The Constitutively Active Orphan G-protein-coupled Receptor GPR39 Protects from Cell Death by Increasing Secretion of Pigment Epithelium-derived Growth Factor*

GPR39 is a constitutively active orphan G-protein-coupled receptor capable of increasing serum response element-mediated transcription. We found GPR39 to be up-regulated in a hippocampal cell line resistant against diverse stimulators of cell death and show that its overexpression protects against oxidative and endoplasmic reticulum stress, as well as against direct activation of the caspase cascade by Bax overexpression. In contrast, silencing GPR39 rendered cells more susceptible to cell death. An array analysis of transcripts induced by GPR39 revealed up-regulation of RGS16 (inhibitor of G-protein signaling 16), which suggested coupling to Gα13 and induction of serum response element-mediated transcription by the small GTPase RhoA. In line with this, co-expression of GPR39 with RGS16, dominant-negative RhoA, or serum response factor abolished cell protection, whereas overexpression of the serum response factor protected from cell death. Further downstream the signaling cascade, GPR39 overexpression leads to increased secretion of the cytoprotective pigment epithelium-derived growth factor (PEDF). Medium conditioned by cells overexpressing GPR39 contained 4-fold more PEDF, and when stripped off it lost most but not all of its protective properties. We conclude that GPR39 is a novel inhibitor of cell death, which might represent a therapeutic target with implications for processes involving apoptosis and endoplasmic reticulum stress like cancer, ischemia/reperfusion injury, and neurodegenerative disease.

vated by a wide variety of natural ligands, and pharmacological alteration of their signaling constitutes one of the most successful approaches to the treatment of human disease, which makes GPCRs the most targeted protein superfamily in pharmaceutical research (2). We recently presented a screening system able to discriminate protective and detrimental receptors involved in oxidative stress (3) to identify targets for the multitude of human diseases caused or aggravated by oxidative stress. In this proof-of-concept study, we noticed that the relative mRNA content of cytoprotective GPCRs (the most prominent being VPAC 2 , a receptor for the neuroprotective vasoactive intestinal peptide VIP) was increased in glutamate-resistant (HT22R) cells generated by repeated exposure of the parental cell line HT22 to high concentrations of glutamate and further propagation of the few surviving cells. HT22 cells are derived from embryonal mouse hippocampal cells and are considered to be a model system of cell death by oxidative stress. In this model system, increased extracellular glutamate blocks the gradientdriven glutamate/cysteine antiporter X c Ϫ , depleting the cells of cysteine. Cysteine is required for the synthesis of the important antioxidant glutathione, and the sequence of events after depletion of intracellular glutathione involves the activation of 12-lipoxygenase, the accumulation of intracellular peroxides, activation of a cyclic GMP-dependent calcium channel, and eventually a caspase-independent cell death with features of necrosis and apoptosis (reviewed in Ref. 4). Cell death mediated by the proapoptotic protein Bax, in contrast, leads to cytochrome c release from the mitochondria and formation of a complex between Apaf-1 and procaspase-9, leading to caspase-9 activation. Caspase 9 in turn then activates the executioner caspases 3 and 7, which results in classical apoptosis. A third pathway leading to cell death is death by endoplasmic reticulum (ER) stress, where the accumulation of excess unfolded proteins in the ER activates, although not exclusively, the ER stress-specific caspase 12, which then activates caspase-9 independently of cytochrome c (reviewed in Ref. 5).
In this contribution, we investigated the expression of 82 orphan GPCR in HT22R and wild type cells, which identified GPR39 as a candidate regulator of cell death. GPR39 indeed protected against glutamate toxicity, ER stress, and Bax-mediated cell death by coupling to G␣ 13 leading to activation of the RhoA pathway and secretion of the cytoprotective pigment epithelium-derived growth factor (PEDF) and other factors.

