Ondansetron and granisetron binding orientation in the 5-HT3 receptor determined by unnatural amino acid mutagenesis

Article abstract of DOI:10.1021/cb300246j

The serotonin type 3 receptor (5-HT₃R) is a ligand-gated ion channel found in the central and peripheral nervous systems. The 5-HT ₃R is a therapeutic target, and the clinically available drugs ondansetron and granisetron inhibit receptor activity. Their inhibitory action is through competitive binding to the native ligand binding site, although the binding orientation of the drugs at the receptor has been a matter of debate. Here we heterologously express mouse 5-HT₃A receptors in Xenopus oocytes and use unnatural amino acid mutagenesis to establish a cation-π interaction for both ondansetron and granisetron to tryptophan 183 in the ligand binding pocket. This cation-π interaction establishes a binding orientation for both ondansetron and granisetron within the binding pocket.

Full text of DOI:10.1021/cb300246j

Articles
pubs.acs.org/acschemicalbiology
Ondansetron and Granisetron Binding Orientation in the 5‑HT3
Receptor Determined by Unnatural Amino Acid Mutagenesis
Noah H. Duffy,† Henry A. Lester,‡ and Dennis A. Dougherty*,†
†Division of Chemistry and Chemical Engineering and ‡Division of Biology, California Institute of Technology, Pasadena, California
91125, United States
ABSTRACT: The serotonin type 3 receptor (5-HT3R) is a
ligand-gated ion channel found in the central and peripheral
nervous systems. The 5-HT3R is a therapeutic target, and the
clinically available drugs ondansetron and granisetron inhibit
receptor activity. Their inhibitory action is through com-
petitive binding to the native ligand binding site, although the
binding orientation of the drugs at the receptor has been a
matter of debate. Here we heterologously express mouse 5-
HT3A receptors in Xenopus oocytes and use unnatural amino
acid mutagenesis to establish a cation-π interaction for both
ondansetron and granisetron to tryptophan 183 in the ligand
binding pocket. This cation-π interaction establishes a binding
orientation for both ondansetron and granisetron within the
binding pocket.
he serotonin type 3 receptor (5-HT3R)1,2 is a ligand-gated
ion channel in the Cys-loop (pentameric) family of
A−F), with loops A−C contributed by the “principal” and β-
strands D−F contributed by the “complementary” subunit
(Figure 1). Docking studies using homology models have found
multiple energetically favorable poses of the antagonist
granisetron within this binding site, and the orientation of
the drug and the identities of the interacting residues are not
constant across poses.11−14 For example, Thompson et al.
reported two classes of poses of granisetron.13 In one class, the
cationic ammonium was oriented between Trp183 (loop B)
and Tyr234 (loop C), while the aromatic indazole was oriented
between Trp90 (β-strand D) and Phe226 (loop C). In the
second class of poses, this orientation was reversed, with the
cationic ammonium toward Trp90 and Phe226. Experimentally,
both Trp183 and Trp90 have been established to be important
for granisetron binding. In other experimental work, Yan and
White found that Trp90 is important for both ondansetron and
granisetron binding, and they interpreted their results as
providing evidence that the cationic ammonium of granisetron
was oriented toward Trp90.14
The accepted pharmacophore model for 5-HT3R includes an
amine group on the ligand. Past studies in our laboratory have
established the primary amine of 5-HT to make a cation-π
interaction with a conserved tryptophan residue on loop B
(Trp183).15 Mutation of Trp183 (as well as Trp90, Glu129
(loop A), and Tyr234) to alanine abolishes binding of
[3H]granisetron.13 Taken together, these results identify
Trp183 as a prime candidate for more detailed studies.
Moreover, the docking studies of Thompson et al. provide us
T
receptors, which also includes GABAA, glycine, and nicotinic
acetylcholine (nACh) receptors.3 The 5-HT3 receptor is a
cation selective channel found in the central and peripheral
nervous systems. Conduction occurs through a central pore,
formed by the pseudosymmetric assembly of five subunits
(Figure 1). There are five known 5-HT3R subunits (A−E),4
with the best characterized receptors being the homomeric 5-
HT3A receptor (5-HT3AR) and the heteromeric 5-HT3AB
receptors.5
The 5-HT3 receptor has been validated as a therapeutic
target; antagonists are currently used to control chemotherapy-
induced nausea, as well as to treat irritable bowel syndrome.6,7
Beyond current clinical uses, there is evidence that compounds
targeting the 5-HT3 receptor could be useful for the treatment
of a variety of disorders including schizophrenia and substance
abuse, as well as management of pain associated with, for
example, fibromyalgia. Prototype antagonists of the 5-HT3
receptor are ondansetron (Zofran) and granisetron (Kytril)
(Figure 2a).
There is no high-resolution structure of the 5-HT3 receptor.
There is, however, structural information from a number of
sources, including cryoelectron microscopy images of the nACh
receptor,8 high resolution structures of the homologous
acetylcholine binding protein,9 and the recently published X-
ray structure of the glutamate-gated chloride channel from C.
elegans, GluCl.10 These data, along with homology modeling
and biochemical studies, have provided a putative binding site
for 5-HT3 antagonists that coincides with the binding site of the
native agonist, serotonin (5-HT, Figure 2b). This binding site is
formed by a series of β-strands and connecting loops (labeled
Received: May 17, 2012
Accepted: August 8, 2012
Published: August 8, 2012
© 2012 American Chemical Society
1738
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
Figure 1. Basic layout of a Cys-loop (pentameric) receptor. The structure is that of the GluCl α subunit10 (PDB: 3RIF). Left: The full receptor, with
two subunits highlighted. One (blue) contributes principal agonist binding site residues, found on loops A (red), B (green), and C (blue). The
complementary subunit (black) contributes loops D (purple), E (orange), and F (yellow). The α-helical region corresponds to the transmembrane
domain; the region above it is extracellular. Right: Detail of agonist binding site, noting approximate locations of key residues considered here.
