Chloroquine Sensitizes GNAQ/11-mutated Melanoma to MEK1/2 Inhibition

Mutational activation of GNAQ or GNA11 (GNAQ/11) occurs in a mutually exclusive manner and comprises less than 2% of driver mutations in all melanoma cases (1). However, GNAQ/11 mutations are uniquely enriched in more than 90% of uveal melanoma, a rare subtype of melanoma that arises from melanocytes of the uveal tract of the eye (2–4). Approximately half of patients with uveal melanoma develop metastatic disease, with an unusually high propensity for lethal liver metastases (5, 6). Unlike patients with metastatic cutaneous melanoma, for which pathway-targeted therapy or immunotherapy have improved survival, there are currently no effective systemic therapy options for patients with metastatic uveal melanoma (7).

GNAQ and GNA11 are homologous genes encoding Gα subunits of heterotrimeric G proteins that are essential for G protein–coupled receptor signaling. The most common mutations in GNAQ/11 lead to amino acid substitutions at glutamine 209 (Q209P/L) that disable the protein's ability to hydrolyze GTP. The subsequent accumulation of active, GTP-bound GNAQ/11 results in constitutive activation of multiple downstream oncogenic signaling pathways, including the RAS-regulated PLCβ>PKC>RAS>RAF>MEK1/2>ERK1/2 MAPK pathway and the TRIO>RHO/RAC-regulated Yes-associated protein (YAP) pathway (8–11). Although MEK1/2 inhibitors, such as binimetinib, cobimetinib, selumetinib, or trametinib are FDA approved, clinical trials of MEK1/2-targeted agents have failed to demonstrate any survival benefit for patients with metastatic uveal melanoma, either alone or in combination with chemotherapy (6, 12, 13).

Recent research on RAS- or BRAF-driven cancers has demonstrated that blockade of RAF>MEK>ERK signaling results in an elevation of autophagy that protects cancer cells from the cytotoxic effects of pathway-targeted inhibition (14–18). Autophagy is the cell's degradative and recycling pathway in which organelles and macromolecules can be recycled for other cellular processes (19–21). Therefore, combined targeting of RAS>RAF>MEK>ERK signaling plus autophagy may be a rational combination strategy for multiple cancer types and has led to several ongoing clinical trials (14–16, 22).

Here, we show that inhibition of GNAQ/11 or of RAF>MEK>ERK signaling can be combined with the antimalarial drug and lysosome inhibitor, chloroquine or its derivative, hydroxychloroquine (HCQ), to enhance cell death and inhibit tumor growth in preclinical models. In addition, combined MEK1/2 inhibition plus chloroquine inhibited YAP signaling, which appears to be important for treatment efficacy. Thus, we conclude that three critical survival nodes: RAF>MEK>ERK signaling; YAP activity; and lysosome function must be targeted for successful treatment of GNAQ/11-mutant melanoma.

Cells were seeded at a density of 7,000–10,000 cells per well in a 96-well plate. Twenty-four hours after plating, media was replaced with 20% [vol for vol (v/v)] full media diluted in 80% Earle balanced salt solution (EBSS) containing DMSO or drug treatments and placed in the IncuCyte Zoom Live-Cell Analysis System (Sartorius). To measure cell death, cells were treated with IncuCyte Cytotox Red Reagent (Sartorius) as per the manufacturer's instructions and imaged every 2 hours for up to 96 hours. Phase and fluorescence confluence were determined using IncuCyte Zoom software. Relative cell death was determined by measuring the AUC of fluorescence confluence (IncuCyte Cytotox Red Reagent) normalized to phase confluence plotted over time.

Cells at 50%–75% confluency were transfected using Lipofectamine 3000 Transfection Reagent according to manufacturer's instructions. Fresh media was replaced at 24 hours posttransfection and drug treatments were added. Cell lysates were analyzed by immunoblotting to confirm successful transfection.

HEK293T cells were seeded 24 hours prior to transfection in a 10 cm dish in DMEM with 1% penicillin/streptomycin and 10%(v/v) FBS. Lentiviral vector DNA (3 μg), psPAX2 (3 μg), and CMV-VSVG (1.5 μg) were combined in 600 μL of sterile PBS with 18 μL FugeneHD (Promega). After a 20-minute incubation at room temperature, the mixture was added dropwise to HEK293T cells. Fresh media was added after 24 hours. After 48 hours, media was collected and filtered through 0.45 μmol/L filters and added directly to target cell lines or snap frozen in dry ice/ethanol to be stored at −80°C. Cell lines were transduced with virus-containing media for 8–16 hours. After 48–72 hours, cells were selected using appropriate fluorescent or antibiotic-resistant selection markers.

