Structures, sensory activity, and dose/response functions of 2,5-diketopiperazines in roasted cocoa nibs (Theobroma cacao)

Article abstract of DOI:10.1021/jf051313m

The taste compounds inducing the blood-like, metallic bitter taste sensation reported recently for a dichloromethane extract prepared from roasted cocoa nibs were identified as a series of 25 diketopiperazines by means of HPLC degustation, LC-MS/MS, and i

Full text of DOI:10.1021/jf051313m

7222 J. Agric. Food Chem. 2005, 53, 7222−7231
Structures, Sensory Activity, and Dose/Response Functions of
2,5-Diketopiperazines in Roasted Cocoa Nibs
(Theobroma cacao)
†
TIMO STARK AND THOMAS HOFMANN*,‡
Deutsche Forschungsanstalt fu¨r Lebensmittelchemie, Lichtenbergstrasse 4,
D-85748 Garching, Germany, and Institut fu¨r Lebensmittelchemie, Universita¨t Mu¨nster,
Corrensstrasse 45, D-48149 Mu¨nster, Germany
The taste compounds inducing the blood-like, metallic bitter taste sensation reported recently for a
dichloromethane extract prepared from roasted cocoa nibs were identified as a series of 25
diketopiperazines by means of HPLC degustation, LC-MS/MS, and independent synthesis. Among
these 25 compounds, 13 cis-configured diketopiperazines, namely, cyclo(L-IIe-L-Phe), cyclo(L-Val-
L-Leu), cyclo(L-Pro-L-Pro), cyclo(L-IIe-L-Pro), cyclo(L-Val-L-Tyr), cyclo(L-Ala-L-Tyr), cyclo(L-Phe-L-Ser),
cyclo(L-Ala-L-IIe), cyclo(L-Leu-L-Phe), cyclo(L-Pro-L-Val), cyclo(L-Pro-L-Thr), cyclo(L-Pro-L-Tyr), and
cyclo(L-Val-L-Val) were identified for the first time in cocoa. In addition, the taste recognition thresholds
for the metallic as well as the bitter taste of the diketopiperazines were determined, and after
quantitative analysis by using two diastereomeric diketopiperazines as the internal standards, the
sensory impact of the diketopiperazines was evaluated on the basis of their dose-over-threshold
(DoT) factors calculated as the ratio of the concentration and the threshold concentration of a
compound. These data revealed DoT factors above 1.0 exclusively for cis-cyclo(L-Pro-L-Val), cis-
cyclo(L-Val-L-Leu), cis-cyclo(L-Ala-L-Ile), cis-cyclo(L-Ala-L-Leu), and cis-cyclo(L-Ile-L-Pro), whereas all
of the other diketopiperazines were present below their individual bitter taste threshold concentrations
and should therefore not contribute to the cocoa taste. Because the DoT factors do not consider the
nonlinear relationship between the concentration and gustatory response of an individual compound,
we, for the first time, report on the recording of dose/response functions describing the human bitter
taste perception of diketopiperazines more precisely.
KEYWORDS: Cocoa; bitter taste; metallic taste; cis-cyclo(L-Pro-L-Val); diketopiperazines; threshold
concentrations; dose/response functions
INTRODUCTION
cocoa (3). Among these dilactams, cis-cyclo(Ala-Gly) and cis-
cyclo(Ala-Val) have been found as the predominating dike-
topiperazines in roasted cocoa samples (3). In contradiction to
these findings, recent studies identified cis-cyclo(Pro-Gly) as
the major diketopiperazine in cocoa (4). Although the dike-
topiperazines cis-cyclo(Val-Phe) and cis-cyclo(Ala-Phe) have
been detected in roasted cocoa only in trace amounts, model
experiments demonstrated that these diketopiperazines, when
mixed with theobromine, induce a bitter taste sensation similar
to that perceived from an aqueous suspension of cocoa powder
(3). High-performance liquid chromatography in combination
with mass spectrometry using a moving belt interface was used
to monitor diketopiperazines in cocoa powders (5), but a
comparison of quantitative data and the sensorially perceived
bitter intensity of the products led to the conclusion that further
systematic studies are necessary to understand the typical taste
of roasted cocoa on a molecular level (4).
Besides the pleasant aroma, the sensory quality of the
fermented and roasted seeds of Theobroma cacoa is mainly
driven by its alluring gustatory profile centering around the
typical bitter taste, a slight sour note, as well as its long-lasting
astringent and mouth-drying sensation. These enhance the
complexity and palate length of cocoa beverages and chocolate
confectionary.
Because of its low detection threshold of 10 mg/L in water
(1), the methylxanthine theobromine, which is present in cocoa
beans in concentrations between 1.8 and 3.8 g/100 g of dry
weight (2), is believed to contribute to the typical bitter taste.
In addition, some diketopiperazines, generated during the
roasting of the fermented cocoa beans from hydrophobic amino
acids, are believed to contribute to the bitter taste of roasted
* To whom correspondence should be addressed. Telephone: +49-251-
83-33-391. Fax: +49-251-83-33-396. E-mail: thomas.hofmann@
uni-muenster.de.
With the aim at investigating the key molecules imparting
the typical bitterness and astringent taste sensation of roasted
cocoa, very recently, an aqueous acetone extract prepared from
† Deutsche Forschungsanstalt fu¨r Lebensmittelchemie.
‡ Institut fu¨r Lebensmittelchemie.
10.1021/jf051313m CCC: $30.25
© 2005 American Chemical Society
Published on Web 08/09/2005

Metallic- and Bitter-Tasting Diketopiperazines in Roasted Cocoa Nibs
J. Agric. Food Chem., Vol. 53, No. 18, 2005 7223
velvety astringent, mouth-drying oral sensation, respectively, and iron-
(II) gluconate (0.03 mmol/L) for metallic taste. The assessors had
participated earlier at regular intervals for at least 2 years in sensory
experiments and were, therefore, familiar with the techniques applied.
Sensory analyses were performed in a sensory panel room at 22-25
°C in three different sessions.
Determination of Taste Threshold Concentrations. Taste recognition
thresholds, which means the concentrations at which the typical taste
qualities of the compounds were just detectable, were determined in
bottled water by means of a triangle test as reported recently (6). The
values between individuals and between three separate sessions differed
by not more than plus or minus one dilution step; that is, a threshold
value of 40 µmol/L for the metallic taste of cis-cyclo(L-Leu-L-Phe)
represents a range from 20 to 80 µmol/L.
Recording of Dose/Response Functions. Serial 1:1 dilutions of the
samples in water were prepared starting at the solution of maximum
solubility of each individual taste compound and ending at the
concentration level two steps below the individual recognition threshold
concentration. To fit the dose/response functions into a five-point
intensity scale, first, the taste intensity of the individual compounds
were compared at the highest concentration level by means of the
recently reported half-tongue tasting method (6, 8), thus offering a direct
comparison of the taste impact and a reliable evaluation of the gustatory
response of different compounds. To achieve this, the solutions of the
individual compounds were applied in binary combinations to one side
of the tongue and the panelist were asked to determine which side
showed the stronger sensation. On a five-point scale, the compound
evaluated with the highest taste intensity at its maximum concentration,
that is, cis-cyclo(L-Val-L-Pro) at its maximum solubility of 81.92
mmol/L (water) in the present study, was set to the maximum score of
5.0. After the taste intensity of each compound at its maximum solubility
had been rated, the taste intensities of the other dilutions were
determined by using the half-tongue tasting method so that one dilution
of an individual compound was rated against the intensity of another
dilution of the same compound and by cross-checking the taste intensity
between the different compounds at the same dilution. The dose/
response functions of three individual sessions were averaged. The
intensity values between trained individuals and separate sessions did
not differ more than (0.3 units.
powdered roasted cocoa nibs and reflecting the typical taste
profile of roasted cocoa was sequentially extracted with dichlo-
romethane and ethyl acetate and, after removing the solvents
in high vacuum, the dichloromethane solubles, the ethyl acetate
extractables, as well as the water solubles were evaluated
sensorially in aqueous solution by means of a comparative taste
profile analysis (6). The highest scores for bitter taste, astrin-
gency, and sour taste were found for the water solubles as well
as the ethyl acetate extractables, whereas the dichloromethane
solubles exhibited a metallic-like bitter taste perceived with high
taste intensity (6). Activity-guided analysis of the key taste
contributors in the water as well as the ethyl acetate fraction
revealed that, besides the theobromine and caffeine, the flavan-
3-ols epicatechin, catechin, procyanidin B-2, procyanidin B-5,
procyanidin C-1, [epicatechin-(4â f 8)]3-epicatechin, and
[epicatechin-(4â f 8)]4-epicatechin were among the key
compounds contributing to the bitter taste as well as the
astringent mouth feel. In addition, a series of quercetin,
naringenin, luteolin, and apigenin glycopyranosides imparting
a smooth, velvety type of astringent sensation as well as a family
of N-phenylpropenoyl amino acids such as (+)-N-[3′,4′-dihy-
droxy-(E)-cinnamoyl]-L-aspartic acid and (-)-N-[3′,4′-dihy-
droxy-(E)-cinnamoyl]-3-hydroxy-L-tyrosine, inducing a puck-
ering type of astringency, have been identified as contributors
to the astringent taste of cocoa (6, 7). In contrast, the molecules
inducing the bloody, metallic bitter taste of the dichloromethane
fraction still need to be investigated.
