Co-targeting Phosphoinositide 3-Kinase and Focal Adhesion Kinase Pathways Inhibits Proliferation of NF2 Schwannoma Cells

Abstract

Neurofibromatosis Type 2 (NF2) is a tumor predisposition syndrome caused by germline inactivating mutations in the NF2 gene encoding the merlin tumor suppressor. Patients develop multiple benign tumor types in the nervous system including bilateral vestibular schwannomas. Standard treatments include surgery and radiation therapy, which may lead to loss of hearing, impaired facial nerve function and other complications. Kinase inhibitor monotherapies have been evaluated clinically for NF2 patients with limited success, and more effective non-surgical therapies are urgently needed. Schwannoma model cells treated with PI3K inhibitors upregulate activity of the focal adhesion kinase family as a compensatory survival pathway. We screened combinations of 13 clinically relevant PI3K and FAK inhibitors using human isogenic normal and merlin-deficient Schwann cell lines. The most efficacious combination was PI3K/mTOR inhibitor omipalisib with SRC/FAK inhibitor dasatinib. Sub-GI₅₀ doses of the single drugs blocked phosphorylation of their major target proteins. The combination was superior to either single agent in promoting a G1 cell cycle arrest and produced a 44% decrease in tumor growth over a two-week period in a pilot orthotopic allograft model. Evaluation of single and combination drugs in six human primary vestibular schwannoma cell models revealed the combination was superior to the monotherapies in 3 of 6 VS samples, highlighting inter-tumor variability between patients consistent with observations from clinical trials with other molecular targeted agents. Dasatinib alone performed as well as the combination in the remaining three samples. Preclinically validated combination therapies hold promise for NF2 patients and warrants further study in clinical trials.

Introduction

Neurofibromatosis Type 2 (NF2) is a tumor predisposition syndrome of the nervous system. It recently was renamed NF2-related schwannomatosis (1). It is caused by pathogenic mutations in the NF2 gene that encodes the merlin tumor suppressor protein (2, 3). Individuals with NF2 develop schwannomas, meningiomas, and ependymomas throughout their nervous systems. Bilateral vestibular schwannomas (VS) on the eighth cranial nerve alone are sufficient for an NF2-schwannomatosis diagnosis (1). Although the tumors are benign and many are slow growing, over time they cause increasing loss of nerve function and can be life threatening if not removed. Current standard of care involves regular monitoring with magnetic resonance imaging, followed by surgery and/or radiotherapy when deemed necessary (4). Surgery often leaves patients with permanent disability. For VS resections, these deficits include loss of hearing, balance, and facial mobility. Over the last ten years, there have been a number of clinical trials of small molecule drugs aimed at controlling tumor growth. However, drug selection based on merlin pathway knowledge has yet to produce a Food and Drug Administration (FDA)-approved drug therapy.

Merlin plays important roles in regulating Schwann cell responses to growth factors and extracellular matrix (5). It links actin dynamics to cell cycle progression, cell differentiation and survival (6). Loss of merlin function leads to aberrant signaling in multiple receptor tyrosine kinase (RTK) activated pathways, such as the phosphoinositide-3 kinase (PI3K), mitogen-activated protein kinase (MAPK), and Hippo pathways (6-9). Merlin also regulates signaling from extracellular matrix (ECM) receptors, integrins, and cell-cell adhesion receptors, cadherins (10-12). Despite decades of research, a complete understanding of the many signaling pathways merlin controls and their effect on Schwann cell proliferation and differentiation during development and neural repair remains elusive. The complexity of the signaling pathways regulated by merlin and the ability of schwannoma cells to rewire these pathways when challenged with a single kinase inhibitor contributes to the difficulty in developing an NF2 monotherapy.

