IGF1R inhibition enhances the therapeutic effects of G_q/11 inhibition in metastatic uveal melanoma progression

Abstract

Uveal melanoma (UM) is the most common intraocular tumor in adults and up to 50% of patients develop metastatic disease, which remains uncurable. Since metastatic UM patients have an average survival of less than 1 year after diagnosis, there is an urgent need to develop new treatment strategies. While activating mutations in Gα_q or Gα₁₁ proteins are major drivers of pathogenesis, therapeutic intervention of downstream Gα_q/11 targets has been unsuccessful in treating UM, possibly due to alternative signaling pathways and/or resistance mechanisms. Activation of the insulin-like growth factor 1 (IGF1) signaling pathway promotes cell growth, metastasis, and drug resistance in many types of cancers including UM where expression of the IGF1 receptor (IGF1R) correlates with a poor prognosis. Here we show that direct inhibition of Gα_q/11 by the cyclic depsipeptide YM-254890 in combination with inhibition of IGF1R by linsitinib, cooperatively inhibits downstream signaling and proliferation of UM cells. We further demonstrate that a 2-week combination treatment of 0.3–0.4 mg/kg of YM-254890 administered by intraperitoneal injection and 25–40 mg/kg linsitinib administered by oral gavage effectively inhibits the growth of metastatic UM tumors in immunodeficient NOD scid gamma (NSG) mice and identify the IGF1 pathway as a potential resistance mechanism in response to Gα_q/11 inhibition in UM. These data suggest that the combination of Gα_q/11 and IGF1R inhibition provides a promising therapeutic strategy to treating metastatic UM.

Introduction

Uveal melanoma (UM) is the most common intraocular tumor in adults with approximately 2,500 new cases per year in the United States and a median age of diagnosis of 62 years (1,2). UM has a high propensity to metastasize and as a result, 36–50% of UM patients develop metastasis, predominantly to the liver (3–6). Once metastasis occurs, patients have a poor prognosis with a median overall survival time of less than one year due to lack of therapeutic options. Unlike cutaneous melanoma (CM), targeted therapy and checkpoint inhibitor clinical trials of metastatic UM have shown little benefit (7,8). While recent studies with a bispecific antibody tebentafusp showed an overall survival benefit in metastatic UM patients (9), there remains an urgent need for additional therapeutic options for metastatic UM.

UM predominantly develops as a result of aberrant activation of the G_q/11 signaling pathway via activating mutations (Q209L/P) in the Gα_q/11 subunit of the heterotrimeric protein (10–12). Approximately 90% of all UM cases contain oncogenic Gα_q/11, which canonically activates phospholipase C β (PLCβ) and downstream effectors, but leads to an upregulation of mitogen activated protein kinase (MAPK), AKT, focal adhesion kinase (FAK), and yes-associated protein (YAP) signaling when constitutively active (13–17). Like CM, the MAPK pathway is highly upregulated in UM, however, this occurs in the absence of the canonical drivers found in CM (13,18). While BRAF and MEK inhibitors show efficacy in treating CM patients, UM remains poorly responsive to similar treatments (19). This may be due to G_q/11 acting as a central node for multiple downstream oncogenic pathways.

Recent evidence suggests that mutant constitutively active Gα_q/11 can be directly targeted by the very similar, naturally occurring cyclic depsipeptides, FR900359 (FR) (20) and YM-254890 (YM) (21,22). These compounds act by allosterically locking the Gα_q/11 subunit in the GDP-bound inactive conformation, preventing downstream signaling (23,24). We and others have shown that YM and FR can effectively inhibit the oncogenic mutant Gα_q/11 in UM, which leads to cell cycle arrest, decreased oncogenic signaling, and increased apoptosis (25–28). Furthermore, recent studies have shown that YM and FR can inhibit growth of primary and metastatic UM tumors in mouse xenograft models (29–31). Moreover, the combination of YM and a MEK inhibitor can successfully treat UM in mice, while additional studies suggest FR may be used in combination with other standard-of-care interventions if necessary (29,30). Nevertheless, both YM and FR are potent inhibitors of wild-type Gα_q/11, which raises concerns over the safety and therapeutic window of these inhibitors in targeting G_q/11 in UM.

Although recent studies have shown that YM and FR can inhibit growth of UM tumors in mouse xenograft models, the subcutaneous tumor microenvironment is significantly different to that of the liver where UM most commonly metastasizes. While YM and FR have shown efficacy as single agents treating tumors subcutaneously, this may not be true of tumors located in the liver where growth factors such as hepatocyte growth factor (HGF) and insulin-like growth factor 1 (IGF1) can potentially provide resistance to G_q/11 inhibition (32). HGF has previously been shown to provide resistance to MEK inhibition in UM, and IGF1 receptor (IGF1R) signaling promotes cell proliferation and survival of UM cells (33,34). To address the potential role of the liver microenvironment, we utilized a metastatic UM mouse model wherein human hepatic-metastatic UM tumors form in the livers of NSG mice. Here, we identify the IGF1 pathway as a potential resistance mechanism to G_q/11 inhibition in UM and find that co-targeting the IGF1R and oncogenic Gα_q/11 works synergistically to enhance the effects of YM on UM cell growth, signaling, and tumor growth in vivo.

Materials and Methods

Drugs and Chemicals

YM-254890 was purchased from Focus Biomolecules (10–1590) and Wako Pure Chemical Industries (257–00631). Linsitinib (OSI-906) was purchased from Selleck Chemical (S1091) and LC Laboratories (L-5814). Picropodophyllin (PPP) and Alpelisib (BYL719) were purchased from Selleck Chemical. IGF1 was purchased from Abcam.

