Synthetic Lethal Screens Reveal Cotargeting FAK and MEK as a Multimodal Precision Therapy for GNAQ-Driven Uveal Melanoma

G protein–coupled receptors (GPCR) are the largest family of cell surface proteins with over 800 members (1), and their dysregulation contributes to some of the most prevalent human diseases (2–4). GPCRs represent the largest family of targets for approved drugs. Strikingly, our recent analysis of human cancer genomes revealed that nearly 30% of human cancers present mutations in G proteins and GPCRs (5). In particular, uveal melanoma can be defined as a Gα_q-driven malignancy. Indeed, approximately 93% of uveal melanoma lesions harbor activating mutations in GNAQ or GNA11, encoding for the alpha subunits Gα_q and Gα₁₁ of the heterotrimeric G protein, respectively (6, 7). An additional 4% harbor mutations in the Gα_q-linked receptor CYSLTR2, also acting as a driver oncogene (8).

Uveal melanoma is diagnosed in about 2,500 adults in the United States every year, and is the most common primary cancer of the eye in adults and the second most common melanoma subtype after skin cutaneous melanoma (9). Although the majority of early-stage uveal melanoma lesions can be treated by irradiation or enucleation, approximately 50% of the patients will metastasize, primarily to the liver, within 5 to 10 years after diagnosis (10). Inactivating mutations or copy loss of the BAP1 gene, which is located on chromosome 3p21, are strongly associated with metastasis in patients with uveal melanoma (11), supporting that BAP1 functions as a metastasis suppressor (12). Most patients with metastatic uveal melanoma (mUM) are refractory to current chemotherapies and immune checkpoint blockade (13). Ultimately, the majority of advanced disease patients succumb within a year due to the suboptimal efficacy of these treatments, often combined with severe toxicities, underlying the high unmet medical need for new therapeutic strategies.

Most of the recent clinical research efforts in uveal melanoma have focused on inhibiting the Gα_q classical signaling pathway, PLC_β–PKC–ERK. MEK inhibitors (MEKi) selumetinib and trametinib have been extensively evaluated for mUM. Despite encouraging results in preclinical studies, MEK inhibition with these agents has been shown to have nearly no impact on the overall survival of patients with mUM, as single-agent or when combined with chemotherapy (14–16). Our recent findings uncovered a noncanonical Gα_q signaling pathway, leading to the Rho-dependent activation of the Hippo/YAP pathway, which contributes to aberrant cancer cell growth (17, 18). By further decoding this oncogenic signaling circuitry, we showed that the nonreceptor focal adhesion kinase (FAK) is an integral node of this noncanonical Gα_q pathway (19). Interestingly, FAK overexpression has already been associated with various cancer types including ovarian, head and neck, and breast cancers (20–22). Furthermore, multiple FAK inhibitors (FAKi), defactinib (VS-6063), PF-562271, GSK2256098, and IN10018, have been already tested in clinical trials, showing manageable toxicity profiles (23), and hence can be considered for mUM treatment. However, single-agent targeted therapies often activate compensatory mechanisms resulting in treatment resistance. This prompted us to perform a kinome-wide CRISPR-Cas9 sgRNA screen to identify synthetic lethal gene interactions in the context of FAKi that can be exploited therapeutically (24). Our study demonstrates that dual inhibition of MEK and FAK act synergistically to promote the conversion of cytostatic to cytotoxic inhibition of tumor growth, thereby identifying a new treatment option for mUM.

DMEM, RPMI1640, and antibiotic/antimycotic solution were purchased from Sigma-Aldrich. Turbofect, DMEM/F12 Glutamax, basic FGF (bFGF), EGF, B-27, and N2 supplements were purchased from Thermo Fisher Scientific. FBS (lot 18H165) and penicillin/streptomycin are from Sigma Aldrich, Polybrene from Millipore, and AquaBluer from MultiTarget Pharmaceuticals.

VS-4718, VS-6063, and VS-6766 were provided by Verastem Oncology. All inhibitors used in this study were purchased from Selleck Chemicals. Anhydrous DMSO, carboxymethyl cellulose (CMC), polyethylene glycol 400 (PEG400), and Tween 80 were purchased from Sigma-Aldrich. Inhibitors were dissolved in DMSO to a stock concentration of 10 mmol/L, aliquoted, and stored at −80°C for in vitro experiments. For in vivo treatments, trametinib was prepared at a stock concentration 5 mg/mL in DMSO and freshly diluted in PBS containing 5.2% PEG400 + 5.2% Tween 80, and a dose of 1 mg/kg was administered once daily by intraperitoneal injection. VS-4718 was freshly diluted at 5 mg/mL in 0.5% CMC (C5678, Sigma-Aldrich) + 0.1% Tween 80 (P1754, Sigma Aldrich) in sterile water (B. Braun Medical) and a dose of 10 mL/kg was administered by oral gavage twice a day.

