Eur Food Res Technol (2004) 218:589–597 DOI 10.1007/s00217-004-0901-x

ORIGINAL PAPER

Yi-Hong Chen · Su-Er Liou · Chu-Chin Chen

Two-step mass spectrometric approach for the identification of diketopiperazines in chicken essence Received: 8 December 2003 / Revised: 29 January 2004 / Published online: 11 March 2004
Springer-Verlag 2004

Abstract Diketopiperazines (DKPs) have been found in a variety of foods and beverages that impart a metallic bitter taste. The presence of DKPs in chicken essence (CE) was first postulated by a two-step monitoring technique, with mass precursor ion scanning from the immonium ion of a specific amino acid first, followed by mass matching with a DKP database. The postulated DKPs were further verified by tandem mass spectrometric (MS/ MS) analysis. It was found that this technique performed well, with about 20 DKPs tentatively identified. Proline was an important constituent of the DKPs in CE, together with other non-polar amino acids, such as glycine, alanine, valine, leucine/isoleucine and phenylalanine, which also showed great magnitude. Some proline-based DKPs in CE were quantified before and after thermal treatment at 130 C for 1 h. The concentrations of the DKPs were determined not only by the relative ease of formation and the stability of the DKPs, but also by past thermal processings. Keywords Diketopiperazine · Cyclic dipeptide · Chicken essence · Protein hydrolysate · Mass spectrometry

Introduction Diketopiperazines (DKPs) are cyclic dipeptides formed from the N-terminal amino acid residues of a linear peptide or protein. A reaction mechanism was proposed which suggested the attack of the N-terminal amino group of a peptide or protein as shown in Fig. 1 [1]. DKP formation by intermolecular reaction of amino acids was also possible, but only under exaggerated reaction conditions [2]. DKPs were identified in several beverages and foods, especially fermented and thermally Y.-H. Chen ()) · S.-E. Liou · C.-C. Chen Food Industry Research and Development Institute, 30099 Hsinchu, Taiwan, Republic of China e-mail: [email protected] Tel.: +886-3-522-3191 Fax: +886-3-521-4016

treated ones, which include aged sake [3], beer [4], cocoa [1], dried squid [5], hydrolyzed vegetable protein [6], roasted coffee [7, 8], roasted malt [9], and others. Degradation of the peptide analogue sweetener aspartame will produce 3-carboxymethyl-6-benzyl-2,5-diketopiperazine, which can be found in diet foods and beverages [10]. The bitter perception of many foods and beverages has been attributed in part to DKPs. A synergistic effect between DKPs and theobromine was observed in cocoa [1, 11]. A mechanism for bitter taste sensibility in DKPs was proposed by Ishibashi et al. [12]. DKPs have also provoked research attention owing to their biological activities. Cyclo(His-Pro) has been of special interest because it derives in part from thyrotropin-releasing hormone [13] and has been associated with several biological functions [14]. Some other DKPs also exhibit biological functions, but only few of them have been tested, rendering this relatively unexplored field of great promise for the future [14]. However, identification of DKPs in foods and beverages is still a low throughput manipulation because of the limited availability of standards. Researchers have synthesized their own DKPs of interest in order to achieve five–seven identifications by mass spectrometry (MS) for each instance [4, 7, 8]. In this paper, we try to identify DKPs in a sample by using mass spectrometry (MS) precursor ion scanning from the immonium ions of constituted amino aids following by tandem MS (MS/MS) verification of the molecular structures of the DKPs. It was supposed that a holistic viewing of the DKPs in a

Fig. 1 Diketopiperazines (DKP) formation mechanism

590

sample could be made, and that more DKPs would be identified.

