Abstract
Lysine residues near each end of the receptor ectodomain (in rat P2X2 Lys69 and Lys308) have been implicated in ATP binding to P2X receptors. We recorded membrane currents from human embryonic kidney cells expressing P2X subunits and found that lysine-to-alanine substitutions at equivalent positions in the P2X3 receptor (Lys63 and Lys299) also prevented channel function. Heteromeric P2X2/3 receptors are formed when P2X2 and P2X3 subunits are expressed together; they can be distinguished by their relatively sustained response to αβ-methylene-ATP. By coexpression of wild-type P2X3 and mutated P2X2 subunit, we found that the heteromeric P2X2/3 channel functioned normally when either lysine in the P2X2 subunit was mutated to alanine (i.e., [K69A] or [K308A]) but not when both lysines were mutated to alanine (i.e., [K69A, K308A]). However, coexpression of wild-type P2X2 with a mutated P2X3 subunit ([K68A] or [K299A]) produced no functional heteromers. The rescue of the single lysine mutant P2X2 subunit by wild-type P2X3 (but not the converse) suggests that the heteromeric channel contains one P2X2 and two P2X3 subunits and that the receptor functions essentially normally as long as two subunits are not mutated. The failure to rescue function in the P2X2 subunit with both lysines mutated by wild-type P2X3 suggests that these residues from two different subunits interact in agonist binding or channel opening.
The heteromeric P2X2/3 receptor is of interest because of its predominant expression on a subset of primary afferent nerves involved in the sensation of chronic noxious damage (Jarvis, 2003), visceral distension (Vlaskovska et al., 2001), hypoxia (Gourine, 2005), and taste (Finger et al., 2005). A key role for these receptor subunits is supported by experiments using P2X3 gene knockouts (Cockayne et al., 2000, 2005), P2X3 RNA suppression (Barclay et al., 2002; Honoré et al., 2002), and pharmacological antagonists selective for P2X3 subunit-containing receptors (Jarvis et al., 2002).
There is good evidence that P2X subunits form channels as trimers (Nicke et al., 1998; North, 2002; Jiang et al., 2003; Barrera et al., 2005). Like other ligand-gated channels, P2X receptors can form by homo- or hetero-oligomeric assembly of subunits (for review, see North, 2002). Although both P2X2 and P2X3 subunits are able to form homomeric channels, the heteromeric P2X2/3 receptor that is also formed can be distinguished by its unique functional properties (Lewis et al., 1995). P2X2 receptors are activated by ATP but not by the analog αβ-methylene-ATP (αβmeATP), and the current is largely sustained through a 2-s agonist application. P2X3 receptors are activated by both ATP and αβmeATP, and the current desensitizes very rapidly (<100 ms). The current through heteromeric P2X2/3 channels can be identified as the sustained component when αβmeATP is the agonist (Lewis et al., 1995).
The molecular operation of P2X subunits is not well understood. Each receptor subunit has intracellular amino and carboxyl termini and two transmembrane domains (TM1: approximately residues 30-50; TM2: approximately residues 330-354) that are joined by an ectodomain of approximately 280 residues. The ectodomain is glycosylated and disulfide-bonded. There are no obvious “canonical” ATP-binding amino acid sequences in the ectodomain. In the case of the P2X2 homomeric receptor, two conserved ectodomain lysine residues have been suggested to play a role in binding the negatively charged phosphate moiety of ATP (Lys69 and Lys308 in rat P2X receptor), but the evidence is most complete for Lys69 (Jiang et al., 2000). A similar role for the corresponding lysines has also been identified in the case of the P2X1 receptor (Lys68 and Lys309; Ennion et al., 2000). These lysines are situated approximately 20 amino acids from the outer end of each TM. Experiments on the P2X1 receptor using partial agonists also suggest a role for aromatic residues Phe185 and Phe291 to coordinate the binding of the adenine ring of ATP (Roberts and Evans, 2004; Vial et al., 2004). An alternative model of ATP binding has been proposed by Yan et al. (2005) in the case of the P2X4 receptor. This is based on secondary structure homology between part of the receptor (Lys180-Lys326) and class II aminoacyl-tRNA synthetases (Freist et al., 1998).
