Naturally occurring hotspot cancer mutations in Gα13 promote oncogenic signaling

Heterotrimeric G-proteins are signaling switches broadly divided into four families based on the sequence and functional similarity of their Gα subunits: Gs, Gi/o, Gq/11, and G12/13. Artificial mutations that activate Gα subunits of each of these families have long been known to induce oncogenic transformation in experimental systems. With the advent of next-generation sequencing, activating hotspot mutations in Gs, Gi/o, or Gq/11 proteins have also been identified in patient tumor samples. In contrast, patient tumor-associated G12/13 mutations characterized to date lead to inactivation rather than activation. By using bioinformatic pathway analysis and signaling assays, here we identified cancer-associated hotspot mutations in Arg-200 of Gα13 (encoded by GNA13) as potent activators of oncogenic signaling. First, we found that components of a G12/13-dependent signaling cascade that culminates in activation of the Hippo pathway effectors YAP and TAZ is frequently altered in bladder cancer. Up-regulation of this signaling cascade correlates with increased YAP/TAZ activation transcriptional signatures in this cancer type. Among the G12/13 pathway alterations were mutations in Arg-200 of Gα13, which we validated to promote YAP/TAZ-dependent (TEAD) and MRTF-A/B-dependent (SRE.L) transcriptional activity. We further showed that this mechanism relies on the same RhoGEF-RhoGTPase cascade components that are up-regulated in bladder cancers. Moreover, Gα13 Arg-200 mutants induced oncogenic transformation in vitro as determined by focus formation assays. In summary, our findings on Gα13 mutants establish that naturally occurring hotspot mutations in Gα subunits of any of the four families of heterotrimeric G-proteins are putative cancer drivers.