Figure 2. Chemical structures of drugs and amino acids used in this study. (a) 5-HT3A receptor antagonists. (b) 5-HT3A receptor agonists. (c)
Tryptophan analogues.
with testable guidelines as to other possible interactions,
specifically, with Trp90.13 In the present work, we set out to
better understand the binding of the high affinity antagonistic
drugs ondansetron and granisetron to the 5-HT3AR. The
cationic center of granisetron is a tertiary ammonium ion in a
granatane moiety (pKa = 9.6), but ondansetron has a
structurally distinct N-akylimidazolium moiety (pKa = 7.4).16
We sought to determine if this structural difference leads to
different binding orientations for the two drugs.
receptor, equilibrium dissociation constants, Kb, provide the
most direct indicator of binding interactions.17 Previous reports
have indicated that granisetron and ondansetron act com-
petitively with 5-HT at the 5-HT3AR.2,18 Both antagonists bind
reversibly, in that after a several minute washout of either
ondansetron or granisetron, agonist responses recovered
completely. Previous studies have established dissociation rate
constants of 0.58 min−1 for ondansetron and 0.13 min−1 for
granisetron,19 consistent with our observations. It has also been
shown that granisetron and ondansetron directly compete with
each other for the same binding site.19 We attempted to
determine Kb for ondansetron and granisetron using Schild
(dose-ratio) analysis;17 this requires measurements of dose−
RESULTS AND DISCUSSION
■
Ondansetron and Granisetron Schild Analysis. In
examining the interaction of a competitive antagonist with a
1739
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
As shown in Figure 3, during our standard 15 s agonist
application (see Methods), the inhibition by ondansetron and
granisetron was insurmountable by 5-HT: full receptor activity
could not be restored even with high concentrations of 5-HT.
We attribute the insurmountable inhibition to the slow off rates.
The granisetron data are better behaved than the ondansetron
data, but in either case we face a requirement of agonist
applications several minutes in duration. However, the 5-
HT3AR desensitizes on this time scale before full equilibrium is
achieved. Thus, for the high-affinity 5-HT3AR antagonists
ondansetron and granisetron, determination of true Kb by
functional measurements is not possible.
Thus, the concentration required for 50% receptor inhibition
(IC50) was the only viable functional measurement, and so
modified procedures were developed to render IC50 values a
good direct measure of antagonist binding. The IC50 measure-
ment is taken in the context of multiple equilibria, including
both agonist and antagonist binding/dissociation, conforma-
tional changes in the protein, and the “gating” equilibria
between the open and closed states of the receptor. We sought
to reduce the contribution from changes in agonist potency,
while retaining an index of antagonist potency. When evaluating
competitive antagonists, it is useful to consider more than one
agonist, so that one can be sure that IC50 determinations are
not distorted by agonist-receptor interacts. To determine IC50
values for ondansetron and granisetron, we studied both 5-HT
and an additional agonist, m-chlorophenylbiguanide (mCPBG,
Figure 2b).21
Agonist Behavior at Trp183 and Trp90. The effective
concentrations for 50% receptor activation (EC50) were
determined for both the native agonist 5-HT, as well as the
potent partial agonist mCPBG. The EC50 values were also
determined for a series of fluorinated tryptophan derivatives
(Figure 2c) introduced by nonsense suppression at Trp183
(Table 1). The EC50 values confirm the previously reported15
cation-π interaction for 5-HT at Trp183. Interestingly, we have
recently found that mCPBG does not respond to fluorination at
Trp183 in the same manner as 5-HT,22 and the data of Table 2
produced for this work confirm that result. Detailed experi-
ments and discussion concerning the activation of 5-HT3
receptors by mCPBG has been presented in a separate
Figure 3. Dose−response curves of wild type 5-HT3A receptor.
Responses to 5-HT with increasing concentrations of (a) ondansetron
and (b) granisetron. Data fit to the Hill equation. Fit parameters: (a)
Mean maximal current: 8 4 μA. [ondansetron] = 0 nM, EC50: 1.3
0.2 μM, nH: 2.4 0.6; [ondansetron] = 0.64 nM, Imax: 63%, EC50: 1.1
0.4 μM, nH: 2 1; [ondansetron] = 4.5 nM, Imax = 61%, EC50: 9
29 μM, nH: 2 8. (b) Mean maximal current: 9 4 μA. [granisetron]
= 0 nM, EC50: 1.5 0.1 μM, nH: 2.0 0.1; [granisetron] = 0.6 nM,
Imax: 25%, EC50: 0.7 0.7 μM, nH: 2 7; [granisetron] = 1.2 nM,
Imax: 4%, EC50: 1 1 μM, nH: 1 2.
