Characterization of initial reaction intermediates in heated model systems of glucose, glutathione, and aliphatic aldehydes

Article abstract of DOI:10.1016/j.foodchem.2019.125482

To understand the effect of lipid degradation on Maillard formation of meaty flavors, initial reaction intermediates in model systems of glucose–glutathione with hexanal, (E)-2-heptenal, or (E,E)-2,4-decadienal were identified by HPLC–MS and by NMR. Besides Amadori compounds, hemiacetals and thiazolidines via addition of sulfhydryl to carbonyl or to the conjugated olefinic bond were found. Concentrations of all intermediates increased with reaction time while degradation of the intermediates with a glutathione moiety helped formation of thiazolidines with cysteinylglycine. The unsaturated aldehydes (E)-2-heptenal and (E,E)-2,4-decadienal exhibited high reactivity against glucose for glutathione, yielding higher levels of intermediate compounds than from glucose. Heating prepared intermediates reversibly released the original aldehydes, which caused various compounds formed by retro-aldol, oxidation, etc. to react with H₂S and NH₃. Among them, formation pathways including 3-nonen-2-one, 2-hexanoylfuran, and six dialkylthiophenes (e.g., 2-ethyl-5-(1-methylbutyl)thiophene) were proposed for the first time.

Full text of DOI:10.1016/j.foodchem.2019.125482

FoodChemistry305(2020)125482
Contents lists available at ScienceDirect
Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem
Characterization of initial reaction intermediates in heated model systems of
glucose, glutathione, and aliphatic aldehydes
⁎
Tianze Wanga, Dawei Zhenb, Jia Tana, Jianchun Xiea, , Jie Chengc, Jian Zhaoa
a Beijing Advanced Innovation Center for Food Nutrition and Human Health, Beijing Technology and Business University (BTBU), Beijing 100048, China
b Beijing Lanjingzhongyu Scientific Development Co. Ltd, Beijing 100048, China
c Institute of Quality Standard and Testing Technology for Agro-products of CAAS, Beijing 100081, China
A R T I C L E I N F O
A B S T R A C T
Keywords:
To understand the effect of lipid degradation on Maillard formation of meaty flavors, initial reaction inter-
mediates in model systems of glucose–glutathione with hexanal, (E)-2-heptenal, or (E,E)-2,4-decadienal were
identified by HPLC–MS and by NMR. Besides Amadori compounds, hemiacetals and thiazolidines via addition of
sulfhydryl to carbonyl or to the conjugated olefinic bond were found. Concentrations of all intermediates in-
creased with reaction time while degradation of the intermediates with a glutathione moiety helped formation of
thiazolidines with cysteinylglycine. The unsaturated aldehydes (E)-2-heptenal and (E,E)-2,4-decadienal ex-
hibited high reactivity against glucose for glutathione, yielding higher levels of intermediate compounds than
from glucose. Heating prepared intermediates reversibly released the original aldehydes, which caused various
compounds formed by retro-aldol, oxidation, etc. to react with H2S and NH3. Among them, formation pathways
including 3-nonen-2-one, 2-hexanoylfuran, and six dialkylthiophenes (e.g., 2-ethyl-5-(1-methylbutyl)thiophene)
were proposed for the first time.
Glutathione
Aliphatic aldehyde
Intermediate
Mechanism
Meat flavor
Maillard reaction
Lipid degradation
1. Introduction
with the amino compounds to form non-volatile intermediates, espe-
cially the Amadori rearrangement compounds, which are critical for
Glutathione (γ-Glu-Cys-Gly, GSH), which exists in most living ani-
mals and plants, functions in protecting cells from toxins such as free
radicals and peroxides, due to its reducing and nucleophilic properties
(Pan et al., 2014; Schumacher et al., 2017). In food systems, similar to
cysteine, GSH is known as an important meaty flavor precursor, which
can release H2S from cysteine and result in formation of various sulfur-
containing meaty flavors (El-massry, Farouk, & El-Ghorab, 2003; Lee, Jo,
& Kim, 2010; Zhou, Grant, Goldberg, Ryland, & Aliani, 2018). Previous
findings demonstrated that lipid degradation affected the Maillard re-
action of cysteine and reducing sugars to produce meaty flavor com-
pounds (Elmore, Campo, Enser, & Mottram, 2002; Farmer & Mottram,
1990; Yang et al., 2015). With the inclusion of a lipid, the amount of
sulfur-containing flavors formed was decreased, while some new alkyl
sulfur compounds (e.g., 2-hexylthiophene) were generated. A recent in-
vestigation (Zhao, Wang, Xie, Xiao, Cheng, et al., 2019) revealed similar
effect of lipids upon the Maillard reaction of GSH and glucose in the
production of sulfur-containing meaty flavors. Nevertheless, it remains
unclear how lipids affect the Maillard reaction to form meaty flavors.
When lipids are included in the Maillard reaction of reducing sugars
and amino compounds such as GSH, initially, reducing sugars will react
development of volatile flavor compounds (Hou et al., 2017; Liu,
Zhang, Huang, Song, & Nsor-Atindana, 2012). On the other hand, the
lipid degradation products of carbonyl compounds also react with the
amino compounds, forming non-volatile intermediates that can play
roles in the development of volatile food flavors (Starkenmann, 2003).
For example, it was reported that 3-S-glutathionylhexanal from reaction
of (E)-2-hexenal and GSH was the precursor of 3-mercapto-1-hexanol in
grape juices (Clark & Deed, 2018; Thibon et al., 2016). Therefore, in
this study, from the point of formation of non-volatile initial inter-
mediates, reaction model systems of GSH and glucose with hexanal, (E)-
2-heptenal, or (E,E)-2,4-decadienal, the typical carbonyls from fat oxi-
dation and degradation (Yang et al., 2015; Zhao, Wang, Xie, Xiao,
Cheng, et al., 2019), were investigated. Structures of the found initial
intermediate compounds in reaction mixtures were characterized by
HPLC–MS and NMR, and the roles of the aldehyde-derived inter-
mediates in development of volatile flavors were exposed by thermal
degradation of the prepared intermediates. The object of the work is to
gain insight into the flavor formation mechanism involving Maillard
reaction with the lipid degradation during processing of meat or meat-
like foods.
⁎ Corresponding author.
E-mail address: xjchun@th.btbu.edu.cn (J. Xie).
https://doi.org/10.1016/j.foodchem.2019.125482
Received 16 March 2019; Received in revised form 4 September 2019; Accepted 4 September 2019
Availableonline05September2019
0308-8146/©2019ElsevierLtd.Allrightsreserved.

T. Wang, et al.
FoodChemistry305(2020)125482
2. Experimental
mixture was loaded on the column. The column was eluted gradually
with 200 mL of 40% ethanol. The target fractions were collected and
freeze-dried, yielding 0.1 g of product M2. Similarly, the product M2
was subjected to HPLC–ELSD, HPLC–MS, and 1H NMR and 13C NMR
analyses.
2.1. Materials and chemicals
L-Glutathione (99%),
tenal (95%), (E,E)-2,4-decDa-dgileuncaolse(9(39%8%), )a,nhdexthaenaalu(t9h5e%nt)ic, (cEh)e-2m-ihceapls-
(≥95%) for identification of volatile flavors were purchased from J&K
Chemical Ltd. (Beijing, China). The n-alkanes (C5–C26) for determina-
tion of retention indices and other chemicals used were all of analytical
grade and purchased from Sinopharm Chemical Reagent Co., Ltd.
(Beijing, China).
