Epithelial membrane proteins induce membrane blebbing and interact with the P2X7 receptor C terminus.

The binding of extracellular ATP to the P2X(7) receptor opens an integral cation-permeable channel; it also leads to membrane blebbing and, in certain immune cells, interleukin-1beta secretion and eventual death. The latter three effects are unique to the P2X(7) receptor; also unique among P2X receptors is the long intracellular C terminus of the protein. We have shown that the C-terminal domain of the P2X(7) receptor is responsible for the cell blebbing phenotype. A screen for proteins that associate with the C-terminal domain of the P2X(7) receptor and might mediate the blebbing phenotype, identified epithelial membrane protein 2 (EMP-2). The interaction between EMP-2 and P2X(7) was confirmed biochemically by co-immunoprecipitation, co-purification, and glutathione S-transferase pull-down assays, and this interaction was entirely dependent on the C-terminal domain of P2X(7). The P2X(7) receptor also interacted with the other members of the epithelial membrane protein family (EMP-1, EMP-3, and PMP-22). All four EMPs were found to be expressed in HEK-293 cells and in THP-1 monocytes, which express P2X(7) receptors. Interestingly, the constitutive overexpression of any of the EMPs in HEK-293 cells led to cell blebbing, annexin V binding, and cell death, by a caspase-dependent pathway. These findings suggest that the P2X(7) C-terminal domain associates with EMPs, and this interaction may mediate some aspects of the downstream signaling following P2X(7) receptor activation.

P2X 7 receptors belong to a family of ion channels gated by extracellular ATP (1), all with the same predicted topology of two transmembrane domains and intracellular N and C termini (1). The P2X 7 receptor shares 40 -45% amino acid identity with the other P2X proteins, but it is structurally distinct at the C terminus, extending for an additional 100 -200 amino acids (2). P2X [1][2][3][4][5][6] receptors are widely distributed in both neuronal and non-neuronal cells, whereas P2X 7 receptors are most highly expressed in immune and epithelial cells (3,4). Brief stimulation (10 -30 s) of the P2X 7 receptor leads to the formation of a channel permeable to large cations, as is also seen for some of the other P2X receptors (2,5). However, more prolonged activation of the P2X 7 receptor results in extensive membrane blebbing from within seconds to minutes (5), and eventual cell death (5)(6)(7), although such responses are not observed with other P2X receptors. We hypothesized that the C terminus of the P2X 7 receptor might engage other cellular proteins to mediate its distinct responses of membrane blebbing and cell death. We tested this hypothesis by using a construct containing the soluble C-terminal domain (Asn 356 -Tyr 595 ) of the rat P2X 7 protein as bait in a yeast two-hybrid screen. Because HEK-293 cells expressing P2X 7 receptors exhibit pronounced blebbing and eventual cell death when stimulated with ATP (5),we sought interacting proteins using a HEK-293 cell library.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The full-length cDNA encoding epithelial membrane protein EMP-2 1 was isolated using PCR from the pACT-2 library cDNA and subcloned into pcDNA3.1ϩmyc/HisA vector, hence incorporating a Myc epitope and His tag at the C terminus of the expressed protein.
Cell Imaging-Fluorescence was measured using a Zeiss Axiovert 100 and Fluar 20x objective with Photonics monochrometer imaging (Photonics, Graefelfing, Germany). HEK-293 cells were loaded for 40 min at 37°C with 2 M Fluo-4AM (Molecular Probes) and 0.02% Pluronic (Molecular Probes). Images were captured at 0.5 Hz. Calcium concentrations were calibrated with standards, where free Ca 2ϩ was calculated using the EQCAL program, in 500 nM ionomycin to allow free exchange between cells and the extracellular medium.