response relationships (Figure 3). Schild analysis of ondanse-
tron applied to preparations of rat vagus nerves gave parallel
shifts, indicative of a competitive interaction, but efforts to
perform similar studies with granisetron were not successful.20
Table 1. EC50 Data for Wild Type and Mutant 5-HT3AR in Response to 5-HT or mCPBG
a
b
c
d
agonist
5-HT
mutation
wild type
N
EC50 (μM)
fold shift
Hill
25
18
11
19
10
20
29
9
1.6 0.1
3.4 0.2
2.1 0.2
2.0 0.2
2.9 0.2
2.2 0.3
W183F1W (5)
2.2
11
W183F2W (5,7)
W183F2W (4,7)
W183F3W (5,6,7)
W90F3W (5,6,7)
W90F4W (4,5,6,7)
wild type
18
36
1
2
23
250 10
160
1/6.8
1/2.4
5
1
0.23 0.01
0.64 0.03
0.50 0.02
5.8 0.4
2.0 0.2
2.3 0.2
2.2 0.1
1.6 0.1
1.8 0.1
1.4 0.1
2.6 0.2
1.5 0.1
3.2 0.5
2.1 0.3
mCPBG
W183F1W (5)
14
9
12
5.3
W183F2W (5,7)
W183F2W (4,7)
W183F3W (5,6,7)
W183F4W (4,5,6,7)
W90F3W (5,6,7)
W90F4W (4,5,6,7)
2.6 0.1
19
13
14
15
7
0.34 0.03
6.6 0.2
1/1.5
13
8.4 0.6
17
0.092 0.004
0.20 0.02
1/5.4
1/2.5
a
b
c
Number of oocytes averaged in EC50 determination. The effective concentration for half-maximal receptor activation. EC50(mutant)/EC50(wild
d
type). The Hill coefficient, nH, as determined from fitting the Hill equation.
1740
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
Table 2. IC50 Data for Wild Type and Mutant 5-HT3AR for Granisetron and Ondansetron in Response to 5-HT or mCPBG
a
b
c
d
e
antagonist
granisetron
agonist
5-HT
mutation
wild type
N
IC50 (nM)
fold shift
Hill
27
17
14
11
9
0.78 0.04
2.9 0.3
100 10
3.4 0.2
−1.4 0.1
−1.4 0.2
−1.7 0.2
−1.6 0.1
−1.6 0.1
−1.1 0.1
−1.8 0.1
−1.5 0.1
−1.4 0.1
−1.5 0.1
−1.3 0.2
−1.2 0.1
−0.9 0.1
−1.2 0.1
−1.3 0.2
−1.6 0.1
−1.5 0.1
−2.2 0.2
−1.7 0.2
−1.8 0.1
−1.3 0.1
−1.7 0.1
−1.5 0.1
−1.9 0.2
−1.3 0.1
−1.7 0.2
−1.1 0.1
−1.1 0.1
−1.5 0.2
−1.3 0.1
W183F1W (5)
3.7
130
4.4
W183F2W (5,7)
W183F2W (4,7)
W183F3W (5,6,7)
W90F3W (5,6,7)
W90F4W (4,5,6,7)
wild type
61
2
78
11
12
27
8
0.20 0.03
0.64 0.03
1.5 0.1
3.3 0.2
140 10
2.4 0.3
1/3.9
1/1.2
granisetron
mCPBG
W183F1W (5)
2.2
93
W183F2W (5,7)
W183F2W (4,7)
W183F3W (5,6,7)
W183F4W (4,5,6,7)
W90F3W (5,6,7)
W90F4W (4,5,6,7)
wild type
16
13
9
1.6
41
2
27
14
7
110 10
0.20 0.02
0.38 0.04
0.87 0.04
6.2 0.4
74
1/7.6
1/4
9
ondansetron
ondansetron
5-HT
21
16
11
9
W183F1W (5)
7.1
160
77
W183F2W (5,7)
W183F2W (4,7)
W183F3W (5,6,7)
W90F3W (5,6,7)
W90F4W (4,5,6,7)
wild type
140 10
67
6
9
160 10
0.40 0.03
1.3 0.1
180
1/2.2
1.4
13
11
23
15
13
13
8
mCPBG
0.91 0.04
W183F1W (5)
10
1
11
160
80
W183F2W (5,7)
W183F2W (4,7)
W183F3W (5,6,7)
W183F4W (4,5,6,7)
W90F3W (5,6,7)
W90F4W (4,5,6,7)
140 10
73
89
6
5
99
15
11
9
590 80
0.37 0.03
1.5 0.1
650
1/2.4
1.6
a
b
c
Agonist used in the presence of antagonist for channel activation. Number of oocytes averaged in IC50 determination. The effective concentration
for half-maximal receptor inhibition. IC50(mutant)/IC50(wild type). The Hill coefficient, nH, as determined from fitting the Hill equation.
d
e
publication.22 For the present purposes, the key point is that
mCPBG responds for fluorination at Trp183 differently than 5-
HT.
When the fluorinated tryptophans F3Trp and F4Trp were
installed at Trp90, both 5-HT and mCPBG show a gain of
function. These data are consistent with no cation-π interaction
for 5-HT or mCPBG with Trp90.
Ondansetron and Granisetron at Trp183 and Trp90.
An important aspect of the complex nature of measuring IC50
in receptors is the concentration of agonist used for receptor
activation. For a competitive interaction, a measured IC50 value
will depend on the agonist concentration.17 In order to make
meaningful comparisons of IC50 values across mutant receptors,
the concentration of agonist used was kept at a constant value
of twice EC50. The choice of a constant ratio of twice EC50 was
made to ensure sufficient signal, even in cases of low receptor
expression. We also emphasize that the series of mutations
being introduced represents a much more subtle variation in
structure than is possible with conventional mutagenesis. This
provides further confidence that no dramatic changes in
receptor−antagonist interactions are occurring in the study.