Cell lines and culture

OMM1.3 cells (RRID:CVCL_C306) were obtained from Dr. Bruce Ksander’s lab and contained a Q209P mutation in GNAQ, 92.1 cells (RRID:CVCL_8607) were from Dr. Martine Jager and contained a GNAQ-Q209L mutation, and OCM3 cells (RRID:CVCL_6937) were obtained from Dr. Bruce Ksander and contained a BRAF-V600E mutation and wild type GNAQ/11. UM001 cells (RRID:CVCL_B6PW) were derived from a liver metastasis of human UM and contained a Q209P mutation in GNAQ as described previously (35). OMM1.3, 92.1, OCM3, and UM001 cells were authenticated by STR profiling prior to their use beginning in 2019. 92.1, OMM1.3, UM001, and OCM3 cells were cultured in RPMI 1640 containing 10% fetal bovine serum (FBS), 2 mM L-glutamine, 0.2 units/ml penicillin and 100 μg/ml streptomycin. Cells were used within 30 passages after thawing. Cells were cultured at 37°C and 5% CO₂ in a humidified incubator. Cells were tested for mycoplasma by PCR in September 2019.

Cell Proliferation and Synergy Assay

92.1, OMM1.3, UM001 and OCM3 cells were plated on poly-L-lysine coated 96-well white bottom plates at 5–10 × 10³ cells per well and allowed to sit overnight at 37°C and 5% CO₂ in a humidified incubator. Cells were then treated with DMSO, YM-254890, and/or linsitinib for 3 days and harvested by removing the media and lysing the cells with 100 μl of cell-grade water for 30 min at 37°C with shaking. 100 μl of 200-fold diluted Quant-iT™ PicoGreen™ (Invitrogen) reagent in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.5) was added to the mixture and incubated for 5 min at room temperature in the dark with shaking. DNA-bound PicoGreen™ fluorescence was detected using a Tecan Infinite F500 microplate reader. Relative cell number was normalized to vehicle (100%) and input into Combenefit software to obtain Loewe synergy and cell proliferation (36).

3D Cultures in Matrigel

The morphogenesis analysis was done using 3D cultures in Cultrex. Briefly, 5 × 10³ OMM1.3 cells treated or not with YM-254890 were seeded on top of a Cultrex matrix (Trevigen, Gaithersburg, MD) as described (37). Cells were cultured under these conditions for 7 to 12 days and the morphology of the colonies was monitored daily by phase contrast microscopy.

Histone H2B-GFP Label Retention

OMM1.3 cells constitutively expressing td-tomato and tet-on-H2B-GFP were treated with 1 μg/ml doxycycline for 3 days in 2D culture to induce H2B-GFP expression. After confirming induction of H2B-GFP, cells were harvested for 3D culture. 3D cell culture assays were performed in 8 well chamber (Falcon) with 400 μl/well of Matrigel™ Matrix Growth Factor Reduced (R&D Systems™ Cultrex). Cells were suspended in complete medium supplemented with 5% Matrigel™, plated at a density of 500 cells/well and incubated at 37°C for 14 days. At day 0, cells were treated with either DMSO or different concentrations of YM-254890. A fresh layer of complete medium supplemented with 5% serum and 3% Matrigel™ and different doses of YM-254890 was added every day to continue the treatment for 14 days. Colonies formed after 5–7 days were monitored daily and at day 14 colonies were quantified in triplicate for % label retention using Nikon Eclipse Ti-S microscope.

Mouse Experiments

NOD.Cg-Prkdc^scid Il2rg^tm1WjI/SzJ (NSG) mice (RRID:IMSR_JAX:005557) were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were housed and bred in ventilated microisolator cages at 22 °C, 60% humidity. Both male and female 6- to 8-week-old mice were used in the study. The animal study was approved by the Institutional Animal Care and Use Committee of Thomas Jefferson University and adhered to the guidelines in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Mice were placed on a heating pad and anesthetized with 3% isoflurane for induction and 2% isoflurane for maintenance. 70% ethanol was sprayed on the abdomen and then it was scrubbed with betadine before surgery. Mice were placed in the supine position. A 1- to 1.5-cm skin incision was made in the upper abdominal wall, followed by a 1-cm incision in the peritoneum to expose the liver. The left lobe of the liver was moved outside the body and placed on a nonwoven absorbent fabric sheet as previously described (38). 1 × 10⁶ UM001-tdTomato cells (RRID:CVCL_B6PZ) in 20 μL of 2:1 RPMI1640/Matrigel (BD Biosciences, Bedford, MA, USA) were gently injected under the surface of the left lobe of the liver. The insertion site of the needle was sealed with absorbable hemostatic material. The liver was returned within the body after UM001-tdTomato cell injection, and the abdominal incision was closed with 5–0 polydioxanone absorbable thread (AD Surgical, Sunnyvale, CA, USA). We had originally planned to use an in vivo imaging system (IVIS) to measure the growth of tumors in the liver, however, this was confirmed to be inaccurate due to variation of distance from the implanted tumors and the detector. Therefore, micro-CT scan (Inveon Micro-CT, Siemens, Germany) was performed pre-treatment and at 2 and 4 weeks after treatment. One day before the CT scan, mice were injected with 100 μL of a contrast agent (ExiTron nano 12000, Militenyi Biotec, Germany) via the tail vein. Tumor volume was calculated using the following formula: 4π/3 × (a/2) × (b/2) × (c/2) and 3-dimensional images of tumor in the liver (Fig. S1). The longest tumor diameter (a) was scaled and then the shortest diameter (b) of the tumor was measured in the axial view. Z-axis of the tumor (c) was scaled with coronal view or sagittal view depending on tumor shape in the liver. CT data was analyzed using Horos software (Horos Project, Annapolis, MD, USA). Analyses of continuous data was performed by the unpaired non-parametric Mann–Whitney U-test. Groups were considered significantly different at p < 0.05. Statistical analyses were performed using Prism 9.

