The First Transmembrane Domain of the P2X Receptor Subunit Participates in the Agonist-induced Gating of the Channel*

Based on pharmacological properties, the P2X receptor family can be subdivided into those homo-oli-gomers that are sensitive to the ATP analog (cid:1)(cid:2) -methyl-ene ATP( (cid:1)(cid:2) meATP) (P2X 1 and P2X 3 ) and those that are not (P2X 2 , P2X 4 , P2X 5 , P2X 6 , and P2X 7 ). We exploited this dichotomy through the construction of chimeric receptors and site-directed mutagenesis in order to identify domains responsible for these differences in the abili-ties of extracellular agonists to gate P2X receptors. Replacement of the extracellular domain of the (cid:1)(cid:2) meATP-sensitive rat P2X 1 subunit with that of the (cid:1)(cid:2) meATP- insensitive rat P2X 2 subunit resulted in a receptor that was still (cid:1)(cid:2) meATP-sensitive, suggesting a non-extracel-lular domain was responsible for the differential gating of P2X receptors by various agonists. Replacement of the first transmembrane domain of the rat P2X 2 subunit with one from an (cid:1)(cid:2) meATP-sensitive subunit (either rat P2X 1 or P2X 3 subunit) converted the resulting chimera to (cid:1)(cid:2) meATP sensitivity. This conversion did not occur when the first transmembrane domain came from a non- (cid:1)(cid:2)

P2X receptors are ligand-gated ion channels activated by extracellular ATP. Although these receptors were first described almost 20 years ago (1), the lack of useful pharmacological tools has greatly hampered the elucidation of the roles that these receptors play in ongoing physiological functions (2,3). Recent advances in the molecular biology of these receptors have led to a resurgence of interest, and have served to illustrate how little is actually known about these proteins. Indeed, results from in situ hybridization and immunochemical studies demonstrate that these subunits have a widespread distribution throughout the body, being present in almost all tissues (for a review, see Ref. 4), suggesting that they may have more extensive functions than appreciated previously.
To date, a total of seven individual subunit genes have been cloned and their products characterized (4). These subunits have been demonstrated to form homo-and/or hetero-oligomeric receptors (5) that are non-selective cation channels with a high permeability to Ca 2ϩ (6), a property that confers the potential for important functions in excitable and secretory cells. Recent investigation into the structural features of the subunits has helped elucidate the role of the transmembrane domains in ion permeation through the channel (7)(8)(9), and have identified a number of extracellular residues as affecting the binding of agonists and antagonists (10 -13). Nevertheless, the domains involved in ligand-induced opening of the channel (gating ) have not yet been delineated, nor have the domains responsible for the pharmacological fingerprints of the various homo-and hetero-oligomeric P2X receptors. Originally, the P2X receptor family was distinguished from other ATP receptors (such as the G-protein-coupled P2Y receptors) by their sensitivity (EC 50 Յ 30 M) to the ATP analog ␣␤-methylene-ATP (␣␤meATP) 1 (14). Characterization of the cloned subunits has now shown that only two of the seven mammalian gene products have appreciable sensitivity to this analog (3). For this study, we sought to use chimeric P2X proteins to exploit the pharmacological differences between the ␣␤meATP-sensitive and -insensitive subunits in an effort to identify the domains and/or amino acid residues involved in determining agonist potencies at P2X subunits. Topological studies have shown that a P2X subunit has two transmembrane domains, with the intervening loop located extracellularly and the N and C termini located inside the cell (15). Previous studies from other groups have provided evidence that residues in the extracellular loop near the first transmembrane domain (TMD1) can affect the binding of ATP to the receptor (12,13), and results from our laboratories (16) suggested that mutations in TMD1 can influence channel gating and agonist potency. In this report we present further evidence that TMD1 determines the potencies of the agonists ATP and ␣␤meATP in gating the P2X 2 receptor, leading to the interpretation that this domain is involved in the transduction pathway linking agonist binding to opening of the channel gate in this receptor family.