Materials and methods Materials Chicken essence (CE) was purchased from a local supermarket. DKP standards of cyclo(Ala-Ser), cyclo(Phe-Pro), cyclo(Pro-Gly), cyclo(Pro-Leu), cyclo(Pro-Thr), cyclo(Pro-Tyr) and cyclo(Pro-Val) were products of Bachem (Bubendorf, Switzerland). Dichloromethane was Anala R grade of BDH (VWR, Poole, England). Sodium chloride was GR grade of Merck (Darmstadt, Germany). Formic acid was purchased from Riedel-de Haen (Seelze, Germany) and acetonitrile (HPLC grade) was purchased from Tedia (Fairfield, Ohio, USA). Methods Preparation of DKP sample from CE. CE (900 mL) was extracted with dichloromethane (340 mL). The extracts were combined and evaporated to dryness in a rotary evaporator. The dried extracts were dissolved in H2O/acetonitrile (1:1) and brought to 5 mL with the same solvent for HPLC-MS/MS analysis. For the DKP quantification experiment, another DKP sample was prepared from thermally reacted CE by heating it in a reaction vessel at 130 C for 1 h before solvent extraction of the DKPs. HPLC-MS/MS analysis of diketopiperazine. A HPLC-MS/MS system including a Waters 600E multi-solvent delivery pump, a Waters 2487 UV/VIS detector (Milford, MA, USA) and a Micromass Quattro LC MS/MS detector (Wythenshawe, UK) at positive electrospray ionization (ESI+) mode was used. The desolvation temperature was 300 C, and the extraction cone voltage was 20 V. The sample was resolved with a 4.650 Chromolith SpeedROD RP-18e column (Merck) before entering the MS/MS detector. The sample size was 10 mL, the flow rate was 0.2 mL/min, and both solvent A (H2O) and solvent B (acetonitrile) contained 0.1% (v/v) formic acid. The gradient was programmed as following: 0% B for 3 min, linear gradient from 20% to 100% B over 37 min, and 100% B for 10 min. Quantification of proline-based DKP in CE. Five proline-based DKPs, cyclo(Phe-Pro), cyclo(Pro-Gly), cyclo(Pro-Leu), cyclo(ProThr) and cyclo(Pro-Val), were quantified in CE with the HPLCMS/MS. Conditions were the same as specified above. The DKPs were monitored with the selected reaction monitoring (SRM) mode of MS/MS, where the transition from precursor ion to the largest product ion was used to represent each DKP, that is, 245>70 for cyclo(Phe-Pro), 155>70 for cyclo(Pro-Gly), 211>70 for cyclo(ProLeu), 199>153 for cyclo(Pro-Thr), and 197>70 for cyclo(Pro-Val). The standard curve of each DKP plotted by concentration (0– 140 mM) versus response peak area showed good linearity (R2>0.99). Quantification of DKPs in samples was made by interpolation.

Results and discussion Establishment of the framework for DKPs identification As is the practice in proteomics, a peptide sequence can be first postulated by its mass profile and isoelectric pH value (pI) followed by identification with MS/MS analysis. Several well-known bioinformatic databases containing the m/z values and pI data for peptides have been