For initial in vivo experiments, YM-254890 was prepared from a 10 mg/ml stock solution in DMSO. Mice were treated with 0.5 mg/kg of YM-254890 (daily intraperitoneal injection) for 4 weeks or with a matched vehicle (2.5% DMSO in 5% dextrose in ddH₂O). For combination in vivo experiments, YM-254890 and linsitinib were prepared from 10 mg/ml stocks. Mice were treated with 0.4 mg/kg or 0.3 mg/kg of YM-254890 (daily intraperitoneal injection; 5 days on, 2 days off), 40 mg/kg or 25 mg/kg of linsitinib (daily oral gavage), a combination of both, or a matched vehicle (2.5% DMSO in phosphate buffered saline + 25 mM tartaric acid). The treatment period was limited to 4-weeks due to the large size of the tumors in many of the mice following the initial 5-week pre-treatment and 4-week treatment/follow-up period.

Immunohistochemistry

Tumor tissues were removed from xenograft mice following 2 weeks of treatment. Tissues were fixed in formalin overnight and then embedded in paraffin. After sectioning, tissue sections were treated with antigen retrieval solution. Sections were incubated with Bioxall blocking solution (Vector Laboratories, Burlingame, CA, USA) followed by anti–phospho-histone H3 (pHH3) (#06–570, Millipore, RRID:AB_310177)) or anti-phospho-S6 (Ser235/236, #4858, Cell Signaling Technology, RRID:AB_916156)) overnight at 4 °C. On the following day, sections were incubated with Immpress AP anti-rabbit IgG reagent (Vector Laboratories) for 1 hour, follow by 10 min incubation with ImmPACT vector Red substrate reagent (Vector Laboratories). Tissue sections were counterstained using Hematoxylin (Vector Laboratories). The intensity of pHH3 and pS6 staining was captured using a Nikon Eclipse 50i microscope with NIS-Elements D3.1 software (Nikon, Melville, NY, USA). The images were captured in randomly selected tumor areas in each group. The positive pHH3 in each section was counted with Celleste 4.1 software (Invitrogen, Waltham, MA, USA). Statistical analyses were performed using Graphpad Prism 9 software (Graphpad Software, San Diego, CA, USA, RRID:SCR_000306). Data were analyzed using Mann-Whitney test.

Reverse Phase Protein Array (RPPA) Analysis

UM001 and OMM1.3 cells were plated in six-well dishes at 1.5 × 10⁵ cells per well. Cells at ~80% confluency were treated with DMSO, 100 nM YM-254890, or 1 μM YM-254890 for 24 hours. Cells were lysed in RPPA lysis buffer (1% Triton X-100, 50 mM HEPES, pH 7.4, 150 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, 100 mM NaF, 10 mM Na pyrophosphate, 1 mM Na₃VO₄, 10% glycerol, and freshly added protease and phosphatase inhibitors from Roche Applied Science), and proteins were prepared and analyzed as previously described (39). RPPA data were analyzed as previously described (25). Results were validated by Western blotting for key targets.

Western Blotting

Cell lysates were separated by SDS-PAGE and proteins were transferred to PVDF membrane. ERK2 (Santa Cruz Biotechnology Cat# sc-1647, RRID:AB_627547) and ERK1/2 (Santa Cruz Biotechnology Cat# sc-514302, RRID:AB_2571739) and pERK1/2 (Cell Signaling Technology Cat# 9101, RRID:AB_331646) antibodies were detected using IRDye 800 CW and IRDye 680RD fluorescent secondary antibodies (LI-COR) with a LI-COR Odyssey and fluorescence was quantified using ImageStudioLite (RRID:SCR_013715). All other immunoreactivity was detected using horseradish peroxidase conjugated secondary antibody (Vector Laboratories Cat# PI-1000, RRID:AB_2336198) and chemiluminescence substrate (ThermoScientific). For other western blots detecting phosphoproteins, the phosphoproteins were probed first then blots were stripped (25 mM glycine, 1 % SDS, pH 2.2) for 5 minutes, washed with TBST 3 times for 10 minutes, and membranes were re-probed for total protein. For phospho-S6 (pS6), the two different pS6 blots were the same sample run on separate gels and then probed for pS6-Ser235/236 or pS6-Ser240/244. IRS2 (#4502, RRID:AB_2125774), pFAK (Y397, #8556, RRID:AB_10891442), Wee1 (#4936, RRID:AB_2288509), pTAZ (S89, #59971, RRID:AB_2799578), TAZ (#2149, RRID:AB_823657), pS6 (Ser235/236, #2211, RRID:AB_331679)), pS6 (Ser240/244, #2215, RRID:AB_331682), pRb (Ser807/811, #9308, RRID:AB_331472), GAPDH (#5174, RRID:AB_10622025), pIGF1R (Y1135, #3918, RRID:AB_10548764), IGF1R (#3027, RRID:AB_2122378), pAKT (S473, #4060, RRID:AB_2315049), AKT (#9272, RRID:AB_329827), pYAP (S127, #13008, RRID:AB_2650553), and Cleaved Caspase-3 (#9664, RRID:AB_2070042) antibodies were all from Cell Signaling Technology. pCDK1 (Thr14/Tyr15, #44686G, RRID:AB_2533722) antibody was from Invitrogen and YAP (#ab52771, RRID:AB_2219141) antibody was from Abcam.