The experiments reported herein focused on the two juxtamembrane lysines proposed to take part in ATP binding. We have found that, for the heteromeric P2X2/3 channel, their importance in channel function depends greatly on whether they are in the P2X2 or the P2X3 subunit.
Materials and Methods
Molecular and Cell Biology. Rat P2X receptor cDNAs were used as described previously (Jiang et al., 2000). Mutations were introduced using QuikChange site-directed mutagenesis (Stratagene, La Jolla CA) and confirmed by sequencing. Wild-type and mutated receptors were transiently expressed in human embryonic kidney 293 cells using Lipofectamine 2000 (Invitrogen, San Diego CA) according to the manufacturer's instructions. Transfected cells were harvested and replated on glass coverslips 24 to 48 h after transfection and kept at 37°C for 18 to 48 h before electrophysiological recording. P2X2 plasmid (0.2 μg) was cotransfected with P2X3 plasmid (1 μg), these ratios having been established to produce equal amounts of protein when detected by Western blot with a common C-terminal epitope tag (Jiang et al., 2003). Enhanced green fluorescent protein cDNA (0.1 μg) was also included. Immunocytochemistry of transfected cells was as described, using a C-terminal EYMPME epitope (Jiang et al., 2000, 2003).
Electrophysiology. Whole-cell patch-clamp recordings were made at 19 to 22°C using a HEKA EPC9 amplifier (Heka, Lambrecht, Germany). The holding potential was -60 mV. Patch pipettes (3-6 MΩ) contained 145 mM NaCl, 10 mM HEPES, and 10 mM EGTA, pH adjusted to 7.3 with NaOH. The external solution contained 147 mM NaC1, 2 mM KC1, 2 mM CaC12, 2 mM MgCl2,10 mM HEPES, and 13 mM glucose, pH adjusted to 7.3 with NaOH. Both internal and external solutions were 290 to 310 mOsM. Agonists were applied using the RSC 200 rapid solution changer (Biologic Science Instruments, Grenoble, France), with flow-pipe tips approximately 150 μm from cell. All chemicals were purchased from Sigma (Poole, UK) and were made up daily from aliquots stored at -20°C.
Results
Homomeric Channels. Previous work on homomeric human P2X1 (Ennion et al., 2000) and rat P2X2 (Jiang et al., 2000) receptors has indicated that Lys69 and Lys308 (rat P2X2 numbering) are essential for channel function. We found that alanine substitution at the equivalent positions in the P2X3 subunit also led to a nonfunctional homomeric channel. We observed no current with applications of αβmeATP (up to 1 mM; n = 5) (Fig. 1) or ATP (up to 1 mM; n = 7). Immunocytochemistry of transfected cells showed no difference in the pattern or distribution between wild-type and mutated subunits. Figure 1 also shows that equivalent mutations in other rat P2X receptors also prevented any response to agonist [P2X1[K68A]: ATP (300 μM) 23 ± 9 pA, n = 8; αβmeATP (300 mM) no current, n = 3; P2X4[K67A]: ATP (1 mM) no current, n = 8; P2X7[K64A]: ATP (3 mM) no current, n = 3].
Neither P2X2[K69A] nor P2X2[K308A] gave any detectable current in response to 1 or even 3 mM ATP (Fig. 1). However, coexpression of these two mutated receptors clearly did (Fig. 2). The currents were small compared with wild-type P2X2 currents (>600 pA/pF) but not obviously different in other properties (Fig. 2). It was striking that coexpression of two “dead” subunits resulted in formation of a channel that was responsive to 100 or 300 μM ATP. If we assume that neither mutation affects expression or assembly, then one would expect three eighths of all trimeric channels to be [K69A]2·[K308A] and three-eighths to be [K69A]·[K308A]2. This result therefore indicates that the effects of these two point mutations are not independent.
