Constitutively Active Gαq and Gα13 Trigger Apoptosis through Different Pathways*

We investigated the effect of expression of constitutively active Gα mutants on cell survival. Transfection of constitutively active Gαq and Gα13 in two different cell lines caused condensation of genomic DNA and nuclear fragmentation. Endonuclease cleavage of genomic DNA was followed by labeling the DNA fragments and subsequent flow cytometric analysis. The observed cellular phenotype was identical to the phenotype displayed by cells undergoing apoptosis. To distinguish between the apoptosis-inducing ability of the two Gα-subunits, the signaling pathways involved in this cellular function were investigated. Whereas Gαq induced apoptosis via a protein kinaseC-dependent pathway, Gα13 caused programmed cell death through a pathway involving the activation of the small G-protein Rho. Both of the pathways leading to apoptosis were blocked by overexpression of bcl-2. In contrast to other apoptosis-inducing systems, expression of constitutively active Gαq and Gα13 triggered apoptosis in high serum as well as in defined medium.

There are two distinct modes of death in cellular systems. In one, an insult leading to irreversible cellular injury causes a phenotype called necrosis that is manifested by the rupture of the cell membrane. In the other, under certain physiological situations, death of cells is triggered by metabolic or developmentally programmed events and may be required for the organisms survival or for differentiation (1). This programmed cell death (apoptosis) occurs when a cell dies by a mechanism initiated by proteins encoded by its own genome.
Apoptosis can be distinguished from necrosis through morphological characteristics including cell shrinkage, chromatin condensation, activation of specific proteases and endonucleases, and fragmentation of genomic DNA (2). Apoptosis has been implicated in many important biological processes including immune defense, growth control, and development (3). In addition some events leading to human disease involve apoptosis (4,5).
The execution of apoptosis is triggered by intrinsic signal transduction events that link changes in physiological conditions to the cell death machinery. Signal transduction through seven pass membrane receptors represents a common mechanism of eukaryotic signaling and physiological control. During this process heterotrimeric G-proteins are responsible for transducing a ligand binding event at the membrane into a cellular response (6,7). There are four different classes of G␣-subunits that contribute to heterotrimeric G-proteins mediated responses (8). In addition to tight regulation of activation of signal transduction pathways the cell also contains a carefully regulated system to shut down signaling after prolonged exposure to ligands. This desensitization is primarily achieved by phosphorylation induced inactivation of the receptor and subsequent internalization of the phoshorylated receptor (9). Recently a desensitization mechanism directly acting at the ␣-subunits was discovered. GAP proteins were cloned that accelerate the intrinsic GTPase activity of ␣-subunits thereby shutting off the signal (10).
In some situations permanently elevated levels of second messengers are observed inside the cells. It is believed that these elevated activities contribute to the execution of pathological phenotypes (11). Recent data point to the involvement of the G␣-subunit, G␣o, in the execution of apoptosis triggered by a mutated and presumably constitutively active amyloid precursor protein (APP) in Alzheimer's disease (12).
To investigate the effect of other constitutively active Gprotein subunits, we examined their ability to cause apoptosis in vivo. Constitutively active mutants of G␣q, G␣12, G␣13, and G␣i2 were expressed in CHO 1 and COS-7 cells, and the resulting cellular phenotypes were observed.

EXPERIMENTAL PROCEDURES
Materials-The dye Hoechst 33258 was obtained from Sigma. The Texas red conjugated goat antibodies to rabbit immunoglobulin G (IgG) and the polyclonal antibodies to ␤-galactosidase were received from Cappel and 5 Prime 3 3 Prime, Inc., Boulder, CO, respectively. Profectin Ca 3 (PO 4 ) 2 Transfection kit was purchased from Promega. Protein kinase C inhibitors and EGTA/AM were received from Calbiochem. Cell culture and transfection reagents including DMEM, FBS, serum free medium (Optimem), LipofectAMINE, and trypsin/EDTA were purchased from Life Technologies, Inc., whereas Ham's F-12 was received from Irvine Scientific. The Apo-Direct kit was obtained from Phoenix Flow Systems. The peptide inhibitor, benzyloxycarbonyl-Val-Ala-Aspfluoromethylketone (z-VAD.fmk), was purchased from Enzyme Systems Products. The in situ cell death detection kit was obtained from Boehringer Mannheim, and the Vectashield mounting medium containing 4Ј,6-diamidino-2-phenylindole was from Vector Laboratories.