EXPERIMENTAL PROCEDURES
Cells, Proteins, siRNA, and Plasmids-HT22 and HT22R cells were cultured and generated as described (6). Immortalized mouse embryonic fibroblasts (a kind gift from Christoph Borner, University of Freiburg, Freiburg, Germany) and human HEK-293T cells were maintained in Dulbecco's modified Eagle's medium (PAA laboratories) supplemented with 10% fetal calf serum and 100 IU/ml penicillin, 100 g/ml streptomycin. Full-length GPR39 and GPR15 cDNA were cloned by high fidelity PCR from HT22 cDNA into pcDNA5.1 and pQCXIP (Invitrogen). RGS16 and dominant-negative RhoA T19N in pcDNA3.1 were obtained from the Missouri University of Science and Technology cDNA Resource Center. Human SRF fused to VP16, the nonfunctional ⌬M mutant, and the dominant-negative SRF in pCS2ϩ were kind gifts from Ulrike Philippar (Institute for Cell Biology, University of Tübingen, Tübingen, Germany). Bax-EGFP in pcDNA3 was a gift from Pawel Kermer (University of Göttingen, Göttingen, Germany). Pools containing four siRNA duplexes against GPR39 were obtained from Dharmacon (MQ-065633-00-0005). MicroRNA was designed against a specific sequence in the first exon of GPR39 (5Ј-AAACTGCTGATTGGCTTTGTA-3Ј) and cloned into pcDNA TM 6.2-GW/EmGFP-miR (Invitrogen).
GPCR Quantification-cDNA from HT22R and wild type HT22 cells was generated as described (3), and the amount of 82 orphan GPCR was quantitated on a 7500 real time PCR system (Applied Biosystems) using the SYBR green I core kit (Eurogentech). Primer sequences used in the screen can be obtained from the authors. For GPR39 the up-regulation was reproduced using exon spanning primers (upper, 5Ј-AGGACAGACCAT-CATATTCCTGAGACT-3Ј; lower, 5Ј-GATCCGTCGGATC-TGATTGG-3Ј) and a 5Ј-5-carboxyfluorescein-and 3Ј-Tamralabeled probe (5Ј-ATTGTGGTGACGTTGGCCGTGTGTT-3Ј) with a qPCR Mastermix (Roche Applied Science). Membrane-bound GPR39 was quantified by flow cytometry. In brief, 10 4 HT22 and HT22R cells were seeded in six-well plates and grown for 24 h, permeabilized, and incubated with rabbit polyclonal ␣-GPR39 (SP4221P, Acris) and stained with an ALEXA-labeled secondary anti-rabbit antibody (Serotec). HEK293T clones were incubated with fluorescein isothiocyanate-labeled mouse monoclonal antibody HA.11 clone 16B12 (Covance). Subsequently, the cells were washed and resuspended in phosphate-buffered saline ϩ 2% fetal calf serum containing 1 g/ml 7-aminoactinomycin D (7-AAD; Calbiochem-Novabiochem), and mean fluorescence at 488 nm was analyzed using a FACSCalibur flow cytometer and quantified using Cell Star TM software (Becton Dickinson). Dead cells were identified by 7-AAD staining and excluded from the analysis.
Transfections-High purity plasmids were prepared using Nucleobond AX 500 columns (Machery and Nagel). For transient transfections, the cells were grown to 80 -90% confluence and transfected with Lipofectamine 2000 (Invitrogen); HT22 cells were nucleofected using the Amaxa system. For this purpose, 10 6 HT22 cells in 100 l of Nucleofector solution L (Amaxa) were mixed with 2 g of the indicated expression con-struct in a cuvette and nucleofected using program A-020. Following nucleofection, the cells were incubated at a concentration of 10 6 cells/ml in complete Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 1U/ml penicillin, 1 g/ml streptomycin. For siRNA transfections, HT22 cells were grown to 80% confluence in a 24-well plate and transfected with a final concentration of 100 nM siRNA with DharmaFECT 2 (Dharmacon). For stable transfections, pQCXIP-GPR39 was transfected in HEK-293T cells and selected with 2.5 g/ml puromycin for 3 weeks. Single colonies were isolated using cloning cylinders (Sigma) and transferred into 96-well plates, where they were further propagated.
Cell Death and Viability Assays-For toxicity experiments, 5000 HT22 or 8000 HEK-293T cells were seeded in 100 l of medium into 96-well plates in the presence of the indicated amounts of PEDF (BioProducts MD) or ROCK1 inhibitor Y-27632 (Sigma). Transfections were carried out 24 h before seeding. Tunicamycin, glutamate, or vehicle was added 24 h after seeding in 50 l of a 1:1 dilution in complete medium to 50 l of medium. Cell viability was measured 24 h later by the amount of blue formazan produced by viable cells from the tetrazolium salt 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (Sigma), which is proportional to the number of viable cells as previously described (3,6).
