Mutagenesis and Molecular Modeling Reveal the Importance of the 5-HT3 Receptor F-loop*

The 5-HT3 receptor is a member of the Cys-loop family of ligandgated ion channels. The extracellular domains of these proteins contain six amino acid loops (A–F) that converge to form the ligand binding site. In this study we have mutated 21 residues in or close to the 5-HT3 receptor F-loop (Ile192 to Gly212) to Ala or to a residue with similar chemical properties. Mutant receptors were expressed in HEK293 cells, and binding affinity was measured using [3H]granisetron. Two regions displayed decreases in binding affinity when mutated to Ala (Ile192–Arg196 and Asp204–Ser206), but only one region was sensitive when mutated to chemically similar residues (Ile192–Val201). Homology modeling using acetylcholine-binding protein crystal structures with a variety of different bound ligands suggests there may be distinct movements of Trp195 and Asp204 upon ligand binding, indicating that these residues and their immediate neighbors have the ability to interact differently with different ligands. The models suggest predominantly lateral movement around Asp204 and rotational movement around Trp195, indicating the former is in a more flexible region. Overall our results are consistent with a flexible 5-HT3 receptor F-loop with two regions that have specific but distinct roles in ligand binding.

The 5-HT 3 2 receptor is a transmembrane protein that is involved in signal transmission in the central and peripheral nervous systems. It is a member of the Cys-loop family of ligand-gated ion channels that also includes the nicotinic acetylcholine (nACh), zinc-activated, glycine, GABA A and GABA C receptors. The functional receptors consist of five symmetrically arranged subunits that surround a centrally located ionconducting pore. Each subunit has a large extracellular N-terminal domain, four transmembrane domains (M1-M4), and a large intracellular loop between M3 and M4. The ligand binding site is located in the extracellular domain and is formed at the interface of two adjacent subunits (1)(2)(3)(4). In the 5-HT 3 receptor, five subunits (A-E) have been identified, although only homomeric 5-HT 3A or heteromeric 5-HT 3A/B subunit complexes have been functionally expressed and characterized (5,6). Current evidence suggests 5-HT 3 receptors in the brain may be predominantly homomeric receptors, whereas in the peripheral nervous system heteromeric receptors may predominate (7).
Biochemical studies have identified a number of key residues involved in binding and have shown that the amino acids responsible for the receptor-ligand interaction are located in six convergent loops (A-F). In addition, structural insight has been gained from cryoelectron microscopy of the nACh receptor and from the crystal structures of the acetylcholine binding protein (AChBP), a soluble protein that is homologous to the extracellular region of Cys-loop receptors (8 -11). However, there are some important structural and functional differences between AChBP and Cys-loop receptors. For example, AChBP lacks a channel and shows little cooperativity in ligand binding, and it is not clear whether AChBP best resembles an activated, resting, or desensitized state of the nACh receptor (9).
Celie et al. (10) have used isothermal titration calorimetry to compare binding in AChBP from Limnea. stagnalis and Bulinus truncatus and found that residues in loops A-E could not account for differences in ligand binding affinity between the two proteins. This suggests that differences in binding in these, and in other related proteins, may be due to differences in their F loops, whose crystal structure is poorly resolved in all of the AChBP structures to date, and whose sequence is poorly conserved both between subunits of the same receptor type and between different members of the same Cys-loop family (1-4, 8, 9). This region is known to have a significant impact on ligand binding and may interact with ligands as they enter or exit the binding site (12)(13)(14). To date, only the GABA A ␣1 and GABA C 1 subunits have received a comprehensive study of the residues in the F-loop region, but due to the high levels of sequence variability in this region it is difficult to make comparisons between homologous residues across the Cys-loop family (13,15). To gain a better understanding of the role of the F-loop in ligand binding in the 5-HT 3 receptor, we have mutated 21 consecutive amino acids in this region (Ile 192 to Gly 212 ) and created a series of structural models of this region using a variety of AChBP crystal structures. Our results demonstrate the importance of the 5-HT 3 receptor F-loop region in ligand binding.

EXPERIMENTAL PROCEDURES
Materials-All cell culture reagents were obtained from Invitrogen, except fetal calf serum, which was from Labtech International (Ringmer, UK). [ 3 H]Granisetron (63.5 Ci/mmol) was from PerkinElmer Life Sciences. All other reagents were of the highest obtainable grade.
