Combination Therapy with c-Met and Src Inhibitors Induces Caspase-Dependent Apoptosis of Merlin-Deficient Schwann Cells and Suppresses Growth of Schwannoma Cells

Neurofibromatosis type 2 (NF2) is a genetic disorder characterized by the development of vestibular schwannomas (VS) and meningiomas. NF2 is caused by mutations in the NF2 gene encoding the tumor suppressor merlin (1, 2). Loss of merlin function in Schwann cells leads to aberrant signaling in molecular pathways involved in cell survival and proliferation, resulting in schwannoma formation (3).

VS treatment relies on surgical removal of tumors, frequently causing nerve damage. A less invasive option, stereotactic radiosurgery, controls tumor growth with hearing preservation rates consistently approaching approximately 20% to 44% at 10 years (4, 5). However, the rate of malignant transformation or secondary malignancies following radiation for NF2 is estimated to be approximately 5%, representing a seven-fold increase compared with NF2 patients without irradiation (6). Clinical trials with off-label use of FDA-approved drugs, such as lapatinib and everolimus (RAD001), have shown moderate success in slowing NF2 schwannoma growth (7, 8). More success has been achieved with bevacizumab, a mAb against the VEGF-A (9). Bevacizumab promotes VS tumor shrinkage and hearing response in approximately 40% to 50% of NF2 patients (10), but has adverse side-effects, including hypertension and proteinuria, and amenorrhea in women (11, 12). Thus, it is imperative to identify drugs suitable for prolonged treatment in NF2 patients. Identifying molecular targets that allow re-purposing of FDA-approved drugs would accelerate development of NF2 therapies.

Microarray and qPCR analysis revealed that expression of c-Met, a receptor tyrosine kinase (RTK), is elevated in human VS compared with nerves or Schwann cells (13, 14). Additionally, hepatocyte growth factor (HGF), the c-Met ligand, is a potential NF2 biomarker (15). Merlin directly interacts with HGF-regulated tyrosine kinase substrate (HRS), supporting a role of merlin in c-Met/HRS signaling (16). The loss of merlin function could contribute to aberrant c-Met signaling, leading to schwannoma formation. C-Met activates several cell proliferation and survival signaling pathways and is mitogenic for Schwann cells (17, 18). Cabozantinib (XL184, Cabometyx™, Exelixis) is a small-molecule inhibitor of c-Met and other RTKs involved in cell motility and metastasis, and inhibits VEGFR2-dependent angiogenesis (19). It is FDA-approved for treatment of tumors with elevated c-Met and HGF expression, such as medullary thyroid cancer (20) and renal cell carcinoma (21). Cabozantinib also reduces proliferation of NF1-associated malignant peripheral nerve sheath tumor (MPNST) cells in vitro and suppresses tumor growth in mouse models (22, 23).

Src-family kinases are another promising drug target for NF2-associated schwannomas. Compared with normal Schwann cells, VS have increased levels of phosphorylated Src and focal adhesion kinase (FAK), resulting in deregulation of cell proliferation pathways (24). Dasatinib (SPRYCEL^(R), Bristol-Myers Squibb) is a type I, ATP-competitive Src/Abl kinase inhibitor. It is FDA-approved for chronic myeloid leukemia and acute lymphocytic leukemia with mutant Abl kinase expression (25, 26) and reduces growth of both solid tumors and blood cancers (27). In tumor cell lines, dasatinib is a cytostatic agent that inhibits cell proliferation, invasion, and metastasis (28). Similar to dasatinib, saracatinib (AZD0530, AstraZeneca) is another type I, ATP-competitive Src/Abl inhibitor, but binds the inactive Src conformation (29). Preclinical studies confirm the ability of saracatinib to reduce proliferation and migration/invasion of tumor cells (30). Moreover, Src inhibition by saracatinib enhances the sensitivity of gastric tumors to c-Met inhibitors (31), supporting the use of Src and c-Met inhibitors in combination.

In this study, we identified a potential combination therapy for NF2-associated schwannomas. Although cabozantinib and saracatinib each promoted G₁ cell-cycle arrest, their combination induced apoptosis and suppressed the growth of merlin-deficient Schwann cell allografts as well as primary human VS cells. Our results support targeting c-Met and Src kinases as a potential treatment for NF2-associated schwannomas.