EdU Incorporation Assays

92.1 and OMM1.3 cells were treated with DMSO, 30 nM YM-254890, 1 μM linsitinib, or both YM-254890 and linsitinib while UM001 cells were treated with DMSO, 3 nM YM-254890, 1 μM linsitinib, or both for 24 h. Where noted, 50 ng/ml IGF1 was added to media for 24 h. Cells were then incubated with 10 μM EdU and then prepared according to the Click-iT™ Plus EdU Alexa Fluor 647 Flow Cytometry Assay Kit (Invitrogen) protocol. EdU was measured and analyzed using a FACS Calibur Flow Cytometer.

Data availability

The data generated in this study are included in the figures. Additional primary data sources supporting the study are available upon request from the corresponding author.

Results

YM-254890 treatment inhibits UM cell proliferation and colony formation

To establish that YM can effectively inhibit proliferation of UM cells, we treated several UM cell lines with varying concentrations of YM over the course of 3 days and determined the relative cell abundance compared to untreated controls (Fig. 1A). The proliferation of primary 92.1 (GNAQ-Q209L) and metastatic OMM1.3 (GNAQ-Q209P) and UM001 (GNAQ-Q209P) cells was effectively inhibited by YM in a dose-dependent manner. In contrast, YM had no effect on the proliferation of OCM3 melanoma cells, containing a BRAF-V600E mutation. This corroborates previous reports of YM inhibition of UM cell lines with activating mutations in Gα_q or Gα₁₁ (27,30).

Figure 1:

Cell growth, colony formation, and MAPK signaling effects of YM. A) Control (OCM3) and UM cell lines were treated with YM for 3 days and relative cell number was detected using the DNA intercalator, PicoGreen. Error bars represent the mean ± SEM from 3 experiments done in triplicate. B) OMM1.3 cells were seeded in 3D Matrigel culture and treated with YM 4 hours later. YM treatment continued for 7 days followed by a 5-day washout. C) OMM1.3 cells positive for td-tomato and tet-on-H2B-GFP expression after doxycycline treatment were grown in 3D Matrigel culture and treated with DMSO or YM for 14 days. Colonies and percent H2B-GFP retention was quantified. Each data point represents a colony across two independent wells per treatment dose. D) UM cells were treated with 1 μM YM for 1–24 hours or DMSO for 24 hours and cell lysates were immunoblotted for ERK1/2 and pERK1/2. Error bars represent the mean ± SEM from 3 independent experiments. P < 0.05 (*), P < 0.005 (**), P < 0.0001 (****) for differences between control and YM treated pERK to ERK ratio, One-way ANOVA.

In order to obtain a better idea of how effectively YM might inhibit cell proliferation in vivo, we next utilized 3D-Matrigel-cell culture, which allows cells to grow in contact with an extracellular matrix and provides positional information. OMM1.3 cells were seeded in the Matrigel culture and allowed to form 3D cellular colonies of at least 20 cells in size, analogous to how they would initiate colonization in the liver. Untreated cells formed colonies within 12 days, while cells treated with 100 nM or 300 nM YM immediately after being seeded were completely incapable of forming colonies over the same time span (Fig. 1B). YM also led to significantly increased H2B-GFP/td-tomato ratio of OMM1.3 cells in 3D Matrigel culture in a dose dependent manner (Fig. 1C). These data suggest YM prevents colony formation and induces growth arrest.

Oncogenic Gα_q inhibition differentially decreases MAPK signaling in UM cell lines

Increased MAPK signaling is commonly observed in UM cells as a result of oncogenic Gα_q/11 (11–13). Gα_q/11 inhibition has been shown to significantly decrease MAPK signaling in UM cells (25,27,29–31). To determine if there are differences in the effects of YM treatment on MAPK signaling in various UM cells, we immunoblotted for pERK1/2 after treating 92.1, OMM1.3, and UM001 cells with 1 μM YM over a 24-hour period (Fig. 1D). ERK1/2 activation in the metastatic cell lines, OMM1.3 and UM001, was effectively reduced within 1 hour after the addition of YM, while ERK1/2 activation in the primary cell line, 92.1, slowly decreased reaching its lowest point by 12 hours. Interestingly, we observed a slight rebound of ERK1/2 activation in the UM cell lines after 24 hours of YM treatment, although it was still significantly inhibited. In contrast, ERK1/2 phosphorylation was unchanged by YM treatment of control OCM3 cells. We further investigated how rapidly ERK1/2 inhibition occurs upon YM treatment of OMM1.3 cells and determined a t_1/2 of ~7.5 min (Fig. S2). These studies demonstrate that YM significantly decreases MAPK signaling in UM cells harboring oncogenic Gα_q, although the rate of this inhibition is cell line dependent.

