Orphan G protein–coupled receptor GPR56 plays a role in cell transformation and tumorigenesis involving the cell adhesion pathway

Cancers are genetic diseases resulting from multiple disease-causing changes in gene expression that involve a variety of biological pathways. The causal expression changes can be revealed by expression profiling coupled with phenotype experiments based on gain or loss of function. The data derived from this type of studies can be used to further elucidate the biological pathways contributing to cell transformation and tumorigenesis mediated by the gene of interest. Additional genes with oncogenic properties in these pathways can also be explored as antagonistic targets for cancer therapy.

Cell adhesion is an important aspect in the process of cellular transformation. Cells interact with their environment via adhesion molecules on the cell surface to the extracellular matrix (ECM) or neighboring cells, thus mediating many critical cellular functions, including survival. A transformed cell usually shows differential expression of a variety of adhesion proteins and thus also manifests different growth (e.g., acquired anchorage-independent growth capacity) and migratory properties from that of their normal counterparts. Attenuating the adhesion properties of cancer cells would, therefore, have potential therapeutic implications. Many proteins are involved in cell adhesion, including surface proteins such as various integrins (whose ligands include ECM or Ig superfamily receptors), cadherins, and intracellular signaling molecules such as Fyn (a proto-oncogene), focal adhesion kinases, and extracellular signal-regulated kinase (ERK; refs. 1, 2). These protein kinases are involved in adhesion-related signaling. Some of the adhesion proteins have been explored as drug targets for the therapeutics of human diseases, including cancer and inflammatory diseases (1, 2).

GPR56, also known as TM7XN1 (7-transmembrane protein with no epidermal growth factor-like NH₂-terminal domains 1; Genbank accession no. NM_005682), is an orphan G protein–coupled receptor (GPCR) belonging to the secretin-like receptor of the GPCR subfamily (also called adhesion GPCR subfamily) with 33 members (3). This subfamily is characterized by an extremely long extracellular domain (4, 5) with a typical GPCR proteolytic site (3). The secretin-like polypeptide ligands include secretin, calcitonin, parathyroid hormone/parathyroid hormone–related peptides, and vasoactive intestinal peptide, all of which activate adenyl cyclase and the phosphatidylinositol-calcium pathway. They also have their own unique “7TM” signature sequence (4, 5).

GPR56 is expressed as a 3-kb mRNA in various peripheral tissues, with the highest levels in the brain, thyroid gland, and heart (4–6). GPR56 has recently been implicated in cortical patterning, and genetic mutations in GPR56 have been associated with the bilateral frontoparietal polymicrogyria, a rare hereditary condition that is characterized by disorganized cortical lamination primarily in the frontal cortex region (7). The mutations include a frameshift mutation, splicing mutations, and several missense point mutations exclusively located on the predicted extracellular regions, including the NH₂-terminal extracellular domain, GPCR proteolytic site, and the extracellular loops of the transmembrane domain. Our recent study clearly showed that all the observed missense mutations result in the failure of the mutated proteins to traffic to the cell surface. These loss-of-function mutations have been implicated in the defect in neuronal migration and proliferation that originate from ventricular and subventricular zones during development of the bilateral frontoparietal polymicrogyria patients. However, no phenotypes outside the central nervous system (CNS) have been observed in these adult patients (7), suggesting the dispensability of GPR56 postdevelopment. Others recently reported increased GPR56 expression in human glioma, a central nervous malignancy, compared with normal brain tissues and suggested a possible role of GPR56 in cell adhesion and glioma tumorigenicity (6). However, no further studies have been reported about this effect on non-CNS malignancies. On the other hand, another study reported that GPR56 suppresses tumor growth and metastasis in melanoma, an opposite effect to that implied from the glioma study (8).