For cell death analysis by flow cytometry, the cells were plated in 24-well plates and transfected with 0.4 g of Bax-EGFP and the indicated constructs. 24 h later, the cells were resuspended in 100 l of annexin V binding buffer (BD-Pharmingen) and stained with 5 l of annexin V-PE (BD-Pharmingen) and 5 l of 7-AAD. Single EGFP-positive cells were gated at 488 nm and analyzed for annexin V and 7-AAD staining. The data were acquired on a FACSCalibur flow cytometer and quantified using Cell Star TM software (Becton Dickinson).
To remove PEDF from GPR39-conditioned medium, 3000 wild type HT22 cells/well were seeded into a 96-well plate, and the medium was replaced 24 h later by serum-free medium conditioned by HT22 wt cells or cells overexpressing GPR39 containing 2 g/ml ␣-PEDF (rabbit polyclonal antiserum; Bio-Products MD) or control antibody (rabbit polyclonal antichicken antiserum; Davids Biotechnology). Tunicamycin in the indicated amounts was added 24 h later, and cell death was measured as described above.
Microarray Analysis-Total RNA from HEK-293T cell clones 1 and 17 was quantified by a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies), and quality was assessed by a 2100 Bioanalyzer (Agilent Technologies). Synthesis of double-stranded cDNA and in vitro transcription were carried out according to Affymetrix protocols. In brief, 3 g of total RNA from cultured cells were reverse transcribed with SuperScriptII and T7-oligo(dT) promotor primer for 1 h at 42°C, followed by 2 h of incubation at 16°C for second strand cDNA synthesis. After sample clean-up, the cDNA served as a template for subsequent in vitro transcription carried out in the presence of T7 RNA polymerase and a biotinylated nucleotide analog/ribonucleotide mix for complementary (cRNA) amplification and biotin labeling for 16 h at 37°C. The quality of cRNA was assessed in a 2100 Bioanalyzer. Biotinylated cRNA was cleaned up, and 15 g were fragmented by metal/heat-induced fragmentation. The fragmented cRNA was hybridized against GeneChip Human genome U133 Plus 2.0 in an Affymetrix GeneChip hybridization oven for 16 h at 45°C and 60 rpm. Each microarray was washed and stained with streptavidin-phycoerythrin using Affymetrix GeneChip Fluidics Station 450 and scanned with Affymetrix Scanner 3000. Raw data were analyzed with Affymetrix MAS5 algorithm, and probe sets with a detection p value of p Ͻ 0.01 were used for further analysis.
PEDF Quantification-PEDF mRNA was quantitated in HEK293 cells by real time PCR 48 h after transfection using the SYBRgreen core kit (Eurogentech) and QuantiTect primer assays (QT00100730, Qiagen) according to the manufacturer's protocol using a 7500 real time PCR system (Applied Biosystems). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as endogenous control, and the results are given as cycles over endogenous control. PEDF concentrations were determined by using a sandwich enzyme-linked immunosorbent assay for PEDF (microwells coated with specific polyclonal antibody for full-length recombinant human PEDF antigen and detection with high titer polyclonal anti-PEDF antibody conjugated to biotin (BioProducts MD). Briefly, the assays were performed incubating the samples (100 l each, diluted 1:50) for 1 h at 37°C into duplicate wells. The samples were aspirated, and the wells were washed. Afterward 100 l of reconstituted PEDF detector antibody was added to each well and incubated 1 h at 37°C. After washing five times, 100 l of streptavidin peroxidase working solution was added to each well and incubated for 30 min at 37°C. After another five washing cycles, 100 l of tetramethylbenzidine substrate was added to each well and incubated for 20 min at room temperature. After this time 100 l of stop solution was added, and optical density read at 450 nm. The concentration of PEDF in each specimen was calculated by interpolation from the standard curve and given as ng/ml.
Immunocytochemistry-For immunocytochemistry, HEK 293T cells or mouse embryonic fibroblasts were transfected in 6-well plates as described above and seeded into four-well chamber slides (Labtek) 24 h later. Again 24 h later, the cells were fixed for 20 min in 4% paraformaldehyde, preincubated in phosphate-buffered saline containing 1% bovine serum albumin and 0.05% Tween 20 for 30 min. Rhodamine-phalloidin was incubated for 1 h and 4Ј,6Ј-diamino-2-phenylindole for 15 min at room temperature. The images were captured with an Olympus IX-81 inverse fluorescence microscope and deconvoluted by CellR software (Olympus).
