Loss of NF2 Induces TGFβ Receptor 1–mediated Noncanonical and Oncogenic TGFβ Signaling: Implication of the Therapeutic Effect of TGFβ Receptor 1 Inhibitor on NF2 Syndrome

This article is featured in Highlights of This Issue, p. 2269

NF2 (Neurofibromatosis type 2) syndrome is a very rare genetic disorder, where patients are afflicted by Schwannoma (benign tumor of Schwann cell) in the peripheral nervous system (1). Loss of hearing is common feature of NF2 syndrome, due to presence of tumor in the vestibular nervous system. Furthermore, tumors are also detected in the spinal ganglion in patients with NF2 (2). Genetic alteration of NF2 is revealed a loss of the NF2/Merlin (2). At the cellular level, NF2/Merlin functions as a linker between the cytoskeleton and membrane (3, 4). In addition, it is suggested to be a negative regulator of the Hippo/YAP pathway (5), Ras signaling cascade, and mTORC2 (6). Thus, according to previous literatures, it is clear that NF2/Merlin is an important tumor suppressor protein. However, these functions are unable to explain why loss of NF2/Merlin induces Schwannoma without the development of a general carcinoma.

In our previous study, we revealed that treatment of silica activates the Erk–Snail signaling and suppresses the p53 pathway via RKIP reduction in mesothelioma (7). In addition, loss of NF2 promotes RKIP reduction and suppresses p53 activation (7). Indeed, NF2 is an important regulator of physical stresses and is frequently found to be deleted in asbestos-induced mesothelioma (5, 6).

During the development of the neuro-system, Schwann cells (peripheral nerve system) and oligodendrocytes (central nerve system) need to physically bind to the axon of neurons. Indeed, contact of Schwann precursor cells and neural axons is an important trigger for differentiation of Schwann cells (8). For Schwann cell differentiation and cell death, TGFβ signaling is critical (9). In particular, there is an association between TGFβ signaling and axonal neurogulin-1 during the attachment of Schwann cell to neural axon. Considering this fact, it is plausible that TGFβ signaling components, including TGFβ receptors, would be associated with the physical contact–induced signaling cascade.

Mesothelioma and Schwannoma commonly occur due to the deregulation of the physical contact–induced signaling cascade. Malignant mesothelioma occurs in the thin layer surrounding the organs, namely mesothelium (10), and some weighted material (maybe asbestos) are implicated as the causative agents of this cancer. Thus, we hypothesize that NF2 would work as an important regulator of physical contact–induced cellular signaling, with an additional involvement of TGFβ signaling.

To examine the physical stimulation induced signaling cascade and relevance with NF2 syndrome, we investigate the pressure-induced cellular response and role of NF2 in this signaling network. Although asbestos is the best trigger for mesothelioma-related physical stimulation, its use is completely forbidden for any purpose. Thus, we used silica as a replacement for asbestos (11–13). In addition, pressure was generated on the cell by overlaying with a coverslip. Our experiments indicate that TβR1, but not TβR2, works as a receptor for silica and pressure; loss of NF2 induces the down regulation of TβR2; in the absence of TβR2, physical stimulation activates TβR1, which in turn induces RKIP reduction and p53 inactivation. Taken together, we believe that inhibition of TβR1 can induce cell death in NF2-deficient Schwannoma as well as mesothelioma.

ACHN (NF2 mutant) and MDA-MB-231 (NF2 mutant) cells were purchased from the Korean Cell Line Bank. A549, HEK293, NCI-H28, NCI-H2452, and MDA-MB-468 cell lines were obtained from ATCC. LNcaP (TβR1 null) was provided by Dr. S.G. Chi (Korea University, Seoul, Korea). HCT116 cell lines were obtained from Dr. B. Vogelstein (Johns Hopkins University, Baltimore, MD) and human Schwannoma cell line HEI-193 (NF2 deficient) was provided by Dr. G. Zadeh (University Health Network, Toronto, Canada). Mouse schwann cells derived from NF2^flox mice (14) were subjected to in vitro Cre-mediated deletion and then cells were transduced with pMSCV-hygro retroviral rescue constructs encoding either full-length Merlin isoform 1(MSchw-WT), the truncated version that was encoded by the SKY-mutant allele (MSchw-SKY), or the empty vector (MSchw-KO). The cell lines were kindly provided from Dr. P. Greer P (Queen University, Ontario, Canada). All cells were maintained in a 5% CO₂ humidified incubator at 37°C. ACHN, HEI-193, HEK293, L132, LNCaP, and MDA-MB-468 were cultured in liquid DMEM supplemented with 10% FBS and 1% antibiotics. A549, HCT116, NCI-H28, NCI-H2452, and MDA-MB-231 were cultured in RPMI1640 supplemented with 10% FBS and 1% antibiotics. Primary tumor cells isolated from the NF2 syndrome model mice were cultured in DMEM supplemented with 15% FBS and 1% antibiotics. Mouse embryonic fibroblast (MEF) cells were isolated from 14.5-day embryos using a standard protocol and cultured in DMEM supplemented with 15% FBS and 1% antibiotics. The appearance and growth characteristics of all cells used in this study were compared with published information to ensure their authenticity.

