Kinetic modelling of Amadori N-(1-deoxy-D-fructos-1-yl)-glycine degradation pathways. Part I - Reaction mechanism

Article abstract of DOI:10.1016/S0008-6215(03)00173-3

The fate of the Amadori compound N-(1-deoxy-D-fructos-1-yl)-glycine (DFG) was studied in aqueous model systems as a function of pH and temperature. The samples were heated at 100 and 120°C with initial reaction pH of 5.5 and 6.8. Special attention was paid to the formation of the free amino acid, glycine; parent sugars, glucose and mannose; organic acids, formic and acetic acid and α-dicarbonyls, 1- and 3-deoxyosone together with methylglyoxal. For the studied conditions decreasing the initial reaction pH with 1.3 units or increasing the temperature with 20°C has the same effect on the DFG degradation as well as on glycine formation. An increase in pH seems to favour the formation of 1-deoxyosone. The lower amount found comparatively to 3-deoxyosone, in all studied systems, seems to be related with the higher reactivity of 1-deoxyosone. Independently of the taken pathway, enolization or retro-aldolization, DFG degradation is accompanied by amino acid release. Together with glycine, acetic acid was the main end product formed. Values of 83 and 55 mol% were obtained, respectively. The rate of parent sugars formation increased with pH, but the type of sugar formed also changed with pH. Mannose was preferably formed at pH 5.5 whereas at pH 6.8 the opposite was observed, that is, glucose was formed in higher amounts than mannose. Also, independently of the temperature, at higher pH fructose was also detected. pH, more than temperature, had an influence on the reaction products formed. The initial steps for a complete multiresponse kinetic analysis have been discussed. Based on the established reaction network a kinetic model will be proposed and evaluated by multiresponse kinetic modelling in a subsequent paper.

Full text of DOI:10.1016/S0008-6215(03)00173-3

Carbohydrate Research 338 (2003) 1651Á1663
/
www.elsevier.com/locate/carres
Kinetic modelling of Amadori N-(1-deoxy-
D
-fructos-1-yl)-glycine
degradation pathways. Part I*Reaction mechanism
/
Sara I.F.S. Martins,a Antonius T.M. Marcelis,b Martinus A.J.S. van Boekela,*
a Department of Agrotechnology and Food Sciences, Product Design and Quality Management Group, Wageningen University, P.O. Box 8129, 6700 EV
Wageningen, The Netherlands
b Department of Agrotechnology and Food Sciences, Organic Chemistry Group, Wageningen University, P.O. Box 8129, 6700 EV Wageningen, The
Netherlands
Received 11 December 2002; accepted 11 April 2003
Abstract
The fate of the Amadori compound N-(1-deoxy-D-fructos-1-yl)-glycine (DFG) was studied in aqueous model systems as a
function of pH and temperature. The samples were heated at 100 and 120 8C with initial reaction pH of 5.5 and 6.8. Special attention
was paid to the formation of the free amino acid, glycine; parent sugars, glucose and mannose; organic acids, formic and acetic acid
and a-dicarbonyls, 1- and 3-deoxyosone together with methylglyoxal. For the studied conditions decreasing the initial reaction pH
with 1.3 units or increasing the temperature with 20 8C has the same effect on the DFG degradation as well as on glycine formation.
An increase in pH seems to favour the formation of 1-deoxyosone. The lower amount found comparatively to 3-deoxyosone, in all
studied systems, seems to be related with the higher reactivity of 1-deoxyosone. Independently of the taken pathway, enolization or
retro-aldolization, DFG degradation is accompanied by amino acid release. Together with glycine, acetic acid was the main end
product formed. Values of 83 and 55 mol% were obtained, respectively. The rate of parent sugars formation increased with pH, but
the type of sugar formed also changed with pH. Mannose was preferably formed at pH 5.5 whereas at pH 6.8 the opposite was
observed, that is, glucose was formed in higher amounts than mannose. Also, independently of the temperature, at higher pH
fructose was also detected. pH, more than temperature, had an influence on the reaction products formed. The initial steps for a
complete multiresponse kinetic analysis have been discussed. Based on the established reaction network a kinetic model will be
proposed and evaluated by multiresponse kinetic modelling in a subsequent paper.
# 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Amadori compound; N-(1-deoxy-D-fructos-1-yl)-glycine; Maillard reaction; Multiresponse kinetic analysis
1. Introduction
ketoseamine, also known as the Amadori rearrangement
product (ARP). The set of reactions that occurs there-
after is of great importance in the processing of foods
for the production of aroma, taste and colour. Further-
more, evidence strongly suggests that this intricate
reaction cascade is involved in the pathology of diabetes
and ageing.2
Due to the complexity of products that are formed
from the degradation of the ARP, recent studies used
the Amadori compound as the initial reactant.3Á6 These
studies showed that the 1,2- and 2,3-enolizations of
Amadori compounds under acid/base catalysis condi-
tions initiate b-elimination reactions, which eventually
lead to the formation of reactive intermediates (Scheme
Since Hodge1 presented the first coherent Maillard
reaction scheme in 1953, the Amadori compound N-
substituted 1-amino-1-deoxy-ketose is believed to be the
key intermediate in the early stages of the reaction. The
accepted mechanism for the formation of Amadori
compounds involves the initial reaction of a reducing
sugar with an amino group to give the corresponding
glycosylamine, that rearranges to the corresponding
* Corresponding author. Tel.: ꢀ
317-483669.
E-mail address: tiny.vanboekel@wur.nl (M.A.J.S. van
Boekel).
/
31-317-482520; fax: ꢀ31-
/
1). The degradation of N-(1-deoxy-D-fructos-1-yl)-gly-
cine (DFG) (1) by 1,2-enolization (pathway A) and 2,3-
0008-6215/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0008-6215(03)00173-3

1652
S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
Scheme 1. Amadori compound N-(1-deoxy-D-fructos-1-yl)glycine degradation pathways: enolization and retro-aldolization.
enolization (pathway B) leads to the formation of 3-
deoxy-2-hexosulose (3) and 1-deoxy-2,3-hexodiulose (5),
respectively. The formation of these intermediates is
accompanied by amino acid release. In parallel, other a-
dicarbonyls can also be formed from ARP enolization.
A recent study6 presented a detailed scheme of Amadori
compound degradation pathways through enolization,
such as the formation of glucosones by transition-metal