With the aim at completing the knowledge on the taste active
nonvolatiles in roasted cocoa, the objectives of the present
investigation were, therefore, to identify and quantify the
metallic- and bitter-tasting compounds in the dichloromethane
solubles isolated from fermented and roasted cocoa, to determine
their human threshold concentrations and to evaluate their taste
contribution based on dose/response considerations.
MATERIALS AND METHODS
Synthesis of 2,5-Diketopiperazines. After a synthetic strategy,
reported in the literature (3, 9) with some modifications, 2,2-dimethox-
ypropane (0.6 mmol) and hydrochloric acid (125 µL, 32%) were added
to a solution of the corresponding dipeptide (1.2 mmol) Z-IIe-Pro-OH,
Z-Ala-Val-OH, Z-Ala-Pro-OH, Z-Pro-Pro-OH, Z-Val-Leu-OH, Z-Ala-
Leu-OH, Z-Val-Tyr-OH, or Z-Ala-IIe-OH in methanol (2.0 mL). After
stirring the mixture for 24 h at room temperature, the solvent was
removed in a vacuum and the residue was dissolved in a minimum
amount of absolute methanol. Addition of diethyl ether (50 mL) resulted
in crystallization of the product; the supernatant diethyl ether was
discarded; the procedure was repeated twice; and then the sample was
freed from solvent in a vacuum. The dipeptide methyl ester obtained
was dissolved in methanol (80 mL), and the N-protecting group was
removed upon hydrogenation with hydrogen (5 bar) using palladium
(10% on charcoal; 100 mg) as the catalyst. After 30 min, the catalyst
was filtered off, triethylamine (2 mmol) was added, and the solution
was stirred for 48 h. The solvent was then removed in a vacuum, and
the residue obtained was taken up in a mixture (20:80, v/v; 5 mL) of
methanol and aqueous formic acid (0.1%, pH 2.5) and was applied
onto the top of a water-cooled glass column (140 × 40 mm i.d.) filled
with a slurry of LiChroprep 25-40 µm RP-18 material (Merck,
Darmstadt, Germany) in the same solvent mixture. When the effluent
was monitored at 220 nm, chromatography was performed starting with
a mixture (20:80, v/v) of methanol and aqueous formic acid (0.1%,
pH 2.5), thereafter increasing the methanol content to 70%. The frac-
tions containing the title compounds were freed from solvent in a
vacuum, freeze-dried, and finally purified by means of automated HPLC
(Jasco, Groâ-Umstadt, Germany) consisted of a PU 1580 HPLC-pump
system with a DG-1580-53 in-line degasser, a LG-1580-02 low-pressure
gradient unit, and a MD 1515 diode array detector (DAD) using a
semipreparative RP-18 column, ODS-Hypersil, 250 × 10 mm, 5 µm
(ThermoHypersil, Kleinostheim, Germany) as the stationary phase and
Chemicals. The following compounds were obtained commer-
cially: caffeine, theobromine, triethylamine (Fluka, Neu-Ulm, Ger-
many), 2,2-dimethoxypropane, palladium 10% on charcoal (Sigma-
Aldrich, Steinheim, Germany), hydrochloric acid (Merck, Darmstadt,
Germany), Z-IIe-Pro-OH, Z-Ala-Val-OH, Z-Ala-Pro-OH, Z-Pro-Pro-
OH, Z-Val-Leu-OH, Z-Ala-Leu-OH, Z-Val-Tyr-OH, Z-Ala-IIe-OH,
Z-Val-Phe-O-CH3, Z-IIe-Phe-O-CH3, Z-Ala-Phe-O-CH3, Z-Asn-Phe-
O-CH3, Z-Ala-Tyr-O-CH3, cis-cyclo(Gly-Pro) (1), cis-cyclo(Phe-Pro)
(2), cis-cyclo(Leu-Pro) (3), cis-cyclo(Gly-Phe) (4), cis-cyclo(Asp-Phe)
(5), cis-cyclo(Leu-Gly) (6), cis-cyclo(Pro-Thr) (7), cis-cyclo(Pro-Tyr)
(8), cis-cyclo(Leu-Phe) (9), cis-cyclo(Pro-Val) (10), cis-cyclo(Phe-Ser)
(11), cis-cyclo(Val-Val) (12), cis-cyclo(Ala-Ala) (13), cis-cyclo(Gly-
Gly) (14), cis-cyclo(Ala-Gly) (15), trans-cyclo(D-Ala-L-Pro) (16), and
trans-cyclo(D-Ala-L-Val) (17) (Bachem, Bubendorf, Switzerland).
Solvents were of HPLC grade (Merck, Darmstadt, Germany). Deuter-
ated solvents were obtained from Euroiso-Top (Gif-Sur-Yvette, France).
Cocoa beans, fermented in Ghana for 5 days, were roasted by the food
industry. Bottled water (Vittel, low mineralization, 405 mg/L) adjusted
to pH 6.0 with aqueous hydrochloric acid (0.1 mol/L) was used for
sensory evaluation.
Sensory Analyses. Training of the Sensory Panel. A total of 12
subjects (5 women and 7 men, ages 25-38 years) with no history of
known taste disorders were trained to evaluate the taste of aqueous
solutions (2 mL each) of the following standard taste compounds in
bottled water adjusted to pH 6.0 with aqueous hydrochloric acid (0.1
mol/L) or aqueous sodium hydroxide solution (0.1 mmol/L) by using
a triangle test as described below: sucrose (25 mmol/L) for sweet taste;
lactic acid (20 mmol/L) for sour taste; NaCl (12 mmol/L) for salty
taste, caffeine (1 mmol/L) for bitter taste, sodium glutamate (3 mmol/
L) for umami taste, gallustannic acid (0.05%) and quercetin-3-O-â-D-
glucopyranoside (0.002 mmol/L) for the puckering astringency and the

7224 J. Agric. Food Chem., Vol. 53, No. 18, 2005
Stark and Hofmann
Figure 1. Chemical structures of cis-configured 2,5-diketopiperazines reported earlier in roasted cocoa.
Table 1. Mass Spectrometric Data of 2,5-Diketopiperazinesa
m/z of
m/z (%) of
product ion
+
2,5-diketopiperazine
cis-cyclo(Gly- -Pro) (1)
-Phe- -Pro) (2)
[M
+
1]
L
155
245
211
205
263
171
199
261
261
197
235
199
143
115
129
169
171
185
219
169
171
262
247
185
235
213
263
261
211
195
247
70 (100), 155 (90), 127 (53), 82 (28), 113 (10), 58 (6)
120 (100), 70 (75), 245 (53), 154 (51), 217 (17), 172 (11), 98 (7)
cis-cyclo(
L
L
L
cis-cyclo(
-Leu-L-Pro) (3)
70 (100), 183 (37), 138 (29), 86 (28), 26 (211), 26 (114), 155 (15), 98 (12)
120 (100), 132 (28), 114 (25), 149 (21), 177 (20), 205 (20), 91 (18)
175 (100), 203 (84), 91 (44), 245 (28), 130 (24), 263 (21), 158 (20), 172 (12), 120 (8)
86 (100), 72 (19), 69 (15), 114 (14), 126 (14), 171 (13), 143 (5)
153 (100), 70 (27), 125 (20), 181 (17), 199 (11), 97 (7), 74 (6)
136 (100), 216 (46), 70 (28), 107 (19), 155 (18), 233 (6)
120 (100), 86 (55), 188 (40), 261 (39), 114 (32), 216 (29), 233 (24), 170 (5)
70 (100), 124 (72), 72 (50), 100 (31), 197 (31), 169 (26), 98 (20), 154 (15), 141 (13)
120 (100), 162 (26), 144 (15), 235 (14), 207 (11), 147 (6)
72 (100), 126 (54), 199 (35), 171 (30), 100 (24), 154 (23)
72 (100), 98 (73), 70 (56), 143 (27), 55 (11), 115 (10)
58 (100), 59 (66), 115 (65), 87 (50)
72 (100), 84 (80), 55 (60), 101 (39), 129 (38)
70 (100), 169 (24), 141 (12), 96 (11), 72 (9), 113 (5), 98 (5)
72 (100), 98 (81), 126 (41), 100 (37), 171 (17), 143 (12)
86 (100), 140 (50), 114 (37), 112 (26), 72 (21), 185 (20), 157 (7)
120 (100), 146 (9), 191 (6), 174 (6), 219 (5), 128 (4), 72 (3)
70 (100), 169 (24), 141 (12), 96 (11), 72 (9), 113 (5), 98 (5)
72 (100), 98 (81), 126 (41), 100 (37), 171 (17), 143 (12)
203 (100), 245 (65), 175 (53), 262 (31), 130 (20), 120 (8), 91 (7)
120 (100), 174 (63), 219 (46), 72 (23), 247 (21), 202 (17), 156 (14), 100 (10), 148 (7)
86 (100), 140 (50), 114 (37), 112 (26), 72 (21), 185 (20), 157 (7)
107 (100), 135 (78), 235 (30), 129 (20), 162 (16), 207 (9)
72 (100), 86 (86), 140 (60), 213 (50), 100 (35), 168 (35), 185 (35), 84 (18), 113 (17)
136 (100), 235 (85), 263 (57), 72 (28), 107 (15)
cis-cyclo(Gly-L-Phe) (4)
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
L
L
L
L
L
L
L
L
L
-Asp-
-Leu-Gly) (6)
L-Phe) (5)
-Pro-
L
-Thr) (7)
-Tyr) (8)
-Phe) (9)
-Pro-
L
-Leu-
L
-Val-L-Pro) (10)
-Phe-L-Ser) (11)
-Val-
L
-Val) (12)
-Ala) (13)
-Ala-
L
cis-cyclo(Gly-Gly) (14)
cis-cyclo( -Ala-Gly) (15)
L
trans-cyclo(
D
-Ala-
L
-Pro) (16)
-Val) (17)
trans-cyclo(
D
-Ala-
L
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
L
L
L
L
L
L
L
L
L
L
L
L
L
-Ala-
-Ala-
-Ala-
-Ala-
L
L
L
L
-Leu) (18)
-Phe) (19)
-Pro) (20)
-Val) (21)
-Phe) (22)
-Phe) (23)
-IIe) (24)
-Tyr) (25)
-Leu) (26)
-Tyr) (27)
-Asn-
L
-Val-
-Ala-
-Ala-
-Val-
-Val-
L
L
L
L
L
-IIe-
L
-Phe) (28)
-Pro) (29)
120 (100), 86 (55), 188 (40), 261 (39), 114 (32), 216 (29), 233 (24), 170 (5)
70 (100), 183 (37), 138 (29), 86 (28), 26 (211), 26 (114), 155 (15), 98 (12)
195 (100), 70 (19), 98 (18)
-IIe-
L
-Pro-
L
-Pro) (30)
trans-cyclo(L-Val-D
-Phe) (31)
120 (100), 174 (63), 219 (46), 72 (23), 247 (21), 202 (17), 156 (14), 100 (10), 148 (7)
+
a MS data were obtained by ESI ionization.
