Neurophannacology,Vol. 36, No. 4[5, pp. 637447, 1997 @ 1997Elsevier Science Ltd. AUrights reserved Printed in Great Britain 0028-3908/97$17.00 + 0.00 P PII: S0028-3908(97)00044-0 Analysis of the Ligand Binding Site of the 5-HT3 Receptor Using Site Directed Mutagenesis: Importance of Glutamate 106 F. G. BOESS,]T L. J. STEWARD,l J. A. STEELE,l D. LIU,l J. REID,* T. A. GLENCORSE’~ and I. L. MARTIN’* IDepartment of Pharmacology,Faculty of Medicine, 9-70 Medical Sciences Building, University of Alberta, Edmonton, T6G 2H7, Canadaand 2GlaxoInstitutefor Molecular Biology S.A., 14 Chemin des Aulx, Case Postale 674, CH-1228Plan-les-Ouates, Geneva, Switzerland (Accepted 15 January 1997) Summary—The 5-HT3 receptor is a ligand-gated ion channel with significant structural similarity to the nicotinic acetylcholinereceptor. Severalregions that form the ligand binding site in the nicotinic acetylcholine receptor are partially conservedin the 5-HT3receptor, presumablyreflecting the conservedsignal transduction mechanism. Specific amino acid differences in these regions may account for their distinct ligand recognition properties. Using site-directed mutagenesis, we have replaced one of these residues, glutamate 106 (EI06), with aspartate (D), asparagine (N), alanine (A) or glutamine (Q) and characterized the ligand-binding and electrophysiologicalpropertiesof the mutantreceptors after transientexpressionin HEK-293cells. The affinity for the selective 5-HT3 receptor antagonist [3H]GR65630 was decreased 14-fold in the mutant E106D (K~= 3.69 + 0.32 nM) when compared to wildtype (WT, E106) 5-HT3receptor (0.27 + 0.03 nM), while the affinity for E106N was unchanged (0.42 + 0.07 nM, means~ SEM,n = 3–10).Decreasedaffinitiesforboth E106D and E106N were observed for the antagonists granisetron, ondansetron and renzapride and for the agonists 5-HT (130- and 30-fold) and 2-methyl-5-HT (250- and 20-fold), respectively. Both mutants still formed 5-HT-activatable ion channels, but the high Hill coefficient of the concentration effect curves in wildtype (2.0) was decreased to unity in both cases. The EC=joof 5-HT was increased seven-fold in E106N (8.7 vM) when compared to wildtype (1.2 PM), but unchanged in E106D, and the potency of the antagonist ondansetron for both mutants was decreased. E106A and E106Q expressed poorly preventing a detailed characterization. These data suggestthat E106 contributes to the ligand-bindingsite of the 5-HT3receptor and may form an ionic or hydrogenbond interaction with the primary ammoniumgroup of 5-HT. @ 1997Elsevier Science Ltd. Keywords—5-HT3receptor, mutagenesis, ligand recognition. for maximum receptor activation (Jackson and Yakel, 1995). In 1991, Maricq and colleagues isolated a cDNA encoding a 5-HT3 receptor subunit from a mouse neuroblastomax chinese hamster brain cell hybrid (NCB20) cDNA library (Maricq et al.). The sequence of the cloned subunit is 20--3O%identical to various yaminobutyric acid A (GABAA), glycine, and nicotinic acetylcholine(nACh) receptor subunits.The members of this superfamily share a common hydropathy profile including a large hydrophilic N-terminal domain that is *To whomcorrespondence shouldbe addressed.Tel: Canada thought to be extracellular and four characteristically (403)-492-0511; Fax: Canada (403)-492-4325; E-mail: spaced hydrophobic regions that have been proposed to Ian.Martin@UAlberta.Ca. span the cell membrane. Biochemical and electron tPresent address: F. Hoffman-LaRoche AG, PharruaDivision, Preclinical Research, P.O. Box, CH-4070 Basel, Switzer- microscopical studies confirm that the quaternary structure of the 5-HT3receptor is similar to that of the nACh land. $Present address: Institute of Biomedical and Life Sciences, receptor (McKernanet aZ.,1990a,b;Lummis and Martin, Division of Molecular Genetics, Robertson Building, 1992; Unwin, 1993; Green et al., 1995; Boess et al., 1992a, 1995). The receptor complex is a pentamer University of Glasgow, Glasgow, G11 6NU, U.K. The biological actions of serotonin (5-hydroxytryptamine, 5-HT) are mediated by at least 14 different receptors, all of which are G-protein coupled, with the exception of the 5-HT3receptor, which is a ligand-gated ion channel (Boess and Martin, 1994).Agonistactivation of the 5-HT3 receptor produces a depolarizing response which desensitizes in the continued presence of agonist. The response exhibits positive cooperativity suggesting that occupation of two agonist binding sites is necessary 637 638 F. G. Boess et al. N 1 ‘A~c Fig. 1. Schematic representation of a ligand-gated ion channel subunit. Dark boxed areas represent the hydrophobic (putative transmembrane) regions, C and N represent the carboxy- and amino-ten-ninusregions, respectively. Loops 1, 2 and 3 are the N-terminal regions contributing to the ligand binding site, as identified in biochemical and site-directedmutagenesisstudies.A comparisonof the amino acid sequencesof the firstproposed recognitionloop, in the mouse nACh ct7and 5-HT3ALreceptor subunitsis shown.Boxed areas indicate conserved amino acids. Tyrosine 93 (Y93) in the nicotinic u subunitshas been identifiedas part of the ligand recognition site of the nACh receptor. Glutamate 106 (E106) in the 5-HT3ALreceptor has been mutated in the present study to investigate its involvement in ligand recognition in the 5-HT3receptor. formed by five identical or related subunits arranged symmetricallyaround a central pore (Boess et al., 1995). In contrast to the other ligand-gated ion channels, the native 5-HT3 receptor may be a homo-oligomer, since only a single subunithas been identifiedin mouse,rat and human (Hope et al., 1993;Isenberg et al., 1993; Uetz et al., 1994; Werner et al., 1994; Belelli et al., 1995; Miyake et al., 1995). However, in the mouse and rat, there are two splice variants, 5-HT3-A~and 5-HT3A~, which differ by the absence or presence of five or six amino acids in the large intracellular loop region, between hydrophobic regions three and four (Hope et al., 1993; Werner et al., 1994; Miyake et al., 1995; Miquel et al., 1995). Evidence obtained from biochemical and mutagenesis studies with the nACh, GABAA and glycine receptors, suggeststhat the binding site for agonistsand competitive antagonistsis located in the N-terminal domain (Kuhseet al., 1995; Dunn et al., 