Cell Culture and Maintenance-Cells were grown in 5% CO 2 and DMEM, 10% FBS (COS-7), or Ham's F-12, 10% FBS (CHO-K1), respectively. Transfection protocols for COS-7 and CHO-K1 cells with Lipo-fectAMINE were described previously (13). Cells were kept in serumfree medium during LipofectAMINE transfection for 5 h and followed by the addition of an equal volume of DMEM, 20% FBS. Transfection of COS-7 cells with the Perfectin kit was performed according to the manufacturer's instructions in DMEM, 10% FBS. Treatment of COS-7 cells with z-VAD.fmk was performed according to the peptide manufacturer's directions. Immediately after transfection of cells using lipo-* This work was supported by National Institutes of Health Grant GM 34236. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  fectAMINE, z-VAD.fmk was added to a final concentration of 50 M. Cells were maintained in medium containing 50 M z-VAD.fmk for approximately 41 h. New aliquots of the peptide inhibitor were added 4 times during the course of the incubation period to ensure that the effectiveness of the inhibitor was maximized.
Expression Plasmids-Expression plasmids for G␣13QL, G␣qRC, LacZ, rasN19, and rhoAN17 were previously described (14 -16). The cDNA of the constitutively active G␣i2QL mutant or of bcl-2 was inserted into the pcDNA expression vector (Invitrogen) or the Pac expression vector to yield an expression plasmid for G␣i2QL or bcl-2, respectively (17).
Analysis of Apoptotic Phenotypes-Cells were grown on glass coverslips and fixed at 48 h after transfection by incubation in methanol/ acetone for 2 min. Immunostaining of ␤-galactosidase and staining of the genomic DNA were performed as described (18). Cells expressing LacZ were counted using a Zeiss Axiovert 35 fluorescence microscope. 200 -1200 cells were counted in each experiment. Preparation of COS-7 cells for flow cytometric analysis were done with the Apo-Direct kit following the manufacturer's instructions. The total cell population (transfected and non-transfected cells) was used for the flow cytometric analysis. The total amount of DNA used in the assays was kept constant. Terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL) assay was performed as described in the in situ detection kit manufacturers' instructions. Coverslips were mounted on glass slides using Vectashield mounting medium containing 4Ј,6-diamidino-2-phenylindole to visualize the nuclear morphology of the cells.

Expression of Constitutively Active G␣13 and G␣q but not
G␣i2 Triggers Apoptosis-Mutations in the catalytic domain of G␣-subunits have been described which inhibit their intrinsic GTPase activity and therefore convert these proteins into con-stitutively active ␣-subunits (19). To investigate the phenotype of cells transfected with constitutively active G␣-subunits, the nuclear morphology of the transfected cells was analyzed. The characteristic nuclear phenotype exhibited by apoptotic cells (nuclear fragmentation and condensation of genomic DNA) was used to distinguish between normal and apoptotic cells (Fig. 1). To eliminate the background of non-transfected cells, cells were cotransfected with expression plasmids for the constitutively active G␣-subunit and LacZ. Only cells expressing LacZ were counted during the experiment. Expression of the various proteins was verified by Western blot analysis (data not shown). In general, a large fraction of singly transfected cells are also cotransfected with a second plasmid. Thus, there is a high probability that LacZ positive cells are also expressing the cotransfected G-protein expression plasmid.
