The Lysophosphatidic Acid 2 Receptor Mediates Down-regulation of Siva-1 to Promote Cell Survival*

Lysophosphatidic acid (LPA) promotes cell survival through the activation of G protein-coupled LPA receptors. However, whether different LPA receptors activate distinct anti-apoptotic signaling pathways is not yet clear. Here we report a novel mechanism by which the LPA2 receptor targets the proapoptotic Siva-1 protein for LPA-dependent degradation, thereby attenuating Siva-1 function in DNA damage response. The carboxyl-terminal tail of the LPA2 receptor, but not LPA1 or LPA3 receptor, specifically associates with the carboxyl cysteine-rich domain of Siva-1. Prolonged LPA stimulation promotes the association of Siva-1 with the LPA2 receptor and targets both proteins for ubiquitination and degradation. As a result, adriamycin-induced Siva-1 protein stabilization is attenuated by LPA in an LPA2-dependent manner, and the function of Siva-1 in promoting DNA damage-induced apoptosis is inhibited by LPA pretreatment. Consistent with this result, inhibition of the LPA2 receptor expression increases Siva-1 protein levels and augments adriamycin-induced caspase-3 cleavage and apoptosis. Together, these findings reveal a critical and specific role for the LPA2 receptor through which LPA directly inactivates a critical component of the death machinery to promote cell survival.

Substantial evidence has shown that LPA can protect cells from serum deprivation-induced apoptosis, prevent chemotherapeutic agents-induced apoptosis, and block death receptor-mediated apoptosis in a variety of cells, suggesting that LPA is a survival factor (7)(8)(9). Several LPA-induced signaling pathways have been defined for these protective effects (8). For example, the activation of AKT/PKB by LPA stimulation leads to the phosphorylation and inactivation of BAD and procaspase-9, thereby suppressing the apoptotic signaling pathways (10,11). In addition, LPA-induced activation of ERKs has been reported to protect fibroblasts from serum deprivation-induced apoptosis (7). LPA also induces gene expression of NF-B, which activates the transcription of several anti-apoptotic proteins to promote cell survival (12). As the levels of LPA are increased in ovarian cancer cells, LPA may contribute to the resistance to apoptosis and tumor progression (13). Thus far, the specificity and mechanisms by which different LPA receptors regulate these signaling events are not yet fully understood.
The carboxyl-terminal tails of the LPA 1 , LPA 2 , and LPA 3 receptors are structurally distinct form each other, suggesting that this region may specifically regulate the unique proteinprotein interactions and functions of each receptor. Therefore, we used the carboxyl-terminal tail of the LPA 2 receptor as bait in a yeast two-hybrid screen to identify molecules specifically involved in the function of the LPA 2 receptor. In addition to TRIP6, a focal adhesion molecule that regulates LPA-induced cell migration (14), the proapoptotic Siva-1 protein was found to interact with the LPA 2 receptor with a high affinity.
Siva-1 is a proapoptotic protein originally identified as a CD27-binding protein (15). The structure of Siva-1 protein contains a death domain homology region in the internal sequences and two zinc finger-like cysteine-rich domains in the carboxyl terminus; however, they lack histidine residues (see Fig. 1D) (15). Several lines of evidence suggest that Siva-1 is a proapoptotic protein. For example, Siva-1 gene expression is activated during DNA damage response and stroke injury (16 -19). Overexpression of a metastasis suppressor, TIP30, activates Siva-1 transcription in small cell lung cancers (19), and the transcription factors, including p53 and E2F1, directly bind to the Siva-1 promoter and transcriptionally activate Siva-1 expression during DNA damage response (17). Siva-1 is essential for p53-dependent neuronal cell death (20). In addition to inducing apoptosis of T lymphocytes through a caspasedependent mitochondrial pathway (21), Siva-1 may negatively regulate NF-B activity in T cell receptor-mediated activationinduced cell death (22). Siva-1 also binds to Bcl-xL through its death domain homology region and inhibits Bcl-xL-mediated protection against UV irradiation-induced apoptosis (23). Previously it has been reported that Siva-2, a minor form of alternative splice variant that lacks exon 2-coding sequences and most of the death domain homology region, is less apoptotic (24). However, a recent report shows that overexpression of Siva-2 similarly induces apoptosis in T lymphocytes (21). Although it has been shown that phosphorylation of Siva-1 by ARG (c-Abl-related gene) kinase at Tyr-34 is required for ARGmediated apoptosis during oxidative stress (25), the mechanisms that regulate Siva-1 function in apoptosis are still very elusive.
In this report, we characterize the functional significance of the interaction between Siva-1 and the LPA 2 receptor. Our results show that this association promotes LPA-dependent ubiquitination and degradation of Siva-1 protein, thereby down-regulating the proapoptotic activity of Siva-1 during DNA damage response. Thus, in addition to activating AKT/ PKB and ERK pathways to promote cell survival, LPA directly inhibits the proapoptotic function of Siva-1 through specific binding of the LPA 2 receptor to Siva-1.
Cellular Co-immunoprecipitation and Immunoblotting-Cellular co-immunoprecipitation of the LPA receptors with the interacting proteins was performed as described before (14).
