STAT3 and HIF1α Signaling Drives Oncogenic Cellular Phenotypes in Malignant Peripheral Nerve Sheath Tumors

Neurofibromatosis type 1 (NF1, OMIM 162200) is an autosomal dominant tumor predisposition syndrome affecting approximately 1 in 3,500 individuals (1). The NF1 gene (17q11.2), encodes neurofibromin, which is highly expressed in the brain and central nervous system. NF1 is clinically characterized by hyperpigmentary abnormalities of the skin (café-au-lait macules and inguinal/axillary freckling), iris hamartomas (Lisch nodules), and the growth of benign cutaneous peripheral nerve sheath tumors (neurofibromas; refs; 1, 2). Neurofibromin functions as a tumor suppressor by acting as a GTPase-activating protein (GAP) toward the small G-protein Ras (3, 4). Consequently, inactivating mutations to NF1 lead to increased signal transduction through Ras to promote uncontrolled cell growth and tumorigenesis (5, 6).

Malignant peripheral nerve sheath tumors (MPNST) are the main cause of death in NF1. MPNSTs usually arise from preexisting plexiform neurofibromas; however, they are also known to occur sporadically (7). MPNSTs represent 10% of all soft tissue sarcomas with 50% occurring in association with NF1. The lifetime risk of developing sporadic MPNST is 0.001%, compared with 5% to 13% for NF1 patients (8). Therapeutic options are limited for NF1-associated tumors. Current treatment options are restricted to complete surgical resection with clear margins; however, the local recurrence of MPNSTs is high (from 32% to 65%; ref. 9).

Although multiple receptor tyrosine kinases are known to be amplified in MPNSTs, clinical trials that target receptor tyrosine kinases for MPNSTs have been challenging (10). For instance, a phase II trial using EGFR inhibitor, erlotinib, was unsuccessful (11). Subsequently, a phase II clinical trial was initiated using sorafenib that was also ineffective (12). Sorafenib is a multikinase inhibitor that blocks receptor tyrosine kinases VEGFR and PDGF and the Raf serine/threonine kinases. Phase II clinical trials with receptor tyrosine kinase inhibitors Gleevec (13) and dasatinib have also been unsuccessful (14). It is probable that the heterogeneity of MPNSTs is an underlying cause for the lack of success in these clinical trials. For example, recent LOH analysis of gene markers linked to MPNST tumor progression indicated high levels of intratumoral molecular heterogeneity, with differences in LOH reported within the same tumor samples (6). In view of these challenges, new therapeutic approaches to treat the heterogeneous MPNST population are required.

Much progress has been made in determining the molecular pathophysiology of MPNSTs. High-throughput whole-genome analysis with microarrays has provided the most insight into new therapeutic options by revealing patterns or molecular signatures common in MPNSTs that are not present in benign neurofibromas, that is, copy number variations and gene expression changes (15, 16). It has been previously determined that the HGF, MET, and PDGFRA genes are frequently amplified in MPNSTs (15, 17). The MET proto-oncogene, receptor tyrosine kinase (MET) is activated through its sole ligand, hepatocyte growth factor (HGF). HGF/MET signaling reduces cell-to-cell adhesion, enhances cell motility, and increases the proteolytic activity of matrix metalloproteases (MMP), promoting tumor cell invasiveness (18, 19).

Although much headway has been made through genomic analysis, it is still unclear what signaling components are frequent drivers of malignancy within the heterogeneous MPNST population. Throughout this study, we utilize four MPNST cell lines to carefully ascertain the level of dependence of known signaling pathways linked to MPNST malignancy. Such work is necessary to identify effective therapeutic strategies for NF1 patients who have MPNSTs with varied cell migratory and invasive signaling profiles. In three of the four MPNST cell lines (ST8814, S1844.1, and S1507.2), exhibiting elevated MET expression, cell migration, and invasion was significantly blocked with either MET or mTOR inhibition, whereas the S462 cells were highly resistant to both treatment. Of importance, our work reveals that STAT3, which is positioned downstream from multiple receptor tyrosine kinase signaling inputs, including MET, functions as a more frequent driver of cell migration, invasion, and tumor formation in these four MPNST cell lines. Furthermore, we uncover that STAT3 regulates these malignant characteristics through the transcription factor hypoxia-inducible factor 1α (HIF1α). This work indicates for the first time that targeting the JAK2/STAT3/HIF1α/VEGF-A signaling axis could be a viable treatment strategy for blocking tumor growth of MPNSTs with variable signaling profiles.