S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
1653
catalyzed oxidation of 1,2-enaminol (2) and 1-amino-
1,4-dideoxy-2,3-diulose by elimination of the C-4 OH
group of the 2,3-enaminol (4).
Another possible mechanism of degradation of Ama-
dori compounds involves a retro-aldol reaction, as
shown in pathway C. Previous studies4,7 have indicated
on reaction conditions; (iii) differentiate between pri-
mary and secondary reaction routes; (iv) propose a
kinetic model based on the established reaction network;
(v) test the hypothesized mechanism; (vi) estimate the
rate constants. The present paper (Part I) deals with the
first three steps. Special attention was paid to the
formation of the free amino acid, glycine; parent sugars,
Glc and Man; organic acids, formic and acetic acid and
a-dicarbonyls, 1- and 3-deoxyosone together with
methylglyoxal. A subsequent paper, Part II,11 deals
with the remaining steps with a focus on the kinetics
of the hypothesized mechanism.
that 1 generates 1-glycine-1-deoxy-D-glyceraldehyde (6)
and glyceraldehyde (7) through a retro-aldol cleavage at
C-3 to C-4. Compound 7 can react with free glycine and
produce more of compound 6 that subsequently under-
goes a b-elimination to form methylglyoxal (8) and
release glycine. Moreover, 8 can also be produced from
7 through the catalytic action of the amino acid.
In spite of all the work done in Maillard reaction, the
reversibility of ARP is still a controversial issue.
Theoretically, as a series of equilibrium reactions, the
Amadori rearrangement is expected to be a reversible
process, that is to undergo a non-enzymatic reversal into
enolamines and subsequently into free sugars and amino
acid, as shown in pathway D. However, other pathways
as side reactions from ARP enolization or aldol-type
condensations between smaller sugar fragments gener-
ated from the decomposition of ARPs can also give rise
to the original aldose. Under physiological conditions
(pH 7.4 and 37 8C), Glc and Man were identified as
major products formed from protein-bound Amadori
compound fructoselysine, along with tetroses, pentoses
and 3-deoxyglucosone.8 Up until recently the formation
of parent sugars had only been reported for physiolo-
gical conditions. However, quantitative data concerning
the formation of parent sugars, Glc (10) and Man (11),
from DFG at the early stage of the reaction have now
been reported for conditions relevant to food proces-
sing.6 Besides 10 and 11, formic and acetic acid were
also detected, together with the free amino acid. Formic
and acetic acid have also been identified as two major
degradation products by heating the isolated Amadori
compound from a glucose/casein system.9 The proposed
mechanism suggests the C-1 to C-2 cleavage of 3 for the
formic acid formation and C-2 to C-3 cleavage of 5 or a
cleavage reaction of trioses intermediates, like 8, for the
formation of acetic acid. However, no attempt was
made to identify and quantify any a-dicarbonyl com-
pound to support such mechanism.
2. Experimental
2.1. Chemicals
The following compounds were obtained commercially:
Glycine, D-Glc, D-fructose (D-Fru), D-Man, formic acid,
ammonium acetate (Merck, Darmstadt, Germany);
AcOH, methylglyoxal, 2-methylquinoxaline, 5-(hydro-
xymethyl)-furan-2-carboxaldehyde
Germany); 1,2-diaminobenzene (78410, Fluka, Switzer-
(SigmaÁAldrich,
/
land).
2.2. N-(1-Deoxy-D
-fructos-1-yl)-glycine (1)12
Glycine (5.25 g, 0.07 mol),
Na2S2O5-sodium disulphite (7.6 g, 0.04 mol) were added
to a 1:1 MeOHÁwater solution (80 mL). The mixture
was then stirred for 20 min at room temperature (rt) and
subsequently refluxed for 8 h. In the end a 1:1 EtOHÁ
D-Glc (50.3 g, 0.28 mol) and
/
/
water solution (250 mL) was added. The final mixture
was applied onto an ion-exchange resin column (Dowex
50W*8, 3ꢁ
/
30 cm, Hꢀ). The column chromatography
of the residue was performed with 1:1 EtOHÁ
/
water
solution (750 mL), water (250 mL) and elution with
ammonia solution (2.5 L, 0.1 mol/L). Fractions of 100
mL were collected. Biochrom 20 Amino Acid Analyzer,
Pharmacia, was used to check the presence of 1. The
fractions containing only 1 were pooled and concen-
trated until dryness at 40 8C in a rotavapor. The
resulting syrup was dissolved in MeOH (using ultrasonic
bath) and kept in the fridge overnight. A white
precipitate was collected by filtration through a glass-
sintered funnel (G4). The final yield was 19% based on
the amount of glycine used. Anal. Calcd for C8H15NO7:
C, 40.50; H, 6.37; N, 5.91. Found: C, 40.13; H, 6.4; N,
5.89. mp: 160Á163 8C. On the Biochrom 20 Amino Acid
Analyser the synthesised compound showed only one
3
peak. In addition, DFG was also kindly offered by
Drs. Blank and Davidek, from Nestle´ Research Center
Lausanne, P.O. Box 44, Vers-Chez-les-Blanc, CH-1000
Lausanne 26, Switzerland.
The aim of the present study was to establish the main
thermal degradation pathways of DFG by applying
multiresponse kinetic modelling. The multiresponse
kinetics analysis considers reaction pathways in more
detail. It provides extra information about the reaction
mechanism since the reactants degradation is analyzed
simultaneously with the intermediates formation. First a
reaction mechanism is proposed and then the multi-
response kinetic analysis is applied.10 The following
steps should be taken into account: (i) identification and
quantification of the reactants and main products
formed; (ii) identification of reaction pathways based
/

