The P2X7 Carboxyl Tail Is a Regulatory Module of P2X7 Receptor Channel Activity*

P2X7 receptors are ATP-gated cation channels composed of three identical subunits, each having intracellular amino and carboxyl termini and two transmembrane segments connected by a large ectodomain. Within the P2X family, P2X7 subunits are unique in possessing an extended carboxyl tail. We expressed the human P2X7 subunit as two complementary fragments, a carboxyl tail-truncated receptor channel core (residues 1-436 or 1-505) and a tail extension (residues 434-595) in Xenopus laevis oocytes. P2X7 channel core subunits efficiently assembled as homotrimers that appeared abundantly at the oocyte surface, yet produced only ∼5% of the full-length P2X7 receptor current. Co-assembly of channel core subunits with full-length P2X7 subunits inhibited channel current, indicating that the lack of a single carboxyl tail domain is dominant-negative for P2X7 receptor activity. Co-expression of the tail extension as a discrete protein increased ATP-gated current amplitudes of P2X7 channel cores 10-20-fold, fully reconstituting the wild type electrophysiological phenotype of the P2X7 receptor. Chemical cross-linking revealed that the discrete tail extension bound with unity stoichiometry to the carboxyl tail of the P2X7 channel core. We conclude that a non-covalent association of crucial functional importance exists between the carboxyl tail of the channel core and the tail extension. Using a slightly shorter P2X7 subunit core and subfragments of the tail extension, this association could be narrowed down to include residues 409-436 and 434-494 of the split receptor. Together, these results identify the tail extension as a regulatory gating module, potentially making P2X7 channel gating sensitive to intracellular regulation.

P2X 7 receptors are ATP-gated cation channels composed of three identical subunits, each having intracellular amino and carboxyl termini and two transmembrane segments connected by a large ectodomain. Within the P2X family, P2X 7 subunits are unique in possessing an extended carboxyl tail. We expressed the human P2X 7 subunit as two complementary fragments, a carboxyl tail-truncated receptor channel core (residues 1-436 or 1-505) and a tail extension (residues 434 -595) in Xenopus laevis oocytes. P2X 7 channel core subunits efficiently assembled as homotrimers that appeared abundantly at the oocyte surface, yet produced only ϳ5% of the full-length P2X 7 receptor current. Co-assembly of channel core subunits with full-length P2X 7 subunits inhibited channel current, indicating that the lack of a single carboxyl tail domain is dominant-negative for P2X 7 receptor activity. Co-expression of the tail extension as a discrete protein increased ATPgated current amplitudes of P2X 7 channel cores 10 -20-fold, fully reconstituting the wild type electrophysiological phenotype of the P2X 7 receptor. Chemical cross-linking revealed that the discrete tail extension bound with unity stoichiometry to the carboxyl tail of the P2X 7 channel core. We conclude that a non-covalent association of crucial functional importance exists between the carboxyl tail of the channel core and the tail extension. Using a slightly shorter P2X 7 subunit core and subfragments of the tail extension, this association could be narrowed down to include residues 409 -436 and 434 -494 of the split receptor. Together, these results identify the tail extension as a regulatory gating module, potentially making P2X 7 channel gating sensitive to intracellular regulation. The P2X 7 receptor, an ATP-gated cation channel, is expressed predominantly in immune cells (such as macrophages and lymphocytes), glial cells, and epithelial cells (for recent reviews, see Refs. 1 and 2). Activation of the P2X 7 recep-tor has been implicated in pivotal inflammatory responses resulting from ATP-stimulated pro-inflammatory cytokine release (particularly of interleukin-1␤ and interleukin-18) through exosomes (3) during cell proliferation and apoptosis (1,2). These functions of the P2X 7 receptor have been attributed to its unusual dual role as a classic ligand-gated channel for small cations and as a cytolytic pore (4,5).
Cloning of the P2X 7 receptor revealed a typical P2X subunit "core" structure, with a short cytoplasmic NH 2 -terminal tail and two transmembrane segments connected by a large N-glycosylated ectodomain. However, the P2X 7 receptor has a carboxyl tail extension that is 120 -200 amino acids longer than that of the other six P2X family members, P2X 1 -P2X 6 (4,5). Because the cytolytic pore-forming ability of P2X 7 is not shared by the other P2X receptor subtypes, this function has been plausibly assigned to the long carboxyl tail. Indeed, truncation of most of the extra portion of the carboxyl tail of P2X 7 prevented cytolytic pore formation, apparently without affecting the function of the receptor as an ATP-gated channel for small cations (4,5). Additional functions distinct from cytolytic pore formation that have been assigned to the carboxyl tail include interactions with adaptor and effector proteins (6,7), intracellular binding of lipopolysaccharide (8,9), and regulation of trafficking of the P2X 7 receptor to the plasma membrane (10,11).
Conflicting data have been published as to whether the cytolytic pore is formed by the P2X 7 receptor itself through progressive dilatation of the cation-conducting pore from 7 Å to up to 40 Å (12), or secondarily by the opening of a distinct accessory channel (13)(14)(15). Support for the view that P2X 7 receptor-induced permeability increases must be secondary to P2X 7 receptor activation was provided by single-channel analysis of P2X 7 receptor kinetics, which showed that pore dilatation observed in macroscopic current recordings had no molecular correlate at the single P2X 7 channel level (16,17). The P2X 7 receptor subunit might be regarded as consisting of two structurally and functionally distinct parts: an NH 2 -terminal receptor core (comprising roughly two-thirds of the polypeptide chain) sufficient for ligand-gated channel opening and cation permeation, and a carboxyl tail (the remaining onethird of the polypeptide) harboring topogenic information and a series of protein recruitment domains, but dispensable for channel gating and cation permeation. Because we previously observed that COOH-terminal truncation of the human P2X 7 (hP2X 7 ) subunit at a position (residue 436) expected to allow for efficient surface expression (10) greatly reduced the current amplitudes compared with the wild type hP2X 7 receptor (18), we asked whether the carboxyl tail contributes in an unknown manner to P2X 7 receptor channel function. Here, we addressed this issue by co-expressing the truncated tail portion and the carboxyl tail-truncated hP2X 7 receptor core as discrete proteins and testing for functional restoration. We show the ability of the truncated tail portion to fully restore the wild type electrophysiological phenotype by physically interacting with the carboxyl tail of the hP2X 7 receptor core. These results define a novel function of the carboxyl tail as a modulator of the gating of the hP2X 7 receptor.