FAK, phospho-FAK, ERK, phospho-ERK, cleaved PARP1, cleaved caspase-3, YAP, Cas9, and GAPDH antibodies were purchased from Cell Signaling Technology. BAP1 and PARP1 were purchased from Santa Cruz Biotechnology. Bromodeoxyuridine (BrdUrd) was purchased from Abcam. All antibodies were diluted in 5% BSA at 1:500 to 1:1,000 before use for immunoblotting and in 3% BSA at 1:100 to 1:400 for IHC. pHAGE PGK-GFP-IRES-LUC-W was from Addgene (#46793).

HEK 293T cells were cultured in High Glucose DMEM and supplemented with 10% FBS and antibiotic/antimycotic solution (100 U penicillin, 0.1 mg/mL streptomycin, and 0.25 μg/mL amphotericin B). Uveal melanoma cell lines were cultured in RPMI1640 media supplemented with 10% FBS (92.1, OMM1.3, OMM1.5, and OMM1.5^Cas9) or 20% FBS (MP41, MP46, MP38, and MM28) and antibiotic/antimycotic solution. Mouse melanoma B2905 cell line was cultured in RPMI1640 medium supplemented with 10% FBS. All cell lines were tested for Mycoplasma and proved to be Mycoplasma free using the MycoAlert PLUS Mycoplasma Detection Kit (Lonza).

Cells were seeded in 96-well ultralow attachment culture dishes (Corning) at 100 cells/well with the indicated concentration of 10 nmol/L trametinib, 1 μmol/L VS-4718, or both. Medium consisted of serum-free DMEM/F12 GlutaMAX supplemented bFGF (20 ng/mL), EGF (20 ng/mL), B-27 (1:50 dilution), and N2 supplement (1:100 dilution). Three weeks after seeding, the number of spheres in each well and their sizes were assessed by bright-field microscopy and quantify using ImageJ (25).

Uveal melanoma cells were harvested at the indicated time points posttreatment with 10 nmol/L trametinib and 1 μmol/L VS-4718. Cell lysates were subjected to SDS/PAGE on 10% acrylamide gels and electroblotted to PVDF membranes. Blocking and primary and secondary antibody incubations of immunoblots were performed in Tris-buffered saline/Tween 20 [10 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, and 0.1% Tween 20] supplemented with 5% (w/v) BSA. The primary antibodies were used according to the manufacturers' instructions. HRP-conjugated donkey anti-rabbit and anti-mouse IgGs were used at a dilution of 1:5,000, and immunoreactive bands were detected using enhanced chemiluminescence. Full blots are shown in Supplementary Fig. S4.

Apoptosis was determined by detecting phosphatidylserine exposure on cell plasma membranes using the fluorescent dye FITC Annexin V (BD Biosciences) according to the manufacturer's protocol. Briefly, at the end of the treatment period, the cells (including floating cells) were harvested and washed twice with cold PBS. After resuspension in 100 μL 1× binding buffer, 5 μL FITC Annexin V was added and the cells were incubated for 15 minutes at room temperature in the dark. Finally, 400 μL 1× binding buffer was added to the cells and flow cytometric analysis was conducted. The experiments were conducted in triplicate.

For IHC, all tissue samples were processed and stained as described previously (26). Slides were scanned using a Zeiss Axioscan Z1 slide scanner equipped with a 20×/0.8 NA objective. All image analyses were performed using the QuPath software (27) to perform cell detection and quantify each marker as a percentage of cells stained.

For lentivirus-Cas9 production, 293T cells were plated in a poly-D-lysine–coated dish and, 16 hours later, transfected with 12 μg pLenti-CAS9-Blast, 8 μg psPAX2, 4 μg pCMV-VSV-G, using 32 μL TurboFect Transfection Reagent, and media was refreshed 6 hours posttransfection. Forty-eight and 72 hours later, the virus-containing media was collected, filtered through a low protein binding filter unit (PVDF, 0.45 μm, Sigma-Aldrich), and stored at 4°C up to 5 days or at −80°C.

