Plasma membrane Ca2+ATPase isoform 4b is cleaved and activated by caspase-3 during the early phase of apoptosis.

The plasma membrane Ca(2+) pump (PMCA) is an essential element in the complex of mechanisms that maintain low intracellular Ca(2+) concentration in the living cell. This pump is tightly regulated by calmodulin through binding to a high affinity calmodulin-binding domain at the C terminus that also serves as an autoinhibitor of the enzyme. Inspection of the C terminus of hPMCA4b, the most widely distributed form of PMCA, revealed a caspase-3 consensus sequence ((1077)DEID(1080)) just a few residues upstream of the calmodulin-binding domain. We demonstrate here that, in the early phase of apoptosis, hPMCA4b is cleaved at aspartic acid Asp(1080) in hPMCA4b-transfected COS-7 cells or in HeLa cells that naturally express this protein. This cleavage of hPMCA4b produces a single 120-kDa fragment that is fully active in the absence of calmodulin, because the whole inhibitory region downstream of the (1077)DEID(1080) sequence is removed. Our experiments show that caspase-3 or a caspase-3-like protease is responsible for the formation of the constitutively active 120-kDa PMCA4b fragment: 1) Pretreatment of the cells with the caspase-3 inhibitor Z-DEVD-FMK (benzyloxycarbonyl-Asp(OMe)-Glu(OMe)-Val-Asp(OMe)-fluoromethyl ketone) was able to block the production of the 120-kDa fragment. 2) In vitro treatment of hPMCA4b with recombinant caspase-3 also generated a 120-kDa cleavage product, consistent with that seen in cells undergoing apoptosis. 3) Mutants in which the caspase-3 consensus sequence was altered ((1077)AEID(1080), (1077)DEIA(1080), and (1077)AEIA(1080) mutants) were resistant to proteolysis. Based on these data, we conclude that hPMCA4b is a newly identified, natural caspase-3 substrate. We suggest that a constitutively active form of this protein, responding much faster to an increase in Ca(2+) concentration than the autoinhibited form, may have an important role in regulating intracellular Ca(2+) concentration in the apoptotic cell.

Apoptosis is a well controlled form of cell death (1), characterized by typical morphological changes of the suicidal cells. These changes include shrinkage of the cytoplasm, blebbing of the plasma membrane, externalization of membrane phos-phatidylserine molecules, chromatin condensation, and DNA fragmentation (2). These phenomena result from a series of different biochemical events in which caspases, a family of cysteine proteases, play a central role (3,4). Caspases are all expressed as inactive zymogens, and after they are activated, mainly by proteolytic cleavage, they cleave a set of cellular proteins in a coordinated manner. They are responsible for the cleavage of many essential proteins such as components of the apoptotic machinery, several structural proteins, and various proteins related to cellular signaling. The proteolysis of a large number of substrates leads to loss of cell function and finally to cell death. Over a dozen different caspases have been identified, which act as a cascade system. Roughly, they are divided into two groups: initiators and effectors. The initiator proteases are involved in the activation of the effectors, which in turn carry out apoptosis by cleaving their cellular substrates. The common feature of all caspases is that they all recognize a tetrapeptide sequence on their substrates and cleave exclusively after an aspartate residue. The four optimal amino acids upstream from the cut-site differ considerably among caspases, which implies different biological roles for each of them. Caspase-3 is one of the key effectors of apoptosis. Activated caspase-3 cleaves several important proteins, such as poly-(ADP-ribose) polymerase (PARP) 1 (5), various protein kinases (e.g. DNA-dependent protein kinase, protein kinase C ␦) (6 -8), ␣-spectrin, huntingtin, and inositol 1,4,5-trisphosphate (IP 3 ) receptor (9 -11) primarily at a DEXD recognition motif (12,13), resulting in either activation or loss of function of the substrates.
Several studies have suggested that changes in intracellular Ca 2ϩ homeostasis play an important role in apoptosis (for a recent review see Berridge et al. (14)). Excessive elevation of the intracellular Ca 2ϩ level by ionophores or by the sarco/ endoplasmic reticulum Ca 2ϩ -ATPase pump inhibitor thapsigargin induces apoptosis or necrosis in a variety of cells (15)(16)(17)(18)(19). Moderate intracellular calcium elevations, however, have been shown to be either apoptotic or antiapoptotic depending on the interplay between the mitochondria and the ER (20 -23). Numerous studies have reported that treatments that minimize increases in intracellular Ca 2ϩ level, i.e. buffering of intracellular Ca 2ϩ by addition of permeant Ca 2ϩ chelators or removal of extracellular Ca 2ϩ with Ca 2ϩ chelators, depress apoptosis (15,24,25). This indicates that an increase in intracellular Ca 2ϩ level is an important cell death signal. Other studies, however, have suggested that a rise in intracellular Ca 2ϩ concentration is not essential in the early phase of apoptosis but is necessary in the final degradation of the DNA (24). Thus, the precise role of Ca 2ϩ is controversial and appears to be highly dependent on the apoptotic stimuli applied and the cell type being studied.
