Amino Acid Residues Involved in Gating Identified in the First Membrane-spanning Domain of the Rat P2X 2 Receptor*

The first hydrophobic segment of the rat P2X 2 recep- tor extends from residue Leu 29 to Val 51 . In the rat P2X 2 receptor, we mutated amino acids in this segment and adjoining flanking regions (Asp 15 through Thr 60 ) individually to cysteine and expressed the constructs in human embryonic kidney cells. Whole-cell recordings were used to measure membrane currents evoked by brief (2-s) applications of ATP (0.3–100 m M ). Currents were normal except for Y16C, R34C, Y43C, Y55C, and Q56C (no currents but normal membrane expression by immuno-histochemistry), Q37C (small currents), and F44C (nor-mal current but increased sensitivity to ATP, as well as ab -methylene-ATP). We used methanethiosulfonates of positive, negative, or no charge to test the accessibility of the substituted cysteines. D15C, P19C, V23C, V24C, G30C, Q37C, F44C, and V48C were strongly inhibited by neutral, membrane-permeant methanethiosulfonates. Only V48C was also inhibited by positively and negatively charged methanethiosulfonates, consistent with an extracellular position; however, accessibility of V48C was increased by channel opening. V48C could disulfide with I328C,

P2X receptors are a family of multimeric membrane proteins that function as ion channels gated by extracellular ATP. Hydrophobicity plots for P2X receptors suggest that two parts of the protein are sufficiently long and hydrophobic to cross the plasma membrane (1). These are the regions, in the P2X 2 receptor, of Leu 29 to Val 51 and of Ile 331 to Leu 353 . Considerable experimental evidence now supports the view that the N and the C termini are intracellular, and the region between Val 51 and Ile 331 faces the extracellular aspect. First, antibodies against N-and C-terminal epitopes work only in permeabilized cells (2). Second, the proteins can be glycosylated at both natural and artificially introduced consensus sequences (NX(S/T)) at several positions in the extracellular domain from Pro 62 to Lys 324 (in the P2X 2 receptor), though not at such positions in the N terminus (positions 9, 16, or 26) (2)(3)(4). Third, concatenated cDNAs in which the C terminus of one construct is joined to the N terminus of a second form functional channels (5).
There is now biochemical evidence that the P2X receptors form channels as trimers (6,7). However, the parts of the individual subunits that contribute to different functions of the receptor are little understood. Mutations of several positively charged residues have been shown to decrease the effectiveness of ATP as an agonist at the P2X 1 (8) and P2X 2 (9) receptors, and these residues occupy corresponding positions (e.g. Lys 69 , Lys 71 , Lys 188 , Arg 290 , Arg 304 , and Lys 308 in rat P2X 2 numbering). The region around Lys 69 and Lys 71 is of particular interest with regard to a possible ATP binding site. The P2X 2 receptor functions normally when Ile 67 is mutated to cysteine (I67C). However, the attachment of a negatively charged methanethiosulfonate ((2-sulfonatoethyl) methanethiosulfonate; MTSES 1 ) led to a parallel rightward shift in the ATP concentration-response curve that was not seen with neutral (methyl methanethiosulfonate; MTSM) or positively charged methanethiosulfonate ([2-(trimethylammonium)ethyl] methanethiosulfonate; MTSET). Point mutations that introduced a negative charge (I67E and I67D), but not those that introduced a positive charge (I67R and I67K), also caused inhibition of the current that could be overcome by increasing the ATP concentration. Together these results provide strong evidence that this region of the receptor contributes to the ATP binding site (9).
When ATP binds to the P2X receptor the protein undergoes a conformational change that results in the opening of a cationpermeable channel. The substituted cysteine accessibility method has also been used to implicate residues in and around the second transmembrane domain in the formation of the ion-conducting pathway (10,11). In particular, P2X 2 -T336C is almost completely blocked by exposure for 8 min to MTSET and MTSES; both negatively and positively charged MTS reagents were effective, suggesting that Thr 336 is located outside the membrane electric field, but the finding that outward currents were blocked more rapidly than inward currents indicates that the attached side chain might directly interfere with permeation (10). Another residue within the second hydrophobic segment (Asp 349 ) showed prominent block by MTSEA, but not by MTSET and MTSES. This block did not occur if the MTSEA was applied without opening, and in view of the fact that MTSEA is quite permeable through the P2X 2 receptor channel, this suggests that Asp 349 is situated internal to the "gate" of the channel (10).