In the present study, we have extensively examined the expression of GPR56 in a variety of tissues and showed that, while ubiquitously expressed, GPR56 was further up-regulated in many cancer types compared with their normal counterparts. In addition, significantly elevated levels were observed in the transformed HeLa cells compared with its isogenic nontransformed revertant, HeLaHF (9, 10), suggesting the correlation between expression and cell transformation/tumorigenicity. Down-regulation of GPR56 reverses transformation phenotypes in many cancer cell lines tested and also led to tumor growth inhibition, including regression in several human xenograft tumor models. Although few GPCRs have been considered as cancer drug targets, the disease-modifying “oncogene” role of GPR56, potentially via a cell adhesion pathway, clearly suggests a therapeutic potential of targeting this gene for human cancers.

A2058, M14, OVCAR8, HeLa, NIH3T3, HCT116, and PC3M cells were obtained from the American Type Culture Collection. HeLaHF cells are kind gifts of Dr. H. Zarbl (Fred Hutchinson Cancer Center, Seattle, WA). A2058, NIH3T3, HeLa, and HeLaHF cells were cultured in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen), 2 mmol/L l-glutamine (Fisher Scientific), 1× nonessential amino acids (Irvine Scientific), and 1% sodium pyruvate (Invitrogen). PC3M, HCT116, M14, and OVCAR8 cells were maintained in RPMI 1640 (Invitrogen) supplemented with 10% fetal bovine serum and 2 mmol/L l-glutamine. All cells were maintained in a humidified incubator with 5% CO₂ at 37°C.

Construction, preparation, and transduction of constitutive lentiviral vector pSD31 and inducible vector pTRIP expressing short hairpin RNAs (shRNA) have been described before (10–12). Primer sequences used in the study are as follows: shGPR56A, 5′-tgctGGATCCAAAAAAgAAATGTGGCTCCAGTTGCTCTcTCTCTTGAAgAGAGCAACTGGAGCCACATTTcAAACAAGGCTTTTCTCCAAGGG-3′; shGPR56B, 5′-tgctGGATCCAAAAAAgAAGGTGCACATGAACCTGCTGcTCTCTTGAAgCAGCAGGTTCATGTGCACCTTcAAACAAGGCTTTTCTCCAAGGG-3′; shCNTL, 5′-tgctGGATCCAAAAAAggcgcgctttgtaggattcgcTCTCTTGAAgcgaatcctacaaagcgcgccAAACAAGGCTTTTCTCCAAGGG-3′; and Fyn, 5′-tgctGGATCCAAAAAAggaatggactcatatgcaaTCTCTTGAAttgcatatgagtccattccAAACAAGGCTTTTCTCCAAGGG-3′. HeLa, A2058, PC3M, OVCAR8, and HCT116 cells stably expressing shRNAs were generated by transduction and selection in medium containing the desired concentrations of puromycin for different cells.

The construction of the expression vector of full-length GPR56 cDNA, FL-WT, was carried out as follows. GPR56 cDNA (clone MGC-1409) was obtained from the American Type Culture Collection. The full-length coding sequence of GPR56 was PCR amplified from MGC-1409 using primers 5′-atAAGCTTgcggccgcGCCACCatgactccccagtcgctgctg-3′ and 5′ gctATCGATActagatgcggctggacgaggt-3′. PCR products were restriction digested with HindIII and ClaI and cloned into the corresponding sites in pLHCX (FL-WT; Clontech). Retroviral vectors of LHCX or FL-WT (LHCX) were transduced into A2058 and murine fibroblast cell line NIH3T3. Stable cells were selected in medium with desired concentration of hygromycin.

Anchorage-dependent and anchorage-independent growth in 96-well, as well as anchorage-independent colony formation in 10-cm plate, were described in detail previously (10, 11, 13). For mitogen-activated protein kinase/ERK kinase inhibitor inhibited cell growth, cells were seeded in 96-well plates. PD98059 (10 μmol/L; Sigma) was added to the culture medium 24 h later, and the cells were incubated for another 4 days. Cell growth was monitored by Alamar Blue staining. Measurement of cell apoptosis and anoikis was also described previously (10). For focus formation assay, cells were seeded in 10-cm dishes and allowed to grow to confluence. Cells were kept in medium until foci formed. Foci were then stained using crystal violet and counted.

Ninety-six–well plates were coated with fibronectin at 10 μg/mL at 4°C for 3 days. A2058 cells (1 × 10⁴) containing either shCNTL or shGPR56A were seeded onto the wells coated with fibronectin. One hour after seeding, the unattached cells were washed away and the attached cells were scored using Alamar Blue staining.