We adopted an unbiased approach to identifying drug targets and drugs that reduce viability of human schwannoma model cells. Previously, we reported results of a proof of principle, ultra-high-throughput screen of compounds in the Library of Pharmacologically Active Compounds (13). This screen identified PI3K inhibitors as an effective drug class that inhibits the well validated merlin-dependent pathway (13). As part of the NF2 Synodos Consortium, two PI3K inhibitors, omipalisib (GSK2126458), a dual PI3K/mTOR inhibitor (14), and CUDC-907, a dual PI3K/HDAC inhibitor, were identified as effective agents in a screen of 40 compounds chosen by participating NF2 clinicians. Kinome analysis of an isogenic pair of wild-type (WT) and merlin-deficient (MD) human schwannoma model cells treated with omipalisib and CUDC-907 for 24 hours revealed activation of kinases involved in integrin dependent adhesion to ECM. These were focal adhesion kinase (FAK/PTK2) and multiple members of SRC family kinases (SFK), such as Fyn, Lyn, and Src. This suggested to us that inhibition of a critical RTK-activated pathway, PI3K/mTOR, led to compensatory activation of ECM adhesion-dependent kinases to enable cell survival (15). Support for co-inhibition of these targets was provided by results of a combination high-throughput screen of the MIPE compound library in human schwannoma model cells that identified several PI3K and FAK/SFK inhibitors as top performing combinations (16). Indeed, cooperative signaling from the ErbB2/3 RTK and ß1 integrins is well-documented in Schwann cell (SC) development. (17). Here, we conducted a small combination screen of PI3K and FAK/SFK inhibitors in clinical use to evaluate their synergy and efficacy in human schwannoma models and VS cells.

Materials and Methods

Cell culture

Human wild-type (WT; HS11) and NF2-deficient cell lines (HS01) were generated as previously described (18, 19). The NF2-deficient human SC line HS02 was generated from the parental cell line HS11 using lenti-NF2-shRNA #75 (Sigma-Mission). Primary adult Schwann cell line immortalized in vitro with telomerase reverse transcriptase (TERT) and cyclin-dependent kinase 4 (cdk4) (Dr. M Wallace, University of Florida) were used to generate NF2-deficient cell lines HS03 and HS04.3. HS03 cells were generated using lenti-NF2-shRNA #45 (Sigma-Mission). HS04.3 cells were generated using CRISPR/Cas9 with NF2sg1 (Broad Institute), Clone #3. The NF2-deficient mouse cell line, MS01, was generated as previously described (12, 20). Cells lines were authenticated by protein expression of human nuclear antigen, S100ß1, proteolipid protein and Sox10, as well as population studies using EdU incorporation and flow cytometry analysis. Cells were used between passage 6 and 18, and routinely monitored for mycoplasma, as previously described (18, 19). Vestibular schwannoma tumors and blood samples were harvested from 6 patients at the time of surgery (UM IRB #20150637) and used to create primary cell lines, as previously described (18, 21). Retrospective chart review was performed to confirm pathologic diagnosis of schwannoma (World Health Organization Grade I). DNA was extracted from blood leukocytes and tumor chunks using DNeasy Blood and Tissue Kit (Qiagen). Whole exome sequencing and multiplex ligation-dependent probe amplification was performed to detect mutations in the NF2, LTZR1, and SMARCB1 genes (20, 21).

Compounds

All compounds were purchased from SelleckChem and 10mM stocks were prepared in DMSO. For the allograft study, omipalisib and dasatinib were solubilized in 1.25% PEG, 2.5% Tween80, and 5% DMSO.

Combination screening

Combination matrix cell viability assays were conducted as previously described (16). Briefly, cells were seeded at 2,000 cells/well in 384-well plates. After overnight adhesion, cells received compounds in 0.1% DMSO and allowed to incubate for 72h at 37°C, 5% CO₂. Cell viability was measured with CellTiterFluor Viability Assay (Promega) and read on a Synergy H1 plate reader (BioTek). Cell viability was normalized to DMSO-only controls. Assay performance was evaluated by calculating the Z’ score for each plate with control wells. All Z’ values were >0.5. Synergy was determined using the open access Combenefit software (22). Biological interactions were determined using the sum weighted score of two standard reference models, Loewe and Highest Single Agent (HSA). Selectivity for merlin-deficient over isogenic normal cells was calculated from change in Loewe sum synergy weigh scores (LSSW_HS01/HS02 – LSSW_HS11 = △LSSW).