Expression of constitutively active G␣i2QL in COS-7 cells does not increase the number of apoptotic cells over that observed in control experiments where LacZ alone was expressed. In contrast, expression of constitutively active G␣qRC and G␣13QL increased the appearance of the apoptotic phenotype dramatically. An approximate three-fold rise in the percentage of apoptotic cells was observed. Both G␣qRC and G␣13QL expression triggered programmed cell death in nearly 30% of the transfected cells ( Fig. 2A). Varying the amount of expression plasmid (0.05-1 g) revealed no differences in the occurrence of the apoptotic phenotype in our assays (data not shown). Thus, differences in the ability to trigger apoptosis does not appear to be dependent on expression levels of the To rule out the effect of cell line specific mutations that could lead to differences in cell survival, CHO cells were also transfected with expression plasmids for G␣qRC, G␣13QL, and G␣i2QL. Again in the CHO cell system the expression of G␣qRC and G␣13QL led to significant apoptotic cell death, whereas G␣i2QL expression did not (Fig. 2B). Thus, the results with the CHO cells correspond to the data obtained using the COS-7 cells and, together they suggest that the ability of G␣13 and G␣q to induce programmed cell death in tissue culture is cell type independent. A second assay system was employed to further confirm the correlation between the expression of G␣qRC and G␣13QL and apoptosis. The Apo-Direct assay (Phoenix Flow Systems) was used to detect increased activity of an endonuclease, an additional characteristic of apoptotic cell death, in the transfected cells. In this assay fluorescently labeled nucleotides are incorporated into the ends of the fragmented, genomic DNA via terminal transferase. Therefore an increase in fluorescence intensity indicates an increase in the amount of apoptotic cells. COS-7 cells expressing G␣qRC and G␣13QL exhibit an increase in incorporation of fluorescently labeled nucleotides (Fig. 3). This is consistent with the observed morphological nuclear changes in the transfected cells and confirms the notion that G␣13QL and G␣qRC expression causes apoptosis.
Induction of Apoptosis by G␣qRC and G␣13QL Occurs in High Serum-Most proteins involved in both apoptotic and non-apoptotic signaling trigger an apoptotic phenotype only in low serum. This is thought to be caused by the imbalance of proliferative signals given by overexpression of these proteins and by cell cycle arrest imposed by serum starvation (20). Whereas activation of G␣q causes a strong proliferative response, G␣13 has only a modest effect on cell proliferation (21). LipofectAMINE transfection requires incubation of the transfection mixture for 5 h in serum-free medium. Although 5 h in serum-free medium is not sufficient for the cells to cease progression through the cell cycle, another transfection protocol was used to keep the cells in high serum throughout the experiment. Therefore, to determine if the expression of G␣13QL and G␣qRC also cause an imbalance in cell cycle signaling, transfections were performed with Ca 3 (PO 4 ) 2 in COS-7 cells in high serum. Even at continuous culture in 10% FBS, expression of G␣13QL and G␣qRC increased the appearance of the apoptotic nuclear phenotype 2-3-fold compared with the amounts observed in control experiments (Fig. 4). These data confirm that constitutive activation of G␣13 and G␣q can activate pathways directly targeting the cell death machinery.
Signaling Pathways Activated by G␣q and G␣13 Leading to Apoptosis Are Different but Converge at a Step Controlled by bcl-2-To investigate the signaling mechanism connecting the expression of G␣13QL and G␣qRC to apoptosis, the effects of inhibitors of different pathways on apoptotic cell morphology were examined in COS-7 cells. Although some of the cellular responses caused by activation of G␣13 and G␣q are similar, G␣q and G␣13 activate different pathways. Whereas the target of G␣q signaling is the family of phospholipase C-␤ enzymes, G␣13 does not activate this class of enzymes (21)(22)(23)(24). In contrast, several groups have established that a major target of G␣13 signaling is the small G-protein Rho (25). Therefore, to test the involvement of the Rho and the phospholipase C-␤ pathways in the induction of apoptosis by G␣qRC and G␣13QL, two different approaches were employed. First cells were cotransfected with expression plasmids for G␣13QL or G␣qRC and a dominant interfering RhoA mutant (RhoAN19) (Fig. 5A). Coexpression of RhoAN19 with G␣13QL compared with G␣13QL expression yielded a significant reduction in the amount of transfected cells displaying an apoptotic nuclear phenotype. This reduction was not observed when G␣qRC and RhoAN19 were coexpressed (Fig. 5A).
To rule out a general effect of the expression of dominant negative small G-proteins, the effect of coexpression of a dominant negative Ras mutant (RasN17) in this assay system was investigated (19). Neither G␣qRCnor G␣13QL-induced apoptosis was reduced by coexpression of RasN17 demonstrating that the effect of RhoAN19 on G␣13QL-triggered apoptosis was specific for Rho.