YM significantly reduces tumor size in metastatic UM mouse models

We next investigated if YM can inhibit tumor growth of metastatic UM in vivo. Others have reported the success of G_q inhibitors, YM and FR, in preventing tumor growth in UM mouse models, however, these studies have all used subcutaneous xenografts (29–31). The subcutaneous tumor microenvironment is significantly different to that of the liver where UM most commonly metastasizes. Therefore, we utilized a model wherein human hepatic-metastasis UM cells, UM001-tdTomato, were injected into the livers of NSG mice and were allowed to form small tumors over a 5-week period followed by a 4-week treatment with 0.5 mg/kg YM (Fig. S3A). CT scans were taken pre-treatment and every 2 weeks in vehicle control and YM treated mice (Fig. S3B). YM treatment did not significantly inhibit tumor growth but did show some evidence of reduced growth, suggesting a higher dose of YM may be more effective (Fig. S3C). Unfortunately, treatment with a higher dose of YM is unfeasible as YM administered at 0.5 mg/kg had adverse effects causing 3 of the 7 YM treated mice to die as early as day 9 over the course of the experiment. While the cause of death is not known, some mice became inactive after administration of YM and subsequently died. These data indicate that a higher dose of YM might be needed to successfully reduce metastatic tumor growth in the liver of mice but has significant toxicity at the 0.5 mg/kg concentration. Therefore, we questioned whether there is an additional therapeutic target that may work synergistically to enhance the effect of YM at a safer concentration.

Insulin-like receptor substrate 2 expression increases upon YM treatment of UM cells

To determine a potential co-target that may work synergistically with G_q/11 inhibition, we used a high-throughput antibody-based reverse phase protein array (RPPA) analysis to help identify proteins whose expression or activation is increased in response to YM treatment. Cell lysates from DMSO, 100 nM YM or 1 μM YM treated UM cells were probed for proteins and phosphoproteins with 466 unique antibodies and analyzed for changes in protein levels. 25 of the proteins that were most affected by YM treatment of UM001 and OMM1.3 cells are shown in Fig. 2A. The changes in expression or phosphorylation for a select number of them were then validated by immunoblotting of DMSO and YM treated cell lysates (Fig. 2B). Many of the downregulated proteins observed here are in line with what we previously reported when using FR in UM cells (25). The expression of proteins that promote cell cycle progression such as pCDK1, cyclin-B1, PLK1, pRb, and Wee1 was substantially decreased by YM treatment, while the decreased phosphorylation of the S6 ribosomal subunit and p-90RSK is indicative of cell growth arrest (40). Upon YM treatment we also observed decreased expression of proteins that are often upregulated in UM and help drive UM pathogenesis such as pAKT, pFAK, and TAZ (YAP homolog). In contrast, OCM3 cells showed no changes in expression or phosphorylation of the validated proteins when treated with YM (Fig. 2B). Interestingly, one of the only proteins observed to be upregulated by YM treatment in both UM cell lines was insulin-like receptor substrate 2 (IRS2). Additionally, RNA-sequencing data shows transcription of IRS2 is significantly increased in OMM1.3 cells treated with YM (30). Taken together, this suggests IRS2 is upregulated as a result of oncogenic G_q/11 inhibition. IRS2 mediates the intracellular response of the IGF1R after activation by the binding of IGF1 or insulin (41), and IRS2 expression has been implicated in many cancers (42–44). IGF1R expression has been shown to significantly correlate with worse prognosis in UM, while IGF1 is produced in the liver which is the main cite of UM metastasis (45–47). We therefore turned our attention to IGF1R as a potential co-target.

Figure 2:

Reverse Phase Protein Array analysis of YM treated UM cells. A) A heat map showing supervised hierarchical clustering of samples using median-centered log2-transformed RPPA expression value data for proteins deemed significant in at least one cell line. Proteins with a Storey FDR < 0.05 and a log-2 ratio > 1 were considered significant. B) Lysates from UM001, OMM1.3, and OCM3 cells were immunoblotted for proteins and phosphoproteins that were most significantly up- or downregulated by YM treatment. Blots are representative of 3 independent experiments.

IGF1 partially rescues cell proliferation, cell cycle, and apoptosis of YM treated UM cells

To determine if the IGF1 pathway can provide resistance to G_q/11 inhibition in UM cells, we first performed a cell proliferation assay in which OMM1.3 and OCM3 cells were treated with YM in complete media with or without IGF1 added (Fig. 3A). We observed increased proliferation of untreated OMM1.3 cells with the addition of IGF1 and a partial rescue of OMM1.3 cell proliferation in the presence of IGF1 under YM treatment conditions. OCM3 cell proliferation appeared to slightly increase with the addition of IGF1, however, no proliferative rescue was observed as OCM3 cells are unaffected by YM treatment. Next, we investigated the effect of IGF1 on cell cycle of YM treated UM cells via EdU incorporation analyzed by flow cytometry (Fig. 3B). In OMM1.3, 92.1, and UM001 cell lines IGF1 significantly increased the EdU incorporation of YM treated cells, suggesting IGF1 provides resistance to G_q/11 inhibition. Additionally, we made similar observations regarding the expression of cleaved caspase-3 (Fig. 3C). Under serum free conditions, IGF1 alone was sufficient to significantly reduce caspase-3 cleavage of YM treated UM cells. These results suggest IGF1 provides protection from cell cycle arrest and apoptosis caused by YM treatment in UM cells and provides further rationale for targeting IGF1R in addition to G_q/11.

Figure 3:

YM effects on UM cell proliferation, cell cycle, and apoptosis are partially rescued by IGF1. A) OMM1.3 and OCM3 cells were treated with YM and additional IGF1 added to complete media for 3 days. Relative cell number was detected by fluorescent DNA intercalator, PicoGreen. Error bars represent the mean ± SEM from 3 independent experiments done in triplicate. B) 92.1 and OMM1.3 cells were treated with 30 nM YM and UM001 cells were treated with 3 nM YM for 24 hours with or without the addition of 50 ng/ml IGF1. EdU incorporation was detected by FACS. Error bars represent the mean ± SEM from 3 independent experiments. C) UM cells were treated with 1 μM YM in serum free media with or without 50 ng/ml IGF1 for 24 hours. Cell lysates were immunoblotted for cleaved caspase-3 and GAPDH. Blots are representative of 3 independent experiments. P < 0.005 (**), P < 0.0005 (***), P < 0.0001 (****), One-way ANOVA.

