Fragmented Inositol 1,4,5-Trisphosphate Receptors Retain Tetrameric Architecture and Form Functional Ca2+ Release Channels*

Background: Proteolytic cleavage and disruption of inositol 1,4,5-trisphosphate receptor (IP3R) architecture may contribute to the initiation/progression of apoptosis. Results: Proteolytic cleavage products of the IP3R remain membrane-associated, and these fragments form functional tetrameric channels. Conclusion: Fragmentation of the IP3R does not inevitably lead to loss of function. Significance: Peptide continuity is not required for IP3R function, and IP3R activation may persist after protease cleavage during apoptosis. Inositol 1,4,5-trisphosphate receptor isoforms are a family of ubiquitously expressed ligand-gated channels encoded by three individual genes. The proteins are localized to membranes of intracellular Ca2+ stores and play pivotal roles in Ca2+ homeostasis. Previous studies have demonstrated that IP3R1 is cleaved by the intracellular proteases calpain and caspase both in vivo and in vitro. However, the resultant cleavage products are poorly defined, and the functional consequences of these proteolytic events are not fully understood. We demonstrate that IP3R1 is cleaved during staurosporine-induced apoptosis, yielding N-terminal fragments encompassing the ligand-binding domain and the majority of the central modulatory domain together with a C-terminal fragment containing the channel domain and cytosolic tail. Notably, these fragments remain associated with the membrane after initiation of apoptotic cleavage. Furthermore, when recombinant IP3R1 fragments, corresponding to those predicted to be generated by caspase or calpain cleavage, are stably coexpressed in cells, they physically associate and form functional channels. These data provide novel insights regarding the regulation of IP3R1 during proteolysis and provide direct evidence that polypeptide continuity is not required for IP3R activation and Ca2+ release.

In addition to the well defined roles of IP 3 Rs in mediating physiological Ca 2ϩ signaling events, burgeoning recent evidence suggests that dysregulated Ca 2ϩ signaling contributes to the pathogenesis of numerous disease states (10,11). A prominent example is the role of intracellular Ca 2ϩ elevations in contributing to programmed cell death, or apoptosis (12)(13)(14). Apoptosis is an evolutionary conserved process vital for normal development and is also implicated in the etiology of many human diseases. Notably, many current cellular models of apoptosis posit that prolonged and sustained elevations of the intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i )can be an important event for the initiation and/or progression of the apoptotic cascade (14,15). Moreover, numerous reports have suggested that altered IP 3 R activity is involved in apoptosis in a variety of cell types (reviewed in Ref. 16). For example, initial reports showed that IP 3 Rs are required to promote apoptosis. However, no consensus exists as to whether particular IP 3 R isoforms are specifically required or, alternatively, whether individual IP 3 R isoforms can function in a redundant fashion (17)(18)(19)(20)(21). Several studies have shown that during apoptosis, IP 3 R1 is proteolytically cleaved in a caspase-dependent manner (22)(23)(24)(25). These studies further suggested that cleavage is crucial to mediating the dysregulated Ca 2ϩ signaling events (24). Specifically, studies have shown that cleavage yields multiple fragments detected by antibodies recognizing the C terminus and, therefore, suggest that caspase cleaves at several sites (23). The emerging paradigm from these studies is that proteolytic cleavage occurs at the junction between the central modulatory domain and the channel domain. In this scenario, cleavage uncouples the N-terminal ligand-binding domain from the channel domain and, thus, terminates IP 3 -mediated Ca 2ϩ release (22,26). In addition, on the basis of the observation that transient expression of the C-terminal 95-kDa channel fragment resulted in the formation of constitutively open leaky channels, it is thought that a second major consequence of IP 3 R1 cleavage is the formation of IP 3 -independent Ca 2ϩ release channels, which may lead to depletion of the ER Ca 2ϩ store (22,26,27). Despite these reports questions remain. For example, what is the fate of the expected N-terminal fragment (16)? It is also not clear whether all of the subunits within individual tetrameric channels are cleaved during caspase proteolysis of IP 3 R1. Moreover, it is uncertain whether cleaved subunits within a single tetrameric channel are selectively and individually removed and dislocated from the ER membrane or, alternatively, whether the proteolyzed tetrameric channel remains assembled. Finally, whether cleaved subunit(s) "disable" the entire channel and how proteolyzed IP 3 R1 contributes to Ca 2ϩ signaling are still open questions.
To begin to address these issues, in this study, we show that IP 3 R1 is fragmented into N-terminal and C-terminal fragments in response to staurosporine treatment. Notably, however, we provide evidence that suggests that, during proteolysis, the fragments are not extracted from intracellular membranes and, surprisingly, that tetrameric IP 3 R1 channels, although cleaved, do not necessarily disassemble during apoptosis. In addition, supporting this idea, we show that stable expression of the N-terminal and C-terminal fragments, expected to result from proteolytic cleavage, associate to form tetrameric channels that are functionally competent to be gated by IP 3 . These data provide insight into the role played by IP 3 R1 during apoptosis and suggest that cleavage of IP 3 R1 might not necessarily abrogate IP 3 -induced Ca 2ϩ release or immediately generate constitutively leaky channels. Rather, it may serve to modify patterns of Ca 2ϩ signals required for driving the apoptotic cascade into completion. Finally, our data provide unequivocal evidence that peptide continuity is not required to gate IP 3 R following binding of IP 3 .

MATERIALS AND METHODS
Reagents-RPMI 1640 medium, DMEM, penicillin/streptomycin, G418 sulfate, ␤-mercaptoethanol, chicken serum, and all reagents used for Native Blue PAGE were purchased from Invitrogen. Fetal bovine serum was from Gemini Bio-products. Staurosporine was from Enzo Life Sciences. Protein A/G Plusagarose beads were from Santa Cruz Biotechnology, Inc. CHAPS was from G Biosciences. Dylight TM 800CW secondary antibodies and enhanced chemiluminescent substrate were obtained from Thermo Scientific. Horseradish peroxidaseconjugated secondary antibodies and reagents used for SDS-PAGE were obtained from Bio-Rad. Cyclopiazonic acid was from Tocris Bioscience. Fura-2AM was from TEFLABS. DNAmodifying enzymes were from New England Biolabs. Unless indicated otherwise, all other chemicals were purchased from Sigma.