Cell lines were transduced with pUltra-Auto encoding a mCherry-eGFP-LC3 fusion protein, thereby resulting in autophagy reporter cell lines (23). Cells were seeded at a density of 50,000 cells per 12-well plate. After 24 hours, media was replaced with 20%(v/v) full media and 80%(v/v) EBSS with various drug treatments. After 48 hours of drug exposure, cells were harvested using Trypsin-EDTA (0.25%), washed once with PBS, and resuspended in 200–300 μL of full media and 1 μg/mL DAPI for dead cell exclusion. mCherry and enhanced green fluorescent protein (eGFP) intensity was assessed by flow cytometry (BD LSRFortessa X-20, BD Biosciences). mCherry:EGFP ratio was calculated using FlowJo (Version 10.4.0) with all samples normalized to the DMSO-treated drug vehicle control.

Cells treated with various agents were washed twice with ice-cold PBS and harvested into 1 mL of ice-cold PBS, placed into a 1.5-mL microcentrifuge tube and pelleted by centrifugation for 10 seconds at 10,000 × g. After aspirating PBS, cells were resuspended in approximately 3× pellet volume of RIPA buffer containing Halt Protease and Phosphatase Inhibitor cocktail (Thermo Fisher Scientific) and incubated on ice for 10–20 minutes. Cellular debris was then pelleted at 14,000 × g for 10 minutes at 4°C and protein lysate was collected and stored at −20°C. Protein concentration was measured using BCA Protein Assay (Thermo Fisher Scientific). A total of 15–100 μg of protein was loaded into NuPAGE 4%–12% Bis-Tris protein gels (Invitrogen), electrophoresed at 100–200 volts, and transferred onto a nitrocellulose membrane using the iBlot Dry Blotting System (Thermo Fisher Scientific). Membranes were blocked in Odyssey Blocking Buffer in TBS (LI-COR) at room temperature for 1 hour and then incubated with primary antibody diluted in Odyssey Blocking Buffer in TBS overnight at 4°C. Immunoblot procedures, as previously described by Silva and McMahon (24), were followed using IRDye 800CW and 680RD secondary antibodies of corresponding species (LI-COR). Membranes were imaged using Odyssey CLx imaging system (LI-COR). For YAP immunoblotting, cells were lysed using Pierce IP Lysis Buffer (Thermo Fisher Scientific) and used standard Western blot procedures, including imaging using Immobilon ECL HRP substrate (Millipore) and X-ray film.

To assess cell density and synergy of drugs in vitro, 10,000 cells were seeded in black/white Nunc Microwell 96-well optical bottom plates. After 24 hours, cells were treated with various drug treatments in 20%(v/v) full media and 80%(v/v) EBSS. After 48–72 hours, cells were washed twice with PBS. The ATP-lite one-step luminescence assay system (Perkin Elmer) was used following manufacturer's instructions. Luminescence was measured using the Synergy HTX Multi-Mode Reader (BioTek). Synergy was determined using the Loewe Model (ComBenefit software).

A pool of four siRNAs against human YAP (LQ-012200-00-0005) and nontargeting control (D-001810-01-20) were purchased from Horizon. Cells were seeded at a density of 10,000 cells per well of a 96-well plate. After 24 hours, target genes were knocked down using the DharmaFECT siRNA transfection protocol (Horizon) in the absence or presence of various drug treatments.

NRG (NOD/RAG1/2^−/−IL2Rγ^−/−) or NOD/SCID mice were obtained from the Preclinical Research Resource at the University of Utah. C57BL/6-Tyr^c-BrdTg(Gnrhr-luc/EGFP, aka Glowing Head) mice were a generous gift from Chi-Ping Day and Glenn Merlino (NCI). All animal experiments were approved by the University of Utah Institutional Animal Care and Use Committee. Animals were bred and maintained in specific-pathogen-free housing at the Huntsman Cancer Institute, University of Utah (Salt Lake City, UT) on a 12-hour light/dark cycle at 22°C. Food and water were available ad libitum.

Following the protocol described by Kageyama and colleagues, the abdomen of mice was shaved 1 day prior to surgery (25). After induction of general anesthesia with 2%(v/v) isoflurane, 0.1 mg/kg buprenorphine was injected subcutaneously in the posterior neck, and 50-μL 0.5% lidocaine was injected subcutaneously into the surgical incision site. After the surgical site was cleaned with betadine or chlorhexidine and alcohol, a 1-cm horizontal incision was made in the left upper abdominal quadrant, exposing the liver using aseptic surgical technique. The left lower liver lobe was gently lifted outside of the abdominal cavity and placed onto sterile gauze. A total of 10⁶ luciferase-expressing OMM2.5 or OMM1 melanoma cells were resuspended in 20-μL 50%(v/v) RPMI and 50%(v/v) Matrigel (Corning) and injected directly into the liver lobe using a 27-gauge needle. Bleeding was cauterized prior to placing the liver lobe back into the abdominal cavity. The surgical incision site was closed using 9-mm wound clips and removed 7–10 days postsurgery.

The back of recipient mice was shaved approximately 1 day prior to injection. A total of 2 × 10⁶ Melan-a/GNAQ^Q209L or Melan-a/GNA11^Q209L cells were suspended in 100-μL 50%(v/v) RPMI and 50%(v/v) of Matrigel and injected subcutaneously into the mouse flank using a 27-gauge needle.