All culture reagents used in the experiments were procured as follows, unless indicated otherwise: Silica (S5631; silicon dioxide), SB431542 and 4-OHT from Sigma Aldrich; adriamycin from Calbiochem; porcine TGF-β1 from R&D Systems; TEW7197 was provided by Dr. S.J. Kim (CHA University, School of Medicine, Seongnam, Korea); other TβR1 kinase inhibitors (alectinib, PF-06463922, certinib, and crizotinib) from Selleck Chemicals.

To know how cells responded to physical stress, we used silica or cover glass. Cells were seeded at a density of 0.5 × 10⁴ cells on a cover glass (12-mm circle; thickness, 0.13–0.17 mm). The cover glass was flipped upside down and incubated for the indicated time in serum-free medium. The intensity of the pressure was changed by placing a second cover glass over the seeded cover glass. Silica was directly administered to 2 × 10⁵ cells under serum-free (SF) conditions. The schematic diagram is described in Supplementary Fig. S1A.

To obtain the silica-resistant A549 (A549T) cell line, A549 cells were seeded in a cell culture dish and exposed to high concentration silica (100 μg/mL). After 3–4 days incubation, the viable cells were washed with media and cultured. This method was repeated for one month to obtain surviving cell population resistant to silica. Experiments involving A549T were performed using the resistant cell line thus developed.

pCMV RKIP-HA was provided by G. Keum (David Geffen School of Medicine at University of California, Los Angeles, CA). pCMV RKIP-T101A-HA and pCMV RKIP-T101D-HA were generated by Dr. J.H. Hwang (Pusan National University, Pusan, Korea). Snail-FLAG vector were provided by Dr. M.C. Hung (MD Anderson Cancer Center, Houston, TX). The pcDNA3 NF2-FLAG, pRK5 TGFβ type 1 receptor-FLAG, pRK5 TGFβ type 1 receptor-CA (T202D)-FLAG, pCMV5b TGFβ type 2 receptor-HA, and pWZL E-cadherin-GFP were obtained from Addgene. HA-p53 vectors were a kind gift from Dr. Kim S (Seoul National University, Seoul, Korea). Transfection was performed using the Jetpei transfection agent (Polyplus New York) for mammalian expression of these vectors. As described previously (15), cells were seeded at a density of 2 × 10⁵ cells per well in 12-well plates, and cultured overnight before transfection. The vector (1.5 μg) was mixed with 1.5 μL of Jetpei reagent in 150 mmol/L NaCl solution. The mixture was incubated for 15 minutes at room temperature, after which it was added to the cells. After 3 hours, the SF medium was replaced by medium supplemented with 10% FBS.

For in vitro gene knockdown, si-RNA against target proteins were generated (Cosmo Genetech).

Si-NF2-1: 5′-ATCATGATCCAGTACCTTCTTGTCC-3′

Si-NF2-2: 5′-CAGCCTGTCTTTCGACTTCAA-3′

Si-Snail: 5′- GCGAGCTGCAGGAC

The INTERFERin transfection reagent (Polyplus New York) for siRNA was used for transfection. Briefly, cells were seeded at a density of 2 × 10⁵ per well in 12-well plates and incubated overnight before transfection. A total of 1.5 pmol/L (21 ng) of siRNA duplexes were mixed with 4 μL of INTERFERin in 100 μL medium without serum. The mixture was incubated for 15 minutes at room temperature to allow the formation of INTERFERin/siRNA complex. This mixture was then added to the cells and incubated for a further 4 hours, after which the SF medium was replaced by medium supplemented with 10% FBS.

