Tyrosine Phosphorylation of HSP90 within the P2X7 Receptor Complex Negatively Regulates P2X7 Receptors*

The purinergic P2X7 receptor not only gates the opening of a cationic channel, but also couples to several downstream signaling events such as rapid membrane blebbing, microvesicle shedding, and interleukin-1β release. Protein-protein interactions are likely to be involved in most of these signaling cascades; and recently, a P2X7 receptor-protein complex comprising at least 11 distinct proteins has been identified. We have studied one of these interacting proteins, HSP90, in human embryonic kidney cells expressing either human or rat P2X7 receptors as well as in rat peritoneal macrophages using biochemical (immunoprecipitation and Western blotting) and functional (membrane blebbing and currents) assays. We found that HSP90 was tyrosine-phosphorylated in association with the P2X7 receptor complex, but not in the cytosolic compartment. The HSP90 inhibitor geldanamycin decreased tyrosine phosphorylation of HSP90 and produced a 2-fold increase in the sensitivity of P2X7 receptors to agonist. Protein expression and tyrosine phosphorylation of a mutant P2X7 receptor in which a tyrosine in the C-terminal domain was substituted with phenylalanine (Y550F) were not changed, but tyrosine phosphorylation of HSP90 associated with this mutant P2X7 receptor complex was significantly greater than that associated with the wild-type complex. P2X7-Y550F receptors showed a 15-fold lower sensitivity to agonist, which was reversed by geldanamycin. We conclude that selective tyrosine phosphorylation of P2X7 receptor-associated HSP90 may act as a negative regulator of P2X7 receptor complex formation and function.

Extracellular ATP activates both ligand-gated ion channels (P2X receptors) and G protein-coupled metabotropic receptors (P2Y). There are seven ATP-gated P2X receptor channels (P2X 1 -P2X 7 ), which can form as functionally distinct homomeric and/or heteromeric channels (1). The P2X 7 receptor, which is prominently expressed in many immune and epithelial cells, differs strikingly from the other P2X receptors in additionally coupling to numerous other signaling cascades (2); these include rapid release of interleukin-1␤ via microvesicle shedding (3), plasma membrane blebbing (3,4), macrophage fusion (5,6), lymphoid cell proliferation (7), and apoptotic/ necrotic cell processes (8). It is reasonable to assume that additional proteins are directly involved with P2X 7 receptor coupling to these events; indeed, in a recent proteomic study, we identified 11 proteins as members of a P2X 7 receptor-protein complex (9). One of these proteins was a tyrosine phosphatase (receptor protein-tyrosine phosphatase-␤); blockade of receptor protein-tyrosine phosphatase-␤ activity with phosphatase inhibitors revealed a significant tyrosine phosphorylation of the P2X 7 receptor itself and an increase in ATP-evoked membrane blebbing and ion current (9). During these experiments, we noted that one other protein in this complex was also tyrosine-phosphorylated; this was a 90-kDa protein that most likely represented HSP90 1 because HSP90 was one of three heat shock proteins identified in the P2X 7 receptor complex (HSP90, HSP70, and HSC70).
HSP90 accounts for 1-2% of total cytosolic protein content, and its role as a molecular chaperone able to prevent protein aggregation and to promote refolding of denatured proteins has been studied in much detail (10). More recently, interest in HSP90, in particular, has expanded to include an apparently mechanistically distinct function involving direct interaction with specific biochemical signaling pathways underlying apoptosis (11)(12)(13). For example, the anti-apoptotic action of HSP90 has been linked to its direct association with Apaf-1 apoptotic protease activation factor-1, rendering Apaf-1 unable to bind to caspase-9, thus inhibiting the subsequent apoptotic "caspase cascade" (14,15). However, it has been difficult to provide firm evidence for HSP90 substrates or biochemical mechanisms of HSP90-substrate interactions in cell death signaling pathways. No signature sequence for substrate recognition in the HSP90 molecule has been identified, and most HSP90-substrate interactions are regulated by low abundance co-chaperones (e.g. Hop and p23). The HSP90-caspase axis may have potential relevance to P2X 7 receptor function because activation of this purine receptor in human and/or rodent macrophages and microglia can result in increased levels of caspase-1, -3, and -8 with consequent apoptotic cell death (16,17), although it should be emphasized that neither the molecular nor cellular mechanisms by which P2X 7 receptor activation leads to activation of caspases are known.