Hepatic tumors were measured weekly using bioluminescence imaging (IVIS Spectrum, Perkin Elmer). Mice were anesthetized and injected intraperitoneally with 165 mg/kg d-luciferin potassium salt (GoldBio) for 10 minutes prior to imaging. Tumor burden was measured by luminoscore as described previously (26). Subcutaneous tumors were measured twice weekly using digital calipers. Tumor size was calculated using (width² × length)/2 (27).

Drug treatment was initiated once tumors reached at least 1 × 10⁸ photons/second for the hepatic colonization model or at least 200 mm³ for the subcutaneous injection model. Drug vehicle contained 0.5%(w/v) hydroxypropyl methylcellulose and 0.5%(v/v) Tween-20 in H₂O. Mice were dosed daily by oral gavage using the following drug concentrations: 1 mg/kg trametinib, 40 mg/kg HCQ, or 100 mg/kg temozolomide. Treatment was discontinued, and mice were euthanized once tumors reached >1 × 10¹¹ photons/second or 1 cm³.

Nuclear and cytosolic fractions were prepared with NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific) according to manufacturer's instructions. Scanning densitometry was used for immunoblot quantification of total YAP, fractionated YAP, and the loading controls β-actin, lamin A/C, and β-tubulin.

OMM2.5 cells were cultured in a 10-cm dish to approximately 50% confluence and treated with indicated drugs for 48 hours. RNA was harvested using RNeasy Mini Kit (74106, Qiagen) according to the manufacturer's protocol. cDNA was generated following the protocol of iScript Reverse Transcription Supermix for qRT-PCR (1708840, Bio-Rad). Quantitative PCR was performed using iTaq Universal SYBR Green Supermix (172-5120, Bio-Rad) in the CFX96 Connect instrument (Bio-Rad) at an initial cycle of 95°C for 30 seconds, followed by 30 cycles of 95°C for 5 seconds and 60°C for 30 seconds.

YAP-mediated transcriptional activity in uveal melanoma cells was measured by a synthetic TEAD luciferase reporter (28). Luciferase assays were performed following the transfection of 8xGTIIC-luciferase reporter plasmid (for YAP transcriptional activity) or Renilla luciferase (control plasmid for transfection efficiency). Both luciferases were detected with the Dual-Luciferase Reporter Assay System (Promega) according to manufacturer's instructions and using the Synergy HTX Multi-Mode Reader (BioTek).

Statistical tests were performed using Graphpad Prism, Version 8. Distribution normality was measured by D'Agostino-Pearson omnibus normality test or Shapiro–Wilk normality test. Statistical significance for AFR and qRT-PCR experiments were calculated using one-way ANOVA with Dunnett multiple comparison test. Statistical significance for relative cell death and luciferase reporter experiments were calculated using one-way ANOVA with Tukey or Dunnet T3 multiple comparison test. Equal variation between groups were confirmed using Brown–Forsythe test. Statistical significance for cell proliferation assays, tumor measurements, and body weight was calculated using repeated measures two-way ANOVA analysis and adjusted P values using the Holm–Sidak method. The log-rank test was used for survival analyses with adjusted P values using the Holm–Sidak method. Protein quantifications were log transformed, and statistical significance was calculated using repeated measures one-way ANOVA with Tukey multiple comparison test.

Inhibition of Gα signaling with the plant (Ardisia crenata)-derived depsipetide, FR900359 (FR), has recently shown effectiveness in inhibiting growth of GNAQ/11-mutated melanoma cell lines by preventing GDP-GTP nucleotide exchange on the mutationally activated oncoproteins (29–31). To determine whether inhibition of Gα signaling induces autophagy, we measured the expression of p62^SQSTM1, a protein that is degraded during autophagy and utilized an mCherry-EGFP-LC3 fluorescent autophagy reporter, as described previously (15, 23, 32). Treatment of the GNAQ-mutated human metastatic uveal melanoma cell line OMM2.5, with FR (25 nmol/L) or trametinib (100 nmol/L) led to inhibition of RAF>MEK>ERK signaling (Fig. 1A) and a significant induction of autophagy that was detected in the cells treated with as little as 6.25 nmol/L FR (Fig. 1B). In contrast, treatment of the BRAF-mutated melanoma cell line, WM793, with FR up to 100 nmol/L failed to induce autophagy, whereas treatment of these cells with trametinib (100 nmol/L) significantly induced autophagy (Fig. 1C). The effects on autophagic flux assessed by flow cytometry were supported by immunoblot analysis of p62^SQSTM1 expression in drug-treated OMM2.5 cells where both FR and trametinib led to decreased p62^SQSTM1 (Fig. 1A). These results demonstrate the specificity of FR for inducing autophagy in a GNAQ-mutated uveal melanoma cell line and further suggests that the induction of autophagy in these cells is due, at least in part, to inhibition of RAF>MEK>ERK signaling.