The purpose of the present experiments was to ascertain whether residues in and around the first hydrophobic segment might also contribute to the ATP binding site or to the permeation pathway. P2X receptors are not well conserved in the regions corresponding to the first 14 amino acids of the P2X 2 receptor, and we therefore began our cysteine substitutions at Asp 15 . We have recently reported the effects of MTS compounds on the region Asp 57 to Lys 71 (9); in the present experiments we ended the cysteine substitutions at Thr 60 (see Fig. 1). As a first approach we used MTSM, a small, neutral methanethiosulfonate, in conjunction with point mutations to cysteine. We reasoned that this might provide a picture of cysteines accessible to the aqueous environment on both the intracellular and extracellular aspects of the receptor. We followed this with tests of positively and negatively charged methanethiosulfonates for those positions at which MTSM caused a large inhibition. In an effort to understand further the mechanism of the inhibition we studied the effect on the ATP concentration-response curve and asked whether the inhibition required channel opening. Finally, we sought to determine whether substituted cysteines in the two transmembrane domains were sufficiently close to form disulfide bonds. EXPERIMENTAL PROCEDURES P2X 2 Receptor cDNA and Mutagenesis-A P2X 2 subunit cDNA carrying a C terminus epitope was used; its source and the methods used for introducing point mutations were as described previously (10). All mutants were sequenced on both strands.
Electrophysiology-Transient transfection of human embryonic kidney 293 cells using Lipofectin or LipofectAMINE 2000 was as described previously (10,(12)(13)(14). Standard whole-cell recordings and fast-flow agonist applications were made as previously described (10,12,13). Internal solution contained (in mM): 145 NaF, 10 EGTA, and 10 HEPES; external solution was (in mM): 147 NaCl, 2 CaCl 2 , 2 KCl, 1 MgCl 2 , 13 glucose, and 10 HEPES. Current-voltage relations were obtained by ramp voltages (1-s duration) from Ϫ120 to 40 mV. The following MTS reagents used were obtained from Toronto Research Chemicals (Ontario, Canada): MTSEA, MTSET, MTSES, MTSM, and butyl methanethiosulfonate (MTSB). MTSM and MTSB were used from a 1 M stock solution that was made in Me 2 SO and kept as frozen aliquots; stock solutions (100 mM) for the other MTS compounds were made daily by dissolving the solid in control external solution kept at 4°C. All MTS compounds were diluted to 1 mM immediately prior (2-5 min) to their application. Numerical estimates of EC 50 values were made for individual cells by least squares curve fitting as previously described (9), using the function I/I max ϭ [ATP] n /(EC 50 n ϩ [ATP] n ), where I is the current as a fraction of the maximum current (I max ). The figures show this function fitted to the mean for all cells tested. Results are shown as means Ϯ S.E. Tests of significance between paired observations were by Student's t test or non-parametric Mann-Whitney test; the effect of MTSM at many different positions was compared with that on the wild-type channel by analysis of variance, followed by Tukey-Kramer multiple comparison test (InStat software; GraphPad, San Diego, CA). Results were considered significant for p Ͻ 0.05.

RESULTS
Effects of Cysteine Substitutions-We introduced cysteine into each position individually and studied the actions of ATP on human embryonic kidney 293 cells expressing the mutated receptors. ATP (30 M) elicited currents not distinguishable from those in cells expressing wild-type receptors (1-8 nA) for all the mutated receptors except Y16C, R34C, Q37C, Y43C, F44C, Y55C, and Q56C. Cells expressing Y16C, R34C, Y43C, Y55C, and Q56C showed no responses to ATP (up to 3 or 10 mM); immunohistochemistry showed staining of the plasma membrane in these cells. For Q37C, ATP-evoked currents were smaller (0.3 to 2 nA), and the EC 50 was about three times higher than for the wild-type receptor (29 Ϯ 5.7 M; n ϭ 4). Cells expressing two further mutations (T18C and L29C) responded well to an initial application of ATP, but the current declined steeply with repeated applications, and they could therefore not be usefully studied. At T18C, the current also declined during the ATP application more rapidly than seen at wild-type channels; at the end of a 2-s application (30 M) the current was 43 Ϯ 7% (n ϭ 5) of its peak for T18C and 89 Ϯ 2.8% (n ϭ 8) for the wild-type receptor. This is similar to the finding of Boué-Grabot et al. (16) for T18A.