Yeast Two-hybrid Screen-A bait construct was made (encoding Asn 356 to Tyr 595 of rat P2X 7 ) by amplification of the rat P2X 7 cDNA (primers: 5Ј-GCCGGGAATTCAACACGTATGCCAGTACCTGC-3Ј and 5Ј-GCCGGCTCGAGTCAGTAGGGATACTTGAAG CC-3Ј). The PCR product was subcloned into pAS2-1 using EcoRI and SalI (XhoI on PCR product) restriction sites, thereby generating a C-terminal fusion to the binding domain of Gal-4. The HEK-293 cell library (CLONTECH), in pACT2 as a fusion with the activation domain of Gal-4, was amplified according to the manufacturer's instructions (CLONTECH). The Matchmaker Two-hybrid System (CLONTECH) was used for screening. ␤-Galactosidase-activating positive clones were isolated, sequenced, and reconfirmed by co-transforming with the bait plasmid into Y190 cells.
Co-purification on Affinity Resin-HEK-293 cells were co-transfected with 1 g of plasmid encoding EMP-2myc/His (or other family members) and 1 g of plasmid encoding P2X 7 EE, using LipofectAMINE 2000, according to the manufacturer's instructions (Invitrogen). After 24 -36 h, cells were harvested by washing with phosphate-buffered saline (PBS) and then solubilized in extraction buffer (50 mM Tris-HCl pH 8, 300 mM NaCl, 5 mM imidazole, 1% Nonidet P-40) for 15 min on ice. Particulate matter was removed by two centrifugations at 16,000 ϫ g for 15 min at 4°C. TALON resin (Cobalt affinity resin, CLONTECH) was washed in extraction buffer, and cell extracts were added and incubated by rotating at 4°C for 1 h. Unbound protein was removed, and the resin was washed four times in extraction buffer for 5 min per wash by rotating. The bound protein was eluted in 30 l of elution buffer (50 mM Tris-HCl, pH 8, 300 mM NaCl, 200 mM imidazole, 1% Nonidet P-40) by incubating at room temperature for 15 min. Eluted proteins were analyzed by using SDS-PAGE and transferred onto polyvinylidene difluoride membranes. Samples were tested for the presence of P2X 7 receptor using a C-terminal polyclonal antibody (Alomone). Samples from original crude cell extracts that had not been TALON-purified were tested for the presence of P2X 7 and EMPs by immunoblotting using the C-terminal polyclonal antibody (Alomone) or c-Myc antibody (9E10 clone from Santa Cruz Biotechnology), respectively. Blots were then reacted with anti-rabbit or anti-mouse horseradish peroxidaseconjugated antibodies (Vector Laboratories) and detected using the ECL-Plus system (Amersham Biosciences). All antibody incubations were carried out in PBS-Tween (0.1%) plus 2% nonfat milk, and washes of the blots were performed in PBS-Tween (0.1%).
Co-immunoprecipitation-HEK-293 cells were transfected with EMP-2myc/His (or other family members) plus P2X 7 EE as described above. Controls were used for EMP-2myc/His co-transfected with Ref 2-1EE, a 25-kDa unrelated protein (8). Cells were washed with PBS and then treated with solubilization buffer (PBS, Nonidet P-40 (1%), and Calbiochem Protease Inhibitor Mixture Set III) for 15 min at 4°C. Particulate matter was removed by two centrifugations at 16,000 ϫ g for 15 min at 4°C. For the pre-clearing step, cell extracts were added to protein G-Sepharose (Amersham Biosciences), pre-washed in solubilization buffer, and incubated by rotating at 4°C for 2 h. Unbound protein was removed and incubated with a monoclonal anti-EE antibody (Covance) for 1 h at 4°C. This reaction product was then added to a new aliquot of protein G-Sepharose, and incubated for 1-2 h by rotating at 4°C. The resin was washed four times in solubilization buffer, and bound protein was eluted in 30 l of 1ϫ SDS-PAGE loading buffer heated to 92°C for 3 min. Eluted proteins were analyzed by SDS-PAGE, and transferred onto polyvinylidene difluoride membrane. Products were tested by immunoblotting for the presence of EMP-2 (or other family members) using the c-Myc antibody, and samples of the original unpurified cell extracts were tested for the presence of P2X 7 and EMP expression by blotting with the P2X 7 C-terminal polyclonal antibody or c-Myc antibody, respectively, as detailed above.
