The role of tyrosine residues in the extracellular domain of the 5-hydroxytryptamine3 receptor.

Aromatic residues play an important role in the ligand-binding domain of Cys loop receptors. Here we examine the role of the 11 tyrosines in this domain of the 5-HT3 receptor in ligand binding and receptor function by substituting them for alanine, for serine, and, for some residues, also for phenylalanine. The mutant receptors were expressed in HEK293 cells and Xenopus oocytes and examined using radioligand binding, Ca2+ imaging, electrophysiology, and immunochemistry. The data suggest that Tyr50 and Tyr91 are critical for receptor assembly and/or structure, Tyr141 is important for antagonist binding and/or the structure of the binding pocket, Tyr143 plays a critical role in receptor gating and/or agonist binding, and Tyr153 and Tyr234 are involved in ligand binding and/or receptor gating. Tyr73, Tyr88, Tyr94, Tyr167, and Tyr240 do not appear to play major roles either in the structure of the extracellular domain or in ligand binding. The data support the location of these residues on a model of 5-HT docked into the ligand-binding domain and also provide evidence for the structural similarity of the extracellular domain to AChBP and the homologous regions of other Cys loop ligand-gated ion channels.

The Cys loop family of ligand-gated ion channels plays a critical role in neuronal transmission and are also the target of many neuroactive agents; understanding their molecular details is therefore a priority in this field. Among the different family members, the 5-HT 3 1 receptor has the closest functional homology with the nicotinic acetylcholine (nACh) receptor. The nACh receptor is particularly well characterized, because although no high resolution x-ray crystal structure is known, a vast array of techniques including chemical modification, sitedirected mutagenesis, electron microscopy, and molecular modeling have provided a reasonable understanding of the structure-function relationships of the receptor. Molecular details of the N-terminal extracellular domain, which harbors the ligandbinding site, have recently been further clarified following publication of the high resolution structure of the acetylcholinebinding protein (AChBP), which is homologous to this domain (1). This work has confirmed that six regions of the sequence (loops A-F), previously indicated to be involved in ligand binding, are located in the binding pocket and has provided some insight into the molecular mechanisms of ligand recognition. More recently, superimposed structures of the AChBP and the nACh receptor have also elucidated gross structural rearrangements that occur upon ligand binding (2).
AChBP also provides a good model for the extracellular domain of the 5-HT 3 receptor (3), although molecular details of the ACh-binding site cannot be directly extrapolated to this protein. There is, however, some overlap between pharmacophore models for the nACh and 5-HT 3 receptors (4), although compounds binding to both receptors can display different affinities for the two receptors or have opposing effects. For example, the 5-HT 3 receptor antagonist d-tubocurarine (dTC) displays differential affinities at the two nonidentical nACh receptor-binding sites, and nicotine, a nACh receptor agonist, behaves as an antagonist at the 5-HT 3 receptor ligand-binding site (5). Despite this, the structural variation between ligandbinding sites responsible for these differences in action may be subtle; one report suggests the replacement of just one residue in the 5-HT 3 receptor (F130N) is sufficient to produce a receptor responsive to ACh (6). The recently published model of the 5-HT 3 receptor-binding site (7), therefore, is probably broadly accurate, even though it is based on a structure than binds ACh; however, because it is purely a homology model, it requires support from experimental data. Aromatic residues have been previously shown to be involved in ligand binding to the 5-HT 3 receptor (3). In particular a role for tyrosine residues has been demonstrated (8): pretreatment of 5-HT 3 receptors in NG108 -15 cells with the tyrosine-modifying reagent tetranitromethane caused a 30% reduction in specific binding of the 5-HT 3 receptor antagonist [ 3 H]zacopride. In addition, mutation of Tyr 94 to alanine caused a small shift in [ 3 H]granisetron binding affinity (9), and alanine substitutions of Tyr 141 , Tyr 143 , and Tyr 153 have been shown to affect both agonist and antagonist affinities and 5-HT-induced currents (10). Here we extend these studies to examine the radioligand binding, electrophysiological, and cell surface expression properties of mutants of each of the 11 tyrosine residues in the mouse 5-HT 3 receptor extracellular domain. The data reveal that specific tyrosine residues play roles in the structure and/or function of the receptor and also support the model of the 5-HT 3 receptor extracellular domain (7).