Live-imaging and cell death assay

Cells were seeded in a 384-well CellBIND (Corning) at 1,000 cells/well with phenol red-free growth media and incubated overnight (~12-14h) at 37°C, 5% CO₂. The cells were then treated with drug treatments at the concentrations listed as well as Caspase 3/7 Green Apoptosis Assay Reagent (Sartorius) at a 1:1000 dilution (n=4 wells/treatment). Cells were imaged using the Incucyte S3 live imaging system (Sartorius). Phase and green contrast images were captured every 4h for 72h. Images were analyzed using the integrated software. Primary human VS cells were seeded in 384-well CellBIND plates (Corning) at 1,000 cells/well with Schwann media and incubated overnight at 37°C and 5% CO₂. Cells were then treated with 0.05% dimethyl sulfoxide (DMSO; control condition), omipalisib (2, 4, 8nM), dasatinib (200, 400, 800nM), or the combination in maintenance media consisting of Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (Seradigm) and 1% penicillin-streptomycin (n=6 per condition). At 96 hrs, the ApoTox-Glo (Promega) assay was used to detect live cell protease activity for viability and cleaved caspase-3/7 activity for apoptosis using the Glomax Discover System (Promega).

Cell cycle assay

HS01 cells were grown in a six-well CellBIND dish at 250,000 cells/well and allowed to incubate overnight (~12-14h) at 37°C, 5% CO₂. Cells were then treated with drugs at the doses listed, and incubated for 24h at 37°C, 5% CO₂. EdU (Click-iT^™ EdU Cell Proliferation Kit, Molecular Probes, ThermoFisher) was added at 10μM to each well and incubated for 3h. Cells were lifted using 0.05% Trypsin-EDTA (Gibco) and stained with LIVE/DEAD^™ Fixable Violet Cell Stain (ThermoFisher). Following staining, cells were fixed, permeabilized, and treated with the Click-iT^™ reaction cocktail according to the manufacturer’s instructions. Cells were then counterstained with FxCycle Red stain and analyzed using a CytoFLEX S flow cytometer (Beckman Coulter) and CytExpert software (Beckman Coulter).

Membrane asymmetry assay

HS01 cells were grown in a six-well CellBIND dish at 250,000 cells/well and allowed to incubate overnight (~12-14h) at 37°C, 5% CO₂. Cells were then treated with drugs at the doses listed, and then incubated for 72h at 37°C, 5% CO₂. Cells were lifted using 0.05% Trypsin-EDTA (Gibco) and stained with Violet Ratiometric Asymmetry assay per the manufacturer’s instructions. Cells were analyzed using a CytoFLEX S flow cytometer (Beckman Coulter) and CytExpert software (Beckman Coulter).

Western blots

Protein extraction was conducted using a modified RIPA buffer (4% sodium dodecyl sulfate, 0.01% bromophenol blue, 10% glycerol, 100mmol/L dithiothreitol). Blots were incubated overnight at 4°C in a 1:1 mixture of tris buffered saline (TBS) and Odyssey blocking buffer (LI-COR Biosciences). The following antibodies were used: rabbit anti-pMLKL #17-10400 from Millipore Sigma, rabbit anti-pFAK(Y577/576) #44614G from Invitrogen, mouse anti-FAK #610087 from BD Biosciences, and rabbit anti-pAKT(T308) #2965, mouse anti-panAKT #2920, rabbit anti-pAKT(S473) #4060, rabbit anti-Src #2109, rabbit anti-pSRC(Y527) #2105, rabbit anti-EphA2 #6997, rabbit anti-pEphA2(S897) #6347, rabbit anti-RIP #3493, rabbit anti-cyclinD1 #2922, rabbit anti-p21(Waf1/Cip1) #2947 from Cell Signaling Technology. Antibodies were diluted 1:1000 unless otherwise stated. The following day, blots were incubated in a 1:1 mixture of 0.1%Tween and 0.02%SDS in TBS (TBS-TS) and Odyssey blocking buffer for 1 hour. Blots were imaged and quantified using the LI-COR Odyssey Imaging System (LI-COR Biosciences, n=3 biological replicates).

Orthotopic allograft schwannoma model

The mouse orthotopic allograft schwannoma model was generated using male and female NSG mice bred in house at 10-11 weeks of age. Care and use were approved by the University of Central Florida (UCF) Institutional Animal Care and Usage Committee (IACUC; #19-08). 5,000 luciferase-expressing merlin-deficient SC were injected into the right sciatic nerve of 10-week-old NSG mice as previously described (18). Mice were divided into treatment or control groups at 8 days post injection. Two mice failed to show sufficient tumor engraftment at this time and were excluded from the study. Control group (n=4) was treated with daily oral gavage of 1.25% PEG, 2.5% Tween80, and 5% DMSO. Mice in the treatment group (n=5) received oral daily gavage of omipalisib at 2mg/kg for the 1^st week and 1mg/kg for the 2^nd week, and dasatinib at 20 mg/kg. The dosage of omipalisib administered was lowered due to weight loss in one mouse that was subsequently euthanized and excluded from the study. Tumor radiance was measured at 8-, 15, and 22-days post-injection using the In Vivo Imaging System (IVIS, Caliper). All mice were euthanized after 14 days of treatment.