Glutathione S-Transferase (GST) Pull-downs-The C terminus of the rat P2X 7 receptor (Asn 356 -Tyr 595 ) was made as a GST fusion construct in the pGEX5X2 vector and overexpressed in BL21(DE3) cells, by in- duction with isopropyl-␤-D-thiogalactopyranoside at 37°C. Cells were harvested, resuspended in PBS/Triton X-100 (1%) with 1 mM phenylmethylsulfonyl fluoride and 1 mM dithiothreitol , and sonicated. The soluble extract was mixed with glutathione-Sepharose (Amersham Biosciences) for 1 h at 4°C, and unbound protein was removed by four washes in PBS/Triton. The purification of GST only or GST-P2X 7 C terminus (GST-P2X 7 CT) was confirmed by eluting an extract with 40 mM reduced glutathione and SDS-PAGE analysis. The GST and GST-P2X 7 CT protein attached to glutathione-Sepharose were mixed with extracts of HEK-293 cells overexpressing each EMP-2myc/His protein or control extracts (mock-transfected HEK-293 or cells transfected with Ref 2-1myc) for 1 h at 4°C by rotating. Glutathione-Sepharose was washed four times in solubilization buffer to remove nonspecifically bound protein. Subsequently, bound protein was removed by elution with 40 mM reduced glutathione. Eluates were analyzed by SDS-PAGE, transferred onto polyvinylidene difluoride membranes, and immunoblotted for the presence of EMPs using the c-Myc antibody. Original HEK-293 extracts were also tested for the presence of EMPs by c-Myc immunoblotting.
Cell Viability Assay-Supernatants from cultured HEK-293 cells were taken at various time points after transfections. At 12 h posttransfection, cell samples (including negative controls) were incubated in low serum (1% fetal calf serum), with or without z-VAD caspase inhibitor (30 M in 0.5% Me 2 SO, Calbiochem) or 0.5% Me 2 SO control.
Supernatants were analyzed for lactose dehydrogenase (LDH) activity using the LDH-optimized kit (Sigma) and compared with control extracts treated with 1% Triton X-100 as the maximal activity. At 48 h after treatment, cells were harvested and cell viability was determined by trypan blue exclusion.
Annexin V Binding-Phosphatidylserine exposure on the extracellular leaflet of the membrane was measured by annexin V binding (9). HEK-293 cells were transfected as described, with plasmids encoding EMP-1, EMP-2, EMP-3, PMP-22, and P2X 2 proteins. Cells were washed and provided with fresh medium 24 h post-transfection and harvested at 48 h. The binding of annexin V conjugated to fluorescein isothiocyanate (Molecular Probes) was measured by using flow cytometry. Cells were also gated by propidium iodide, used in this case as a marker of non-viable, necrotic cells. The number of cells binding annexin but not taking up propidium was normalized to that observed with staurosporine (1 M), which is well known to stimulate apoptosis in HEK-293 cells (10).
Immunocytochemistry-HEK-293 cells were washed in PBS, fixed in Zamboni's fixative (20 min), then blocked in 5% goat serum in PBS with 2% Triton X-100 (30 min), and incubated with primary antibody (anti c-Myc, anti-EE, or anti-P2X 7 ectodomain (11)) for 1-2 h at room temperature, or overnight at 4°C. Cells were washed, and secondary antibody was added for 1 h at room temperature. Other product sizes were as expected. B, message for PMP-22, EMP-1, EMP-2, and EMP-3 was detected by RT-PCR of THP-1 mRNA, using the same methods as for A. C, overexpression of four EMPs (with myc/His tags) with P2X 7 in HEK-293 cells was followed by purification using a cobalt-affinity resin to bind the His tag. This resulted in the copurification of P2X 7 with EMP-2, EMP-1, EMP-3, and PMP-22.