Antibodies against phosphorylated STAT3 at Tyr705 (#9145), and Ser727 (#9134), total STAT3 (#4904) and total β-actin (13E5, #4970) were purchased from Cell Signaling Technology Inc. (Danvers). Antibodies against phosphorylated MET (Tyr1234/1235; #05-900) and VEGF-A (#07-1376) were purchased from Merck Millipore U.K. Limited, while total MET (#sc162) was bought from Santa Cruz Biotechnology Inc.. HIF1α antibodies (#610959) were obtained from BD transduction laboratories, and HIF2α antibodies (#NB100-122) were from Novus Biologicals.

ST8814 MPNST-derived cell lines and HEK293 were purchased from ATCC (distributed by LGC Standards). The S462, S1844.1, and S1507.2 MPNST cell lines were a kind gift from Prof. Mautner, (University of Hamburg, Germany) and the late Prof. Guha, (University of Toronto, Canada). Benign neurofibromas (for mRNA extraction) were obtained from patients. This project was approved by the local ethics committee. Informed consent for sample collection was obtained according to protocols approved by this committee. Cell lines were cultured and maintained in DMEM supplemented with 10% (v/v) FBS and 1% (v/v) penicillin-streptomycin in a humidified incubator (5% CO₂ at 37°C). All cells were routinely screened for mycoplasma and were decontaminated where necessary (MycoAlert Detection Kit from Lonza Biologicals Plc.).

PF-04217903 was a kind gift from Pfizer Limited. Cells were pretreated for 30 minutes before HGF stimulation. All experiments were performed in triplicate. Cucurbitacin-I, 5,15-DPP, FLLL31, SU11274, rapamycin, and all other reagents unless otherwise stated were purchased from Sigma-Aldrich Company Ltd., IL6, HGF, and PDGF were purchased from R&D systems.

Cells were lysed using NP40 lysis buffer [50 mmol/L Tris-HCl, 0.5 mol/L NaCl, pH 7.4, 50 mmol/L β-glycophosphate, 5 mmol/L MgCl₂, 10% (v/v) glycerol and 1% (v/v) Nonidet-P40] supplemented with a Mini complete Protease Inhibitor Cocktail (Roche Diagnostics Ltd.). Protein concentrations were assessed with use of Bradford reagent, in accordance with the manufacturer's protocol (Sigma-Aldrich Company Ltd.). MET protein was immunopreciptated using anti-MET antibodies coupled to G-Sepharose beads (GE Healthcare Lifesciences), washed three times in NP40 lysis buffer, and resolved by SDS-PAGE. Samples for STAT3 and HIF1α analysis were prepared by direct lysis in sample buffer [62.5 mmol/L Tris-HCL, 50 mmol/L DTT, 2% (w/v) SDS, 10% (v/v) glycerol and 0.1% (w/v) Bromophenol blue, pH 7.6] and sonicated (Bioruptor from Diagenode) for 5 × 40 second cycle pulses.