EXPERIMENTAL PROCEDURES
Chemicals-Molecular biology enzymes were purchased from New England Biolabs (Schwalbach, Germany). ATP (sodium salt) was purchased from Roche (Mannheim, Germany). Unless otherwise specified in the text below, all other chemicals were obtained from Sigma.
Protein Expression in Xenopus laevis Oocytes-Capped cRNA was synthesized from linearized templates with SP6 RNA polymerase (Epicenter Biotechnologies, Madison, WI). It was then purified and quantified based on its absorbance at 260 nm. Fully defolliculated stage V-VI X. laevis oocytes were obtained by collagenase treatment as previously described (16). The cRNAs were injected singly or in various pairwise combinations at 1 or 50 ng/oocyte for patch clamping or two-electrode voltage clamping, respectively. Until use 1-3 days later, the oocytes were maintained at 19°C in a modified Barth solution containing the following (in mM): 100 NaCl, 1 KCl, 1 CaCl 2 , 1 MgCl 2 , and 5 or 10 Hepes-NaOH, pH 7.4, supplemented with 100 units/ml penicillin and 100 g/ml streptomycin or 50 g/ml gentamycin.
Two-electrode Voltage-clamp and Single-channel Recordings-One to 3 days after cRNA injection, membrane currents were recorded by two-electrode voltage clamping at Ϸ22°C and a holding potential of Ϫ40 mV, exactly as described previously (18). Single hP2X 7 channel recordings were performed on excised outsideout patches from X. laevis oocytes, and analyzed as previously described (16). Data are given as mean Ϯ S.D. of n measurements if not otherwise stated. The statistical significance (p Ͻ 0.05) of the differences between means was determined by one-way analysis of variance, followed by a Bonferroni multiple comparison t test using Jandel Sigmastat statistical software (SPSS, Chicago, IL). The Sigmaplot program (SPSS) was used for function fitting and graphical representation of the data.
Metabolic After an additional 24-or 48-h chase period, plasma membrane-bound proteins were selectively labeled using the amino-reactive fluorescent dye Cy5 N-hydroxysuccimide (NHS) ester (GE Healthcare), which is membrane-impermeant due to its two sulfonic acid groups. Shortly before Cy5 dye addition, oocytes were washed in oocyte-phosphatebuffered saline at pH 8.5 (in mM: 20 sodium phosphate, 110 NaCl, 1 MgCl 2 ), and then incubated for 30 min at ambient temperature (21-24°C) with the Cy5 dye, which was diluted 200-fold from a dimethyl sulfoxide stock solution to a final concentration of 50 g/ml. The reaction was terminated by washing the cells with oocyte-phosphate-buffered saline, followed by membrane protein extraction with digitonin and receptor purification (see below). In some experiments, hP2X 7 receptor constructs at the plasma membrane were selectively labeled with 125 I-labeled sulfosuccinimidyl-3-(4-hydroxyphenyl)propionate exactly as described previously (21,22).
Chemical Cross-linking-Oocytes were lysed in 0.1 M sodium phosphate buffer, pH 8.0, supplemented with 1% digitonin and one of two reversible, homobifunctional cross-linkers (Pierce): the membrane-impermeable NHS ester DTSSP (3,3Ј-dithiobis(sulfosuccinimidylpropionate)) or its membrane-permeable analogue DSP (dithiobis(succinimidylpropionate)). DTSSP and DSP were dissolved just before use in sodium citrate buffer, pH 5.0, or dry dimethyl sulfoxide, respectively, and diluted to the desired concentration in 0.1 M sodium phosphate buffer, 1% digitonin. The cross-linking reaction was initiated by adding this reaction mixture to the cells and immediately lysing the cells by passing them through a 200-l pipette tip. After 30 min of incubation at 21°C, excess cross-linker was quenched by the addition of Tris/HCl, pH 7.5, to a final concentration of 50 mM. A digitonin extract was then prepared from the cells, from which hP2X 7 receptor proteins were purified by affinity chromatography as detailed below. 5 The abbreviations used are: EGFP, enhanced green fluorescent protein; DTSSP, 3,3Ј-dithiobis(sulfosuccinimidylpropionate); DSP, dithiobis(succinimidylpropionate); NHS, N-hydroxysuccimide; Ni-NTA, nickel-nitrilotriacetic acid.
Affinity Purification and PAGE-Proteins from digitonin (1.0%) extracts of oocytes were purified by Ni-NTA-agarose (Qiagen, Hilden, Germany) or by Strep-Tactin TM Sepharose (IBA, Göttingen, Germany) affinity chromatography. Proteins were eluted from Ni-NTA-agarose or Strep-Tactin Sepharose with the appropriate non-denaturing elution buffer consisting of 0.5% digitonin and 250 mM imidazole/HCl, pH 7.4, or 2.5 mM D-desthiobiotin in 100 mM Tris/HCl, pH 8.0, 150 mM NaCl, and 1 mM EDTA, respectively. Blue native PAGE (23,24) was carried out on the same day as purification using gradient gels (4 -20% acrylamide) as described previously (21,22,25). For partial dissociation of natively purified hP2X 7 receptor complexes into lower-order oligomers (down to single polypeptides), samples were treated for 1 h at 37°C with 0.1% SDS with or without 0.1 M dithiothreitol as indicated before loading onto the blue native PAGE gel.