As described previously (16), proteins were extracted from cells with RIPA buffer (50 mmol/L Tris-Cl, pH 7.5, 150 mmol/L NaCl, 1% NP-40, 0.1% SDS, and 10% sodium deoxycholate). Samples were applied to SDS-PAGE and transferred onto a polyvinylidene difluoride membrane. Blotted membranes were incubated with primary antibodies for 1 hour to overnight at 4°C, followed by incubation with horseradish peroxidase (HRP)-conjugated species-matched secondary antibodies for 1 hour at room temperature. Peroxidase activity was detected by chemiluminescence using the ECL kit (Intron). For immunoprecipitation (IP) analysis, whole-cell lysates were incubated first with the appropriate antibodies for 4 hours at 4°C, and then with protein A/G agarose beads (Invitrogen) for 2 hours at 4°C. After centrifugation and washing with RIPA, the precipitated immunocomplexes were subjected to SDS-PAGE and Western blot analysis. The following antibodies were used in this study: HA (sc-7392), GFP (sc-9996), GST (sc-138), RKIP (sc-5426), Snail (sc-28199), p53 (sc-126), NF2 (sc-331), TGFβ R1(sc-398), TGFβ R2(sc-400), Vimentin (sc-7557), Actin (sc-1616), and E-cadherin (sc-8426), all purchased from Santa Cruz Biotechnology. Anti-phospho-Erk (9101L), anti-total-Erk (9102), and anti-p-smad2/3 (8828S) were obtained from Cell Signaling Technology. Anti-FLAG (F3165) was obtained from Sigma-Aldrich.

Cells were seeded on a cover glass and transfected with the indicated vectors or treated with the appropriate chemicals. After fixing with Me-OH for 30 minutes, the cells were incubated with blocking buffer [PBS + anti-human-antibody (1:500)] for 1 hour. Cells were then washed with PBS, incubated with anti-γ-tubulin antibody in blocking buffer (1: 100 to 200) for 4 hours, and subsequently with FITC-conjugated or Rhodamine-conjugated secondary antibodies in blocking buffer (1: 500) for 2 hours. Nuclear staining was achieved with DAPI. After washing with PBS, cover glasses were mounted with mounting solution (Vector Laboratories), and the immunofluorescence signal was detected through fluorescence microscopy (Zeiss; ref. 17).

Glutathione S-transferase (GST)-pull down assay and immunoprecipitation (IP) assay were performed to evaluate the protein–protein interaction. For GST-pull down, agarose-bead–conjugated GST (negative control) or GST-target protein was incubated with cell lysate or recombinant protein in RIPA buffer for 1 hour at 4°C. IP assay was performed with cell lysate or recombinant protein with RIPA buffer. The whole lysates were incubated with appropriate primary antibodies for 2 hours at 4°C, and reacted with agarose bead–conjugated protein A/G (Invitrogen) for 2 hours. After centrifugation, the precipitates were washed with RIPA buffer twice and subjected to SDS-PAGE and Western blot analysis (17).

For analyzing the association of silica with cell surface proteins, precipitation assay was performed using heavy materials (FeO₂ and silica). The silica was treated to react with cell surface protein 4 hours before lysis using RIPA buffer. To avoid artifacts, the same amount of heavy material (40 μg/mL) was added to the control sample. Furthermore, protein A/G-agarose (Invitrogen) was also added to the control sample and the whole-cell lysate. Immediately, centrifugation was performed. The precipitated material was washed with RIPA buffer and analyzed by SDS-PAGE and Western blot analysis (Fig. 1E; Supplementary Fig. S2A and S2B).

To check the phosphorylation of RKIP by TβR1, whole-cell lysates transfected with Flag-TβR1 were incubated with 15 mg of GST-fused RKIP protein for 4 hours at 37°C in 25 mL of 50 mmol/L Tris (pH 8.0) buffer containing 1 mmol/L ATP, 10 mmol/L MgCl₂, and 10 mmol/L 2-mercaptoethanol. Western blot analysis was performed to detect phosphorylated RKIP using anti-phospho-Smad2 (Ser465/467)/Smad3 (Ser423/425) antibody (Cell Signaling Technology; refs. 18, 19).