In a recent study Szalai et al. (20) have demonstrated that a coincident occurrence of both proapoptotic stimuli and Ca 2ϩ signals induces apoptosis in permeabilized HepG2 cells. They showed that the rise in intracellular Ca 2ϩ caused by IP 3 -producing agonists leads to mitochondrial Ca 2ϩ uptake but not normally to apoptosis. This series of events is turned into an apoptotic signal by ceramide, which facilitates the Ca 2ϩ -induced opening of the mitochondrial permeability transition pore, resulting in the release of cytochrome c from the mitochondrial matrix. Cytochrome c activates caspase-9, which in turn activates the executioner caspases, leading to a finely controlled execution of cells. After the decay of the Ca 2ϩ spike, the mitochondria reseal to maintain mitochondrial metabolism and to provide the ATP necessary to support the following steps of apoptosis. In the absence of this ATP supply, a critical breakdown of energy metabolism unavoidably would lead to necrotic and inflammatory processes.
In a recent paper Pinton et al. (25) showed that ceramide alone, without IP 3 -producing agonists, induces a drastic loss of Ca 2ϩ from the ER in HeLa cells, and, consequently, an increase in Ca 2ϩ concentration both in the cytosol and in the mitochondrial matrix. They also showed that the Ca 2ϩ content of the ER can be an important modulator of apoptosis; reduction of the ER Ca 2ϩ content as well as buffering of changes in intracellular Ca 2ϩ concentration by Ca 2ϩ chelators protected HeLa cells from ceramide-induced apoptosis. The same authors suggested that the anti-apoptotic action of Bcl-2 is based on a Bcl-2-dependent reduction of the ER Ca 2ϩ content.
Plasma membrane calcium ATPase (ATP2B, referred to here as PMCA) is a P-type ATPase that plays a crucial role in regulation of cell calcium homeostasis. Its function is to remove excess calcium from the cytosol to maintain the resting low intracellular calcium concentration and to prevent cells from a lethal overload of calcium. The pump is coded by four different genes and alternative splicing of the primary transcripts produces at least 15 isoform variants. Two of the four PMCAs (PMCA1 and PMCA4) are ubiquitous, whereas the expression of the two other isoforms (PMCA2 and PMCA3) is cell-and tissue-specific (26).
PMCAs are calmodulin-regulated enzymes. They have a special C terminus, which in the absence of calmodulin interacts with the catalytic core, thus inhibiting the calcium transport activity of the pump (27). Binding of Ca 2ϩ -calmodulin to the calmodulin-binding domain (located between residues 1086 and 1113 of hPMCA4b) abolishes this interaction, and the pump becomes activated. The C-terminal regions of PMCAs differ substantially from one another. Not only do the four genes differ from each other in this region, but each of them has an alternate splice option in the middle of the sequence coding for the calmodulin-binding domain (28). This alternative splice generates two or more versions (a, b, c, etc.) of the C-terminal tail that result in very different modulation of the pump's activity.
PMCA4b, although it is widely distributed, appears to be a specialized form of PMCA: 1) It has the lowest calmodulin affinity among the PMCA "b" forms, and 2) its activation by calmodulin is the slowest of all isoforms studied so far; this results in a very slow activation by Ca 2ϩ (29). Additional reports have suggested that this slow response of PMCA4b is necessary for full development of the Ca 2ϩ spike upon stimu-lation of cells expressing this isoform. Differences in the response of PMCAs to changes in intracellular Ca 2ϩ concentration should affect substantially the shape of Ca 2ϩ spikes. In a recent paper (30) we concluded that the type of PMCA expressed corresponds well with the speed of Ca 2ϩ signals in the cell, so that cells that need to respond faster to Ca 2ϩ spikes express faster-responding PMCA isoforms. In good accordance with this theory we found that fast pumps (PMCA2b, 3f, 2a, 4a) are located in tissues such as heart, skeletal muscle, and brain whereas PMCA4b was abundant in Jurkat cells.