Purity of the product M2 was 99% according to relative peak area
by HPLC–ELSD analysis. HPLC-(ESI+) MS: 420 ([M+H]+). 1H NMR
(600 MHz, D2O, ppm): δ 0.75–0.82 (m, CH3-17), 1.15–1.36 (m, CH2-15,
CH2-16), 2.01–2.13 (m, CH2-3, CH2-4, CH2-14), 2.84–2.91 (m, CH2-10),
3.75 (t, J = 6.30 Hz, CH-2), 3.90 (s, CH2-8), 4.45–4.54 (m, CH-6),
4.99–5.11 (m, CH-11), 8.35–8.47 (NH-CO); others: Others: 1.45–1.60
(m, CH2-14′), 2.95–3.08 (m, CH2-10′), 3.12–3.24 (m, CH2-11′). 13C
NMR (150 MHz, D2O, ppm): δ 13.21 (C-17), 21.80 (C-16), 25.31 (C-3),
25.92 (C-15), 31.11 (C-10), 31.23 (C-4), 34.12 (C-14), 41.42 (C-8),
53.68 (C-2), 55.57 (C-6), 89.26 (C-11), 125.82 (C-12), 129.14 (C-13),
172.41–175.10 (C-1, C-5, C-7, C-9), 177.61, 181.86 (C-1 or C-9 for
COO−); others: 24.66 (C-14′), 28.11 (C-10′), 38.59 (C-11′), 39.51 (C-
13′), 57.52 (C-6′), 206.38 (C-12′).
2.2. Model reactions
Three model reaction systems, i.e., glucose, glutathione, and hex-
anal (Glu-GSH-Hex); glucose, glutathione, and (E)-2-heptenal (Glu-
GSH-Hept); and glucose, glutathione, and (E,E)-2,4-decadienal (Glu-
GSH-Dec); along with the control systems, i.e., glutathione and glucose
(GSH-Glu); glutathione and hexanal (GSH-Hex); glutathione and (E)-2-
heptenal (GSH-Hept); and glutathione and (E,E)-2,4-decadienal (GSH-
Dec), were performed.
2.3.3. M3
GSH (0.460 g, 1.5 mmol) and (E)-2-heptenal (0.067 g, 0.6 mmol)
were first dissolved in 5 mL of phosphate buffer (0.2 M, pH 6.5) in a 15-
mL pressure-resistant tube, and then heated at 90 °C for 80 min. The
resulted reaction mixture was separated on a column (3.0 cm × 60 cm)
packed with polyamide resin (125–150 μm) (Mitsubishi Chemical
Corporation, Tokyo, Japan) using the above automatic liquid chroma-
tographic system with UV monitoring at 220 nm. The reaction mixture
was entirely loaded the column. The column was first washed with
300 mL of water, and then eluted gradually with 500 mL of ammonium
hydroxide (0.3 M). The target fractions were collected and freeze-dried,
yielding 0.06 g of product M3. Similarly, the product M3 was subjected
to HPLC–ELSD, HPLC–MS, and 1H NMR and 13C NMR analyses.
Purity of the product M3 was 94% in relative peak area by
HPLC–ELSD analysis. HPLC-(ESI+) MS: 580 ([M+H]+). 1H NMR
(600 MHz, D2O, ppm): δ 0.75 (t, J = 6.6 Hz, CH3-4′), 1.10–1.33 (m,
CH2-2′, CH2-3′), 1.38–1.58 (m, CH2-1′), 1.94–2.10 (m, CH2-3, CH2-4,
CH2-12), 2.60–2.66 (m, CH-11), 2.68–3.07 (m, CH2-10), 3.10–3.34 (m,
CH2-14), 3.59–3.75 (m, CH-2, CH2-8, CH-15, CH2-17), 4.39–4.48 (m,
CH-6), 4.73–4.80 (m, CH-13). 13C NMR (150 MHz, D2O, ppm): δ 13.35
(C-4′), 21.76 (C-3′), 26.08 (C-3), 27.96 (C-2′), 30.51 (C-4), 31.38 (C-1′),
33.36 (C-10), 34.80 (C-14), 38.68 (C-12), 42.98 (C-8), 43.43 (C-17),
44.45 (C-11), 52.52 (C-2), 57.37 (C-6), 64.75 (C-13), 66.29 (C-15),
171.66–176.28 (C-1, C-5, C-7, C-9, C-16, C-18).
The amounts of reactants used were GSH (0.30 mmol), glucose
(0.30 mmol), and aldehyde (0.70 mmol, each). The reactants were
weighed according to the respective systems and dissolved in 5 mL of
phosphate buffer (0.2 M, pH 6.5) in a 15-mL pressure resistant tube.
Then the tubes were sealed and heated at 140 °C while stirring for 0.16,
0.25, 0.5, 1, 2, 3, or 4 h. Two replicates were performed. The reaction
mixtures were subjected to HPLC–MS analysis, as described in Section
2.5.
2.3. Preparation of the intermediate compounds (L1, M2, M3, N4, and N5)
2.3.1. L1
GSH (0.184 g, 0.6 mmol) and hexanal (0.060 g, 0.6 mmol) were first
dissolved in 5 mL of phosphate buffer (0.2 M, pH 6.5) in a 15-mL
pressure-resistant tube, and then heated at 90 °C for 80 min. The ob-
tained reaction mixture was separated on a column (1.6 cm × 40 cm)
packed with AG 50W-X4 cation exchange resin (63–150 μm) (Bio-Rad
Co., Ltd., Shanghai, China) by a MD-99 automatic liquid chromato-
graphic system (Shanghai Qingpu Huxi Instruments Co., Shanghai,
China) with UV monitoring at 220 nm. The reaction mixture was en-
tirely loaded on the column. The column was first washed with 100 mL
of water, and then eluted gradually with 200 mL of ammonium hy-
droxide (0.3 M). The target fractions were collected and freeze dried,
yielding 0.1 g of product L1. The product L1 was subjected to
HPLC–ELSD, HPLC–MS, and 1H NMR and 13C NMR analyses. Regarding
the NMR analysis, an AV 600 NMR spectrometer (Bruker, Switzerland)
was used with D2O as the solvent and tetramethylsilane as the internal
standard.
2.3.4. N4 and N5
GSH (0.460 g, 1.5 mmol) and (E,E)-2,4-decadienal (0.228 g,
1.5 mmol) were first dissolved in 5 mL of phosphate buffer (0.2 M,
pH 6.5) in a 15-mL pressure-resistant tube, and then heated at 90 °C for
5 min. The resulted reaction mixture was separated on a column
(3.0 cm × 60 cm) packed with polyamide resin (125–150 μm)
(Mitsubishi Chemical Corporation, Tokyo, Japan) using the above au-
tomatic liquid chromatographic system with UV monitoring at 220 nm.
The reaction mixture was entirely loaded on the column. The column
was first washed with 300 mL of water, and then eluted gradually with
500 mL of ammonium hydroxide (0.3 M). The target fractions were
freeze-dried and collected, yielding 0.02 g of product N4 and 0.08 g of
product N5, respectively. Similarly, the products N4 and N5 were
subjected to HPLC–ELSD, HPLC–MS, and 1H NMR and 13C NMR ana-
lyses.
Purity of the product L1 was 99% according to relative peak area by
HPLC–ELSD analysis. HPLC-(ESI+
)
MS: 261([M+H]+). 1H NMR
(600 MHz, D2O, ppm): δ 0.78 (t, J = 6.4 Hz, CH3-11), 1.20–1.43 (m,
CH2-8, CH2-9, CH2-10), 1.66–1.86 (m, CH2-7), 3.05–3.24 (m, CH2-5),
3.67–3.80 (m, CH2-2, CH-4), 4.61–4.65 (m, CH-6). 13C NMR (150 MHz,
D2O, ppm): δ 13.20 (C-11), 21.75 (C-10), 25.56 (C-8), 30.76 (C-9),
35.37 (C-5), 37.58 (C-7), 43.20 (C-2), 64.91 (C-6), 69.84 (C-4), 172.01,
176.45 (C-1, C-3), 184.36 (C-1, COO−).