For SDS-PAGE, proteins were supplemented with the appropriate SDS-PAGE sample buffer containing 20 or 50 mM dithiothreitol, followed by heating to 56°C for 15 min. Denatured samples were electrophoresed in parallel with 14 C-labeled molecular mass markers (Rainbow TM , GE Healthcare) and blue-stained mass markers (Precision Plus Protein All Blue Standard, Bio-Rad). Cy5-labeled proteins were visualized by scanning of the wet SDS-PAGE gel with a fluorescence scanner (Typhoon, GE Healthcare). For subsequent visualization of radiolabeled proteins, the SDS-PAGE gel was fixed, dried, exposed to a PhosphorImager screen and scanned using a Storm 820 PhosphorImager (GE Healthcare). Individual bands were quantified with ImageQuant software. All biochemical experiments were performed at least in triplicate. Fig. 1, B-F, shows typical traces of currents induced by 1 mM free ATP (ATP 4Ϫ ) and recorded by the two-electrode voltage-clamp method from intact X. laevis oocytes expressing the indicated hP2X 7 constructs. The ATPinduced wild type current consists of two components, an exponentially saturating current a (see labeling in Fig. 1B) and a linearly increasing current b that are directly hP2X 7 receptormediated and, evidently, secondary to activation of oocyte-endogenous channels, respectively (18). Based on concentrationresponse analysis, it is known that high-affinity and low-affinity sites for ATP 4Ϫ (K D values of ϳ4 and ϳ220 M) are responsible for ϳ10 and ϳ90%, respectively, of the exponentially saturating current a (18). The deactivating current may also be separated into two distinct components characterized by fast and slow exponential decay (c and d in Fig. 1B).

Macroscopic hP2X 7 Receptor Current Kinetics Are Strongly Affected by Truncation, but Are Restored by Co-expression of the Truncated Carboxyl Tail-
Truncation of the hP2X 7 receptor at position 436 strongly diminished the ATP-induced current amplitude, causing a leftward shift in the concentration-response curve for ATP (Fig.  1N), and abolishing the exponentially saturating current a as well as the linearly increasing and the fast deactivating current components (Fig. 1C). All of these effects may be explained kinetically by a selective functional loss of ATP activation at the low-affinity site; the leftward shift of the concentration-response curve results from the unchanged current activation of the high-affinity site (18). Remarkably, co-expression of the hP2X 7 1-436 receptor core with the hP2X 7 434 -595 carboxyl tail as discrete polypeptides increased the ATP-induced current amplitude ϳ15-fold, reaching a value near that of the non-split full-length hP2X 7 (Fig. 1, D, G, and H). Similar results were obtained for a mutant hP2X 7 truncated at position 505, hP2X 7 1-505 (Fig. 1, E and F). Two observations indicate that this stimulating effect specifically involves the carboxyl tail-truncated hP2X 7 receptor cores: (i) expression of the hP2X 7 434 -595 tail alone did not lead to ATP-activated currents (data not shown), and (ii) currents mediated by the wild type hP2X 7 receptors were not significantly affected by co-expression of the carboxyl tail (cf. Fig. 5D, left two bars).
To provide a quantitative basis for this observation, the activating part of the hP2X 7 receptor current (I act (t)) during ATP 4Ϫ application was fitted according to the following equation.
In other words, the current time course was approximated by the sum of an exponentially saturating and a linearly increasing component (18), where I act,∞ is the amplitude of the exponentially saturating current a (cf. Fig. 1, B-F) after an infinite time of agonist application, act is the activation time constant, s is the slope of the linearly rising current, and I 0 is the steady-state current without ATP 4Ϫ application.
The best approximation of the deactivating current (I deact (t)) during washout of ATP 4Ϫ was achieved using a bi-exponentially decaying function, where I 0 has the same meaning as in Equation 1, I deact,1 and I deact,2 are the initial amplitudes, and deact,1 and deact,2 are the time constants of the slow and fast deactivating component, respectively. The loss of both the linearly increasing and the fast decaying current component upon COOH-terminal truncation of the hP2X 7 receptor at position F436 is substantiated by the statistical analysis ( Fig. 1, G-M).
The current amplitude I act,6s elicited by 6-s applications of different concentrations of ATP 4Ϫ ([ATP 4Ϫ ]) (see Fig. 1B) was used to define the ATP 4Ϫ concentration dependence of the full-length and split hP2X 7 receptors. To account for cell-tocell variation in receptor expression, the activating current I act,6s ([ATP 4Ϫ ]) was normalized to the current measured at 1 mM ATP 4Ϫ . The concentration-response curves of the calculated relative amplitudes I rel ([ATP 4Ϫ ]) of the full-length hP2X 7 receptor and the co-expressed split hP2X 7 constructs were fitted to a model of two equal high-affinity and two equal lowaffinity non-cooperative activating sites (18) (Fig. 1N) as shown, where I rel,∞,1 and I rel,∞,2 are the maximal relative current compo-nents contributing to I rel ([ATP 4Ϫ ]) after saturating of the effector sites at infinite agonist concentrations with apparent dissociation constants K D,1 and K D,2 , respectively. A coefficient of 2 yielded higher correlation coefficients than models having one, three, or more than three equal effector sites. This model fitted the data significantly better (26) than a simpler model assuming only one effector site (I rel,∞,2 ϭ 0). The calculated pK D values for the highaffinity and low-affinity ATP 4Ϫ effector sites were not significantly different from the full-length hP2X 7 receptor (pK D,1 ϭ Ϫ5.5 Ϯ 0.4, pK D,2 ϭ Ϫ3.6 Ϯ 0.2, mean Ϯ S.E.) or the hP2X 7 1-436 truncation mutant co-expressed with the hP2X 7 434 -595 carboxyl tail (pK D,1 ϭ Ϫ5.6 Ϯ 0.8, pK D,2 ϭ Ϫ3.9 Ϯ 0.3). Curve-fitting further indicated that the high-affinity and low-affinity effector sites accounted, in both cases, for ϳ10 and ϳ90% of the ATP-induced current, respectively. The concentration-response curve for the truncation mutant alone could best be described by a function with only one kind of effector site (I rel,∞,2 ϭ 0) displaying high-affinity for ATP 4Ϫ . The calculated pK D,1 ϭ Ϫ5.4 Ϯ 0.1 was not significantly different from that of the apparent high-affinity effector site of the wild type hP2X 7 receptor or the hP2X 7 1-436 receptor core co-expressed with the hP2X 7 434 -595 tail domain. The apparent leftward shift of the concentration-response curve is fully compatible with the view that activation of the hP2X 7 1-436 receptor core is mediated only by the highaffinity effector site and that the gating effect of the low-affinity site is abolished by the truncation. The more right-handed position of the concentration-response curve for the (intact or split) fulllength P2X 7 receptor is predominantly determined by the low-affinity effector site, which accounts for ϳ9-fold more current than the high-affinity site.