MATERIALS AND METHODS
DNA Constructs-Several different types of mutant receptors were engineered, all of them through the use of overlap polymerase chain reaction as described previously (5). Briefly, chimerae were constructed by performing two successive polymerase chain reaction amplifications using 5Ј and 3Ј primers containing overlapping regions between rP2X 2 and the specified P2X subunit (rP2X 1 , rP2X 3  with that of rP2X 2 we used oligonucleotides encoding aa VYEK/SYQD (corresponding to aa 50 -53 for rP2X 1 , aa 54-57 for rP2X 2 ) for the post-TMD1 overlap and oligos encoding aa AGKF/DIIP (aa 322-325 from rP2X 2 , aa 327-330 from rP2X 1 ) for the pre-TMD2 overlap. The final construct (named 1.2.1) consists of the N terminus and TMD1 of rP2X 1 followed by the extracellular region of rP2X 2 up to 2-4 aa from TMD2, at which point the protein reverts back to rP2X 1 until the final stop codon. Additional chimerae were then generated that consisted of a rP2X 2 backbone in which TMD1 (aa 30 -53) was replaced by the cognate sequence from one of the following: rP2X 1 , rP2X 3 , or zP2X 3 . Point mutations were introduced into the appropriate construct using previously described methods (8). All mutants and chimeras were tagged at the C terminus with the FLAG epitope and their sequences verified by DNA sequencing using the Terminator kit from Amersham Pharmacia Biotech.
Transfection of HEK 293 Cells-cDNAs encoding the various P2X subunit constructs were expressed individually in HEK 293 cells. 35-mm dishes containing 3 ϫ 10 5 cells were incubated with 1 g of total cDNA mixed with 6 l of LipofectAMINE (Life Technologies, Inc.) in 1 ml of serum-free medium. After 5 h at 37°C, the medium was replaced with minimal essential medium and cells were incubated for another 40 -48 h. All constructs were co-transfected with an eGFP-expressing plasmid to aid in the detection of transfected cells.
Electrophysiological Recordings-Whole-cell currents were recorded at room temperature using the perforated-patch method. The pipette solution contained (in mM): 130 cesium methanesulfonate, 24 CsCl, 1 MgCl 2 , 1 CaCl 2 , 10 HEPES, and amphotericin B (200 g/ml) (17). The extracellular solution contained (in mM): 154 NaCl, 1 MgCl 2 , 2 CaCl 2 , 10 glucose, and 10 HEPES. Cells predicted to express P2X receptor protein were identified using fluorescence microscopy to detect the presence of eGFP. Different concentrations of ATP were applied by positioning the cell in front of one of a number of inlet tubes arranged side-by-side in the bath chamber as described previously (18). Successive applications of ATP were separated by at least 3 min to minimize receptor desensitization. Concentration-response curves were generated from the nonlinear, least squares fit to the Hill equation as implemented in version 4.0 of Igor Pro (www.wavemetrics.com) using the averaged data computed for each concentration of ATP. All values given are means Ϯ standard deviations.

RESULTS
The potency of ATP at the homo-oligomeric rP2X receptors falls into three categories: high (EC 50 ϭ 1-3 M, seen at rP2X 1 and rP2X 3 ), intermediate (EC 50 ϭ 10 -30 M, seen at rP2X 2 , rP2X 4 , and rP2X 5 ), and low (EC 50 Ն 100 M, seen at rP2X 7 ) (3). ATP has no action at P2X 6 , as this subunit does not appear to form homo-oligomeric receptors (5,19). In contrast, the ATP analog ␣␤meATP exhibits high potency and efficacy at only two subunits (rP2X 1 and rP2X 3 ; EC 50 Х 1 M), while having little or no potency at the other five subunits even at concentrations approaching the millimolar level (3). There are two simple hypotheses that could explain the inability of ␣␤meATP to operate five of the seven subunits: first, that ␣␤meATP does indeed bind with appreciable affinity to all receptors but is unable to induce the channel to gate in five of them (i.e. it is functionally equivalent to an antagonist), and second, that ␣␤meATP does not interact at the agonist binding domain with appreciable affinity and thus its binding energy is not sufficient to cause the degree of tertiary/quaternary structure perturbation necessary to result in the gating of the channel. It was important to determine which of the two hypotheses explains the ␣␤meATP data so that experiments to investigate the structural basis for the pharmacological differences among subunits could be designed in a rational fashion.
The first hypothesis, that ␣␤meATP is a functional antagonist, was the most straightforward question to test, and we did so by examining the ability of a relatively high (100 M) concentration of ␣␤meATP to reduce the efficacy of a co-applied dose of 10 M ATP at the homomeric rP2X 2 receptors, at which ATP has moderate and ␣␤meATP little potency (Fig. 1a, ATP EC 50 ϭ 22 M; ␣␤meATP EC 50 Ͼ 300 M). As shown in Fig. 1b, co-application of ␣␤meATP did not alter the degree to which a sub-EC 50 concentration of ATP activated the receptor. This result rules out the hypothesis that ␣␤meATP binds to the agonist site on the rP2X 2 receptor with appreciable affinity. This leaves the second hypothesis, that ␣␤meATP's inability to gate the rP2X 2 subunit is due to a low binding affinity for the agonist binding domain, to be tested. This low ␣␤meATP affinity would presumably result from differences in the threedimensional structure of the P2X 2 receptor compared with the P2X 1 or P2X 3 receptors that arise from the evolutionarily derived amino acid variations present in their predicted primary sequences. We therefore exploited this sequence divergence in testing the low affinity hypothesis by constructing chimeric receptors derived from ␣␤meATP-sensitive and -insensitive subunits.