a mixture of methanol and aqueous formic acid (0.1% in water, pH
2.5) as the mobile phase. After the solvents were removed in high
vacuum (10-6 Pa), the title compounds were suspended in water (10
mL) and freeze-dried 3 times to afford the corresponding 2,5-diketo-
piperazines as white, amorphous powders in high purities of more than
99%.
(3S,6S)-3-Methyl-6-(2-methylpropyl)-piperazine-2,5-dione, cis-cyclo-
(L-Ala-L-Leu), 18 (Figure 1). Yield: 0.82 mmol (68%). LC-MS (ESI+)
m/z: 185 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table 1. 1H NMR
(400 MHz, DMSO-d6, COSY) δ: 0.86 [d, 3H, J ) 6.6 Hz, H-C(10/
11)], 0.88 [d, 3H, J ) 6.6 Hz, H-C(10/11)], 1.27 [d, 3H, J ) 7.0 Hz,
H-C(7)], 1.47 [m, 1H, H-C(8a)], 1.62 [m, 1H, H-C(8b)], 1.83 [m, 1H,
H-C(9)], 3.77 [m, 1H, H-C(6)], 3.86 [m, 1H, H-C(3)], 8.06 [s, 1H,
H-N(4)], 8.07 [s, 1H, H-N(1)]. 13C NMR (100 MHz, DMSO-d6, HMQC,
HMBC) δ: 19.7 [C(7)], 22.0 [C(10/11)], 23.1 [C(10/11)], 23.8 [C(9)],
42.7 [C(8)], 50.1 [C(3)], 52.8 [C(6)], 168.5 [C(5)], 169.0 [C(2)].
(3S,6S)-3-Methyl-6-(phenylmethyl)-piperazine-2,5-dione, cis-cyclo-
(L-Ala-L-Phe), 19 (Figure 2). Yield: 1.04 mmol (87%). LC-MS (ESI+)
m/z: 219 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table 1. 1H NMR
(400 MHz, DMSO-d6, COSY) δ: 0.49 [d, 3H, J ) 7.0 Hz, H-C(7)],
2.87 [dd, 1H, J ) 5.0, 13.4 Hz, H-C(8a)], 3.12 [dd, 1H, J ) 3.8, 13.4
Hz, H-C(8b)], 3.63 [m, 1H, H-C(6)], 4.17 [m, 1H, H-C(5)], 7.22 [m,
5H, H-C(10-14)], 7.98 [s, 1H, H-N(1)], 8.06 [s, 1H, H-N(4)]. 13C NMR
(100 MHz, DMSO-d6, HMQC, HMBC) δ: 19.6 [C(7)], 38.2 [C(8)],
49.6 [C(6)], 55.3 [C(3)], 126.5 [C(12)], 127.9 [C(11, 13)], 130.3 [C(10,
14)], 136.0 [C(9)], 165.7 [C(2)], 167.6 [C(5)].

Metallic- and Bitter-Tasting Diketopiperazines in Roasted Cocoa Nibs
J. Agric. Food Chem., Vol. 53, No. 18, 2005 7225
Figure 2. Chemical structures of cis-configured 2,5-diketopiperazines previously not reported in roasted cocoa.
(3S,6S)-3-Methyl-hexahydropyrrolo-[1,2-a]-pyrazine-1,4-dione, cis-
cyclo(L-Ala-L-Pro), 20 (Figure 1). Yield: 0.72 mmol (60%). LC-MS
(ESI+) m/z: 169 (100, [M + 1]+); LC-MS/MS (ESI+): cf. Table 1.
1H NMR (400 MHz, MeOH-d4, COSY) δ: 1.38 [d, 3H, J ) 6.8 Hz,
H-C(10)], 1.90-2.01 [m, 3H, H-C(7a, 8)], 2.31 [m, 1H, H-C(7b)], 3.52
[m, 2H, H-C(9)], 4.18 [m, 1H, H-C(3)], 4.25 [m, 1H, H-C(6)]. 13C
NMR (100 MHz, DMSO-d6, HMQC, HMBC) δ: 15.7 [C(10)], 23.6
[C(8)], 29.2 [C(7)], 46.4 [C(9)], 52.1 [C(3)], 60.5 [C(6)], 169.0 [C(2)],
172.6 [C(5)].
(3S,6S)-3-Methyl-6-(1-methylethyl)-piperazine-2,5-dione, cis-cyclo-
(L-Ala-L-Val), 21 (Figure 1). Yield: 0.42 mmol (35%). LC-MS (ESI+)
m/z: 171 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table 1. 1H NMR
(400 MHz, DMSO-d6, COSY) δ: 0.85 [d, 3H, J ) 6.6 Hz, H-C(9/
10)], 0.95 [d, 3H, J ) 7.2 Hz, H-C(9/10)], 1.28 [d, 3H, J ) 6.9 Hz,
H-C(7)], 2.15 [m, 1H, J ) 3.3, 6.9 Hz, H-C(8)], 3.68 [m, 1H, H-C(6)],
3.88 [q, 1H, J ) 6.9 Hz, H-C(3)], 7.95 [s, 1H, H-N(1)], 8.01 [s, 1H,
H-N(4)]. 13C NMR (100 MHz, DMSO-d6, HMQC, HMBC) δ: 16.4
[C(9/10)], 18.0 [C(9/10)], 19.6 [C(7)], 30.6 [C(8)], 49.2 [C(3)], 58.9
[C(6)], 166.1 [C(5)], 168.2 [C(2)].
H-C(10)], 0.91 [d, 3H, J ) 7.2 Hz, H-C(11)], 1.18 [m, 1H, H-C(9a)],
1.27 [d, 3H, J ) 7.0 Hz, H-C(7)], 1.41 [m, 1H, H-C(9b)], 1.85 [m,
1H, H-C(8)], 3.74 [m, 1H, H-C(6)], 3.88 [m, 1H, H-C(3)], 7.93 [s,
1H, H-N(1)], 8.07 [s, 1H, H-N(4)]. 13C NMR (100 MHz, DMSO-d6,
HMQC, HMBC) δ: 12.9 [C(10)], 15.1 [C(11)], 20.0 [C(7)], 24.3 [C(9)],
38.1 [C(8)], 49.8 [C(3)], 58.9 [C(6)], 166.7 [C(5)], 168.7 [C(2)].
(3S,6S)-3-[(4-hydroxyphenyl)methyl]-6-methyl-piperazine-2,5-di-
one, cis-cyclo(L-Ala-L-Tyr), 25 (Figure 2). Yield: 0.60 mmol (50%).
LC-MS (ESI+) m/z: 235 (100, [M + 1]+). LC-MS/MS (ESI+): cf.
Table 1. 1H NMR (400 MHz, DMSO-d6, COSY) δ: 0.54 [d, 3H, J )
7.1 Hz, H-C(14)], 2.73 [dd, 1H, J ) 4.8, 13.6 Hz, H-C(7a)], 3.01 [dd,
1H, J ) 3.6, 13.6 Hz, H-C(7b)], 3.60 [m, 1H, H-C(6)], 4.07 [m, 1H,
H-C(3)], 6.65 [d, 2H, J ) 8.5 Hz, H-C(10, 12)], 6.93 [d, 2H, J ) 8.5
Hz, H-C(9, 13)], 7.97 [s, 1H, H-N(1)], 8.02 [s, 1H, H-N(4)]. 13C NMR
(100 MHz, DMSO-d6, HMQC, HMBC) δ: 20.0 [C(14)], 37.8 [C(7)],
49.9 [C(6)], 55.8 [C(3)], 115.0 [C(10, 12)], 126.1 [C(8)], 131.5 [C(9)],
156.3 [C(11, 13)], 166.1 [C(2)], 167.9 [C(5)].
(3S,6S)-3-(1-Methylethyl)-6-(2-methylpropyl)-piperazine-2,5-
dione, cis-cyclo(L-Val-L-Leu), 26 (Figure 2). Yield: 0.67 mmol (56%).