Although numerous studies have shown both serine and threonine phosphorylation of HSP90 and the functional consequences (18,19), there appears to be only a single report demonstrating tyrosine phosphorylation of HSP90 (20). In the present study, we have used biochemical and functional assays to examine the interaction between HSP90 and P2X 7 receptors in human embryonic kidney (HEK) cells heterologously expressing either human or rat P2X 7 receptors as well as in rat peritoneal macrophages expressing native P2X 7 receptors. We have specifically addressed the question of tyrosine phosphorylation of HSP90 in the P2X 7 receptor complex and its possible functional role in HSP90-P2X 7 receptor interactions. Our data suggest a non-chaperone role for tyrosine-phosphorylated HSP90 as a P2X 7 receptor repressor.

EXPERIMENTAL PROCEDURES
Materials and Cells-Geldanamycin (Calbiochem) was dissolved in Me 2 SO to obtain a 2 mM stock and then diluted in the appropriate buffer or in the culture medium. Potassium bisperoxo(1,10-phenanthroline) oxovanadate (Calbiochem) was dissolved in water and then used at the appropriate concentration in the different buffers. BzATP was from Sigma. Dodecyl ␤-D-maltoside was from Calbiochem. Immunoblot polyvinylidene difluoride membrane and Immun-star substrate were from Bio-Rad. During the immunoprecipitation procedure, Gamma Bind G-Sepharose (Amersham Biosciences) and protease inhibitors (Complete TM , Roche Applied Science) were used. Polyvinylidene difluoride membranes were stripped using Restore Western blot stripping buffer (Pierce). The following antibodies (Abs) were used for co-immunoprecipitation, Western blotting, and immunocytochemistry: anti-rat P2X 7 (rP2X 7 ) and anti-P2X 2 polyclonal Abs (Alomone Labs), anti-HSP90 and anti-phosphotyrosine (PY20) Abs (Transduction Laboratories), anti-Glu-Glu epitope tag (GLNEYMPME; referred to as EE) monoclonal Ab (BabCo), alkaline phosphatase-conjugated goat anti-rabbit and antimouse IgG (HϩL) Abs (Bio-Rad), and fluorescein isothiocyanate-conjugated goat anti-rabbit and anti-mouse Abs (Sigma). The monoclonal Ab directed against the extracellular domain of rP2X 7 was described previously (21). The anti-human P2X 7 polyclonal Ab was kindly provided by Dr. J. Barden (University of Sydney, Sydney, Australia).
HEK293 cells stably or transiently transfected with rat or human P2X 7 receptors were used as detailed previously, and transient transfection was accomplished using LipofectAMINE 2000 and cotransfections with enhanced green fluorescent protein (22). In all single cell experiments using transient transfections (electrophysiology and membrane blebbing), green fluorescent protein-positive cells were assumed to express also the P2X 7 receptor; subsequent immunostaining for P2X 7 receptors revealed that this was true for 85-95% of the green fluorescent protein-positive cells. Peritoneal macrophages were obtained from young adult rats as described previously (21). All experiments on macrophages were carried out after a 4-h incubation with lipopolysaccharide (1 g/ml), as this has been shown to up-regulate expression of P2X 7 receptors (23).
Immunoprecipitation, Immunoblotting, and Immunohistochemistry-Cells were incubated in 500 l of solubilization buffer (20 mM dodecyl ␤-D-maltoside in phosphate-buffered saline containing protease inhibitors) for 2 h under gentle agitation, after which cell debris was discarded, and background proteins were pre-absorbed with 50 l of Gamma Bind G-Sepharose. The appropriate Ab (anti-P2X 7 , 80 ng/ml; or anti-EE, 10 g/ml) was applied, followed by 50 l of Gamma Bind G-Sepharose; and immunoprecipitation was carried out overnight at 4°C. Resin was then washed (three to five washes) with 10 volumes of solubilization buffer, and proteins were separated from the Sepharose by boiling for 5 min in 50 l of SDS sample buffer. Proteins were separated by SDS-PAGE (12%) and electrophoretically transferred to polyvinylidene difluoride membranes. Initial experiments used total cell lysates as positive controls for immunoblotting with the anti-HSP90 and/or anti-P2X 7 antibody and for estimating comparable amounts of protein to load onto the gels for experiments as illustrated in Figs. 1C and 3B. They were obtained by lysing cells in 100 l of solubilization buffer and denaturing in an equal volume of SDS loading buffer, after which 20 l was loaded onto the SDS-polyacrylamide gel. Therefore, the amount of samples loaded represented ϳ50% of the total proteins from immunoprecipitation, but only 10% of the total proteins from crude cell lysates.