For PC3M cells, 5 × 10⁶ cells with inducible shRNA cassettes for either pTRIP-shCNTL or pTRIP-shGPR56A were used for xenograft experiment; for A2058 cells, cells with either pTRIP-shCNTL or pTRIP-shGPR56B were used. Cells were injected s.c. into athymic nude mice. Drinking water with 5% sucrose plus or minus doxycycline (2 mg/mL) was provided on day 0 (PC3M) or day 1 (A2058; the induction schedule for the early-stage tumor) or on day 16 after implantation (the induction schedule for advanced-stage PC3M tumors). Tumor volume was measured twice weekly starting on day 7 (1/2 × length × width²) when the tumors became measurable. For the advanced-stage PC3M tumor schedule, the animals were grouped into two groups with the similar mean tumor volume (∼270 mm³) and with tumor size >100 mm³. Then, one of the two groups was treated with doxycycline and another was not.

mRNA preparation from cell lines and tumor samples was described earlier (10, 12). The mRNA levels of GPR56 and Fyn were determined by Taqman real-time reverse transcription-PCR (RT-PCR) as described previously (10, 11). Dual-labeled fluorogenic probe and primers were synthesized at Eurogentec North America, Inc. and their sequences were as follows: GPR56 probe, 5′-AGCACCAGCTACAGCCGAAGAATGTGAC-3′; GPR56 forward primer, 5′-GGTACAGAACACCAAAGTAGCCAAC-3′; and GPR56 reverse primer, 5′-TCAACCCAGAACACACATTGC-3′. The detailed procedure and data analysis for Affymetrix GeneChip array experiment is described in Supplementary Data.³

For real-time RT-PCR (Taqman) and biological phenotype (Alamar Blue, DNA fragmentation, etc.), data were presented as mean of triplicates. SDs were also provided. P values reflecting samples versus control were calculated using two-tailed t test. Usually, more than two independent experiments have been done for the same experiment, and data from one of the multiple experiments with consistent results were presented. For xenograft experiments, 4 to 10 animals per group were used as indicated in the figure legends and the text. The mean of the groups and the SEs are shown in the figures along with the P values for statistical significance.

An important piece of evidence that GPR56 may be associated with cell transformation came from our earlier global gene expression profiling studies of the isogenic cell pair HeLa and HeLaHF, which exhibit distinct transformation phenotypes (10, 14, 15). HeLaHF is a nontransformed revertant that has lost the capacity for anchorage-independent growth and tumorigenicity of the parental HeLa cells while retaining similar anchorage-dependent growth properties (9). GPR56 (accession no. NM_005682; Affychip probe_set: 206582_s_at) was shown to be up-regulated in HeLa cells, as well as in the tumors derived from the retransformed HeLaHF (16), by 6- to 9-fold over the nontransformed HeLaHF cells (Fig. 1). This up-regulation of GPR56 was also confirmed by Taqman real-time RT-PCR (data not shown). The observed close correlation of GPR56 expression level to the transformation phenotype of HeLa cells suggested a possible role of GPR56 involvement in HeLa cell transformation.

Next, we attempted to investigate the possible relevance of GPR56 in human cancers by examining molecular epidemiology information. We first mined the data from the National Center for Biotechnology Information Virtual Northern database and found that GPR56 was widely expressed in nearly all human tissues, including heart, brain, kidney, breast, liver, pancreas, and lung, confirming the earlier report (4–6). In particular, the in silico data also showed up-regulation in several human cancers, including breast cancer, pancreatic cancers, non–small cell lung, brain tumor, and renal carcinoma, compared with their normal counterparts. The broad expression of GPR56 in various tissues and elevated expression in cancer tissues from the in silico data were further confirmed by us through analyzing RNAs from individual patient tumor samples along with the normal tissue samples using Taqman real-time RT-PCR. The up-regulation of GPR56 expression was observed in many cancer types, including ovarian cancers (4 of 5 patients), pancreatic cancer (5 of 5 patients), colon cancer (6 of 10 patients), and non–small cell lung cancer (7 of 10 patients), compared with their normal counterparts (Table 1). The observation of GPR56 up-regulation in the tumors outside the CNS, in line with a previous report of the elevated expression in glioma compared with normal brains (6), may point to a possible role of GPR56 in tumorigenicity in general.