1994; Changeux et al., 1992; Karlin and Akabas, 1995). A chimeric protein, comprising the N-terminal domain of the U7 neuronal nAChR with the remaining sequence derived from the 5-HT3 receptor, showed ligand-gated cation channel activity with ct7-likepharmacology, proving that the N-terminal domain is essential for ligand-bindingspecificity(Eise16 et al., 1993). Labelling studies with irreversible ligands identified three regions (“loops”) in the N-terminal domain of the nAChR a subunits that interact directly with ligands; additional residues in the d and y subunits were labelled by some ligands (reviewed in Changeux et al., 1992). A role for several of these residues in ligand binding has been confirmedby site-directedmutagenesis (reviewed in Changeux et al., 1992; Karlin and Akabas, 1995). Acetylcholine mustard aziridinium is an agonist in which a reactive crosslinking group is located in the position of the positivelycharged quaternary ammonium. In To~edo electric organ nAChR, this irreversibleligand labels tyrosine 93 in the LXsubunit (Cohen et al., 1991). The same residue is also labelled by the photoactivatable antagonist p-dimethylamino-benzenediazonium fluoroborate (DDF; Galzi et al., 1990).Mutationsof this Tyr to Phe, Trp, Ser, Thr and several unnatural phenykdanine derivativessuggestedthat the aromatichydroxylgroup of Tyr forms a salt bridge or acts as hydrogen bond donor (Grdziet al., 1991a;Aylwin and White, 1994;Sine et al., 1994; Nowak et al., 1995). All 5-HT3 ligands possess a positively charged nitrogen, frequently embedded in an aliphatic or aromatic heterocycle, that may bind to the 5-HT3 receptor in a position homologous to the site of interaction of the quaternary ammonium group of acetylcholine with the nACh receptor. Amino acid sequence alignment of the 5-HT3 receptor with the U7 nAChR in the region containing tyrosine 93 is shown in Fig. 1. While significantsequence homology is apparent, there are marked differences in the 5-HT3 receptor sequencein the immediatevicinity of a7 nAChR tyrosine 93. Antagonist modelling studies predict that the positively charged nitrogen in high affinity 5-HT3 receptor ligands will form an ion pair with a negatively charged aspartate or glutamate residue on the receptor (Gozlan and Langlois, 1992).In order to examine the role of glutamate 106 in the recognition of agonists and antagonists, we have modified a 5-HT3AL receptor sequence isolated from NG108-15 cells (Werner et al., 1994) using site-directed mutagenesis. Whole-cell patch clamp and radioligand binding have been used to characterizewildtypeand mutantreceptors after transient expression in human embryonic kidney (HEK 293) cells. A preliminary report of this work was presented at the British Pharmacological Society (Steward et al., 1996). MATERIALS AND METHODS Site-directed mutagenesisof the 5-HT3-ALcDNA A cDNA isolated from NG108-15 cells encoding the mouse 5-HT3-AL receptor (Werner et al., 1994) was Mutagenesis of the 5-HT3receptor generously provided by Dr Eric Kawashima (Glaxo Molecular Biology Institute, Geneva). The cDNA sequence was subcloned into the phagemid pAlter (Promega, Madison, WI, U.S.A.) and the eukaryotic expression vector pRC/CMV (Promega). Mutagenesis was performed using the Altered Sites Mutagenesis Kit (Promega). Glutamate (E) 106 was mutated to aspartate 639 50 mg/1 and phenylmethyl sulphonyl fluoride (PMSF) 100 pM) and homogenized using an ultra-turax (20000 rpm, 10 see). The homogenate was centrifuged (27000g for 20 tin at 4“C), the resulting membrane pellet gently resuspended in ice-cold 10 mhl HEPES (4(2-hydroxyethyl) -l-piperazineethanesulphonic acid) buffer (pH 7.5) and frozen at –20°C until further use. For the (D), asparagine (N), glutamine (Q), or alanine (A) using assay, the whole cell membranes were recentrifuged the following mutagenic oligonucleotides(WT is shown (27OOOgfor 20 tin at 4°C), and resuspended in ice-cold HEPES. for comparison); WT 5’ AGA-CTT-CCC-CAC-GTC-CAC-AAA-CTC-A'lT-GAT-GAG-AAT-GTC 3’ E106D 5’ AGA-CTT-CCC-CAC-GTC-GAC-AAA-GTC-ATI'-GAT-GAG-AAT-GTC 3’ E106N 5’ AGA-CH'-CCC-CAC-GTC-GAC-AAA-ATT-ATT-GAT-GAG-AAT-GTC 3’ E106Q 5’ AGA-CTT-CCC-CAC-GTC-GAC-AAA-CTG-ATT-GAT-GAG-AAT-GTC 3’ EI06A 5’ AGA-CR-CCC-CAC-GTC-GAC-AAA-CGC-A~-GAT-GAG-AAT-GTC 3’ A silent mutation, present in each oligonucleotide, introduced a new Sal 1 restriction site to facilitate mutant screening. The sequence between two Bsu 361restriction sites in the 5-HT3-AL sequence encodes the aminoterminus and the firsttwo transmembranesegmentsof the 5-HT3 receptor. Mutant pAlter-5-HT3-ALvectors were digested with Bsu 361and the resulting 891 bp fragments were ligated into cut pRC/CMV-5-HT3-ALvector, to replace the wildtype Bsu 361 fragment. Mutations were confirmed by sequence analysis using an automated fluorescent sequencing system (Applied Biosystems, Foster City, CA, U.S.A.) for the coding region between and including the two Bsu 361restriction sites. Transientexpression and preparation of membranes Human embryonic kidney (HEK 293) cells were cultured in Dulbecco’s modified Eagles medium (DMEM) containing 10% fetal bovine serum, 100U/ml penicillin, 100 pghnl streptomycinand 2 mM glutamine. The cells were incubated in a humidified 7% COZ atmosphereat 37°C, in 150 mm diameterplates or 35 mm diameter plates (for electrophysiologicalstudies). HEK 293 cells were transiently transected with 1 pg of pRC/CMV-5-HT3-ALor mutant cDNAs for electrophysiologicalstudies using a modificationof the calcium phosphate coprecipitation method (Chen and Okayama, 1988)and incubated under 3% C02. Eighteen hours after transection, the cells were washed with Hanks buffered salt solution, fresh media (DMEM) added and the cells subsequentlyincubated under 770 C02. Recordingswere made 1–2days after transection. For radioligandbinding studies, HEK 293 cells were transiently transected with 35-45 pg of pRC/CMV-5-HT3-ALor mutant cDNAs. Forty-eight hours after transection the cells were harvested in ice-cold protease inhibitor buffer (Tris 10 mM pH 7.5, ethylenediaminetetra-aceticacid (EDTA) 1 rdvl, bacitracin 50 mg/1, soybean trypsin inhibitor [3H]GR65630radioligandbinding assay For [3H]GR65630 binding, assay tubes contained 800 PI of competing drug (non-specific binding was defined in the presence of 300PM