Statistical analysis

Significance was determined using GraphPad Prism Version 7.04 or SAS Studio (9.3M2). Two-way analysis of variance (ANOVA) and Bonferroni’s multiple comparisons post-test was used to assess treatment groups for cell cycle assay and IVIS radiance fold change of allograft tumors. A one-way ANOVA with Dunnett’s multiple comparisons post-test was used to assess significance of treatment groups compared to DMSO in western blots. Tumor weight significance between the treatment and control group was assessed using an unpaired t-test. ANOVA with Tukey post-hoc testing and Bonferroni adjustment was used to analyze viability and cleaved caspase-3/7 activity. To compare the relationship of combination therapies to monotherapies in individual primary VS cultures, 95% confidence intervals were used. Pearson’s coefficient was utilized to determine the strength and the direction of the relationship between phospho-protein expression and viability.

Data Availability

The data generated in this study are included as supplemental data files or will be made available upon request.

Results

Omipalisib and Dasatinib Synergize to Reduce Human Schwannoma Cell Viability

We tested whether co-inhibition of PI3K and FAK/SFK reduced viability of human schwannoma model cells. We selected six PI3K/mTOR and seven FAK/SFK pathway inhibitors to screen in 10x10 dose response combination matrices using human and mouse schwannoma model cell lines. Inhibitors were either FDA approved, in clinical trials, or in advanced pre-clinical development (Table S1). The initial combination screens were conducted in two human schwannoma model cells, HS01 and HS02. Biological interactions were determined using Loewe Additivity and Highest Single Agent (HSA) standard reference models. The top three combinations were assessed for selectivity in the isogenic human WT-SC line (HS11; Table S2) and screened in two additional human (HS03, HS04.3) and one mouse (MS01) schwannoma model cell lines (Table S3).

The most synergistic inhibitor combination was the PI3K/mTOR inhibitor, omipalisib with Src/Abl inhibitor, dasatinib (LSSW = 49.4 and 55.8 for HS01 and HS02, respectively; Table 1, Fig. 1A). The GI50 for omipalisib and dasatinib alone in HS01 cells were 50-80nM and >10μM, respectively (Table 1). The synergistic doses were 39-156nM for omipalisib and 156-625nM for dasatinib (Fig. 1A). At the lowest doses tested (39nM omipalisib and 156nM dasatinib), the combination produced a 44% and 49% reduction in cell viability in HS01 and HS02 cells, respectively. The combination was selective for merlin-deficient human Schwann cells over isogenic normal Schwann cells (ΔLSSW = 23.8 and 29.4 for HS01 and HS02, respectively). The low synergism scores in the HS03 and HS04.3 cells could be attributed to their immortalization with CDK4/6 and TERT. The combination was also effective in the mouse schwannoma model cell line later used in the in vivo study (MS01; Table S3).

Table 1.

Top 10 Synergic PI3K/mTOR and FAK/SRC Combinations Reducing Viability of Human Schwannoma Model Cells

