-Arellano, D., 2006. Formation of polymerization stability and thermal polymerization of heated oils. European Journal of Lipid

compounds during thermal oxidation of cottonseed oil, partially hydrogenated Science and Technology 110, 455–464.

cottonseed oil and their blends. Grasas y Aceites 57, 284–291. Winkler, J.K., Warner, K., 2008b. Effect of phytosterol structure on thermal polymer-

Stevenson, S.G., Vaisey-Genser, M., Eskin, N.A.M., 1984. Quality control in the use of ization of heated soybean oil. European Journal of Lipid Science and Technology

deep frying oils. Journal of the American Oil Chemists’ Society 61, 1102–1108. 110, 1068–1077.

Tabee, E., Jägerstad, M., Dutta, P.C., 2008. Lipids and phytosterol oxidation products Winkler, J.K., Warner, K., Glynn, M.T., 2007. Effect of deep-fat frying on phytosterol

in commercial potato crisps commonly consumed in Sweden. European Food content in oils with differing fatty acid composition. Journal of the American Oil

Research and Technology 227, 745–755. Chemists’ Society 84, 1023–1030.

Q. Zhang et al. / Chemistry and Physics of Lipids 165 (2012) 662–681 681

Wu, C., Chen, S., 1992. Volatile compounds in oils after deep frying or stir frying and Zamora, R., Francisco, J.H., 2005. Coordinate contribution of lipid oxidation and

subsequent storage. Journal of the American Oil Chemists’ Society 69, 858–865. Maillard reaction to the nonenzymatic food browning. Critical Reviews in. Food

Yamane, M., 2002. High-performance liquid chromatography–thermospray Science and Nutrition 45, 49–59.

ionization-mass spectrometry of the oxidation products of polyunsaturated- Zamora, R., Hidalgo, F.J., 2008. Contribution of lipid oxidation products to acrylamide

fatty acids. Analytica Chimica Acta 465, 227–236. formation in model systems. Journal of Agricultural and Food Chemistry 56,

Yasuda, K., Reddy, B., Chang, S., 1968. Chemical reactions involved in the deep fat 6075–6080.

frying of foods. IV. Identiﬁcation of acidic volatile decomposition products of Zamora, R., Hidalgo, F.J., 2011. The maillard reaction and lipid oxidation. Lipid Tech-

hydrogenated cottonseed oil. Journal of the American Oil Chemists’ Society 45, nology 23, 59–62.

625–628. Zhang, Y., Dong, Y., Ren, Y., Zhang, Y., 2006. Rapid determination of acrylamide

Yeo, J.D., Park, J.W., Lee, J.H., 2011. Evaluation of antioxidant capacity of sesamol and contaminant in conventional fried foods by gas chromatography with electron

free radical scavengers at different heating temperatures. European Journal of capture detector. Journal of Chromatography A 1116, 209–216.

Lipid Science and Technology 113, 910–915. Zhang, Y., Zhang, G., Zhang, Y., 2005. Occurrence and analytical methods of acry-

Yu, T.H., Wu, C.M., Ho, C.T., 1993. Volatile compounds of deep-oil fried, microwave- lamide in heat-treated foods: Review and recent developments. Journal of

heated and oven-baked garlic slices. Journal of Agricultural and Food Chemistry Chromatography A 1075, 1–21.

41, 800–805. Zhang, Y., Ho, C.T., 1989. Volatile compounds formed from thermal interaction of

Yurawecz, M.P., Sehat, N., Mossoba, M.M., Roach, J.A.G., Kramer, J.K.G., Ku, Y., 1999. 2,4-decadienal with cysteine and glutathione. Journal of Agricultural and Food

Variations in isomer distribution in commercially available conjugated linoleic Chemistry 37, 1016–1020.

acid. Fett/Lipid 101, 277–282. Publishing

AUSTRALIAN JOURNAL OF CHEMISTRYCHEMISTRY AN INTERNATIONAL JOURNAL FOR CHEMICAL SCIENCE

publishing research papers from all fields of chemical Volume 55, 2002 science, including synthesis, structure, new materials, © CSIRO 2002 macromolecules, supramolecular chemistry, biological chemistry, nanotechnology, surface chemistry, and analytical techniques.

All enquiries and manuscripts should be directed to:

Dr Alison Green Australian Journal of Chemistry – an International Journal for Chemical Science CSIRO PUBLISHING PO Box 1139 (150 Oxford St) Collingwood, Vic. 3066, Australia

Telephone: +61 3 9662 7630 Fax: +61 3 9662 7611 E-mail: [email protected]

Published by CSIRO PUBLISHING for CSIRO and the Australian Academy of Science

www.publish.csiro.au/journals/ajc Review

Aust. J. Chem. 2002, 55, 299–310

CH02076 NewReJ. A. cent GerrGe De rard r v elopment sCocern n ing the MaillardR eaction Aspects of an AGEing Chemistry— Recent Developments Concerning the Maillard Reaction

Juliet A. Gerrard

Department of Plant & Microbial Sciences, University of Canterbury, Christchurch, New Zealand (e-mail: [email protected]).

The chemistry and consequences of the Maillard reaction network—initiated by the condensation of an amine with a carbonyl group, often from a reducing sugar—are reviewed. This chemistry has consequences that pervade the two disparate literatures of food science and medicine. The Maillard reaction is responsible for many aspects of the colour, flavour and texture of processed foods. It is also a key player in the ageing process, especially amongst diabetics. Recent developments have led to detailed characterization of some of the many Maillard reaction products and a sufficient understanding of their mechanisms of formation to pique the interest of mainstream chemists. Much remains to be elucidated, but the potential rewards of understanding, and ultimately controlling, this chemistry are enormous.

Manuscript received: 18 April 2002. Final version: 6 June 2002.

Introduction literature on the effects and implications of this reaction. Recent attention has particularly focused on the role of Unlike the many named reactions in organic chemistry, the Maillard chemistry in ageing, especially in diabetics, where Maillard reaction represents neither a synthetic procedure, sugar levels are poorly regulated.[2] This has led to the notion nor a mechanistic description, but a complex network of that inhibitors of Maillard chemistry may play a role as anti- chemical reactions. This cascade is initiated by the ageing therapeutics.[3,4] deceptively simple condensation of an amine with a carbonyl Despite the evident importance of Maillard chemistry, group, often within a reducing sugar. and the potential applications of unravelling and controlling The reaction was first reported in 1912, in a brief paper to its many individual steps, the complexity of the reaction the French Academy by Louis-Camille Maillard.[1] He network has made elucidation of the detailed reaction described a very simple observation: upon gently heating pathways too daunting for the mainstream chemical sugars and amino acids in water, a yellow-brown colour community, until recently. It is only over the last few years developed. Although no chemical explanation was provided that aspects of Maillard chemistry have started to become to account for this phenomenon, Maillard was astute enough characterized with a sufficient degree of chemical rigour to to predict the far-reaching consequences of a facile reaction invite the interest of those who trained as chemists, as well as between two very common biological moieties. The those studying multifarious consequences of this fascinating predicted ramifications of Maillard chemistry have since sequence of reactions. been observed and analysed in many biological systems, and A seminal review, covering all aspects of Maillard the reaction has become particularly important in food chemistry, was published by Ledl and Schleicher in 1991.[5] science and medicine, which each harbour an expansive Since then, certain aspects of the reaction have been covered

Born in the U.K. and educated at Oxford, Juliet Gerrard is now a permanent resident of New Zealand. Currently a Senior Lecturer in Biochemistry at the University of Canterbury, she spent several years teaching at the Oxford colleges and five years as a researcher with Crop & Food Research, New Zealand. Juliet’s research interests include the Maillard reaction of proteins in food and biology and the enzymes of lysine biosynthesis.

in some detail: for example, Maillard chemistry pertaining to H H O H NCH2R H NHCH2R food has been described by Ames,[6–8] Nursten,[9] and HO NHCH2R H OH H OH H OH OH [10–16] others; the chemistry of Amadori rearrangement HO H H2N–CH2–R HO H HO H HO H products was surveyed in 1994;[17] kinetic aspects have been H OH amino H OH H OH H OH OH group OH OH [18] H H OH H H recently explored by van Boekel; and the capacity of CH2OH CH2OH CH2OH CH2OH Maillard chemistry for synthesis of protein–polysaccharide typical Schiff base conjugates has been described.[19] In the medical arena, sugar Baynes[20] and Monnier[21] have written accounts of the role of the Maillard reaction in ageing, and the potential of H H H O H NHCH2R H NHCH2R Maillard reaction inhibitors as anti-ageing drugs has been OH O O reviewed.[4,22,23] The role of the Maillard reaction in diabetes OH HO H H H [2,24] H OH H OH H OH has also been reviewed. Maillard chemistry is the subject H OH H OH H OH of several texts, including a recent monograph on the CH2OH CH2OH CH2OH [25–27] aminoketose— analysis of the reaction in food. the Amadori product For many years, research into the Maillard chemistry of foods and that into the Maillard chemistry that takes place in H the body were pursued separately, with different CH3 H NHCH2R nomenclatures and sparse cross-references between the two O O O O sets of literature. In recent years, the two lines of inquiry have H OH H H cross-fertilized one another, as scientists realize that the H OH H OH CH OH CH OH same chemistry which is ultimately responsible for browning 2 2 a steak on a barbecue also takes place in the body during ageing, albeit at a much slower rate. In this review, a summary of the chemistry involved in the Maillard reaction Fig. 1. The early stages of the Maillard reaction, adapted from Ledl is provided to orient the chemical reader, followed by and Schleicher.[5] highlights of developments in the last five years. Selected research from both the food science and medical literature, as well as the increasing number of research papers appearing in core chemical and biochemical journals, is foods and are found on tissue proteins in the human body, included. Finally, future prospects for research in this field and at higher levels in diabetic patients, with high blood are considered. glucose concentration.[5] The Amadori product is an important intermediate in The Early Stages of the Maillard Reaction Maillard chemistry, and the chemistry of Amadori The starting materials for Maillard chemistry have generally rearrangement provides a key to understanding the been considered as any biological amine, especially amino intermediate phase of this reaction network. Although acids and the lysine residue of proteins, with any sugar. generally drawn as the straight-chain molecule for Historically, attention has chiefly been focused on simplicity, the Amadori product (along with many other of monosaccharides, especially glucose, fructose and xylose, the early Maillard products) can also adopt cyclic furanose and disaccharides, especially maltose and lactose.[5] and pyranose forms.[17] Formation of the Amadori product However, work has also been carried out on poly- is reversible, but the half-life under biological conditions is saccharides,[19] and considerable recent research has been 6–8 months. carried out on the reactive breakdown products of sugars, such as methylglyoxal.[28] Additionally, substantial research Intermediate Stages of the Maillard Reaction is underway to elucidate the role of fat breakdown products Amadori compounds undergo many of the characteristic in Maillard chemistry.[29] These developments are discussed reactions of their component sugar and amino in detail in later sections. For simplicity, the chemistry of the acid—enolizations, eliminations, aldol and retro-aldol Maillard reaction is here described for the best-studied reactions, carbonyl migrations, nucleophilic additions, case—the reaction of glucose with a simple amine. decarboxylations, autoxidations, Michael additions, The early stages of the Maillard reaction were elucidated Canizarro rearrangements, and so on—all taking place in an by Maillard himself, Amadori, Kuhn and Weygand, Simon aqueous system. The combined set of possibilities, as an idle and Kraus, and Heyns.[5] They are summarized in Figure 1. doodle will confirm, is much larger than the decomposition In a series of reversible reactions, a carbonyl group, often products of the sugars and amino acids alone. Thus, the derived from a sugar molecule, forms a Schiff base with a apparently simple reaction of an amino group with a biological amine, typically an amino acid, or the lysine monosaccharide leads to a huge variety of products, in residue of a protein. The resulting Schiff base is labile and varying yields. Some substances may be formed in yields of may undergo two sequential rearrangements, yielding a up to 30% under favourable conditions; many others have reasonably stable aminoketose—dubbed the Amadori been detected in the parts per billion range, yet still influence product. Amadori products have been identified in processed qualities such as the aroma of food.[5] Recent Developments Concerning the Maillard Reaction 301

Any attempt to gain a complete picture of Maillard amino acid Amadori sugar chemistry demands that this plethora of information be product organized in a coherent manner. The first attempt to do this was by Hodge in 1953[30] and his scheme, outlined in Figure 2, amino acid Amadori sugar is still widely used today as a summary of the basic reaction fragmentation fragmentation fragmentation processes.[8,9] Recent research has elucidated pathways, pool pool pool notably those involving free-radical reactions,[31–33] which are not incorporated in the original scheme, but it remains a reasonable summary of the basic processes involved.[8] More recent attempts to conceptualize the Maillard reaction include that of Yaylayan,[34] as summarized in Figure 3. This depiction of Maillard chemistry aims to provide a system that uses chemical pools to consolidate and categorize the scattered information in the literature, and PARENT reduce the complexity of the system. Similar approaches POOL (one of have been adopted by those who use predictive modelling to above three) try and control the reaction with respect to a particular outcome,[18] for example if modelling the kinetics of the primary browning process. fragmentation An increasing number of the molecules in this pool intermediate phase of the Maillard reaction have now been isolated and characterized. 5-Hydroxymethylfurfural is a INTERACTION POOLS typical reaction product, which has been detected in a wide (of the parent pool, with one variety of Maillard studies. It is used as an indicator of the or both of the other pools) extent of Maillard chemistry that has taken place in various processed food systems,[35,36] and may occur in [37] concentrations as high as 1 g/kg of foodstuff. This has heterocycles, polymers, raised concerns, due to the potential mutagenicity of this advanced Maillard products compound.[37] A complete catalogue of Maillard reaction products is Fig. 3. An alternative conceptualization of the Maillard network, beyond the scope of this review, but a small selection is adapted from Yaylayan.[34] included in Figure 4. It must be emphasized that these compounds represent a minute percentage of all reported Maillard reaction products, and are simply intended to illustrate the range of structures that results from the simple interests of space, products of more complex amines, such as condensation of an aldehyde and a primary amine. In the proline or tryptophan are not included; nor have products that form in the presence of sulfur-containing compounds, such as cysteine, which include very important aroma compounds. Amadori product For many years, the Maillard literature documented newly fission –3H2O isolated and characterized structures as they were

–2H2O acetol, diacetyl, pyruvaldehyde, etc. discovered, but rarely contained definitive experiments that Schiff base of hydroxymethylfurfural shed light on their mechanisms of formation. Over the last 5 or furfural +amino reductones years, huge progress has been made in the isolation and acid, –CO2 [+2H] [–2H] purification of Maillard reaction products, providing enough –amino sugars compound dehydroreductones aldehydes material for the definitive elucidation of their structures

+H O using a variety of techniques, especially nuclear magnetic 2 [38–53] +amino resonance (NMR) spectroscopy. Hofmann is one of the compound aldols and nitrogen pioneers in this area, and has made an outstanding hydroxymethylfurfural [10,54–59] or furfural free polymers contribution to the field. In addition to the structural analysis of Maillard reaction products derived from model +amino +amino +amino +amino compound compound compound compound systems and food extracts, his group has provided mechanistic insights into the formation of many of the melanoidins identified molecules, which are often also of relevance to Fig. 2. The Amadori product and beyond, adapted from Ames,[8] those studying the reaction in vivo. based on the original Hodge scheme.[30] Note that in addition to The use of labelled molecules in elegant model studies taking part in many of the reactions in Maillard chemistry, amines has allowed pathways of formation of isolated compounds to also act as catalysts of sugar rearrangements and dehydrations. be proposed. For example, in Figure 5, the proposed pathway 302 J. A. Gerrard

OOH for formation of an isolated chromophore (Z)-2-[furfurylidene]-5,6-di(2-furyl)-6H-pyran-3-one is illustrated.

CH3 Whereas many published pathways to Maillard reaction products remain speculative, the use of specifically labelled O 13 OH O C-labelled glucose molecules and unequivocal O O identification of the position of the labels in the product H CH3 O O O O molecule provides substantive evidence for the proposed HO [59] O O OH pathway. 5-hydroxymethylfurfural

CH The Late Stages of the Maillard Reaction 3 O HN O Many of the low molecular weight compounds that form N HO during the Maillard reaction ultimately react together to form O O O HO OH heterogeneous polymers during the later stages. These high molecular weight, highly coloured, materials are called CH2OH melanoidins, and are extremely difficult to separate and H OH H H H OH characterize from any Maillard reaction mixture. The N N HO R O O structural elucidation of the melanoidins thus remained a H H [5] N N challenge over many decades, with little progress made. pyralline OH O O Recently however, researchers including Tressl[52] have R made substantial progress on the structural elucidation of N O— OH CH3 melanoidins by carrying out model experiments in which the N N N+ H3C N N polycondensations of variously substituted pyrroles were NH R HO 2 N N examined in combination. The structures of the resulting mixtures were analysed by mass spectrometry and NMR Fig. 4. A smorgasbord of Maillard reaction products.[5,6,16,59,110,154] spectroscopy. 13C-labelling experiments were also employed, to characterize the products further, leading to the suggestion that regular oligomers containing up to 15–30 methine-

HO HO HO O O OH OH OH HO HO O HO H O O OH O OH O OH OH– OH

OH HO HO HO H HO O HO OH O HO OH O O O O O O O HO H HO OH OH O

H+ H2O

HO O O O OH O O O O O O O CH O O H O O 2 OH 2 O O O

O H O O O O O O O O O O H2O

Fig. 5. Proposed pathway for formation of chromophore (Z)-2-[furfurylidene]-5,6-di(2-furyl)- 6H-pyran-3-one. ᭹ = 13C-labelled carbon atom from D-1-[13C]glucose; ᭿ = 13C-labelled carbon atom from D-6-[13C]glucose.[59] Reproduced from Fayle and Gerrard[27] with permission. Recent Developments Concerning the Maillard Reaction 303

CO2H

N H NH CO2H N n O NH

NCOH N N 2 HO2C N

n HO C CO H 2 CO2H CO2H 2 N N

O HN N O N N O OCO2H N CO2H O CO2H CO2H

CO2H N

HO2C R N

HO2C HO2C CO2H CO2H N HO2C N R CHO N N n N N N H HO2C N N CO2H

HO2C N CO2H CO2H

Fig. 6. Proposed structure of melanoidin-like polymer.[52] Reproduced from Fayle and Gerrard[27] with permission.

bridged pyrroles were formed. These results were used to fluorescent moiety that is believed to form through the propose both a structure for native melanoidins, as illustrated condensation of a lysine residue with an arginine residue and in Figure 6, and a mechanistic pathway for their formation a reducing sugar. The exact mechanism of its formation has (Fig. 7). been the subject of considerable debate, but pathways have In addition to polymeric material formed from repeated recently been proposed from glyceraldehyde, by Chellan and condensation of small molecules, many Maillard products Nagaraj,[33] and from pentoses and hexoses, by Biemel et have been characterized that remain protein-bound. Such al.;[65] these are illustrated in Figure 8. moieties are often a result of the reaction of a protein-bound A selection of reported AGEs is illustrated in Figure 9. amine (usually lysine or arginine) with a sugar, or sugar-like Not all of these represent cross-linked structures, for molecule. These reactions are known collectively as protein example carboxymethyllysine (CML)[66] is a common glycation, and the end-products have thus become known as product of the Maillard reaction of lysine in tissue proteins advanced glycation end-products, or AGEs, especially by that has been used as a biomarker for various ageing those researching the chemistry of ageing. processes,[67] particularly in diabetics.[68] Some of these A subset of the AGEs that has attracted particular structures are thought to derive not directly from sugar attention is that which results in protein crosslinking. Such molecules, but from sugar breakdown products such as crosslinks have been isolated from long-lived proteins such glyoxal and methylglyoxal. MOLD (methylgloxal lysine as collagen,[60] and lens crystallins,[61] where they are dimer)[69] and GOLD (glyoxal lysine dimer)[70] both fit into thought to modify the function of the protein by structural this category. change. It has been further postulated that protein crosslinks Many of the well-characterized AGEs are acid stable. have deleterious consequences throughout the body, and play This is no coincidence: most AGEs were initially isolated an important role in the ageing process.[62] One of the first from a protein that had been subject to an acidic hydrolysis protein-derived Maillard reaction products isolated and prior to characterization. However, confirmation that the characterized was the crosslink pentosidine,[63,64] a AGEs are genuine Maillard reaction products, not just 304 J. A. Gerrard

+ H H + H N N N N R R R R O O O

+ 2H H N N –H2O N N N N R R R R R OH R OH O

H N N R R N N R R NR O NR H

O OH N N + 2H R R N N NR –H2O R R NR etc. NR NR

etc. H N N N N R R R R O NR NR H H

N N N N N N O O R R R R R R NR NR NR NR NR

Fig. 7. Proposed mechanism of formation of melanoidin-like Maillard polymers from pyrroles.[52] Reproduced from Fayle and Gerrard[27] with permission. artefacts of the analytical procedures used to isolate them, cascade are unravelled, many challenging synthetic targets has been obtained by recent results from enzymatic digestion have arisen, which have begun to pique the interests of those of proteins in place of the acidic digestion step,[65,71] and by looking for a synthetic challenge. immunological detection of products in situ.[72] Despite Amadori products have been synthesized by refluxing an these advances, it seems that many as yet uncharacterized amino acid in the presence of excess sugar for several hours, AGEs are likely to be unstable in acid, and thus have escaped with subsequent isolation and purification of the product by the attention of those using the acid hydrolysis procedures. A laborious chromatography. Other techniques, such as fusion recently reported example of such an acid-labile AGE is Nω- of the starting materials in their dry state, have also been carboxymethylarginine,[43] which was identified after employed.[17] Such methods are limited by the instability of enzymatic hydrolysis of collagen. the products, which are prone to oxidation in air, and persistently react to form products further down the chemical Synthesis of Maillard Reaction Products cascade. However, the Amadori products derived from One approach to unravelling the intricacies of Maillard proline and tryptophan can be prepared in reasonable chemistry is to synthesize pure samples of putative products yield,[17] and it has recently been suggested that Amadori to provide authentic reference samples. This tactic is chemistry is under-utilized as a preparative method in natural particularly powerful if coupled with immunological product chemistry.[73] approaches to locate the product of interest in situ in a During early work into Maillard chemistry, a range of biological sample. There are two basic strategies that can be crystalline Amadori products was obtained from the reaction employed to generate pure samples of synthetic Maillard of 4,6-o-benzylidene glucose with the appropriate amine, products. In theory, at least, the course of the reaction facilitating their study in isolation.[17] Reaction of amines between an amine and a carbonyl group can be steered, in with 2,3-4,5-diisopropylidenefructose also provides a order to increase the yield of a particular molecule, which successful approach to Amadori product synthesis and study can then be purified from the resulting reaction mixture. of subsequent chemistry.[74] More recently, glucose-derived Alternatively, these capricious starting materials may be Amadori products have been synthesized by incorporating abandoned, and a novel synthesis of a target structure devised fully protected sugar derivatives into solid-phase peptide in the conventional manner. Many of the early reaction synthesis,[75] and generation of the glucose Amadori product products are small and unstable, and, like many of tyrosine has been reported under Pictet–Spengler biochemicals, make irritating targets for the synthetic conditions.[76] Glomb and Pfahler report the synthesis of a organic chemist. However, as the later stages of the reaction key glucose breakdown product, 1-deoxy-D-erythro-hexo- Recent Developments Concerning the Maillard Reaction 305

H H O H NH lysine residue H O H NH lysine residue H OH O O + H O O lysine residue NH2 [O] 2 HO H HO H HO H HO H – lysine residue NH H OH H OH H OH 2 H OH HOH HOH HOH HOH R R R R

–3H2O

H NH2 N arginine residue + NH2 former lysine residue other products N N H N HO N former arginine residue pentosinane

oxidation –H2O

former lysine residue H N N NH N former arginine residue pentosidine

–H2O former lysine residue H N N NH HO N former arginine residue

–H2O former lysine residue former lysine residue former lysine residue H H HO N N –glyceraldehyde N N HN N NH NH NH former arginine residue N former arginine residue HO N former arginine residue N GODIC (glyoxal-derived lysine residue NH imidazoline crosslink) 2

O HN NH N former arginine residue

HO [O] HO HO HN HN –HCHO HO HN NH O HO NH NH H N arginine residue 2 N former arginine residue N former arginine residue

Fig. 8. Proposed pathways for pentosidine formation in vivo, from pentoses or hexoses,[65] or from glyceraldehyde.[33]

2,3-diulose, and demonstrate its ability to modify Factors Influencing the Maillard Reaction proteins.[41] Independent synthesis of putative AGE structures has In stark contrast to traditional preparative chemistry, the huge potential in the advancement of our knowledge in this significance of the Maillard reaction in a particular situation area, which is just beginning to be realized. Structural is determined not by the yield of a compound, but by its characterization of pentosidine, by NMR and mass effect. A few µg of product may be sufficient to characterize spectrometry, has been assisted by synthesis of this molecule the aroma of a heated food;[5] AGE formation within the using a variety of approaches,[77,78] providing authentic proteins comprising an eye-lens may alter the opacity.[80] samples of the compound for comparison to Maillard Furthermore, the product pattern is critically dependent on reaction product mixtures. Similarly, independent syntheses the precise reaction conditions under which the reaction is of GOLA and GALA,[71] lysine–arginine crosslinks formed occurring. Although the scope for changing Maillard from pentoses and hexoses[65] and aminophospholipid- reaction conditions in vivo is clearly limited, food processors linked Maillard products[79] were all used in recent studies to have invested considerable effort in investigating the effect confirm the identity of AGEs that had been isolated from of process variables on the chemistry and its consequences, biological samples. in terms of product quality. 306 J. A. Gerrard

H R the kidney of a diabetic patient. The vast number of H3C N NR NH R R O interfering substances present in even the simplest of NH N N HO H O biological systems, coupled with the acute sensitivity of – HO CO2 OH CH3 Maillard reaction product ratios to reaction conditions, renders this problem a considerable challenge. Thus, carboxymethyllysine argpyrimidine pyrraline glycolic acid– (CML) lysine–amide whether the conclusions reached in model studies in any way (GALA) represent reality is open to question. There remains a large R R R R gap in our knowledge that must be filled in order to relate the H3C N NH N N N H HO NR results of chemically defined experiments to those less ONH H N N N defined studies that better mimic a given physiological or R R R HO food-processing situation. The use of antibodies for the MOLD GOLD glucosepane glyoxal–lysine–amide (GOLA) detection of Maillard products has begun to find wide application in the medical field,[47,102–108] and goes some way CH2OH CH3 to addressing this problem, but ultimately the reaction must HO H H R N R N N H be studied in the situation of interest—usually in food or in HO H NR N OH HO HO H N HN the human body. OH HO NH [27] N+ R The Maillard Reaction in Food R N HO R Maillard chemistry takes place in nearly all food, especially crosslines DOGDIC MODIC food that contains high levels of sugars or fat, which has been deoxyglucosone-derived methylglyoxal-derived [8] imidazoline crosslink imidazoline crosslink cooked. It affects many of the sensory properties of food during storage and processing. These include: changes in Fig. 9. A selection of reported advanced glycation end-products that have been identified in tissue proteins.[23,64,65,71,130,155,156] colour, particularly browning and, to a lesser extent, fluorescence; production of aroma and flavour compounds; production of bioactive compounds, both beneficial and toxic; loss of nutritional quality, especially of proteins; and changes The relationship between the conditions under which the in texture.[26] In order to correlate changes in each of these Maillard reaction takes place and its specific consequences individual properties of a particular food with Maillard is, not surprisingly, complex. The temperature–time chemistry, quantitative methods are required to measure each combination is an important parameter. For example a long, property of interest. Quantifying sensory properties is no easy gentle heating may produce a more desirable product, in task, since many of the properties that make food appealing terms of colour and flavour parameters, than a short, high- are subjective qualities, but food scientists have established temperature process. Water activity and pH also alter a raft of methods with which to measure individual properties Maillard reaction pathways, and may strongly influence the of foods with some precision.[27] product profile, a fact that may prove useful for food The Maillard reaction is so commonly associated with processors.[8,81] As a general rule, model studies have colour formation that it is often dubbed ‘the browning focused on buffered systems in the pH range 4–7.5,[5] but reaction’ or ‘non-enzymic browning’, the latter to unbuffered systems have also been studied; buffers distinguish this phenomenon from the equally common food themselves have been found to influence the reaction browning reactions caused by polyphenol oxidase and other pathway,[82–85] so model studies should be interpreted with enzymes.[16] Often, this browning is an integral part of the caution. Pressure has also been found to affect Maillard food processing of interest—and ‘brownness’ is the quality chemistry of model systems, and in a pH-dependent way.[8] to which the consumer responds, in bread crust, for example. Substances that may well be present during Maillard Elsewhere, browning has been used as a measurable chemistry, such as metals,[86] atmospheric oxygen,[86,87] and symptom of the Maillard reaction, making the large, yet Maillard inhibitors such as sulfite,[88] may all result in simplifying, assumption that the extent of browning provides dramatic changes in the observed product ratios. a quantitative indicator of the extent of chemical reaction. Hofmann[58,109,110] has developed the ‘colour activity The Maillard Reaction In Situ concept’ to address the surprising lack of knowledge as to the Our knowledge of the Maillard reaction, as summarized key chromophores that evoke Maillard browning. This above, has largely been derived from the study of model method facilitates evaluation of the most intensely coloured systems—laboratory simulations of a particular situation of chromophores of any Maillard reaction, without the need for interest, containing a limited number of components. knowledge of their chemical structures. Early results suggest Typically, these contain simple mixtures of compounds, such that, despite the complexity of the network of browning as an individual amino acid and a sugar.[49,55,89–95] Other reactions, large proportions of the total colour observed may researchers have extended this concept to the use of proteins be accounted for by a few key chromophores. This allows the and sugars in model systems.[64,96–101] Ultimately, the challenging structural elucidation experiments, such as those validity of these results in the situation of interest must be summarized in Figure 5, to focus on those compounds tested, whether this be in a food processing plant or within known to be responsible for producing colour. Recent Developments Concerning the Maillard Reaction 307

In addition to the often desirable colour of foods that loss of nutritional quality during the processing of foods derives from the products of the Maillard reaction, many containing reducing sugars and proteins has long been a aroma and flavour compounds are to be found amongst this concern, especially in dairy products.[6] The importance of complex network of molecules. Maillard chemistry plays a the Maillard reaction of proteins, or protein glycation, in role in forming the distinctive flavours of many foods and vivo, became clear upon the discovery of glycated beverages—among them chocolate,[111] coffee[112] and bakery haemoglobin.[5] As was noted in an early article by Furth and products.[113] In fact, the aroma of most foods that are Harding: in the food industry, ‘glycation is bad news, subjected to baking, roasting and grilling will contain Maillard because proteins that are modified by sugars tend to turn reaction products. Almost invariably, hundreds of volatile yellowy-brown on standing. … It now seems that we too go compounds have been isolated from each food studied.[5] yellowy-brown on standing—on ageing, that is.’[129] Not The food industry has invested great effort in trying to surprisingly, the problems associated with sugar damage to create synthetic flavours and aromas by reconstituting proteins are exacerbated in diabetics, and protein glycation is combinations of these compounds, so called ‘aromagrams’, a very active area of research in the diabetes field.[2,24,130] during processing. This approach has met with limited It is worth pausing to note that glycation chemistry in vivo success, since the subtleties of flavour perception are many is fundamentally different from the better studied and varied, and the detection of these compounds, however phenomenon of protein glycosylation, in that the latter is sophisticated the separation system might be, must by carefully controlled by enzymes, whereas the former is a necessity use a human nose as a detector.[114] Considerable result of the spontaneous reaction of a sugar and a protein in progress has been made using the concept of ‘odour a Maillard event.[129] Protein glycation happens at a much activity’, analogous to the colour activity concept described slower rate than protein glycosylation, but is still significant above, which is a parameter relating to the threshold over the lifetime of a protein. Glucose, the predominant concentration at which a compound must be present before sugar in the body, is the least reactive of the aldohexoses to its odour is detectable.[115] Maillard chemistry, due to its preference to exist in an Although the effect of the Maillard reaction on colour, unreactive cyclic form. This limits the amount of deleterious flavour and aroma is well documented, the effect on food Maillard chemistry that takes place in cells. Indeed, there has texture has attracted less attention. However, increasing been suggestion that this factor may have played a role in evidence suggests that protein crosslinking chemistry takes evolution, with glucose emerging as the primary metabolic place within food systems[116] which may have profound fuel in biology due to its comparatively low reactivity effects on food texture.[117–121] This emerging area of towards protein.[131] research is likely to offer new methods for the manipulation Amadori products and AGEs have been detected in long- of texture in processed foods. lived proteins in vivo, particularly in collagen and the Amongst the catalogue of Maillard reaction products, crystalline proteins of the lens, but are also found in shorter- there lurk many bioactive compounds, some of which lived plasma proteins and in intracellular enzymes, in fact in may be beneficial, for example, with anti-oxidant any protein exposed to glucose. The Amadori product levels properties,[16,122–124] and some anti-nutritional[125] or have been measured and appear to remain almost constant in toxic.[126] Although our understanding of the chemistry of a normal ageing lens, but occur at much higher levels in formation of these compounds remains in its infancy, diabetic and senile cataractous lenses.[33] Many tissues there are enormous potential benefits of understanding develop increased AGE content during ageing, making and controlling the Maillard reaction in such a way that moieties such as pentosidine and carboxymethyllysine useful beneficial compounds could be produced in the absence biomarkers for Maillard damage to proteins in vivo.[132,133] of unwanted or toxic compounds. Crosslinking AGEs are thought to influence the elasticity of In addition to the effects of toxic, or anti-nutritional,[16] the extracellular matrix. Increasingly, high levels of AGEs products of the Maillard reaction, the nutritional quality of are being associated with specific pathologies, such as foods may be damaged by the Maillard reaction during cataract formation, Alzheimer’s disease, amyloidosis, processing. This can occur for two main reasons. First, artherosclerosis, diabetic retinopathy,[33] pulmonary fibrosis aggregation of proteins on heating is known to decrease and male erectile dysfunction.[130,134] protein digestibility and is exacerbated by Maillard The chemistry of the Maillard reaction in vivo is the subject reactions.[127] Second, the Maillard reaction often leads to the of increasing literature scrutiny. In early work, the role of loss of amino acids, especially lysine. Since lysine is often glucose as the participating sugar in protein crosslinking the limiting nutrient in a diet, its irreversible loss during reactions was the primary focus, along with related molecules, Maillard chemistry can have a damaging effect on the overall e.g. ribose and ascorbate. [63,135,136] However, more recently, nutritional value of a food.[16] There is also mounting attention has been directed toward α-dicarbonyl compounds evidence that AGEs consumed in the diet of diabetics pose such as 3-deoxyglucosone, and glycolytic intermediates such an added risk factor for those with renal impairment.[128] as sugar phosphates, methylglyoxal and glyoxal, as these have been found to be active protein crosslinkers in vitro and in The Maillard Reaction in Medicine vivo.[137–140] The mechanisms of glyoxal and methylglyoxal The study of the Maillard reaction of proteins has its origins generation in vivo have been investigated, with a number of in food science, where problems of colour development and works describing the generation of dicarbonyls by glucose 308 J. A. Gerrard autoxidation[141,142] or by degradation of an Amadori remains an understanding of the underlying chemistry product.[143] Researchers have suggested that methylglyoxal behind these many and varied processes. Chemists are and glyoxal can be generated from the degradation of the ideally placed to take up this continuing challenge. Schiff base adduct that is initially formed on reaction of sugars with an amine (Fig. 1).[28] Methylglyoxal can also be formed Acknowledgments by enzymatic routes, from dihydroxyacetone phosphate and I thank the many research students who have worked with me glyceraldehyde 3-phosphate.[144] on the Maillard reaction over the last few years; Dr Susie Another emerging area in both the food and medical arena Meade, and Antonia Miller, Plant & Microbial Sciences is the role of lipids and their oxidation products in Maillard Department, University of Canterbury, for assistance with chemistry in vivo. Many products of lipoxidation are compilation of the bibliography; Daniel Owen for evidently able to undergo Maillard-type chemistry to form commodious office facilities, inter alia; and Peter Steel, what have been dubbed ‘ALEs’, or Advanced Lipoxidation Chemistry Department, University of Canterbury, for proof- End-products. Whether a particular product that has been reading this manuscript. The University of Canterbury is isolated from a biological specimen is an AGE or an ALE gratefully acknowledged for its generous study leave can sometimes be rather arbitrary, which has led to a rather provisions, which have made the writing of this review stylish new acronym—EAGLE—Either an Advanced possible. Glycation or an advanced Lipoxidation End-product.[20] The relative importance of ALEs and AGEs is still under debate, References as is the relative impact of ‘traditional’ Maillard-type reactions and free-radical chemistry in the ageing [1] L.-C. Maillard, C. R. Acad. Sci. Ser. 2 1912, 154, 66. process.[20,24,33] It is becoming evident that glycated proteins [2] A. J. Furth, Br. J. Biomed. Sci. 1997, 54, 192. [33,145] [3]S. Vasan, P. G. Foiles, H. W. Founds, Expert Opin. Investig. are themselves a source of radical chemistry. AGEs Drugs 2001, 10, 1977. can bind to specific receptors on cells—RAGES—and cause [4] D. L. Price, P. M. Rhett, S. R. Thorpe, J. W. Baynes, J. Biol. oxidative stress, usually resulting in damage to the affected Chem. 2001, 276, 48967. tissues[33] by triggering the production of reactive oxygen [5] F. Ledl, E. Schleicher, Angew. Chem., Int. Ed. Engl. 1990, 29, 565. species.[71] [6] J. M. Ames, The Maillard Reaction in Foods, in Biochemistry of Food Proteins (Ed. B. J. F. Hudson) 1992 (Elsevier Science: The body has mechanisms to deal with harmful protein U.S.A.). chemistry. Most proteins are turned over in the body on a [7] J. M. Ames, in The Maillard Reaction. Consequences for the timescale that prevents a build-up of damaged tissues. Thus Chemical and Life Sciences (Ed. R. Ikan) 1996 (John Wiley & it is only in in vitro simulations or in long-lived proteins that Sons: Chichester). accumulated deleterious chemistry can be witnessed. Other [8] J. M. Ames, Food Chem. 1998, 62, 431. [9] H. E. Nursten, Volatiles Produced by the Maillard Reaction, in in vivo mechanisms to repair Maillard damage have recently Frontiers of Flavour Science (Eds P. Schieberle and K.-H. Engel) been discovered. Fructose amines, a product of the Maillard 2000 (German Centre for Food Chemistry: Garching, Germany). reaction of fructose, have recently been found to be [10] T. Hofmann, ACS Symp. Ser. 2001, 775, 134. metabolized in vivo.[146,147] The Amadoriases are a novel [11] L. Manzocco, S. Calligaris, D. Mastrocola, M. C. Nicoli, C. R. class of enzyme that catalyse the decomposition of Amadori Lerici, Trends Food Sci. Technol. 2000, 11, 340. [148–152] [12] V. Fogliano, Ind. Aliment. 2001, 40, 385. products. To date, characterized enzymes act upon [13] S. Martins, W. M. F. Jongen, M. van Boekel, Trends Food Sci. glycated amino acid substrates, with little or no activity on Technol. 2000, 11, 364. glycated proteins. However, this is most likely to be a [14] L. W. Kroh, A. Schulz, Nahrung 2001, 45, 160. reflection of the substrates used to screen organisms for their [15] F. Hayase, Recent Res. Dev. Agric. Biol. Chem. 1997, 1, 351. activity. The discovery of an Amadoriase that acts on [16] M. Friedman, J. Agric. Food Chem. 1996, 44, 631. [17] V. A. Yaylayan, A. Huyghues-Despointes, Crit. Rev. Food Sci. glycated protein in vivo cannot be far away. Nutr. 1994, 34, 321. [18] M. van Boekel, Nahrung 2001, 45, 150. Future Directions [19] A. Kato, Dev. Food Sci. 2001, 41(Novel Macromolecules in Food As the chemistry of ageing, and the biochemistry of repair, Systems), 385. become increasingly well understood, the potential for [20] J. W. Baynes, Biogerontology 2000, 1, 235. [21] V. M. Monnier, J. F. Fogarty, C. S. Monnier, D. R. Sell, Methods reversing the chemistry of AGEing increases, raising the Aging Res. 1999, 657. tantalizing possibility of anti-ageing therapeutics. [22] R. G. Khalifah, J. W. Baynes, B. G. Hudson, Biochem. Biophys. Amadoriases have already been patented due to their possible Res. Comm. 1999, 257, 251. role in anti-ageing gene therapies. Additionally, several new [23] S. Dukic-Stefanovic, R. Schinzel, P. Riederer, G. Munch, compounds have been described in the literature which are Biogerontology 2001, 2, 19. [24] T. J. Lyons, A. J. Jenkins, Diabetes Rev. 1997, 5, 365. purported to prevent or reverse in vivo Maillard chemistry, and [25] R. Ikan, (Ed.) The Maillard Reaction: Consequences for the thence alleviate the symptoms of ageing. Some of these Chemical and Life Sciences 1996 (Wiley: New York) compounds are in clinical trials and undoubtedly reduce [26] J. O’Brien, H. E. Nursten, M. J. C. Crabbe, J. M. Ames, The pathological effects in animal models,[153] although their Maillard Reaction in Foods and Medicine 1998 (Royal Society of mechanism of action remains hotly debated.[3,4,22] Chemistry: London). [27] S. E. Fayle, J. A. Gerrard, The Maillard Reaction, Royal Society In both food science and medicine, the bottleneck in of Chemistry Food Analysis Monographs 2002 (Royal Society of controlling the Maillard reaction and its many consequences Chemistry: London). Recent Developments Concerning the Maillard Reaction 309