EXPERIMENTAL PROCEDURES
Materials-All of the cell culture reagents were obtained from Invitrogen, except fetal calf serum which was from Labtech International (Ringmer, UK). [ 3 H]granisetron (81 Ci/mmol) was from PerkinElmer Life Sciences. The 5-HT 3A receptor N-terminal domain antiserum, pAb120, was generated as previously described (11). 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 and 7% CO 2 in a humidified * This work was supported by funds from the Wellcome Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: 5-HT, 5-hydroxytryptamine; nACh, nicotinic acetylcholine; HEK, human embryonic kidney; WT, wild type; atmosphere. They were cultured in Dulbecco's modified Eagle's medium/Nutrient Mix F-12 (1:1) with GLUTAMAX I™ containing 10% fetal calf serum and passaged when confluent. The cells were grown in 90-mm-diameter dishes for radioligand binding studies and on 22-mmdiameter coverslips for immunocytochemistry and Ca 2ϩ imaging studies. The cells were transfected using calcium phosphate precipitation (12) at 80 -90% confluency. Following transfection the cells were incubated for 3-4 days before assay.
Site-directed Mutagenesis-The mutagenesis reactions were performed using the method developed by Kunkel (13) using the 5-HT 3A(b) subunit DNA as previously described (14). The oligonucleotide primers were designed according to the recommendations of Sambrook et al. (15) and using some suggestions of the Primer Generator (16) (www.med-.jhu.edu/medcenter/primer/primer.cgi). A silent restriction site was incorporated in each to assist rapid identification. The primers are shown in Table I with base changes shown in bold type.
Radioligand Binding-Radioligand binding was as described previously (17) with minor modifications. Briefly, transfected HEK293 cells that had been transfected with wild type (WT) or mutant DNA were washed twice with phosphate-buffered saline at room temperature. They were 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. The harvested cells were washed in HEPES with proteinase inhibitor and frozen at Ϫ20°C. After thawing, they were washed twice with HEPES buffer and resuspended, and 50 g was incubated in 0.5 ml HEPES buffer containing [ 3 H]granisetron (81 Ci/ mmol; PerkinElmer Life Sciences). Initially, single-point radioligand binding assays were performed using 1 nM and sometimes 20 nM [ 3 H]granisetron to test for specific binding. If specific binding was present, saturation binding (8 point) assays were performed on at least three separate plates of transfected cells for each mutant. Nonspecific binding was typically determined by the addition of 1 M dTC. It was noticed, however, that for assays on some mutants (Y141A and Y153A), nonspecific binding was higher than expected. For these mutants nonspecific binding was determined using 100 M dTC or 1 M quipazine, which were able to reduce nonspecific binding to expected levels. The reactions were incubated for 1 h at 4°C and terminated by rapid vacuum filtration using a Brandel cell harvester onto GF/B filters presoaked for 3 h 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. The data were analyzed by iterative curve fitting (GraphPad, PRISM, San Diego, CA) according to the following equation: B ϭ (B max [L] n )/([L] n ϩ K n ), where B is bound radioligand, B max is maximum binding at equilibrium, K is the equilibrium dissociation constant, [L] is the free concentration of radioligand, and n is the Hill coefficient Ca 2ϩ Imaging-Ca 2ϩ imaging was as described previously (18) with minor modifications. Briefly, transfected HEK293 cells were washed once with HEPES-buffered medium (HBM: 115 mM NaCl, 5 mM KCl, 0.5 mM MgCl 2 , 2 mM CaCl 2 , 25 mM HEPES, 15 mM glucose, pH 7.4) and then incubated for 30 min at room temperature in HBM containing 2 M fura-2/acetoxymetyl ester and 1 mg/ml bovine serum albumin. After washing (twice with 1 ml of HBM) and at least 30 min further incubation, the cells were washed (twice with 1 ml) in Na ϩ -free HBM where Na ϩ was replaced by N-methyl-D-glucamine, and coverslips were placed in a perfusion chamber on the stage of a Nikon Diaphot inverted microscope. 340/380-nm image pairs were collected at 5-s intervals using Metafluor software (Universal Imaging Corp.), and fluorescence ratios were formed by dividing pairs of images; these images were converted to [Ca 2ϩ ] i by reference to a look-up table created using standard solutions (Molecular Probes). Data from concentration-response experiments were normalized to maximal current (I max ) readings and, using Prism software (GraphPad, San Diego, CA), were fitted to the following Hill equation: where I max is the maximal current, [A] is the agonist concentration, n is the Hill coefficient, and EC 50 is the agonist concentration that induces half-maximal response. Significance was calculated using a one-way analysis of variance and a Dunnett post-test.