Wild-type (WT) and merlin-deficient (MD) mouse Schwann cells (MSC) were generated and authenticated as previously described (ref. 32; MD-MSCs created in-house in 2010) and routinely tested for Mycoplasma contamination (LookOut Mycoplasma PCR Detection Kit; Sigma). VS were obtained according to the Institutional Review Board-approved human subject protocol with patient informed consent at University of Miami Miller School of Medicine through the Tissue Bank Core Facility and were used to prepare primary VS cultures (33). A description of the genetic analysis of human VS samples and the generation of human SC lines is provided in Supplementary Methods.

MD-MSCs were grown in six-well CellBind plates (Corning) at 200,000 cells/well for 24 hours and then received transduction growth media with 8 μg/mL of polybrene. Lentiviral luciferase particles [prepared from pLenti PGK V5-LUC Neo (Addgene #21471)] were added at five multiplicity of infection (MOI) for 18 hours, followed by growth media for 24 hours prior to selection media containing 1 μg/mL puromycin. The same protocol was followed for the lentiviral shRNA knockdown of c-Met (Sigma-Adrich catalog no. SHCLNV-NM_008591- TRCN0000023529 for c-Met shRNA and catalog no. SHC202V for scrambled shRNA viruses).

Cabozantinib, saracatinib (Selleckchem and MedChemExpress), and dasatinib (Selleckchem) were dissolved in DMSO (10 mmol/L stock) for in vitro experiments. MD-MSCs and WT-MSCs were seeded in a 384-well CellBind plate at 1,000 and 3,250 cells/well, respectively, and treated with drug for 48 hours, followed by paraformaldehyde fixation and Hoechst dye staining. Images were acquired with Image Express (micro-automated microscope) and analyzed using the Definiens platform (Definiens, Inc.).

MD-MSCs were seeded in 384-well CellBind plates at 2,500 cells/well in phenol red-free (prf) growth media. WT-MSCs were seeded at 15,000 to 20,000 cells/well in 96-well plates coated with poly-l-lysine (200 μg/mL) and laminin (25 μg/mL). After attachment, cells were treated with drugs or DMSO for 48 hours. Cell viability was determined using the CellTiter-Fluor Assay (Promega). Primary VS cells were seeded in a 96-well plate at 5,000 cells/well. After 24 hours, cells were treated with drugs or DMSO for 48 hours. Cell viability was determined with a crystal violet assay as previously described (32). For caspase activation assays, MD-MSCs were seeded in 384-well CellBind plates at 5,000 cells/well in prf growth media and were treated with drugs for 18 hours. The Apo-ONE Homogeneous Caspase 3/7 Assay (Promega) was used per the manufacturer's instructions with fluorescence measurement using an H1 Synergy plate reader.

MD-MSCs were seeded as described for cell viability assays. Cells were treated for 24 hours with drugs and/or inhibitors of apoptosis: Fas(human):Fc(human) (ALX-522-002, Enzo Life Sciences), Z-DEVD-FMK Caspase-3 Inhibitor, Z-LEHD-FMK Caspase-9 Inhibitor (#550378, #550381, BD Pharmingen), and Z-IETD-FMK Caspase-8 Inhibitor (#FMK007, R&D Systems). The inhibitors were dissolved at 20 mmol/L in DMSO. Fas:Fc was dissolved at 1 mg/mL in water. A cross-linking enhancer (#ALX-203-001, Enzo Life Sciences) was used with the Fas:Fc treatment. Cell viability was measured using the CellTiter-Fluor assay.

Genomic DNA was isolated and purified using Trizol (Invitrogen). Samples were subjected to NF2 multiplex ligation-dependent probe amplification (MLPA) using the NF2 MLPA Kit (MRC-Holland) per the manufacturer's instructions.

MD-MSCs were grown in six-well CellBind plates to approximately 80% confluence and treated with drugs or DMSO for 19 hours. Cells were harvested with 0.05% Trypsin and resuspended in 1 mL of HBSS/million cells. The Violet Ratiometric Assay (Invitrogen) was used according to the manufacturer's instructions. Cell populations were measured with a Cytoflex (Beckman Coulter) flow cytometer and analyzed with CytExpert (Beckman Coulter) software.