[28] P. J. Thornalley, A. Langborg, H. S. Minhas, Biochem. J. 1999, [68] P. Shibayama, N. Araki, R. Nagai, S. Horiuchi, Diabetes 1999, 344, 109. 48, 1842. [29] P. A. Voziyan, T. O. Metz, J. W. Baynes, B. G. Hudson, J. Biol. [69] E. Brinkmann, K. J. Wells-Knecht, S. R. Thorpe, J. Baynes, J. Chem. 2002, 277, 3397. Chem. Soc., Perkin Trans. 1 1995, 2817. [30] J. E. Hodge, J. Agric. Food Chem. 1953, 1, 928. [70] K. Wells-Knecht, E. Brinkmann, J. Baynes, J. Org. Chem. 1995, [31] T. Hofmann, W. Bors, K. Stettmaier, J. Agric. Food Chem. 1999, 60, 6246. 47, 379. [71] M. Glomb, C. Pfahler, J. Biol. Chem. 2001, 276, 41638. [32] V. V. Mossine, M. Linetsky, G. V. Glinsky, B. J. Ortwerth, M. S. [72] Y. Al-Abed, R. Bucala, Bioconjugate Chem. 2000, 11, 39. Feather, Chem. Res. Toxicol. 1999, 12, 230. [73] T. M. Wrodnigg, B. Eder, Top. Curr. Chem. 2001, 215, 115. [33] P. Chellan, R. H. Nagaraj, J. Biol. Chem. 2001, 276, 3895. [74] D. Xenakis, N. Moll, B. Gross, Synthesis 1983, 541. [34] V. A. Yaylayan, Trends Food Sci. Technol. 1997, 8, 13. [75] A. Jakas, S. Horvat, J. Chem. Soc., Perkin Trans. 2 1996, 789. [35] A. Ramirez-Jimenez, E. Guerra-Hernandez, B. Garcia- [76] P. Manini, M. d’Ischia, G. Prota, J. Org. Chem. 2001, 66, 5048. Villanova, J. Agric. Food Chem. 2000, 48, 4176. [77] H. Sugiyama, F. Yokokawa, T. Shioiri, N. Katagiri, O. Oda, H. [36] M. D. del Castillo, M. Villamiel, A. Olano, N. Corzo, J. Agric. Ogawa, Tetrahedron Lett. 1999, 40, 2569. Food Chem. 2000, 48, 4217. [78] F. Yokokawa, H. Sugiyama, T. Shioiri, N. Katagiri, O. Oda, H. [37] C. Janzowski, V. Glaab, E. Samimi, J. Schlatter, G. Eisenbrand, Ogawa, Tetrahedron 2001, 57, 4759. Food Chem. Toxicol. 2000, 38, 801. [79] C. M. Utzmann, M. O. Lederer, Carbohydrate Res. 2000, 325, [38] S. Diem, M. Herderich, J. Agric. Food Chem. 2001, 49, 2486. 157. [39] T. Shimamura, H. Ukeda, M. Sawamura, J. Agric. Food Chem. [80] N. J. Akhtar, T. X. Sun, J. J. N. Liang, Current Eye Res. 1999, 18, 2000, 48, 6227. 270. [40] R. G. Bailey, J. M. Ames, J. Mann, J. Agric. Food Chem. 2000, [81] J. M. Ames, Trends Food Sci. Technol. 1990, 1, 150. 48, 6240. [82] L. N. Bell, Food Chem. 1997, 59, 143. [41] M. A. Glomb, C. Pfahler, Carbohydrate Res. 2000, 329, 515. [83] L. N. Bell, K. L. White, Y. H. Chen, J. Food Sci. 1998, 63, 785. [42] J. S. Elmore, D. S. Mottram, J. Agric. Food Chem. 2000, 48, [84] F. Tessier, I. Birlouez-Aragon, Glycoconjugate J. 1998, 15, 571. 2420. [85] A. Lapolla, D. Fedele, R. Aronica, L. Baldo, M. D’Alpaos, R. [43] K. Iijima, M. Murata, H. Takahara, S. Irie, D. Fujimoto, Biochem. Seraglia, P. Traldi, Rapid Commun. Mass Spectrom. 1996, 10, J. 2000, 347, 23. 1512. [44] A. Ravagli, G. Boschin, L. Scaglioni, A. Arnoldi, J. Agric. Food [86] F. Hayase, T. Shibuya, J. Sato, M. Yamamoto, Biosci. Biotechnol. Chem. 1999, 47, 4962. Biochem. 1996, 60, 1820. [45] S. de Gennaro, G. G. Birch, S. A. Parke, B. Stancher, Food Chem. [87] J. E. Litchfield, S. R. Thorpe, J. W. Baynes, Int. J. Biochem. Cell 2000, 68, 179. Biol. 1999, 31, 1297. [46] F. J. Morales, A. Arnoldi, Food Chem. 1999, 67, 185. [88] R. Iyengar, A. J. McEvily, Trends Food Sci. Technol. 1992, 3, 60. [47] T. Oya, N. Hattori, Y. Mizuno, S. Miyata, S. Maeda, T. Osawa, K. [89] L. P. Leong, B. L. Wedzicha, Food Chem. 2000, 68, 21. Uchida, J. Biol. Chem. 1999, 274, 18492. [90] M. Bristow, N. S. Isaacs, J. Chem. Soc., Perkin Trans. 2 1999, [48] T. Hofmann, Carbohydrate Res. 1998, 313, 203. 2213. [49] J. M. Ames, R. G. Bailey, J. Mann, J. Agric. Food Chem. 1999, [91] O. Mandin, S. C. Duckham, J. M. Ames, J. Agric. Food Chem. 47, 438. 1999, 47, 2355. [50] M. F. Wang, Y. Jin, J. G. Li, C. T. Ho, J. Agric. Food Chem. 1999, [92] A. Sensidoni, D. Peressini, C. M. Pollini, J. Sci. Food Agric. 47, 48. 1999, 79, 317. [51] P. C. Howard, M. I. Churchwell, L. H. Couch, M. M. Marques, D. [93] Y. Al-Abed, P. Ulrich, A. Kapurniotu, E. Lolis, R. Bucala, Biorg. R. Doerge, J. Agric. Food Chem. 1998, 46, 3546. Med. Chem. Lett. 1995, 5, 2929. [52] R. Tressl, G. T. Wondrak, L. A. Garbe, R. P. Kruger, D. Rewicki, [94] Y. Al-Abed, D. Callaway, A. Kapurniotu, T. Holak, W. Voelter, R. J. Agric. Food Chem. 1998, 46, 1765. Bucala, Polish J. Chem. 1999, 73, 117. [53] R. Tressl, G. T. Wondrak, R. P. Kruger, D. Rewicki, J. Agric. Food [95] A. D’Agostina, M. Negroni, A. Arnoldi, J. Agric. Food Chem. Chem. 1998, 46, 104. 1998, 46, 2554. [54] T. Hofmann, J. Agric. Food Chem. 1999, 47, 4763. [96] V. Fogliano, S. M. Monti, A. Visconti, G. Randazzo, A. M. [55] T. Hofmann, S. Heuberger, Z. Lebensm. Unters. Forsch. 1999, Facchiano, G. Colonna, A. Ritieni, Biochim. Biophys. Acta 1998, 208, 17. 1388, 295. [56] T. Hofmann, Eur. Food Res. Technol. 1999, 209, 113. [97] R. L. Bunde, E. J. Jarvi, J. J. Rosentreter, Talanta 2000, 51, 159. [57] H. Ottinger, A. Bareth, T. Hofmann, J. Agric. Food Chem. 2001, [98] L. Pellegrino, I. De Noni, S. Cattaneo, Nahrung 2000, 44, 193. 49, 1336. [99] R. Stockmann, R. K. Weerakkody, Aust. J. Dairy Technol. 2000, [58] O. Frank, T. Hofmann, J. Agric. Food Chem. 2000, 48, 6303. 55, 108. [59] O. Frank, S. Heuberger, T. Hofmann, J. Agric. Food Chem. 2001, [100] J. A. Gerrard, S. E. Fayle, K. H. Sutton, J. Agric. Food Chem. 49, 1595. 1999, 47, 1183. [60] V. M. Monnier, O. Bautista, D. Kenny, D. R. Sell, J. Fogarty, W. [101] S. E. Fayle, J. A. Gerrard, L. Simmons, S. J. Meade, E. A. Reid, Dahms, P. A. Cleary, J. Lachin, S. Genuth, Diabetes 1999, 48, A. C. Johnston, Food Chem. 2000, 70, 193. 870. [102] Y. Sato, T. Kondo, T. Ohshima, Histopathology 2001, 38, 217. [61] R. H. Nagaraj, C. Sady, FEBS Lett. 1996, 382, 234. [103] N. Nyhlin, Y. Ando, R. Nagai, O. Suhr, M. El Sahly, H. Terazaki, [62] S. Zarina, H. R. Zhao, E. C. Abraham, Mol. Cell. Biochem. 2000, S. Yamashita, M. Ando, S. Horiuchi, J. Internal Med. 2000, 247, 210, 29. 485. [63] D. R. Sell, V. Monnier, J. Biol. Chem. 1989, 264, 21597. [104] F. A. Shamsi, R. H. Nagaraj, Current Eye Res. 1999, 19, 276. [64] D. G. Dyer, J. A. Blackledge, S. R. Thorpe, J. W. Baynes, J. Biol. [105] N. Miyazawa, Y. Kawasaki, J. Fujii, M. Theingi, A. Hoshi, R. Chem. 1991, 266, 11654. Hamaoka, A. Matsumoto, N. Uozumi, T. Teshima, N. Taniguchi, [65] K. M. Biemel, O. Reihl, J. Conrad, M. O. Lederer, J. Biol. Chem. Biochem. J. 1998, 336, 101. 2001, 276, 23405. [106] T. Matsuse, E. Ohga, S. Teramoto, M. Fukayama, R. Nagai, S. [66] M. U. Ahmed, S. R. Thorpe, J. W. Baynes, J. Biol. Chem. 1986, Horiuchi, Y. Ouchi, J. Clin. Pathol. 1998, 51, 515. 261, 4889. [107] K. Uchida, O. T. Khor, T. Oya, T. Osawa, Y. Yasuda, T. Miyata, [67] K. Ikeda, T. Higashi, H. Sano, Y. Jinnouchi, M. Yoshida, T. Araki, FEBS Lett. 1997, 410, 313. S. Ueda, S. Horiuchi, Biochemistry 1996, 35, 8075. [108] Y. Al-Abed, R. Bucala, Bioconjugate Chem. 2000, 11, 39. 310 J. A. Gerrard

[109] T. Hofmann, O. Frank, Abstr. Pap. Am. Chem. Soc. 1999, 218, 26. [134] J. C. Morris, unpublished results 2002. [110] T. Hofmann, J. Agric. Food Chem. 1998, 46, 3912. [135] S. H. Slight, M. S. Feather, B. J. Ortwerth, Biochim. Biophys. [111] C. Q. Chen, E. Robbins, ACS Symp. Ser. 2000, 754, 286. Acta 1990, 1038, 367. [112] A. N. Wijewickreme, D. D. Kitts, Food Chem. Toxicol. 1998, 36, [136] N. G. Watkins, S. R. Thorpe, J. W. Baynes, J. Biol. Chem. 1985, 543. 260, 10629. [113] W. Grosch, P. Schieberle, Cereal Chem. 1997, 74, 91. [137] P. Chellan, R. H. Nagaraj, Arch. Biochem. Biophys. 1999, 368, [114] J. P. Dufour, L. P. Albela, Abstr. Pap. Am. Chem. Soc. 1999, 218, 98. 91. [138] F. E. Brinkmann, T. P. Degenhardt, S. R. Thorpe, J. W. Baynes, J. [115] P. Schieberle, T. Hofmann, J. Agric. Food Chem. 1997, 45, 227. Biol. Chem. 1998, 273, 18714. [116] L. K. Peng, N. Ismail, A. M. Easa, J. Food Sci. Technol.-Mysore [139] M. A. Glomb, G. Lang, J. Agric. Food Chem. 2001, 49, 1493. 2000, 37, 188. [140] H. Odani, T. Shinzato, J. Usami, Y. Matsumoto, E. B. Frye, J. W. [117] J. A. Gerrard, M. P. Newberry, M. Ross, A. J. Wilson, S. E. Fayle, Baynes, K. Maeda, FEBS Lett. 1998, 427, 381. S. Kavale, J. Food Sci. 2000, 65, 312. [141] K. J. Wells-Knecht, D. V. Zyzak, J. E. Litchfield, S. R. Thorpe, J. [118] J. A. Gerrard, S. E. Fayle, A. J. Wilson, M. P. Newberry, M. Ross, W. B ay n e s , Biochemistry 1995, 34, 3702. S. Kavale, J. Food Sci. 1998, 63, 472. [142] S. P. Wolff, R. T. Dean, Biochem. J. 1987, 245, 243. [119] I. N. A. Ashie, T. C. Lanier, J. Food Sci. 1999, 64, 704. [143] D. V. Zyzak, J. M. Richardson, S. R. Thorpe, J. W. Baynes, Arch. [120] N. Seki, H. Nozawa, S. W. Ni, Fish Sci. 1998, 64, 959. Biochem. Biophys. 1995, 316, 547. [121] H. Y. Ting, S. Ishizaki, M. Tanaka, J. Muscle Food 1999, 10, 279. [144] P. J. Thornalley, Mol. Aspects Med. 1993, 14, 287. [122] M. Anese, L. Manzocco, M. C. Nicoli, C. R. Lerici, J. Sci. Food [145] M. B. Yim, H. S. Yim, C. Lee, S. O. Kang, P. B. Chock, Ann. N.Y. Agric. 1999, 79, 750. Acad. Sci. 2001, 928, 48. [123] D. Mastrocola, M. Munari, J. Agric. Food Chem. 2000, 48, 3555. [146] G. Delpierre, F. Vanstapel, V. Stroobant, E. Van Schaftingen, [124] Y. Yoshimura, T. Iijima, T. Watanabe, H. Nakazawa, J. Agric. Biochem. J. 2000, 352, 835. Food Chem. 1997, 45, 4106. [147] B. S. Szwergold, S. Howell, P. J. Beisswenger, Diabetes 2001, 50, [125] M. Alonso, J. Zapico, J. Food Biochem. 1995, 18, 393. 2139. [126] Z. Balogh, J. I. Gray, E. A. Gomaa, A. M. Booren, Food Chem. [148] C. Gerhardinger, M. S. Marion, A. Rovner, M. Glomb, V. M. Toxicol. 2000, 38, 395. Monnier, J. Biol. Chem. 1995, 270, 218. [127] R. F. Hurrell, P. A. Finot, in Digestibility and Amino Acid [149] A. K. Saxena, P. Saxena, V. M. Monnier, J. Biol. Chem. 1996, Availability in Cereals and Oilseeds (Eds J. W. Finley and D. T. 271, 32803. Hopkins) 1985 (American Association of Cereal Chemists: MN, [150] M. Takahashi, M. Pischetsrieder, V. M. Monnier, J. Biol. Chem. U.S.A.). 1997, 272, 3437. [128] T. Koschinsky, C. J. He, T. Mitsuhashi, R. Bucala, C. Liu, C. [151] M. Takahashi, M. Pischetsrieder, V. M. Monnier, J. Biol. Chem. Buenting, K. Heitmann, H. Vlassara, Proc. Natl Acad. Sci. U.S.A. 1997, 272, 12505. 1997, 94, 6474. [152] X. Wu, M. Takahashi, S. G. Chen, V. M. Monnier, Biochemistry [129] A. Furth, J. Harding, New Scientist 1989, September, 26. 2000, 39, 1515. [130] A. Stitt, Br. J. Ophthalmol. 2001, 85, 746. [153] M. Asif, J. Egan, S. Vasan, G. N. Jyothirmayi, M. R. Masurekar, [131] H. F. Bunn, P. J. Higgins, Science 1981, 213, 222. S. Lopez, C. Williams, R. L. Torres, D. Wagle, P. Ulrich, A. [132] M. A. Friedlander, Y. C. Wu, C. P. Randle, G. P. Baumgardner, P. Cerami, M. Brines, T. J. Regan, Proc. Natl Acad. Sci. U.S.A. B. DeOreo, V. M. Monnier, in J. O’Brien, H. E. Nursten, M. J. C. 2000, 97, 5679. Crabbe, J. M. Ames, The Maillard Reaction in Foods and [154] A. Arnoldi, E. A. Corain, L. Scaglioni, J. M. Ames, J. Agric. Medicine 1998 (Royal Society of Chemistry: London). Food Chem. 1997, 45, 650. [133] Y. Ueda, T. Miyata, T. Hashimoto, H. Yamada, Y. Izuhara, H. [155] Y. Al-Abed, T. Mitsuhashi, P. Ulrich, R. Bucala, Biorg. Med. Sakai, K. Kurokawa, Biochem. Biophys. Res. Commun. 1998, Chem. Lett. 1996, 6, 1577. 245, 785. [156] M. O. Lederer, R. G. Klaiber, Bioorg. Med. Chem. 1999, 7, 2499. This article was downloaded by: [UNAM Ciudad Universitaria] On: 09 April 2014, At: 15:45 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Food Reviews International Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/lfri20 TOXICOLOGY AND ANTIOXIDANT ACTIVITIES OF NON- ENZYMATIC BROWNING REACTION PRODUCTS: REVIEW Kwang-Geun Lee a & Takayuki Shibamoto b a Department of Environmental Toxicology , University of California, Davis , One Shields Avenue, Davis, CA, 95616, U.S.A. b Department of Environmental Toxicology , University of California, Davis , One Shields Avenue, Davis, CA, 95616, U.S.A. Published online: 02 Nov 2011.

To cite this article: Kwang-Geun Lee & Takayuki Shibamoto (2002) TOXICOLOGY AND ANTIOXIDANT ACTIVITIES OF NON- ENZYMATIC BROWNING REACTION PRODUCTS: REVIEW, Food Reviews International, 18:2-3, 151-175, DOI: 10.1081/ FRI-120014356 To link to this article: http://dx.doi.org/10.1081/FRI-120014356

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

FOOD REVIEWS INTERNATIONAL Vol. 18, Nos. 2 & 3, pp. 151–175, 2002

TOXICOLOGY AND ANTIOXIDANT ACTIVITIES OF NON-ENZYMATIC BROWNING REACTION PRODUCTS: REVIEW

Kwang-Geun Lee and Takayuki Shibamoto*

Department of Environmental Toxicology, University of California, Davis, One Shields Avenue, Davis, CA 95616

ABSTRACT

This article summarizes toxicity and antioxidative activity of non-enzymatic browning reaction products. The subject focuses on the formation and toxicity testing of mutagenic Maillard reaction products (MPRs) formed in food model systems and in actual foods. The MRPs have also been investigated for nutritional, physiological, and biological activities. The antioxidant properties of MRPs are also reviewed in terms of protection potential against oxidative damage associated with a wide variety of diseases such as diabetes and cancer, and pathological conditions such as aging.

Key Words: Maillard reaction; Non-enzymatic browning reaction; Muta- gens; Toxicity; Heterocyclic compounds; Heterocyclic aromatic amines; Toxicity tests; Model systems; Maillard reaction products Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014

INTRODUCTION

Food is a mixture of numerous chemicals, including proteins, amino acids, carbohydrates, lipids, vitamins, and minerals. It is well known that heat treatment,

*Corresponding author. Fax: (530) 752-3394; E-mail: [email protected]

151

DOI: 10.1081/FRI-120014356 8755-9129 (Print); 1525-6103 (Online) Copyright q 2002 by Marcel Dekker, Inc. www.dekker.com MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

152 LEE AND SHIBAMOTO

such as cooking, promotes complex chemical reactions among food components. The chemical changes in food components caused by high heat treatment have raised questions about the subsequent consequences of reducing nutritive value and formation of some toxic chemicals.[1,2] Among the many reactions occurring in processed foods, the non-enzymatic browning reaction or the Maillard reaction plays the most important role in the formation of various chemicals, including toxic ones.[3] The Maillard reaction refers to the interaction between a free amine group in peptides or proteins and a carbonyl group in sugars or carbohydrates. This reaction proceeds without catalytic enzymes and gives condensation products. Any mixtures containing an amine group (amino acids, proteins, etc.) and a carbonyl group (sugars, carbohydrates, etc.) produce a large number of the so-called Maillard reaction products (MRPs).[4] MRPs include volatile compounds (generally low molecular- weight compounds) of hydrocarbons, alcohols, ketones, aldehydes, esters, ethers, and heterocyclic compounds.[5] Less volatile compounds with medium to high molecular weights—polyphenols, peptide polymers, and complex brown pigments—are also formed.[6] The MRPs are primarily known to provide color and ﬂavor to cooked foods.

TOXICOLOGICAL EFFECTS

The nutritional, physiological, and biological activities of MRPs have received much attention in the last three decades.[4,6 – 10] Among these, many researchers have focused on the toxicity of MRPs because some MRPs were reportedly associated with a variety of diseases such as diabetes and cancer.[11] For example, it was hypothesized that mutagenic and carcinogenic heterocyclic amines have been formed from amino acids and proteins through non-enzymatic browning reactions in various broiled foods.[12,13] b-Carboline derivatives, such as 3,4-dihydro-1-methyl-9H-pyrido[3,4-b]indole-3-carboxylic acid, isolated from a browning reaction mixture of carbohydrate and L-tryptophan were proved to be [12] premutagens. D-Isoglucosamine, a model substance produced by the nonenzymatic browning reaction, reportedly damaged DNA.[13] Also, partially

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 substituted reducing sugars derived from the non-enzymatic browning reaction promoted DNA strand cleavage.[14] The Maillard reaction in vivo was reported to have opposite effects on collagen structure, which related to sequel of aging and diabetes.[15,16] Feeding a reaction mixture from a casein/glucose model system increased urinary loss and levels of zinc and copper in the kidney.[17,18] Also, a decreased level of zinc in urine was found to correlate with the level of MRPs formed in the casein/glucose mixture.

Toxicity Tests

Genotoxicity studies of MRPs in vivo and in vitro are a recent development.[19,20] The mutagenic activities of food components have been studied MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 153

in a number of test systems using bacteria (Salmonella typhimurium, Escherichia coli ), yeast (Saccharomyces cerevisiae ), insects (Drosophila melanogaster ), and mammalian species in vitro (Chinese hamster ovary cells) and in vivo (mouse). Mutagenicity testing systems are performed based on three types of alterations: (1) point mutation, a small submicroscopic change in the DNA affecting one or a few nucleotide bases; (2) chromosomal effects, a gross chromosomal change which can be seen with a light microscopic, i.e., chromosomal breaks, deletions, and translocations; (3) cell transformation in cultured human or other mammalian cells, i.e., tumors produced by implantation in animals.[21,22] All three tests have been used in evaluating the mutagenicity of MRPs. Bacterial systems are useful as a prescreening method to detect mutagens capable of inducing point mutation. The most widely used assay for point mutation employs specially constructed strains of S. typhimurium mutants in which the pathway to histidine biosynthesis is blocked. This assay system, known as the Ames assay, is used to detect reverse point mutations of base-pair substitution or the frame-shift type.[23– 25] In order to mimic the mammalian metabolism of certain compounds, a fraction of the homogenized liver tissue (S9) is incorporated into the test system. The S9 fraction increases the rate of enzymatic metabolism toward the agent. Bacterial tester strains—S. typhimurium TA153, TA1537, and TA 1538—have been developed for base-pair mutations. Other more sensitive strains—S. typhimurium TA98 and TA100—are less speciﬁc to the type of mutation caused.[26,27] For example, ether extracts of nitrite-treated casein– glucose mixtures gave mutagenic responses to strains TA 100, TA 102, and TA [28] 104 with S9 mixture. Addition of nitrite to MRPs from carbohydrates and L- tryptophan resulted in increased mutagenicity in S. typhimurium TA 100.[29] Soy sauce heated with nitrite exhibited mutagenic activities against S. typhimurium TA 100 and TA 98, with or without S9.[30 – 32]

The Maillard Reaction in Model Systems

The study of the Maillard reaction involves many complex substances

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 including reactants, intermediates, and products, some of which have an unstable nature. To investigate the nature of the Maillard reaction, a simple model system (the so-called browning reaction model system) consisting of an amine compound, such as amino acid or protein, and a carbonyl compound, such as a sugar or carbohydrate, has been widely used. During the last two decades, many model systems have been employed to study the Maillard reactions occurring in complex food systems. The model systems consisting of a sugar and an amino acid were most commonly used to study the nature of cooked foods. Typical model systems used for investigating browning reactions are as follows:[3,33] . Reactions involving reducing sugars and amino acids. . Thermal degradation of the Amadori rearrangement products. Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit 154

Table 1. Mutagenicity of Reactants and Products of Maillard Reaction

Compounds Chemical Structure Model System or Resources Mutagenicity Test References M

Dicarbonyl compounds ARCEL Glyoxal Coffee, baked cereals þ to TA 98 and TA 100 with and without S9 mix [43–45]

Methylglyoxal Coffee þ to TA 100 with and without S9 mix [42,44,45] EKKER,

Diacethyl Coffee, baked cereals þ to TA 98 and TA 100 with and without S9 mix [41,45] I NC . •270 Maltol Coffee þ to TA 98 and TA 100 with and without S9 mix [41] M ADISON

Furans

Furfural Lysine–arabinose, Arginine– þ to TA 100 and TA 104 with S9 mix [50,51,54] A

arabinose Lysine–xylose VENUE • Arginine–xylose N

5-Hydroxymethyl-2- Lysine–arabinose, Arginine– þ to TA 100 with S9 mix, DNA breaking [52,53] SHIBAMOTO AND LEE furfural arabinose Lysine–xylose activity in the presence of Cu2þ EW

Arginine–xylose Y ORK ,

5-Methylfurfural Lysine–arabinose, Arginine– þ to TA 100 with S9 mix [51] NY arabinose Lysine–xylose 10016 Arginine–xylose ten permissionofMarcelDekker,Inc.

5-Sulfooxylmethyl- Metabolite of 5-hydroxymethyl þ to TA 100 [56] furfural furfural Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit OIOOYO ALADRATO RDCS155 PRODUCTS REACTION MAILLARD OF TOXICOLOGY

2,5-Dimethyl-4- Soy sause þ to TA 100 with and without S9 mix, induced [57,58] hydroxy -3(2H)- micronucleated mouse peripheral reticulocytes furanone

4-Hydroxy-2-ethyl- Soy sause þ to TA 100 without S9 mix, induced micro- [58] M

5-methyl -3(2H)- nucleated mouse peripheral reticulocytes ARCEL furanone

Pyrroles D EKKER,

1-Nitro-2-acethyl- Reactant from 2-acethylpyrrole þ to TA 100 and TA 98 without S9 mix [63]

pyrrole and nitrite and cytotoxic to mouse cell I NC . •270

1,3,5-Trinitro-2- Reactant from 2-acethylpyrrole þ to TA 100 and TA 98 without S9 mix [63] acethylpyrrole and nitrite and cytotoxic to mouse cell M ADISON

Dithianes

1,3-Dithiane Broiled beef extracts þ to TA 100 and TA 98 with and without S9 mix [65] A VENUE •

1,4-Dithiane Broiled beef extracts þ to TA 100 and TA 98 without S9 mix [65] N EW

Y ORK Thiazolidines , NY

Thiazolidine Cysteamine–glucose þ to TA 100 and TA 98 with and without S9 mix [61,66] 10016 ten permissionofMarcelDekker,Inc.

(continued) Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit 156 Table 1. Continued

Compounds Chemical Structure Model System or Resources Mutagenicity Test References

þ M 2-Methylthiazolidine Cysteamine–glucose to TA 100 and TA 98 with and without S9 mix [61,66] ARCEL

D EKKER, 2-Ethylthiazolidine Cysteamine–glucose þ to TA 100 and TA 98 with and without S9 mix [61,66]

I NC . •270 2-Propylthiazolidine Cysteamine–glucose þ to TA 100 and TA 98 with and without S9 mix [61,66] M ADISON

2-(1,2,3,4,5-Penta- Cysteamine–glucose þ to TA 100 and TA 98 with and without S9 mix [61,66]

hydroxy)-pentylthia- A

zolidineyl VENUE •

N-nitrosothiazolidine Cysteamine–glucose–nitrite þ to TA 100 without S9 mix [66,67,69] N E N SHIBAMOTO AND LEE EW

Y ORK , NY

N-Nitroso-2- Cysteamine–glucose–nitrite þ to TA 100 without S9 mix [66,67,69] 10016 alkylthiazolidine ten permissionofMarcelDekker,Inc. Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit OIOOYO ALADRATO RDCS157 PRODUCTS REACTION MAILLARD OF TOXICOLOGY

Imidazoles

2-Amino-1-methyl- Creatine with L-threonine þ to TA 98, TA 100 and TA 1538 with S9 mix [77,78] 5-propylideneimi-

dazol-4-one M ARCEL

2-Amino-5-ethy- Creatine with L-threonine þ to TA 98, TA 100 and TA 1538 with S9 mix [77,78] D

lidene-1-methyl- EKKER, imidazol-4-one

I NC

Pyrazines . •270 2-Methylpyrazine Rhamnose-ammonia S. cerevisiae strain D5, and induced chromo- [80] somal aberrations in Chinese hamster ovary cells M ADISON

2-Ethylpyrazine Rhamnose–ammonia S. cerevisiae strain D5, and induced chromo- [80]

somal aberrations in Chinese hamster VENUE • ovary cells

2,5-Dimethyl- Rhamnose–ammonia S. cerevisiae strain D5, and induced chromo- [80] N EW pyrazine somal aberrations in Chinese hamster

ovary cells ORK , NY 2,6-Dimethyl- Rhamnose–ammonia Saccharomyces cerevisiae strain D5, and induced [80] pyrazine chromosomal aberrations in Chinese hamster 10016 ovary cells ten permissionofMarcelDekker,Inc. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

158 LEE AND SHIBAMOTO

. Thermal degradation of sugars. . Pyrolysis of a-amino acids and dipeptides. . Pyrolysis of vitamins and related substances. . Reaction of b-dicarbonyl compounds and aldehydes in the presence of either a-amino acids or ammonia and/or hydrogen sulﬁde. . Other miscellaneous model systems. The above model systems have led to the discovery of many compounds formed in non-enzymatic browning reactions.

Mutagenicity of Maillard Reaction Products

Although the occurrence of the Maillard reaction was recognized at the beginning of the 20th century,[34] MRPs were not thoroughly investigated until the 1960s. This may be due to a lack of advanced analytical techniques for MRPs. With the development of analytical instruments such as gas chromatography (GC), high- performance liquid chromatography (HPLC), and mass spectrometry (MS), the analysis of MRPs advanced significantly.[5,22,35,36] Consequently, many volatile compounds have been identified in the MRPs over the past three decades. Among them, heterocyclic compounds, such as furans, thiophenes, thiazoles, pyridines, pyrazines, pyrroles, oxazoles, and quioxaline have received much attention as cooked flavors.[36,37] Mutagenicity of reactants of the MRPs was given in Table 1. On the other hand, many toxic heterocyclic compounds, such as 2-amino-3- methylimidazo[4,5-f]quinoline (IQ), have been isolated and identified.[27,38 – 40] In the 1980s, the mutagenicity of these heterocyclic compounds was reported.[27]

Dicarbonyl Compounds

The degradation of sugar accelerated with amine compounds produces many toxic carbonyl compounds, including formaldehyde, acetaldehyde, acrolein, glyoxal, methyl glyoxal, malonaldehyde, and maltol. Reactive dicarbonyl Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 compounds generated during the Maillard reaction play an important role in the formation of heterocyclic aroma compounds.[35] Maltol and glyoxal have been reported in coffee, soybeans, baked cereals, and bread crust.[41 – 43] The a- dicarbonyl compounds, such as glyoxal, methyl glyoxal, diacetyl, and 2,3- pentanedione, were mutagenic in S. typhimurium TA98 and TA100 with and without S9 activation.[42,44,45] The degree of mutagenicity of a-dicarbonyl compounds in S. typhimurium TA100 without S9 activation was as follows: glyoxal . methyl glyoxal . phenyl glyoxal @ 1,2-cyclohexanedione @ diacetyl . 3,4-cyclohexanedione.[46] Recently, it has been reported that the mutagenic activities of a-dicarbonyl compounds were related to their chemical reactivities with puric bases.[47] Malonaldehyde, a typical b-dicarbonyl compound, is also mutagenic toward Salmonella strains.[48 – 50] Malonaldehyde is well known as a MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 159

major lipid peroxidation product. Lipids have also been used as one of the carbonyl reactants in browning reaction model systems.[6]

Furans

Furans, the most abundant products of the Maillard reaction, account for the caramel-like odor of heated carbohydrates.[35] Furan derivatives, in particular furfural,[51] were the first heterocyclic compounds found in the Maillard reaction. They were reported in products arising from sugar–amino acid interactions, degradation of Amadori intermediates, thermal degradation of sugars, and reactions of sugars with hydrogen sulfide and/or ammonia. Furans were demonstrated to be weakly mutagenic in the Salmonella/microsomal assay in comparison to other strong mutagens such as benzo[a]pyren.[52] Furfural, 5-hydroxymethylfurfural (5-HMF), and 5-methylfurfural were identified in a reaction mixture of sugars with amino acids (lysine–arabinose, arginine–arabinose, lysine–xylose, and arginine– xylose). These furan compounds were mutagenic in S. typhimurium TA100 with S9 activation, and possessed DNA breaking activities with the presence of Cu2þ.[51,53,54] In other research, furfural gave equivocal responses in S. typhimurium TA100 and TA104.[55] Although 5-HMF—an intermediate product in the Maillard reaction—has been reported to possess cytotoxic, genotoxic, and tumorigenic activities, it did not pose any health risks under conditions of use as a flavor ingredient.[56] 5-Sulfooxymethylfurfural, a possible metabolite of 5-HMF, induced dose-dependent activity in S. typhimurium TA100.[57] DNA breaking activity and superoxide formation by 2,5-dimethyl-4-hydroxy-3(2H)-furanone (2,5-DMHF) and 4-hydroxy-2-ethyl-5-methyl-3(2H)-furanone were investigated using DNA fragments obtained from P53 tumor suppressor genes.[54,58,59]

Pyrroles

Many pyrrole derivatives are also formed via the Maillard reaction during Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 processing. Some of these pyrrole derivatives have pleasant ﬂavors. For example, pyrrole-2-carboxaldehyde gives a sweet and cornlike odor, while 2-acetylpyrrole has a caramel-like ﬂavor.[60 – 62] The carbonyl pyrroles such as 2-acetylpyrrole, pyrrole-2-carboxaldehyde, and pyrrole-2-carboxylic acid were not mutagenic toward S. typhimurium TA97, TA98, TA102, and TA104, with and without S9 activation.[62,63] However, N-nitropyrrole compounds produced from 2-acethylpyrrole with nitrite (1-nitro-2-acetyl-pyrrole and 1,3,5-trinitro-2-acetylpyrrole) have demonstrated moderate mutagenicity toward the Salmonella strains TA98 and TA100 in the absence of a mammalian activation system, and exhibited cytotoxicity toward mouse cells.[63] These results suggest that the formation of direct-acting mutagens from nitro-derivatives may take place in nitrite-containing foods or in vivo by nitrosation following ingestion of 2-acethylpyrrole. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

160 LEE AND SHIBAMOTO

Dithianes

Dithianes such as 1,3-dithiane and 1,4-dithiane are the sulfur-containing MRPs. They have been identiﬁed in broiled beef extracts.[64] 1,3-Dithiane exhibited strong mutagenic activity in S. typhimurium strains TA98 and TA100 with and without S9 activation, whereas 1,4-dithiane demonstrated weak mutagenicity, similar to that of 2,6-dimethylpyrazine.[65]

Thiazoles, Thiazolidines, and N-Nitrosothiazolidines

Thiazoles have received much attention as meat ﬂavor ingredients following their isolation from cooked meat. Thiazoles are formed in many Maillard model systems consisting of a sugar and sulfur-containing amino acids such as cysteine. Cysteine produces many thiazoles from a reaction with monosaccharides, but the formation of thiazolidine has not been observed. However, cysteamine, a decarboxylated derivative of cysteine, produced a series of alkylthiazolidines.[66] Several thiazolidines isolated from a heated cysteamine–glucose model system were examined for their mutagenic activities in test strains of S. typhimurium TA98 and TA100, with and without S9 activation. Unsubstituted, 2- methyl-, 2-ethyl-, and 2-propylthiazolidine found in a dichloromethane extract showed weak mutagenic activity. On the other hand, 2-(1,2,3,4,5-pentahydroxy)- pentylthiazolidine found in an aqueous fraction exhibited some mutagenic activity.[62,66] Since the discovery of carcinogenic N-nitrosamines, much research has been done to isolate and identify N-nitrosoamines in Maillard reaction systems. However, to date, there has been no report of ﬁnding N-nitrosoamines in Maillard reactions. Some heterocyclic compounds that contain a possible source of nitrosatable nitrogen, such as thiazolidines, can react with nitrite-rich foods to form mutagenic or carcinogenic N-nitroso-2-alkylthiazolidine.[67,68] N-nitroso-2- methylthiazolidine and N-nitroso-2-ethylthiazolidine obtained from a cysteamine–glucose–nitrite model system had much stronger mutagenic activities than Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 their thiazolidine counterparts in S. typhimurium strains TA100 without S9 activation.[66,67,69] The mutagenic activity of N-nitrosothiazolidines decreases in the following order: unsubstituted . isopropyl . propyl . ethyl . butyl . isobutyl . methyl:

Imidazoles

Although imidazoles comprise the second largest fraction of the volatile MRPs after pyrazines, they have not received much attention from flavor chemists due to their lack of flavors.[35] They are sometimes recognized in cooked foods as compounds responsible for off-flavors. Several imidazoles are formed in the MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 161

[70] Maillard reaction from L-rhamnose and ammonia. According to the research on mutagenic activities of imidazoles and nitroimidazoles, two out of the 18 imidazoles examined were mutagenic, whereas 31 of the 33 nitroimidazoles showed mutagenic activities.[71 – 75] On the other hand, some alkylimidazoles (1- methyl-, 1,2-dimethyl-, 2-methyl-, 1-ethyl-, 4-methyl-, and 4,5-dimethylimida- zole) exhibited no mutagenic activities toward S. typhimurium strains TA100.[76] A new class of low molecular weight amino-methylimidazole-4-one mutagens was identiﬁed from the reaction of creatinine with the amino acid L- threonine.[77,78] This class of mutagens is important as IQ (amino-methylimida- zolequinolins)-like mutagens. The IQ have been known as powerful genotoxic carcinogens. Two IQ-like mutagens, 2-amino-1-methyl-5-propylideneimidazol-4- one and 2-amino-5-ethylidene-1-methylimidazol-4-one, were isolated and were positive in all IQ-sensitive Ames tester strains.[79]

Pyrazines

Pyrazines are the most abundant heterocyclic compounds found in MRPs, and are well characterized as the compounds contributing roasted or toasted ﬂavors to cooked foods. In addition, pyrazine formation in the Maillard reaction is related to color formation: as pyrazine products increase, the color of the mixture changes from colorless to yellow.[70] There has been no report that pyrazines have reasonable mutagenic activity toward S. typhimurium strains with or without microsomal activation. However, four alkylpyrazines—2-methyl-, 2-ethyl-, 2,5- dimethyl-, and 2,6-dimethylpyrazine—did respond to S. cerevisiae strain D5 and induced an increase in frequency of chromosomal aberrations in Chinese hamster ovary cells.[80] Antimutagenic activities of the pyrazines in extracts from amino acid/sugar model systems toward mutagenicity of 2-amino-3-methylimidazo[4,5- f]quinoline (IQ) were investigated using S. typhimurium TA98 with S9 activation.[81] This work suggests an important role of pyrazines in the dual functionality of mutagenicity and antimutagenicity. Recently, the pyrazine cation radical generated in the initial stage of the Maillard reaction of glucose and amino Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 acid was studied as a key intermediate of imidazoquinoxaline-type mutagens.[82] Because of this, the pyrazine cation radical can be used as a marker to test for antimutagenic activities of chemicals toward the formation of imidazoquinoxaline-type mutagens.

Polycyclic Aromatic Amines

In the 1970s, the presence of potent mutagens other than polycyclic aromatic hydrocarbons (PAHs) was recognized ﬁrst in charred foods. Later, heterocyclic amines were isolated and identiﬁed in the heated proteins as a cause of mutagenicity in charred foods, in addition to PAHs.[27,38 – 40,83] Heterocyclic MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

162 LEE AND SHIBAMOTO

amines are small molecules formed when components of food proteins and creatine/creatinine (compounds in muscle) are exposed to high temperature. Heterocyclic amines are divided into two groups.[38,84 – 86] Heating a mixture of creatine/creatinine, amino acids, and sugars produces one, the 2-amino-3- methylimidazo quinoline (IQ)-type. The IQ- type possesses an imidazole ring, i.e., that is formed from creatinine. The IQ-type heterocyclic amines are 2-amino-3- methylimidazo[4,5-f]quinoline (IQ), 2-amino-3,4-dimethylimidazo[4,5-f]quinoline (MeIQ), 2-amino-3-methylimidazo[4,5-f]quinoxaline (IQx), 2-amino-3,8- dimethylimidazo[4,5-f]quinoxaline (MeIQx), 2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline (4,8-DiMeIQx), and 2-amino-1- methyl-6-phenylimidazo[4,5-b]pyridine (PhIP). The formation of the IQ-type heterocyclic amines is a result of heat treatment that causes cyclization of creatinine to form the imidazole moiety. The remaining moieties of the structure arise from pyridines and pyrazines formed through the Strecker degradation of amino acids and breakdown products of b-dicarbonyl products formed by the Maillard reaction. The other heterocyclic amines, the so-called nonIQ type, are pyrolysis products formed from tryptophan: 3-amino-1,4-dimethyl-5H-pyrido[4,5-b ]indole (Trp-P-1), 3-amino-1-methyl-5H-pyrido[4,5-b ]indole (Trp-P-2); from glutamic acid: 2-amino-6-methyldipyrido[1,2-a:30,20-d ]imidazole (Glu-P-1); 2-aminodipyrido[1,2-a:30,20-d ]imidazole (Glu-P-2); and from soya bean globulin: 2-amino- 9H-pyrido[2,3-b ]indole (AaC), 2-amino-3-methyl-9H-pyrido[2,3-b ]indole (MeAaC). Mutagenicities of heterocyclic aromatic amines and those of other representative mutagens/carcinogens in S. typhimurium strains are given in Table 2. Some of these heterocyclic aromatic amines, such as IQ and MeIQ, exhibit stronger mutagenic activities than well-known carcinogens like aflatoxin B1 and benzo[a]pyrene. The most important factors in the formation of these mutagenic heterocyclic aromatic amines in foods are cooking temperature and time. For instance, the mutagenicity of crust, pan residue, and smoke from pan-broiled pork patties increased five-fold when the pan temperature was increased from 200 to 3008C.[87] In the past three decades, researchers have confirmed that isolated heterocyclic aromatic amines cause cancer in various tissues of animal models when fed at high

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 levels. The carcinogenic potency of heterocyclic aromatic amines in mice and rats is given in Table 3. The major target organs are the small and large intestines, liver, lung, and blood vessels.