established for molecular biologists and other researchers, to facilitate a high throughput identification of peptides in a mixture. However, all the databases frequently encountered in biochemical researches deal with linear peptides with or without modifications. A method of a two-step protocol similar to that used in proteomics was adopted for the identification of DKPs in a sample. The first step was to postulate if a specific DKP is present in the sample, by monitoring the parent ion mass scanned from the immonium ion of an amino acid. A table of DKPs’ mass values under the ESI+ mode of mass analysis was therefore established (Table 1). Although there were 400 possible combinations of any two amino acid residues selected from the frequently encountered 20 in foods, there were only 102 mass values. In most of the cases where DKPs share the same mass value, they are isomeric counterparts. Differentiation of DKP isomers with mass analysis requires a different approach. Therefore, both DKP isomers were represented by a single DKP in this study, as has also been done by others [7, 8]. An immonium ion is an internal fragment of a peptide with just a single side chain formed by the combination of a-type and y-type cleavage during MS/MS analysis [15, 16, 17, 18]. The immonium ion has been served as an indicator for the presence or absence of a specific amino acid residue in the peptide sequences of an unseperated peptide mixture with precursor ion scanning [19]. Data from Gautschi et al. [4] and Ginz and Engelhardt [7, 8] also showed that m/z 70, the immonium ion mass of proline, is the major product ion mass of the proline-based DKPs they had tested. A list of immonium ion masses of the 20 common amino acids can be obtained elsewhere [14, 16]. To test the postulation process, a mixture of cyclo(AlaSer), cyclo(Phe-Pro), cyclo(Pro-Tyr) and cyclo(Pro-Val) (0.1 mole each in 50/50 H2O/acetonitrile) was applied to the mass system for precursor ion scanning of the 20 immonium ion masses. Results showed only channels of m/z 44 (Ala), 60 (Ser), 70 (Pro), 72 (Val), 120 (Phe) and 136 (Tyr) gave peak responses and the precursor ions scanned are shown in Fig. 2. There was only an m/z 159 ion obtained from m/z 44 (Ala) scanning (Fig. 2a), from which it could be postulated that cyclo(Ala-Ser) or cyclo(Ser-Ala) was present in the mixture because serine was the only possible combination in alanine-based DKPs to give m/z 159 (Table 1). This postulation could be partly verified by looking at the m/z 60 channel of Ser (Fig. 2b) where the retention time (RT) and mass value of the scanned precursor ion corresponded to those in the m/z 44 channel. Although it is quite certain that cyclo(Ala-Ser) or cyclo(Ser-Ala) was present in the mixture, a positive identification would not be made until product ion scanning for m/z 159 was performed. The chromatogram of the m/z 70 channel (Fig. 2c) showed three peaks, and the corresponding mass spectrum of each peak was also shown. The peak at RT=4.81 min consisted of the m/z 116 ion, which did not match any mass value of the prolinebased DKP (Table 1). This m/z 116 ion was therefore

221 235 251 261 263 265 267 277 277 278 279 292 292 293 295 301 311 320 327 350 214 228 244 254 256 258 260 270 270 271 272 285 285 286 288 294 304 313 320 343 205 219 235 245 247 249 251 261 261 262 263 276 276 277 279 285 295 304 311 334 195 209 225 235 237 239 241 251 251 252 253 266 266 267 269 275 285 294 301 324 189 203 219 229 231 233 235 245 245 246 247 260 260 261 263 269 279 288 295 318 187 201 217 227 229 231 233 243 243 244 245 258 258 259 261 267 277 286 293 316 186 200 216 226 228 230 232 242 242 243 244 257 257 258 260 266 276 285 292 315 172 186 202 212 214 216 218 228 228 229 230 243 243 244 246 252 262 271 278 301 171 185 201 211 213 215 217 227 227 228 229 242 242 243 245 251 261 270 277 300 171 185 201 211 213 215 217 227 227 228 229 242 242 243 245 251 261 270 277 300 161 175 191 201 203 205 207 217 217 218 219 232 232 233 235 241 251 260 267 290 159 173 189 199 201 203 205 215 215 216 217 230 230 231 233 239 249 258 265 288 157 171 187 197 199 201 203 213 213 214 215 228 228 229 231 237 247 256 263 286 155 169 185 195 197 199 201 211 211 212 213 226 226 227 229 235 245 254 261 284 145 159 175 185 187 189 191 201 201 202 203 216 216 217 219 225 235 244 251 274 129 143 159 169 171 173 175 185 185 186 187 200 200 201 203 209 219 228 235 258 115 129 145 155 157 159 161 171 171 172 173 186 186 187 189 195 205 214 221 244 Gly Ala Ser Pro Val Thr Cys Ile Leu Asn Asp Gln Lys Glu Met His Phe Arg Tyr Trp

Tyr Arg Phe His Met Glu Lys Asn Leu Ile Cys Thr Val Pro Ser Ala Gly

Table 1 Calculated mass values of DKPs under positive electrospray ionization mode of mass spectrometry

244 258 274 284 286 288 290 300 300 301 302 315 315 316 318 324 334 343 350 373 186 200 216 226 228 230 232 242 242 243 244 257 257 258 260 266 276 285 292 315 173 187 203 213 215 217 219 229 229 230 231 244 244 245 247 253 263 272 279 302