Antibodies-␣-IP 3 R1 antibodies used were as follows, and the epitopes are depicted in Fig. 1A. Rabbit polyclonal antibody (CT-1) raised against the most carboxyl-terminal 19 amino acid (aa) residues was a gift from Dr. Richard Wojcikiewicz (Upstate Medical University). Mouse monoclonal antibody (UCD) directed against aa residues 2680 -2749 was purchased from UC Davis/NIH NeuroMab. Rabbit monoclonal antibody (CS) against an epitope centered around aa 40 in human IP 3 R1 was from Cell Signaling Technologies, Inc. Rabbit polyclonal antibody (SC) against aa residues 1894 -1973 in human IP 3 R-1 was obtained from Santa Cruz Biotechnology, Inc. Rabbit ␣-IP 3 R2 antisera were generated by Pocono Rabbit Farms and Laboratories. CT2 was raised against the extreme carboxyl aa 2686 -2701, and NT2 was raised against aa 320 -338 ( Fig. 1D) (28). Rabbit polyclonal anti-sarcoendoplasmic reticulum ATPase and GAPDH were obtained from Santa Cruz Biotechnology, Inc., and the anti-PKD1; Protein Kinase D1 antibody was from Cell Signaling Technology, Inc.
Cell Transfection and Culture Conditions-DT40 -3ko cells, a chicken B lymphocyte line with targeted deletion of the three endogenous IP 3 R isoforms (18), were grown in RPMI 1640 medium supplemented with 1% chicken serum, 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 39°C with 5% CO 2 and subcultured every 3 days. DT40 -3ko cell transfection and generation of stable cell lines was performed as described previously using the Amaxa nucleofector (Lonza Laboratories) (28,29). HEK293 cells were maintained in DMEM supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C with 5% CO 2 . HEK293 cells were transfected using Lipofectamine 2000 (Invitrogen) following the protocol of the manufacturer.
Plasmid Construction-To create truncation fragments of IP 3 R1 and IP 3 R2, cDNAs encoding rat IP 3 R1 or mouse IP 3 R2 flanked by the NheI and NotI sites at the 5Ј and 3Ј ends in pcDNA3.1ϩ, respectively, were used as templates. All PCR modifications described here were conducted using Pfu Ultra II Hotstart 2ϫ Master Mix (Agilent) and pairs of complementary primers that contained the NheI and NotI sites (underlined in the text below) to facilitate subcloning. Only forward primers are shown here. To generate constructs coding for N-terminal and C-terminal fragments predicted to result from caspase cleavage of IP 3 R1 at the DEVD1891 consensus site (23,24) (Fig.  1B), IP 3 R1 cDNA was modified by PCR (forward, 5Ј-GAAAG-ATGATGAAGTGGACTAGAATTCGCGGCCGCGCTAGCAT-GCGGGATGCCCCATCCCGAA 3Ј). This modification introduced a stop codon after residue Asp-1891 and also a Kozak sequence and an initiation methionine in-frame with the sequence coding for the membrane fragment, designed to ensure efficient expression. For the creation of the caspase cytosolic fragment, R1 casp sol, the plasmid was digested with the NotI enzyme to delete the sequence coding for the membrane fragment, and the remaining construct was self-ligated. To generate the membrane construct, R1 casp mem, the cytosolic component was deleted from the plasmid by digestion with NheI followed by self-ligation. To coexpress both fragments, we originally utilized the pIRES (Clontech) expression vector. However, initial experiments using this vector resulted in poor expression from the second cloning site. Therefore, we assembled a new vector with two expression cassettes derived from pcDNA3.1ϩ (Invitrogen) and pIRES as follows. The IRES sequence in pIRES was deleted by digesting the plasmid with NheI and XbaI and then ligating the NheI-and XbaI-compatible ends. This vector was modified further (forward, 5Ј-CTA-CACTTGCCAGATCTCTAGCGCCCGCTC-3Ј) to introduce a new BglII site just downstream of the polyadenylation signal (underlined is the BglII site). The modified construct was then digested with BglII, and the BglII fragment encompassing the cytomegalovirus (CMV) promoter, cloning site, and polyadenylation signal was subcloned into the BglII site in pcDNA3.1ϩ, resulting in a two-promoter vector (TPV) with two independent expression cassettes. To obtain a TPV encoding both N-and C-terminal fragments, the NheI-NotI cytosolic fragment was inserted into the TPV digested with NheI and PspOMI (Not1 and PspOMI have compatible cohesive ends). The NotI-NotI fragment coding for R1 casp mem was inserted into the TPV that was digested with NotI. IP 3 R1 fragments corresponding to cytosolic (R1 calp sol) and membrane (R1 calp mem) fragments, predicted to result from calpain cleavage rIP 3 R1 after glutamic acid 1917 (30) (Fig. 1C), were constructed in an identical manner using 5Ј-CCGGGATCAGCTCTTGGAATAGAATTCG-CGGCCGCGCTAGCATGGCATCTGCTGCCACCAGAAA-AGCC-3Ј, which introduced a stop codon after glutamic acid 1917 and a Kozak sequence and initiation methionine in-frame with the membrane fragment to ensure efficient expression. A TPV plasmid encoding both fragments (R1 calp sol/mem) was generated as described above. The same PCR and subcloning strategies were followed to make other plasmid constructs. To make cDNA construct encoding tryptic fragments of IϩII (aa residues 1-922) and IIIϩIVϩV (aa residues 923-2749) of IP 3 R1 (31), PCR was performed with 5Ј-GGCAGCAACGTG-ATGAGATAGGCGGCCGCGCTAGCATGTCTATCCATG-GAGTTGG-3Ј (forward), which introduced a stop codon after arginine 922. To introduce a stop codon after arginine1582 and to generate a construct encoding tryptic fragments IϩIIϩIII (aa residues 1-1582) and IVϩV (aa residues 1583-2749), 5Ј-CTGGCGG-TTATCAGCCCGCTAGGCGGCCGCGCTAGCATGAACG-CTGCTCGTAGAGAC-3Ј (forward) was used. These primer pairs also contained the NotI and NheI sites and introduced a Kozak sequence and initiation methionine in-frame with the remainder of the coding sequence to ensure efficient expression. To create fragmented IP 3 R2 (R2-frags) (Fig. 1D), the following primer was used: 5Ј-GGAATGAAAGGGCAGTTAAC-AGAATAGAATTCGCGGCCGCGCTAGCATGGCGTCTT-CAGCCACATCC-3Ј (forward). This primer introduced a stop codon after aa 1869. The NheI-NotI cytosolic fragment was inserted into the TPV that was digested with NheI and PspOMI. The NotI-NotI fragment coding for the membrane region was inserted into the TPV that was digested with NotI.