Food Systems Associated with the Maillard Reaction

Identification and quantification of mutagens in processed foods have processed slowly due to many problems. For instance, time-consuming analytical procedures, requirements for large sample amounts, and problems of quantification make the studies difficult.[88] Development of advanced instrumentation such as gas chromatography/mass spectrometry (GC/MS) has spurred the characterization of MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 163

Table 2. Mutagenicity of Heterocyclic Amines and Typical Carcino- gens in S. typhimurium TA98 and TA100 with S9 Microsomal Activation (Adapted from Refs. [22,37])

Number of Revertants

Mutagens TA98 TA100 IQ 433,000 7,000 MeIQ 661,000 30,000 IQx 75,000 1,500 MeIQx 145,000 14,000 4,8-DiMeIQx 183,000–206,000 8,000 7,8-DiMeIQx 163,000–189,000 9,900 PhIP 1,800–2,000 120 Aﬂatoxin B1 6,000 28,000 Benzo[a]pyrene 320 660

IQ: 2-amino-3-methylimidazo[4,5-f ]quinoline; MeIQ: 2-amino-3,4- dimethylimidazo[4,5-f ]quinoline; IQx: 2-amino-3-methylimidazo[4,5- f ]quinoxaline; MeIQx: 2-amino-3,8-dimethylimidazo[4,5-f ]quinoxaline; 4,8-DiMeIQx: 2-amino-3,4,8-trimethylimidazo[4,5-f ]quinoxaline; 7,8- DiMeIQx: 2-amino-3,7,8-trimethylimidazo[4,5-f ]quinoxaline; PhIP: 2- amino-1-methy-6-phenylimidazo[4,5-b ]pyridine ( ).

mutagens, such as heterocyclic aromatic amines, found in food samples.[84] Isolation and identification of mutagens in food systems began in the late 1970s. The high levels of mutagenicity found in the charred parts of beefsteak and grilled fish was first observed in 1977.[89] Heated beef or beef extract was also found to be mutagenic in the same era. It was reported that foods with low water and high protein contents yielded higher mutagenicity. Identification of heterocyclic aromatic amines, potent mutagens, IQ, MeIQ, and MeIQx in beef extract was conducted using several analytical methods, such as electrochemical, photodiode UV array, and on-line scintillation in conjunction with HPLC.[90–93] Since the late 1980s, GC/MS and LC/MS have become effective for identification and quantification of mutagens in processed foods.[94–96] Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014

ANTIOXIDANT ACTIVITIES

It is well known that the Maillard reaction inﬂuences the oxidative stability of food products.[97] Lipid peroxidation is the primary mechanism by which food deteriorates upon storage in the presence of oxygen. This process of oxidation can be initiated by enzyme catalysts, metal ion catalysis, or photochemical processes. Free radicals such as peroxyl, alkoxyl, and hydroxyl have been implicated in the mechanism of lipid peroxidation. The changes in the quality of processed foods are manifested by deterioration of ﬂavor, aroma, color, texture, nutritive value, and the formation of toxic compounds such as aldehydes and epoxides.[98] These Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit 164 Table 3. Carcinogenic Effects of Heterocyclic Amines (Adapted from Refs. [22,26,37,87])

Chemicals Structure Species Dose (%) in Diet Target Organs TD50a (mg/kg/day)

Trp-P-1 Rats 0.015 Liver 0.1 M ARCEL

D EKKER, Mice 0.02 Liver 8.8 Trp-P-2 Rats 0.01 Liver —

Mice 0.02 Liver 2.7 NC . •270 M

Glu-P-1 Rats 0.05 Liver, small and large intestines 0.8 ADISON Mice 0.05 Liver, blood vessels 2.7

A VENUE •

Glu-P-2 Rats 0.05 Liver, small and large intestines 5.7 Mice 0.05 Liver, blood vessels 4.9 N E N SHIBAMOTO AND LEE EW

Y ORK AaC Mice 0.08 Liver, blood Vessels 15.8 , NY 10016 ten permissionofMarcelDekker,Inc. MeAaC Rats 0.02, 0.01 Liver 6.4 Mice 0.08 Liver, blood vessels 5.8 Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit OIOOYO ALADRATO RDCS165 PRODUCTS REACTION MAILLARD OF TOXICOLOGY

IQ Rats 0.03 Liver, small and large intestines, skin 0.7 Mice 0.03 Liver, lung 14.7 M ARCEL

MeIQ Rats 0.03 Large intestines, mammary gland 0.1 EKKER, Mice 0.04, 0.01 Liver, forestomach 8.4

I NC . •270 M MeIQx Rats 0.04 Liver, skin 0.7 ADISON Mice 0.06 Liver, lung 11.0

A VENUE •

PhIP Rats 0.04 Large intestines, mammary gland, prostate 2.2 N

Mice 0.04 Lymphoid tissue 64.6 EW

Y ORK

0 0 Trp-P-1: 3-amino-1,4-dimethyl-5H-pyrido[4,5-b ]indole; Trp-P-2: 3-amino-1-methyl-5H-pyrido[4,5-b ]indole; Glu-P-1: 2-amino-6-methyldipyrido[1,2-a:3 ,2 - , NY d ]imidazole; Glu-P-2: 2-aminodipyrido[1,2-a:30,20-d ]imidazole; Aa C: 2-amino-9H-pyrido[2,3-b ]indole; MeAaC: 2-amino-3-methyl-9H-pyrido[2,3- b ]indole; IQ: 2-amino-3-methylimidazo[4,5-f ]quinoline; MeIQ: 2-amino-3,4-dimethylimidazo[4,5-f ]quinoline; MeIQx: 2-amino-3,8-dimethylimidazo[4,5- 10016 f ]quinoxaline; PhIP: 2-amino-1-methy-6-phenylimidazo[4,5-b ]pyridine ( ). ten permissionofMarcelDekker,Inc. a TD50 is the dosage (mg/kg body weight/day) causing toxic effect in 50% of the exposed animals. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

166 LEE AND SHIBAMOTO

changes in food are significant to both food stability and food safety. Traditionally, lipid peroxidation has been prevented by the use of synthetic antioxidants such as BHA and BHT. Recently, however, the safety of these compounds with respect to human health has been questioned and, therefore, the discovery of safer natural alternatives is in order.[99] Most studies on antioxidative activities of MRPs have been conducted using a model system consisting of MRPs and lipids. The antioxidant activity of nonenzymatic browning reaction products is given in Table 4. The antioxidative activity of the MRPs was first observed in early 1950s.[100] However, the exact nature of the antioxidants formed is not yet well known.[101] Higher molecular weight substances, such as melanoidins, are thought to be the major antioxidative products formed by the non-enzymatic browning reactions. For example, the addition of sugar and/or amino acids to baked foods, such as cookies, enhances the browning reaction and subsequently increases the stability against oxidative rancidity.[102] Heat treatment of various milk products has also been reported to result in improved oxidative stability.[103] In the case of heat-treated milk products, the formation of antioxidative sulfyhydryl groups from the milk protein has been considered to play a role in the products’ stability. The oxidative stability of various cereals such as wheat, corn, and oats has also been improved by heat treatment.[6,101] In the case of coffee, roasted coffee brews were found to have higher antioxidative activity than unroasted coffee brews, which contained higher concentrations of known polyphenolic antioxidants.[104] This suggests that compounds, in addition to polyphenols, are responsible for the antioxidative activity of the roasted coffee brews. Although comprehensive investigations have been performed on the antioxidative properties of MRPs, the problem of expedient and profitable utilization of MRPs as effective antioxidants in various food products remains unsolved. This may be due to the lack of knowledge of the specific structures of MRPs with antioxidative activity. The most commonly recognized MRP with antioxidative activity are the high molecular weight melanoidines. It has been reported that the melanoidin prepared from D-xylose and glycine showed an antioxidative activity comparable to BHA and BHT.[105] Furthermore, melanoidin exhibited a synergistic effect with tocopherol, BHA, or

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 BHT in inhibiting the autoxidation of linoleic acid. Identification of the Maillard reaction antioxidants has focused primarily on the higher molecular weight melanoidines. The Maillard reaction, however, also produces hundreds of volatile compounds that are responsible for the flavors of cooked foods. Recently, volatile compounds obtained from a glucose/cysteine browning model system were reported to possess certain antioxidative activities.[106] Also, column chromatographic fractions prepared from a dichloromethane extract of a glucose/cysteine browning model system demonstrated the ability to inhibit the oxidative transformation of hexanal to hexanoic acid.[107] Additionally, several nitrogen- and/or sulfur-containing heterocyclic compounds, which are the major flavor compounds formed by the Maillard reaction, exhibited antioxidative activity in two separate testing systems.[108] For Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 ©2002 MarcelDekker,Inc.Allrightsreserved.Thismaterialmaynotbeusedorreproducedinanyformwithouttheexpresswrit OIOOYO ALADRATO RDCS167 PRODUCTS REACTION MAILLARD OF TOXICOLOGY M ARCEL

Table 4. The Antioxidant Activity of Non-enzymatic Browning Reaction Products D EKKER, Description of Non-enzymatic Browning Reaction Products Tested for Antioxidant Activity Antioxidant Activity Assessment References

I Pregelatinized starch, glucose, and lysine Peroxide scavenging test based on crocin bleaching [6] NC Cysteine and glucose Aldehyde/carboxylic acid conversion assay [10,107] . •270 Glucose or lactose with lysine, alanine, or glycine 2,2-Diphenyl-1-picrylhydrazyl radical assay (DPPH) [97]

Heated tomato derivatives and roasted coffee Chain-breaking capacity in lipid peroxidation [98] M Heterocyclic compounds formed in Maillard reactions Aldehyde/carboxylic acid conversion assay and lipid/malonaldehyde assay [104,106,108] ADISON Corn oil and glycine Aldehyde/carboxylic acid conversion assay [109] Glucose–lysine and fructose–lysine Deoxyribose oxidative degradation assay [110]

Fructose and arginine Thiobarbituric acid assay (TBA) [111] VENUE • Glucose or fructose and glutamic acid Chain breaking and oxygen scavenging activity [112] Heated pasta Chain breaking and oxygen scavenging activity [113] N EW

Y ORK , NY 10016 ten permissionofMarcelDekker,Inc. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

168 LEE AND SHIBAMOTO

example, alkylthiophenes, 2-thiophenethiol, 2-methyl-3-furanthiol, furfuryl mercaptan, 2-thiothiazoline, and imidazole, which are all found in coffee, inhibited hexanal oxidation for up to 30 days, and also exhibited antioxidative activities measured in lipid peroxidation systems and in the thyrosyl radical scavenging assay. 1-Methylpyrrole and several of its 2-alkyl homologues inhibited pentanal and hexanal oxidation.[109] 2-Methylfuran exhibited the greatest activity among volatile furans.[108] Similarly, 2-methylthiophene exhibited the greatest activity among the volatile thiophenes. In general, thiazoles were ineffective. 2-Acetylpyrrole, 2- methylfuran, and 2-methylthiophene exhibited one-tenth the antioxidative activity of the synthetic antioxidant BHT. The cells of the human body are subjected to oxidative damage, which is associated with certain biological complications such as aging and carcinogenesis. Although living systems are protected from active oxidants by enzymatic systems, additional antioxidants such as ascorbic acid and a-tocopherol in fruits and vegetables are required to protect living cells from oxidation. Dietary antioxidants, including vitamins C and E, as well as carotene, may protect against cancer, heart disease, and cataracts. Moreover, it is obvious that humans consume some quantity of antioxidants formed in heat-treated foods. These antioxidants may play a role in protection against oxidative damage associated with the diseases described above.

CONCLUSION

The Maillard reaction is responsible for the formation of mutagenic chemicals formed in several degradation pathways or by the Maillard reaction intermediates combining with food matrix. The diversity of the mutagens ranges form simple dicarbonyl compounds and heterocyclic volatiles to heterocyclic aromatic amines. The advanced analytical developments and genotoxic testing techniques used to identify and quantify such compounds in foods and to assess their mutagenicity have provided information to establish relationships between dietary intake and carcinogenesis. In contrast to toxic effects of the MRPs, they also have antioxidant properties. Research in these areas is improving and Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 continues to be assessed.

REFERENCES

1. Manzocco, L.; Calligaris, S.; Mastrocola, D.; Nicoli, M.C.; Lerici, C.R. Review of Non-enzymatic Browning and Antioxidant Capacity in Processed Foods. Trends Food Sci. Technol. 2000, 11, 340–346. 2. Chang, W.C.; Chen, C.C.; Tseng, T.H.; Huang, H.P.; Hsu, J.D.; Wang, C.J. Tumor Promotion of N-Nitroso-N-(3-Keto-1,2-butanediol)-30-Nitrotyramine Derived from Nitrosation of Maillard Reaction Product in CD-1 Mice. Toxicol. Appl. Pharmacol. 2000, 166, 51–58. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 169

3. Shibamoto, T. Genotoxicity Testing of Maillard Reaction Products. In The Maillard Reaction in Aging, Diabetes, and Nutrition: Proceedings of an NIH Conference on the Maillard Reaction in Aging, Diabetes, and Nutrition, Held in Bethesda, Maryland, September 22-23, 1988; Baynes, J.W., Monnier, V.M., Eds.; A.R. Liss: New York, 1989; Vol. xxvii, 359–376. 4. Martins, S.; Jongen, W.M.F.; van Boekel, M. A Review of Maillard Reaction in Food and Implications to Kinetic Modelling. Trends Food Sci. Technol. 2000, 11, 364–373. 5. Shibamoto, T.; Eiserich, J.; Dinnauer, M.; Jennings, W. Maillard Reaction Products From Microwave Heating of Model Systems Exploiting Gc Variables. Abstr. Papers Am. Chem. Soc. 1992, 204, AFGD 41. 6. Mastrocola, D.; Munari, M.; Cioroi, M.; Lerici, C.R. Interaction Between Maillard Reaction Products and Lipid Oxidation in Starch-Based Model Systems. J. Sci. Food Agric. 2000, 80, 684–690. 7. Sensidoni, A.; Munari, M.; Lerici, C.R. Early Indicators of Chemical Changes in Foods Due to Enzymic or Nonenzymic Browning Reactions. 3. Study on Heat Treated Model Systems by Hplc Separation of Some Maillard Reaction Products. Food Science and Technology-Lebensmittel-Wissenschaft & Technologie 1991, 24, 218–225. 8. Marko, D.; Kemeny, M.; Bernady, E.; Habermeyer, M.; Weyand, U.; Meiers, S.; Frank, O.; Hofmann, T. Studies on the Inhibition of Tumor Cell Growth and Microtubule Assembly by 3-Hydroxy-4-[(E )-(2-Furyl)Methylidene]Methyl-3- Cyclopentene-1,2-dione, An Intensively Coloured Maillard Reaction Product. Food Chem. Toxicol. 2002, 40, 9–18. 9. Umano, K.; Hagi, Y.; Nakahara, K.; Shyoji, A.; Shibamoto, T. Volatile Chemicals Formed in the Headspace of a Heated D-Glucose/L-Cysteine Maillard Model System. J. Agric. Food Chem. 1995, 43, 2212–2218. 10. Eiserich, J.P.; Macku, C.; Shibamoto, T. Volatile Antioxidants Formed From an L- Cysteine D-Glucose Maillard Model System. J. Agric. Food Chem. 1992, 40, 1982–1988. 11. Sugimura, T. Overview of Carcinogenic Heterocyclic Amines. Mutat. Res.— Fundam. Mol. Mech. Mutagen. 1997, 376, 211–219. 12. Johansson, M.A.E.; Fay, L.B.; Gross, G.A.; Olsson, K.; Jagerstad, M. Inﬂuence of Amino Acids on the Formation of Mutagenic/Carcinogenic Heterocyclic Amines in

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 a Model System. Carcinogenesis 1995, 16, 2553–2560. 13. Kashige, N.; Yamaguchi, T.; Ohtakara, A.; Mitsutomi, M.; Brimacombe, J.S.; Miake, F.; Watanabe, K. Structure–Activity Relationships in the Induction of Single-Strand Breakage in Plasmid Pbr322 DNA by Amino Sugars and Derivatives. Carbohydr. Res. 1994, 257, 285–291. 14. Shimamura, T.; Takamori, A.; Ukeda, H.; Nagata, S.; Sawamura, M. Relationship Between the Reduction of Tetrazolium Salt XTT and DNA Strand Breakage with Aminosugars. J. Agric. Food Chem. 2000, 48, 1204–1209. 15. Chellan, P.; Nagaraj, R.H. Protein Crosslinking by the Maillard Reaction: Dicarbonyl-Derived Imidazolium Crosslinks in Aging and Diabetes. Arch. Biochem. Biophys. 1999, 368, 98–104. 16. Ulrich, P.; Cerami, A. Protein Glycation, Diabetes, and Aging. Recent Prog. Horm. Res. 2001, 56, 1–21. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

170 LEE AND SHIBAMOTO

17. Furniss, D.E.; Vuichoud, J.; Finot, P.A.; Hurrell, R.F. The Effect of Maillard Reaction Products on Zinc Metabolism in the Rat. Br. J. Nutr. 1989, 62, 739–749. 18. Navarro, P.; Aspe, T.; Seiquer, I. Zinc Transport in Caco-2 Cells and Zinc Balance in Rats: Influence of the Heat Treatment of a Casein–Glucose–Fructose Mixture. J. Agric. Food Chem. 2000, 48, 3589–3596. 19. de Maagd, P.G.J.; Tonkes, M. Selection of Genotoxicity Tests for Risk Assessment of Effluents. Environ. Toxicol. 2000, 15, 81–90. 20. Evans, H.J. Tests for Genotoxicity—Principles and Findings in Relation to Omeprazole. Digestion 1990, 47, 45–48. 21. Anderson, D. Genetic Toxicology. In Experimental Toxicology: The Basic Issues; Anderson, D., Conning, D.M., Eds.; Royal Society of Chemistry: Cambridge, 1993; 242–286. 22. Wong, J.W.; Shibamoto, T. Genotoxicity of Maillard Reaction Products. In The Maillard Reaction: Consequences for the Chemical and Life Sciences; Ikan, R., Ed.; Wiley: New York, 1996; 129–159. 23. Ames, B.N.; McCann, J.; Yamasaki, E. Methods for Detecting Carcinogens and Mutagens with the Salmonella/Mammalian-Microsome Mutagenicity Test. Mutat. Res. 1975, 31, 347–364. 24. Gee, P.; Maron, D.M.; Ames, B.N. Detection and Classification of Mutagens: A Set of Base-Specific Salmonella Tester Strains. Proc. Natl Acad. Sci. USA 1994, 91, 11606–11610. 25. Maron, D.M.; Ames, B.N. Revised Methods for the Salmonella Mutagenicity Test. Mutat. Res. 1983, 113, 173–215. 26. Sugimura, T. Nutrition and Dietary Carcinogens. Carcinogenesis 2000, 21, 387–395. 27. Wakabayashi, K.; Nagao, M.; Esumi, H.; Sugimura, T. Food-Derived Mutagens and Carcinogens. Cancer Res. 1992, 52, 2092s–2098s. 28. Brands, C.M.; Alink, G.M.; van Boekel, M.A.; Jongen, W.M. Mutagenicity of Heated Sugar–Casein Systems: Effect of the Maillard Reaction. J. Agric. Food Chem. 2000, 48, 2271–2275. 29. Kinae, N.; Saito, M.; Sawada, S.; Yamashita, M.; Uto, H.; Tomita, I.; Kanamori, H. Mutagenicity of Nondialyzable Melanoidins Prepared From Carbohydrates and L- Tryptophan Before and After Nitrite Treatment. Chem. Pharm. Bull. (Tokyo) 1985, 33, 2438–2444.

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 30. Higashimoto, M.; Yamamoto, T.; Kinouchi, T.; Handa, Y.; Matsumoto, H.; Ohnishi, Y. Mutagenicity of Soy Sauce Treated with Nitrite in the Presence of Ethanol or Alcoholic Beverages. Mutat. Res. 1995, 345, 155–166. 31. Nagahara, A.; Ohshita, K.; Nasuno, S. Investigation of Soy Sauce Treated with Nitrite in the Chromosomal Aberration Test In Vitro and the Micronucleus Test In Vivo. Mutat. Res. 1991, 262, 171–176. 32. Kimura, S.; Okazaki, K.; Higashimoto, M.; Ohnishi, Y. Mutagenicity of Nitrite- Treated Soy Sauce in Chinese Hamster V79 Cells. Tokushima J. Exp. Med. 1990, 37, 31–34. 33. Obrien, J.; Morrissey, P.A. Nutritional and Toxicological Aspects of the Maillard Browning Reaction in Foods. Crit. Rev. Food Sci. Nutr. 1989, 28, 211–248. 34. Hodge, J.E. Chemistry of Browning Reactions in Model Systems. J. Agric. Food. Chem. 1953, 1, 928–943. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 171

35. Shibamoto, T. Volatile Falvor Chemicals Formed by the Maillard Reaction. In Thermal Generation of Aromas: ACS Symposium Series 409; Parliment, T.H., McGorrin, R.J., Ho, C.-T., Eds.; American Chemical Society: Washington, DC, 1989; Vol. xii, 134–142. 36. Shibamoto, T. Heterocyclic Compounds in Browning and Browning/Nitrite Model Systems: Occurrence, Formation Mechanisms, Flavor Characteristics and Mutagenic Activity. In Instrumental Analysis of Foods; Charalambous, G., Inglett, G.E., Eds.; Academic Press: New York, 1983; 229–278. 37. Mottram, D.S. Flavour Formation in Meat and Meat Products: A Review. Food Chem. 1998, 62, 415–424. 38. Sugimura, T.; Nagao, M.; Wakabayashi, K. How We Should Deal with Unavoidable Exposure of Man to Environmental Mutagens: Cooked Food Mutagen Discovery, Facts and Lessons for Cancer Prevention. Mutat. Res.— Fundam. Mol. Mech. Mutagen. 2000, 447, 15–25. 39. Felton, J.S.; Malfatti, M.A.; Knize, M.G.; Salmon, C.P.; Hopmans, E.C.; Wu, R.W. Health Risks of Heterocyclic Amines. Mutat. Res.—Fundam. Mol. Mech. Mutagen. 1997, 376, 37–41. 40. Robbanabarnat, S.; Rabache, M.; Rialland, E.; Fradin, J. Heterocyclic Amines— Occurrence and Prevention in Cooked Food. Environ. Health Perspect. 1996, 104, 280–288. 41. Aeschbacher, H.U.; Wolleb, U.; Loliger, J.; Spadone, J.C.; Liardon, R. Contribution of Coffee Aroma Constituents to the Mutagenicity of Coffee. Food Chem. Toxicol. 1989, 27, 227–232. 42. Bjeldanes, L.F.; Chew, H. Mutagenicity of 1,2-Dicarbonyl Compounds: Maltol, Kojic Acid, Diacetyl and Related Substances. Mutat. Res. 1979, 67, 367–371. 43. Kasai, H.; Kumeno, K.; Yamaizumi, Z.; Nishimura, S.; Nagao, M.; Fujita, Y.; Sugimura, T.; Nukaya, H.; Kosuge, T. Mutagenicity of Methylglyoxal in Coffee. Gann 1982, 73, 681–683. 44. Lederer, M.O.; Klaiber, R.G. Cross-Linking of Proteins by Maillard Processes: Characterization and Detection of Lysine–Arginine Cross-Links Derived from Glyoxal and Methylglyoxal. Bioorg. Med. Chem. 1999, 7, 2499–2507. 45. Kim, S.B.; Hayase, F.; Kato, H. Desmutagenic Effect of Alpha-Dicarbonyl and Alpha-Hydroxycarbonyl Compounds Against Mutagenic Heterocyclic Amines. Mutat. Res. 1987, 177, 9–15. 46. Dorado, L.; Ruis Montoya, M.R.; Rodriguez Mellado, J.M. A Contribution to the Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 Study of the Structure–Mutagenicity Relationship for Alpha-Dicarbonyl Compounds Using the Ames Test. Mutat. Res. 1992, 269, 301–306. 47. Rodriguez Mellado, J.M.; Ruiz Montoya, M. Correlations Between Chemical Reactivity and Mutagenic Activity Against S. Typhimurium Ta100 for Alpha- Dicarbonyl Compounds as a Proof of the Mutagenic Mechanism. Mutat. Res. 1994, 304, 261–264. 48. Shamberger, R.J.; Andreone, T.L.; Willis, C.E. Antioxidants and Cancer. IV. Initiating Activity of Malonaldehyde as a Carcinogen. J. Natl Cancer Inst. 1974, 53, 1771–1773. 49. Basu, A.K.; O’Hara, S.M.; Valladier, P.; Stone, K.; Mols, O.; Marnett, L.J. Identiﬁcation of Adducts Formed by Reaction of Guanine Nucleosides with Malondialdehyde and Structurally Related Aldehydes. Chem. Res. Toxicol. 1988, 1, 53–59. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

172 LEE AND SHIBAMOTO

50. Marnett, L.J.; Hurd, H.K.; Hollstein, M.C.; Levin, D.E.; Esterbauer, H.; Ames, B.N. Naturally Occurring Carbonyl Compounds Are Mutagens in Salmonella Tester Strain TA104. Mutat. Res. 1985, 148, 25–34. 51. Khan, Q.A.; Shamsi, F.A.; Hadi, S.M. Mutagenicity of Furfural in Plasmid DNA. Cancer Lett. 1995, 89, 95–99. 52. Shinohara, K.; Kim, E.H.; Omura, H. Furans as the Mutagens Formed by Amino- Carbonyl Reactions. In Amino-Carbonyl Reactions in Food and Biological Systems: Proceedings of the 3rd International Symposium on the Maillard Reaction, Susono, Shizuoka, Japan, 1–5 July 1985; Fujimaki, M., Namiki, M., Kato, H., Eds.; Kodansha: New York, 1986; Vol. xvi, 353–362. 53. Janzowski, C.; Glaab, V.; Samimi, E.; Schlatter, J.; Eisenbrand, G. 5- Hydroxymethylfurfural: Assessment of Mutagenicity, DNA-Damaging Potential and Reactivity Towards Cellular Glutathione. Food Chem. Toxicol. 2000, 38, 801–809. 54. Yamashita, N.; Murata, M.; Inoue, S.; Hiraku, Y.; Yoshinaga, T.; Kawanishi, S. Superoxide Formation and DNA Damage Induced by a Fragrant Furanone in the Presence of Copper(II). Mutat. Res. 1998, 397, 191–201. 55. Peterson, L.A.; Naruko, K.C.; Predecki, D.P. A Reactive Metabolite of Furan, cis- 2-Butene-1,4-dial, Is Mutagenic in the Ames Assay. Chem. Res. Toxicol. 2000, 13, 531–534. 56. Adams, T.B.; Doull, J.; Goodman, J.I.; Munro, I.C.; Newberne, P.; Portoghese, P.S.; Smith, R.L.; Wagner, B.M.; Weil, C.S.; Woods, L.A.; Ford, R.A. The FEMA GRAS Assessment of Furfural Used as a Flavour Ingredient. Flavor and Extract Manufacturers’ Association. Food Chem. Toxicol. 1997, 35, 739–751. 57. Surh, Y.J.; Liem, A.; Miller, J.A.; Tannenbaum, S.R. 5-Sulfooxymethylfurfural as a Possible Ultimate Mutagenic and Carcinogenic Metabolite of the Maillard Reaction Product, 5-Hydroxymethylfurfural. Carcinogenesis 1994, 15, 2375–2377. 58. Hiramoto, K.; Asoo, R.; Ni-iyama, H.; Hikage, S.; Kato, T.; Kikugawa, K. DNA Strand Break by 2,5-Dimethyl-4-hydroxy-3(2H)-furanone, a Fragrant Compound in Various Foodstuffs. Mutat. Res. 1996, 359, 17–24. 59. Li, X.; Hiramoto, K.; Yoshida, M.; Kato, T.; Kikugawa, K. Identiﬁcation of 2,5- Dimethyl-4-hydroxy-3(2H)-furanone (DMHF) and 4-Hydroxy-2(or 5)-ethyl-5(or

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 2)-methyl-3(2H)-furanone (HEMF) with DNA Breaking Activity in Soy Sauce. Food Chem. Toxicol. 1998, 36, 305–314. 60. Martin, F.L.; Ames, J.M. Formation of Strecker Aldehydes and Pyrazines in a Fried Potato Model System. J. Agric. Food Chem. 2001, 49, 3885–3892. 61. Maga, J.A. Pyrazines in Foods: An Update. Crit. Rev. Food Sci. Nutr. 1982, 16, 1–48. 62. Omura, H.; Jahan, N.; Shinohara, K.; Murakami, H. Formation of Mutagens by the Maillard Reaction. In The Maillard Reaction in Foods and Nutrition; Waller, G.R., Feather, M.S., Eds.; American Chemical Society: Washington, DC, 1983; 537–563. 63. Wang, C.J.; Lin, Y.L.; Lin, J.K. Mutagenicity and Cytotoxicity of Nitropyrrole Compounds Derived from the Reaction of 2-Acetyl Pyrrole with Nitrite. Food Chem. Toxicol. 1994, 32, 839–844. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 173

64. Lee, H.; Lin, M.Y.; Chan, S.C. Formation and Identiﬁcation of Carcinogenic Heterocyclic Aromatic Amines in Boiled Pork Juice. Mutat. Res. 1994, 308, 77–88. 65. Lee, H.; Bian, S.S.; Chen, Y.L. Genotoxicity of 1,3-Dithiane and 1,4-Dithiane in the CHO/SCE Assay and the Salmonella/Microsomal Test. Mutat. Res. 1994, 321, 213–218. 66. Mihara, S.; Shibamoto, T. Mutagenicity of Products Obtained from Cysteamine— Glucose Browning Model Systems. J. Agric. Food Chem. 1980, 28, 62–66. 67. Umano, K.; Shibamoto, T.; Fernando, S.Y.; Wei, C.I. Mutagenicity of 2- Hydroxyalkyl-N-Nitrosothiazolidines. Food Chem. Toxicol. 1984, 22, 253–259. 68. Loury, D.J.; Byard, J.L.; Shibamoto, T. Genotoxicity of N-Nitrosothiazolidine in Microbial and Hepatocellular Test Systems. Food Chem. Toxicol. 1984, 22, 1013–1014. 69. Miller, A.J.; Pensabene, J.W.; Doerr, R.C.; Fiddler, W. Apparent Mutagenicity of N-Nitrosothiazolidine Caused by a Trace Contaminant. Mutat. Res. 1985, 157, 129–134. 70. Shibamoto, T.; Bernhard, R.A. Formation of Heterocyclic Compounds from the Reaction of L-Rhamnose with Ammonia. J. Agric. Food Chem. 1978, 26, 183–187. 71. Voogd, C.E.; van der Stel, J.J.; van Bruchem, M.C. Increased Mutagenicity of Some Nitroimidazoles by Non-mutagenic Nitrotoluene on Klebsiella pneumoniae (ﬂuctuation Test). Mutat. Res. 1992, 282, 73–77. 72. Voogd, C.E.; van der Stel, J.J.; Jacobs, J.J. The Mutagenic Action of Nitroimidazoles. IV. A Comparison of the Mutagenic Action of Several Nitroimidazoles and Some Imidazoles. Mutat. Res. 1979, 66, 207–221. 73. Voogd, C.E.; Van der Stel, J.J.; Jacobs, J.J. The Mutagenic Action of Nitroimidazoles. II. Tinidazole, Ipronidazole, Panidazole and Ornidazole. Mutat. Res. 1977, 48, 155–161. 74. Voogd, C.E.; Van der Stel, J.J.; Jacobs, J.J. The Mutagenic Action of Nitroimidazoles. II. Effects of 2-Nitroimidazoles. Mutat. Res. 1975, 31, 149–152. 75. Voogd, C.E.; Van der Stel, J.J.; Jacobs, J.J. The Mutagenic Action of Nitroimidazoles. I. Metronidazole, Nimorazole, Dimetridazole and Ronidazole. Mutat. Res. 1974, 26, 483–490. 76. Shibamoto, T. Occurrence of Mutagenic Products in Browning Model Systems.

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 Food Technol. 1982, 59–62. 77. Jones, R.C.; Weisburger, J.H. Inhibition of Aminoimidazoquinoxaline-Type and Aminoimidazol-4-one-Type Mutagen Formation in Liquid Reﬂux Models by L- Tryptophan and Other Selected Indoles. Jpn. J. Cancer Res. 1988, 79, 222–230. 78. Jones, R.C.; Weisburger, J.H. Characterization of Aminoalkylimidazol-4-one Mutagens from Liquid-Reﬂux Models. Mutat. Res. 1989, 222, 43–51. 79. Weisburger, J.H.; Jones, R.C. Nutritional Toxicology: On the Mechanisms of Inhibition of Formation of Potent Carcinogens During Cooking. Prog. Clin. Biol. Res. 1989, 304, 377–390. 80. Stich, H.F.; Stich, W.; Rosin, M.P.; Powrie, W.D. Mutagenic Activity of Pyrazine Derivatives: A Comparative Study with Salmonella typhimurium, Saccharomyces cerevisiae and Chinese Hamster Ovary Cells. Food Cosmet. Toxicol. 1980, 18, 581–584. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

174 LEE AND SHIBAMOTO

81. Jenq, S.N.; Tsai, S.J.; Lee, H. Antimutagenicity of Maillard Reaction Products From Amino Acid/Sugar Model Systems Against 2-Amino-3-methylimidazo-[4,5- f]Quinoline: The Role of Pyrazines. Mutagenesis 1994, 9, 483–488. 82. Kikugawa, K.; Hiramoto, K.; Kato, T.; Yanagawa, H. Effect of Food Reductones on the Generation of the Pyrazine Cation Radical and on the Formation of the Mutagens in the Reaction of Glucose, Glycine and Creatinine. Mutat. Res. 2000, 465, 183–190. 83. Eisenbrand, G.; Tang, W. Food-Borne Heterocyclic Amines. Chemistry, Formation, Occurrence and Biological Activities. A Literature Review. Toxicology 1993, 84, 1–82. 84. Toribio, F.; Galceran, M.T.; Puignou, L. Separation of Heteroaromatic Amines in Food Products. J. Chromatogr. B Biomed. Sci. Appl. 2000, 747, 171–202. 85. Skog, K.I.; Johansson, M.A.; Jagerstad, M.I. Carcinogenic Heterocyclic Amines in Model Systems and Cooked Foods: A Review on Formation, Occurrence and Intake. Food Chem. Toxicol. 1998, 36, 879–896. 86. Skog, K. Cooking Procedures and Food Mutagens: A Literature Review. Food Chem. Toxicol. 1993, 31, 655–675. 87. Overvik, E.; Gustafsson, J.A. Cooked-Food Mutagens: Current Knowledge of Formation and Biological Signiﬁcance. Mutagenesis 1990, 5, 437–446. 88. Nishimura, S. Chemistry of Mutagens and Carcinogens in Broiled Food. Environ. Health Perspect. 1986, 67, 11–16. 89. Kasai, H.; Nishimura, S.; Nagao, M.; Takahashi, Y.; Sugimura, T. Fractionation of a Mutagenic Principle from Broiled Fish by High- Pressure Liquid Chromatog- raphy. Cancer Lett. 1979, 7, 343–348. 90. Skog, K.; Solyakov, A.; Arvidsson, P.; Jagerstad, M. Analysis of Nonpolar Heterocyclic Amines in Cooked Foods and Meat Extracts Using Gas Chromatography-Mass Spectrometry. J. Chromatogr. A 1998, 803, 227–233. 91. Skog, K.; Augustsson, K.; Steineck, G.; Stenberg, M.; Jagerstad, M. Polar and Non- Polar Heterocyclic Amines in Cooked Fish and Meat Products and Their Corresponding Pan Residues. Food Chem. Toxicol. 1997, 35, 555–565. 92. Jagerstad, M.; Skog, K.; Grivas, S.; Olsson, K. Formation of Heterocyclic Amines Using Model Systems. Mutat. Res. 1991, 259, 219–233. 93. Jagerstad, M.; Skog, K. Formation of Meat Mutagens. Adv. Exp. Med. Biol. 1991, 289, 83–105.

Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 94. Gooderham, N.J.; Watson, D.; Rice, J.C.; Murray, S.; Taylor, G.W.; Davies, D.S. Metabolism of the Mutagen MeIQx In Vivo: Metabolite Screening by Liquid Chromatography-Thermospray Mass Spectrometry. Biochem. Biophys. Res. Commun. 1987, 148, 1377–1382. 95. Edmonds, C.G.; Sethi, S.K.; Yamaizumi, Z.; Kasai, H.; Nishimura, S.; McCloskey, J.A. Analysis of Mutagens from Cooked Foods by Directly Combined Liquid Chromatography-Mass Spectrometry. Environ. Health Perspect. 1986, 67, 35–40. 96. Yamaizumi, Z.; Kasai, H.; Nishimura, S.; Edmonds, C.G.; McCloskey, J.A. Stable Isotope Dilution Quantiﬁcation of Mutagens in Cooked Foods by Combined Liquid Chromatography-Thermospray Mass Spectrometry. Mutat. Res. 1986, 173,1–7. 97. Morales, F.J.; Jimenez-Perez, S. Free Radical Scavenging Capacity of Maillard Reaction Products as Related to Colour and Fluorescence. Food Chem. 2001, 72, 119–125. MARCEL DEKKER, INC. • 270 MADISON AVENUE • NEW YORK, NY 10016

TOXICOLOGY OF MAILLARD REACTION PRODUCTS 175

98. Nicoli, M.C.; Anese, M.; Parpinel, M.T.; Franceschi, S.; Lerici, C.R. Loss and/or Formation of Antioxidants During Food Processing and Storage. Cancer Lett. 1997, 114, 71–74. 99. Ford, S.M.; Hook, J.B.; Bond, J.T. The Effects of Butylated Hydroxyanisole and Butylated Hydroxytoluene on Renal Function in the Rat. I. Effects on Fluid and Electrolyte Excretion. Food Cosmet. Toxicol. 1980, 18, 15–20. 100. Franzke, C.; Ivanisky, H. Antioxidant Capacity of Melanoidin. Dtsch. Lebensm.- Rundsch 1954, 50, 251–254. 101. Manzocco, L.; Calligaris, S.; Mastrocola, D.; Nicoli, M.C.; Lerici, C.R. Review of Non-enzymatic Browning and Antioxidant Capacity in Processed Foods. Trends Food Sci. Technol. 2001, 11, 340–346. 102. Bressa, F.; Tesson, N.; Dallarosa, M.; Sensidoni, A.; Tubaro, F. Antioxidant Effect of Maillard Reaction Products—Application to a Butter Cookie of a Competition Kinetics Analysis. J. Agric. Food Chem. 1996, 44, 692–695. 103. Morales, F.J.; van Boekel, M. A Study on Advanced Maillard Reaction in Heated Casein/Sugar Solutions: Colour Formation. Int. Dairy J. 1998, 8, 907–915. 104. Fuster, M.D.; Mitchell, A.E.; Ochi, H.; Shibamoto, T. Antioxidative Activities of Heterocyclic Compounds Formed in Brewed Coffee. J. Agric. Food Chem. 2000, 48, 5600–5603. 105. Hayase, F.; Takahashi, Y.; Tominaga, S.; Miura, M.; Gomyo, T.; Kato, H. Identiﬁcation of Blue Pigment Formed in a D-Xylose–Glycine Reaction System. Biosci. Biotechnol. Biochem. 1999, 63, 1512–1514. 106. Eiserich, J.P.; Shibamoto, T. Antioxidative Activity of Volatile Heterocyclic Compounds. J. Agric. Food Chem. 1994, 42, 1060–1063. 107. Shaker, E.S.; Ghazy, M.A.; Shibamoto, T. Antioxidative Activity of Volatile Browning Reaction Products and Related Compounds in a Hexamal/hexanoic Acid System. J. Agric. Food Chem. 1995, 43, 1017–1022. 108. Eiserich, J.P.; Wong, J.W.; Shibamoto, T. Antioxidative Activities of Furan- and Thiophenethiols Measured in Lipid Peroxidation Systems and by Tyrosyl Radical Scavenging Assay. J. Agric. Food Chem. 1995, 43, 647–650. 109. Macku, C.; Shibamoto, T. Volatile Antioxidants Produced from Heated Corn Oil Glycine Model System. J. Agric. Food Chem. 1991, 39, 1990–1993. 110. Jing, H.; Kitts, D.D. Comparison of the Antioxidative and Cytotoxic Properties of Glucose–Lysine and Fructose–Lysine Maillard Reaction Products. Food Res. Int. 2000, 33, 509–516. Downloaded by [UNAM Ciudad Universitaria] at 15:45 09 April 2014 111. Ide, N.; Lau, B.H.S.; Ryu, K.; Matsuura, H.; Itakura, Y. Antioxidant Effects of Fructosyl Arginine, a Maillard Reaction Product in Aged Garlic Extract. J. Nutr. Biochem. 1999, 10, 372–376. 112. Anese, M.; Manzocco, L.; Nicoli, M.C.; Lerici, C.R. Antioxidant Properties of Tomato Juice as Affected by Heating. J. Sci. Food Agric. 1999, 79, 750–754. 113. Anese, M.; Nicoli, M.C.; Massini, R.; Lerici, C.R. Effects of Drying Processing on the Maillard Reaction in Pasta. Food Res. Int. 1999, 32, 193–199. 3152 J. Agric. Food Chem. 2000, 48, 3152−3158

Contribution of Pyrrole Formation and Polymerization to the Nonenzymatic Browning Produced by Amino-Carbonyl Reactions

Rosario Zamora, Manuel Alaiz, and Francisco J. Hidalgo*

Instituto de la Grasa, CSIC, Avenida Padre Garcı´a Tejero 4, 41012 Sevilla, Spain

Recent studies have hypothesized that pyrrole formation and polymerization may be contribute to the nonenzymatic browning produced in both oxidized lipid/protein reactions and the Maillard reaction. To develop a methodology that would allow investigation of the contribution of this browning mechanism, the kinetics of formation of color, fluorescence, and pyrrolization in 4,5(E)-epoxy-2(E)- heptenal/lysine and linolenic acid/lysine model systems were studied. In both cases similar kinetics for the three measurements were observed at the two temperatures assayed (37 and 60 °C), and there was a high correlation among color, fluorescence, and pyrrolization measurements obtained as a function of incubation time. Because the color and fluorescence production in the 4,5(E)-epoxy- 2(E)-heptenal/lysine system is a consequence of pyrrole formation and polymerization, the high correlations observed with the unsaturated fatty acid also suggest a contribution of the pyrrole formation and polymerization to the development of color and fluorescence observed in the fatty acid/lysine system. Although the contribution of other mechanisms cannot be discarded, all of these results suggest that when the pyrrole formation and polymerization mechanism contributes to the nonenzymatic browning of foods, a high correlation among color, fluorescence, and pyrrolization measurements should be expected.