DKP identification in CE

Trp Gln

thought to be a contaminant rather than a proline-based DKP. This ion was not included in the product ion scanning for final identification. The peak at RT= 26.70 min showed three precursor ions of m/z 70, among which m/z 197 matched the mass value of cyclo(Pro-Val) from Table 1, m/z 261 matched that of cyclo(Pro-Tyr), and m/z 302 was later found to be cyclo(Pro-Tyr) with acetonitrile adducted. Similar results were observed in the peak at RT=29.67 min, where m/z 245 matched cyclo(Pro-Phe) [or cyclo(Phe-Pro)] and m/z 286 for its acetonitrile adduct. Postulation of the presence of cyclo(Pro-Val), cyclo(Phe-Pro) and cyclo(Pro-Tyr) in the mixture could be supported by the precursor ion scanning results from channels of m/z 72 (Val), 120 (Phe) and 136 (Tyr), respectively (Fig. 2d, e, f). Although it is quite possible that cyclo(Ala-Ser), cyclo(Pro-Val), cyclo(Phe-Pro) and cyclo(Pro-Tyr) existed in the mixture, their identities were not verified until product ion scanning of m/z 159, 197, 245 and 261 were performed and their low mass ions interpreted from the DKPs’ precursor ions (Fig. 3). Generally, the [M+H]+ and immonium ions of the constituent amino acids were always among the major product ions. Other low mass ions could also be found at the mass values corresponding to [M-nCO-nNH3+H]+, where n=0, 1 or 2.

Asp

591

The protocols described hereto were applied to the identification of DKPs in a CE. CE is an oil-removed, clear solution made of steam extract or protease hydrolysate of chicken, and has always been considered a good means of providing amino acids and peptides to the human consumer [20]. Identification of several proline-based DKPs was exemplified here for viewing the holistic DKP spectrum of CE. Precursor ion scanning on the immonium ion of proline (m/z 70) showed five major signals at m/z 169, 195, 197, 211 and 245 that were postulated to be cyclo(Pro-Ala), cyclo(Pro-Pro), cyclo(Pro-Val), cyclo (Pro-Leu/Ile) and cyclo(Pro-Phe), respectively (Fig. 4a). It was also found that minor signals at m/z 155 [cyclo (Pro-Gly)], 185 [cyclo(Pro-Ser)], 199 [cyclo((Pro-Thr)], 212 [cyclo(Pro-Asn)], 227 [cyclo(Pro-Glu)], 229 [cyclo (Pro-Met)], 261 [cyclo(Pro-Tyr)] and 284 [cyclo(ProTrp)] matched the mass values of the proline-based DKPs in Table 1. However, signals were not observed at m/z 201 [cyclo(Pro-Cys)], 213 [cyclo(Pro-Asp)], 226 [cyclo(ProGln or Pro-Lys)], 235 [cyclo(Pro-His)] or 254 [cyclo(ProArg)] even when the spectrum was scrutinized at high attenuation. Mass values of the five major proline-based DKPs, m/z 169, 195, 197, 211 and 245, were found to be cross referenced from precursor ions scanned for m/z 44 (Ala) (Fig. 4b), m/z 70 (Pro) (Fig. 4a), m/z 72 (Val) (Fig. 4c), m/z 86 (Leu/Ile) (Fig. 4d) and m/z 120 (Phe) (Fig. 4e), respectively. It was interesting to note that each of the five major proline-based DKPs was also among the major

592

Fig. 2a–f Precursor ion spectra of the DKP mixture (described in the Results and discussion section) scanned for a specific immonium ions. a m/z 44 (Ala). b m/z 60 (Ser). c m/z 70 (Pro). d m/z 72 (Val). e m/z 120 (Phe). f m/z 136 (Tyr)

ones in its corresponding spectrum. RTs for m/z 169, 195, 197, 211 or 245 in the precursor ion scanning chromatograph of m/z 70 were also coincident to that in the chromatographs of m/z 44, 70, 72, 86 and 120, respectively (data not shown). Product ions were then scanned for m/z 169, 195, 197, 211 and 245. The low mass ion spectra clearly verified that the m/z 169 is cyclo(Pro-Ala), m/z 195 is cyclo(Pro-Pro), m/z 197 is cyclo(Pro-Val), m/z 211 is cyclo(Pro-Leu/Ile), and m/z 245 is cyclo(Pro-Phe) (Fig. 5).