Cell Lysis-Following treatment, cells were harvested by centrifugation, washed once with ice-cold PBS, and solubilized in cell lysis buffer containing 10 mM Tris-HCl, 10 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM NaF, 20 mM Na 4 P 2 O 7 , 2 mM Na 3 VO 4 , 1% Triton X-100 (v/v), 0.5% sodium deoxycholate (w/v), and 10% glycerol with a mixture of protease inhibitors (Roche). After 30 min on ice, cell lysates were precleared by centrifuga-tion at 16,000 ϫ g for 10 min at 4°C. Cleared lysates were transferred into fresh tubes, and protein concentrations were measured using a D C protein assay kit (Bio-Rad). Proteins were resolved on 5-7.5% SDS-PAGE gels, transferred to nitrocellulose, and processed for immunoblotting with the indicated primary antibodies and corresponding secondary antibodies. Proteins were detected with chemiluminescence or using an Odyssey infrared imaging system (LI-COR Biosciences).
Coimmunoprecipitation-Cells were harvested, washed once with ice-cold PBS, and lysed in Igepal lysis buffer containing 50 mM Tris-HCl, 120 mM NaCl, 0.5% Igepal (v/v), and 1 mM EDTA (pH 8) with protease inhibitors. After 30 min on ice, cell lysates were precleared by centrifugation at 16,000 ϫ g for 10 min at 4°C. Cleared supernatants were rocked with the indicated antibodies for 3 h in the cold room followed by the addition of 30 l of protein A/G slurry for an additional 2 h. Immunocomplexes were washed five times, resuspended in gel loading buffer, and processed on SDS-PAGE for immunoblotting.
Subcellular Fractionation-After harvesting and washing with ice-cold PBS, control or staurosporine-treated cells were resuspended in homogenization buffer containing 20 mM HEPES, 5 mM NaN 3 , 0.5 mM EGTA, and 320 mM sucrose (pH 7.4) supplemented with protease inhibitors. Cells were then homogenized using a Teflon glass homogenizer. Homogenates were cleared by centrifugation at 1000 ϫ g for 10 min at 4°C. The resulting supernatants were centrifuged at 100,000 ϫ g at 4°C for 1 h. The supernatants designated as the cytosolic fraction were removed, and the microsomal pellet was resuspended in lysis buffer. Equivalent amounts of proteins were fractionated and processed for immunoblot analyses with the indicated antibodies.
Native Blue PAGE Analysis-DT40 cells expressing various constructs were harvested by centrifugation and washed two times in ice-cold PBS. Cells were lysed in 100 l of lysis buffer containing 10 mM CHAPS, 25 mM HEPES, 40 mM NaCl, and 1 mM EDTA (pH 7.4) and supplemented with protease inhibitors. After 20 min on ice, cell lysates were centrifuged at 16,000 ϫ g for 10 min at 4°C. 70 l of cleared lysates were mixed with 5 l of 5% G-250 sample additive and 25 l of 4ϫ sample buffer. Protein samples were then resolved on 3-12% NativePAGE TM Novex using dark cathode buffer at 150 V for 1 h and then at 250 V for 1.5 h using light cathode buffer. Proteins were then transferred to a polyvinylidene difluoride membrane and immunoblotted with the indicated antibodies.
[Ca 2ϩ ] i Measurement-Ca 2ϩ imaging was performed as described before (29,32,33). Briefly, DT40 cells expressing the desired constructs were washed with imaging buffer containing 10 mM HEPES, 5.5 mM glucose, 1.26 mM Ca 2ϩ , 137 mM NaCl, 0.56 mM MgCl 2 , 4.7 mM KCl, and 1 mM Na 2 HPO 4 (pH 7.4) and then loaded with 2 M Fura-2AM on a coverslip held to a perfusion chamber at room temperature for 20 min to allow for cell attachment to the coverslip. Loaded cells were then perfused with imaging buffer, and image acquisition was performed using a Till Photonics imaging system. Experiments were repeated at least three times. Data are presented as mean Ϯ S.E.