To measure the cell viability, cells were treated with the indicated chemicals for 4 days. For the MTT assay, cells were incubated with 0.5 mg/mL of MTT solution (Calbiochem) for 4 hours at 37°C. After removing the excess solution, the precipitated material was dissolved in 200 μL DMSO and quantified by measuring the absorbance at 540 nm.

To estimate the transcriptional activity of TGFβ signaling, 3TP-Luc vectors were transfected into cells for 24 hours, and cells were exposed to the indicated chemicals for appropriate duration. After washing with wash buffer (Promega), the cells were lysed using lysis buffer (Promega), and luciferase activity was determined by a luminometer (MicroDigital).

For the analysis of in vitro cell migration, transwell assay was performed using Polycarbonate Membrane Transwell Inserts (3422; Corning). Briefly, 0.6-mL media containing 10% FBS was added to each well of the well plate. Cells were resuspended in serum-free medium, and 0.1 mL of the cell suspension was added to the inside compartment. The plate was incubated with or without silica (10 μg/mL, 60 hours) in 5% CO₂ incubator. Cells in the top chamber were then discarded, and the attached cells in the bottom section were fixed using 4% PFA for 30 minutes. After fixation, migratory cells were stained by 0.05% Trypan blue solution (Gibco) and quantified. The migration rate was quantified by counting the migration cells in six random fields using a light microscope.

For RT-PCR, total cellular RNA was extracted using the QiagenRNA extraction kit. After measuring the RNA concentration, 1 μg of total RNA was reverse transcribed into cDNA using the MMLV RT (Invitrogen) and a random hexamer. RT-PCR was performed with the specific primers. The primer sequences were as follows:

HMGCS1 Forward 5′-TCCCACTCCAAATGATGACA-3′

HMGCS1 Reverse 5′-CTTCAGGTTCTGCTGCTGTG-3′

LDLR Forward 5′-TCTGTCGTGTGTGTTGGGAT-3′

LDLR Reverse 5′-ACGACAAGATTGGGGAAGTG-3′

INSIG1 Forward 5′-CAACACCTGGCATCATCG-3′

INSIG1 Reverse 5′-CTCGGGGAAGAGAGTGACAT-3′

DDIT4 Forward 5′-CGAGTCCCTGGACAGCAG-3′

DDIT4 Reverse GGTCACTGAGCAGCTCGAAG-3′

MMP2 Forward 5′-CCCTGATGTCCAGCGAGTG-3′

MMP2 Reverse 5′-ACGACGGCATCCAGGTTATC-3′

GAPDH Forward 5′-ATCTTCCAGGAGCGAGATCCC-3′

GAPDH Revere 5′-AGTGAGCTTCCCGTTCAGCTC-3′

All experimental procedures using laboratory animals were approved by the animal care committee of Pusan National University (Busan, Republic of Korea). NF2 (FVB/NJ) mice were obtained from Dr. D.W. Clapp (Indiana University, Indianapolis, IN). Before commencing the experiments, all mice were maintained under temperature- and light-controlled conditions (20–23°C, 12/12-hour light/dark cycle) and provided autoclaved food and water ad libitum. The Postn-cre transgene was detected by PCR analysis using the following primers:

Postn-Cre Forward 5′-ATG-TTT-AGC-TGG-CCC-AAA-TG-3′

Postn-Cre Reverse 5′-CGA-CCA-CTA-CCA-GCA-GAA-CA-3′

Nf2^flox.flox and Nf2^Δ bands were detected by PCR analysis with the following primer.

Full-length NF2 FORWARD 5′-CTTCCCAGACAAGCAGGGTTC-3′

Full-length NF2 Reverse5'-GAAGGCAGCTTCCTTAAGTC-3′

ΔNF2 Forward 5′-CTCTATTTGAGTGCCTGCCATG-3′

ΔNF2 Reverse 5′- GAAGGCAGCTTCCTTAAGTC-3′

Band sizes and primers have been previously described by Giovannini and colleagues (14, 20).

NF2 (3-month-old, N = 10) mice were intraperitoneally (i.p.) administered the carrier, TEW7197 (5 mg/kg), 3 times a week. After terminating the experiment of each group, mice were dissected and tumors were isolated. Throughout the experimental period, the body weight were measured 3 times a week.