Keywords: Nonenzymatic browning; Maillard reaction; oxidized lipid/protein reactions; aminocarbonyl reactions; pyrrole polymerization; color; fluorescence; oxidative stress

INTRODUCTION intermediates for the nonenzymatic browning produced between oxidized lipids and amino acids and proteins Nonenzymatic browning reactions of amino acids and (Hidalgo and Zamora, 1993). These compounds have proteins with carbohydrates and oxidized lipids cause been shown to be produced in the reaction of 4,5-epoxy- modification of food during storage and processing 2-alkenals with the amino groups of amino acids and together with simultaneous formation of both deleteri- proteins, and they are always accompanied by the ous and beneficial compounds (Friedman, 1996; Hutch- ings, 1994; Narayan, 1998; Sapers, 1993). They embrace formation of N-substituted pyrroles (II) (Zamora and Hidalgo, 1994, 1995). These last compounds are much a whole network of different reactions that are only > partially understood due to the high reactivities of the more stable and have been found in 20 fresh food reactants and products, the intertwining reaction routes, products, including meats, fishes, vegetables, and nuts and the diversity of products (Ames et al., 1999; Ikan, (Zamora et al., 1999). However, the determination of 1996; Namiki, 1988; O’Brien et al., 1998). The products N-substituted 2-(1-hydroxyalkyl)pyrroles is much more that have been characterized are preponderantly those complex because they polymerize rapidly and spontane- that are stable and do not undergo further change to ously to produce brown macromolecules with fluorescent any great extent either in the reaction mixture or during melanoidin-like characteristics (Hidalgo and Zamora, isolation or purification (Ledl and Schleicher, 1990). 1993). This polymerization reaction, which is shown in Such compounds can serve as indicator substances for Figure 1, was characterized for 2-(1-hydroxyethyl)-1- certain reaction pathways in the nonenzymatic brown- methylpyrrole and produces sequentially dimers, tri- ing. However, difficulties are encountered in the isola- mers, tetramers, and, lately, melanoidin-like polymers tion of reactive intermediates, because they are present (Hidalgo and Zamora, 1993). only in very low concentrations in the reaction mixture One possibility for studying the formation of N- and they usually react further during isolation. Never- substituted 2-(1-hydroxyalkyl)pyrroles, and its later theless, these compounds are precisely those of the polymerization, in oxidized lipid/amino acid mixtures greatest significance for these reactions because they is to follow the pyrrole formation and polymerization play an important role in the formation of browning in these mixtures. Pyrrole rings can be easily deter- products, aroma compounds, and high molecular weight mined by using the Ehrlich reagent (Hidalgo et al., substances. 1998). However, this procedure has been neither used Model studies carried out in this laboratory have nor demonstrated to be useful for the study of pyrrole identified two N-substituted 2-(1-hydroxyalkyl)pyrroles polymerization. The present investigation was under- (compound I with R2 ) H in Figure 1) as potential key taken to apply the Ehrlich reaction to the study of pyrrole formation and polymerization in oxidized lipid/ * Author to whom correspondence should be addressed amino acid mixtures to evaluate the contribution of [telephone +(3495) 461 1550; fax +(3495) 461 6790; e-mail pyrrole polymerization to the nonenzymatic browning [email protected]]. produced in these mixtures. 10.1021/jf991090y CCC: $19.00 © 2000 American Chemical Society Published on Web 06/30/2000 Pyrrole Polymerization and Nonenzymatic Browning J. Agric. Food Chem., Vol. 48, No. 8, 2000 3153

Figure 2. Chemical structures of model pyrroles used in this study. 2-(1-Hydroxyethyl)pyrrole (IX) and 2-(1-hydroxyethyl)-1- methylpyrrole (X) were prepared from compounds VII and VIII, respectively, according to the procedure described previously for the synthesis of compound X (Hidalgo and Zamora, 1993). Briefly, compounds VII and VIII (4 mmol) were reduced with sodium borohydride (4 mmol) in methanol (5 mL) for 1.5 h at room temperature. The reactions were fractionated by column chromatography with ether/hexane (1:1) as eluent, and pure compounds IX and X were obtained (60 and 98%, respectively); their identities were confirmed by 1H and 13C nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS). Spectroscopic characterization of compound X was described previously (Hidalgo and Zamora, 1993). 1 Spectral data obtained for compound IX: H NMR [(CD3)2CO] δ 1.49 d (3H, J ) 6.5 Hz, H2′), 4.86 q (1H, J ) 6.5 Hz, H1′), 5.98 m (1H, H3), 6.01 q (1H, J ) 2.8 Hz, H4), 6.67 dt (1H, JH3,H5 ) 1.7 Hz, JH5,NH ) 2.6 Hz, JH4,H5 ) 2.6 Hz, H5), and 13 9.13 s, br (1H, NH); C NMR [(CD3)2CO] δ 23.61 q (C2′), 63.93 d (C1′), 104.33 d (C3), 107.80 d (C4), 117.56 d (C5), and 136.78 s (C2); MS of the trimethylsilyl derivative (70 eV), m/z (%, ion + + + structure) 183 (7, M ), 168 (38, M - CH3), 93 [76, M - (CH3)3SiOH], and 75 (100). 1,5-Dimethyl-2-pyrrolecarboxaldehyde (XIV) was prepared by reduction of compound XIII with lithium aluminum hy- Figure 1. Pyrrole formation and polymerization mechanism. dride. A solution of 240 mg (2 mmol) of compound XIII in Pyrroles are produced in the reaction between oxidized fatty diethyl ether (1.2 mL) was treated with 50 mg of AlLiH4, which acids having the 4,5-epoxy-1-oxo-2-pentene group (including was added in small amounts. The reaction mixture was 4,5-epoxy-2-alkenals) and primary amino groups of amines, allowed to react for 15 min at room temperature, then diluted amino acids, and proteins. Nonenzymatic browning is a with 4 mL of ether, and, finally, treated slowly with methanol consequence of the polymerization of N-substituted 2-(1- to eliminate the excess of hydride. The solids were removed, hydroxyalkyl)pyrroles (I). N-Substituted pyrroles (II) have and the resulting solution was evaporated and fractionated been detected in fresh food products (Zamora et al., 1999) as by column chromatography on silica gel 60 using hexane/ well as volatile pyrroles (Zamora et al., 1994). acetone (4:1) as eluent. Pure compound XIV was obtained (51.2 1 mg, 21%): TLC, Rf 0.39 (hexane/acetone, 4:1); H NMR [(CD3)2- EXPERIMENTAL PROCEDURES CO] δ 2.25 s (3H, CH3), 3.83 s (3H, NCH3), 6.01 dd (1H, JH4,CHO ) 0.6 Hz, JH3,H4 ) 3.9 Hz, H4), 6.84 d (1H, JH3,H4 ) 3.9 Hz, 13 Materials. 4,5(E)-Epoxy-2(E)-heptenal was prepared in a H3), and 9.40 d (1H, JH4,CHO ) 0.6 Hz, CHO); C NMR [(CD3)2- manner analogous to that of 4,5(E)-epoxy-2(E)-decenal (Zamo- CO] δ 11.87 q (CH3), 32.27 q (NCH3), 110.02 d (C4), 124.68 d ra and Hidalgo, 1995). Pyrrole (III), 1-methylpyrrole (IV), 2,5- (C3), 132.76 s (C2), 141.26 s (C5), and 178.78 d (CHO); MS dimethylpyrrole (V), 1,2,5-trimethylpyrrole (VI), 2-acetylpyr- (70 eV), m/z (%, ion structure) 123 (100, M+), 122 (100, M+ - + + role (VII), 2-acetyl-1-methylpyrrole (VIII), pyrrole-2-carbox- 1), 108 (43, M - CH3), 94 (89, M - CHO), 67 (58), and 53 aldehyde (XI), 1-methyl-2-pyrrolecarboxaldehyde (XII), and (84). 1,5-dimethyl-2-pyrrolecarbonitrile (XIII) were purchased from Determination of Absorption Maxima and Extinction Aldrich (Milwaukee, WI). Structures for the model pyrroles Coefficients of Ehrlich Adducts of Model Pyrroles. Model employed in this study are collected in Figure 2. Other re- pyrroles were dissolved in water and derivatized with p- agents and solvents were of analytical grade and were pur- (dimethylamino)benzaldehyde (Mattocks, 1968; Liddell et al., chased from reliable commercial sources. 1993) using the conditions described previously (Hidalgo et 3154 J. Agric. Food Chem., Vol. 48, No. 8, 2000 Zamora et al. al., 1998). Briefly, the solution of pyrrole in water (800 µL) Table 1. Absorbance Maxima and Extinction Coefficients was introduced into a 1.5 mL microtube and treated with 128 of Ehrlich Adducts Obtained from Model Pyrroles µLof2%p-(dimethylamino)benzaldehyde in 3.5 M HCl/ethanol absorbance maximum (extinction coefficient)a (4:1). The tube was closed and heated at 45 °C for 30 min, and, finally, the maximum at 450-600 nm was determined compound at 450-500 nm at ∼520 nm at ∼560 nm spectrophotometrically. III 496 (18000) 557 (37000) Reaction of 4,5(E)-Epoxy-2(E)-heptenal with Lysine. IV 565 (38000) Lysine (41 mg, 0.28 mmol) was dissolved in 7 mL of 0.3 M V 521 (65000) sodium phospate buffer, pH 7.4, and treated with 18 mg (0.14 VI 523 (56000) mmol) of 4,5(E)-epoxy-2(E)-heptenal. The reaction was main- VII 481 (4) 511 (5) 557 (3) tained at 37 or 60 °C, and, at different intervals of time, VIII 485 (20) 517 (24) 563 (15) samples (500 µL) were withdrawn for analytical determina- IX 541 (2500) tions. Samples were extracted with 500 µL of CHCl3/MeOH X 557 (6800) (2:1) and centrifuged at 2250g for 5 min, and the aqueous XI 553 (100) phase was employed for the determination of color, fluores- XII 563 (800) cence, and pyrrole content. XIII 490 (230) 521 (240) The color of the solutions was determined by using the XIV 528 (5900) weighted-ordinate method (Hunter, 1973). Tristimulus values a Wavelength in nm; extinction coefficient in M-1. (X, Y, Z) were calculated from the transmittances (T) obtained in a Beckman spectrometer. Transmittances were recorded at peared at 541 nm. In addition, compounds VII and VIII constant intervals (10 nm) from 400 to 700 nm using 1 cm - glass cells. These readings were then converted by means of a also exhibited a similar maximum at 511 517 nm, but computer program into the corresponding tristimulus and the two maxima had very small extinction coefficients. CIELAB L* a* b* color values (CIE, 1978). The difference of On the contrary, if there was no proton at the R-position, color (∆E) between CIELAB L* a* b* determined at the initial the main maximum of the Ehrlich adduct appeared at time and that determined at each time was calculated, 521-528 nm. Additionally, some derivatives exhibited according to Hunter (1973), using the following equation: other maxima at wavelengths slightly below 500 nm. It occurred with the pyrrole and the derivatives con- ∆E ) [(∆a*)2 + (∆b*)2 + (∆L*)2]1/2 (1) taining ketone and nitrile groups. Thus, the pyrrole exhibited a second maximum at 496 nm, which had an Fluorescence spectra were recorded on a Perkin-Elmer LS-5 extinction coefficient lower than the main maximum. fluorescence spectrometer of 25-50 µL of aqueous phase This maximum was also present in compounds VII, diluted to 2.5 mL with 50 mM sodium phosphate buffer, pH VIII, and XIII but with much smaller extinction coef- 7.4. A slit width of 5 nm was used, and the instrument was ficients. standardized with quinine sulfate (0.1 µMin0.1NH2SO4)to give fluorescence intensity of 100 at 450 nm, when excitation The highest extinction coefficients were obtained was at 350 nm. Results are given for 25 µL of sample. when only hydrogen or alkyl groups were present as Determination of pyrrole amino acids was carried out as substituents. These were compounds III)VI, and the described previously. The pyrrole content was determined extinction coefficients ranged from 37000 for compounds spectrophotometrically at the maximum at ∼564 nm by using III and IV to 60000 for compounds V and VI. The an extinction coefficient of 37000. introduction of a hydroxyalkyl group decreased the Reaction of Linolenic Acid with Lysine. Lysine (41 mg, extinction coefficient, and it was smaller if the substitu- 0.28 mmol) was dissolved in 7 mL of 0.3 M sodium phospate ent was an aldehyde. Nevertheless, the smallest extinc- buffer, pH 7.4, and treated with 39 mg (0.14 mmol) of linolenic tion coefficients were observed when the substituent acid. The reaction was maintained at 37 or 60 °C, and, at was a ketone. different intervals of time, samples (500 µL) were withdrawn for determination of color, fluorescence, and pyrrole content. These results suggested that in pyrrole mixtures, for Samples were extracted with 500 µL of CHCl3/MeOH (2:1) and which the maximum observed is the sum of the maxima centrifuged at 2250g for 5 min. The aqueous phase was of the different compounds, this maximum would mainly employed for the determinations, which were carried out as be the sum of the unsubstituted pyrroles and the described above. The only difference was the measurement of pyrroles substituted only with alkyl groups because pyrrole amino acids, which were determined spectrophoto- these are the pyrroles with higher extinction coef- ∼ metrically at the maximum at 525 nm by using an extinction ficients. By using an extinction coefficient of 37000 (for coefficient of 60000. the maximum at ∼560 nm) or 60000 (for the maximum at ∼520 nm) in pyrrole mixtures, it is possible to RESULTS determine a minimum concentration of pyrroles in the Determination of Absorption Maxima and Ex- mixture, which will mainly correspond to the sum of the tinction Coefficients of Ehrlich Adducts of Model unsubstituted and alkyl-substituted pyrroles present. Pyrroles. As a first step in the determination of pyrrole In the case of pyrrole polymerization, 2-(1-hydroxyal- formation and polymerization using the Ehrlich reagent, kyl)pyrroles are converted to alkyl-substituted polypyr- different model pyrroles were derivatized with p-(di- roles (Hidalgo and Zamora, 1993), and these last methylamino)benzaldehyde and studied spectrophoto- compounds should contribute significantly to the maxima metrically. The reaction of a pyrrole ring with the of the Ehrlich adducts. Ehrlich reagent always produced at least one main Reaction of 4,5(E)-Epoxy-2(E)-heptenal with maximum of absorbance at 500-600 nm. Both the Lysine. To study how the nonenzymatic browning position and the intensity of this maximum depended produced as a consequence of pyrrole formation and on the number and the type of the substituents in the polymerization may be followed by using the Ehrlich pyrrole ring. Table 1 collects the absorption maxima and reagent, the reaction between 4,5(E)-epoxy-2(E)-hepte- extinction coefficients of model pyrroles III)XIV. Most nal with lysine was selected because it has been well of the pyrroles that had at least one proton in the characterized and it is known that the color and R-position exhibited the main maximum at 553-565 fluorescence produced are a consequence of the forma- nm. However, for compound IX, this maximum ap- tion of N-substituted 2-(1-hydroxyalkyl)pyrroles and Pyrrole Polymerization and Nonenzymatic Browning J. Agric. Food Chem., Vol. 48, No. 8, 2000 3155

Ehrlich adducts, very good correlations were observed among the three measurements (Table 2). Analogous results were obtained at 60 °C (Figure 4). This temperature increased the reaction rates when compared with the results obtained at 37 °C, but analogous Ehrlich adducts were produced. In addition, similar kinetics for the formation of color, fluorescence, and pyrroles were also observed. Analogously to the results obtained at 37 °C, color, fluorescence, and pyrrolization measurements could also be adjusted by using the Boltzmann equation (eq 2). The widths of the curves were 0.25, 0.64, and 0.17 h for color, fluorescence, and pyrrolization, respectively. As expected because the reaction occurred more rapidly, these values were smaller than those obtained at 37 °C but they increased following a similar order. In addition, color, fluorescence, and pyrrolization measurements were highly correlated analogously to those observed at 37 °C (Table 2). All of these results suggest that, when nonenzymatic browning is a consequence of pyrrole production and polymerization, as occurs in the epoxyalkenal/lysine system, analogous kinetics for these three measurements should be expected. This can be analyzed by determining the correlations among these three measurements, which should be very high. In addition, and by using the Boltzmann equation, some small differences in the width of the adjusted curves should also be expected. In this case, the smallest dx values should Figure 3. Time course of (A) color, (B) fluorescence, and (C) correspond to the curve of pyrrolization, because pyrrole pyrrolization in a reaction of 4,5(E)-epoxy-2(E)-heptenal and formation occurs prior to the development of color and lysine at 37 °C. fluorescence. In addition, formation of color seems also to be slightly faster than fluorescence production. their subsequent polymerization (Hidalgo and Zamora, Reaction of Linolenic Acid with Lysine. Analog- 1995a; Zamora and Hidalgo, 1994). ously to the reaction between 4,5(E)-epoxy-2(E)-heptenal When a 4,5(E)-epoxy-2(E)-heptenal/lysine reaction and lysine, the reaction between linolenic acid and was heated at 37 °C, the production of color, fluores- lysine also produced color, fluorescence, and pyrroliza- cence, and pyrrole was observed. Figure 3 shows the tion. However, the Ehrlich adducts produced in this time course of these three measurements. Analogously reaction exhibited the main absorbance maximum at to color and fluorescence, Ehrlich adducts increased ∼526 nm. This maximum suggested that trisubstituted with incubation time and always exhibited analogous ∼ pyrroles were produced preferentially from long-chain absorbance spectra with the main maximum at 564 fatty acids. Figure 5 collects the results obtained for a nm (data not shown). The measurements of color, linolenic acid/lysine reaction mixture heated at 37 °C. fluorescence, and pyrrolization followed very similar Similar to that observed in the 4,5(E)-epoxy-2(E)- kinetics, which could be adjusted by using the Boltz- heptenal/lysine reaction mixture, analogous kinetics for mann equation (Microcal Origin, v. 4.10, Microcal color, fluorescence, and pyrrole production were also Software, Northampton, MA) observed in this fatty acid/lysine system. In this case,

- the reaction took place more slowly because the oxida- ) - + (x x0)/dx + y [(A1 A2)/(1 e ] A2 (2) tion of the fatty acid had to be produced prior to its reaction with the amino acid. When the curves were where A1 is the initial Y value, A2 is the final Y value, adjusted by using the Boltzmann equation (eq 2), the x0 is the x value at Y50, and dx is the width. The dx curve widths obtained were 99.73, 199.22, and 80.53 h values obtained for color, fluorescence, and pyrrolization for color, fluorescence, and pyrrolization, respectively. curves were 0.98, 1.49, and 0.43 h, respectively, sug- As expected, these values were much higher than the gesting that although the three measurements changed values obtained for the 4,5(E)-epoxy-2(E)-heptenal/ parallelly, the pyrrolization was produced slightly faster lysine reaction mixture. However, they followed an order than the formation of color and both of them were a bit similar to that obtained for the epoxyalkenal/lysine faster than fluorescence. These results are in agreement system, and the correlations among color, fluorescence, with the mechanism of color and fluorescence production and pyrrolization measurements were also very high suggested for this reaction: the reaction of an epoxy- (Table 2). alkenal with lysine produces in a first step both N- When the linolenic acid/lysine reaction mixture was substituted 2-(1-hydroxyalkyl)pyrroles and N-substitut- studied at 60 °C (Figure 6), the production of color, ed pyrroles, and the rapid polymerization of the fluorescence, and pyrroles was faster than at 37 °C, but hydroxyalkylpyrroles is the origin of the color and analogous kinetics for the three measurements were fluorescence produced (Hidalgo and Zamora, 1993; also observed. In addition, the three curves could be Zamora and Hidalgo, 1994). Nevertheless, because both adjusted by using the Boltzmann equation (eq 2), and pyrrole formation and polymerization is almost simul- the dx values obtained were 108.00, 156.04, and 68.45 taneous and pyrrole polymerization is likely to increase h, for color, fluorescence, and pyrrolization, respectively. 3156 J. Agric. Food Chem., Vol. 48, No. 8, 2000 Zamora et al.

Table 2. Correlations among Color, Fluorescence, and Pyrrolization correlation coefficient (significance) reaction temp, °C color/fluorescence color/pyrrolization fluorescence/pyrrolization epoxyalkenal/lysine 37 0.964 (0.000115) 0.883 (0.0037) 0.955 (0.000223) 60 0.977 (<0.0001) 0.956 (<0.0001) 0.986 (<0.0001) linolenic acid/lysine 37 0.983 (<0.0001) 0.976 (<0.0001) 0.989 (<0.0001) 60 0.984 (<0.0001) 0.945 (<0.0001) 0.975 (<0.0001)

Figure 4. Time course of (A) color, (B) fluorescence, and (C) pyrrolization in a reaction of 4,5(E)-epoxy-2(E)-heptenal and Figure 5. Time course of (A) color, (B) fluorescence, and (C) lysine at 60 °C. pyrrolization in a reaction of linolenic acid and lysine at 37 °C. Therefore, and analogous to the results obtained at 37 the formation of colored low molecular weight com- °C, the smallest dx values corresponded to pyrrolization, pounds (Chio and Tappel, 1969; Kikugawa and Ido, followed by color and fluorescence. In addition, there 1984; Nakamura et al., 1998). were very good correlations among color, fluorescence, and pyrrolization measurements (Table 2). The results obtained in the present study propose a methodology, which may be applied to food systems, to investigate the contribution of pyrrole formation and DISCUSSION polymerization mechanism to the nonenzymatic brown- The first mechanism proposed for the nonenzymatic ing produced by oxidized lipid/protein reactions. By browning produced by oxidized lipid/protein reactions studying the 4,5(E)-epoxy-2(E)-heptenal/lysine model was a repeated aldol condensation of the Schiff bases reaction, in which color and fluorescence have been produced between the carbonyl compounds derived from shown to be a consequence of pyrrole formation and unsaturated lipids and the protein-free amino groups polymerization (Hidalgo and Zamora, 1993, 1995a; (Belitz and Grosch, 1999; Frankel, 1998; Gardner, 1979; Zamora and Hidalgo, 1994), it has been found that the Montgomery and Day, 1965). It has not been until much formation of color, fluorescence, and pyrrolization fol- more recently that an additional mechanism has also lowed parallel kinetics. In addition, the correlation been proposed, suggesting that a pyrrole formation and among these three measurements was always very high polymerization was taking place and contributed to the independent of the temperature of the reaction, sug- production of both color and fluorescence in these gesting that if pyrrole polymerization is the main reactions (Hidalgo and Zamora, 1993, 1995a; Zamora mechanism responsible for the formation of color and and Hidalgo, 1994). The main problem for testing if fluorescence in these reactions, a very high correlation either of these mechanisms was occurring in foods was among color, fluorescence, and pyrrolization should be the lack of methods that allowed one to follow these expected. polymerization reactions, which might also be influ- Very high correlations were also observed in the enced by the polymerization described for oxidized lipids reaction between linolenic acid and lysine, suggesting to produce brown oxypolymers (Buttkus, 1975; Khayat that pyrrole polymerization was also contributing to the and Schwall, 1983; Venolia and Tappel, 1958) and by nonenzymatic browning produced in this system. Al- Pyrrole Polymerization and Nonenzymatic Browning J. Agric. Food Chem., Vol. 48, No. 8, 2000 3157

the formation of color in the linolenic acid/lysine system or other pyrroles different from those produced in the alkenal/lysine system and with a much higher browning capacity are being produced in the fatty acid/lysine reaction. Both mechanisms described by these hypoth- eses are likely to contribute to the nonenzymatic browning of this system. In fact, most of pyrroles obtained from linolenic acid were different from the pyrroles obtained from the epoxyalkenal because their Ehrlich adducts exhibited different maxima. On the other hand, the linolenic acid/lysine system produced >5 times the fluorescence produced by the epoxyalkenal. Additional studies are needed to isolate the pyrroles produced in the linolenic acid/lysine systems and to investigate how these pyrroles contribute to the formation of color and fluorescence in these reactions. All of these results may also be useful for studying the color and fluorescence produced in the Maillard reaction between carbohydrates and proteins. In these latter reactions, a pyrrole polymerization mechanism, similar to that proposed for the oxidized lipid/protein reactions, has also been proposed (Tressl et al., 1998a,b), which might be competing with the polymerization of Amadori compounds also suggested (Kato and Tsuchida, 1981; Olsson et al., 1981). If pyrrole polymerization is contributing to the development of color and fluorescence in the Maillard reaction, good correlations among color, fluorescence, and pyrrolization should also be found in sugar/protein systems. These studies are being Figure 6. Time course of (A) color, (B) fluorescence, and (C) developed at present in this laboratory. pyrrolization in a reaction of linolenic acid and lysine at 60 °C. ACKNOWLEDGMENT though oxidized fatty acids were previously found to We are indebted to Mr. J. L. Navarro for technical react with the -amino group of lysine residues to assistance. produce both pyrrolized fatty acids and short-chain pyrroles (Hidalgo et al., 1995b,c; Zamora and Hidalgo, LITERATURE CITED 1995), this is the first time that the formation of pyrroles is determined when an unoxidized fatty acid is incu- Ames, J. M.; Bailey, R. G.; Mann, J. Analysis of furanone, bated in the presence of an amino acid. pyranone, and new heterocyclic colored compounds from Additional confirmation that the pyrrole formation sugar-glycine model Maillard systems. J. Agric. Food Chem. 1999, 47, 438-443. and polymerization were taking place was obtained from Belitz, H.-D.; Grosch, W. Food Chemistry, 2nd ed.; Springer: the curves adjusted by using the Boltzmann equation Berlin, Germany, 1999. (eq 2). The width of the curves obtained for the pyr- Buttkus, H. A. Fluorescent Lipid Autoxidation Products. J. rolization of the amino acid was always the smallest of Am. Oil Chem. Soc. 1975, 23, 823-825. the three, suggesting that the formation of pyrroles is Chio, K. S.; Tappel, A. L. Synthesis and characterization of step immediately prior to the formation of color and the fluorescent products derived from malondialdehyde and fluorescence, which is in accordance with the mecha- amino acids. Biochemistry 1969, 8, 2821-2827. nism proposed. In fact, the maximum value of pyrroles CIE (Commission Internationale de l’Eclairage). Recommenda- was attained prior to the maximum value of color or tions on Uniform Color Spaces, Color-Differences Equations, fluorescence, suggesting that pyrrole formation, and Psychometric Color Terms; CIE Publication 15 (Supplement perhaps some polymerization, finished before the maxi- 2); Bureau Central de Commission Internationale de mum color or fluorescence was reached. This maximum l’Eclairage: Paris, France, 1978. Frankel, E. N. Lipid Oxidation; The Oily Press: Dundee, color and fluorescence might be related to an increase Scotland, 1998. in the conjugation, and this might not influence signifi- Friedman, M. Food browning and its prevention: An overview. cantly the formation of Ehrlich adducts. J. Agric. Food Chem. 1996, 44, 631-653. With the data obtained in this study it is not possible Gardner, H. W. Lipid hydroperoxide reactivity with proteins to conclude if pyrrole formation and polymerization are and amino acids: A review. J. Agric. Food Chem. 1979, 27, the only, or even the main, mechanisms involved in the 220-229. development of color and fluorescence in the linolenic Hidalgo, F. J.; Zamora, R. Fluorescent pyrrole products from - acid/lysine system. Considering that the two studied carbonyl-amine reactions. J. Biol. Chem. 1993, 268, 16190 reactions at the two assayed temperatures are compa- 16197. rable, the reaction between 4,5(E)-epoxy-2(E)-heptenal Hidalgo, F. J.; Zamora, R. Characterization of the products ∼ formed during microwave irradiation of the nonenzymatic and lysine produced 15 times more pyrroles than the browning lysine/(E)-4,5-epoxy-(E)-2-heptenal model system. linolenic acid/lysine reaction, but the color difference J. Agric. Food Chem. 1995a, 43, 1023-1028. obtained by the epoxyalkenal was only double than Hidalgo, F. J.; Zamora, R. In vitro production of long chain attained with the fatty acid. This might be a conse- pyrrole fatty esters from carbonyl-amine reactions. J. Lipid quence of either additional mechanisms contributing to Res. 1995b, 36, 725-735. 3158 J. Agric. Food Chem., Vol. 48, No. 8, 2000 Zamora et al.

Hidalgo, F. J.; Zamora, R. Epoxyoxoene fatty esters: Key O’Brien, J., Nursten, H. E., Crabbe, M. J. C., Ames, J. M., Eds. intermediates for the synthesis of long-chain pyrrole and The Maillard Reaction in Foods and Medicine; The Royal furan fatty esters. Chem. Phys. Lipids 1995c, 77,1-11. Society of Chemistry: Cambridge, U.K., 1998. Hidalgo, F. J.; Alaiz, M.; Zamora, R. A spectrophotometric Olsson, K.; Pernemalm, P. A.; Theander, O. Reaction products method for the determination of proteins damaged by and mechanism in some simple model systems. Prog. Food oxidized lipids. Anal. Biochem. 1998, 262, 129-136. Nutr. Sci. 1981, 5,47-55. Hunter, R. S. The Measurement of Appearance; Hunter As- Sapers, G. M. Browning of foods: control by sulfites, antioxi- sociates Laboratory: Fairfax, VA, 1973. dants, and other means. Food Technol. 1993, 47 (10), 75- Hutchings, J. B. Food Colour and Appearance; Blackie: Lon- 84. - don, U.K., 1994; pp 417 433. Tressl, R.; Wondrak, G. T.; Kru¨ ger, R.-P.; Rewicki, D. New Ikan, R., Ed. The Maillard Reaction. Consequences for the melanoidin-like Maillard-polymers from 2-deoxypentoses. J. Chemical and Life Science; Wiley: New York, 1996. Agric. Food Chem. 1998a, 46, 104-110. Kato, H.; Tsuchida, H. Estimation of melanoidin structure by Tressl, R.; Wondrak, G. T.; Garbe, L.-A.; Kru¨ ger, R.-P.; pyrolysis and oxidation. Prog. Food Nutr. Sci. 1981, 5, 147- Rewicki, D. Pentoses and hexoses as sources of new mel- 156. anoidin-like Maillard polymers. J. Agric. Food Chem. 1998b, Khayat, A.; Schwall, D. Lipid oxidation in seafood. Food 46, 1765-1776. Technol. 1983, 37 (7), 130-140. Kikugawa, K.; Ido, Y. Studies on peroxidized lipids. V. Forma- Venolia, A. W.; Tappel, A. L. Brown-colored oxypolymers of - tion and characterization of 1,4-dihydropyridine-3,5-dicar- unsaturated fats. J. Am. Oil Chem. Soc. 1958, 35, 135 baldehydes as model of fluorescent components in lipofuscin. 138. Lipids 1984, 19, 600-608. Zamora, R.; Hidalgo, F. J. Modification of lysine amino groups Ledl, F.; Schleicher, E. New aspects of the Maillard reaction by the lipid peroxidation product 4,5(E)-epoxy-2(E)-heptenal. in foods and in the human body. Angew. Chem., Int. Ed. Lipids 1994, 29, 243-249. Engl. 1990, 29, 565-594. Zamora, R.; Hidalgo, F. J. Linoleic acid oxidation in the Liddell, P. A.; Forsyth, T. P.; Senge, M. O.; Smith, K. M. presence of amino compounds produces pyrroles by carbonyl Chemical synthesis of a GSA-pyrrole and its reaction with amine reactions. Biochim. Biophys. Acta 1995, 1258, 319- Ehrlich’s reagent. Tetrahedron 1993, 49, 1343-1350. 327. Mattocks, A. R. Toxicity of pyrrolizidine alkaloids. Nature Zamora, R.; Rıós, J. J.; Hidalgo, F. J. Formation of volatile 1968, 217, 723-728. pyrrole products from epoxyalkenal/protein reactions. J. Sci. Montgomery, M. W.; Day, E. A. Aldehyde-amine condensation Food Agric. 1994, 66, 543-546. reaction: A possible fate of carbonyls in foods. J. Food Sci. Zamora, R.; Alaiz, M.; Hidalgo, F. J. Determination of -N- 1965, 30, 828-832. pyrrolylnorleucine in fresh food products. J. Agric. Food Nakamura, T.; Hama, Y.; Tanaka, R.; Taira, K.; and Hatate, Chem. 1999, 47, 1942-1947. H. A new red coloration induced by the reaction of oxidized lipids with amino acids. J. Agric. Food Chem. 1998, 46, - 1316 1320. Received for review October 6, 1999. Revised manuscript Namiki, M. Chemistry of Maillard reactions: Recent studies received April 22, 2000. Accepted May 2, 2000. This study was on the browning reaction mechanism and the development supported in part by the Comisioń Interministrial de Ciencia of antioxidants and mutagens. Adv. Food Res. 1988, 32, y Tecnologıá of Spain (Project ALI97-0358) and the Junta de - 115 184. Andalucıá (Project AGR 0135). Narayan, K. A. Biochemical aspects: Nutritional bioavailabil- ity. In Food Storage Stability; Taub, I. A., Singh, R. P., Eds.; CRC Press: Boca Raton, FL, 1998; pp 125-174. JF991090Y 4890 J. Agric. Food Chem. 2000, 48, 4890−4895

Approaches to Wine Aroma: Release of Aroma Compounds from Reactions between Cysteine and Carbonyl Compounds in Wine

Ste´phanie Marchand, Gilles de Revel,* and Alain Bertrand

Faculte´ d’Œnologie, Unite´ Associe´e INRA/Universite´ Victor Segalen Bordeaux 2, 351 Cours de la Libe´ration, F-33405 Talence Cedex, France

Under conditions close to those of wine, that is, low pH, aqueous medium, and low temperatures, this work examines the role of carbonyl (acetoin and acetol) and dicarbonyl (glyoxal, methylglyoxal, diacetyl, and pentane-2,3-dione) compounds associated with cysteine in the formation of odorous products. In particular, thiazole, 4-methylthiazole, 2-acetylthiazole, and trimethyloxazole and two sulfur and oxygenated heterocyclic compounds, 2-furanmethanethiol and thiophene-2-thiol, are examined. For thiophene-2-thiol, the reactional mechanism is proposed. Attempts were made to detect these compounds in wines from various origins. Certain molecules were identified for the first time in wine.

Keywords: Cysteine; carbonyl compounds; Strecker degradation; wine aroma; thiazole; trimethyloxazole; 2-acetylthiazole; thiophene-2-thiol; 2-furanmethanethiol

INTRODUCTION Chart 1. Heterocyclic Compounds Studied Cysteine is one of the most remarkable amino acids from a chemical point of view. The thiol function carried by the side chain is suitable to form dimerous units for the formation of sulfur and disulfur bridges of the tertiary structure of proteins. In musts and wines, cysteine is present in variable amounts and is more or less metabolized by the various microorganisms. We were interested in the specific involvement of cysteine in the genesis of wine flavors. Recently, Tominaga et al. (1998) showed that cysteine could be involved in the varietal flavors of Sauvignon wine. Some flavor compounds were found in the Sauvignon must in the form reactions (Pripis-Nicolau et al., 2000). The present study of S-cysteine conjugate precursors. The work presented led us to identify new odorous compounds from model here deals with the involvement of this amino acid in solutions with cysteine. All of them were known prod- the chemical reactions occurring in wine, more or less ucts of the Maillard reaction. This reaction occurs in early in or during the aging process. These reactions the agrofood industry and leads to roasted food flavors, may contribute to the typical flavor of certain wines. but in our mild conditions its mechanisms are poorly According to Strecker, cysteine degradation leads to known. Among all of the compounds identified in the the formation of small very reactive molecules such as solutions (substituted alkyl- and acetylpyrazines, furanes, hydrogen sulfide, ammonia, or ethanal. These molecules thiophenes, and thiazoles), five of the most abundant of low molecular mass could be reagents in the aromatic and most odorous were identified in wines and their heterocyclic compound formation studied in this work. contents specified: thiazole, 2-acetylthiazole, trimethyl- In addition, the nitrogen-containing function of cysteine oxazole, 2-furanmethanethiol, and thiophene-2-thiol is an electrodonor, which is very reactive toward the (Chart 1). electrophilic carbonyl functions. Many carbonyl and R-dicarbonyl compounds are present in wines; their origins have been studied in our laboratory and their MATERIALS AND METHODS contents specified (de Revel and Bertrand, 1994). The Materials. Cysteine, carbonyl compounds (glyoxal, methyl- R-hydroxyketones and the R-dicarbonyls present char- glyoxal, diacetyl, pentane-2,3-dione, acetoin, acetol, and etha- acteristic and remarkable odors. Previous work in our nal), and compounds used for the identification of the final laboratory showed that some carbonyl compounds placed products alkylpyrazines, thiazoles, alkylthiazoles, 2-acetylthi- in solution with an amino acid reacted, even if kept in azole, 2-acetyl-2-thiazoline, 2-furanmethanethiol, and alde- a reducing medium and under soft conditions similar hydes were purchased from Sigma Aldrich Chemical Co.; to those of in-bottle wine aging (low pH, low tempera- trimethyloxazole, thiophene-2-thiol, and 2,4-dimethylthiazole ture, and aqueous medium). The reaction products were purchased from Lancaster. Inorganic reagents and solvents were all commercial products of analytical grade. The presented odors close to those developed during Maillard mixture of a carbonyl compound and an amino acid in an aqueous ethanolic solution (12% volume), red wine or white * Author to whom correspondence should be addressed (fax wine, was prepared in stoichiometric conditions (20 mM) and 33 5 56846468, e-mail [email protected]). adjusted to pH 3.5 with1NH3PO4 and 1 N NaOH (Pripis- 10.1021/jf000149u CCC: $19.00 © 2000 American Chemical Society Published on Web 09/07/2000 Reaction of Cysteine and Carbonyl Compounds J. Agric. Food Chem., Vol. 48, No. 10, 2000 4891

Table 1. Heterocyclic Compounds in Wines thiazole 4-methylthiazole trimethyloxazole 2-acetylthiazole thiophene-2-thiol Alsace no. of values 5 5 5 5 9 (white wines) range (µg/L) 0-10-10 0-4 0.7-20-4 Burgundy no. of values 8 8 8 8 11 (white wines) range (µg/L) 0-19 0-10-20-70-1 Burgundy no. of values 5 5 5 5 8 (red wines) range (µg/L) 0-70-30-70-0.7 0-0.6 Champagne no. of values 7 7 7 7 8 range (µg/L) 0-23 0-00-50-30-4 Provence/Languedoc no. of values 3 3 3 3 4 (white wines) range (µg/L) 0-10-00-00-0.6 0-0 Provence/Languedoc no. of values 3 3 3 3 4 (red wines) range (µg/L) 0.4-30-0.6 0-00-50-1 Medoc/Graves no. of values 6 6 6 6 7 (red wines) range (µg/L) 0.6-20-60-70-70-0.3 Graves no. of values 2 2 2 2 3 (white wines) range (µg/L) 1.1-1.2 0-00-00-00-2 Pomerol/Saint Emilion no. of values 7 7 7 7 9 (red wines) range (µg/L) 0.9-14 0-00-10-14 0-5 botrytized wines no. of values 4 4 4 4 7 range (µg/L) 0-30-11 0-70-12 0-3 fortified wines no. of values 4 4 4 4 6 range (µg/L) 3-34 0-00-00-13 0-7

Nicolau et al., 2000). The solutions were stored at 25 °C in the initial step lasting 1 min, a rate of 2 °C/min to 100 °C and the dark and under nitrogen during a 4-week storage period. at a rate of 5 °C/min to 220 °C, and the final step lasting 30 Flavor modifications were examined daily, and H2S determi- min. The carrier gas was helium (1.5 mL/min). The injector nation was performed. Heterocyclic compounds were deter- was a splitless system. A quantitative determination of mined after the storage period. To study the effect of temper- thiophene-2-thiol and 2-furanmethanethiol was done in the ature on reactions and flavor modifications, solutions containing SIM mode selecting ions of m/z 71 and 116 for thiophene-2- cysteine and methylglyoxal were stored at 10, 20, and 40 °C. thiol and m/z 81 and 114 for 2-furanmethanethiol, and m/z Analytical Procedures. After the addition of the internal 115 for the internal standard (N,N-diethylacetamide). Quan- standard, each reaction mixture (50 mL) was extracted by titative determination was done by comparison with a stand- various solvents and was analyzed by GC-FPD, GC-NPD, and ard solution containing pure thiophene-2-thiol and pure GC-MS (Pripis-Nicolau et al., 2000). 2-furanmethanethiol at the concentration of 1 µg/L. GC-FPD Analysis. Two gas chromatographs (Hewlett-Pack- HPLC Analysis. Amino acid analysis was carried out by ard) were coupled with a flame photometric detector (FPD). reversed-phase HPLC using a Hewlett-Packard (HP 1050) The least volatile sulfur compounds were determined according liquid chromatograph. Samples were submitted to automatic to the method of Pripis-Nicolau et al. (2000) by an initial gas derivatization with o-phthaldialdehyde (OPA) in the presence chromatograph. H2S was determined by a second gas chromatograph (Hewlett-Packard) also according to the method of of 2-sulfanylethanol and a second derivatization with iodo- Pripis-Nicolau et al. (2000). acetic acid (IDA) for specific determination of sulfur amino GC-NPD Analysis. For the quantitative determination of acids. Solvents and gradient conditions were described by thiazole derivatives, pyrazines, and trimethyloxazole, a gas Anocibar Beloqui (1998). Separations were performed with two chromatograph (Hewlett-Packard) was coupled with an NPD octadecyl Lichrocart cartridges mounted in series containing detector; separation was carried out with an HP5 column (50 an RP 18 Lichrospher column and the same type of precolumn. m × 0.32 mm, 0.52 µm). The oven temperature was pro- Detection was done by a fluorometric detector (Jasco-821-FP) ) ) grammed from 60 to 200 °C at a rate of 2 °C/min. The final at λex 356 and λem 445, and the data were acquired on an isothermal time was 20 min. The carrier gas was helium U, HP Chemstation. R-Dicarbonyl compound analysis was carried the splitless time was 20 s, and the split vent was 30 mL/min. out by a type C18 phase HPLC using a Varian 5000 liquid The internal standard was N,N-diethylacetamide. Calibration chromatograph as described by de Revel et al. (2000). Samples curves were established by analysis of five different concentra- were derived with o-diaminobenzene at pH 8. For wine tions of pure products between 0.1 and 100 µg/L. analysis the quinoxalines were extracted by dichloromethane GC-MS Analysis. A first gas chromatograph (Hewlett- (2 × 5 mL, 5 min) at pH 2. After drying and the addition of 5 Packard) was coupled with a mass spectrometer (HP 5972; mL of methanol, 20 µL was injected. The quinoxalines were electronic impact, 70 eV; eMV, 2.7 kV). The column was a BP detected by spectrophotometry at λ ) 313 nm. 21 (SGE), (50 m × 0.25 mm, 0.25 µm). The oven temperature Determination of Odor Descriptors. These were gener- was programmed from 40 to 220 °C at a rate of 2 °C/min, the ated by four trained tasters (researchers from the laboratory) initial step lasting 1 min and the final step lasting 20 min. by smelling standard solutions in different dilutions in glasses The carrier gas was helium (1.5 mL/min). The injector was a (AFNOR). A data bank of odors was constituted. The odors of splitless system: the splitless time was 20 s and split vent 30 the model solutions at different times were compared to the mL/min. A qualitative determination of 2-acetylthiazole, data bank. 2-acetyl-2-thiazoline, and 2-acetylthiazolidine was done in the Determination of Olfactory Perception Thresholds. selected ion monitoring (SIM) mode selecting ions of m/z 89, 103, 112, 127, and 129; trimethyloxazole ions of m/z 111 and For each compound detected in wine, thiazole, 2-acetylthiazole, 139; thiazole, 2-methylthiazole, 2,4-dimethylthiazole, and 2,5- trimethyloxazole, 2-furanmethanethiol, and thiophene-2-thiol, dimethylthiazole ions of m/z 58, 85, 99, 113, and 114; and the odor threshold was determined as the minimum concen- pyrazine, methylpyrazine, 2,5-dimethylpyrazine, trimethyl- tration below which 50% of the tasters failed to smell the pyrazine, and tetramethylpyrazine ions of m/z 80, 81, 94, 108, difference from the control by the triangle test at five concen- 122, and 136. For the other heterocyclic compounds, a second trations in mineral water. Smelling of the solutions placed in gas chromatograph (Hewlett-Packard) was coupled with a glasses corresponding to AFNOR standards was done by a 20- mass spectrometer (HP 6890; electronic impact, 70 eV; eMV, person jury (Boidron et al., 1988). 2.7 kV). The column was a BP1 (50 m × 0.32 mm, 0.2 µm). Wines. Seventy-two wines from different regions of France The oven temperature was programmed from 40 to 220 °C, and Europe were analyzed and are shown in Table 1. 4892 J. Agric. Food Chem., Vol. 48, No. 10, 2000 Marchand et al.