At F44C, cysteine substitution caused three distinct changes in the properties of the receptor. First, ATP-evoked currents returned back to the baseline level more slowly than normal after a brief (2-s) application. The times required to return to half the peak current at the end of ATP application were 0.44 Ϯ 0.04 s (n ϭ 15) for wild-type receptors (30 M ATP) and 1.2 Ϯ 0.8 (n ϭ 9; 3 M ATP) or 1.6 Ϯ 0.05 s (n ϭ 29; 30 M ATP) for F44C. Second, there was 10-fold increase in sensitivity to ATP (EC 50 0.72 Ϯ 0.1 M; n ϭ 4). Third, there was a remarkable increase in effectiveness of ␣␤meATP. The wild-type P2X 2 receptor is essentially insensitive to ␣␤meATP (13); indeed, we found that 100 and 300 M ␣␤meATP evoked currents in cells expressing wild-type P2X 2 receptors were, respectively, 1.1 Ϯ 0.2 and 8.1 Ϯ 1.3% (n ϭ 5) of the currents evoked by 100 M ATP. In contrast, at F44C receptors ␣␤meATP activated currents with an EC 50 value of 10.8 Ϯ 0.5 M (n ϭ 4), and the maximum current evoked by ␣␤meATP (100 M) (1.7 Ϯ 0.2 nA; n ϭ 5) was similar to that evoked by a maximal concentration of ATP (3 M) (1.2 Ϯ 0.2 nA; n ϭ 5). There was no difference in the holding current between cells expressing F44C and wildtype receptors.
Accessibility to MTSM-We used MTSM for an initial screen of this segment of the receptor; it is small and uncharged, and we expected that it would cross the membrane readily and react with accessible residues on either the cytoplasmic or the extracellular part of the receptor protein. Fig. 1 illustrates representative ATP-evoked currents prior to and during an 8-min application of MTSM at 1 mM and after washing for 4 min. The effects of MTSM are summarized in Fig. 2. There was no significant effect on ATP-evoked currents at wild-type receptors. We included T336C as a control mutation, and found that the inhibition (Ͼ80%) was similar to that previously reported for MTSEA, MTSET, and MTSES (10). At 8 of the 39 cysteine-substituted receptors that responded to ATP, MTSM caused a large (Ն 60%) inhibition of the current that was significantly different from the wild-type (p Ͻ 0.001); these were D15C, P19C, V23C, V24C, G30C, Q37C, F44C, and V48C (Fig. 3). This inhibition did not reverse on washing out the MTSM for up to 10 min (Fig. 1). These effects of MTSM were mimicked closely by another neutral MTS derivative, MTSB. At 1 mM (for 8 min), the inhibitions by MTSB were as follows: V15C, 90 Ϯ 3% (n ϭ 3); P19C, 89 Ϯ 7.8% (n ϭ 4); V23C, 82 Ϯ 4.1% (n ϭ 3); and V24C, 97 Ϯ 1.1% (n ϭ 4). The inhibition of the current by MTSM was not obviously dependent on membrane potential as judged from ramp current-voltage plots.
Previous studies on the second transmembrane domain found no effects by MTSEA, MTSET, and MTSES on the positions on the C-terminal side of D349C (W350C, I351C, L352C, L353C, and T354C). These positions are clearly at the inner aspect of the second transmembrane domain, and we re-examined them using MTSM. We found that MTSM (1 mM, 8 min) gave rise to significant inhibition (p Ͻ 0.001) at I351C (76.5 Ϯ 6.2%; n ϭ 3) and L352 (69.8 Ϯ 5.1%; n ϭ 3).
Effects Taken together, the results with MTSM, MTSES, and MT-SET are consistent with the topology currently proposed for the P2X 2 receptor. Introduction of cysteine at positions Asp 15 , Pro 19 , Val 23 , Val 24 , and Gly 30 (before the first hydrophobic segment) and Ile 351 and Leu 352 (end of the second hydrophobic segment) led to significant inhibition by the membrane permeant MTSM but little or no inhibition by charged MTS derivatives. Conversely, cysteine substitution at Val 48 (at the outer edge of the first hydrophobic segment) resulted in strong inhibition by MTSM, MTSES, and MTSET as we have previously described for three residues (Ile 328 , Asn 333 , and Thr 336 ) at the beginning of the second hydrophobic domain (10).