Heteromeric Channels. For the P2X2/3 receptors, there was a striking difference in the effect of a single point mutation of either of the lysine residues, depending whether it was present in the P2X2 or P2X3 subunit. The currents in cells expressing mutant P2X2 receptors with wild-type P2X3 subunits were essentially the same as those observed in wild-type heteromeric receptors (Fig. 3). The αβmeATP concentrations giving half-maximal currents (EC50) were: for P2X2/3 wild-type, 27 ± 3.3 μM (n = 5); for P2X2[K69A] + P2X3, 22 ± 2.6 μM (n = 5), and for P2X2[K308A] + P2X3, 44 ± 2.6 μM (n = 8) (the last two values are significantly different, p < 0.0001). In contrast, cells expressing wild-type P2X2 with P2X3[K63A] subunits showed no currents in response to αβmeATP up to 100 μM. At such high concentrations, P2X2 homomeric receptors are sensitive to αβmeATP (n = 5) (Fig. 3) (see also Jiang et al., 2003), and the small current observed was presumably passing through such receptors. Likewise, coexpression of wild-type P2X2 subunits with P2X3 [K299A] gave no currents in response to αβmeATP (up to 100 μM, n = 6) (Fig. 3). These results indicate that a wild-type P2X3 subunit can rescue function when it enters a heteromer with a mutated P2X2 subunit, but not vice versa.
This rescue did not occur when both ectodomain lysines were mutated in the same P2X2 subunit. Coexpression of P2X2 [K69A,K308A] with P2X3 produced no sustained currents to αβmeATP (n = 9) (Fig. 3). However, a fast desensitizing current typical of homomeric P2X3 receptors was usually observed (Fig. 3). A schematic summary of the expression of the further P2X2 and P2X3 subunits (either wild-type or mutant) that were coexpressed, and the results of the functional studies, is shown in Fig. 4.
Discussion
The two lysines that are the focus of the present study are very highly conserved among P2X subunits. Lys69 is present in 41 of 43 vertebrate sequences (including seven human, seven rat, seven mouse, one guinea pig, three chick, two Xenopus laevis, one bullfrog, nine zebrafish, and six pufferfish); the zebrafish P2X2 receptor has Asp at this position, and it cannot be activated by ATP when heterologously expressed (Kucenas et al., 2003). This lysine is also present in the one invertebrate (schistosome) sequence that is known to operate as an ATP-gated channel (Agboh et al., 2004). Lys308 is even more highly conserved; it is found in all the above 44 sequences and in two of the five related Dictyostelium discoideum sequences. Of the eight positively charged residues in the P2X2 receptor ectodomain, Lys69 and Lys308 are the two where replacement by alanine causes the greatest loss of channel function; even replacement by arginine results in more than 100-fold reduction in sensitivity to ATP (Jiang et al., 2000). The present experiments confirm that the lysine in the first of these positions is critical also for function of other P2X receptors, including P2X3.
The heterotrimeric P2X2/3 receptor must contain either one or two copies of the P2X3 receptor subunit. The simplest interpretation of the one-way rescue of P2X2 mutants by P2X3 (but not vice versa; Fig. 3), is that the P2X2/3 receptors activated by αβmeATP have two copies of P2X3 and one copy of P2X2. In other words, a functional channel can be formed and activated by αβmeATP even though it contains one dead P2X2 subunit (e.g., P2X2[K69A]) but no opening can be elicited from a receptor comprising two dead P2X3 subunits with a wild-type P2X2 subunit. The rescue of wild-type P2X3 was complete in the case of P2X2[K69A] and slightly less so for P2X2[K308A] (Fig. 3), indicating that the two P2X2 lysines do not have precisely equivalent roles.
This result contrasts with earlier findings in which a highly conserved intracellular C-terminal lysine was mutated (Chaumont et al., 2004). This lysine is required in homomeric P2X2 and P2X3 receptors for retention in the plasma membrane. In that case, wild-type P2X3 subunits could restore the appearance and retention in the membrane of mutant P2X2 subunits (P2X2[K366A]), and this rescue was fully reciprocal because wild-type P2X2 also restored the appearance of P2X3[K357A]) (Chaumont et al., 2004). Taken together, these results imply that only a single wild-type subunit is needed for successful retention in the membrane, but two wild-type (P2X3) subunits are needed for channel function. The conclusion that the functional heteromer contains two P2X3 subunits and one P2X2 subunit agrees well with the results of a completely independent approach in which subunits were joined by disulfide bonds between engineered cysteine substitutions at the outer ends of the TMs (Jiang et al., 2003). The finding is also consistent with the observation that the limiting slope of the Hill plot for activation of the heteromeric P2X2/3 receptor is close to two, whereas for the homomeric P2X3 and homomeric P2X2 receptors, it is close to three (Jiang et al., 2003). However, the present analysis can not distinguish between different numbers of ligand binding sites and different degrees of intersubunit cooperativity in the heteromeric versus the homomeric channels.