We also exploited the difference in the binding modes of 5-HT
and mCPBG to control for possible artificial trends in IC50
measurements.
Figure 4. These measurements were performed on the wild
type 5-HT3AR as well as for a series of fluorinated tryptophan
derivatives at Trp183. Agonist concentrations were 2 × EC50
for each agonist at each receptor, and data were fit to the Hill
equation (Figure 4). The resultant IC50 values are presented in
Table 2. Inhibition data for F4Trp could not be gathered using
5-HT as the agonist, because of channel block by high
concentrations of 5-HT.
The effect of fluorine substitution in modulating a cation-π
interaction has been well established.15,23−25 For both
ondansetron and granisetron, incremental substitutions of
fluorine to Trp183 increased IC50. As in previous studies of
fluorination trends, IC50 fold-shift values were plotted against
cation-π binding ability of fluorinated indoles, producing the
“fluorination plots” shown in Figure 5. Ondansetron inhibition
linearly correlates with the energy of cation-π binding,
regardless of whether 5-HT or mCPBG was used as an agonist.
Granisetron also displayed a strong correlation with respect to
degree of fluorination, regardless of agonist identity.
We interpret these results as establishing a cation-π
interaction between each drug and Trp183. The results of
Figure 5 highlight the value of having two distinct agonists to
evaluate an antagonist. With 5-HT as the agonist, we see linear
fluorination plots for the antagonists. The agonist alone, 5-HT,
shows a similar plot in a study of its EC50. We corrected for this
by always using a 5-HT does of 2 × EC50 for the particular
fluorination mutant. Nevertheless, there could be concern
Dose−response/inhibition relations were determined for
ondansetron and granisetron, using both 5-HT and mCPBG as
agonists, with representative voltage-clamp traces shown in
1741
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
Figure 4. Representative voltage-clamp traces and dose-inhibition curves for antagonists for wild type and Trp183 mutant 5-HT3A receptors.
Representative traces for inhibition of (a) wild type and (b) W183F3W receptors by increasing doses of granisetron. Time of application and
concentration noted by black bars. Channels opened by addition of mCPBG (red bars) at (a) 1 μM and (b) 13 μM. Hashes indicate wash times.
Ondansetron inhibition curves shown for receptors activated by (c) mCPBG and (d) 5-HT. Granisetron inhibition curves shown for receptors
activated by (e) mCPBG and (f) 5-HT. Data fit to the Hill equation; fit parameters (IC50 and nH) are in Table 2.
about deconvoluting the effect of fluorination on the agonist
versus the antagonist. With mCPBG as the agonist, there is no
such concern, as it does not respond to fluorination at Trp183
in a manner consistent with a cation-π interaction. Thus, the
fluorination trends seen in Figure 5 can confidently be assigned
as reflecting the response of the antagonist to the mutation. We
also note that the fluorination effect remains regardless of the
structural identity of the cation. Ondansetron, with a N-
akylimidazolium moiety, and granisetron, with a tertiary
ammonium, show similar trends at the same residue, which is
evidence that their binding orientations are similar.
by the other derivatives, especially for granisetron. The effect is
evident, but much less pronounced, in studies of ondansetron.
We considered the possibility that an additional unique,
perhaps steric, feature of 5,7-F2Trp was influencing the analysis.
As such, we prepared 4,7-F2Trp (Scheme 1), which should
have the same cation-π binding ability, but different steric
requirements. Synthesis began by direct formation of 4,7-
difluoroindole (1) in a Bartoli reaction between the appropriate
nitrodifluorobenzene and vinyl magnesium bromide.26,27 In a
sequence similar to Gilchrist et al.28 the difluoroindole (1) was
then reacted with ethyl-3-bromo-2-hydroxyiminopropanoate to
yield the oxime (2), which was then reduced using aluminum/
mercury amalgam. The amine of the resulting amino acid ester
was protected with the photocleavable 2-nitroveratryloxycar-
bonyl (NVOC), and the ester was hydrolyzed with sodium
While the general trends in our data are clear, the detailed
behaviors of the F2Trp unnatural amino acids present
interesting details. In the fluorination plots, 5,7-F2Trp, a
residue that we have used extensively, deviates from the line set
1742
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
Figure 5. Fluorination plots. Calculated cation-π binding ability versus log[IC50/IC50(wt)] for a series of fluorinated tryptophan derivatives at
Trp183. Ondansetron fluorination plots for receptors activated by (a) mCPBG and (b) 5-HT. Granisetron fluorination plots for receptors activated
by (a) mCPBG and (b) 5-HT. Red lines are linear fits (y = mx + b) inclusive of all points. Fit parameters: (a) m = −0.15 0.03, b = 5.1 0.6, R =
0.94; (b) m = −0.18 0.03, b = 6.0 0.8, R = 0.96; (c) m = −0.11 0.04, b = 3 1, R = 0.77; (d) m = −0.14 0.06, b = 5 2, R = 0.82.
Scheme 1
hydroxide. Conversion to the cyanomethyl ester (3) gave
material suitable for acylation of the dinucleotide dCA and for
preparation of tRNA necessary for incorporation into the 5-
HT3R (methods described previously).29 The unnatural amino
acid prepared by this route is, of course, racemic, but only the
natural L configuration will be incorporated; the ribosome of
the Xenopus oocyte in effect performs a kinetic resolution.