The marked changes in the ATP 4Ϫ -gated membrane currents might result from a reduction in FIGURE 1. Co-expressed hP2X 7 carboxyl tail profoundly changes electrophysiological characteristics of the carboxyl-truncated hP2X 7 receptors, hP2X 7 1-436 and hP2X 7 1-505 . A, schemiatic of hP2X 7 constructs. The carboxyl tail-truncated hP2X 7 subunit core (designated hP2X 7 1-436 ) and the separately expressed carboxyl tail (hP2X 7 434 -595 ) are indicated in black and blue, respectively. The full-length hP2X 7 subunit consists of amino acids 1-595 and includes the hP2X 7 channel subunit core (demarcated by dashed lines) and the gray-colored carboxyl tail. Numbers beside the red lines indicate the amino acid sites of the truncations. B-F, representative current traces elicited by 1 mM ATP 4Ϫ in oocytes expressing the indicated hP2X 7 receptor construct(s). The exponentially saturating current component mediated directly by the hP2X 7 receptor is marked by a. Please note the different ordinate scales. In B-D, activating and deactivating current traces induced by ATP (horizontal line) and subsequent washout were approximated (gray lines) by Equations 1 and 2, respectively. G-M, for a statistical comparison, current traces (as shown in B-F) from 5 to 23 oocytes were fitted by Equations 1 and 2 to assess the following electrophysiological parameters: G, amplitude of the exponentially saturating current; H, relationship of the linearly increasing to the exponentially saturating current; I, time constant of the exponentially saturating current; K, relative amplitudes of fast deactivating currents (with I deact ϭ I deact,1 ϩ I deact,2 ); L, time constant of fast deactivating current; M, time constant of slow deactivating current. A significant difference from the full-length hP2X 7 receptor or the truncated hP2X 7 receptor is indicated by * or #, respectively. N, ATP 4Ϫ concentrationresponse curves. Currents were elicited by 6-s ATP 4Ϫ applications. I rel was calculated, and mean data were approximated by Equation 3. Data are mean Ϯ S.D. from 5 to 15 oocytes. receptor synthesis, assembly, surface trafficking, channel function, or a combination of these factors. Carboxyl tail-truncated hP2X 7 receptors efficiently assembled stable homotrimers as evidenced by their migration in blue native PAGE gels (Fig. 2, A  and B), thus excluding an assembly defect as an explanation for the weak receptor function. To assess cell surface expression, we labeled plasma membrane-bound hP2X 7 receptors with a membrane-impermeant, amino-reactive fluorophore, Cy5 NHS ester, and visualized the SDS-PAGE-separated receptor subunits by fluorescence scanning (Fig. 2C). Because the synthesized proteins were also metabolically labeled with [ 35 S]methionine, the total receptor pool (surface plus internal) could be visualized by phosphorimaging autoradiography of the same SDS-PAGE gel (Fig. 2D). Quantification of SDS-PAGE-resolved, surface-labeled receptors established that the large increase in ATP-gated membrane current upon co-expression of hP2X 7 1-436 and hP2X 7 434 -595 was not paralleled by a proportional increase in the cell surface expression (Fig. 2C) or total expression (Fig. 2D) of the hP2X 7 1-436 truncation mutant. In fact, co-expression of the hP2X 7 434 -595 carboxyl tail decreased rather than increased cell surface abundance of the truncated hP2X 7 1-436 receptor (Fig. 2,  C, lanes 5 and 6, and E). We consider this effect to result specifically from the interaction of the hP2X 7  receptor with the hP2X 7 434 -595 carboxyl tail and an ensuing hP2X 7 434 -595 domain-dependent change of the intracellular trafficking of the holoprotein toward the trafficking properties of the wild type hP2X 7 receptor. Together, these findings indicate that the carboxyl tail domain boosts hP2X 7 1-436 channel function by a mechanism that is independent of plasma membrane incorporation.