2.3.2. M2
GSH (0.276 g, 0.9 mmol) and (E)-2-heptenal (0.067 g, 0.6 mmol)
were first dissolved in 5 mL of phosphate buffer (0.2 M, pH 6.5) in a 15-
mL pressure-resistant tube, and then heated at 90 °C for 10 min. The
resulting reaction mixture was separated on a column (1.6 cm × 40 cm)
packed with HP 20SS macroporous resin (Mitsubishi Chemical
Corporation, Tokyo, Japan) using the above automatic liquid chroma-
tographic system with UV monitoring at 220 nm. The entire reaction
N4 and N5 both had a purity of 94% in relative peak area by
HPLC–ELSD analysis. Regarding N4, HPLC-(ESI+) MS: 460 ([M+H]+).
1H NMR (600 MHz, D2O, ppm): δ 0.70–0.82 (m, CH3-20), 1.12–1.37 (m,
CH2-17, CH2-18, CH2-19), 1.94–2.20 (m, CH2-3, CH2-4, CH2-16),
2.74–2.97 (m, CH2-10), 3.54–3.80 (m, CH-2, CH2-8), 4.42–4.55 (m, CH-
6), 5.13–5.23 (m, CH-11), 5.28–5.40 (m, CH-15), 5.48–5.56 (m, CH-
12), 5.57–5.67 (m, CH-14), 6.13–6.21 (m, CH-13), 7.29, 8.36 (NH-CO);
2

T. Wang, et al.
FoodChemistry305(2020)125482
others: 1.42–1.55 (m, CH2-15′), 3.08–3.29 (m, CH2-10′, CH2-10″),
4.20–4.28 (m, CH-6′), 6.76–6.89 (m, CH-13′). 13C NMR (150 MHz, D2O,
ppm): δ 13.34 (C-20), 21.83 (C-19), 25.45 (C-3), 26.09 (C-17), 30.54
(C-4), 31.37 (C-18), 37.22 (C-10), 38.69 (C-16), 43.25 (C-8), 54.05 (C-
2), 55.60 (C-6), 78.96 (C-11), 128.59 (C-14), 128.63 (C-13), 128.75 (C-
12), 133.98 (C-15), 171.65–176.24 (C-1, C-5, C-7, C-9); others: 25.30
(C-13″), 28.00 (C-15′), 33.36 (C-10′), 35.40 (C-14′), 58.30 (C-6′), 64.68
(C-6″), 81.82 (C-12″), 122.34 (C-12′), 149.08 (C-13′).
2.6. Analysis of volatile compounds from degradation of the intermediates
As in the case of the model reactions performed in Section 2.2, the
intermediate compounds (L1, M2, M3, N4, and N5) (0.06 M, each)
dissolved in phosphate buffer were heated at 140 °C for 2 h. Then the
resulting solutions were analyzed by solid-phase microextraction
(SPME) and gas chromatography–mass spectrometry (GC–MS) as fol-
lows.
Regarding N5, HPLC-(ESI+
)
MS: 620 ([M+H]+). 1H NMR
Samples were pre-equilibrated at 50 °C for 10 min in a 15-mL SPME
vial, and then extracted with the fiber (DVB/CAR/PDMS, 2 cm, 50/
30 μm) (Supelco Inc., Bellefonte, PA) at the same temperature for
20 min in headspace mode. A 7890B gas chromatograph coupled with a
(600 MHz, D2O, ppm): 0.68–0.78 (m, CH3-7′), 1.07–1.33 (m, CH2-4′,
CH2-5′, CH2-6′), 1.90–2.13 (m, CH2-3, CH2-3′, CH2-4, CH2-12),
2.68–2.90 (m, CH2-10), 2.92–3.12 (m, CH2-14), 3.14–3.25 (m, CH-11),
3.49–3.80 (m, CH-2, CH2-8, CH-15, CH2-17), 4.35–4.56 (m, CH-6),
5.06–5.16 (m, CH-13), 5.44–5.49 (m, CH-2′), 5.52–5.62 (m, CH-1′),
6.11–8.31 (NH-CO). 13C NMR (150 MHz, D2O, ppm): δ 13.31 (C-7′),
21.90 (C-6′), 26.32 (C-3), 28.05 (C-4′), 30.38 (C-4), 31.40 (C-5′), 35.34
(C-10), 36.42 (C-3′), 38.55 (C-14), 39.77 (C-12), 43.19 (C-8), 43.33 (C-
17), 52.48 (C-11), 53.20 (C-2), 54.06 (C-6), 65.22 (C-13), 67.55 (C-15),
128.81 (C-1′), 135.67 (C-2′), 171.77–176.30 (C-1, C-5, C-7, C-9, C-16,
C-18).
5975C mass spectrometer and
a
DB-Wax capillary column
(30 m × 0.25 mm × 0.25 μm; Agilent Technologies, Santa Clara, CA)
was used in analysis of the volatiles. The fiber was desorbed at 250 °C
for 3 min in splitless mode. The GC oven was initially set at 35 °C for
2 min; raised to 60 °C at 2 °C/min, held for 2 min; and raised to 230 °C at
4 °C/min. Carrier gas was helium at 1.0 mL/min. Temperature of elec-
tron impact ion source (70 eV) was 230 °C. The mass detector was op-
erated at 150 °C. Mass spectra were recorded over the range of m/z
33–450.
Identification of the volatile compounds was performed by search of
NIST2015 mass database, comparison of the RIs (retention indices)
relative to n-alkanes (C5–C26) with those documented in literature, and
co-injection of available authentic chemicals. Amount of the volatile
compounds was expressed as relative peak area (%), which was ob-
tained by normalization of peak areas of the found compounds.
2.4. HPLC–ELSD analysis
An Agilent 1260 HPLC system coupled with a 1260 Infinity eva-
porative light scattering detector (ELSD) (Agilent Technologies, Santa
Clara, CA) was used. The temperature of the ELSD evaporation tube was
40 °C. The flow rate of nitrogen (99.99%) for the evaporation tube was
3.5 mL/min. Chromatographic separation was performed by an Xbridge
Amide column (4.6 mm × 150 mm, 3.5 μm; Waters Co., Milford, MA).
Mobile phase was composed of acetonitrile and ammonium formate
(10 mM; 75:25, v/v), flowing in isocratic elution mode at 1 mL/min.
Column temperature was kept at 40 °C. The injection volume of sample
was 1 μL.
3. Results and discussion
3.1. HPLC–MS analysis
Reaction conditions were used according to our former investiga-
tion on the model systems of GSH and glucose with fat or oxidized fat
(Zhao, Wang, Xie, Xiao, Cheng, et al., 2019). The resulting reaction
mixtures were analyzed by HPLC–MS, while Fig. 1(a–g) showed total
ion current (TIC) chromatograms of the 0.16-h reaction mixtures. As
aforementioned, in the initial reaction of GSH, glucose, and aldehyde,
both glucose and aldehyde can react with the GSH, forming non-volatile
intermediate compounds. By comparison with the control reaction
systems of GSH-glucose (GSH-Glu), GSH-hexanal (GSH-Hex), GSH-(E)-
2-heptenal (GSH-Hept), or GSH-(E,E)-2,4-decadienal (GSH-Dec), it was
seen that formation of G1, G2, G3, or G4 involved glucose, while that of
L1, M1–M3, and N1-N6 involved hexanal, (E)-2-heptenal, and (E,E)-2,4-
decadienal, respectively. The co-eluted peaks (e.g. G1 and M2 in
2.5. HPLC–MS analysis
An LCQ-DECA XP MAX LC–MS system equipped with HPLC and trap
tandem mass spectrometer (Thermo-Electron Company, San Jose, CA)
was used. HPLC conditions used were identical to those described in the
above HPLC–ELSD analysis. Electrospray ionization (ESI) was operated
in a positive mode. Ion source voltage was 4.5 kV. Ion source current
was 80 μA. Capillary temperature was 350 °C. Flow rates of the sheath
gas and auxiliary gas were 60 arb and 20 arb, respectively. Full scan
range of mass spectra was set over m/z 50–2000 while for MS/MS de-
tection the collision energy was set at 24%. Data were acquired with
Xcalibur software system.