LC-MS (ESI+) m/z: 213 (100, [M + 1]+). LC-MS/MS (ESI+): cf.
Table 1. 1H NMR (400 MHz, DMSO-d6, COSY) δ: 0.85 [d, 3H, J )
7.1 Hz, H-C(8/9)], 0.86 [d, 3H, J ) 6.6 Hz, H-C(12/13)], 0.88 [d, 3H,
J ) 6.6 Hz, H-C(12/13)], 0.95 [d, 3H, J ) 7.1 Hz, H-C(8/9)], 1.45 [m,
1H, H-C(10a)], 1.63 [m, 1H, H-C(10b)], 1.85 [m, 1H, H-C(11)], 2.11
[m, 1H, H-C(7)], 3.62 [m, 1H, H-C(3)], 3.76 [m, 1H, H-C(6)], 8.01 [s,
1H, H-N(4)], 8.13 [s, 1H, H-N(1)]. 13C NMR (100 MHz, DMSO-d6,
HMQC, HMBC) δ: 17.5 [C(8/9)], 18.9 [C(8/9)], 21.9 [C(12/13)], 23.2
[C(12/13)], 23.7 [C(11)], 31.6 [C(7)], 44.1 [C(10)], 52.6 [C(6)], 59.7
[C(3)], 167.0 [C(2)], 168.6 [C(5)].
(3S,6S)-3-[(4-Hydroxyphenyl)methyl]-6-(1-methylethyl)-piperazine-
2,5-dione, cis-cyclo(L-Val-L-Tyr), 27 (Figure 2). Yield: 0.29 mmol
(24%). LC-MS (ESI+) m/z: 263 (100, [M + 1]+). LC-MS/MS
(ESI+): cf. Table 1. 1H NMR (400 MHz, DMSO-d6, COSY) δ: 0.37
[d, 3H, J ) 7.0 Hz, H-C(15/16)], 0.69 [d, 3H, J ) 7.0 Hz, H-C(15/
16)], 1.74 [m, 1H, H-C(14)], 2.77 [dd, 1H, J ) 4.8, 13.8 Hz, H-C(7a)],
3.01 [dd, 1H, J ) 4.5, 13.8 Hz, H-C(7b)], 3.51 [m, 1H, H-C(6)], 4.10
[m, 1H, H-C(3)], 6.62 [d, 2H, J ) 8.4 Hz, H-C(10, 12)], 6.96 [d, 2H,
J ) 8.4 Hz, H-C(9, 13)], 7.85 [s, 1H, H-N(1)], 7.95 [s, 1H, H-N(4)],
9.10 [s, 1H, HO-C(11)]. 13C NMR (100 MHz, DMSO-d6, HMQC,
HMBC) δ: 16.5 [C(15/16)], 18.5 [C(15/16)], 31.3 [C(14)], 37.4 [C(7)],
55.5 [C(3)], 59.4 [C(6)], 114.9 [C(10, 12)], 126.5 [C(8)], 131.3 [C(9,
13)], 156.3 [C(11)], 166.6 [C(5)], 166.9 [C(2)].
(3S,6S)-3-Phenylmethyl-6-amidocarboxymethyl-piperazine-2,5-di-
one, cis-cyclo(L-Asn-L-Phe), 22 (Figure 1). Yield: 1.08 mmol (90%).
LC-MS (ESI+) m/z: 262 (100, [M + 1]+); LC-MS/MS (ESI+): cf.
1
Table 1. H NMR (400 MHz, DMSO-d6, COSY) δ: 1.42 [dd, 1H, J
) 8.6, 15.9 Hz, H-C(14a)], 2.29 [dd, 1H, J ) 4.1, 15.9 Hz, H-C(14b)],
2.96 [dd, 1H, J ) 4.9, 13.7 Hz, H-C(7a)], 3.10 [dd, 1H, J ) 4.9, 13.7
Hz, H-C(7b)], 4.02 [m, 1H, H-C(6)], 4.18 [m, 1H, H-C(3)], 6.89 [s,
2H, H-N(16)], 7.25 [m, 5H, H-C(9-13)], 7.58 [s, 1H, H-N(1)], 8.08
[s, 1H, H-N(4)]. 13C NMR (100 MHz, DMSO-d6, HMQC, HMBC) δ:
38.0 [C(7)], 38.1 [C(14)], 51.1 [C(6)], 55.1 [C(3)], 126.6 [C(11)], 128.0
[C(10, 12)], 130.0 [C(9, 13)], 136.3 [C(8)], 166.3 [C(2)], 166.8 [C(5)],
171.5 [C(15)].
(3S,6S)-3-(1-Methylethyl)-6-(phenylmethyl)-piperazine-2,5-dione, cis-
cyclo(L-Val-L-Phe), 23 (Figure 1). Yield: 0.72 mmol (60%). LC-
MS (ESI+) m/z: 247 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table
1
1. H NMR (400 MHz, DMSO-d6, COSY) δ: 0.31 [d, 3H, J ) 6.9
Hz, H-C(8/9)], 0.67 [d, 3H, J ) 6.9 Hz, H-C(8/9)], 1.70 [m, 1H,
H-C(7)], 2.88 [dd, 1H, J ) 5.0, 13.4 Hz, H-C(10a)], 3.14 [dd, 1H, J )
4.3, 13.4 Hz, H-C(10b)], 3.53 [m, 1H, H-C(6)], 4.20 [m, 1H, H-C(3)],
7.21 [m, 5H, H-C(12-16)], 7.88 [s, 1H, H-N(1)], 8.06 [s, 1H, H-N(4)].
13C NMR (100 MHz, DMSO-d6, HMQC, HMBC) δ: 17.1 [C(8/9)],
19.1 [C(8/9)], 31.9 [C(7)], 38.7 [C(10)], 55.9 [C(3)], 60.1 [C(6)], 127.3
[C(14)], 128.8 [C(13, 15)], 131.2 [C(12, 16)], 137.2 [C(11)], 167.3
[C(2)], 167.4 [C(5)].
(3S,6S)-3-[(1S)-1-Methylpropyl]-6-(phenylmethyl)-piperazine-2,5-di-
one, cis-cyclo(L-IIe-L-Phe), 28 (Figure 2). Yield: 1.02 mmol (85%).
LC-MS (ESI+) m/z: 261 (100, [M + 1]+). LC-MS/MS (ESI+): cf.
Table 1. 1H NMR (400 MHz, DMSO-d6, COSY) δ: 0.54 [d, 3H, J )
7.0 Hz, H-C(10)], 0.57 [m, 3H, H-C(9)], 0.66 [m, 2H, H-C(8)], 1.38
(3S,6S)-3-Methyl-6-[(1S)-1-methylpropyl)]-piperazine-2,5-dione, cis-
cyclo(L-Ala-L-IIe), 24 (Figure 2). Yield: 0.42 mmol (35%). LC-MS
(ESI+) m/z: 185 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table 1.
1H NMR (400 MHz, DMSO-d6, COSY) δ: 0.85 [t, 3H, J ) 7.4 Hz,

7226 J. Agric. Food Chem., Vol. 53, No. 18, 2005
Stark and Hofmann
[m, 1H, H-C(7)], 2.86 [dd, 1H, J ) 5.0, 13.5 Hz, H-C(11a)], 3.15 [dd,
1H, J ) 4.1, 13.5 Hz, H-C(11b)], 3.57 [m, 1H, H-C(3)], 4.20 [m, 1H,
H-C(6)], 7.21 [m, 5H, H-C(13-17)], 7.84 [s, 1H, H-N(4)], 8.05 [s,
1H, H-N(1)]. 13C NMR (100 MHz, DMSO-d6, HMQC, HMBC) δ: 12.0
[C(9/10)], 14.7 [C(9/10)], 23.4 [C(8)], 37.8 [C(11)], 38.0 [C(7)], 55.3
[C(6)], 59.0 [C(3)], 126.7 [C(15)], 128.1 [C(14, 16)], 130.7 [C(13, 17)],
136.5 [C(12)], 166.6 [C(5)], 166.7 [C(2)].
(3S,6S)-3-[(1S)-1-Methylpropyl]-hexahydropyrrolo-[1,2-a]-pyrazine-
1,4-dione, cis-cyclo(L-IIe-L-Pro), 29 (Figure 2). Yield: 1.06 mmol
(88%). LC-MS (ESI+) m/z: 211 (100, [M + 1]+). LC-MS/MS
(ESI+): cf. Table 1. 1H NMR (400 MHz, DMSO-d6, COSY) δ: 0.82
[t, 3H, J ) 7.6 Hz, H-C(12)], 0.98 [d, 3H, J ) 7.1 Hz, H-C(13)], 1.31
[m, 2H, H-C(11)], 1.81 [m, 3H, H-C(7a, 8)], 2.03 [m, 1H, H-C(10)],
2.14 [m, 1H, H-C(7b)], 3.36 [m, 2H, H-C(9)], 3.95 [s, 1H, H-C(3)],
4.11 [t, 1H, J ) 7.7 Hz, H-C(6)], 7.91 [s, 1H, H-N(4)]. 13C NMR (100
MHz, DMSO-d6, HMQC, HMBC) δ: 12.5 [C(12)], 15.2 [C(13)], 22.2
[C(8)], 24.1 [C(11)], 28.1 [C(7)], 35.1 [C(10)], 44.8 [C(9)], 58.4 [C(6)],
59.4 [C(3)], 165.5 [C(2)], 170.3 [C(5)].