For lentivirus-Cas9 infection, OMM 1.5 cells were plated in a 6-well plate and, 16 hours later, transduced using 1 mL virus-containing media + 1 mL complete media + 10 μg/mL polybrene. The plate was centrifuged for 15 minutes at 1,200 rpm. The same process was repeated after 24 hours, and 48 hours later, the cells were selected with blasticidin (10 μg/mL).

Cas9 editing efficiency was measured by quantifying the editing frequency of the safe-harbor locus AAVS1(TGCCTAACAGGAGGTGGGGGTTAGACCCAATATCAGGAGACTAGGAAGGAGGAGGCCTAAGGATGGGGCTTTTCTGTCACCAATCCTGTCCCTAGTGGCCCCACTGTGGGGTGGAGGGG) by next-generation sequencing (NGS) using two separate sgRNA sequences (sgT1 and sgT2). Briefly, cells were transduced with either AAVS1 sgRNA and, after hygromycin-B selection, genomic DNA was isolated using the Qiagen DNeasy Blood and Tissue Kit (cat # 69504). The AAVS1 region was amplified, DNA gel-purified, and barcoded using the NEBNext Multiplex Oligos for Illumina Kit (E7335) for multi-plex sequencing. Analysis of genome editing was performed using CRISPResso (28).

In total, 3,052 unique sgRNAs targeting 763 human kinome genes for 4 guides per target were used for pooled CRISPR screens [Brunello Human Kinome CRISPR Knockout Library (29); Addgene cat #75312]. Lenti-X 293T cells were seeded in 15-cm plates and transfected the next day with pooled CRISPR library DNA (21 μg), PAX2 (14 μg), and VSV-g (7 μg). Viral media was collected 48- and 72 hours posttransfection, pooled and concentrated using an ultracentrifugation protocol, and stored at −80°C in aliquots.

Several aliquots of concentrated viral stock were used to infect separate 15-cm plates containing 4.8 × 10⁶ OMM1.5-Cas9 cells that were seeded the day before. Noninfected plates for OMM 1.5 parental and Cas9 were also plated as antibiotic selection controls. Forty-eight hours posttransduction, cells were treated with 2 μg/mL puromycin for 72 hours. After puromycin selection, complete cell death was achieved in uninfected control plates. The number of puromycin-resistant cells in each infected plate was counted using the Countess Automated Cell Counter to calculate the functional titer of the viral stock. Cells for screening were infected with a virus dose to achieve a MOI of 0.4 to 0.5 with sufficient cell numbers plated to obtain a screening depth of >1,000 cells per sgRNA.

A total of 4.8 × 10⁶ cells were seeded into 15-cm plates a day before CRISPR library transduction. Forty-eight hours posttransduction, cells were subject to puromycin selection (2 μg/mL; 72 hours). Cells were amplified and seeded into 15-cm plates (4 × 10⁶ cells/plate) divided into two treatment arms: 3 replicate plates for either vehicle/DMSO or VS-4718 treatments. A total of 4 × 10⁶ cells from each individual plate were passaged into a new plate containing DMSO or 0.5 μmol/L VS-4718 every 3 to 4 days for a total of 10 days of treatment. A total of 5 × 10⁶ cells were aliquoted from each plate at the beginning and end of the screen and stored at −80°C for sgRNA quantification.

Genomic DNA was isolated using the Qiagen DNeasy Blood and Tissue Kit (cat # 69504) according to the manufacturer's protocol. DNA concentration was measured using a Qubit assay (Thermo Fisher Scientific; cat #Q33226). Amplification of sgRNAs for NGS was performed according to the Broad Institute's recommended protocol. NGS read counts were processed, aligned, and analyzed using PinAPL-Py (30).

sgRNAs were quantified immediately prior to the start of the screen to evaluate CRISPR library representation: sgRNA representation was 99.5% and gene representation was 100%.

sgRNAs were quantified from each replicate plate at the end of the screen and analyzed to identify sgRNAs depleted in VS-4718–treated cells relative to DMSO-treated cells.