Polyvinylidene difluoride membranes were rinsed in 10 mM Tris and 250 mM NaCl (pH 7.5) (TBS) and then blocked with 1% bovine serum albumin in TBS and 0.1% Tween 20 (TBS/Tween) for 1 h at room temperature. Membranes were washed with TBS/Tween, and primary Abs were applied at the appropriate concentration and incubated overnight at 4°C. Membranes were washed twice with TBS/Tween, and the secondary Ab was added to the blocking solution for 2 h at room temperature. Membranes were then washed three times, and Immun-star substrate was applied for 5 min in the dark. In most experiments, membranes were stripped and reblotted to confirm directly tyrosine phosphorylation of HSP90. The stripping procedure was as follows. Membranes were washed twice with phosphate-buffered saline and incubated in Restore Western blot stripping buffer for 1 h at 50°C. They were then incubated in TBS/Tween and 1% bovine serum albumin blocking solution for 1 h at room temperature, and the standard procedure described above for Western blotting was repeated, but inverting the antibodies. The anti-rP2X 7 receptor Ab (300 ng/ml), anti-HSP90 Ab (250 ng/ml), anti-phosphotyrosine Ab PY20 (200 ng/ml), and anti-human P2X 7 receptor polyclonal Ab (1:250 dilution) were the primary Abs used. The secondary Abs were used at a 1:3000 dilution. Densitometric semiquantitation of Western blot bands was performed using the Gene Genius imaging system; images were acquired from films and analyzed using GeneSnap and Genetools software (Syngene). The data represent the ratio of the calculated density of each sample to the control untreated sample or the IgG band present on the same film. The total number of cells was constant for all conditions in a given experiment.
Immunostaining methods have been described in detail previously (21). Briefly, cells were fixed with Zamboni's buffer for 20 min at room temperature, blocked in 5% goat serum in phosphate-buffered saline with 2% Triton X-100 for 30 min, incubated with the primary Ab (anti-P2X 2 , 0.6 g/ml; anti-rP2X 7 , 0.12 g/ml; or anti-HSP90, 0.5 g/ml) for 2-3 h at room temperature, rinsed, and incubated with the secondary Ab (fluorescein isothiocyanate-conjugated anti-rabbit or antimouse, 0.25 g/ml) for 1 h. Cells were viewed under a fluorescence microscope (Olympus U-RFL-T) and photographed using a digital camera (SpotCam, Cambridge, UK). Numbers of blebs/cell were counted, and data represent a total of 300 cells measured from three separate experiments, in each of which 100 cells were examined. Confocal microscopy (Nikon PCM2000 with EZ 2000 software) was also used to examine plasma membrane versus intracellular localization of P2X 7 receptors.
ATP Secretion-Cells were plated on polylysine-coated 24-well culture plates and maintained until confluent (usually 2 days), at which time they were incubated for 30 min in the presence of test or control solution. The supernatant was removed, and the ATP concentration was determined by luciferin/luciferase assay using a Lucy Anthos-1 luminometer (Lab Tech International).
Electrophysiology, Fura-2, and Blebbing Measurements-Standard whole-cell patch-clamp methods were used; all methods and internal/ external solutions were identical to those detailed in several recent studies (9,21,24). To take account of the occasional "run-up" of agonist sensitivity at this receptor (25,26), BzATP concentration-response curves were obtained as follows. For each cell, two consecutive sets of agonist applications (from low to high concentrations) were carried out; results from the second set were used to plot control curves. However, in this series of experiments, significant run-up of responses was observed in Ͻ10% of the cells and so is unlikely to influence interpretations of the results obtained. Time to onset of membrane blebbing was measured as described in detail previously (22).
In separate experiments, changes in the cytoplasmic free calcium concentration were measured with the fluorescent indicator Fura-2/ AM. Cells were loaded with 2 M Fura-2/AM for 30 min in the presence of 250 M sulfinpyrazone, rinsed, and resuspended in physiological saline solution at a final concentration of 10 6 /ml. Changes in cytosolic calcium concentrations were determined in a thermostat-controlled, magnetically stirred cuvette fluorometer (LS50, PerkinElmer Life Sciences, Beaconsfield, UK) with an excitation ratio of 340/380 nm and an emission wavelength of 505 nm.
