Probing the Binding Pocket of the Broadly Tuned Human Bitter Taste Receptor TAS2R14 by Chemical Modification of Cognate Agonists

Article abstract of DOI:10.1111/cbdd.12734

Sensing potentially harmful bitter substances in the oral cavity is achieved by a group of ^?25 receptors, named TAS2Rs, which are expressed in specialized sensory cells and recognize individual but overlapping sets of bitter compounds. The receptors differ in their tuning breadths ranging from narrowly to broadly tuned receptors. One of the most broadly tuned human bitter taste receptors is the TAS2R14 recognizing an enormous variety of chemically diverse synthetic and natural bitter compounds, including numerous medicinal drugs. This suggests that this receptor possesses a large readily accessible ligand binding pocket. To allow probing the accessibility and size of the ligand binding pocket, we chemically modified cognate agonists and tested receptor responses in functional assays. The addition of large functional groups to agonists was usually possible without abolishing agonistic activity. The newly synthesized agonist derivatives were modeled in the binding site of the receptor, providing comparison to the mother substances and rationalization of the in vitro activities of this series of compounds.

Full text of DOI:10.1111/cbdd.12734

Chem Biol Drug Des 2016; 88: 66–75
Research Article
Probing the Binding Pocket of the Broadly Tuned
Human Bitter Taste Receptor TAS2R14 by Chemical
Modification of Cognate Agonists
Rafik Karaman1, Stefanie Nowak2,†, Antonella Di
The human sense of taste facilitates the recognition of
Pizio3,4,†, Hothaifa Kitaneh1, Alaa Abu-Jaish1,
beneficial and potentially harmful food components prior to
ingestion (1). Taste cells distributed in the oral cavity are
equipped with taste receptors devoted to the detection of
Wolfgang Meyerhof2, Masha Y. Niv3,4 and
Maik Behrens2,*
the five basic taste modalities: sweet, sour, salty, umami,
1Bioorganic Chemistry Department, Faculty of Pharmacy,
Al-Quds University, P.O. Box 20002, Jerusalem, Israel
and bitter (2). The recognition of the numerous and struc-
2Department of Molecular Genetics, German Institute of
turally diverse bitter compounds found in nature (structures
and references accessible via BitterDB database (3)) is
believed to protect humans from involuntary ingestion of
potentially harmful substances, although a moderate bitter
taste of food items known to be safe for consumption is
frequently enjoyed (4). Within the oral cavity, bitter sub-
stances are released from foodstuff resulting in the activa-
tion of one or several of the ~25 functional human bitter
taste receptors of the taste 2 receptor family (TAS2R) that
belongs to the large family of G protein-coupled receptors
(GPCRs) (5). Previous screening efforts of heterologously
expressed human TAS2Rs resulted in the deorphanization
of 21 of the 25 receptors (6,7). The ability to associate the
majority of TAS2Rs with activating bitter substances
revealed several important features of the TAS2R gene
family. Firstly, a single bitter compound can activate multi-
ple TAS2Rs and, vice versa, TAS2Rs can respond to sev-
eral bitter substances. Secondly, the TAS2Rs’ agonist
recognition profiles range from broadly to narrowly tuned.
A recent screening experiment using more than 100 natu-
ral and synthetic bitter compounds demonstrated that the
three most broadly tuned receptors TAS2R10, TAS2R14,
and TAS2R46 each recognized about one-third of the
tested substances and that their combined response pro-
files accounted for the detection of already one-half of all
substances (6). Hence, these receptors may contribute
disproportionally to human bitter tastant recognition. Two
of these receptors, the TAS2R46 (8) and the TAS2R10 (9),
have been subjected to detailed structure-function analy-
ses to elucidate the architecture of their binding pockets
which enable the recognition of so many structurally differ-
ent bitter substances while maintaining a high level of
selectivity. It was shown that these TAS2Rs possess single
ligand binding pockets (8) tailored to recognize multiple
diverse bitter agonists (9). Although TAS2R14 has been
among the first deorphanized TAS2Rs (10), similar experi-
ments for this receptor are lacking. In fact, the original
report already recognized that this receptor may exhibit an
exceptionally broad tuning, with no apparent common
Human Nutrition Potsdam-Rehbruecke, 14558 Nuthetal,
Germany
3The Institute of Biochemistry, Food and Nutrition, The
Robert H Smith Faculty of Agriculture, Food and
Environment, The Hebrew University, Rehovot 76100,
Israel
4The Fritz Haber Center for Molecular Dynamics, The
Hebrew University, Jerusalem 91904, Israel
*Corresponding author: Maik Behrens, behrens@dife.de
†Authors contributed equally.
Sensing potentially harmful bitter substances in the
oral cavity is achieved by a group of ~25 receptors,
named TAS2Rs, which are expressed in specialized
sensory cells and recognize individual but overlapping
sets of bitter compounds. The receptors differ in their
tuning breadths ranging from narrowly to broadly
tuned receptors. One of the most broadly tuned human
bitter taste receptors is the TAS2R14 recognizing an
enormous variety of chemically diverse synthetic and
natural bitter compounds, including numerous medici-
nal drugs. This suggests that this receptor possesses
a large readily accessible ligand binding pocket. To
allow probing the accessibility and size of the ligand
binding pocket, we chemically modified cognate ago-
nists and tested receptor responses in functional
assays. The addition of large functional groups to ago-
nists was usually possible without abolishing agonistic
activity. The newly synthesized agonist derivatives
were modeled in the binding site of the receptor, pro-
viding comparison to the mother substances and
rationalization of the in vitro activities of this series of
compounds.