Beside these five major proline-based DKPs, the other eight minor proline-based DKPs of m/z 155, 185, 199, 212, 227, 229, 261 and 284 (Fig. 4a) were also analyzed for their product ions to verify their identities if ions of the same mass values were also observed at the same RTs in corresponding spectra of precursor ion scanning. However, verification of those DKPs with low responses by product ion scanning may be rendered more difficult by background interference. The minor proline-based DKPs with m/z of 227 and 229, for example, were

593

Fig. 2d–f

difficult to verify (data not shown). Postulations of m/z 155, 185, 199, 212, 261 and 284 as cyclo(Pro-Gly), cyclo (Pro-Ser), cyclo(Pro-Thr), cyclo(Pro-Asn), cyclo(Pro-Tyr) and cyclo(Pro-Trp), respectively, were later confirmed with their product ion profiles (Table 2). The same procedures were performed on all other immonium ions of the common amino acids. The DKPs tentatively identified in CE are listed in Table 2, where proline-based DKPs contributed more than half of them. Several immonium ions, especially those with charged side chains, did not give signals for their precursor ions (data not shown). DKPs based on asparagine or amino

acids with polar, uncharged side chains, such as serine, threonine and tyrosine, were found only when these amino acids coupled to proline. The others consisted of amino acids with nonpolar aliphatic or aromatic side chains, such as glycine, alanine, valine, leucine/isoleucine and phenylalanine. It is interesting to find that non-polar amino acids, especially proline, were important in DKPs identified in this study as well as some others [4, 7, 8]. Gautschi et al. [4] suggested that the results were presumably determined by the relative ease of formation and stability of the DKPs in question. We believe this may result from the special molecular structure of proline that

594 Fig. 3a–d Product ion spectra of a m/z 159, cyclo(Ala-Ser); b m/z 197, cyclo(Pro-Val); c m/z 245, cyclo(Phe-Pro); and d m/z 261, cyclo(Pro-Tyr), in the peptide mixture specified in the Results and discussion section

served as a parking point during peptide bond breakage by heat or proteases when the CE was manufactured. The Nterminal proline thus exposed was then cycled with the second amino acid residue to form a proline-based DKP. Quantification of some proline-based DKPs in CE Five proline-based DKPs of CE were further quantified in CE and thermally reacted CE (TRCE). By using the SRM

mode of MS/MS, only the DKP of interest or the DKP with a few additional peaks were present in each MS/MS chromatogram (data not shown). Therefore, the reliability of quantification of this method was thought to be superior to other techniques [21, 22]. Results showed that all the five proline-based DKPs were in the parts per million level (Table 3). DKPs formed from proline and non-polar amino acids (Phe, Leu or Val) showed higher amounts than those formed from proline and polar amino acids (Gly or Thr) in CE. Although heating was expected

595 Fig 4a–e Precursor ion spectra of CE for the immonium ions of a m/z 70 (Pro), b m/z 44 (Ala), c m/z 72 (Val), d m/z 86 (Leu/Ile), and e m/z 120 (Phe). Possible DKPs for major signals were specified in the spectra

596 Fig 5a–e Product ion spectra of a m/z 169, cyclo(Pro-Ala); b m/ z 195, cyclo(Pro-Pro); c m/z 197 cyclo(Pro-Val);. d m/z 211 cyclo(Pro-Leu/Ile); and e m/z 245 cyclo(Pro-Phe), in chicken essence

597 Table 2 DKPs tentatively identified in chicken essence DKP

RT (min)

m/z

Product ion (%)