Fig. 1(d)) were resolved by extracted ion chromatogram with [M+H]+
,
based on the identified structures of the intermediate compounds
shown in Figs. 2–5. In Fig. S1, the scheme for reaction of GSH, glucose,
and the aldehydes to form the intermediate compounds is presented.
Concentrations of the intermediate compounds in the reaction
mixtures of glucose-GSH-hexanal (Glu-GSH-Hex), glucose-GSH-(E)-2-
heptenal (Glu-GSH-Hept), and glucose-GSH-(E,E)-2,4-decadienal (Glu-
GSH-Dec) were determined according to Clark and Deed (2018), re-
lative to the internal GSH, since authentic substances of most inter-
mediate compounds were unavailable. As shown in Fig. 1(h–j), con-
centrations of all the intermediates first rose and then fell with reaction
time. Regarding the glucose-derived intermediates (G1–G4), the max-
imum concentration occurred earliest for G4, then G3, and finally G2 or
G1 for the reaction systems. According to chemical structures of the
intermediates shown in Fig. 2, this peaking time order could be ex-
plained by that G4 was constructed from the starting materials of GSH
and glucose, whereas G3 was from glucose and the cysteinylglycine
(Cys-Gly) that was hydrolyzed from GSH or from a molecule with a GSH
moiety later, and G1 and G2 were from Cys-Gly and a C3 or C4-frag-
ment split from glucose or from a molecule with a glucose skeleton
later. Following the same principle, as shown in Fig. 1(h–j), L1 (0.25 h)
Concentrations of the intermediate compounds in reaction mixtures
of the Glu-GSH-Hex, Glu-GSH-Hept, and Glu-GSH-Dec systems were
determined as follows. Selected ion monitoring (SIM) mode by mon-
itoring the respective [M+H]+ ion was adopted in MS detection, while
GSH was used as the internal standard. First, HPLC–MS analyzed
samples of 0.16-h reaction mixtures from the three model systems that
were 0, 4, 10, 20, 30, and 100-folds diluted with distilled water, in
order to get response factors (ρ) of the found intermediates vs GSH. The
ρ was calculated by ki / k0, where ki and k0 were slopes measured by
correlating the diluted folds with peak areas of the intermediate com-
pounds and GSH, respectively. Second, calibration factor of GSH (f0)
according to concentration (mmol/L) divided by peak area, was ob-
tained by analysis of the 0.16-h reaction mixtures spiked with a known
amount of GSH. Finally, concentration of the found intermediate
compounds (Ci, mmol/L) in reaction mixtures was calculated by
Ci = (ρ / n) × f0 × Ai, where Ai was peak area of an intermediate
compound in HPLC–MS analysis of the reaction mixtures, and n was
equivalent numbers of cysteines in the molecule of the intermediate
compound.
3

T. Wang, et al.
FoodChemistry305(2020)125482
Fig. 1. (a–g). HPLC–MS total ion current chromatograms of 0.16-h reaction mixtures from both the target and the control reaction systems involving GSH, Glu, and/
or aldehydes (hexanal, (E)-2-heptenal, or (E,E)-2,4-decadienal); and (h–j) trends of concentrations of the detected intermediate compounds with reaction time in the
model systems of GSH, Glu, and the aldehydes (hexanal, (E)-2-heptenal, or (E,E)-2,4-decadienal).
peaked later than G4 in the Glu-GSH-Hex system. M2 or M3 was the
earliest (0.16 h) to arrive at maximum concentration, followed by M1
(0.5 h) for the Glu-GSH-Hept system. For the Glu-GSH-Dec system, N4
or N5 was the earliest (0.16 h) to arrive at maximum concentration,
followed by first N6 (0.25 h), then N1 and N3, and finally N2 (1 h). N2
had the latest peaking time because its formation involved not only Cys-
Gly, but also hexanal that was later derived from the (E,E)-2,4-
decadienal by retro-aldol reaction. On the other side, since GSH tended
to hydrolyze the glutamyl, the earlier peaking intermediates with a GSH
moiety (e.g., M2 and N4) exhibited a fast degradation, as shown in
Fig. 1(h–j). However, the later peaking intermediates (e.g., M1 and N1)
constructed with Cys-Gly decreased slowly along the timeline. Besides,
with sharp degradation of the earlier peaking intermediates (e.g., M2 or
N4), concentration of the later peaking intermediates (e.g., M1 or N1)
Fig. 2. MS/MS spectra and structural elucidation process of the detected glucose-derived intermediate compounds G1–G4. Note. ▲ Elucidation of the ion peak is
shown in the molecule structure below.
4

T. Wang, et al.
FoodChemistry305(2020)125482
Fig. 3. MS/MS spectra and structural elucidation process of the aldehyde-derived intermediate compounds of thiazolidines (L1, M1, and N1) found in Glu-GSH-Hex,
Glu-GSH-Hept, and Glu-GSH-Dec systems, respectively. Note. ▲ Elucidation of the ion peak is shown in the molecule structure below.
in the reaction solution was observed to increase, suggesting degrada-
tion of the former might help formation of the latter.
the carbohydrate chain in collision-induced dissociation (CID) during
MS/MS analysis. Besides, the ions m/z 215 for G1, m/z 245 for G2, and
m/z 323 and 305 for G3 with loss of one or two molecules of water from
the carbohydrate moiety of the respective [M+H]+ were revealed with
a strong peak. The ion m/z 203 due to loss of one molecule of water
from the carbohydrate moiety of the aforementioned m/z 221 ion also
exhibited a considerable abundance in G1, G2, and G3. Otherwise, the
ion of Cys-Gly (178) for G2 or G3 and the ion m/z 227 with loss of three
molecules of water from the [M+H]+ ion for G2 were also detected.
With regards to the Amadori compound G4, similarly, the m/z 452,
434, and 416 ions were attributed to consecutive loss of water from the
carbohydrate moiety of the [M+H]+ (470) (Linetsky, Shipova,
Legrand, & Argirov, 2005). The relatively strong peak of m/z 386 re-
presented a stable immonium ion with removal of three molecules of
water and one molecule of formaldehyde from the carbohydrate moiety
of the [M+H]+ (Jerić & Horvat, 2009; Jerić, Versluis, Horvat, & Heck,
2002). The strong peak of m/z 274 ion might be acquired by loss of Cys-
Gly (178) as well as one molecule of water from the [M+H]+ (470)
(Jerić & Horvat, 2009). The ion m/z 274 through first loss of one mo-
lecule of water and then another, could lead to the m/z 256 ion and the
m/z 238 ion, respectively. The m/z 256 ion with cleavage of one mo-
lecule of formaldehyde from the carbohydrate chain could lead to the
m/z 226 ion.
Formation of the intermediate compounds (L1, M1–M3, and N1–N6)
with GSH or the hydrolyzed Cys-Gly indicated inhibition from the al-
dehydes (hexanal, (E)-2-heptenal, and (E,E)-2,4-decadienal) on the
GSH-glucose reaction. In the Glu-GSH-Hex system, the glucose-derived
intermediates G1–G4 had a concentration higher than the hexanal-de-
rived intermediate L1. However, for the Glu-GSH-Hept or Glu-GSH-Dec
system, the sum of concentrations of the aldehyde-derived inter-
mediates (M1–M3, or N1–N6) was much above that of the glucose-de-
rived intermediates (G1–G4). These suggested that (E)-2-heptenal and
(E,E)-2,4-decadienal exhibited higher inhibition on the GSH-glucose
reaction than hexanal, because of higher reactivity of the unsaturated
aldehydes to react with GSH against glucose. Otherwise, according to
our former research (Cao et al., 2017; Hou et al., 2017; Zhao, Wang,
Xie, Xiao, Du, et al., 2019), except for the Amadori compound G4, the
glucose-derived intermediates G1–G3 should also have inhibited the
GSH-glucose reaction in development of sulfur-containing volatile fla-
vors, because of their relatively stable thiazolidine structures. This was
evidenced from their tardy degradation with time, as obviously seen
from Fig. 1(h).