Key words: bitter taste receptor, chemical ligand design,
functional calcium imaging, in silico homology modeling,
TAS2R binding pocket
Received 28 October 2015, revised 15 December 2015 and
accepted for publication 13 January 2016
66
ª 2016 John Wiley & Sons A/S. doi: 10.1111/cbdd.12734

Probing the TAS2R14 Binding Pocket
chemical motifs among the identified agonists. By com-
bined in silico prediction of potentially agonistic molecules
and subsequent functional screening of candidate ago-
nists, numerous additional and previously unknown activa-
tors of TAS2R14 were revealed (11). The data suggested
that, instead of a single common pharmacophore, a num-
ber of different chemical scaffolds may exist among
TAS2R14 agonists. Moreover, it turned out that a consid-
erable number of the newly identified agonists of this
receptor represent important drugs with anti-inflammatory,
analgesic, and antitumor activities (11). As it was shown
that TAS2Rs are expressed in several non-gustatory tis-
sues, including the respiratory tract, the alimentary canal,
the thyroid, and the heart (for a recent review see (12)),
knowledge about the chemical features of agonists is of
outmost importance for identifying potential off-target
activities of medical drugs. The fact that TAS2R14 is the
most abundant TAS2R in human heart tissue (13) under-
scores the importance for a precise knowledge on struc-
ture–activity relationships among its agonists. Moreover,
an improved knowledge about chemical structures com-
mon to agonists may help to repurpose existing drugs, or
rationally design novel TAS2R14 antagonists. Although at
present only few reasonably specific TAS2R antagonists
have been identified (14–19), the majority of these mole-
cules share common chemical structures with the corre-
sponding receptor agonists (14,18).
purity dichloromethane, THF, and diethyl ether (>99%)
were purchased from Biolab (Israel). Silica gel (Silica gel 60
(0.040–0.063 mm)) and thin layer chromatography (TLC)
(TLC Silica gel 60F254) sheets were purchased from
Merck Ltd (Miami Lakes, FL, USA). The LC/ESI-MS/MS
system used was Agilent 1200 series liquid chromatogra-
phy coupled with a 6520 accurate mass quadruple-time of
flight mass spectrometer (Q-TOF LC/MS). The analysis
was performed in the positive electrospray ionization
mode. The capillary voltage was 4.0 kV, and the scanned
mass range was 200–540 m/z (MS). The high pressure liq-
uid chromatography (HPLC) system consisted of an Alli-
ance 2695 module equipped with 2996 Photodiode array
detector from Water (Milford, MA, USA). Data acquisition
and control were carried out using EMPOWER 2TM software
(Water). Analytes were separated on a 4.6 mm 9 150 mm
XBridgeꢀ C18 column (5 lm particle size) used in con-
junction with a 4.6 9 20 mm, XBridgeꢀ C18 guard col-
umn. Microfilters of 0.45 lm porosity were normally used
(Acrodiscꢀ GHP, Water). pH meter model HM-30G: TOA
electronicsTM was used in this study to measure the pH
value for the buffers. The Sep-Pack C18 6 cc (1 g) car-
tridges were purchased form Water. 1H-NMR experiments
were performed with a Bruker AvanceII 400 spectrometer
equipped with a 5 mm BBO probe. Infrared spectra (FTIR)
for solid compounds were obtained from a KBr matrix and
for liquid compounds as neat (4000–400 cmÀ1) using a
PerkinElmer Precisely, Spectrum 100.
To elucidate the structural requirements for TAS2R14 ago-
nists and to probe the size of the TAS2R14 binding
pocket, we performed functional calcium imaging experi-
ments. In contrast to previous studies, which mostly relied
on testing commercially available compounds, we tailored
novel test compounds, based on known agonists, by
chemical syntheses (20). The agonist activities of the newly
synthesized substances were determined and directly
compared with those of the corresponding mother sub-
stances. Computational methodologies were used to ana-
lyze the results, allowing the investigation of binding
pocket requirements in an extraordinary broadly tuned
receptor.
Preparation of mefenamic acid and diclofenac
alkyl esters
In a 250 mL round-bottom flask, diclofenac sodium or
mefenamic acid (10 mmol) was dissolved in dry dioxane
(50 mL), solutions of 2 g sodium carbonate in dry dioxane
(50 mL) and 40 mmol alkyl halide (1-iodohexane, 1-iodo-
decane, or benzyl bromide) in dry dioxane (50 mL) were
added, the resulting solution was refluxed while stirring for
3 days (Figure 1). The reaction mixture was cooled to
room temperature and evaporated using vacuum pump to
dryness. Fifty mL 1N HCl and 50 mL dry ether were
added to the dry mixture, and the resulting mixture was
transferred to separator funnel and shacked. The two lay-
ers were separated, and the acidic aqueous layer was
extracted twice with dry ether (50 mL). The combined
ether fractions were dried over MgSO4 anhydrous, filtered,
and evaporated to dryness using vacuum pump. The dry
residue was subjected to column chromatography, and
the desired product was dried and characterized by FTIR,
H-NMR, and LC-MS analysis (see Supporting Information).