To detect cellular interaction of Siva-1 and the LPA 2 receptor at physiological levels, mouse embryonic fibroblasts (MEFs) deficient in LPA 1 and LPA 2 were isolated from lpa 1 Ϫ/Ϫ double knock-out mice (27) and immortalized with SV 40 large T antigen. These cells were stably transfected with a puromycin resistance gene expression vector and either pCMV-FLAG empty vector (Stratagene) (designated DKOmock) or pCMV-FLAG-LPA 2 (designated DKO-LPA 2 ). Semiquantitative RT-PCR analysis was performed to select cell lines expressing FLAG-LPA 2 mRNA at the levels comparable with the endogenous LPA 2 in wild-type MEFs. Five mg of total lysates were isolated from DKO-mock and DKO-LPA 2 MEFs treated with 20 M MG-132 or not for 2 h, and the LPA 2 receptor was immunoprecipitated with anti-FLAG M2 monoclonal antibody-conjugated agarose beads (Sigma). The co-immunoprecipitated endogenous Siva-1 was detected with an affinitypurified anti-Siva-1 polyclonal antibody raised against human Siva-1 peptide (aa 30 -44). The blot was then reprobed with an anti-human LPA 2 receptor polyclonal antibody (28).
To detect ubiquitinylated Siva-1 and the LPA 2 receptor, HAubiquitin was expressed with FLAG-Siva-1 or FLAG-LPA 2 receptor in HEK 293T cells. The cells were starved overnight and then pretreated with 20 M MG-132 (Calbiochem) or vehicle (Me 2 SO) for 1 h. Subsequently, the cells were incubated with LPA for 60 min and harvested in 1ϫ radioimmune precipitation assay buffer. FLAG-tagged Siva-1 or LPA 2 receptor was immunoprecipitated with anti-FLAG M2 monoclonal antibody-conjugated agarose beads (Sigma), and the immunoblot was probed with an anti-HA polyclonal antibody (Santa Cruz Biotechnology) to detect ubiquitinylated proteins.
To determine whether Siva-1 is expressed in the nucleus, NIH 3T3 cells transiently expressing Siav-1 without or with FLAG-LPA 2 were starved in 0.1% fatty acid BSA-containing medium overnight followed by the incubation with 2 M LPA for 1 h and then harvested for subcellular fractionation in hypotonic solution (20 mM Tris, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 , 0.5 mM dithiothreitol, and 0.1% Nonidet P-40). The enriched nuclei were dissolved in 1% SDS lysis buffer. 1% SDS was also added to the supernatant, which includes cytosol, subcellular organelles, and plasma membrane. After sonication, the total cell lysates, nuclear extract, and supernatant were subjected to immunoblotting using an anti-Siva-1 antibody. The blot was then probed with an anti-vinculin antibody (BD Biosciences) and an anti-histone antibody (Roche Applied Science) as a cytosolic and nuclear marker, respectively.
To study the effect of LPA on the turnover of Siva-1 and the LPA 2 receptor, NIH 3T3 cells were transfected with vectors expressing FLAG-Siva-1 or FLAG-LPA 2 receptor by electroporation (Gene Pulser; Bio-Rad). The cells were starved for 2 h, pretreated with 20 g/ml cycloheximide for 5 min, and then incubated with LPA for various times. FLAG-Siva-1 in the whole cell lysates and the immunoprecipitated FLAG-LPA 2 receptor were detected by immunoblotting using an anti-FLAG polyclonal antibody (Sigma). The immunoblots were probed with either an anti-GAPDH antibody (Alexis) or an anti-␤-actin antibody (Sigma) as a loading control. Similar procedure was performed to determine the turnover rate of FLAG-Siva-1 in NIH 3T3 cells after treatment with 10 M LPA and 10 M adriamycin for 20 h.
The RH7777 rat hepatoma cell line was purchased from the American Type Culture Collection and stably transfected with pCMV-FLAG empty vector or pCMV-FLAG-LPA 2 . The positive clones were pooled to establish RH7777-mock and RH7777-LPA 2 stable cell lines. The expression of the LPA 2 receptor mRNA was verified by RT-PCR analysis. The effect of LPA on the turnover rate of transfected HA-Siva-1 was determined as described above. To determine the effect of adriamycin on the turnover rate of Siva-1, these cells were treated with adriamycin for 20 h in serum-containing medium, and the turnover rate of Siva-1 was determined.
GST Pull-down Assay-To detect in vitro binding of Siva-1 and the carboxyl-terminal tail of the LPA 2 receptor, GST, GST-LPA 2 -CT, or GST-Siva-1 was expressed in (BL21) (DE3) (LysS) Escherichia coli and purified by immobilizing the proteins on glutathione-Sepharose beads (Amersham Biosciences). GST-Siva-1 was further digested with PreScission Protease (Amersham Biosciences Bioscience) to cleave GST. 0.1 g of Siva-1 was immediately incubated with 1 g of GST or GST-LPA 2 -CT for 2h at 4°C. Siva-1 protein pulled down by GST-LPA 2 -CT was detected with an anti-Siva-1 polyclonal antibody.
Immunocytochemistry-To examine co-localization of Siva-1 with the LPA 2 receptor, GFP-Siva-1 was co-expressed with the FLAG-LPA 2 receptor or FLAG-LPA 2 -⌬C in NIH 3T3 cells. The cells were starved overnight in 0.1% fatty acid-free BSA-containing medium. After fixation, permeabilization, and blocking, the cells were incubated with an anti-FLAG M2 monoclonal antibody (Sigma) followed by the incubation with a Texas Red X-conjugated anti-mouse secondary antibody (Molecular Probes) to detect FLAG-LPA 2 receptor. The nuclei were stained with Hoechst 33258 (Sigma). The images of GFP and GFP-Siva-1 were directly captured by fluorescence microscopy (Axioplan 2, Zeiss). A similar procedure was performed to detect co-localization of FLAG-Siva-1 with HcRed1-LPA 2 or GFP-LPA 2 with HcRed1-Siva-1.