Next, we investigated whether GPR56 expression has causal effect for cell transformation in model systems. We first set out to confirm the endogenous GPR56 expression in several experimental cancer cell lines. Real-time RT-PCR revealed GPR56 expression in almost all the experimental cell lines tested, including melanoma cell lines A2058 and M14, colon carcinoma HCT116, non–small cell lung cancer cell line NCI-H460, prostate cancer cell line PC3, and ovarian cancer cell line OVCAR3 and OVAR8 (data not shown).

We then investigated the role of GPR56 in cell transformation by RNA interference (RNAi)-mediated silencing using a previously described lentiviral RNAi vector pSD31 expressing shRNA (shGPR56; refs. 10, 11). When we transduced pSD31-shGPR56 into several cancer cell lines, including HeLa, A2058, M14, PC3, and HCT116, cytopathic effects were transiently observed in some cancer cells, such as HCT116 colon carcinoma line and M14 melanoma line (data not shown), whereas no cytopathic effect was observed for the pSD31-shCNTL control vector on the same cells. We confirmed that the observed cytopathic effect was due to the induction of apoptosis as determined by the DNA fragmentation–based ELISA assay. Figure 2 shows an example of apoptosis induction in the transiently transduced M14 cells. Apoptosis induction was also observed in several other cancer cell lines tested, including HeLa and HCT116 (Table 2), although no obvious cytopathic effects were seen. This apoptosis resulted in accumulated survival/growth reduction that can be measured by Alamar Blue (Fig. 2B). These observations suggest that GPR56 may have a prosurvival role by preventing intrinsic apoptosis in cancer cells.

Although apoptosis induction was observed, stably transduced cancer cells with significant GPR56 silencing were nevertheless obtained in all the cancer cells tested [e.g., shGPR56-HeLa (Fig. 3A), shGPR56-A2058, and shGPR56-PC3 cells (Fig. 3C), confirmed at both RNA (Fig. 3A and C) and protein levels (Western blot analysis for total GPR56 protein or fluorescence-activated cell sorting for surface localized protein; Fig. 3E and F)]. The stably silenced cancer cells manifested a certain degree of growth reduction (varying from cell to cell) when compared with the shCNTL cells, most likely due to the level of apoptosis. We have previously observed a striking difference in anchorage-independent growth between the HeLaHF versus HeLa cell pair, where GPR56 is significantly lower in HeLaHF than the parental HeLa cells. We therefore evaluated the GPR56-silenced cancer cells for their transformation potential in terms of anchorage-independent growth. The soft agar colony formation assay, both in the 96-well format (measuring overall cell growth) and the traditional format in 100-mm plate format (measuring cloning efficiency by QCount), was used in this study (11, 13). A correlated reduction in both formats was observed for GPR56-silenced cells compared with the control transduced cells, as shown in Fig. 3B for HeLa cells and Fig. 3D for A2058 and PC3 cells. Similar results were obtained with additional cancer cell lines, including M14, OVCAR3, OVCAR8, NCI-H460, and pancreatic cell line AsPC1 (Table 2). This showed that GPR56 plays a direct role in the transformation phenotype of various cancer cells.

A major concern when doing RNAi-based experiments is the potential off-target effects of a particular siRNA/shRNA. One way to ensure that the effect caused by a RNAi is truly due to the specific silencing of the intended gene is to use multiple shRNAs. To this end, we have used an additional shRNA vector, as shown for A2058 and PC3 cells in Fig. 3D. Another approach to rule out off-target effects is to rescue the induced phenotype by overexpressing the target gene in the silenced cells. We expressed full-length GPR56 (FL-WT) in A2058 cells. The expression vector shGPR56 and the control shCNTL vector were introduced into A2058 cells stably transduced with either the vector alone or FL-WT (full-length GPR56; refs. 10, 11). Cell growth in liquid culture was determined after 4 days. Whereas shGPR56 reduced cell growth dramatically, overexpression of GPR56 cDNA was able to rescue the growth almost to the level of cells transduced with the control shRNA vector (Fig. 4). This further confirms that the reduced cell growth phenotype rendered by shGPR56 is GPR56 specific.