Western blots were performed as previously described (34). Cells were extracted with 4% SDS, 0.01% bromophenol blue, 10% glycerol, and 100 mmol/L dithiothreitol. The following antibodies were used: anti-cleaved caspase 3 (17 to 19 kDa), total caspase 3 (35 kDa), p-Src (Y416), Src (60 kDa), p-FAK (Y576), p-c-Met (Y1234/Y1235), β-actin (45 kDa), p-VEGFR2/3 (230 kDa), p-ERK1/2, ERK1/2(42–44 kDa), p-Akt (T308), Akt (60 kDa), cyclin D₁(36 kDa), and P27^Kip1 (27 kDa) from Cell Signaling Technology; FAK (125 kDa) and paxillin (68 kDa) from BD Biosciences; Fas (45 kDa) from Santa Cruz Biotechnology; p-paxillin (Y118) and c-Met(145 kDa) from Thermo Fisher Scientific. Primary antibodies were prepared in 1:1 Odyssey Blocking Buffer (TBS-0.1% Tween) and incubated overnight at 4°C or for 1 hour at room temperature. Secondary antibodies were prepared similarly but with 0.02% SDS and incubated for 45 to 60 minutes in the dark at room temperature. Western blots were quantified using the Odyssey System (LI-COR Biosciences).

NSG (NOD.Cg-Prkdcscid Il2rgtmlWjl/SzJ) mice were cared for as approved by the University of Central Florida (UCF) Institutional Animal Care and Usage Committee. The animal facility at UCF is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Three mice were dosed by oral gavage (20 mg/kg dasatinib, 25 mg/kg saracatinib, or 40 mg/kg cabozantinib in a vehicle containing 1.25% polyethylene glycol, 2.5% Tween-80, and 5% DMSO) and one mouse received vehicle only. At the indicated times, mice were sacrificed. Blood and sciatic nerves were collected and analyzed by mass spectrometry at Sanford Burnham Prebys Medical Discovery Institute (Lake Nona, FL).

Male and female NSG mice were bred in house and used at 6 to 8 weeks of age. Following anesthesia with isoflurane (1–3% in oxygen), the right sciatic nerve was injected with 5,000 luciferase-expressing MD-MSCs in 3 μL of L-15 prf media. Wounds were sealed with Vetbond. Mice were imaged with an In Vivo Imaging System (IVIS, Caliper) 15 minutes after intraperitoneal injection of luciferin (150 mg/kg) to confirm engraftment, were assigned to treatment groups (Day 0) and began daily treatment with the vehicle (as above), saracatinib (25 mg/kg), cabozantinib (12.5 mg/kg), or a combination of saracatinib and cabozantinib (25 and 12.5 mg/kg, respectively). Mice were imaged with IVIS every 7 days and were sacrificed at day 14. Tumor grafts and contralateral sciatic nerves were removed, weighed and photographed, followed by fixation overnight in 4% paraformaldehyde and storage in 30% sucrose/0.02% azide.

Fixed grafts and nerves were embedded in paraffin and cut into 4-μm sections. Antigen retriever reaction was performed by heating sections in 0.1 mol/L sodium citrate, pH 6.0 in a pressure cooker for 30 minutes. To block endogenous peroxidase activity, sections were treated with 3% hydrogen peroxide for 15 minutes. Following blocking (Super Block solution, ScyTek Lab), sections were incubated with primary antibody at 4°C overnight. After washing, an UltraTek anti-polyvalent biotinylated secondary antibody (ScyTek Lab) was added to the sections for 10 minutes, followed by serial treatment with UltraTek HRP (ScyTek Lab), an AEC substrate, and hematoxylin counterstaining. Stained sections were mounted with Immu-mount (Thermo Fisher Scientific) and photographed under a Nikon Eclipse microscope. Images were taken from multiple areas in each section, and representative images are shown.

Statistical analysis was performed using GraphPad Prism 6. Comparisons were made using ANOVA with Bonferroni posttest or Kruskal–Wallis test with Dunn's comparison. For mouse study, statistical analysis was performed using SAS version 9.4 (SAS Institute Inc. SAS/STAT 9.4 User's Guide, SAS Institute Inc.; 2014) and GraphPad Prism 6. Bonferroni adjustments were used when performing multiple comparisons and significance was fixed at the 5% level.