Cell Viability Assay-100,000 cells/well were seeded in 24-well plates in 0.4 ml of fresh medium and were allowed to recover for 2-3 h. Cells were then either treated as controls or treated with 15 g/ml anti-chicken IgM (Southern Biotech) for 17 h. Afterward, 50 l of CellTiter 96 AQ ueous One Solution (Promega) was added to each well, and 1 h later, absorbance at 490 nm was recorded as described previously (34). Fig. 1A shows a schematic illustrating the basic IP 3 R1 domain architecture and depicts the N-terminal suppressor domain and IP 3 binding core, the intervening region containing the modulatory domain, and the channel domain consisting of six transmembrane segments and the C-terminal tail (35). Also shown are cleavage sites in IP 3 R1 resulting from limited exposure to trypsin (31,36) together with the locations of the various epitopes targeted by the antibodies used in this study. Previous studies indicated that IP 3 R1 is cleaved by caspase-3 at multiple sites in vitro and in vivo, one of which was determined to be after a caspase consensus site (DEVD) at aa residues 1888 -1891 (23,24). Fig. 1B depicts this site and the expected N-terminal and C-terminal IP 3 R1 truncation fragments generated from this cleavage. Fig. 1C shows the predicted N-terminal and C-terminal fragments resulting from calpain cleavage of IP 3 R1 after glutamic acid residue 1917 (30). Finally, the membrane and soluble fragments of IP 3 R2, generated at a site corresponding to calpain cleavage of IP 3 R1, together with the epitope locations of ␣-IP 3 R2 antibodies used in this study, are shown in Fig.  1D.

Characterizing IP 3 R1 Fragmentation Patterns during Apoptosis-
An initial series of experiments was performed to examine the status of IP 3 R1 in cells undergoing apoptosis. As a model system, apoptotic cell death was triggered in DT40 cells stably expressing rat IP 3 R1 by exposure to 2 M staurosporine, a general kinase inhibitor which has been shown to result in caspase 3 activation (23,24). Cell lysates were then processed for immunoblot analysis. Fig. 2A shows that full-length IP 3 R1 immunoreactivity decreased with time in response to staurosporine. Notably, IP 3 R1 was partially cleaved into three faint but detectible fragments that corresponded to ϳ215 kDa and ϳ170 kDa fragments detected with CS antibodies that recognize the N terminus of the IP 3 R1 protein (Fig. 2B). In addition, a ϳ95-kDa fragment was recognized with UCD antibodies, which are raised against the C terminus of IP 3 R1 (Fig. 2C). In some experiments, the appearance of a complementary 120-to 140-kDa fragment was also observed (see Fig. 2C). The IP 3 R1 fragmentation became more evident at 12 h, and these products persisted even after 24 h. Lysates from cells stably expressing the IP 3 R1 truncation constructs corresponding to the predicted caspase cleavage products of IP 3 R1, N-terminal fragment (aa 1-1891) and tryptic fragments IϩIIϩIII (aa 1-1581) (31) were used as controls. The cleavage fragments from staurosporinetreated lysates detected with CS migrated at molecular weights comparable with that of recombinant fragments loaded as controls. No cleavage products were detected in untreated cells or in DT40 -3ko cells, which do not express IP 3 R. These data demonstrate that during staurosporine-induced apoptosis, IP 3 R1 FIGURE 1. Schematic diagrams of the IP 3 R structure. A, representation of the IP 3 R1 basic domain structure with functional domains: the suppressor domain (aa residues 1-226), the IP 3 binding core (aa residues 227-577), the modulatory domain (residues 578 -2275), the channel domain (aa residues 2276 -2589), and the cytosolic tail (aa residues 2589 -2749). Also shown are the fragments that result from limited tryptic digestion of IP 3 R1, with residues constituting the boundaries between fragments indicated by the numbers below the schematic. The location of antibody epitopes used in this study are also denoted: CS centered around aa residue 40, SC within aa residues 1894 -1973, UCD within aa residues 2680 -2749, and CT-1 within aa residues 2731-2749. B and C, representations of the N-terminal and C-terminal fragments expected to result from caspase and calpain cleavage of IP 3 R1, respectively. D, the recombinant N-terminal and C-terminal fragments of IP 3 R2 with ␣-IP 3 R2 antibody epitope locations: NT-2 within aa residues 320 -338 and CT-2 within aa residues 2686 -2701.
undergoes proteolysis in a manner consistent with previous studies in DT40 cells and several other cell types (22)(23)(24)(25). More importantly, our experiments suggest that during apoptosis, IP 3 R1 is cleaved into two overlapping N-terminal fragments encompassing the suppressor domain, the ligand-binding domain, and most of the coupling domain and a C-terminal fragment containing the channel domain and the cytosolic tail.
Cleavage Fragments Remain Associated with Membranes-We next asked whether the fragments generated during staurosporine treatment were dislocated or extracted from intracellular membranes following proteolysis. DT40 cells stably expressing IP 3 R1 were either treated with vehicle as a control or incubated with 2 M staurosporine for 12 h when significant cleavage is readily observed. Cytosolic and membrane fractions were prepared by ultracentrifugation. Equivalent amounts of proteins were subjected to Western blot analysis with the indi-cated antibodies. As shown in Fig. 3, IP 3 R1 fragments generated during apoptosis detected by either the N-terminally directed antibody, CS (Fig. 3A), or by the C-terminal antibody, CT-1 (Fig. 3B), remained localized exclusively in the membrane fractions. These data suggest that although IP 3 Rs are cleaved, the fragmented receptor remains associated with the ER and is not dislocated or extracted from ER membranes. To confirm efficient subcellular fractionation, GAPDH and SERCA were used as markers for cytosolic and membrane fractions, respectively (Fig. 3C).