Cells were seeded at a density of 5 × 10³ to 1 × 10⁴ cells/well in 96-well white plates. Eight different dilutions of each inhibitor were assayed in technical triplicates for 72 hours in each experiment. Cell viability was measured with the AquaBluer Cell Viability Reagent on a Spark microplate reader (Tecan). Using the GraphPad Prism v8.2.0 software, the half-maximal inhibitor concentration values (GI₅₀) were determined from the curve using the nonlinear log (inhibitor) versus response–variable slope (three parameters) equation. GI₅₀ values were only determined for compounds that inhibited growth by more than 50%.

The Chou–Talalay method (31) was used to determine possible synergistic effects of selected kinase inhibitor combinations. Briefly, cells were seeded at a density of 5 × 10³ to 1 × 10⁴ cells/well in 96-well white plates (CulturePlate; PerkinElmer Inc.). Cells were treated with either single inhibitors or combinations thereof using eight different dilutions of each inhibitor and in technical triplicates. Cell viability was measured, after 72-hour treatment, with the AquaBluer Cell Viability Reagent on a Spark microplate reader (Tecan). Combination index (CI) values showing either synergy (<1) or antagonism (>1) were calculated using the following equation:

CI = (D)₁/(D_x)₁ + (D)₂/(D_x)₂, where D_x equals the concentration of the tested substance used in the single treatment that was required to decrease the cell viability by x% and D equals the concentration of the tested substance 1 in combination with the concentration of the tested substance 2 that when combined decreased the cell number by x%.

The Bliss independence model (32) assumes a stochastic process in which two drugs elicit their effects independently, and the expected combination effect was calculated using the following equation: IAB = IA + IB – IA x IB, where IA and IB are the single-agent inhibition levels at fixed concentrations. If the experimentally measured effect of the drug combination was equal to, higher than, or lower than the expected effect (IAB), the combination was considered to be additive (ΔBliss = 0), synergistic (<0), or antagonistic (>0), respectively.

All animal studies were approved by the Institutional Animal Care and Use Committee of University of California, San Diego (San Diego, CA) with protocol S15195. Female 4- to 6-week-old NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (SCID-NOD) mice were purchased from the UCSD in-house breeding program. Mice were injected subcutaneously in both flanks with either 2 × 10⁶ or 2.5 × 10⁶ 92.1 or OMM1.3 cells, respectively. Mice were monitored twice weekly for tumor development. Tumor growth analysis was assessed as LW²/2, where L and W represent length and width of the tumor. VS-4718 (5 mg/mL) was prepared in 0.5% CMC (Sigma-Aldrich) and 0.1% Tween 80 (Sigma-Aldrich) in sterile water. Trametinib (0.1 mg/mL) was prepared in 5.2% polyethylene glycol 400 (Sigma-Aldrich) and 5.3% Tween 80 (Sigma-Aldrich) in DPBS. Mice were administered 50 mg/kg VS-4718 (Verastem Oncology) twice daily by oral gavage and 1 mg/kg trametinib once daily by intraperitoneal injection; control group was treated with each vehicle. Mice were euthanized at the indicated time points and tumors were isolated for sequencing, histologic, and IHC evaluation. Results of mice experiments were expressed as mean ± SEM of a total of tumors analyzed.

For lentivirus-GFP-Luc production, 293T cells were plated in a poly-D-lysine–coated 15-cm dish and, 16 hours later, transfected with 30 μg pHAGE PGK-GFP-IRES-LUC-W, 3 μg VSV-G, 1.5 μg Tat1b, 1.5 μg Rev1b, and 1.5 μg Gag/Pol using 25.2 μL p3000 buffer and 25.2 μL of Lipofectamine 3000 transfection reagent, and media was refreshed 16 hours posttransfection. Forty-eight and 72 hours later, the virus-containing media was collected, filtered through a low protein binding filter unit (PVDF, 0.45 μm, Sigma-Aldrich), and stored at 4°C up to 5 days or at −80°C.

For lentivirus-Cas9 infection, 92.1 cells were plated in a 6-well plate and, 16 hours later, transduced using 1 mL virus-containing media + 1 mL complete media + 10 μg/mL polybrene. The plate was centrifuged for 15 minutes at 1,200 rpm. GFP expression was validated by fluorescent microscopy.

Mice were injected with 1 × 10⁶ 92.1 GFP-Luc cells in the spleen, followed by removal of the spleen at 2 minutes postinjection. Tumor implantation by bioluminescence was assessed twice weekly by bioluminescence images captured using the In Vivo Imaging System (IVIS) Spectrum (PerkinElmer). To this end, mice received an intraperitoneal injection of 200 mg/kg D-luciferin firefly potassium salt diluted in PBS 15 minutes before imaging (GoldBio). Total bioluminescence was determined upon subtracting the background from the region of interest. Vehicle, trametinib, VS-4718, or trametinib/VS-4718 were administered, starting 7 days postsurgery, with the abovementioned dosing.