As discussed above, several studies showed that various apoptotic stimuli might change the intracellular calcium concentration; however, the involvement of the PMCAs in apoptosis has not been demonstrated. Therefore, in the present study we investigated the possible role of PMCA4b in apoptotic cells. Transiently transfected COS-7 cells expressing either wild type hPMCA4b or its mutants and epithelial HeLa cells expressing hPMCA4b endogenously were examined. Here we investigated mutants to the predicted caspase cut site, which is only a few residues upstream of the calmodulin-binding domain. We show that cells that expressed wild type hPMCA4b, subjected to either spontaneous-or staurosporine (STS)-induced apoptosis, contain a degraded form of the hPMCA4b, from which the C-terminal regulator region is cut. The apoptotic degradation of the mutants was compared with that of the wild type pump. In situ and in vitro experiments proved that PMCA4b is a substrate of caspase-3 and that the cleavage by this protease irreversibly activates the pump.

EXPERIMENTAL PROCEDURES
Chemicals-STS and calmodulin were obtained from Sigma Chemical Co. Recombinant caspase-3 was from Upstate Biotechnology Inc., caspase-3 inhibitor (Z-DEVD-FMK) was obtained from Calbiochem. Annexin V fluorescein conjugate was from Molecular Probes. Lipo-fectAMINE and OPTI-MEM were obtained from Invitrogen. Polyclonal anti-PARP antibody was purchased from Biomol Research Laboratories, Inc. All other chemicals used for this study were of reagent grade.
Construction of the hPMCA4b Mutants-1077 AEID 1080 , 1077 DEIA 1080 , and 1077 AEIA 1080 mutants were made by the double-PCR method. In this method, the first PCR product was made by amplification of an hPMCA4b sequence using a primer containing the desired mutation and a primer including a unique restriction site (BamHI) already present in the sequence. The product of this PCR was then used as primer for a second round of PCR with the other primer, including the second unique site (NsiI). The PCR products were cloned using the blunt PCR cloning kit from Invitrogen and sequenced by the Mayo Molecular Biology Core Facility. The NsiI-BamHI piece was cut out and placed into the wild type hPMCA4b in pSP72. The full-length SalI-KpnI piece was then cut out of pSP72 and ligated into expression vector pMM2.
Cell Culture and Transfection-COS-7 and HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. All cells were kept at 37°C, 5% CO 2 in a humidified atmosphere. COS-7 cells were transfected as described before (31) using the LipofectAMINE reagent based on the protocol as recommended by the manufacturer (Invitrogen). Briefly, cells were grown in 175-cm 2 flasks to 70 -80% confluence. Cells were incubated at 37°C with DNA-LipofectAMINE complex (formed by incubating 8 g of DNA and 100 l of LipofectAMINE in 3.6 ml of serum-free OPTI-MEM medium for 30 min) in 14.5 ml of serum-free OPTI-MEM medium. After 5 h of incubation, cells were supplemented with serum; 24 h later the incubation medium was replaced with fresh tissue culture medium, and cells were harvested after an additional 24 h.
Apoptosis Induction-COS-7 cells were grown on six-well plates and transfected as above. 48 h after transfection, floating cells were removed and precipitated by 6% ice-cold trichloroacetic acid then resuspended in electrophoresis sample buffer. In adherent cells apoptosis was induced by adding 1 M STS in fresh tissue culture medium. Optionally 120 M of caspase-3 inhibitor (Z-DEVD-FMK) was added 1 h prior to STS treatment. Cells were precipitated by 6% ice-cold trichloroacetic acid after different times of incubation then resuspended in electrophoresis sample buffer. HeLa cells were treated similarly, except for the transfection procedure, because these cells express hPMCA4b endogenously.
Staining Cells with Annexin V Conjugates-Apoptosis was induced in COS-7 or HeLa cells as described above. After 5-h treatment with STS, cells were harvested and prepared for flow cytometry analysis according to the annexin V binding protocol recommended by the manufacturer (Molecular Probes). Briefly, cells were washed in cold phosphate-buffered saline, then resuspended in annexin-binding buffer (10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl 2 , pH 7.4) to 10 6 cells/ml. 5 l of annexin V conjugate and 5 g/ml propidium iodide (PI) (an indicator of dead cells) were also added to 100 l of cell suspension. After 15-min incubation at room temperature, 400 l of annexin-binding buffer was added, and then cells were analyzed on FACSCalibur flow cytometer (Becton Dickinson Biosciences, San Jose, CA). List mode data of 10,000 cells were collected using the CellQuest software.