RESULTS
Previous studies on the P2X 7 receptor have shown that truncation of the receptor by removal of most of the intracellular C terminus reduces YOPRO-1 dye uptake and channel dilatation (2). We have now verified that the C terminus is required for the P2X 7 receptor-mediated cell blebbing. In HEK-293 cells expressing the wild-type (WT) rP2X 7 receptor, cells underwent membrane disruption within 1 min of activation with 100 M Bz-ATP, whereas cells transfected with the truncated (⌬418) rP2X 7 receptor did not show any significant membrane disruption over a 4-min stimulation with 100 M Bz-ATP (Fig. 1,  A-C). Both the WT rP2X 7 receptor and the truncated receptor were expressed on the cell surface as determined by immunocytochemistry (Fig. 1A). In agreement with previous electrophysiological recordings (2), the ⌬418 truncated rP2X 7 receptor was shown to form a functional channel, because cells loaded with the calcium indicator Fluo-4AM displayed a prolonged Ca 2ϩ rise after stimulation with 100 M Bz-ATP (Fig. 1B). Therefore, the P2X 7 receptor C terminus is required for ATPmediated cell blebbing. For this reason we used this intracellular domain to probe for interacting proteins, which might mediate this response.
EMP-2 was identified as a strong positive hit in the yeast two-hybrid screen of a HEK-293 cell cDNA library with the P2X 7 receptor C terminus (P2X 7 CT). The interaction was confirmed by isolation of the EMP-2 Gal-4 activation domain fusion plasmid, retransformation into Y190 cells and assaying for growth on His Ϫ plates, and ␤-galactosidase activity. The interaction between full-length P2X 7 receptor and EMP-2 was further tested using biochemical approaches. EMP-2myc/His was cotransfected with the rP2X 7 receptor into HEK-293 cells; P2X 7 receptor was copurified from cell extracts with EMP-2 attached to a cobalt affinity resin (Fig. 2A, lane 5). In the absence of EMP-2myc/His, only a low background level of rP2X 7 bound the cobalt affinity resin (Fig. 2A, lane 4). A complementary experiment was performed whereby EMP-2myc/His was coimmunoprecipitated with P2X 7 EE using an EE monoclonal antibody (Fig. 2B). No EMP-2 was immunoprecipitated in the absence of P2X 7 EE (Fig. 2B, lane 2), nor did EMP-2 co-immunoprecipitate with an EE-tagged mRNA export factor, Ref 2-1EE (Fig. 2B, lane 3).
EMP-2 is a member of a small family of epithelial membrane proteins, which also includes PMP-22, EMP-1, and EMP-3. RT-PCR using primers specific for each family member on extracts of HEK-293 cells indicated the presence of all four mRNAs (Fig. 3A). Contamination by genomic DNA is unlikely because primers to ␤-actin (that span a short intron) gave a product of the expected size of the cDNA, and all EMP primers were chosen to span introns in the genomic sequences. RT-PCR products were also confirmed by sequencing. The RT-PCR was repeated on extracts from THP-1 monocytes, and mRNAs encoding all four EMP members were detected (Fig. 3B). Each of the family members (Myc/His-tagged) were co-expressed in HEK-293 cells with the P2X 7 receptor. The P2X 7 receptor copurified with each of the His-tagged proteins on the cobalt affinity resin (Fig. 3C).
We also sought to confirm directly an interaction between the C terminus of the P2X 7 receptor and the EMPs. Extracts were prepared from HEK-293 cells transfected with PMP-22, EMP-1, EMP-2, EMP-3, and Ref-2-1 (each Myc/His-tagged). Expression of the proteins was confirmed by direct analysis of samples of the extracts by immunoblotting with the Myc antibody (Fig. 4A). Several bands were observed for the Myc-tagged EMPs because they exhibit different glycosylation states (12). Extracts were incubated either with GST (control) or GST-P2X 7 CT. The PMP-22, EMP-1, EMP-2, and EMP-3 proteins were pulled down by GST-P2X 7 CT although not by GST alone (Fig. 4B). No pull-down of any Myc-labeled products in the presence of GST-P2X 7 CT was observed using control extracts of HEK-293 mock-transfected or Ref 2-1myc-transfected cells. A deletion mutant of the P2X 7 receptor lacking the C terminus (to residue N356, ⌬CT-P2X7EE) was not capable of co-immunoprecipitating EMP-2 (Fig. 4C). The truncated receptor was expressed at the cell surface in a similar pattern to the WT P2X 7 receptor (Fig. 4D). This indicates that EMPs specifically interact with the C terminus of the P2X 7 receptor.