Adriamycin-induced Siva-1 Expression, Caspase-3 Cleavage, and Apoptosis-NIH 3T3 cells transiently expressing HA-Siva-1 or not were incubated in 0.1% fatty acid-free BSA-containing Dulbecco's modified Eagle's medium without or with 5 M LPA for 6 h followed by the addition of 5 M adriamycin for another 24 h. The cells were fixed in 70% ethanol and DNA was stained with propidium iodide (Roche Applied Science). The population of sub-G 1 cells was analyzed by flow cytometry (UAB Flow Cytometry Core Facility). Similar procedures were performed in NIH 3T3 cells overexpressing a scrambled siRNA, an LPA 2 siRNA, or a Siva-1 siRNA except that cells were pretreated with 10 M LPA for 6 h followed by the addition of 10 M adriamycin for 20 h. Immunoblotting was performed to detect endogenous Siva-1, HA-Siva-1, procaspase-3, and ␤-actin using a Siva-1-specific polyclonal antibody, an anti-HA antibody (Santa Cruz Biotechnology), an anticaspase-3 antibody (Santa Cruz Biotechnology), and an anti-␤-actin antibody (Sigma), respectively. Apoptosis was also determined by annexin V-fluorescein isothiocyanate staining (BD Biosciences) followed by fluorescence-activated cell sorter analysis.

RESULTS
The Carboxyl-terminal Tail of the LPA 2 Receptor Interacts with the Carboxyl Cysteine-rich Domain of Siva-1 Protein-In an attempt to identify the molecules specifically involved in the regulation of the LPA 2 receptor, a fusion protein containing the carboxyl-terminal tail (aa 296 -351) of the LPA 2 receptor (designated LPA 2 -CT) and the Gal4 DNA-binding domain was used as bait to screen a HeLa cell cDNA library (14). Among the four million clones screened, one positive clone containing six base pairs 5Ј to the translation start sites, the entire Siva-1 coding sequences, and the 3Ј-untranslated region was identified. The interaction has been verified by selective growth of yeast cells co-expressing full-length Siva-1 with LPA 2 -CT, but not LPA 1 -CT or LPA 3 -CT, on plates lacking tryptophan, leucine, and histidine (Fig. 1A). To further examine a direct interaction between these two molecules, purified full-length Siva-1 was incubated with GST protein or GST fusion protein of LPA 2 -CT in vitro. As shown in Fig. 1B, Siva-1 was pulled down by GST-LPA 2 -CT but not GST, indicating a direct interaction between Siva-1 and the carboxyl-terminal tail of the LPA 2 receptor.
Previously we have shown that the carboxyl-terminal tail of the LPA 2 receptor directly interacts with the LIM domains of TRIP6 (14). In comparison with the two cysteine-rich zinc finger-like motifs (aa 114 -170) of Siva-1, the second zinc finger motif of the TRIP6-LIM1 domain (aa 307-331) shares 32% identity with amino acids 114 -138 of Siva-1, and the second zinc finger motif of the TRIP6-LIM3 domain (aa 432-460) shares 21% identity with amino acids 142-170 of Siva-1 (Fig.  1C). Given the structural similarity between the carboxyl cysteine-rich domains of Siva-1 and TRIP6, we speculated that the zinc finger-like motifs of Siva-1 might be responsible for LPA 2 receptor binding. To test this hypothesis, a number of cDNA constructs encoding GFP fusion proteins of Siva-1, Siva-2, and several Siva-1 deletion mutants (Fig. 1D) were generated and transfected without or with a Myc-LPA 2 receptor expression vector into HEK 293T cells. Because the expression of some Siva-1 mutants was very low, here the GFP fusion proteins of Siva-1 and Siva-1 deletion mutants were employed to ensure comparable expression levels of each protein. Fusion to GFP was able to increase the expression of Siva mutants, but GFP itself did not nonspecifically associate with the Myc-LPA 2 receptor (Fig. 1E). Our result showed that the LPA 2 receptor co-immunoprecipitated with Siva-1, Siva-2, Siva-1-⌬N, Siva-1-CT, and Siva-1-⌬C1. In contrast, the receptor binding with either Siva-1-⌬C2 or Siva-1-⌬C3 was greatly diminished, indicating that the carboxyl cysteine-rich domain (aa 139 -175) of Siva-1 is responsible for this interaction.