3.2. Structure identification by MS
As shown in Fig. 3, L1, M1, and N1 were thiazolidine derivatives
generated by hexanal, (E)-2-heptenal, or (E,E)-2,4-decadienal reacted
with both the groups of sulfhydryl and amino of Cys-Gly. Since (E,E)-
2,4-decadienal could transform into hexanal via retro-aldol reaction,
the intermediate compound (named N2 herein) with structure of L1 was
also detected in the reaction systems involving (E,E)-2,4-decadienal.
Moreover, as shown in Fig. 4, M2 and N4 had the chemical structures of
hemiacetals formed by sulfhydryl of the GSH attacked by the carbonyl
of (E)-2-heptenal and (E,E)-2,4-decadienal, respectively. As shown in
Fig. 5, M3, N3, N5, and N6 were complex thiazolidine derivatives
formed from (E)-2-heptenal or (E,E)-2,4-decadienal reacted with one
MS spectra corresponding to the above HPLC–MS analysis and
structural elucidation process were shown in Figs. 2–5. According to
Fig. 2, G1, G2, and G3 were identified to be thiazolidine derivatives
formed from reaction of Cys-Gly with a fragment of glucose (G1, G2) or
glucose itself (G3). G4 was the Amadori compound formed from reac-
tion of glucose and GSH. Among them, the chemical structures of G3
and G4 had been found from a Glu-GSH system by Jerić and Horvat
(2009), but no report exists for those of G1 and G2. In Fig. 2, the ion m/
z 221 due to scission in the carbohydrate chain of the [M+H]+ ion all
had a good abundance for G1, G2, and G3, indicating vulnerability of
5

T. Wang, et al.
FoodChemistry305(2020)125482
Fig. 4. MS/MS spectra and structural elucidation process for the aldehyde-derived intermediate compounds of hemiacetals (M2 and N4) found from Glu-GSH-Hept
system and Glu-GSH-Dec system, respectively. Note. ▲ Elucidation of the ion peak is shown in the molecule structure below.
molecule of glutathione and one molecule of Cys-Gly (M3 and N5), two
molecules of Cys-Gly (N3), or one molecule of glutathione and two
molecules of Cys-Gly (N6). Notably, formation of L1, M1, and N1 in-
volved nucleophilic addition of both the sulfhydryl and the amino
groups towards the aldehyde group, and formation of M2 and N4 in-
volved nucleophilic addition of the sulfhydryl towards the aldehyde
group; all of which created a chiral center and led to a mixture of two
diastereoisomers (Block et al., 2018; Fernandez et al., 2001). In addi-
tion, formation of the complex thiazolidine derivatives M3, N3, N5, and
N6 not only involved the aforementioned classical carbonyl addition,
but also the 1,4-conjugate addition (Michael addition) of sulfhydryl
towards the conjugated carbon–carbon double bond of the α,β-un-
saturated aldehydes, which created another chiral center and led to the
mixture with a doubled number of diastereoisomers (Clark & Deed,
2018).
of L1 and M1, respectively, were detected. The resulted high abundance
with the m/z 244 and 256 ions might be due to the presence of a re-
latively stable structure of conjugated amide in the two ions. Moreover,
the ions m/z 158 for L1, m/z 170 for M1, and m/z 210 for N1 could all
be acquired by cleavage of the amide chain from the respective [M
+H]+ ion. The ion m/z 189 for M1 or N1 could be acquired by removal
of the alkene chain from the respective [M+H]+ ion.
As shown in Fig. 4, either M2 or N4 was characterized by a strong
peak of [M+H−H2O]+ because of the hydroxyl group of hemiacetal.
However, the weak peak of [M+H−H2O]+ for M3, N5, and N6 (shown
in Fig. 5) was ascribed to the presence of the GSH moiety, since GSH
also had such a weak peak of [M+H−H2O]+ during MS detection
(data not shown). Overall, the hemiacetals (M2 and N4) and the com-
plex thiazolidines (M3, N3, N5, and N6) mainly had the following types
of scission in MS/MS analysis: cleavage of the glutamyl (e.g., m/z 291
for M2, 331 for N4, 451 for M3, 491 for N5, and 669 for N6), reversible
loss of the Cys-Gly (e.g., m/z 313 for N3 and m/z 620 for N6), reversible
loss of the GSH (e.g., m/z 273 for M3, 313 for N5, and 491 for N6), or
reversible loss of both the Cys-Gly and the GSH (e.g., m/z 313 for N6).
Consequently, in addition to the ions of GSH (m/z 308) and Cys-Gly (m/
Different from G1, G2, and G3, L1, M1, and N1 all had a strong peak
for the [M+H]+ ion (e.g., m/z 261 for L1) (see Fig. 3), due to presence
of a hydrocarbon chain in these thiazolidine derivatives, instead of that
of a carbohydrate chain. Besides, another strong peak of m/z 244 for L1
and of m/z 256 for M1 with elimination of NH3 from the [M+H]+ ions
6

T. Wang, et al.
FoodChemistry305(2020)125482
Fig. 5. MS/MS spectra and structural elucidation process of the aldehyde-derived intermediate compounds of complex thiazolidines (M3, N3, N5, and N6) found
from Glu-GSH-Hept system and Glu-GSH-Dec system, respectively. Note. ▲ Elucidation of the ion peak is shown in the molecule structure below.
z 179), the fragment ions with a structure of M1 (m/z 273) or N1 (m/z
313) were also detected from the hemiacetals (M2 and N4) and the
complex thiazolidines (M3, N3, N5, and N6). Regarding the hemiacetals
(M2 and N4), possibly the ions m/z 291 for M2 and m/z 331 for N4 (see
Fig. 4) with cleavage of the glutamyl from the [M+H]+ ions went
through such a reaction, i.e., the hemiacetal carbon was attacked by the
amino group, followed by elimination of water, resulting in ions with
the structures of M1 (m/z 273) and N1 (m/z 313), respectively.
Michael addition reaction of the sulfhydryl group to the olefinic bond of
(E)-2-heptenal (Starkenmann, 2003). For N5, the chemical shifts (ppm)
of δC 128.81/δH 5.57 (C-1′) and δC 135.67/δH 5.47 (C-2′) corresponded
to an olefinic bond. The chemical shifts (ppm) of δC 52.48/δH 3.16 (C-
11) revealed that the sulfhydryl group was added at the unsaturated
carbon at the β-position to the aldehyde group of (E,E)-2,4-decadienal,
while those of δC 39.77/δH 2.03 (C-12) corresponded to the saturated
carbon with addition of hydrogen to the α-position unsaturated carbon
of the olefinic bond (Starkenmann, 2003).