Tetrameric IP 3 R1s Remain Assembled following Proteolysis-To examine whether tetrameric channels are disassembled in response to apoptotic stimulus, proteins from control and staurosporine-treated cells were fractionated on native gels. A complex with a molecular mass of ϳ1.15 MDa was detected with both N-terminal and C-terminal ␣-IP 3 R1 antibodies in both control and staurosporine-treated cells (Fig. 3D). Compared FIGURE 2. Characterizing IP 3 R1 fragmentation pattern during staurosporine-induced apoptosis. A, DT40 cells stably expressing IP 3 R1 were treated either as a control or incubated with 2 M staurosporine for the indicated times. Cells were then harvested and lysed, and equivalent amounts of proteins were fractionated and processed for immunoblot analyses with the CS antibody. The blot was reprobed with an antibody recognizing PKD as a loading control (lower panel). B, DT40 cells stably expressing IP 3 R1 were treated either as a control or incubated with 2 M staurosporine for the indicated times. Cells were then harvested and lysed, and equivalent amounts of proteins were fractionated and processed for immunoblot analyses with the CS antibody. Lysates from DT40 -3ko cells, HEK293 cells expressing an IP 3 R1 truncation fragment (1-1891), or IP 3 R1 truncation mutant fragI-III (residues 1-1582) were used as migration controls. The right panel is a longer exposure of the portion of the same gel. C, a duplicate gel with the same lysates as in B but probed with C-terminal UCD antibody. The right panel is a longer exposure of the portion of the same gel. B and C, bands were visualized using chemiluminescence. Representative experiments are shown. The asterisks denote nonspecific bands seen in 3ko lysates. with control untreated cells, there was a decrease in the density of the 1.15-MDa band, consistent with IP 3 R1 degradation induced by staurosporine. Nevertheless, immunoreactivity of lower molecular weight species, suggestive of segregated cleaved fragments or disassembly of the tetrameric channel, were not readily observed. Taken together, these data indicate that during staurosporine-induced apoptosis, IP 3 R1 was cleaved at multiple sites, giving rise to stable N-terminal fragments encompassing the ligand-binding domain and the regulatory domain and a C-terminal fragment containing the channel domain. Importantly, these fragments were not extracted from ER membranes, and IP 3 R1 largely maintained its tetrameric integrity.
Next, we asked whether caspase-generated fragments of IP 3 R1 could interact with each other when coexpressed. We therefore generated truncated IP 3 R1 constructs on the basis of the canonical caspase cleavage site, corresponding to the N-terminal 1-1891 aa acids (R1 casp sol) and C-terminal 1892-2749 aa acids (R1 casp mem). These constructs were stably transfected into DT40 -3ko, either individually or in combination (R1 casp sol/mem) using the TPV vector. Whole cell lysates were generated from these stable cell lines, and proteins were processed for immunoblot analyses. As expected, the N-terminal CS antibody detected an ϳ215-kDa band in R1 casp sol and R1 casp sol/mem. Similarly, the C-terminal antibody UCD detected an ϳ95-kDa band in lysates expressing R1 casp mem and R1 casp sol/mem (Fig. 4A). There was a lower band detected with the C-terminal antibody UCD but not with SC (an epitope located within residues aa 1894 -1973), and, therefore, we concluded that this lower band could result from an alternative translation or a degradation product. IP 3 R1 is also a known substrate for calpain, and cleavage has been reported to result in fragments, including an ϳ95-kDa membrane retained fragment analogous to those produced by caspase activity on IP 3 R1 (30,37,38). Further, recent reports have suggested that calpain-mediated IP 3 R1 proteolysis is implicated in the disruption of Ca 2ϩ homeostasis during ischemic brain injury (30,39). Therefore, to extend these studies, IP 3 R1 constructs encoding fragments on the basis of the putative calpain cleavage site after aa residue 1917 were generated. Stable cell lines were then established to either individually express R1 calp sol or R1 calp mem, corresponding to the N-terminal aa 1-1917 and the C-terminal aa 1918 -2749, respectively, or express both fragments in combination using the TPV (R1 calp sol/mem). Fig. 4B shows immunoblot analyses from stable cells expressing these various constructs. Using a similar approach, we also generated two truncated IP 3 R2 constructs homologous to IP 3 R1 fragments generated by calpain. The N-terminal fragment consisted of aa 1-1869, and the C-terminal peptide corresponded to aa 1870 -2701. Stable cell lines were generated expressing both fragments, as demonstrated in Fig. 4C.
To this point, our data suggest that the proteolytic fragments of IP 3 R1 remain in the ER membrane following cleavage. To determine whether the various N-and C-terminal IP 3 R fragments can physically interact in this environment, coimmunoprecipitation experiments were performed. DT40 expressing R1 casp sol/mem, R1 calp sol/mem, or R2-frags were harvested and lysed, and proteins were either mock-immunoprecipitated with protein A/G-agarose beads or immunoprecipitated with fragment-specific antibodies. Fig. 5A shows that the N-terminal fragment R1 casp sol was specifically coimmunoprecipitated with

. Establishment of stable DT40 cells lines expressing IP 3 R constructs.
A, whole cell lysates were prepared from either DT40 -3ko (3ko) cells or DT40 cells stably expressing rat IP 3 R1 (R1), R1 casp sol, R1 casp mem, or R1 casp sol/mem. Lysates were fractionated on 5% gel and probed with CS (left panel) or UCD (right panel). B, whole cell lysates were prepared from either 3ko or DT40 cells stably expressing rat IP 3 R1 (R1), R1 calp sol, R1 calp mem, or R1 calp sol/mem. Lysate were fractionated on 5% gel and probed with CS (left panel) or UCD (right panel). C, whole cell lysates from 3ko cells, DT40 cells stably expressing mouse IP 3 R2 (R2), or complementary IP 3 R2 fragments (R2frags) were processed on 5% gel and probed with NT-2 (left panel) or CT-2 (right panel). The arrowheads point to the position of the expressed constructs. The arrow indicates a band that likely represents a degradation or an alternative translation product. The asterisks denote nonspecific bands. the C-terminal fragment. Similarly, the C-terminal R1 calp mem was coimmunoprecipitated with N-terminal R1 calp sol (Fig. 5B), and, likewise, the N-terminal fragment of IP 3 R2 was coimmunoprecipitated with its C-terminal fragment (Fig. 5C). These data show that complementary split IP 3 R polypeptides retain the ability to interact in cells and further indicate the potential that IP 3 R1 fragments generated during apoptosis could remain physically associated after caspase and calpain proteolysis.