Immunofluorescent Localization-Immunofluorescent localization was as described previously (11). Briefly, transfected cells were washed with three changes of Tris-buffered saline (TBS: 0.1 M Tris/HCl, pH 7.4, 0.9% NaCl) and fixed using ice-cold 4% paraformaldehyde in phosphate buffer (66 mM Na 2 HPO 4 , 38 mM NaH 2 PO 4 , pH 7.2). After two TBS washes, the cells were incubated in pAb120 antisera at 1:1000 dilution in TBS to determine membranous expression. Intracellular receptor expression was determined by inclusion of 0.3% Triton X-100 (TTBS) for membrane permeabilization. Primary antibody incubation was overnight at 4°C. Biotinylated anti-rabbit IgG (Vector) and fluorescein isothiocyanate avidin D (Vector) were used to detect bound antibody as per the manufacturer's instructions. The coverslips were mounted in Vectashield mounting medium (Vector), and immunofluorescence was observed using a Nikon optiphot or confocal microscope.
Oocyte Preparation-Oocyte preparation was as described previously (19). Harvested stage V-VI Xenopus oocytes were washed in four changes of OR2 (82.5 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 5 mM HEPES, pH 7.5), defolliculated in 1 mg/ml collagenase for ϳ1 h, washed again in four changes of OR2, and transferred to 70% Leibovitz medium (Invitrogen) buffered with 10 mM HEPES, pH 7.5. The following day they were injected with 5 ng mRNA produced by in vitro transcription using the mMESSAGE mMACHINE kit (Ambion) from DNA subcloned into pGEMHE (20) as previously described (19). Electrophysiological measurements were performed 24 -72 h post-injection.
Electrophysiological Recordings-Two-electrode voltage clamping of Xenopus oocytes was performed using standard electrophysiological procedures as previously described (19), with minor modifications. Briefly, a GeneClamp 500B amplifier was connected to a PC running CLAMPEX version 6.0.3 software via a DigiData1200 Series Interface (all Axon Instruments, Inc.). Glass microelectrodes were pulled from GC150TF-10 glass capillaries (Harvard Apparatus) using a P-87 micropipette puller (Sutter) to a resistance of 0.5 M⍀ and back-filled with 3 M KCl. 3 M KCl agar bridges connected ground electrodes to the bath. The oocytes were maintained at a holding potential of Ϫ30mV unless stated otherwise and perfused continuously with calcium-free frog ringer (115 mM NaCl, 2.5 mM KCl, 1.8 mM MgCl 2 , 10 mM HEPES, pH 7.5) at a rate of 3-4 ml/min. Serotonin (creatinine sulfate complex; Sigma) was diluted in calcium-free frog ringer and applied to the bath using a Valve Bank 8 II system (Automate Scientific, Inc.). Doses of 5-HT were applied at 3-min intervals to allow for recovery from desensitization. Concentration-response curves and parameters were obtained using Prism software as described above.

Radioligand Binding Properties of Wild Type and Mutant
Receptors-Radioligand binding assays were initially performed with the alanine and serine mutants of the 11 tyrosine residues. Specific, saturable binding was detected for 12 of the 22 mutants but could not be detected for Y50A, Y50S, Y91A,   Table II.

Immunofluorescent Localization of Wild Type and Mutant
Receptors-To ascertain whether the lack of binding observed for some tyrosine mutants (Y50F, Y50A, Y50S, Y91A, Y91S, Y141S, Y153S, Y234A, and Y234S) was due to an ablation of ligand binding ability or to a lack of cell surface receptor expression, the localization of these mutant receptors was investigated using an antisera specific to the extracellular domain of the 5-HT 3 receptor (11). Intracellular staining was observed in all cells transfected with WT or mutant receptors (Fig. 1, lefthand panels) but not mock transfected cells. Conversely cell surface staining, visualized as a well defined ring of fluorescence and indicating that the receptors are expressed on the plasma membrane (11,(21)(22)(23), was observed only for some mutant receptors (Fig. 1, right-hand panels). Thus, cells transfected with Y50F, Y50A, Y50S, Y91A, and Y91S mutants showed reduced levels or an absence of cell surface staining compared with cells transfected with WT subunit DNA (which had a robust ring of fluorescence around the cells), indicating that these subunits are not efficiently transported to the cell surface.