(3S,6S)-Octahydrodipyrrolo-[1,2-a; 1′,2′-d]-pyrazine-5,10-dione, cis-
cyclo(L-Pro-L-Pro), 30 (Figure 2). Yield: 0.89 mmol (74%). LC-MS
(ESI+) m/z: 195 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table 1.
1H NMR (400 MHz, DMSO-d6, COSY) δ: 1.70-1.90 [m, 6H, H-C(7a,
10a, 8, 11)], 2.13 [m, 2H, H-C(7b, 10b)], 3.36 [m, 4H, H-C(9, 12)],
4.27 [t, 2H, J ) 8.0 Hz, H-C(3, 6)]. 13C NMR (100 MHz, DMSO-d6,
HMQC, HMBC) δ: 23.0 [C(8, 11)], 27.3 [C(7, 10)], 44.7 [C(9, 12)],
59.9 [C(3, 6)], 166.1 [C(2, 5)].
(3S,6R)-3-(1-Methylethyl)-6-(phenylmethyl)-piperazine-2,5-dione,
trans-cyclo(L-Val-D-Phe), 31. Yield: 0.38 mmol (32%). LC-MS (ESI+)
m/z: 247 (100, [M + 1]+). LC-MS/MS (ESI+): cf. Table 1. 1H NMR
(400 MHz, DMSO-d6, COSY) δ: 0.76 [d, 3H, J ) 6.9, H-C(8/9)],
0.83 [d, 3H, J ) 6.9 Hz, H-C(8/9)], 2.04 [m, 1H, H-C(7)], 2.88 [dd,
1H, J ) 4.9, 13.5 Hz, H-C(10a)], 2.97 [m, 1H, H-C(3)], 3.14 [dd, 1H,
J ) 4.1, 13.5 Hz, H-C(10b)], 4.17 [m, 1H, H-C(6)], 7.21 [m, 5H,
H-C(12-16)], 7.93 [s, 1H, H-N(4)], 8.21 [s, 1H, H-N(1)]. 13C NMR
(100 MHz, DMSO-d6, HMQC, HMBC) δ: 16.8 [C(8/9)], 18.3 [C(8/
9)], 31.7 [C(7)], 38.1 [C(10)], 54.9 [C(6)], 59.1 [C(3)], 126.7 [C(14)],
128.1 [C(13, 15)], 130.3 [C(12, 16)], 136.3 [C(11)], 167.2 [C(5)], 167.5
[C(2)].
exchange chromatography following the procedure described above.
The residue was taken up in a mixture (50:50, v/v; 3 mL) of methanol
and aqueous formic acid (0.1% in water, pH 2.5), and after membrane
filtration, aliquots (5-20 µL) were analyzed by means of LC-MS on
a 150 × 2 mm i.d., 4 µm, Synergi Fusion-RP column (Phenomenex)
using the same gradient described above. All of the 2,5-diketopipera-
zines containing the aromatic amino acids L-phenylalanine or L-tyrosine,
respectively, were quantified by using trans-cyclo(L-Val-D-Phe) (31)
as the internal standard. The remaining 2,5-diketopiperazines were
quantified by using trans-cyclo(D-Ala-L-Val) (17) as the internal
standard. The amounts of the individual 2,5-diketopiperazines were
calculated using response curves that had been determined by analysis
of solutions containing defined amounts of the internal standards and
the target compounds in different concentrations.
Liquid Chromatography/Mass Spectrometry (LC-MS/MS). Ali-
quots (5-20 µL) were analyzed by means of LC-MS on a 150 × 2
mm i.d., 4 µm, Synergi Fusion-RP column (Phenomenex) coupled to a
TSQ-Discovery-MS (Thermo Finnigan MAT GmbH, Bremen, Ger-
many) using positive (ESI+) electronspray ionization. Starting with
aqueous formic acid (0.1% in water, pH 2.5) at a flow rate of 0.2 mL/
min, after 10 min, the methanol content was increased to 60% within
40 min, increased to 100% within 10 min, and finally, held at 100%
for 10 min. Using the selected reaction monitoring (SRM) mode and
the multiple reaction monitoring (MRM) mode, the 2,5-diketopipera-
zines were detected by measuring the corresponding mass traces as
given in Table 1.
Nuclear Magnetic Resonance (NMR) Spectroscopy. 1H, 13C, and
DEPT-135 NMR experiments were performed on Bruker AV-360
1
spectrometers. H, COSY, HMQC, and HMBC measurements were
performed on a Bruker AMX 400-III spectrometer (Bruker, Rhein-
stetten, Germany). Evaluation of the experiments was carried out using
1D- and 2D-WIN NMR as well as XWin-NMR software (version 3.5;
Bruker, Rheinstetten, Germany). DMSO-d6 and MeOH-d4 were used
as solvents, and tetramethylsilane (TMS) was used as the internal
standard.
RESULTS AND DISCUSSION
With the aim at identifying the key molecules imparting the
metallic and bitter taste sensation observed recently for a
dichloromethane isolate of roasted cocoa (6), roasted cocoa nibs
were fully extracted with acetone/water and the aqueous solution
obtained was extracted with dichloromethane closely following
the procedure reported recently (6). Analysis of this dichlo-
romethane extract by means of HPLC/DAD and HPLC-MS
revealed theobromine as the major constituent, followed by a
large series of compounds each in comparatively low concentra-
tions. HPLC/degustation revealed that these compounds im-
parted a significant bitter taste accompanied by a blood-like
metallic taste sensation.
Identification of Compounds Imparting Metallic and
Bitter Taste. To further characterize the chemical structure of
these bitter compounds, the major amounts of theobromine
needed to be removed first. To achieve this, the dichloromethane
solubles were placed on the top of a cation-exchange column
preconditioned with 0.001 mol/L aqueous hydrochloric acid.
Using 0.001 mol/L aqueous hydrochloric acid as the eluent, the
effluent was collected in five fractions, freeze-dried, and
analyzed by means of LC-MS/MS. Fractions 1-4 containing
the compounds under investigation were combined, and the
structure of the individual compounds were evaluated by means
of LC-MS/MS experiments.
Isolation and Identification of 2,5-Diketopiperazines in Roasted
Cocoa. Roasted cocoa nibs (100 g) were frozen in liquid nitrogen,
crushed in a grinding mill, and then extracted with n-pentane (5 ×
300 mL) at room temperature for 30 min. After centrifugation, the
residual cocoa material was then extracted 5 times with acetone/water
(70:30, v/v; 300 mL each) for 45 min at room temperature while stirring.
After filtration, the liquid layer was freed from acetone under reduced
pressure at 30 °C, and the aqueous solution obtained was extracted
with dichloromethane (5 × 150 mL) and then freeze-dried to give the
dichloromethane extractables. An aliquot (100 mg) of the dichlo-
romethane extract was suspended in aqueous hydrochloric acid (0.001
N HCl in water; 5 mL) and then applied on top of a glass column (170
× 15 mm i.d.) filled with a slurry of Dowex 50WX 8, 50-100 mesh
(Merck, Darmstadt, Germany), which was preconditioned with 0.1
mol/L aqueous hydrochloric acid (500 mL), followed by 0.001 mol/L
aqueous hydrochloric acid (500 mL). Using 0.001 mol/L aqueous
hydrochloric acid (500 mL) as the eluent, five fractions (100 mL each)
were collected, freeze-dried, and analyzed by means of LC-MS/MS
on a 150 × 2 mm i.d., 4 µm, Synergi Fusion-RP column (Phenomenex,
Aschaffenburg, Germany). Starting with aqueous formic acid (0.1%
in water, pH 2.5) at a flow rate of 0.2 mL/min, after 10 min, the
methanol content was increased to 60% within 40 min, increased to
100% within 10 min, and finally, held at 100% for 10 min. A
comparison of retention time and spectroscopic data with those obtained
for the corresponding reference compounds led to the unequivocal
identification of the 2,5-diketopiperazines 1-12 and 18-30 (Figures
1 and 2) in the roasted cocoa nibs.
Each of the compounds detected showed a characteristic mass
spectrum that was well in line with the mass spectrometric data
reported for 2,5-diketopiperazines (10). For example, the MS/
MS spectrum of the pseudomolecular ion m/z 247 detected for
the diketopiperazine cis-cyclo(L-Val-L-Phe) (23) revealed the
characteristic ions with m/z 72 and 120, which are well in line
with the expected immonium ions formed from the valine or
Quantitative Analysis of 2,5-Diketopiperazines (DKPs). Ground,
roasted cocoa nibs (50.0 g) were spiked with the internal standards
trans-cyclo(D-Ala-L-Val) (1.4 mg) and trans-cyclo(L-Val-D-Phe) (1.9
mg) dissolved in methanol (2 mL). After homogenization and equilibra-
tion, the dichloromethane extractables were isolated as reported above,
and an aliquot (100 mg) of this isolate was fractionated by cation-

Metallic- and Bitter-Tasting Diketopiperazines in Roasted Cocoa Nibs
J. Agric. Food Chem., Vol. 53, No. 18, 2005 7227
+
Figure 3. MS/MS (ESI ) spectrum of cis-cyclo(
L-Val-L-Phe).