Schwannoma Model Human Cell Line			HS01			HS02
Ranking	Drug Name	Target	GI50 (μM)	Synergy Model		GI50 (μM)	Synergy Model
				Loewe	HSA		Loewe	HSA
1	Omipalisib	p110α/β/δ/γ	0.05	49.9	54.9	0.08	55.7	55.9
	Dasatinib	Abl/Src/c-Kit	>10			>10
2	Pictilisib	p110α/δ	>10	46.1	49.5	>10	52.9	53.7
	Dasatinib	Abl/Src/c-Kit	>10			>10
3	Gedatolisib	p110α/γ/mTOR	0.08	35.3	38.4	0.09	19.6	21.8
	PND-1186	FAK	>10			>10
4	Buparlisib	p110α/β/δ/γ	4.3	33.6	38.1	4.9	20	23.8
	PND-1186	FAK	>10			>10
5	Omipalisib	p110α/β/δ/γ	0.02	28.3	29.2	0.02	27.6	28.9
	PND-1186	FAK	>10			>10
6	Buparlisib	p110α/β/δ/γ	5.3	27.7	32.9	3.6	25.4	27.6
	Dasatinib	Abl/Src/c-Kit	>10			>10
7	Omipalisib	p110α/β/δ/γ	0.3	26.1	29.6	0.33	15	21.6
	TAE226	FAK	1			2.1
8	Pictilisib	p110α/δ	3	25.7	30	>10	15.2	19.2
	PF-00562271	FAK	>10			6.5
9	Voxtalisib	p110γ/mTOR	>10	23.8	26.7	>10	21.7	25.15
	PND-1186	FAK	>10			>10
10	Omipalisib	p110α/β/δ/γ	0.02	22.8	26.5	0.01	16.5	19.9
	PF-00562271	FAK	2.2			3.8

Figure 1. Low Doses of Omipalisib and Dasatinib Synergize to Reduce Cell viability, AKT and FAK/SFK phosphorylation in HS01 cells.

A) Drug-combination matrix of two human MD-SC lines, HS01, HS02. Left matrices are viability results expressed as % of DMSO control. Right panels show synergy matrix analyses, Loewe additivity and HSA models respectively. B) Individual dose responses from the single-point viability matrix assay. C) Western blots of whole cell lysates of HS01 cells treated for 24h as indicated. Graphs of phospho-targets are shown relative to loading control. One-way ANOVA with Dunnett’s multiple comparison's posttest. *p<0.05, **p<0.005, ***p<0.0005, ****p<0.00001, n=3.

We verified that omipalisib (4 and 16nM) and dasatinib (100 and 400nM) inhibited phosphorylation of their expected targets in HS01 cells both alone and in combination after 24 hours of treatment. Phospho-AKT (T308 and S473) decreased by 91-99% at both doses in HS01 cells treated with omipalisib alone and in combination with dasatinib (Fig. 1C). Dose-dependent decreases in phosphorylated FAK-Tyr576 and Src-Tyr527 were observed in HS01 cells treated with dasatinib alone and in combination confirming inhibition of Src/FAK (Fig. 1C). We observed that omipalisib alone increased phosphorylated Src-Tyr527 levels compared to vehicle in HS01 cells (Fig. 1C). Dasatinib produced a small decrease in total and phosphorylated EphA2-Ser897, but the reduction was not superior to omipalisib alone (Fig. S1A, S1B).

Omipalisib and Dasatinib Promote a G1 Cell Cycle Arrest in Human Schwannoma Model Cells

Cell cycle assays analyzed by flow cytometry indicated that the percentage of HS01 cells arrested in G1 of the cell cycle was higher in cultures treated with the combination than with either single drug at the same concentrations (Fig. 2A, 2B). Cyclin D1 levels were reduced in omipalisib, and combination treated cells, consistent with a G1 cell cycle arrest. A concomitant increase in p21 levels was not observed (Fig. 2C). In dasatinib treated cells, cyclin D1 levels remained consistent but an increase in p21 was observed. To further refine the combination doses, we selected sub-GI₅₀ concentrations of each drug that synergized to reduce HS01 viability by 50% but would be ineffective alone. We tested omipalisib at 4, 8, and 16nM and dasatinib at 100, 200, and 400nM as monotherapies and in combination, respectively. Results of 72h live imaging confirmed that the combination of omipalisib and dasatinib at all doses was superior to either single agent in blocking cell proliferation as assessed by well confluence, indicating a significant cytostatic effect of the combination. Cleaved caspase-3/7 was not observed in any condition (Fig. 2D), nor was significant apoptosis or cell death observed by Violet Ratiometric flow cytometry assay (Fig. S2). Additionally, no significant modulation of necroptosis markers such as phosphorylated mixed-lineage kinase domain-like pseudokinase (pMLKL) or receptor-interacting serine/threonine protein kinase 1 (RIPK1) were observed in any condition in HS01 cells (Fig. S1). Based on this evidence, the combination of omipalisib and dasatinib produces a significant cytostatic, but not cytotoxic response in the human schwannoma model cells.