Materials and Methods
Chemical syntheses and analyses
Inorganic salts were of analytical grade and were used
without further purification. Organic buffer components
were distilled or recrystallized. Distilled water was redistilled
twice before use from all-glass apparatus. Benzoin,
diclofenac sodium, mefenamic acid, iodohexane, iodode-
cane, benzyl bromide, sodium hydride, sodium hydroxide,
hydrochloric acid, sodium carbonate, anhydrous MgSO4,
dioxane, dimethylformamide, and ethyl acetate were pur-
chased from Aldrich Chemicals Ltd and were used without
further purification. HPLC grade solvents of methanol, ace-
tonitrile, and water were purchased from Sigma-Aldrich
(Sigma-Aldrich Chemie GmbH, Munich, Germany). High
Preparation of benzoin alkyl ethers
In a 250 mL round-bottom flask, benzoin (10 mmol) was
dissolved in dry dioxane (50 mL), 60% sodium hydride
(30 mmol) was added to the solution, the reaction was stir-
red at room temperature until hydrogen gas production
ended, benzyl bromide, or hexyl iodide (20 mmol) was
Chem Biol Drug Des 2016; 88: 66–75
67

Karaman et al.
Computational analyses
Homology modeling
b2 adrenergic receptor in its active state (PDB ID: 3SN6)
(21) was used as template for constructing TAS2R14
model. Sequence alignment (Figure 4) was performed with
MP-T (22) and manually adjusted; homology modeling was
carried out with MEDELLER (23). Hydrogen atoms and
side chain orientations of the receptor were optimized at
physiological pH with the Protein Preparation Wizard tool
€
in Maestro (version 10.2, Schrodinger, LLC, New York,
NY, 2014) and the Predict Side Chains tool in Prime (ver-
€
sion 4.0, Schrodinger, LLC, New York, NY, USA 2014),
respectively. Ramachandran plots of the modeled receptor
attested that the prediction of backbone dihedral angles of
amino acid residues in the models is reliable. Throughout
the article, transmembrane (TM) residues are identified by
a superscript number system according to the Balles-
teros–Weinstein numbering method (24).
Figure 1: Chemical structures of the cognate TAS2R14 agonists
and schematic representation of syntheses performed for
chemical modifications. The chemical structures of the two anti-
inflammatory drugs, mefenamic acid and diclofenac, and of the
natural bitter substance benzoin are depicted. The synthesis
strategies for mefenamic acid and diclofenac alkyl esters,
mefenamic acid hexyl, mefenamic acid decyl, mefenamic acid
benzyl, diclofenac hexyl, and diclofenac benzyl esters as well as
benzoin alkyl ethers, benzoin hexyl, and benzoin benzyl ethers,
respectively, are indicated. R-X = Hexyl iodide, decyl iodide (only
for mefenamic acid modification), benzyl bromide.
Docking
Ligands were manually built using the Built facility in Mae-
stro and prepared for the docking using LIGPREP (version
€
3.4, Schrodinger, LLC, 2014) that generates 3D struc-
slowly added, and the reaction was mixed for overnight
and then refluxed for 2 h (Figure 1). The reaction mixture
was cooled to room temperature, few drops of 0.1 N HCl
were added slowly to destroy the excess of sodium
hydride, and then, the reaction mixture was evaporated to
dryness. 1N NaOH (50 mL) and dry ether (50 mL) were
added to the dry mixture; the resulting mixture was trans-
ferred to separator funnel and shacked. The two layers
were separated, and the basic aqueous layer was
extracted twice with dry ether (50 mL). The combined ether
fractions were dried over MgSO4 anhydrous, filtered, and
evaporated to dryness using a vacuum pump. The dry resi-
due was subjected to column chromatography, and the
desired product was dried and characterized by FTIR, H-
NMR, and LC-MS analysis (see Supporting Information).
tures, stereoisomers (except where stereochemistry of a
compound is defined), tautomers and protonation states
at pH 7.0 Æ 0.5. Glide (Grid-based Ligand Docking with
€
Energetics) software (version 6.7, Schrodinger, LLC, 2014)
was used for the Induced Fit Docking (25,26). The grid
box was centroid of residues Trp893.32 and Phe2476.55
,
and the van der Waals scalings of receptors and ligands
ꢀ
were 0.5. Side chains of residues within 5.0 A of the
ligand were refined. The docking was performed with the
Standard Precision (SP) mode of Glide. Docking poses
showing similar binding modes among all compounds
were selected and re-scored with the Extra Precision (XP)
mode of Glide. In the case of the diclofenac derivatives,
the binding poses were obtained aligning the ligand struc-
tures to the diclofenac in its docking pose with Phase
€
Shape Screening (Phase, version 4.3, Schrodinger, LLC,
2014). The resulting complexes between TAS2R14 and
diclofenac derivatives were first minimized with MACROMODEL
Functional calcium imaging experiments
€
Functional heterologous expression of TAS2R14 was car-
ried out as before (e.g. (6,7,11)). Briefly, TAS2R14 cDNA
cloned in frame with a 5’ located sequence coding for the
first 45 amino acids of rat somatostatin receptor 3 and a
3’ located sequence coding for the herpes simplex virus
glycoprotein D epitope in the vector pcDNA5FRT was
transiently transfected into HEK 293T-Ga16gust44 cells.
