Structural and electrostatic properties of the 5-HT3 receptor pore revealed by substituted cysteine accessibility mutagenesis.

5-HT(3) receptors are members of the Cys loop family of ligand-gated ion channels. We used the substituted cysteine accessibility method to identify amino acid residues in the channel forming domain, M2 that face the water-accessible surface and to locate their position in the ion conduction pathway. Cysteine was substituted for each residue, one at a time, in the M2 segment (Asp(274)-Asp(298)). 5-Hydroxytryptamine EC(50) values for functional mutants did not vary from wild type (1.4 +/- 0.2 microm) by more than 10-fold, and five mutants were nonfunctional. Covalent modification of the mutant receptors with sulfydryl reagents revealed 11 residues to be water-accessible, with a pattern consistent with an alpha-helix except at Leu(285) and Leu(293). The data suggest that charge selectivity begins at a more cytoplasmic level than Val(291). Modification at some positions (Val(291), Leu(293), Ile(294), Leu(287), and Ser(280)) resulted in channels that were locked open. Reaction rates with accessible cysteines were voltage-dependent at some residues, suggesting that access occurs via the ion channel. Overall the data observed are similar but not identical to that reported for other members of the family and confirms the high degree of structural and functional homology between receptors in the Cys loop receptor family.

The 5-HT 3 1 receptor is a member of the Cys loop family of ligand-gated ion channels, which includes nicotinic acetylcholine (nACh), GABA A and glycine receptors. These receptors are pentamers, usually formed by the co-assembly of one to four different subunits each with a large extracellular N-terminal region and four putative transmembrane domains (M1-M4). Two 5-HT 3 receptor subunits, 5-HT 3A (1) and 5-HT 3B (2), have been identified so far, and receptors can function as either homo-oligomeric (A only) or hetero-oligomeric receptors (2). Evidence suggests that the Cys loop family of receptors is modular in design, with the extracellular N-terminal domain containing the ligand binding site and the transmembrane regions containing the pore (3). There is good evidence from a variety of studies that the second transmembrane segment, M2, lines the channel (4). Studies on acetylcholine receptors, for example, have identified rings of residues that alter conductance (5) or the selectivity among monovalent (5,6) or divalent (7) cations or channel gating (8). The high resolution structure of a protein homologous to the extracellular domain of the acetylcholine receptor was recently determined (9); however, so far details of the complete structure of any of this family of receptors are lacking.
The substituted cysteine accessibility method (SCAM) has been used to identify systematically the residues that line an ion channel. Here residues in a membrane-spanning segment are individually mutated to cysteine and each mutant receptor expressed in Xenopus oocytes. If the mutant receptors have similar properties to wild type, it can be assumed that their structure is similar to that of wild type. The accessibility of each residue can then be determined by examining the ability of small sulfydryl-specific reagents to react with the cysteine. The information gained is able to provide information on the secondary structure of channel-lining segments and the location of ion channel gates and selectivity filters and to map binding sites within the channel (10). SCAM has been used to identify pore-lining residues in a variety of ion channels, including the nACh and GABA A receptors. The M2 regions in these receptors gave a similar but not identical pattern of labeling and supported previous studies suggesting that this region is largely ␣-helical. There are, however, some discrepancies, particularly in the region surrounding the conserved central leucine residue.
The amino acid sequence of the 5-HT 3A receptor subunit displays strong sequence similarity with nACh receptor subunits, especially in the M2 region (e.g. the ␣ 1 nACh receptor subunit as illustrated in Fig. 1). We therefore wanted to confirm that the water-accessible residues in this receptor are similar to those in the nACh and GABA A receptors and also explore whether the use of homomeric receptors could provide additional information about the structure and function of M2. The data revealed a similar but not identical pattern of wateraccessible residues in the pore to those previously observed for other members of the Cys loop ligand-gated ion channel family and also provided new information about the structure and role of certain residues located in and around the pore.