metoclopramide), or vehicle (HEPES 10 mM buffer, pH 7.5) and 100 pl [3H]GR65630 (0.04-12 nM for saturation studies and 0.41–1.85nM for competition studies). The assay tubes were preincubatedfor 2 min at O“Cbefore the additionof 100pl cell membranes (equivalent to approximately 100pg protein) to initiate binding which was allowed to proceed at O°C for 2 hr. It was terminated by rapid filtration under vacuum through Whatman GF/B filters (pretreated with 0.3% v/v polyethyleneimine in HEPES buffer), followedby washingwith ice-cold HEPES buffer for 8 sec. Bound radioactivity remaining on the filters was determined in 4.5 ml Ecoscint A (National Diagnostics) by liquid scintillation spectroscopy, at an efficiency of approximately6090. Electrophysiology Membrane currents from single cells were recorded under voltage clamp with the whole-cell configurationof the patch clamp technique using an Axopatch-lD amplifier (Axon Instruments, Foster City, CA, U.S.A.). All currents were subject to initial run-down within the first few minutes of recording and were allowed to stabilize before data collection began. Data acquisition, storage and analysis were performed using pClamp 6 (Axon). Cells expressing receptors could be identified visuallyusing phase contrastmicroscopy.These cells had clusters (up to six) of small (approximately 2 pm in diameter) phase bright “inclusion bodies”; 90’%0 of these cells generated measurable currents. Borosilicate glass pipettes were heat-polished and had resistances of 2– 5 M!2 The membrane potential was held at –60 mV and inward currents are displayed downwards. Ligands were applied to the cell using a gravity-feed rapid perfusion F. G. Boess et al, 640 Materials HEK 293 cells were from the American Type Culture Collection (ATCC, Rockville, MD, U.S.A.), DMEM, Hank’s buffered salt solution (HBSS) and penicillin/ streptomycin from BioWhittaker (Marysville, MD, U.S.A.), and fetal bovine serum (Life Technologies, Burlington,Ontario, Canada). 5-Hydroxytryptaminewas obtained from Sigma, meta-chlorophenylbiguanide, 2methyl-5-HT,metoclopramidefrom Research Biochemical Inc. (Natick, MA, U.S.A.), [3H]GR65630 (61.464.4 Ci/mmol) from NEN Dupont, ondansetron was donated by Glaxo (Ware, U.K.) and renzapride and granisetron were donated by SmithKline Beecham Pharmaceuticals(Harlow,U.K.). All drugs were prepared in HEPES 10 mM (pH 7.5). Other chemicals and reagents were purchased from Sigma Chemical Company, BDH (Toronto, Ont., Canada) and FischerBiotech (Nepean, Ont., Canada). (A) [3H]GR65630 Free [nM] -600 z = 500 % 400 g.- 300 ~ 200 r% 100 -# ob o ‘B) 20 40 ‘/0 60 60 100 Specific Fig. 2. [3H]GR65630saturation binding in HEK 293 cells transiently transected with 5-HT3-ALwildtype (WT = E106 (~)) or mutant (E106D (0)) cDNA. (A) Results shown are from a single experiment. K~ values were determined with an iterative curve fitting program. Data is from a single representative experiment which was repeated 3–10 times. Mean K~values are found in Table 1. For E106D, an additional point was obtained at 11.76nM representing over 90% saturation. Specific binding was defined by 300 PM metoclopramide. (B) Scatchard transformation with normalized specific binding (I&u = 100%),to compensate for differences in transection efficiencies for WT and E106D. system based on that described by Carbone and Lux (1987). The cell was continuouslyperfused and solution changes were effected by a manually-operated valve which housed a manifold connected to solution reservoirs. The bath solution contained (in mM): NaCl, 130; KC1,5; CaC12,1.8; MgC12,1.2; HEPES, 10; glucose, 5; pH 7.4. The pipette solution contained (in mM): CSC1, 135; MgC12, 0.5, l,2-bis(2-aminophenoxy)ethaneN,N,N’,N’-tetraaceticacid tetrapotassium salt (BAPTA), 10; HEPES, 10; pH 7.2 with CSOH.All recordings were made at room temperature (19–21°C). Data analysis Radioligandbinding data were analyzed by computerassisted iterative curve fitting (Kaleidagraph, Synergy Software, Reading, PA, U.S.A.) according to the equation:B = (Bm,X[L]”)/([L]n+ (K)”), where B is bound radioligand, B~~Xis maximum binding at equilibrium, K is the equilibrium dissociation constant for saturation studies or the molar concentration of competing compound to reduce the specific binding by 50% for competition studies (IC50),[L] is the free concentration of radioligand for saturation studies or molar concentration of competing compound for competition studies and n is the Hill coefficient. The Cheng–Prusoff equation was used to calculate both the Ki values of the competing drugs in radioligand binding studies and the Kb values for antagonists of 5HT-induced currents in whole-cell patch clamp studies (Craig, 1993).For radioligand binding studies Ki = IC5~ (1+ ([L]/K~)),where IC~ois the inhibitory concentration of the competing compound to reduce the specific binding by 50%, [L] is the free concentration of radioligand and Kd the equilibrium dissociation constant of the radioligand. For whole-cell patch clamp studies, Kb = IC5~(1+ ([A]/EC50)),where IC50is the concentration of antagcinistrequired to inhibit the agonist response by 50%, [A] is the agonist concentration used to induce Table 1. Affinities of mutant and wildtype 5-HT3receptors in transiently transected HEK 293 cells Ligand [3H]GR65630(Kd) 5-HT 2-Methyl-5-HT m-CPBG Granisetron Ondansetron Renzapride WT 0.27 + 15.56 ~ 173 + 5.01 + 0.28 + 0.58 + 1.29 ~ 0.03 (10) 3.22 (8) 18 (9) 1.09 (6) 0.03 (6) 0.10 (5) 0.17 (4) E106D 3.69 i 2056 ~ 42711 + 8.37 ~ 232 ~ 15.98 ~ 28.87 + 0.32 (3) 564 (6) 9117 (3) 3.08 (3) 69 (3) 1.5 (3) 9.97 (3) E106N 0.42 + 469 + 2947 t 2.89 t 154 * 5.56 ~ 16.97 t 0.07 (3) 25 (3) 456 (3) 0.37 (3) 86 (3) 1.14 (5) 3.82 (3) Ki values of the ligands were obtained by determination of IC50values obtained in competition with [’H] GR65630,together with the Kd values obtainedby saturationwith [3H]GR65630.All values are mean (nM)~ SEM (rI determinations). 641 Mutagenesis of the 5-HT3receptor u ~ 1( I m ■ : 5 * $ 0zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA , ‘ Fig. 3. 