Identification of enterodiol as a masker for caffeine bitterness by using a pharmacophore model based on structural analogues of homoeriodictyol

Article abstract of DOI:10.1021/jf301335z

Starting from previous structure-activity relationship studies of taste modifiers based on homoeriodictyol, dihydrochalcones, deoxybenzoins, and trans-3-hydroxyflavones as obvious analogues were investigated for their masking effect against caffeine. The most active compounds of the newly investigated taste modifiers were phloretin, the related dihydrochalcones 3-methoxy-2′,4,4′-trihydroxydihydrochalcone and 2′,4- dihydroxy-3-methoxydihydrochalcone, and the deoxybenzoin 2-(4-hydroxy-3- methoxyphenyl)-1-(4-hydroxyphenyl)ethanone. Starting with the whole set of compounds showing activity >22%, a (Q)SAR pharmacophore model for maskers of caffeine bitterness was calculated to explain the structural requirements. After docking of the pharmacophore into a structural model of the broadly tuned bitter receptor hTAS2R10 and docking of enterolactone and enterodiol as only very weakly related structures, it was possible to predict qualitatively their modulating activity. Enterodiol (25 mg L^-1) reduced the bitterness of the 500 mg L^-1 caffeine solution by about 30%, whereas enterolactone showed no masking but a slight bitter-enhancing effect.

Full text of DOI:10.1021/jf301335z

Article
pubs.acs.org/JAFC
Identification of Enterodiol as a Masker for Caffeine Bitterness by
Using a Pharmacophore Model Based on Structural Analogues of
Homoeriodictyol
Jakob P. Ley,*,† Marco Dessoy,‡ Susanne Paetz,† Maria Blings,† Petra Hoffmann-Lucke,†
̈
Katharina V. Reichelt,† Gerhard E. Krammer,† Silke Pienkny,‡ Wolfgang Brandt,‡
and Ludger Wessjohann‡
†Symrise AG, Flavor & Nutrition, Research & Technology, P.O. Box 1253, D-37601 Holzminden, Germany
‡Leibniz Institute of Plant Biochemistry, Weinberg 3, D-06120 Halle, Germany
S
* Supporting Information
ABSTRACT: Starting from previous structure−activity relationship studies of taste modifiers based on homoeriodictyol,
dihydrochalcones, deoxybenzoins, and trans-3-hydroxyflavones as obvious analogues were investigated for their masking effect
against caffeine. The most active compounds of the newly investigated taste modifiers were phloretin, the related
dihydrochalcones 3-methoxy-2′,4,4′-trihydroxydihydrochalcone and 2′,4-dihydroxy-3-methoxydihydrochalcone, and the deoxy-
benzoin 2-(4-hydroxy-3-methoxyphenyl)-1-(4-hydroxyphenyl)ethanone. Starting with the whole set of compounds showing
activity >22%, a (Q)SAR pharmacophore model for maskers of caffeine bitterness was calculated to explain the structural
requirements. After docking of the pharmacophore into a structural model of the broadly tuned bitter receptor hTAS2R10 and
docking of enterolactone and enterodiol as only very weakly related structures, it was possible to predict qualitatively their
modulating activity. Enterodiol (25 mg L−1) reduced the bitterness of the 500 mg L−1 caffeine solution by about 30%, whereas
enterolactone showed no masking but a slight bitter-enhancing effect.
KEYWORDS: flavor modifiers, bitter masking, dihydrochalcones, deoxybenzoins, trans-3-hydroxyflavanones, enterodiol
Figure 2), and their shorter chain analogues, the deoxybenzoins
(Figure 3), and some 3-hydroxyflavanones (dihydroflavonols,
Figure 4) were prepared and tested for their masking ability. In
a second step, a pharmacophore model was calculated based on
structure−activity relationships using the reduction of caffeine
bitterness. In a final step, the nonobvious mammalian lignan
metabolites enterodiol and enterolactone were tested in the
pharmacophore model as well as in sensory tests for their
caffeine bitterness modulating activity.
INTRODUCTION
■
Because of the increasing importance of healthier products,
sometimes, ingredients are added that can improve the health
status of people but show deficits in taste. Another challenge for
the taste of food products is the reduction of some common
ingredients such as sucrose or other bulk carbohydrates to
lower the caloric intake. Unfortunately, in most cases, the flavor
of the products is sacrificed; therefore, a huge demand for flavor
modifiers showing activity as off-taste maskers1 or sweet taste
enhancers2 has developed during the past 10 years. Some
potent new bitter masking molecules such as the 1-
carboxymethyl-5-hydroxy-2-hydroxymethylpyridinium inner
salt,2,3 C-glycosides of catechins against bitterness of
proanthocyanidins,4 some azo-dyes,5 and 4-(2,2,3-
trimethylcyclopentyl)butanoic acid6 against bitterness of sweet-
eners and some derivatives related to the flavanone
homoeriodictyol (HED, 1, Figure 1)7 were developed:
hydroxybenzoic acid vanillylamides8 and short chain ginger-
diones related to hispolone.9 During earlier studies, the vanillyl
group was identified as a common structural element for a
certain masking activity.
MATERIALS AND METHODS
■
Phloretin, naringenin-7-O-glucosid, and phloridzin [each >95%, high-
performance liquid chromatography (HPLC)] were from Kaden
Biochemicals (Hamburg), ampelopsin was from Changsha Sunfull
Biotech Co., Ltd. (China), 2-(3,4-dihydroxyphenyl)-1-(2,4-dihydrox-
yphenyl)-ethanon was from Toronto Chemicals (Ontario, Canada),
and enterodiol and enterolactone were from Phytolab (Verstenbergs-
greuth, Germany). All other test compounds were synthesized
according to principally known procedures as outlined in the
Supporting Information.