Stably Expressed Split IP 3 R1s Form Tetrameric Channels-We then examined whether separate fragmented polypeptides can associate to reconstitute IP 3 R1 tetrameric channels. To explore this idea, lysates from DT40 -3ko cells and cells expressing wt IP 3 R1, R1 casp mem, R1 casp sol, or R1 casp sol/mem were processed for immunoblot analyses using native PAGE where protein complexes retain their native structure. A protein band of ϳ1.15 MDa was detected with both N-terminal and C-terminal directed antibodies in cells expressing WT-IP 3 R corresponding to tetrameric IP 3 R1 (40). Interestingly, a similar band was observed in R1 casp sol/mem but not in cells expressing either fragment alone (Fig. 6, A and B.) Moreover, although R1 casp sol appeared as a prominent band migrating at ϳ200 kDa, reflecting monomeric species, R1 casp mem migrated as a smear extending from ϳ380 kDa to the top of the native gel, perhaps resulting from protein aggregation or higher order molecular assemblies. These data are consistent with the idea that expression of the N-terminal fragment somehow "rehabilitates" the aggregation-prone C-terminal fragment and traps it within the tetrameric structure. Parallel protein samples, resolved on denaturing 5% SDS-PAGE and processed for immunoblot analyses with the same antibodies, confirmed the expression of the intended protein fragments. Overall, these findings indicate that complementary IP 3 R1 fragments can assemble to form tetrameric structures and suggest, in line with data shown in Fig. 3D, that a proportion of caspase-fragmented IP 3 R1 channels have potential to remain assembled during apoptosis.
Stably Expressed Split IP 3 R1 Fragments Form Functional Channels-We next sought to address whether the channels assembled from the expressed fragments are functional in terms of Ca 2ϩ release. Cells expressing various constructs were loaded with Fura-2 AM, and the Ca 2ϩ profile was monitored as changes in 340/380 fluorescence ratio. When stimulated with 500 nM trypsin to activate the endogenous G protein-coupled receptor protease-activated 2 receptor (PAR2) (41), cells expressing R1 casp sol or R1 casp mem were not responsive, indicating, as expected, that both truncation fragments, when expressed individually, fail to form any functional Ca 2ϩ release channel (Fig. 7B). Remarkably, coexpression of R1 casp sol and R1 casp mem fragments together responded similarly to cells expressing WT-IP 3 R1 (compare Fig. 7, A and B), suggesting that caspase-fragmented IP 3 R1s retain function as IP 3 -gated Ca 2ϩ release channels. These experiments indicate that cleaved IP 3 R1s generated during apoptosis may maintain the capacity to function as IP 3 -gated Ca 2ϩ channels. Furthermore, although DT40 cells expressing the R1 calp sol or R1 calp mem construct alone were also not responsive to PAR2 activation, cells coexpressing both calpain fragments responded robustly (Fig. 7C), FIGURE 5. Interaction of recombinant IP 3 R fragments. A, DT40 cells expressing R1 casp sol/mem, R1 calp sol/mem, or R2-frags were lysed, and proteins were subjected to mock or fragment-specific immunoprecipitation (IP) followed by Western blot analysis. A, R1 casp sol/mem lysates were immunoprecipitated with CT-1 and probed with either CS specific to the N-terminal fragment (upper panel) or C-terminal fragment-specific antibody (CT-1) (lower panel). B, R1 calp sol/mem lysates were immunoprecipitated with N-terminal fragment antibody (SC) and probed with either CS (upper panel) or CT-1 (lower panel). C, R2-frags lysates were immunoprecipitated with C-terminal antibody (CT-2) and probed with either N-terminal fragment-specific antibody (NT-2) (upper panel) or C-terminal specific CT-2 (lower panel). FIGURE 6. Fragmented IP 3 Rs migrate as tetrameric complexes on native gels. Whole cell lysates were prepared from either DT40 -3ko (3ko) cells or DT40 cells stably expressing rat IP 3 R1 (R1), R1 casp sol, R1 casp mem, or R1 casp sol/ mem using CHAPS lysis buffer. Lysates were then resolved on 3-12% native gel as described in Fig. 3D. Proteins were immunoblotted with CS in A or CT-1 in B. Parallel aliquots were fractionated on denaturing 5% SDS-PAGE and probed with CS in C or CT-1 in D. The asterisks denote nonspecific bands.
suggesting similarly that the calpain-fragmented IP 3 R1s also function as IP 3 -gated Ca 2ϩ release channels. These results indicate that functional IP 3 R channels can be assembled from separate complementary polypeptides. To confirm this, mIP 3 R2 fragments, homologous to calpain fragments of IP 3 R1, also formed functional channels when coexpressed in the same cells (Fig. 7D). Pooled data for the various constructs is shown in Fig.   7E. Previous studies showed that transient expression of the C-terminal 95-kDa-fragment resulted in a modest depletion of ER stores (22,27,30). However, examination of resting Ca 2ϩ levels and the content of intracellular Ca 2ϩ stores, as judged by the rate and magnitude of cyclopiazonic acid -induced Ca 2ϩ release, showed no significant difference between these cell lines, expressing channel-only fragments and IP 3 R1 cells or FIGURE 7. Robust IP 3 R-mediated Ca 2؉ release activity in cells expressing protease-generated IP 3 R fragments. DT40 -3ko (3ko), or DT40 cells stably the expressing various IP 3 R constructs as indicated were loaded with the Ca 2ϩ indicator Fura-2AM and stimulated with 500 nM trypsin to induce IP 3 formation. Ca 2ϩ release was measured as a change in the 340/380 fluorescence ratio. Shown are averaged Ca 2ϩ traces of 3ko and IP 3 R1 expressing cells in A; R1 casp sol, R1 casp mem and R1 casp sol/mem in B; R1 calp sol, R1 calp mem, and R1 calp sol/mem in C; and IP 3 R2 (R2) and R2-frags in D. E, histograms depicting the average change over the basal 340/380 fluorescence ratio resulting from trypsin stimulation in cells expressing the indicated constructs. Experiments were repeated at least three times with more than 40 cells imaged in each run. Data are presented as mean Ϯ S.E. DT40 -3ko (Fig. 8, A-C). That we did not observe any difference in the ER Ca 2ϩ store content might stem from the low expression level of the C-terminal 95kDa fragment or reflect some compensatory gene regulation induced in the stable cell lines we generated. Nevertheless, in total, these data suggest that protease-cleaved IP 3 R1s can assemble and function as ligand-gated Ca 2ϩ release channels.