Heterotrimeric G-proteins are signaling switches broadly divided into four families based on the sequence and functional similarity of their Ga subunits: G s , G i/o , G q/11 , and G 12/13 . Artificial mutations that activate Ga subunits of each of these families have long been known to induce oncogenic transformation in experimental systems. With the advent of next-generation sequencing, activating hotspot mutations in G s , G i/o , or G q/ 11 proteins have also been identified in patient tumor samples. In contrast, patient tumor-associated G 12/13 mutations characterized to date lead to inactivation rather than activation. By using bioinformatic pathway analysis and signaling assays, here we identified cancer-associated hotspot mutations in Arg-200 of Ga 13 (encoded by GNA13) as potent activators of oncogenic signaling. First, we found that components of a G 12/13 -dependent signaling cascade that culminates in activation of the Hippo pathway effectors YAP and TAZ is frequently altered in bladder cancer. Up-regulation of this signaling cascade correlates with increased YAP/TAZ activation transcriptional signatures in this cancer type. Among the G 12/13 pathway alterations were mutations in Arg-200 of Ga 13 , which we validated to promote YAP/ TAZ-dependent (TEAD) and MRTF-A/B-dependent (SRE.L) transcriptional activity. We further showed that this mechanism relies on the same RhoGEF-RhoGTPase cascade components that are up-regulated in bladder cancers. Moreover, Ga 13 Arg-200 mutants induced oncogenic transformation in vitro as determined by focus formation assays. In summary, our findings on Ga 13 mutants establish that naturally occurring hotspot mutations in Ga subunits of any of the four families of heterotrimeric G-proteins are putative cancer drivers.
Heterotrimeric G-proteins are critical transducers of signaling triggered by a large family of G-protein-coupled receptors (GPCRs). Essentially, GPCRs promote GTP loading on the a-subunits of G-proteins (1,2), which triggers signaling downstream. Heterotrimeric G-proteins are composed of a nucleotide-binding Ga subunit and an obligatory Gbg dimer, and they are classified into four families based on the nature of the Ga subunits. These four families are G s , G i/o , G q/11 , and G 12/13 , and Ga subunits of each one of them have distinct actions on specific effectors. For example, G s members stimulate adenylyl cyclase activity, whereas G i/o family members tend to inhibit it; G q/11 members stimulate phospholipase C enzymes and a sub-group of RhoGEFs; and G 12/13 members stimulate a different subgroup of RhoGEFs (3,4). Signaling is terminated upon GTP hydrolysis mediated by the intrinsic GTPase of Ga subunits.
The role of heterotrimeric G-proteins in cancer-related signaling has been documented for decades. Early studies identified cancer-associated mutations in Ga s that disrupted its GTPase activity, rendering the G-protein constitutively active (5). This seminal finding spurred a wave of studies exploring whether analogous mutations introduced artificially in other Ga subunits would also promote their ability to induce oncogenic transformation. It was found that GTPase-deficient mutants of most, if not all, Ga subunits tested led to oncogenic transformation in vitro, regardless of the G-protein family they belonged to. For example, Ga i2 , Ga o , and Ga z (G i/o family); Ga q (G q/11 family); and Ga 12 and Ga 13 (G 12/13 family), in addition to Ga s (G s family), all promoted oncogenic transformation as assessed by in vitro assays using fibroblasts (6)(7)(8)(9)(10)(11)(12)(13). In most cases, transformation in vitro correlated well with tumor growth in vivo using mouse xenografts. Thus, one theme emerging from these studies was that enhancement of GPCR/ G-protein signaling tends to favor oncogenicity.
Despite these initial observations and the identification of some mutations in G-proteins in tumors (5,14), only with the recent advent of deep-sequencing techniques has it become obvious that dysregulation of the GPCR/G-protein signaling axis in cancer is highly prevalent (15)(16)(17). The mutational landscape of GPCR/G-protein signaling components in cancer supports the theme that G-protein hyperactivation in cancer tends to be pro-oncogenic. There are many examples of GPCRs that are either overexpressed or contain activating mutations (15,(17)(18)(19)(20), and negative regulators of G-protein activity have also been shown to bear loss-of-function mutations (21). As for Gproteins themselves, it is now known that hyperactive G-protein mutants can be very frequent in certain types of cancers. The most striking example is uveal melanoma, in which ;90% of tumors contain activating mutations in Ga q (encoded by GNAQ) or Ga 11 (GNA11) (22,23). Similarly, activating mutations in Ga s (GNAS) can be as frequent as 70% in certain subtypes of pancreatic ductal carcinomas (24,25), and activating mutations in Ga i2 can be as frequent as 24% in epitheliotropic intestinal T-cell lymphomas (26). Thus, for representative members of three of four families of G-proteins (G q/11 , G s , and G i/o ) the oncogenic activity in vitro caused by artificially introduced mutations has found a counterpart in prevalent mutations in cancer. Interestingly, findings so far suggest that the remaining family of G-proteins (G 12/13 ) might be an exception to this trend. For example, mutations in Ga 13 in some types of lymphoma are frequent, but they lead to inactivation rather than activation (27,28). This suggests that, at least in these lymphomas, Ga 13 activity is tumor-suppressive. This is the opposite of the oncogene function previously suggested from experiments in vitro with a constitutively active artificial Ga 13 mutation (7,9). Activation of Ga 12 or Ga 13 proteins leads to activation of RhoA-dependent transcriptional programs, including those mediated by the activation of the Hippo pathway effectors YAP and TAZ. The cascade of events involves the direct activation of a subgroup of RhoGEFs, composed of p115-RhoGEF, PDZ-RhoGEF, and LARG, by active, GTP-loaded Ga subunits of G 12/13 proteins, which in turn activates RhoA, RhoB, and RhoC GTPases (29). Through mechanisms that involve the remodeling of the actin cytoskeleton, Rho GTPases induce transcriptional responses that include those regulated by YAP/TAZ, which serve as co-factors for the TEA domain-containing transcription factor family (TEADs), and via myocardin-related transcription factors A and B (MRTF-A/B), which serve as coactivators for the transcriptional factor SRF (30)(31)(32)(33)(34)(35)(36). Although the G 12/13 -YAP/TAZ signaling axis has been shown to be oncogenic in some contexts (35)(36)(37), no cancer-associated mutation that activates Ga 12 or Ga 13 has been characterized to date. Prompted by the finding that the G 12/13 -YAP/TAZ signaling axis is up-regulated in bladder cancer, here we characterized the effect of hotspot mutations in Ga 13 identified in this cancer type. We found that mutations in the Arg-200 of Ga 13 , a residue required to hydrolyze GTP, lead to activation of YAP/ TAZ-dependent and MRTF-A/B-dependent transcription through a RhoGEF-Rho GTPase cascade and that they promote oncogenic transformation in vitro. This implies that naturally occurring hotspot mutations in Ga subunits of any of the four families of heterotrimeric G-proteins are putative cancer drivers.