EXPERIMENTAL PROCEDURES
Drugs and Reagents-The sulfydryl reagents [2-(trimethylammonium)ethyl] methanethiosulfonate bromide (MTSET) and sodium (2-sulfonatoethyl) methanethiosulfonate (MTSES) were obtained from Biotium, Inc. (Haywood, CA). To minimize hydrolysis of these reagents, stock solutions in water were made daily, stored on ice, and diluted in buffer to the appropriate concentration just before use. All other reagents were of the highest quality. * This work was supported by funds from the Wellcome Trust (to S. C. R. L. and D. R.) and the National Institutes of Health (to M. H. A. and E. N. G.). 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  Mutagenesis and Preparation of cRNA and Oocytes-Mutant 5-HT 3A receptor subunits were developed using the eukaryotic expression vector pcDNA 3.1 (InVitrogen, Abingdon, UK) containing the complete coding sequence for the 5-HT 3A(b) subunit from NIE-115 cells as described previously (11). Mutagenesis reactions were performed using the Kunkel method (12) and confirmed by DNA sequencing. Wild type (WT) and mutant receptor subunit coding sequences were then subcloned into pGEMHE plasmid (13). This was linearized with NheI (New England Biolabs) and cRNA synthesized as described previously (14). Oocytes from Xenopus laevis, prepared and maintained as described previously (14), were injected with 50 nl of cRNA (10 -200 ng/l), and the experiments were performed 1-6 days following injection.
Characterization of Mutant Receptors-5-HT-induced currents were recorded from individual oocytes using two-voltage electrode clamp as described previously (14). All experiments were performed at room temperature (22-25°C). Agarose cushion electrodes (15) were filled with 3 M KCl and had a resistance of less than 2 M⍀. The ground electrode was connected to the bath by a 3 M KCl/agar bridge. During experiments the oocytes were continuously perfused (6 ml/min) in calcium-free frog Ringer solution (CFFR) of the following composition: 115 mM NaCl, 2.5 mM KCl, 1.8 mM MgCl 2 , and 10 mM HEPES, pH 7.5. The holding potential was maintained at Ϫ60 mV unless otherwise specified. To determine EC 50 and maximal 5-HT concentrations, 5-HT doseresponse curves were fitted to the following Hill equation.
where I max is the maximal peak current, [A] is the concentration of agonist, and n is the Hill cooefficient.
Reaction with Sulfydryl Reagents-The positively charged -SCH 2 CH 2 N(CH 3 ) 3 ϩ moiety is added to cysteine following reaction with MTSET. Following reaction with MTSES the negatively charged -SCH 2 CH 2 SO 3 Ϫ group is added to cysteine. The reactions of wild type and mutant 5-HT 3 receptors with 1 mM MTSET or 10 mM MTSES were tested using alternating pulses of maximal (5-HT MAX ) and EC 50 (5-HT 50 ) concentrations of 5-HT before and after addition of the sulfydryl reagent. Changes in the peak current induced by the EC 50 5-HT test pulses are more sensitive to effects of modification on gating kinetics, whereas changes in the peak current induced by the maximal 5-HT test pulses are more sensitive to the effects of modification on conductance (16).
Thus a typical protocol would be: 5-HT MAX   Data for each mutant were compared with wild type by one-way analysis of variance applying the Dunnett post-hoc test (p Ͻ 0.05) using Prism v3.0 (Graphpad Software Inc, CA). We infer that the sulfydryl reagents have reacted with a cysteine if the subsequent 5-HT-induced currents are significantly different than the initial 5-HT-induced currents. The MTS reagents react 5 ϫ 10 9 times faster with ionized thiolates than with thiols (17). We assumed that only cysteines on the water-accessible surface of a protein ionize to any significant extent; thus we inferred that reactive cysteines were on the water-accessible surface of the protein. In addition, MTSET ϩ and MTSES Ϫ are membrane-impermeant (18,19), and thus when we applied them extracellularly they only had access to residues that were on the extracellular water-accessible surface of the protein.