Data Analysis-All data are shown as means Ϯ S.E. Tests of significance were carried out by Student's t test using GraphPAD Instat software; p values Ͻ0.005 were considered significant.

RESULTS
HSP90 Is Tyrosine-phosphorylated in the P2X 7 -associated Protein Complex-After immunoprecipitation with the anti-P2X 7 Ab, immunoblotting with Ab PY20 revealed a tyrosinephosphorylated protein at 90 kDa. After treatment with the tyrosine phosphatase inhibitor potassium bisperoxo(1,10-phenanthroline) oxovanadate, this band was unchanged, whereas a second band at 76 kDa was detected (Fig. 1A). We have previously shown that the 76-kDa band present after tyrosine phosphatase inhibition is the P2X 7 receptor itself (9). Immunoblotting with Ab PY20, followed by stripping and reprobing with the anti-HSP90 antibody, directly confirmed that the 90-kDa band was tyrosine-phosphorylated HSP90 (Fig. 1B). In contrast, no tyrosine phosphorylation of uncomplexed (cytosolic) HSP90 could be detected; Fig. 1C shows that, although the quantity of both P2X 7 and HSP90 that we loaded onto the gel (see "Experimental Procedures") was approximately the same in the cell lysates (lanes 1 and 3) and in the co-immunoprecipitated samples (lanes 2 and 4), tyrosine phosphorylation of HSP90 was detected only in the immunoprecipitated samples (lanes 2 and 4). This result was obtained with HEK cells stably expressing rP2X 7 receptors using the anti-ectodomain rP2X 7 Ab for immunoprecipitation (Fig. 1C, lanes 1 and 2) as well as with HEK cells transiently transfected with an EE epitopetagged rP2X 7 receptor and immunoprecipitated using the anti-EE Ab (Fig. 1C, lanes 3 and 4), thus ruling out possible technical artifacts due to the anti-P2X 7 Ab. Activation of P2X 7 receptors with a near-maximal concentration of BzATP (30 M for 5 min) produced no significant alteration in the amounts of HSP90 that co-immunoprecipitated with the P2X 7 receptor or in the levels of tyrosine phosphorylation of the precipitated HSP90, nor did treatment with various kinase inhibitors (the Src kinase inhibitor protein phosphatase-2, the phosphatidylinositol 3-kinase inhibitor wortmannin, and the protein kinase C inhibitor staurosporine) (data not shown).
Geldanamycin Inhibits HSP90 Tyrosine Phosphorylation-Geldanamycin specifically inhibits HSP90 by binding to the ATP hydrolysis site on the molecule with an affinity Ͼ500 times greater than for ATP, thus effectively displacing ATP and so disrupting HSP90-substrate interactions (27)(28)(29). We therefore examined P2X 7 receptor and HSP90 profiles on Western blots after treatment for 0.5 and 16 h with geldanamycin (1-5 M). Geldanamycin did not alter the total amount of HSP90 protein immunoprecipitated within the P2X 7 receptor complex, but significantly decreased the levels of tyrosine phosphorylation of HSP90 (Fig. 2B); geldanamycin also significantly increased the amounts of immunoprecipitated P2X 7 re-ceptor protein when the anti-ectodomain P2X 7 Ab was used, but not when the anti-EE epitope tag Ab was used. The most likely explanation for this difference is that a proportion of the membrane-localized P2X 7 receptor is in a conformation or complex unfavorable to immunoprecipitation by the anti-ectodomain Ab, whereas tyrosine dephosphorylation of the complexed HSP90 alters the conformation/complex of these P2X 7 receptors in such a way as to allow more efficient binding to this Ab. Such an interpretation would be in keeping with known roles of HSP molecules to alter the folding and conformation of their client proteins (37)(38)(39). Confocal microscopic examination of immunostained cells using either the anti-ectodomain P2X 7 Ab or the anti-EE Ab revealed that geldanamycin did not alter the plasma membrane localization of the receptor, nor did it reveal any increased intracellular (i.e. endoplasmic reticulum/Golgi) immunostaining or altered subcellular localization (see below and Fig. 4).