Next, we investigated whether Siva-1 interacts with the LPA 2 receptor at physiological levels. We reconstituted a FLAG-tagged human LPA 2 receptor in immortalized MEFs lacking both LPA 1 and LPA 2 receptors (designated DKO-LPA 2 ) and ensured that the reconstituted LPA 2 receptor was comparably expressed as the endogenous LPA 2 receptor in wild-type MEFs by semi-quantitative RT-PCR analysis ( Fig. 2A, left panel). We found that endogenous Siva-1 specifically associated with the LPA 2 receptor in DKO-LPA 2 MEFs ( Fig. 2A, middle panel); however, we could barely detect endogenous Siva-1 expression in the total lysates of DKO-LPA 2 cells ( Fig. 2A, middle panel). Because Siva-1 mRNA was comparably expressed in DKOmock and DKO-LPA 2 MEFs ( Fig. 2A, left panel), we speculated that the stability of Siva-1 protein might be different in these two cell lines. Indeed, when both cell lines were incubated with MG-132 for 2 h to inhibit proteasomal degradation, they expressed similar levels of Siva-1 protein ( Fig. 2A, right panel), and under this condition, endogenous Siva-1 was found to coimmunoprecipitate with the LPA 2 receptor in DKO-LPA 2 MEFs ( Fig. 2A, right panel).
Previously we have shown that LPA stimulation rapidly recruits TRIP6 to the activated LPA 2 receptor but not other LPA receptors (14). To investigate whether Siva-1 also binds to the LPA 1 and LPA 3 receptors, co-immunoprecipitation of Siva-1 with different LPA receptors was performed in HEK 293T cells without or with LPA stimulation for 5 min. As shown   in Fig. 2B, Siva-1 preferentially bound to the LPA 2 receptor. In contrast to TRIP6, Siva-1 associated with the LPA 2 receptor in the absence of serum, and this interaction was not affected by LPA stimulation for 5 min. Siva-1 also bound to the LPA 1 receptor with a much lower affinity but barely interacted with the LPA 3 receptor. Because the carboxyl-terminal tail of the LPA 1 receptor does not interact with Siva-1 (Fig. 1A), this result suggests that the LPA 1 receptor may bind to Siva-1 indirectly or through another weak binding site.

Siva-1-CT
Although a brief stimulation with LPA has no effect on the association of Siva-1 with the LPA 2 receptor, prolonged LPA stimulation affects the expression and interaction of these two Ϫ/Ϫ double knock-out fibroblasts stably expressing a human FLAG-LPA 2 . pCMV-FLAG-LPA 2 or pCMV-FLAG was stably transfected into immortalized MEFs deficient in LPA 1 and LPA 2 (designated DKO-LPA 2 and DKO-mock, respectively). The left panel is a semi-quantitative RT-PCR analysis showing the mRNA expression of LPA 2 , Siva-1 and GAPDH. In the middle panel, the LPA 2 receptor was immunoprecipitated (IP) with anti-FLAG M2 monoclonal antibody-conjugated agarose beads from 5-mg lysates of DKO-LPA 2 or DKO-mock cells. The immunoblot (IB) was probed with an anti-Siva-1 polyclonal antibody to detect co-immunoprecipitated endogenous Siva-1. The blot was then reprobed with an anti-LPA 2 polyclonal antibody. The result shown is the ϳ37-kDa LPA 2 monomer and the modified forms of LPA 2 (ϳ75-200 kDa). In the right panel, DKO-mock and DKO-LPA 2 cells were treated with MG-132 for 2 h. Co-immunoprecipitation of endogenous Siva-1 and the reconstituted FLAG-LPA 2 receptor was performed as described above. B, Siva-1 predominantly associated with the LPA 2 receptor in HEK 293T cells. GFP-Siva-1 was co-expressed with one of the FLAG-LPA receptors in HEK 293T cells as indicated. The cells were starved overnight followed by the incubation with LPA for 5 min. The LPA receptors were immunoprecipitated with anti-FLAG M2 monoclonal antibody-conjugated agarose beads, and the immunoblot was probed with an anti-GFP polyclonal antibody to detect co-immunoprecipitated GFP-Siva-1. The blot was stripped and reprobed with an anti-FLAG polyclonal antibody to detect the LPA receptors. The result shown is the LPA receptor monomer (ϳ37-40 kDa). The bottom panel shows the expression of GFP-Siva-1 in the whole cell lysates. C, prolonged LPA stimulation regulates the expression and association of the LPA 2 receptor and Siva-1. NIH 3T3 cells expressing GFP-Siva-1 alone or GFP-Siva-1 with the FLAG-LPA 2 receptor were split evenly into different plates. The cells were either cultured in serumcontaining medium or starved overnight followed by the incubation with LPA for 2 h. Co-immunoprecipitation of the FLAG-LPA 2 receptor with GFP-Siva-1 was performed as described above. The expression of GFP-Siva-1 and TRIP6 in the total lysates was detected with an anti-GFP polyclonal antibody and an anti-TRIP6 monoclonal antibody, respectively. The levels of GFP-Siva-1 and FLAG-LPA 2 were quantified by NIH IMAGE J software and normalized by the intensity of TRIP6. D, the levels of Siva-1 were reduced by prolonged LPA stimulation only when it was co-expressed with the LPA 2 receptor, but not LPA 1 or LPA 3 receptor. HEK 293T cells transiently expressing Myc-Siva-1 with either FLAG-tagged LPA 1 , LPA 2 , or LPA 3 were starved overnight followed by the incubation with LPA for 1 h. The results shown are the Myc-Siva-1 and ␤-actin in the total lysates and the FLAG-LPA receptor monomer immunoprecipitated from the same amounts of total lysates. The intensity of Myc-Siva-1 and FLAG-LPA receptors was quantified and normalized by the intensity of ␤-actin and was compared without or with LPA treatment in each set. The result shown in each figure is a representative from three independent experiments. WT, wild type.