The NuPage Novex gel system was used for electrophoresis as described in the manufacturer's protocol (Life Technology). Depending on the protein size, protein samples were resolved on either 3% to 8% Tris-acetate or 4% to 12% Bis-Tris gels. Proteins were then transferred to a polyvinylidene fluoride membrane purchased from Merck Millipore, blocked in 5% (w/v) dry milk powder in standard TBS supplemented with 0.1% (v/v) Tween (as recommended by Cell Signaling Technology Inc.) for 4 hours. Membranes were incubated at 4°C overnight in primary antibody in 2% (w/v) BSA in TBS-T, then washed three times for 4 minutes in TBS-T, and incubated in the appropriate HRP-conjugated secondary antibody (1:10,000 dilution in TBS-T) for a minimum of 30 minutes at room temperature. Membranes were washed four times for 3 minutes with TBS-T and then incubated in Enhanced Chemiluminesence solution (GE Healthcare Lifesciences) for 1 minute. Konica Medical Film was used to visualize the signal and the exposed films were developed using a Konica Minolta SRX-101A developer.

TRIzol (Life Technology) was used as described in the manufacturer's protocol for mRNA extraction from benign neurofibromas and MPNSTs obtained from patients. Cell lines were harvested in RNA-protect buffer (Qiagen) and centrifuged at 5,000 rpm for 5 minutes. mRNA was extracted from the pellet using the Qiagen mRNA extraction kit in accordance with manufacturer's protocol. QIA shredders were utilized to homogenize the pellet (Qiagen). mRNA concentration and purity was assessed by using a nanodrop spectrophotometer. Total RNA from each sample (1 μg) was transcribed into complementary cDNA using a Quantitect reverse transcription kit (Qiagen) in a thermal cycler (Applied Biosystems). MET primers used were forward 5′-CCACCACAGTCCCCAGAGT-3′ and reverse 5′-AGATCACATGAACACAGGA-3′, with an amplicon size of 51 base pairs. β-actin with an amplicon size of 104 base pairs (cat no. QT01680476) was purchased from Qiagen who retain the right to withhold primer sequence information. Quantitative real-time PCR reactions were conducted in 96-well plates using appropriate primer assays and SYBR Green PCR Master mix (Qiagen). Assays were performed as follows: Initial denaturation step (95°C, 15 minutes), 40 cycles of denaturation (94°C, 15 seconds), annealing step (55°C, 30 seconds), extension step (72°C, 40 seconds). The amplification products were quantified during the extension step in the fortieth cycle. The results were then determined using the ΔΔC= method, and standardized to β-actin control. A dissociation step was performed, which verified that only one PCR product was produced with each primer set and shows their specificity. The correct amplicon size of PCR products was also verified by resolution on 1% (v/v) agarose gels.

Cells were seeded in 60 mm plates and left to reach 100% confluency. Cells were then synchronized in 1% (v/v) FBS DMEM for 24 hours and “wounded” with a pipette tip. Dead cells were removed by washing with PBS. Cells were pretreated for 30 minutes with either rapamycin, MET, or STAT3 inhibitors (where indicated) before cytokine stimulation. Pictures were taken before treatment and 12 to 18 hours after treatment using an inverted AMG EVOS microscope equipped with an Olympus camera.

Transwell permeable supports with 6.5 mm diameter inserts, 8.0 μm pore size, and a polycarbonate membrane (Corning, cat no: 3428) were used to perform migration assays. Cells were grown in a 75-cm² flask with standard medium (10% (v/v) FBS) until confluent. Cells were then harvested using Trypsin-EDTA. Cells were counted using a haemocytometer. A total of 1 × 10⁶ cells and were resuspended in DMEM containing 1% (v/v) FBS. These cells were then seeded in the upper chamber of the Transwell; the lower chamber was filled with 600 μL of standard culture medium (10% (v/v) FBS) and 5 mg/mL fibronectin (R&D systems), as an adhesive substrate. Cells were incubated at 37°C 5% CO₂ for 24 hours. The percentage of adherent cells was then determined by fixing the cells with methanol and acetone (1:1) for 20 minutes at −20°C. Cells were then stained with Crystal Violet (5 mg/mL) in ethanol for 10 minutes, followed by a stringent wash with dH₂O until the water ran clear. Crystal Violet stained cells were eluted with 1% (w/v) SDS and the absorbance was read at 550 nm on a Genova MK3 Lifescience Analyzer (Jenway Scientific). For invasion assays, a similar protocol was used; however, the top chamber of the Transwell was filled with 300 μL of BD Matrigel Basement Membrane Matrix (1 mg/mL). The Matrigel was incubated at 37°C for 4 hours to allow it to gel. Cells were then seeded and incubated as described for migration assay for 3 days. The number of invaded cells was determined by fixation staining and elution of crystal violet with 1% (w/v) SDS, as before.