Determination of Reaction Rate Constants-Rate constants for reaction of MTSET and MTSES with substituted cysteines were determined as described previously (20). Briefly, second order rate constants for MTSET were obtained using a voltage-step protocol as shown in Fig.  2C. Current because of leak and the effect of desensitization were subtracted from the currents recorded during application of 5-HT plus MTSET. The decrease in current was approximately linear, and the second order rate constant (k) was estimated from the following equation.
where I is current, t is time, x is the concentration of sulfydryl reagent, and the subscripts refer to the beginning and end of each voltage-step interval.
Reaction with MTSES caused a large and irreversible decrease in inactivation rate; therefore we used a different protocol to determine the rate of reaction with this reagent. The following sequence of reagents was applied repeatedly (see  (46) and are oriented with the putative pore-lining face toward the respective residue numbers in the plane of the page. The rings of residues after Changeux and co-workers (4) are also shown. Shorthand notation for each equivalent position is given in the center between the residue numbers (32).
at each holding potential. Because the reagent was in excess, we determined the second order rate constant by fitting the peaks from the test applications of 5-HT to the following equation.
where the subscripts refer to zero time and infinite time (complete reaction), respectively, and the other parameters are as above.
Calculation of Electrical Distance-The calculations were performed as described previously (20). Briefly, to determine the electrical distance (␦) from the bath solution to the site of reaction, the second order rate constant was plotted as a function of the holding potential ( M ) and fit to the following equation.
where F is the Faraday constant, R is the molar gas constant, T is the absolute temperature, and z is the algebraic charge on the reagent, with all other parameters as above. MTSET is poorly permeant through the 5-HT 3 channel, but when it does reach the residues near the cytoplasmic end of the channel, it was assumed that the rate of reagent leaving to the cytoplasm was much greater than the rate of leaving to the extracellular bath. Thus in the case of E277C the apparent ␦ ϭ (␦ Ϫ 0.5) (20). 3 Receptors-Transfected HEK293 cells maintained as described previously (21) were washed with three changes of Tris-buffered saline (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). To label the 5-HT 3 receptor N-terminal domain pAb120 antiserum (21) was used at 1:1000 dilution in Tris-buffered saline. 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. Coverslips were mounted in Vectashield mounting medium (Vector), and immunofluorescence was observed using a Medical Research Council Radiance confocal microscope.

RESULTS
Effect of Mutations to Cysteine-Cysteine was substituted at 25 consecutive positions (Asp 274 -Asp 298 ), one at a time, and the function of the mutants was tested following expression in oocytes. Twenty of the mutants expressed 5-HT-activated cur-rents in Xenopus oocytes following injection of mRNA, indicating that in general the mutations were well tolerated. Application of 5-HT to oocytes expressing WT receptors resulted in rapid inward currents averaging 806 Ϯ 71 nA (n ϭ 6) at a holding potential of Ϫ60 mV. Typical responses are shown in Fig. 2B. The average I max for the mutants ranged from 131 nA Ϯ 11 nA (n ϭ 6, I295C) to 2976 Ϯ 327 nA (n ϭ 6, F281C).
Dose-response curves were determined for all mutants, and the EC 50 values are shown in Fig. 3. None of the mutants had an EC 50 value more than 10-fold different from that of wild type receptors (1.4 Ϯ 0.2 M, n ϭ 6).
Nonfunctional Mutants-The expression of G288C, Y289C, S290C, F292C, and D298C mutant receptors were examined using several batches of cRNA, but no currents were detected during application of 5-HT up to a concentration of 1 mM (Ϸ1000 ϫ WT EC 50 ). To rule out the possibility that the lack of functional channels for these mutants was due to spontaneously formed disulfide bonds, as was observed in the GABA A receptor at 17Ј and 20Ј positions (22), they were perfused with 10 mM dithiothreitol for 3 min; no 5-HT induced currents were observed following this treatment. We used immunofluorescent labeling of transiently transfected HEK293 cells to determine whether these mutants, which were also shown to be nonfunctional in HEK293 cells using calcium imaging techniques (11), formed cell surface receptors. Using the 5-HT 3 selective antiserum, pAb120, raised against the extracellular domain (21), WT, and all of the mutants had a band of fluorescence on the cell surface when examined in nonpermeabilized cells. Untransfected cells were nonfluorescent as described previously (21). Typical results for G288C are shown in Fig. 2A (inset). Thus the data suggest that the receptors were correctly synthesized, assembled and targeted to the cell surface, but the mutation resulted in loss of function.