Syntheses. 2-(4-Hydroxy-3-methoxyphenyl)-1-(4-
hydroxyphenyl)ethanone (18). Guajacol (2-methoxyphenol, catechol
monomethylether, 2.0 g, 16.1 mmol) and 4-hydroxyphenylacetic acid
(2,4 g, 16.1 mmol) were cooled to 0 °C, and 35 mL (322 mmol)
boron trifluoride etherate was added. The mixture was slowly heated
The previously published structure−activity concepts7,8
starting from HED (1), eriodictyol (2), and naringenin (3)
(Figure 1) were extended to evaluate the potential to predict
the bitter masking activity of nonobviously related natural
products against caffeine by using a pharmacophore model.
First of all, some other more obvious structural analogues of the
parent flavanones, namely, the related dihydrochalcones (see
Received: March 28, 2012
Revised: June 4, 2012
Accepted: June 7, 2012
Published: June 7, 2012
© 2012 American Chemical Society
6303
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
Figure 1. Investigated structural classes related to HED (1), eriodictyol (2), and naringenin (3).
Figure 2. Investigated dihydrochalcones 4−13 [NHDC (4), phloridzin (5), trilobatin (6), phloretin (7), and davidigenin (8)].
to 90 °C for 13 h (TLC control CHCl3/MeOH 10:1, Rf = 0.4). BF3-
etherate was distilled, and the cooled residue was quenched with
water/ethyl acetate (100 mL, 1:1) under vigorous stirring. The organic
phase was separated, and the water phase was extracted again with 50
mL of ethyl acetate. The combined organic phases were washed with
NaHCO3 solution and brine, and dried over Na2SO4 concentrated,
keeping the product in solution. After treatment with ethyl ether, the
white product precipitated as a white solid (purity by H NMR, 97%),
which was recrystallized from ethyl acetate. Yield, 59 mg (14%).
HPLC-MS (RP-18, APCI-): m/z = 257.38 (100%, [M − H]−). HRMS
(ESI−, M − H) calcd for C15H13O4, 257.0819; found, 257.0818; (ESI−,
NMR (100 MHz; DMSO-d6, internal standard TMS): δ = 196.1 (C,
C-α), 155.8 (C, C-4′), 151.6 (C, C-4), 147.4 (C, C-3), 130.2 (2 × CH,
C-2′, C-6′), 128.0 (C, C-1), 125.6 (C, C-1′), 123.5 (CH, C-6), 115.0 (2
× CH, C-3′, C-5′), 114.8 (CH, C-5), 111.5 (CH, C-2), 55.5 (CH3, O−
CH3), 43.3 (CH2, C-β) ppm.
2-(4-Hydroxyphenyl)-1-(2,4,6-trihydroxyphenyl)ethanone (20). 4-
Hydroxybenzylcyanide (2.60 g, 20 mmol) and 1,3,5-trihydroxybenzol
(3.78 g, 30 mmol) were mixed with 1,4-dioxan (30 mL) and treated
with dry HCl for 5 h. The mixture was stirred at room temperature for
another 16 h. The precipitate was filtered off, and the raw crystalline
material (6.34 g) was diluted with ice water (15 mL) and refluxed for 1
h. After it was cooled, the precipitate was filtered off, washed with
water, and dried in vacuo. Yield, 3.0 g (colorless crystals, 11.5 mmol,
58% of theory). HPLC-MS (RP-18, APCI-): m/z = 259.44 (100%, [M
− H]−), 518.88 (14%, [2M − H]−). HRMS (ESI−, M − H) calcd for
C14H11O5, 259.0612; found, 259.0623; (ESI+, M + H) calcd for
1
M − H) calcd for C15H15O4, 259.0965; found, 259.0976. H NMR
(400 MHz, DMSO-d6, internal standard TMS): δ = 9.97 (1H, bs,
OH), 9.24 (1H, bs, OH), 7.60 (1H, dd, J = 8.3 Hz, J = 2.0 Hz, H-2),
7.48 (1H, d, J = 2.0 Hz, H-6), 7.05 (2H, m, H-2′, H-6′), 6.85 (1H, d, J
= 8.3 Hz, H-3), 6.68 (2H, m, H-3′, H-5′), 4.12 (2H, s, H-a) ppm. 13C
6304
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
Figure 3. Investigated deoxybenzoins 14−21.
Figure 4. Structures of trans-3-O-hydroxyflavanones 22−26 [blumeatin A (23) and ampelopsin (26)].
1
C14H13O5, 261.0757; found, 261.0762. H NMR (400 MHz, DMSO-
d6, internal standard TMS): δ = 12.23 (ca. 2H, s, OH), 10.37 (1H, s,
OH), 9.19 (1H, s, OH), 7.01 (2H, m, H-2′, H-6′), 6.67 (2H, m, H-3′,
H-5′), 5.81 (2H, s, H-3, H-5), 4.21 (2H, s, H-α) ppm. 13C NMR (100
MHz; DMSO-d6, internal standard TMS): δ = 202.9 (C, CO),
164.7 (C, C-4), 164.1 (2 × C, C-2, C-6), 155.7 (C, C-4′), 130.4 (2 ×
CH, C-2′, C-6′), 125.8 (C, C-1′), 114.8 (2 × CH, C-3′, C-5′), 103.5 (C,
C-1), 94.6 (2 × CH, C-3, C-5), 47.9 (CH2, C-α) ppm.
chloride (1.5 and 1.8 g L−1), sucrose (8.0 and 10.0 g L−1), and
monosodium glutamate (3.6 and 4.4 g L−1) using paired comparison
test.