STAT3 shRNA (Clone ID: NM_003150.3-458s21c1), HIF1α shRNA (Clone ID: NM_001530.x-3867s1c1), and non-target control MISSION shRNA (Clone ID: SHCO16) in pLKO.1 vector (Sigma-Aldrich Company Ltd.) were packaged into lentivirus using HEK293T cells and cotransfected (lipofectamine 2000, Life Technology) with pLP1, pLP2, and pLP (VSVG). Confluent MPNST cell lines were infected with shRNA containing lentivirus (STAT3, HIF1α, or nontarget control) and selected over 2 weeks with 5 μg/mL puromycin (Life Technology). Puromycin selected mixed cell populations were then used for cell migration and tumor formation assays as described above.

Two-layered soft agar assays were carried out in 6-well plates. MPNST cell lines were plated in complete DMEM media in 0.35% (v/v) agar at (3 × 10^×5) over a 0.6% (v/v) agar layer. The agar was then overlaid with complete DMEM media and MPNST spheroids were grown for 14 days at 37°C in 5% CO₂. Media were changed twice a week. Pictures were taken using an inverted AMG EVOS microscope equipped with an Olympus camera. Volume of tumor spheroids was measured using ImageJ (v1.48) software.

Current evidence implies that the HGF/MET signaling pathway plays an important role in the tumor growth and invasive properties of MPNSTs and enhancement of MET expression has previously been observed (18, 20). We examined MET gene expression using qPCR in multiple NF1-MPNST tumor samples and benign neurofibromas. As Fig. 1A indicates, we observed that eight out of fourteen MPNST tissues had elevated levels of MET mRNA compared with the benign neurofibromas. We next examined MET protein levels in four MPNST-derived cell lines (ST8814, S462, S1844.1, and S1507.2) in comparison with a benign control (Fig. 1B), and again observed variation in the level of MET protein expression. MET was elevated in the ST8814, S1844.1, and S1507.2 MPNST cell lines (highest expression in the ST8814 cells), but not the S462 cells. This variation in MET expression highlights the heterogeneous status of MPNSTs.

Given the limited success of clinical trials using single drug agents, and the heterogeneous nature of MPNSTs, we hypothesized that these four MPNST cell lines may have differential dependency on migratory cell signaling pathways. To test this hypothesis with clinical implication, we initially analyzed the effects of MET inhibition on signaling pathways that could contribute to malignancy. We used two MET inhibitors, SU11274 and PF-4217903. SU11274 is a first-generation MET inhibitor that competes with ATP to bind to the activation loop of MET. PF-4217903 is a novel ATP-competitive small-molecule inhibitor of MET. PF-4217903 previously showed efficacy in a patient with MET driven papillary renal cell carcinoma (RCC; ref. 21) and subsequently reached phase I clinical trials.