Cell Culture-Human embryonic kidney (HEK) 293 cells were maintained on 90-mm tissue culture plates at 37°C and 7% CO 2 in a humidified atmosphere. They were cultured in Dulbecco's modified Eagle's medium/Nutrient Mix F12 (1:1) with GlutaMAX TM I media containing 10% fetal calf serum and passaged when confluent. For radioligand binding studies, cells in 90-mm dishes were transfected using calcium phosphate precipitation at 80 -90% confluency and incubated for 3-4 days before use (16,17). For functional studies cells were transfected by electroporation using the AMAXA system (Amaxa GmbH, Cologne, Germany), plated on 96-well plates, and incubated 1-2 days before assay.
Site-directed Mutagenesis-Mutagenesis reactions were performed using the method described by Kunkel (18). The 5-HT 3A subunit DNA (accession: AY605711) has been described previously (19). Oligonucleotide primers were designed according to the recommendations of Sambrook et al. (20) and some suggestions of the Primer Generator (Ref. 21; www.med.jhu.edu/medcenter/primer/primer.cgi). A silent restriction site was incorporated into each primer to assist rapid identification.
Radioligand Binding-This was undertaken as previously described (22) with minor modifications. Briefly, transfected HEK293 cells were washed twice with phosphate buffered saline and then scraped into 1 ml of ice-cold HEPES buffer (10 mM, pH 7.4) containing the following proteinase inhibitors: 1 mM EDTA, 50 g/ml soybean trypsin inhibitor, 50 g/ml bacitracin, and 0.1 mM phenylmethylsulfonyl fluoride, and frozen at Ϫ20°C. After thawing, they were washed twice with HEPES buffer, resuspended, and 50 g of cell membranes was incubated in 0.5 ml of HEPES buffer containing the 5-HT 3 receptor antagonist [ 3 H]granisetron. Saturation binding (eight point) assays were performed on at least three separate plates of transfected cells for each mutant using 0.1-40 nM [ 3 H]granisetron. Nonspecific binding, determined using 1 M quipazine, was routinely 5-10% of total binding. Reactions were incubated for 1 h at 4°C and terminated by rapid vacuum filtration using a Brandel cell harvester onto GF/B filters pre-soaked in 0.3% polyethyleneimine followed by two rapid washes with 4 ml of ice-cold HEPES buffer. Radioactivity was determined by scintillation counting (Beckman LS6000sc). Protein concentration was estimated using the Bio-Rad Protein Assay with bovine serum albumin standards. Data were analyzed by iterative curve fitting ( Immunofluorescence-This was as described previously (23). Briefly, transfected cells were washed with three changes of Trisbuffered saline (TBS: 0.1 M Tris, pH 7.4, 0.9% NaCl) and fixed using ice-cold 4% paraformaldehyde in phosphate buffer (PB, 66 mM Na 2 HPO 4 , 38 mM NaH 2 PO 4 , pH 7.2). After two TBS washes, cells were incubated overnight at 4°C in pAb120 at 1:1600 in TBS. Biotinylated anti-rabbit IgG (Vector Laboratories, Burlingame, CA) and fluorescein isothiocyanate avidin D (Vector) were used to detect bound antibody as per the manufacturer's instructions. Coverslips were mounted in Vectashield mounting medium (Vector). Immunofluorescence was observed using an UltraVIEW TM LCI Confocal Imaging System (PerkinElmer Life Sciences).
FLEXstation TM Analysis-This was as previously described (24). Briefly, cells were gently rinsed twice with buffer (10 mM HEPES, 115 mM NaCl, 1 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, pH 7.4). To each well 50 l of buffer and 50 l of fluorescent membrane potential dye (diluted 2:5 in buffer) were added, and the cells were incubated at room temperature for 45 min before assay. Fluorescence was measured in a FLEXstation TM (Molecular Devices Ltd., Wokingham, UK) every 2 s for 200 s using the acquisition software SOFTmax PRO version 4.3. Control (buffer alone) or 5-HT (0.001-30 M) was added to each well at 20 s. Data were analyzed by iterative curve fitting using Prism version 4.03 (GraphPad Software) according to the equation, F ϭ F min ϩ((F max Ϫ F min )/1 ϩ 10 (log EC50 Ϫ log [L])nH) ), where F min is the baseline fluorescence at 20 s, F max is the maximal fluorescence, EC 50 is the concentration required for a half-maximal response, [L] is the concen-tration of ligand, and nH is the Hill coefficient. Data are reported as means Ϯ S.E. Statistical analysis was performed using Student's t test.