G 12/13 pathway up-regulation correlates with increased Yap/ TAZ transcriptional activity in bladder cancer
We mined data from the Cancer Genome Atlas (TCGA) through cBioportal to explore genomic alterations in components of a G 12/13 -YAP/TAZ pathway (Fig. 1A). More specifically, we queried the G-proteins Ga 12 (GNA12) and Ga 13 (GNA13); the RhoGEFs p115-RhoGEF (ARHGEF1), PDZ-Rho-GEF (ARHGEF11), and LARG (ARHGEF12); the Rho GTPases RhoA (RHOA), RhoB (RHOB), and RhoC (RHOC); and the Hippo pathway effectors YAP (YAP1) and TAZ (WWTR1). We found that these genes were altered in a large portion (;40%) of the TCGA bladder cancers (TCGA-BLCA) (Fig. 1B). The alterations appeared to be largely mutually exclusive and trending toward up-regulation. For example, both heterotrimeric Gproteins, two of the three RhoGEFs, and both Hippo effectors displayed amplifications as the dominant feature. For RhoA Figure 1. G 12/13 -YAP/TAZ signaling axis is up-regulated in many bladder cancers. A, diagram of a G 12/13 -YAP/TAZ signaling cascade. B, many bladder cancers show genetic alterations in the G 12/13 -YAP/TAZ signaling cascade that collectively suggest up-regulation. Data mined from the TCGA-BLCA data set is displayed as an oncoprint representation. C, up-regulated expression of the G 12/13 pathway correlates with YAP/TAZ activation in bladder cancers. The correlation between the expression of G 12/13 pathway components and a 24-gene transcriptional signature regulated by YAP/TAZ was assessed by ssGSEA. The solid line represents the linear regression fit of the data, with 95% confidence intervals indicated in gray. D, comparison of the correlation coefficient observed in C (vertical dotted line) with a null distribution generated from bootstrapping correlation coefficients for the G 12/13 pathway and 10,000 randomly selected 24-gene signatures. The statistical significance p value was determined as described under "Experimental procedures." (RHOA) and RhoB (RHOB), the main feature was that they were mutated, and several of these mutations are classified as putative drivers in cBioportal (38). Although not all RhoA/ RhoB mutations have been characterized, some of them have been previously proposed to lead to signaling activation, like Ala-161 mutations in RhoA (39) or the E172K mutation in RhoB (40). Thus, although LARG (ARHGEF12) and RhoC (RHOC) are exceptions to the overall trend, these observations suggest that the G 12/13 -YAP/TAZ pathway might be up-regulated in bladder cancer.
Motivated by these observations, we carried out a bioinformatic analysis of gene expression data to establish a potential correlation between up-regulation of the G 12/13 pathway and YAP/TAZ activation. For this, we turned to a previously characterized 24-gene signature that depends on YAP/TAZ (41) and analyzed its relationship to the expression levels of the rest of the upstream components of the proposed G 12/13 pathway. We used single-sample gene set enrichment analysis (ssGSEA) to quantify relative enrichment of each pathway across over 400 primary tumors in the TCGA-BLCA RNA-Seq data set. We found a strong correlation between the activation scores of the G 12/13 pathway and the activation scores for YAP/TAZ (Fig.  1C). We then tested the observed correlation coefficient against a null distribution of correlations between ssGSEA-quantified activity of the G 12/13 pathway and 10,000 random 24-gene signatures, resulting in a significant p value of 1e24 (Fig. 1D). Taken together, these observations indicate that up-regulation of the G 12/13 pathway in bladder cancer correlates with increased transcriptional output of the downstream effectors YAP/TAZ.