We performed a high-content drug screen to assess the effect of c-Met inhibitors on proliferation of MD-MSCs. All inhibitors tested reduced proliferation of MD-MSCs after 48 hours (Supplementary Fig. S1). We further evaluated cabozantinib, a c-Met/VEGFR2 inhibitor, because it is FDA-approved and would target Schwann cells and the vasculature. Cabozantinib significantly reduced viability of MD-MSCs after 48 hours of treatment, with IG₅₀ of 2.2 μmol/L, an 85% maximum effect, and five-fold selectivity over WT-MSCs (IG₅₀ = 10 μmol/L; Fig. 1A). MD-MSCs treated with increasing cabozantinib concentrations had a dose-dependent decrease in c-Met (Y1234/1235) phosphorylation (Fig. 1B), but no change in the VEGFR2/3 phosphorylation levels (Fig. 1C). Knockdown of c-Met in MD-MSCs decreased their growth rate by approximately 30% over 72 hours (Fig. 1D), confirming that c-Met contributes to enhanced MD-MSC proliferation. Downstream of c-Met, cabozantinib decreased the phosphorylation levels of Erk1/2 and Akt(T308) in MD-MSCs within 6 hours, and the effect was maintained at 24 hours (Fig. 1E and F). These changes were accompanied by decreased cyclin D₁ and increased p27 levels at 24 hours of treatment (Fig. 1G), consistent with a G₁ cell-cycle arrest (35). To assess suitability for in vivo testing, we performed a pharmacokinetic analysis of cabozantinib in NSG mice. Plasma and nerve concentrations of cabozantinib peaked at 4 hours (6,500 and 1,650 ng/ml, respectively) and were detectable for at least 17 hours with a t_1/2 of approximately 7 hours following a single 40 mg/kg oral dose (Fig. 1H).

High-content screening of drug combinations identified that combination treatment of cabozantinib with the Src inhibitor, dasatinib, selectively reduced the number of MD-MSCs compared with WT-MSCs (Fig. 2A). Synergy scores revealed that cabozantinib and dasatinib synergized in MD-MSCs at low concentrations (Fig. 2B). In addition, the combination index (CI) of cabozantinib and dasatinib was 0.6 (Fig. 2C), indicating a synergistic relationship of the two inhibitors and supporting their use in combination for these studies.

Antibody array analysis revealed increased Src phosphorylation in human VS compared with normal Schwann cells (33), supporting Src as a potential target for NF2 therapeutics (24). The Src inhibitor, dasatinib, reduced MD-MSC viability with IG₅₀ = 9 nmol/L, a maximum effect of approximately 80%, and 100-fold selectivity over WT-MSCs (IG₅₀ = 1.45μmol/L; Fig. 3A). In addition to reducing Src(Y416) phosphorylation, dasatinib decreased Src-dependent phosphorylation of FAK(Y576) and Src/FAK-dependent phosphorylation of paxillin(Y118) in a dose-dependent manner (Fig. 3B). Dasatinib promoted a G₁ cell-cycle arrest in MD-MSCs, as evidenced by decreased cyclin D₁ and increased p27 levels at 24 hours of treatment (Fig. 3C). A pharmacokinetic analysis revealed that dasatinib (20 mg/kg oral dose) had poor nerve penetrance and it peaked in the plasma at 1 hour (200 ng/mL). Consistent with previous reports (36), the half-life (t_1/2) of dasatinib in plasma was approximately 2.5 hours (Fig. 3D). These results do not support in vivo evaluation of dasatinib.

A pharmacokinetic study of saracatinib (25 mg/kg oral dose) in NSG mice revealed that plasma and nerve concentrations of saracatinib peaked at 0.5 hours with a t_1/2 of 4.4 hours, but remained constant in the nerve at approximately 300–400 ng/g for at least 8 hours, the longest time point measured (Fig. 3D). Saracatinib selectively reduced MD-MSC viability (IG₅₀ = 0.3 μmol/L) but had a low maximum effect of approximately 40%–50%, compared with an 80% maximum effect for dasatinib (Fig. 3E). Saracatinib decreased the levels of Src(Y416) phosphorylation compared with total Src levels, and also decreased Src-dependent phosphorylation of FAK(Y576), and Src/FAK-dependent phosphorylation of paxillin(Y118) at 6 hours of treatment (Fig. 3F). After 24 hours of saracatinib treatment, MD-MSCs had increased p27 levels compared with controls (Fig. 3G).