SRE Reporter Assay-HEK 293T cells were transiently transfected in a 48-well plate with the serum response element luciferase reporter plasmid (pSRE-Luc, Stratagene) and the indicated expression constructs. 48 h after transfection cells were washed and lysed in 200 l cell lysis buffer (Promega), lysate was centrifuged at 12,000 ϫ g for 1 min, and 20 l of supernatant were transferred to a white 96-well microtiter plate. 100 l of luciferase assay buffer (Promega) were injected to each well directly before measurement. Luminescence was measured by a Genios Pro microplate reader (Tecan) and integrated for 10,000 ms.
Statistical Analysis-The data were summarized as the means Ϯ S.D., and the statistical significance was assessed using two-tailed t tests or analysis of variance with Tukey's multiple comparison test as indicated.

HT22R Cells Are Resistant against Cell Death Caused by
Diverse Cytotoxic Pathways-HT22R cells have been described before; they are resistant against oxidative stress elicited by glutamate and hydrogen peroxide by up-regulation of the glutamate/cysteine exchanger xCT (6) and the cytoprotective G-protein-coupled receptor VPAC 2 (3). We now extended the description of their protective phenotype to cell death elicited by ER stress and overexpression of Bax. Treating HT22 wild type cells with tunicamycin, a compound that inhibits protein glycosylation in the ER thus generating an unfolded protein response, led to a prominent induction of cell death even more pronounced than the cell death caused by glutamate toxicity. HT22R cells, in contrast, were protected (ϳ18.7% difference at 5 g/ml; Fig. 1A). Transfection of a Bax-EGFP fusion construct and quantification of early and late apoptotic cells by measuring phosphatidylserine translocation with annexin V and 7-AAD staining of dead cells in EGFP-positive transfected cells also demonstrated the protective phenotype of HT22R cells (ϳ15.8% difference after 24 h; Fig. 1B). In summary, we conclude that HT22R cells are not only protected against oxidative glutamate toxicity but also against cell death caused by ER stress and Bax overexpression.
HT22R Cells Differ in the Expression of Orphan GPCRs-To identify novel drug targets implicated in protection against oxidative stress and cell death in general, we investigated the expression of 82 orphan GPCRs in wild type and HT22R cells by quantitative PCR. VPAC 2 served as positive control. Half of all investigated transcripts did not yield reliable amplification curves, probably because of lack of expression. Another 36 transcripts were up-regulated less than 4-fold, and only five, including the positive control, were up-regulated more than 4-fold in HT22R cells (Fig. 1C). The expression of GPR37 and GPR101 differed significantly in the three different mRNA preparations used in the analysis and were not analyzed further. GPR15 and GPR39, however, were reproducibly regulated more then 4-fold, with GPR39 being more abundant. Both receptors were initially identified for their homology to known receptors (7,8) with GPR39 being closely related to the neurotensin receptors and GPR15 to the angiotensin and bradykinin receptors (9). Analysis of membrane localization of GPR39 using flow cytometry and a specific antibody directed against the third extracellular loop revealed an ϳ2-fold increase corroborating the observed regulation at the mRNA level (Fig. 1D). Western blotting resulted in no visible bands, probably because of the still low expression level and the inherent difficulties of detecting GPCRs in Western blots.