Reaction with MTSET-The mutants were examined for their susceptibility to react with MTSET applied in the presence of 5-HT by comparing the mean of a pair of responses to

FIG. 2. Expression of WT and mutant 5-HT 3 receptors in Xenopus oocytes.
A, sample concentration-response curves for WT homomeric 5-HT 3A receptors. Inset, immunolabeling of 5-HT 3A receptors reveals they are located at the plasma membrane. This example is the G288C mutant and is representative of WT and all nonfunctional mutants; untransfected cells were unlabeled. The scale bar represents 10 m. B, sample protocol for determination of the effect of sulfydryl reagents on cysteine substitution mutants. The data shown are the typical responses from T284C. The four test pulses of 5-HT before and after sulfydryl reagent are at 5-HT MAX , 5-HT 50 , 5-HT MAX , and 5-HT 50 (see text for details). Mean responses from these test pulses were compared before and after application of reagent. C, voltage-step protocol for the determination of rate constants of reaction with MTSET. Panel i, the sequence of potentials used during each application of reagent. Panel ii, a typical set of recordings from T284C used to calculate the rate constants. 5-HT and MTSET, where applied, were present throughout the recording. The voltage steps were carried out at the same point just after peak current was reached. 5-HT before MTSET application to the mean of a pair of responses after MTSET application (see "Experimental Procedures" and Fig. 2). The effect of MTSET was examined by the effect on currents induced by 5-HT test pulses at EC 50 and at maximal concentrations (Figs. 2 and 4). For wild type 5-HT 3A receptors, a 1-min application of 1 mM MTSET with 10 M 5-HT had no effect on the subsequent 5-HT-induced currents. Thus we infer either that MTSET does not react with the endogenous cysteines or that reaction with them has no functional effect. For the cysteine mutants, following a 1-min application of 1 mM MTSET with 10 M 5-HT, the subsequent 5-HT EC 50 currents were irreversibly altered with D274C, G276C, E277C, S280C, T284C, L285C, L287C, V291C, L293C, I294C, and I295C. With the majority of these residues MTSET interaction resulted in inhibition of the subsequent response. With two mutants, D274C and I295C, MTSET modification caused enhancement of the subsequent 5-HT currents. At two other positions, L287C and S280C, reaction with MTSET in the presence of 5-HT appeared to lock the channels in the open state. Thus following washout of the MTSET and 5-HT, the currents remained elevated and did not return to their initial base-line level even after 30 min of observation in the absence of MTSET and 5-HT.
Reaction with MTSES-The mutants were examined for their susceptibility to react with the negatively charged MT-SES in a similar way to MTSET, both at the EC 50 and at a maximal concentration of 5-HT. For wild type 5-HT 3A receptors a 1-min application of 10 mM MTSES with 10 M 5-HT had no effect on the subsequent 5-HT-induced currents (Fig. 5). In contrast to the inhibition of 5-HT-induced current after application of MTSET with 5-HT at V291C, L293C, and I294C, near the extracellular end of M2, application of MTSES with 5-HT locked the channels in the open state (Fig. 5). Thus following washout of the MTSES and 5-HT, the current remained ele-vated and did not return to its initial base-line level even after 30 min of observation. Application of 300 M diltiazem to the MTSES locked-open channels formed by V291C reversibly blocked the current by 37% Ϯ 9% (n ϭ 4). Diltiazem (300 M), however, had no effect on locked-open channels formed by L293C or I294C (data not shown).