Figure 2. Omipalisib and Dasatinib in Combination Inhibit Cell Proliferation but Do Not Induce Cell Death.

A) Flow cytometry assay of EdU incorporation in HS01 cells treated for 24h with 0.1 μM omipalisib and dasatinib alone and in combination. B) Plot shows mean +/− SEM of 3 independent experiments. **p<0.01 and ****p<0.0001 determined by two-way ANOVA and Bonferroni’s multiple comparisons post-test. C) Western blots of whole cell lysates of HS01 cells treated for 24h with indicated drug (0.1 μM). D) Representative images of cell confluence (phase contrast, shown in yellow) and apoptosis (green object count, shown in pink) using live-cell imaging with indicated treatments E) Time course graphs of cell confluence and apoptosis over 72h (mean ±SEM, n=4). Untreated and DMSO treated cells initiate apoptosis at high cell density.

Inhibitor Combination Slows Intraneural Growth of Mouse Schwannoma Models Cells

We conducted a pilot in vivo drug efficacy study of the combination using an orthotopic nerve allograft schwannoma model in NSG immune-deficient mice (Fig. 3). After the 2-week regimen, the average weight of the combination-treated allografts (46.8 mg), was reduced by 61% when compared to the average weight of the vehicle treated mice (118.7 mg; p=0.04). When assessing the fold increase in radiance of the allografts over time, the combination treated mice had a 44% reduction in radiance when compared to the vehicle (p=0.09). The in vivo study had several limitations, such as small group number further exacerbated by the loss of two mice (one per group, not related to drug toxicity). This finding supports conducting a larger study to assess superiority of the combination over the single agents.

Figure 3. Pilot in vivo Allograft Study Demonstrates Efficacy of Drug Combination.

A) Bioluminescent signals (IVIS) were measured at indicated times beginning 7 days post implantation of mouse merlin-deficient SC (MS01) expressing luciferase into the right sciatic nerve of NSG mice. B) Tumor weights at conclusion of the study (mean ±SD). Significance assessed by an unpaired t-test. C) Mean fold increase in radiance in individual animals (mean ±SEM). Statistical significance determined using two-way ANOVA with Bonferroni’s multiple comparisons post-test.

Combination Therapy is More Effective at Reducing Viability of Primary Vestibular Schwannoma Cells than Omipalisib and Dasatinib Monotherapies

Multiplex ligation-dependent probe amplification analysis (MLPA) and whole exon sequencing (WES) was performed on blood leukocytes and tumors from six patients undergoing surgery for VS. Genetic testing revealed that all VS tumors contained at least one mutant NF2 allele and no LTZR1 or SMARCB1 gene mutations (Table S4). Cell viability and cleaved caspase-3/7 dependent apoptosis were assessed in the primary VS cells treated with omipalisib (4, 8, and 16nM) and dasatinib (200, 400, and 800nM) alone and in combination. Because clinical trials testing therapies for NF2-associated VS typically use a reduction of ≥20% as a threshold for response, we used this standard to assess drug response in our viability assays. When drug response was analyzed by group, the mean fold change (MFC) in viability and cleaved caspase-3/7 of all VS tumors showed that omipalisib (16nM) and dasatinib (200, 400, and 800nM) monotherapies initiated significant reductions in viability ≥20%. Overall, dasatinib was more effective than omipalisib at the monotherapy concentrations tested. (Fig. 4A). The combination therapies induced a more robust reduction in MFCs in viability at the three doses tested compared to the monotherapies alone. The highest combination dose (16nM omipalisib + 800nM dasatinib) promoted a substantial induction of cleaved caspase-3/7 as well. An equivalent increase in cleaved caspase-3/7 was not observed with either monotherapy. (Fig. 4A). When drug responses were analyzed in individual VS, we observed that each VS demonstrated greater decreases in viability with dasatinib than with omipalisib monotherapy (Fig. 4B and 4C).

Figure 4. Cell Viability and Caspase 3/7-Dependent Apoptosis in Primary Human Vestibular Schwannoma (VS) Cells Treated with Omipalisib and Dasatinib.

A) Viability and cleaved caspase-3/7 assays at 96hrs. Mean Fold Changes (MFC) were normalized to vehicle (0.05% DMSO). B) Viability for six VS samples at 96hrs. Results expressed as fold change compared to DMSO control. C) Cleaved Caspase-3/7 for six VS samples at 96hrs. Results expressed as change (△) in relative luminescence units (RLU). Line represents mean. Error bars represent standard error mean. *p<0.05. **p<0.01. ***p<0.001. n=6 replicates for each experiment.