G_q/11 and IGF1R inhibition synergistically affects cell cycle and proliferation of UM cells

Using the IGF1R/IR selective inhibitor, linsitinib (OSI-906) (48) and YM, we next investigated the effect of the combination treatment on UM cell proliferation (Fig. 4A). Linsitinib acts by inhibiting the autophosphorylation of IGF1R (48). Cells were treated with a combination of the two drugs up to 1 μM each for three days and the relative number of cells was normalized to that of untreated cells (left side matrix of each panel in Fig. 4A). These data were used to determine synergy scores of each combination (right side matrix of each panel in Fig. 4A). The combination treatment synergically inhibited the cell proliferation of 92.1, OMM1.3, and UM001 cells, while neither drug alone nor in combination had any effect on OCM3 cells. Linsitinib alone decreased cell proliferation of OMM1.3 and UM001 cells, however, not to the extent of YM alone. 92.1 cells were unaffected by IGF1R inhibition alone. We also observed that the addition of linsitinib inhibited IGF1 resistance to YM in UM cells and that the combination treatment was significantly more effective at inhibiting cell cycle than the single agents in 92.1 and OMM1.3 cells (Fig. 4B). These data suggest that combined inhibition of G_q/11 and IGF1R acts synergistically to prevent UM cell growth.

Figure 4:

IGF1R/IR inhibitor, linsitinib, acts synergistically with YM on cell proliferation and cell cycle. A) Synergy plots of UM cells treated with linsitinib and YM. Left panels show relative cell number as a percent of untreated control cells (100%) and the right panels show the corresponding Loewe synergy plots created using Combenefit software. B) UM cells were treated with DMSO, linsitinib, YM, or YM + linsitinib for 24 hours in complete media with or without the addition of 50 ng/ml IGF1. EdU incorporation was detected via FACS. Note that some of the data from Fig. 3B are replotted here to enable a better comparison of all conditions. Error bars represent the mean ± SEM from 3 independent experiments. P < 0.05 (*), P < 0.005 (**), P < 0.0005 (***), P < 0.0001 (****), One-way ANOVA.

Synergy assays combining YM and picropodophyllin (PPP), an IGF1R selective inhibitor that shows no activity towards the insulin receptor, or alpelisib (BYL719), a PI3K (downstream effector of IRS2) inhibitor approved for breast cancer (49), further suggest the inhibition of the IGF1 pathway works synergistically with YM treatment to inhibit UM cell proliferation (Fig. S4). While an apparent synergistic decrease in cell proliferation was also observed in OCM3 cells treated with YM and PPP, this was not entirely unexpected as PPP was previously shown to inhibit the growth of OCM3 and other BRAF-driven melanomas (34,50). Additionally, cell proliferation was decreased to a more drastic extent in OMM1.3 and UM001 cells. The minimal synergistic effects of YM and BYL179 seen in OCM3 are negligible as cell proliferation is decreased at most by only about 6%. Together, these data highlight the particular susceptibility of G_q/11-driven UM to the inhibition of the IGF1 pathway and the enhanced effects in combination with YM treatment.

G_q/11 and IGF1R inhibition combinatorially reduces UM oncogenic signaling

To further investigate how the combined treatment may affect UM cell growth and viability, we next evaluated potential changes in downstream oncogenic signaling in UM cells. 92.1, OMM1.3, and UM001 cells were treated with DMSO, linsitinib, YM, or a combination of YM and linsitinib for 24 hours in serum free media, IGF1 supplemented media, or FBS supplemented media. Cell lysates were immunoblotted for various relevant proteins (Fig. 5) and protein levels were quantified (Fig. S5–S7). All three cell lines express IGF1R, which is phosphorylated when IGF1 or FBS is present and is inhibited by the addition of linsitinib. IRS2 expression increases with YM treatment as observed in Fig. 2, particularly when IGF1 is abundant. AKT and FAK phosphorylation is combinatorially downregulated with the drug combination, while ERK activation is only affected by YM treatment. This is not surprising as it was previously determined that IGF1 signals specifically through the AKT pathway in UM (35). Looking further downstream of IGF1R and AKT activation, we observed a combinatorial downregulation of the phospho-S6 ribosomal subunit, particularly at the S240/44 residue. Interestingly, instead of simply promoting the phosphorylation/inactivation of YAP and TAZ as expected, the combination treatment led to a decrease in the total protein expression. Additionally, we observed a significant increase of cleaved caspase-3 with the combination treatment compared to either single agent treatment. These data suggest that the combination treatment of YM and linsitinib act combinatorially to downregulate proteins involved in UM pathogenesis and promote apoptosis of UM cells.

Figure 5:

YM in combination with linsitinib shows synergistic effects on downstream signaling and apoptosis of UM cells. Cells were treated with DMSO (D), linsitinib (L), YM (Y), or YM + linsitinib (YL) in serum free (S.F.), serum free + 50 ng/ml IGF1, or FBS conditioned media for 24 hours. Cell lysates were immunoblotted for proteins and phosphoproteins that have been implicated in UM. Blots are representative of 3 independent experiments.