Detachment of cells from a substratum often leads to anoikis, which is one of the key factors prohibiting cell anchorage-independent growth (14, 17). We previously showed that the reduced anchorage-independent growth in NR4A2-silenced cells is primarily due to an increase in anoikis (10). We therefore investigated the specific role of anoikis in the reduction of anchorage-independent growth of the GPR56-silenced HeLa and A2058 cells. The cells were plated either in standard tissue culture medium (attached) or in methylcellulose semiliquid medium (detached) and incubated for 16 h. Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling–based ELISA assay was used to quantitate DNA fragmentation, an indication of apoptosis (anoikis). The results showed that significantly elevated levels of anoikis were observed for GPR56-silenced HeLa and A2058 cells (Fig. 5A and B), confirming that the reduction in soft agar cloning efficiency results at least partly from an increase in anoikis.

If inactivation of the GPR56 gene reduces cell transformation in cancer cell lines, it would be interesting to know whether overexpression of GPR56 would enhance transformation. Our observation indicated no obvious enhanced growth phenotypes in vitro by overexpression of full-length GPR56 (FL-WT; data not shown) in melanoma cell line A2058, which can be interpreted as the cells having sufficient basal GPR56 expression for full transformation. This observation prompted us to investigate GPR56 expression in nontransformed cells. NIH3T3 is an immortalized but nontransformed mouse fibroblast cell line that has been widely used to assess the transformation potential of small molecules and biological agents because it is believed that only one hit is needed to transform this cell line. Normally, the cells grow as a monolayer in regular liquid culture and form foci when transformed due to loss of contact inhibition. When GPR56 is overexpressed in NIH3T3 cells, it is posttranslationally processed similarly as in human cells. Although FL-WT-3T3 cells did not exhibit any apparent change in cell growth over the vector-transduced cells (data not shown), they showed pronounced increase in focus formation, which measures loss of contact inhibition of cell growth on monolayer culture (Fig. 6). This observation provided further evidence of the oncogenic properties of GPR56.

GPCRs usually interact with ligands, either extracellularly or on cell surface of neighboring cells, which transduce signal into cells via G proteins to modulate downstream gene expression and function. We were interested in identifying the biological pathways affected by GPR56 function, particularly those influencing the observed transformation phenotype. We therefore compared expression profiles of A2058 cells with or without GPR56 silencing using Affymetrix GeneChip expression arrays (see Supplementary Data for detailed analysis). Using stringent analytic criteria, ∼600 genes were found to be down-regulated at the mRNA level in the GPR56-silenced cells, whereas >200 genes were up-regulated. Among the down-regulated genes (in addition to GPR56, which is the number one silenced gene) are proto-oncogene Fyn, Ras family proteins, and integrins α4,6 and β3,5, consistent with the oncogenic nature of GPR56 (Supplementary Table S1; Supplementary Fig. S1). Correlated down-regulation of Fyn and GPR56 has also been confirmed by real-time RT-PCR, and silencing of Fyn mediated by shRNA also caused a similar phenotype as that seen for GPR56 silencing (i.e., reduced anchorage-independent growth; Supplementary Fig. S2), suggesting a role of Fyn in the GPR56-mediated cell transformation pathway (18, 19).