1654
S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á1663
/
2.3. 2-Methyl-3-(1,2,3-trihydroxypropyl)quinoxaline
(5?)13
pH 2.8 for 16 min and column temperature 80 8C;
lithium citrate buffer pH 3.0 for 21 min and column
temperature 75 8C; lithium citrate buffer pH 3.15 for 12
min and column temperature 75 8C; lithium citrate
buffer pH 3.55 for 21 min and column temperature
75 8C; lithium hydroxide solution for 6 min and column
temperature 85 8C; lithium citrate buffer pH 2.8 for 8
min and column temperature 85 8C; lithium citrate
buffer pH 2.8 for 20 min and column temperature
75 8C; lithium citrate buffer pH 2.8 for 5 min and
column temperature 80 8C. All buffers were provided by
Biochrom, England. After postcolumn derivatization
with ninhydrine, the reaction products were determined
photometrically at 570 nm: 1 (tr 9.5 min); glycine (tr 38.8
min). Quantification was done using external standards.
A mixture of
D-Glc (196 mmol), L-alanine (183 mmol)
and 1,2-diaminobenzene (183 mmol) in phosphate
buffer (800 mL, 0.5 M, pH 6.8) was refluxed for 12 h.
A very black mixture was obtained, with a crude type
precipitate at the bottom. After cooling to rt, the
reaction mixture was extracted with methylene chloride
(5ꢁ200 mL). The organic layers were then combined
/
and dried over Na2SO4. After concentration till dryness,
the residual was dissolved in ethyl acetate (EtOAc) and
the separation was performed by column chromatogra-
phy (35ꢁ400 mm) on silica gel (200 g, silica gel 60,
/
Merck) conditioned with EtOAc. Elution was per-
formed first with EtOAc (1.5 L) followed by a 99:1
EtOAcÁ
and analysed by thin layer chromatography (TLC) on
silica gel (20ꢁ20 cm, 0.5 mm, Merck) using EtOAcÁ
MeOH (99:1) as the mobile phase. The target compound
was found at Rf ꢂ0.3Á0.4. It was visualized by UV and
/
MeOH (1 L). Fractions of 30 mL were collected
2.6. Quantification of 1-deoxy-2,3-hexodiulose (5), 3-
deoxy-2-hexosulose (3) and methylglyoxal (8)14
/
/
At each reaction time, 1 mL of sample was withdrawn
and put into 1 mL of water and 2 mL of a methanolic
solution of 1,2-diaminobenzene (1 mol/L). The mixtures
were maintained overnight at rt (25 8C). With previous
experimental tests it was established that this treatment
ensured a complete derivatization of the target com-
pounds into their quinoxaline form (Scheme 2). The
mixtures were analysed by RP-HPLC using a solvent
gradient starting with a mixture (10/90; v/v) of acetoni-
trile and ammonium acetate buffer (pH 3.5; 20 mmol/L)
and increasing the acetonitrile content to 30% within 50
min. The flow rate was 0.4 mL/min. By monitoring the
effluent at a wavelength of 320 nm a similar chromato-
gram to the used reference was obtained. The peaks
were identified by direct comparison of their retention
times with that of authentic samples and by spiking the
samples with standards. By comparison with 5? the 2-
(2,3,4-trihydroxybutyl)quinoxaline (3?) was identified.
The quantification was done using 5? and methylqui-
noxaline (10?) as external standards. Both quinoxalines
have the same extinction coefficient. The same extinc-
tion coefficient was found for 2-methyl-3-(1,2,3-trihy-
droxypropyl)quinoxaline and 2-(1,2,3,4-tetrahydroxy-
butyl)quinoxaline.15 Increasing the column temperature
to 40 8C optimised the peak separation. Compound 5, 3
and 8 were detected as their corresponding quinoxaline
derivatives at the retention times (tr) given in parenth-
esis: 2-methyl-3-(1,2,3-trihydroxypropyl)quinoxaline (tr
11.3 min); 2-(2,3,4-trihydroxybutyl)quinoxaline (tr 12.4
min); 2-methylquinoxaline (tr 38.8 min).
/
/
spraying with a KMnO4 solution. Pure fractions were
combined and concentrated till dryness and then
recrystalised from MeOH. In the end we obtained fine,
1
white yellowish crystals (0.02% in yield). H NMR (200
MHz; CD3OD) d 2.86 (s, 3 H, Ã
H, ÃCH2OH), 4.08 (m, 1 H, ÃCH(OH)Ã
CCH(OH)Ã), 7.77 (m, 2 H, ÄCHÃ), 7.98 (m, 1 H, Ä
CHÃ), 8.096 (m, 1 H, ÄCHÃ) which is in agreement with
/
CH3), 3.81Á
/
3.92 (m, 2
/
/
/
), 5.10 (d, 1 H,
Ã
/
/
/
/
/
/
/
/
the reference used. Also the high performance liquid
chromatography (HPLC) detection method revealed
only one peak.
2.4. Preparation of the reaction mixtures
Compound 1 (0.237 g, 10 mmol) was dissolved in
phosphate buffer (100 mL, 0.1 M K2HPO4/KH2PO4,
pH 6.8 and 5.5) and heated at 100 8C for a maximum of
4 h and at 120 8C for a maximum of 3 h, in an oil bath,
in screw-capped glass tubes (Schott, 16ꢁ160 mm). At
/
predetermined heating times, samples were taken and
immediately cooled in ice water, prior to analyses. Each
reaction mixture was prepared, heated and analysed in
at least duplicate.
2.5. Quantification of DFG (1) and glycine
The reaction mixtures were diluted with water (1:10) and
the pH adjusted with loading buffer (LB, Biochrom,
England) to 2.2 in the proportion 100 mL of diluted
sample to 900 mL of LB. The capsules were then loaded
with 20 mL of sample, 20 mL of internal standard
(glutamic acid; 25 mmol/mL) and 20 mL of LB. The
analysis was performed on a Biochrom 20 Amino Acid
Analyser, Pharmacia, England. The separation was
done using the following program: lithium citrate buffer
2.7. Quantification of sugars (D-Glc (10) and D-Man
(11)) and organic acids (formic and acetic acid)
Diluted samples (1:10) were analysed by HPLC, using
an ion-exchange column (ION-300, Interaction Chro-
matography Inc., San Jose, CA). Eluent 2.5 mmol/L