In 3 of 6 primary VS (VSA21, VSA23, VSA26), combination therapies initiated significantly greater reductions in viability compared to their equivalent monotherapies (p<0.05). Combination therapy at the highest dose caused a detectable increase in cleaved caspase-3/7 in VSA18, VSA19, and VSA21. Western blots were conducted using protein extracts from untreated tumor chunks for 5 of 6 VS (Fig. 5); there was insufficient sample to assess VSA26. VS tumors expressed p-EphA2, p-FAK, p-AKT, and p-SRC at varying levels. There were inverse relationships between target protein expression (i.e., p-EphA2, p-Src, p-AKT and p-FAK) and cell viability in response to omipalisib and dasatinib monotherapies (Fig. S3); however, the relationship did not meet statistical significance with Pearson’s correlation coefficient.

Figure 5. Western blots for targeted pathway kinases in untreated vestibular schwannomas tumors.

Phospho-protein levels were normalized to total actin expression.

Discussion

The findings confirm results of single and combination high-throughput compound screens that identified PI3K and FAK/SFK as potential co-targets for reducing schwannoma cell viability (13, 15, 16). Omipalisib and dasatinib emerged as the most synergistic combination using multiple human and mouse schwannoma model cell lines and stringent advancement criteria. Omipalisib and dasatinib synergized at sub-GI₅₀ concentrations and had a high selectivity for merlin-deficient schwannoma model cells over their isogenic normal Schwann cells. At sub-GI₅₀ concentrations, the combination of omipalisib and dasatinib reduced cell confluence when compared to either drug at the same dose and decreased phosphorylation of their respective targets at 24h of treatment. Dasatinib at sub-maximal doses has been used successfully in multiple clinical trials for treatment of chronic myeloid leukemia (23, 24). Combination therapy with a drug cocktail that synergizes to produce disease control at lower doses than a monotherapy is becoming the norm for various conditions including melanoma and breast cancer, hypertension, and HIV infection (25, 26). By administering a combination of drugs at doses below the maximum tolerable dose (MTD), the risk of developing drug resistance or adverse effects seen with higher doses of monotherapies are reduced or eliminated (27-29). This study demonstrates the efficacy of this concept in human schwannoma model cells and supports further research into the use of low dose combination treatments targeting these pathways.

The combination therapy completely inhibited proliferation by promoting a strong G1 cell cycle arrest but did not induce cell death in model schwannoma cells. Thus, omipalisib and dasatinib have a cytostatic effect on the model cells rather than a cytotoxic effect as seen with CUDC907 (21). Emerging research highlights the advantages of both therapeutic strategies, and currently outlines parameters in which one strategy might be superior to the other (30). The pilot mouse schwannoma allograft model demonstrated that a two-week treatment with the combination reduced intraneural growth of mouse schwannoma model cells (MS01). Despite limitations due to small sample size, the average increase in graft size by weight for the omipalisib and dasatinib treated mice was 61% less than those from vehicle treated mice. In patient VS samples, the combination of omipalisib and dasatinib was superior to single drugs in three of the six VS, depending on the dose, whereas dasatinib alone performed statistically as well as the combination in the remaining VS samples. Overall, a greater number of VS responded to dasatinib alone than to omipalisib alone. Cleaved caspase 3/7 was detected in three of 6 VS samples treated with the combination at the highest dose. Caspase 3/7 activation was not observed in the human schwannoma model cells. This inconsistency between primary VS cells and the schwannoma model cells could be due to several reasons including: 1) the pathogenic NF2 gene mutations in VS cells compared to stable shRNAi knockdown of merlin in model cells, and 2) possible cellular stress pathways present in freshly isolated primary VS cells compared to culture-adapted schwannoma model cells (31).

Further investigations are warranted to determine if target phospho-protein levels can be used as reliable biomarkers for drug response and to determine how tumor heterogeneity impacts effectiveness of drug treatment. These results and the outcomes of NF2 clinical trials thus far supports development of patient-specific drug screening (18-21, 32). Previously, this lab conducted a small proof-of-principle drug study using an immortalized schwannoma cell line from a schwannomatosis patient with a germline SMARCB1 mutation (33). Generation of a patient-derived xenograft was not feasible due to slow in vivo growth but indicates that personalized models can be developed as is increasingly done for breast, prostate, and colorectal cancers (34-37).