proteins. We found that in NIH 3T3 cells co-expressing GFP-Siva-1 and the FLAG-LPA 2 receptor, both Siva-1 and the LPA 2 receptor were expressed at much higher levels by serum starvation overnight (Fig. 2C, lane 3 compared with lane 1 and lane  4 compared with lane 2). However, the expression of both proteins was reduced by further treatment with LPA for 2 h. Under this condition, similar amounts of Siva-1 were co-immunoprecipitated with the LPA 2 receptor compared with that in the absence of serum (Fig. 2C, lane 5 compared with lane 4). This result may suggest that the association of Siva-1 and LPA 2 is increased by prolonged LPA treatment. Our result also showed that in serum-containing medium, Siva-1 was expressed at lower levels by co-expressed with the LPA 2 receptor (Fig. 2C, lane 2 compared with lane 1). Because serum contains micromolar concentrations of LPA (30), it is likely that the transfected LPA 2 receptor promotes LPA-induced reduction of Siva-1 protein in the presence of serum. Although TRIP6 also interacts with the LPA 2 receptor, the levels of TRIP6 were not affected by overexpression of the LPA 2 receptor.
We also found that LPA stimulation for 1 h attenuated the expression of LPA 1 , LPA 2 , and LPA 3 in HEK 293T cells (Fig.  2D). Although the LPA 1 receptor was able to bind to Siva-1 weakly (Fig. 2B), LPA reduced the expression of Siva-1 only when Siva-1 was co-expressed with the LPA 2 receptor. Thus, these results suggest that LPA 2 , but not LPA 1 or LPA 3 , specifically regulates the expression of Siva-1 protein in an LPA-dependent manner.
The association of Siva-1 with the LPA 2 receptor was further confirmed by co-localization of GFP-Siva-1 with FLAG-LPA 2 (Fig. 3A) or HcRed1-Siva-1 with GFP-LPA 2 (Fig. 3B) in serumfree conditions and by co-localization of FLAG-Siva-1 with HcRed1-LPA 2 after LPA stimulation for 2 h (supplemental Fig.  S1C). The result showed that Siva-1 was predominantly expressed in the nucleus, although it could also be found in the cytosol and plasma membrane ( Fig. 3 and supplemental Fig.  S1A). In contrast to Siva-1, the LPA 2 receptor was only present in the cytosol or on the plasma membrane ( Fig. 3 and supplemental Fig. S1B). Strikingly, the cytosolic and membrane localization of Siva-1 was greatly increased in cells overexpressing the LPA 2 receptor, where Siva-1 appeared in clusters that coincided precisely with the distribution of the LPA 2 receptor (Fig.  3 and supplemental Fig. S1C). In contrast to the LPA 2 receptor, the LPA 2 -⌬C mutant, which lacks the carboxyl-terminal tail to bind to Siva-1, failed to capture Siva-1 in the cytosol (Fig. 3A). This result suggests that the LPA 2 receptor binds to Siva-1 and prevents cytosolic Siva-1 from translocation into the nucleus.
To verify whether Siva-1 is indeed present in the nucleus, NIH 3T3 cells expressing Siva-1 without or with the LPA 2 receptor were treated with LPA for 1 h, and differential centrifugation was performed to separate nuclei and the supernatant, which contains subcellular organelles, cytosol and the plasma membrane. Consistently, the result showed that Siva-1 was predominantly present in the nuclear extract (supplemental Fig.  S1D). When co-expressed with the LPA 2 receptor, the levels of Siva-1 were significantly lower and were further reduced by LPA.

The LPA 2 Receptor Promotes LPA-dependent Degradation of
Siva-1 Protein-Many G protein-coupled receptors are endocytosed after ligand-induced activation (31). The internalized receptors are either recycled back to the plasma membrane in a process known as resensitization or undergo proteolytic degradation following prolonged ligand stimulation. Because the expression of LPA 2 receptor and Siva-1 was both reduced by prolonged LPA stimulation (Fig. 2, C and D), we speculated that both proteins might undergo LPA-dependent degradation concomitantly. Therefore, we set out to examine LPA-induced turnover of the LPA 2 receptor and Siva-1 in NIH 3T3 cells by pretreating cells with cycloheximide to inhibit protein synthesis. Our result showed that the turnover rate of the FLAG-LPA 2 receptor was enhanced by LPA stimulation in NIH 3T3 cells (Fig. 4A). Likewise, Siva-1 was more stable in serum-free conditions, whereas LPA induced a rapid turnover of endogenous Siva-1 (Fig. 4B) and transfected FLAG-Siva-1 (Fig. 4C). To investigate the effect of LPA 2 on Siva-1 turnover, next we examined the turnover of a transfected HA-Siva-1 in RH7777-mock rat hepatoma cells that do not express endogenous LPA 1 , LPA 2 , and LPA 3 (32) and in RH7777-LPA 2 cells stably expressing the FLAG-LPA 2 receptor. In general, the expression of Siva-1 was much lower in RH7777-LPA 2 cells than in RH7777-mock cells when equal amounts of Siva-1 cDNA were transfected (Fig.  4D). Our results showed that LPA treatment did not alter the expression of HA-Siva-1 in RH7777-mock cells but significantly attenuated its expression in RH7777-LPA 2 cells (Fig.  4D), suggesting that the LPA 2 receptor promotes LPA-induced Siva-1 degradation.