RESULTS AND DISCUSSION Study of New Odorous Aromatic Heterocyclic Compounds. The model solutions containing cysteine and carbonyl compounds developed odors that can be defined according to three families of descriptors: odors of the type “popcorn” or “roasted”; odors of the type “herbaceous” or “potato”; and odors of the type “sulfur” or “rotten eggs”. In addition, the odors developed by the solutions with an R-dicarbonyl compound (glyoxal, methylglyoxal, diacetyl, and pentane-2,3-dione) were always more intense and complex than those developed in the presence of ethanal or a hydroxy ketone (acetoin and acetol). The study of the model solution composition after 4 weeks of reaction led to the identification of many nitrogen-containing, sulfur-containing, and oxygenated heterocyclic compounds. Alkylpyrazines were identified in the solutions of cysteine and dicarbonyl compounds. In the presence of methylglyoxal, there appeared 2,5-dimethylpyrazine and 2,6-dimethylpyrazine: the first has an “earthy” or “raw potato” odor (Chastrette et al., 1997); the second has a “roast” or “chocolate” odor (Maga, 1982). Tri- Figure 1. (A) Relationship between cysteine and thiazole in methylpyrazine, which has a “roasted hazelnut” odor Pomerol and Saint-Emilion wines. (B) Relationship between (Chastrette et al., 1997), was encountered with the cysteine and thiophene-2-thiol in Champagne. addition of diacetyl. 5-Diethyl-3,6-dimethylpyrazine and of the solutions containing cysteine were concerned. This 2,6-diethyl-3,5-dimethylpyrazine, which have odors of molecule is characterized by a persistent “roasted “roast” and “nut” (Maga, 1982), were identified in the hazelnut” odor, and its threshold in water was 3 µg/L. solutions of pentane-2,3-dione. Koehler et al. (1971) The wines from Pomerol and Saint-Emilion presented determined the odor threshold for smelling alkylpyr- higher 2-acetylthiazole levels on average than the other azines, which vary from 10 to 100 mg/L. Among these, wines studied (Table 1). Nine wines from these regions 2,6-dimethylpyrazine has a threshold of 54 mg/L and were analyzed; they contained >3 µg/L on average, tetramethylpyrazine a threshold of 10 mg/L. Thus, the which is the odor threshold value in water. White direct influence of these molecules on wine flavor is Burgundy wines, Champagne, and Alsace contained on undoubtedly weak, so they were not studied further in average between 1.4 and 1.8 µg/L of 2-acetylthiazole. this work. On the other hand, red wines from Me´doc, Burgundy, Methylthiazoles. In the model solutions containing and Provence wines and fortified wines had <1 µg/L of R-diketone (diacetyl or pentane-2,3-dione) and cysteine, 2-acetylthiazole. This molecule was identified for the there appeared various methylthiazoles, in particular first time in wine. The frequency of occurrence of 4-methylthiazole, 2,4-dimethylthiazole, and 2,5-dimeth- 2-acetylthiazole and its concentrations were variable, ylthiazole. 4-Methylthiazole has a “green hazelnut” odor, but it is obvious that among the nitrogen-containing its threshold being 55 µg/L. 2,4-Dimethylthiazole has heterocyclic compounds, it was the molecule most “oxidized beer”, “roasted red meat”, or “coffee” odors. 2,5- frequently observed at concentrations with olfactory Dimethylthiazole has “meat” or “hazelnut” odors. These activity values (OAV is the quotient of the concentration molecules could not be detected in the wines. divided by the odor threshold in water) >1. This Thiazole was formed in all of the solutions including molecule could play a role in wine flavor. Small quanti- glyoxal or methylglyoxal. Thiazole has an odor of ties of 2-acetyl-2-thiazoline were detected in the wines “popcorn” and “peanut”. Its odor threshold was 38 µg/ richest in 2-acetylthiazole, which thus strengthened the L. Therefore, this molecule could be detected in the assumption that 2-acetyl-2-thiazoline could be the re- wines. The wines from Pomerol and Saint-Emilion duced precursor of 2-acetylthiazole. A mode of formation showed some correlation (R ) 0.7) between thiazole and was proposed by Griffith and Hammond (1988) for the cysteine levels (Figure 1). The maximum concentration origin of the aromatic components of Swiss cheese. encountered was 14 µg/L for a Pomerol wine of 15 years Trimethyloxazole, which was detected in all of the of age (Table 1); however, these compounds rarely R-diketone solutions, has an aggressive “very ripe fruit” exceeded 4 µg/L for wines of >10 years age. In addition, odor. Its odor threshold in water was 17 µg/L. Tri- the fortified wines such as Port or Banyuls had high methyloxazole was the only non-sulfur heterocyclic levels of thiazole, up to 34 µg/L for a 40-year-old Port compound that we found in wines. It is most frequently bottled in 1968 and 6 µg/L in old Banyuls (Hors d’Age). met in wines from the Pessac-Le´ognan, Bordeaux, and On the other hand, these wines after a long aging Me´doc regions (Table 2). All of these vineyards are presented very low levels in cysteine: 0.5 mg/L in located in the same geographical area, and the wines Banyuls and 0 mg/L in Port wine. These wines have are vinted according to very similar processes. Sixty- specific manufacturing and aging processes (fortified seven percent of the wines resulting from this area with alcohol, high sugar concentration, and long and contained between 2 and 7 µg/L of trimethyloxazole. oxidative aging) that could explain the contents encoun- However, trimethyloxazole was detected in some red tered. Moreover, the wine odor impact could be impor- and white wines from other areas (Table 1): 38% of the tant in Tawny Port. Burgundy wines analyzed contained between 1.5 and 3 2-Acetylthiazole was one of the molecules synthesized µg/L. Finally, 4 µg/L of trimethyloxazole was found in a in the greatest amounts in the model solutions, and all late harvest 1995 wine from the Alsace region and 1 Reaction of Cysteine and Carbonyl Compounds J. Agric. Food Chem., Vol. 48, No. 10, 2000 4893

Table 2. Trimethyloxazole Levels in Different Wines Scheme 1. Hypothesis for Formation of Thiophene- 2-thiol from Cysteine (Ho, 1996) wine origin trimethyloxazole (µg/L) red wines Saint Julien 1994 7 Saint Julien 1994 2 Pessac-Le´ognan 1994 6 Bordeaux 1994 0 Saint Este`phe 1994 4 white wines Bordeaux 1983 0 Barsac 1983 7 Graves 1996 0 Pessac-Le´ognan 1996 2 Table 3. Influence of Temperature on Cysteine/ µg/L in a Saint-Emilion wine (great vintage of 1994) and Methylglyoxal Model Solution Odorsa in a Champagne-rose´ wine. The other wines analyzed did not contain trimethyloxazole. This molecule has days never been mentioned before in wine. This oxygenated temp, °C odor 2 5 7 9 11 15 and nitrogen-containing heterocyclic compound was 10 sulfurous * ** ** **** *** ** described by Vernin and Metzger (1981) as the product roasted * * ** ** *** R of the condensation of -aminoketone, resulting from 20 herbaceous ***** the degradation of an amino acid according to the sulfurous *** *** *** *** **** *** Strecker mechanism, and from acetaldehyde. Acetalde- roasted * * * hyde (or ethanal) can be abundant in wine (up to 100 pop corn ** mg/L), although it is in the form of a bisulfite combina- 40 herbaceous ** tion. sulfurous *** *** *** *** *** 2-Furanmethanethiol was detected only in R-diketone roasted ** ** ** * * solutions and only when they were kept at 40 °C. Its carmel * organoleptic descriptor is “roasted coffee” or “burned a *, weak; **, moderate; ***, intense; ****, very intense. rubber” at high levels. The organoleptic importance of 2-furanmethanethiol could be very great because of its This molecule was one of the most abundant products odor threshold, estimated at 1 ng/L. According to our when experimentation was conducted at pH 2.2; in tests, only the wines from Pomerol and Saint-Emilion, addition, its formation is favored by the presence of the Crus Classe´s of Burgundy wines, and some Cham- carbonyl compounds. Ho (1996) proposed a protocol of pagnes presented 2-furanmethanethiol (Table 1). The cysteine decomposition leading to acetaldehyde and average contents reached 350 ng/L for the wines of 2-sulfanylethanal. By condensation of two aldehydes, Pomerol and Saint-Emilion, 70 ng/L for the wines of 4-sulfanylbuten-2-al was formed, which, in the presence Burgundy, and 10 ng/L for Champagnes. These contents of hydrogen sulfide, could lead to thiophene-2-thiol by mean that the OAV index is 350, 70, and 10, respec- cyclization (Scheme 1). This sulfur heterocyclic com- tively, for these three types of wines. We could not detect pound was identified in wine for the first time. The 2-furanmethanethiol in any other wines. Schieberle search for the origin of these molecules is underway in (1993) defined 2-furanmethanethiol as the product of our laboratory. reaction between sugars or derivatives of sugars and All of the molecules studied in this work are known cysteine or hydrogen sulfide. This furan-type sulfur to be products of the Maillard reaction (Shibamoto and molecule was also cited in wine by Blanchard et al. Russell, 1976; Farmer and Mottram, 1989; Schieberle (1999), who attributed its presence in wine to oak wood. and Hofmann, 1996). They all present odorous notes Thiophene-2-thiol was observed in the solutions with close to those developed in roasted food. The presence each one of the carbonyl compounds studied and when of thiazoles, pyrazines, and trimethyloxazole has al- cysteine was kept at 25 °C. This sulfur aromatic ready been reported by Pripis-Nicolau et al. (2000) as heterocyclic compound is very odorous. Its threshold in reactions between amino acids and carbonyl compounds water was ∼0.8 µg/L. It has an odor of “burned”, “burned under wine conditions. The heterocyclic compounds rubber”, or “roasted coffee”. It was detected in wines for detected in model solutions could also exist in wine. The the first time. Champagne and some Pomerol and Saint- contents of seven compounds, that is, thiazole, 4-meth- Emilion wines were among the richest in thiophene-2- ylthiazole, 2-acetylthiazole, 2-acetyl-2-thiazoline, tri- thiol. It was detected at levels of 1.2 µg/L for Pomerol methyloxazole, 2-furanmethanethiol, and thiophene-2- and Saint-Emilion wines and 1 µg/L for Champagne thiol (Scheme 1), were specified for wines from various wines, thus giving OAV indices of 1.7 and 1.4, respec- areas of France and Europe, with various types of tively. Aging of wines from Pomerol and Saint Emilion cultivars and which are vinted according to various is done in oak barrels. Thus, like 2-furanmethanethiol, methods (Table 1). thiophene-2-thiol could be released by wood in still Temperature Influence. Model solutions with cys- wines. Eight Champagne wines were analyzed. For five teine and methylglyoxal were kept at 10, 20, or 40 °C. wines, the thiophene-2-thiol levels were bound to the The notes developed are shown in Table 3. At 10 °C, cysteine contents (Figure 1B); three others contained there was a regular development of the sulfur notes, no or very low levels of thiophene-2-thiol despite cys- arising from the presence of hydrogen sulfide due to the teine concentrations ranging between 2 and 4 mg/L. The decomposition of cysteine according to the Strecker cysteine concentration of the wines could be a determin- mechanism. At the same time, weak “roasted” notes ing factor for this molecule. Thiophene-2-thiol was appeared. However, the odors remained fairly simple. described by Shu et al. (1985) in a study of the pH effect At 20 °C, the “hydrogen sulfide” note was great on the on thermal degradation of cysteine in aqueous solution. second day of reaction, but its intensity increased very 4894 J. Agric. Food Chem., Vol. 48, No. 10, 2000 Marchand et al.

wine, and model solution) over the same reaction time did not reveal any notable difference. Nevertheless, the red wine solutions contained lower quantities of hydrogen sulfide in their headspace than the other solutions. This was perhaps due to combinations. This work confirms the high reactivity of cysteine with carbonyl and dicarbonyl molecules and its particularly interesting role in the genesis of flavors. We now show the presence of thiazole, 4-methylthiazole, 2-acetylthiazole, and trimethyloxazole and that of two sulfured and oxygenated heterocyclic compounds, 2-furanmethanethiol and thiophene-2-thiol. These molecules could play an important part because of their very low olfactory thresholds. Hydrogen sulfide is an intermediate in the formation of these odorous heterocyclic compounds. Molecules with “roasted coffee”, “roasted hazelnut”, “popcorn”, “burned”, and “roasted” notes were found in various wines. Among these, we found 2-acetyl- Figure 2. Evolution of H2S in the headspace of cysteine/ thiazole in many wines, thiazole particularly in Pomerol methylglyoxal solutions stored at 10, 20, and 40 °C. and Saint-Emilion wines and in fortified wines, trimethyloxazole in Graves and Medoc wines, 2-furan- little with time. Figure 2 shows the daily results of H2S methanethiol in some Champagne and in the great contained in the headspace. At 20 °C, the H2S level vintages of Burgundy, and thiophene-2-thiol particu- tended to stabilize. On the other hand, at 10 °C, it larly in Champagne, Pomerol, and Saint-Emilion wines. continued to increase. Part of the hydrogen sulfide In some cases there seemed to be a correlation between produced at 20 °C might be reused in chemical reactions cysteine and thiophene-2-thiol levels in Champagne which could induce the “roasted”, “popcorn”, and “her- wines and cysteine and thiazole in Pomerol and Saint- baceous” notes appearing at this temperature. At 40 °C, Emilion wines. These molecules described for the first the intensity of the “hydrogen sulfide” note was equiva- time in wines could play a great part in their flavor. lent and even lower than the perceived intensity at 20 °C. Moreover, H2S quantified in the headspace con- LITERATURE CITED firmed these observations. However, at 40 °C, the balance between the liquid and vapor phases shifted in Anocibar Beloqui, A. Contribution to the study of sulfur favor of the latter. Therefore, by supposing that cysteine compounds of the red wines. Thesis 611, Universite´ Victor Segalen Bordeaux 2, France, 1998. produced at least as much hydrogen sulfide at 40 °C as Blanchard, L.; Darriet, P.; Bouchilloux, P.; Tominaga, T.; at other temperatures, the quantity of hydrogen sulfide Dubourdieu, D. Caracte´risation de la fraction volatile de measured in the headspace should be greater, yet the nature soufre´e dans les vins de Cabernet Sauvignon et de opposite was observed. Thus, consumption of hydrogen Merlot. Etude de son e´volution au cours de l’e´levage en sulfide by chemical reaction is greater at 40 °C than at barriques. In Œnologie, Proceedings of the 6th Symposium 20 °C. Figure 3 shows that on the second day, the International d’Œnologie, June 10-12, 1999; Lonvaud- hydrogen sulfide content was greater in the solution Funel, A., Coord.; Tec & Doc Ed.: Bordeaux, France, 1999; kept at the highest temperature. The phenomenon was pp 501-505. reversed when the reaction time increased. In addition, Boidron, J. N.; Chatonnet, P.; Pons, M. Influence du bois sur “caramel” and “popcorn” notes appeared in the solutions certaines substances odorantes des vins. Connaiss. Vigne Vin 1988, 22, 275-294. at 20 and 40 °C. These odors are close to those developed Chastrette, M.; El Aı¨di, C.; Cretin, D. Structure-odour rela- during the roasting of food. Hydrogen sulfide could be tionship for bell-pepper, green and nutty notes in pyrazines a breakdown product of cysteine and a reagent in the and thiazoles. Comparison between neural networks and formation of the odorous compounds responsible for the similarity searching. SAR QSAR Environ. Res. 1997, 7, “roasted” “popcorn”, and “caramel” notes. 233-258. Influence of Medium. A comparative sensory study de Revel, G.; Bertrand, A. Dicarbonyl compounds and their of the model solutions and white or red wines was reduction products in wine. Identification of wine aldehydes. carried out. The influence of the matrix seemed to be In Trends in Flavour Research, Proceedings of the 7th - rather low on the variety of the nuances perceived but Weurman Flavour Research Symposium, June 15 18, 1993; Maarse H., van der Heij D. G., Eds.; Elsevier: Zeist, The was quite high on their intensity. In wine, the sensory Netherlands, 1994; pp 353-361. threshold of the molecules responsible for the “roasted” de Revel, G.; Pripis-Nicolau, L.; Barbe J. C.; Bertrand A. The and “popcorn” notes could even be higher. This phe- detection of R-dicarbonyls compounds in wine by formation nomenon was noted with most of the molecules. In white of quinoxaline derivatives. J. Sci. Food Agric. 2000, 80, 102- wine, the “popcorn” and “roasted” notes described during 108. the Maillard reactions developed more quickly than in Farmer, L. J.; Mottram, D. S.; Whitfield, F. B. Volatiles com- red wine. In the case of pentane-2,3-dione solutions, the pounds produced in Maillard reactions involving cysteine, “popcorn” note, which was well identified previously in ribose and phospholipid. J. Sci. Food Agric. 1989, 49, 347- the synthetic medium, was perceptible in the white wine 368. supplemented as of the third day, whereas it was Griffith, R.; Hammond, G. Generation of Swiss cheese flavor > components by the reaction of amino acids with carbonyl necessary to wait 6 days in red wine. When the compounds. J. Dairy Sci. 1988, 72, 604-613. reaction time increased, the odors developed in the Ho, C.-T. Thermal generation of Maillard Aromas. In The white and red wine solutions were equivalent. Simul- Maillard Reaction Consequences for the Chemical and Life taneous analysis of the “light” sulfur compounds of the Sciences; Ikan, R., Ed.; Wiley: Chichester, U.K., 1996; pp various mixtures in the three media (white wine, red 27-53. Reaction of Cysteine and Carbonyl Compounds J. Agric. Food Chem., Vol. 48, No. 10, 2000 4895

Koehler, P. E.; Mason, M. E.; Odell, G. V. Odor threshold levels Shibamoto, T.; Russell, G. F. Study of meat volatiles associated of pyrazine compounds and assessment of their role in the with aroma generated in a D-glucose/H2S/ammonia model flavor of roasted food. J. Food Sci. 1971,36, 816-818. system. J. Agric. Food Chem. 1976, 24, 843-846. Maga, J. A. Pyrazine in foods: an update. CRC Crit. Rev. Food Shu, C. K.; Hagedorn, M. L.; Mookherjee, B. D.; Ho, C. T. pH Sci. Nutr. 1982, 16,1-48. effect on the volatile components in the thermal degradation Pripis-Nicolau, L.; de Revel, G.; Bertrand, A. Formation of of cysteine. J. Agric. Food Chem. 1985, 33, 442-446. flavor components by the reaction of R-amino acids and Tominaga, T.; Peyrot des Gachons, C.; Dubourdieu, D. A new carbonyls compounds in mild conditions. J. Agric. Food type of flavor precursors in Vitis vinifera L. cv. Sauvignon Chem. 2000, 48, 3761-3766. Blanc: S-cysteine conjugates. J. Agric. Food Chem. 1998, Schieberle, P. Studies on the flavour of roasted white sesame. 46, 5215-5219. In Progresses in Flavour Precursor Studies; Schreier, P., Vernin, G.; Metzger, J. La chimie des aroˆmes: les he´te´rocycles. Winterhalter, P., Eds.; Allured Publishing: Carol Stream, Bull. Soc. Chim. Belg. 1981, 90, 553-587. IL, 1993; pp 343-360. Schieberle, P.; Hofmann, T. Identification of the Key Odorants in Processed Ribose-Cysteine Maillard Mixtures by Instru- Received for review February 1, 2000. Revised manuscript mental Analysis and Sensory Studies; RSC Special Publica- received June 8, 2000. Accepted June 16, 2000. tion; Royal Society of Chemistry: London, U.K., 1996; pp 175-181. JF000149U Food Chemistry 132 (2012) 1316–1323

Contents lists available at SciVerse ScienceDirect

Food Chemistry

journal homepage: www.elsevier.com/locate/foodchem

Mechanism of formation of sulphur aroma compounds from L-ascorbic acid and L-cysteine during the Maillard reaction ⇑ Ai-Nong Yu , Zhi-Wei Tan, Fa-Song Wang

School of Chemistry & Environmental Engineering, Hubei University for Nationalities, Enshi, Hubei 445000, China article info abstract

13 13 Table 1 and L-[4- C] ascorbic acid (99 atom-% C) were from Omicron Bio- Model reactions. chemicals, Inc. (South Bend, IN). Cys was from Shanghai Yuanju Biolog-

Table 2 2.2. Degradation of Cys Degradation products of L-cysteine at pH8.

OH OH OH OH O O O O HOH2CHC O Feather, 1993 Rizzi, 2005 Barham, et al., 2010 or

HO OH OH OH O OH O OH OH

L-[1,4-13C]Ascorbic acid Aldopentose 1-Deoxypentosone Acetol Acetol

Rizzi, 2005 Rizzi, 2005 O Rizzi, 2005 OH O O O

Acetaldehyde O O O O Butadione Hydroxybutadione Pyruvaldehyde Pyruvaldehyde

a LRI calculated for a DB-5 capillary column; mean values. b Co-injection with authentic sample. A.-N. Yu et al. / Food Chemistry 132 (2012) 1316–1323 1319

O O

S S S S S S S

1a 1b 2 3a 3b 4a 4b S

S S S S S S

5a O 5b O 5c O 6aO 6b O 7

N N N N N N N

S S S S S S S

8a 8b 9a 9b 9c 10a O 10b O

OH OH OH A

-2H H2S -2H2O OH S S S OH H+ O OH O H 2 1b O H Red. O

H+ + H2S O O O Formaldehyde Pyruvaldehyde OH O O H

S OH H2S -H2O OH S S OH H+ H OH OH S S S

-2H2O -2H

S S S 7

O O B

Red. H2S H+ -2H2O S HO S H O O OH O O O S H+ 3b H O OH O O O OH O O Red. H2S -H2O Acetaldehyde Pyruvaldehyde OH S S OH H+ H 4b

O O O O C O H O

Red. H2S -H2O Red. S -H O S OH O OH HO SH HO 2 O O O O O O O 5b Pyruvaldehyde

OH OH O O O OH O OH O

H -H2O Red. H2S S -H2O S O OH OHHO OHHS O 6a O Hydroxybutadione Hydroxyacetone O O

13 13 Fig. 3. Proposed formation pathway for thiophenes from L-[4- C] ascorbic acid and L-Cysteine (d = C). Compound nos. correspond to Fig. 2.

NH3 HCHO CH2=NH A O O O OH HN H2N H2S CH2=NH CH CH SH 2 2 HO S HO S O O H Butadione OH

NH N N N -H2O -H2O CH 2 HO HO HO S S S S 8a

O O

H2S O NH N O SH 2 N OH -H2O Butadione S O S S NH OH OH NH3 9b H H Acetaldehyde

O B O H SH SH O S NH N O CH2 CH2 -CO2 -4H Pyruvaldehyde CH2 CH NH2 CH N S S -H2O CH2 N COOH O O COOH 10a

OH OH OH OH OH O Red. H2S H2SO -H2O, Red. H S S S S O O OH OH OH SH SH S S Hydroxyacetone 11 and 12

Bart Vanderhaegen *, Hedwig Neven, Hubert Verachtert, Guy Derdelinckx

Centre for Malting and Brewing Science, Katholieke Universiteit Leuven, Kasteelpark Arenberg 22, B-3001 Heverlee, Belgium

Received 1 November 2004; received in revised form 4 January 2005; accepted 4 January 2005

Abstract

Currently, the main quality problem of beer is the change of its chemical composition during storage, which alters the sensory properties. A variety of flavours may arise, depending on the beer type and the storage conditions. In contrast to some wines, beer aging is usually considered negative for flavour quality. The main focus of research on beer aging has been the study of the cardboard-flavoured component (E)-2-nonenal and its formation by lipid oxidation. Other stale flavours are less described, but may be at least as important for the overall sensory impression of aged beer. Their origin has been increasingly investigated in recent years. This review summarizes current knowledge about the chemical origin of various aging flavours and the reaction mechanisms responsible for their formation. Furthermore, the relationship between the production process and beer flavour stability is discussed. Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: Beer; Staling; Aging; Flavours; Brewing

1. Introduction ers have in recognizing the flavour of just the particular brand of beer that they generally drink. To meet the con- As for other food products, also for beer, several qual- sumers expectations, the flavour of a certain beer brand ity aspects may be subject to changes during storage. Beer must be constant. However, as the expected flavour is nor- shelf-life is mostly determined by its microbiological, col- mally the flavour of the particular fresh beer, as a result of loidal, foam, colour and flavour stabilities. In the past, the beer aging, such flavour may change, and the expected fla- appearance of hazes and the growth of beer spoilage vour is lost. This should mainly be considered as the most micro-organisms were considered as the main trouble- important reason that beer staling is undesirable. causing phenomena. However, with progress in the field Starting from the 1960s, several studies have focussed of brewing chemistry and technology, these problems on the chemical aspects of beer staling. Notwithstanding are now largely under good control. Most of the interest 30–40 years of research, beer aging remains difficult to has shifted to factors affecting the changes in beer aroma control. With the increasing export of beer, due to mar- and taste, as beer flavour is regarded as the most impor- ket globalisation, shelf-life problems may become extre- tant quality parameter of the product. However, bearing mely important issues for some breweries. Beer aging is in mind that de gustibus et colouribus non est disputandum, a very complex phenomenon. This overview on the consumers do not necessarily dislike the flavour of an chemistry of beer aging intends to illustrate the complex- aged beer. Indeed, a study (Stephenson & Bamforth, ity of the aging reactions. 2002) with consumer trials pointed out that aging flavours are not always regarded as off-flavours. More important for appreciation of a beer were the expectations consum- 2. Sensory changes in beer during storage

* Corresponding author. Tel.: +32 16321460; fax: +32 16321576. E-mail address: [email protected] (B. Van- The literature on beer staling reveals only few reports derhaegen). dealing with the actual sensory changes during beer

tial burnt flavours in dark beers may mask the develop- Bitter taste Ribes ment of aging flavours and result in a better flavour stability of this beer type. However, as will be explained Sweet aroma further on, other factors probably also account for this observation. Contact of beer with oxygen causes a rapid deterioration of the flavour and the type of flavour changes depends on the oxygen content of bottled beer. For Intensity instance, there is a close correlation between the ribes odour and headspace air, and this flavour can be Sweet taste & toffee-like aroma and flavor avoided in the absence of excessive contact with air (Clapperton, 1976). Furthermore, it is found that beer Cardboard flavor staling still occurs at oxygen levels as low as possible (Bamforth, 1999b), which suggests that beer staling is Time partly a non-oxidative process. Fig. 1. Sensory changes during beer aging according to Dalgliesh Apart from oxygen concentration, storage tempera- (1977). ture affects the aging characteristics of beer, by affecting the many chemical reactions involved. The reaction rate storage. Dalgliesh (1977) described the changes in the increase for a certain temperature increase depends on most detail. However, the Dalgliesh plot (Fig. 1)isa the reaction activation energy. This activation energy generalization of the sensory evolution during beer stor- differs with the reaction type, which means that the rates age and is by no means applicable to every beer. A con- of different reactions do not equally increase with stant decrease in bitterness is observed during aging. increasing temperature. Consequently, beer storage at This is partly due to sensory masking by an increasing different temperatures does not generate the same rela- sweet taste. In contrast to an initial acceleration of sweet tive level increase of staling compounds. Some sensory aroma development, the formation of caramel, burnt- studies confirm this prediction. According to Furusho sugar and toffee-like aromas (also called leathery) coin- et al. (1999), cardboard flavour shows different time cides with the sweet taste increase. Furthermore, a very courses during lager beer storage at 20 and 30 °C. In rapid formation of what is described as ribes flavour is the early phase of beer aging, this results in a sensory observed. The term ribes refers to the characteristic pattern with relatively more cardboard character when odour of blackcurrant leaves (Ribes nigrum). After- beer is stored at 30 °C compared to 20 °C. This agrees wards, the intensity of the ribes flavour decreases. with the findings of Kaneda, Kobayashi, Furusho, Sa- According to Dalgliesh (1977), cardboard flavour devel- hara, and Koshino (1995b) that lager beer aged at 25 ops after the ribes aroma. On the other hand, according °C tends to develop a predominantly caramel character to Meilgaard (1972), cardboard flavour constantly in- whereas, at 30 or 37 °C, more cardboard notes are creases to reach a maximum, followed by a decrease. Be- dominant. sides these general findings, other reported changes in From these examples, it follows that the Dalgliesh flavour are harsh after-bitter and astringent notes in plot (Fig. 1) is a simplification of the sensory changes taste (Lewis, Pangborn, & Tanno, 1974) and wine- and during storage. The nature of flavour changes is com- whiskey-like notes in strongly aged beer (Drost, Van plex and mainly depends on the beer type, the oxygen Eerde, Hoekstra, & Strating, 1971). Positive flavour concentration and the storage temperature. attributes of beer, such as fruity/estery and floral aroma tend to decrease in intensity. For the overall impression, the decrease of positive flavours may be just as impor- 3. Chemical changes in beer during storage tant as development of stale flavours (Bamforth, 1999b; Whitear, Carr, Crabb, & Jacques, 1979). 3.1. General Often beer staling is presented as just being related to cardboard flavour development. While, in some cases, Flavour deterioration is the result of both formation and especially in lager beers, cardboard flavour is the and degradation reactions. Formation of molecules, at major manifestation of beer staling, this can not be gen- concentrations above their respective flavour threshold eralized. Aging flavours vary between beer types and leads, to new noticeable effects, while degradation of certainly, for speciality beers, other stale flavours are of- molecules to concentrations below the flavour threshold ten more prominent. Whitear (1981) reported aging may cause loss of initial fresh beer flavours. Further- notes of a strong ale as burnt, alcoholic, caramel, liquo- more, interactions between different aroma volatiles rice and astringent flavours, whereas cardboard and may enhance or suppress the flavour impact of the mol- metallic flavours were not found. Moreover, strong ini- ecules (Meilgaard, 1975a). B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 359

3.2. Volatile compounds search. In the following years, other studies (Drost et al., 1971; Meilgaard, Ayma, & Ruano, 1971; Wohleb, Jen- 3.2.1. Analysis nings, & Lewis, 1972) confirmed the results, but all re- With the introduction of gas chromatography in the ferred to heated and acidified (pH 2) beer. Such 1960s, it became possible to study the changes in beer extreme storage conditions were initially used to obtain volatiles during storage. In the late 1960s, several studies detectable levels, as research on beer carbonyls is com- (Ahrenst-Larsen & Levin Hansen, 1963; Engan, 1969; plicated due the extremely low levels at which many of Jamieson, Chen, & Van Gheluwe, 1969; Trachman & these compounds occur. However, it is questionable Saletan, 1969; von Szilvinyi & Pu¨spo¨k, 1969) reported whether the results are representative of real storage the formation of staling-related compounds. Table 1 conditions. In general, it remains important that steps shows a classification of the volatiles currently known in the analytical procedure are avoided, which might al- as being related to concentration changes during beer ter or form compounds of interest. aging. Direct analysis by gas chromatography of either a In recent years, new techniques, such as aroma headspace or a solvent extract of non-treated beer is extraction dilution analysis (AEDA), have been devel- not applicable because other, more abundant, volatiles oped to evaluate the relevance of detected volatiles to frequently obscure the carbonyl compound peaks. Wang odour perception in foods. (Belitz & Grosch, 1999). and Siebert (1974) first developed a method to follow the Using this method, several staling compounds have been (E)-2-nonenal concentration increase under more nor- identified in beer (Gijs, Chevance, Jerkovic, & Collin, mal storage conditions (6 days at 38 °C). The technique 2002; Schieberle, 1991; Schieberle & Komarek, 2002). was based on extraction of beer with dichlorometh- In this technique, a flavour extract of beer is sequentially ane, followed by derivatisation of (E)-2-nonenal with diluted and each dilution is analyzed by GC–O (gas 2,4-dinitrophenylhydrazine (DNPH) under acidic condi- chromatography/olfactometry) by a small number of tions. The treated beer extract was subjected to separa- judges. The extraction method is very important, as it tion and concentration steps by means of column and is essential to ensure that extracts with an odour repre- thin-layer chromatography, and finally analysed by high sentative of the original product are obtained. The fla- performance liquid chromatography. With this method, vour dilution (FD) of an odorant corresponds to the there was a concentration increase in levels of (E)-2-non- maximum dilution at which that odorant can be per- enal to levels above the flavour threshold of 0.1 lg/L. In ceived by at least one of the judges. Consequently, the other studies (Greenhoff & Wheeler, 1981a; Greenhoff & FD factors give an estimation of the importance of vol- Wheeler, 1981b; Hashimoto & Eshima, 1977; Jamieson atiles for the perceived flavour of a beer sample. The & Chen, 1972; Stenroos, Wang, Siebert, & Meilgaard, method should be regarded as a first step in the screen- 1976) on aldehydes in beer, similar analysis techniques, ing for staling compounds and not to obtain conclusive based on carbonyl 2,4-dinitrophenylhydrazone forma- results about the relevance of flavour compounds. tion, were used, and although isolation techniques were usually different, they confirmed the increase of (E)-2- 3.2.2. Carbonyl compounds nonenal and other linear C4–C10 alkenals and alkanals From the start of research on staling compounds, in beer during storage. Due to the growing importance carbonyls attracted most attention. Such compounds of (E)-2-nonenal and other carbonyls in beer, various were known to cause flavour changes in food products techniques have been proposed to measure their concen- such as milk, butter, vegetables and oils. Hashimoto trations in beer. Many methods remain based on deri- (1966) was the first to report a remarkable increase in vatisation of the carbonyls in order to decrease the the level of volatile carbonyls in beer during storage, interference caused by the beer matrix. Derivatisation in parallel with the development of stale flavours. Acet- agents, such as o-(2,3,4,5,6-pentafluorobenzyl)hydroxyl- aldehyde was one of the first compounds for which a amine (PFBOA) (Gro¨nqvist, Siirila¨, Virtanen, Home, & concentration increase was observed in aged beer (En- Pajunen, 1993; Ojala, Kotiaho, Siirila¨, & Sihvonen, gan, 1969) and further research (Meilgaard, Elizondo, 1994; Angelino et al., 1999) or hydroxylamine hydro- & Moya, 1970; Meilgaard & Moya, 1970; Palamand & chloride (Barker, Pipasts, & Gracey, 1989) have been Hardwick, 1969) on alkanals and alkenals revealed their applied to liquid–liquid extracts of beer and the deriva- high flavour potency in beer. In that context, Palamand tives were eventually analysed using GC–MS or GC– and Hardwick (1969) first described (E)-2-nonenal as a ECD. These procedures remain laborious and time- molecule, which on addition to beer, induces a card- consuming. More recent methods, using solid-phase board flavour similar to such flavour in aged beer. A micro-extraction (Vesely, Lusk, Basarova, Seabrooks, & year later, the identification, in heated acidified beer, Ryder, 2003) or stir bar sorptive extraction (Ochiai, of (E)-2-nonenal, by Jamieson and Van Gheluwe Sasamoto, Daishima, Heiden, & Hoffmann, 2003) with (1970), as the molecule responsible for cardboard fla- on-site PFBOA derivatisation of carbonyls, have been vour, was considered a breakthrough in beer flavour re- developed. Other techniques include the extraction of 360 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381