GPR39, but Not GPR15, Protects against Diverse Stimuli of Cell Death-Both candidate receptors GPR15 and GPR39 were cloned from HT22R cDNA and overexpressed in wild type HT22 cells. GPR39, but not GPR15, protected against glutamate toxicity (ϳ12% increase in viability at 40 mM glutamate), against direct oxidative stress caused by hydrogen peroxide (ϳ19.5% at 1.25 mM H 2 O 2 ), and tunicamycin (ϳ19.6% at 5 mg/ml) ( Fig. 2A). GPR39 also protected against direct activation of the caspase cascade by overexpression of Bax (ϳ31.4%) speaking in favor of a more general protective mechanism (Fig.  2B). In these experiments, only EGFP-positive transfected cells were counted, whereas in the experiments investigating cell death caused by treatment with glutamate, H 2 O 2 , or tunicamycin viability was quantitated by 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide assays in a mixed population with an app. transfection efficiency of 50%, thus explaining the apparently more prominent protection against Bax-induced cell death. The reverse experiment, transfecting HT22R cells with a pool of four different siRNAs in contrast increased susceptibility to glutamate toxicity (ϳ19.8% decrease in viability at 40 mM glutamate; Fig. 2C). Knockdown efficiency was measured by quantitative PCR and constituted 1.79 ⌬C T values over GAPDH (⌬⌬C T ), corresponding to a ϳ3.5-fold reduction in GPR39 mRNA abundance. To exclude the possibility that the effect of the pooled siRNA corresponded to off target effects, we repeated this experiment with vector-expressed microRNA, which yielded very similar results. Overexpression of GPR39 microRNA driven by RNA polymerase II promoter increased cell death of HT22R cells caused by 40 mM glutamate by ϳ16.5% (Fig. 2C). In this approach, GPR39 mRNA was reduced ϳ3-fold. The differences in glutamate susceptibility between the siRNA and the vector-driven microRNA experiments are most likely caused by the different transfection agents used. These data strongly suggest that GPR39 has a protective phenotype, because overexpression and silencing (using two different approaches) resulted in opposite effects on cell viability. The protective phenotype was not due to mutations in the GPR39 coding region because the cDNA cloned from HT22R cells corresponded to the sequence deposited with NCBI.
GPR39 Overexpression Induces Its Inhibitor RGS16-We next investigated the effect of stable overexpression of GPR39 tagged with an N-terminal hemagglutinin on cell death in human HEK293 cells. We determined the surface expression of 36 monoclonal colonies by flow cytometry using a fluorescein isothiocyanate-labeled monoclonal hemagglutinin antibody and chose six clones exhibiting low, middle, and top surface expression for glutamate toxicity studies, which yielded a pos- itive correlation between GPR39 surface expression and resistance against glutamate toxicity (Pearson r2 ϭ 0.66, p Ͻ 0.05). In subsequent studies using again hydrogen peroxide and tunicamycin, the highest expressing clone 17 always exhibited a more pronounced protection against cell death when compared with the lowest expressing clone 1 (ϳ6.7% at 1 mM H 2 O 2 , respectively, ϳ4.95% at 5 mg/ml tunicamycin (Tn); see Fig. 3A). We compared the transcriptome of clones 1 and 17 using Affymetrix chip human genome U133 Plus 2.0 containing 47000 different human probe sets to identify the downstream targets of GPR39 that mediate the protective effect. We chose this approach over comparison with wild type or vector-transfected cells because clones 1 and 17 only differ 2.9-fold in the amount of GPR39 mRNA and of course by the site of genomic integration. We figured that comparing these very similar clones would result in more meaningful data then simply comparing GPR39-overexpressing vector-expressing cells. Transcripts relatively increased in clone 17 but also expressed in clone 1 are shown in Fig. 3B in descending order. The corresponding proteins are enzymes like enolase (ENO2), tissue plasminogen activator (PLAT) or the putative serine protease PRSS23, proteins of the extracellular matrix like the integrinbinding protein (EDIL3) or collagen II alpha 1 (COL2A1), the major histocompatibility complex-binding nuclear protein MAGEB2, and zinc finger protein TRIM38. Furthermore, proteins involved in transcription and signal transduction like PEDF, the putative high mobility group transcription factor TNRC9, and the transcriptional activator Cited1 were identified (Fig. 3B). To verify the array results, we confirmed the upregulation of PEDF by quantitative PCR, which yielded similar results (Fig. 3C). Of utmost interest was, however, the increased expression of the regulator of G-protein signaling RGS16, because some RGS proteins attenuate signals that induce their expression in a feedback desensitization loop, as shown for RGS16 in lysophatidic acid signaling (10) or for RGS1 in the stimulation of B-lymphocytes by platelet-activating factor (11 Fig. 4A). RGS16 directly binds to G␣ 13 and translocates the complex to detergent-resistant membranes (10). This reduces p115Rho-GEF binding and leads to a reduced capacity of p115Rho-GEF to catalyze the nucleotide exchange on the small GTPase RhoA. In line with this, dominant-negative RhoA T19N (vector ϳ14.65% versus GPR39 ϳ26.2%, p Ͻ 0.05, versus GPR39ϩRhoA T19N ϳ19.1% versus vector (not significant) analysis of variance; Fig. 4B) as well completely abolished the cytoprotective effect similar to pharmacological inhibition of the Rho-associated coiled-coil forming protein serine/threonine kinase p160ROCK, which lies downstream of RhoA, with the specific inhibitor Y-27632 (12) (vector ϳ2.3% versus GPR39 ϳ15.8%, p Ͻ 0.05, versus GPR39ϩY-27632 ϳ4.9% (not significant) analysis of variance; Fig. 4C).