After a ~24 h incubation period, cells were loaded with
Fluo-4 AM in the presence of 2.5 mM probenecid, then
washed several times with C1-buffer and placed in a fluo-
rometric imaging plate reader. After application of test
stimuli, changes in fluorescence were monitored. As nega-
tive controls, cells transfected with an empty expression
vector were used and directly compared to receptor trans-
fected cells.
(version 10.8, Schrodinger, LLC, 2014) to a derivative con-
ꢀ
vergence of 0.05 kJ/mol-A using the Polak-Ribiere Conju-
gate Gradient (PRCG) minimization algorithm, the
OPLS2005 force field, and the GB/SA water solvation
model; and then used as input for Glide Pose refinement.
Glide XP was used as scoring function for all complexes
(see Supporting Information, Table S1).
QM partial charges calculations
The Becke three-parameter, hybrid functional combined
with the Lee, Yang, and Parr correlation functional,
denoted B3LYP, were employed in the calculations using
density functional theory (DFT). All calculations were car-
ried out based on the restricted Hartree-Fock method
68
Chem Biol Drug Des 2016; 88: 66–75

Probing the TAS2R14 Binding Pocket
using the quantum chemical package Gaussian-2009 (27).
The starting geometries of all calculated molecules were
obtained using the Argus Lab program (28) and were ini-
tially optimized at AM1, PM3, HF/6-31G level of theory,
followed by optimization at the B3LYP/6-31G(d,p).
the clear signals observed upon treatment with 300 lM
mefenamic acid decyl ester (Figure 2A) demonstrate per-
sistent responsiveness of the receptor. To determine
whether chemical modification of the structurally related,
but less potent agonist diclofenac shows similar tolerance,
we tested the corresponding benzyl and hexyl esters of
this compound as well. As evident from Figure 2B, diclofe-
nac derivatives fail to activate TAS2R14 transfected cells
at all applicable concentrations. Hence, performing the
same modifications in agonists with lower potency resulted
in pronounced changes in receptor activation.
Results
Mefenamic acid (2-(2,3-dimethylphenyl)aminobenzoic acid)
and diclofenac (2-(2,6-dichloranilino)phenylacetic acid) have
been identified as novel TAS2R14 agonists in a recent in sil-
ico screening of agonists and confirmed by functional
assays (11). Both substances represent non-steroidal anti-
inflammatory drugs and are structurally similar to flufenamic
acid, a previously identified TAS2R14 agonist (6) which was
used as a template for the in silico prediction of a ligand-
based pharmacophore model (11). Benzoin (2-hydroxy-1,2-
di(phenyl)ethanone), which represents the simplest aromatic
hydroxyl ketone and is a natural component in bitter almond
oil, also belongs to the group of cognate TAS2R14 agonists
(6) sharing the presence of two phenyl ring systems with
mefenamic acid and diclofenac (Figure 1).
Functional heterologous expression assays do not allow
assessing binding of agonist molecules to receptors.
Instead, receptor activation is monitored. To test whether
the lack of TAS2R14 activation by diclofenac benzyl ester
has resulted from a loss of its receptor binding capacity or
whether steric hindrance has prevented occupation of the
receptor’s binding pocket by the compound, we set up
competition experiments using the potent TAS2R14 ago-
nist aristolochic acid (6,29). We reasoned that, if diclofenac
benzyl ester would not fit into the receptor’s binding
pocket, aristolochic acid treatment of TAS2R14 expressing
cells should result in unimpaired receptor responses. If,
however, diclofenac benzyl ester is able to enter the bind-
ing pocket without causing receptor activation, competitive
inhibition of aristolochic acid activation should occur. We
therefore challenged TAS2R14 expressing cells with
increasing concentrations of diclofenac benzyl ester in the
presence of a signal saturating concentration of aris-
tolochic acid (3 lM) and monitored calcium responses.
The calcium traces of aristolochic acid treated TAS2R14
expressing cells (Figure 2C) showed no reduction of maxi-
mal amplitudes even in the presence of 100 lM diclofenac
benzyl ester excluding the possibility that this compound
acts as an inhibitory ligand (antagonist) for the receptor.
To analyze structural requirements of TAS2R14 agonists,
we first modified the most potent of the 3 agonists, mefe-
namic acid, by the addition of a third phenyl group linked
to the carboxyl group via an ester bond (Figure 1) and
tested the compound in functional assays. Surprisingly,
the resulting mefenamic acid benzyl ester retained agonis-
tic properties, although the lack of receptor activation
below a concentration of 10 lM indicates reduced potency
(Figure 2A). The occurrence of receptor-independent cal-
cium signals already at 30 lM prevented the use of higher
compound concentrations in this assay. The addition of an
aliphatic group with an identical number of carbon atoms,
leading to mefenamic acid hexyl ester, similarly resulted in
a
moderate reduction of the potency, yet, receptor
To confirm and extend the above observations, we modi-
fied benzoin, a TAS2R14 agonist with an even lower
potency compared to diclofenac. Much to our surprise,
the addition of aliphatic hexyl or aromatic benzyl side
chains resulted in the generation of more potent agonists.