For bitter modulating tests, the panelists had to rate the intensity
(I) of the bitterness on a scale of 1 (nothing) to 10 (extremely strong).
The mean rating of 500 kg L−1 caffeine solution ranged between 5 and
6. For other bitter tastants (quinine, KCl, and salicin), concentrations
were used at which all panelists can perceive a pronounced bitterness
(mean ratings about 4−6). Modulation effects (MEs, as %) were
calculated relative to the blind probe using the equation:
Sensory Studies. Bitter taste modulating studies were done using a
duo-comparison method with a panel of trained healthy adults (7
males and 15 females; ages 25−48 years) without any reported taste
disorder. Tasting sessions were carried out in the morning hours 1−2
h after breakfast, during which time the panelists were asked not to
drink black or green tea or coffee due to adaptation to caffeine. Duo
tests were presented to the testers in a randomized order; in case of
discoloration, samples were covered with aluminum foil or tasted
under colored light. Panel members were trained to evaluate the taste
of aqueous solutions (5 mL each) of the following standard taste
compounds: citric acid (0.4 g L−1) for sour taste, caffeine (0.3 g L−1)
for bitter taste, sodium chloride (1.3 g L−1) for salty taste, sucrose (6.0
g L−1) for sweet taste, and monosodium glutamate (0.4 g L−1) for
umami taste. Difference tests were carried out for the basic tastes: citric
acid (0.5 and 0.6 g L−1), caffeine (0.25 and 0.35 g L−1), sodium
ME = (Itest − Iblind)/Iblind
Negative ME values were interpreted as inhibitory, and positive values
were interpreted as enhancing effects. Statistical analysis was
performed using internal functions of MS Excel 97 (Microsoft
Corp.). Student's t test (double-sided, paired) was used to calculate
statistical significance.
Molecular Modeling. All 3D structures of the compounds were
constructed with MOE10 and were subsequently minimized with the
Merck MMFF94 force field.11−15 Twenty-one compounds with a
bitter masking activity higher than 22% (see Table 1) were considered
as active and used for three-dimensional superposition. For this
6305
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
ethylate solution and subsequent evaporation of the solvents.
The dihydrochalcones 9−13 were synthesized starting with
Knoevenagel condensation of the acetophenons with the
appropriate benzaldehydes and subsequent hydrogenolysis of
the intermediate chalcones. Deoxybenzoins are still much more
rarely reported as natural compounds, for example, deoxy-
vanilloin (14) as a trace compound in the root wood of
Melicope semecarpifolia.26 The commercially unavailable
deoxybenzoins 15−21 were synthesized according to the
well-known BF3·Et2O-catalyzed Friedel−Crafts acylation of
phenols with the appropriate phenylacetic acids.27,28 In the case
of the tetrahydroxylated deoxybenzoin 14, the alternative route
via the corresponding phenyl acetonitrile facilitated by the use
of dry HCl was much more successful.29 The deoxybenzoins 18
and 21 were never described in the literature. The investigated
trans-3-hydroxyflavanones are all known from the literature.
Taxifolin (24) is a well-known natural compound, and
blumeatin A (23) was described in Blumea balsamiferia.30
The vanilloid isomer 22 occurs as trace component (5 mg/kg)
besides HED (1) in Lychnophora granmongolense (“Brazil
arnica”),31 whereas the isomer 25, which is related to the
flavanone sterubin, was also found in Blumea balsamiferia.32
Dihydromyricetine or ampelopsin (26) was found in
Ampelopsis cantoniensis.33 The synthesis of the racemic
compounds was performed starting with perbenzylation of
the parent flavanones HED (1), hesperetin, or sterubin,
subsequent epoxidation in alkaline medium using hydrogen
peroxide, and finally hydrogenation to the corresponding trans-
3-hydroxyflavanones 22, 23, and 25.
Table 1. Evaluation of MEs for Bitter Taste of
Dihydrochalcones as Compared to HED (1), Eriodictyol
a
(2), and Naringenin (3) against 500 mg L−1 Caffeine
bitter modulation 500 mg L−1
caffeine
panelists all/
masking
ME for bitter
taste
profile (100 mg L−1 in 5%
compd
sucrose)
b
1
10/10
−43%*
−46%**
−9%
−8%
+7%
weak, sweet, vanillic, phenolic
neutral
2
3
4
16/12
9/5
dry-dusty, fatty, creamy
neutral, long lasting sweetness
16/8 (3 mg L−1
)
16/6 (30 mg L−1
)
5
6
16/4
+10%
+21%
−29%*
−6%
bitter
16/2 (50 mg L−1
16/3 (50 mg L−1
16/9
)
)
vanillin, honey, fruity, weak
sweet, mouth feel, dusty
balsamic, weak, bitter, licorice
weak, fruity, bitter
7
8
9
15/7
0%
10
16/8
−9%
weak, smooth, balsamic,
smoky
11
15/11 (50
−24%
neutral
mg L−1
)
12
13
16/9
15/12 (50
mg L−1
−7%
−27%*
weak, fruity, herbal, dusty
neutral
)
a
Test concentration, 100 mg L−1; *, significant (p < 0.05); **, p <
b
0.005; ND, not determined. Ref 7.
purpose, the “pharmacophore elucidator” module of MOE was used.