YM and linsitinib combination treatment synergistically inhibits tumor growth in vivo

We next tested the combination of YM with linsitinib in our metastatic UM mouse model. UM001-tdTomato cells were injected into NSG mouse livers and allowed to form tumors over five weeks. The mice were then treated with vehicle control, 0.4 mg/kg YM, 40 mg/kg linsitinib, or a combination of the drugs for two weeks. CT scans were taken pre-treatment and post-treatment to determine tumor growth (Fig. 6A). Tumors of mice treated with YM or linsitinib alone showed no significant difference in tumor growth compared to the vehicle control mice, while tumor growth of the combination treated mice was significantly reduced (Fig. 6B). Unfortunately, each treatment condition had observed toxicity including jaundice with linsitinib and severe inactivity after YM administration. We, therefore, repeated the experiment using lower doses (0.3 mg/kg YM and 25 mg/kg linsitinib) to reduce toxicity. CT scans were taken pre- and post-treatment (Fig. 6C). Again, tumors of mice treated with YM or linsitinib alone showed no difference in tumor growth compared to the vehicle control mice, while tumor growth of the combination treated mice was significantly reduced even at these lower doses (Fig. 6D). Nevertheless, each treatment condition still resulted in lethal toxicity. After the two-week treatment, tumors were excised, fixed and processed for IHC staining of phosphorylated histone H3 (pHH3), a marker of proliferation, which only expresses during mitosis and late G2 phase (51) and phosphorylated S6 ribosomal subunit (pS6), a downstream effector of the IGF1/AKT pathway (Fig. 6E). There were significantly fewer pHH3-positive cells in each treatment condition alone, but the combination treatment resulted in the fewest number of pHH3-positive cells. The combination treatment also resulted in fewer pS6-positive cells compared to vehicle and single agent treatments. This suggests that while we were unable to avoid toxicity, the combination treatment can successfully reduce tumor growth and allow for a lower dose of YM to be effective. Taken together, these data suggest that combined inhibition of IGF1R and G_q/11 may be a suitable therapeutic strategy if toxicity can be minimized.

Figure 6:

Combination treatment of YM and linsitinib synergistically inhibits tumor growth in a hepatic-metastasis mouse model. A) Representative CT scans of mouse livers containing UM tumor pre-treatment and post 2-week treatment. UM001-tdTomato cells were injected into mouse livers and formed tumors over 5 weeks. The mice were treated with 0.4 mg/kg YM and/or 40 mg/kg linsitinib. Tumors are circled. L = liver, G = gall bladder, S = spleen. B) Tumor growth fold change compared to vehicle control over 2 weeks. C) Representative CT scans of mouse livers containing UM tumor pre-treatment and post 2-week treatment. Mice were treated with 0.3 mg/kg YM and/or 25 mg/kg linsitinib. Tumors are circled. L = liver, S = spleen. D) Tumor growth fold change compared to vehicle control over 2 weeks. E) Tumors were excised from mice after 2-week treatment and fixed. Ten areas of tumor tissues were selected for IHC staining of phosphorylated histone H3 (pHH3) and phosphorylated S6 (pS6). F) Quantification of pHH3-positive cells under each treatment condition. P < 0.05 (*), P < 0.0001 (****), Mann-Whitney test.

Discussion

Uveal melanoma is the most common intraocular tumor in adults and has a high rate of metastasis. This leaves patients with a short survival time as UM mainly metastasizes to the liver, and there remains a lack of suitable therapeutic options (52,53). UM is predominantly driven by activating mutations in Gα_q/11 proteins, which promotes increased signaling in various cell proliferative pathways, such as the MAPK, AKT, and FAK/YAP pathways (10–17). To date, targeting proteins downstream of G_q/11 has yielded unsuccessful results in clinical trials (53). Because UM is largely defined by mutational activation of the G_q/11 pathways and harbor low mutational burden, it stands to reason that direct inhibition of Gα_q/11 may be a highly effective treatment strategy for most patients with oncogenic G_q/11-driven UM. Thus far, current studies support this hypothesis (17,25,27,30,31).

Here we show that inhibition of G_q/11 in UM inhibits further cell proliferation through cell cycle arrest, decreases MAPK, FAK and YAP signaling, and promotes apoptosis. This suggests G_q/11 inhibition may be a promising treatment strategy. To this end, recent studies have shown YM and FR can significantly reduce tumor size in UM subcutaneous xenograft models (29–31). However, these models are limited by the use of primary or metastatic cell lines in subcutaneous tissue. Considering the hepatic tropism of UM, an orthotopic hepatic tumor model is preferred to investigate treatment efficacies in the liver microenvironment. Here we show the first instance of using such a model to test the efficacy of a G_q/11 inhibitor in treating metastatic UM. To illustrate the importance of using such a model, we identified the IGF1 pathway as a potential resistance mechanism of G_q/11 inhibition. IGF1 production is specific to the liver while it has also been shown that UM cells can produce and secrete their own IGF1 (35,46,47). Previous studies have demonstrated that IGF1 signaling through IGF1R might be a key mechanism for growth of uveal melanoma in the liver (35). We determined that IGF1R inhibition enhances the effects of G_q/11 inhibition on oncogenic signaling in UM and metastatic UM tumor growth in the liver.

While our orthotopic hepatic tumor model is more biologically relevant than subcutaneous models, it still differs from the human disease. Unlike the appearance of several individual nodules or miliary infiltration in the liver of metastatic UM commonly seen in humans (54), we observed one large tumor develop in the liver of the mice, likely as a result of injecting UM cells in a single area of the liver. Based on the effectiveness of YM in our 3D Matrigel studies, it is reasonable to believe that YM can more effectively reduce the proliferation of several small individual nodules than that of a larger single tumor so long as the drug is distributed evenly throughout the liver. It is likely easier for drugs to penetrate smaller, less dense tumors than the ones observed in our model. In addition, since our mouse studies are limited to a single UM cell line, future work using PDX models of metastatic uveal melanoma would be of value (55).