5-HT competition for [3H]GR65630binding (l.592.80 nM) in HEK 293 cells transiently transected with 5-HT3ALwildtype (WT) or mutated (E106N)cDNA. Data are from a single representative experiment which was repeated three to eight times. Mean Ki values are found in Table 1. Fig. 4. Granisetroncompetingfor [3H]GR65630binding (0.5& 0.57 nM) in HEK 293 cells transiently transected with 5-HT3ALwildtype (WT) or mutated (E106D)cDNA. Data are from a single representative experiment which was repeated three to six times. Mean Ki values are found in Table 1. the response and the concentration of agonist required to produce a half-maximal response. Concentration-effect curves were fitted to the equaby computer-assisted iteration: Z= 1~,.J(l + ( tive curve fitting as described previously, where Z and z~= are the currents at a given agonist concentration and the maximal value, respectively. is the agonist concentration required to obtain half-maximal current and n is the Hill coefficient. showed no significant change in affinity for the mutant receptors when compared to WT (Table 1). Hill coefficients for all agonists were approximately 1, indicating that in the present study there was no evidence of agonist cooperativity in radioligand binding for WT or mutant receptors. The affinities of several 5-HT3 receptor antagonists were also altered. The largest change was observed for granisetron, with an 800-fold and 550-fold decrease in R 1 nA Radioligand binding characterization of mutant 5-HT3 receptors E106D and E106N In saturation studies with membranes obtained from HEK 293 cells transiently expressing wildtype (WT), E106D or E106N mutant 5-HT3 receptors, [3H]GR65630 labelled a homogeneous population of binding sites (Fig. 2). The binding site density varied with the efficiency of both transection and expression. Therefore, for ease of comparison, the saturation data are normalized to 11~,, in the Scatchard transformations (Fig. 2B). The affinity of [3H]GR65630 was decreased by 14-fold for the mutant E106D (Kd = 3.69 f 0.32 uM, mean ~ SEM; n =3) when compared to WT (0.27 i 0.03 nM, n = 10; Table 1, Fig. 2.); however, there was no appreciable difference in the affinity of [3H]GR65630 for the mutant E106N (Table 1). Competition for [3H]GR65630 binding by a variety of 5-HT3 receptor agonist and antagonist ligands demonstrated changes in affinity for the mutant receptors, when compared to WT. 5-HT showed a 130-fold and a 30-fold decrease in affinity for the mutant receptors E106D (m= 2056 f 564 uM, n = and E106N (469 f 25 nM, n = 3), respectively, when compared to WT (15.56 ~ 3.22 nM, n =8; Fig. 3., Table 1). Similarly, the affinity of the 5-HT3 receptor agonist 2-methyl-5-HT was decreased 250-fold for E106D and 17-fold for E106N (Table 1). However, the high affinity 5-HT3 receptor agonist meta-chlorophenylbiguanide (mCPBG) “0’” ~ 300 PA “ ~ “ ~ “0”~ 100 pA 150 pA 50 PA 2 sec Fig. 5. A comparisonof 5-HT-inducedinward currents in HEK 293 cells transiently transected with 5-HT3-AL wildtype (E106) or mutated (E106D, E106N, E106Q, E106A) cDNAs. The horizontal bar indicates the duration of the 5-HT application (WT, 50 ,uM; E106D, 100,uM; E106N, 100pM; E106Q, 100YM; E106A, 10 mM). Currents were recorded from cells voltage clam~ed at a holding Dotentialof –60 mV. F. G. Boess et al. 642 on application of 100 VM 5-HT, while both E106D and 1- E106N mutant receptors exhibited maximal current levels some 10-fold lower (Fig. 5). This may be due to low expression levels as a similar reduction was observed in radioligand binding studies. There was no reduction in the respective currents for WT and mutant receptors with application of up to 10 rnM 5-HT at a holding potential of –60 mV indicating that agonist dependent block did not occur. In concentration-effect curve studies, the EC511for 5HT was increased seven-fold (8.7 PM, n =6) for the 0.8$ 0.6E s o.40.20 w 1 1O-s 1o-7 10-s 10-5 10-4 ,.-3 5-HT (M) Fig. 6. Concentration-effect curves for 5-HT3-ALwildtype and mutant E106N receptors in transiently transected HEK 293 cells. Data are means (n = 5–10cells). Data were fittedwith the Hill equation Z= Z~J(l + (EC5~[5-HT])”)where Zand Z~u are the currents at a given 5-HT concentration [5-HT] and the maximal value respectively, and n is the Hill coefficient. affinity for E106D (232*69 nM, n = 3) and for E106N (154*86 uM, n =3) when compared to WT (0.28 f 0.03 nM, n =6; Fig. 4, Table 1). Renzapride and ondansetron showed smaller decreases in affinity of approximately 10-30-fold (Table 1). Radioligand binding characterization of mutant 5-HT3 receptorsE106A and E106Q The expression of E106A and E106Q was poor compared to WT. Specific binding of [3H]GR65630(l– 2 nM) was too low to permit acceptable radioligand binding characterization of E106A and E106Q precluding accurate Ki or Kd determinations.A direct comparison of membranes from untransfected and transected HEK 293 cells yielded “specific” [3H]GR65630 (1.8 nM) binding of 603 (wildtype), 75 (E106A), 69 (E106Q) and 408 (NG108-15 cells; fmol/mg protein; single determination). A structurallydistinct radioligand [3H]-(S)-zacopride (data not shown) also yielded low specificbinding for E106A and E106Q. Electrophysiologicalcharacterizationof wildtype 5-HT3 receptor and mutants E106D and EI06N The whole-cell patch clamp technique was used to examine the functional properties of WT and mutant receptors. WT 5-HT3 receptors transiently expressed in HEK 293 cells produced maximal currents up to 2.5 nA asparagine mutant E106N when compared to WT (1.2 PM; n =9; Fig. 6, Table 2). There was no change in EC50for 5-HT for the aspartate mutant (E106D; Table 2). The Hill coefficient for WT was 2.0 indicating positive cooperativity of agonist activation (Table 2), whereas for both mutant receptors E106D and E106N, the Hill coefficientwas significantlydifferent to wildtype and close to unity, suggesting a loss in positive cooperativity (Fig. 6, Table 2). meta-Chlorophenylbiguanide(mCPBG) is a full agonist in NIE-115 cells (Septilvedaet al., 1991) and has a lower EC50 than the natural agonist 5-HT. In our experiments, mCPBG acted as a full agonist at WT and E106D receptors. There was no significantchange in the EC50values of mCPBG for E106D (Table 2) which was consistentwith the radioligandbinding studies.However, for mCPBG, the Hill coefficient observed in concentration-effect curve with E106D was significantlyreduced compared to WT (Table 2). In order to determine antagonist potencies, relatively high concentrationsof 5-HT were utilized because of the poor expression