Modeling-This was performed as described previously (12,25). Briefly, the three-dimensional models of the extracellular region of the 5-HT 3 receptor were built using MODELLER 6v2 (26) based on the crystal structure of AChBP in the unbound, agonist-bound, and antagonist-bound states (PDB codes: 2byn, 1uv6, 2byq, 1yi5, and 2byr). The pentamer was generated by superimposing the 5-HT 3 subunit on to each protomer of AChBP, with care taken not to alter the coordinate axes of reference. The generated pentameric model was energy minimized in SYBYL version 6.8 using the AMBER force field by moving side chains alone, to relieve short contacts at the inter-protomer interfaces (27). Electrostatic terms were included in these minimization cycles.

RESULTS
Effects of Mutations-Each amino acid along a sequence of 21 residues was mutated to either Ala or an amino acid with properties similar to the wild-type amino acid (subsequently referred to as a conserved amino acid change). The position of these residues within the linear sequence of the 5-HT 3 receptor is shown in Fig. 1 and is compared with AChBP and other Cys-loop receptors in Fig. 2.
The [ 3 H]granisetron binding affinity of the mutants is shown in Tables 1 and 2. Changing 10 of the 21 residues resulted in no significant change in affinity for either the Ala or conserved mutation, suggesting these residues do not play a role in ligand binding (Ser 197 , Glu 199 , Arg 202 , Ser 203 , Ile 207 , Phe 208 , Ile 209 , Asn 210 , Gln 211 , and Gly 212 ). For the remaining 11 residues there were differences in binding affinities compared with wild type for one or both of the substitutions, suggesting that these residues have some role in  (Table 1). These changes were clustered into two groups at the N and C termini of the linear sequence. The changes to the N-terminal cluster extended from Ile 192 , which lies immediately adjacent to the B-loop, through to Arg 196 . From Ser 197 to Ser 203 all of the substitutions displayed a [ 3 H]granisetron binding affinity similar to wild type. The C-terminal cluster was Asp 204 , Lys 205 , and Ser 206 . D204A displayed no specific binding at the concentrations of [ 3 H]granisetron used in this study. Immunofluorescence confirmed that this mutant was expressed at the cell surface, indicating that the mutation did not impair receptor assembly or membrane trafficking (Fig. 3). Binding was similar to wild type for Ala mutations from Ile 207 through to Gly 212 .
Effects of Conserved Mutations-Amino acid substitutions to residues with conserved properties revealed changes in the [ 3 H]granisetron binding affinity for 6 of the 21 residues ( Table 2). The N-terminal cluster of residues, centered around Trp 195 , that was sensitive to Ala mutation remained sensitive to conserved mutations and was extended to also include Pro 198 , Glu 200 , and Val 201 . The C-terminal cluster displayed different properties to the Ala mutants, as conserved mutations at Asp 204 , Lys 205 , and Ser 206 revealed wild-type binding affinities.   (n ϭ 8). An increase of ϳ10-fold in the EC 50 value for W195A receptors was significantly greater than wildtype receptors (3.3 Ϯ 0.5 M, n ϭ 4, p Ͻ 0.05) and was similar to previously published increases for W195Y (4-fold increase) and W195S (9-fold increase) mutants (28).

5-HT concentration of 30 M
F-loop Modeling-To gain further insight into the importance of the F-loop, a series of models were generated, based upon crystal structures of AChBP in different states; unbound (PDB ID: 2byn), agonist-bound (carbamylcholine, PDB ID: 1uv6 and epibatidine, PDB ID: 2byq), and antagonist-bound (2-methyllcaconitine, PDB ID: 2byr and ␣-cobratoxin, PDB ID: 1yi5) (8,29,30). The lowest energy state for each model is shown in Fig. 4. An overlay of the backbone carbons (Fig. 4F) for each of the models suggests that the F-loop does not undergo rigid body movements but displays conformational flexibility that is dependent upon the bound ligand. To demonstrate this point, the inter-subunit distances between the ␣-carbon of Trp 183 (principal subunit) and the ␣-carbon of either Trp 195 or Asp 204 (complementary subunit) were calculated (Table 3). Trp 183 was chosen as a reference point, because there is considerable evidence regarding the importance of this residue and it is known to be centrally located within the binding pocket (12,25,28,31,32). The distances between ␣-carbons of Trp 183 residues of adjacent subunits were included as an internal control for widespread disruption of the whole binding site.