Regarding M2, chemical shifts (ppm) of δC 125.82 (C-12) and
129.14 (C-13) indicated the presence of an olefinic bond. Regarding N4,
chemical shifts (ppm) of δC 128.75/δH 5.53 (C-12), δC 128.63/δH 6.18
(C-13), δC 128.59/δH 5.63 (C-14), and δC 133.98/δH 5.33 (C-15) in-
dicated the presence of a conjugated carbon–carbon double bond (Zhao
et al., 2018). Due to disappearance of the aldehyde group, the above
values of olefinic bond chemical shifts for M2 and N4 were obviously
smaller than those of the original (E)-2-heptenal or (E,E)-2,4-decadienal
(data not shown). Moreover, the assignment of δC 89.26/δH 5.04 ppm
(C-11) for M2 and δC 78.96/δH 5.18 ppm (C-11) for N4, corresponding
to a hetero hemiacetal with S atom and O atom, agreed with the lit-
erature (Block et al., 2018). Otherwise, it seemed that tautomerization
might occur for the hemiacetals in D2O. For M2 (see Fig. S5), signals
(ppm) of δC 206.38 (C-12′), δC 57.52 (C-6′), δC 39.51 (C-13′), δC 38.59/
δH 3.17 (C-11′), δC 28.11/δH 3.02 (C-10′), and δC 24.66/δH 1.52 (C-14′)
might be ascribed to the tautomer with a ketone group from the
hemiacetal. For N4 (see Fig. S6), signals (ppm) of δC 149.08/δH 6.79 (C-
13′), δC 122.34 (C-12′), δC 58.30/δH 4.23 (C-6′), δC 35.40 (C-14′), δC
33.36/δH 3.19 (C-10′), and δC 28.00/δH 1.50 (C-15′) might be ascribed
to the structure of tautomer (i) by transfer of hydrogen proton of the
hydroxy group of the hemiacetal from C-11 to C-15. Also for N4 (see
Fig. S6), signals (ppm) of δC 81.82 (C-12″), δC 64.68 (C-6″), δC 25.30 (C-
13″), and δH 3.19 (C-10″) might be ascribed to the structure of tautomer
(ii), due to transfer of hydrogen proton of the hydroxy group of the
hemiacetal from C-11 to C-13. Anyway, the aforementioned tauto-
merization needs to be further proved, since some NMR signals of the
tautomers were not discerned, as their peaks could be overlapped with
3.3. Structure identification by NMR
Preparation of the aldehyde-derived intermediates was performed
by separation of GSH-aldehyde reaction mixtures, in order to further
identify structures of the found intermediates by NMR and to subse-
quently investigate roles of the intermediate compounds in develop-
ment of volatile flavors. As a result, the substances of L1, M2, M3, N4,
and N5 with identities as those in the above HPLC–MS analysis were
obtained with a high purity by HPLC analysis using the universal de-
tection of ELSD.
The processed NMR data were presented in Section 2.3, while the
NMR spectra are shown in Figs. S2–S6. Comparing with those of the
original aldehydes (hexanal, (E)-2-heptenal, and (E,E)-2,4-decadienal)
and those of the GSH, 1H NMR and 13C NMR data of the prepared in-
termediate compounds revealed most information from the skeletons of
the original aldehydes or the GSH, which were not discussed in the
following. Besides, there were signals of trace impurities or environ-
mental contaminants found in the NMR spectra. However, obvious
signals of the aldehyde group or the sulfhydryl group were not detected
for the intermediate compounds. With regards to L1, chemical shifts
(ppm) of δC 64.91/δH 4.64 (C-6), δC 35.37/δH 3.16 (C-5), and δC 69.84/
δH 3.74 (C-4) corresponding to the thiazolidine ring were consistent
with the literature (Corbi et al., 2008; Fernandez et al., 2001;
Starkenmann, 2003); likewise for M3 and N5. Moreover, for M3, both
1H NMR and 13C NMR signals did not reveal any information of an
olefinic bond, whereas chemical shifts (ppm) of δC 44.45/δH 2.63 (C-
11) and δC 38.68/δH 2.03 (C-12) indicated the saturated carbons after
7

T. Wang, et al.
FoodChemistry305(2020)125482
Fig. 6. Possible pathways for (a) formation of (E,E)-2,4-dienals and acetone by radical oxidation of the (E)-2-alkenals. (b) formation of some new aliphatic com-
pounds from (E)-2-heptenal via retro-aldol condensation, oxidation, reduction, etc. (c) formation of some new aliphatic compounds in particular (E)-3-nonen-2-one
(No. 1) and 2-hexanoylfuran (No. 2) from (E,E)-2,4-decadienal via retro-aldol condensation, oxidation, reduction, etc. (d) formation of the dialkyl thiophenes,
namely, 2,5-dipropylthiophene (No. 4), 2-ethyl-5-propylthiophene (No. 5), 2-sec-butyl-5-(1-methylbutyl)thiophene (No. 8), 2-butyl-5-ethylthiophene (No. 9), and 2-
ethyl-5-(1-methylbutyl)thiophene (No. 10) by H2S reacting with the resulting aliphatic products, especially the new aldehydes from (E)-2-heptenal or (E,E)-2,4-
decadienal, as shown in (a–c).
those of the hemiacetals.
configuration of
a di-substituted carbon–carbon double bond.
Above all, according to the empirically estimated values, chemical
shifts (ppm) of the allylic carbons by 13C NMR, i.e., 89.26 (C-11) and
34.12 (C-14) for M2, 78.96 (C-11) and 38.69 (C-16) for N4, and 52.48
(C-11) and 36.42 (C-3′) for N5 were consistent with those of the (E)-
Therefore, probably M2, N4, and N5 kept the (E)-configuration as the
original used (E)-2-heptenal or (E,E)-2,4-decadienal.
8

T. Wang, et al.
FoodChemistry305(2020)125482
3.4. Volatile flavor compounds formed from degradation of the intermediate
compounds
acids, or the GSH, the aforementioned reaction of (E,E)-2,4-heptadienal
and (E,E)-2,4-decadienal could lead to 2-ethylpyridine and 2-pentyl-
pyridine, respectively (Mottram, 1998; Zhang & Ho, 1989).
Solutions of the prepared intermediate compounds (L1, M2, M3, N4,
and N5) were heated as the model reaction systems for 2 h, as they were
observed able to be completely degraded in 2 h, as shown in Fig. 1(h–j).
In fact, GC–MS detected more than a hundred chromatographic peaks
in each heated solution; it was found that identifications between M2
and M3 solutions or between N4 and N5 solutions were almost the
same, as shown in Table S1. The original aldehydes of hexanal, (E)-2-
heptenal, and (E,E)-2,4-decadienal or the (E)-2-heptenoic acid oxidized
from (E)-2-heptenal were revealed with the highest level in the re-
spective solution. This indicated that during heating, these intermediate
compounds were reversibly reacted to release the original reactants.
Fig. 6 shows possible formation pathways of some identified com-
pounds. Overall, the released aldehydes (hexanal, (E)-2-heptenal or
(E,E)-2,4-decadienal) undertook retro-aldol or aldol condensation,
oxidation, cyclization, and reduction, etc., leading to the various new
aldehydes, ketones, alcohols, furans, etc., listed in Table S1. For ex-
ample, aldol condensation of hexanal could account for the major
compound 2-butyl-(E)-2-octenal found in L1 solution. Regarding the
high level of 3-nonen-2-one in N4 and N5 solutions, it was probably
formed from aldol condensation of hexanal and acetone, as shown in
Fig. 6(c) for compound No. 1. Among them, hexanal was formed from
(E,E)-2,4-decadienal by retro-aldol reactions, while acetone was from
(E,E)-2,4-decadienal through a series of reactions including retro-aldol,
oxidation, β-scission, tautomerization, and reduction, as shown in
Fig. 6(c) and (a). Moreover, 2-pentyl-2-cyclopenten-1-one was formed
from cyclization of (E,E)-2,4-decadienal (Adams, Kitrytė, Venskutonis,
& De Kimpe, 2011). The high level of 2-hexanoylfuran (No. 2) could be
formed from (E,E)-2,4-decadienal through alkene epoxidation and ring
opening, cyclization, loss of water, and oxidization. The identified 2-
propylfuran, 2-butylfuran, and 2-pentylpyran/2-hexylfuran from M2,
M3, N4, or N5 solution could result from cyclization of (E,E)-2,4-hep-
tadienal, (E,E)-2,4-octadienal, and (E,E)-2,4-decadienal, in turn
(Adams, Bouckaert, Van Lancker, De Meulenaer, & De Kimpe, 2011;
Mottram, 1998). Among them, (E,E)-2,4-heptadienal and (E,E)-2,4-oc-
tadienal (No. 3, see Fig. 6(c)) were oxidized from (E)-2-heptenal and
(E)-2-octenal, respectively, according to the proposed Way (ii) reactions
in Fig. 6(a); while the (E)-2-octenal could be formed from (E,E)-2,4-
decadienal via retro-aldol reaction, as shown in Fig. 6(c).