In all four MPNST-derived cell lines, the MET inhibitors, SU11274 and PF-4217903, were sufficient to block HGF-induced MET activation, as observed by a reduction in Tyr1234/1235 phosphorylation of MET (Fig. 2A). Of interest, both PDGF and IL6 also potently activated MET to a degree similar to HGF (Fig. 2A) revealing signaling cross-talk within the MPNSTs between MET and the PDGF and IL receptors (PDGF-R and ILR, respectively). This is in concordance with previous studies identifying crosstalk between receptor tyrosine kinases and MET in cancer cell lines (22, 23). We show that HGF-induced wound healing in the ST8814, S1844.1, and S1507.2 cells was significantly reduced after inhibition of MET with both the SU11274 and PF-4217903 inhibitors (Fig. 2B, see Fig. 2C for representative images of wound healing in ST8814 and S462 MPNST cell lines). In contrast wound healing in the S462 cell line was completely insensitive to MET inhibition during the 18-hour wound-healing assay (Fig. 2B). These results suggest that instead of utilizing the migratory HGF/MET signaling pathway for wound healing, the S462 cell line is more dependent on other migratory signaling pathways. Critically, these data reveal that MPNST cell lines can exhibit varied sensitivity to MET inhibition, suggesting that therapeutic targeting of MET may not be a suitable treatment strategy for all NF1 patients. These data reveal for the first time that PDGF and IL6 can also activate MET in MPNSTs, and imply that growth factor receptor signaling cross-talk likely contributes to the malignancy of MPNSTs.

One mechanism by which MET drives cell motility is through activation of the JAK2/STAT3 signaling pathway. The JAK2/STAT3 pathway is also downstream of other receptor tyrosine kinases, including PDGF-R (24), ILR (21, 25), and EGF-R (24, 25), which are known to promote tumorigenesis in MPNSTs. STAT3 is considered an oncogene and promotes transcription of genes linked to cancer progression, driving cell migration, invasion, and survival (26). In the ST8814 cells, tyrosine phosphorylation of STAT3 was robustly induced after just 30 minutes of HGF stimulation and maintained throughout the 3-hour time course. This is in contrast with the S462 cells (Supplementary Fig. S1) where HGF stimulation resulted in a much slower and less pronounced STAT3 response (Supplementary Fig. S1). This difference likely reflects increased HGF/MET signaling in the ST8814 cells when compared with the S462 cell line. Cells were also stimulated with IL6 for 30 minutes as a positive control for STAT3 activation; also illustrating that signaling through multiple receptor tyrosine kinases in MPNSTs (such as ILR) can activate JAK2/STAT3 signaling. Of interest, IL6 serum levels were reportedly elevated in NF1 patients with MPNSTs (25). It is therefore likely that STAT3 is activated via multiple mechanisms in NF-1, making it an appealing therapeutic target.

We next wanted to examine whether STAT3 is a common driver of cell migration within our panel of MPNST cells. We used three different STAT3 inhibitors; Cucurbitacin-I, 5,15-diphenylporphyrin (5,15-DPP) and FLLL31. Cucurbitacin-I is derived from cucurbitane and FLLL31 is derived from curcumin, the bioactive compound in the spice turmeric. Both of these inhibitors selectively bind to and inhibit the tyrosine kinase JAK2, which is immediately upstream of STAT3. 5,15-DPP is a selective STAT3 Src homology-2 domain (SH2) antagonist, preventing STAT3 SH2 domain-mediated ligand binding, dimerization, and signal transduction. All three inhibitors suppressed HGF-induced JAK2-mediated Tyr705 phosphorylation of STAT3 in all cell lines (Fig. 3A). Ser727 phosphorylation of STAT3, which is directly phosphorylated by mTORC1 (27), was less sensitive to drug treatments. Next we analyzed the effects of STAT3 inhibition on cell migration using a wound-healing assay. As Fig. 3B and C shows, FLLL31 significantly blocked wound healing in all cell lines, suggesting that STAT3 is a key mediator of cell migration in MPNSTs. 5,15-DPP was ineffective at suppressing wound healing in three out of the four MPNST cell lines (ST8814, S462, and S1507.2). Cucurbitacin-I also significantly blocked wound healing; however, it also greatly altered the morphology of the S1844.1 and S1507.2 cell lines, (Supplementary Fig. S2) causing cellular detachment. This suggests that Cucurbitacin-I may have additional nonspecific downstream signaling effects.