Despite promising preclinical results, PI3K inhibitors have had limited success in clinical trials. We found that low dose omipalisib alone increased pTyr527-Src levels compared to DMSO control in model schwannoma cells, confirming our previous kinome results. This cautions against using PI3K inhibitors alone for NF2-Schwannomatosis and highlights how quickly human schwannoma cells activate compensatory cell survival kinases. Despite setbacks, the FDA has approved several PI3K inhibitors; some only when used in combination with other inhibitors or after relapse with another therapy (38). Omipalisib has high potency for all PI3K isoforms and mTOR, due to the strong interactions of a charged sulfonamide with conserved lysine residues in all p110 isoforms. It has been evaluated for pulmonary fibrosis (NCT01725139) and solid tumors (NCT00972686) but appears to no longer be under development. Dasatinib is FDA approved for use in some patients with Philadelphia chromosome-positive acute lymphoblastic leukemia and chronic myeloid leukemia. It is well tolerated and is in numerous clinical trials alone and in combination including with crizotinib (NCT01744652, NCT01644773) and erlotinib (NCT00444015, NCT00826449). We assessed the efficacy of dasatinib and saracatinib, another Src inhibitor, in combination with FDA-approved c-MET inhibitor cabozantinib (20). The combination of saracatinib and cabozantinib promoted apoptosis of mouse schwannoma cells and reduced allograft growth by 85% (20).

Although omipalisib and dasatinib as a combination therapy may not be translatable into the clinic due to toxicity, the results indicate that co-targeting FAK/SFK and PI3K is a valid strategy for NF2 schwannomas. FAK is an important cancer target and plays essential roles in SC development and has been shown to be a viable target for NF2 schwannomas (16, 39-44). FAK overexpression and increased integrin/FAK/Src/Ras signaling was reported in human schwannoma cells and contributed to cell proliferation (43). In a study of ALK- and MET-inhibitor crizotinib, FAK1 was identified as a critical effector for the drug’s mechanism of action in NF2 mouse and human schwannoma model cells (44). In a subsequent study, brigatinib, a second generation ALK inhibitor, and dasatinib were identified as one of three top performing drug combinations tested in human schwannoma cells. This combination was not as effective as brigatinib monotherapy when tested in Postn-Cre; Nf2^flox/flox mice. However, this could be due to the reduced dose of brigatinib in the combination compared to the monotherapy as a result of toxicity (16). Brigatinib was found to inhibit FAK1/2, as well as FER and TNK in the human schwannoma model cells used here. Both crizotinib and brigatinib are currently in clinical trials for treatment of NF2 (NCT04283669 and NCT04374305, respectively.

PI3K is activated downstream of multiple RTKs that drive Schwann and schwannoma cell proliferation. These include: ErbB2/3, platelet-derived growth factor receptor, and c-MET (45, 46). In addition to the PI3K pathway, activation of RTKs triggers signaling from multiple parallel pathways including RAS-MAPK and SFK signaling. Thus, inhibitors of RTKs are expected to have more pleiotropic effects than inhibitors that block further downstream such as at the level of PI3K/AKT/mTOR. Indeed, we reported in mouse schwannoma models cells, that the c-MET inhibitor, cabozantinib reduced phosphorylation of the receptor as well as ERK and AKT (19). Integrin-dependent adhesion to ECM plays a critical role in SC development and is elevated in schwannomas (10). Cooperative signaling between RTK and ß1 integrins occurs in Schwann cells (10). This could explain why FAK/SFK activation occurs so readily when PI3K is inhibited and serves to block anoikis in adhesion-dependent schwannoma cells. Indeed, in silico studies revealed that among cancer cell types, those with loss of merlin are particularly sensitive to FAK inhibitors (47). Overall, a strategy that employs a monotherapy with broad downstream effects, such as brigatinib currently in phase 2 trials for NF2 patients or a dual inhibitor such as fimepinostat, that targets PI3K as well as histone deacetylase with expected broad transcriptional effects, may be preferable to highly specific mono-kinase inhibitors.