A functional analysis of the changes was conducted using the informatics tools Ingenuity Pathways Analysis.⁴

The implication of GPR56 in the cell adhesion pathway based on gene expression profiling data can be further argued from the following considerations. First, several other GPCR members in the same subfamily as GPR56 have also been implicated in adhesion functions. Second, the increase in anoikis or decrease in anchorage-independent growth (see above) caused by GPR56 silencing is a typical cell adhesion response. Third, previous observations have also implicated adhesion phenotypes in CNS system (6, 7). To verify the relationship of GPR56 to cell adhesion in non-CNS cancer cells, we examined the effect of GPR56 gene inactivation on cell adhesion properties. We compared A2058 cells with and without GPR56 silencing by plating them onto fibronectin-coated tissue culture plates. Fibronectin is a major ECM protein commonly used to assay cell adhesion functions (20). The cells that adhere to fibronectin after incubation for 1 h were then quantitated. We found that the GPR56-silenced cells exhibited significant reduction in cell adhesion to fibronectin (Fig. 7), confirming a role for GPR56 in cell adhesion.

Genes or pathways that may contribute to GPR56-mediated cell adhesion functions were next examined. Phosphatidylinositol 3-kinase-ERK signaling pathways have been implicated in anchorage-dependent cell cycle progression mediated by growth factor receptor and/or integrin receptor signaling. It is also involved in anchorage-independent proliferation and anoikis mediated by the cell adhesion responses of integrin receptor signaling (21). We have observed that ERK was partially dephosphorylated (inactivated) in GPR56-silenced cells (Supplementary Table S2), similar to observed when using mitogen-activated protein kinase/ERK kinase inhibitors (PD98059), supporting the assumption that GPR56 is also involved in ERK activation.

The above in vitro experimental results thus far implicate the “oncogenic” nature of GPR56. We next attempted to assess the relevance of this property to the in vivo tumorigenicity. We tested whether induced GPR56 silencing would affect human tumor growth in xenograft athymic (nu/nu) mouse models. One critical issue in cancer gene target validation using xenograft model is that if an essential gene for cancer cell growth is stably silenced, no viable cell is available for in vivo engraftment. We therefore chose a staged tumor model using an inducible RNAi system, pTRIP, which we previously described (12), because these models could potentially provide more predictive variables for therapeutic outcomes of naturally occurring human cancers. This model makes it possible to measure the response of different staged tumors to GPR56 inactivation, thus mimicking a commonly used clinical end point. The model is established by transplanting human cancer cells containing a tetO-based inducible lentiviral vector into mice, and the RNAi against GPR56 was induced via oral dose of the inducer doxycycline when tumors were established at varying stages of interest.

We investigated the melanoma xenograft model by first generating A2058 cells stably transduced with pTRIP-shCNTL and pTRIP-shGPR56 vectors, respectively. The pooled transduced cells were tested for growth in medium with (induced) or without (noninduced) doxycycline, an inducer of Tet-regulated expression. With the induced phenotype confirmed in vitro (data not shown), we transplanted the transduced cells into athymic nude mice s.c. under noninduced conditions and allowed the tumors to progress. The pTRIP-shGPR56 cells formed tumors similar to those by pTRIP-shCNTL cells. When the tumors grew to ∼100 mm³ in size, these animals were orally given with doxycycline at 2 mg/mL via their drinking water. The induction of GPR56 silencing in xenograft tumors 3 days after induction in vivo in shGPR56 tumors was compared with that in A2058-shCNTL tumors, and ∼60% down-regulation was observed (Fig. 8A). We next assessed the early-stage tumor response of A2058 melanoma tumors to the induced silencing in vivo by continuous dosing of doxycycline starting on day 1 after implantation, allowing tiny tumors (early stage) to establish before induction of the silencing (see above). The tumor response to GPR56 silencing was assessed by tumor volumes measured twice weekly = 1/2 (length × width × width). Two groups of animals were transplanted with either A2058-shCNTL or A2058-shGPR56 cells (with eight tumors per group). The results showed that induction of GPR56 silencing resulted in significant (P < 0.05) tumor growth inhibition for shGPR56 tumors compared with shCNTL tumors [(1 − T/C)% ∼40%, where T is the tumor volume of treated (shGPR56) and C is the tumor volume of control (shCNTL)], which suggests that GPR56 does play a causal role in in vivo tumorigenicity (Fig. 8B). It is possible that a higher degree of silencing in vivo in the tumor cells could yield even more robust tumor response.