Using this model, we observed combined inhibition of oncogenic G_q/11 and IGF1R by YM and linsitinib, respectively, significantly reduced UM tumor growth. The effects observed are likely tumor-cell specific and not a result of host tissues being affected. Mice treated with the closely related and more potent FR compound showed no evidence of anemia, platelet deficiency, or liver dysfunction (29). It has been suggested that YM has a high rate of metabolic degradation and is considered a high clearance compound in the liver, while intratracheal administration may be preferable to IP for FR (56). In addition, no severe side effects were reported with high doses of YM, at least in C57BL/6J and CB17-SCID mice (30). Moreover, linsitinib is largely well tolerated in clinical trials (57–59), and does not affect kidney or liver function in immunocompromised mice (60).

While combined treatment reduced tumor growth, our treatment strategy was not without its limitations as we observed toxic effects of YM in our NSG mice. This was not entirely unexpected as YM is a potent inhibitor of physiologically active wild-type Gα_q/11 and the mutationally active Gα_q/11 found in UM. Onken et al. recently determined an LD₅₀ of ~0.6 mg/kg for the closely related FR compound in NSG mice (29). Therefore, it is reasonable that at 0.5 mg/kg YM we observed mouse death as early as 1.5 weeks of treatment. Similarly, in our combination in vivo studies we observed mouse death within the 2-week treatment period at both 0.4 mg/kg and 0.3 mg/kg YM. On the other hand, Hitchman et al. were able to treat mice with YM concentrations 5–10 times higher than our maximum dose without observing severe side effects (30), suggesting that YM might not be as toxic as it appears. One potential reason for this might be differences in the mouse strains used. While we used NSG mice with severe immunodeficiency, Hitchman et al. used CB17-SCID mice from Taconic, which retain functional natural killer cells, macrophages, and granulocytes, unlike our NSG mice. Additionally, we observed toxic effects of linsitinib treatment. This was unexpected as others have treated NSG mice with similar concentrations of the drug over the same period of time without observed toxicity (61). Linsitinib has also been used in multiple clinical trials, suggesting it is safe (57–59). It may be possible that the presence of a tumor in the liver impairs the liver’s ability to process and metabolize these compounds, leading to toxic effects.

Additionally of note, we observed differences in the effect of YM on the expression of homologues, YAP and TAZ. The dephosphorylation of YAP/TAZ, which facilitates translocation to the nucleus to promote oncogenic gene transcription, is driven by constitutively active Gα_q/11 (14,15). Perturbation of G_q/11 signaling results in the phosphorylation of YAP, followed by its sequestration in the cytoplasm and eventual degradation (14). YAP and TAZ are believed to perform redundant functions, so it is curious that upon G_q/11 inhibition, TAZ expression but not YAP expression is significantly downregulated. We also observed the homologues are differentially expressed across UM cell lines. With the current disagreements over the importance of YAP signaling in UM pathogenesis (27,62,63), it may be important to include TAZ in future studies that are focused on the YAP pathway in UM, as most studies tend to only focus on YAP expression and function and fail to include analysis of TAZ. Additionally, we also observed that G_q/11 inhibition by YM causes MAPK downregulation at varying rates across different UM cell lines, in some cases this is a rapid process. While the mechanism of these differences is unclear, it has been shown that MAPK signaling is not durably suppressed by YM or FR, as seen by the beginnings of a rebound in pERK levels within 24 hours after inhibitor treatment (Fig. 1) (29,30).

Our results suggest that current G_q/11 inhibitors are required at potentially toxic doses to significantly inhibit metastatic tumor growth in the liver. It should be noted that inhibition of oncogenic G_q/11 requires higher concentrations of YM compared to that of receptor activated wild type G_q/11 (30). This is likely due to the defective nature of GTP hydrolysis of the oncogenic protein, resulting in more GTP-bound protein that may be less effectively inhibited by YM as the drug acts by stabilizing the GDP-bound protein. Given the many physiological roles of G_q/11, local administration or a targeted delivery approach would be desired and may allow YM to be delivered at higher doses to promote durable arrest without encountering substantial dose-limiting toxicity. Another strategy, as we have outlined, would be to combine a lower dose of YM with an additional treatment that works to enhance the effects of YM, such as an IGF1R inhibitor. It has become apparent that the liver microenvironment may be capable of overcoming even the widespread inhibitory effects on UM by YM as a monotherapy. Directly inhibiting oncogenic G_q/11 at the very least may provide a window to treat patients with an additional therapeutic.

Abbreviations:

BYL719

Alpelisib

CM

cutaneous melanoma

CT

computerized tomography

DMSO

dimethyl sulfoxide

FAK

focal adhesion kinase

FR

FR900359

HGF

hepatocyte growth factor

IGF1

insulin like growth factor 1

IGF1R

insulin like growth factor 1 receptor

IRS2

insulin-like receptor substrate 2

MAPK

mitogen-activated protein kinase

MEK

mitogen-activated protein kinase kinase

NSG mice

NOD.Cg-Prkdcscid Il2rgtm1WjI/SzJ mice

PLCβ

phospholipase C β

PPP

Picropodophyllin

RPPA

reverse phase protein array

UM

uveal melanoma

YAP

yes associated protein

YM

YM-254890