S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
1655
Scheme 2. a-Dicarbonyl derivatization with 1,2-diaminobenzene: 3-deoxy-2-hexosulose (3); 1-deoxy-2,3-hexodiulose (5); methyl-
glyoxal (8) and the respective quinoxaline derivatives: 2-(2,3,4-trihydroxybutyl)quinoxaline (3?); 2-methyl-3-(1,2,3-trihydroxypro-
pyl)quinoxaline (5?); 2-methylquinoxaline (8?).
H2SO4 in milipore water, flow 0.4 mL/min, column
temperature 85 8C. Sugars were detected by monitoring
the refractive index: Glc (tr 15.1 min); Man (tr 16.4 min);
and organic acids by their absorbance at 210 nm: formic
acid (tr 21.6 min); and AcOH (tr 23.7 min). All
compounds were quantified by external standards.
2.10. Melanoidins quantification
The browning intensity of the heated reaction mixtures
was determined by measuring the absorbance at 470 nm
with a spectrophotometer (Pharmacia Biotech, Upsala,
Sweden). When necessary the samples were diluted with
demi water. The absorbance was then recalculated to the
concentration of the melanoidins by using the equation
2.8. Quantification and separation of
(11) and -Fru (13)
D-Glc (10), D-Man
of LambertÁ
/
Beer. The extinction coefficient formed in
D
Glc-glycine was measured to be 0.649
/0.03 l/mmol cm at
470 nm.17 The concentration of melanoidins is thus
expressed as moles of sugar incorporated in the brown
polymers.
After heating, diluted samples (1:100) were analysed by
a Dionex system (Sunnyvale, CA) using a CarboPack
PA100 column. The separation was done using the
following gradient: starting mixture (16/0/84, v/v) of
NaOH (0.1 M), sodium acetate (NaOAc) (1 M in NaOH
0.1 M) and water for 20 min. After NaOAc increased to
(0/100/0) and was kept isocratic for 10 min, after which
the gradient changed to (100/0/0) in 1 min and was kept
isocratic for 5 min. The gradient was then brought to the
initial starting mixture and kept for 25 min, before the
next injection. The flow rate was 1 mL/min. Detection
was performed by an electrochemical detector (model
3. Results and discussion
The effect of the combination of pH and temperature on
the decomposition of 1 was studied at 100 and 120 8C,
with initial pH values of 5.5 and 6.8. By analysing the
reactant degradation and the formation of the main
products, under different reaction conditions, the reac-
tion routes could be established. The following steps
were taken.
ED-40):
-Fru (tr 16.59 min). All compounds were quantified by
external standards.
D-Glc (tr 12.95 min); D-Man (tr 13.9 min); and
D
3.1. Identification and quantification of the reactant 1
(DFG) and main products formed
2.9. Determination of heterocyclic compounds16
3.1.1. Degradation of DFG. The rate of degradation of 1
increased with temperature and pH (Fig. 1). After
heating for 30 min at 120 8C and pH 6.8, 1 was
completely degraded whereas after 180 min at 100 8C
and pH 5.5, only approximately 57% of the initial
concentration had reacted. Moreover, when comparing
the system at 100 8C and pH 6.8 with the system at
120 8C and pH 5.5, a similar degradation rate was
observed. These results indicate that decreasing the
initial reaction pH with 1.3 units or increasing the
5-(Hydroxymethyl)furan-2-carboxaldehyde (HMF), 4-
hydroxy-2-(hydroxymethyl)-5-methyl-3(2H)-furanone
(HHMF) and 2,3-dihydro-3,5-dihydroxy-6-methyl-4H-
pyran-4-one (DDMP) were determined by RP-HPLC
(Lichrosorb RP-18, Merck). The eluent was 7.5:100
MeOHÁwater solution, flow 0.6 mL/min, UV detection
/
at 280 nm. Only HMF was quantified by external
calibration line with 5-(hydroxymethyl)furan-2-carbox-
aldehyde (SigmaÁAldrich, Germany).
/

1656
S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á1663
/
the thermal degradation of 1 was compared at different
pH, either in water or in phosphate buffer solution. It
was observed that the yield of glycine in water increased
with increasing pH, whereas in phosphate buffer solu-
tion the opposite was observed. Also NaOH was added
to keep the pH constant throughout the reaction period.
The authors suggested that phosphate slowed down the
degradation of glycine at lower pH, mainly pH 5 and 6.
However, it has been reported18 that phosphate ions act
as a catalyst leading to enhanced degradation of
Amadori compounds with an optimum in the pH range
between 5 and 7, which contradicts their suggestion. A
possible reason for the observed decrease of the glycine
yield with increasing pH in phosphate buffer solution
can be the increasing addition of NaOH with increasing
pH. It may enhance the formation of alkyl(1H)pyrazi-
nones, which can retain more than one glycine mole-
cule.7 We will come back to glycine degradation
pathways in the following step.
Fig. 1. DFG thermal degradation at 100 (^, pH 5.5;', pH
6.8) and at 120 8C (k, pH 5.5; m, pH 6.8).
temperature with 20 8C has the same effect on the DFG
degradation. However, this observation is only valid in
the pH range between 5 and 7.6
3.1.4. a-Dicarbonyls formation. (a) The deoxyosones, 3
(3-DG) and 5 (1-DG) were identified and quantified in
their quinoxaline form. As shown in Fig. 3, at both
studied pH values and temperatures, both deoxyosones
were formed. Compared to the amount of DFG reacted,
a quite high concentration of 1-DG and 3-DG was
already present after 5 min at pH 5.5 and temperature
100 8C (34%). Independently of the temperature, at
lower pH 3-DG was present in larger amounts relatively
to 1-DG. At pH 6.8 that difference decreased. An
increase in pH seems to favour the 1-DG formation.
These findings are in line with what has been reported.19
Under more alkaline conditions, relative to the pKa of
the ARP (DFG pKa is 8.2), Amadori compound is
believed to undergo 2,3-enolization, as well, with 1-DG
formation. Also, independently of the pH and tempera-
ture, as the reaction proceeded the amount of 1-DG
decreased much faster than the one of 3-DG. (b)
Another identified and quantified a-dicarbonyl was 8
(MG) as its quinoxaline derivative. Contrary to the
temperature increase, the increase of pH had almost no
influence on the MG formation (Fig. 4). A very low
increase in the peak area of methylquinoxaline, when the
pH increased from five to seven, was also observed in a
glucose/b-alanine reaction mixture heated in phosphate
buffer for 12 h.20 However, the influence of temperature
is quite significant. An increase from 100 to 120 8C more
than doubled the formation of MG. No decrease in its
amount was observed with time.
3.1.2. Main products formed. Special attention was paid
to the formation of the free amino acid, glycine; a-
dicarbonyls, 1- and 3-deoxyosone together with methyl-
glyoxal, parent sugars, Glc and Man; and organic acids,
formic and acetic acid.
3.1.3. Free amino acid formation. Similar to DFG
degradation, glycine formation increased with pH and
temperature (Fig. 2). Also a decrease in the initial
reaction pH of 1.3 units had a similar effect in glycine
formation as increasing the temperature with 20 8C.
Moreover, when comparing the yield of glycine (amount
of glycine formed/amount of DFG that reacted), we
observed that at the early stage of the reaction the yield
of glycine increased with time (Table 1). However, as the
reaction proceeded a decrease was observed. Moreover,
in all the studied systems the decrease in the yield of
glycine was followed by an increase again. Also, the
yield of glycine increased with increasing pH, indepen-
dently of the temperature. In a recently published study6
3.1.5. Parent sugars formation. Compound 10 and 11
have recently been reported as the DFG parent sugars.6
No other sugar was mentioned in that study, even
though Fru (13) can also be formed by isomerisation of
Glc and Man.21 Based on those results,6 initial sugar
analyses were carried out using an Ion exchange HPLC
Fig. 2. Glycine formation during thermal degradation of DFG
at 100 (^, pH 5.5;', pH 6.8) and 120 8C (k, pH 5.5; m, pH
6.8).