of mutant receptors. Therefore, 10 and 20 pM 5-HT were used for E106D and E106N, respectively,to generate an appropriateagonist response, while 10PM 5-HT was used for similar studies with WT. The antagonist potency of ondansetron was reduced for both E106D and E106N mutant receptors when compared to WT receptors (Table 2). Blockade by ondansetron was readily reversed by washing (Fig. 7). The antagonist potency of renzapride was decreased for E106N (IC50, 121nM; Kb 36.7 nM; Table 2) when compared to WT (IC50,11.5nM; Kb 1.23nM), while no change was found for E106D (Table 2). These decreases in antagonistpotencies were similar to those produced in radioligand binding studies with the exception of the Table 2. Whole-cell patch clamp data for mutant and wildtype 5-HT3receptors in transiently transected HEK 293 cells m-CPBG 5-HT Receptor WT E106D E106N Renzapride Ondansetron EC50(MM) Hill EC~o(pM) Hill IC50 (nM) Kb (nM) IC50 (nM) Kb (nM) 1.2 (1.1–1.3) 1.2 (1.0-1.6) 8.7 (6.8-11.2) 2.0 (1.62.3) 1.1 (0.9-1.3) 1.0 (0.7–1.2) 0.3 (0.3-0.4) 0.4 (0.3-0.6) ND 1.6 (1.3-1.9) 0.8 (0.6-1.0) ND 6.0 (4.2-8.3) 104 (64-169) 74 (52-104) 0.64 11.1 22.4 11.5 (8.1-16.6) 15.4 (12.4-19.1) 121 (99-148) 1.23 1.65 36.5 EC511 and Hill coefficient values are given for the agonists 5-HT and mCPBG. IC~IIand Kb values are given for the antagonists renzapride and ondansetron.Data for the antagonists were generated in the presence of 10AM 5-HT for E106 and E106D, and 20 pM 5-HI for E106N. Values are given as means with 95% confidence intervals in parentheses. Data were collected from 6-13 cells for determination of each value. ND= not determined. Mutagenesis of the 5-H’T3receptor A 5-HT bidansetron 5-HT 643 5-HT 300PA 1 ““DL 100PA ~Q’75pA 4SSC B 5-HT 9rsnisetron 5-HT ~ 5-HT -L.-. --l’”pA zyxwvutsrqponmlk ““’”ti 4sec Fig. 7. Ondansetron(A) and granisetron (B) blockade of 5-HT-inducedinward currents recorded from HEK 293 cells transiently transected with wildtype or mutantE106Dor E106NcDNAs. (A) Cells were prepulsedfor 30 sec with ondansetron (WT, 10 nM; E106D, 1 PM; E106N, 100nM), or (B) prepulsed with granisetron (WT, 50 nM; E106D, 100nM) and then either 10pM (WT, E106D)or 20 PM (E106N)5-HTwas coappliedwith the antagonist. After approximately 1 hr of washing subsequentto granisetronapplication,the maximal current was still reduced, compared to initiat pre-granisetron5-HT current. This precluded a detailed study of this antagonist. renzapride interaction with E106D. The 5-HT3 receptor of up to 2PM failed to antagonize 10 mM mCPBGantagonist, granisetron, blocked the current response to induced responses for E106A receptors. The EC50of 55-HT (10 PM; Fig. 7) for E106D and WT, but recovery HT for E106Q mutant receptors was approximately from the antagonism was very slow precluding its ICSCI 50 pM (n= 7). However, mCPBG behaved as an determination.Similar experimentshave not been carried antagonist with this mutant, blocking 100 pM 5-HTinduced inward current with an IC50of approximately out with E106N. 50 nM (n= 5) but producing no response when applied Electrophysiologicalcharacterizationof mutants E106A alone. and E106Q DISCUSSION As observed in the radioligand binding studies, it appeared that E106A and E106Q mutant receptors In the nicotinic acetylcholine receptor, amino acids expressed poorly. Currents produced with 5-HT for involved in ligand recognition are concentrated in three E106Q and E106A were greatly reduced (<200 pA) when “loops” of the a subunitsand additional sites are present compared to WT (Fig. 5) preventing their detailed study. on the y and d subunits (reviewed in Changeux et al., However, estimated values for 5-HT and mCPBG for 1992; Karlin and Akabas, 1995). A comparison of the E106A indicated that their EC50values were greater than amino acid sequence of the 5-HT3 receptor and nAChR 1 mM (n= 3), an approximate 1000-foldincrease for 5- a subunits shows that there is significant sequence HT and 30000-fo~d-increase for mCPBG, when com- conservation in the first of these recognition loops (Fig. pared to WT. In addition. ondansetron at concentrations 1). However, there are also major differences, for 644 F. G. Boess et al. example, the triplet YNS of the nAChR IX7sequence aligns with NEF of the 5-HT3 receptor sequence. The tyrosine residue is conserved in all nAChR ci subunit sequencesknown to date, with the exception of &5which is functionally atypical. Irreversible labelling and mutagenesis experiments have confirmed that this residue is important for ligand binding to the nAChR, where it may interact with the diffuse positive charge of the quatemary ammonium group of acetylcholine (reviewed in Changeux et al., 1992; Karlin and Akabas, 1995). The glutamate residue at the adjacent position of the 5-HT3 receptor sequence (E106) may subserve a similarly important role in recognition and form an ionic interaction with the ammonium group of the natural agonist 5HT or the positively charged nitrogen, present in all 5HT3 receptor ligands. To explore this possibility, we mutated glutamate 106 to the similarly charged amino acid aspartate, the uncharged but polar analogues glutamine or asparagine and the small neutral amino acid alanine. We examined these mutants with both radioligand binding, to characterize the recognition properties of the desensitized receptor state and wholecell patch clamp electrophysiology to study the active state of the receptor. Our data show that shortening the side chain of glutamate by one methylene in E106D, reduces the high affinity recognition of both 5-HT and 2-methyl-5-HTby over two orders of magnitude, though the EC50for 5-HT is unchanged. However, we find essentiallyno change in the high affinitybinding of the agonistmCPBG or its ECW determined electrophysiologically. Removal of the charge on the aspartate in the mutant E106N produces broadly the same change but the magnitudeis less; again mCPBG is unaffected by this mutation. Generally, antagonist recognition is similarly affected for both E106D and E106N mutants. The alanine and glutarnine mutants express very poorly and while we have reported some data on their electrophysiologicalcharacterization, radioligand binding studies proved impracticable. These observations suggest that changing E106 interferes with either the formation, or the