To investigate whether a similar tumor response can also be observed in other cancer types, we also experimented with prostate cancer cell line PC3, a model we previously reported for assessing staged tumor response to the induced mammalian target of rapamycin silencing (12). After the stably transduced PC3 cells were similarly characterized in vitro, the cells were transplanted into athymic mouse similar to that of the A2058 experiment. In this PC3 model, two induction schedules were applied to evaluate both early-stage and the advanced-stage tumor response.

In the first induction schedule, PC3 cells with either pTRIP-shCNTL vector or pTRIP-shGPR56 were transplanted into athymic nude mice s.c., and the animals were continuously dosed with doxycycline starting on the same day of implantation, allowing tiny tumors (early stage) to establish before induction of the silencing. The tumors were found to grow until day 10, after which regression was observed for tumors containing pTRIP-shGPR56, whereas all the pTRIP-shCNTL tumors continue to grow. By day 17, no shGPR56 tumors could be detected, and all the 10 animals remained tumor-free until day 37 when the experiments were terminated (Fig. 8C). Thus, GPR56 is essential for the growth of early-stage PC3 prostate tumors.

In the second dosing schedule, tumors with shGPR56 were allowed to grow under noninduced conditions until day 16 to advanced stages. Animals with tumor sizes >100 mm³ (14 of 20 mice) for shGPR56 were chosen and regrouped into two new groups so that the mean tumor sizes were similar between the groups (∼270 mm³). Continuous oral dosing of doxycycline commenced on day 16 after implantation in one group, whereas the other group was left untreated. On induction, the shGPR56 tumors ceased to grow and tumor regression was observed at a slow but convincing rate for 80% (six of seven) of the animals until day 28, after which some tumors started to regrow (Fig. 8D). In contrast, all shCNTL control tumors continued to grow (12). The 100% regression response (10 of 10) observed for the early-stage tumors and 80% (6 of 7) regression response for the advanced-stage tumors suggest that GPR56 is a potent target for cancer therapy.

We also assessed an early-stage HCT116 colon cancer model for response to the induced GPR56 silencing, similar to the A2058 experiment, and comparable responses were obtained (data not shown; Table 3). These in vivo observations correlate well with the in vitro testing and imply an important role for GPR56 in many human cancers.

GPR56 is an orphan GPCR with unknown functions. It has, however, been implicated in the development of the frontal cortex of the human brain (7) according to the genetic analysis of a rare central nervous condition, bilateral frontoparietal polymicrogyria. Because there is no apparent phenotype outside the CNS in the GPR56-mutated patients, little is known about its function in nonbrain tissues, although GPR56 is widely expressed as observed by us and others. A recent separate study showed that GPR56 was overexpressed in human gliomas and suggested that it might affect glioma cell adhesion (6). Our data reveal that GPR56 is also overexpressed in many non-CNS cancers and provide the first experimental evidence for the oncogenic properties of this gene in non-CNS malignancies and potential pathway information.

Interestingly, another recently published study reported that GPR56 played a role in down-regulating melanoma cell survival and metastatic potential (8), which is exactly opposite to what we have observed. In their study, gene expression profiling was conducted on a metastatic melanoma cell and its nonmetastatic parental cell. GPR56 expression was found to be down-regulated in the metastatic cells (4, 6). Overexpression of GPR56 suppresses melanoma tumor growth and metastasis in an in vivo xenograft model (8). Hence, according to this study, GPR56 functions as a “tumor suppressor.” It would be interesting to investigate whether the GPR56 mutant populations have higher or lower cancer incidence. These contradictory observations remain to be addressed. Nevertheless, our GPR56-overexpressing FL-WT-A2058 cells did not show any apparent growth phenotype under both anchorage-dependent and anchorage-independent conditions. It would be interesting to know whether overexpression of GPR56 would affect A2058 tumorigenicity or metastatic potential in vivo.