Characterization of a Split hP2X 7 Receptor at the Single-channel Level-To unequivocally exclude any contribution of a contaminating current component to the effects observed at the macroscopic whole cell level, we recorded ATP 4Ϫ -induced single-channel activity in excised outside-out patches from oocytes expressing the full-length and a split hP2X 7 receptor. Oocytes expressing the wild type hP2X 7 receptor (Fig. 3A) or co-expressing the truncated hP2X 7 1-505 receptor and the hP2X 7 434 -595 carboxyl tail (Fig. 3B) displayed typical ATPgated single-channel activity (16), with similar kinetics and indistinguishable voltage dependences (Fig.  3, C and E). The single-channel conductance of the full-length hP2X 7 receptor (11.4 Ϯ 0.7 pS, n ϭ 11 patches) was not significantly different from that of the split hP2X 7 1-505 /hP2X 7 434 -595 receptor (12.6 Ϯ 0.9 pS, n ϭ 3 patches). Due to a slight inward rectification of hP2X 7 receptor-dependent single channel currents, fitting over a more negative potential range here (Ϫ120 to Ϫ40 mV) than before (Ϫ80 to 0 mV) accounts for the slightly higher single channel conductance of 11.4 pS compared with the previously reported 9.2 pS (17). Gaussian fits of amplitude histograms (Fig. 3C) yielded a mean open probability (P o ) for the split hP2X 7 receptor of 0.05 Ϯ 0.004 (S.E., n ϭ 3 patches) at 30 M ATP 4Ϫ , which is not significantly different from the P o value of 0.04 Ϯ 0.01 for the intact hP2X 7 receptor (16). As for the full-length hP2X 7 receptor, P o was voltage-independent at the negative membrane potentials studied.
In six of eight outside-out patches from oocytes expressing the split receptor, such single-channel events could be recorded. In contrast, in oocytes expressing the truncated hP2X 7 1-505 receptor alone, ATP-gated single-channel events could not be resolved (Fig. 3D) (n ϭ 30 patches). At high expres-  35 S-labeled receptor pool). His-tagged EGFP (shown in red) was co-expressed as a cytosolic indicator protein to assess the selectivity of the labeling of plasma membrane-bound proteins by the Cy5 NHS ester. His-EGFP remained unlabeled by Cy5 fluorescence, indicating that this dye did not enter the oocytes. E, the fluorescence intensity of plasma membrane-bound proteins in C was quantified and normalized to that of the full-length His-hP2X 7 receptor co-expressed with His-EGFP. Open bars and filled bars, normalized surface expression of the indicated hP2X 7 constructs alone or co-expressed with the carboxyl tail, respectively. Note that carboxyl tail-truncated hP2X 7 constructs were more abundant in the plasma membrane than the wild type hP2X 7 receptor, and that co-expression of the COOH-terminal tail domain diminished surface expression. sion levels, an increased ATP-dependent current noise was observed (data not shown), resulting presumably from short single-channel openings with very low current amplitudes; this could not be resolved with our recording system. hP2X 7 1-436 Receptor Cores and the hP2X 7 Carboxyl Tail Interact Physically-To provide further insight into the mechanism of activation of the hP2X 7 1-436 channel core by the co-expressed hP2X 7 434 -595 domain, we performed cross-linking experiments using the amino-reactive, thiol-cleavable reagents DSP or DTSSP. Initial control experiments showed that hP2X 7 1-436 and hP2X 7 434 -595 proteins in the non-His-tagged forms bound the Ni-NTA resin despite the presence of 20 mM imidazole in the binding and washing buffers (data not shown). Therefore, metal affinity chromatography did not appear to allow for unambiguous co-purification of non-His-tagged prey proteins with His-tagged bait proteins. As an alternative, we co-expressed His-hP2X 7 1-436 -StrepII (a protein bearing a COOH-terminal, nineamino acid StrepII tag in addition to the NH 2 -terminal hexahistidine tag) as a bait protein, together with the His-hP2X 7 434 -595 tail as the prey. Proteins were purified from aliquots of the same digitonin extracts of cells using metal affinity chromatography or Strep-Tactin chromatography to verify the expression of the two proteins (Fig.  4A) and to screen for the presence of co-purified His-hP2X 7 434 -595 tail protein (Fig. 4B), respectively. His-hP2X 7 1-436 -StrepII (calculated mass 52 kDa without N-glycans) and His-hP2X 7 434 -595 (calculated mass 19.5 kDa) could be isolated from the cells in which they were expressed (Fig. 4A). The more intense labeling of the His-hP2X 7 1-436 -StrepII protein compared with the His-hP2X 7 434 -595 carboxyl tail can be attributed mainly to its 4-fold higher number of methionine residues (8 versus 2 residues, respectively).
Purification by Strep-Tactin chromatography led to the co-isolation of the non-StrepII-tagged His-hP2X 7 434 -595 protein (Fig. 4B, lanes  3-8), particularly when cell lysis was performed in the presence of the water-soluble, membrane-impermeable cross-linker DTSSP (lanes 4 -6). The water-insoluble, membrane-permeant cross-linker DSP was somewhat less efficient in trapping   A and B, oocytes expressing the indicated hP2X 7 constructs singly or in combination were labeled with [ 35 S]methionine overnight, chased for 48 h, and then lysed in the absence or presence of the lysine-reactive cross-linker DSP or DTSSP as indicated. Proteins were isolated in parallel from aliquots of the same digitonin extracts by Ni-NTA chromatography or Strep-Tactin chromatography, as indicated. Shown are PhosphorImager scans of reducing SDS-PAGE gels. Cross-linking followed by Strep-Tactin chromatography resulted in the specific co-isolation of the non-StrepII-tagged hP2X 7 carboxyl tail (prey in B) with the StrepII-tagged truncated hP2X 7 1-436 receptor used as bait. C, oocytes were processed as above, except that a Cy5 surface labeling step was added before lysis. Proteins were purified by Ni-NTA or Strep-Tactin chromatography as indicated, resolved by reducing SDS-PAGE, and visualized by Typhoon fluorescence scanning (upper two panels) and 35 S phosphorimaging (lower panel). Using the StrepII-tagged hP2X 7 434 -595 carboxyl tail as bait, the non-StrepIItagged hP2X 7 1-436 channel core was isolated. In contrast, the co-expressed full-length hP2X 7 1-595 receptor could not be co-isolated. The asterisk and the open arrowhead indicate a nonspecific background band and the migration position of the full-length hP2X 7 subunit, respectively. the His-hP2X 7 434 -595 carboxyl tail (lanes 7 and 8). The trapped His-hP2X 7 434 -595 carboxyl tail only became visible after the crosslinks were broken by reducing the internal disulfide bond of DTSSP or DSP (non-reducing SDS-PAGE gel not shown). The strong dependence of the amount of trapped tail domain on an amino-reactive cross-linker excludes the possibility that the tail was anchored to the receptor core by an interchain disulfide bond formed under oxidizing cytosolic conditions between the numerous cysteine residues of the two fragments. When expressed alone, the His-hP2X 7 434 -595 carboxyl tail was not detected (lane 2), indicating that the Strep-Tactin resin does not directly bind the His-hP2X 7 434 -595 tail domain, thus confirming the suitability of this method.