S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
1657
Table 1
Glycine yield from DFG heated in phosphate buffer (0.1 M) at different reaction conditions
Heating time (min)
Glycine yield (%) a
A (T 100 8C pH 5.5)
B (T100 8C pH 6.8)
C (T120 8C pH 5.5)
D (T120 8C pH 6.8)
5
10
43.40
n.a.
n.a.
n.a.
25.87
62.05
59.56
69.79
66.20
71.95
n.a.
n.a.
83.92
94.72
95.82
92.36
84.41
n.a.
15
30
57.89
53.61
78.23
79.69
81.46
72.39
65.03
n.a.
71.02
93.85
87.44
84.91
n.a.
45
60
75
90
78.42
76.59
80.14
81.45
74.20
70.18
n.a.
84.05
86.39
n.a.
120
150
180
78.31
n.a.
n.a.
n.a.: not analyzed.
a
The yield of glycine was calculated based on the amounts of DFG reacted.
column, knowing that with this method Man and Fru
have the same retention time, but providing a good
separation of these sugars with Glc. It was assumed that
the Fru/Man peak was mainly Man. However, to check
the assumption that no Fru was formed, samples from
all the systems studied taken at 120 min were also
analysed using the Dionex method, with which it was
possible to separate Fru from Man, besides Glc.
Contrary to our expectations, Fru was also observed
in addition to Glc and Man, in samples heated at pH
initial value 6.8. To gain insight whether Fru would be
an isomerisation product from Glc and Man, systems at
pH 6.8 heated at 100 and 120 8C were analysed with the
Dionex method. As shown in Fig. 5 the formation of
Fru showed no lag phase, indicating that it can be
formed directly from DFG rather than from isomerisa-
tion of Glc and Man. Moreover, independently of the
temperature, at pH 5.5, Man was formed in higher
amounts than Glc (Fig. 5 A and C). Also, as the reaction
proceeded at 120 8C (Fig. 5C) Man reached a maximum
after 30 min, decreasing afterwards, whereas Glc did not
decrease after 120 min of heating. At pH 6.8 the
opposite was observed, that is, Glc not only was formed
in higher amounts than Man but also seemed to be more
reactive, showed by the fast decrease at 120 8C (Fig. 5 B
and D). Fru, even though formed in considerable
Fig. 4. Methylglyoxal formation during thermal degradation
of DFG at 100 (^, pH 5.5;', pH 6.8) and 120 8C (k, pH 5.5;
m, pH 6.8).
Fig. 3. Deoxyosones formation during thermal degradation of DFG at 100 (A) and 120 8C (B): 1-DG (^, pH 5.5; ', pH 6.8) and
3-DG (k, pH 5.5; m, pH 6.8).

1658
S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á1663
/
Fig. 5. Sugars formation during thermal degradation of DFG at pH 5.5 (---) and pH 6.8 (Á Á Á). A and B heated at 100 8C; C and D
heated at 120 8C. -Glc (m); -Man ('); -Fru (ꢁ).
D
D
D
/
amounts was always lower in concentration than the
other two sugars. Also independently of the reaction
conditions no decrease in the Fru amount was observed.
organic acids, in particular acetic acid, are important
end products of the Maillard reaction. Independently of
the reaction conditions, acetic acid was always formed in
higher amounts than formic acid. The difference was
more significant at pH 6.8.
3.1.6. Organic acids formation. The formation of formic
acid and acetic acid was particularly favoured with
increasing pH (Fig. 6). At pH 5.5 the yield of formic and
acetic acid was considerably lower relative to pH 6.8.
Also an increase of temperature enhanced formation of
both acids. The sum of the yields of both acids reached
up to 0.8 mmol/mmol of DFG, which indicates that
In the present study the reaction pH was not kept
constant. As the reaction proceeded a pH drop was
observed at both studied initial pHs. In Fig. 7 the results
are shown for initial pH 6.8. After 120 min at 120 8C the
pH drop was more significant than at 100 8C after 180
min, 0.21 and 0.19, respectively. Due to the low buffer
Fig. 6. Organic acids formation during thermal degradation of DFG at 100 (A) and 120 8C (B): acetic acid (k, pH 5.5; m, pH6.8);
Formic acid (^, pH 5.5;', pH 6.8).