From an electrostatic point of view, 5,7-F2Trp and 4,7-F2Trp
are, to first order, indistinguishable, and so they should be
equivalent in a cation-π interaction. This is born out in the EC50
data for serotonin (Table 1), where the two F2Trp residues
differ by only a factor of 2, while the full fluorination series
spans more than a factor of 150. In contrast, the EC50 values for
mCPBG differ by 8-fold for the two F2Trps. Recall that
mCPBG does not make a cation-π interaction to the Trp. This
again suggests that specific steric interactions at Trp183 may be
involved.
special steric effect with 5,7-F2Trp, but 5,6,7-F3Trp and 4,5,6,7-
F4Trp both have fluorine atoms at the positions in 5,7-F2Trp
yet follow the trend indicative of electrostatics as the major
determinant to binding. As such, we cannot provide a simple
rationalization of the behaviors of the two difluoro-Trp
residues. Nevertheless, the consistent linear trend of Trp,
F1Trp, F3Trp, and F4Trp (Figure 5) provides compelling
evidence for a cation-π interaction to Trp183 for both
granisetron and ondansetron.
Both ondansetron and granisetron either increase their
potencies or retain their potencies when Trp90 is mutated to
F3Trp or F4Trp, respectively. This holds for both 5-HT and
mCPBG used as the agonist, which indicates no cation-π
interaction at that site. A loss of potency would be expected if a
cation-π interaction were present. We noted above that Yan
and White concluded that Trp90 is important for binding of
ondansetron and granisetron.14 Based on the observation that
granisetron and ondansetron responded differently to a W90F
mutation, the authors concluded that the bicyclic amine of
In the granisetron IC50 plots, 4,7-F2Trp gives a quite different
response than 5,7-F2Trp. We have suggested the possibility of a
1743
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
were obtained for ≥8 concentrations of granisetron or ondansetron on
≥8 cells. All EC50 and Hill coefficient values were obtained by fitting
dose−response relations to the Hill equation (Inorm = 1/[ 1 + (EC50/
[antagonist])n]) and are reported as means standard error of the fit.
For Schild analysis, the protocol for EC50 determinations was
repeated with the following changes: during the course of the
experiment after each minimal EC50 curve was determined, running
buffer containing granisetron or ondansetron was used for the
subsequent EC50 determinations. Agonist applications in the
subsequent EC50 determinations contained the same concentration
of antagonist as the running buffer.
granisetron interacts with Trp90. Our results position this
moiety in contact with Trp183, and we conclude that the
importance of Trp90 is for reasons other than a cation-π
interaction.
The present results provide evidence that the cationic centers
of ondansetron and granisetron are oriented toward Trp183
and not toward Trp90. Establishing a cation-π interaction with
ondansetron and granisetron at Trp183 determines a binding
orientation for these antagonists. Docking studies of granise-
tron performed in other laboratories have generated a series of
poses, some which are consistent with the cation pointed
toward Trp183. Our data provide evidence that these poses are
the most viable, while those with the cation pointed away from
Trp183 are not likely to be relevant.
Summary. We have identified a cation-π interaction with
the antagonists ondansetron and granisetron to Trp183 in the
5-HT3AR. This interaction is consistent with the binding mode
of 5-HT, but not mCPBG. The use of agonists with alternate
binding modes validates our data as direct measurements of
ondansetron and granisetron. Thus, the common antagonists
follow the basic pharmacophore established by 5-HT and not
the structurally dissimilar agonist mCPBG.
Synthesis of 3 (Scheme 1). 4,7-Difluoroindole (1). A solution of
3.5 mL (32.3 mmol) of 1,4-difluoro-2-nitrobenzene in 30 mL of dry
THF was cooled in an acetone/dry ice bath to −78 °C under argon. A
1 M solution of vinylmagnesium bromide in THF (100 mL, 100
mmol, 3 equiv) was added via cannula over 20 min. The reaction was
stirred for 1 h at −78 °C. Reaction quenched by the addition of 20 mL
of saturated aq NH4Cl. Upon warming to RT, 20 mL of water was
added, which formed a thick emulsion. The reaction mixture was
filtered through a layer of sand and washed copiously with ethyl
acetate. The organic layer was separated and dried over Na2SO4. The
solvent was removed under reduced pressure to yield a brown oil
containing multiple compounds. Purification by silica gel chromatog-
raphy using a gradient of 3% to 10% ethyl acetate in hexanes yielded a
slightly volatile amber oil: 871 mg (18%). Silica TLC (4% EtOAc in
1
hexanes) Rf = 0.26, stains red/pink using p-anisaldehyde. H NMR
METHODS
(300 MHz, CDCl3): δ 8.47 (br, 1H), 7.18 (t, J = 2.8 Hz, 1H), 6.78
(ddd, J = 10.3, 8.6, 3.5 Hz, 1H), 6.71−6.60 (m, 2H). 19F NMR (282
MHz, CDCl3): δ −124.1 to −129.8 (m), −139.4 to −142.3 (m). 13C
NMR (126 MHz, CDCl3): δ 152.2 (dd, J = 239, 2.4 Hz), 145.8 (dd, J
= 238.6, 3.0 Hz), 126.0 (dd, J = 15.9, 11.5 Hz), 124.7, 119.9 (dd, J =
24.9, 5.7 Hz), 106.4 (dd, J = 18.9, 8.2 Hz), 103.9 (dd, J = 18.9 8.2 Hz),
99.8. HRMS EI(+) m/z for C8H5NF2 found 153.0395, calculated
153.0390 (M+•).