Significantly, our data provide for the first time unequivocal, direct evidence that polypeptide continuity is not necessary for IP 3 R activity in intact cells. There have been suggestions that trypsin-fragmented IP 3 R1 could mediate IP 3 -induced Ca 2ϩ release in crude cerebellar microsomes (31). However, it should be noted that these studies did not demonstrate complete unambiguous digestion of cerebellar IP 3 R1. Specifically, it was not discerned if cleavage of each monomer within the every native channel had occurred. Therefore, it was not possible to ascertain whether residual IP 3 -induced Ca 2ϩ release in trypsinized microsomes resulted from fully intact channels or tetramers consisting of both cleaved and intact monomeric subunits. To further extend the idea that peptide continuity is not a prerequisite for gating the IP 3 R, we generated truncation mutants on the basis of trypsin cleavage sites on IP 3 R1 (Fig. 1A). Stable DT40 cells expressing individual constructs were established, and expression of expected fragments was confirmed by Western blot analysis (data not shown). These stably expressed constructs were all refractory to stimulation of the PAR2 receptor. Fig. 9A shows that functional channels were reconstituted when tryptic fragments IIIϩIVϩV were transiently expressed in cells stably harboring complementary IϩII fragments. Likewise, functional channels were restored when tryptic fragments IVϩV were transiently introduced into cells stably expressing complementary IϩIIϩIII fragments (Fig. 9B). Finally, functional channels could also be reconstituted when the IP 3 R2 N-terminal fragment or R1 casp sol was coexpressed with R1 calp mem in the same cells (Fig. 9, C and D) or the N-terminal fragment of rat IP 3 R3 and C-terminal fragment of rat IP 3 R1 (data not shown). It is noteworthy that the channel reconstituted from coexpressed R1 casp sol with R1 calp mem was missing 26 amino acids spanning the distance between the putative caspase and calpain cleavage sites.
In a final series of experiments, we asked whether Ca 2ϩ release mediated by complementary soluble and membrane fragments of caspase-cleaved IP 3 R1 could support apoptosis induced in a physiologically relevant manner. Immature B cells undergo apoptosis following strong stimulation of the B cell receptor. This process is proposed to contribute to the deletion of autoreactive immature cells (42). Apoptosis was therefore induced by exposure to IgM (15 g/ml) for 17 h, and cell viability was monitored as described under "Materials and Methods." IgM exposure resulted in significantly more cell death in DT40 cells expressing IP 3 R1 in comparison with DT40 -3ko cells. Importantly, a similar loss of cell viability occurred in cells expressing both R1 casp sol and R1 casp mem (Fig 10). These data indicate that Ca 2ϩ release through complemented IP 3 R is sufficient for its role in apoptosis.

DISCUSSION
Studies investigating the role of IP 3 R in apoptosis have reached disparate and often seemingly paradoxical conclusions. For example, IP 3 R1 was shown to undergo caspase 3-dependent cleavage in various cell types, and this cleavage led to the formation of a C-terminal constitutively leaky channel termed a "channel-only domain" that mediates Ca 2ϩ release essential for complete execution of the apoptotic program (22,24,27). In contrast, it has been reported that IP 3 R1 is not cleaved in HEK293, HeLa, and Jurkat cells undergoing apoptosis in response to exposure to staurosporine, TNF␣, Trail, or FIGURE 8. Stable expression of IP 3 R membrane fragments does not significantly alter ER store content. DT40 cells expressing the indicated constructs were exposed to cyclopiazonic acid (CPA) following perfusion in Ca 2ϩfree buffer (1 mM EGTA) following the protocol shown in A. The 340/380 ratio (average of the initial 10 s of recording) corresponding to basal [Ca 2ϩ ] was measured prior to removal of extracellular Ca 2ϩ . Pooled data for the basal ratio are shown in B. C, pooled data of the maximum ratio achieved following cyclopiazonic acid treatment as an indicator of store content. Four experimental runs were conducted for each construct, in which Ͼ 30 cells were analyzed per run.
UV irradiation (43). Moreover, these latter studies concluded that IP 3 R1 was, if at all, a very late caspase substrate during apoptosis. A recent study has also proposed that caspase activity is not required for the apoptotic Ca 2ϩ increase and that IP 3 -mediated Ca 2ϩ signals, rather than the consequences of IP 3 R1 cleavage, were solely responsible for intracellular Ca 2ϩ elevations associated with staurosporine-induced apoptosis in HeLa cells and human primary fibroblasts (44). An additional scenario was proposed by Khan et al. (34) in light of evidence that staurosporine-induced increases in [Ca 2ϩ ] i occurred even in DT40 cells expressing the channel-only domain or fulllength IP 3 R1 with an inactivating channel pore mutation. These authors concluded that the cytosolic tail of IP 3 R1 was required for apoptosis and proposed that IP 3 R1 plays an ion channelindependent role, perhaps by facilitating Ca 2ϩ entry across the plasma membrane or by providing a platform for protein-protein interactions required for the apoptotic process (34). Indeed, the reasons for the variable susceptibility to caspase cleavage and disparate outcomes of cleavage are not known but might result from different experimental conditions or reflect tissue-specific mechanisms for induction and regulation of apoptosis. A recent quantitative proteomic study has reinforced this view by showing that individual proteins are variably susceptible to caspase cleavage in different cell lines and even in the same cell in response to a distinct apoptotic challenge (45).