For both benzoin derivatives, we observed robust receptor
responses already at concentrations of 10 lM, whereas
responses were not abolished (Figure 2A). Even the com-
pound mefenamic acid decyl ester robustly activated
TAS2R14-expressing cells starting from a concentration of
~100 lM (Figure 2A). Thus, for a potent agonist such as
mefenamic acid, the substantial increase of the molecular
mass impaired TAS2R14 activation considerably; however,
Figure 2: Chemical structures and receptor activating properties of mefenamic acid and diclofenac derivatives. Chemical structures of
mefenamic acid (A) and diclofenac (B) derivatives are provided together with molecular formulas and the corresponding molecular weights.
Side chains added to the original agonists are highlighted in red. Changes in fluorescence after agonist application and applied agonist
concentrations are shown on the right. Black traces were obtained from HEK 293T-Ga16gust44 cells transiently transfected with
TAS2R14 cDNA, and gray traces represent negative controls obtained by stimulation of cells transfected with empty expression vector. (C)
To investigate competition for the ligand binding site of TAS2R14 different mixtures of aristolochic acid and diclofenac benzyl ester were
added to HEK 293T-Ga16gust44 cells transiently transfected with TAS2R14 cDNA and fluorescence changes were monitored. Note the
stable amplitudes obtained in
a representative experiment for the agonist aristolochic acid alone and together with increasing
concentrations of diclofenac benzyl ester. Colored traces were obtained from HEK 293T-Ga16gust44 cells transiently transfected with
TAS2R14 cDNA, and gray traces represent negative controls obtained by stimulation of cells transfected with empty expression vector.
Next to the overlay traces that show the same color coding as seen for the individual traces the peak signal amplitudes and standard
deviations in relative fluorescence units (RFU) of two independent experiments performed in triplicates are provided using the same colors
as for the labeling of the traces. The observed differences were not statistically significant (Student’s t-test).
Chem Biol Drug Des 2016; 88: 66–75
69

Karaman et al.
A
B
C
the necessary concentration of unmodified benzoin
required to activate TAS2R14 was ~10-fold higher
(Figure 3A). Hence, despite the pronounced chemical simi-
larities between mefenamic acid, diclofenac and benzoin,
it appears that their binding modes differ.
To identify the chemical structural factors associated with
TAS2R14 activation and to evaluate the ability of the com-
pounds to interact with the receptor’s binding pocket, the
compounds were docked into
a TAS2R14 homology
model. TAS2Rs are seven transmembrane (7TM) proteins,
70
Chem Biol Drug Des 2016; 88: 66–75

Probing the TAS2R14 Binding Pocket
Figure 3: Chemical structures and receptor activating properties of benzoin derivatives. Chemical structures of benzoin (A), benzoin bexyl
ether (B) and benzoin benzyl ether (C) are provided together with molecular formulas and the corresponding molecular weights. Side
chains added to the original agonists are highlighted in red. Changes in fluorescence after agonist application and the applied agonist
concentrations are shown on the right. Black traces were obtained from HEK 293T-Ga16gust44 cells transiently transfected with
TAS2R14 cDNA, and gray traces represent negative controls obtained by stimulation of cells transfected with empty expression vector.
Note that the addition of hydrophobic side chains to benzoin results in elevated potencies of the modified derivatives.
Figure 4: Transmembrane sequence alignment of modeled TAS2R14 receptor and the template X-ray structure of b2 adrenergic
receptor (PDB ID: 3SN6). Sequence identity and similarity of TAS2R14 and 3SN6 sequences are of 14% and 25%, respectively. Identical
residues are shown in red and similar residues in blue. Positions X.50 according to the Ballesteros–Weinstein numbering system are
shaded yellow.
considered as a subgroup of the Class A G protein-
coupled receptors (GPCRs) (30). As the docking simula-
tions focus on bitter agonists, we chose to model the
receptor based on the published X-ray structure of the b2
adrenergic receptor in its active state, bound to an agonist
and the Gs protein (PDB ID: 3SN6) (21) (Figure 4). This
template was successfully used to build a TAS2R10 model
(9). Low-resolution homology models were found to be
capable to capture ligand–protein interactions in docking
simulations (e.g. (30–32)). Induced Fit Docking (25,26)
technique is used here to investigate the binding modes of
all compounds presented above with the TAS2R14 recep-
tor, to allow flexibility of the side chains in the binding site.
The mefenamic acid and diclofenac derivatives, which
have decreased activity relative to the corresponding
mother compounds, assume TAS2R14-bound conforma-
tions, in which the substituents point toward the aniline
nitrogen of the compound skeleton structure. Benzoin
derivatives, that have improved activity relative to benzoin,
have TAS2R14-bound conformations, in which the sub-
stituent is far from the carbonyl group.
Chem Biol Drug Des 2016; 88: 66–75
71

Karaman et al.