stability of functional 5-HT3 receptor complexes. E106 may be important for the proper folding of subunits or the assembly of five subunits to a homopentamer. Folding or expression clearly place significant constraints at this position and we are only able to address functionality within those limits. In the wildtype 5-HT3 receptor, we initially proposed that the negatively charged carboxylate group of glutamate 106 may form a direct ionic interaction with the positivelycharged primary ammoniumgroup of 5-HT and the positively charged secondary, tertiary or quatemary ammonium groups present in the majority of high affinity 5-HT3 receptor antagonists (Gozlan and Langlois, 1992).However, our observationsindicate that in the asparagine mutant E106N, where the charge has been removed, the binding of 5-HT and 2-methyl-5-HTis less affected than for E106D. It appears unlikely, therefore, that the interaction of glutamate (E106) with 5-HT and 2-methyl-5-HT(Fig. 8) is ionic, as the similarly charged aspartate produces greater changes in binding affinity than the polar residue asparagine. These compounds may instead form a hydrogen bond with glutamate 106, and both aspartate and asparagine may be able to partially substitute for glutamate in this interaction. Replacement of E106 by aspartate or asparagine did not alter the affinity of the 5-HT3 receptor agonist mCPBG, suggesting that interaction with E106 is not essential for high affinity mCPBG binding. In the protonated form of this ligand, the positive charge is presumablydelocalizedover the whole molecule (Fig. 8). Interactions with the aromatic side chains of tyrosine (Tyr) and tryptophan(Trp) residuesof the 5-HT3receptor may be more importantfor the binding of mCPBG. It has been suggested that the ‘mild electronegative’ character of the n electron systems of Tyr and Trp residues serves to complex the diffuse positive charge of the quatemary ammoniumgroup of acetylcholine(Galzi and Changeux, 1995). This idea is supported by the observation that a synthetic aromatic host strocture can bind acetylcholine (Dougherty and Stauffer, 1990). Several of these aromatic residues are conserved in the 5-HT3 receptor (W67, W98, W160, Y211) and one of them (W67) has already been shown to contribute to ligand binding (Schulte et al., 1995). In the nAChR, a tyrosine residue (Y93) in the correspondingregion (loop 1) is labelled five times more intensely by the photoactivatable,irreversible antagonist DDF, if labelling is carried out in the presence of meproadifen, a non-competitive antagonist that is thought to shift the equilibrium to the desensitized state of the receptor (Galzi et al., 1991b). This led to the suggestion that confirmational changes in the binding site upon desensitization render this residue more accessible. The irreversible cholinergic agonist acetylcholine mustard aziridinium, that is thought to interact with the high affinity desensitized state labels only Y93 (Cohen et al., 1991).In analogy with the nAChR, the 5HT3 receptor can probably assume several confirmational states, including a resting closed, an open and several desensitized states (Changeux et al., 1992; Jackson and Yakel, 1995). Agonist binding promotes the transitionfrom the resting to the open state and in the continuingpresence of agonists the receptor desensitizes and assumes one of several possible closed, desensitized conformations, that display a higher affinity for the agonist. In equilibrium binding experiments with agonists, the affinity of the desensitized high affinity state is measured. Several groups have observed, that the concentrationof agonist for the induction of desensitization in the 5-HT3receptor is lower than that for channel activation (Bartrup and Newberry, 1996; van Hooft and Vijverberg, 1996).In NG108-15 cells, the IC50values for desensitizationby mCPBG (20 nM), 5-HT (50 nM) and 2-methyl-5-HT (600 nM) (Bartrup and Newberry, 1996) Mutagenesis of the 5-HT3receptor JNH2 I#y /-’NH2 2-m3thyl-5-H’I 5-HT 645 6H3 GR65630 mCPBG CH3 Ondansetron 7H3 CH3 CH3 Renzapnde Fig. 8. Chemical structures of 5-HT3 receptor agonists (5-HT, 2-methyl-5-HT and mCPBG) and antagonists (GR65630,ondansetron,granisetron and renzapride). are very similar to the Ki values (14, 170 and 810 nM, respectively) determined in radioligand binding assays with intact cells under physiologicalconditions(Boess et al., 1992b).While in our studies, there was more than a 100-fold decrease in the affinity of 5-HT for the mutant E106D measured by radioligand binding, the EC50of 5HT observed in electrophysiological experiments with this receptor did not change. For E106N, a 30-fold decrease in binding affinity,butjust a seven-foldincrease in EC50was found. Our data suggestthat the interactionof 5-HT with the receptor, which is responsible for the transition from the closed resting to the open state, is retained when glutamate 106 is substitutedby aspartate. However, an asparagine, glutamine, or alanineresidue in this position reduces the potency of 5-HT. We have not observed any clear changes in the desensitizationkinetics of the responses elicited by 5-HT in cells expressing E106D or E106N mutant receptors.If the reduced affinity in radioligandbinding studies reflects a lower affinityfor the desensitized receptor, a detailed analysis of the IC5CI values of agonists for the induction of desensitization would perhaps reveal differences between these mutants and the wildtype receptor. Concentration-response studies revealed a reduction of the Hill coefficientsfrom a value of two in the wildtype to unity in both E106D and E106N mutant receptors. High Hill coefficientsare generally interpreted as a sign of cooperativity between different binding sites on the same receptor complex. In the nAChR, binding sites are present on the subunit interfaces, i.e. different parts of two subunits contribute to a single binding site. Because the nAChR complex is a heteropentamer, binding sites are not equivalent (Pedersen and Cohen, 1990). Occupation of two binding sites promotes an allosteric transition from a closed resting to an open state of the receptor (Jackson, 1994). In homopentameric receptors such as recombinant IX7nAChR expressed in Xenopus oocytes, the existence of five binding sites for the competitive 646 F. G. Boess