Isolation of Microsomes from COS-7 Cells-Crude microsomal membranes from COS-7 cells were prepared as described previously (32) with the following modifications. Cells were washed with ice-cold phosphate-buffered saline solution then harvested in the same medium containing 0.1 mM phenylmethylsulfonyl fluoride, 6 g/ml aprotinin, 2.2 g/ml leupeptin, and 1 mM EGTA, pH 7.4. After centrifugation, cells were resuspended in an ice-cold hypotonic solution containing 10 mM Tris-HCl, pH 7.4, 1 mM MgCl 2 , 0.5 mM EGTA, 4 g/ml aprotinin, 2 g/ml leupeptin, and 4 mM dithiothreitol. After lysis, homogenization, and centrifugation, the final pellet was resuspended in a solution of 0.25 M sucrose, 0.15 M KCl, 10 mM Tris-HCl, pH 7.4, 2 mM dithiothreitol, 20 M CaCl 2 , and the suspension was stored in liquid N 2 .
In Vitro Digestion with Recombinant Caspase-3 and -Calpain-10 g of microsomal membrane preparation isolated from COS-7 cells transfected with the appropriate construct was preincubated for 2 min at 37°C in a medium containing 100 mM KCl, 25 mM TES-triethanolamine, pH 7.2, 0.1 mM CaCl 2 , 0.09 mM EGTA, 8.5% sucrose, 5 mM dithiothreitol, 20 g/ml aprotinin, 20 g/ml leupeptin. Proteolysis was initiated by the addition of 0.25 g of recombinant caspase-3 and stopped by 6% ice-cold trichloroacetic acid. The precipitate was supplemented with 100 g of bovine serum albumin, washed once with distilled water, and then dissolved in electrophoresis sample buffer. Alternatively, the Ca 2ϩ transport activity of the digested membrane preparation was measured as described below. Digestion with -calpain was carried out in a medium containing 50 g/ml microsomal membrane preparation, 0.2 mM CaCl 2 , 0.1 mM EGTA, 25 mM TES-triethanolamine, pH 7.2, 5 mM dithiothreitol, 1 mM MgCl 2 , 0.05% Triton, 80 mM KCl. After preincubation for 2 min at 37°C, proteolysis was initiated by 5 g/ml -calpain and stopped by the addition of 6% ice-cold trichloroacetic acid. The precipitate was supplemented with 100 g of bovine serum albumin, washed once with distilled water, then dissolved in electrophoresis sample buffer. Ca 2ϩ Transport Assay-Ca 2ϩ uptake by microsomal membrane vesicles was carried out as described previously (33,34) in a reaction mixture containing 100 mM KCl, 25 mM TES-triethanolamine, pH 7.2, 7 mM MgCl 2 , 100 M CaCl 2 (labeled with 45 Ca), 40 mM KH 2 PO 4 / K 2 HPO 4 , pH 7.2, 200 nM thapsigargin, 4 g/ml oligomycin, enough EGTA to obtain the desired Ca 2ϩ concentration and optionally 4 g/ml calmodulin. The free Ca 2ϩ is calculated as previously described (35,36). The results of these calculations are nearly identical to those obtained with MaxChelator (www.stanford.edu/ϳcpatton/maxc.html). Microsomes of 20 g/ml concentration were added, and Ca 2ϩ uptake was initiated by the addition of 5 mM ATP. The reaction was terminated by rapid filtration of the microsomes using Millipore membrane filters (0.45-m pore size).

Transfected hPMCA4b Is Cleaved and Activated in Detached COS-7 Cells That Undergo Spontaneous Apoptosis-After
COS-7 cells were transiently transfected with hPMCA4b and cells became confluent, many of them detached from the flask. Several reports have shown that loss of cell adhesion might result in suspension-induced apoptosis (38,39). Immunoblot analysis of the floating cell lysates also supported this idea. We used an anti-PARP polyclonal antibody that specifically recognizes both the intact 116-kDa PARP protein and its 85-kDa fragment generated by caspase-3 in the beginning of apoptotic events. As shown in Fig. 1A, adherent COS-7 cells express the intact PARP protein, whereas in detached cells only the apoptosis-induced 85-kDa fragment was found. The detached cells didn't lose their membrane integrity, because they were not stained by the DNA dye trypan blue. This indicated that the cells were undergoing apoptosis rather than necrosis.
We also examined the hPMCA4b protein produced in transfected cells to see whether suspension-induced apoptosis affected its expression. Immunostaining by the anti-PMCA monoclonal antibody 5F10 revealed that the full-length 134-kDa hPMCA4b protein, typically expressed by the attached cells, was degraded to an ϳ120-kDa fragment in the floating apoptotic cells (Fig. 1B). The ratio of the intact and degraded PMCA4b protein varied from one preparation to the other in the detached cells, so that the protein remaining intact was typically between about 5 and 30% of the total expressed pump (for a comparison see also Fig. 3). To explore if the pump was truncated C-or N-terminally, Western blot analysis of the detached cell lysates was also performed by two other PMCAspecific antibodies. Monoclonal antibody JA9 recognizes both PMCA4 variants (PMCA4a and 4b) at a common N-terminal sequence, while monoclonal antibody JA3 binds to a specific C-terminal region of PMCA4b. In contrast to monoclonal antibody JA9, which stained the 120-kDa apoptotic fragment well, the JA3 antibody did not recognize this fragment (data not shown) indicating that the C-terminal regulatory domain was cut.