Mefenamic acid establishes p–p interactions with
His943.37, Phe1865.46, and Phe2436.51, and H-bonds with
Asn933.36 and Gln2667.39 side chains of TAS2R14
(Figure 5A). Mefenamic acid derivatives interact with the
receptor with a similar binding mode, but to accommodate
the ester moiety; they are shifted compared to the original
molecule (Figure 5D,G): because of the ester substitution,
interaction with Gln2667.39 is not possible and, because of
the increased size, ligands cannot engage in all stacking
interactions observed for mefenamic acid. Interestingly, in
case of the mefenamic acid benzyl ester, the additional
aromatic ring can re-establish one of these interactions,
with improved activity compared to the other derivatives.
Discussion
The TAS2R14 is a highly promiscuous human bitter taste
receptor (6,10). The wide tuning breadth of the receptor
may be achieved by architecture of the ligand binding
pocket that provides a diverse array of contact points for
ligand interactions as well as sufficient space to accom-
modate the various agonists. Moreover, as we have shown
for the broadly tuned human bitter taste receptor
TAS2R10, the ability to interact with many structurally
diverse bitter substances may come with the trade-off of
rather low-affinity binding between receptor and agonist
(9). To characterize the binding pocket of TAS2R14 in
more detail, we chemically synthesized derivatives of
known TAS2R14 agonists and determined their ability to
activate cells heterologously expressing the receptor. This
approach demonstrated that TAS2R14 tolerated the addi-
tions of large chemical groups to the chemical backbone
of ligand molecules or, in case of the benzoin derivatives,
even showed improved activation. Moreover, ligands are
not only required to establish contacts within the receptor
binding pockets, they first need to gain access through
extracellular loops which are decorated with at least one,
and in case of TAS2R14, with two, asparagine-linked,
oligosaccharide side chains (33). Sequence deviation from
crystallized GPCRs prevents reliable inclusion of loops in
homology models and hence understanding of the con-
straints for entering the binding pocket. Nevertheless, pre-
vious observations (8,34) indicate that receptor residues
located in or close to extracellular loop areas contribute to
receptor selectivity. Moreover, recent molecular dynamics
simulations of agonist binding to human TAS2R46 sug-
gested the presence of a transient, vestibular ligand bind-
ing site that may contribute to ligand selectivity in addition
to the orthosteric binding site (35). For the TAS2R14, steric
exclusion of agonists seems to represent a minor con-
straint, because additions of large side chains retained
agonistic activity or even became more potent.
Diclofenac has a methyl spacer between the carboxyl
group and the aromatic ring, and has chloride atoms in
ortho positions of the phenyl group (Figure 1). These struc-
tural modifications affect the binding mode: the carboxyl
can interact with Asn933.36 but not with Gln2667.39
;
His943.37, establishes both hydrophobic and polar interac-
tions with the ligand, but there is no interaction with
Phe1865.46 (Figure 5B). As the methyl spacer is causing a
shift in diclofenac binding, the addition of the ester moiety
could increase steric hindrance, negatively affecting bind-
ing. Docking results of diclofenac derivatives did not fur-
nish any poses comparable to the diclofenac pose
described above. Diclofenac derivatives were aligned to
diclofenac in its docking conformation, and then minimized
to adjust the position of the alkoxy group. Both the hexyl
and the benzyl esters have clashes with the backbone of
TM7. In particular, bad contacts were observed between
the hydrogen of the ligand alkoxy groups and the back-
bone Ile2627.35 (shown as red arrows in the 2D represen-
tation of binding modes in Figure 5E,H). This is despite
diclofenac derivatives being closer to TM3 than other com-
pounds (see the 3D representation of binding modes in
Figure 5E,H). Thus, diclofenac derivatives cannot be
accommodated in the binding pocket because of steric
hindrance.
Furthermore, structural analysis suggests that improved
activity of benzoin derivatives compared to benzoin relates
to hydrophobic interactions in general, and p–p stacking
with His943.37 in particular. As mefenamic acid benzyl
ester exhibited a somewhat reduced potency compared to
mefenamic acid, the interaction with Gln2667.39 is of
importance.
Benzoin and its derivatives were synthetized and tested
in racemic mixture. The docking results of the
R
stereoisomers (R)-benzoin, (R)-benzoin benzyl ether, and
(R)-benzoin hexyl ether showed better performance and
are reported here. (R)-benzoin establishes an H-bond
with Asn933.36 and aromatic interactions with Trp893.32
and Phe2476.55 (Figure 5C). Interestingly, the benzoin
derivatives can interact also with His943.37 (Figure 5F,I).
In addition, the Mulliken atomic charges for benzoin,
benzoin benzyl ether, and benzoin hexyl ether were cal-
culated by both DFT and AM1 methods. The net nega-
tive charge value around the two oxygen atoms in both
ethers is higher than in benzoin, enabling these oxygens
to participate in stronger interactions than in benzoin.
This is in agreement with the functional data showing
that ether derivatives being more potent agonists than
benzoin.