HSP90-P2X 7 Interactions in Native Cells-It is well known that the actions of HSP90 as well as geldanamycin are highly dependent on the cellular context; in particular, many downstream actions of geldanamycin, such as its anti-apoptotic or antitumor effects, are apparent only in certain types of cells (30,31). In this regard, our previous proteomic study examined only heterologously expressed rP2X 7 receptors, raising the possibility that HSP90-P2X 7 receptor interactions may not even occur in native cells expressing endogenous P2X 7 receptors (9). Therefore, we repeated the above experiments on rat peritoneal macrophages because macrophage cells have been a focus of several studies on the actions of P2X 7 receptors in immune cells (4,5,21). Qualitatively similar results were obtained, as illustrated in the representative experiment shown in Fig. 3A. HSP90 co-immunoprecipitated with the native rP2X 7 receptor, and this complexed HSP90 (but not the cytosolic fraction) was tyrosine-phosphorylated. Geldanamycin (5 M, 30 min) produced small decreases in the levels of HSP90 tyrosine phospho- FIG. 1. HSP90 is tyrosine-phosphorylated within the P2X 7 receptor complex. A, rP2X 7 immunoprecipitated from HEK cells transfected with the receptor and immunoblotted with the antiphosphotyrosine Ab (Ab PY20). Lanes 1 and lane 2, in the absence and presence of the tyrosine phosphatase inhibitor potassium bisperoxo(1,10-phenanthroline) oxovanadate (bpV; 100 M), respectively. Note that the 90-kDa band was present in both cases, whereas the 76-kDa band (which is the P2X 7 receptor protein) was detected only after tyrosine phosphatase inhibition. B, immunoprecipitation with the anti-P2X 7 Ab and blotting with the anti-HSP90 antibody, followed by stripping and reprobing with Ab PY20 (anti-PY), confirm tyrosine-phosphorylated HSP90 in the P2X 7 receptor complex. The 52-kDa band in A and B is IgG. C, tyrosine-phosphorylated HSP90 (PY-Hsp90) is detected only in the immunoprecipitated P2X 7 complex (lanes 2 and 4), but not in total cell lysates (lanes 1 and 3). Lanes 1 and 2, HEK cells stably expressing rP2X 7 receptors using the anti-P2X 7 Ab for immunoprecipitation; lanes 3 and 4, an identical experiment performed on HEK cells transiently transfected with EE-tagged P2X 7 receptors and immunoprecipitated with the anti-EE Ab. In A-C, all blots shown are representative of six separate experiments.
rylation, but the changes were not significantly different from the control levels.
In view of the known species differences in the pharmacology of P2X 7 receptors (1, 26, 32), we additionally asked whether the HSP90-P2X 7 receptor interaction observed above for native and cloned rP2X 7 receptors held for the human ortholog. We were not able to examine endogenous human P2X 7 receptors from native cells because the anti-ectodomain P2X 7 monoclonal Ab was species-specific for rat versus human in immunoprecipitation, whereas the other P2X 7 Abs available for Western blotting and immunostaining were not effective for immunoprecipitation. 2 Therefore, we transiently transfected HEK cells with EE epitope-tagged human P2X 7 receptor cDNA and used the anti-EE Ab for immunoprecipitation. Here, too, HSP90 co-immunoprecipitated with human P2X 7 ; the complex-associated HSP90 was tyrosine-phosphorylated; and the intensity of tyrosine phosphorylation of HSP90 was significantly decreased after treatment with geldanamycin (Fig. 3B).
Functional Assays for HSP90-P2X 7 Receptor Interactions-We used geldanamycin to investigate whether there is any functional evidence for HSP90-P2X 7 receptor interactions in HEK cells transiently or stably expressing P2X 7 receptors. Geldanamycin alone (1 or 5 M, 2-30 min) did not evoke any calcium influx (n ϭ four preparations) or inward currents (n ϭ 35 cells), thus demonstrating that it is not an agonist for P2X 7 receptors; it also had no significant effect on agonist-evoked membrane blebs (30 M BzATP; n ϭ 10 cells), currents (n ϭ 35 cells), or calcium influx (n ϭ 10 preparations). However, when P2X 7 (but not untransfected or P2X 2 ) receptor-expressing cells were exposed to geldanamycin for 0.5-16 h, significant membrane blebbing was observed (Fig. 4, A and B). The characteristics of these blebs were indistinguishable from those induced by activation of P2X 7 receptors with agonist ( Fig. 4A) (3), although it should be emphasized that membrane blebbing in response to geldanamycin was observed only after prolonged incubation (Ͼ30 min), whereas P2X 7 receptor activation by ATP or BzATP produced membrane blebbing within 30 -60 s (3).