As shown in Fig. 1B, the size of the apoptotic fragment matched exactly the size of a C-terminally truncated hPMCA4b mutant (termed ct120), which was constructed by deletion of the C-terminal autoinhibitory sequences of the pump. Conse- quently, this mutant is fully active without calmodulin (32), which raises the question whether the apoptotic fragment in detached COS-7 cells is also released from the autoinhibition. We prepared microsomal membrane fraction from the floating cells and determined the transport activity of the truncated hPMCA4b protein in the absence of calmodulin. In these experiments we used those preparations in which more than 90% of the expressed PMCA4b protein was degraded, so that the characteristics of the intact protein did not interfere with those of the 120-kDa fragment. The results are summarized in Fig.  1C. In contrast to the full-length hPMCA4b, both the ct120 mutant and the 120-kDa apoptotic fragment showed maximal activity without calmodulin.
Caspase-3, one of the key effector molecules of apoptosis, prefers substrates with a DEXD recognition sequence. Analyzing the primary amino acid sequence of hPMCA4b we found a caspase-3 consensus motif ( 1077 DEID 1080 ), which is located a few residues upstream of the N terminus of the C-terminal regulatory domain (Fig. 2).
STS-induced Apoptosis Results in a Caspase-3-related hPMCA4b Fragment in COS-7 Cells-To investigate the apoptosis-induced fragmentation of hPMCA4b, we treated transiently transfected COS-7 cells with the protein kinase inhibitor STS, which has been shown to induce apoptosis in various cells (40 -42). After 1.5-to 5-h incubation with 1 M STS, floating cells were removed and only the attached cells were harvested. Immunoblot analysis of the STS-treated cell lysates with the antibody 5F10 revealed the degradation of the fulllength hPMCA4b in a time-dependent manner. Parallel to the decay of the intact pump, increasing formation of the 120-kDa fragment could be detected (Fig. 4A). We also followed the course of STS-induced apoptosis by observing the degradation of the 116-kDa PARP protein (Fig. 4B). The cleavage of hPMCA4b correlated well with the caspase-3-dependent generation of the 85-kDa PARP cleavage product.
To examine whether caspase-3 is involved in hPMCA4b cleavage, we pretreated the cells with a cell-permeable caspase-3 inhibitor (Z-DEVD-FMK). Before STS treatment, cells were preincubated with 120 M caspase inhibitor for 1 h. This was followed by STS treatment for additional 5 h. The results indicated that caspase-3 inhibitor prevents STS-induced degradation of hPMCA4b (Fig. 4C).
To verify that the caspase-3 consensus sequence in hPMCA4b is required for apoptosis-dependent cleavage of the pump, we performed experiments in STS-treated cells with the mutant hPMCA4b proteins. Transfection was carried out with either single ( 1077 AEID 1080 and 1077 DEIA 1080 ) or double ( 1077 AEIA 1080 ) mutant DNAs. COS-7 cells expressing the mutant proteins were exposed to a 5-h STS treatment, then cells were harvested, lysed, and analyzed as described above. As before, STS-mediated apoptosis could be easily detected in each sample by the specific caspase-3-mediated cleavage of the PARP protein, independently of which mutant the cells were   expressing (Fig. 5B). In contrast, immunoblot analysis by 5F10 showed no degradation of the mutant pump proteins (Fig. 5A). Based on these results, we considered that the cut of hPMCA4b occurs at the caspase-3 consensus sequence 1077 DEID 1080 during STS-induced apoptosis.
Degradation of hPMCA4b in STS-treated HeLa Cells-To determine whether endogenously expressed hPMCA4b has the same properties during apoptosis as the transiently expressed pump protein, we investigated the STS-induced apoptosis of non-transfected epithelial HeLa cells, which originally contain hPMCA4b and hPMCA1b isoforms. We performed the same experiments as before, except that we excluded the hPMCA1b isoform by analyzing the blots with monoclonal antibody JA9 instead of 5F10. Antibody JA9 recognizes exclusively the PMCA4 isoform at the N-terminal portion, thus we could follow the time-dependent degradation solely of hPMCA4b in STStreated HeLa cells. The results presented in Fig. 6 (A and B) were very similar to those of the COS-7 cells transfected with hPMCA4b: 1) The pump protein was digested in a time-dependent manner; 2) The PARP protein simultaneously gave its apoptosis-related 85-kDa fragment; 3) The specific caspase-3 inhibitor (Z-DEVD-FMK) inhibited the degradation of hPMCA4b fully and that of PARP only partially. The partial inhibition of PARP may be due to difficulties in penetration of the inhibitor to the site required for its action. The typical 120-kDa fragment was also seen in HeLa cells, which were spontaneously detached from the flask's surface (data not shown), similarly to that found in the case of the COS-7 cells.