GraphPad Prism version 8 for Windows (GraphPad Software) was used to perform data analyses, variation estimation, and validation of test assumptions. Statistical analysis was performed using either a paired Student t test or one-way ANOVA.

As the kinome has been the target of most drug discovery efforts, the identification of kinases whose activity are essential for uveal melanoma survival in the context of FAK inhibition may facilitate the discovery of new cotargeting strategies. To perform an efficient loss-of-function screen, we used a two-vector CRISPR system, expressing a Cas9 transgene in representative uveal melanoma cells (OMM1.5, originally derived from metastasis, hereby called OMM1.5^Cas9), in which gene editing efficiency was validated (Fig. 1; Supplementary Fig. S1A and S1B). OMM1.5^Cas9 cells were infected with the Human Kinome Brunello pooled sgRNAs library (29), targeting 763 kinase genes and containing 3,052 unique sgRNAs along with 100 nontargeting controls, and subjected to puromycin selection. Surviving cells were treated with either selective FAKi (VS-4718) or vehicle for 10 days, followed by the collection of cellular DNA for gRNA analysis by NGS (Fig. 1A). Depleted sgRNAs (dropouts) suggest that inhibition of their gene targets could sensitize cells to FAKi treatment. Our analysis of synthetic lethal interactors of FAKi revealed a significant enrichment of Gα_q-PLC and MAPK components of the classical Gα_q signaling cascade (Fig. 1B; Supplementary Fig. S1C). Using the pan-PKC inhibitor, Go-6983 (33), we found that inhibition of the canonical PLCβ–PKC–ERK pathway decreased uveal melanoma cell viability and that this inhibitory effect was further enhanced by combining Go-6983 with VS-4718 (Fig. 1C). However, PKC inhibition with Go-6983 was less effective in reducing active ERK (pERK) than the MEKi trametinib (Fig. 1D). This is aligned with prior studies supporting only partial pERK reduction after prolonged PKC inhibition (34), which may explain its very modest clinical activity in mUM (35). Thus, we focused on direct MEK inhibitors for further studies. Interestingly, inhibition of FAK by VS-4718 led to a gradual increase in pERK levels (Fig. 1D and E) and uveal melanoma cells stably expressing the constitutive active MEK mutant (MEK-S218/222D, MEK-DD) were more resistant to VS-4718 (Fig. 1F), suggesting a possible compensatory mechanism that may contribute to drug resistance.

Cotargeting of MEK and FAK inhibits both canonical and noncanonical Gα_q signaling pathways and may provide a suitable drug combination for clinical application in mUM. We used the effect-based ΔBliss model and the dose-effect–based Chou–Talalay CI to assess synergistic, additive, or antagonistic drug interaction in a given combination. We confirmed that trametinib (MEKi) and VS-4718 (FAKi) decrease uveal melanoma cell growth as single agents. CI and ΔBliss scores both converged to demonstrate a synergistic interaction between trametinib and VS-4718 (Fig. 2A). Synergistic antiproliferative effects were observed using multiple clinically relevant MEKi (trametinib, cobimetinib, and selumetinib) as well as the second-generation RAF/MEK inhibitor VS-6766, combined with two different FAKi (VS-4718 and the clinically relevant defactinib). All combinations showed remarkable synergistic activity at relevant doses supporting a general drug–drug class pharmacodynamics interaction (Fig. 2B). Using a panel of GNAQ-mutant uveal melanoma cells lines with different expression levels of the metastasis suppressor protein BAP1 (12), we confirmed similar synergistic profiles in BAP1 wild-type or null cells (Fig. 2C and D), suggesting that this combination may also be active in mUM.

Several FAKi inhibitors have been evaluated in the clinic and demonstrated primarily a cytostatic effect as single agents (23). Given the high synergistic activity of MEKi and FAKi combination in vitro, we next evaluated the ability of trametinib and VS-4718 to induce cell death. As shown in Fig. 3A, FACS analysis suggests that trametinib induces an increase in the fraction of apoptotic cells (Annexin V⁺), while VS-4718 has no apoptotic activity as a single agent. Nonetheless, when both drugs were combined, at a ratio of 1:100 based on our isobologram analysis (Supplementary Fig. S2), the apoptotic response was significantly increased.