To determine whether the induction of autophagy might serve as a cytoprotective mechanism, OMM2.5 cells were treated with FR either alone or in combination with autophagy or lysosome inhibition. Genetic inhibition of autophagy in OMM2.5 cells was achieved using tetracycline-inducible expression of a dominant-negative (C74>A) form of the key autophagy cysteine protease ATG4B/Autophagin 1 (ATG4B^DN). As expected, treatment of OMM2.5/Tet-ATG4B^DN cells with doxycycline led to increased ATG4B^DN expression and an accumulation of both p62^SQSTM1 and LC3 expression (Fig. 1D). Cell death was measured using fluorescence live-cell imaging every 2 hours for up to 48 hours and quantified over this time period using area under the curve. Treatment of OMM2.5/Tet-ATG4B^DN cells with FR in the presence of ATG4B^DN yielded significantly higher cell death compared with either condition alone (Fig. 1E). Moreover, inhibition of lysosome function, and thereby the end-stage of autophagy using either bafilomycin A1 (Fig. 1F and G) or chloroquine (Fig. 1F and H) also enhanced cell death when combined with FR in OMM2.5 cells, as well as MP41 cells, a human GNA11-mutated primary uveal melanoma cell (Supplementary Fig. S1A and S1B). These findings suggest that autophagy serves as an adaptive mechanism to protect cell viability in the face of inhibition of Gα signaling with FR.

Because of the current challenges of FR isolation or synthesis, in vivo studies with this agent are limited, and it is currently not a clinically feasible drug candidate. To evaluate pathways downstream of mutationally activated GNAQ/11 that could be pharmacologically targeted, mouse Melan-a melanocytes, oncogenically transformed with GNAQ^Q209L or GNA11^Q209L, were compared with nontransformed cells (33). Consistent with prior studies, expression of activated GNAQ/11 led to increased RAF>MEK>ERK signaling (pERK1/2; Fig. 2A) and YAP through increased YAP-mediated transcription and downstream target, CTGF, expression compared with nonmutant cells (Fig. 2B and C; refs. 11, 33). Inhibition of RAF>MEK>ERK signaling with MEK1/2 inhibitors (trametinib or binimetinib) or ERK1/2 inhibition (SCH772984), but not inhibition of YAP with the benzoporphyrin derivative, verteporfin, induced autophagy (Fig. 2D and E). Furthermore, immunofluorescence imaging of trametinib-treated OMM2.5 cells revealed the presence of numerous p62^SQSTM1-positive autophagic vesicles that were larger than those observed in DMSO-treated cells, demonstrating increased autophagy activity (Supplementary Fig. S2A–S2D). These findings of MAPK pathway inhibition-specific induction of autophagy are consistent with our prior study of RAS-driven cancers (15); thus, we next tested whether MEK1/2 inhibition plus 4-aminoquinolines such as chloroquine or HCQ might be an effective treatment strategy, because these are FDA-approved agents that could be rapidly deployed in the clinic.

Treatment of uveal melanoma cell lines, MP41 (GNA11^Q209L), MEL270 (GNAQ^Q209P), or OMM2.5 (GNAQ^Q209P) with increasing concentrations of MEK1/2 inhibitor inhibited RAF>MEK>ERK signaling in a dose-dependent manner (Supplementary Fig. S3A–S3C). To determine whether GNAQ/11-mutated uveal melanoma cell lines were sensitive to the antiproliferative effects of MEK1/2 inhibition, we treated OMM2.5, MEL270, or MP41 with trametinib or binimetinib (Supplementary Fig. S3D–S3F). Compared with BRAF^V600E-driven WM793 cells, which are highly sensitive to MEK1/2 inhibition, all GNAQ- or GNA11-mutated cell lines grew, albeit more slowly, in the presence of either MEK1/2 inhibitor (Supplementary Fig. S3G and S3H). These data indicate that GNAQ/11-mutated uveal melanoma cell lines are relatively resistant to MEK1/2 inhibitor monotherapy, an observation consistent with the failure of MEK1/2 inhibitors in uveal melanoma clinical trials (12, 34).

To investigate the effects of MEK1/2 inhibition plus chloroquine to elicit cell death, OMM2.5 cells were treated for 48 hours with DMSO, trametinib (T), chloroquine (CQ), or trametinib plus chloroquine (T/CQ; Fig. 2F). Treatment of either OMM2.5 (GNAQ^Q209P) or OMM1 (GNA11^Q209L) uveal melanoma cells with T/CQ resulted in increased cell death compared with vehicle or single agents. In addition, similar observations were made using Melan-a/GNAQ^Q209L or Melan-a/GNA11^Q209L cells (Fig. 2F; ref. 33). Furthermore, the antiproliferative effects of T/CQ treatment were calculated to be synergistic in all four cell lines (Fig. 2G). These results demonstrated that T/CQ is effective in eliciting synergistic cytotoxicity in GNAQ/11-mutated melanoma cell lines in vitro.