LPA Stimulation Increases Ubiquitination of Both Siva-1 Protein and the LPA 2 Receptor-Several G protein-coupled receptors have been shown to undergo ligand-dependent ubiquitination and degradation through the lysosomal or proteasomal pathways (33). We hypothesized that Siva-1 might undergo concomitant ubiquitination and degradation together with the LPA 2 receptor in response to LPA stimulation. To test this possibility, HEK 293T cells expressing HA-ubiquitin with FLAG-LPA 2 or FLAG-Siva-1 were starved overnight and pretreated with MG-132 for 1 h before stimulation with LPA for another hour. Our result showed that ubiquitination of the LPA 2 receptor was found even in the absence of LPA (Fig. 5A). However, the levels of ubiquitinylated LPA 2 were increased by MG-132 pretreatment, which was further enhanced when followed by LPA stimulation (Fig. 5A). Similarly, LPA stimulation promoted Siva-1 ubiquitination, and the levels of ubiquitinylated Siva-1 were further increased by MG-132 (Fig. 5B). Using MG-132 to prevent protein degradation, we found that the ubiquitinylated Siva-1 and LPA 2 receptor were present in the same complex, and this complex formation was increased by LPA treatment for 60 min (Fig. 5C). Together, these results suggest that prolonged LPA stimulation promotes the association of Siva-1 with the LPA 2 receptor and targets both proteins for proteasomal ubiquitination and degradation.
DNA Damage-induced Siva-1 Stabilization Is Attenuated by LPA in an LPA 2 Receptor-dependent Manner-It has been shown that transcription of Siva-1 mRNA can be induced by treatment with the DNA damaging agents camptothecin and cisplatin (17,18). Further evidence reveals that Siva-1 is a direct transcriptional target for p53 and E2F1 (17). Consistently, our result showed that adriamycin (Doxorubicin) transcriptionally activated Siva-1 mRNA in NIH 3T3 cells; however, this induction was not affected by LPA (Fig. 6A). Adriamycin not only induced endogenous Siva-1 protein expression but also stabilized the transfected Siva-1, particularly in serum-free conditions; however, this effect was attenuated by LPA pretreatment (Fig. 6B). The counteracting role of LPA in regulating adriamycin-mediated Siva-1 stability was further demonstrated by assessing the turnover rate of transfected FLAG-Siva-1 in NIH 3T3 cells treated with adriamycin and LPA (Fig. 6C). Because cycloheximide can facilitate apoptosis induction in some cases (34), here we also found that the addition of cycloheximide to inhibit protein synthesis further enhanced adriamycin-mediated stabilization of Siva-1 protein. Nonetheless, this effect was attenuated by LPA.
Next, we assessed the effect of LPA 2 receptor on the turnover of Siva-1 during DNA damage response. As shown in Fig. 6D, treatment with adriamycin and cycloheximide induced a robust activation and stabilization of Siva-1 in RH7777-mock cells but not RH7777-LPA 2 cells, suggesting a role for LPA 2 in mediating the down-regulation of Siva-1 during DNA damage response.
If the down-regulation of Siva-1 protein expression is mediated by LPA 2 binding, overexpression of the LPA 2 receptor should not alter the expression of Siva-1-⌬C3 mutant that lacks the LPA 2 receptor-interacting domain (Fig. 1E). To test this hypothesis, FLAG-Siva-1 or FLAG-Siva-1-⌬C3 was expressed without or with the Myc-LPA 2 receptor in NIH 3T3 cells, and then cells were treated with adriamycin for 20 h in serum-containing medium. Although the exogenous Siva-1 mRNA and Siva-1-⌬C3 mRNA were constitutively expressed, Siva-1 and Siva-1-⌬C3 proteins were significantly induced only when treated with adriamycin (Fig. 6E). This result suggests that adriamycin must have stabilized these proteins through the regulation of the amino-terminal region of Siva-1, which is independent on LPA 2 binding. However, only Siva-1, but not Siva-1-⌬C3, was attenuated by LPA 2 receptor overexpression, suggesting that the carboxyl cysteine-rich motif of Siva-1 mediates LPA 2 receptor-dependent Siva-1 degradation (Fig. 6E). LPA Attenuates Siva-1 Function in Adriamycin-induced Apoptosis-It has been shown that overexpression of Siva-1 in itself is sufficient to activate caspase-3 activity and induce apoptosis in several cell lines (15,21,(23)(24)(25). In contrast, LPA protects NIH 3T3 fibroblasts from serum starvation-induced apoptosis (7). Because LPA targets Siva-1 for degradation, it might block the proapoptotic function of Siva-1 during DNA damage response. Indeed, our results showed that inhibition of the LPA 2 receptor expression in NIH 3T3 cells increased Siva-1 levels and attenuated the protective effect of LPA against adriamycin-induced caspase-3 cleavage (Fig. 7A). Likewise, overexpression of Siva-1 signif-icantly enhanced adriamycin-induced caspase-3 cleavage; however, this effect was reduced by LPA pretreatment (Fig. 7B). Consistent with these observations, overexpression of an LPA 2 siRNA or Siva-1 protein promotes adriamycin-induced apoptosis ( Fig. 7C and supplemental Fig. S2). In contrast, LPA protected cells from apoptosis in all of the transfected cells except for the LPA 2 siRNA-expressing cells, in which LPA showed a much less protective effect (Fig. 7C). Perhaps because adriamycin only induced a mild apoptosis in NIH 3T3 cells, knocking down the expression of Siva-1 either did not protect cells from adriamycin-induced caspase-3 cleavage (Fig. 7B) and apoptosis (annexin V staining; Fig. 7C) or only mildly reduced apoptosis (propidium iodide staining; supplemental Fig.  S2B). Taken together, these results demonstrate that LPA not only serves as a survival factor but also plays an anti-apoptotic role by directly downregulating the expression and proapoptotic activity of Siva-1 through a specific interaction of the LPA 2 receptor with Siva-1.