Table 1 Overview of the currently known volatile compounds formed during storage of beer Chemical class Compounds Linear aldehydes Acetaldehyde (E)-2-octenal/(E)-2-nonenal (I)/(E,E)-2,6-nonadienal/(E,E)-2,4-decadienal Strecker aldehydes 2-Methyl-butanal/3-methyl-butanal/2-phenylacetaldehyde/benzaldehyde/3-(methylthio)propionaldehyde (II) Ketones (E)-b-damascenone (III) 3-Methyl-2-butanone/4-methyl-2-butanone/4-methyl-2-pentanone (IV) Diacetyl (V)/2,3-pentanedione Cyclic acetals 2,4,5-Trimethyl-1,3-dioxolane (VI)/2-isopropyl-4,5-dimethyl-1,3-dioxolane/2-isobutyl-4,5-dimethyl- 1,3-dioxolane/2-sec butyl-4,5-dimethyl-1,3-dioxolane Heterocyclic compounds Furfural (VII)/5-hydroxymethyl-furfural/5-methyl-furfural/2-acetyl-furan/2-acetyl-5-methyl-furan/ 2-propionylfuran/furan/furfuryl alcohol Furfuryl ethyl ether (VIII)/2-ethoxymethyl-5-furfural/2-ethoxy-2,5-dihydrofuran Maltol (IX) Dihydro-5,5-dimethyl-2(3H)-furanon/5,5-dimethyl-2(5H)-furanon 2-Acetylpyrazine (X)/2-methoxypyrazine/2,6-dimethylpyrazine/trimethylpyrazine/tetramethylpyrazine Ethyl esters Ethyl 3-methyl-butyrate (XI)/ethyl 2-methyl-butyrate/ethyl 2-methyl-propionate Ethyl nicotinate (XII)/diethyl succinate/ethyl lactate/ethyl phenylacetate/ethyl formate/ethyl cinnamate Lactones c-Nonalactone (XIII)/c-hexalactone S-compounds Dimethyl trisulﬁde (XIV) 3-Methyl-3-mercaptobutylformate (XV)

carbonyls by vacuum distillation, followed by GC–MS crease during storage is by no means ubiquitous. analysis (Lermusieau, Noel, Liegeois, & Collin, 1999) According to Van Eerde and Strating (1981),(E)-2-non- or steam distillation, solid phase extraction and isocratic enal increased at 40 °C in several beers within a few days HPLC–UV detection (Santos et al., 2003). to levels above its threshold whereas, at 20 °C this was Since the Jamieson and Van Gheluwe (1970) publica- not found, even after 4 months of storage. A similar re- tion, expectations were high that (E)-2-nonenal could be sult was obtained by Narziss, Miedaner, and Graf considered as the molecule responsible for beer staling (1985). Moreover, recent publications (Foster, Samp, and that methods to reduce its concentration would fi- & Patino, 2001; Narziss, Miedaner, & Lustig, 1999; nally solve the problem of beer staling. However, its in- Schieberle & Komarek, 2002; Vesely et al., 2003) B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 361 mention no significant increases in (E)-2-nonenal con- ketones whose concentrations increase with beer age centration during beer aging. In contrast, other authors are 3-methyl-butan-2-one and 4-methylpentan-2-one (Lermusieau et al., 1999; Liegeois, Meurens, Badot, & (Hashimoto & Kuroiwa, 1975; Lustig, Miedaner, Nar- Collin, 2002; Santos et al., 2003) continue to report its ziss, & W., 1993) and the vicinal diketones; diacetyl formation and observations that it occurs independently and 2,3-pentanedione. This is more pronounced at of the oxygen concentration in a bottled beer (Narziss higher oxygen levels and diacetyl may even surpass et al., 1985; Noel et al., 1999; Walters, Heasman, & its flavour threshold (Wheeler et al., 1971). Hughes, 1997b). Despite such seemingly contradictory reports, there 3.2.3. Cyclic acetals are indications that some carbonyl compounds are Particularly when beer is in contact with oxygen, the important in flavour staling. This statement can be illus- cyclic acetals, 2,4,5-trimethyl-1,3-dioxolane, 2-isopro- trated by HashimotoÕs demonstration (Hashimoto, pyl-4,5-dimethyl-1,3-dioxolane, 2-isobutyl-4,5-dimethyl- 1981) that carbonyl scavengers, such as hydroxylamine, 1,3-dioxolane and 2-sec-butyl-4,5-dimethyl-1,3-dioxo- immediately diminish certain aspects of the aging fla- lane, increase during storage (Vanderhaegen et al., vour in beer. 2003b). A flavour threshold of 900 lg/l and a maximum Other linear aldehydes have flavour properties similar concentration in beer of around 100 lg/l were reported to those of (E)-2-nonenal (Meilgaard, 1975b). The for 2,4,5-trimethyl-1,3-dioxolane (Peppard & Halsey, involvement of these linear C4–C10 alkanals, alkenals 1982). and alkedienals in beer aging has been studied to a lesser extent. In a study of Greenhoff and Wheeler (1981a, 3.2.4. Heterocyclic compounds 1981b), the levels of all linear C4–C10 2-alkenals in- Heterocyclic compounds, some with carbonyl func- creased. In particular, longer chain 2-alkenals, starting tions, represent a large group of compounds subject to from 2-heptenal, surpassed their threshold during beer concentration changes during beer aging. The follow- storage. Only the shorter chain linear alkanals; butanal, ing furans are formed (Lustig et al., 1993; Madigan, pentanal and hexanal, were significantly formed. Haray- Perez, & Clements, 1998; Varmuza, Steiner, Glinsner, ama, Hayase, and Kato (1994) reported that the alkedie- & Klein, 2002): furfural, 5-hydroxymethyl-furfural nals, (E,Z)-2,6-nonadienal and (E,E)-2,4-decadienal (HMF), 5-methyl-furfural, 2-acetyl-furan, 2-acetyl-5- take part in flavour staling. methyl-furan, 2-propionylfuran, furan and furfuryl Other aldehydes formed during beer storage are the alcohol. Although they generally remain far below so-called Strecker aldehydes: 2-methyl-propanal their flavour thresholds, they are mentioned as sensi- (Wheeler, Pragnell, & Pierce, 1971; Bohmann, 1985b; tive indicators of beer flavour deterioration (Bernstein Vesely et al., 2003), 2-methyl-butanal (Miedaner, Nar- & Laufer, 1977; Brenner & Khan, 1976; Shimizu ziss, & Eichhorn, 1991; Vesely et al., 2003), 3-methyl- et al., 2001b). Furfural and HMF levels may increase butanal (Miedaner et al., 1991; Vesely et al., 2003; with time at an approximately linear rate, which varies Wheeler et al., 1971), benzaldehyde (Miedaner et al., logarithmically with the storage temperature (Madigan 1991; Wheeler et al., 1971), phenylacetaldehyde et al., 1998). Oxygen seems without effect. Interest- (Miedaner et al., 1991; Vesely et al., 2003) and meth- ingly, a close correlation is found between their in- ional (Gijs et al., 2002; Vesely et al., 2003). Generally, crease and the sensory scores for stale flavour. their concentrations increase more at elevated oxygen Therefore, these compounds can be used as indicators concentrations (Bohmann, 1985a; Miedaner et al., of heat-induced flavour damages to beer. Recently, we 1991; Narziss et al., 1985). AEDA of aged beer re- reported that furfuryl ethyl ether can also function as vealed that methional (cooked potato-like) (Gijs such an indicator (Vanderhaegen et al., 2004a). Fur- et al., 2002; Schieberle & Komarek, 2002) and phenyl- thermore, this furanic ether may increase to levels acetaldehyde (sweet, honey-like) (Schieberle & Komar- above the flavour threshold (6 lg/l) during storage, ek, 2002) are relevant for the sensory profile of aged inducing a solvent-like stale flavour in the beer (Van- beer. The other Strecker aldehydes do not seem impor- derhaegen et al., 2003b). tant for stale flavour formation, but can be considered According to Lustig et al. (1993), the following fura- as suitable markers for beer oxidation (Narziss, Mieda- nones also appear during beer aging: dihydro-5,5-di- ner, & Eichhorn, 1999a, 1999b). methyl-2(3H)-furanone, 5,5-dimethyl-2(5H)-furanone, For ketones, an AEDA study revealed that a carot- dihydro-2(3H)-furanone, 3-methyl-2(5H)-furanone and enoid-derived compound,b-damascenone (rhubarb, red 5-methyl-2(5H)-furanone. Furanones generally have fruits, strawberry) affects beer flavour during aging burnt flavours, but no data are available on their impor- (Chevance, Guyot-Declerck, Dupont, & Collin, 2002; tance for beer staling. Gijs et al., 2002). Carotenoid-derived flavour compo- Pyrazines form another group of heterocyclic mole- nents had already been suspected to be staling compo- cules subject to changes during storage. According to nents by Strating and Van Eerde (1973). Other Qureshi, Burger, and Prentice (1979), the concentrations 362 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 of some pyrazines decrease very rapidly (pyrazine, 2- pto-4-methyl-penta-2-one (Tressl, Bahri, & Kossa, ethyl-6-methylpyrazine, 2-ethyl-5-methylpyrazine) and 1980). some even completely disappear (2-acetylpyrazine, 2,3- dimethylpyrazine, 2,5-dimethylpyrazine, 2-ethyl-3,6- 3.3. Non-volatile compounds dimethylpyrazine, 2-ethyl-3,5-dimethylpyrazine). The concentrations of other pyrazines appreciably increase, Non-volatile compounds in beer can be important for e.g., 2,6-dimethylpyrazine, trimethylpyrazine and tetra- taste and mouthfeel. Changes in concentration may methylpyrazine. This is somewhat contradictory to the therefore induce important sensory alterations. AEDA results of Gijs et al. (2002) who considered 2- Iso-a-acids, the main bitterness substances in beer, acetylpyrazine (sweet, candy floss, caramel), 2-methoxy- are particularly sensitive to degradation during storage pyrazine (cereal roasted) and also maltol (caramel, (De Cooman, Aerts, Overmeire, & De Keukeleire, roasted) relevant for the sensory profile of aged beer (5 2000; King & Duineveld, 1999; Walters et al., 1997b), days, 40 °C). which results in a decrease in sensory bitterness. The iso-a-acids comprise six major components: the trans- 3.2.5. Esters and cis-isomers of isocohumulone, isohumulone and Volatile esters introduce fruity flavour notes and are isoadhumulone. The trans-isomers are much more sensi- considered highly positive flavour attributes of fresh tive to degradation than the cis-isomers. The concentra- beer. Isoamyl acetate, produced by yeast, e.g., gives a tion ratio trans/cis isomer was proposed as a good banana-like flavour. However, during storage, the con- marker for the flavour deterioration of beer (Araki, centration of this ester can decrease to levels below its Takashio, & Shinotsuka, 2002). threshold level (Neven, Delvaux, & Derdelinckx, 1997; Apart from iso-a-acids, polyphenols are some of the Stenroos, 1973) which results in a diminished fruity fla- more readily oxidized beer constituents (Kaneda, Kano, vour of beer. Osawa, Kawakishi, & Kamimura, 1990). McMurrough, In contrast, certain volatile esters (ethyl 3-methyl- Madigan, Kelly, and Smyth (1996) measured decreases butyrate, ethyl 2-methyl-butyrate, ethyl 2-methyl- of the flavanols (+)-catechin, ()-epicatechin, prodel- propionate, ethyl nicotinate, diethyl succinate, ethyl phinidin B3 and procyanidin B3 concentration during lactate, ethyl phenylacetate, ethyl formate, ethyl furo- storage at 37 °C. The loss was highest during the first ate and ethyl cinnamate) are synthesized during beer four to five weeks but continued at a decreased rate aging (Bohmann, 1985b; Gijs et al., 2002; Lustig throughout prolonged periods. Dimeric flavanols disap- et al., 1993; Miedaner et al., 1991; Williams & Wag- peared more rapidly than monomers. In contrast, after a ner, 1978). Williams and Wagner (1978) related the lag period of about 5 weeks, the levels of tannoids began formation of ethyl 3-methyl-butyrate and 2-methyl- to increase (McMurrough, Madigan, & Kelly, 1997) and butyrate to the development of winy flavours. The the changes in the polyphenol contents were associated importance of these molecules for the flavour of aged with the appearance of harsh/astringent tastes. beer was also recently reported using AEDE experi- There are only few reports on beer storage-related ments (Schieberle & Komarek, 2002) and Gijs et al. changes in amino acids. In general, a slight decrease is (2002) confirmed this also for ethyl cinnamate (fruity, observed of some individual amino-acids (Basarova, Sa- sweet). vel, Janousek, & Cizkova, 1999) and glutamine has been Finally, lactones or cyclic esters, such as c-hexalac- proposed as a staling marker (Hill, Lustig, & Sawatzki, tone and c-nonalactone (peach, fruity) tend to 1998). increase in concentration (Eichhorn, Komori, Mieda- ner, & Narziss, 1989) and the latter molecule is con- 3.4. Chemical origin of beer flavour deterioration sidered important for the flavour of aged beer (Gijs et al., 2002). A closer look at the changes in chemical constituents of aging beer reveals the enormous complexity 3.2.6. Sulfur compounds of the beer staling phenomenon. Early on, it was as- Sulfur compounds generally have an extremely low sumed that mainly (E)-2-nonenal was responsible for flavour threshold in beer and small concentration sensory changes, but now it is evident that a myriad changes may have noticeable effects on flavour. Di- of flavour compounds is responsible. A stale flavour methyl trisulfide (fresh-onion-like) may increase to is considered the result of formation and degradation above its flavour threshold of 0.1 lg/l (Gijs et al., reactions. With AEDA some new compounds have al- 2002; Gijs, Perpete, Timmermans, & Collin, 2000; Wil- ready been discovered although it remains necessary liams & Gracey, 1982). An AEDA experiment re- to study their flavour-affecting properties, using spik- vealed that also 3-methyl-3-mercaptobutyl formate ing experiments and sensory evaluations. An overview (catty, ribes)(Schieberle, 1991) is involved. Another of the chemical reactions involved in beer staling may ribes flavour-linked molecule in aged beer is 4-merca- help to better understand the beer-aging phenomenon. B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 363

4. Reaction mechanisms of aging processes in beer eda et al., 1988; Uchida & Ono, 1996; Uchida & Ono, 1999) and chemiluminescence (CL) (Kaneda, Kano, 4.1. General mechanisms Osawa, Kawakishi, & Koshino, 1991) analysis made it possible to unravel the initial oxygen-dependent reac- Chemically, beer can be considered as a water-etha- tions (Fig. 2). 3 nol solution with a pH of around 4.2 in which hundreds Oxygen in the ground state ( O2) is quite stable and of different molecules are dissolved. These originate will not easily react with organic molecules. In the pres- from the raw materials (water, malt, hops, adjuncts) ence of ferrous iron (Fe2+) in beer, oxygen can capture and the wort production, fermentation and maturation an electron and form the superoxide anion ðO2 Þ and processes. However, the constituents of freshly bottled Fe3+. Copper ions probably have the same behaviour beer are not in chemical equilibrium. Thermodynami- and Cu+ is oxidized to Cu2+ (Kaneda, Kobayashi, Tak- cally, a bottle of beer is a closed system and will thus ashio, Tamaki, & Shinotsuka, 1999). It is believed that strive to reach a status of minimal energy and maximal Cu+/Cu2+ and Fe2+/Fe3+ ions are part of a mixed func- entropy. Consequently, molecules are subjected to many tion oxidation system in which polyphenols, sugars, iso- reactions during storage, which eventually determine the humulones and alcohols might act as electron donors type of the aging characteristics of beer. (Kaneda et al., 1992). The superoxide anion can be pro- Although many conversions are thermodynamically tonated to form the perhydroxyl radical (OOHÅ), which possible, their relevance to beer aging is mainly deter- has much higher reactivity. The pKa of this reaction is mined by the reaction rates under practical storage con- 4.8, which means that, at the pH of beer, the majority ditions. The reaction rate is a function of substrate of the superoxide will be in the perhydroxyl form. The concentrations and rate constants, which differ between superoxide anion can also be reduced by Fe2+ or Cu+ 2 reaction types and which are temperature-dependent. In to the peroxide anion ðO2 Þ. In beer, this anion is readily practice, reaction rates increase with higher substrate protonated to hydrogen peroxide (H2O2). Hydroxyl rad- Å concentrations and storage temperatures. icals (OH ) can then be produced from H2O2 or the superoxide anion O2 by metal-induced reactions, such 4.2. Reactive oxygen species in stored beer as the Fenton and the Haber–Weiss reaction. The reactivity of the oxygen species increases with Oxygen, in particular, causes a rapid deterioration of their reduction status (superoxide anion < perhydroxyl beer flavour, meaning that oxygen must initiate some radical < hydroxyl radical). The concentration of free very important aging reactions. The importance of reac- radicals during the aging of beer increases with increas- tive oxygen species (ROS) in beer staling was first indi- ing iron/copper ion concentrations, with increasing oxy- cated by Bamforth and Parsons (1985). In recent gen concentrations or with higher storage temperatures years, studies using electron spin resonance (ESR) with (Kaneda et al., 1992; Kaneda, Kano, Osawa, Kawaki- spin trapping reagents (Andersen & Skibsted, 1998; shi, & Kamada, 1989). Furthermore, the free radicals Kaneda, Kano, Koshino, & Ohyanishiguchi, 1992; Kan- are not always generated just after the start of the aging

RH R RH R Reaction A: Fenton reaction 2+ → 3+ - Fe + H2O2 Fe + OH + OH 3+ → 2+ - + Fe2+ Fe3+ Fe2+ Fe3+ Fe + H2O2 Fe + O2 + 2H - - + - 2- Net: 2H2O2 OH + OH + O2 + 2H 3 O O O2 2 2 Reaction B:Haber-Weiss reaction pKa 4.5-4.9 pKa 16-18 3+ 2- → 2+ Fe + O Fe + O2 Fe2+ + H O → Fe3+ + OH + OH- - 2 2 2- - HO HO2 Net: O + H O → O + OH + OH 2 Reaction A 2 2 2

pKa 11.6 Fe2+ Fe3+

H2O2 OH

Fe2+ Fe3+

Reaction B

Fig. 2. Reactions producing reactive oxygen species (ROS) in beer (Kaneda et al., 1999). 364 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 process, but can be formed after a definite time period, concentrations of polyphenols. Furthermore, the reac- called the ‘‘lag time’’ of free-radical generation (Uchida tivity of alcohols decreases with their molecular weight. & Ono, 1996; Uchida, Suga, & Ono, 1996). The ‘‘lag- Irwin, Barker, and Pipasts (1991) found this pathway time’’ seems related to the endogenous antioxidant irrelevant in the formation of (E)-2-nonenal because of activity of beer and can be used as an objective tool the very low efficiency (0.2%) of 2-nonenol to nonenal for its evaluation. conversion in model systems. Hydroxyl radicals are one of the most reactive species Nonetheless, it was recently reported that the that have been identified. Therefore, it was suggested 1-hydroxyethyl radical is quantitatively the most impor- that they non-selectively react with ethanol in beer be- tant radical in stale beer due to hydroxyl radicals react- cause it is the second most abundant compound of beer ing with ethanol (Andersen & Skibsted, 1998). A main and a good radical scavenger. The findings of Andersen degradation product of the radical is acetaldehyde and Skibsted (1998), which revealed the 1-hydroxyethyl (Fig. 3). Even though ethanol is more abundant in beer radical as quantitatively the most important radical in than any other organic molecule, ROS may react in a beer, support this. The 1-hydroxyethyl radical arises in similar manner with the main higher alcohols. From this the reaction of ethanol with the hydroxyl radical. Gen- perspective, the formation of ROS and their reaction Å erally, the reactive oxygen species (O2 , HOO ,H2O2 with alcohols can be regarded as a generalization of and HOÅ) react with all kinds of organic molecules in the reaction mechanism proposed by Hashimoto (1972). beer, such as polyphenols, isohumulones and alcohols, resulting in various changes in the sensory profile of 4.3.3. Strecker degradation of amino acids beer. Amino acids in stored beer may be a source of aldehydes. Blockmans, Devreux, and Masschelein (1975) ob- 4.3. Aging reactions producing carbonyl compounds served an increased formation of 2-methyl-propanal and 3-methyl-butanal when either valine or leucine were 4.3.1. Importance of carbonyl mechanisms added to beer and oxygen was present. The reaction Soon after the importance of carbonyl compounds was catalysed by Fe and Cu ions. This was explained for beer staling was revealed, pathways for their forma- by a Strecker reaction between amino acids and a-dicar- tion were suggested. From the beginning, reaction mech- bonyl compounds (Fig. 4). The reaction involves trans- anisms leading to (E)-2-nonenal have been the focus of amination, followed by decarboxylation of the this research. Many routes have been studied in beer subsequent a-ketoacid, resulting in an aldehyde with model systems and it therefore remains difficult to tell one carbon atom less than the amino acid. to what extent a particular reaction mechanism is rele- Additional a-dicarbonyl compounds in beer are pos- vant under normal storage conditions. sibly formed by the Maillard reaction, the oxidation of reductones or the oxidation of polyphenols. Thum, 4.3.2. Oxidation of higher alcohols Miedaner, Narziss, and Black (1995) mention that Strec- The most important alcohols in beer are ethanol, 2- ker degradation is only important at strongly increased methyl-propanol, 2-methyl-butanol, 3-methyl-butanol amino acids contents, but not at the amino acid concen- and 2-phenyl-ethanol. Various researchers have re- trations normally present in beer (±1 g/l). ported that the concentrations of the corresponding aldehydes increase during beer aging, in particular when 4.3.4. Aldol condensation oxygen was present (see above). Hashimoto and Kuroiwa (1975) suggested that aldol Hashimoto (1972) studied the increased formation of condensation of carbonyl compounds is possible under aldehydes due to exposure of beer to higher oxygen lev- the mild conditions existing in beer during storage. For els. High temperatures, low pH and the supplementation example, (E)-2-nonenal was formed by aldol condensa- of additional higher alcohols to beer led to higher con- tion of acetaldehyde with heptanal in a model beer centrations of aldehydes. Moreover, direct oxidation stored for 20 days at 50 °C and containing 20 mmol/l of alcohols by molecular oxygen was not possible in beer of proline (Fig. 5). In these reactions, the amino acids model systems, unless melanoidins were present. A reac- may be the basic catalysts through the formation of an tion mechanism was proposed in which alcohols transfer imine intermediate. This pathway can produce car- electrons to reactive carbonyl groups of melanoidins. bonyl compounds with lower flavour thresholds from Molecular oxygen accelerates the oxidation of the alco- carbonyls present in beer which are less flavour active, hols, probably because the melanoidins are transformed and which can be formed by other pathways. Although in such a way that the reactive carbonyl groups are in- the aldol condensation pathway seems plausible, it is volved in the electron-transfer system. not clear whether the amounts of reaction products Devreux, Blockmans, and vande Meersche (1981) are sufficiently high to reach threshold concentrations doubted the importance of this pathway as they ob- under normal beer storage conditions (Bamforth, served the requirements of light and inhibition by low 1999b). B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 365

ethanol HO + CH3CH2OH

85% 13%

CH2CH2OH CH3CHOH

O O2 2

H C C OH OO CH2 CH2 OH 3 H

bimolecular reactions

acetaldehyde CH3CHO HOCH CH OH 2 2 + HOCH2CHO + CH O + O + H O HOO HOO 2 2 2 2 + + other minor byproducts

Fig. 3. Reaction of ethanol with the hydroxyl radical in beer according to Andersen and Skibsted (1998).

Strecker degradation of amino acids

R amino acid CC N COOH H O O C C RCCOOH C H O NH2 H2O

N COOH H O C C 2 RCCOOH C R O O CC H N O 2 CO2

R CHO Strecker aldehyde

Formation of dicarbonyl compounds OH OH R OH Amino acid + Sugar R' R OH O O Amadori rearrangement 2 O2

CC O O a-di carbonyl compound

Fig. 4. Formation of aldehydes in beer by Strecker degradation of amino acids (Thum et al., 1995).

4.3.5. Degradation of hop bitter acids indications that some of them are involved in the The degradation of hop bitter acids (iso-a-acids, a- appearance of aging ﬂavours. Indeed, Hashimoto and acids and b-acids) not only decreases sensory bitterness, Eshima (1979) reported that beer brewed without hops but also results in the formation of products. There are hardly develops a typical stale ﬂavour, even after a long 366 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381

heptanal acetaldehyde O O O O O O R C + R H3C C H H O OH O OH amino-acid OH

H O α-acid β-acid 2 R H2 N CH H C O O C C O O COOH OH H R R H O HO 2 HO OH OH O O

H2O amino-acid

H C O α C C (E)-2-nonenal trans-iso- -acid cis-iso-α-acid H H R = a CH(CH3)2 R = Fig. 5. Formation of (E)-2-nonenal in beer by aldol condensation of b CH2CH(CH3)2 acetaldehyde and heptanal according to Hashimoto and Kuroiwa Rc = CH(CH )CH CH (1975). 3 2 3 Fig. 6. Most important hop bitter acids in beer. shelf storage. The exact degradation mechanism for hop acids and the chemical structures of the volatiles formed, tone, 2-methyl-propanal, 3-methyl-butan-2-one, 4- have not been completely elucidated. Fig. 6 presents the methyl-pentan-2-one and 2-methyl-3-buten-2-ol were structure of the most important beer hop bitter acids. also identified as oxidation products. Moreover, Wil- Iso-a-acids quickly degrade in the presence of ROS liams and Wagner (1979) showed that degradation of (Kaneda et al., 1989), the trans-isomer being much more the carbonyl side-chain of a-acids and b-acids releases sensitive than the cis-isomer (De Cooman et al., 2000). 2-methyl-propionic acid, 2-methyl-butyric acid and 3- However, recent research (Huvaere et al., 2003) indi- methyl-butyric acid. As will be explained later, these cates that electrons are released from iso-a-acids in the acids are precursors in the formation of staling esters. presence of suitable electron acceptors, which do not necessarily require the involvement of oxygen species. 4.3.6. Oxidation of unsaturated fatty acids As a result, oxygen- and carbon-centred radicals are 4.3.6.1. General mechanisms and intermediates. From the formed. These radicals are very reactive and lead to start of research on beer staling, the oxidation of unsat- products of varying nature; however, all lack the tricar- urated fatty acids received more attention than any bonyl chromophore. The double bonds in the side-chains other reaction. Soon after the cardboard flavour of beer of the hop acids are less reactive toward oxidation than was linked to (E)-2-nonenal formation, it was suggested was commonly thought. It is now clear that iso-a-acids that its formation and that of other saturated and unsat- can be subject to oxidative-type degradation in the urated aldehydes was due to lipid oxidation (Dale & absence of molecular oxygen. Pollock, 1977; Drost et al., 1971; Jamieson & Van Ghe- The reduced side-chain iso-a-acids, used to impart luwe, 1970; Tressl, Bahri, & Silwar, 1979). This was cer- light resistance, have fewer structural positions sensitive tainly related at that time to the extensive research and to radical formation. Consequently, they show more knowledge of the oxidative breakdown of lipids in resistance to oxidative breakdown. foods, leading to carbonyl compounds and rancidity. According to Hashimoto and Eshima (1979), volatile In beer and wort, the only lipid substrates of significance degradation products of iso-a-acids in beer model sys- are linoleic acid (C18:2) and linolenic (C18:3) acid, aris- tems are carbonyl compounds with various chain ing from malted barley. These acids are mainly released lengths, such as C3 to C11 2-alkanones, C2 to C10 alka- from triacylglycerols by the activity of barley and malt nals, C4 to C7 2-alkenals and C6 to C7 2,4-alkedienals. lipases (Baxter, 1984). During malting, slight changes In an earlier study (Hashimoto & Kuroiwa, 1975), ace- in fatty acids and lipid composition occur. Hydrolysis B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 367 of triacylglycerols to fatty acids occurs mainly during are both important for the oxidation of linoleic acid mashing. Malt lipase remains active through much of and the subsequent release of (E)-2-nonenal. There are the mashing process (Schwarz, Stanley, & Solberg, two possible oxidation routes: auto-oxidation or an 2002). enzymatic oxidation with lipoxygenases (LOX) only At present, there is strong evidence that lipid oxida- during mashing and malting. There is a great deal of tion does not occur in beer after bottling. (E)-2-nonenal controversy concerning the relative importance of both is released from other precursors, in a non-oxidative routes for (E)-2-nonenal release in finished beer (Bam- process, as the oxygen concentration of bottled beer forth, 1999a; Stephenson, Biawa, Miracle, & Bamforth, does not influence the (E)-2-nonenal release (Ler- 2003). This is partly related to the use, in brewing re- musieau et al., 1999; Noel et al., 1999). search, of small-scale equipment in which the oxygen in- Oxidation intermediates of linoleic acid, trihydroxy gress is much larger than in industrial scale brewing. fatty acids, have been investigated as possible precursors Recent studies (Lermusieau et al., 1999; Liegeois et al., (Drost et al., 1971; Graveland, Pesman, & Van Eerde, 2002) using wort spiked with deuterated (E)-2-nonenal, 1972). However, they convert to (E)-2-nonenal only in revealed that 70% of the (E)-2-nonenal released during acidified beer (pH 2), thus excluding them as plausible beer staling was initially produced during boiling, and precursors in beer (Stenroos et al., 1976). the other 30% during mashing. However, these results Generally, it is now agreed that, during wort produc- do not exclude the possibility that some oxidation inter- tion, enzymatic and non-enzymatic oxidation, mainly of mediates of fatty acids, such as hydroxy fatty acids and linoleic acid, generates a ‘‘(E)-2-nonenal potential’’ initi- hydroperoxy fatty acids, produced by LOX during ating the (E)-2-nonenal formation during beer storage. mashing, may be converted non-enzymatically to (E)- Drost, van den Berg, Freijee, van der Velde, and Holle- 2-nonenal under the extreme conditions of wort boiling. mans (1990) defined the nonenal potential as the potential of wort to release (E)-2-nonenal after its treatment 4.3.6.2. Auto-oxidation of fatty acids. Auto-oxidation of for 2 h at 100 °C at pH 4.0, under an argon atmosphere. a fatty acid is initiated by the abstraction of a H-atom Noe¨l and Collin (1995) found strong evidence that from the molecule by free radicals. As previously men- (E)-2-nonenal in wort forms SchiffÕs bases (imines) with tioned, hydroxyl radicals (HOÅ) are exceptionally reac- amino acids or proteins which pass into beer. During tive toward many molecules found in food. In the storage, (E)-2-nonenal is then released and this is en- complex wort environment it is, however, debatable hanced at low pH (Lermusieau et al., 1999). ‘‘Free’’ whether a hydroxyl radical can reach a fatty acid before (E)-2-nonenal in wort is reduced to nonenol by yeast fer- it finds much more abundant molecules (e.g., sugars). mentation and nonenol is not significantly re-oxidized Therefore, auto-oxidation is more likely to be initiated during beer storage (Irwin et al., 1991). by the slower-reacting peroxy radicals (ROOÅ), abstract- There has been some debate about whether (E)-2- ing the most weakly bound H-atom in the fatty acid. Be- nonenal is similarly released from non-volatile bisulfite sides the peroxy radicals that are produced in the adducts during storage. The sulfite formed by yeast dur- pathway (hence the auto-catalytic character), perhydr- ing fermentation would form reversible adducts with oxyl radicals (HOOÅ) may also abstract the H atoms. (E)-2-nonenal (Barker, Gracey, Irwin, Pipasts, & Leiska, With linoleic acid, the methylene group at position 11 1983). Sulfite can add on to the carbonyl function and is activated, especially by the two neighbouring double on to the double bound of (E)-2-nonenal. As during bonds (Fig. 7). Hence, this is the initial site for H storage, the sulfite concentration gradually decreases, abstraction, leading to a pentadienyl radical, which is the (E)-2-nonenal would be released. However, Dufour, then stabilized by the formation of two hydroperoxides Leus, Baxter, and Hayman (1999) recently showed that, at positions 9 and 13 (9-LOOH and 13-LOOH), each while bisulfite addition to the carbonyl function is retaining a conjugated diene system. The monoallylic reversible, bisulfite addition to the double bond in (E)- groups in linoleic acid (positions 8 and 14) also react 2-nonenal is irreversible. Due to the irreversible nature to a small extent and form four hydroperoxy acids of the bisulfite addition to the double bond and the sta- (8-,10-, 12- and 14-LOOH), each isomer having two iso- bility of such adducts, (E)-2-nonenal cannot be released lated double bonds. The proportion of these minor from these non-volatile species. This supports the obser- hydroperoxy acids is about 4% of the total. vations of other authors (Kaneda, Takashio, Osawa, The hydroperoxy acids can be further subject to non- Kawakishi, & Tamaki, 1996; Lermusieau et al., 1999), enzymatic oxidation or degradation processes leading to describing a minimal formation of reversible (E)-2-non- a variety of compounds such as volatiles. Several reac- enal-bisulfite adducts during fermentation. tion mechanisms have been suggested to explain their The increase of the cardboard-flavoured compound formation. Ohloff (1978) proposed an ionic mechanism (E)-2-nonenal in aging beer is thus probably linked to for the formation of (E)-2-nonenal from 9-LOOH and oxidation processes earlier in the production process, hexanal from 13-LOOH in aqueous systems (Fig. 8). A mainly in the brewhouse. Mashing and wort boiling heterolytic cleavage is initiated by protonation of the 368 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381

H3C (CH2)4 CH CH CH CH CH (CH2)7 COOH linoleic acid H RO2 ROOH 13 9

H3C (CH2)4 CH CH CH CH CH (CH2)7 COOH

OO OO H C (CH ) CH H3C (CH2)4 CH CH CH CH CH (CH2)7 COOH 3 2 4 CH CH CH CH (CH2)7 COOH

RH RH R R

OOH OOH H C (CH ) CH CH CH CH CH (CH ) COOH 3 2 4 2 7 H3C (CH2)4 CH CH CH CH CH (CH2)7 COOH

9-hydroperoxyoctadeca-10,12-dienoic acid 13-hydroperoxyoctadeca-9,11-dienoic acid (9-LOOH) (13-LOOH)

Fig. 7. Formation of the hydroperoxy fatty acids 9-LOOH and 13-LOOH by autoxidation of linoleic acid (Belitz and Grosch, 1999).

H hydroperoxy fatty acid + OH O O + O H H R1 CH CH CH CH CH R2 R1 CH CH CH CH CH R2

+ O + R1 CH CH CH CH CH R2 R1 CH CH CH CH O C R2 H H O O OH R1 CH CH CH CH O CR2 R1 CH CH CH CH OH + CR2 H H O

R1 CH2 CH CH C 9-LOOH 13-LOOH H

R1: R1: (CH2)4 CH3 (CH2)7 COOH

R2: (CH2)7 COOH R2: (CH2)4 CH3

Fig. 8. Proton-catalysed cleavage of 9-LOOH and 13-LOOH hydroperoxy acids of linoleic acid according to Ohloff (1978). hydroperoxide group. After elimination of a water mol- Newly formed unsaturated aldehydes are susceptible ecule, the oxo-cation formed is subjected to an oxygen to further oxidation reactions, which in turn produce atom insertion reaction exclusively on the C–C linkage other carbonyl compounds. adjacent to the double bound. The carbenium ion is hydroxylated and then splits into an oxo-acid and an 4.3.6.3. Enzymatic breakdown of fatty acids. Germinated aldehyde (2-nonenal or hexanal). barley contains two lipoxygenase enzymes, namely On the other hand, in the oil phase of some foods, a LOX-1 and LOX-2 (Baxter, 1982). LOX-1 is present b-cleavage of hydroperoxy acids is the predominant deg- in raw barley and increases during germination, whereas radation reaction. It involves a homolytic cleavage with LOX-2 only develops during germination (Yang & Sch- a formation of short-lived alkoxy radicals. The cleavage warz, 1995). They can oxidize fatty acids with a cis,cis- further away from the double bond is energetically pre- 1,4-pentadiene system, such as linoleic acid and linolenic ferred, since it yields resonance-stabilized compounds. acid, to their hydroperoxy acids. Linoleic acid is stereo- In this way, 2,4-decadienal is formed by the degradation and regio-specifically oxidized to 9-LOOH by LOX-1 of 9-LOOH. and to 13-LOOH by LOX-2 (Doderer, Kokkelink, van B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 369 der Veen, Valk, & Douma, 1991; Garbe, Hu¨bke, & heat-stable than LOX-1. This enzymatic factor seems re- Tressl, 2003; Hugues et al., 1994). Both enzymes are very lated to peroxygenase (POX), a member of the plant cyto- heat-sensitive, with LOX-1 being somewhat more heat- chrome P450-containing systems that use hydroperoxide resistant than LOX-2. Therefore, during kilning, most fatty acid as a substrate and catalyze the hydroxylations LOX activity is destroyed and the remaining activity without NADPH or molecular oxygen. In turn, the new in malt is mainly due to LOX-1 (Yang & Schwarz, hydroxy acids can be broken down non-enzymatically 1995). The remaining activity seems to be the main cause to various carbonyl compounds (Tressl et al., 1979), of fatty acid oxidation during mashing. This is in accor- but considerable levels remain present in the finished beer dance with an observed concentration ratio of 9-LOOH/ (Kobayashi et al., 2000a). Furthermore, it was revealed 13-LOOH of 10/1 in wort during mashing at 52 °C that 9-LOOH is also transformed to (E)-2-nonenal by (Walker, Hughes, & Simpson, 1996). LOX enzymes be- 9-hydroperoxide lyase-like activity during mashing come completely inactivated at temperatures above (Kuroda, Furusho, Maeba, & Takashio, 2003). 65 °C. Both enzymes exhibit a pH optimum at 6.5. During the germination of barley, a hydroperoxy acid LOX-1 has a broad pH range with the activity falling isomerase appears, which catalyzes the transformation to 50% at pH 5. The pH-range of LOX-2 is much nar- of hydroperoxy acids to ketols. The ketols can be con- rower and this enzyme is almost completely inactive at verted non-enzymatically to mono-, di- and trihydroxy pH 5 (Doderer et al., 1991). A reduced formation of acids. Although hydroperoxy acid isomerase is found hydroperoxy fatty acids was observed at higher initial in malt, Schwarz and Pyler (1984) reported that it is mashing temperatures (Kobayashi, Kaneda, Kano, & tightly bound to the insoluble barley grist and is not re- Koshino, 1993b) or when the pH was lowered from leased in the soluble fraction of the mash. Therefore, it 5.5 to 5.0 (Kobayashi, Kaneda, Kano, & Koshino, can be assumed that this enzyme is not involved in 1993a). The hydroperoxy fatty acids are subject to hydroperoxy acid transformations during mashing. further enzymatic or non-enzymatic breakdown Finally, linoleic and linolenic acid, esterified in triac- (Kobayashi, Kaneda, Kano, & Koshino, 1994) and ylglycerol, can also be oxidized by LOX enzymes (Garbe (E)-2-nonenal can be formed from 9-LOOH. et al., 2003; Holtman, VredenbregtHeistek, Schmitt, & Mono-, di- and trihydroxy fatty acids accumulate Feussner, 1997), LOX-2 having a higher activity than during mashing and are possibly formed by enzymatic LOX-1. The finding of esterified hydroxyfatty acids in breakdown of hydroperoxy fatty acids (Kobayashi triacylglycerols or phospholipids in barley and malt et al., 2000b). Recently, Kuroda, Kobayashi, Kaneda, (Wackerbauer & Meyna, 2002; Wackerbauer, Meyna, Watari, and Takashio (2002) showed that, during mash- & Marre, 2003) and concentration increases during storing, linoleic is transformed to di- and trihydroxy acids age gave evidence for lipid oxidation by LOX enzymes. by LOX-1 and an additional enzyme, which is more These oxidized lipids are also likely precursors of

LOX-2 Lipids

O2 oxidized lipase triacylglycerols

Linoleic acid

LOX-2 LOX-1

O2 O2

13-LOOH 9-LOOH

9-hydroperoxide enzymatic factor lyase-like activity

mono, di, trihydroxy acids (E)-2-nonenal mono, di, trihydroxy acids

non-enzymatically non-enzymatically

carbonyl compounds carbonyl compounds

Fig. 9. Overview of the currently known enzymatic oxidation pathways of linoleic acid leading to carbonyl compounds. 370 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 carbonyl compounds in wort. The currently described & Halsey, 1982). In beer, an equilibrium between enzymatic pathways for oxidation of lipids during beer 2,4,5-trimethyl-1,3-dioxolane, acetaldehyde and 2,3- production are summarized in Fig. 9. butanediol is reached quite rapidly. As a result, the increase in the acetaldehyde concentration during aging 4.3.7. Formation of (E)-b-damascenone causes the 2,4,5-trimethyl-1,3-dioxolane concentration (E)-b-damascenone belongs to a class of carotenoid- to increase very similarly (Vanderhaegen et al., 2003b). derived carbonyl compounds. Potential precursors of damascenone in beer are allene triols and acetylene diols 4.5. Maillard reaction formed by degradation of neoxanthin, which is present in the basic ingredients of beer (Chevance et al., 2002). Many heterocyclic compounds found in aged beers Moreover, it was proven that the appearance of (E)-b- are well known products of the Maillard reaction. The damascenone in beer increased during aging when a b- diverse and complex reactions between reducing sugars glucosidase was added. The non-enzymatic release in beer and proteins, peptides, amino acids or amines, as well was enhanced at low pH (Gijs et al., 2002). As the pro- as the numerous consecutive reactions, are all classified posed precursors are linked to sugars, as observed in as Maillard reactions. In contrast to lipid oxidation, wines (Bureau, Baumes, & Razungles, 2000), (E)-b-dama- studies of Maillard reactions related to beer aging are scenone might also result from a chemical hydrolysis of scarce. Such limited interest may stem from observa- glycosides during beer aging. Consequently, glycosides tions that the currently known Maillard products in may also be considered as important sources of flavours aged beer (e.g., furfural and 5-hydroxymethyl furfural), related to beer aging (Biendl, Kollmannsberger, & Nitz, remain below their flavour threshold. On the other 2003). This can be of great significance in the production hand, the typical flavour of many food products is due of speciality beers, which are often characterized by the to Maillard reactions. Studies with model systems con- use fruits or herbs, rich in glycosides. taining a single type of sugar and amino acid revealed the formation of a myriad of Maillard compounds (Hof- 4.4. Acetalization of aldehydes mann & Schieberle, 1997; Umano, Hagi, Nakahara, Shyoji, & Shibamoto, 1995), suggesting that in food, The cyclic acetals (2,4,5-trimethyl-1,3-dioxolane, including beer, an even greater variety of products can 2-isopropyl-4,5-dimethyl-1,3-dioxolane, 2-isobutyl-4,5- be formed. Probably, the list of Maillard products in dimethyl-1,3-dioxolane and 2-sec butyl-4,5-dimethyl- the previous section is only a small reflection of the ac- 1,3-dioxolane) originate from a condensation reaction tual number of beer aging-related compounds. Some of (Fig. 10) between 2,3-butanediol (up to 280 mg/l in beer) them might merit more interest, as it was found recently and an aldehyde (acetaldehyde, isobutanal, 3-methyl- that the Maillard reaction is responsible for the develop- butanal and 2-methyl-butanal, respectively) (Peppard ment of bready, sweet and wine-like flavour notes dur-

H3C H2C OH ethanol

CH3 CH + CH 3 + H 3 C + H + H C C O O H HO H O OH H acetaldehyde

HO OH H3C CH3

+ H3C CH3 - H - H O 2,3-butanediol 2

CH3

H3C CH3 2,4,5-trimethyl-1,3-dioxolane

Fig. 10. Formation mechanism of 2,4,5-trimethyl-1,3-dioxolane in beer (Vanderhaegen, 2004). B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 371 ing beer staling. This was demonstrated through the use During storage, some Maillard intermediates may re- of the specific Maillard reaction inhibitor, aminoguani- act with typical beer constituents to give staling com- dine (Bravo et al., 2001b). pounds. Furfuryl ethyl ether arises in beer due to an Quantitatively, one of the most important heterocy- acid-catalysed condensation reaction (Fig. 12) of furfu- clic staling compound is 5-hydroxymethyl-furfural. Its ryl alcohol and ethanol (Vanderhaegen et al., 2004a). formation by the Maillard reaction is shown in Fig. In the production of beer, furfuryl alcohol is formed 11. The reaction starts with a SchiffÕs base formation be- by Maillard reaction mainly during malt kilning and tween a carbonyl group of a hexose sugar and an amino wort boiling. Evidence was found that a Maillard reac- group, leading to an imine. This undergoes an Amadori tion of maltose and a-(1,4)-oligoglucans is responsible. rearrangement to a more stable 1-amino-1-deoxyketose Reduction of furfural by yeast may further increase (also called Amadori product). However, at the pH of the furfuryl alcohol concentration during fermentation beer, the Amadori product is subject to enolization (Vanderhaegen et al., 2004b). and subsequent release of an amine, which leads to 3- deoxy-2-hexosulose (3-DH). This reactive a-dicarbonyl 4.6. Synthesis and hydrolysis of volatile esters compound can (among various secondary products) give rise to HMF. Starting from a pentose, furfural is Chemical condensation reactions between ethanol formed. At this stage HMF, furfural and other com- and beer organic acids occur at significant rates during pounds are merely intermediates of the Maillard reac- beer storage. For example, 3-methyl-butyric acid and tion. They are subject to further reactions 2-methyl-butyric lead to ethyl 3-methyl-butyrate and (condensation, dehydration, cyclisation, isomerisation,) ethyl 2-methyl-butyrate (Williams & Wagner, 1979). producing brown pigments of high molecular weight, The precursor acids are produced by oxidation of hop the melanoidins. a- and b-acids in beer as mentioned previously. In con- According to Rangel-Aldao et al. (2001), 3-DH is trast, some esters, such as iso-amyl acetate, can be present in considerable quantities in fresh beer (230 hydrolysed and their contribution to the flavour de- lM) and degrades during storage at 28 °C. Further- creases. Chemical hydrolysis and esterification are more, the concentration of a degradation intermediate acid-catalysed processes, but the activity of enzymes (3-DDH) also decreases strongly (Bravo, Sanchez, with esterase activity, sometimes detected in beer, can Scherer, & Rangel-Aldao, 2001a). In contrast, other also affect the ester profile. Neven (1997) showed that deoxyosones, 1-deoxy-2,3-hexodiulose (1-DH), 1,4-dide- some esterases are released by yeast into beer as a result oxyhexosulose (1-DDH) and 1,4-dideoxy-2,3-pentodiu- lose (1-DDP) increase (Bravo et al., 2001b). 1-DH is HC O OH formed by degradation of the Amadori product of hex- HC OH H O C oses, whereas 1-DDH and 1-DDP are probably formed O HO CH by Strecker degradation of 1-DH and 1-deoxy-pentodiu- HO O CH CHOH lose (1-DP). (CHOH ) HC OH 2 HO OH CH OH C OH 2 NHR NHR H2 H O H NR maltose pentose C C CH CH2 CHOH + RNH2 CHOH C O C OH (CHOH ) ( ) Maillard Maillard 3 CHOH 3 (CHOH ) (CHOH ) - H O 3 3 2 Reaction Reaction CH2OH CH OH CH OH 2 CH2OH 2 hexose Amadori product 1,2-enaminol

furfuryl H OH C H O H O alcohol C O C C O H CHOH 2 O + H O C O C O 2 C OH furfural CH2 CH (CHOH ) - H O - - RNH 3 2 H2O CH CH OH 2 CHOH CH + 3 2 CH2OH ACID CHOH CHOH ethanol catalysed SN2 CH OH CH OH 2 2 substitution 3-deoxy-2-hexosulose 3,4-dideoxyhexosulose-3-ene (3-DH) (3-DDH) - H2O