To determine whether MEK1/2 inhibition plus a 4-aminoquinoline might be effective in suppressing tumor growth in vivo, we tested trametinib in combination with HCQ, an analogue of chloroquine that has superior safety and bioavailability (35). Because more than 90% of metastatic uveal melanoma presents in the liver, we first utilized a hepatic colonization model. Cell lines derived from human metastatic uveal melanoma (OMM2.5 and OMM1) were engineered to express an EGFP-luciferase fusion protein to allow in vivo imaging of tumor burden, and were injected into the lower left lobe of the liver of recipient immunocompromised mice. Luminoscore, a surrogate measure for tumor burden using photon flux, was measured weekly using bioluminescent imaging (Fig. 3A; ref. 26). In our experience, a luminoscore of 10¹¹ photons/second corresponded to a tumor size of approximately 1 cm³ (Supplementary Fig. S4A). Effective trametinib-mediated inhibition of RAF>MEK>ERK signaling and increased autophagy in tumors was inferred by immunoblotting of tumor lysates that showed decreased pERK1/2 and p62^SQSTM1 expression (Supplementary Fig. S4B).

Treatment of mice bearing either OMM2.5 or OMM1 tumors with trametinib plus HCQ (T/HCQ) significantly inhibited tumor growth compared with vehicle control or single agents (Fig. 3A and B). Indeed, single-agent HCQ had little or no effect on tumor growth compared with vehicle-treated mice, and trametinib treatment alone led to a modest delay in the growth in OMM2.5 tumors, but had no effect on OMM1 tumors. Of note, temozolomide, a conventional cytotoxic chemotherapy approved for use in patients with metastatic uveal melanoma, but not shown to be beneficial in clinical trials, was also compared and showed no decrease in tumor growth nor survival benefit (Fig. 3B and D; Supplementary Fig. S5A and S5B; ref. 36). Strikingly, T/HCQ treatment provided a significant survival benefit compared to all other treatments such that 100% of tumor-bearing mice receiving this treatment remained alive at study conclusion (Fig. 3C and E). Mice bearing OMM1 tumors on trametinib alone also had a statistically significant survival benefit compared to vehicle treatment, but this was inferior to T/HCQ treatment (Fig. 3E). Moreover, at least in immunocompromised mice, T/HCQ combination treatment was well tolerated, as demonstrated by no significant weight changes compared with other treatments (Supplementary Fig. S4C and S4D).

Next, we tested T/HCQ in a syngeneic immunocompetent mouse model in which Melan-a/GNAQ^Q209L or Melan-a/GNA11^Q209L cells were injected subcutaneously into the flanks of albino C57BL/6 mice (Fig. 3F). Because these cells lines synthesize melanin, albino hosts allowed for distinct visualization of pigmented tumor margins. Similar to the hepatic tumor model, mice with subcutaneous GNAQ- or GNA11-driven Melan-a tumors displayed stable disease when treated with the T/HCQ combination (Fig. 3G–I) and had a significant survival advantage compared with monotherapy or vehicle control (Fig. 3H–J). Hence, using four different models of GNAQ- or GNA11-driven melanoma, we demonstrate that T/HCQ combination inhibits melanoma growth, either subcutaneous or in the liver, with or without the presence of an intact immune system, and also provides a survival advantage to T/HCQ-treated tumor-bearing mice.

4-aminoquinolines such as chloroquine and HCQ prevent the end stage of the autophagy process by preventing autophagosome-lysosome fusion and the acidification of lysosomes through inhibition of the vacuolar-type H(+)-ATPase (V-ATPase). However, these agents are known to have quite complex and pleiotropic effects on numerous additional biochemical processes (37). Hence, to determine whether the ability of chloroquine to enhance cell death in combination with trametinib is dependent on its role as an autophagy inhibitor, we utilized genetic inhibition of autophagy using OMM2.5/Tet-ATG4B^DN mutant cells (Figs. 1D and 4A). Surprisingly, treatment with trametinib in the presence of ATG4B^DN did not recapitulate the cytotoxic effects seen with T/CQ (Fig. 4B).

To determine whether inhibition of other lysosome-dependent recycling processes may act in combination with trametinib to promote cell death, we next examined macropinocytosis and mitophagy (Supplementary Fig. S6A, S6C, and S6D). However, inhibition of macropinocytosis with ethylisopropyl amiloride or inhibiting mitophagy with deguelin did not increase cell death in combination with trametinib (Supplementary Fig. S6B–S6E; refs. 38, 39). Because chloroquine exerts its intracellular effects on the lysosome, we next tested whether inhibition of lysosome acidification would cooperatively elicit cell death in combination with trametinib. However, the combination of bafilomycin A1 (Baf), another inhibitor of the V-ATPase, with trametinib (T/Baf), did not promote death of OMM2.5 cells and, unlike T/CQ, the T/Baf combination did not display synergistic antiproliferative activity in vitro (Figs. 2F, 4C, and D). This observation was also noted in MP41 cells in which the T/Baf combination did not enhance cell death, whereas the cells were sensitive to the T/CQ combination (Supplementary Fig. S1C). These results suggest that the effects of chloroquine on lysosome function are insufficient to explain the synergistic antiproliferative effects of T/CQ against GNAQ- or GNA11-driven melanoma cells.