Figure 4. Reaction sequence used for the synthesis of 2,5-diketopiperazines.
the phenylalanine moiety, respectively (Figure 3). Using the
characteristic immonium ions expected for the individual
diketopiperazines containing proline (m/z 70), alanine (m/z 44),
valine (m/z 72), phenylalanine (m/z 120), tyrosine (m/z 136),
leucine/isoleucine (m/z 86), threonine (m/z 74), and serine (m/z
60), the cocoa extract has been screened for the following 14
cis-configured 2,5-diketopiperazines, which were reported earlier
as cocoa ingredients (3-5, 11): cis-cyclo(Gly-L-Pro) (1), cis-
cyclo(L-Phe-L-Pro) (2), cis-cyclo(L-Leu-L-Pro) (3), cis-cyclo-
(Gly-L-Phe) (4), cis-cyclo(L-Asp-L-Phe) (5), cis-cyclo(L-Leu-
Gly) (6), cis-cyclo(L-Ala-Gly) (15), cis-cyclo(L-Ala-L-Leu) (18),
cis-cyclo(L-Ala-L-Phe) (19), cis-cyclo(L-Ala-L-Pro) (20), cis-
cyclo(L-Ala-L-Val) (21), cis-cyclo(L-Asn-L-Phe) (22), cis-cyclo-
(L-Val-L-Phe) (23), and cis-cyclo(L-Pro-L-Asn). With the ex-
ception of cis-cyclo(L-Ala-Gly) and cis-cyclo(L-Pro-L-Asn), all
of these cis-configured diketopiperazines could be detected in
the dichloromethane fraction isolated from roasted cocoa nibs
(Figure 1). In addition, the presence of 13 additional cis-
configured diketopiperazines, namely, cis-cyclo(L-Pro-L-Thr)
(7), cis-cyclo(L-Pro-L-Tyr) (8), cis-cyclo(L-Leu-L-Phe) (9), cis-
cyclo(L-Pro-L-Val) (10), cis-cyclo(L-Phe-L-Ser) (11), cis-cyclo-
(L-Val-L-Val) (12), cis-cyclo(L-Ala-L-IIe) (24), cis-cyclo(L-Ala-
L-Tyr) (25), cis-cyclo(L-Val-L-Leu) (26), cis-cyclo(L-Val-L-Tyr)
(27), cis-cyclo(L-IIe-L-Phe) (28), cis-cyclo(L-IIe-L-Pro) (29), and
cis-cyclo(L-Pro-L-Pro) (30), were suggested on the basis of their
characteristic immonium ions detected by MS/MS experiments
(Figure 2).
Finally, the identity of the diketopiperazines 1-12 was
confirmed by comparison of LC-MS/MS data, retention times,
and sensory activity with those obtained from commercially
available reference compounds. To confirm the chemical
structures of the compounds 18-30, these diketopiperazines
were synthesized following the reaction sequence given in
Figure 4. The corresponding N-protected dipeptide was esteri-
fied with 2,2-dimethoxypropane and hydrochloric acid, and then
the N-protecting group was removed by hydrogenation using
palladium on charcoal as the catalyst. After triethylamine-
calalyzed cyclization, the target compounds 18-30 (Figures 1
and 2) were isolated by column chromatography on RP-18
material, followed by a final HPLC purification, yielding the
corresponding cis-configured diketopiperazines as white, amor-
phous powders in a high purity of more than 99%.

7228 J. Agric. Food Chem., Vol. 53, No. 18, 2005
Stark and Hofmann
highest threshold concentrations of 3810 and 1950 µmol/L. Cis-
cyclo(Gly-Gly) (14) and cis-cyclo(L-Ala-L-Ala) (13) did not
show any metallic taste sensation in concentrations below 7000
µmol/L. For the bitter taste threshold, cis-cyclo(L-Leu-L-Phe)
(9), cis-cyclo(L-IIe-L-Phe) (28), and cis-cyclo(L-Val-L-Tyr) (27)
were evaluated with the lowest threshold concentration of 190
µmol/L for bitter taste, whereas cis-cyclo(L-Ala-Gly) (15)
exhibited with 3910 µmol/L the highest threshold concentration.
A comparison of the thresholds of the tastant pairs cis-cyclo-
(L-Ala-L-Phe) (19) and cis-cyclo(L-Ala-L-Tyr) (25) or cis-cyclo-
(L-Val-L-Phe) (23) and cis-cyclo(L-Val-L-Tyr) (27) revealed that
the threshold concentration of the metallic sensation of phenyl-
alanine-containing diketopiperazines (40 or 140 µmol/L for 23
or 19) is below the threshold concentration of a diketopiperazine
containing a tyrosine residue (100 or 430 µmol/L for 27 or 25).
Vice versa, tyrosine-containing diketopiperazines were found to
show lower taste thresholds for bitterness when compared to
the corresponding diketopiperazines bearing a phenylalanine
moiety; for example, cis-cyclo(L-Val-L-Phe) (23) and cis-cyclo-
(L-Ala-L-Phe) (19) did not show any bitter taste below 1000
and 570 µmol/L, respectively, whereas cis-cyclo(L-Val-L-Tyr)
(27) and cis-cyclo(L-Ala-L-Tyr) (25) already showed bitter taste
at the threshold concentrations of 190 and 530 µmol/L (Table
2).
To study the influence of the second amino acid residue on
the bitter taste threshold, we compared the threshold concentra-
tions of the alanine-containing 2,5-diketopiperazines 13, 15, 18-
21, 24, and 25 differing in the structure of the second amino
acid moiety. The data showed that the taste threshold concentra-
tion for bitter taste increases with increasing polarity of the
second amino acid residue; e.g., cis-cyclo(L-Ala-L-Tyr), cis-
cyclo(L-Ala-L-IIe), cis-cyclo(L-Ala-L-Phe), and cis-cyclo(L-Ala-
L-Leu) showed threshold concentrations between 530 and 680
µmol/L, and cis-cyclo(L-Ala-L-Val) and cis-cyclo(L-Ala-L-Pro)
showed thresholds of 1470 and 1490 µmol/L, whereas cis-cyclo-
(L-Ala-Gly) and cis-cyclo(L-Ala-L-Ala) exhibited threshold
concentrations of 3910 or higher. This observation is well in
line with literature data pinpointing hydrophobic amino acids
such as phenylalanine, tyrosine, and leucine as the molecular
key responsible for the bitter taste of peptides (20-24).
To finally investigate the influence of the stereochemistry
on the taste threshold concentrations of diketopiperazines, we
compared the threshold concentrations of cis-cyclo(L-Val-L-Phe)
(23), cis-cyclo(L-Ala-L-Val) (21), and cis-cyclo(L-Ala-L-Pro)
(20) with those determined for trans-cyclo(L-Val-D-Phe) (31),
trans-cyclo(D-Ala-L-Val) (17), and trans-cyclo(D-Ala-L-Pro)
(16), respectively. As summarized in Table 2, the taste
thresholds determined for bitter taste as well as the metallic
taste sensation were not significantly influenced by the stere-
ochemistry, thus confirming earlier literature reports (3); for
example, compounds 23 and 31 both showed a threshold
concentration of 40 µmol/L and very close thresholds for
bitterness at 1000 and 500 µmol/L (Table 2).
Table 2. Human Taste Recognition Thresholds of
2,5-Diketopiperazines
taste threshold concentration
mol/L)a for
(µ
compound (number)
metallic sensation
bitter taste
cis-cyclo(Gly-Gly) (14)
>8000
50
>8000
610
cis-cyclo(Gly-
L
-Phe) (4)
-Pro) (1)
cis-cyclo(Gly-
L
384
1950
>7000
406
168
30
144
387
430
510
120
100
40
320
529
40
120
131
40
3250
3910
>7000
1470
540
680
570
1490
530
1260
470
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
-Ala-Gly) (15)
-Ala-
-Ala-
-Ala-
-Ala-
-Ala-
-Ala-
-Ala-
-Val-
-Val-
-Val-
-Val-
-Val-
L
L
L
L
L
L
L
L
L
L
L
L
-Ala) (13)
-Val) (21)
-IIe) (24)
-Leu) (18)
-Phe) (19)
-Pro) (20)
-Tyr) (25)
-Val) (12)
-Leu) (26)
-Tyr) (27)
-Phe) (23)
-Pro) (10)
190
1000
1280
590
-Leu-Gly) (6)
-Leu-
L
-Phe) (9)
-Pro) (3)
190
-Leu-
L
1190
1020
210
3810
190
-Phe-
-Phe-
-Phe-
L
-Pro) (2)
-Ser) (11)
-Asp) (5)
L
L
3810
40
-IIe-L-Phe) (28)
-Asn-
L
-Phe) (22)
380
631
760
190
120
387
188
40
960
-Pro-
-Pro-
-Pro-
L
-Thr) (7)
-Pro) (30)
-Tyr) (8)
1261
2580
480
480
830
L
L
-IIe-L-Pro) (29)
trans-cyclo(
trans-cyclo(
trans-cyclo(
D
-Ala-
L
-Pro) (16)
-Val) (17)
-Phe) (31)
D-Ala-
L
818
500
L-Val-
D
a Taste recognition threshold concentrations were determined by means of a
triangle test in bottled water.
Among all of these diketopiperazines identified in the present
study in cocoa, compounds 2, 3, 10, 29, and 30 were previously
reported in roasted malt (12), compound 2 in aged sake (13),
compounds 2, 3, 10, 20, 29, and 30 in beer (14), compounds
1-3, 7, 9, 10, 18, 20, 21, 23, 24, 26, and 28-30 in chicken
essence (10), compounds 1-3, 9, 10, 22, 23, and 28-30 in
roasted coffee (15, 16), compounds 2, 3, 10, 20, and 30 in comte
cheese (17), and compounds 9, 12, and 23 in hydrolyzed
vegetable protein (18). Although the structures are known in
synthetic chemistry (19), to the best of our knowledge, the com-
pounds cis-cyclo(L-Phe-L-Ser) (11), cis-cyclo(L-Val-L-Tyr) (27),
and cis-cyclo(L-Ala-L-Tyr) (25) identified in the present study
in cocoa have never been reported before in any food product.