One of the most common signaling cascades involved in apoptosis is the activation of a highly specialized family of cysteinyl-aspartate proteases (caspases; ref. 36). Caspase-mediated cell death is achieved through the cleavage of multiple key proteins that are essential to cell survival. PARP-1 is one of the substrates of caspases and a well-established marker for apoptosis (37). Aligned with our prior results, cleaved PARP levels were strongly increased by the combination of FAKi and MEKi (Fig. 3B), while single agents were less active. We then evaluated the ability of VS-4718 and trametinib to induce apoptosis in BAP1-null mUM cells. FAKi alone showed a minimal apoptotic response, and the combination of VS-4718 and trametinib resulted in increased cleaved PARP compared with trametinib alone (Fig. 3B).

Cancer-initiating cells are believed to play a central role in drug resistance and metastasis (38). A typical feature of these cells is their ability to form 3D tumorspheres when cultured in suspension in a stem cell medium. Interestingly, we found that the metastatic-derived OMM1.3 cells, but not 92.1 cells originating from a primary tumor, were able to form melanospheres. Using this model, we found that both VS-4718 and trametinib significantly reduced melanosphere formation as single agents when compared with vehicle (Fig. 3C), and this effect was further enhanced when both drugs were combined. Our results demonstrate that MEKi/FAKi combination acts by inhibiting cancer cell proliferation, including cancer-initiating cells, and enhancing apoptotic cell death. This suggests that this multimodal targeted therapy could be effective on primary and mUM.

We used uveal melanoma xenograft models to evaluate the anticancer activity of the MEKi/FAKi combination in vivo. Tumor-bearing mice were randomly divided into four groups: control, trametinib, VS-4718, and trametinib/VS-4718. In this regard, whereas VS-4718 has better pharmacokinetics than defactinib in mice, the latter is better able to inhibit tumor FAK activity with tolerability in patient with cancer (39). Therefore, VS-4718 is used as a surrogate FAKi in murine preclinical models. As shown in Fig. 4A and B, in the 92.1 xenograft model, although trametinib and VS-4718 effectively induced tumor stasis as single agents, only the combination of trametinib with VS-4718 was able to induce tumor regression. No significant difference in body weight was observed between the control and any of the treated groups (Fig. 4C), suggesting that all treatments were well tolerated by the mice. These results were further confirmed using another uveal melanoma cellular system harboring GNAQ mutations, OMM1.3, albeit in this case MEKi alone was quite potent (see Supplementary Fig. S3). We extended these studies in allograft experiments using the recently developed GNAQ-driven B2905 syngeneic mouse melanoma model (40). These cells are highly sensitive to the Gαq inhibitor FR900359, as we reported in a large panel uveal melanoma cells (41), and to the combined inhibition of MEK and FAK (Supplementary Fig. S4). These cells exhibit limited responses to FAKi in vivo, likely reflecting the complexity of genetic alterations of these cells (40). MEKi limited tumor growth, but prolonged inhibition resulted in acquisition of resistance and rapid tumor regrowth, thereby compromising animal survival. Remarkably, the combination therapy was quite effective in promoting tumor regression, and no resistance was observed for the prolonged duration of the mouse experiments (>40 days).

We next evaluated MEK/ERK and FAK/YAP pathway activities in the uveal melanoma xenograft tumor specimens by IHC (Fig. 4D and E). As expected, pERK1/2 was strongly decreased only in the trametinib and trametinib/VS-4718–treated groups. We monitored YAP nuclear exclusion as a readout for YAP inactivation upon FAKi treatment (18) and observed decreased nuclear YAP in the VS-4718 and trametinib/VS-4718 treatment groups exclusively. The percentage of proliferating BrdUrd-positive cells was significantly decreased in all treated groups, with the VS-4718 and trametinib/VS-4718 treatment groups showing the most significant inhibition. Using cleaved caspase-3 (cl-Casp3) as a marker for apoptosis, we detected a significant increase of cl-Casp3 in the trametinib/VS-4718 group compared with the single-agent groups, suggesting that the MEKi/FAKi combination induces a switch from cytostatic to cytotoxic activity, consistent with the increased cl-PARP levels observed in vitro.