Because the geldanamycin-induced blebbing was specific for cells expressing P2X 7 receptors (Fig. 4B) and because ATP has been shown to be released from HEK cells (33), we hypothesized that the blebbing resulted from activation of these receptors by constitutively released ATP. Inclusion of the ectonucleotidase apyrase (4 units/ml) or the P2X 7 receptor-selective antagonist Coomassie Brilliant Blue G (300 nM) (24) with geldanamycin significantly reduced the membrane blebbing ( Fig. 4B). We also observed an average 4-fold increase in the quantity of ATP secreted into the medium by P2X 7 receptorexpressing cells after a 30-min exposure to geldanamycin compared with cells exposed to Me 2 SO vehicle (Fig. 4C).
Involvement of P2X 7 Tyr 550 with HSP90 -In our previous study, site-directed mutagenesis was used to individually substitute each tyrosine with phenylalanine in the intracellular N and C termini and the putative second transmembrane domain (total of 16 residues) of the P2X 7 receptor in an attempt to identify residues that may be directly involved in tyrosine phosphorylation of this protein; none of the mutants showed altered protein expression or plasma membrane localization, and only two mutants showed altered functional phenotypes: Tyr 343 and Tyr 550 (9). Our biochemical and functional assays then provided fairly conclusive evidence that Tyr 343 (but not Try 550 ) was directly phosphorylated and that this phosphorylation could account for the altered function of the Tyr 343 (but not Try 550 ) mutant (9). We therefore investigated additional biochemical and functional properties of the mutant P2X 7 -Y550F receptor with regard to potential HSP90 interaction.
Co-immunoprecipitation and Western blot experiments confirmed that immunoprecipitated P2X 7 -Y550F protein expression was not significantly different from that of the wild-type receptor (Fig. 5A) and was similarly localized to the plasma membrane and that tyrosine phosphorylation of this mutant P2X 7 receptor was similar to that of the wild-type receptor (data not shown, but see Fig. 5 in Ref. 9). The protein levels of HSP90 within the mutant and wild-type receptor complexes were also not significantly different (Fig. 5, A and B). However, there was a striking increase in the level of tyrosine phosphorylation of HSP90 associated with the mutant receptor (Fig. 5,  A and B). In contrast, the P2X 7 -Y343F receptor complex showed no differences from the wild-type receptor in either the amounts of protein or the levels of tyrosine phosphorylation of HSP90 (data not shown).
BzATP (30 and 100 M) produced multiple membrane blebs in Ͼ90% of the transiently transfected wild-type P2X 7 receptorexpressing cells, with time to onset of first bleb averaging 57 Ϯ 2.2 s (n ϭ 34 from three separate transfections) and 34 Ϯ 2.6 s (n ϭ 40 from four separate transfections) for 30 and 100 M BzATP, respectively (Fig. 5C). However, application of these concentrations of agonist for up to 5 min evoked membrane blebbing in Ͻ5% of the cells transfected with the P2X 7 -Y550F receptor (n ϭ 62 from four transfections) (Fig. 5C). After a 10-min application of geldanamycin (1 M), BzATP (30 M) evoked membrane blebbing in ϳ80% of the cells transfected with the mutant receptor, with time to initial onset of 78 Ϯ 4 s (n ϭ 44 from three separate transfections) (Fig. 5C).
We investigated this apparent geldanamycin-induced shift in sensitivity to BzATP in the mutant receptor by constructing BzATP concentration-response curves for membrane current in wild-type and mutant Y550F receptors in the absence and presence of geldanamycin. Geldanamycin produced a small but significant leftward shift of ϳ2-fold in the dose-response curve of the wild-type receptor (Fig. 6A); half-maximal concentrations (EC 50 values) in control and geldanamycin-treated samples were 31 Ϯ 5 M (n ϭ 5) and 13 Ϯ 3 M (n ϭ 6), respectively. In the mutant receptor, the BzATP concentration-response curve was shifted to the right by Ͼ15-fold compared with the wild-type receptor, and geldanamycin produced a dramatic leftward shift to levels not significantly different from control wild-type responses (Fig. 6B); EC 50 values in the P2X 7 -Y550F receptor in the absence and presence of geldanamycin were 559 Ϯ 42 M (n ϭ 6) and 58 Ϯ 12 M (n ϭ 6), respectively. Maximal current amplitudes recorded in the absence and presence of geldanamycin were not significantly different in either the wild-type or mutant receptor.