The 5-HT 3 Receptor F-loop
Comparison of the Trp 183 to Trp 183 distances for each of the models revealed that these are similar (26.20 -27.07 Å) and indicates that any variability in the measurements of Trp 195 and Asp 204 is unlikely to be the result of macroscopic disruptions of the binding pocket region. Lateral movements of the Trp 195 ␣-carbon positions are small across the range of models, but there are significant rotational movements (Fig. 4, A-E). These are most pronounced for the antagonist-bound models, which show that the side chain of Trp 195 rotates clockwise from a 12 o'clock position to a three or six o'clock positions (Fig. 4, E and D). Distances between the ␣-carbon of Asp 204 and the ␣-carbon of Trp 183 in both the agonist-bound structures are greater than in the unbound form, suggesting that there is movement in this region during binding (Table 3). In contrast, binding of the small antagonist does not cause lateral movement, although the greatest movement of Asp 204 is seen upon binding of the large antagonist, which causes Asp 204 to move away from Trp 183 and toward the solvent.

DISCUSSION
This study describes the affects of substitutions across a 21-amino acid stretch that incorporates the 5-HT 3 receptor F-loop. Residues were mutated to either Ala or to an amino acid with chemical properties similar to the wild-type residue. Mutation of 11 of the amino acids caused changes in the binding affinity of [ 3 H]granisetron, suggesting that they have a role here. The data revealed two clusters of residues of particular importance, centered around Trp 195 and Asp 204 . Modeling, based upon the structures of AChBP bound to a variety of ligands, supports the hypothesis that the F-loop forms a flexible structure and suggests that upon ligand binding there are specific, but distinct movements of Trp 195 and Asp 204 .
Trp 195 has previously been identified as a potentially important binding residue in the 5-HT 3 receptor: both EC 50 values for 5-HT and K d values for [ 3 H]granisetron were substantially increased when Trp 195 was mutated to Tyr or Ser, similar to the observations presented here (28). As proposed in the previous study, our results support a role for this residue in binding rather than gating, and indeed it may contribute to the aromatic box that is critical for all Cys-loop receptors. Because Trp 195 is important, changes in binding affinity observed when mutating the neighboring amino acids may reflect indirect effects resulting from their proximity to Trp 195 . However, it is more likely, given the range of residues that cause a change in binding affinity when mutated (Ile 192 -Val 201 ), that other residues in this region also contribute to ligand binding, either directly or via contributions to the structure of the binding site. Our homology models show Trp 195 and Arg 196 are in close association with residues from other regions of the protein, and therefore may be involved in interactions that permit the correct folding of the binding pocket. For example, Trp 195 is Ͻ5 Å from Thr 64 and Ser 66 with which it might hydrogen bond (Table 3). Perhaps surprisingly, the charged side chains of Glu 199 and Arg 202 do not appear to be important, because both Ala and conserved substitutions do not significantly alter the binding affinity in these mutant receptors. Even an Ala substitution at Pro 198 was well tolerated, which was unexpected as studies have shown that a number of 5-HT 3 receptor Pro residues are critical for binding and/or function (33). There is not an ideal conservative substitution for proline, although changing it to His (which also has a 5-membered ring but is slightly larger) did cause an increase in K d , indicating that size may be important here. Size may also play a role at Glu 200 and Val 201 , where a change in binding affinity was observed with conservative but not Ala substitutions.
The second cluster of amino acids consists of Asp 204 , Lys 205 , and Ser 206 . D204A was the only Ala mutation in the F-loop that completely ablated binding, indicating the importance of this residue. No functional response was observed with this mutant receptor, although immunofluorescence showed it was capable of reaching the cell surface. This is consistent with the lack of function being due solely to the lack of binding. Interestingly, Asp 204 and Lys 205 displayed wild-type binding affinities when their positive charges were preserved in the conserved mutations, suggesting the formation of salt-bridges. However, eliminating the charge in the D204N and K205M mutants showed that is not the case. The amino acids aligning with Asp 204 in the GABA A ␣ subunit (Val 180 ) and nACh receptor ␥ (Asp 203 ), ␦ (Asp 208 ), and ⑀ (Asp 202 ) subunits have all been implicated in ligand binding, and the Lys 205 equivalent (Ala 181 ) in GABA A receptor ␣ subunits has also been shown to be important (4, 13, 34 -36). Therefore, it is likely that this region contributes to ligand binding in all Cys-loop receptors.