S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
1659
formed then is approximately the same as we used as
reactant in the present study.
3.2. Identification of the reaction pathways based on
reaction conditions
3.2.1. Degradation of DFG. Two possible degradation
pathways of DFG are the enolization pathway and the
retro-aldolization pathway. As mentioned before,
through enolization the aminoenol intermediates can
lead to either 3 or 5. Although only small amounts may
be present at a given time, large amounts of ARP may
degrade via one or both of these pathways. The reaction
conditions, especially pH, are believed to determine the
degradation pathway taken.23 At lower pH, 1,2-enoliza-
tion is believed to be favoured and therefore a higher
yield of 3-DG is expected, which is in agreement with the
results obtained in the present study. On the other hand
as the pH increases to neutral/alkaline values, 2,3-
enolization is believed to be favoured and therefore 1-
DG should be present in higher amounts. In literature
the results are however contradictory. By heating 1-
Fig. 7. pH drop during thermal degradation of DFG at initial
reaction pH 6.8. (k, 100 8C; ꢁ, 120 8C).
/
capacity at initial pH of 5.5 the pH drop for the same
reaction conditions was 0.8 and 0.9, respectively, even
though the sum of acids formed was lower. Davidek and
co-workers6 observed in their study that the total
amounts of acetic and formic acid were generally lower
than the amounts of NaOH added to keep the pH
constant, suggesting that other acids might be formed. It
would be interesting though to study the reaction mass
balance to see how much is missing. One of the
consequences of allowing the pH to fall is the inhibition
of the reaction by the pH drop. However, the level-off
observed on organic acids formation (Fig. 6B) matched
the complete degradation of DFG, which implies direct
involvement of DFG or its early degradation product in
the formation of organic acids.
deoxy-1-propylamino-D-fructose in phosphate buffer
for 10 h under reflux the 1-DG/3-DG ratio was 8:5 at
pH 4.5 and 20:1 at pH 7.24 For quantification of
deoxyosones several researchers reacted Maillard mix-
tures in the presence of excess amounts of 1,2-diamino-
benzene.15,20,24 This in situ trapping technique leads to
an accumulation of the quinoxalines, but does not offer
insights into the relative changes in the concentration of
the deoxyosones or any other dicarbonyl compound.25
We, therefore, derivatized the deoxyosones produced
with 1,2-diaminobenzene after rapid cooling of the
thermally treated DFG. From the results presented in
the previous section, independently of the pH, 3-DG
was always present in higher concentration than 1-DG.
Hofmann and co-workers13,14 in a heated aqueous
solution of Glc and l-alanine under reflux at pH 7
found 3-DG in four times higher amounts than 1-DG.
In the present study at 100 8C and pH 6.8 a ratio of 3.2
was found.
3.1.7. Melanoidins formation. The concentration of
melanoidins (nitrogenous brown polymers) was calcu-
lated from the absorbance data through the extinction
coefficient. As observed in Fig. 8, pH more than
temperature had a strong influence in melanoidins
formation from DFG. Moreover, the observed colour
formation is much less than when one starts with Glc/
glycine system,22 even though the amount of DFG
To gain insight whether the amount of 1-DG is lower
because its formation is not favoured or because it is
more reactive than 3-DG, compounds formed by
cyclization/condensation of the deoxyosones were also
analysed. They include HHMF, 2,3-dihydro-3,5-dihy-
droxy-6-methyl-4(H)-pyran-4-one (DDMP) and 5-(hy-
droxymethyl)-2-furfural (HMF). Both the furanone
(HHMF) and the dihydropyranone (DDMP) derive
from 1-DG and are regarded as the 2,3-enolization
indicator.16 In the present study both HHMF and
DDMP were identified at both pHs, which indicates
that 2,3-enolization occurs not only at pH 6.8 but also at
pH 5.5. Unfortunately, their quantification was not
possible since no reference material was available.
However, judging by the response factor of HMF they
Fig. 8. Melanoidins formation during thermal degradation of
DFG at 100 (^, pH 5.5;', pH 6.8) and 120 8C (k, pH 5.5;
m, pH 6.8).

1660
S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á1663
/
were formed in the order of magnitude of mmoles. On
the other hand, HMF was not formed at pH 6.8 and
only mmole amounts at pH 5.5. The relatively low
reactivity of 3-deoxyglucosone as well as its relatively
poor methylated pyrazine precursor has already been
observed in previous studies.26,27 Also HMF formation
has only been reported in more acidic conditions, such
as pH 3.5.28 Besides enolization, intact Amadori pro-
ducts can also undergo retro-aldol reactions or acid/
base-catalyzed thermal degradations without deoxyo-
sone formation, to produce a variety of other reactive
intermediates that retain the amino acid, such as 1-
amino-1,4-dideoxy-2,3-dicarbonyl compounds.29 These
compounds can release the free amino acid later by b-
elimination, which can justify the oscillation observed in
the yield of glycine.
glycine. Moreover, MG is also believed to degrade
further into formic and acetic acid.4
3.2.4. Parent sugars formation. The formation of Glc
and Man together with the free amino acid, at early
stages of the reaction could be considered as sufficient
evidence for the reversibility of the rearrangement, as
shown in pathway D in Scheme 1. However, this
pathway does not explain the formation of Fru at the
early stage of the reaction. The fact that no lag phase
was observed made us exclude the isomerisation possi-
bility from Glc and Man. According to the LobryÁ
/
de-
all
BruynÁAlberdaÁvan-Eckenstein-rearrangement
/
/
three sugars, Glc, Man and Fru are in equilibrium
with the same intermediate, the 1,2-endiol. However,
Fru is also in equilibrium with the 2,3-endiol.21 The
formation of 1,2-endiol from the respective enaminol (2)
in Scheme 1 is not so likely to happen, however, by
release of the amino acid the enaminol 4 can originate its
endiol, 2,3-endiol (12), which through enolization can
lead to sugars formation,19 in particular Fru (13). This
indicates that Fru can be formed from DFG by its 2,3-
enolization step, whereas Man and Glc can only be
formed via the 1,2-enaminol, through the schiff base (9
and 9?). Moreover, it could also be argued that the
sugars might arise by aldol-type condensations between
smaller sugar fragments generated from the decomposi-
tion of DFG. This hypothesis has, however been
excluded by labelling studies.33 To get a better insight
in the sugars enolization pathways two additional
experiments were performed. Glc and Fru were heated
alone in an aqueous phosphate buffer (0.1 mol/L)
solution at 120 8C and pH 6.8. It was observed (results
not shown) that Glc isomerised preferably into Fru,
whereas Man was only formed in small amounts. On the
other hand, the amount of Glc and Man formed from
Fru was very low relatively to its decrease. Fru degraded
preferably into acids. It is suggested that not only 2,3-
enolization is favoured under these conditions with Fru
preferably formed from Glc, but also that the reversi-
bility of 2,3-endiol into 1,2-endiol occurs in a quite low
extent. These findings are in agreement with what we
have observed before, namely that 2,3-enolization gains
importance in the reaction mechanism as pH and
temperature increased.
3.2.2. Formation of glycine. Under the studied condi-
tions the formation of glycine from DFG can occur via
both 1,2- and 2,3-enolization. In theory, 1 mol of DFG
should yield 1 mol of glycine. The results obtained in the
present study show that the yield of glycine increased in
the beginning of the reaction and decreased as the
reaction proceeded. However, the decrease in the yield
of glycine was followed by an increase again. This
indicates that glycine is first liberated and then reacts
with other compounds present in the reaction mixture
(e.g., a-dicarbonyls and hydroxycarbonyls). These reac-
tions may include: (i) the formed a-dicarbonyls can
further react with glycine through Strecker degradation
to produce the corresponding amines, carboxylic acids
and Strecker aldehydes; (ii) formation of pyrazinones
and pyrazines;30 (iii) chain elongation of reactive C-2
and C-3 a-dicarbonyls by one carbon unit originating
from C-2 atom of glycine;31 (iv) incorporation of glycine
into melanoidins. However, because after a decrease an
increase was observed, glycine is suggested to react with
other compounds present in the reaction mixture and
subsequently released again. These reactions may in-
clude not only the conversion of carbohydrate frag-
ments by catalytic action of glycine, but also glycine can
be formed directly from DFG with the formation of 6
that subsequently undergoes b-elimination to produce
MG and release free glycine.4,7
3.2.3. Methylglyoxal formation. In a previous study27
a
mechanism for the MG formation was proposed
through retro-aldolization of both 1- and 3-deoxyo-
sones. This would justify the absence of pH influence
observed in MG formation. Another possibility would
be the direct cleavage of DFG into MG as presented in
Scheme 1. This process of conversion of glyceraldehyde
into MG by catalytic action of amino acid was also
observed with Glc/glycine model systems.32 Literature
results thus give us some indication that DFG degrades
also by retro-aldolization into MG with release of
3.2.5. Organic acids formation. Labelling experiments34
suggest that Glc, Man and Fru together with 1-DG and
3-DG can be regarded as aliphatic acid precursors
(Scheme 3). From the experiments where Glc and Fru
were heated alone we observed that the main acid
formed in the beginning of the reaction was formic
acid. Acetic acid formation showed a lag phase,
especially in the Glc system. These results were com-
pared with the ones obtained from heating Glc and Fru,
separately with glycine. When the amino acid was