Functional Characterization of Wild Type and Mutant Receptors-Mutant receptors for those tyrosine residues suspected of being involved in the function of 5-HT 3 receptors were examined using calcium imaging of DNA expressed in HEK293 cells and, for some, also in oocytes. For WT receptors the concentration of 5-HT required to evoke a half-maximal response (EC 50 ) was 1.47 Ϯ 0.42 M in HEK293 cells and 2.39 Ϯ 0.23 M (n ϭ 5) in oocytes (Tables III and IV). The responses were blocked by the 5-HT 3 receptor antagonist granisetron and desensitized in the continued presence of agonist as previously described (22,24,25). Typical responses from WT receptors expressed in oocytes are shown in Fig. 2.
Calcium imaging studies revealed that responses to 5-HT (up to 1 mM) could not be detected in HEK293 cells transfected with the alanine and serine mutations of Tyr 50 , Tyr 91 , Tyr 141 , or Tyr 234 . In addition no responses to 5-HT were observed with cells expressing Y50F-containing receptors, but responses with EC 50 s not significantly different from WT were observed in cells expressing Y91F and Y141F receptors, with values significantly greater than WT being observed for Y234F receptors. Both Y153A and Y153S receptors also had EC 50 values significantly larger than WT receptors, although not as great as Y143S receptors, which had an EC 50 Ͼ 500 M. The data are shown in Table III. The wave forms of all the functional receptors were similar to WT, suggesting no major changes in the kinetics of the response.
For mutations that we suspected may affect receptor assembly (Tyr 50 and Tyr 91 ), receptors were also expressed in oocytes. Oocytes are generally more tolerant than mammalian cells to expression of ion channel proteins, which probably require longer periods to fold correctly; the difference may arise primarily from the fact oocytes are incubated at lower temperatures (26). Here responses to 5-HT could not be detected in oocytes injected with Y50A and Y50F 5-HT 3 receptor mRNA, but oocytes injected with Y50S mRNA responded to 5-HT with an EC 50 and Hill coefficient not significantly different from WT (Fig. 3). The responses also had similar wave forms to WT receptor, indicating no significant change in response kinetics. I max values for this mutant, however, were noticeably reduced, with maximal currents only 13% those of WT. A comparison of the responses obtained is shown in Table IV. Y91A, Y91S, and Y91F mutant 5-HT 3 receptors expressed in Xenopus oocytes elicited inward currents upon application of 5-HT. Y91F receptors appeared similar to WT, but, as shown in Table III, the EC 50 values for the activation of Y91A and Y91S were significantly larger than WT, with a ϳ6-fold increase in the EC 50 of Y91A and an ϳ24-fold increase in the EC 50 of Y91S compared with WT (Fig. 3). There was no significant change in Hill coefficients or wave forms for any of these mutants. DISCUSSION To identify the roles of tyrosine residues in the extracellular domain of the 5-HT 3A receptor, we substituted the 11 tyrosines in this domain for alanine, for serine, and, for some residues, also for phenylalanine. The binding, function, and cell surface expression of the mutant receptors were then examined using the HEK293 and Xenopus oocyte expression systems. The data suggest that more than half of these residues play important roles in either the structure or the function of the receptor.

Tyrosine Residues in the 5-HT 3 Receptor Extracellular Domain
Thus, Tyr 50 and Tyr 91 are critical for correct receptor assembly and/or structure, whereas Tyr 141 , Tyr 143 , Tyr 153 , and Tyr 234 play roles in the structure and/or function of the receptor. These residues are discussed further below. Alanine and serine substitution of the tyrosine residues Tyr 73 , Tyr 88 , Tyr 94 , Tyr 167 , and Tyr 240 resulted in mutant re-ceptors that had [ 3 H]granisetron binding properties not significantly different from WT receptors. This competitive antagonist binds with high affinity in the same binding pocket as 5-HT (e.g. 10) and is one of a selection of radioligands that have been used in similar conditions to probe the binding pocket (6,9,21); thus, we can been reasonably confident that this pocket is not significantly altered and that Tyr 73 , Tyr 88 , Tyr 94 , Tyr 167 , and Tyr 240 do not play major roles in either the structure of the extracellular domain or in ligand binding.