Chimeras Formed from P2X 1 and P2X 2 Subunits Show Altered ␣␤meATP Sensitivity-The first chimeric receptor constructed was composed of the ␣␤meATP-sensitive rP2X 1 subunit and the ␣␤meATP-insensitive rP2X 2 subunit. This initial chimera, labeled 1.2.1, was a rP2X 1 subunit chain in which only the extracellular domain (presumptive site of the ATP-binding domain) was replaced with the equivalent sequence of the rP2X 2 subunit (schematic shown in Fig. 2a). It was anticipated that this "domain swap" would yield a homo-oligomeric recep-tor with the pharmacology of the P2X 2 receptor. In contrast to these expectations, the receptor formed from this chimeric subunit exhibited increases in both ATP and ␣␤meATP sensitivity (ATP: EC 50 ϭ 0.93 M, ␣␤meATP: EC 50 ϭ 49 M) (Fig. 2b). This unexpected result demonstrated that the extracellular domain of rP2X 2 was indeed capable of forming a binding pocket at which ␣␤meATP could bind with both high affinity and efficacy and, by extension, that non-extracellular regions of a P2X subunit protein can profoundly influence the agonist binding and/or gating properties of the receptor.
Due to the nature of the 1.2.1 chimera, the increased agonist sensitivities observed would have to be attributable to one or more of the amino acid residue(s) present in the rP2X 1 portion of the molecule. There are four distinct domains of rP2X 1 in 1.2.1: the cytosolic N terminus, TMD1, TMD2, and the cytosolic C terminus. Recent work from our laboratories has found that point mutations in the TMD1 of rP2X 2 can change the potency of ATP at the receptor as well as influence the gating properties of the channel (16). In addition, data from two other groups (12,13) strongly suggest that amino acids present in the extracellular domain just C-terminal of TMD1 are involved in ATP binding. Together, those findings suggested to us that the increased potencies of both ATP and ␣␤meATP seen for chimera 1.2.1 might originate from amino acids present in the rP2X 1 -TMD1 region. To test this hypothesis, we constructed a new chimera that comprised the rP2X 2 subunit in which the only alteration was that its TMD1 domain was replaced by the TMD1 of rP2X 1 (this chimera is labeled TM1(r1)). As seen in Fig. 3a, this chimera also had increased sensitivity to both ATP To determine if the change observed in agonist potencies was found for antagonists as well, we tested the actions of the antagonist trinitrophenyl-ATP (TNP-ATP) on ATP-induced currents in cells expressing TM1(r1). This compound has high affinity for the P2X 1 and low affinity for the P2X 2 subunit (IC 50 Х 3 nM versus 1 M, respectively; Ref. 19) and therefore should be a sensitive tool for detecting any global change in ligand interactions with the chimeric receptor. As shown in Fig. 3b, the affinity (IC 50 ϭ 1.7 M) and the Hill slope (n H ϭ Ϫ1.1) for TNP-ATP inhibition of ATP-gated currents at TM1(r1) matched the reported values seen at the wt-rP2X 2 and substantially different from what has been observed for rP2X 1 (20). Thus, the effects of the TMD1 exchange were limited to the action of agonists at the protein.
The Effects of TMD1s from Other P2X Subunits on Agonist Potencies-It was important to assess whether the data ob-tained with the rP2X 1 -containing chimeras were due to rP2X 1specific sequences, or whether results were indicative of P2X receptor structure/function in general. We therefore constructed a chimera in which the TMD1 of rP2X 2 was replaced with that of the rP2X 3 subunit (this chimera is labeled TM1(r3)). rP2X 3 was chosen because it is the only other rat subunit at which ␣␤meATP has high affinity and potency. As seen in Fig. 4a, both ␣␤meATP and ATP again displayed appreciably increased potencies at TM1(r3) (ATP: EC 50 ϭ 0.39 M; ␣␤meATP:EC 50 ϭ 8.9 M) when compared with wt-P2X 2 .