For the homomeric P2X1 and P2X2 receptors, there is evidence that Lys69 and Lys308 contribute to an ATP binding site (Ennion et al., 2000; Jiang et al., 2000; Roberts and Evans, 2004). Therefore, our present finding that the heteromeric channel operates normally even though it incorporates a mutated P2X2 subunit may be interpreted in this context. This would imply that a heterotrimeric P2X receptor could open from a less than fully liganded state. For example, the tetrameric α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (Rosenmund et al., 1998) and the pentameric glycine receptor (Beato et al., 2002) can open from less than fully liganded states. But other interpretations are possible. For example, the P2X3 (though not P2X2) subunit lysines may be critical for multimeric channel assembly. Or the mutation might impede the conformational change of gating: a lysine-to-alanine substitution in a single P2X2 subunit of the trimer might provide insufficient energetic impediment, whereas two such independent substitutions in the P2X3 subunits might prevent it completely.
We were surprised to find that cells cotransfected with P2X2[K69A,K308A] subunit and wild-type P2X3 subunit produced no detectable heteromeric current (Fig. 3). In other words, the wild-type P2X3 subunits can rescue function in a channel containing a P2X subunit with a single mutation ([K69A] or [K308A]) (a dead P2X2) but not in a channel containing both those mutations (a “double-dead” P2X2). This could have a simple energetic explanation of the type introduced above; namely, both P2X3 subunits become liganded by αβmeATP, but the mutations provide too great an energy barrier for opening to occur. Or it could indicate that the effects of the lysines are not independent and that an interaction occurs between Lys69 and Lys308. This would be consistent with the finding that in homomeric P2X2 receptors, ATP can elicit currents when P2X2[K69A] and P2X2[K308A] are coexpressed, but not when either is expressed alone.
If one assumes that at least two agonist binding sites are required for channel activation (see above, and Ding and Sachs, 1999; Jiang et al., 2003), then these observations can be explained if both Lys69 of one subunit and Lys308 of another (or equivalent for the P2X3 subunits) are needed at one ATP binding site. The presence of a dead P2X2 subunit would still provide for two αβmeATP binding sites (Fig. 4B, left and middle), but a double-dead P2X2 subunit would prevent αβmeATP binding at two sites and therefore prevent channel function (Fig. 4B, right). The interpretation that the agonist site is contributed from two different subunits in a heteromeric channel has parallels with nicotinic acetylcholine receptors (Sine, 2002). Moreover, P2X2 subunits carrying mutations of an allosteric zinc binding site have been covalently joined in tandem from concatenated cDNAs; these experiments indicated that histidines from different subunits contribute to the zinc binding site (Nagaya et al., 2005).
The conclusion that the αβmeATP binding sites forms at the P2X3/P2X3 and P2X2/P2X3 interfaces is consistent with the observations that both nucleotide (Virginio et al., 1998; Burgard et al., 2000) and non-nucleotide antagonists (Jarvis et al., 2002, 2004) mostly fail to discriminate between P2X3 homomeric receptors and P2X2/3 heteromeric receptors. A more complete understanding of the agonist binding site will inform the further development of small molecules that act as competitive antagonists at the P2X2/3 heteromeric receptors; such molecules are likely to have value as pain-relieving drugs.
Acknowledgments
We thank Helen Broomhead, Louise Almond, and Sandy Dossi for their technical assistance.
Footnotes
-
The work was supported by the Wellcome Trust
-
ABBREVIATIONS: αβmeATP, αβ-methylene-ATP
-
↵1 Current affiliation: Faculty of Biological Sciences, University of Leeds, Leeds, United Kingdom.
- Received May 15, 2006.
- Accepted July 13, 2006.
- The American Society for Pharmacology and Experimental Therapeutics