In summary, the current work demonstrated that chemical
modification of TAS2R14 agonists allows to probe the
spatial capacity of the binding pocket: the receptor
TAS2R14 is able to accommodate agonists with a wide
range of sizes, as typical for multispecific GPCRs (36),
indicating that agonist-receptor contact points do not
envelop the ligand tightly. This is in agreement with the
finding that TAS2R promiscuity correlates with the binding
site surface (37). Future work elucidating the contact
points between TAS2R14 binding site residues and its
72
Chem Biol Drug Des 2016; 88: 66–75

Probing the TAS2R14 Binding Pocket
A
D
G
B
E
H
C
F
I
Figure 5: 3D and 2D representations of the docking pose of mefenamic acid (A), diclofenac (B), benzoin (C) and their derivatives (benzyl
derivatives in D, E, and F; hexyl derivatives in G, H, and I) with TAS2R14 receptor. Mefenamic acid and its derivatives are shown in violet sticks
(with the exception of mefenamic decyl ester represented in red line, panel G), diclofenac and its derivatives in orange, benzoin, and their
derivatives in green. H-bond interactions are shown in magenta, p–p interactions in green and bad contacts are indicated by red arrows. In the
3D representations, only polar interactions are reported, and the distance between the ligands and Ca of the residues Ile1795.39 and Ile2627.35
are reported in dashed gray lines. In 2D representations, polar and hydrophobic residues are shown as cyan and green spheres, respectively.
Chem Biol Drug Des 2016; 88: 66–75
73

Karaman et al.
agonists is necessary to reveal the molecular basis for the
promiscuity of this receptor aside from its apparent spa-
cious ligand binding site.
12. Behrens M., Meyerhof W. (2011) Gustatory and extra-
gustatory functions of mammalian taste receptors.
Physiol Behav;105:4–13.
13. Foster S.R., Blank K., See Hoe L.E., Behrens M.,
Meyerhof W., Peart J.N., Thomas W.G. (2014) Bitter
taste receptor agonists elicit G-protein-dependent
negative inotropy in the murine heart. FASEB
J;28:4497–4508.
14. Brockhoff A., Behrens M., Roudnitzky N., Appendino
G., Avonto C., Meyerhof W. (2011) Receptor agonism
and antagonism of dietary bitter compounds. J Neu-
rosci;31:14775–14782.
Acknowledgments
The research was supported by the German Research
Foundation (DFG) (ME1024/8-1 to WM, MB, MYN, RK).
Lady Davis fellowship to ADP is gratefully acknowledged.
MYN participates in the European COST action CM1207
(GLISTEN).
15. Ley J. (2008) Masking bitter taste by molecules. Che-
mosens Percept;1:58–77.
Conflict of Interest
16. Pydi S.P., Jaggupilli A., Nelson K.M., Abrams S.R.,
Bhullar R.P., Loewen M.C., Chelikani P. (2015) Absci-
sic acid acts as a blocker of the bitter taste G protein-
coupled receptor T2R4. Biochemistry;54:2622–2631.
17. Pydi S.P., Sobotkiewicz T., Billakanti R., Bhullar R.P.,
Loewen M.C., Chelikani P. (2014) Amino acid deriva-
tives as bitter taste receptor (T2R) blockers. J Biol
Chem;289:25054–25066.
18. Roland W.S., Gouka R.J., Gruppen H., Driesse M.,
van Buren L., Smit G., Vincken J.P. (2014) 6-methoxy-
flavanones as bitter taste receptor blockers for
hTAS2R39. PLoS ONE;9:e94451.
All authors declare no conflict of interests.
References
1. Lindemann B. (1996) Taste reception. Physiol
Rev;76:718–766.
2. Yarmolinsky D.A., Zuker C.S., Ryba N.J. (2009) Com-
mon sense about taste: from mammals to insects.
Cell;139:234–244.
3. Wiener A., Shudler M., Levit A., Niv M.Y. (2012) Bit-
terDB: a database of bitter compounds. Nucleic Acids
Res;40:D413–D419.
4. Rozin P., Vollmecke T.A. (1986) Food likes and dislikes.
Annu Rev Nutr;6:433–456.
5. Behrens M., Meyerhof W. (2013) Bitter taste receptor
research comes of age: from characterization to modu-
lation of TAS2Rs. Semin Cell Dev Biol;24:215–221.
6. Meyerhof W., Batram C., Kuhn C., Brockhoff A., Chu-
doba E., Bufe B., Appendino G., Behrens M. (2010)
The molecular receptive ranges of human TAS2R bitter
taste receptors. Chem Senses;35:157–170.
7. Thalmann S., Behrens M., Meyerhof W. (2013) Major
haplotypes of the human bitter taste receptor TAS2R41
encode functional receptors for chloramphenicol. Bio-
chem Biophys Res Commun;435:267–273.
8. Brockhoff A., Behrens M., Niv M.Y., Meyerhof W.
(2010) Structural requirements of bitter taste receptor
activation. Proc Natl Acad Sci USA;107:11110–11115.
9. Born S., Levit A., Niv M.Y., Meyerhof W., Behrens M.
(2013) The human bitter taste receptor TAS2R10 is
tailored to accommodate numerous diverse ligands.
J Neurosci;33:201–213.