In the second approach toward analyzing these pathways, protein kinase C inhibitors were added to cells expressing constitutively active G␣qRC and G␣13QL mutants. Calphostin C and Ro31-8220 were shown to inhibit all isoforms of PKC at IC 50 of 50 or 10 nM, respectively. Thus, 50 -100-fold higher concentrations of the respective inhibitors were used in these assays. Inhibition of other kinases by these inhibitors is effective only at approximately 500 -1000-fold higher concentrations (26,27). Whereas both Calphostin C and Ro31-8220 diminished the amount of apoptotic cells caused by G␣qRC expression 2-3-fold, G␣13QL-induced programmed cell death was not significantly altered by the addition of the PKC inhibitors (Fig. 5B). This indicates that the PKC inhibitors specifically interfere with G␣q signaling which leads to apoptosis.
Although G␣13 and G␣q obviously activate different signaling mechanisms, these pathways may converge to initiate the apoptotic cellular response. Signal transduction pathways triggering the activation of the cell death machinery share common steps, which in many cases are inhibited by expression of bcl-2 (28,29). Expression of bcl-2 was shown to inhibit apoptosis in several cell systems by sequestering molecules of the apoptotic pathway. To test the effect of bcl-2 on G␣q and G␣13 activated pathways, COS-7 cells were cotransfected with expression plasmids for G␣13QL or G␣qRC and bcl-2. Both G␣13QL-and G␣qRC-induced apoptosis were greatly reduced by overexpressed bcl-2. The observed 3-fold reduction again confirms the connection of pathways activated by G␣13 and G␣q to the cell death machinery (Fig. 6). This also reveals that both pathways must converge upstream of a point controlled by bcl-2 (Fig. 7).
Activation of the interleukin-1␤-converting enzyme-like family of cysteine proteases is thought to play a central role in the programmed cell death pathway (for review see Ref. 5). Proteolysis of their many target substrates may be responsible for the eventual demise of the cell. The interleukin-1␤-converting enzyme-like family of proteases or caspases can be inhibited by specific peptide inhibitors (30). To determine if activation of caspases is involved in the observed increase in cell death induced by G␣13QL and G␣qRC, COS-7 cells were cotransfected with LacZ and G␣13QL or G␣qRC in the presence of 50 M of z-VAD.fmk, an irreversible inhibitor of the caspase family of cysteine proteases. The presence of the peptide inhibitor prevented the induction of cell death by both G␣13QL and G␣qRC when measured using the TUNEL assay, supporting the notion that activation of a caspase(s) is involved in G␣13QLand G␣qRC-induced apoptosis (Table I). Although, the peptide inhibitor prevented DNA fragmentation, nuclear integrity still seemed to be affected. Many cells appeared to display some of the morphological characteristics of apoptosis, nuclear fragmentation and DNA condensation. These results are similar to findings previously reported demonstrating that in BAX-induced programmed cell death inhibition of interleukin-1␤-converting enzyme-like proteases with z-VAD.fmk prevented DNA fragmentation but not many of the other morphological changes associated with apoptosis (31). DISCUSSION Collectively, our results suggest that expression of constitutively active G␣13QL or G␣qRC mutants trigger apoptosis in higher eukaryotic cell lines. Different cell lines contain different mutations that interfere with the cell death machinery and render these lines immortal. G␣13QL and G␣qRC expression caused programmed cell death in two different cell lines, thus affecting a common pathway leading to apoptosis. In addition, the observed increase in apoptotic cell number correlates with data obtained with the expression of other proteins or with extracellular stimuli causing apoptosis, e.g. with expression of FIG. 4. Expression of G␣qRC and G␣13QL triggers apoptosis also via Ca 3 (PO 4 ) 2 transfection in high serum. COS-7 cells were cotransfected with pCisLacZ and the expression plasmid for the constitutively active G␣-subunits or an empty expression vector (total 1.5 g DNA, 1 ϫ 10 5 cells). Transfection was performed with Ca 3 (PO 4 ) 2 and continuous culture in DMEM with 10% FBS as noted under "Experimental Procedures." 48 h after transfection, transfected cells were counted, and the percentage of cells displaying an apoptotic nuclear phenotype was determined. MEKK1 in PC12 cells or tumor necrosis factor-␣ stimulus in U937 cells (18,32). The ability of G␣qRC and G␣13QL to activate the cell death machinery also in high serum distinguish our data from other studies where apoptosis was only achieved by simultaneous expression of the respective protein and incubation in low serum. Therefore, G␣13QL and G␣qRC seem to activate signaling pathways directly linked to the cell death machinery.