DISCUSSION
LPA is a growth factor-like phospholipid that has been shown to mediate cell survival through the activation of phosphatidylinositol 3-kinase-AKT and ERK signaling pathways and the induction of NF-B gene expression (7-9, 12, 35). Our study further reveals a novel mechanism by which prolonged stimulation with LPA not only promotes ubiquitination and degradation of the LPA 2 receptor but also concomitantly down-regulates LPA 2 -interacting Siva-1 protein. Consequently, the proapoptotic function of Siva-1 in DNA damage response is attenuated by LPA. Fig. 7D provides a model by which the interaction of Siva-1 with the LPA 2 receptor promotes LPA-dependent down-regulation of Siva-1 function during DNA damage response. Because the function of Siva-1 protein is important for inducing caspase-3 activity and apoptosis, an appropriate suppression of Siva-1 activity through LPA-induced Siva-1 protein degradation may serve as a mechanism for normal cell growth.
Previously it has been demonstrated that in primary chronic lymphocytic leukemia cells in which the LPA 1 receptor is upregulated, LPA protects cells from spontaneous apoptosis through an LPA 1 -dependent activation of AKT/PKB pathways (36). In contrast, in IEC-6 intestinal epithelial cells that express high levels of LPA 2 and lesser amounts of LPA 1 , an LPA 2 -selective agonist, FAP-12, protects cells from camptothecin-induced apoptosis (37). Moreover, another LPA 2 receptor-selective ago- nist, octadecenyl thiophosphate, specifically protects LPA 1null mice, but not LPA 2 -null mice, from radiation-induced apoptosis of intestinal epithelium (38). It remains to be tested whether octadecenyl thiophosphate elicits the anti-apoptotic effect through an LPA 2 -dependent down-regulation of Siva-1 expression. Together, these findings suggest that LPA 1 and LPA 2 may utilize distinct anti-apoptotic signaling mechanisms to promote cell survival. Thus far, a number of G protein-coupled receptors such as the ␤ 2 -adrenergic receptor, the CXCR4 chemokine receptor, the protease-activated receptor 2 and the V2 vasopressin receptor have been shown to undergo ubiquitination and degradation following prolonged agonist stimulation (39 -42). In contrast, several other receptors such as the plateletactivating factor receptor, the opioid receptor, and the thyrotropin-releasing hormone receptor are ubiquitinated in an agonist-independent manner because of misfolding or incomplete folding of the receptor during synthesis (42)(43)(44). Our results show that the LPA 2 receptor is ubiquitinated in the absence of ligand; however, in the presence of MG-132, LPA stimulation further promotes ubiquitination of the LPA 2 receptor. Thus far, only a very limited number of proteins that directly associate with the G protein-coupled receptors, such as ␤-Arrestin1, ␤-Arrestin2, and G protein-coupled receptor kinase 2, have been reported to undergo ligand-dependent ubiquitination and degradation (39,(45)(46)(47). Therefore, our results would provide another example of G protein-coupled receptor-mediated regulation of associated proteins through ligand-dependent ubiqutination and degradation. Whether LPA targets the LPA 2 receptor and Siva-1 to the same E3 ligase for ubiquitination remains to be determined.
Through the yeast two-hybrid screen, we have now identified Siva-1 and TRIP6 as the LPA 2 receptor-interacting proteins. The carboxyl-terminal cysteine-rich motif of Siva-1 and the second zinc finger motif of TRIP6-LIM3 domain share some structural similarity and are both important for the interaction B, adriamycin-induced Siva-1 protein stabilization is attenuated by LPA pretreatment. NIH 3T3 cells transfected with Siva-1 or not were pretreated with 10 M LPA in 0.1% fatty acid-free BSA-containing medium for 6 h followed by the addition of 10 M adriamycin for another 20 h. Immunoblotting (IB) was performed to determine the protein levels of Siva-1 and GAPDH in the whole cell lysates. The relative fold difference of Siva-1 expression was quantified as described above. C, LPA plays a counteracting role in regulating the turnover rate of Siva-1 during DNA damage response. NIH 3T3 cells transiently expressing FLAG-Siva-1 were pretreated with LPA or not for 6 h followed by the addition of adriamycin for another 20 h. The turnover rate of FLAG-Siva-1 was determined as described above. D, adriamycin induces and stabilizes Siva-1 in RH7777 cells, whereas this effect is greatly reduced in RH7777 cells stably expressing the LPA 2 receptor. RH7777-mock and RH7777-LPA 2 cells were treated with 10 M adriamycin for 20 h in serum-containing medium. After cycloheximide pretreatment for 5 min, the turnover rate of Siva-1 was determined as described above. E, adriamycin-mediated stabilization of Siva-1, but not Siva-1-⌬C3, is attenuated by overexpression of the LPA 2 receptor. FLAG-Siva-1 or FLAG-Siva-1-⌬C3 was expressed alone or co-expressed with the Myc-LPA 2 receptor in NIH 3T3 cells. The cells were treated with 5 M adriamycin for 20 h in serum-containing medium. The expression of FLAG-Siva-1 in 30 g of cell lysates or FLAG-Siva-1-⌬C3 in 150 g of cell lysates was detected by immunoblotting with an anti-FLAG antibody, and the blot was reprobed with an anti-␤-actin antibody. The Myc-LPA 2 receptor monomer was detected as described in Fig. 1E. The bottom three panels show the mRNA expression of FLAG-Siva-1, FLAG-Siva-1-⌬C3, total LPA 2 receptors, and GAPDH. The data shown in each figure are representative results from three independent experiments.