Particularly, six dialkylthiophenes (see Table S1) were found at a
high or considerable level, namely, 2,5-dipropylthiophene, 2-ethyl-5-
propylthiophene, 2-butyl-5-propylthiophene, 2-butyl-5-ethylthiophene,
and 2-sec-butyl-5-(1-methylbutyl)thiophene in M2 or M3 solution, and
2-ethyl-5-(1-methylbutyl)thiophene in N4 or N5 solution. As shown in
Fig. 6(d), regarding the dialkylthiophenes found in M2 or M3 solution,
possibly, 1-propanol and ethanol reacting with the aforementioned 2-
propylthiophene via electrophilic substitution reaction formed 2,5-di-
propylthiophene (No. 4) and 2-ethyl-5-propylthiophene (No. 5), re-
spectively. 2-Butenal (No. 6, in Fig. 6(b)) reacting with acetaldehyde
via aldol condensation followed by elimination of water formed (E,E)-
2,4-hexadienal, which subsequently reacted with H2S to form 2-ethyl-
thiophene (No. 7, in Fig. 6(d)). The 2-ethylthiophene reacting with
ethanol, acetaldehyde, and acrolein followed by reduction led to 2-sec-
butyl-5-(1-methylbutyl)thiophene (No. 8). 1-Butanol reacting with 2-
ethylthiophene via electrophilic substitution reaction formed 2-butyl-5-
ethylthiophene (No. 9). Regarding the 2-ethyl-5-(1-methylbutyl)thio-
phene (No. 10) in N4 or N5 solution, as shown in Fig. 6(d), it was
proposed that first 2-ethylthiophene was formed, then electrophilic
substitution by ethanol, addition of acrolein, and reduction occurred.
However, herein, the (E,E)-2,4-hexadienal in formation of 2-ethylthio-
phene might be originated from the (E,E)-2,4-decadienal by two-times
retro-aldol reactions and two-times Way (ii) reactions, as shown in
Fig. 6(c) for compound No. 11, while the Way (ii) reactions were shown
in Fig. 6(a).
4. Conclusions
Initial reaction intermediate compounds were identified from the
model systems composed of glucose, GSH, and hexanal, (E)-2-heptenal,
or (E,E)-2,4-decadienal by HPLC-MS as well as NMR. Except for the
Amadori compounds and hemiacetals, all the others had relatively
stable structures of thiazolidines. In addition, degradation of the in-
termediates with a GSH moiety by removal of the glutamyl contributed
to formation of those relatively stable thiazolidines. The unsaturated
aldehydes (E)-2-heptenal and (E,E)-2,4-decadienal exhibited much
stronger reactivity against glucose to react with GSH, which resulted in
the sum of concentrations of intermediate compounds being over more
than the glucose-derived intermediates or above the glucose-derived
intermediates. During heating, the aldehyde-derived intermediates
mainly released the original aldehydes, which induced formation of
various new aliphatic products through retro-aldol or aldol condensa-
tion, oxidation, cyclization, and reduction, etc. Both the original alde-
hydes and the new aliphatic compounds could react with H2S or NH3 to
form alkyl sulfur-containing or nitrogen-containing compounds.
Particularly, formation pathways of 3-nonen-2-one, 2-hexanoylfuran,
and six 2,5-dialkylthiophenes, found at high or considerable levels in
the heated solutions, were proposed for the first time. The work is
helpful for gaining insight into formation of meat flavor in processing of
meat or meat-like foods.
Like our previous findings from GSH-glucose with fat or oxidized fat
model reaction systems (Zhao, Wang, Xie, Xiao, Cheng, et al., 2019), as
shown in Table S1, several alkyl sulfur- or nitrogen-containing com-
pounds (e.g., 2-hexylthiophene and 2-ethylpyridine) were found. Worth
mentioning, these listed alkyl compounds were also detected in the
corresponding model systems of GSH and aldehyde (hexanal, (E)-2-
heptenal or (E,E)-2,4-decadienal) with or without glucose (data not
shown). This indicated that these model reaction systems passed in the
same way via these intermediates to form the volatile flavors. During
heating, thermal degradation of GSH in particular the Strecker de-
gradation of cysteine from GSH induced by the lipid degradation pro-
ducts (e.g. (E)-2-heptenal) can release H2S and NH3 (Gallardo et al.,
2008; Hidalgo & Zamora, 2016). The identified 1-hexanethiol in L1
solution could be formed from reaction of 1-hexanol and H2S, while
reduction of hexanal could lead to the 1-hexanol (Zhao, Wang, Xie,
Xiao, Cheng, et al., 2019). (E,E)-2,4-Heptadienal and (E,E)-2,4-dec-
adienal via epoxidation, addition of H2S, and oxidization could form the
1-(2-thienyl)-1-propanone in M2 and M3 solutions and the 1-(2-
thienyl)-1-hexanone in N4 and N5 solutions, respectively (Zhao, Wang,
Xie, Xiao, Cheng, et al., 2019). 1-(2-Pyridinyl)-1-pentanone could be
generated in the same way as 1-(2-thienyl)-1-hexanone, with NH3 in-
volved instead of H2S. (E,E)-2,4-Heptadienal, (E,E)-2,4-octadienal, and
(E,E)-2,4-decadienal reacting with H2S could form the 2-propylthio-
phene, 2-butylthiophene, and 2-pentylthiopyran/2-hexylthiophene, in
turn (Elmore & Mottram, 2000). But in the presence of NH3, amino
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.foodchem.2019.125482.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
The present work was supported by the National Natural Science
Foundation of China (31671895), Beijing Municipal Natural Science
9

T. Wang, et al.
FoodChemistry305(2020)125482
Foundation (6172004), and National Key Research and Development
Program of China (2017YFD0400106).
doi.org/10.1080/10408398.2012.761173.
Hou, L., Xie, J., Zhao, J., Zhao, M., Fan, M., Xiao, Q., ... Chen, F. (2017). Roles of different
initial Maillard intermediates and pathways in meat flavor formation for cysteine-
xylose-glycine model reaction systems. Food Chemistry, 232, 135–144. https://doi.
org/10.1016/j.foodchem.2017.03.133.
Compliance with ethical standards
Jerić, I., & Horvat, Š. (2009). Screening for glucose-triggered modifications of glu-
tathione. Journal of Peptide Science, 15(8), 540–547. https://doi.org/10.1002/psc.
1159.
This article does not contain any studies with human participants or
animals performed by any of the authors.
Jerić, I., Versluis, C., Horvat, Š., & Heck, A. J. R. (2002). Tracing glycoprotein structures:
Electron ionization tandem mass spectrometric analysis of sugar-peptide adducts.
Journal of Mass Spectrometry, 37(8), 803–811. https://doi.org/10.1002/jms.337.
Lee, S. M., Jo, Y.-J., & Kim, Y.-S. (2010). Investigation of the aroma-active compounds
formed in the Maillard reaction between glutathione and reducing sugars. Journal of
Agricultural and Food Chemistry, 58(5), 3116–3124. https://doi.org/10.1021/
jf9043327.
References
Adams, A., Bouckaert, C., Van Lancker, F., De Meulenaer, B., & De Kimpe, N. (2011).
Amino acid catalysis of 2-alkylfuran formation from lipid oxidation-derived α,β-un-
saturated aldehydes. Journal of Agricultural and Food Chemistry, 59(20),
11058–11062. doi:https://doi.org/10.1021/jf202448v.
Linetsky, M. D., Shipova, E. V., Legrand, R. D., & Argirov, O. O. (2005). Glucose-derived
Amadori compounds of glutathione. Biochimica et Biophysica Acta, 1724(1–2),
181–193. https://doi.org/10.1016/j.bbagen.2005.04.003.