■
Procedures for incorporating unnatural amino acids, expressing
receptors in Xenopus oocytes, and characterization by electro-
physiology followed established protocols.29
Protein Expression in Xenopus Oocytes. The mouse 5-HT3A
receptor in the pGEMHE vector was linearized with the restriction
enzyme Sbf I (New England Biolabs). mRNA was prepared by in vitro
transcription using the mMessage Machine T7 kit (Ambion).
Unnatural mutations were introduced by the standard Stratagene
QuickChange protocol using a TAG mutation at W183 and W90.
Stage V−VI Xenopus laevis oocytes were injected with mRNA. Each
cell was injected with 50 nL containing only mRNA (5 ng) for wild
type 5-HT3AR or a mixture of mRNA (5−32 ng, typically ∼12 ng) and
tRNA (18−30 ng, typically ∼18 ng) for unnatural amino acid.
Uncharged full length tRNA was injected as a negative control.
Electrophysiology. Electrophysiological experiments were per-
formed 24−48 h after injection using the OpusXpress 6000A
instrument (Axon Instruments) in two-electrode voltage clamp
mode at a holding potential of −60 mV. The running buffer was
Ca2+-free ND96 solution (96 mM NaCl, 2 mM KCl, 1 mM MgCl2,
and 5 mM HEPES, pH 7.5). Serotonin hydrochloride (5-HT) was
purchased from Alfa Aesar. 1-(3-Chlorophenyl)biguanide (mCPBG)
was purchased from Sigma-Aldrich. Granisetron hydrochloride and
ondansetron hydrochloride were purchased from Tocris Bioscience.
For EC50 determinations, oocytes were superfused with running
buffer at 1 mL/min for 30s before application of 5-HT or mCPBG for
15 s followed by a 116 s wash with the running buffer. Data were
sampled at 125 Hz and filtered at 50 Hz. Dose−response data were
obtained for ≥9 concentrations of 5-HT or mCPBG on ≥9 cells. All
EC50 and Hill coefficient values were obtained by fitting dose−
response relations to the Hill equation (Inorm = 1/[ 1 + (EC50/
[agonist])n]) and are reported as means standard error of the fit.
For IC50 determinations, oocyte response to either 5-HT or
mCPBG at 2 × EC50 for each receptor was measured before
application of antagonists by application of the agonist for 15 s
followed by 116 s of wash with the running buffer. Granisetron or
ondansetron doses were then preapplied, and the oocyte was allowed
to incubate for 60 s, followed by application of a mixture of the
antagonist dose with 5-HT or mCPBG at 2 fold EC50. The oocytes
were then washed with the running buffer for 116 s. Every 4 antagonist
doses, the oocytes were washed for 10 min, and oocyte response
reconfirmed using either 5-HT or mCPBG at 2 × EC50. Oocytes that
did not give consistent responses to 5-HT or mCPBG alone
throughout the experiment were discarded. Dose−response data
Ethyl 3-(4,7-Difluoro-1H-indol-3-yl)-2-(hydroxyimino)propanoate
(2). A solution of 424 mg (2.8 mmol, 2 equiv) of 4,7-difluoroindole in
10 mL of CH2Cl2 was added to 290 mg (1.4 mmol, 1 equiv) of ethyl 3-
bromo-2-(hydroxyimino)propanoate and 205 mg (1.9 mmol, 1.4
equiv) of Na2CO3. The mixture was stirred overnight under argon at
RT. The reaction was diluted with 50 mL of CH2Cl2 and 50 mL of
ethyl acetate and washed with 50 mL of water and 50 mL of brine. The
organic phase was separated and dried over Na2SO4. Purification
performed by silica chromatography, gradient 25% to 40% EtOAc in
1
hexanes to yield a white solid: 195 mg (50%). H NMR (300 MHz,
CD3CN): δ 9.93 (s, 1H), 9.64 (br, 1H), 7.01 (m, 1H), 6.81 (ddd, J =
10.5, 8.5, 3.5 Hz, 1H), 6.67 (ddd, J = 10.7, 8.5, 3.2 Hz, 1H), 4.21 (q, J
= 7.2 Hz, 2H), 4.12 (d, J = 1.1 Hz, 2H), 1.23 (t, J = 7.1 Hz, 3H). 19F
NMR (282 MHz, CD3CN): δ −130.99 to −131.20 (m), −141.02 to
−141.19 (m). 13C NMR (126 MHz, CDCl3): δ 164.80, 153.77 (dd, J =
240.4, 2.3 Hz), 152.58, 146.84 (dd, J = 238.1, 2.9 Hz), 127.18 (dd, J =
16.4, 12.1 Hz), 125.12, 119.55 (dd, J = 22.1, 5.7 Hz), 110.05 (dd, J =
3.7, 1.6 Hz), 106.94 (dd, J = 19.7, 9.1 Hz), 104.42 (dd, J = 22.9, 7.5
Hz), 62.28, 21.96 (d, J = 3.1 Hz), 14.33. HRMS FAB(+) m/z for
C13H13O3N2F2 found 283.0888, calculated 283.0894 (M + H).