Given that autophagy inhibition is sufficient to enhance cell death in the presence of an inhibitor of Gα signaling (FR), but not an inhibitor of RAF>MEK>ERK signaling (trametinib), and that Gα signaling is known to activate YAP signaling, we hypothesized that chloroquine may have effects on the MST1/2>LATS>YAP/TAZ/TEAD pathway. YAP is a transcriptional coactivator that binds to the TEAD family of transcription factors to elicit activation of progrowth and prosurvival pathways (40). As such, the presence of YAP in the nucleus is generally indicative of pathway activation (41, 42).

Interestingly, as single agents, neither trametinib nor chloroquine had striking effects on total YAP expression, as measured by immunoblotting, whereas the T/CQ combination resulted in an approximately twofold reduction in total YAP in OMM2.5 and OMM1 cells (Fig. 5A and B; Supplementary Fig. S7A and S7B). Moreover, T/CQ treatment decreased nuclear YAP accumulation as assessed by measuring nuclear:cytoplasmic YAP. Single-agent trametinib also led to a significant decrease in nuclear YAP in OMM2.5 and OMM1, whereas chloroquine alone led to decreased nuclear YAP in OMM1 cells only (Fig. 5C and D; Supplementary Fig. S7C and S7D). These data suggest that trametinib and chloroquine each have the ability to, at least partially, inhibit YAP nuclear localization.

To determine whether YAP-mediated transcription output was inhibited by T/CQ, we utilized a YAP/TEAD luciferase reporter. Consistent with the YAP expression data, T/CQ treatment led to decreased levels of YAP/TEAD-mediated luciferase activity. Indeed, T/CQ decreased YAP activity to levels comparable to the established YAP inhibitor, verteporfin (Fig. 5E; Supplementary Fig. S7E; ref. 43). Correspondingly, T/CQ treatment also led to a decrease in mRNAs of known YAP target genes, CTGF and CYR61, similar to the effects of verteporfin (Fig. 5F). Consequently, these results suggest that the T/CQ combination is an inhibitor of YAP/TEAD-mediated transcription.

To determine the significance of YAP inhibition for cell death induced by T/CQ, we used pharmacologic and genetic approaches to inhibit YAP in the presence of trametinib. Because verteporfin elicits reactive oxygen species in response to red light (∼689 nm), we were unable to use the IncuCyte live-cell imaging approach that was used previously for cell death measurements (44). Instead, under dark conditions, we used an endpoint ATP-luciferase assay at 48 hours as a readout of cell proliferation. The combination of trametinib plus verteporfin elicited a modest, but insignificant decrease in cell density compared with trametinib or verteporfin alone (Fig. 6A). T/Baf did not result in changes in cell density compared with trametinib alone, which is consistent with our previous data (Figs. 4C and 6A). Verteporfin combined with bafilomycin A1 had a similarly modest inhibitory effect on cell proliferation compared with T/Baf. However, the triple combination of trametinib, verteporfin plus bafilomycin A1 resulted in a significant decrease in cell proliferation similar to that observed with the T/CQ combination, or staurosporine, a multikinase inhibitor and potent inducer of cell death (Fig. 6A). A similar analysis of the response of MP41 cells to the same treatments revealed that this cell line was more sensitive to verteporfin than OMM2.5 cells. Consistent with OMM2.5, verteporfin in combination with either trametinib or bafilomycin A1 did not decrease cell density. However, the combination of verteporfin, trametinib plus bafilomycin A1 resulted in a decrease in cell proliferation that was similar to that observed with T/CQ (Supplementary Fig. S8A). To complement these studies, we employed a siRNA approach to inhibit YAP expression. In cells in which YAP expression was inhibited by siYAP-mediated protein knockdown (Fig. 6B), there was a decrease in cell density comparable with verteporfin (Fig. 6A and C). In addition, treatment with T/Baf yielded a significant decrease in cell proliferation in siYAP knockdown cells versus control cells (Fig. 6C). Taken together, these data are consistent with the hypothesis that inhibition of uveal melanoma cell proliferation requires combined inhibition of: (i) RAF>MEK>ERK; (ii) YAP/TEAD transcriptional output plus; (iii) Inhibition of lysosome function.