Overexpression of PMP-22 and EMP-2 in NIH-3T3 cells has previously been shown to cause cell rounding, blebbing, and a reduction in survival rate (13)(14)(15). We therefore sought to determine whether overexpression of all four members of the family mediate similar effects in HEK-293 cells. Overexpression of each protein led to a significant increase in cell death. A significantly higher release of LDH occurred in EMP-transfected cells after 36 h of treatment in low serum (Fig. 5A). In addition, an increase in trypan blue uptake was observed as determined by cell counting (Fig. 5B), compared with control mock-transfected cells also maintained in low serum. Fluorescence-activated cell sorter analysis of HEK-293 cells 48 h after transfection showed that each member of the EMP family resulted in a significant proportion of annexin-positive cells (Fig. 5, C and D). Mock-transfected cells and cells transfected with P2X 2 receptor cDNAs showed no significant annexin binding. The high percentage of propidium-positive cells in the mock-transfected control (Fig. 5C) is likely caused by the addition of transfection reagent. The LDH release and trypan blue uptake in EMP-overexpressing cells was significantly inhibited by the addition of the caspase inhibitor, z-VAD (Fig.  5, A and B).
Immunocytochemistry of HEK-293 cells transfected with each of the EMP cDNAs (Myc/His tagged) showed predominant staining at the cell surface (Fig. 6). All cells transfected with PMP-22, EMP-1, EMP-2, or EMP-3 exhibited several large membrane blebs on their surfaces (Fig. 6). DISCUSSION The P2X 7 receptor is functionally distinct compared with the other members of the ATP-gated ion channel family because its activation leads to membrane blebbing (5). Here we have demonstrated that the unique intracellular C terminus of the P2X 7 receptor is required for this response. The removal of the C-terminal domain of P2X 7 also greatly reduces the uptake of dyes such as YOPRO-1, but it does not grossly alter the ATPgated cation channel function (2).
The search for proteins that interact with the C-terminal domain of P2X 7 , and may be involved in the membrane-blebbing phenotype, led to the identification of EMP-2. EMP-2 consists of 167 amino acids containing four predicted transmembrane domains (16) and exhibits 40% amino acid identity to PMP-22. In addition, two other related proteins have been identified by random sequencing of a mouse intestine library (EMP-1 (17)) and from related expressed sequence tag (EST) sequences (EMP-3 (16)). Direct interaction of EMP-2 and the C terminus of the P2X 7 receptor was verified by using biochemical approaches-copurification, coimmunoprecipitation, and GST pull-down assays. In addition, the related proteins EMP-1, EMP-3, and PMP-22 were also found to interact with the P2X 7 CT using these same techniques. Notably these interactions did not depend on activation of the P2X 7 receptor. Consistent with a role in the membrane-blebbing phenotype of P2X 7 and not its channel functions, overexpression of EMP-2 in HEK-293 cells had no significant effect on P2X 7 -activated cation currents, YOPRO-1 uptake, or a significant shift in the EC 50 value for Bz-ATP (data not shown). P2X 7 has previously been found to associate with a multiprotein complex including cytoskeletal proteins, heat shock proteins, and two integral membrane proteins, receptor-phosphotyrosine phosphatase-␤ (RPTP␤) and integrin ␤2, although the domains of the P2X 7 receptor responsible for most of these interactions have not yet been determined (18). The C terminus of the P2X 7 receptor is generally thought to be entirely intra- FIG.7. EMPs share homology to the ␥ subunits of voltage-activated Ca 2؉ channels. Alignment of the human five exon ␥5 and ␥7 to human PMP-22, EMP-1, EMP-2, and EMP-3 using the ClustalW algorithm. Bars above the sequence represent predicted transmembrane spanning domains; solid triangles (above sequence) show exon-intron boundaries for PMP-22, EMP-1, EMP-2, and EMP-3; open triangles (below sequence) show exon-intron boundaries for ␥5 and ␥7 subunits. The lower line provides the consensus sequence for the alignment. cellular, whereas EMPs have a predominantly intramembrane disposition with only a small part of the protein likely to be intracellular (amino acids 86 -94 in EMP-2), although this domain may be sufficient for the observed interactions. Alternatively, there is a significantly hydrophobic segment in the P2X 7 receptor C terminus (amino acids 512-536), raising the possibility that the main interaction occurs between hydrophobic regions within the membrane.