In addition, to verify that STS-induced formation of the 120-kDa fragment was the result of apoptosis, we examined another early apoptotic event, the exposure of the phosphatidylserine molecules on the cell surface. The flipping of these molecules from the inside of the cell membrane to the outside is quite an early step of the apoptotic processes, and it can be easily detected by the specific binding of the fluorescein isothiocyanate-labeled annexin V protein (43). To distinguish between early apoptotic cells and late apoptotic or necrotic cells, we used annexin V in combination with PI and analyzed cells by flow cytometry. Annexin V-positive and PI-negative staining refers to the early stage of apoptosis, whereas PI-positive cells have lost their membrane integrity and consequently PI-positive staining represents late apoptosis and necrosis. Table II indicates that 12% of untreated cells were stained by annexin V alone (background staining), whereas after 5 h of STS treatment 55% of the cells became annexin V-positive and PI-negative. The ratio of PI-positive HeLa cells was unaffected by STS treatment. Taking these data into consideration, we concluded that STS induced apoptosis in HeLa cells. We got very similar results with transfected COS-7 cells (data not shown).
In Vitro Cleavage of hPMCA4b and Its Mutants with Recombinant Caspase-3-To further demonstrate that hPMCA4b is cleaved and activated by caspase-3, a mixed microsomal membrane preparation of the transfected COS-7 cells was treated with recombinant caspase-3 in vitro for various periods of time. At the end of each period, a portion of the samples was analyzed by immunostaining, whereas another portion was measured for transport activity in the presence or in the absence of calmodulin. Results are represented in Fig. 7 (A and B). An immunoblot of the samples shows a time-dependent decay of the intact pump and consequent formation of the 120-kDa  apoptotic fragment. The maximal rate of transport activity of each corresponding sample was measured in the presence of calmodulin whereas the activity of caspase-3-mediated fragment was measured in the absence of calmodulin. (The basal activity of the intact pump protein is between 10 and 20% of the maximum.) The increasing quantity of the 120-kDa fragment resulted in increasing transport activity without calmodulin. This is in good accordance with the results shown in Fig. 1C, in which we showed the Ca 2ϩ dependence of activity for a fully proteolyzed enzyme without calmodulin.
To confirm that the cleavage site of hPMCA4b is at the 1077 DEID 1080 consensus sequence of the pump, membrane preparations containing either the wild type pump protein or its mutants ( 1077 AEID 1080 , 1077 DEIA 1080 , or 1077 AEIA 1080 ) were incubated with 0.25 g of caspase-3 for 1 h (Fig. 7C). As expected, the enzyme was unable to cleave any of the mutant proteins, whereas the cleavage of the wild type pump was pronounced.
We considered it necessary to establish that the 120-kDa apoptotic fragment of hPMCA4b could not result from the activity of calpain. For this reason, we digested microsomal membrane preparations containing hPMCA4b with either recombinant caspase-3 or purified porcine erythrocyte -calpain in vitro. Fig. 8 demonstrates that we could separate the caspase-3 and calpain-related fragments (120 and 126 kDa, respectively) on SDS-PAGE. Moreover, we could also produce the 126-kDa fragment by digestion of the 1077 DEIA 1080 mutant hPMCA4b with -calpain, indicating that digestion with caspase-3 occurs indeed at a different location of the C terminus and the cleavage with -calpain occurs at an independent site. DISCUSSION We report here for the first time that the plasma membrane Ca 2ϩ ATPase isoform 4b (PMCA4b) is cleaved by caspase-3 in cells undergoing apoptosis. We identify the caspase-3-cleavage site in PMCA4b and demonstrate that cleavage at this site causes irreversible activation of the Ca 2ϩ transport activity of the enzyme.