MD-MSCs treated with saracatinib (0.5 μmol/L) together with increasing cabozantinib concentrations had a three-fold decrease in the IG₅₀ of cabozantinib (0.6 μmol/L, compared with 2.2 μmol/L for cabozantinib alone), and more than five-fold selectivity over WT-MSCs (IG₅₀ = 3.4 μmol/L; Fig. 4A). Treatment of MD-MSCs with saracatinib (0.5 μmol/L) and increasing cabozantinib concentration resulted in decreased phosphorylation levels of c-Met(Y1234/1235), Erk1/2, and Akt(T308) after 6 hours (Fig. 4B). After 24 hours of the combination treatment, MD-MSCs had decreased cyclin D₁ levels and increased p27 levels at lower cabozantinib concentrations than those treated with cabozantinib alone (Fig. 4C).

We investigated the efficacy of this drug combination in decreasing viability of primary VS cells obtained from two patients. VS01 was a sporadic VS with a heterozygous duplication of NF2 exon 5 and a homozygous duplication of exon 7. VS02 was from an NF2 patient with an exon 14 deletion. VS cells were cultured with cabozantinib (2 μmol/L) and saracatinib (2 μmol/L), alone or in combination, for 48 hours. For both VS01 and VS02, the combination treatment reduced VS cell viability by approximately 35%–40% compared with vehicle (0.3% DMSO) and was significantly more effective than saracatinib alone. In addition, for VS01 cells, the combination treatment was more effective than cabozantinib alone (Fig. 4D).

To evaluate the efficacy of cabozantinib and saracatinib in vivo, luciferase-expressing MD-MSCs were grafted into the sciatic nerves of NSG mice, and the graft size was monitored by bioluminescence imaging (BLI). Upon confirmation of successful grafting (Supplementary Fig. S2A), mice were assigned into vehicle, cabozantinib (12.5 mg/kg/day), saracatinib (25 mg/kg/day), and combination (cabozantinib and saracatinib at 12.5 mg/kg/day and 25 mg/kg/day, respectively) treatment cohorts. BLI showed that grafts in the combination-treated group had a significantly slower growth rate compared with those in the single-agent groups (Fig. 5A and B). Although the vehicle-treated allografts had a 160-fold increase in BL signal over 14 days, the grafts treated with saracatinib or cabozantinib had a 50- and 60-fold increase in BL signal, respectively. Significantly, the allografts from the combination group had only a 25-fold increase in BL signal after 14 days of treatment (Fig. 5C). The reduction in BL signals correlated with lower tumor weights in the combination-treated group compared with the vehicle group (Fig. 5D; Supplementary Fig. S2B).

To assess the signaling pathways modulated by drug treatments, allograft sections were analyzed by immunohistochemistry. Similar to the in vitro findings, saracatinib decreased the FAK phosphorylation levels, and cabozantinib decreased the ERK1/2 phosphorylation levels compared with vehicle controls (Fig. 5E). These changes were also observed in the combination-treated allografts. Moreover, CD31 staining of endothelial cells was decreased in grafts treated with cabozantinib compared with vehicle controls, indicating that cabozantinib targets the vasculature in vivo as well (Fig. 5E). Grafts from mice treated with cabozantinib or saracatinib, alone or in combination, had fewer cyclin D₁ and Ki67-positive cells compared with those from vehicle-treated mice. In addition, grafts in the combination-treated group had elevated cleaved caspase 3 staining, compared with those in the vehicle and single treatment groups (Fig. 5E), supporting the conclusion that the drug combination induces apoptosis of MD-MSCs in vivo.

To confirm the in vivo findings, we studied caspase-dependent apoptosis in cultured MD-MSCs. Caspase 3/7 activity was triggered by cabozantinib or saracatinib alone only when administered at high doses (3 μmol/L saracatinib or 10 μmol/L cabozantinib) for 19 to 24 hours (Fig. 6A). However, cotreatment of MD-MSCs with saracatinib (0.5 μmol/L) and increasing cabozantinib concentrations induced caspase 3/7 activity at 100-fold lower concentrations of cabozantinib compared with cabozantinib administered alone (0.1 μmol/L vs. 10 μmol/L; Fig. 6A). Similarly, MD-MSCs treated with cabozantinib (0.5 μmol/L) and increasing saracatinib concentrations induced caspase 3/7 activity at approximately 30-fold lower saracatinib concentrations compared with saracatinib administered alone (0.1 μmol/L vs. 3 μmol/L; Fig. 6A). A membrane asymmetry assay confirmed that the drug combination induced a larger apoptotic cell population than either drug alone (Fig. 6B). Treatment with 2 μmol/L saracatinib resulted in 6.2% apoptotic cells, 4.8% dead cells, and 88.4% live cells, whereas treatment with 2 μmol/L cabozantinib had 7.7% apoptotic cells, 9% dead cells, and 82.8% live cells. The combination treatment increased the apoptotic population to 20% after 19 hours (Fig. 6C). When administered alone, saracatinib did not induce cleavage of caspase 3. Cabozantinib induced caspase cleavage at 1μmol/L whereas in the presence of 0.5 μmol/L saracatinib, caspase cleavage was detected at 0.1 μmol/L cabozantinib (Fig. 6D). Collectively, our results indicate that although the individual drugs promote a G₁ cell-cycle arrest of MD-MSCs, saracatinib and cabozantinib combination is cytotoxic and promotes caspase 3/7-dependent apoptosis.