10. Behrens M., Brockhoff A., Kuhn C., Bufe B., Winnig
M., Meyerhof W. (2004) The human taste receptor
hTAS2R14 responds to a variety of different bitter
19. Slack J.P., Brockhoff A., Batram C., Menzel S., Son-
nabend C., Born S., Galindo M.M. et al. (2010) Modu-
lation of bitter taste perception by a small molecule
hTAS2R antagonist. Curr Biol;20:1104–1109.
20. Karaman R. (2013) Prodrugs for masking bitter taste of
antibacterial drugs–a computational approach. J Mol
Model;19:2399–2412.
21. Rasmussen S.G., Devree B.T., Zou Y., Kruse A.C.,
Chung K.Y., Kobilka T.S., Thian F.S. et al. (2011) Crys-
tal structure of the beta(2) adrenergic receptor-Gs pro-
tein complex. Nature;477:549–555.
22. Hill J.R., Deane C.M. (2013) MP-T: improving mem-
brane protein alignment for structure prediction. Bioin-
formatics;29:54–61.
23. Kelm S., Shi J., Deane C.M. (2010) MEDELLER:
homology-based coordinate generation for membrane
proteins. Bioinformatics;26:2833–2840.
24. Ballesteros J., Weinstein H. (1995) Integrated methods
for the construction of three-dimensional models of
structure-function relations in G protein-coupled recep-
tors. Methods Neurosci;25:366–428.
25. Friesner R.A., Banks J.L., Murphy R.B., Halgren T.A.,
Klicic J.J., Mainz D.T., Repasky M.P., Knoll E.H., Shel-
ley M., Perry J.K., Shaw D.E., Francis P., Shenkin P.S.
(2004) Glide: a new approach for rapid, accurate
docking and scoring. 1. Method and assessment of
docking accuracy. J Med Chem;47:1739–1749.
26. Zhu K., Shirts M.R., Friesner R.A. (2007) Improved
methods for side chain and loop predictions via the
protein local optimization program: variable dielectric
model for implicitly improving the treatment of polariza-
tion effects. J Chem Theory Comput;3:2108–2119.
compounds.
Biochem
Biophys
Res
Com-
mun;319:479–485.
11. Levit A., Nowak S., Peters M., Wiener A., Meyerhof
W., Behrens M., Niv M.Y. (2014) The bitter pill: clinical
drugs that activate the human bitter taste receptor
TAS2R14. FASEB J;28:1181–1197.
74
Chem Biol Drug Des 2016; 88: 66–75

Probing the TAS2R14 Binding Pocket
27. Frisch M.J., Trucks G.W., Schlegel H.B., Scuseria
G.E., Robb M.A., Cheeseman J.R., Scalmani G. et al.
(2009) Gaussian 06, Revision A.1.
28. Casewit C.J., Colwell K.S., Rappe A.K. (1992) Applica-
tion of a universal force field to organic molecules.
J Am Chem Soc;114:10035–10046.
29. Sainz E., Cavenagh M.M., Gutierrez J., Battey J.F.,
Northup J.K., Sullivan S.L. (2007) Functional character-
ization of human bitter taste receptors. Biochem
J;403:537–543.
bitter taste receptor.
11:4439–4449.
J
Chem Theory Comput;
36. Levit A., Beuming T., Krilov G., Sherman W., Niv M.Y.
(2014) Predicting GPCR promiscuity using binding site
features. J Chem Inf Model;54:184–194.
37. Di Pizio A., Niv M.Y. (2015) Promiscuity and selectivity
of bitter molecules and their receptors. Bioorg Med
Chem;23:4082–4091.
30. Di Pizio A., Niv M.Y. (2014) Computational studies of
smell and taste receptors. Isr J Chem;54:1205–1218.
31. Yarnitzky T., Levit A., Niv M.Y. (2010) Homology mod-
eling of G-protein-coupled receptors with X-ray struc-
tures on the rise. Curr Opin Drug Discov Dev;13:317–
325.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Appendix S1. Analytical data for the synthesized organic
compounds.
32. Skolnick J., Zhou H., Gao M. (2013) Are predicted pro-
tein structures of any value for binding site prediction
and virtual ligand screening? Curr Opin Struct
Biol;23:191–197.
33. Reichling C., Meyerhof W., Behrens M. (2008) Func-
tions of human bitter taste receptors depend on N-gly-
cosylation. J Neurochem;106:1138–1148.
34. Pronin A.N., Tang H., Connor J., Keung W. (2004)
Identification of ligands for two human bitter T2R
receptors. Chem Senses;29:583–593.
Appendix S2. Xyz coordinates for the predicted
TAS2R14-bound conformations of mefenamic acid anion,
mefenamic benzyl ester, mefenamic hexyl ester, mefe-
namic decyl ester, diclofenac anion, diclofenac benzyl
ester, diclofenac hexyl ester, benzoin, benzoin benzyl ether
and benzoin hexyl ether.
Table S1. Glide XP scores for mefenamic acid, diclofenac,
benzoin and their derivatives in complex with TAS2R14
structure.
35. Sandal M., Behrens M., Brockhoff A., Musiani F., Gior-
getti A., Carloni P., Meyerhof W. (2015) Evidence for a
transient additional ligand binding site in the TAS2R46
Chem Biol Drug Des 2016; 88: 66–75
75