DISCUSSION
This study has shown that protein-protein interaction between rat and human P2X 7 receptors and a subpopulation of HSP90 molecules occurs not only in heterologous expression systems, but also in native macrophages expressing endogenous P2X 7 receptors. The HSP90 specifically associated with the P2X 7 receptor complex is tyrosine-phosphorylated, and the levels of tyrosine phosphorylation of this HSP90 are associated with changes in at least some of the functional properties of the P2X 7 receptor, where increased tyrosine phosphorylation of complexed HSP90 was associated with decreased P2X 7 receptor function and vice versa. This correlation between HSP90 tyrosine phosphorylation and P2X 7 receptor function was observed using pharmacological (geldanamycin) and molecular (site-directed mutagenesis) manipulations. If we assume that this correlation suggests a direct physiological role for HSP90 regulation of P2X 7 receptor function, it is unlikely to involve the more established role of HSP90 as a chaperone because we found no evidence to suggest altered trafficking, membrane localization of the receptor, or receptor degradation.
We did not detect any tyrosine phosphorylation of HSP90 on immunoblots of whole-cell lysates even though the HSP90 that was immunoprecipitated with the P2X 7 receptor from these same lysates consistently showed high levels of tyrosine phosphorylation (Figs. 1-3). The most likely explanation for this result is that the P2X 7 -associated HSP90 composes a small fraction of the total cellular content of this protein, with only the complexed HSP90 being tyrosine-phosphorylated. The immunoprecipitation procedure would be expected to select and so concentrate this fraction of the protein, thus providing a detectable amount of tyrosine-phosphorylated protein for Ab PY20. The only previous report of tyrosine phosphorylation of HSP90 was also obtained with an immunoprecipitated complex comprising nitric-oxide synthase-Akt and tyrosine-phosphorylated HSP90, although no comparison with the total cellular HSP90 phosphorylation state was made (20). Because HSP90 is ubiquitously abundant in cells and because tyrosine phosphorylation per se is most commonly associated with proteinprotein interactions (e.g. Refs. 34 -36), it is tempting to speculate that tyrosine phosphorylation of HSP90 is the means by which it is targeted into large protein complexes and/or targets other proteins into specific complexes. This speculation could be readily tested by measuring tyrosine phosphorylation of HSP90 in other systems where it is known to form parts of protein complexes, e.g. in the large N-methyl-D-aspartate receptor-channel complex obtained from brain neurons (34). It also remains a possibility that there is another 90-kDa protein, in addition to HSP90, that co-immunoprecipitates with the P2X 7 receptor and that it is this other 90-kDa protein that is FIG. 5. P2X 7 -Y550F receptor shows increased tyrosine-phosphorylated HSP90 in the receptor complex and decreased function. A, co-immunoprecipitation experiments performed on wildtype P2X 7 (WT) and mutant P2X 7 -Y550F receptors. The amounts of P2X 7 and HSP90 protein did not differ, but the levels of tyrosine phosphorylation of HSP90 (PY) were significantly increased, as shown in B, which plots the densitometric results obtained from six separate experiments. C, BzATP-induced membrane blebbing from HEK cells expressing wild-type P2X 7 or mutant P2X 7 -Y550F receptors. Shown is the number of cells displaying membrane blebs in response to a 2-min (for the wildtype receptor) or 5-min (for the P2X 7 -Y550F receptor) application of BzATP (30 M for the wild-type receptor and 100 M for the P2X 7 -Y550F receptor). The numbers of cells counted in each case is given in parentheses above each bar. PY-Hsp90, tyrosine-phosphorylated HSP90; GA, geldanamycin.
tyrosine-phosphorylated. However, we have been unable to obtain any evidence from mass spectroscopy analysis for a second protein running at this molecular mass; from nine peptide fragments obtained, all were consistent with HSP90␤, although none of these fragments appeared phosphorylated (9). If it is HSP90 that is tyrosine-phosphorylated, then our results do indicate compartmentalization of tyrosine-phosphorylated HSP90; this potentially has important implications in terms of interpretation of functional results obtained in studies with geldanamycin, which presumably disrupts heat shock protein function non-selectively.