Rate Constants for Reaction with Sulfydryl Reagents-We determined the reaction rates with MTSET or MTSES for each of the accessible mutants. For mutants where reaction with MTSET caused inhibition of subsequent 5-HT-induced currents, the rates were measured using the procedure illustrated in Fig. 2C, and the results were analyzed using Equation 3. For some mutants the voltage dependence of the reaction rates were also examined, and these data are shown in Fig. 6 (A and  B).
To determine the reaction rates of MTSES with V291C, L293C, and I294C, we used the protocol illustrated in Fig. 7A of brief sequential applications of MTSES and 5-HT followed by test pulses of 5-HT. We analyzed the results by fitting the peak currents induced by the 5-HT test pulses as a function of cumulative exposure time to MTSES with an exponential decay function as in Equation 4 to calculate the pseudo-first order rate constant (Fig. 7B). The second order rate constant was determined by dividing the pseudo-first order rate constant by the MTSES concentration (Fig. 7C).
These rate constants are slower than the reaction rates with cysteine in free solution (23) but are generally faster than the rates of reaction with cysteines found in protein clefts (e.g. in the binding site of the dopamine D2 receptor) (24). The reaction rates of MTSET with L293C and I294C, however, were similar to the rates measured with cysteines in protein clefts (24). We also examined these rate constants at a variety of different membrane potentials, and the data show that the reactions of MTSET with E277C, S280C, T284C, V291C, L293C, and I294C were voltage-dependent (Fig. 6, A and B) and that the reactions of V291C, L293C, and I294C with MTSES were voltagedependent (Fig. 7C).
Electrical Distance-Because the sulfydryl reagents are charged and their reactions with cysteine are assumed to be fast, the rate-limiting step of their reaction with the thiolate is their movement into the ion channel; this should be voltagedependent. The magnitude of the voltage dependence depends on the fraction of the electrical field through which the reagent moves to reach the cysteine (20). Based on the voltage dependence of the reaction rates, we calculated the electrical distance (␦) from the bulk solution to the target cysteine residues using Equation 5, for those reactions whose rates were sufficiently high to use for this analysis. The results are shown in Fig. 6C. The data reveal that the electrical distance is greatest toward the cytoplasmic end and decreases toward the extracellular end. DISCUSSION We identified the water-accessible residues in and flanking the M2 segment of the 5-HT 3A receptor subunit using the substituted cysteine accessibility method. Twenty of the twenty five cysteine substitution mutants were functional, and the EC 50 values of these mutant receptors did not vary from wild type by more than 10-fold. We infer that their structures were similar to the structure of wild type. Thus insertion of five cysteine residues (one for each subunit) was tolerated in these mutants. For 11 of the cysteine substitution mutants, application of MTSET in the presence of 5-HT irreversibly altered the subsequent 5-HT-induced currents, and we infer that the engineered cysteine, in at least one of the five subunits, covalently reacted with MTSET. We cannot determine whether reaction occurred with more than one of the five engineered cysteines

5-HT 3 Receptor Channel-lining Residues
that is present in each functional receptor. Because the rate of reaction of these sulfydryl reagents with a thiolate anion (RSϪ) is 5 ϫ 10 9 faster than the rate with the uncharged thiol (RSH) (17) and the extent of reaction of a sulfydryl reagent with an engineered cysteine directly correlates with the surface accessibility of the corresponding wild type (25), we assume that the 5-HT 3A cysteine mutants that react with MTSET and MTSES are on the water-accessible surface of the receptor. Thus in the presence of 5-HT, the residues Asp 274 , Gly 276 , Glu 277 , Ser 280 , Thr 284 , Leu 285 , Leu 287 , Val 291 , Leu 293 , Ile 294 , and Ile 295 in the M2 transmembrane region are on the water-accessible surface of the protein at least part of the time. The voltage dependence of the MTSET reaction rates with many of these residues implies that the access pathway to them is via the ion channel. This supports our inference that at least some of these residues form the channel lining. In the presence of 5-HT the channels undergo transitions between the open, desensitized, and closed states. We cannot distinguish in which state reaction is occurring. We did not determine MTSET reactivity in the absence of 5-HT because without data on the spontaneous open probability of each mutant it is difficult to interpret closed state reactivity data with sulfydryl reagents.