To further investigate the requirement of YAP inhibition in T/CQ efficacy, OMM2.5 cells were engineered to stably express a constitutively active form of YAP (YAP^5SA) in which five sites of serine phosphorylation (S61A, S109A, S127A, S164A, S381A), which are required for inhibition of YAP activity by upstream LATS protein kinases, are substituted to alanine (45). Consequently, cells expressing YAP^5SA are predicted to have constitutively active nuclear YAP even in the absence of upstream signaling. First, we noted that expression of YAP^5SA sustains YAP activity in OMM2.5 cells, even in the presence of T/CQ (Fig. 6E). Furthermore, expression of YAP^5SA in OMM2.5 cells significantly diminished the cytotoxic activity of T/CQ compared with vector-expressing cells (Fig. 6F). To test these observations in another uveal melanoma setting, MP41 cells were transiently transfected with YAP^5SA with the mRNA expression of YAP targets, CTGF and CYR61 measured in the presence or absence of T/CQ. T/CQ-treated cells had decreased YAP activity, and YAP^5SA-transfected cells had two- to threefold increased YAP activity that was predominantly sustained in the presence of T/CQ (Supplementary Fig. S8B). Consistent with YAP^5SA-expressing OMM2.5 cells, YAP^5SA-expressing MP41 cells had reduced cell death in the presence of T/CQ compared with vector-expressing cells (Supplementary Fig. S8C). These data are therefore consistent with a role for YAP in protecting uveal melanoma cells against the effects of combined inhibition of RAF>MEK>ERK signaling plus autophagy/lysosome inhibition.

Data presented here are consistent with the observation that inhibition of lysosome function by chloroquine is insufficient to promote cell death in the presence of MEK1/2 inhibition. It appears that inhibition of YAP/TEAD transcriptional output is also required for the cell death that is observed in response to trametinib plus chloroquine. Thus, we identify three critical survival nodes in uveal melanoma cells: RAF>MEK>ERK signaling, lysosome function, and YAP/TEAD signaling, that must be targeted to promote cytotoxicity in GNAQ/11-mutant melanoma, which is achieved by the novel trametinib plus chloroquine combination therapy.

Despite encouraging initial results, a phase III clinical trial of MEK1/2 inhibition (selumetinib) in combination with dacarbazine failed to provide clinical benefit to patients with metastatic uveal melanoma (12). However, there is increasing evidence from preclinical studies that other combination strategies may be more efficacious (46–51). Here we show that the combination of the more potent MEK1/2 inhibitor, trametinib, plus chloroquine increases antiproliferative activity in vitro, and suppresses tumor growth leading to prolonged survival of mice bearing melanomas driven by mutationally activated GNAQ/11. We and others previously showed that this combination strategy is effective in certain RAS-mutated malignancies, including melanoma and pancreatic cancer, due to chloroquine's ability to inhibit cytoprotective autophagy that is elicited in response to inhibition of RAF>MEK>ERK signaling (14, 15, 18). Interestingly, and in contrast to RAS-mutated cancers, we found that autophagy inhibition was insufficient to elicit cell death in combination with trametinib in GNAQ/11-mutant melanoma. Although this may be due to technical limitations of our autophagy inhibition assays, we also demonstrated that inhibition of other lysosome-dependent mechanisms, such as mitophagy or macropinocytosis, did not enhance cell death in combination with trametinib. We also identified YAP signaling as a survival pathway that is inhibited by T/CQ in melanomas driven by mutationally activated GNAQ/11.

YAP has been associated with MEK1/2 inhibitor resistance in BRAF- and NRAS-mutated melanoma (52). MEK1/2 inhibition has also been shown to elevate YAP signaling in primary uveal melanoma cell lines at low concentrations of trametinib (10 nmol/L; ref. 51). Hence, the decrease in YAP activity in response to MEK1/2 inhibition in uveal melanoma cell lines observed here suggests a dose-dependent effect of MEK1/2 inhibition or that YAP may be differentially regulated in different uveal melanoma cell lines, perhaps revealing additional complexity and heterogeneity of YAP signaling in uveal melanoma in response to targeted therapy.

There are currently no defined mechanisms of inhibition of YAP signaling in response to chloroquine. We and others have shown that chloroquine reduces, to a modest extent, YAP nuclear localization, suggesting a role for chloroquine in stabilizing YAP in the cytoplasm (37, 53, 54). Future studies should identify mechanisms of YAP regulation through MAPK activity and chloroquine. Of note, verteporfin, although used as a YAP inhibitor in several studies, is not an entirely specific or selective YAP inhibitor (43, 55–57). Nevertheless, the consistent effects of verteporfin and siYAP on YAP/TEAD-driven transcription and the ability of YAP^5SA to bypass the effects of verteporfin suggests that the inhibitory effect of verteporfin is most likely due to YAP inhibition.

In our preclinical models, we observed stabilization of tumor growth leading to prolonged survival of mice bearing GNAQ/11-driven melanoma, but we did not see evidence of frank tumor regression as was observed in certain RAS-driven cancers treated with T/HCQ (15). This may be due to unique survival pathways activated by Gαq/11 oncoproteins. Of note, BRCA1-associated protein 1 (BAP1) is a deubiquitinase that is commonly silenced in hepatic metastases derived from primary uveal melanoma (58, 59). Because the cell lines used in this study are BAP1 proficient, further studies of T/CQ in BAP1-deficient settings are warranted. Although it is not clear whether BAP1 influences autophagy, it has been reported that BAP1 loss has pleiotropic effects on cells including inhibition of apoptosis, metabolic changes, increases in endoplasmic reticulum stress, and accumulation of DNA damage—all of which may promote autophagy (60–63). Interestingly, in pancreatic ductal adenocarcinoma, loss of BAP1 has been shown to activate Hippo-mediated YAP signaling through LATS2 degradation (64). Thus, the autophagy and YAP inhibitory mechanisms of T/CQ may be particularly relevant in BAP1-deficient tumors.