GPR56, as a member of GPCRs, is responsible for transduction of signals into cells and was suggested to physically interact with the G protein Gα_q/11 and with the tetraspanins CD81 and CD9 in a biochemical study in which these proteins were overexpressed (22). It was suggested that intracellular scaffolding involving heterotrimeric GPCRs achieves specificity in signaling through a few heterotrimeric G proteins. However, down-regulation of CD81 in A2058 cells in our experiments did not affect cell growth and/or cell survival as measured by soft agar culture (data not shown), which implies that CD81 is not required for cell transformation mediated by GPR56, at least in our experimental systems. Although the proteins directly interacting with GPR56 for signal transduction and mediating cell transformation remain elusive, our analyses of A2058 cells identified several genes downstream of GPR56 involved in the activation of cell adhesion pathways. These include transcriptional regulation of Fyn, integrins, etc. and activation of ERK via phosphorylation.

Cell adhesion to the ECM plays a vital role in cell survival. Epithelial cells unable to adhere to the ECM undergo detachment-triggered apoptosis (anoikis). Transformation usually enhances integrin-mediated cell adhesion, resulting in increased cell motility and invasiveness and greater resistance to apoptosis. Integrins have been known to have prosurvival roles in epithelial cells and endothelial cells through adhesion to the ECM. Their down-regulation results in the detachment of cells from the ECM, which usually leads to apoptosis (anoikis). Because GPR56 seems to regulate integrin expression and signaling, its role in transformation could potentially be attributed at least in part to the integrin-mediated cell adhesion pathways. Murine GPR56 is preferentially expressed in neuronal progenitor cells of the cerebral cortical ventricular and subventricular zones during periods of neurogenesis, suggesting it may be involved in neuronal proliferation and may be relevant to our observed prosurvival and adhesion nature of GPR56.

Although GPR56 is broadly expressed in many normal human adult tissues, its overexpression has been observed in various non-CNS malignancies in patients, particularly in breast, pancreatic, brain, and renal cancers according to National Center for Biotechnology Information Virtual Northern database and our own data (Table 1). In this study, we showed that inhibition of GPR56 suppresses cancer cell growth (particularly under anchorage-independent growth conditions) and tumorigenicity in vivo, which indicates that GPR56 could potentially serve as a target for cancer therapeutics. We have previously described an inducible RNAi system to assess xenograft tumor response to induced target silencing exemplified by targeting mammalian target of rapamycin, a well-known cancer target that is under evaluation in several clinical trials. Using this system, we also evaluated GPR56. Our current data in PC3 model seem to show more potent effect of targeting GPR56 than previously observed when targeting mammalian target of rapamycin, suggesting the possibility of a novel and more effective strategy for cancer therapy.

GPR56 has additional features that make it an attractive drug target. (a) Because there are no apparent non-CNS or postdevelopmental defects resulting from the loss-of-function mutations of GPR56 in humans (7), targeting GPR56 in adult patients should potentially be feasible. (b) Interestingly, GPR56 silencing induced a much more profound level of apoptosis in cancer cells than in several nontransformed primary cells tested using the transient transduction. The normal cells tested included primary endothelial human umbilical vascular endothelial cell and fibroblasts, such as WI38 and IMR90 (data not shown). This may suggest that GPR56 is not required for normal cell survival, consistent with the lack of non-CNS and non–postdevelopment-related phenotype for the null mutants. (c) p53 independent because PC3 cells lack wild-type p53 function and still responded to GPR56 inactivation. Therefore, GPR56 is a potential target for p53-deficient cancers, which comprise >50% all cancers and display poor prognosis and higher resistance to drugs and radiation therapy. (d) Although few GPCRs are targeted for cancer therapy, GPCRs are traditionally considered highly attractive as potential drug targets either by small molecules or antibody agents, particularly for its characteristic long NH₂-terminal extracellular domain with the likely polypeptide ligand-binding properties. Identification of the ligand(s) and elucidation of the protein structure of GPR56 would greatly facilitate the discovery of these drugs.

We thank Dr. H. Zarbl for kindly providing HeLaHF cells; Ingenuity Pathways Analysis for access to their tools for this study; Drs. Johnny Peppers, Demin Zhou, Xiuyuan Hu, Hongwen Ma, Gisela Claassen, Gundo Diedrich, and Jon E. Chatterton for useful suggestions; and Aaron Albers, Lindsey Tsugawa, Jing Zhang, and Violet Abraham for excellent technical assistance.