To this point, results from the studies of agonists on the various chimeras suggested that the agonist pharmacology of a given receptor was determined, in large part, by the structure of its TMD1. If this were true, then replacement of the rP2X 2 TMD1 with a TMD1 from a different ␣␤meATP-insensitive subunit should not yield a chimeric protein with enhanced agonist sensitivities. To test this hypothesis, we chose to use the TMD1 from the zebrafish P2X 3 (zP2X 3 ) subunit as previous work from our labs has demonstrated that ␣␤meATP has much lower potency at zP2X 3 (EC 50 Ͼ 100 M) than at rP2X 3 , while having similar ATP sensitivity (EC 50 ϭ 1.5 M) (21). Therefore, a chimera, labeled TM1(z3), containing the TMD1 of the zP2X 3 subunit transplanted into rP2X 2 was constructed, and doseresponse curves for ATP and ␣␤meATP were determined. As was predicted from our working model, this construct exhibited ATP and ␣␤meATP sensitivities (Fig. 4b)  Dependence of Agonist Efficacy on TMD1 Amino Acid Composition-Although the TM1(r3) and TM1(z3) chimeras had quite different sensitivities to ␣␤meATP, their respective TMD1 domains differed by only five residues (Fig. 5a). Thus, a point mutation strategy was used in an effort to elucidate which of the amino acid differences provided the structural underpinnings of the pharmacological differences observed between the two chimeras. In this approach, the chimera TM1(z3) was used as the template in order that the end point (␣␤meATP sensitivity) would be a gain of function. We tested mutations beginning with the N-terminal portion of the domain to determine which one(s) has a role in enhancing agonist potency. This was done by comparing whole cell currents elicited by 30 and 100 M ␣␤meATP to that of 30 M ATP (a concentration that elicits maximal whole cell currents). As seen in Fig. 5, the I34A and L41S mutations did not show any increased sensitivity to ␣␤meATP. Mutant C45G was not functional (data not shown), FIG. 5. No single amino acid difference between the TMD1 s of the rat and zebrafish P2X 3 subunits accounts for the differences observed at the chimeras TM1(r3) and TM1(z3). a, alignment of the amino acid sequences present in the TMD1 regions of the rat (upper) and zebrafish (lower) P2X 3 subunits used to construct the chimeric receptors; boxed residues are identical; ␣␤meATP sensitivity is indicated by ϩ or Ϫ. b, currents evoked from cells expressing the indicated point mutations in the chimera TM1(z3). Traces shown are those induced by 30 M ATP (square), 30 M ␣␤meATP (inverted triangle), and 100 M ␣␤meATP (triangle), and are representative of those seen in at least three cells for each mutation. and therefore we made the double mutant I44V/C45G to compensate for the lethality of the single C45G mutation. This construct was functional but did not show an enhanced ␣␤meATP response either, thus effectively ruling out either residue as being essential for an enhanced sensitivity to agonists. Additionally, neither the double mutation I34A/L41S or the triple mutation L41S/I44V/C45G yielded increased sensitivity to low concentrations of ␣␤meATP. Thus, it did not appear that any one of the four N-terminal differences between the rat and zebrafish TMD1s was responsible for the increased potency of ␣␤meATP seen at the TM1(r3) chimera. The last single mutant, M49L, was found to have an increased ␣␤meATP sensitivity (also Fig. 5), although ␣␤meATP was not as potent as ATP. As this result suggested that the C-terminal portion of TMD1 might be important in altering agonist potencies, we made the triple mutant I44V/C45G/M49L to test whether we could get full conversion of the ␣␤meATP response to that seen in TM1(r3). This triple mutant gave currents resembling the single mutant M49L, and dose-response curves (Fig. 6) show that ␣␤meATP acted as a full agonist. The enhanced potency (EC 50 ϭ 155 M) seen at the triple mutant, while greater than the value obtained at the parent chimera was still less than that observed at TM1(r3). The EC 50 for ATP was not appreciably altered (1.5 M versus the 2.15 M seen at TM1(z3)). DISCUSSION In this study, we provide evidence that the variation in ATP and ␣␤meATP potencies exhibited at various P2X homo-oligomeric receptors is derived from the structure of the first transmembrane domain present in the monomer. By simply replacing the TMD1 of an ␣␤meATP-insensitive subunit with the equivalent domains of ␣␤meATP-sensitive subunits, it was possible to increase the potencies of both ␣␤meATP and ATP in the chimeric receptor. Additional site-directed mutagenesis experiments indicated that this effect was not the direct result of any single amino acid substitution, but rather the product of a cumulative effect upon the overall structure of TMD1 induced by the multiple amino acid differences seen across individual subunit primary sequences.