The EMPs share 22-25% similarity to the "stargazin" or voltage-gated channel ã subunits, with common exon-intron boundaries and alignment of the four transmembrane domains (Fig. 7). This is of interest, as the ␥ subunits (␥1-␥8) have been shown to interact with neuronal voltage-dependent calcium channels (19,20), indicating that the P2X 7 receptor-EMP interaction is not unique, and a number of ion channels may form complexes with tetra-span membrane proteins.
The normal physiological role of EMPs is not well understood. Genetic studies and the generation of PMP-22-deficient mice have established that it is responsible for a set of inherited peripheral neuropathies in mice and humans. One type of Charcot-Marie Tooth disease (CMT1A) results from a mutation in PMP-22 (21), and Dejerine-Sottas syndrome, which also involves impaired nerve conductions, is a product of a deletion within the PMP-22 gene (22). In rodents, point mutations of PMP-22 are associated with the "Trembler" phenotype due to impaired conduction in peripheral nerves (23). However, the wide tissue distribution of PMP-22 suggests that it may play a more general role than in myelin formation. PMP-22 was originally isolated from NIH-3T3 cells and was first named gas-3 for its growth arrest-dependent expression (24). Northern analysis has shown that each member of the EMP family, including PMP-22, is widely expressed (16,24), and there is considerable overlap with the tissue distribution of P2X 7 receptors. For example, many epithelial cells express both EMPs and P2X 7 receptors (4). PMP-22 is highly expressed in Schwann cells (25,26), where there is also good functional evidence for P2X 7 receptors (27,28).
The RT-PCR analysis confirmed that mRNAs encoding all members of the EMP family are present in both HEK-293 cells and in THP-1 monocytes. P2X 7 receptors are strongly expressed in THP-1 monocytes, and their activation is an important trigger for interleukin-1␤ release (29). After prolonged stimulation, P2X 7 receptors mediate ATP-induced killing in human macrophages (30,31). In NIH-3T3 cells, overexpression of PMP-22 also causes cell rounding, blebbing, and a reduction in the survival rate (13,14), and recently bleb formation has been reported following overexpression of EMP-2 (15). We have now extended these observations, and shown that membrane blebbing is induced and cell survival is decreased when any one of the EMPs is overexpressed in HEK-293 cells, as evidenced by increased LDH release, trypan blue uptake, annexin V binding, and immunofluorescence. The cell death could be inhibited by the addition of the caspase inhibitor, z-VAD, indicating that overexpression of EMP-1, EMP-2, EMP-3, or PMP-22 results in a caspase-dependent apoptotic-like phenotype.
Although overexpression of EMPs leads to constitutive cell blebbing, the normal cellular trigger for an EMP to manifest this response is currently unknown. This study indicates that EMPs can associate with other integral membrane proteins such as the P2X 7 receptor, which under the appropriate condi-tions (i.e. ATP stimulation), can provide that trigger and promote blebbing, although the mechanism by which the P2X 7 -EMP complex could promote cell blebbing is not understood at present. Because EMPs are widely expressed, even in cells lacking P2X 7 , this would suggest they may have additional integral membrane protein partners that link EMP-like proteins to apoptosis pathways. Identification of such proteins will be an important goal for the future together with understanding the mechanisms involved in the constitutive cell blebbing observed when EMP proteins are overexpressed.