H CHO O CH C C 3 H O HOH C CHO O - 2 2 H H HOH2C O OH O 2 2 + H 5-hydroxymethyl furfural furfuryl ethyl ether (5-HMF) Fig. 12. Formation mechanism of furfuryl ethyl ether during beer Fig. 11. Formation of 5-hydroxymethyl furfural by Maillard reaction. storage and its precursor, furfuryl alcohol, during beer production. 372 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 of cell autolysis during fermentation and maturation. can easily react with many beer constituents, leading Such esterase activity is strain dependent and top-fer- to rapid changes in the flavour profile. Among these menting yeasts are more active than bottom fermenting processes are the oxidation of alcohols, hop bitter yeasts. The optimal activity in beer is between 15 and 20 compounds and polyphenols. Since oxygen is very °C. Furthermore, the enzyme is largely inactivated by detrimental for the flavour of beer, brewers have tried beer pasteurisation. Horsted, Dey, Holmberg, and Kiel- to minimize the oxygen pick-up in finished beer. Mod- land-Brandt (1998) showed that an extracellular esterase ern filling equipment can achieve total oxygen levels in of Saccharomyces cerevisiae had an optimal activity at the bottle of less than 0.1 mg/l. At such low oxygen pH 4–5 and identified TIP1 as the structural gene. The levels, it is debatable whether the formation of reac- activity of such esterases leads to biochemical aging pro- tive oxygen species (ROS) is the determining factor cesses in parallel with chemical aging reactions. Another in the aging of these beers. Indeed, other molecules effect of biochemical aging is due to proteases in beer present in beer have enough reactivity to interact which, by protein hydrolysis, cause less foam stability and form staling compounds. Beer staling is often (Ormrod, Lalor, & Sharpe, 1991). Especially, bottle- regarded as only the result of oxidation, but non- refermented beers, in contact with an inactive yeast layer oxidative processes may be just as important, espe- during storage, may become more susceptible to bio- cially at the low oxygen levels reached in modern chemical transformations (Vanderhaegen et al., 2003a). breweries. Non-oxidative reactions causing flavour deterioration 4.7. Formation of dimethyltrisulfide are esterifications, etherifications, Maillard reactions, glycoside and ester hydrolysis. Even (E)-2-nonenal, a Various precursor molecules in beer may trigger the compound long suspected to be the main cause of oxi- formation of dimethyltrisulfide (DMTS). According to dized flavour, paradoxically appears to arise by non- Peppard (1978), the reaction between methanesulfenic oxidative mechanisms in beer. This explains why beer acid and hydrogen sulfide leads to DMTS during beer staling is possible in the absence of oxygen. On the other storage. Methanesulfenic acid is formed by b-elimination hand, although some compounds result from oxidation from S-methylcysteine sulfoxide, introduced to beer from reactions, it is at present not really clear which com- hops. Other DMTS precursors may be 3-methylthiopro- pound(s) is/are responsible for the oxidation off-flavour pionalehyde and its reduced form, 3-methylthiopropanol of beer. (Gijs & Collin, 2002; Gijs et al., 2000). The production of In conclusion, some reactions have received consider- DMTS is enhanced at low pH (Gijs et al., 2002). ably more attention than others, partly due to historical factors. However, it remains important to evaluate the 4.8. Degradation of polyphenols relevance of specific reported reactions in the overall beer aging process. Such assessment has scarcely been Polyphenols in beer easily react with ROS and free done and it is currently not clear how important a spe- radicals. The structural changes due to oxidation have cific aging reaction is for the changes in flavour percep- not been completely elucidated. It is believed that simple tion of a particular beer. polyphenols polymerize to high molecular weight species Nonetheless, better knowledge of reaction mecha- (tannins), either by acid catalysis, or by oxidative mech- nisms involved in staling phenomena, allows closer anisms (Gardner & McGuinness, 1977). Possibly, poly- study of the effects on particular reactions of wort and phenols are first oxidized to quinones or semi-quinone beer production parameters. radicals, which interact with other phenolic compounds. Furthermore, polymerisation reactions also can be induced by acetaldehyde, formed by yeast or by ethanol 5. Inhibiting and promoting effects on beer aging reactions oxidation, through the formation of ethyl bridges between flavanols (Delcour, Dondeyne, Trousdale, & Sin- 5.1. Types of reactions gleton, 1982; Saucier, Bourgeois, Vitry, Roux, & Glories, 1997). Apart from polymerisation, ring opening Chemical and biochemical processes, which occur in oxidised phenols was proposed as an alternative deg- during beer storage, proceed simultaneously although radation mechanism (Cilliers & Singleton, 1990). During at different rates. To what extent certain reactions take beer storage, phenolic polymers interact with proteins place depends on storage conditions and by competition and form insoluble complexes and hazes. and interaction of pathways. This also applies to reactions during the brewing process, which determine the 4.9. Oxidative versus non-oxidative beer aging precursor concentrations for staling reactions in the final beer. Several methods have been suggested to control, to From the previous considerations, it becomes clear some degree, the reactions responsible for flavour dete- that oxygen triggers the release of free radicals, which rioration during beer storage. B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 373

5.2. Oxidative beer aging reactions (g) the redox potential (Buckee, Mom, Nye, & Hamond, 1997; Galic, Palic, & Cikovic, 1994; 5.2.1. General van Strien, 1987); Especially in bottled beer, excessive amounts of oxy- (h) the capacity to delay methyl linoleate oxidation in gen may cause a rapid change in aroma and taste. In re- lipidic media and at high temperature, followed by cent years, it became evident that levels of oxygen gas chromatography (Boivin et al., 1993; Maillard throughout the brewing process can also affect the beer & Berset, 1995); shelf-life downstream. Minimizing the formation and (i) linoleic acid hydroperoxide in a Fenton-type reac- Å Å activity of ROS (O2 , HOO ,H2O2 and HO )inbeer tion (Bright, Stewart, & Patino, 1999); and wort, must be a first step for improving beer flavour (j) the inhibition time of 2,20-azobis(2-amidinopro- stability. pane) dihydrochloride-induced oxidation of an Molecular oxygen itself is not very reactive but its ini- aqueous dispersion of linoleic acid (Liegeois, Ler- tial concentration determines the level of ROS. In the musieau, & Collin, 2000). activation of oxygen, transition metal ions (Cu+ and Fe2+) act as electron donors. Consequently, process The major endogenous anti- or pro-oxidants in wort and technological parameters should be adapted to min- and beer are discussed below. imize wort and beer oxygen pick-up and the copper and iron ion concentrations. 5.2.2. Sulfite The activation of oxygen can be stimulated by pro- Conversion of sulfate by yeast (from water and raw oxidant molecules, which are generally able to reduce materials) is the major endogenous source of sulfite in metal ions. In this process the pro-oxidant itself may be- beer. A study (Andersen, Outtrup, & Skibsted, 2000) come a radical, which reacts with other constituents or using the ESR lag phase method (e) highlighted sulfite degrades and may produce off-flavours. Actually, oxida- as one of the most effective antioxidants in beer. Its pres- tive reactions in wort and beer must be regarded as a ence postpones the formation of free radicals (mainly chain of redox agents involved in electron transfer reac- the 1-hydroxyethyl radical) measured by ESR spin trap- tions. On the other hand, the effects of oxygen can be ping. The effectiveness of sulfite seems to be due to its inhibited by certain beer or wort components (anti-oxi- two electron non-radical-producing reaction with dants). Generally, the anti-oxidant activity is based on peroxides. the capture of ROS and free radicals. The capture of metal ions with some chelating agents is another anti-oxi- 5.2.3. Polyphenols dative approach. Polyphenolic compounds are important antioxi- In the past few years, the anti- or pro-oxidative activ- dants in many systems. Generally, in beer, 70–80% ity of wort and beer has been investigated by various of the polyphenol fraction originates from barley malt methods including the determination of: and another 20–30% from hop. Lower molecular weight polyphenols, in particular, are excellent anti- (a) the capacity to reduce the iron-(II)-dipyridyl com- oxidants. With increasing molecular weight, the reduc- plex (Chapon, Louis, & Chapon, 1981); ing power decreases (Buggey, 2001). Polyphenols can (b) the ability to scavenge the radical cation of react with free radicals to produce phenoxy radicals 2,20-azinobis-(3-ethylbenzothiazoline-6-sulfonate) (Fig. 13), which are relatively stable due to delocaliza- (ABTS) in an aqueous phase (Araki et al., 1999); tion of the free radical over the aromatic ring. Some (c) the 1,1-diphenyl-2-picrylhydrazyl (DPPH) free polyphenols are also anti-oxidants, by their ability to radical-scavenging activity (Kaneda, Kobayashi, chelate transition metal ions. On the other hand, cer- Furusho, Sahara, & Koshino, 1995a, 1995b); tain polyphenols behave as pro-oxidants due to their (d) chemiluminescence (CL), either directly or after ability to transfer electrons to transition metal ions reaction with the radical scavenger, 2-methyl-6- (Bamforth, 1999b). phenyl-3,7-dihydroimidazo(1,2-a)pyrazin-3-one There is some controversy concerning the relevance (Kaneda, Kano, Kamimura, Kawaskishi, & of polyphenols as anti-oxidants in beer and wort. ESR Osawa, 1991; Kaneda, Kano, Kamimura, lag phase studies (Andersen et al., 2000; Andersen & Osawa, & Kawakishi, 1990a, Kaneda, Kano, Kamimura, Osawa, & Kawakishi, 1990b; Kan- eda, Kano, Osawa, Kawakishi, & Koshino, O O O O 1994); (e) free radicals by electron spin resonance (ESR) C C (Kaneda et al., 1989; Kaneda et al., 1988); (f) 2-thiobarbituric acid-reactive substances (Grigsby C & Palamand, 1976); Fig. 13. Stabilization of phenoxy radical by delocalization. 374 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381

Skibsted, 2001) showed no significant effect of polyphe- is a typical reductone, but it is not produced by Maillard nols on the formation of free radicals in beer during reactions. However, in the production of beer, ascorbic storage or in wort during brewing. This was attributed acid is often used as an exogenous anti-oxidant. Re- to the extreme reactivity of hydroxyl radicals and their cently, ESR studies questioned the relevance of ascorbic non-selective elimination through reaction with other acid for flavour stability. On addition to beer rather a prominent compounds of beer (ethanol) or wort (sug- pro-oxidative activity was found due to the formation ars). This avoids radical-scavenging with polyphenols of more free radicals (Andersen et al., 2000). present in relatively small concentrations. However, it A limited number of studies (Wijewickreme, Kitts, & is not clear whether this applies also to other radicals, Durance, 1997; Wijewickreme & Kitts, 1998b; Wijewick- such as, e.g., fatty acid oxidation radicals in wort. reme, Krejpcio, & Kitts, 1999) are related to the antiox- In beer, polyphenols contribute up to 60% of the idant properties of melanoidins and conclusions endogenous reducing power measured in the iron-(II)- concerning the structural features responsible for anti- dipyridyl test (a) and the DPPH test (c) (Kaneda oxidative activity are difficult to draw. Moreover, mela- et al., 1995a; McMurrough et al., 1996). Partial removal noidins or its precursors may also present pro-oxidative of the polyphenol fraction by polyvinylpolypyrrolidone properties as Hashimoto (1972) showed their involve- (PVPP) treatment diminishes the reducing power by 9– ment in the oxidation of alcohols to aldehydes during 38%, but does not make the beer more susceptible to beer storage. The levels of antioxidants resulting from oxidative damage (McMurrough et al., 1996). This was Maillard reactions and sugar caramelization are low in confirmed in sensory experiments by Mikyska, Hrabak, light malts, but significant in dark speciality malts Haskova, and Srogl (2002). The PVPP-treated beers (Bright et al., 1999; Coghe, Vanderhaegen, Pelgrims, developed a less astringent character. Moreover, accord- Basteyns, & Delvaux, 2003; Griffiths & Maule, 1997). ing to Walters et al. (1997b) and Walters, Heasman, and The higher reducing power of wort and beer produced Hughes (1997a), (+)-catechin and ferulic acid reduce the from darker coloured malts may then contribute to a formation of particular carbonyl compounds in beer at better flavour stability, often reported for such beers. high oxygen levels, but not at low oxygen levels. Fur- thermore, ferulic acid levels determine whether it is 5.2.5. Chelating agents pro-oxidant (low concentration) or anti-oxidant (high- Apart from polyphenols, various other compounds in concentration). wort and beer, including amino acids, phytic acid and The main effect of polyphenols on flavour stability is melanoidins (Wijewickreme & Kitts, 1998a), may func- probably situated in the mashing and wort boiling steps tion as sequestration agents for metal ions. In wort (Liegeois et al., 2000; Mikyska et al., 2002). In particu- and beer, an equilibrium exists between free and che- lar, polyphenols extracted from hop during wort boiling lated metal ions. Depending on chelator type, bound significantly contribute to the reducing power and metal ions have either less or more capability to promote effectively diminish the nonenal potential of wort (Ler- oxygen radical formation (Bamforth, 1999b). musieau, Liegeois, & Collin, 2001). Sensory experiments (Mikyska et al., 2002) also confirm the positive effects of 5.3. Enzymatic oxidation of fatty acids hop polyphenols, during brewing, on flavour stability. Many strategies have been proposed for reducing the enzymatic oxidation of fatty acids. Lipoxygenase activ- 5.2.4. Melanoidins and reductones ity during mashing can be controlled by several techno- Malt kilning (up to 80 °C), malt roasting (110–250 logical parameters. LOX enzyme activity during °C) and wort boiling generate antioxidants through mashing is influenced by the temperature regime and Maillard reactions (Boivin et al., 1993). These antioxi- the wort pH. Mashing-in at high temperatures (>65 dants include reductones and melanoidins. The reducing °C) effectively inhibits LOX enzymes (Kobayashi et al., power of reductones is due to the endiol group (Fig. 14), 1993a). However, this condition is not very acceptable, which can generate carbonyls. Ascorbic acid (vitamin C) as this temperature inactivates other indispensable malt enzymes: amylases, glucanases or proteases. Lowering the mash pH from 5.4 to 5.1 seems more efficient for reducing LOX activity (Kobayashi et al., 1993a). C O C O Lipoxygenase in wort also appears to be inhibited by Ox. C OH C O polyphenols (Goupy, Hugues, Boivin, & Amiot, 1999). Another approach consists of limiting the extraction C OH C O into the mash of lipoxygenase enzymes. This is possible by selecting malts with low LOX contents or by reduc- endiol group ing the LOX activity by kilning at intense regimes. Mill- Fig. 14. Oxidation of a reductone. ing regimes that leave the embryo intact, were also Table 2 Strategies for dealing with beer staling according to Bamforth (2000b) Process stage Barley Malting Grist Milling Mashing & Wort Boiling & Fermentation & Downstream Packaging & collection clarification conditioning processing distribution High relevance – Avoidance of Strategies to Oxygen Minimum oxygen low cost/risk excessively enhance sulfur minimization uptake in package prolonged dioxide within 357–381 (2006) 95 Chemistry Food / al. et Vanderhaegen B. heating legal limits Use of sulfur dioxide Minimum pick-up of iron and copper High relevance – Refrigerated high cost/risk stockholding and transportation Stock rotation and logistics Oxygen-scavenger crown corks Low relevance – Suppression of Increased use of Suppressing Bright worts Use of low cost/risk embryo development sugar adjuncts oxygen ingress ascorbic acid – e.g., rootlet inhibitors Maximized kilning Use of coloured High mashing- Use of reduced temperatures malts in temperatures iso-a-acids Reduced mashing pH Low relevance – Selection of Sparging of grist Milling regimes high cost/risk barleys with with inert gas which minimize low LOX embryo damage potential Anaerobic milling 375 376 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 proposed (van Waesberghe, 1997). Later, Bamforth On the other hand, Maillard reactions are inhibited (1999a) suggested that the availability of oxygen in the by sulfite (Wedzicha & Kedward, 1995). Together with mash is more likely to limit lipoxygenase activity than its ability to inhibit radical formation and to bind with the availability of enzymes. This was confirmed by carbonyl compounds, sulfite seems to be a very good Kobayashi et al. (2000b). A prevention of oxygen in- inhibitor of beer staling. gress during mashing reduces the enzymatic oxidation of fatty acids. 6. Concluding remarks 5.4. Non-oxidative beer aging reactions Optimisation of the brewing process with respect to flavour stability requires a clear insight of the types of Non-oxidative reactions in stored beer are of very dif- flavour changes during storage and the nature of the ferent natures. Consequently, the effects of production molecules involved. This may, however, vary between and storage parameters are highly variable. Neverthe- beer types (e.g., pilsner beers and speciality beers). Sec- less, several non-oxidative aging processes, such as the ondly, it is necessary to clarify the reaction pathways release of aldehydes from imines, esterification, etherifi- in beer leading to the staling compounds. Finally, the cation, ester hydrolysis, dimethyltrisulfide formation influence of the production process on the staling reac- and glycoside hydrolysis, are all promoted at a low beer tions must be made clear. pH. A low pH may also enhance oxidative reactions by Knowledge of the aging phenomenon in a particu- protonation of superoxide ðO2 Þ radicals to the more Å lar type of beer can be used to develop appropriate reactive perhydroxyl radicals (HOO ). All this supports technological process improvements to control its par- the sensory findings that beer ages faster at low pH ticular flavour stability. Besides their relevance for fla- (Kaneda, Takashio, Tomaki, & Osawa, 1997). Shimizu vour stability, the investment costs for suggested et al. (2001b) correlated the decrease in pH during fer- process modifications must be evaluated and a balance mentation with the cellular size of the pitching yeast. should be made between better and longer flavour sta- A higher pH was obtained with yeast large-cell sizes bility and costs. Table 2, presented by Bamforth and the resulting beers showed better flavour stability (2000a, 2000b), is helpful in summarizing such consid- (Shimizu, Araki, Kuroda, Takashio, & Shinotsuka, erations. Although the strategies seem straightforward, 2001a). Yeast also has an important role in decreasing practical experience may soon show that flavour sta- the amount of staling compounds and its precursors. bility is still hard to control. There might be one Yeast metabolism during the fermentation is mainly fer- explanation for this: the knowledge of staling compo- mentative. This results in an excess of reduced coen- nents is still incomplete, not only concerning the num- zymes NADH and NADPH. To regenerate these ber and types of compounds, involved but also the coenzymes, they are used by several aldoketoreductases reaction mechanisms. (Debourg, Laurent, Dupire, & Masschelein, 1993; De- bourg, Verlinden, Van De Winkel, Masschelein, & Van Nedervelde, 1995; Van Iersel, Eppink, Van Berkel, Rombouts, & Abee, 1997; Van Nedervelde, Oudjama, References Desmedt, & Debourg, 1999; Van Nedervelde, Verlinden, Philipp, & Debourg, 1997). This so-called ‘‘yeast reduc- Ahrenst-Larsen, B., & Levin Hansen, H. (1963). Gaschromatograph- ing power’’ results in the reduction of wort aldehydes to ische Untersuchungen u¨ber die Geschmaksstabilita¨t von Bier. alcohols during fermentation. Recently, an interesting Brauwissenschaft, 16, 393–397. yeast NADPH-dependent reductase was described (San- Andersen, M. L., Outtrup, H., & Skibsted, L. H. (2000). Potential antioxidants in beer assessed by ESR spin trapping. Journal of chez et al., 2001), capable of reducing a-dicarbonyl Agricultural and Food Chemistry, 48, 3106–3111. intermediates of the Maillard reaction, such as 3-deoxy- Andersen, M. L., & Skibsted, L. H. (1998). Electron spin resonance 2-hexosulose (3-DH). However, further investigations spin trapping identification of radicals formed during aerobic must demonstrate whether, under beer fermentation forced aging of beer. Journal of Agricultural and Food Chemistry, conditions, yeast can reduce such Maillard reaction 46, 1272–1275. Andersen, M. L., & Skibsted, L. H. (2001). Modification of the levels dicarbonyls. of polyphenols in wort and beer by addition of hexamethylenetet- Maillard products such as furfural increased more ramine or sulfite during mashing. Journal of Agricultural and Food during beer aging when malt was kilned at higher tem- Chemistry, 49, 5232–5237. peratures and the thermal load on wort was higher dur- Angelino, S. A. G. F., Kolkman, J. R., van Gemert, L. J., Vogels, J. T. ing production (Narziss, Back, Miedaner, & Lustig, W. E., van Lonkhuijsen, H. J., & Douma, A. C. (1999). Flavour stability of pilsener beer. Proceedings of the European Brewery 1999). The staling compound, furfuryl ethyl ether, Convention Congress, 103–112. showed the same behaviour (Vanderhaegen et al., Araki, S., Kimura, T., Shimizu, C., Furusho, S., Takashio, M., & 2004b). Shinotsuka, K. (1999). Estimation of antioxidative activity and its B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 377

relationship to beer flavor stability. Journal of the American Society Bright, D., Stewart, G. G., & Patino, H. (1999). A novel assay for of Brewing Chemists, 57, 34–37. antioxidant potential of specialty malts. Journal of the American Araki, S., Takashio, M., & Shinotsuka, K. (2002). A new parameter Society of Brewing Chemists, 57, 133–137. for determination of the extent of staling in beer. Journal of the Buckee, G. K., Mom, M., Nye, J. W. S., & Hamond, R. V. (1997). American Society of Brewing Chemists, 60, 26–30. Measurement and significance of oxidation–reduction levels in Bamforth, C. W., & Parsons, R. (1985). New procedures to improve beer. Proceedings of the European Brewery Convention Congress, the flavor stability of beer. Journal of the American Society of 607–614. Brewing Chemists, 43, 197–202. Buggey, L. A. (2001). A review of polyphenolic antioxidants in hops, Bamforth, C. W. (1999a). Enzymic and non-enzymic oxidation in the brewing and beer. The brewer international, 4, 21–25. brewhouse: A theoretical consideration. Journal of the Institute of Bureau, S. M., Baumes, R. L., & Razungles, A. J. (2000). Effects of Brewing, 105, 237–242. vine or bunch shading on the glycosylated flavor precursors in Bamforth, C. W. (1999b). The science and understanding of the grapes of Vitis vinifera L. Cv. Syrah. Journal of Agricultural and flavour stability of beer: a critical assessment. Brauwelt Interna- Food Chemistry, 48, 1290–1297. tional, 98–110. Chapon, L., Louis, C., & Chapon, S. (1981). Estimation of the Bamforth, C. W. (2000a). Beer quality: oxidation. BrewersÕ Guardian, reducing power of beer using an iron-dipyridyl complex. Proceed- 129, 31–34. ings of the European Brewery Convention Congress, 13, 307–322. Bamforth, C. W. (2000b). Making sense of flavor changes in beer. Chevance, F., Guyot-Declerck, C., Dupont, J., & Collin, S. (2002). MBAA Technical Quarterly, 37, 165–171. Investigation of the beta-damascenone level in fresh and aged Barker, R. L., Gracey, D. E. F., Irwin, A. J., Pipasts, P., & Leiska, E. commercial beers. Journal of Agricultural and Food Chemistry, 50, (1983). Liberation of staling aldehydes during storage of beer. 3818–3821. Journal of the Institute of Brewing, 89, 411–415. Cilliers, J. J. L., & Singleton, V. L. (1990). Autoxidative phenolic ring- Barker, R. L., Pipasts, P., & Gracey, D. E. F. (1989). Examination of opening under alkaline conditions as a model for natural polyphe- beer carbonyls as their oximes by gas chromatography–mass nols in food. Journal of Agricultural and Food Chemistry, 38, spectrometry. Journal of the American Society of Brewing Chemists, 1797–1798. 47, 9–14. Clapperton, J. F. (1976). Ribes flavour in beer. Journal of the Institute Basarova, G., Savel, J., Janousek, J., & Cizkova, H. (1999). Changes in of Brewing, 82, 175–176. the content of the amino-acids in spite of the natural aging of beer. Coghe, S., Vanderhaegen, B., Pelgrims, B., Basteyns, A., & Delvaux, Monatsschrift Fur Brauwissenschaft, 52, 112–118. F. R. (2003). Characterization of dark specialty malts: new insights Baxter, E. D. (1982). Lipoxidases in Malting and Mashing. Journal of in colour evaluation and pro- and antioxidative activity. Journal of the Institute of Brewing, 88, 390–396. the Amercian Society of Brewing Chemists, 61, 125–132. Baxter, E. D. (1984). Recognition of 2 lipases from barley and green Dale, R. S., & Pollock, J. R. A. (1977). A means to reduce formation of malt. Journal of the Institute of Brewing, 90, 277–281. precursors of 2-trans-nonenal in brewing. Journal of the Institute of Belitz, H. D., & Grosch, W. (1999). Food chemistry. Berlin Heidelberg: Brewing, 83, 88–91. Springer Verlag. Dalgliesh, C. E. (1977). Flavour stability. Proceedings of the European Bernstein, L., & Laufer, L. (1977). Further studies on furfural: the Brewery Convention Congress, 623–659. influence of raw materials, processing conditions and pasteuriza- De Cooman, L., Aerts, G., Overmeire, H., & De Keukeleire, D. (2000). tion temperatures. Journal of the American Society of Brewing Alterations of the profiles of iso-alpha-acids during beer ageing, Chemists, 35, 21–35. marked instability of trans-iso-alpha-acids and implications for Biendl, M., Kollmannsberger, H., & Nitz, S. (2003). Occurence of beer bitterness consistency in relation to tetrahydroiso-alpha-acids. glycosidically bound flavour compounds in different hop products. Journal of the Institute of Brewing, 106, 169–178. Proceedings of the European Brewery Convention Congress, Debourg, A., Laurent, M., Dupire, S., & Masschelein, C. A. (1993). 252–259. The specific role and interaction of yeast enzymatic systems in the Blockmans, C., Devreux, A., & Masschelein, C. A. (1975). Formation removal of flavour-potent wort carbonyls during fermentation. de compose´s carbonyles et alteration du gouˆt de la bie`re. Proceedings of the European Brewery Convention Congress, Proceedings of the European Brewery Convention Congress, 437–444. 699–713. Debourg, A., Laurent., M., Verlinden, V., Van De Winkel, L., Bohmann, J. J. (1985a). Aging behavior of beer. 4. Aging trials Masschelein, C. A., & Van Nedervelde, L. (1995). Yeast physio- combined with gassing by carbon-dioxide, nitrogen, air and logical state on carbonyl reduction. EBC-Symposium immobilized oxygen. Monatsschrift Fur Brauwissenschaft, 38, 175–180. yeast applications in the brewing industry. Monograph XXIV (pp. Bohmann, J. J. (1985b). Determination of the aging behavior of beer. 158–174). 2. The behavior of some volatile compounds under heat-treatment. Delcour, J. A., Dondeyne, P., Trousdale, E. K., & Singleton, V. L. Monatsschrift Fur Brauwissenschaft, 38, 79–85. (1982). The reactions between polyphenols and aldehydes and the Boivin, P., Allain, D., Clamagirant, V., Maillard, M. N., Cuvelier, M. influence of acetaldehyde on haze formation in beer. Journal of the E., Berset, C., Richard, H., Nicolas, J., & Forget-Richard, F. Institute of Brewing, 88, 234–243. (1993). Mesure de lÕactivite´ antioxyge`ne de lÕorge et du malt: Devreux, A., Blockmans, C., & vande Meersche, J. (1981). Carbonyl approche multiple. Proceedings of the European Brewery Conven- compounds formation during aging of beer. EBC-Flavour sympo- tion Congress, 397–404. sium. Monograph VII (pp. 191–201). Bravo, A., Sanchez, B., Scherer, E., & Rangel-Aldao, R. (2001a). Doderer, A., Kokkelink, I., van der Veen, S., Valk, B. E., & Douma, Highly sensitive chemical indices of beer aging. Proceedings of the A. C. (1991). Purification and characterization of lipoxygenase European Brewery Convention Congress, 595–601. from germinating barley. Proceedings of the European Brewery Bravo, A., Scherer, E., Madrid, J., Herrera, J., Virtanen, H., & Rangel- Convention Congress, 109–116. Aldao, R. (2001b). Identification of a-dicarbonylic compounds in Drost, B. W., van den Berg, R., Freijee, F. J. M., van der Velde, E. G., aged beers: their role in beer aging process. Proceedings of the & Hollemans, M. (1990). Flavor stability. Journal of the American European Brewery Convention Congress, 602–611. Society of Brewing Chemists, 48, 124–131. Brenner, M. W., & Khan, A. A. (1976). Furfural and beer colour as Drost, B. W., Van Eerde, P., Hoekstra, S. F., & Strating, J. (1971). indices of beer flavor deterioration. Journal of the American Society Fatty acids and staling of beer. Proceedings of the European of Brewing Chemists, 34, 14–19. Brewery Convention Congress, 451–458. 378 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381

Dufour, J. P., Leus, M., Baxter, A. J., & Hayman, A. R. (1999). Hashimoto, N. (1972). Oxidation of higher alcohols by melanoidins in Characterization of the reaction of bisulfite with unsaturated beer. Journal of the Institute of Brewing, 78, 43–51. aldehydes in a beer model system using nuclear magnetic resonance Hashimoto, N. (1981). Flavour stability of packaged beers. In J. R. A. spectroscopy. Journal of the American Society of Brewing Chemists, Pollock (Ed.), Brewing Science (pp. 347–405). London: Academic 57, 138–144. Press. Eichhorn, P., Komori, T., Miedaner, H., & Narziss, L. (1989). Hashimoto, N., & Eshima, T. (1977). Composition and pathway of Alterungscarbonyle im sub-ppb bereich. Proceedings of the Euro- formation of stale aldehydes in bottled beer. Journal of the pean Brewery Convention Congress, 717–724. American Society of Brewing Chemists, 35, 145–150. Engan, S. (1969). Some changes in beer flavour during aging. Journal Hashimoto, N., & Eshima, T. (1979). Oxidative degradation of of the Institute of Brewing, 75, 371–376. isohumulones in relation to flavour stability of beer. Journal of Foster, R. T., Samp, E. J., & Patino, H. (2001). Multivariate modeling the Institute of Brewing, 85, 136–140. of sensory and chemical data to understand staling in light beer. Hashimoto, N., & Kuroiwa, Y. (1975). Proposed pathways for the Journal of the American Society of Brewing Chemists, 59, 201–210. formation of volatile aldehydes during storage of bottled beer. Furusho, S., Kobayashi, N., Nakae, N., Takashio, M., Tamaki, T., & Proceedings of the American Society of Brewing Chemists, 33, Shinotsuka, K. (1999). A developed descriptive sensory test reveals 104–111. beer flavor changes during storage. MBAA Technical Quarterly, 36, Hill, P., Lustig, S., & Sawatzki, V. (1998). The amino acid glutamin as 163–166. a parameter for the determination of the extent of staling in beer. Galic, K., Palic, A., & Cikovic, N. (1994). On the redox potential in Monatsschrift Fur Brauwissenschaft, 51, 36–38. brewing process – a review. Monatsschrift Fur Brauwissenschaft, 47, Hofmann, T., & Schieberle, P. (1997). Identification of potent aroma 124–128. compounds in thermally treated mixtures of glucose/cysteine and Garbe, L., Hu¨bke, H., & Tressl, R. (2003). Oxygenated fatty acids and rhamnose/cysteine using aroma extract dilution techniques. Journal flavour stability – new insights. Proceedings of the European of Agricultural and Food Chemistry, 45, 898–906. Brewery Convention Congress, 740–748. Holtman, W. L., VredenbregtHeistek, J. C., Schmitt, N. F., & Gardner, R. J., & McGuinness, J. D. (1977). Complex phenols in Feussner, I. (1997). Lipoxygenase-2 oxygenates storage lipids in brewing: a critical survey. MBAA Technical Quarterly, 14, embryos of germinating barley. European Journal of Biochemistry, 250–260. 248, 452–458. Gijs, L., Chevance, F., Jerkovic, V., & Collin, S. (2002). How low pH Horsted, M. W., Dey, E. S., Holmberg, S., & Kielland-Brandt, M. C. can intensify beta-damascenone and dimethyl trisulfide production (1998). A novel esterase from Saccharomyces carlsbergensis, a through beer aging. Journal of Agricultural and Food Chemistry, 50, possible function for the yeast TIP1 gene. Yeast, 14, 793–803. 5612–5616. Hugues, M., Boivin, P., Gauillard, F., Nicolas, J., Thiry, J. M., & Gijs, L., & Collin, S. (2002). Effect of the reducing power of a beer on Richardforget, F. (1994). 2 Lipoxygenases from germinated barley dimethyltrisulfide production during aging. Journal of the American – heat and kilning stability. Journal of Food Science, 59, 885–889. Society of Brewing Chemists, 60, 68–70. Huvaere, K., Andersen, M. L., Olsen, K., Skibsted, L. H., Heyerick, Gijs, L., Perpete, P., Timmermans, A., & Collin, S. (2000). 3- A., & De Keukeleire, D. (2003). Radicaloid-type oxidative decom- Methylthiopropionaldehyde as precursor of dimethyl trisulfide in position of beer bittering agents revealed. Chemistry-a European aged beers. Journal of Agricultural and Food Chemistry, 48, Journal, 9, 4693–4699. 6196–6199. Irwin, A. J., Barker, R. L., & Pipasts, P. (1991). The role of copper, Goupy, P., Hugues, M., Boivin, P., & Amiot, M. J. (1999). Antiox- oxygen, and polyphenols in beer flavor instability. Journal of the idant composition and activity of barley (Hordeum vulgare) and American Society of Brewing Chemists, 49, 140–148. malt extracts and of isolated phenolic compounds. Journal of the Jamieson, A. M., Chen, E. C., & Van Gheluwe, J. E. A. (1969). A Science of Food and Agriculture, 79, 1625–1634. study of the cardboard flavour in beer by gas chromatography. Graveland, A., Pesman, L., & Van Eerde, P. (1972). Enzymatic Proceedings of the American Society of Brewing Chemists, oxidation of linoleic acid in barley suspensions. MBAA Technical 123–126. Quarterly, 9, 98–104. Jamieson, A. M., & Chen, E. C. H. (1972). Determination of carbonyl Greenhoff, K., & Wheeler, R. E. (1981a). Analysis of beer carbonyls at compounds by flash exchange gas chromatography. Proceedings of the part per billion level by combined liquid chromatography and the American Society of Brewing Chemists, 92–93. high pressure liquid chromatography. Journal of the Institute of Jamieson, A. M., & Van Gheluwe, J. E. A. (1970). Identification of a Brewing, 86, 35–41. compound responsible for cardboard flavor in beer. Proceedings of Greenhoff, K., & Wheeler, R. E. (1981b). Evaluation of stale flavour the American Society of Brewing Chemists, 192–197. and beer carbonyl development during normal and accelerated Kaneda, H., Kano, Y., Kamimura, M., Kawaskishi, S., & Osawa, T. aging utilizing liquid and high pressure liquid chromatography. (1991). A study of beer staling using chemiluminescence analysis. Proceedings of the European Brewery Convention Congress, Journal of the Institute of Brewing, 97, 105–109. 405–412. Kaneda, H., Kano, Y., Kamimura, M., Osawa, T., & Kawakishi, S. Griffiths, D. J., & Maule, A. P. (1997). The effects of different cereal (1990a). Detection of chemiluminescence produced during beer coloured malts on reducing power during the brewing process. oxidation. Journal of Food Science, 55, 881–882. Proceedings of the European Brewery Convention Congress, Kaneda, H., Kano, Y., Kamimura, M., Osawa, T., & Kawakishi, S. 267–274. (1990b). Evaluation of beer deterioration by chemiluminescence. Grigsby, J. H., & Palamand, S. R. (1976). The use of 2-thiobarbituric Journal of Food Science, 55, 1361–1364. acid in the measurement of beer oxidation. Journal of the American Kaneda, H., Kano, Y., Koshino, S., & Ohyanishiguchi, H. (1992). Society of Brewing Chemists, 34, 49–55. Behavior and role of iron ions in beer deterioration. Journal of Gro¨nqvist, A., Siirila¨, J., Virtanen, H., Home, S., & Pajunen, E. (1993). Agricultural and Food Chemistry, 40, 2102–2107. Carbonyl compounds during beer production and in beer. Pro- Kaneda, H., Kano, Y., Osawa, T., Kawakishi, S., & Kamada, K. ceedings of the European Brewery Convention Congress, 421–428. (1989). The role of free radicals in beer oxidation. Journal of the Harayama, K., Hayase, F., & Kato, H. (1994). Evaluation by a American Society of Brewing Chemists, 47, 49–53. multivariate-analysis of the stale flavor formed while storing beer. Kaneda, H., Kano, Y., Osawa, T., Kawakishi, S., & Kamimura, M. Bioscience Biotechnology and Biochemistry, 58, 1595–1598. (1990). Effect of free radicals on haze formation in beer. Journal of Hashimoto, N. (1966). Rep. Res. Lab. Kirin Brewery Co., 9,1. Agricultural and Food Chemistry, 38, 1909–1912. B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 379

Kaneda, H., Kano, Y., Osawa, T., Kawakishi, S., & Koshino, S. Liegeois, C., Meurens, N., Badot, C., & Collin, S. (2002). Release of (1991). Role of active oxygen on deterioration of beer flavor. deuterated (E)-2-nonenal during beer aging from labeled precur- Proceedings of the European Brewing Convention Congress, sors synthesized before boiling. Journal of Agricultural and Food 433. Chemistry, 50, 7634–7638. Kaneda, H., Kano, Y., Osawa, T., Kawakishi, S., & Koshino, S. Lustig, S., Miedaner, H., Narziss, L., & W., B. (1993). Untersuchungen (1994). Role of beer components on chemiluminescence production flu¨chtiger aromastoffe bei der bieralterung mittels multidimensio- during beer storage. Journal of the American Society of Brewing naler gaschromatographie. Proceedings of the European Brewery Chemists, 52, 70–75. Convention Congress, 445–452. Kaneda, H., Kano, Y., Osawa, T., Ramarathnam, N., Kawakishi, S., Madigan, D., Perez, A., & Clements, M. (1998). Furanic aldehyde & Kamada, K. (1988). Detection of free-radicals in beer oxidation. analysis by HPLC as a method to determine heat-induced flavor Journal of Food Science, 53, 885–888. damage to beer. Journal of the American Society of Brewing Kaneda, H., Kobayashi, M., Furusho, S., Sahara, H., & Koshino, S. Chemists, 56, 146–151. (1995a). Reducing activity and flavor stability. MBAA Technical Maillard, M. N., & Berset, C. (1995). Evolution of antioxidant activity Quarterly, 32, 90–94. during kilning – role of insoluble bound phenolic-acids of barley Kaneda, H., Kobayashi, M., Takashio, M., Tamaki, T., & Shinotsuka, and malt. Journal of Agricultural and Food Chemistry, 43, K. (1999). Beer staling mechanism. MBAA Technical Quarterly, 36, 1789–1793. 41–47. McMurrough, I., Madigan, D., & Kelly, R. J. (1997). Evaluation of Kaneda, H., Kobayashi, N., Furusho, S., Sahara, H., & Koshino, S. rapid colloidal stabilization with polyvinylpolypyrrolidone (PVPP). (1995b). Chemical evaluation of beer flavor stability. MBAA Journal of the American Society of Brewing Chemists, 55, 38–43. Technical Quarterly, 32, 76–80. McMurrough, I., Madigan, D., Kelly, R. J., & Smyth, M. R. (1996). Kaneda, H., Takashio, M., Osawa, T., Kawakishi, S., & Tamaki, T. The role of flavanoid polyphenols in beer stability. Journal of the (1996). Behavior of sulfites during fermentation and storage of American Society of Brewing Chemists, 54, 141–148. beer. Journal of the American Society of Brewing Chemists, 54, Meilgaard, M. (1972). Stale flavor carbonyls in brewing. Brewers 115–120. Digest, 47, 48–57. Kaneda, H., Takashio, M., Tomaki, T., & Osawa, T. (1997). Influence Meilgaard, M., Ayma, M., & Ruano, J. I. (1971). A study of carbonyl of pH on flavour staling during beer storage. Journal of the Institute compounds in beer. IV. Carbonyl compounds identified in heat- of Brewing, 103, 21–23. treated wort and beer. Proceedings of the American Society of King, B. M., & Duineveld, C. A. A. (1999). Changes in bitterness as Brewing Chemists, 219–229. beer ages naturally. Food Quality and Preference, 10, 315–324. Meilgaard, M., Elizondo, A., & Moya, E. (1970). A study of Kobayashi, N., Kaneda, H., Kano, Y., & Koshino, S. (1993a). Lipid carbonyl compounds in beer – 2. Flavor and flavor thresholds of oxidation during wort production. Proceedings of the European aldehydes and ketones added to beer. MBAA Technical Quarterly, Brewery Convention Congress, 405–412. 7, 143–150. Kobayashi, N., Kaneda, H., Kano, Y., & Koshino, S. (1993b). The Meilgaard, M., & Moya, E. (1970). A study of carbonyl compounds in production of linoleic acid hydroperoxides during mashing. Journal beer – 1. Background and literature review. MBAA Technical of Fermentation and Bioengineering, 76, 371–375. Quarterly, 7, 135–142. Kobayashi, N., Kaneda, H., Kano, Y., & Koshino, S. (1994). Behavior Meilgaard, M. C. (1975a). Flavor chemistry of beer; Part I: flavor of lipid hydroperoxides during mashing. Journal of the American interaction between principal volatiles. MBAA Technical Quarterly, Society of Brewing Chemists, 52, 141–145. 12, 107–117. Kobayashi, N., Kaneda, H., Kuroda, H., Kobayashi, M., Kurihara, Meilgaard, M. C. (1975b). Flavor chemistry of beer; Part II: flavor and T., Watari, J., & Shinotsuka, K. (2000a). Simultaneous determi- threshold of 239 aroma volatiles. MBAA Technical Quarterly, 12, nation of mono-, di-, and trihydroxyoctadecenoic acids in beer and 151–168. wort. Journal of the Institute of Brewing, 106, 107–110. Miedaner, H., Narziss, L., & Eichhorn, P. (1991). Einige faktoren der Kobayashi, N., Kaneda, H., Kuroda, H., Watari, J., Kurihara, T., & geschmacksstabilita¨t – sensorische und analytische bewertung. Shinotsuka, K. (2000b). Behavior of mono-, di-, and trihydroxy- Proceedings of the European Brewery Convention Congress, octadecenoic acids during mashing and methods of controlling 401–408. their production. Journal of Bioscience and Bioengineering, 90, Mikyska, A., Hrabak, M., Haskova, D., & Srogl, J. (2002). The role of 69–73. malt and hop polyphenols in beer quality, flavour and haze Kuroda, H., Furusho, S., Maeba, H., & Takashio, M. (2003). stability. Journal of the Institute of Brewing, 108, 78–85. Characterization of factors involved in the production of 2(E)- Narziss, L., Back, W., Miedaner, H., & Lustig, S. (1999). Investiga- nonenal during mashing. Bioscience Biotechnology and Biochemis- tions to influence the taste stability by varying technological try, 67, 691–697. parameters when making beer. Monatsschrift Fur Brauwissenschaft, Kuroda, H., Kobayashi, N., Kaneda, H., Watari, J., & Takashio, M. 52, 192–206. (2002). Characterization of factors that transform linoleic acid into Narziss, L., Miedaner, H., & Eichhorn, P. (1999a). Studies on taste di- and trihydroxyoctadecenoic acids in mash. Journal of Bioscience stability of beer. Monatsschrift Fur Brauwissenschaft, 52, 80–85. and Bioengineering, 93, 73–77. Narziss, L., Miedaner, H., & Eichhorn, P. (1999b). Studies on taste Lermusieau, G., Liegeois, C., & Collin, S. (2001). Reducing power of stability of beer – Part 1. Monatsschrift Fur Brauwissenschaft, 52, hop cultivars and beer ageing. Food Chemistry, 72, 413–418. 49–57. Lermusieau, G., Noel, S., Liegeois, C., & Collin, S. (1999). Nonox- Narziss, L., Miedaner, H., & Graf, H. (1985). Carbonyls and aging of idative mechanism for development of trans-2-nonenal in beer. beer. 2. Influence of some technical parameters. Monatsschrift Fur Journal of the American Society of Brewing Chemists, 57, 29–33. Brauwissenschaft, 38, 472–477. Lewis, M. J., Pangborn, R. M., & Tanno, L. A. S. (1974). Sensory Narziss, L., Miedaner, H., & Lustig, S. (1999). The behaviour of analysis of beer flavor. MBAA Technical Quarterly, 11, 83–86. volatile aromatic substances as beer ages. Monatsschrift Fur Liegeois, C., Lermusieau, G., & Collin, S. (2000). Measuring antiox- Brauwissenschaft, 52, 164–175. idant efficiency of wort, malt, and hops against the 2,20-azobis(2- Neven, H. (1997). Evolution of top fermented beer esters during bottle amidinopropane) dihydrochloride-induce oxidation of an aqueous conditioning and storage: chemical versus enzymatic hydrolysis. dispersion of linoleic acid. Journal of Agricultural and Food Dissertationes de Agricultura 333, Katholieke Universiteit Leuven, Chemistry, 48, 1129–1134. Belgium. 380 B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381