We next analyzed the effectiveness of both MET and STAT3 inhibition upon cell migration (Fig. 4A) and invasion (Fig. 4B). In the ST8814, S1844.1, and S1507.2 cell lines, inhibition of either MET (SU11274) or STAT3 (FLLL31) significantly decreased both cell migration and invasion. Conversely, cell migration and invasion in the S462 cells were only suppressed by STAT3 inhibition. The insensitivity of the S462 cells to MET inhibition indicates that the migratory and invasive properties of these cells are less dependent on MET signaling. These data suggest that STAT3 may be a common mediator of cell migration in a wide range of genetically heterogeneous MPNSTs, making it a more appealing therapeutic target than MET.

To confirm STAT3 as a common mediator of oncogenic processes within these four different MPNST cell lines and to validate our studies with the STAT3 inhibitors, we next performed shRNA-mediated knockdown of STAT3 (or non-target control) to create stable STAT3 knockdowns. As expected, knockdown of STAT3 robustly inhibited wound closure by over 50% (Fig. 5A) in all four MPNST cell lines. STAT3 knockdown also prevented cellular migration (Fig. 5B) and invasion (Fig. 5C) by over 60%. Given that STAT3 is thought to be involved in many aspects of tumorigenesis (26); we next examined the ability of MPNSTs to form tumors in soft agar (using a tumor spheroid assay) after STAT3 knockdown. Of interest, STAT3 knockdown caused a significant reduction in the average tumor spheroid volume in all of the MPNST cell lines, with the S1507.2 cell lines showing the biggest reduction of approximately 80% (Fig. 5D). These data support the central involvement of STAT3 in promoting tumorgenesis and malignancy in a range of different MPNSTs. From our laboratory, Dodd and colleagues recently showed that STAT3 was essential for driving HIF1α expression within non-cancer cell line models (28). There are a number of reported genes involved in MPNST pathology that are also linked to HIF. For example, IGF2 is known to be transcriptionally regulated by HIF1α and promotes tumor cell survival (29) and is reportedly involved in the malignant transformation of MPNSTs (30). VEGF-A is also a direct target of both HIF1α and HIF2α and a key mediator of tumor angiogenesis. A recent histologic study found that VEGF expression was upregulated in MPNSTs but not in peripheral neurofibromas, suggesting that an angiogenic drive may be contributing to the malignant transformation of NF1-associated MPNSTs (21, 31). Furthermore, the angiogenic factor midkine is a direct target for HIF1α (27), and is thought to have both angiogenic and mitogenic activity within NF-1–deficient Schwann cells (32). Given these possible signaling connections between STAT3 and HIF-mediated gene-expression in MPNSTs, we examined HIF1α and HIF2α and VEGF-A protein expression levels. Figure 5E shows that expression of both HIF1α and HIF2α was ablated upon knockdown of STAT3. Furthermore, expression of VEGF-A was also strikingly impaired (Fig. 5E). These very low levels of HIF1α, HIF2α, and VEGF-A protein expression in all of the STAT3 knockout MPNST cell lines strongly imply that STAT3 is a dominant signaling component promoting the angiogenic response in multiple MPNSTs.

While we previously showed that mTORC1 is a pivotal kinase involved in HIF1α/VEGF-A signaling in HEK293 cells and an upstream kinase of STAT3 (28). A recent study indicated that HIF1α expression levels in MPNSTs were fairly resistant to mTORC1 inhibition with rapamycin (33), indicating that mTORC1 is less involved in promoting HIF1α in MPNSTs. We therefore hypothesized that JAK2/STAT3 might be the dominant pathway that promotes signal transduction through HIF1α in MPNSTs. To confirm this new JAK2/STAT3/HIF1α signaling axis in MPNSTs, we examined HIF1α protein expression after treatment with the JAK2 inhibitors, Cucurbitacin-I, 5,15-DPP, and FLLL31, and with the mTORC1 inhibitor, rapamycin (Fig. 6A). Inhibition with Cucurbitacin-I and FLLL31 showed a significant reduction in HIF1α levels in all MPNST cell lines. Consistent with our previous observations, the 5,15-DPP inhibitor was less effective. Apart from in the S1507.2 cells, inhibition of mTORC1 with rapamycin was insufficient to decrease the protein levels of HIF1α in these MPNST cell lines. Phosphorylation of ribosomal protein S6 (rpS6) was used as a readout for mTORC1 activity and confirmed efficacy of rapamycin. In all MPNST cell lines, STAT3 and mTORC1 inhibition reduced VEGF-A protein levels, where presumably the mTORC1 input through VEGF-A is likely through its translational regulation as previously reported by ourselves (28).