Despite the lack of structural information, there is evidence from the GABA A receptor that the F-loop may undergo a structural re-arrangement upon ligand binding (13). Using pentobarbital (which activates the channel at a site remote from the ligand binding site), methanethiosulfonate ethylammonium modification of cysteine mutants indicated that Val 180 (which aligns with Asp 204 in the 5-HT 3 receptor), Ala 181 (Lys 205 ), and Arg 186 (Asn 210 ) undergo structural changes during ligand binding and/or channel opening. It was also noted that Val 178 (Arg 202 ), Val 180 (Asp 204 ), and Asp 183 (Ile 207 ) are likely to contribute to the receptor binding site. Our homology modeling suggests that there is also movement in the 5-HT 3 receptor F-loop region, around the two clusters of residues that were identified as important in our mutagenesis studies. Our data indicate rotational movement in the region of Trp 195 and both lateral and rotational movement around Asp 204 . Trp 195 is located immediately

The 5-HT 3 Receptor F-loop
downstream of the ␤8-sheet, and lateral movement here may be limited by the anchoring effect of this rigid ␤-sheet structure. Interactions between this ␤-sheet and the region around Trp 195 could be important, because the adjacent ␤9 and ␤10 strands are part of the C-loop. Therefore, movement around Trp 195 could directly affect the C-loop and have an impact on ligand binding or subsequent receptor movement. More structural information, especially comparisons of the closed and open states of the receptor, is needed to support this hypothesis. It is also noteworthy that in some studies researchers have subdivided the F-loop (e.g. Ref. 37). Our data would support the idea that there is more than one important region in this loop, but given the structural evidence from AChBP it seems logical to consider the region as a whole until further evidence suggests otherwise.
Comparisons of AChBP and nACh receptor structure indicate that conformational changes in the binding site are predominantly within the C-and F-loop regions. The C-loop is proposed to undergo a rigid body movement that changes its position from one that is extended toward the solvent when in the resting state, to one that contracts around the ligand in the agonist-bound conformation (8,10,30). In contrast, there is strong evidence that the F-loop is flexible, and in many of the crystallization studies of AChBP the F-loop is poorly resolved due to its unordered structure (9, 10). Where we do have structural information, comparisons of the F-loop in AChBP from different species has revealed a high level of variability, supporting the hypothesis that the F-loop may play a role in subunit specific pharmacological characteristics (13,38,39).
Our models show that there are significant differences in the orientation and positioning of the residues Trp 195 and Asp 204 that depend upon the ligand but not necessarily on whether the ligand is an agonist or antagonist. The distance of Asp 204 from Trp 183 (located in the center of the binding site) was similar in both the unbound and small antagonist-bound models, consistent with thinking that antagonists lock the receptor in the resting state. However, the presence of a large antagonist appeared to cause a large lateral displacement of Asp 204 away from Trp 183 (5.96 Å), providing some support for the recent proposal that large antagonists push the components of the binding pocket further apart (30). The change in the position of Trp 195 was smaller for both small and large antagonists (2.18 and 1.63 Å). Agonists, on the other hand, appear to cause very little displacement of Trp 195 (Ͻ0.5 Å) but do appear to displace Asp 204 ; in the two agonist-bound models Asp 204 is 1.82 and 2.52 Å further away from Trp 183 than in the unbound state. These movements may not only contribute to conformational changes within the F-loop but may also influence other closely located binding loops, such as the adjacent ␤9 and ␤10 region of the C-loop.
In summary, we have shown that residues within the F-loop region, and in particular those centered around Trp 195 and Asp 204 , are critical for antagonist binding in the 5-HT 3 receptor. Combining these data with our modeling studies suggests F-loop residues are important contributors to the binding of all ligands and also have the potential to cause or influence conformational changes in or close to the binding pocket. Such changes may be common to all Cys-loop receptors.