S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
1661
Scheme 3. Reaction scheme for organic acids formation from primary thermal degradation products of DFG (adapted from Ginz
and co-workers34).
present, the formic acid amount decreased considerably
and acetic acid became the dominant acid (results not
shown). It is worth noting that no glycolic acid or lactic
acid were detected. These results suggest that formic and
acetic acid can also be formed by sugars enolization.
However, in a study35 where Glc was heated alone at
100 8C (pH 10), 3-DG was estimated to be only 14% of
the total a-dicarbonyls detected. In the present study we
have therefore considered 1-DG and 3-DG to be mainly
formed by DFG degradation. Through dicarbonyl
cleavage the deoxyosones become the main precursors
of organic acids formation. The fact that no MG
degradation was observed made us assume that a C-1
to C-2 cleavage of MG would be a secondary route in
organic acids formation.
3.3. Differentiate between primary and secondary
reaction routes
Based on the discussion above we suggest the following
primary and secondary routes.
3.3.1. Primary routes.
. DFG degradation through enolization occurs
through two intermediates, designed as E1 and E2,

1662
S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á1663
/
which can be the Schiff’s base, the cation form of the
all studied systems, seems to be related with the higher
reactivity of 1-deoxyosone. Independently of the path-
way taken, enolization or retro-aldolization, DFG
degradation is accompanied by amino acid release.
Together with glycine, acetic acid was the main end
product formed. At this stage we are able to propose a
kinetic model based on the established reaction network.
In Part II11 the following steps for a complete multi-
response kinetic analysis will be taken.
Schiff’s base, the 1,2-enaminol or the 2,3-enaminol.
These intermediates have not been isolated yet from
the Maillard reaction due to their reactivity, however,
according to previous studies36,37 ARPs undergo 1,2-
or 2,3-enolizations while the amino acid moiety is still
attached. E1 is favoured at lower pH whereas E2 is
favoured at higher pH.
. The intermediates E1 and E2 by release of the amino
acid lead to the formation of 3-DG and 1-DG,
respectively.
. Glc, Man and 3-DG are formed through the same
intermediate (E1) whereas Fru is formed by E2.
Moreover, Man and Glc can isomerise into each
Acknowledgements
The authors would like to thank the Fundacao para a
¸˜
other as well as degrade into Cn (n 5
compounds.
/
6) carbonyl
Cieˆncia e Tecnologia, Ministe´rio da Cieˆncia e Tecnolo-
gia (Portuguese foundation for science and technology,
Ministry of Science and Technology) for its financial
support.
. DFG degradation through retro-aldol cleavage leads
to MG formation, with amino acid release.
. DFG, 3-DG and 1-DG due to their reactive func-
tional group can easily degrade to produce reactive
Cn carbonyl compounds.
References
. Formic acid and acetic acid are formed from DFG
degradation pathways through E1 by 3-DG dicar-
bonyl cleavage and through E2 by 1-DG dicarbonyl
cleavage, respectively.
. Melanoidins formation results from the interaction of
Cn carbonyl compounds with glycine.
1. Hodge, J. E. J. Agric. Food Chem. 1953, 1, 928Á943.
/
2. The Maillard Reaction in Aging, Diabetes and Nutrition;
Baynes, J. W.; Monnier, V. M., Eds. Progress in Clinical
and Biological Research; Vol. 304; Alan R. Liss, Inc: New
York, 1989.
3. Staempli, A. A.; Blank, I.; Fumeaux, R.; Fay, L. B. Biol.
Mass Spectrom. 1994, 23, 642Á
4. Huyghues-Despointes, A.; Yaylayan, V. A. J. Agric. Food
Chem. 1996, 44, 672Á681.
5. Coleman, W. M., III; Chung, H. L. J. Anal. Appl.
Pyrolysis 2002, 62, 215Á223.
6. Davidek, T.; Nathalie, C.; Aubin, S.; Blank, I. J. Agric.
Food Chem. 2002, 50, 5472Á5479.
7. Keyhani, A.; Yaylayan, V. A. J. Agric. Food Chem. 1996,
44, 2511Á2516.
8. Zyzak, D. V.; Richardson, J. M.; Thorpe, S. R.; Baynes, J.
W. Arch. Biochem. Biophys. 1995, 316, 547Á554.
/
646.
3.3.2. Secondary routes.
/
. Products formed by cyclization/condensation of the
deoxyosones that include HMF, HHMF and
DDMP.
. Incorporation of glycine in pyrazinones and pyra-
zines, as well as in chain elongation of a-dicarbonyls.
Also degradation of glycine through Strecker degra-
dation to produce the corresponding amines, car-
boxylic acids and Strecker aldedehydes, which can be
included in the melanoidins formation.
. MG formation from deoxyosones retro-aldol reac-
tion and its degradation into carboxylic acids.
. Isomerisation of Glc and Man into Fru, as well as
direct degradation of Glc into 3-DG.
. Fru formation from E1 as well as its degradation into
reactive Cn carbonyl compounds.
/
/
/
/
9. Brands, C. M. J. Ph.D. Thesis, Wageningen University,
The Netherlands, 2002.
10. Martins, S. I. F. S.; Jongen, W. M. F.; Van Boekel, M. A.
J. S. Trends Food Sci. Technol. 2001, 11, 364Á
11. Martins, S. I. F. S.; Van Boekel, M. A. J. S., Kinetic
modelling of the Amadori N-(1-deoxy- -fructos-1-yl)-
glycine degradation pathways. Part II*Kinetic analysis,
Carbohydr. Res. 2003 (in press).
12. Kroh, L. W.; Schroder, R.; Mugge, C.; Wesrphal, G.;
/373.
D
/
. Carboxylic acids formation from sugars enolization
or scission products from Cn carbonyl compounds.
¨
¨
Baltes, W. Z. Lebensm.-Unters.-Forsch. 1992, 194, 216Á
/
221 (in German).
13. Hofmann, T.; Bors, W.; Stettmaier, K. J. Agric. Food
Chem. 1999, 47, 379Á
14. Hofmann, T. Eur. Food Res. Technol. 1999, 209, 113Á
15. Morita, N.; Inoue, K.; Takagi, M. Agric. Biol. Chem.
1981, 45, 2665Á2668.
16. Knerr, T.; Severin, T. Z. Lebensm.-Unters.-Forsch. 1993,
196, 366Á369.
/
390.
4. Conclusions
/121.
The main degradation pathways for DFG have been
established. pH, more than temperature, has an influ-
ence on the reaction products formed. An increase in pH
seems to favour the formation of 1-deoxyosone. The
lower amount found comparatively to 3-deoxyosone, in
/
/
17. Martins, S. I. F. S.; Van Boekel, M. A. J. S. Extinction
coefficient of melanoidins in the glucose/glycine Maillard