Sensory Evaluation of 2,5-Diketopiperazines. Prior to
sensory analysis, trace amounts of solvents were removed in
high vacuum, and after triple freeze drying, the purity of all of
1
the compounds was checked by LC-MS as well as H NMR
spectroscopy. To study the sensory activity of the diketopip-
erazines, the human sensory recognition thresholds were
determined in water by means of the triangle test. The oral
sensation imparted by the diketopiperazines was described as
metallic and bitter, with threshold concentrations ranging from
30 to 1950 µmol/L for the blood-like metallic sensation and
from 190 to 3910 µmol/L for the bitter taste (Table 2). For all
of these compounds, the taste threshold for the metallic sensation
was found to be equal or below the threshold concentration
determined for the bitter taste. For the metallic taste sensation,
the diketopiperazines cis-cyclo(L-Ala-L-Leu) (18) showed the
lowest threshold concentration of 30 µmol/L, whereas cis-cyclo-
(L-Phe-L-Asp) (5) and cis-cyclo(L-Ala-Gly) (15) showed the
Quantitative Analysis of Diketopiperazines and Calcula-
tion of Dose-over-Threshold (DoT) Factors. With the aim at
evaluating the taste contribution of the individual diketopip-
erazines, 25 cis-configured diketopiperazines were quantitatively
determined in the roasted cocoa nibs by means of a diastereomer
dilution assay. For the quantitative analysis, the cocoa nibs were
spiked with synthetic trans-cyclo(L-Val-D-Phe) as the internal
standard for the quantification of the 2,5-diketopiperazines
containing the aromatic amino acids L-phenylalanine or L-
tyrosine, respectively, and with synthetic trans-cyclo(D-Ala-L-
Val) as a second internal standard for the quantification of the
remaining 2,5-diketopiperazines bearing aliphatic side chains.
After isolation of the dichloromethane fraction from the

Metallic- and Bitter-Tasting Diketopiperazines in Roasted Cocoa Nibs
J. Agric. Food Chem., Vol. 53, No. 18, 2005 7229
Figure 5. Quantification of selected diketopiperazines in roasted cocoa by means of LC−MS using trans-cyclo(L-Val-D-Phe) (31) as the internal standard
(IS).
powdered cocoa nibs and removal of theobromine by means of
cation-exchange chromatography as detailed above, quantifica-
tion was performed by analytical RP-HPLC-MS as shown for
example in Figure 5 for analysis of the phenylalanine- and
tyrosine-containing diketopiperazines. By far, the highest con-
centrations of 8878.0 and 1357.0 µmol/kg were found for cis-
cyclo(L-Val-L-Pro) and cis-cyclo(L-Ala-L-Pro), followed by cis-
cyclo(L-Val-L-Leu), cis-cyclo(L-Leu-L-Pro), cis-cyclo(L-Ala-L-
Leu), cis-cyclo(L-Ala-L-Ile), cis-cyclo(L-Ala-L-Val), and cis-
cyclo(L-Ile-L-Pro) with concentrations between 537.0 and 817.1
µmol/kg. In contrast, just very low concentrations of 0.7 and
1.6 µmol/kg were determined for cis-cyclo(L-Asp-L-Phe) and
cis-cyclo(L-Ala-L-Tyr) (Table 3).
To gain first insights into the taste contribution of these
compounds, these were rated in their sensory impact based on
the ratio of the concentration and the taste recognition threshold
of a compound. Calculation of these DoT factors (25) revealed
that exclusively the concentrations of cis-cyclo(L-Pro-L-Val)
(10), cis-cyclo(L-Val-L-Leu) (26), cis-cyclo(L-Ala-L-Ile) (24),
cis-cyclo(L-Ala-L-Leu) (18), and cis-cyclo(L-Ile-L-Pro) (29) in
the roasted cocoa nibs exceeded their taste threshold concentra-
tions by a factor of 6.9, 1.7, 1.2, and 1.1, respectively (Table
3). All of the other diketopiperazines were present below their
individual bitter taste threshold concentrations. Because the DoT
factors do not consider the nonlinear relationship between the
concentration and gustatory response of an individual compound
as well the maximum solubility of such a compound, in the
following, dose/response functions describing the human bitter
taste perception of diketopiperazines were recorded for some
selected compounds.
Human Dose/Response Functions. As independent from
their sensory training status, panellists have difficulties in
memorizing the intensity of a taste compound for a longer period
of time; they are known to give different ratings for the same
solution of the test compound tasted at different time intervals
(26). Consequently, recording dose/response functions with
standard sensory methodologies usually leads to unreliable
curves with very high error margins. In contrast, the recently
reported half-tongue testing (6, 8) offers the possibility of a
direct comparison of the taste impact of two samples, thus
overcoming the memory problem mentioned above. In this test,
each compound is applied onto one side of the tongue in random
order and the panellist is asked to determine which side shows
the stronger sensation. To adopt this technique to the recording
of dose/response functions, serial 1:1 dilutions of the samples
in water were prepared starting at the solution of maximum
solubility of the individual compound and ending at the
concentration level two steps below the individual recognition
threshold concentration. To fit the dose/response functions into
a five-point intensity scale, first, the taste intensity of the
individual compounds was compared at the highest concentra-
tion level by means of the half-tongue tasting method, thus
offering a direct comparison of the taste impact and a reliable
evaluation of the gustatory response of different compounds.
On this five-point scale, the compound evaluated with the
highest taste intensity at its maximum concentration, that is cis-
cyclo(L-Val-L-Pro) (10) at its maximum solubility of 81.92
mmol/L (water), was set to the maximum score of 5.0. After
the taste intensity of each compound at its maximum solubility
had been rated, the taste intensities of the other dilutions were

7230 J. Agric. Food Chem., Vol. 53, No. 18, 2005
Stark and Hofmann
topiperazines follows rather different dose/response functions.
In particular, the perception of diketopiperazines cis-cyclo(L-
Val-L-Pro), cis-cyclo(L-Leu-L-Pro), and cis-cyclo(L-Ile-L-Pro),
consisting of L-proline and a second hydrophobic amino acid
moiety, is reflected in rather high slopes and high taste intensities
at higher concentration levels. The highest taste intensity of 5.0
was found for an aqueous solution of cis-cyclo(L-Val-L-Pro),
exceeding the threshold concentration by 64-fold. Also, cis-
cyclo(L-Leu-L-Pro) and cis-cyclo(L-Ile-L-Pro) reached a high
maximum bioresponse with scores of 4.6 and 4.2; a further
increase of taste activity was prevented by the limited solubility
of these compounds. In comparison, the cis-cyclo(L-Ala-L-Pro)
containing the less hydrophobic amino acid L-alanine did not
reach the same taste intensity as found for cis-cyclo(L-Leu-L-
Pro) or cis-cyclo(L-Ile-L-Pro) and was just perceived with an
intensity score of 2.3 in 32-fold threshold concentration. It is
interesting to note that all of the tested diketopiperazines lacking
proline in the molecule were very limited in solubility; for
example, cis-cyclo(L-Leu-L-Phe) or cis-cyclo(L-Leu-L-Val) bear-
ing a phenylalanine or a valine moiety instead of the proline
showed maximum solubility already at the taste threshold
concentration of 190 µmol/L (9) or after exceeding the threshold
concentrations just by a factor of 8.
Table 3. Concentrations and DoT Factors of 2,5-Diketopiperazines
Identified in Roasted Cocoa Nibs
concentrationa in
DoT factorsb
compound (number)
(mg/kg)
(µ
mol/kg)
for bitter taste
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
L
L
L
L
L
L
L
L
L
L
L
L
L
L
L
-Val-
-Val-
-Ala-
-Ala-
L
L
L
L
-Pro) (10)
-Leu) (26)
-IIe) (24)
1742.3
173.5
117.8
135.8
112.9
228.2
149.6
107.8
12.5
16.0
47.1
14.3
15.7
15.7
2.3
8877.8
817.1
639.5
734.0
537.0
1357.0
711.2
633.5
47.9
61.5
237.6
58.0
72.0
64.3
8.8
11.3
14.6
22.9
32.5
4.0
0.7
1.6
3.0
2.1
6.9
1.7
1.2
1.1
1.1
0.9
0.6
0.4
0.3
0.3
0.2
0.1
0.1
-Leu) (18)
-IIe-
-Ala-
-Leu-
L
-Pro) (29)
-Pro) (20)
L
L
-Pro) (3)
-Ala-
L
-Val) (21)
L
-Leu-
-Phe) (9)
-IIe-
L
-Phe) (28)
-Val-
-Val-
-Ala-
L
L
L
-Val) (12)
-Phe) (23)
-Phe) (19)
-Phe-
-Val-
-Phe) (4)
-Asn- -Phe) (22)
L
-Pro) (2)
0.1
L
-Tyr) (27)
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
<0.1
cis-cyclo(Gly-
L
2.3
3.8
4.5
6.3
1.0
0.2
0.4
0.7
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
cis-cyclo(
L
L
L
L
L
L
L
L
-Pro-
-Pro-
-Pro-
-Asp-
-Ala-
L
-Thr) (7)
-Pro) (30)
-Tyr) (8)
L
L
L
-Phe) (5)
It appears that the detection threshold concentration of a taste
compound alone does not enable a reliable evaluation of the
taste impact of a compound. This value just considers the very
beginning of the dose/response function, but there seems to be
no correlation between the threshold and the maximal gustatory
activation of a taste compound.