Geldanamycin resulted in a consistent significant decrease in immunoprecipitated HSP90 phosphotyrosine levels without a detectable change in the HSP90 protein itself (Figs. 2 and 3). In the wild-type P2X 7 receptor, dephosphorylation of HSP90 with geldanamycin led to a small but significant 2-fold increase in the affinity of the receptor for BzATP (i.e. within 10 min). Prolonged treatment with geldanamycin (hours) resulted in activation of P2X 7 receptors and consequent membrane blebbing; activation of these receptors was due to release of ATP from the HEK cells because membrane blebbing was inhibited when ATP was broken down by co-application of apyrase. We also detected an increased release of ATP specifically from P2X 7 -expressing cells during short-term treatment with geldanamycin. It appears difficult to reconcile the small increase in receptor affinity and the increased, but still submicromolar concentrations of ATP released into the medium with the observed P2X 7 receptor-dependent membrane blebbing because receptor activation sufficient to cause blebbing would be expected only at very high micromolar amounts (3,4). However, we observed that P2X 7 receptor activation was dependent not only on agonist concentration, but also on duration of agonist application, 2 and several studies have shown that changes in external ion concentration (25, 40 -42) or repeated brief application of agonist (25,26) can significantly increase the affinity of the P2X 7 receptor for ATP. A high (low to submicromolar) and a low (Ͼ100 M) affinity state for ATP binding to the human P2X 7 receptor have been demonstrated (43). It therefore remains to be determined whether the concentrations of ATP as those measured in ATP secretion assays (7,33) are able to activate P2X 7 receptors if present for prolonged periods, although such experiments will be difficult given the rapid breakdown of ATP and the current lack of non-hydrolyzable P2X 7 receptor agonists.
The rather unimpressive actions of HSP inhibition by geldanamycin on P2X 7 receptor function in the wild-type re- ceptor stand in dramatic contrast to those observed in the Try 550 mutant receptor. Here, the affinity of the receptor for BzATP was decreased by ϳ15-fold and rapidly (within 10 min) reverted to wild-type values in the presence of geldanamycin (Fig. 6). It is highly unlikely that this residue contributes to the extracellular agonist-binding site given its location in the intracellular C terminus; a conformational change in the protein or protein complex is more likely. This mutant receptor was also associated with significantly increased tyrosine phosphorylation of HSP90, although the levels of neither the immunoprecipitated P2X 7 receptor nor the co-immunoprecipitated HSP90 were altered, nor was the tyrosine phosphorylation of the P2X 7 receptor itself (Fig. 5) (9). How then might this tyrosine residue in the C terminus of the P2X 7 receptor be involved with HSP90 tyrosine phosphorylation and/or association? Further studies will be required to address this question and the mechanisms underlying these observations, but it may be naive to imagine that only a single protein (HSP90) is involved. We know that both HSP70 and HSC70 are part of the P2X 7 receptor complex (9,44), whose functions are also disrupted by geldanamycin (45). The HSC70 protein has been shown to be tyrosine-phosphorylated in T cells in response to activation by antigen (46), although our preliminary experiments have provided no evidence for altered tyrosine phosphorylation of HSC70 or HSP70 in the wild-type or mutant P2X 7 receptor. 2 Just as HSP90 couples to numerous transcription regulation and signal transduction pathways (12,13), so, too, does the P2X 7 receptor (1, 2); in many cases, the same transduction cascades are represented. For example, geldanamycin has been shown to decrease activation of caspases, NF-B, and stressactivated protein kinase/c-Jun N-terminal kinase kinases (31,47,48), whereas P2X 7 receptor activation leads to increased activation of these same processes (16,17,49). In contrast, we have observed a facilitatory action of geldanamycin in the P2X 7 receptor, although we have examined only two events in the P2X 7 signaling cascade, ionic current and membrane blebbing; these are among the earliest cell processes associated with this receptor (1). For these rapidly induced processes, increased tyrosine phosphorylation of P2X 7 receptor-complexed HSP90 was associated with decreased currents and blebbing. It remains to be determined whether HSP90, and particularly tyrosine phosphorylation of HSP90, is associated with more downstream events in the myriad signaling pathways of this receptor. Nevertheless, our present results are compatible with a non-chaperone role for tyrosine-phosphorylated HSP90 as a negative repressor of P2X 7 receptor function.