To acquire [¹⁸F] -FDG PET/CT image, each mouse was fasted at least 6 hours. [¹⁸F] FDG (500 ± 23 μCi) was intravenously administered through tail vein. After administration of [¹⁸F] FDG, mouse was placed in the dimmed lighted cage for 60 minutes. Each mouse was maintained under anesthesia with isoflurane (2.5% flow rate) for the duration of the scan. Animals were positioned prone in the standard mouse bed. Limbs were positioned lateral to the body to acquire uniform CT images. Whole-brain CT images were acquired with a Micro-PET/CT scanner (nanoPET/CT, Bioscan Inc.). For CT image acquisition, the X-ray source was set to 200 μA and 45 kVp with 0.5 mm. The CT images were reconstructed using cone-beam reconstruction with a Shepp filter with the cutoff at the Nyquist frequency and a binning factor of 4, resulting in an image matrix of 480 × 480 × 632 and a voxel size of 125 μm (21).

At the end of the experimental period, the mice were sacrificed, and the tumors were dissected; tissues were fixed using 10 % formalin in PBS for 24 hours and embedded in paraffin blocks according to the basic tissue processing procedure. For histologic analysis, the embedded tissues were cut in 5-μm slices using a Leica microtome, and transferred onto adhesive-coated slides (Marienfeld Laboratory Glassware). After deparaffinization and rehydration, sections were stained with hematoxylin and eosin for routine examination. For IHC staining, the rehydrated tissue sections were incubated with antibodies to TGFβ R1 (sc-398) and TGFβ R2 (sc-400). Antigen retrieval was performed using 10 mmol/L sodium citrate (pH 6.0), twice at 95°C for 10 minutes each, and endogenous peroxidase activity was blocked with 3% hydrogen peroxidase for 10 minutes. The treated slides were then dehydrated following a standard procedure, and sealed with cover glass using mounting solution (22, 23).

For differential gene expression analysis by NF2 transfection and TEW7197 treatment, microarray was performed using the Affymetrix GeneChip (Human Gene 2.0 ST Array; DNA Link, Inc). Total RNA obtained from HEI-193 transfected with NF2 (about 60% transfection yield) or treated with TEW7197 (10 μmol/L for 12 or 24 hours) was quantified and used for analysis. Array data extract were performed using Affymetrix GeneChip Command Console Software (AGCC). For differentially expressed gene (DEG) analysis, probes were selected if the difference in expressions were 1.5 fold compared with the HEI-193 control, and were statistically filtered by t test (P < 0.05). Fold value was calculated in this way [absolute fold change = (case/control) if case > control, (control/case) if control > case]. For the functional analysis, only DEG probes were used. Web tool DAVID (24, 25), database of which are Gene Ontology, KEGG, BIOCARTA, and OMIM_DISEAE, was used for functional grouping of probes. All statistical and functional analysis were performed by DNA Link, Inc. The full dataset is available in the NCBI's Gene Expression Omnibus (GEO) database (GEO GSE115359; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE115359).

All results are expressed as the mean + SEM and performed at least n = 3 per group. Student t test was performed for obtaining the statistical significance.

To generate physical stress, cells were seeded on cover glass, which was subsequently flipped upside down (Supplementary Fig. S1A). The effect of NF2 on pressure-induced cell viability was assessed first by using mouse embryonic fibroblast (MEF), obtained from Ubc-Cre-ER; NF2^f/f mouse. In the absence of 4-OHT, NF2 was normally expressed (Supplementary Fig. S1B) and viability of MEFs gradually decreased with increasing incubation time (Fig. 1A). However, elimination of NF2 expression by 4-OHT treatment (Supplementary Fig. S1B) increased the cell viability (Fig. 1A). Pressure-induced p53 and RKIP induction were not observed in NF2-deficient MEFs (Fig. 1B). To confirm this, the same experiment was performed using HEI-193 cells (NF2-deficient Schwannoma cell line; refs. 26, 27). Resistance to pressure-induced growth suppression was restored by reexpression of NF2 (Fig. 1C). In addition, pressure-induced RKIP and p53 reduction was abolished by NF2 transfection in HEI-193 cells (Fig. 1D). We also observed that the reduction of p53 and RKIP in response to silica treatment (Supplementary Fig. S1A) that induced RKIP reduction in mesothelioma, was completely blocked by NF2 restoration (Fig. 1E). Numerous cell lines were evaluated in this experiment. In nontransformed L132, pressure suppressed the cell viability. In contrast, ACHN cells (NF2-mutated kidney cancer cell line; 28) were resistant to physical stimulus (Supplementary Fig. S1C), and HCT116 was partially resistant to physical stimulation (29). Our results indicate that elimination of NF2 or RKIP provides the resistance to pressure-induced growth suppression in L132 (Supplementary Fig. S1D), indicating that RKIP and NF2 are critical for physical stress–induced signaling.