Effects of MTSM Modification on the ATP Concentrationresponse Curve-The shape of the concentration-response curve for ATP might provide information on the mechanism by which the current is inhibited (9,15) , n ϭ 3, respectively). These values are close to those for the wild-type receptor (7.9 Ϯ 1.1 M; n ϭ 8). In other words, MTSM modification at these positions results in a simple depression of the maximum current evoked by ATP, with little change in the EC 50 . This is similar to the result observed with T336C (see Fig. 4 and Refs. 9 and 10).
Dependence of MTS Inhibition on Channel Opening by ATP-V24C was rapidly and completely inhibited by MTSM (Figs. 1, 3, and 5). This inhibition was essentially the same even when ATP applications were discontinued during the presence of the MTSM (Fig. 5A, left). On the other hand, Fig.  5B (left) shows that the positively charged MTSET produced little or no inhibition of the currents at V24C unless ATP was repeatedly applied. We interpret this to indicate that Val 24 is situated on the intracellular aspect of the receptor, but it can be accessed by MTSET entering through the open channel. This is the same result, and the same conclusion, as we made previously for inhibition at D349C by MTSEA (10).
In the case of V48C, inhibition was observed with MTSM, MTSET, and MTSES (Fig. 3). However, the effectiveness of MTSM and MTSET was considerably greater when the ATP was repeatedly applied than when it was not applied during the presence of the MTS derivative (Fig. 5, A and B, right). This result implies that conformational changes associated with ATP binding and channel opening moves V48C into a position in which it is much more readily accessible to reaction with MTS derivatives. In other words, Val 48 moves as a result of channel opening, and by moving it becomes more accessible to MTS derivatives.
Disulfide Formation between V48C and I328C-We have previously presented evidence that T336C is located in the outer vestibule of the ionic channel; the evidence for this was that outward currents were inhibited more rapidly than inward currents as the MTSET reacted with the cysteine. The Following a 2-min wash, the second ATP concentration-response curve was obtained (filled circles). The duration of MTSM application was altered to achieve a significant but not complete inhibition; it was 0.5 min for V24C, 2 min for P19C, V23C, and F44C, and 4 min for V48C and T336C. Broken lines indicate mean EC 50 values for ATP before applying MTSM (derived by fitting to individual cells; see "Experimental Procedures"; n ϭ 3-6 cells for each case). present work indicates that Val 48 is situated at the outer edge of the membrane, and we therefore asked whether these residues were sufficiently close to form disulfides that altered the properties of the channel. We expressed the double mutants V48C/I328C, V48C/N333C, and V48C/T336. The current elicited by ATP (30 M) at the V48C/I328C receptor was much smaller (243 Ϯ 70 pA; n ϭ 11) than wild-type, V48C, or I328C receptors (Fig. 6). We also observed relatively large inward currents when the cells were held at Ϫ60 mV; it normally required less than Ϫ50 pA to hold a human embryonic kidney 293 cell at Ϫ60 mV, but for V48C/I328C this was Ϫ235 Ϯ 52 pA (n ϭ 8). This suggested that the P2X 2 receptor channel was constitutively open in this mutated receptor. Dithiothreitol (10 mM) greatly increased the amplitude of the current evoked by ATP (about 6-fold) over 20 min and progressively reduced the sustained holding current in the absence of ATP (Fig. 6, A and  C). The ATP-evoked current increased exponentially with time constant () of 5.9 Ϯ 0.8 min (n ϭ 8) with 10 mM dithiothreitol, whereas the sustained inward holding current declined rather more slowly ( ϭ 19.3 Ϯ 7.4 min; n ϭ 6) and had reached Ϫ76 Ϯ 41 pA at 20 min. A further reducing agent, bismercaptoethanol (5 mM), also potentiated the ATP-evoked currents (Fig. 6, B and C). Its action was somewhat more rapid than that of dithiothreitol ( ϭ 1.2 Ϯ 0.5 min; n ϭ 4). Bismercaptoethanol also reduced the inward holding current from Ϫ299 Ϯ 115 to Ϫ48 Ϯ 17 pA (n ϭ 4) during a 20-min application ( ϭ 4.3 Ϯ 1.6 min; n ϭ 3).
ATP-evoked currents for V48C/N333C and V48C/T336C were similar to those observed for single cysteine mutants V48C, N333C, or T336C (range 1-8 nA), and application of dithiothreitol at 10 mM for 20 min had no effect on the currents (see Fig. 6C).