Five of the M2 segment cysteine substitution mutants did not form functional receptors when expressed in Xenopus oocytes. The nonfunctional mutants G288C, Y289C S290C, F292C, and D298C appeared to be targeted to the cell surface, based on the immunofluorescent antibody labeling experiments in HEK 293 cells. This implies that they were correctly assembled. Thus for these five the mutations probably do not cause a gross change in receptor structure; rather the cysteine residues at these positions appeared to prevent channel opening.

FIG. 4. Irreversible effects of MT-SET on 5-HT induced currents.
The percentage of change in mean peak current recorded from oocytes expressing each cysteine substitution mutant after a 1-min application of 1 mM MTSET as in Fig. 2B is shown. The left panel shows the effect on maximal peak current, and the right panel shows the effect on peak current at the appropriate EC 50 for each mutant. The data are the means Ϯ S.E. (n Ն 3). The filled bars denote significantly different to WT, p Ͻ 0.05 in the Dunnett test after one-way analysis of variance. NC, no currents could be recorded from this mutant.

FIG. 5. Irreversible effects of MT-SES on 5-HT-induced currents.
The percentage of change in mean peak current recorded from oocytes expressing each cysteine substitution mutant after a 1-min application of 10 mM MTSES as in Fig. 2B is shown. The left panel shows the effect on maximal peak current, and the right panel shows the effect on peak current at the appropriate EC 50 for each mutant. The data are the means Ϯ S.E. (n Ն 3). The filled bars denote significantly different to WT, p Ͻ 0.05 in the Dunnett test after one-way analysis of variance. NC, no currents could be recorded from this mutant.
At most reactive positions, MTSET inhibited the subsequent 5-HT-induced currents. We do not know whether the inhibition of macroscopic currents occurred because of a change in gating kinetics or to a change in single channel conductance or both. Single channel studies of each mutant would be necessary to determine this. In two of the mutants, D274C and I295C, MTSET modification potentiated the subsequent currents. This is most likely due to a change in gating kinetics because placing a positive charge on the wall of a cation conducting channel is unlikely to increase single channel conductance. An alternative explanation suggested by preliminary data from another group who have also performed SCAM on the 5-HT 3 receptor suggests that the potentiation at D274C may be due to relief of inhibition by extracellular magnesium ions (26).
Charge Selectivity-MTSET, which is positively charged, reacted with 11 cysteine mutants along the entire length of the M2 segment, whereas the negatively charged MTSES only reacted at three positions near the extracellular end of M2. The most cytoplasmic of these is the 13Ј position, V291C. This implies that both anions and cations can enter the extracellular end of the 5-HT 3 channel. At V291C the ratio of the reaction rate with MTSET to the rate with MTSES is ϳ10 (Figs. 6 and  7), similar to the ratio of the reaction rates with 2-mercaptoethanol in free solution (23). This implies that the access pathway to this residue is not charge-selective. Thus the residues lining the extracellular vestibule and the extracellular ring of charge (5) do not serve a major role in the charge selectivity process. The fact that an anionic reagent reacted with M2 segment cysteine mutants to the 13Ј level implies that the discrimination between cations and anions occurs at a more cytoplasmic position than V291C. This is consistent with results of similar experiments in the nACh receptor (27), although in the anion-selective GABA A receptor cationic reagents were able to react with engineered cysteines as far down as the ␣ 1 Thr 261 , the 6Ј level (28). In the cationic members of the gene superfamily, mutation of the 13Ј residue and of two others in the M1-M2 loop changes the charge selectivity and vice versa (4, 29 -31). This suggests that the major determinants of charge selectivity are near the cytoplasmic end of the channel.
The Structure of M2-The residues that react with MTSET form a pattern that is consistent with a significant proportion of M2 being in an ␣-helical conformation (Fig. 8, A and B). This is similar, although not identical, to SCAM data previously reported for nACh and GABA A receptor subunits (Fig. 8C). It is also consistent with other studies that indicate that the M2 segments in all ligand-gated ion channel superfamily subunits are predominantly ␣-helical (for review see Ref. 4).