We next ascertained how effective rapamycin was at blocking wound healing (Fig. 6B), cell migration (Fig. 6C), and invasion (Fig. 6D) in our four MPNST cell lines. Similar to our observations with MET inhibition, the malignant properties of the S462 cells to wound heal, migrate, and invade through extracellular matrix were particularly insensitive to mTORC1 inhibition, in contrast with the ST8814, S1844.1, and S1507.2 cells. These data again highlight the variation of signaling pathways involved in malignancy within the heterogeneous population of MPNST cells.

As current evidence only loosely implicates HIF in malignant progression of MPNSTs, we next wanted to determine the significance of HIF1α in these oncogenic signaling processes. To do this, we carried out shRNA-mediated HIF1α knockdown. Phenocopying STAT3 knockdown, we observed that knockdown of HIF1α in all four MPNST cell lines markedly reduced their ability to wound heal (Fig. 7A), migrate (Fig. 7B), invade through the extracellular matrix (Fig. 7C), and to form tumor spheroids in soft agar (Fig. 7D). Western blotting confirms that HIF1α was efficiently knocked down in these MPNST cells (Fig. 7D). These data indicate for the first time that HIF1α is a necessary downstream component of STAT3 that drives tumor growth and malignancy in a range of MPNST cell lines with differing signaling profiles.

Although it is known that somatic NF1 gene inactivation results in aberrant Ras signaling, Ras activation is insufficient to induce malignant transformation alone. Although many advances have been made in determining the molecular pathophysiology of MPNSTs, it is still unclear which pathways are common drivers of malignancy in the heterogeneous MPNST population. In this study, we aimed to identify commonly upregulated signaling pathways involved in migration, invasion, and tumor formation which could be targeted for a wide range of MPNSTs. Although we show that the MET and mTORC1 signaling pathways are both important in three of the four MPNST cell lines tested, insensitivity within the S462 cell line indicates that targeting MET and/or mTORC1 alone may not be a suitable therapeutic strategy for the heterogeneous NF-1-MPNST population. Of importance, we reveal that cell migration, invasion, and tumor formation within all MPNST cell lines tested were acutely sensitive to STAT3 and HIF1α inhibition. Our work indicates that STAT3/HIF1α activation may be a common driver of tumorigenesis in MPNSTs. Supporting this notion, several receptor tyrosine kinases that converge on and activate the JAK2/STAT3 pathway are known to drive tumor progression in MPNSTs, including MET (20), PDGF receptors (PDGF-R; ref. 24), ILR (25), and EGF-R (25, 34).

Although MET targeting has previously been shown to inhibit the migratory, invasive, and angiogenic characteristics of MPNST cells (20), we demonstrate that STAT3 is a crucial mediator of these oncogenic activities downstream of MET. Our work indicates that targeting STAT3 further downstream might have therapeutic benefit for a larger percentage of the heterogeneous MPNST population, in contrast with targeting MET alone. Interestingly, we observed that PDGF stimulation also caused robust activation of MET, this is suggestive of signaling cross-talk between PDGF-R and MET. PDGFR-β expression is known to transform Schwann cells lacking NF1 (15, 35), implying that PDGF-Rβ amplification might be involved in Schwann cell hyperplasia. In a more recent study using a panel of 11 MPNST tumors, PDGF-Rα, PDGF-Rβ, and EGF-R were shown to be overexpressed when compared with benign controls correlating with a higher level of PI3K/Akt/mTORC1 signal transduction (24, 34). It is therefore possible that some of the transforming potential through PDGF-R might be via activation of MET.