In this process, all single bonds were allowed to be rotatable, and
pharmacophores like hydrogen bond donor and acceptor sites as well
as hydrophobic areas were applied in the unified scheme of MOE. On
the basis of these calculations and superposition of all active structures,
a pharmacophore model was derived manually.
Sensory Studies. All compounds were first tested in 5%
sucrose and 0.5% sodium chloride solution for the general taste
profile (for screening data, see Tables 1−3). Only in the
A homology model of the human hTas2R10 based on the X-ray
structure 1U1016 (bovine rhodopsine) has been taken from the model
database MODBASE17 (Q9NYW0, model original ID: EN
ENSP00000240619). Because analysis with PROCHECK18 and
PROSA II19 indicated a not perfectly folded 3D structure, the protein
model has been improved by a molecular dynamics simulation (50 ns,
including a membrane composed of PEA in a water box) with
YASARA20,21 using the force field AMBER0322 and the md-
runmembrane protocol of YASARA. Here, the conformation of
extracellular loop 2 has changed most in comparison to the original
MODBASE model. Overall, the quality of the model has improved
according to the evaluation with PROCHECK (amino acids in most
favored region from 85.2 to 85.5%, two outliers) and the z score of the
pair potential of Cα and Cβ of PROSA II (from 0.53 to −1.83).
The ligands enterodiol and enterolactone have been docked into the
active site of the hTas2R10 model using GOLD23 with the Nε of Trp
88 as the center of a sphere with a radius of 12 Å. The used Fitness
Function was GoldScore.
Table 2. Evaluation of MEs for Bitter Taste of
a
Deoxybenzoins against 500 mg L−1 Caffeine
bitter modulation 500 mg L−1
caffeine
panelists all/
masking
ME for bitter
taste
profile (100 mg L−1 in 5%
compd
sucrose)
14
16/10
−23%
vanillin, spicy, clove, woody,
balsamic
15
16
17
18
15/7
−10%
−9%
−15%
neutral
14/8
sweet, dry-dusty, balsamic
neutral
14/10
15/12
−29%*
vanillin, phenolic, balsamic,
somewhat bitter
(50 mg L−1
)
(p < 0.06)
19
20
21
15/9
−18%
vanillin, balsamic, woody
very bitter, astringent
vanillin, phenolic, sweet
ND
16/10
−8%
(50 mg L−1
)
a
Test concentration, 100 mg L−1; *, significant (p < 0.05); ND, not
RESULTS AND DISCUSSION
■
determined.
Syntheses. Most of the investigated compounds were not
commercially available and were therefore synthesized by
principally known procedures. With the exception of bitter-
sweet phloretin glycosides from apple (Malus ssp.), dihy-
drochalcones were only rarely found in nature, for example,
davidigenin (9) as a trace in Artemisia dracunculus.24 The
naturally occurring trilobatin (6) (e.g., isolated from Malus
trilobata, Vitis ssp., and Symplococus ssp.25) was isolated as a
side product from a mild acidic hydrolysis of naringin
dihydrochalcone. The spectral data of the product correspond
to the published data from literature. Phloretin sodium salt was
prepared by exchange of protons by using 1 equiv of sodium
absence of strong off-tastes, the compounds were tested for
their bitter modulation activity. Caffeine was selected as a
screening vehicle for bitter masking effects due to its activation
spectrum of various bitter receptors including the broadly
tuned hTAS2R10,34 acceptable taste, and low toxicity. From
previous studies, it is known that HED (1) and its structural
relatives depicted in Figure 1 show a comparable pattern of
activity against various bitter compounds, for example, quinine,
which is similar to caffeine, can activate several human bitter
receptors including hTAS2R10.34 Until now, the mechanism
6306
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
for pharmaceutical actives has been known for a long time,35
but the strong sweet taste at higher concentrations often cannot
be tolerated in all applications.
Table 3. Evaluation of MEs for Bitter Taste of trans-3-
a
Hydroxyflavanones against 500 mg L−1 Caffeine
bitter modulation 500 mg L−1
The most promising bitter taste modulators 7, 11, 13, and 18
for caffeine were tested in more detail. First of all, dose−
response plots were determined for caffeine with different
concentrations of the test compounds (Figure 5). All four
components showed an increasing dose response up to 50 mg
L−1, but then, the activity decreased somewhat at 100 mg L−1;
probably, the masking activity is competed by arising intrinsic
bitterness at higher concentrations (as found, for example, for
compound 18 at 100 mg L−1), which may compete with their
masking effect. Because of the number of 25 different hTas2R
bitter receptors in human,34,36 this might be an overlapping
effect of antagonistic and agonistic activity as described for
certain sesquiterpene lactones37 for different receptor types,
which should be investigated in the future.