Neven, H., Delvaux, F., & Derdelinckx, G. (1997). Flavor Schwarz, P., Stanley, P., & Solberg, S. (2002). Activity of lipase during evolution of top fermented beers. MBAA Technical Quarterly, mashing. Journal of the American Society of Brewing Chemists, 60, 34, 115–118. 107–109. Noe¨l, S., & Collin, S. (1995). trans-2-Nonenal degradation products Shimizu, C., Araki, S., Kuroda, H., Takashio, M., & Shinotsuka, K. during mashing. Proceedings of the European Brewery Convention (2001a). Yeast cellular size and metabolism in relation to the flavor Congress, 483–490. and flavor stability of beer. Journal of the American Society of Noel, S., Metais, N., Bonte, S., Bodart, E., Peladan, F., Dupire, S., & Brewing Chemists, 59, 122–129. Collin, S. (1999). The use of Oxygen 18 in appraising the impact of Shimizu, C., Nakamura, Y., Miyai, K., Araki, S., Takashio, M., & oxidation process during beer storage. Journal of the Institute of Shinotsuka, K. (2001b). Factors affecting 5-hydroxymethyl furfural Brewing, 105, 269–274. formation and stale flavor formation in beer. Journal of the Ochiai, N., Sasamoto, K., Daishima, S., Heiden, A. C., & Hoffmann, American Society of Brewing Chemists, 59, 51–58. A. (2003). Determination of stale-flavor carbonyl compounds in Stenroos, L., Wang, P., Siebert, K., & Meilgaard, M. (1976). Origin beer by stir bar sorptive extraction with in situ derivatization and and formation of 2-nonenal in heated beer. MBAA Technical thermal desorption–gas chromatography–mass spectrometry. Jour- Quarterly, 13, 227–232. nal of Chromatography A, 986, 101–110. Stenroos, L. E. (1973). The effects of time, temperature, and air on Ohloff, G. (1978). Recent developments in the field of naturally- various finished beer components. Journal of the American Society occurring aroma components. Fortschritte der Chemie Organischer of Brewing Chemists, 50–56. Naturstoffe, 35, 431–527. Stephenson, W. H., & Bamforth, C. W. (2002). The impact of Ojala, M., Kotiaho, T., Siirila¨, J., & Sihvonen, M. (1994). Analysis of lightstruck and stale character in beers on their perceived quality: aldehydes and ketones from beer as O-(2,3,4,5,6-pentafluoroben- A consumer study. Journal of the Institute of Brewing, 108, zyl)hydroxylamine derivates. Talanta, 41, 1297–1309. 406–409. Ormrod, I. H. L., Lalor, E. F., & Sharpe, F. R. (1991). The release of Stephenson, W. H., Biawa, J. P., Miracle, R. E., & Bamforth, C. W. yeast proteolytic enzymes into beer. Journal of the Institute of (2003). Laboratory-scale studies of the impact of oxygen on Brewing, 97, 441–443. mashing. Journal of the Institute of Brewing, 109, 273–283. Palamand, S. R., & Hardwick, W. A. (1969). Studies on the relative Strating, J., & Van Eerde, P. (1973). The staling of beer. Journal of the flavor importance of some beer constituents. MBAA Technical Institute of Brewing, 79, 414–415. Quarterly, 6, 117–128. Thum, B., Miedaner, H., Narziss, L., & Black, W. (1995). Bildung von Peppard, T. L. (1978). Dimethyltrisulphide, its mechanism of forma- Alterundscarbonylen – mo¨geliche Mechanismen und Bedeutung tion in hop oil and effect on beer flavour. Journal of the Institute of bei der Bierlagerung. Proceedings of the European Brewery Brewing, 84, 337–340. Convention Congress, 491–498. Peppard, T. L., & Halsey, S. A. (1982). The occurrence of 2 Trachman, H., & Saletan, L. T. (1969). Simplified determination of geometrical-isomers of 2,4,5-trimethyl-1,3-dioxolane in beer. Jour- beer volatiles by gas chromatography. Proceedings of the American nal of the Institute of Brewing, 88, 309–312. Society of Brewing Chemists, 19–28. Qureshi, A. A., Burger, W. C., & Prentice, N. (1979). Polyphenols and Tressl, R., Bahri, D., & Kossa, M. (1980). In G. Charalambous (Ed.), pyrazines in beer during aging. Journal of the American Society of The analysis and control of less desirable flavors in food and Brewing Chemists, 37, 161–163. beverages (pp. 293–318). New York: Academic Press. Rangel-Aldao, R., Bravo, A., Galindo-Castro, I., Sanchez, B., Tressl, R., Bahri, D., & Silwar, R. (1979). Bildung von aldehyden Reverol, L., Scherer, E., Madrid, J., Ramirez, J. L., Herrera, J., durch lipidoxidation und deren bedeutung als ‘‘off-flavor-kompon- Penttila¨, M., Vehkoma¨ki, M., Vidgren, V., Virtanen, H., & Home, enten in bier. Proceedings of the European Brewery Convention S. (2001). A new technology for beer flavor stabilization: control of Congress, 27–41. key intermediates of the Maillard reaction. Proceedings of the Uchida, M., & Ono, M. (1996). Improvement for oxidative flavor European Brewery Convention Congress, 576–585. stability of beer – role of OH-radical in beer oxidation. Journal of Sanchez, B., Reverol, L., Galindo-Castro, I., Bravo, A., Ramirez, J. the American Society of Brewing Chemists, 54, 198–204. L., & Rangel-Aldao, R. (2001). BrewersÕs yeast oxidoreductase Uchida, M., & Ono, M. (1999). Determination of hydrogen peroxide with activity on Maillard reaction intermediates of beer. in beer and its role in beer oxidation. Journal of the American Proceedings of the European Brewery Convention Congress, Society of Brewing Chemists, 57, 145–150. 456–464. Uchida, M., Suga, S., & Ono, M. (1996). Improvement for oxidative Santos, J. R., Carneiro, J. R., Guido, L. F., Almeida, P. J., Rodrigues, flavor stability of beer – rapid prediction method for beer flavor J. A., & Barros, A. A. (2003). Determination of E-2-nonenal by stability by electron spin resonance spectroscopy. Journal of the high-performance liquid chromatography with UV detection – American Society of Brewing Chemists, 54, 205–211. assay for the evaluation of beer ageing. Journal of Chromatography Umano, K., Hagi, Y., Nakahara, K., Shyoji, A., & Shibamoto, T. A, 985, 395–402. (1995). Volatile chemicals formed in the headspace of a heated D- Saucier, C., Bourgeois, G., Vitry, C., Roux, D., & Glories, Y. (1997). glucose/L-cysteine Maillard model system. Journal of Agricultural Characterization of (+)-catechin-acetaldehyde polymers: a model and Food Chemistry, 43, 2212–2218. for colloidal state of wine polyphenols. Journal of Agricultural and Van Eerde, P., & Strating, J. (1981). trans-2-Nonenal. EBC-Flavour Food Chemistry, 45, 1045–1049. symposium. Monograph VII (pp. 117–121). Schieberle, P. (1991). Primary odorants of pale lager beer - Differences Van Iersel, M. F. M., Eppink, M. H. M., Van Berkel, W. J. H., to other beers and changes during storage. Zeitschrift Fur Rombouts, F. M., & Abee, T. (1997). Purification and character- Lebensmittel-Untersuchung Und-Forschung, 193, 558–565. ization of a novel NADP-dependent branched-chain alcohol Schieberle, P., & Komarek, D. (2002). Changes in key aroma dehydrogenase from Saccharomyces cerevisiae. Applied and Envi- compounds during natural beer aging. In K. R. Cadwallader & ronmental Microbiology, 63, 4079–4082. H. Weenen (Eds.), Freshness and Shelf Life of Foods (pp. 70–79). Van Nedervelde, L., Oudjama, Y., Desmedt, F., & Debourg, A. (1999). Washington DC: ACS. Further characterization and amplification of the aldoketoreduc- Schwarz, P., & Pyler, R. E. (1984). Lipoxygenase and hydroperoxide tase activity during fermentation: impact on aldehyde reduction. isomerase activity of malting barley. Journal of the American Proceedings of the European Brewery Convention Congress, Society of Brewing Chemists, 42, 47–53. 711–718. B. Vanderhaegen et al. / Food Chemistry 95 (2006) 357–381 381

Van Nedervelde, L., Verlinden, V., Philipp, D., & Debourg, A. (1997). of hydroperoxy fatty acids in malting and the protein rest stage of Purification and characterization of yeast 3-methyl butanal reduc- mashing. Journal of the Science of Food and Agriculture, 70, tases involved in the removal of wort carbonyls during fermenta- 341–346. tion. Proceedings of the European Brewery Convention Congress, Walters, M. T., Heasman, A. P., & Hughes, P. S. (1997a). Comparison 447–454. of (+)-catechin and ferulic acid as natural antioxidants and their van Strien, J. (1987). Direct measurement of the oxidation–reduction impact on beer flavor stability. 1. Forced-aging. Journal of the conditions of wort and beer. Journal of the American Society of American Society of Brewing Chemists, 55, 83–89. Brewing Chemists, 45, 77–79. Walters, M. T., Heasman, A. P., & Hughes, P. S. (1997b). Comparison van Waesberghe, J. W. M. (1997). New milling and mahing-in systems, of (+)-catechin and ferulic acid as natural antioxidants and their developed to avoid the initiation of lipid oxidation at the start of impact on beer flavor stability. 2. Extended storage trials. Journal the lagerbeer brewing process. Proceedings of the European Brewery of the American Society of Brewing Chemists, 55, 91–98. Convention Congress, 247–255. Wang, P. S., & Siebert, K. J. (1974). Determination of trans-2-nonenal Vanderhaegen, B. (2004). Flavor stability of beer – essential role of in beer. MBAA Technical Quarterly, 11, 110–117. furfuryl ethyl ether in staling. Dissertationes de agricultura 618, Wedzicha, B. L., & Kedward, C. (1995). Kinetics of the oligosaccha- Katholieke Universiteit Leuven, Belgium. ride-glycine-sulfite reaction - relationship to the browning of Vanderhaegen, B., Neven, H., Coghe, S., Verstrepen, K. J., Derde- oligosaccharide mixtures. Food Chemistry, 54, 397–402. linckx, G., & Verachtert, H. (2003a). Bioflavoring and beer Wheeler, R. E., Pragnell, M. J., & Pierce, J. S. (1971). The refermentation. Applied Microbiology and Biotechnology, 62, identification of factors affecting flavour stability in beer. Proceed- 140–150. ings of the European Brewery Convention Congress, 423–436. Vanderhaegen, B., Neven, H., Coghe, S., Verstrepen, K. J., Verachtert, Whitear, A. L. (1981). Factors affecting beer stability. EBC-Flavour H., & Derdelinckx, G. (2003b). Evolution of chemical and sensory symposium. Monograph VII (pp. 203–210). properties during aging of top-fermented beer. Journal of Agricul- Whitear, A. L., Carr, B. L., Crabb, D., & Jacques, D. (1979). The tural and Food Chemistry, 51, 6782–6790. challenge of flavour stability. Proceedings of the European Brewery Vanderhaegen, B., Neven, H., Daenen, L., Verstrepen, K. J., Convention Congress, 13–25. Verachtert, H., & Derdelinckx, G. (2004a). Furfuryl ethyl ether: Wijewickreme, A. N., & Kitts, D. D. (1998a). Metal chelating and Important aging flavor and a new marker for the storage antioxidant activity of model Maillard reaction products. In F. conditions of beer. Journal of Agricultural and Food Chemistry, Shahidi, C. T. Ho, & N. van Chuyen (Eds.), Proces induced 52, 1661–1668. chemical changes in food (pp. 245–254). New York: Plenum. Vanderhaegen, B., Neven, H., Verstrepen, K. J., Delvaux, F. R., Wijewickreme, A. N., & Kitts, D. D. (1998b). Oxidative reactions of Verachtert, H., & Derdelinckx, G. (2004b). Influence of the model Maillard reaction products and alpha-tocopherol in a fluor- brewing process on furfuryl ethyl ether formation during beer lipid mixture. Journal of Food Science, 63, 466–471. aging. Journal of Agricultural and Food Chemistry, 52, 6755–6764. Wijewickreme, A. N., Kitts, D. D., & Durance, T. D. (1997). Reaction Varmuza, K., Steiner, I., Glinsner, T., & Klein, H. (2002). Chemo- conditions influence the elementary composition and metal chelat- metric evaluation of concentration profiles from compounds ing affinity of nondialyzable model Maillard reaction products. relevant in beer ageing. European Food Research and Technology, Journal of Agricultural and Food Chemistry, 45, 4577–4583. 215, 235–239. Wijewickreme, A. N., Krejpcio, Z., & Kitts, D. D. (1999). Hydroxyl Vesely, P., Lusk, L., Basarova, G., Seabrooks, J., & Ryder, D. (2003). scavenging activity of glucose, fructose, and ribose-lysine model Analysis of aldehydes in beer using solid-phase microextraction Maillard products. Journal of Food Science, 64, 457–461. with on-fiber derivatization and gas chromatography/mass spec- Williams, R. S., & Gracey, D. E. F. (1982). The significance to flavor of trometry. Journal of Agricultural and Food Chemistry, 51, thioesters and polysulfides in canadian beers. Journal of the 6941–6944. American Society of Brewing Chemists, 40, 68–71. von Szilvinyi, A., & Pu¨spo¨k, J. (1969). Gaschromatographische Williams, R. S., & Wagner, H. P. (1978). The isolation and Untersuchungen an Leichtflu¨chtigen Bierkomponenten. Brauwis- identification of new staling related compounds form beer. Journal senschaft, 16, 204–208. of the American Society of Brewing Chemists, 36, 27–31. Wackerbauer, K., & Meyna, S. (2002). Free and triglyceride-bonded Williams, R. S., & Wagner, H. P. (1979). Contribution of hop bitter hydroxy fatty acids in barley and malt, I. (2002). Influence of substances to beer staling mechanisms. Journal of the American variety, provenience and harvest year. Monatsschrift Fur Brauwis- Society of Brewing Chemists, 37, 13–19. senschaft, 55, 52–57. Wohleb, R., Jennings, W. G., & Lewis, M. J. (1972). Some observa- Wackerbauer, K., Meyna, S., & Marre, S. (2003). Hydroxy fatty acids tions on (E)-nonenal in beer as determined by a headspace as indicators for ageing and the influence of oxygen in the sampling technique for gas–liquid chromatography. Proceedings brewhouse on the flavour stability of beer. Monatsschrift Fur of the American Society of Brewing Chemists, 1–3. Brauwissenschaft, 56, 174–178. Yang, G. S., & Schwarz, P. B. (1995). Activity of lipoxygenase Walker, M. D., Hughes, P. S., & Simpson, W. J. (1996). Use of isoenzymes during malting and mashing. Journal of the American chemiluminescence HPLC for measurement of positional isomers Society of Brewing Chemists, 53, 45–49. J. Agric. Food Chem. 1995, 43, 2917-2921 2917

Relative Reactivities of Amino Acids in the Formation of Pyridines, Pyrroles, and Oxazoles

Hui-Ing Hwang,+Thomas G. Hartman,$ and Chi-Tang Ho*it Department of Food Science and Center for Advanced Food Technology, Cook College, New Jersey Agricultural Experiment Station, Rutgers, The State University of New Jersey, New Brunswick, New Jersey 08903

The contributions of 15N-labeledglycine and tested amino acids (glutamine, glutamic acid, asparagine, aspartic acid, lysine, arginine, phenylalanine, and isoleucine) to pyridine, pyrrole, and oxazole formation were investigated. Ten pyridines, nine pyrroles, two oxazoles, three amines, and one benzonitrile were identified in the present study. The quantities of pyridines, pyrroles, and oxazoles in the reaction mixture of glycine and aspartic acid were the highest. Aspartic acid, lysine, and asparagine had the highest contribution in pyridine, pyrrole, and oxazole formation, respectively. In the presence of glycine, glutamic acid showed the least contribution, whereas asparagine had the highest contribution to the formation of all nitrogen-containing compounds among the tested amino acids. While lysine was able to increase the reactivity of glycine, arginine inhibited the capability of glycine to produce nitrogen-containing volatile compounds.

Keywords: Model Maillard reaction; amino acid reactivities; heterocyclic flavor compounds

INTRODUCTION glucose, ~-glycine-a-amine-'~N,and tested amino acid (L- glutamine, L-glutamic acid, L-asparagine, L-aspartic acid, It is well-known that the Maillard or nonenzymatic L-lysine, L-arginine, L-phenylalanine, or L-isoleucine) were browning reaction has a profound contribution to food mixed with 150 mL of deionized water and adjusted to pH 7 flavors. The effect of different amino acids on the by using hydrochloric acid or sodium hydroxide. After being formation of Maillard types of flavor compounds has freeze-dried, the solid mixture was placed in the upper level been widely studied (Koehler et al., 1969; Shigematsu of a desiccator; a Pyrex dish containing 20 mL of deionized et al., 1972). However, most of the studies have been water was placed in the lower level to adjust the moisture focused on the reaction between a single amino acid and content of the samples back to 12-14%. The samples were sugar (Piloty and Baltes, 1979; Fry and Stegink, 1982; further transferred into a reaction vessel and heated at 180 Ashoor and Zent, 1984). Flavor formation involving the "C for 1 h. interaction of more than one amino acid in the Maillard The heated samples (2 g of each) were packed in the center reaction has not been investigated. In an earlier paper of glass tubes, and silanized glass wool was placed at the two (Hwang et al., 19951, we reported the competition in ends of the tubes. One microliter of 1.001 mg/mL deuterated toluene was spiked into the tubes as the internal standard. formation between flavor different amino acids and 15N- The tubes were further sealed in a Scientific Instrument labeled glycine in a reaction system containing glucose Services (SIS) solid sample purge-and-trap apparatus (Ringoes, by examining the relative reactivities of eight different NJ), and the volatiles were purged with nitrogen at a flow rate amino acids (glutamine, glutamic acid, asparagine, of 40 mumin to silanized glass-lined stainless steel desorption aspartic acid, lysine, arginine, phenylalanine, and iso- tubes (4.0 mm i.d. x 10 cm length). The desorption tubes were leucine) on the formation of pyrazines. In the present also from SIS and consisted of a 3-cm bed volume of Tenax paper, we further report the relative reactivities of these TA adsorbent and a 3-cm bed volume of Carbotrap adsorbent. eight different amino acids on the formation of py- This volatile isolation was carried out at 80 "C for 1 h. ridines, pyrroles, and oxazoles. Volatile Analysis by Gas Chromatography-Mass Spec- trometry (GC-MS). The volatile analysis was conducted EXPERIMENTAL PROCEDURES according to the same method as described by Hwang et al. (1993). Linear retention indices for the volatiles were deter- Materials. L-Glycine, L-glutamine, L-glutamic acid, L- mined through the use of a Cj-CZb n-paraffin standard, asparagine, L-aspartic acid, L-lysine, L-arginine, L-phenylala- according to the method of Majlat et al. (1974). All mass nine, L-isoleucine, and wheat starch were purchased from spectra obtained were identified by utilizing an on-line com- Sigma Chemical Co. (St. Louis, MO). Glucose and deuterated puter library (NIST) or published literature. toluene (toluene-&),the internal standard, were obtained from Calculations for the Relative Contribution of 14N Aldrich Chemical Co. (Milwaukee, WI). Glycine-a-~mine-~~N Nitrogen and lSN Nitrogen to Flavor Formation. The was purchased from Isotec, Inc. (Miamisburg, OH) with a flavors monitored in this study were pyridines, pyrroles, stated purity of 99%. Tenax TA (2,6-diphenyl-p-phenylene oxazoles, amines, and benzonitrile. The molecular weights of oxide) adsorbent (60-80 mesh) was obtained from Alltech those flavor compounds will increase 1 mass unit when 15N Associates (Deerfield, IL). Carbotrap (activated graphitized atoms are incorporated, instead of 14N atoms, in heterocyclic carbon) adsorbent (20-40 mesh), Cj-C2j n-paraffin standard, rings which only contain one nitrogen atom in each compound. and silanized glass wool were purchased from Supelco Inc. (Bellefonte, PA). Thus, each flavor compound may have two different molecular Volatile Generation and Isolation. Twenty grams of weights, denoted W1 and Wz. W1 represents one 14N nitrogen atom in the ring, and has one 15N nitrogen atom in the wheat starch and an equal amount (2.66 pmol of each) of WZ ring. The simultaneous equations below are used to solve the contribution of each component (W1 and WZ) present in a Department of Food Science. mixture. This detailed explanation was previously reported t Center for Advanced Food Technology. by Hwang et al. (1993). 0021-8561/95/1443-2917$09.00/0 0 1995 American Chemical Society 2918 J. Agric. Food Chem., Vol. 43, No. 11, 1995 Hwang et al.

(1) ditional sources of ammonia. However, the yield of the glutamine reaction mixture was about the same as that of the glutamic acid reaction mixture, and the yield of the asparagine reaction mixture was even lower than (M - l),, M,, and (M + l), are the experimental relative that of aspartic acid reaction mixture. Therefore, these abundances of the ion peaks of the flavor compounds from the results might imply that the a-amino groups of amino reaction of nonlabeled glycine, tested amino acid, and glucose. acids prefer to condense with carbonyl compounds via Mexpand (M + l)expare the experimental abundances of the a one-step reaction to form pyridines rather than to ion peaks of the flavor compounds generated from the reaction degrade to ammonia and later incorporate into a pyri- of glycineJ5N, tested amino acid, and glucose. After the relative contributions of the two different com- dine ring. pounds (W1 and W2) were calculated, the percent of the Pyrroles identified in this study included four alkyl- contribution from a tested amino acid and labeled glycine could pyrroles, two acylpyrroles, two pyrrolealdehydes, and be determined by using eqs 3 and 4. As mentioned above, W1 one furfurylpyrrole. These pyrroles tend to contribute contains one 14N nitrogen atom in the ring; therefore, the baked cereal product notes or smoky notes (Ohloff and nitrogen of component W1 is from tested amino acids. On the Flament, 1978) and have been reported in various other hand, component Wz contains one I5N nitrogen atom from heated foods, especially coffee (Flament, 1991; Fors, I5N labeled glycine. 1983). It was also found that acylpyrroles have sweet, % contribution of tested amino acid = smoky, and slightly medicine-like odors (Shigematsu et [W,/(W, + W,)] x 100% (3) al., 1972,1977). Pyrrolealdehydes seem to have an odor analogous to xylene, cinnamaldehyde, or benzaldehyde % contribution of labeled glycine = (Kato, 19671, while l-furfurylpyrrole has a green hay- [W,/(W, + W,)l x 100% (4) like aroma (Walradt et al., 1970). Alkylpyrroles may have undesirable intense petroleum-like odors; however, RESULTS AND DISCUSSION they give a sweet, slightly burnt-like aroma on extreme dilution (Fors, 1983). There are two pathways to form In an earlier paper (Hwang et al., 1995), we reported pyrroles. One results from the interaction between an a total of 56 pyrazines in the reaction systems contain- amino acid and a 3-deoxyhexosone through the Strecker ing 15N labeled glycine and 8 other amino acids. In this degradation followed by dehydration and ring closure paper, we further report another 25 nitrogen-containing (Kato and Fujimaki, 1968). The other pathway is the reaction products (Table 1): 10 pyridines, 9 pyrroles, 2 reaction of furans with amines or amino acids. This oxazoles, 3 amines, and 1 benzonitrile. In comparison reaction requires a carbonyl function in position 2 of the with pyrazine formation, these nitrogen-containing furan derivative (Rizzi, 1974). heterocyclic compounds represent a relatively small The yields and relative contributions of the tested proportion. However, they possess some unique sensory properties which may contribute significantly to the amino acids to pyrrole formation are shown in Figure 2. The overall contributions of the tested amino acids flavor of processed foods. The 10 identified pyridines included 7 alkylpyridines, were below 35% in generating pyrroles when they 2 acylpyridines, and 1phenylpyridine. These pyridines competed with glycine. This result is due to glycine, have been reported in coffee, barley, roasted lamb, and which, having no side chain, is more flexible than other amino acids for involvement in the formation of pyrroles. meat (Buttery et al., 1977; Mottram, 1991; Suyama and Adachi, 1980). Some pyridines possess pleasant odors; For a comparison of each system for the contribution of amino acids to the formation of pyrroles, phenylalanine however, most pyridines have green, bitter, astringent, roasted, burnt, pungent vegetable, or phenolic proper- and isoleucine had the least contribution, whereas ties (Maga, 1981a; Pittet and Hruza, 1974). In general, asparagine, aspartic acid, and lysine had the highest contribution. The yield from the reaction mixture of alkylpyridines possess a less desirable odor (Fors, 19831, whereas acylpyridines have more pleasant aromas; for aspartic acid and glycine was the highest. example, 2-acetylpyridine has a cracker-type aroma If we further examine the relative contributions of the (Buttery et al., 1971). The formation of pyridine may tested amino acids to the formation of l-methylpyrrole involve the condensation of aldehydes, ketones, or a#- and 1-methylpyrrole-2-carboxaldehyde,more than 90% unsaturated carbonyl compounds with ammonia which of the nitrogen atoms were from glycine. Moreover, the is degraded from amino acids (Suyama and Adachi, yields of l-methylpyrrole and 1-methylpyrrole-2-car- 1980). Thus, various types of substituted pyridines can boxaldehyde were the two highest in most of the be produced from different combination of aldehydes, reaction mixtures (Table 1). As mentioned above, ketones, or carbonyl compounds (Table 1). The yields glycine can produce l-methyl-substituted pyrrole or and relative contributions of tested amino acids to pyrrolealdehyde by either reacting with 3-deoxyhex- pyridine formation are shown in Figure 1. Aspartic acid osone through the Strecker degradation or exchanging had the highest contribution and isoleucine the lowest the oxygen atom of the corresponding furan. The other contribution in the formation of pyridines. Aspartic acid amino acids might require an extra cleavage step to also generated the highest quantity of pyridines, while form l-methyl-substituted pyrrole or pyrrolealdehyde. arginine had the lowest quantity of pyridine in the These results indicate that glycine is superior to other presence of glycine. Although it has been suggested that amino acids in the production of pyrroles. the availability of ammonia is the determining factor The yields and relative contributions of the tested in generating pyridines (Baines and Moltkiewicz, 1984), amino acids to oxazole formation are shown in Figure this hypothesis is not supported in our study. If the 3. Unlike the other nitrogen-containing heterocyclic availability of ammonia was the determining factor to compounds, the quantities of oxazoles were very few in form pyridine, the yields of the glutamine and aspar- this study (Table 1). Oxazoles existing in cooked foods agine reaction mixtures would be higher than those of have also been reported, though in relatively small the glutamic acid and aspartic acid reaction mixtures amounts (Baltes et al., 1989). There was no oxazole which have the labile amide side chains as the ad- identified in the reaction mixture containing glycine Reactivities of Amino Acids in Formation of N-Containing Compounds J. Agric. Food Chem., Vol. 43, No. 11, 1995 2919

Table 1. Pyridines, Pyrroles, Oxazoles, and Other Nitrogen-Containing Compounds Identified in the Reaction of Glucose, Gly~ine-a-amine-l~N,and Tested Amino Acids yield (mg/g of glucose) compound Ctrla Glna Lysa Ama Phea Glua Aspa Ilea Arg pyridines pyridine - 6.3 - - 8.5 - - 4-methylpyridine - - - - 14.5 - - 2-methylpyridine - - - - 4.6 - - 2-ethylpyridine - - - - 8.3 2.2 - 2,5-dimethylpyridine - - - - 1.4 - - 3-ethyl-2,6-dimethylpyridine 11.6 5.1 4.5 6.5 - - - 3-butylpyridine 4.3 - - 9.6 1.7 - - 2-acetylpyridine 31.5 14.3 10.2 32.2 18.0 5.5 6.0 2-propionylpyridine 1.6 0.8 - - 6.5 - - 2,6-diphenylpyridine - - 11.8 - - - - pyrroles pyrrole - 9.3 - 6.1 10.0 5.0 4.8 1-methylpyrrole 30.2 26.7 58.9 59.0 131.5 26.8 11.8 2,5-dimethylpyrrole - 4.4 - 3.9 4.0 4.7 1.8 tetramethylpyrrole - - - 2.2 - - - 2-acetylpyrrole 17.0 9.2 - 15.8 8.7 - 16.8 1-methyl-2-acetylpyrrole 29.5 27.0 69.1 54.1 52.0 25.0 27.8 1-methylpyrrole-2-carboxaldehyde 6.4 - 10.0 - - 7.5 6.0 1-ethylpyrrole-2-carboxaldehyde 20.1 6.1 14.2 14.5 9.6 18.6 8.3 1-(2-furanylmethyl)pyrrole 8.2 6.9 6.1 9.2 10.2 - 6.4 oxazoles 4,5-dimethyloxazole 9.8 - - 13.9 14.0 - 7.7 trimethyloxazole - - - - 9.0 - - 2-acetyl-4,5-dimethyloxazole - - - - 10.1 - - other nitrogen-containing compounds benzonitrile - - 2.4 - - - - N-(2-methylbutylidene)-2-methylbutylamine - - - - - 69.7 - bis(2-methylbutybamine - - - - - 65.9 - tris(2-methy1butyl)amine - - - - - 83.0 - totals 170.2 116.1 187.2 227.0 322.4 313.9 97.4 Ctrl, glycine only; Gln, labeled glycine and glutamine; Lys, labeled glycine and lysine; Asn, labeled glycine and asparagine; Phe, labeled glycine and phenylalanine; Glu, labeled glycine and glutamic acid; Asp, labeled glycine and aspartic acid; Ile, labeled glycine and isoleucine; Arg, labeled glycine and arginine. Not observed.

250

200 F - 150 CI) P s loo -0 Q, F 50

IGIY E3Gly Tested Amino Acid Tested Amino Acid Figure 1. Total yields and relative contributions of pyridines Figure 2. Total yields and relative contributions of pyrroles generated from reaction systems containing tested amino acids generated from reaction systems containing tested amino acids and glycine. Ref, glycine only; Gln, labeled glycine and and glycine. Abbreviations and explanations are as in Figure glutamine; Glu, labeled glycine and glutamic acid; Asn, labeled 1. glycine and asparagine; Asp, labeled glycine and aspartic acid; Lys, labeled glycine and lysine; Arg, labeled glycine and from heat treatment (Flament, 19911, the exact mech- arginine; Phe, labeled glycine and phenylalanine; Ile, labeled anism of oxazole formation is not quite known. 4,5- glycine and isoleucine. "he numbers on the tops of the columns Dimethyloxazole was the most abundant oxazole iden- show the percent contributions of each tested amino acid. tified in this study. There are two possible pathways for the formation of 4,5-dimethyloxazole as shown in only or in the reaction mixture containing glycine and Scheme 1. The first is the reaction of glycine directly phenylalanine. The yield of oxazoles in the reaction with diacetyl which originates from glucose to form an ' mixture of glycine and aspartic acid was the largest. unstable Schiff base followed by decaboxylation, ring Asparagine had the highest contribution in the genera- closure, and aromatization to produce 4,5-dimethylox- tion of oxazoles. Although oxazoles are formed only azole. In this case, the source of nitrogen atoms would 2920 J, Agric. food Chem., Vol. 43, No. 11, 1995 Hwang et al.

3000 40 I

9 2250 8 -a cn p 20 1500 0) 5 v3 v 9 rrQ) Q) Y F 10 750

0 0

D Gly 0Gly Tested Amino Acid Tested Amino Acid Figure 4. Total yields and relative contributions of all Figure 3. Total yields and relative contributions of oxazoles nitrogen-containing compounds generated from reaction sys- generated from reaction systems containing tested amino acids tems containing tested amino acids and glycine. Abbreviations and glycine. Abbreviations and explanations are as in Figure and explanations are as in Figure 1. 1. Scheme 1. Proposed Pathways for the Formation of oxazole and trimethyloxazole observed in the present 4,B-Dimethyloxazole study have been identified in various processed foods such as coffee, cocoa, roasted green tea, meat, and baked CH, potato (Coleman et al., 1981; Flament, 1991; Maga, c= 0 1981b). Besides those nitrogen-containing heterocyclics, we found three amines from the isoleucine mixture and one benzonitrile which were specific products of the phenylalanine mixture (Table 1). Since these amines were only identified from the reaction mixture containing isoleucine, they might be generated from the interaction of 2-methylbutanal with the amino group of isoleucine or glycine. About 96% of the nitrogen atoms in benzonitrile were contributed from phenylalanine. This implies that benzonitrile is mainly the direct degradation product of phenylalanine. The yields and relative contributions of the tested amino acids to the formation of all nitrogen-containing compounds, including pyrazines, pyridines, pyrroles, CHI H* oxazoles, amines, and benzonitrile, are summarized in ‘5- NiCHz Figure 4. Glutamic acid was the lowest contributor, CHIA 0-6 while asparagine was the highest contributor, to flavor formation among the tested amino acids in the presence of labeled glycine. Figure 4 also shows that the yield of all nitrogen-containing flavors from labeled glycine in the lysine reaction mixture is the highest and more than that in the glycine reaction mixture alone. The yields of all nitrogen-containing flavors from labeled glycine in the arginine reaction mixture were the lowest and even less than that in the glycine reaction mixture alone. This suggests that the lysine acts as a synergist to increase the reactivity of other amino acids (glycine in this case); however, the arginine could depress specifically pyrroles, oxazoles, and pyridines at the expense of other products. be solely from glycine. The other pathway is through Another interesting phenomenon was that the yield Strecker degradation. Strecker degradation of amino of nitrogen-containing compounds in the asparagine acids and diacetyl yields 2-amino-3-butanone. 4,5- reaction mixture was not the highest one even though Dimethyloxazole is then produced by the condensation asparagine had the highest contribution. Theoretically, of 2-amino-3-butanone with formaldehyde. In this case, those amino acids consisting of side-chain nitrogen could the source of nitrogen atoms would be either glycine or result in a higher quantity of nitrogen-containing flavors the other tested amino acids. In our study, it seemed due to two nitrogen sources. The participation of side- that the former pathway was more favorable because chain nitrogen of glutamine and lysine in pyrazine more than 80% of the nitrogen atoms in the 4,5- formation has been proved by Hwang et al. (1993,1994). dimethyloxazole ring come from glycine. 4,5-Dimethyl- Therefore, this result seems to reveal that amino acids Reactivities of Amino Acids in Formation of N-Containing Compounds J. Agric. Food Chem., Vol. 43, No. 11, 1995 2921 possessing side-chain nitrogen except lysine are not only Hwang, H. I.; Hartman, T. G.; Ho, C.-T. Relative reactivities involved in flavor generation but also reduce the reac- of amino acids in pyrazine formation. J. Agric. Food Chem. tivity of glycine. Unlike the amide side chains of 1996,43, 179-184. asparagine and glutamine as well as the 8-guanidino Kato, H. Chemical studies on amino-carbonyl reaction: part 111. Formation of substituted pyrrole-2-aldehydes by reaction group of arginine, the +amino group of lysine can of aldoses with alkylamines. Agric. Biol. Chem. 1967, 31, participate in the Maillard reaction and catalyze the 1086-1090. sugar fragmentation (Hwang et al., 1994). This unique Kato, H.; Fujimaki, M. Formation of N-substituted pyrrole-2- property of lysine may be the reason that lysine is aldehydes in the Browning reaction between D-XylOSe and recognized as the most reactive amino acid in the amino compounds. J. Food Sci. 1968,33,445-449. Maillard reaction and why it produced the largest Koehler, P. E.; Mason, M. E.; Newell, J. A. Formation of quantities of volatile compounds in the present study. pyrazine compounds in sugar-amino acid model systems. J. Agric. Food Chem. 1969,17, 393-396. Maga, J. A. Pyridines in foods. J. Agric. Food Chem. 1981a, ACKNOWLEDGMENT 29, 895-898. We thank the Center for Advanced Food Technology Maga, J. A. The chemistry of oxazoles and oxazolines in foods. Mass Spectrometry Facility for providing instrumenta- CRC Crit. Rev. Food Sci. Nutr. 1981b, 11, 295-307. Majlat, P.; Erdos, Z.; Takacs, J. Calculation and application tion support. We also acknowledge Dr. Robert Rosen of retention indices in programmed temperature gas chro- and Mr. J. Lech for technical assistance. matography. J. Chromatogr. 1974, 91, 89-103. Mottram, D. S. Meat. In Volatile Compounds in Foods and LITERATURE CITED Beverages; Maarse, H., Ed.; Dekker: New York, 1991; pp 107-165. Ashoor, S. H.; Zent, J. B. Maillard browning of common amino Ohloff, G.; Flament, I. Heterocyclic constituents of meat aroma. acids and sugars. J. Food Sci. 1984,49,1206-1207. Heterocycles 1978, 11, 663-695. Baines, D. A.; Mlotkiewicz, J. A. The chemistry of meat flavor. Piloty, M.; Baltes, W. Investigations on the reaction of amino In Recent Advances in the Chemistry of Meat; Bailey, A. J., acids with a-dicarbonyl compounds. I. Reactivity of amino Ed.; Royal Society of Chemistry: London, 1984; pp 119- acids in the reaction with a-dicarbonyl compounds. 2. 164. Lebensm. Unters. Forsch. 1979, 168, 368-373. Baltes, W.; Kunert-Kirchhoff, J.; Reese, G. Model reactions on Pittet, A. 0.; Hruza, D. E. Comparative study of flavor generation of thermal aroma compounds. In Thermal G'en- properties of thiazole derivatives. J. Agric. Food Chem. eration of Aromas; Parliment, T. H., McGorrin, R. J., Ho, 1974,22, 264-269. C.-T., Eds.; ACS Symposium Series 409; American Chemical Rizzi, G. P. Formation of N-alkyl-2-acylpyrroles and aliphatic Society: Washington, DC, 1989; pp 143-155. aldimines in model nonenzymatic browning reactions. J. Buttery, R. G.; Seifert, R. M.; Guadagni, D. G.; Ling, L. C. Agric. Food Chem. 1974,22, 279-282. Characterization of additional volatile compounds of tomato. Shigematsu, H.; Kurata, T.; Kato, H.; Fujimaki, M. Volatile J. Agric. Food Chem. 1971,19, 524-529. compounds formed on roasting m-a-alanine with D-glUCOSe. Buttery, R. G.; Lin, L. C.; Teranishi, R.; Mon, T. R. Roast lamb Agric. Biol. Chem. 1972, 36, 1631-1637. fat: basic volatile components. J. Agric. Food Chem. 1977, Shigematsu, H.; Shibata, S.; Kurata, T.; Kato, H.; Fujimaki, 25, 1227-1229. M. Thermal degradation products of several Amadori com- Coleman, E. C.; Ho, C.-T.; Chang, S. S. Isolation and identi- pounds. Agric. Biol. Chem. 1977, 41, 2377-2385. fication of volatile compounds from baked potatoes. J. Agric. Suyama, K.; Adachi, S. Origin of alkyl-substituted pyridines Food Chem. 1981,29, 42-48. in food flavor: formation of the pyridines from the reaction Flament, I. Coffee, cocoa, and tea. In Volatile Compounds in of alkanals with amino acids. J. Agric. Food Chem. 1980, Foods and Beverages; Maarse, H., Ed.; Dekker: New York, 28, 546-549. 1991; pp 617-653. Walradt, J. P.; Lindsay, R. C.; Libbey, L. M. Popcorn flavor: Fors, S. Sensory properties of volatile Maillard reaction identification of volatile compounds. J. Agric. Food Chem. products and related compounds: a literature review. In The 1970,18,926-928. Maillard Reaction in Foods and Nutrition; Waller, G. R., Feather, M. S., Eds.; ACS Symposium Series 215; American Chemical Society: Washington, DC, 1983; pp 185-286. Received for review February 28, 1995. Revised manuscript Fry, L. K.; Stegink, L. D. Formation of Maillard reaction received June 23, 1995. Accepted June 30, 1995.@ This is products in parenteral alimentation solutions. J. Nutr. 1982, publication D-10535-2-94 of the New Jersey Agricultural 112, 1631-1637. Experiment Station supported by State Funds and the Center Hwang, H. I.; Hartman, T. G.; Rosen, R. T.; Ho, C.-T. for Advanced Food Technology (CAFT). CAFT is a New Jersey Formation of pyrazines from the Maillard reaction of glucose Commission on Science and Technology Center. This work and gl~tamine-amide-~~N.J. Agric. Food Chem. 1993, 41, was also supported in part by the US. Army Research Office. 2112-2115. Hwang, H. I.; Hartman, T. G.; Rosen, R. T.; Lech, J.; Ho, C.- JF950123G T. Formation of pyrazines from the Maillard reaction of glucose and ly~ine-a-amine-'~N.J. Agric. Food Chem. 1994, @ Abstract published in Advance ACS Abstracts, Oc- 42, 1000-1004. tober 15, 1995.