L
-Tyr) (25)
L
-Phe-
-Ser) (11)
cis-cyclo(Gly-
L-Pro) (1)
0.3
<0.2
cis-cyclo( -Leu-Gly) (6)
L
<1.2
a The quantitative data are given as the mean of triplicates. The standard
deviation of the quantitative data for the individual diketopiperazines was less than
10%. b DoT factors are calculated as the ratio of the concentration and the taste
recognition threshold of a compound.
Another advantage of recording these dose/response functions
is that the gustatory response at the factor-over-threshold given
on the x axis in the dose/response functions reflects the taste
intensity of a compound at the DoT factor calculated on the
basis of the threshold concentration and the concentration of
an individual compound in the cocoa. This opens the possibility
to reliably elucidate the taste impact of each individual
compound at the “natural” concentration level in cocoa. For
example, the cis-cyclo(L-Val-L-Pro) calculated with a DoT factor
of 6.9 exhibits a taste intensity of 1.8, whereas at the DoT factors
of 1.7, 1.2, or 1.1, the diketopiperazines cis-cyclo(L-Val-L-Leu),
cis-cyclo(L-Ala-L-Ile), and cis-cyclo(L-Ile-L-Pro) just reach taste
intensity scores of about 0.5. On the basis of these results, it
can be concluded that, out of the group of diketopiperazines,
the cis-cyclo(L-Val-L-Pro) derivative is the most important bitter
compound contributing to the bitter taste of roasted cocoa.
With the aim at demonstrating the contribution of individual
compounds to the overall taste of roasted cocoa nibs, taste
reconstruction and omission experiments using these compounds
in their “natural” concentrations as well as the recording of
additional dose/response functions for other cocoa taste com-
pounds are currently in progress and will be published else-
where.
Figure 6. Human dose/response functions of selected diketopiperazines.
determined by using the half-tongue tasting method so that one
dilution of an individual compound was rated against the
intensity of another dilution of the same compound and by cross-
checking the taste intensity between the different compounds
at the same dilution. Although this methodology is laborious,
it gives much more reliable results with much smaller error
margins than classical sensory tools used to measure dose/
response curves. For example, the intensity values between
trained individuals and separate sessions did not differ more
than (0.3 units.
Using this methodology, dose/response functions were re-
corded for cis-cyclo(L-Val-L-Pro) (10), cis-cyclo(L-Val-L-Leu)
(26), cis-cyclo(L-Leu-L-Phe) (9), cis-cyclo(L-Ala-L-Ile) (24), cis-
cyclo(L-Ala-L-Pro) (20), and cis-cyclo(L-Ile-L-Pro) (29), and cis-
cyclo(L-Leu-L-Pro) (3). The results, outlined in Figure 6, clearly
demonstrated that the gustatory response for different dike-
LITERATURE CITED
(1) Ney, K. H. Cocoa aroma: Bitter compounds as important taste
ingredients (in German). Gordian 1986, 5, 84-88.
(2) Matissek, R. Evaluation of xanthine derivates in chocolates
Nutritional and chemical aspects. Z. Lebensm.-Unters. Forsch.
1997, 205, 175-184.
(3) Pickenhagen, W.; Dietrich, P.; Keil, B.; Polonsky, J.; Nouaille,
F.; Lederer, E. Identification of the bitter principle of cocoa.
HelV. Chim. Acta 1975, 58, 1078-1086.
(4) Bonvehi, S. J.; Coll, V. F. Evaluation of purine alkaloids and
diketopiperazines contents in processed cocoa powder. Eur. Food
Res. Technol. 2000, 210, 189-195.

Metallic- and Bitter-Tasting Diketopiperazines in Roasted Cocoa Nibs
J. Agric. Food Chem., Vol. 53, No. 18, 2005 7231
(5) van der Greef, J.; Tas, A. C.; Nijssen, L. M.; Jetten, J.; Ho¨hn,
M. Identification and quantification of diketopiperazines by liquid
chromatography-mass spectrometry, using a moving belt in-
terface. J. Chromatogr. 1987, 394, 77-88.
(6) Stark, T.; Bareuther, S.; Hofmann, T. Sensory-guided decom-
position of roasted cocoa nibs (Theobroma cocoa) and structure
determination of taste-active polyphenols. J. Agric. Food Chem.
2005, 53, 5407-5418.
(7) Stark, T.; Hofmann, T. Isolation, structure determination,
synthesis, and sensory activity of N-phenylpropenoyl-L-amino
acids from cocoa (Theobroma cacao). J. Agric. Food Chem.
2005, 53, 5419-5428.
(8) Scharbert, S.; Holzmann, N.; Hofmann, T. Identification of the
astringent taste compounds in black tea infusions by combining
instrumental analysis and human bioresponse. J. Agric. Food
Chem. 2004, 52, 3498-3508.
cheese by liquid carbon dioxide. J. Food Sci. 1993, 58, 1005-
1009.
(18) Eriksen, S. A.; Fagerson, I. S. Nonvolatile nitrogen compounds
in hydrolyzed vegetable protein. In Cereals for Food and
BeVerages: Recent Progress in Cereal Chemistry and Technol-
ogy; Inglett, G. E., Munck, L., Eds.; Academic Press: New York,
1980, pp 395-408.
(19) Kopple, K. D.; Marr, H. H. Conformations of cyclic peptides.
The folding of cyclic peptides containing aromatic side chain.
J. Am. Chem. Soc. 1967, 89, 6193-6200.
(20) Matoba, T.; Hata, T. Relationship between bitterness of peptides
and their chemical structures. Agric. Biol. Chem. 1972, 36,
1423-1431.
(21) Ishibashi, N.; Kouge, K.; Shinoda, I.; Kanehisa, H.; Okai, H.
Bitterness of phenylalanine- and tyrosine-containing peptides.
Agric. Biol. Chem. 1987, 51, 3309-3313.
(9) Rachele, J. R. The methyl esterification of amino acids with 2,2-
dimethoxypropane and aqueous hydrogen chloride. J. Org. Chem.
1963, 28, 2898.
(22) Ishibashi, N.; Kouge, K.; Shinoda, I.; Kanehisa, H.; Okai, H.
Bitterness of leucine-containing peptides. Agric. Biol. Chem.
1987, 51, 2389-2394.
(10) Chen, Y.-H.; Liou, S.-E.; Chen, C.-C. Two-step mass spectro-
metric approach for the identification of diketopiperazines in
chicken essence. Eur. Food Res. Technol. 2004, 218, 589-597.
(11) Rizzi, G. P. Heat-induced flavor formation from peptides. In
Thermal Generation of Aromas; Parliament, T. H., McGorrin,
R. J., Ho, C. T., Eds.; ACS Symposium Series 409; American
Chemical Society, Washington, DC, 1989, pp 172-181.
(12) Sakamura, S.; Furukawa, K.; Kasai, T. Bitter diketopiperazines
in roasted malts for beer brewing. Agric. Biol. Chem. 1978, 42,
607-612.
(13) Takahashi, K.; Tadenuma, M.; Kitamoto, K. L-Prolyl-L-leucine
anhydride a bitter compound formed in aged sake. Agric. Biol.
Chem. 1974, 38, 927-932.
(14) Gautschi, M.; Schmid, J. P. Chemical characterization of
diketopiperazines in beer. J. Agric. Food Chem. 1997, 45, 3183-
3189.
(23) Ishibashi, N.; Kouge, K.; Shinoda, I.; Kanehisa, H.; Okai, H.
Role of the hydrophobic amino acid residue in the bitterness of
peptides. Agric. Biol. Chem. 1988, 52, 91-94.
(24) Wieser, H.; Belitz, H.-D. Studies on the structure/activity
relationship of bitter tasting amino acids and peptides. Z.
Lebensm.-Unters. Forsch. 1976, 160, 383-392.
(25) Scharbert, S.; Hofmann, T. Molecular definition of black tea taste
by means of quantitative studies, taste reconstitution, and
omission experiments. J. Agric. Food Chem. 2005, 53, 5377-
5384.
(26) Galopin, C.; Tigani, L., Furrer, S. Structure property relationship
of chemosensates. In State-of-the-Art in FlaVour Chemistry and
Biology; Hofmann, T., Rothe, M., Schieberle, P., Eds.; Proceed-
ings of the 7th Wartburg Symposium: Eisenach, Germany, 2005;
pp 61-67.
(15) Ginz, M.; Engelhardt, U. H. Identification of proline-based
diketopiperazines in roasted coffee. J. Agric. Food Chem. 2000,
48, 3528-3532.
(16) Ginz, M.; Engelhardt, U. H. Identification of new diketopipera-
zines in roasted coffee. Eur. Food Res. Technol. 2001, 213,
8-11.
Received for review June 3, 2005. Revised manuscript received July
15, 2005. Accepted July 15, 2005. This research project was supported
by a grant from FEI (Forschungskreis der Erna1hrungsindustrie
e.V., Bonn, Germany), the AiF (Arbeitsgemeinschaft industrieller
Forschung), and the German Ministry of Economics (project number
13235 N).
(17) Roudot-Agaron, F.; Le Barbs, D.; Einhorn, J.; Adda, J.; Gripon,
J. C. Flavor constituents of aqueous fraction extracted from comte
JF051313M