N-(2-Nitroveratryloxycarbonyl)-4,7-difluorotryptophan Cyano-
methyl Ester (3). In a beaker 1.5 g of 8−20 mesh aluminum was
stirred under 15 mL of 2 M NaOH for 5 min. After decanting, the
aluminum was rinsed with water, and 15 mL of a 2% HgCl2 solution
was added and stirred slowly. The solution was decanted after
formation of the Hg-Al (∼10 min) and added to 195 mg (0.69 mmol)
of 2 in 20 mL of 9:1 dioxane/water. The reaction was stirred slowly at
RT overnight (∼24 h). The reaction was filtered through fluted paper
and then applied to a silica plug, eluting with EtOAc followed by 4%
MeOH in EtOAc. After concentration under reduced pressure, the
resulting oil was used directly. The oil was dissolved in 20 mL of 1:1
THF/water, and 151 mg of Na2CO3 (1.42 mmol) and 248 mg of 4,5-
dimethoxy-2-nitrobenzyl chloroformate (0.9 mmol) were added. The
reaction was stirred at RT for 3 h, followed by dilution with 20 mL of
CH2Cl2 and 20 mL of 1 N HCl. The organic phase was separated,
washed with brine, and dried over Na2SO4. Initial purification by silica
1744
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745

ACS Chemical Biology
Articles
chromatography 20% to 40% EtOAc in hexanes did not separate the
product from nitroveratryl side products. This mixture was dissolved in
3 mL dioxane and 3 mL of 2 N NaOH and stirred for 15 min. The
reaction was quenched with 6 mL of 1 N HCl and diluted with 20 mL
of EtOAc. The organic phase was separated, and the aqueous phase
was washed with 20 mL of CH2Cl2. The combined organic phases
were dried over Na2SO4, concentrated under reduced pressure, and
filtered through a silica plug eluting with EtOAc followed by 0.5%
acetic acid in EtOAc. This residue (∼35 mg, 0.07 mmol) was dissolved
in 1 mL of DMSO and added to a reaction flask containing 0.5 mL of
chloroacetonitrile (7.9 mmol) and 0.5 mL of triethylamine (3.6
mmol). The reaction was allowed to stir at RT for 5 h. The reaction
was poured onto a dry column of silica and eluted with EtOAc to
(10) Hibbs, R. E., and Gouaux, E. (2011) Principles of activation and
permeation in an anion-selective Cys-loop receptor. Nature 474, 54−
60.
(11) Joshi, P. R., Suryanarayanan, A., Hazai, E., Schulte, M. K.,
Maksay, G., and Bikadi, Z. (2006) Interactions of Granisetron with an
agonist-free 5-HT3A receptor model. Biochemistry 45, 1099−1105.
(12) Maksay, G., Bikadi, Z., and Simonyi, M. (2003) Binding
interactions of antagonists with 5-hydroxytryptamine(3A) receptor
models. J. Recept. Signal Transduct. 23, 255−270.
(13) Thompson, A. J., Price, K. L., Reeves, D. C., Chan, S. L., Chau,
P. L., and Lummis, S. C. R. (2005) Locating an antagonist in the 5-
HT3 receptor binding site using modeling and radioligand binding. J.
Biol. Chem. 280, 20476−20482.
(14) Yan, D., and White, M. M. (2005) Spatial orientation of the
antagonist granisetron in the ligand-binding site of the 5-HT3
receptor. Mol. Pharmacol. 68, 365−371.
(15) Beene, D. L., Brandt, G. S., Zhong, W. G., Zacharias, N. M.,
Lester, H. A., and Dougherty, D. A. (2002) Cation-pi interactions in
ligand recognition by serotonergic (5-HT3A) and nicotinic acetylcho-
line receptors: The anomalous binding properties of nicotine.
Biochemistry 41, 10262−10269.
(16) Delorenzi, F. G., Bridal, T. R., and Spinelli, W. (1994) Block of
the delayed rectifier current (I-K) by the 5-HT3 antagonists
ondansetron and granisetron in feline ventricular myocytes. Br. J.
Pharmacol. 113, 527−535.
(17) Wyllie, D. J. A., and Chen, P. E. (2007) Taking the time to study
competitive antagonism. Br. J. Pharmacol. 150, 541−551.
(18) Barnes, N. M., Hales, T. G., Lummis, S. C. R., and Peters, J. A.
(2009) The 5-HT3 receptor - the relationship between structure and
function. Neuropharmacology 56, 273−284.
(19) Rojas, C., Stathis, M., Thomas, A. G., Massuda, E. B., Alt, J.,
Zhang, J., Rubenstein, E., Sebastiani, S., Cantoreggi, S., Snyder, S. H.,
and Slusher, B. (2008) Palonosetron exhibits unique molecular
interactions with the 5-HT3 receptor. Anesth. Analg. 107, 469−478.
(20) Ito, H., Akuzawa, S., Tsutsumi, R., Kiso, T., Kamato, T., Nishida,
A., Yamano, M., and Miyata, K. (1995) Comparative study of the
affinities of the 5-HT3 receptor antagonists, YM060 YM114 (KAE-
393), granisetron and ondansetron in rat vagus nerve and cerebral
cortex. Neuropharmacology 34, 631−637.
(21) Kilpatrick, G. J., Butler, A., Burridge, J., and Oxford, A. W.
(1990) 1-(m-Chlorophenyl)-biguanide, a potent high-affinity 5-HT3
receptor agonist. Eur. J. Pharmacol. 182, 193−197.
(22) Miles, T. F.; Bower, K. S.; Lester, H. A.; Dougherty, D. A.,
(2012) A coupled array of noncovalent interactions impacts the
function of the 5-HT3A serotonin receptor in an agonist-specific way.
ACS Chem. Neurosci. In press. DOI: 10.1021/cn3000586.