To test whether hP2X 7 1-436 /hP2X 7 434 -595 protein complexes exist at the plasma membrane as expected from the electrophysiological experiments, we performed reciprocal co-purification experiments using hP2X 7 434 -595 -StrepII and His-hP2X 7 1-436 as bait and prey proteins, respectively. Just before cross-linking, the cells were surface-labeled with the membrane-impermeant Cy5 dye to detect plasma membrane-bound His-hP2X 7 1-436 receptors. In the non-reducing gel, very little co-isolated His-hP2X 7 1-436 protein was visible (data not shown). However, breaking of the cross-links by reducing the DTSSP-inherent disulfide bridge clearly revealed co-iso-lated, plasma membrane-bound His-hP2X 7 1-436 polypeptide (Fig.  4C, middle panel, lane 6). Significant amounts of co-isolated His-hP2X 7 1-436 protein were seen in the 35 S PhosphorImager scan if the StrepII-tagged bait was co-expressed (Fig. 4C, lower panel, lanes 5 and 6) complexes exist not only in undefined compartments in the cell interior, but also at the plasma membrane.
Electrophysiological recordings revealed no significant functional interaction between the hP2X 7 434 -595 tail domain and the co-expressed fulllength hP2X 7 receptor (Fig. 5D, left two bars). Therefore, we hypothesized also that no physical interaction occurs, and used the full-length hP2X 7 receptor as a possible negative control prey to assess the specificity of the crosslinking assay. Indeed, although clearly expressed in the cells (Fig.  4C, upper panel, lanes 1-3), the full-length hP2X 7 receptor could not be co-isolated with the co-synthesized hP2X 7 434 -595 -StrepII tail domain (Fig. 4C, middle and lower panels, lanes 2 and 3). The absence of detectable cross-linking of the full-length hP2X 7 receptor provides strong support for the view that DTSSP-assisted co-isolation of the hP2X 7 1-436 receptor with the hP2X 7 434 -595 tail reflects a specific association of the two polypeptides and not random collisional cross-linking. Moreover, it can be concluded that the tail binding site is blocked within the non-split full-length hP2X 7 subunit by a stable intrachain interaction between the hP2X 7 receptor core and the contiguous carboxyl tail, but it is unoccupied and hence accessible in the truncated hP2X 7 1-436 receptor lacking the distal carboxyl tail.
Mapping the Interaction Site of the Split hP2X 7 Receptor Fragments-The MEMSAT3 program (27) predicts the second transmembrane domain of the hP2X 7 subunit to extend from residues Asp 329 to Asp 356 . Accordingly, the truncated hP2X 7 1-436 subunit has a residual carboxyl tail of ϳ80 residues (Thr 357 -Phe 436 ) for interaction with the hP2X 7 434 -595 tail domain. Truncation at His 408 to remove 28 more residues from the residual carboxyl-terminal tail completely abrogated the physical interaction with (Fig. 5B, lanes 3 and 4) and stimulation by the hP2X 7 434 -595 tail domain (Fig. 5D). Akin to hP2X 7  , the hP2X 7 1-408 truncation mutant was also abundantly expressed at the cell surface (Fig. 5A, lanes 3 and 4) as a 1-408 and hP2X 7 1-436 subunits were abundantly expressed (A), but only the hP2X 7 1-436 subunit could be isolated as prey by the StrepII-tagged hP2X 7 434 -595 bait (arrows in B). *, nonspecific background bands. C, mapping of the interaction site on the hP2X 7 434 -595 tail domain. Oocytes were processed exactly as above, but only proteins purified by Strep-Tactin binding are shown. Co-isolation of the hP2X 7 1-436 prey subunit by the carboxyl tail subfragment hP2X 7 465-595 , but not by the smaller subfragment hP2X 7 495-595 , suggests that residues 465-494 are required for the core-tail interaction. D, current amplitude restoration test. Bars are mean Ϯ S.D. of 1 mM ATP 4Ϫ -gated current amplitudes recorded by two-electrode voltage clamping from n oocytes per group (as indicated by the numbers above the columns) expressing the indicated hP2X 7 receptor construct(s). E, schematic map of the co-expressed hP2X 7 constructs and summary of the functional and biochemical results. M1 and M2 delineate transmembrane domains 1 and 2, respectively.
To narrow down the part of the tail domain required for interaction with the receptor core, progressively shorter carboxyl tail fragments were co-expressed as baits with the prey hP2X 7 1-436 subunit (see scheme in Fig. 5E). The hP2X 7 1-436 subunit could be co-isolated with the subfragment hP2X 7

465-595
as efficiently as with the initial hP2X 7 434 -595 tail domain (Fig.  5C, lanes 2 and 4). However, deletion of an additional 30 residues, as realized with the hP2X 7 495-595 construct, resulted in a short-lived polypeptide, for which no interaction with the hP2X 7 1-436 prey subunit could be detected using the established cross-linking procedure (Fig. 5C, lane 6). Pulse-chase experiments revealed rapid degradation (data not shown) as an explanation for the low amount of hP2X 7 495-595 polypeptide present on day 3 after cRNA injection. In coisolation experiments performed on days 1 and 2, the hP2X 7 495-595 bait polypeptide demonstrated greater expression, but co-isolated hP2X 7 1-436 polypeptide was never detected (data not shown).