S.I.F.S. Martins et al. / Carbohydrate Research 338 (2003) 1651Á
/1663
1663
reaction: influence of pH and temperature, Food Chem.
2003 (in press).
26. Weenen, H.; van der Ven, J. G. M.; van der Linde, L. M.;
van Duynhoven, J.; Groenewegen, A. C4, C5 and C6 3-
Deoxyglycosones: structures and reactivity. In: O’Brien, J.;
Nursten, H. E.; Crabbe, M. J. C.; Ames, J. M., (Eds.); The
Maillard reaction in foods and medicine, The Royal Society
18. Potman, R. P.; van Wijk, T. A. In Thermal Generation of
Aromas; Parliament, T. H.; McGorrin, R. J.; Ho, C-T.,
Eds. Mechanistic studies of the Maillard reaction with
emphasis on the phosphate-mediated catalysis; American
of Chemistry, 1998; 223, pp. 57Á
27. Weenen, H. Food Chem. 1998, 62, 393Á
28. Ghiron, A. F.; Quack, B.; Mawhinney, T. P.; Feather, M.
S. J. Agric. Food Chem. 1998, 36, 677Á680.
/64.
/
401.
Chemical Society, 1989; pp 182Á
/195.
19. Anet, E. F. L. J. Adv. Carbohydr. Chem. 1964, 19, 181Á
/
/
218.
20. Nevidek, W.; Ledl, F.; Fischer, P. Z. Lebensm.-Unters.-
29. Vernin, G.; Debrauwer, L.; Vernin, G. M. F.; Zamkotsian,
R.-M.; Metzger, J.; Larice, J. L.; Parkanyi, C. In Off-
Flavors in Foods and Beverages; Charalambous, G.; Ed.
Heterocycles by thermal degradation of Amadori inter-
Forsch. 1992, 194, 222Á228.
/
21. De Bruin, J.M. Ph.D. Thesis, University of Delft, The
Netherlands, 1986.
mediates; Elsevier: Amesterdam, 1992; pp. 567Á
30. Oh, Y-C.; Shu, C-K.; Ho, C-T. J. Agric. Food Chem. 1992,
40, 118Á121.
31. Wnorowski, A.; Yaylayan, V. A. J. Agric. Food Chem.
2000, 78, 3549Á3554.
32. Olsson, K.; Pernemalm, P. A.; Theander, O. Acta Chem.
Scand. B. 1978, 32, 249Á256.
33. Yaylayan, V. A.; Huygues-Despointes, A. Carbohydr. Res.
1996, 286, 179Á187.
34. Ginz, M.; Balzer, H. H.; Bradbury, A. G. W.; Maier, H.
G. Eur. Food Res. Technol. 2000, 211, 404Á410.
35. Morita, N.; Takagi, M. Dev. Food Sci. 1986, 13, 59Á
36. Hodge, J. E.; Mills, F. D. Carbohydr. Res. 1976, 51, 9Á
37. Yaylayan, V.; Sporns, P. Food Chem. 1987, 26, 283Á305.
/623.
22. Martins, S. I. F. S.; Van Boekel, M. A. J. S. Role of
browning precursors in the glucose/glycine Maillard reac-
tion: Multiresponse analysis. In: Empis, J.A. (Ed.);
Functionalities of pigments in food, Federation of Eur-
/
/
opean Chemical Societies, 2002; 258, pp. 199Á
23. Tressl, R.; Nittka, C.; Kersten, E. J. Agric. Food Chem.
1995, 43, 1163Á1169.
24. Beck, J.; Ledl, F.; Severin, T. Carbohydr. Chem. 1988,
177, 240Á243.
/
202.
/
/
/
/
25. Martins, S. I. F. S.; Hofmann, T.; Van Boekel, M. A. J. S.
In situ trapping technique applied to the identification of
carbohydrate degradation products. In COST Action 919
/
/65.
/21.
Á
/
Melanoidins in food and health, 2002; Vol. 3, pp. 18Á23.
/
/