To examine the apoptotic pathway activated by dual inhibition of c-Met and Src, MD-MSCs were cultured in the presence of 2 μmol/L saracatinib and 2 μmol/L cabozantinib, alone and in combination, for 24 hours with or without inhibitors for caspase 3 (Z-DEVD-FMK), caspase 8 (Z-IETD-FMK), and caspase 9 (Z-LEHD-FMK). The three caspase inhibitors significantly prevented the loss of viability of MD-MSCs treated with cabozantinib and saracatinib together (Fig. 6E), suggesting that the combination treatment activates both the extrinsic and intrinsic apoptotic pathways. Incubation of MD-MSCs with the Fas ligand (FasL) decoy Fas:Fc partially prevented the loss of MD-MSC viability (Fig. 6E), supporting the conclusion that the Fas receptor is activated by the combination treatment, leading to activation of caspases and cell death. Furthermore, basal expression of Fas receptor was higher in MD-MSCs compared with WT-MSCs (Supplementary Fig. S3), suggesting a possible explanation for drug selectivity.

Individuals with NF2 develop multiple meningiomas and schwannomas over their lifetime. Although there is a low incidence of malignancy, repetitive surgeries to remove tumors increase morbidity and considerably decrease a patient's lifespan (10). Removal or irradiation of bilateral VS that occur in all NF2 patients can lead to deafness and facial nerve paralysis. This outcome in particular leads to social isolation and decreased quality of life. Currently, there is no approved pharmacologic treatment for NF2 tumors; however, patients are increasingly treated off-label with cancer drugs. The majority of drug therapies in clinical use for NF2 are cytostatic or target the tumor vasculature (7–9).

To date, only an HDAC and PI3K inhibitors have been found to induce apoptosis of MD-MSCs and human VS cells (32, 37, 38). A 3-phosphoinositide-dependent protein kinase-1 (PDK-1) inhibitor, OSU-03012, induced caspase-9–dependent apoptosis in NF2-associated Schwann cells and malignant schwannomas (38). Our screen of the Library of Pharmaceutically Active Compounds (LOPAC) revealed that multiple PI3K inhibitors reduced viability of MD-MSCs by inducing caspase-dependent apoptosis and autophagy (32). Currently, there are over 300 ongoing clinical trials of PI3K inhibitors due to their relevance to oncology (clinicaltrials.gov). However, only one PI3K inhibitor, Idelalisib (Zydelig, Gilead Sciences) has been FDA-approved (39). This drug, however, carries a black box warning for adverse effects (40). Recently, six clinical trials of drug combinations with Idelalisib have been halted due to toxicity (41).

Here, we demonstrate that simultaneous inhibition of c-Met and Src with cabozantinib and saracatinib, respectively, reduced the viability of MD-MSCs in vitro and in nerve allografts by inducing caspase-dependent apoptosis. Moreover, the drug combination inhibited growth of primary VS cells carrying genetic inactivation of NF2. Treatment of MD-MSCs with cabozantinib reduced activation of ERK1/2 and Akt signaling pathways downstream of c-Met. In vivo, cabozantinib also reduced microvessel density. Both dasatinib and saracatinib inhibited Src activity in MD-MSCs as evidenced by reductions in phosphorylated FAK and paxillin. FAK signaling is elevated in NF2-associated VS compared with normal nerves (24). A previous study showed that FAK inhibition with crizotinib reduced allograft growth in mice (42). Merlin modulates activity of the β1 integrin pathway activated by extracellular matrix and transduced by Src and FAK (43, 44). This pathway mediates extracellular matrix-dependent cell survival, and converges with proliferative signals from mitogen receptor-dependent activation of ERK1/2 and PI3K/Akt (3). The absence of merlin activates these pathways, leading to enhanced cell survival and proliferation. Simultaneous inhibition of c-Met-ERK1/2-Akt and Src-FAK-paxillin could prevent schwannoma cells from using a compensatory mechanism of adhesion-dependent cell survival.