A subset of the antagonistic pattern of the modulators was
investigated as shown in Figure 6; modulating activities lower
than 10% are not considered for the discussion. Again,
phloretin (7) and the dihydrochalcone 11 showed a non-
significant activity against quinine, whereas the remaining
candidates were not active at the tested concentrations. Only 7
was weakly active as a masker for salicin, whereas the activity of
all compounds against KCl or peptides was not detectable. As a
summary, phloretin (7) seems to be a relatively broadly active
bitter taste reducer comparable to HED (1), whereas the other
compounds found in the screening seem to be active only
against caffeine. However, in contrast to 1, phloretin seems to
be limited at a higher dosage due to its own bitter taste at these
concentrations.
caffeine
panelists all/
masking
ME for bitter
taste
profile (100 mg L−1 in 5%
compd
sucrose)
22
23
15/6
16/9
(50 mg L−1
+15%
fatty, oily
−14%
gujacol, vanillin, sweet
)
24
25
14/4
ND
+18%
very bitter
herbal, bitter, phenolic, woody,
balsamic
26
14/4
+16%
dry dusty, grape, grape seed
a
Test concentration, 100 mg L−1; *, significant (p < 0.05); ND, not
determined.
behind the activity of 1 and its analogues has not been known,
but they seem to block at least one of these bitter receptors,
which are activated by a larger set of bitter molecules.
The best bitter taste reducers for caffeine were the
dihydrochalcones phloretin (7), 11, and 13 and the
deoxybenzoin 18. Whereas 11, 13, and 18 show the vanilloyl
moiety similar to 1, the activity of phloretin is somewhat
surprising. In earlier studies, the simple para-hydroxy-
substituted derivatives [e.g., naringenin (3), 2,4-dihydrobenzoic
acid 4-hydroxybenzylamide, [2]-gingerdione, Figure 1] did not
show any significant masking activity.7,9,29 The short-chain
analogue 20 is only moderately active as a flavor modifier, and
the glucoside phloridzin (5) induces some additional bitterness.
Unfortunately, from all tested trans-3-hydroxyflavanones, only
23 showed a moderate bitter masking activity toward caffeine.
The other analogues exhibit bitter or even bitter-enhancing
effects in contrast to their corresponding flavanones. Interest-
ingly, NHDC (4) showed no masking activity up to 30 mg L−1
against caffeine bitterness. The use of NHDC as a bitter masker
Structure−Activity Relationships. The concept of
structural variation of the parent HED (1) seems to work in
principle but is not valid for all cases. Therefore, the vanilloyl
moiety is probably not solely responsible for the bitter masking
Figure 5. Dose−response plots of masking activities of phloretin (7) (A), dihydrochalcones 11 (B), 13 (C), and deoxybenzoin 18 (D) in 500 mg
L−1 caffeine solution (n = 15−16). *, significant (p < 0.05).
6307
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
Figure 6. Masking activity of phloretin (7) (A, 30 mg L−1), dihydrochalcones 11 (B, 50 mg L−1), 13 (C, 50 mg L−1), and deoxybenzoin 18 (D, 50
mg L−1) for various bitter compounds (relative expressions).
effects of 1. To study the SAR in a more general sense, a first
simplified computational study based on the active (and
inactive) compounds was done, whereby 22% bitter masking,
that is, ca. half of the HED activity value, was set as a lower
threshold for compounds defined as active.
The best result for the superposition of the structures is
displayed in Figure 7 together with the pharmacophore
Eriodictyol (2) shows the highest activity with a bitter
masking rate of 46% (see Table 1). On the basis of the
pharmacophore model, this high activity can be perfectly
explained by the model (Figure 8A). Except for the
hydrophobic binding site F5, all other areas defined by the
suggested pharmacophore are accessible by a low energy
conformation of the compound. Both hydroxyl groups of ring B
can interact with the proton acceptor feature F4, with the meta-
hydroxyl group being preferred for such an interaction. For
comparison, in HED (1), the methyl group of the 3′-methoxyl
moiety of ring B can interact with the hydrophobic area F5 by
rotating the phenyl group by 180°. This explains the similar
activities but also the differences of 1 and 2. The removal of the
3′-hydroxyl group in ring B (naringenin, 3) will lead to an
almost complete loss of interaction with both the proton
acceptor site F4 and the hydrophobic site F5, resulting in a
considerably reduced activity (9%).
Among the compounds 4−13 (Table 1), only compounds 7
(29%), 11 (24%), and 13 (27%) exhibit significant bitter
masking activity. Because of the conformational flexibility of
these compounds, the ring B is able to adopt a conformation in
which the para-hydroxyl group can service the proton acceptor
site F4 (Figure 8B). Additionally, and in difference to
compounds 1−3, the additional hydroxyl group at ring A in
compound 7 is able to interact with the proton acceptor/donor
site F3 quite nicely. Except for the hydrophobic binding site F5,
all pharmacophores are satisfied, which may explain the rather
high activity of 7 but also of 11 where F5 is occupied instead F3
(Figure 8C).
Figure 7. Superposition of all 21 active compounds and derived
pharmacophores F1−F9. Red spheres (features F2, F4, and F9)
represent proton acceptor sites, blue ones (F1 and F3) represent
either proton acceptor and/or donor sites, and the green spheres (F5,
F6, and F8) represent preferred hydrophobic interaction areas.
The inactivity of compounds 4−6 can be simply explained by
steric overlap especially with binding sites at pharmacophores
F2, F9, or/and F1. The lower activity of compound 10 in
comparison to 11 might be explainable by slightly different
conformational behavior caused by the additional hydroxyl
group in ring A. Whereas the ethylene moiety in 11 is highly
flexible, a steric hindrance with both ortho-hydroxyl groups of
ring A in 10 leads to a more perpendicular orientation of the
(“gustophore”) model, a simplified counterpart of the
mutualized overlay of reasonable conformations of the
structures of active compounds. Herein, red spheres (the
features F2, F4, and F9) represent proton acceptor sites, blue
ones (F1 and F3) either proton acceptor and/or donor sites,
and the green ones (F5, F6, and F8) preferred hydrophobic
interaction areas.