Cyclo(Ala-Ala) Cyclo(Pro-Gly) Cyclo(Gly-Val) Cyclo(Pro-Ala) Cyclo(Ala-Val) Cyclo(Pro-Ser) Cyclo(Ala-Leu/Ile) Cyclo(Pro-Pro) Cyclo(Pro-Val) Cyclo(Pro-Thr) Cyclo(Pro-Leu/Ile) Cyclo(Pro-Asn) Cyclo(Leu/Ile-Val) Cyclo(Leu/Ile-Leu/Ile) Cyclo(Pro-Phe) Cyclo(Phe-Val) Cyclo(Pro-Tyr) Cyclo(Leu/Ile-Phe) Cyclo(Pro-Trp) Cyclo(Phe-Phe)

8.15 9.65 11.72 16.03 21.19 9.38 28.59 25.90 27.38 11.85 29.16 29.03 30.52 30.85 29.20 30.12 25.33 31.20 29.73 32.37

143 155 157 169 171 185 185 195 197 199 211 212 213 227 245 247 261 261 284 295

143 155 157 169 171 185 185 195 197 199 211 212 213 227 245 247 261 261 284 295

(17.4), 125 (3.5), 115 (3.8), 98 (41.5), 72 (36.2), 70 (16.2), 55 (3.9), 44 (100) (100), 138 (0.6), 127 (30.2), 85 (1.8), 82 (15.7), 70 (46.2), 58 (1.0), 30 (0.1) (47.7), 129 (27.5), 112 (42.0), 100 (43.4), 84 (24.8), 83 (45.9), 72 (100), 55 (31.4) (85.0), 141 (8.0), 113 (2.4), 98 (11.7), 96 (5.5), 72 (9.5), 70 (100), 44 (5.4) (19.6), 143 (11.8), 126 (23.8), 100 (15.3), 98 (39.2), 86 (7.2), 72 (100), 44 (15.3) (93.3), 167 (12.1), 157 (39.4), 140 (33.4), 98 (21.2), 86 (84.2), 70 (100), 60 (27.9) (22.2), 157 (10.0), 140 (33.9), 114 (22.4), 112 (31.8), 86 (100), 72 (5.4), 44 (21.7) (100), 167 (0.1), 151 (0.6), 139 (0.0), 125 (0.0), 98 (19.2), 82 (0.1), 70 (56.1) (100), 169 (23.6), 141 (8.0), 124 (10.8), 100 (6.0), 98 (14.3), 72 (48.1), 70 (34.9) (44.6), 181 (37.6), 171 (4.3), 153 (100), 125 (40.7), 97 (3.6), 74 (4.2), 70 (54.2) (100), 183 (15.5), 155 (4.5), 138 (6.7), 114 (6.0), 98 (9.2), 86 (38.0), 70 (38.7) (100), 184 (15.0), 156 (3.6), 139 (5.3), 115 (3.4), 98 (4.6), 87 (16.0), 70 (24.0) (53.5), 185 (27.1), 168 (37.4), 140 (54.6), 114 (17.3), 100 (9.3), 86 (70.9), 72 (100) (27.6), 199 (18.1), 182 (24.5), 154 (23.2), 126 (1.0), 114 (12.8), 98 (2.3), 86 (100) (66.1), 217 (12.5), 197 (9.3), 172 (2.7), 154 (8.5), 120 (96.7), 98 (11.5), 70 (100) (22.2), 219 (19.5), 202 (9.1), 174 (29.2), 156 (1.8), 120 (100), 100 (2.8), 72 (24.5) (24.2), 233 (4.0), 197 (100), 169 (3.3), 155 (5.0), 136 (31.0), 107 (5.7), 70 (27.7) (26.9), 233 (20.2), 216(16.4), 188 (22.4), 170 (1.3), 120 (100), 114 (4.5), 86 (54.2) (15.5), 239 (1.1), 170 (11.0), 159 (1.3), 132 (9.3), 130 (100), 115 (3.6), 70 (2.5) (14.9), 278 (0.3), 267 (13.2), 250 (2.0), 222 (6.6), 204 (1.1), 148 (0.8), 120 (100)

Table 3 Proline-based DKPs in chicken essence (CE) and thermally reacted chicken essence (TRCE)

3.