Among the P-type ATPases, PMCAs have a unique property: Calmodulin and some additional cellular regulators such as protein kinases and calpain control its activity. In the case of hPMCA4b a sequence of about 120 residues at the C terminus is responsible for all these regulations. The N-terminal region of this sequence has a 28-residue-long high affinity calmodulinbinding domain, which also serves as an autoinhibitor of the enzyme. At low intracellular Ca 2ϩ concentrations, when calmodulin is removed, the calmodulin-binding domain interacts with the catalytic core and inhibits its activity. Upon stimulation, when the intracellular Ca 2ϩ concentration rises inside the cell, calmodulin binds to the calmodulin-binding domain. That binding frees the catalytic core from the intramolecular interaction, and the pump becomes activated. It has been demonstrated that removal of the C terminus by proteolysis or mutagenic truncation causes irreversible activation of hPMCA4b. Among the proteases studied so far, calpain is the only physiologically relevant enzyme that cuts the pump at two or three different locations at the C terminus and causes partial or complete activation (44 -46). This is the first report showing that a caspase-3 consensus sequence 1077 DEID 1080 exists just five residues upstream of the sequence of the calmodulin-binding domain of hPMCA4b. The DEXD consensus sequence appears to be unique to PMCA4, because the corresponding sequences in the other isoforms (PMCA1-3) are all EEID. Consistent with this difference, they appear to be more resistant to caspase cleavage. 2  clearly demonstrated that caspase-3 plays a key role in mammalian apoptosis. Caspase-3 is activated by a wide variety of apoptotic challenges, and an increasing number of substrates have been identified. Sometimes cleavage results in activation of the substrates, as in the case of apoptosis-related endonucleases (e.g. caspase-activated deoxyribonuclease (CAD) and its inhibitor (ICAD)) (47,48) and protein kinases (e.g. PKC␦, PKC, and PKC) (7,49,50), often by removal of an inhibitory domain or subunit.
Our first evidence indicating that PMCA4b is cut and activated upon apoptosis was that in transiently transfected COS-7 cells, which were spontaneously detached from the surface, the expressed PMCA4b was degraded to a 120-kDa fragment. We showed that these cells were undergoing spontaneous, suspension-induced apoptosis rather than necrosis. A similar fragment of 120 kDa was seen in COS-7 cells expressing PMCA4b and in HeLa cells containing endogenously expressed PMCA4b when apoptosis was induced by STS. The kinetics of the formation of the PMCA4b fragment was similar to those of the fragmentation of PARP, a widely known caspase-3 substrate, indicating that degradation of PMCA4b is an early apoptotic event. The molecular size of the 120-kDa fragment corresponded well to that of the ct120 mutant, and it showed similar characteristics: It was fully active in the absence of calmodulin, and no additional activation with calmodulin occurred. The ct120 mutant ends just five residues downstream of the 1077 DEID 1080 caspase-3 consensus sequence (Fig.  2); therefore, we presumed that hPMCA4b is degraded by caspase-3 or caspase-3-like proteases. This was confirmed by the following experiments: 1) When caspase-3 inhibitor was added to the incubation media before inducing apoptosis by STS, the inhibitor prevented the degradation of the pump. 2) In vitro digestion of the pump by a recombinant caspase-3 enzyme gave the same 120-kDa fragment as in apoptotic COS-7 and HeLa cells. 3) Site-directed mutagenesis of the aspartate residues of the 1077 DEID 1080 consensus sequence prevented the appearance of the 120-kDa fragment. The apparent molecular weight of this fragment was lower than that seen when hPMCA4b was cleaved by calpain. The calpain fragment(s) were definitely higher with a molecular mass of about 124 -126 kDa. Thus, our experiments demonstrated that hPMCA4b is a newly identified substrate for caspase-3 and the cleavage site is at the 1077 DEID 1080 caspase-3 consensus sequence. Cleavage of the pump by caspase-3 resulted in the formation of a 120-kDa constitutively active fragment.
Several lines of evidence have indicated that alteration in intracellular calcium homeostasis plays an important role in apoptosis (14 -25). The effect of moderate elevation of calcium concentration was shown to be either pro-or antiapoptotic, whereas a sustained excessive increase in intracellular Ca 2ϩ concentration could be apoptotic or necrotic depending on the cell type. The functions of many important elements of calcium signaling pathways during apoptosis were investigated by various research groups. It has been demonstrated that depletion of intracellular calcium stores can directly induce apoptosis in Chinese hamster ovary cells (51). A moderate decrease in the ER Ca 2ϩ content, such as that caused by Bcl-2, on the other hand, protected HeLa cells from apoptosis due to reduced Ca 2ϩ release from the ER during cell stimulation and consequently to a smaller increase in both intracellular and mitochondrial Ca 2ϩ levels (25). The IP 3 Ca 2ϩ release channels, type 1 and 3, have also been implicated as critical mediators of apoptosis. T cells deficient in type 1 receptor were resistant to apoptosis (52), and antisense constructs to receptor type 3 prevented programmed cell death in lymphocytes (53) and dorsal root ganglia neurons (54). It has also been reported (11) that the IP 3 receptor isoform 1 is a specific substrate for caspase-3 and that cleavage of the receptor inactivates its channel function. Although plasma membrane calcium ATPases have a significant role in calcium homeostasis, their destiny and function during apoptosis have not been considered.