6308
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
Figure 8. Minimized conformation and interaction of pharmacophore model and (A) eriodictyol (2), (B) phloretin (7), (C) dihydrochalcone (11),
(D) compound 10, (E) deoxybenzoin (14), and (F) deoxybenzoin (18). The orientation within the model is inverted with respect to the A and B
ring position as compared to the best fit of derivative 14. (G) Blumeatin A (23) and (H) comparison of the conformations of HED 1 (green) and 3-
hydroxyflavanone 22 (orange carbon atoms).
ethylene group with respect to ring A, which causes difficulties
for ring B to interact especially with pharmacophore F5 (Figure
8D).
when the compound is posed in an opposite orientation, that is,
ring A and ring B exchange their position within the
pharmacophore as compared to all previously discussed
compounds. Then, the hydroxyl and methoxyl groups of ring
A interact with features F4 and F5, and the hydroxyl group of
ring B interacts with either F2 or F9. This hypothesis is
supported by the introduction of an additional methoxyl group
in compound 17, which leads to reduced affinity in comparison
to compound 18 because of the missing hydrophobic feature
close to F2 and/or F9.
The trans-3-hydroxyflavanones (Table 3) adopt a slightly
different conformation in comparison to the flavanones of
Table 1 since the 3-hydroxy group causes some repulsion with
ring B (Figure 8H). In compound 22 (−15%), the 3-hydroxyl
group will cause repulsive interactions with the hydrophobic
pharmacophore F8. An interaction of the meta-methoxyl group
of ring B with the hydrophobic site F5 is also reduced, which
obviously contributes to the inactivity of this compound. This
holds true also for the inactive compounds 24 and 26. The
special importance of site F5 is supported by the recovered
The deoxybenzoins very likely exhibit a slightly different
binding mode in the pharmacophore in comparison to the
flavonoids. Because of the reduced conformational flexibility
(methylene bridge instead ethylene), their ring B can only
interact with F4 or F5 of the pharmacophore, if the carbonyl
group simultaneously interacts with F3 but not with F1. Only in
this way, the activity of 14 (23%, Table 2 and Figure 8E) can be
explained based on the pharmacophore model. Compounds 15
and 16 lack this hydroxyl, which might be a reason for their
reduced activities (10 and 9%, respectively). In compound 21
(8%), the para-methoxyl group of ring A leads to missing or
repulsive interaction with the proton acceptor site F9, whereas
this site is occupied in compound 19 (18%), leading to higher
activity.
Since in 18 a 3-methyl group is introduced in ring A but no
hydrophobic feature exists close to F2 and F9 (Figure 8F), an
optimal superposition with the pharmacophore is only possible,
6309
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
Figure 9. Poses of enterolactone (27, carbon atoms in cyan) and enterodiol (28, carbon atoms in gray) docked in the active site of the hTas2R10
protein model (carbon atoms in yellow) and additional interaction of enterodiol's hydroxyl group with the carbonyl group of Trp 162, which is
missing in enterolactone.
activity of compound 23 (14%) in which the para-methoxyl
group is now able to interact with this pharmacophore feature
(Figure 8G).
as an inhibitor for the bitter receptor, whereas 27 will enforce a
conformational change of the receptor to get a better fit. To
verify the prediction, enterolactone (27) and enterodiol (28)
were also tested for their bitter modulating activity (Table 4) .
Pharmacophore Docking and Prediction of Activity.
In a second step, the mammalian lignans enterolactone (27)
and enterodiol (28, Figure 9) were chosen as related but not
obvious test vehicles for prediction of bitter masking activity.
A protein model of one of the broadly tuned bitter receptors,
hTas2R10, based on the X-ray structure of bovine rhodopsine
in its inactive conformation has been developed. Docking of the
structures of 27 and 28 into the active site of hTas2R10 led to
similar poses for both but for enterodiol (28) without
significant distortions of the protein.
The structures used for the pharmacophore (Figure 7) have
been aligned with the docking poses of both lignans. Now, the
pharmacophore features compare favorably to certain amino
acid side chains of the bitter receptor model. The proton
acceptor features F2 and F9 localize with the side chain of Asn
173 and the backbone of Ser 139. The proton acceptor and
donor features F1 and F3 are in the places of either the side
chains of Gln 93 and Asn 92 or the backbone of Trp 162. The
proton acceptor feature F4 is located near the side chain of Asp
163. Hydrophobic interaction areas as represented by F5−8
find a counterpart in the protein model with the amino acids
Met 243, Ile 247, Leu 259, and Met 263.
The suggested pharmacophore model of bitter masking
compounds derived or related to flavonoids is able to explain
the structure−activity relationships qualitatively. To check the
value of the model, the activities of enterodiol (28) and
enterolactone (27) were predicted. Both structures interact
with the pharmacophore sites F1, F4, and F7. The meta-
hydroxyl group of 28 interacts with the proton acceptor site F2,
whereas the meta-hydroxyl group of 27 interacts with the
proton acceptor site F9. Enterodiol (28) but not enterolactone
(27) is additionally able to interact with the proton acceptor/
donor site F3 and makes contact to the hydrophobic
interaction sites F6 and F8. This indicates that 28 might act
Table 4. Evaluation of MEs for Bitter Taste of Enterolactone
a
(27) and Enterodiol (28) against 500 mg L−1 Caffeine
bitter modulation 500 mg L−1 caffeine
panelists all/
compd
masking
ME for bitter taste
29%
additional descriptors
27
28
14/4
14/4
dull, earthy, mold, soapy
−31% (p = 0.08) somewhat dull, earthy
a
Test concentration, 25 mg L−1; ND, not determined.