To assess the efficacy of the MEKi/FAKi combination treatment on a murine system that better represents the advanced disease, we developed a preclinical liver metastasis model. Luciferase/GFP–expressing 92.1 cells (92.1-Luc) were injected into the spleen and allowed hematogenous dissemination followed by complete splenic resection (Fig. 5A). Mice were sacrificed 8 weeks postinjection and necropsies showed macroscopic hepatic metastases with no sign of dissemination to any other organs (Fig. 5B), confirming the strong liver tropism of uveal melanoma cells. Remarkably, when tumor-bearing mice were treated with trametinib, VS-4718, or trametinib/VS-4718, the metastatic burden was reduced by MEKi and FAKi treatment and their combination (Fig. 5C and D). FAKi alone was cytostatic, whereas MEKi and MEKi/FAKi combination induced tumor regression. However, detailed analysis showed that the residual disease in the trametinib alone–treated group represented 24.2% ± 0.6% of the initial tumor compared with 7.5% ± 0.2% in the combination-treated group (P = 0.0002), achieving none versus 40% complete responses, respectively (Fig. 5E ). Resistance to trametinib in patients has led to the failure of this agent in uveal melanoma clinical trials. To further investigate resistance to trametinib as a single agent or as part of a combination with FAKi in our metastatic model, we adjusted the dose of trametinib to reach maximum plasma levels similar to those achieved in humans at the recommended therapeutic dose, 2 mg/day, which is approximately 22 ng/mL upon repeated dosing (42). Mice treated with 0.1 mg/kg trametinib, achieving approximately 20 ng/mL (42), become resistant to treatment with a progressive increase of metastatic burden, whereas tumor burden remains barely undetectable in mice treated with the combination. Taken together, these findings indicate that treatment with trametinib combined with VS-4718 inhibited uveal melanoma tumor growth in subcutaneous and liver metastasis models.

The limited responses of uveal melanoma to immunotherapies (40) and the lack of an FDA-approved standard of care for patients with mUM poses an urgent unmet medical need that necessitates the development of new targeted treatment options. Most clinical studies in uveal melanoma have focused on canonical kinases activated downstream from G_q/G₁₁ with MEK as the major therapeutic target (43). MAPK/ERK pathway inhibitors have been proven to be effective treatments in various cancer types, but their effectiveness is often short-lived, with resistance developing often after the start of treatment (44–47). Uveal melanoma was no exception and despite encouraging results in an initial phase II trial, with the MEK inhibitor selumetinib (AZD6244) showing an improved progression-free survival (PFS) compared with dacarbazine or temozolomide (15), the subsequent phase III double-blind trial failed to show improvement in PFS and overall survival with selumetinib + dacarbazine compared with dacarbazine alone (16). Here, we found that adaptive activation of MEK–ERK represents a compensatory resistance mechanism to FAK inhibition, and that in turn concomitant targeting FAK and MEK–ERK converts the cytostatic effects of FAKi into cytotoxic, resulting in uveal melanoma cell death. These findings suggest that dual inhibition of MEK and FAK may represent a promising therapeutic option for patients with advanced and metastatic uveal melanoma.

Our previous studies revealed that Gα_q triggers the activation of the FAK/Hippo–YAP pathway, which represents a major driver in uveal melanoma progression (18, 19). Given that, we aimed at identifying synthetic lethal interactors that could potentiate FAKi efficiency and simultaneously reduce the risk of developing drug resistance (24), which is often observed in single-drug therapy. Our kinome-wide CRISPR/Cas9 screen unveiled potent synthetic lethality between FAKi and the MEK/ERK pathway, uncovering the therapeutic potential of a cotargeting strategy. This idea is reinforced by a recent study demonstrating that treatment of uveal melanoma cells with the MEKi trametinib increased YAP activity and cell proliferation (48), and our findings that FAK/Hippo–YAP pathway inhibition leads to a gradual increase of pERK. These results suggest that both Gα_q-activated pathways, FAK/Hippo–YAP and MEK/ERK, can be part of compensatory feedback processes, and that horizontal inhibition of FAK and MEK might be necessary to achieve tumor regression, and prevent treatment resistance and tumor relapse (Fig. 6). In addition, concomitant inhibition of FAK and MEK triggers the activation of apoptotic cell death programs, whose full elucidation may facilitate the identification of mechanistic biomarkers predicting therapeutic response. Of interest, most uveal melanoma and mUM models used in our studies are driven by the GNAQ oncogene. However, based on its high similarity with the GNA11 oncogene, our findings are also expected to be applicable to GNA11-mutant uveal melanoma lesions (6, 7), which should be nonetheless tested experimentally.