et al. antagonistmethyllycaconitinehas been suggested(Palma et al., 1996). The recombinant 5-HT3 receptor is a homopentamerand the five potentialbinding sites on the subunit interfaces should initially be equivalent. The cooperativity observed at wildtype recombinant 5-HT3 receptors expressed in HEK 293 cells suggests that binding of at least two agonist moleculesis necessary for the efficient activation of the receptor, as in the nAChR. This coupling of two binding sites on different subunit interfaces may be eliminated after replacement of E106 by either aspartate or asparagine. In the resting state of the 5-HT3receptor pentamer, E106 may participatein an inter-subunit bond, that is changed during the allosteric transition after agonist binding. All the highly selective 5-HT3 receptor antagonists examined in the present study blocked 5-HT current activation, indicating that the mutant receptors retain 5HT3 receptor-like recognitionproperties.The changes in affinity of these antagonists were generally similar in both radioligand binding and electrophysiological studies. However, there were subtle differences, even for example between the structural analogues GR65630 and ondansetron.The affinity of ondansetronand its potency to inhibit agonist activation, was similarly reduced for both E106D and E106N, while the affinity of GR65630 was unaffected in the mutant E106N compared to WT, but was reduced 15-fold in E106D. In these ligands, the positive charge which is proposed to interact with E106, is delocalized within an imidazoleheterocycle (Fig. 8). It is possible that the restricted confirmational flexibility imposedby the additionalring (C; Fig. 8) of ondansetron, imposeslimitson its effectiveinteractionwith asparagine which is not experienced by GR65630. The 5-HT3 receptor antagonist granisetron showed a markedly reduced affinity for both the aspartate and asparagine mutants (800–500-fold) in contrast to the other antagonists which we have investigated. In this Iigand, the basic nitrogen is contained in a highly substitutedheterocycle, but unlike GR65630 and ondansetron the charge is not delocalized.It is possiblethat it is interacting with E106 in a similar manner to 5-HT and 2methyl-5-HT.Many studieshave exploredthe systematic modificationsof this heterocyclebut no clear conclusions have been possible as to the nature of the interaction of the basic nitrogen with the receptor (Gozlan and Langlois, 1992). It is interesting, however, that the affinity of renzapride, in which the basic nitrogen is contained within a different, but similarly bulky heterocycle, was only reduced 10-20-fold. In addition, renzapride showed differential effects for E106D with no change in its ability to antagonizereceptor activation but a reduction in binding affinity, whereas for E106N both antagonist potency and affinity were reduced. Our results suggest that glutamate 106 is important in ligand recognitionfor 5-HT3receptor antagonistsand for the agonists in the high affinity desensitized state of the receptor. Its involvement with agonist activation is markedly less pronouncedbut its mutation results in loss of cooperativity. Our data are consistent with an interaction between this residue and the basic nitrogen present in all 5-HT3 receptor ligands and further exploration of the vicinal residues seems warranted. Acknowledgements—The authors would like to thank Drs Tom Blackburn (SmithKline Beecham) and Gavin Kilpatrick (Glaxo) for their kind gifts of drugs. Supported by Glaxo Canada and MRC Canada. L.J.S. holds a MRC/PMAC Pfizer (Canada) Fellowship. REFERENCES Aylwin M. L. and White M. M. (1994) Ligand-receptor interactions in the nicotinic acetylcholine receptor probed using multiple substitutions at conserved tyrosines on the a subunit. FEBS Letters 349: 99–103. Bartrup J. T. and Newberry N. R. (1996) Electrophysiological consequences of ligand binding to the desensitized 5-HT3 receptor in mammalian NG108-15 cells. Journal of Physiology 490.3: 679–690. Belelli D., Balcarek J. M., Hope A. G., Peters J. A., Lambert J. J. and Blackbum T. P. (1995) Cloning and functional expression of a human 5-hydroxytryptamine type 3A, receptor subunit. Molecular Pharmacology48: 1054–1062. Boess F. G., Lurnmis S. C. R. and Martin I. L. (1992a) Molecular properties of 5-hydroxytryptarnine3receptor-type binding sites purified from NG108-15 cells. Journal of Neurochemistry 59: 1692–1701. Boess F. G., SeptilvedaM.-I., Lummis S. C. R. and Martin I. L. (1992b) 5-HT3 receptors in NG108-15 neuroblastoma x glioma cells: effect of the novel agonist l-(m-chlorophenyl)biguanide. Neuropharmacology31: 561–564. Boess F. G. and Martin I. L. (1994)Molecular biology of 5-HT receptors. IVeuropharmacology33: 275–317. Boess F. G., Beroukhim R. and Martin I. L. (1995) Ultrastmcture of the 5-HT3receptor. Journal of Neurochemistry 64: 1401-1405. Carbone E. and Lux H. D. (1987) Kinetics and selectivity of a low-voltage-activated calcium current in chick and rat sensory neurones. Journal of Physiology 386: 547–570. Changeux J.-P., Galzi J.-L., Devilliers-Thi&y A. and Bertrand D. (1992) The functional architecture of the acetylcholine nicotinic receptor explored by affinity labelling and sitedirected mutagenesis. Quarterly Reviews of Biophysics 25: 395432. Chen C. A. and Okayama H. (1988) Calcium phosphatemediated gene transfer: a highly efficienttransection system for stably transforming cells with plasmid DNA. BioTechniques 6 (7): 632-637. Cohen J. B., Sharp S. D. and Liu W. S. (1991) Structure of the agonist-binding site of the nicotinic acetylcholine receptor. Journal of Biological Chemistry 266: 23354-23364. Craig D.A. (1993)The Cheng–Pmsoffrelationship: something lost in the translation. Trends in Pharmacological Sciences 14: 89-91. Dougherty D. A. and Stauffer D. A. (1990) Acetylcholine binding by a synthetic receptor: implications for biological recognition. Science 250: 1558–1560. Dunn S. M. J., BatesonA. N. and Martin I. L. (1994)Molecular neurobiologyof the GABA~ receptor. International Review of Neurobiology 36: 51–96. Eisel& J.-L., Bertrand S., Galzi J.-L., Devillers-Thi&y A., Mutagenesis of thIe 5-HT3receptor Changeux J.-P. and Bertrand D. (1993) Chimaenc nicotinic serotonergic receptor combines distinct ligand binding and channel specificities. Nature 366: 479483. Galzi J.