Indeed, 28 (25 mg L−1) showed a reasonably good masking
activity against bitterness of a 500 mg L−1 caffeine solution in
the duo comparison, whereas the test solutions containing
caffeine and enterolactone (27) were judged more bitter by the
panelists as compared to the control. These results demonstrate
the predictive power of the pharmacophore model.
ASSOCIATED CONTENT
* Supporting Information
Syntheses and materials and methods. This material is available
free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*Tel: +49(0)5531 90 1883. Fax: +49(0)5531 90 48883. E-mail:
jakob.ley@symrise.com.
■
Notes
The authors declare no competing financial interest.
Parts of this study were presented as a poster at the 12th
Weurman Flavor Research Symposium, July 1−4, 2008,
Interlaken, Switzerland.
6310
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311

Journal of Agricultural and Food Chemistry
Article
(19) Sippl, M. J. Recognition of Errors in Three-dimensional
Structures of Proteins. Proteins: Struct., Funct., Bioinf. 1993, 17 (4),
355−362.
(20) Krieger, E.; Darden, T.; Nabuurs, S. B.; Finkelstein, A.; Vriend,
G. Making Optimal Use of Empirical Energy Functions: Force-field
Parameterization in Crystal Space. Proteins: Struct., Funct., Bioinf. 2004,
57 (4), 678−683.
(21) YASARA, version 11.4.18; Radboud University: Nijmegen,
Netherlands, 2011.
ACKNOWLEDGMENTS
■
We thank the analytical department (Dr. B. Weber, M. Roloff,
and Dr. A. Degenhardt) for support and Dr. H.-J. Bertram for
initiation of this work.
REFERENCES
■
(1) Ley, J. P. Masking Bitter Taste by Molecules. Chemosensory
Perception 2008, 1 (1), 58−77.
(22) Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.; Xiong, G.; Zhang,
W.; Yang, R.; Cieplak, P.; Luo, R.; Lee, T. A Point-charge Force Field
for Molecular Mechanics Simulations of Proteins Based on
Condensed-phase Quantum Mechanical Calculations. J. Comput.
Chem. 2003, 24 (16), 1999−2012.
(2) Soldo, T.; Blank, I.; Hofmann, T. (+)-(S)-AlapyridaineA
general taste enhancer? Chem. Sens. 2003, 28, 371−379.
(3) Soldo, T.; Hofmann, T. Application of Hydrophilic Interaction
Liquid Chromatography/Comparative Taste Dilution Analysis for
Identification of a Bitter Inhibitor by a Combinatorial Approach Based
on Maillard Reaction Chemistry. J. Agric. Food Chem. 2005, 53 (23),
9165−9171.
(23) GOLD, version 4.1; The Cambridge Crystallographic Data
Centre: Cambridge, United Kingdom, 2009.
(24) Logendra, S.; Ribnicky, D. M.; Yang, H.; Poulev, A.; Ma, J.;
Kennelly, E. J.; Raskin, I. Bioassay-guided isolation of aldose reductase
inhibitors from Artemisia dracunculus. Phytochemistry 2006, 67 (14),
1539−1546.
(25) Tanaka, T.; Yamasaki, K.; Kohda, H.; Tanaka, O.; Mahato, S. B.
Dihydrochalcone-glucosides as sweet principles of Symplocos ssp.
Planta Med. 1980, No. Suppl., 81−83.
(26) Chou, H.-C.; Chen, J.-J.; Duh, C.-Y.; Huang, T.-F.; Chen, I.-S.
Cytotoxic and anti-platelet aggregation constituents from the root
wood of Melicope semecarpifolia. Planta Med. 2005, 71 (11), 1078−
1081.
(27) Fokialakis, N.; Lambrinidis, G.; Mitsiou, D. J.; Aligiannis, N.;
Mitakou, S.; Skaltsounis, A. L.; Pratsinis, H.; Mikros, E.; Alexis, M. N.
A new class of phytoestrogens: Evaluation of the estrogenic activity of
deoxybenzoins. Chem. Biol. 2004, 11 (3), 397−406.
(28) Goto, H.; Terao, Y.; Akai, S. Synthesis of various kinds of
isoflavones, isoflavanes, and biphenyl-ketones and their 1,1-diphenyl-2-
picrylhydrazyl radical-scavenging activities. Chem. Pharm. Bull. 2009,
57 (4), 346−360.
(29) Ley, J. P.; Kindel, G.; Paetz, S.; Krammer, G. Hydroxydeox-
ybenzoins and the use thereof to mask a bitter taste. EP 1,868,452-B1,
2006.
(30) Osaki, N.; Koyano, T.; Kowithayakorn, T.; Hayashi, M.;
Komiyama, K.; Ishibashi, M. Sesquiterpenoids and Plasmin-Inhibitory
Flavonoids from Blumea balsamifera. J. Nat. Prod. 2005, 68 (3), 447−
449.
(31) Grael, C. F. F.; Vichnewski, W.; De Souza, G. E. P.; Lopes, J. L.
C.; Albuquerque, S.; Cunha, W. R. A study of the trypanocidal and
analgesic properties [of substances] from Lychnophora granmongolese
(Duarte) Semir & Leitao Filho. Phytother. Res. 2000, 14 (3), 203−206.