with the LPA 2 receptor but not other LPA receptors. However, there are some functional discrepancies between these two molecules in LPA signaling. TRIP6 specifically interacts with the LPA 2 receptor upon LPA stimulation for 5-10 min. This association is probably transient because LPA treatment for 15-20 min induces the translocation of TRIP6 to focal adhesion plaques where the LPA 2 receptor is not present (14). In contrast, Siva-1 binds to the LPA 2 receptor in the absence of LPA (Fig. 2, B and C). However, using MG-132 to inhibit proteasomal degradation, we have demonstrated that LPA stimulation for 60 min promotes the complex formation of ubiquitinylated LPA 2 receptor and Siva-1 (Fig. 5C). This result suggests that Siva-1 may preferentially bind to the internalized LPA 2 receptor. The interaction of TRIP6 with the LPA 2 receptor promotes LPA-dependent association of TRIP6 with several focal adhesion molecules, thereby enhancing LPA-induced cell migration (14), whereas the interaction of Siva-1 with the LPA 2 receptor captures Siva-1 in the perinuclear region and promotes LPAdependent degradation of Siva-1. In contrast to Siva-1, LPA stimulation for 60 min does not promote ubiquitination and degradation of TRIP6 (data not shown). This is possibly due to a transient interaction of TRIP6 with the LPA 2 receptor. It should be noted that our results do not exclude the possibility that LPA may mediate Siva-1 degradation through other transcriptional and post-translational mechanisms involved in cell survival and anti-apoptosis, and Siva-1 may be down-regulated by other serum factors in addition to LPA (Fig. 6E). Nonetheless, the interaction of Siva-1 with the LPA 2 receptor would provide a direct mechanism to promote LPA-induced Siva-1 degradation. DNA damage response such as adriamycin treatment transcriptionally regulates Siva-1 gene expression through the activation of p53 and E2F1 (17) and also post-translationally stabilizes Siva-1 protein (Fig. 6). As a result, Siva-1 enhances DNA damage-induced apoptosis (Fig. 7, B and C). In addition to the LPA 2 receptor, Siva-1 has been shown to interact with two other cell surface receptors, including CD27 and GITR (the glucocorticoid-induced tumor necrosis factor receptor familyrelated gene) (15,48). In contrast to the LPA 2 receptor that mediates LPA-induced cell survival and cell proliferation, CD27 and GITR are involved in T lymphocyte apoptosis (15,49). Thus, these receptors may compete for Siva-1 binding in different cellular conditions. Intriguingly, Siva-1 shuttles between nucleus and cytosol and is present predominantly in FIGURE 7. LPA attenuates Siva-1 function in adriamycin-induced caspase-3 cleavage and apoptosis. A, the effects of LPA2 siRNA and Siva-1 siRNA on the expression of Siva-1 and caspase-3 cleavage in cells treated with adriamycin and LPA. NIH 3T3 cells transiently expressing a scrambled siRNA (siScramble), LPA 2 siRNA (siLPA 2 ), or Siva-1 siRNA (siSiva-1) were pretreated with LPA for 6 h followed by the incubation with adriamycin for 20 h. The cell lysates were subjected to immunoblotting (IB) using an antibody specific to Siva-1, caspase-3, or GAPDH, respectively. B, Siva-1 augments adriamycin-induced caspase-3 cleavage, whereas this function is inhibited by LPA. NIH 3T3 cells expressing HA-Siva-1 or not were pretreated with LPA for 6 h followed by the incubation with adriamycin for another 20 h. Immunoblotting was performed to determine the levels of HA-Siva-1, procaspase-3, and GAPDH in the total lysates. C, inhibition of the LPA 2 receptor expression or overexpression of Siva-1 promotes adriamycin-induced apoptosis. NIH 3T3 cells expressing a control GFP siRNA (siGFP), LPA 2 siRNA, Siva-1 siRNA, or Siva-1 protein were pretreated with LPA for 6 h followed by the incubation with adriamycin for 20 h. Apoptosis was determined by annexin V-fluorescein isothiocyanate staining followed by fluorescence-activated cell sorter analysis. The data shown are the means Ϯ S.E. of three to four independent experiments. *, p Ͻ 0.05 compared with the siGFP control with adriamycin treatment. **, p Ͻ 0.05 compared with that without LPA pretreatment in each set of samples, Student's t test. D, a model for the LPA 2 receptor-dependent down-regulation of Siva-1 to promote cell survival. the nucleus; however, all three Siva-1-interacting receptors are cell surface receptors. Moreover, Siva-1 induces apoptosis via a caspase-3-dependent mitochondrial pathway, although the detailed mechanism has yet to be elucidated. Whether nuclear Siva-1 plays an intrinsic role in apoptosis or has a completely different function remains to be explored.