These observations are supported by the placement of these residues on a recent model of the 5-HT 3 receptor extracellular domain based on the structure of AChBP (7). Docking of 5-HT to the ligand-binding site on this model revealed seven possible agonist orientations, of which only two were supported by experimental evidence. These two only differ slightly (in the orientation of the primary amine), and because we cannot yet determine which of these is most favorable, we have chosen the orientation designated model 4 in Reeves et al. (7). This model supports our proposal that a number of tyrosine residues do not have significant roles. Thus, Tyr 73 and Tyr 88 are located some distance from the ligand-binding site and are not positioned in   Tyr 167 and Tyr 240 are not situated in the proposed binding site or near the subunit interface, although interestingly Tyr 167 is in the Cys-Cys loop, which has been implicated to interact with the extracellular linker between M2 and M3 to propagate the allosteric transition leading to channel opening (27). Thus, although our data indicate that this residue does not play a role in either the global structure or the structure of the receptor binding pocket, they do not preclude the possibility that it may be involved in transducing agonist binding into channel opening. However, because tyrosine is not well conserved at this position in other Cys loop receptor subunits (see Fig. 5), it is unlikely that this residue plays a significant role.
Tyr 50 -Our data suggest that Y50 mutant receptor subunits are not efficiently transported to the cell surface. No specific binding of [ 3 H]granisetron could be detected in HEK293 cells transfected with Y50A, Y50S, or Y50F mutant receptor DNA, and immunocytochemical experiments with pAb120, under conditions where it would only label cell surface 5-HT 3 receptors, revealed that few, if any, receptors were expressed at the plasma membrane.
However, expression of Y50S in oocytes revealed functional receptors, albeit at apparently low concentrations as judged by the low I max . This supports previous studies that have shown that complex vertebrate proteins such as ion channels, particularly when modified in some way, may express more reliably in oocytes than in a vertebrate expression system (28). The explanation for this is not yet clear, although it may be a result of oocytes being routinely incubated at lower temperatures than HEK293 cells, thus favoring correct folding (26). It also suggests that the assembly of such proteins is perhaps less tightly regulated in oocytes. Whatever the explanation, the data with Y50S strongly indicate that this residue plays a structural role in which the hydroxyl group is critical. Lack of a hydroxyl group, as in the Y50A and Y50F receptors, resulted in nonfunctional receptors, and we propose that removal of the hydroxyl renders them structurally inadequate to reach the plasma membrane. This hypothesis is supported by the model. Here Tyr 50 is positioned at the end of the ␣-helix at the top of the structure, where its closest neighbor (3.1 Å distant) is Asp 118 . If tyrosine is replaced by serine its hydroxyl is more distant to Asp 118 but is relatively close (3.2 Å distant) to S119 (Fig. 4). We postulate that a hydrogen bond interaction between Tyr 50 and Asp 118 is critical for the structural integrity of the receptor. A hydrogen bond might still be able to form when Tyr 50 is substituted with serine, although it would be less favorable, because the distance between the hydroxyl and its partner residue would be less optimal. This hypothesis would explain the WT EC 50 of Y50S receptors and also the lack of expression of Y50A and Y50F receptors; structurally inaccurate or unassembled receptors do not normally reach the cell surface (29).
An alternative explanation is that a hydrogen bond here forms an intersubunit contact. Tyr 50 faces the subunit interface and therefore has the potential to form a head-to-tail interaction with the ␣-helix of the adjacent subunit. In the model adjacent subunits are not close enough to interact at this point (Ͼ5 Å distant), but because the model is based on the desensitized state of the receptor it may be inaccurate in this region, or alternatively these locations may be closer in the nascent protein, allowing this region to interact to ensure correct folding. In support of this hypothesis a potential role in receptor oligomerization of the nACh receptor ␣7 subunit residues Gly 23 -Asn 46 , which includes the tyrosine residue homologous to Tyr 50 , has recently been reported (30).