DISCUSSION
Effects of Introducing Cysteines-The rat P2X 2 receptor was tolerant of cysteine introduced in all but 5 of the 46 positions examined between Asp 15 and Thr 60 (Figs. 2 and 7). Y16C, Y43C, Y55C, and Q56C were non-functional; these residues are completely conserved among all mammalian P2X receptors. R34C also did not express channels; arginine is found in all subunits except P2X 7 , where it is replaced by tryptophan. In four positions the introduction of cysteine led to an obviously altered phenotype. In the case of T18C and L29C, the response to ATP declined markedly when ATP was applied more than once. Thr 18 in the P2X 2 receptor has been shown by Boué-Grabot et al. (16) to be phosphorylated by protein kinase C, and this alters the desensitization kinetics. Leu 29 has not previously been mutated, but we note that it lies very close to the inner edge of the first hydrophobic domain. Q37C expressed more poorly (smaller maximum currents) than the other mutations and was less sensitive to ATP than wild-type receptors.
The effect of mutating Phe 44 to cysteine was surprising in that it resulted in a 10-fold increase in sensitivity to ATP and an even larger increase in sensitivity to ␣␤meATP. Insensitivity to ␣␤meATP has come to be regarded as a major distinguishing feature among the different subtypes of P2X receptor, with those containing P2X 1 and P2X 3 subunits sensitive (P2X 1 and P2X 3 homomers and P2X 2/3 and P2X 1/5 heteromers) and those not containing these subunits several hundred-fold less sensitive (17). There have not been previous reports of point mutations conferring sensitivity to ␣␤meATP in P2X receptors of the insensitive subclasses (P2X 2 , P2X 4 , P2X 5 , and P2X 7 ). Phenylalanine is found in this position in the P2X 2 and P2X 3 subunits, but in the others the residue is leucine, valine, or isoleucine. One interpretation of the increased effectiveness is that this position contributes to the ATP binding site. This seems somewhat unlikely in view of the fact that it is situated well within the first hydrophobic domain. A more likely explanation might be that ␣␤meATP normally can bind to the P2X receptor in much the same way as ATP but that it has very low efficacy to induce the conformational change leading to channel opening. From the results for the wild-type channel, the EC 50 for ␣␤meATP can be very crudely estimated as around 1 mM, but this is difficult to verify experimentally because of doubts that the ␣␤meATP might contain small amounts of ATP. In F44C, the EC 50 value for ATP shifted about 10-fold (from 8 to 0.7 M); the EC 50 value for ␣␤meATP must then have shifted about 100-fold (from about 1000 to 10 M). A direct effect of this mutation on channel gating was also indicated by the observation that the currents evoked by ATP took longer to decline to the baseline in F44C than in wild-type channels; this is consistent with slowed channel closing. Further analysis of this position with other substitutions, and single channel recordings, is likely to provide insight into the mechanisms of gating. One corollary of this interpretation is that ␣␤meATP should act as a competitive antagonist of ATP at wild-type P2X 2 receptors. Although this has not been reported, there is evidence that ␣␤meATP is an antagonist of ATP action at P2X receptors in sympathetic neurons (18). Ϫ ] side chain on the receptor protein. MTSM had no significant effect on the wild-type P2X 2 receptors but gave strong inhibition of ATP-evoked currents in eight of the cysteinesubstituted receptors (Figs. 2 and 7). Where MTSM has no effect, we cannot say whether the cysteine is not accessible to an aqueous solution or whether the cysteine is modified, but this modification does not change the channel properties in any way that we have studied. The rapid rate at which MTSM reacted with some cysteines at intracellular locations (e.g. G30C and V24C) indicates that there is little obstacle to its passage across the plasma membrane. Other methanethiosulfonates did not have significant effects on cysteines at intracellular positions in the N-terminal region of the receptor, except for V24C. In this case, the action of MTSET was dependent on ATP application, suggesting that it entered the cell through the open channel. We have shown previously that MTSET is about 16% as permeable as sodium through P2X 2 receptors (10); we have not used MTSEA, because we (10) and others (19) have found that it can enter the cell quite readily, presumably in its uncharged state. Cysteine substitutions at positions at the inner end of the second transmembrane domain (Trp 350 to Thr 355 ) have previously been shown to be unreactive to MTSET; the present work showed that two of them were accessible to MTSM, and this is consistent with an intracellular location. In general, the results with MTSM and MTSET are as would be expected on the basis of the topological models currently proposed for the receptor (Fig. 7).