Inhibition of these two pathways which converge on PI3K/Akt in Schwann cells may contribute to induction of apoptosis. Akt modulates transcription factors that regulate cell cycle and apoptosis (45), and may mediate the cytostatic and cytotoxic effects observed in cabozantinib and saracatinib-treated MD-MSCs. Our results suggest that the efficacy of the combination treatment is due, in part, to activation of the Fas/FasL cell death pathway. Inhibition of Fas with a FasL decoy partially prevented cell death induced by the drug combination. We observed activation of both the extrinsic and intrinsic apoptotic pathways downstream of Fas receptor, as evidenced by the ability of caspase 8, 9, and 3 inhibitors to prevent MD-MSC death caused by the drug combination. This could be attributed to decreased pro-survival and pro-proliferative signaling involving ERK and Akt, leading to upregulation of FasL (46). In addition, stimulation of the Fas receptor by FasL activates both the extrinsic and intrinsic apoptotic pathways, resulting in cytochrome c release from mitochondria and apoptosome formation with caspase-9 (47). Merlin loss in SCs leads to accumulation of multiple receptors in the plasma membrane (48), potentially explaining the elevated basal levels of Fas receptor in MD-MSCs compared with WT-MSCs. A proposed pathway summarizing our findings is shown in Fig. 6F.

This study demonstrates that simultaneous c-Met and Src inhibition induced apoptosis of merlin-deficient mouse Schwann cells in culture and in allografts, and reduced growth of primary VS cells with NF2 mutations. The drug combination also decreased the viability of two merlin-deficient human SC lines and a human benign meningioma line in vitro (Supplementary Fig. S4), confirming that the effects are not cell-line or species-dependent. Both ERK and Src-FAK kinases converge onto the PI3K/Akt pathway; inhibitors of which induce apoptosis in NF2-associated cells. Considering the severe toxicity associated with direct targeting of PI3K with Idelalisib, and until additional PI3K inhibitors are approved, simultaneous inhibition of c-Met-ERK and Src-FAK pathways with cabozantinib and dasatinib/saracatinib may promote schwannoma cell apoptosis with fewer adverse effects. This study provides preclinical data supporting further investigation of dual inhibition of c-Met and Src as a potential therapy for NF2-associated schwannomas.

C. Fernandez-Valle is an unpaid research advisory board member for the Children's Tumor Foundation. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.A. Fuse, C.T. Dinh, R. Mittal, J.N. Soulakova, X.Z. Liu, L.-S. Chang, M.C. Franco, C. Fernandez-Valle

Development of methodology: M.A. Fuse, R. Mittal, L.-S. Chang, M.C. Franco, C. Fernandez-Valle

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.A. Fuse, S.K. Plati, S.S. Burns, C.T. Dinh, O. Bracho, D. Yan, R. Mittal, R. Shen, L.-S. Chang, M.C. Franco, C. Fernandez-Valle

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.A. Fuse, S.K. Plati, S.S. Burns, C.T. Dinh, D. Yan, R. Mittal, R. Shen, J.N. Soulakova, A.J. Copik, X.Z. Liu, L.-S. Chang, M.C. Franco, C. Fernandez-Valle

Writing, review, and/or revision of the manuscript: M.A. Fuse, S.K. Plati, C.T. Dinh, R. Mittal, J.N. Soulakova, X.Z. Liu, F.F. Telischi, L.-S. Chang, M.C. Franco, C. Fernandez-Valle

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.A. Fuse, S.K. Plati, R. Mittal, M.C. Franco, C. Fernandez-Valle

Study supervision: C.T. Dinh, X.Z. Liu, C. Fernandez-Valle

We thank Dr. Jacques Morcos for harvesting human VS tumors and Cenix Bioscience for conducting the c-Met inhibitor screen and synergy studies through the Industry-Academia collaboration program.

This study was supported by the Children's Tumor Foundation (to C. Fernandez-Valle and M.A. Fuse, YIA 2015-01-012), DOD NF140044 (to C. Fernandez-Valle), and DOD NF150080 (to L.-S. Chang).

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.