Activation of G␣q can be correlated with an increase in PKC activity and elevation of internal Ca 2ϩ concentration as a result of phospholipase C-␤ activation. The role of PKC activation in apoptosis is not completely clear. In some cell systems PKC activation inhibits apoptosis, whereas in others PKC activation can be correlated with the onset of apoptosis (33)(34)(35). In our assay system, the PKC inhibitors totally abolished the apoptosis-inducing ability of G␣q. This reveals that PKC signaling is an essential step in G␣q triggered apoptosis. Elevation of internal Ca 2ϩ levels is often observed in cells dying from apoptosis (36). Therefore, to determine if G␣qRC induced an in-crease in cytoplasmic Ca 2ϩ levels that may be associated with apoptosis, COS-7 cells expressing G␣qRC were incubated with the Ca 2ϩ chelator EGTA/AM. Incubation with 10 M of this Ca 2ϩ chelator provokes only a slight decrease in the number of apoptotic cells, presumably due to the low concentration of EGTA/AM used (data not shown). Higher concentrations of the Ca 2ϩ chelator could not be used because of its cytotoxic action at higher doses during prolonged incubation (37). Interestingly, a neural disorder, amyotrophic lateral sclerosis, causing neural cell death presumably by apoptosis is accompanied by an elevation in PKC activity and cytoplasmic Ca 2ϩ levels (11). The cause of this disease is currently unknown.
Sustained activation of G␣q was recently shown to cause transformation of NIH3T3 cells (38). This study also showed that a large fraction of the transfected cells died due to expression of constitutively active G␣q. These data can now be explained by the correlation between G␣qRC and apoptosis that we found. The transformed phenotype might occur in cells that accumulate mutations affecting the cell death machinery. Recently it was demonstrated that sustained activation of Rho triggers programmed cell death in low serum (39). In contrast, our data show that coexpression of RhoAN19 and G␣13QL lead to a significant but not total reduction of apoptosis to background levels in the cell population (Fig. 5A). Therefore, it is likely that RhoAN19 expression levels are too low to completely prevent Rho-induced apoptosis. On the other hand, G␣13 might activate an additional pathway that contributes to the Rho signaling in inducing apoptosis in high serum.
Our data clearly demonstrate that G␣13QLand G␣qRCtriggered apoptosis can be blocked by overexpression of bcl-2. Thus, these pathways enter the cell death signaling cascade upstream of a point controlled by bcl-2. In conclusion, our results demonstrate the ability of G␣-subunits to induce programmed cell death through different pathways that, however, converge before the final execution of cell death.
We also performed experiments where constitutively active G␣12QL was expressed in COS-7 cells. A significant fraction of the cells showed the apoptotic phenotype, however, the amount was well below the fraction of apoptotic cells obtained with expression of G␣qRC and G␣13QL (20% Ϯ 6.08; data not shown). Recently it was demonstrated that G␣13 and G␣12 recruit different signaling pathways to activate Na ϩ /H ϩ ex-changers (40). In addition, null mutation of G␣13 in mice displays a distinct phenotype, which also argues against redundant functions of G␣13 and G␣12 (41). Therefore, it is likely that activated G␣13 transmits signals to the cell death machinery through pathways not recruited by G␣12.
Our study clearly demonstrates that sustained activation of G␣q or G␣13 can trigger apoptosis in different cell systems. Several human diseases are thought to be caused by a sustained stimulus that cannot be shut down through desensitization processes. The role of sustained signaling through G␣q or G␣13 in development and in disease remains to be more clearly defined. Expression of constitutively active G␣q and G␣13 trigger different pathways that lead to activation of apoptosis. However, these pathways converge upstream of a signaling event controlled by bcl-2 before initiation of the cell death program.

TABLE I
G␣13QLand G␣qRC-induced apoptosis is inhibited by treatment with z-VAD.fmk COS-7 cells were transfected with LacZ and G␣13QL or G␣qRC with and without the addition of 50 M of z-VAD.fmk. z-VAD.fmk was added four times over the course of the incubation period to maximize the effectiveness of the protease inhibitor. DNA fragmentation was assessed using the TUNEL assay to identify cells undergoing apoptosis. The percentage of TUNEL positive cells was calculated from the total number of cells transfected as determined by antibody staining of ␤-galactosidase. Results represent duplicate samples for each condition.