For all the modified cysteines, the reduction in the ATPevoked current occurred without change in the EC 50 value. In other words, increasing the ATP concentration could not overcome the inhibition of the current resulting from methanethiosulfonate application. One can distinguish broadly between a reduced affinity of the closed channel for ATP (i.e. binding), an impaired ability of the channel to open and stay open when ATP is bound (gating), and a reduced current through the open channel (permeation) (14). Impairment of binding or gating would usually produce a rightward parallel shift in the concentration-response curve before the maximum is reduced, whereas reduction in open channel current would not. It is conceivable that mutations P19C, V23C, V24C, F44C, and V48C (Fig. 4) all directly affect permeation, but independent direct measurements would be required to show this. In no FIG. 7. Schematic summary of results with methanethiosulfonates on cysteine-substituted rat P2X 2 receptor. All the residues shown have been substituted by cysteine and tested with methanethiosulfonates. Shaded residues indicate positions at which cysteine substitution results in strong inhibition. Open letters indicate positions where cysteine substitution results in non-functioning channels. Bold circles indicate two positions at which functional channels were expressed, but methanethiosulfonates could not be studied because of profound run-down of response when ATP application was repeated. Val 48 and Ile 328 can be disulfided. The present work reports results from Asp 15 to Thr 60 and Trp 350 to Thr 354 . Other results included are from Refs. 9 and 10. cases was there, for example, any obvious effect on the rectification of the whole-cell current after cysteine modification.
Movement of Val 48 with Channel Opening-The inhibition of current observed in V48C closely resembled that which we have previously found for I328C, N333C, and T336C (10), all of which are located close to the outer end of the second transmembrane domain. The finding that all three methanethiosulfonates (positive, neutral, and negative) cause strong inhibition indicates that this position is situated outside the membrane electric field. We were surprised therefore to observe that the reaction at V48C occurred much more rapidly when ATP was repeatedly applied than when it was not applied (Fig. 5). This result implies that the cysteine in the position of Val 48 moves to become more accessible when the channel is opened. A second unique feature of V48C was the finding that MTSET treatment caused an inward current to persist following the ATP application. This was not observed for the uncharged reagent MTSM, even though that caused similar inhibition of the current. We interpret this to indicate that channel closing is inhibited by the attachment of a positively charged moiety in this position; a similar observation was made previously for T336C (10). In that earlier work, there were three positions at the outer edge of the second transmembrane domain that were reactive with MTSET (Ile 328 , Asn 333 , and Thr 336 ). Because Val 48 is situated at the outer edge of the first membranespanning domain, we hypothesized that they were sufficiently close to form a disulfide bond. This was tested directly by expressing the doubly mutated receptors V48C/I328C, V48C/ N333C, and V48C/T336C. The latter two combinations resulted in currents that were not different from wild-type receptors, but the currents in cells expressing the V48C/I328C combination were very much reduced in amplitude (Fig. 6). At the same time, these cells exhibited large steady inward currents at Ϫ60 mV. Application of either dithiothreitol or bismercaptoethanol greatly increased the ATP-induced current and concomitantly decreased the persistent inward current. Taken together, these observations suggest that a disulfide formed between V48C and I328C results in a channel that is constitutively open and cannot be opened by applying ATP. The small currents that we observed prior to adding the reducing agent might indicate that the disulfide bond had not formed in all the channels or that the channels could operate, though poorly, with the disulfide bond in place. The impairment of channel opening by the V48C/ I328C disulfide is quite consistent with the conclusion reached above that Val 48 moves during channel opening. Moreover, the finding that the V48C channel does not open when this position is "tethered" to I328C suggests that the movement of Val 48 is not "fortuitous" at some incidental part of the protein, but is a necessary component of the gating mechanism.
It is not possible to conclude from the present work whether the disulfide is formed between V48C and I328C on the same receptor subunit or on different receptor subunits that contribute to the multimeric channel. However, the results do put constraints on models for the channel. In Fig. 7 the two membrane-spanning domains are depicted as ␣-helices. There is no evidence for this, except to say that it is strongly favored by secondary structure prediction algorithms (20,21). The accessible residues in the first transmembrane domain are located at one side of the helix. The proximity of Val 48 and Ile 328 indicated by the present results is fully consistent with this structure. For both transmembrane segments virtually all of the accessible residues mapped by cysteine scanning can be aligned along one face of an ␣-helix (Fig. 7). Key residues involved in gating the mechanosensitive channel of Escherichia coli (which also has two membrane-spanning domains per subunit) have a similar relative orientation (22). The proposed model suggests several opportunities for future experiments to increase our understanding of the modus operandi of P2X receptors.