Two possible mechanisms could explain these data. The first would be that the agonist binding site conformation had changed as a result of the chimerism such that ATP and ␣␤meATP now interacted with higher affinity at the binding site present on the chimeras versus the wild type rP2X 2 . The second possibility would be that the transduction pathway linking agonist binding to opening of the channel gate had been altered in the chimeric protein such that a less precise fit of the agonist into the binding pocket could induce gating. As has been cogently argued (22), when performing mutational analyses of ligand-gated channels, it is extremely difficult to differentiate between mutational effects on agonist binding and those on channel opening, as both steps are dependent on, and affected by, one another. One approach to resolving this dilemma is to compare at least two agonists, differing in their efficacy, at each mutant receptor rather than comparing one mutant versus another using agonists of equal efficacy. This approach was used in our study, and, based on the findings, we do not favor the hypothesis that the observed increases in agonist potencies seen at a number of the chimeras resulted from changes only at the agonist binding site. Instead, the data provides a better fit to the hypothesis that the transduction properties of the receptor have been altered.
This interpretation is based on the following observations. First, ATP potency was shifted at the majority of the chimeras, whereas the ␣␤meATP response was increased only at a few constructs (summarized in Table I). Second, the ␣␤meATP response at 1.2.1 did not reach the maximum obtained with ATP, which suggests that it is acting as a partial agonist at this construct. However, at TM1(r1), TM1(r3), and the triple mutant TM1(z3) I44V/C45G/M49L, ␣␤meATP was as efficacious as ATP and thus was a full agonist at those constructs. Therefore, the findings from all the chimeras strongly imply that the effect of the chimerism on agonist potencies was mediated by an alteration in gating rather than a change in ligand binding. This argument is based on the logic that a mutation affecting FIG. 7. Alignment of the TMD1s from all seven rat P2X subunit sequences. Boxed residues are indicate residues that exhibit identity across the TMD1 at the 65% level. ␣␤meATP sensitivity is indicated by a ϩ or Ϫ.  binding only would be predicted to affect agonist potency but not alter efficacy at the mutant when compared with the parent receptor. On the other hand, if the gating efficiency (i.e. transduction) was affected, then different agonists could have differing efficacies at a mutant compared with the parent receptor due to intrinsic differences in their ability to induce channel gating; such a dependence of channel gating efficiencies on protein structure has been reported for other ligand-gated channels as well (23). A third finding was that the Hill slopes for ATP were shifted toward 1.0 for many of the chimeras (Table I), again a result that is easier to explain by a change in gating efficiency of a ligand rather than a simple change in binding affinity. Thus, the simplest interpretation of all of these results is that the outcome of transplanting the rP2X 1 or rP2X 3 TMD1 into rP2X 2 was an enhancement in the transduction, or gating, of the channel in response to occupancy of the agonist binding site rather than solely an increase in agonist affinity for the binding site. This interpretation does not rule out the possibility that agonist affinity was also altered; it just excludes it as a sole reason for the observed results. In light of the lack of a high degree of conservation in the primary sequences of the P2X 1 and P2X 3 TMD1s, and across P2X subunits in general (Fig. 7), it would appear that the overall topological shape of TMD1, rather than a specific amino acid or two, is an important determinant of the agonist-induced gating behavior of the P2X receptor channel. The findings presented here, when integrated with previous findings from our labs (8,16) and others (12,13), now provide a preliminary conceptual mechanism for understanding how ligand gating of the P2X receptor occurs, with TMD1 playing the role of a "trigger" in actuating channel gating. In this model, interaction of ATP with extracellular amino acid residues that are just C-terminal of TMD1 (and perhaps with other extracellular residues as well) would result in perturbations of structure that would be transmitted to the immediately adjacent proximal TMD1 region. This disturbance of TMD1 would then induce a local structural change that would influence TMD2, causing it to move in such a way as to allow opening of the gate and subsequent ion permeation. If this model is accurate, then it could be predicted that replacement of rP2X 2 -TMD2 with the TMD2 of rP2X 1 should also yield a receptor (TM2(r1)) with altered agonist properties. As seen in Fig. 8, this is what was observed as both ATP and ␣␤meATP exhibited increased potencies (EC 50 values of 0.72 and 102 M, respectively) at this chimera. Interestingly, although ␣␤meATP is a full agonist at TM1(r1), it is only a partial agonist at TM2(r1), showing that the effect of an exchange of TMD2 on agonistinduced gating is not as robust as one of TMD1. It is tempting to speculate that this is because TMD2 has a less direct role in the transduction mechanism. Regardless, this result is in keeping with the hypothesis that TMD1 and TMD2 interact in the membrane, and that this interaction plays a role in the gating efficiencies of agonists at the protein.