Adams, A., Kitrytė, V., Venskutonis, R., & De Kimpe, N. (2011). Model studies on the
pattern of volatiles generated in mixtures of amino acids, lipid-oxidation-derived
aldehydes, and glucose. Journal of Agricultural and Food Chemistry, 59(4), 1449–1456.
https://doi.org/10.1021/jf104091p.
Liu, P., Zhang, X., Huang, M., Song, S., & Nsor-Atindana, J. (2012). Formation mechanism
of cross-linking Maillard compounds in peptide-xylose systems. Journal of Peptide
Science, 18(10), 626–634. https://doi.org/10.1002/psc.2443.
Mottram, D. S. (1998). Flavour formation in meat and meat products: A review. Food
Chemistry, 62(4), 415–424. https://doi.org/10.1016/S0308-8146(98)00076-4.
Pan, K.-L., Huang, W.-J., Hsu, M.-H., Lee, H.-L., Liu, H.-J., Cheng, C.-W., Tsai, M.-H.,
Shen, M.-Y., & Lin, P. (2014). Identification of trans,trans-2,4-decadienal metabolites
in mouse and human cells using liquid chromatography-mass spectrometry. Chemical
Research in Toxicology, 27(10), 1707–1719. doi:https://doi.org/10.1021/
tx500199b.
Block, E., Dethier, B., Bechand, B., Cotelesage, J. J. H., George, G. N., Goto, K., ... Vogt, L.
(2018). Ajothiolanes: 3,4-Dimethylthiolane natural products from garlic (Allium sa-
tivum). Journal of Agricultural and Food Chemistry, 66(39), 10193–10204. https://doi.
org/10.1021/acs.jafc.8b03638.
Cao, C., Xie, J., Hou, L., Zhao, J., Chen, F., Xiao, Q., ... Fan, M. (2017). Effect of glycine on
reaction of cysteine-xylose: Insights on initial Maillard stage intermediates to develop
meat flavor. Food Research International, 99, 444–453. https://doi.org/10.1016/j.
foodres.2017.06.012.
Schumacher, F., Neuber, C., Finke, H., Nieschalke, K., Baesler, J., Gulbins, E., & Kleuser,
B. (2017). The sphingosine 1-phosphate breakdown product, (2E)-hexadecenal, forms
protein adducts and glutathione conjugates in vitro. Journal of Lipid Research, 58(8),
1648–1660. https://doi.org/10.1194/jlr.M076562.
Clark, A. C., & Deed, R. C. (2018). The chemical reaction of glutathione and trans-2-
hexenal in grape juice media to form wine aroma precursors: The impact of pH,
temperature, and sulfur dioxide. Journal of Agricultural and Food Chemistry, 66(5),
1214–1221. https://doi.org/10.1021/acs.jafc.7b04991.
Starkenmann, C. (2003). Analysis of a model reaction system containing cysteine and (E)-
2-methyl-2-butenal, (E)-2-hexenal, or mesityl oxide. Journal of Agricultural and Food
Chemistry, 51(24), 7146–7155. https://doi.org/10.1021/jf0304268.
Corbi, P. P., Andrade, F. C., Massabni, A. C., Heinrich, T. A., Souza, P. P. C., & Costa-Neto,
C. M. (2008). Lithium thiazolidine-4-carboxylate: Synthesis, spectroscopic char-
acterization and preliminary in vitro cytotoxic studies in human HeLa cells.
Spectrochimica Acta Part A, 71(3), 929–931. https://doi.org/10.1016/j.saa.2008.02.
013.
Thibon, C., Böcker, C., Shinkaruk, S., Moine, V., Darriet, P., & Dubourdieu, D. (2016).
Identification of S-3-(hexanal)-glutathione and its bisulfite adduct in grape juice from
Vitis vinifera L. cv. Sauvignon blanc as new potential precursors of 3SH. Food
Chemistry, 199, 711–719. https://doi.org/10.1016/j.foodchem.2015.12.069.
Yang, Z., Xie, J., Zhang, L., Du, R., Cao, C., Wang, M., ... Sun, B. (2015). Aromatic effect of
fat and oxidized fat on a meat-like model reaction system of cysteine and glucose.
Flavour and Fragrance Journal, 30(4), 320–329. https://doi.org/10.1002/ffj.3248.
Zhang, Y., & Ho, C.-T. (1989). Volatile compounds formed from thermal interaction of
2,4-decadienal with cysteine and glutathione. Journal of Agricultural and Food
Chemistry, 37(4), 1016–1020. https://doi.org/10.1021/jf00088a044.
Zhao, J., Wang, T., Xie, J., Xiao, Q., Cheng, J., Chen, F., ... Sun, B. (2019a). Formation
mechanism of aroma compounds in a glutathione-glucose reaction with fat or oxi-
dized fat. Food Chemistry, 270(1), 436–444. https://doi.org/10.1016/j.foodchem.
2018.07.106.
El-massry, K., Farouk, A., & El-Ghorab, A. (2003). Volatile constituents of glutathione-
ribose model system and its antioxidant activity. Amino Acids, 24(1–2), 171–177.
https://doi.org/10.1007/s00726-002-0318-4.
Elmore, J. S., Campo, M. M., Enser, M., & Mottram, D. S. (2002). Effect of lipid compo-
sition on meat-like model systems containing cysteine, ribose, and polyunsaturated
fatty acids. Journal of Agricultural and Food Chemistry, 50(5), 1126–1132. https://doi.
org/10.1021/jf0108718.
Elmore, J. S., & Mottram, D. S. (2000). Formation of 2-alkyl-(2H)-thiapyrans and 2-al-
kylthiophenes in cooked beef and lamb. Journal of Agricultural and Food Chemistry,
48(6), 2420–2424. https://doi.org/10.1021/jf000124v.
Farmer, L. J., & Mottram, D. S. (1990). Interaction of lipid in the Maillard reaction be-
tween cysteine and ribose: The effect of a triglyceride and three phospholipids on the
volatile products. Journal of the Science of Food and Agriculture, 53(4), 505–525.
https://doi.org/10.1002/jsfa.2740530409.
Zhao, J., Wang, T., Xie, J., Xiao, Q., Du, W., Wang, Y., ... Wang, S. (2019b). Meat flavor
generation from different composition patterns of initial Maillard stage intermediates
formed in heated cysteine-xylose-glycine reaction systems. Food Chemistry, 274,
79–88. https://doi.org/10.1016/j.foodchem.2018.08.096.
Fernandez, X., Duñach, E., Fellous, R., Lizzani-Cuvelier, L., Loiseau, M., Dompe, V., ...
Schippa, C. (2001). Identification of thiazolidines in guava: Stereochemical studies.
Flavour and Fragrance Journal, 16(4), 274–280. https://doi.org/10.1002/ffj.996.
Gallardo, E., de Schutter, D. P., Zamora, R., Derdelinckx, G., Delvaux, F. R., & Hidalgo, F.
J. (2008). Influence of lipids in the generation of phenylacetaldehyde in wort-related
model systems. Journal of Agricultural and Food Chemistry, 56(9), 3155–3159. https://
doi.org/10.1021/jf800094k.
Zhao, J.-Y., Ding, J.-H., Li, Z.-H., Feng, T., Zhang, H.-B., & Liu, J.-K. (2018). Two new
polyols from cultures of the fungus Daedaleopsis tricolor. Natural Product Research,
32(9), 1076–1079. https://doi.org/10.1080/14786419.2017.1380013.
Zhou, R., Grant, J., Goldberg, E. M., Ryland, D., & Aliani, M. (2018). Investigation of low
molecular weight peptides (< 1 kDa) in chicken meat and their contribution to meat
flavor formation. Journal of the Science of Food and Agriculture, 99(4), 1728–1739.
https://doi.org/10.1002/jsfa.9362.
Hidalgo, F. J., & Zamora, R. (2016). Amino acid degradations produced by lipid oxidation
products. Critical Reviews in Food Science and Nutrition, 56(8), 1242–1252. https://
10