Preclinical data presented here makes the case that clinical testing of the combination of trametinib plus HCQ may be a promising approach in uveal melanoma—all the more so as direct inhibitors of Gα oncoprotein drivers remain unavailable. Per current National Comprehensive Cancer Network guidelines, trametinib is already indicated as a treatment option for metastatic uveal melanoma (34, 65). Chloroquine/HCQ have been approved for medical use since the 1950s for many medical indications, including malaria and some autoimmune conditions, hence the toxicity profile of these agents is well documented. Of note, HCQ has greater bioavailability than chloroquine and is more commonly prescribed (35). Because trametinib and HCQ are FDA approved, there is significant potential for rapid clinical translation, all the more so due to the urgent, unmet clinical need for effective treatments for metastatic uveal melanoma. Indeed, there are already multiple ongoing clinical trials investigating MEK1/2 inhibition plus HCQ in the setting of RAS-mutated cancers (e.g., NCT03825289, NCT03979651, and NCT04132505). Hence, these findings provide the foundation for a potentially effective treatment strategy for metastatic uveal melanoma that warrants future clinical trials testing the efficacy of combined trametinib plus HCQ for GNAQ/11-mutated melanoma.

A. Truong reports grants from NIH during the conduct of the study. C.G. Kinsey reports nonfinancial support from Novartis (Novartis provides free trametinib for the THREAD Clinical Trial) during the conduct of the study; grants and personal fees from Deciphera Pharmaceuticals, Inc., personal fees from SpringWorks Therapeutics, nonfinancial support from Novartis, and grants from BioMed Valley Discoveries, Inc. outside the submitted work. M.D. Onken reports grants from NIGMS during the conduct of the study. K.J. Blumer is listed as a co-inventor on a provisional patent application on TARGETED PHARMACOLOGICAL THERAPEUTICS IN UVEAL MELANOMA that is owned by Washington University in St. Louis. S.J. Odelberg reports grants from NIH/NCI during the conduct of the study; and nonpaid association with a pharmaceutical company (Navigen, Inc. and its subsidiary A6 Pharmaceuticals) that does not work on any of the drugs mentioned in this manuscript but does have an interest in uveal melanoma. M. McMahon reports grants from NCI during the conduct of the study. No potential conflicts of interest were disclosed by the other authors.

A. Truong: Conceptualization, resources, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing-original draft. J.H. Yoo: Resources, data curation, investigation, methodology, writing-review and editing. M.T. Scherzer: Investigation, writing-review and editing. J.M.S. Sanchez: Investigation, methodology, writing-review and editing. K.J. Dale: Data curation, validation, investigation, writing-review and editing. C.G. Kinsey: Conceptualization, resources, supervision, writing-review and editing. J.R. Richards: Resources, investigation, methodology, writing-review and editing. D. Shin: Resources, investigation, writing-review and editing. P.C. Ghazi: Investigation, writing-review and editing. M.D. Onken: Conceptualization, resources, writing-review and editing. K.J. Blumer: Conceptualization, resources, writing-review and editing. S.J. Odelberg: Data curation, formal analysis, supervision, writing-review and editing. M. McMahon: Conceptualization, resources, supervision, funding acquisition, writing-review and editing.

This study was funded by grants from the NIH (CA176839) to M. McMahon. HCI shared resources are funded, in part, by a Cancer Center support grant from the NCI (P30 CA042014). Research reported in this publication was also supported by the National Center for Research Resources of the NIH (S10 RR026802). A. Truong was supported by awards from: The American Skin Association Hambrick Medical Student Grant Targeting Melanoma & Skin Cancer Research; An NIH training grant (T32 HD007491); An NIH predoctoral fellowship F30CA235964. J.H. Yoo is supported by NCI/NIH (K99 CA230312) and the Melanoma Research Foundation (Cure OM Junior Fellowship award). J.R. Richards is supported by the NIH/NCI predoctoral fellowship (F30 CA217184). S.J. Odelberg is supported by a grant from the NIH/NCI (R01 CA202778). This research utilized both University of Utah and Huntsman Cancer Institute shared research resources including the Flow Cytometry, Preclinical Research Resource and DNA/Peptide Synthesis Facility. We extend our deepest appreciation to Sheri Holmen, Kevin Jones, Benjamin Spike, and Alana Welm for their insightful feedback. We thank Glenn Merlino and Chi-Ping Day for generously providing mice, and Boris Bastian, Scott Woodman, J. William Harbour, Sergio Roman Roman, and Kang Zhang for providing cell lines. We thank Claire Bensard for donating reagents, Mona Foth for technical assistance, and Aria Vaishnavi for critical review of this manuscript.

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.