There are, however, some discrepancies that reveal interesting features of the structure. Reaction at L285C (7Ј) is inconsistent with a straight ␣-helical secondary structure in the middle of M2. Based on the accessibility of the aligned residue or a neighboring residue in SCAM studies of the GABA A and nACh receptors (Fig. 8C), it was hypothesized that there may be a kink in the middle of M2 (27, 28). Furthermore, in the 9 Å resolution cryo-electron microscopy structure of the Torpedo nACh receptor the channel-lining segments appeared to have a bend in the middle with the ends angling away from the channel axis (29). It was hypothesized that the 9Ј leucine that is conserved in almost all ligand-gated ion channel subunits is located at the bend and that it forms the channel gate (29). However, other data are not consistent with this hypothesis. In the nACh receptor Wilson and Karlin (34) using SCAM experiments inferred that the channel gate is located between Gly 240 and Thr 244 (Ϫ2Ј and 2Ј; see Fig. 1). Also, mutation of the 9Ј leucine to smaller more polar residues (e.g. replacement with serine in the nACh receptor; Ref. 34) that might be expected to prevent gate formation has been shown to produce functional receptors. In addition, substitutions for the 9Ј leucine significantly modify channel gating in nACh, GABA A , and 5-HT 3 receptors and can result in stabilization of the open state relative to the resting state (35)(36)(37). Our data are also consistent with a critical role of this leucine residue; reaction of MTSET with the L287C (9Ј) mutants results in a receptor that appears to be locked in the open state. Thus a non-␣-helical structure in this region would allow access to L285C.
Near the extracellular end, assuming that M2 is ␣-helical, L293C is not then predicted to be on the channel-lining face of M2, but it reacted with MTSET (Fig. 8). There are several possible explanations for its accessibility. First, this part of M2 may extend above the level of the hydrophobic interior of the bilayer into the aqueous phase either continuously or tran- siently. This would allow the MTS reagents to gain access to non-channel-lining engineered cysteines. Alternatively, these residues may be on the back side of the M2 segment facing a water-filled crevice that extends into the interior of the membrane-spanning domain. Such a crevice was inferred to exist based on SCAM studies of the GABA A receptor ␣ 1 M3 segment (38). Furthermore, L293 is aligned with a residue in the GABA A receptor that when mutated reduces the efficacy of general anesthetics and may form part of the GABA A receptor anesthetic binding site (39). It is interesting to speculate this same protein region may be involved in general anesthetic interactions with 5-HT 3 receptors, although there is currently no evidence for this. Consistent with L293C not being in the channel, the reaction rates with MTSET and MTSES were slow and not significantly voltage-dependent. In fact the rates were similar to rates that have been measured with engineered cysteines in the dopamine D2 receptor binding crevice (24). Another, although less likely possibility, is that this region of M2 is not ␣-helical. SCAM studies of the acetylcholine and GABA A receptors (27, 28) are consistent with an ␣-helical secondary structure in this region, as is the acetylcholine receptor cryo-electron microscopy data (33).
MTSES modification locked the V291C, L293C, and I294C mutants in the open state. There are several possible mechanisms by which this may occur. Based on disulfide trapping experiments (22) it was hypothesized that channel opening B, peak currents from the 5-HT test applications in A plotted against cumulative reaction time with MTSES. The data are fit with the exponential function (Equation 4) to determine the second order rate constant. C, the second order rate constants for reaction of MTSES with cysteine substitution mutants (determined as in A and B) plotted against the holding potential. The data were fit by Equation 5. The slopes are in the opposite direction to those in Fig. 6 as expected for a negatively charged reagent. The data are the means Ϯ 32 S.E. (n Ն 3).