Thus, our data suggest that Tyr 50 is critical for the assembly of the receptor, perhaps by allowing the formation of the correct structure at the subunit interface. In support of this important role, this residue is highly conserved among the Cys loop receptor subunits (Fig. 5; see also www.pasteur.fr/recherche/ banques/LGIC/LGIC.html), suggesting a function common to all subunits in this receptor family.
Tyr 91 -Data obtained from Y91 mutant receptors suggest that an aromatic residue is essential here for expression in HEK293 cells; no function and poor cell surface expression were observed when it was replaced with alanine or serine, although wild type EC 50 values were obtained when it was replaced by phenylalanine. All three mutant receptors, however, were functional in oocytes, perhaps indicating that this residue plays a role in receptor assembly. In the oocyte expression system there was a larger shift in EC 50 for the serine mutant compared with the alanine mutant, suggesting that a hydrophobic substitution is less disruptive than a hydrophilic one, and indeed the equivalent residue is conserved as hydrophobic in all ligand-gated ion channel subunits (Fig. 5). The model of the binding site places Tyr 91 at more than 5 Å from 5-HT, and thus it is unlikely that it would participate directly in ligand binding. However, there is evidence that the adjacent residues, Trp 90 and Arg 92 , which are within 5 Å of 5-HT, are involved in such an interaction (9,22). Previous data suggest that this region has a ␤-sheet composition (9); thus, if Trp 90 and Arg 92 are involved in ligand interaction, then Tyr 91 would face away from the binding site. We therefore propose that Tyr 91 is in a hydrophobic location and is important for the correct structure of this part of the binding pocket, possibly thereby permitting correct assembly.
Tyr 141 , Tyr 143 , and Tyr 153 -The [ 3 H]granisetron binding data suggest that aromatic residues may be required at positions 141 and 153 for antagonist binding; large increases in K d are seen for Y141A and Y153A mutants (28-fold and 11-fold, respectively), whereas Y141S and Y153S mutants do not appear to be able to bind [ 3 H]granisetron. Similarly Venkataraman et al. (10) observed decreases in [ 3 H]granisetron binding affinity for Y141A (Y142A using their numbering) and Y153A (Y152A) mutant receptors. Interestingly they report a 50-fold decrease in affinity for dTC (50-fold) in Y141A receptors, the antagonist that we initially used in the present study to determine nonspecific binding. This therefore clarifies why specific binding of [ 3 H]granisetron was difficult to detect with Y141A receptors (see "Experimental Procedures") unless higher concentrations of dTC were used or it was replaced by quipazine. Tyr 143 , however, does not appear to be required for antagonist binding; [ 3 H]granisetron binding affinity is not significantly different from WT when this residue is replaced with phenylalanine, alanine, or serine. This differs from the reports of Venkataraman et al. (10), but they do report only a relatively small decrease in affinity (ϳ3-fold).
Tyr 141 , Tyr 143 , and Tyr 153 align with residues in the nACh receptor binding loop E. Extensive studies on the nACh receptor and the ␥-aminobutyric acid A receptor have indicated the importance of residues within this binding loop as determinants of antagonist affinity (31)(32)(33)(34)(35)(36). The lack of sequence conservation in this region in particular is thought to confer the differences in antagonist selectivity. Our data support the role of this region in antagonist binding and suggest that Tyr 141 and Tyr 153 are either directly involved in an interaction with granisetron or shape the binding site such that granisetron binding is possible.
The functional data, however, suggest a different pattern of critical residues for agonist binding and/or gating. Thus, Y141A and Y141S receptors did not function when expressed in HEK293 cells, as previously observed for Y141A receptors (9). Tyr 141 is proposed to be more than 5 Å from 5-HT and located just outside the binding pocket; it would therefore be unlikely to participate directly in agonist binding, although it could still bind larger antagonists. It is also possible that this residue may be involved with the structure or assembly of the ligand binding microdomain. Y143A receptors were also nonfunctional, as shown previously (9), but Y143S receptors did respond to high concentrations of 5-HT. This large change in EC 50 (a value that incorporates both agonist binding and gating) combined with no change in antagonist binding affinity strongly suggests that this residue is involved solely in receptor gating. However, because Tyr 143 is probably located within 5 Å of 5-HT and is in the binding pocket, a specific role for this residue in the binding of agonists cannot yet be excluded. Y153A and Y153S receptors were functional (as has been shown previously for Y153A receptors (9)) but showed large increases in their EC 50 values. Interestingly, however, the relative changes in EC 50 and K d compared with WT were different for the two mutants, suggesting that Tyr 153 may be involved in both binding and gating.