The carboxyl tail subfragments were also tested electrophysiologically with respect to restoration of the ATP-gated current on day 2 after cRNA injection. Subfragment hP2X 7 465-595 increased the current amplitude of the co-expressed hP2X 7 1-436 core ϳ3-fold, which is significantly less than the ϳ15-fold increase produced by the initial hP2X 7 434 -595 tail domain (Fig. 5D). No current increase was produced by the shortest subfragment, hP2X 7 495-595 ( Fig. 5D). Together, these data confine the functional and physical core-tail interactions to residues 409 -436 and 434 -494 of the hP2X 7 1-436 core and hP2X 7 434 -595 tail domain, respectively.
Interaction Stoichiometry of the Carboxyl Tail-truncated hP2X 7 Receptor and hP2X 7 Carboxyl Tail-To assess the stoichiometry of the interaction between the carboxyl tailtruncated receptor and the co-expressed carboxyl tail, we quantified the ratio of co-isolated hP2X 7  polypeptide to the hP2X 7 434 -595 domain at various cross-linker concentrations from the 35 S Phosphor-Imager scan shown in Fig. 6A (taking into consideration the different numbers of methionine residues of the two polypeptides). Plotting this ratio against the concentration of the cross-linker DTSSP yielded a saturation curve with a plateau at ϳ0.85 (Fig.  6B), as calculated from the Hill equation. A justifiable conclusion from these data is that the truncated hP2X 7 1-436 receptor and the hP2X 7 434 -595 carboxyl tail interact with a 1:1 stoichiometry.
To assess how many carboxyl tails a trimeric hP2X 7 receptor needs for full channel activity, we co-expressed the full-length hP2X 7 subunit with an excess of the truncated hP2X 7 1-436 subunit. Because P2X subunit assembly domains are located in the ectodomain and second transmembrane domain (22), efficient co-assembly of the full-length and truncated hP2X 7 1-436 subunits was possible (Fig. 6C). Co-expression of the full-length hP2X 7 subunit and truncated hP2X 7 1-436 mutant drastically diminished the ATP-gated current amplitude from 100% for Shown is a 35 S PhosphorImager scan of a reducing SDS-PAGE gel. B, the molar ratio of prey/bait was calculated using the PhosphorImager data of the scan shown in A and taking into account that prey and bait contain eight and two methionine residues per molecule, respectively. Inset, example of a quantitative scan used to assess the relative amounts of the His-hP2X 7 1-436 polypeptide and the hP2X 7 434 -595 carboxyl tail based on cross-linking experiments. C, co-assembly of full-length hP2X 7 1-595 subunits and carboxyl tail-truncated hP2X 7 1-436 subunits. Receptors were purified from Cy5-labeled oocytes. The expressed hP2X 7 constructs contained either His or StrepII tags as indicated in the legend. Prey and bait, non-StrepII-tagged and StrepII-tagged polypeptide, respectively. D, dominant-negative effect of the truncated hP2X 7 1-436 subunit on full-length hP2X 7 receptor function. Currents gated by 1 mM ATP 4Ϫ were recorded by two-electrode voltage clamping from oocytes injected with equal amounts of the indicated cRNAs 2 days earlier. Bars are mean Ϯ S.D. from 18 oocytes in each group. Open and filled circles schematically illustrate trimeric receptor complexes consisting of full-length and/or truncated hP2X 7 1-436 subunits, respectively. Numbers indicate the probability that the corresponding trimer is formed. the full-length hP2X 7 receptor alone to 8.9 Ϯ 6.2% (Fig. 6D). Assuming a surface expression ratio of 1:1 for hP2X 7 1-436 receptor to full-length hP2X 7 receptor (Fig. 6C, lane 2) and that full-length hP2X 7 1-595 and truncated hP2X 7 1-436 subunits combine with each other with the same probability, fully functional, full-length, homotrimeric hP2X 7 receptors should occur with a probability of 12.5%. The probability of heterotrimers consisting of two full-length and one truncated subunit and vice versa is 37.5% for each. The measured relative mean current amplitude of 8.9% clearly implies that only homotrimers of fulllength hP2X 7 subunits are fully functional. Incorporation of a single truncated subunit into the trimer abolishes channel function, implying a dominant-negative effect of truncated hP2X 7 1-436 subunits on channel function. This result indicates that stimulation of channel activity by the carboxyl tail occurs not incrementally, but in an "all-or-nothing" fashion: any increase in channel activity above the basal level necessitates the presence of three carboxyl tails per trimeric hP2X 7 receptor.