(23) Xiu, X. A., Puskar, N. L., Shanata, J. A. P., Lester, H. A., and
Dougherty, D. A. (2009) Nicotine binding to brain receptors requires
a strong cation-pi interaction. Nature 458, 534−537.
(24) Dougherty, D. A. (2008) Physical organic chemistry on the
brain. J. Org. Chem. 73, 3667−73.
(25) Dougherty, D. A. (2008) Cys-loop neuroreceptors: structure to
the rescue? Chem. Rev. 108, 1642−53.
(26) Choi-Sledeski, Y.-M.; Gardner, C. J.; Liang, G.; Poli, G. B.;
Shum, P. W.-K.; Stoklosa, G. T.; Zhao, Z. (Sanofi-Aventis, Fr.)
Preparation of aminomethylphenylpiperidinylindolylmethanone deriv-
atives for use as tryptase inhibitors. Patent WO 2011022449.
(27) Ricci, A., and Fochi, M. (2003) Reactions between organo-
magnesium reagents and nitroarenes: Past, present, and future. Angew.
Chem., Int. Ed. 42, 1444−1446.
(28) Gilchrist, T. L., Lingham, D. A., and Roberts, T. G. (1979) Ethyl
3-Bromo-2-hydroxyiminopropanoate, a reagent for the preparation of
ethyl esters of alpha-amino acids. J. Chem. Soc., Chem. Commun.,
1089−1090.
1
recover 23 mg of a yellow solid (6%, 4 steps). H NMR (300 MHz,
DMSO-d6): δ 11.71 (s, 1H), 8.21 (d, J = 7.7 Hz, 1H), 7.68 (s, 1H),
7.23 (d, J = 2.0 Hz, 1H), 7.09 (s, 1H), 6.92−6.80 (m, 1H), 6.74−6.62
(m, 1H), 5.39−5.21 (m, 2H), 4.99 (s, 2H), 4.48−4.35 (m, 1H), 3.85
(s, 3H), 3.83 (s, 3H), 3.32−2.99 (m, 2H). 19F NMR (282 MHz,
DMSO-d6): δ −130.18 (dd, J = 22.4, 9.9 Hz), −138.53 (ddd, J = 22.6,
10.6, 3.1 Hz). 13C NMR (126 MHz, DMSO-d6): δ 171.56, 156.03,
153.83, 152.62 (d, J = 239.6 Hz), 148.15, 146.10 (dd, J = 238.5, 2.2
Hz), 139.55, 128.09, 126.70 (dd, J = 16.1, 12.2 Hz), 126.36, 118.82−
118.30 (m), 116.08, 110.61, 109.18, 108.59, 106.23 (dd, J = 18.8, 8.7
Hz), 103.90−103.23 (m), 63.15, 56.55, 55.30, 49.97, 27.84. HRMS
FAB(+) m/z for C23H20O8N4F2 found 518.1271, calculated 518.1249
(M+•).
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: dadougherty@caltech.edu.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank S. Lummis for helpful discussions and input. This
work was supported by the National Institutes of Health (NS
34407 and DA19375).
REFERENCES
■
(1) Nichols, D. E., and Nichols, C. D. (2008) Serotonin receptors.
Chem. Rev. 108, 1614−1641.
(2) Thompson, A. J., and Lummis, S. C. R. (2006) 5-HT3 receptors.
Curr. Pharm. Design 12, 3615−3630.
(3) Thompson, A. J., Lester, H. A., and Lummis, S. C. R. (2010) The
structural basis of function in Cys-loop receptors. Q. Rev. Biophys. 43,
449−499.
(4) Niesler, B., Walstab, J., Combrink, S., Moller, D., Kapeller, J.,
Rietdorf, J., Bonisch, H., Gothert, M., Rappold, G., and Bruss, M.
(2007) Characterization of the novel human serotonin receptor
subunits 5-HT3C, 5-HT3D, and 5-HT3E. Mol. Pharmacol. 72, 8−17.
(5) Jensen, A. A., Davies, P. A., Brauner-Osborne, H., and
Krzywkowski, K. (2008) 3B but which 3B? And that’s just one of
the questions: the heterogeneity of human 5-HT(3) receptors. Trends
Pharmacol. Sci. 29, 437−444.
(6) Machu, T. K. (2011) Therapeutics of 5-HT3 receptor
antagonists: Current uses and future directions. Pharmacol. Ther.
130, 338−347.
(7) Thompson, A. J., and Lummis, S. C. (2007) The 5-HT3 receptor
as a therapeutic target. Expert Opin. Ther. Targets 11, 527−40.
(8) Unwin, N. (2005) Refined structure of the nicotinic acetylcholine
receptor at 4 angstrom resolution. J. Mol. Biol. 346, 967−989.
(9) Hibbs, R. E., Sulzenbacher, G., Shi, J. X., Talley, T. T., Conrod, S.,
Kem, W. R., Taylor, P., Marchot, P., and Bourne, Y. (2009) Structural
determinants for interaction of partial agonists with acetylcholine
binding protein and neuronal alpha 7 nicotinic acetylcholine receptor.
EMBO J. 28, 3040−3051.
(29) Nowak, M. W., Gallivan, J. P., Silverman, S. K., Labarca, C. G.,
Dougherty, D. A., and Lester, H. A. (1998) In vivo incorporation of
unnatural amino acids into ion channels in Xenopus oocyte expression
system. Methods Enzymol. 293, 504−529.
1745
dx.doi.org/10.1021/cb300246j | ACS Chem. Biol. 2012, 7, 1738−1745