Also, overexpression of GPR39 in serum-starved mouse embryonal fibroblasts resulted in prominent actin stress fiber polymerization, which was inhibited by co-expression of RGS16, further proving the activation of the G␣ 13 /RhoA axis by GPR39 and its inhibition by RGS16 (Fig. 4D). RhoA regulates SRE-mediated gene expression; accordingly, GPR39 increased SRE but not cAMP response element induction 4-fold using a luciferase reporter read-out. We even observed a decrease in cAMP response element-mediated transcription, which was enhanced by RGS16 (Fig. 4E, upper panel). We used exactly the same conditions used for cell death assays, meaning in the presence of serum, which might explain the more pronounced induction of SRE described by Holst et al. (13). The induction of SRE-Luc was completely abolished by co-expression of RGS16 or a dominant-negative RhoA (T19N) (Fig. 4E, lower panel), suggesting that the protective effect of GPR39 overexpression is mediated by SRE induction. This view is strengthened by the fact that overexpression of a constitutively active serum response factor, but not of a construct lacking the DNA-binding domain (SRF⌬M-VP16), similarly protected against ER stress (SRF, ϳ40.9%; vector, ϳ7.6%; SRF⌬M-VP16, ϳ20.6%; viability at 10 mg/ml Tn; Fig. 4F), as shown previously for apoptosis elicited with the natural toxin jasplakinolide (14). Moreover, co-expression of a dominant-negative mutant of SRF (15) completely abolished the protective effect of GPR39 and was even detrimental on its own (Fig. 4F, right panel). We infer that the protective action of GPR39 is mediated by a G␣ 13 /RhoA/p160ROCK/SRF-mediated increase in SREdependent transcription.
PEDF Mediates Part of the Cytoprotective Action of GPR39-PEDF is a soluble cytoprotective factor (reviewed in Ref. 16) found increased in clone 17, which could mediate the protective action of GPR39. PEDF was also up-regulated about 4-fold after transient overexpression of GPR39, which was abolished and even reduced by co-expression of dominant-negative SRF. Overexpression of SRF-VP16 but not SRF lacking the DNAbinding domain (⌬M) resulted in a similar up-regulation (Fig.  5A). At the protein level, cells overexpressing GPR39 secreted ϳ4.5-fold more PEDF than control cells (Fig. 5B). PEDF dosedependently protected against ER stress although not as potently as GPR39 overexpression (Fig. 5C). In these experiments, we used only 2% serum to avoid degradation of PEDF and limit the effect of PEDF contained in the serum, which makes a side-by-side comparison of these experiments difficult. The reverse experiment of removing PEDF decreased the protective effect of medium conditioned by GPR39-overexpressing cells, suggesting a prominent role for PEDF in GPR39-mediated B, PEDF protein secreted into the medium of GPR39-overexpressing HT22 cells determined by enzyme-linked immunosorbent assay. C, viability of HT22 cells treated with PEDF and exposed to the indicated amounts of Tn. D, viability of HT22 cells incubated with serum-free medium conditioned by HT22 cells overexpressing GPR39 (GPR39-CM) or empty vector (EV-CM) containing 2 g/ml ␣-PEDF or ␣-IgY antibody as indicated for 24 h before exposure to the indicated amounts of tunicamycin.
cell protection. On the other hand, even conditioned medium stripped of PEDF retained a clear-cut protective effect, suggesting additional factors (Fig. 5D). In summary, we believe that PEDF is at least one and maybe the major downstream effector of GPR39-induced cell protection.