FIG. 8. Helical wheel representation of water-accessible residues. A, M2 residues of 5-HT 3A subunits plotted on a helical net; B, the same residues plotted on a helical wheel. Residues at which reaction with MTSET or MTSES caused a significant irreversible change in peak current are indicated by filled squares. Residues where no change in peak current was detected are indicated by open circles. Mutants for which no current could be recorded are underlined. C, alignment of residues in the ␣ subunit of the nACh and GABA A receptor and the 5-HT 3A receptor subunit. Asterisks denote residues that reacted with sulfydryl reagents and therefore are assumed to be water-accessible. D, model of the 5-HT 3A receptor subunit M2 domain based on our data. The ion path is indicated above those residues that reacted with sulfydryl reagents and were therefore deemed water-accessible. The non-␣-helical region in the center of M2 is also shown. involves a rotation of the M2 segments. Modification of engineered cysteine residues in multiple subunits in their open state position may prevent the M2 segments from rotating back to their closed state conformation, thereby locking the channels open. Lock open was not seen in SCAM studies of nACh and GABA A receptor (16,27,28), but in these experiments there were only one or two engineered cysteines in each receptor, whereas in the homomeric 5-HT 3 receptor there are five. Of note, in the nACh receptor, MTSEA modification of ␤V266C, which aligns with Val 291 in the 5-HT 3 receptor, caused a marked increase in single channel open time (16). The fact that MTSET caused almost complete inhibition of macroscopic currents at V291C might seem to contradict this hypothesis. It should be remembered, however, that placement of multiple positive charges on the channel wall by MTSET reaction with multiple channel-lining cysteines might lock the channels in the open state but prevent flux of cations through the channel. MTSES places multiple negative charges on the channel wall that may not block cation conduction.
To further characterize the locked-open state that resulted from MTSES modification of V291C, L293C, and I294C, we examined the ability of diltiazem, an open channel blocker of 5-HT 3 receptors (40), to block the locked-open current. Diltiazem partially inhibited the locked-open current in MTSESmodified V291C channels, but it did not inhibit the currents in L293C or I294C MTSES-modified channels. There are two possible explanations for this; either Leu 293 and Ile 294 are part of the diltiazem binding site and MTSES modification prevents binding, or these residues are inaccessible to the reagent, perhaps in a protein cleft.
The Electric Field in the Channel-Our data (Figs. 6, A and B, and 7C) show that the reaction of MTSET and MTSES with a number of the mutant channel residues is voltage-dependent, thus confirming that access to these residues does indeed occur via the ion channel. To further explore the location of these residues relative to the electric field in the channel, we used the information from the second order rate constants of reaction with MTSET to calculate the electrical distance from the bath solution to the reactive residues. These data are plotted in Fig.  6C and show that electrical distance is low at the extracellular end and high at the cytoplasmic end. Similar data have been calculated for the nACh receptor (20) and are as expected for these ion channels, thus providing yet more confirmatory evidence that M2 residues line the pore.
Conclusions-Thus in conclusion we have probed the accessibility and the electrostatic potential of the 5-HT 3 receptor ion channel using two differently charged sulfydryl derivatives. Some unexpected data have arisen that, with further experimentation, may clarify details of both the structure of M2 and the changes that occur during channel activation. Thus we observed that a number of mutant receptors were locked into the open state following reaction with the sulfydryl reagents. Mutations in M2 that result in receptors being locked open have been previously reported both in 5-HT 3 receptors (41) and in other ligand-gated ion channels (42)(43)(44), but this is the first report of being able to attain this state using sulfydryl reagents. Further experiments, for example using the 5-HT 3B subunit, could clarify the role of these residues. We also provide data that we have interpreted as suggesting that a protein cleft is located on the non-channel-lining side of M2 toward the extracellular end. If further work confirms the existence of such a cleft there are a number of candidate molecules, such as steroids and anesthetics, for which it might form a binding site (45). Overall, however, the data suggest that M2 is mostly ␣-helical and that charge selectivity occurs at a more intracellular level than Val 291 (13Ј). Thus the data are similar, although not identical, to those reported for nACh and GABA A receptors, further exemplifying the high degree of structural and functional homology between receptors in the Cys loop ligand-gated ion channel family.