DISCUSSION
Large COOH-terminal Tail Truncations Strongly Impair hP2X 7 Cation Channel Function-Consistent with previous reports (4, 10, 18), we found that carboxyl tail-truncated P2X 7 receptors are capable of mediating ATP-gated currents carried by small inorganic cations. However, taking into account their elevated cell-surface abundance, the truncated hP2X 7 1-408 and hP2X 7 1-436 receptors produced ϳ20-fold less inward current than the full-length hP2X 7 receptor. The higher cell-surface abundance of carboxyl tail-truncated hP2X 7 receptors compared with the full-length hP2X 7 receptor can be attributed to the increased total receptor expression combined with the elimination of two regulatory trafficking motifs in the carboxyl tail between residues 551 and 582: a more proximally located ER retention motif and a more distally located export signal that overrides the ER retention motif (10). Altogether, our data clearly indicate that the small currents mediated by the hP2X 7 receptor core are due to a strong reduction in channel function and not to a trafficking defect. Similar small currents were observed in our previous study, but receptor levels at the plasma membrane were not determined (18). Kinetically, the low current activation of the truncated hP2X 7 1-436 receptor can be described by the functional loss of the low-affinity ATP 4Ϫ effector site, i.e. the site responsible for ϳ90% of the exponentially saturating current, the entire linearly increasing current, most of the fast deactivating current (18,28) and the entire 9.2 pS (here 11.4 pS) single channel activity (16). Truncated hP2X 7 versions are not only of biophysical interest, but also seem to play a physiological role. A splice variant of the hP2X 7 receptor lacking the COOH-terminal part of the second transmembrane domain and the entire carboxyl tail is highly expressed in various human tissues and was found to mediate ATP-gated Ca 2ϩ influx and membrane depolarization when recombinantly expressed in HEK-293 cells (29). A further truncated P2X 7 receptor variant, designated P2X 7-j , lacking part of the ectodomain, the entire second transmembrane domain, and the intracellular carboxyl terminus was identified to occur naturally in cervical cancer cells (30). The P2X 7-j var-iant interacted with full-length P2X 7 subunits in a manner consistent with hetero-oligomerization and blocked P2X 7 receptor-mediated actions, similar to what we observed in the present study for the truncated hP2X 7 1-436 construct. The Carboxyl Tail Domain Fully Rescues hP2X 7 Channel Function and Targeting Information-The above data clearly indicate that both domains, the hP2X 7 receptor core, and the cytoplasmic carboxyl tail are required for normal cation channel function. This view is further strengthened by the finding that co-expression of the tail domain as a separate protein rescues carboxyl tail-truncated hP2X 7 1-436 receptors, forming split receptor channels that are indistinguishable in their electrophysiological phenotype from wild type hP2X 7 receptors consisting of contiguous subunits. This observation, together with the co-purification of non-cross-linked and cross-linked split fragments of the hP2X 7 receptor, reveals a functionally essential non-covalent association between the remaining cytoplasmic carboxyl tail of the hP2X 7 subunit core (residues 357-436) and the truncated tail domain (residues 434 -595). The association forces between the hP2X 7 receptor core and tail domain are weaker than those mediating the assembly and stability of trimeric P2X receptor complexes, which require SDS or urea for disassociation (21,22).
By analyzing a slightly shorter truncation mutant, hP2X 7 1-408 , and subfragments of the carboxyl tail domain, the proximal and distal tail association sites could be limited to include residues 409 -436 and 465-494, respectively. Biochemically, deletion of residues 434 -464 from the carboxyl tail domain did not impair the interaction with the truncated hP2X 7 1-436 receptor. Electrophysiologically, however, residues 434 -464 contributed significantly to the stimulatory effect of the carboxyl tail domain on the ATP-gated current of the hP2X 7 1-436 receptor. This indicates that the split hP2X 7 receptor requires virtually the entire carboxyl tail for full channel activity.
We consider it unlikely that the interaction between proximal and distal tail domains contributes significantly to homotrimerization for the following reasons: (i) assembly domains of P2X subunits are in the ectodomain and the second transmembrane segment (22); amino and carboxyl tails have consistently been found to be nonessential for P2X subunit assembly (22,31); and (ii) carboxyl tail-truncated hP2X 7 receptors migrate almost entirely as stable homotrimers in blue native PAGE gels, indicating that the carboxyl tail is indeed dispensable for homotrimerization. Because the receptor core and tail domain are normally covalently connected in a single polypeptide chain, this association clearly must serve other purposes than as a mere physical link to keep the receptor assembled.
Possible Role of the hP2X 7 Carboxyl Tail in Cation Channel Function-An important role of cytoplasmic domains in channel gating has been recognized for P2X 2 receptors (32-35), various K ϩ channels (36 -38), hyperpolarization-activated cyclic nucleotide-modulated channels (39), ClC-type anion channels (40), and voltage-gated Ca 2ϩ channels (41). Models of how these cytoplasmic domains regulate the opening of ion channels have been deduced by combining x-ray crystallographic and functional evidence. By comparing crystal structures in closed and open states, the coupling of COOH-termi-nal tail conformation to gating transitions has been particularly well established for the prokaryotic Ca 2ϩ -gated K ϩ channel MthK (42). MthK harbors, on the intracellular side of the pore, an octameric gating ring formed by Ca 2ϩ -binding carboxylterminal domains that expand upon Ca 2ϩ binding. This expansion can exert a lateral force on the channel pore that opens the channel.
The cytoplasmic domains of hyperpolarization-activated cyclic nucleotide-modulated and MthK channels have obvious roles in transducing the binding of intracellular cyclic nucleotides or Ca 2ϩ , respectively, into channel gating. The physiological meaning of the gating modulation of the P2X 7 receptor by the COOH-terminal domain is less obvious. We suggest that the P2X 7 receptor represents a variation of the above channel modulation theme, such that the P2X 7 receptor consists of three modules: an extracellular ATP-sensing module, a membrane-embedded pore region, and a cytoplasmic signal-sensing module that is directly connected with the second transmembrane domain, the inner region of which is critical for channel function (22,43). This view places gating of the P2X 7 channel pore under the control of two modules, one in the extracellular and one in the intracellular environment. A variety of proteins have already been shown to interact with the carboxyl tail domain of the P2X 7 receptor (6, 15), but it is unknown whether these protein-protein interactions affect channel gating. A structure consisting of three carboxyl tails, each of which is needed for full channel activity, would be consistent with the observation that the lack of a single carboxyl tail domain is dominant-negative for hP2X 7 receptor activity.
In conclusion, we propose that the physical interaction observed between the carboxyl tail domain and the remainder of the P2X 7 subunit connects signals from the intracellular and extracellular environments to channel gating. This link may also operate in the reverse direction to signal the gating state to the carboxyl tail domain, thereby regulating intracellular protein-protein interactions such as those leading to cytolytic pore formation and interleukin release.