In vitro experiments on PMCA4b indicate that an irreversible, proteolytic activation of the protein substantially changes its response to increases in Ca 2ϩ concentration. Early work of Scharff and Foder (55) showed that activation of PMCA4b is slow. We have demonstrated that, because of the slow response of the pump to calmodulin, the rate of response to increases in Ca 2ϩ concentration is also slow. In contrast to the intact PMCA4b, the activation by Ca 2ϩ of a truncated mutant (ct120) that lacks the whole regulatory region of PMCA4b was immediate (30). Because this mutant ends five residues downstream of the 120-kDa caspase fragment, we conclude that the caspase-3 fragment will also respond quickly to the increase in Ca 2ϩ concentration. An experiment by Foder and Scharff (56) may shed light on the physiological consequences of the formation of a fast responding PMCA4b fragment. Using red cell membrane ghosts, they showed that when the erythrocyte pump (which is mainly PMCA4b) was activated by tryptic digestion, no increase in intravesicular Ca 2ϩ concentration occurred if Ca 2ϩ entry was triggered by the Ca 2ϩ ionophore A23187. In contrast, slow activation of the intact protein by calmodulin allowed a transient increase in intravesicular Ca 2ϩ concentration. Moreover, when the activity of the pump was inhibited by calmodulin antagonist, the intravesicular Ca 2ϩ increased quickly to a high steady-state level, as expected. Thus, in a model system in vitro, activation of the pump by proteolysis was able to diminish completely a rather excessive intravesicular Ca 2ϩ load.
The situation in vivo is far more complex, and the distribution of different PMCA isoforms as well as the mechanism of Ca 2ϩ signaling varies greatly from one cell type to another. Therefore, based on the in vitro experiments we cannot propose a simple model for the role of PMCA in cells undergoing apoptosis. We may hypothesize, however, that cells in which PMCA4b is the major PMCA isoform will respond quicker to the increase in intracellular Ca 2ϩ concentration if they produce a constitutively active fragment of this protein during apoptosis. This mechanism may defend these cells from an excessive load of Ca 2ϩ in the cytosol and in the mitochondria, protecting them from a more rapid destruction. Our hypothesis, that an increased activity of PMCA may protect cells from destruction, is supported by two independent observations: 1) Overexpression of PMCA4b in HeLa cells prevented mitochondrial damage and protected these cells from ceramide-induced apoptosis (25). The authors attributed this effect to a reduced ER Ca 2ϩ content due to overexpression of PMCA4b. 2) Additionally, PMCA4boverexpressing clones of PC12 cells were less vulnerable to Ca 2ϩ -mediated cell death whereas antisense clones were more susceptible (57).
Finally, the PMCA "b" variants, isoforms PMCA1b, PMCA2b, PMCA3b, and PMCA4b, have a PDZ binding sequence at the extreme C terminus that allows them to interact with specific PDZ domain-containing proteins. It has been demonstrated that the C-terminal four residues (ETSV) of PMCA4b interact strongly with PDZ domains of membrane-associated guanylate kinase (MAGUK)-related proteins (58,59). These workers also demonstrated the exclusive presence and co-localization of PMCA4b and SAP97 in the basolateral membrane of polarized Madin-Darby canine kidney cells. In a recent report, Zabe and Dean (60) suggested that the PMCA4b C-terminal PDZ binding sequence is involved in the association with the actin cytoskeleton of platelets and that the association in-creases dramatically upon activation with thrombin. The PDZbinding tail of PMCA4b was also shown to interact with the PDZ domain of nitric-oxide synthase and through this interaction PMCA4b dramatically inhibits nitric oxide synthesis (61). Removal of the PDZ-binding motif of PMCA4b by caspase-3 will disrupt these interactions that might result in the redistribution of this protein in the plasma membrane and/or an increased nitric oxide production upon apoptosis.
In conclusion, we showed here for the first time that a critical element of the intracellular Ca 2ϩ homeostasis, the plasma membrane Ca 2ϩ pump, is cleaved and activated by caspase-3mediated proteolysis during apoptosis. This cleavage removes the entire calmodulin-binding domain and frees the pump from autoinhibition. Based on previous experiments we suggest that this constitutively active form of PMCA4b will remove Ca 2ϩ quickly from the cytosol and thus, will prevent Ca 2ϩ accumulation more efficiently in the apoptotic cells.