In this study, we have shown that IP 3 R1 is degraded during staurosporine-induced cell death, consistent with previous studies. During this process, IP 3 R1 was partially cleaved, resulting in fragments corresponding to the IP 3 R1 cytoplasmic arm encompassing the ligand-binding domain together with most of the modulatory region and a residual C-terminal "stump" encompassing the channel region. It appears that upon initiation of apoptosis, IP 3 R1 was attacked at two locations. One site corresponds to the caspase consensus site (DEVD) at aa 1891, generating the N-terminal ϳ215-kDa and C-terminal ϳ95-kDa fragment. The protease responsible for the second fragment is unclear. However, on the basis of the molecular weight of this N-terminal fragment (ϳ170 kDa), the cleavage site is likely in the vicinity of the solvent-accessible junction between tryptic fragments III and IV. A caveat associated with these studies is that we cannot exclude that IP 3 R1 is cleaved at additional sites by caspases or other proteases that generate fragments that were not detected with the antibodies utilized in this study. It also remains to be formally determined whether IP 3 Rs cleaved at multiple sites can form functional channels. At present, it is not clear from our data whether the two locations are attacked simultaneously or sequentially or whether the alternative cleavage is mediated by caspase 3 or another protease. That cleavage at the caspase consensus site precedes proteolysis at other sites has been suggested. For example, Assefa et al. (24) detected a 95-kDa fragment using C-terminal directed antibodies that was generated by caspase 3 both in vitro and in vivo. Moreover, double mutations of the caspase consensus site (Asp-1888I and Asp-1891A) inhibited apoptosis and completely abolished IP 3 R1 degradation and production of the membrane fragment (24).
We present several observations that suggest that caspasecleavage per se might not necessarily, or at least not initially, abolish IP 3 -induced release or inevitably result in the formation of unregulated, leaky channels. Firstly, subcellular fractionation and native gel experiments showed that IP 3 R1 cleavage fragments were retained in membranes and that IP 3 R1 channels can remain assembled as tetrameric channels during staurosporine-induced fragmentation. These data are consistent with earlier experiments showing that fragments of IP 3 R1 in cerebellar microsomes remained associated with membranes following trypsin exposure (31,36). Secondly, when recombinant IP 3 R1 fragments, which mimic caspase-cleaved IP 3 R1 monomers in all four subunits, were coexpressed, they associate with each other to form IP 3 R1 complexes migrating similarly to native tetrameric channels. Finally, Ca 2ϩ imaging studies showed robust IP 3 R1 activity in cells expressing both complementary cleavage fragments, which is sufficient to support B cell receptor-mediated apoptosis. Supporting this conclusion was the finding that IP 3 R1 truncation mutants corresponding to calpain cleavage products and similar fragments of IP 3 R2 also associated to reconstitute functional channels. Although this is surprising at first glance, several studies have shown that the effect of caspase and calpain-mediated cleavage is proteinspecific and sometimes cell-specific (45,46). Examples are also present in the literature where caspase cleavage results in a protein that retains activity (46 -48). In addition, caspase or calpain cleavage can trigger protein activity or regulate localization, its interaction with other proteins, and sensitivity to other modulators and not necessarily lead to immediate protein destruction (46, 47, 49 -51). It can be envisaged that caspase or calpain cleavage of IP 3 R1 at the C-terminal part of the transducing domain serves to modulate IP 3 -mediated Ca 2ϩ release activity rather than totally inactivating the receptor or generating constitutively leaky channels. However, these models are not mutually exclusive. Further studies are required to determine whether caspase or calpain cleavage of IP 3 R1 alters single channel properties, its protein-protein interactions, posttranslational modification, or even its susceptibility to other proteases.
An obvious question is how the fragments physically associate in a cellular context to form functional channels in vivo. Earlier studies have established that the five fragments produced following controlled tryptic digestion of IP 3 R1 in vitro remain associated following cleavage (31,36), indicating that the modular nature of the IP 3 R allows the structure to be maintained following proteolysis. In addition, a 68-kDa tryptic fragment corresponding to the IP 3 binding domain could be immunoprecipitated with a 94-kDa fragment constituting the C-terminal channel domain (36). Further experiments have established that the N and C termini are physically associated within assembled tetrameric channels and that this association can potentially be mediated by both intersubunit or intrasubunit interactions (7). Subsequent studies have shown that critical residues in the suppressor domain and the cytosolic loophelix between transmembrane segments 4 and 5 are required for this interaction (52,53). These studies provide a structural basis to explain our data whereby the expressed fragmented receptor can associate through non-covalent interactions in vivo. Mutagenesis of key residues in the suppressor domain and the transmembrane linker also indicates that these interactions are pivotally important for transducing the conformational change induced by IP 3 binding to opening of the pore (5,(52)(53)(54). These data indicate that a conformational change transmitted through the entire linear peptide sequence of the cytosolic domain of the protein is not necessary for gating. Our data, demonstrating functional channels reconstituted from N and C-terminal fragments, which are not necessarily linearly continuous or indeed complementary, clearly provide strong evidence that peptide continuity is not a prerequisite for channel gating and is consistent with interactions between the N and C termini being sufficient for channel opening following IP 3 FIGURE 10. Expression of complementary IP 3 R fragments supports apoptosis. DT40 cells expressing the indicated constructs were incubated for 17 h in the presence or absence of 15 g/ml IgM, and the IgM-mediated change in cell viability was monitored using the CellTiter A queous One assay (Promega) as described under "Materials and Methods." IgM incubation of cells expressing either IP 3 R1 or complementary caspase membrane and soluble fractions resulted in a comparable loss of cell viability, which was significantly different from DT40 -3ko cells (p Ͼ 0.05, Student's t test). The graph shows mean Ϯ S.E. for three experiments performed in triplicate.
binding. Further experiments are necessary to address whether the interaction of the N and C termini, as opposed to long-range conformational changes, transduce other pivotal forms of IP 3 R channel modulation, such as Ca 2ϩ or ATP regulation, to gating of the pore.