Our work indicates that STAT3 and HIF1α are attractive “common” therapeutic targets for MPNSTs with heterogeneity in their migratory/invasive signaling profiles. Recently, Upadhyaya and colleagues (2012) used Affymetrix SNP 6.0 Array analysis to examine the genetic profile of MPNSTs compared with benign neurofibromas (15). The study reported MPNST-specific upregulation of seven Rho-GTPase pathway genes that are thought to be critically involved in MPNST development and metastasis. Signal transduction through STAT3 is also a critical driver of Rho and therefore may be contributing to this elevation (36, 37). It is apparent that many genetic alterations found in MPNSTs lead to amplification of signal transduction pathways that enhance either JAK2/STAT3 or mTORC1/STAT3 signaling, where HIF1α lies downstream of both STAT3 and mTORC1 (28). For instance, PTEN loss is known to occur in both MPNSTs and epithelioid sarcomas (38, 39) that gives rise to aberrant signaling through PI3K/mTORC1 (see Fig. 8 for signaling diagram). mTORC1 activation has also been implicated in MPNST tumorigenesis (33). Although mTORC1 inhibitors have shown some success in NF1 patients (40, 41), tumor regression did not occur via the usual mechanisms, with the proangiogenic factor HIF1α remaining elevated under rapamycin treatment (33). Of interest, we also observed that HIF1α protein levels were not completely ablated with rapamycin treatment in all four MPNST cell lines tested. Given that HIF1α, HIF2α, and VEGF-A protein levels where instead markedly reduced after STAT3 knockdown, our data indicate that the HIF/VEGF pathway is predominantly regulated through JAK2/STAT3 in a range of MPNST cell lines. We previously demonstrated that both the mTORC1/STAT3 and JAK2/STAT3 signaling pathways converge on STAT3 for maximal STAT3 activation, with JAK2 mediating Tyr705 phosphorylation and mTORC1 mediating Ser727 phosphorylation (28). These data taken alongside our evidence suggest that targeting JAK2/STAT3 and mTORC1 pathways in parallel would likely be required to suppress STAT3 and HIF1α and reduce the angiogenic phenotype in MPNSTs.

Although evaluation of molecular abnormalities in tumors on an individual basis might help design tailor made therapy, there are limitations to this approach with regards to therapy with MPNSTs. As a result of intratumoral molecular heterogeneity in MPNSTs, tumor profiling may not be possible. Instead, it may be more feasible to develop a therapeutic strategy that targets multiple signaling pathways, which are commonly dysregulated in MPNSTs. Given the high level of dependency on both STAT3 and HIF1α for cell migration, invasion, and tumor formation in multiple MPNSTs, therapeutic strategies that target the STAT3/HIF/VEGF-A pathway or pathways that converge on STAT3 could be a viable option for treating NF1 patients.

No potential conflicts of interest were disclosed.

Conception and design: E. Rad, M. Upadhyaya, A.R. Tee

Development of methodology: E. Rad, L.E. Thomas

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Rad

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Rad, K.M. Dodd, A.R. Tee

Writing, review, and/or revision of the manuscript: E. Rad, K.M. Dodd, L.E. Thomas, M. Upadhyaya, A.R. Tee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Rad, M. Upadhyaya

Study supervision: M. Upadhyaya, A.R. Tee

The authors thank Pamela Bramble from Pfizer for providing the MET inhibitor, PF-4217903 and Professors Mautner and Guha for MPNST cell lines.

This work was funded by the Association for International Cancer Research (Career Development Fellowship (No. 06-914/915; to A.R. Tee), Tuberous Sclerosis Association (to K.M. Dood and A.R. Tee) and the Ian Owen Trust (to M. Upadhyaya). E.M. Rad's Ph.D. studentship is funded through the Cancer Genetics Biomedical Research Unit (CGBRU), which is supported by the NIH and Social Care Research (NISCHR) from the Welsh government. Further support was provided via Wales Gene Park through NISCHR.

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.