DKP

4.

Cyclo(Phe-Pro) Cyclo(Pro-Gly) Cyclo(Pro-Leu) Cyclo(Pro-Thr) Cyclo(Pro-Val) a b

CE

TRCE

ppm

RSDa (%)

ppm

RSDa (%)

1.42b 0.78 2.66 0.06 2.75

9.17 4.35 8.52 6.45 8.42

1.38b 1.66 3.55 0.16 3.39

5.49 9.2 10.8 14.15 8.37

n=3 Statistically not significant at P<0.05

5. 6.

7. 8. 9.

to induce DKPs formation, not all proline-based DKPs were increased in TRCE. The amount of cyclo(Phe-Pro) in CE was not statistically different from that in TRCE (P>0.05), while the other four proline-based DKPs were significantly increased (1.23–2.67 times) after thermal treatment. It is suggested that, besides the relative ease of formation and stability of a DKP, the past thermal processing of the CE was also a major factor in determining the DKP formation. Acknowledgements Financial support for this study from the Ministry of Economic Affairs (92-EC-17-A-18-R7-0332), Taiwan, the Republic of China, is highly appreciated.

References 1. Pickenhagen W, Dietrich P, Keil B, Polonsky J, Nouaille F, Lederer E (1975) Helv Chim Acta 58:1078–1086 2. Rizzi GP (1989) Heat-induced flavor formation from peptides. In: Parliament TH, McGorrin RJ, Ho CT (ed) Thermal

10.

11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

generation of aromas. ACS Symposium Series 409, American Chemical Society, Washington, DC, USA, pp 172–181 Takahashi K, Tadenuma M, Kitamoto K, Sato S (1974) Agric Biol Chem 38:927–932 Gautschi M, Schmid JP, Peppard TL, Ryan TP, Tuorto RM, Yang X (1997) J Agric Food Chem 45:3183–3189 Kawai T, Ishida Y, Kakiuchi H, Ikeda N, Higashida T, Nakamura S (1991) J Agric Food Chem 39:770–777 Eriksen SA, Fagerson IS (1980) Nonvolatile nitrogen compounds in hydrolyzed vegetable protein. In: Inglett GE, Munck L (eds) Cereals for food and beverages: recent progress in cereal chemistry and technology. Academic, New York, pp 395–408 Ginz M, Engelhardt UH (2000) J Agric Food Chem 48:3528– 3532 Ginz M, Engelhardt UH (2001) Eur Food Res Technol 213:8-11 Sakamura S, Furukawa K, Kasai T (1978) Agric Biol Chem 42:607–612 Tateo F, Triangeli L, Panna E, Berte F, Verderio E (1988) Stability and reactivity of aspartame in cola-type drinks. In: Charalambous G (ed) Frontiers of flavor. Elsevier, Amsterdam, pp 217–231 Bonvehi JS, Coll FV (2000) Eur Food Res Technol 210:189– 195 Ishibashi N, Kouge K, Shinoda I, Kanehisa H, Okai H (1988) Agric Biol Chem 52:819–827 Prasad C, Peterkofsky A (1976) J Biol Chem 251:3229–3234 Prasad C (1995) Peptides 16:151–164 Falick AM, Hines WM, Medzihradszky KF, Baldwin MA, Gibson BW (1993) J Am Soc Mass Spectrom 4:882–893 Johnson RS, Martin SA, Biemann K, Stults JT, Watson JT (1987) Anal Chem 59:2621–2625 Papayannopoulos IA (1995) Mass Spectrom Rev 14:49–73 Roepstorff P, Fohlman J (1984) Biomed Mass Spectrom 11:601 Wilm M, Neubauer G, Mann M (1996) Anal Chem 68:527–533 Wu HC, Shiau CY (2002) J Food Drug Anal 10:170–177 Soga T, Heiger DN (2000) Anal Chem 72:1236–1241 Chen Y-H, Shih L-L, Liou S-E, Chen C-C (2003) Food Sci Technol Res 9:276–282