-L., Revah R., Black D., Goeldner M., Hirth C. and ChangeuxJ.-P. (1990)Identificationof a novel amino-acidutyrosine 93 within the cholinergic ligand binding sites of the acetylcholine receptor by photoaffinitylabeling. Additional evidence for a three-loop model of the cholinergic ligandbinding sites. Journal of Biological Chemistty 265: 1043010437. Galzi J.-L., Bertrand D., Devillers-Thi&y A., Revah R., Bertrand S. and Changeux J.-P. (1991a) Functional significanceof aromatic amino acids from threepeptide loops of the ct7 neuronal nicotinic receptor investigated by sitedirected mutagenesis. FEBS Letters 349: 99–103. Galzi J.-L., Revah R., Bouet F., M6nez A., Goeldner M., Hirth C. and Changeux J.-P. (1991b) Allosteric transitions of the acetylcholine receptor probed at the amino acid level with a photolabile cholinergic ligand. Proceedings of the National Academy of Science U.S.A. 88: 5051–5055. Galzi J.-L. and Changeux J.-P. (1995) Neuronal nicotinic receptors: Molecula organization and regulations. Neuropharrnacology34: 563-582. Gozlan, H. and Langlois, M. (1992) Structural analysis of receptor–ligand interactions for the mapping of 5-HT3 antagonist binding site. In Central and Peripheral 5-HT3 Receptors, ed. Hamon, pp. 59-87. Academic Press Ltd. Green T., Stauffer K. A. and Lummis S. C. R. (1995) Expression of recombinant homo-oligomeric 5-hydroxytryptamine3 receptors provides new insights into their maturation and structure. Journal of Biological Chemistry, 270 (11): 6056-6061. Hope A. G., Downie D. L., SutherlandL., Lambert J. J., Peters J. A. and Burchell B. (1993) Cloning and functional expression of an apparent splice variant of the murine 5HT3receptor A subunit.EuropeanJournal of Pharmacology (Molecular Pharmacology Section) 245: 187-192. Isenberg K. E., Ukhun I. A., Holstad S. G., Jafri S., Uchiro U., ZorumskiC. F. and Yang J. (1993)Partial cDNA cloning and NGF regulation of a rat 5-HT3receptor subunit.NeuroReport 5: 121-124. Jackson M. B. (1994) Single channel currents in the nicotinic acetylcholine receptor: a direct demonstration of allosteric transitions. Trends in Biochemical Sciences 19 (10): 396399. Jackson M. B. and Yakel J. L. (1995) The 5-HT3 receptor channel. Annual Reviews in Physiology 57: 447-468. Karlin A. and Akabas M. H. (1995) Toward a structural basis for the function of nicotinic acetylcholinereceptors and their cousins. Neuron 15: 1232–1244. Kuhse J., Betz H. and Kirsch J. (1995) The inhibitory glycine receptor: architecture, synaptic localization and molecular pathology of a postsynaptic ion-channel complex. Current Opinion in Neurobiology 5: 318-323. Lummis S. C. R. and Martin I. L. (1992) Solubilization, purification, and functional reconstitution of 5-hydroxytryptamine3 receptors from NIE-115 neuroblastoma cells. Molecular Pharmacology41: 18-23. Maricq A. V., Peterson A. S., Brake A. J., Myers R. M. and Julius D. (1991) Primary structure and functional expression of the 5-HT3receptor, a serotonin-gatedion channel. Science 254: 432437. McKeman R. M., Biggs C. S., Gillard N. P., Quirk K. and 647 Ragan C. I. (1990a) Molecular size of the 5-HT3 receptor solubilizedfrom NCB-20 cells. Biochemistry 269: 623-628. McKeman R. M., Gillard N. P., Quirk K., Kneen C. O., Stephenson G. I., Swain C. J. and Ragan C. I. (1990b) Purificationof the 5-hydroxytryptamine5-HT3receptor from NCB20 cells. Journal of Biological Chemistry 265: 13572– 13577. Miquel M.-C., Ement M. B., Gingrich J. A., Nosjean A., Hamon M. and El Mestikawy S. (1995) Developmental changesin the differentialexpressionof two serotonin5-HT3 receptor splice variantsin rat. Journal of Neurochemistry65: 475483. MiyakeA., MochizukiS., TakemotoY. and Akuzawa S. (1995) Molecularcloning of human 5-hydroxytryptamine3receptor: heterogeneity in distribution and function among species. American Society of Pharmacology and Experimental Therapeutics48: 407-416. Nowak M. W., Keamey P. C., Sampson J. R., Saks M. E., Labarca C. G., Silverman S. K., Zhong W., Thorson J., Abelson J. N., Davidson N., Schultz P. G., DoughertyD. A. and Lester H. A. (1995) Nicotinic receptor binding site probed with unnatural amino acid incorporation in intact cells. Science 268: 439442. Palma E., Bertrand S., Binzoni T. and Bertrand D. (1996) Neuronalnicotinic ct7receptor expressedin Xenopus oocytes presents five putative binding sites for methyllycaconitine. Journal of Physiology 491.1: 151–161. Pedersen S. E. and Cohen J. B. (1990)d-Tubocurarinebinding sites are located at a-y and cd subunit interfaces of the nicotinic acetylcholinereceptor. Proceedingsof the National Academy of Science U.S.A. 87: 2785–2789. SchulteM., Bloom K. E. and White M. M. (1995)Evidence for the involvement of tryptophanin the binding of curare to 5HT3receptors. Society of Neuroscience Abstracts 21: 30.6. SeptilvedaM. I., Lummis S. C. R. and Martin I. L. (1991) The agonist properties of meta-chlorophenylbiguanide and 2methyl-5-hydroxytryptamineon 5-HT3receptors in NIE-I 15 neuroblastoma cells. British Journal of Pharmacology 104: 536-540. Sine S. M., Quiram P., PapanikolaouF., KreienkampH.-J. and Taylor P. (1994)Conserved tyrosines in the a subunit of the nicotinic acetylcholine receptor stabilize quaternary ammonium groups of agonists and curariform antagonists.Journal of Biological Chemist~ 269: 8808–8816. Steward L. J., Boess F. G., Steele J. A. and Martin I. L. (1996) Agonist/antagonist recognition properties of the 5-HT3 receptor: importance of glutamate 106. British Journal of Pharmacology.Proceedings supplement. 117: 96P. Uetz P., Abdelatty F., Villarroel A., Rappold G., Weiss B. and Koenen M. (1994) Organisation of the murine 5-HT3 receptor gene and assignment to human chromosome 11. FEBS Letters 339 (3): 302–306. Unwin N. (1993) Nicotinic acetylcholine receptor at 9 ~ resolution. Journal of Molecular Biology 229: 1101–1124. van Hooft J. A. and Vijverberg H. P. M. (1996) Selection of distinct confirmational states of the 5-HT3receptor by full and partial agonists. British Journal of Pharmacology 117: 839-846. Werner P., Kawashima E., Reid J., Hussy N., Lundstrom K., Buell G., HumbertY. and Jones K. A. (1994)Organisationof the mouse 5-HT3 receptor gene and the functional expression of two splice variants. Molecular Brain Research 26: 233-241.