Ga13 Arg-200 mutants induce YAP/TAZ activity via a RhoGEF-Rho GTPase axis
Although overexpression of WT G 12/13 family Ga proteins has been found before to be sufficient to promote transformation (7,8), a recent study also found that the mutation frequency of GNA13 in the TCGA-BLCA data set is statistically higher than background mutation frequency (q = 0.007) (42). Moreover, the distribution of mutations across the sequence of Ga 13 suggested a hotspot at Arg-200 ( Fig. 2A). The presence of an arginine in this position is absolutely conserved across Ga subunits ( Fig. 2A), and its mutation in several other Ga subunits leads to increased activity and favors oncogenic transformation (5,11,13,14). From studies in other Ga proteins, it has been found that this arginine is crucial for GTPase activity and that it cannot be replaced by other amino acids, even if they preserve the positive charge of the side chain like in the case of lysine (5,(43)(44)(45). Thus, we hypothesized that Ga13 Arg-200 mutations identified in bladder cancer would similarly induce the formation of an active G-protein that increases downstream signaling, including YAP/TAZ (Fig. 2B). Because mutation of this arginine to any other residue is expected to have similar consequences (5, 43), we focused our efforts on characterizing Ga 13 R200K and Ga 13 R200G because these are the two mutants most frequently found in bladder cancer. Before assessing the impact of these mutants in cell signaling assays, we validated that they adopted an active conformation by using a well-validated assay that relies on protection from trypsin hydrolysis (Fig. S1) (44, 46). Next, we expressed Ga 13 R200K and Ga 13 R200G in HEK293T cells and assessed activation of YAP/ TAZ using a TEAD reporter assay (Fig. 2C). We compared the effect of expressing these two mutants with that of Ga13 WT as well as with that of Ga13 Q226L, an artificial mutant previously shown to enhance downstream signaling including YAP/TAZ-dependent TEAD transcriptional activity (34,37). Whereas expression of Ga 13 WT led to a modest increase of TEAD activity, expression of Ga 13 R200K and Ga 13 R200G led to a significantly larger increase comparable with that observed in cells expressing the control mutant Ga 13 Q226L (Fig. 2C).
To determine whether the observed increase in TEAD activity by Ga 13 mutants was mediated by YAP/TAZ, we knocked down both proteins simultaneously using a previously validated siRNA sequence (47,48). As expected, depletion of YAP and TAZ led to a large suppression of TEAD activation by Ga 13 R200K, R200G, or Q226L (Fig. 2D). To further map the cascade of events leading to YAP/TAZ activation by Ga 13 mutants, we blocked the pathway that putatively operates in bladder cancer at different levels. First, inhibition of the Rho GTPases RhoA, RhoB, and RhoC by expression of Clostridium botulinum C3 toxin efficiently suppressed TEAD activation by Ga 13 R200K, R200G, or Q226L (Fig. 2E). Then we tested the effect of a fragment of p115-RhoGEF that works as a dominant-negative by preventing the binding of active Ga 13 to its target RhoGEFs that operate upstream of Rho GTPases in the pathway (49). Expression of this dominant-negative construct, consisting of p115-RhoGEF's RGS homology (RH) domain (p115 RH ), but not a control construct, inhibited TEAD activation by Ga 13 R200K, R200G, or Q226L (Fig. 2F). To further validate the specificity of these manipulations, we tested their impact on Ga 13 -mediated activation of another transcriptional output not controlled by YAP/TAZ but still dependent on Rho GTPase activation (i.e. the transcriptional activation of SRF via MRTF-A/B) (Fig. 2B). As expected, bladder cancer-associated mutants Ga 13 R200K and R200G led to robust activation of the SRF reporter, comparable with that of the positive control Ga 13 Q226L, which was suppressed by inhibition of the activation of RhoGEFs or Rho GTPases but not upon YAP/TAZ depletion (Fig. 2, G-I). Taken together, these results demonstrate that Ga 13 hotspot mutations in Arg-200 found in bladder cancer are bona fide activating mutations that lead to induction of YAP/TAZ-dependent transcription via a RhoGEF-Rho GTPase cascade.