Several potent FAKis have been tested in the clinic in multiple cancer types, showing target inhibition and manageable toxicity profiles. Indeed, based on our findings, a phase I clinical trial using IN10018 as a potent FAKi has been recently initiated in mUM (NCT04109456). The use of FAKis as single agents may act primarily as cytostatic, suggesting the possibility of identifying suitable combination strategies to reach a greater clinical efficacy. Our study supports the clinical potential of cotargeting FAK and its sensitizing pathway, MEK/ERK, as a precision therapy approach in GNAQ-driven uveal melanoma, achieving tumor regression. In this regard, the FAKi defactinib has been recently combined with a RAF/MEKi (VS-6766) in multiple cancer types (NCT03875820). Through the use of an intermittent dosing schedule, the defactinib/VS-6766 combination has shown a manageable safety profile with initial clinical activity in early clinical trial results to date (49). Altogether, we believe our study presents a strong potential for clinical translation as a multimodal signal transduction–based precision therapy for mUM.

J.S. Paradis reports a patent for 16/824639 pending. M. Acosta reports a patent for 16/824639 pending. A. Kishore reports grants from NIH during the conduct of the study, as well as a patent for US20200323863A1 pending. N. Arang reports a patent for US20200323863A1 pending to UCSD. P. Mali is a scientific co-founder of Shape Therapeutics, Boundless Biosciences, Seven Therapeutics, Navega Therapeutics, and Engine Biosciences, which have no commercial interests related to this study. The terms of these arrangements have been reviewed and approved by the University of California, San Diego in accordance with its conflict of interest policies. J.A. Pachter is an employee and shareholder of Verastem Oncology. A.E. Aplin reports ownership interest in patent number 9880150 and a commercial research grant from Pfizer Inc. (2013–2017). J.S. Gutkind reports grants from NIH/NCI and grants from NIH/NCI during the conduct of the study, as well as a patent for 16/824639 pending, and is a consulting member of the scientific advisory board of Oncoceutics, Vividion Therapeutics, and Domain Therapeutics. No disclosures were reported by the other authors.

J.S. Paradis: Conceptualization, formal analysis, funding acquisition, validation, investigation, methodology, writing–original draft. M. Acosta: Formal analysis, validation, investigation, writing–original draft. R. Saddawi-Konefka: Conceptualization, investigation, writing–review and editing. A. Kishore: Conceptualization, formal analysis, writing–review and editing. F. Gomes: Investigation, writing–review and editing. N. Arang: Investigation, writing–review and editing. M. Tiago: Investigation, writing–review and editing. S. Coma: Conceptualization, methodology, writing–review and editing. S. Lubrano: Investigation, writing–review and editing. X. Wu: Conceptualization, writing–review and editing. K. Ford: Conceptualization, writing–review and editing. C.-P. Day: Resources. G. Merlino: Resources. P. Mali: Conceptualization, writing–review and editing. J.A. Pachter: Conceptualization, resources, writing–review and editing. T. Sato: Writing–review and editing. A.E. Aplin: Writing–review and editing. J.S. Gutkind: Conceptualization, resources, supervision, validation, methodology, writing–original draft, project administration, writing–review and editing.

pHAGE PGK-GFP-IRES-LUC-W was a generous gift from Dr. Schoenberger (La Jolla Institute for Immunology). We thank Drs. Alfredo Molinolo for insightful discussion and Helen T. Michael for characterization of mouse melanoma cells. We thank the La Jolla Institute Microscopy and Histology Core Facility for their support with the IHC data presented in this article. We thank the Moores Cancer Center Microscopy Core for the in vivo imaging support. The cartoon was created with BioRender.com. This work was supported, in whole or in part, by NIH Grant R33CA225291 and U54CA209891 to JSG and R01CA196278 and R01CA182635 to AEA. J.S. Paradis was supported by a fellowship from the Canadian Institute for Health Research. K. Ford was supported by the National Science Foundation Graduate Research Fellowship Program (DGE-1650112). S. Lubrano was supported by funding from AIRC and from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement no. 800924. N. Arang was supported by the National Science Foundation Graduate Research Fellowship Program (DGE-1650112). Verastem Oncology provided defactinib, VS-4718, and VS-6766.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.