DISCUSSION
We conclude that GPR39 protects against oxidative, ER, and mitochondrial stress by inducing PEDF and probably other protective transcripts in a G␣ 13 /RhoA/SRE-dependent manner (summarized in Fig. 6). This pathway is similar to that induced by other GPCRs like those activated by lysophosphatidic acid or the protease-activated receptors that bind thrombin. For both ligands conflicting results regarding their role in cell death have been described. Lysophatidic acid activates a large variety of GPCRs that can couple to distinct G-proteins, which renders the elucidation of a particular pathway difficult (reviewed in Ref. 17). In the case of thrombin, dose seems to play the major role. In low concentrations thrombin protects astrocytes and other cell types against cell death, whereas high concentrations are detrimental (reviewed in Ref. 18). Both of these responses are, however, dependent on RhoA. In apoptotic concentrations (1 M thrombin), RhoA activity rapidly increases within minutes, whereas low concentrations (10 nM) lead to a small but consistent increase over hours (19). In the case of GPR39, the ligand is not known and can therefore not be added in higher concentrations to examine a similar dichotomy. We think that the almost linear correlation of GPR39 expression, and its ability to protect against cell death observed by us suggests that GPR39 is constitutively active and leads to low level and thus protective activation of the RhoA pathway. The constitutive activation of GPR39 was already suggested by Holst et al. (13), who also observed that GPR39 dose-dependently activates SREdependent transcription. It can, however, not be excluded that the unknown ligand of GPR39 is present either in the serum or produced by the cells in an auto-or paracrine manner. Alternatively, GPR39 might be reversely regulated, e.g. by a natural inverse agonist.
Zhang et al. (20) reported that GPR39 binds obestatin, a peptide derived by posttranslational cleavage from the grehlin prepropetide, and stimulates in Chinese hamster ovary and HEK293 cells cAMP production and only to a lower degree SRE activation. Obestatin is supposed to suppress food intake, inhibit jejunal contraction, and decrease body weight gain (20). The activation of GPR39 by obestatin has been challenged recently, and it is now generally accepted that obestatin is not the ligand of GPR39. Holst et al. (21) found no reproducible effect of obestatin in GPR39-expressing cells, whereas zinc ions stimulated inositol phosphate turnover, cAMP production, arrestin mobilization, as well as cAMP response element dependent, and SRE-dependent transcriptional activity in GPR39 expressing cells. Also, no specific binding of obestatin could be detected in two different types of GPR39-expressing cells using three different radio-iodinated forms of obestatin. In our hands, obestatin also had no effect on the viability of wild type HEK-293T cells or GPR39-overexpressing cells, whereas zinc killed dose-dependently (data not shown) in the time frame of our experiments as generally accepted (reviewed in Ref. 22). When treated for hours as in our experiments, the detrimental properties of zinc probably override its modulatory effect on GPR39 signaling, which was previously studied in experiments of much shorter duration.
GPR39 consists of two exons separated by a large intron, which contains, according to a recent report, an alternative C terminus resulting in a truncated 5TM receptor (23). The second exon also corresponds to the 3Ј-untranslated region of LYPD1, a putative signaling protein probably expressed both as a membrane-associated and as a secreted protein. It was suggested by the authors that the GPR39 mRNA previously reported to be expressed in nervous tissue mainly corresponds to LYPD1, whereas GPR39 is predominantly expressed in the gastrointestinal system. Our screening primers were indeed designed on the part of GPR39 shared with LYPD1, because this fact was not known at the time we began our study. Control exon-spanning primers resulted in a slightly decreased expression corresponding to the lower expression found by flow cytometry. However, all of the subsequent experiments were done with full-length 7TM GPR39, and the silencing constructs are based solely on the first exon unique for GPR39, thereby excluding any effect of the splice isoform or LYPD1. Egerod et al. (23) suggested that the truncated form of GPR39 might modulate the function of the full-length protein similar to the 5TM form of the closely related ghrelin receptor, which seems to be capable to suppress the function of the full-length recep-α β γ tor (24). A similar modulation of the effect of GPR39 on cell death remains to be investigated.
In summary, we conclude that constitutive activation of the SRE pathway and not activation by obestatin mediates the protective action of GPR39 overexpression against endogenous and exogenous oxidative stress, ER stress, and activation of the caspase cascade by Bax overexpression. GPR39 is a novel inhibitor of cell death, which might represent a therapeutic target with implications for processes involving apoptosis and ER stress like cancer, ischemia/reperfusion injury, and neurodegenerative disease.