Ga 13 Arg-200 mutants induce oncogenic transformation in vitro
Finally, we sought to determine whether the Ga 13 hotspot mutations in Arg-200 described above would be sufficient to promote oncogenic transformation in vitro. For this, we used focus formation assays with NIH3T3 cells. This widely used system is particularly well-suited to analyze the putative oncogenic activity of Ga 13 Arg-200 mutants because it has been used for the vast majority of Ga oncogenic mutants reported to date as a good proxy for tumor growth in mice, including for the oncogenic activity of artificial activating mutations introduced in Ga 13 (7). First, we assessed whether Ga 13 R200K and ACCELERATED COMMUNICATION: GNA13 Arg-200 oncogenic mutants Ga 13 R200G mutants also lead to increased signaling activity in NIH3T3 cells. Surprisingly, we found that whereas Ga 13 R200K and Ga 13 R200G led to robust increases in the MRTF-A/B-dependent SRE.L reporter, they had no significant effect on the activity of the YAP-TAZ-dependent TEAD reporter (Fig. S2). These results confirm that Ga 13 R200K and Ga 13 R200G behave as active G-proteins but that the downstream signaling consequences are cell type-specific. Next, we generated NIH3T3 cell lines stably expressing Ga 13 WT, Ga 13 R200K, and Ga 13 R200G at comparable levels by lentiviral transduction and selection with the appropriate agents (Fig. 3A). Both Ga 13 R200K and Ga 13 R200G induced the formation of numerous foci, whereas Ga 13 WT only had a modest effect (Fig. 3, B and C).

Conclusions
Recent reports have suggested that mutations in Ga 13 are putative oncogene drivers in bladder cancer based on bioinformatics predictions (17,42,50), but no other experimental evidence to support the predictions has been provided. The results presented here provide the missing experimental evidence that supports the idea of Ga 13 hotspot mutations as putative drivers in bladder cancer and suggest that pharmacological blockade of the pathway activated downstream might be a viable therapeutic avenue. Moreover, our findings on Ga 13 mutants establish that naturally occurring hotspot mutations in Ga subunits of any of the four families of heterotrimeric G-proteins (i.e. in G s , G i/o , G q/11, and, now, G 12/13 ) are putative cancer drivers, thereby providing definitive confirmation of a long-held tenet.

Data processing
Data for the oncoprint in Fig. 1B were obtained through cBioportal (38) by querying the term GNA13 on March 23rd, 2020 in the data set "Bladder Cancer (TCGA, Cell 2017)." For the lollipop plot in Fig. 2A, data were obtained from all of the data sets classified as Bladder Urothelial Carcinoma in cBioportal. TCGA-BLCA RNA-Seq count matrix (generated with STAR 2-Pass and HTSeq-Counts) and available metadata were downloaded through the Genomic Data Commons gdc-client (51,52). We performed a variance-stabilizing transformation of the data using the R package DESeq2 (version 1.23.10) followed by a log transformation (53).

Pathway-level correlation analysis
Pathway activity in TCGA-BLCA was measured for G 12/13 and YAP/TAZ signatures-represented by key selected genes (for G 12   ACCELERATED COMMUNICATION: GNA13 Arg-200 oncogenic mutants TGFB2, and RBMS3)-through gene set variation analysis using the ssGSEA method and a Gaussian kernel from the R package GSVA (version 1.34.0) (54). We then measured the Pearson correlation between the activities of each pathway across all primary tumors. We further tested the significance of the observed correlation coefficient by comparing it with a null distribution generated through bootstrapping 10,000 random 24-gene signatures and measuring the correlation of their ssGSEA-quantified activity with the G 12/13 pathway.

Trypsin protection assays
This assay was carried out as described previously (29), except that HEK293T lysates were treated with trypsin for 5 min.

NIH3T3 focus formation
Lentivirus packaging, transduction, and selection (1 mg/ml puromycin) were carried out as described previously (57). All surviving clones were pooled and maintained in the presence of 0.5 mg/ml puromycin. For focus formation assays, NIH3T3 cell lines were seeded (200,000 cells/plate) in 10-cm plates coated with 0.1% gelatin. Medium was replaced at days 5, 7, and 9 after seeding. At day 10, cells were washed with cold PBS, fixed with cold 100% methanol, and stained with crystal violet (0.05% (w/v) in 20% (v/v) methanol). After washing with 20% (v/v) methanol, plates were imaged using a flatbed scanner, and distinct visible colonies (;1.5 mm 2 or larger) with dark blue staining were manually counted in the whole plate.

Data availability
All data are contained in the article except the raw data used for the genomics analysis, which corresponds to the Cancer Genome Atlas data set named TCGA-BLCA and was accessed and/or downloaded directly from cBioPortal or Genomic Data Commons as indicated under "Experimental procedures." Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.