In other Cys loop receptors, residues in this region have been suggested to play a role in gating, although their precise roles have not yet been elucidated. Data from nACh and ␥-aminobutyric acid A receptors, for example, indicate that E loop residues are involved in the allosteric transitions leading to channel opening (37,38). This region of the sequence lies adjacent to the Cys-Cys loop and could therefore provide a direct link between the receptor-ligand interactions at the binding site and the conformational rearrangement that enables channel gating. The pattern of residues identified by affinity labeling, mutagenesis, and cysteine substitution of the nACh receptor is consistent with an anti-parallel ␤-sheet (with the two strands linked by a turn around the central glycine residue), and such a structure is seen in the equivalent region of AChBP (1). The model and our data are consistent with a similar ␤-strand in the 5-HT 3 receptor, which would position Tyr 143 and Tyr 153 on the same face and pointing into the binding pocket. They would therefore be in a good position both to interact with agonists and also to propagate a conformational change to the nearby Cys-Cys loop. The data therefore suggest that Tyr 141 plays a role in antagonist binding and/or the structure of the binding pocket, Tyr 143 is critical for receptor gating and/or agonist binding, and Tyr 153 is involved in both binding and gating.
Tyr 234 -Tyr 234 aligns with nACh receptor ␣ subunit Tyr 198 , a residue shown to be important for binding as part of binding loop C. This residue is conserved as aromatic in almost all the subunits of this superfamily (Fig. 5); thus, it is not surprising that changing it to serine or alanine in the 5-HT 3 receptor ablates both binding and function. Substitution with an alter- native aromatic, phenylalanine, resulted in changed functional receptor characteristics (an ϳ10-fold increase in EC 50 ) but not in binding affinity (K d not significantly different from WT), thus supporting a role for the hydroxyl group in receptor gating. Extensive site-directed mutagenesis of the residue at this location has shown that it is important in the nACh receptor (39), the ␥-aminobutyric acid A receptor (40), and the glycine receptor (41), although there is some debate as to whether it affects the binding affinity (42) or the gating constant (43). The binding of dTC and its analogues to the nACh receptor has been particularly well studied, and there was evidence that these compounds might interact via cation-interactions with Tyr 198 (35). Subsequent incorporation of fluorinated tyrosine derivatives revealed that this does not occur for the nACh receptor (44), but interestingly the equivalent residue does appear to form such an interaction in the MOD-1 receptor (45). This suggests that even with similar binding pockets, there are differences between the mechanisms by which related receptors bind their ligands in the Cys loop receptor family. Thus, although we cannot yet determine the exact role of Tyr 234 in the 5-HT 3 receptor, our data show that it plays an important role in both binding and gating of this receptor.
Conclusion-The work presented here aimed to assess the contribution of tyrosine residues in the 5-HT 3 receptor extracellular domain to receptor structure and function. The radioligand binding data have allowed us to explore the integrity of the binding site in the extracellular domain and also to examine residues that may be involved in antagonist binding, whereas functional studies with 5-HT have probed agonist binding and/or gating. The data suggest that Tyr 73 , Tyr 88 , Tyr 94 , Tyr 167 , and Tyr 240 are not critical; nonconservative mutations do not affect [ 3 H]granisetron binding. However, all the remaining tyrosine residues do appear important for correct receptor structure and/or function. Thus, Tyr 50 and Tyr 91 appear critical for correct receptor assembly and/or structure, and Tyr 141 may also fall into this category, although it also plays a role in antagonist binding. Finally Tyr 143 , Tyr 153 , and Tyr 234 are involved in binding and/or gating of the receptor. The data support the location of these residues on a model of 5-HT docked into the ligand-binding domain and also indicate the strong structural similarity of the extracellular domain with AChBP and the homologous regions of other members of the Cys loop ligand-gated ion channel family.