(32) Ali, D. M.; Wong, K. C.; Lim, P. K. Flavonoids from Blumea
balsamifera. Fitoterapia 2005, 76 (1), 128−130.
(33) Ye, J.; Guan, Y.; Zeng, S.; Liu, D. Ampelopsin prevents
apoptosis induced by H2O2 in MT4-lymphocytes. Planta Med. 2008,
74, 252−257.
(34) Meyerhof, W.; Batram, C.; Kuhn, C.; Brockhoff, A.; Chudoba,
E.; Bufe, B.; Appendino, G.; Behrens, M. The Molecular Receptive
Ranges of Human TAS2R Bitter Taste Receptors. Chem. Senses 2010,
35 (2), 157−170.
(35) Cano, J.; Mintijano, H.; Lopez-Cremades, F.; Borrego, F.
Masking the bitter taste of pharmacetuicals. Manufacturing Chemist
2000, No. July, 16−17.
(36) Meyerhof, W. Elucidation of mammalian bitter taste. Rev.
Physiol., Biochem. Pharmacol. 2005, 154, 37−72.
(37) Brockhoff, A.; Behrens, M.; Roudnitzky, N.; Appendino, G.;
Avonto, C.; Meyerhof, W. Receptor agonism and antagonism of
dietary bitter compounds. J. Neurosci. 2011, 31, 14775−14782.
(4) Degenhardt, A.; Ullrich, F.; Hofmann, T.; Stark, T. Flavonoid
sugar addition products, method for manufacture and use thereof as
flavoring materials. EP 1,856,988.
(5) Matuschek, M.; Jager, M. B.; Kleber, A.; Krohn, M.; Zinke, H.
Method for modulating the taste of material compisitions containing at
least one high intensity sweetener (his). WO 2008 102,018, 2008.
(6) Slack, J. P.; Brockhoff, A.; Batram, C.; Menzel, S.; Sonnabend, C.;
Born, S.; Galindo, M. M.; Kohl, S.; Thalmann, S.; Ostopovici-Halip, L.;
Simons, C. T.; Ungureanu, I.; Duineveld, K.; Bologa, C. G.; Behrens,
M.; Furrer, S.; Oprea, T. I.; Meyerhof, W. Modulation of Bitter Taste
Perception by a Small Molecule hTAS2R Antagonist. Curr. Biol. 2010,
20 (12), 1104−1109.
(7) Ley, J. P.; Krammer, G.; Reinders, G.; Gatfield, I. L.; Bertram, H.
J. Evaluation of Bitter Masking Flavanones from Herba Santa
(Eriodictyon californicum (H. & A.) Torr., Hydrophyllaceae. J. Agric.
Food Chem. 2005, 53 (15), 6061−6066.
(8) Ley, J. P.; Blings, M.; Paetz, S.; Krammer, G. E.; Bertram, H.-J.
New Bitter-Masking Compounds: Hydroxylated Benzoic Acid Amides
of Aromatic Amines as Structural Analogues of Homoeriodictyol. J.
Agric. Food Chem. 2006, 54 (22), 8574−8579.
(9) Ley, J. P.; Paetz, S.; Blings, M.; Hoffmann-Lucke, P.; Bertram, H.-
̈
J.; Krammer, G. E. Structural Analogues of Homoeriodictyol as Flavor
Modifiers. Part III: Short Chain Gingerdione Derivatives. J. Agric. Food
Chem. 2008, 56 (15), 6656−6664.
(10) MOE [2008.10]; Chemical Computing Group Inc.: Montreal,
Canada, 2009.
(11) Halgren, T. A. Merck molecular force field 0.1. Basis, form,
scope, parameterization, and performance of MMFF94. J. Comput.
Chem. 1996, 17, 490−519.
(12) Halgren, T. A. Merck molecular force field 0.2. MMFF94 van
der Waals and electrostatic parameters for intermolecular interactions.
J. Comput. Chem. 1996, 17, 520−552.
(13) Halgren, T. A. Merck molecular force field 0.3. Molecular
geometries and vibrational frequencies for MMFF94. J. Comput. Chem.
1996, 17, 553−586.
(14) Halgren, T. A.; Nachbar, R. B. Merck molecular force field 0.4.
Conformational energies and geometries for MMFF94. J. Comput.
Chem. 1996, 17, 587−615.
(15) Halgren, T. A. Merck molecular force field 0.5. Extension of
MMFF94 using experimental data, additional computational data, and
empirical rules. J. Comput. Chem. 1996, 17, 616−641.
(16) Okada, T.; Sugihara, M.; Bondar, A. N.; Elstner, M.; Entel, P.;
Buss, V. The Retinal Conformation and Its Environment in Rhodopsin
in Light of a New 2.2 Å Crystal Structure. J. Mol. Biol. 2004, 342 (2),
571−583.
(17) Pieper, U.; Eswar, N.; Braberg, H.; Madhusudhan, M. S.; Davis,
F. P.; Stuart, A. C.; Mirkovic, N.; Rossi, A.; Marti-Renom, M. A.; Fiser,
A. MODBASE, a Database of Annotated Comparative Protein
Structure Models, and Associated Resources. Nucleic Acids Res. 2004,
32 (Suppl. 1), D217−D222.
(18) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J.
M. PROCHECK: A Program to Check the Stereochemical Quality of
Protein Structures. J. Appl. Crystallogr. 1993, 26 (2), 283−291.
6311
dx.doi.org/10.1021/jf301335z | J. Agric. Food Chem. 2012, 60, 6303−6311