Although physical stresses reduce NF2 and RKIP, it is as yet unknown how cells recognize these stimuli. To address this, we checked the association of silica with cell surface proteins. Among tested proteins (E-cadherin, TβR1, TβR2, and Her2/Neu), only TβR1 (Fig. 1F; Supplementary Fig. S1F) showed dose-dependent expression to silica (Supplementary Fig. S2A). To avoid artifacts, we the precipitation assay was repeated using FeO₂ and protein A/G-agarose. Similar to silica, FeO₂ coprecipitated with TβR1 (but not Her2/Neu; Supplementary Fig. S2B). However, protein A/G-agarose beads showed no interaction with TβR1 (Supplementary Fig. S2B). Treatment of silica reduced the RKIP and NF2 expression in TβR2-deficient HCT116, but not in TβR1-deleted LNCaP (refs. 30–32; Fig. 1G), suggesting that TβR1 is responsible for physical stimulation. This result is consistent with our previous result that HCT116 is resistance to pressure-induced growth suppression (Supplementary Fig. S1C). Indeed, TβR1 transfection, but not TβR2, suppressed the RKIP and NF2 expression and induced Snail expression (Fig. 1H). Moreover, constitutively active TβR1 was more effective on RKIP, NF2, and Snail expression than wild-type TβR1 (Supplementary Fig. S2C). Furthermore, we observed a dose-dependent reduction of RKIP and NF2 in response to TβR1 (Supplementary Fig. S2D). Silica-induced RKIP and NF2 reduction were also observed in elevated TβR1-expressed cell lines, as compared with TβR2 (Supplementary Fig. S2E). In addition, TβR1 expression was induced by silica treatment (H28, HEI-193 in Supplementary Fig. S2E). However, silica treatment did not activate 3TP-luciferase (Supplementary Fig. S3A) or induce the R-Smad phosphorylation (Supplementary Fig. S3B) or translocation (Supplementary Fig. S3C). Next, we monitored the involvement of Hippo/Yap pathway via analysis of the promoter activity of CTGF, target of YAP-TEAD in NF2-deficient human Schwannoma (HEI-193) and mouse Schwann cell (Mschw). But, treatment of silica did not alter the promotor activity (Supplementary Fig. S3D). This result indicated that silica did not activate Hippo/Yap pathway. Although our results indicate that physical stress uses the unusual TβR1-mediated signaling, it is as yet unknown how TβR1 recognizes silica or heavy material.

To evaluate the effect of NF2 on TβR1-mediated signaling, we assessed the effect of siNF2 on TβR1 and TβR2 expression. The endogenous and exogenous TβR2 expressions were reduced by eliminating NF2 using siRNA in HEK293 and HCT116 cells (Fig. 2A and B; Supplementary Fig. S4A), and a similar reduction was also observed in NF2 knockout MEFs (Fig. 2C). NF2-deficient MEFs showed the reduction of RKIP and TβR2 and induction of p-Erk in response to silica (Fig. 2D). Conversely, the restoration of NF2 into HEI-193 induced the expression of TβR2 expression as well as RKIP and p53 (Supplementary Fig. S4B). We obtained the similar results from NF2-transfected ACHN (Supplementary Fig. S4C) and si-NF2–transfected L132 (Supplementary Fig. S4D), where NF2 enhanced the TβR2 expression. Pulse-chase analysis revealed that NF2 extended the half-life of TβR2 (Fig. 2E), but not TβR1 (Supplementary Fig. S4E and S4F). Indeed, the proteasome inhibitor, MG132 could block the reduction of NF2, RKIP, and TbR2 without altering the mRNA expression (Supplementary Fig. S4G). To confirm the role of NF2 in TβR2 expression and RKIP, we examined the expression of these proteins in MSchw cell (mouse NF2-deficient Schwann cell line) and its stable transfected cell lines with empty (MSchw-EV), wild-type NF2 (MSchw-WT), and mutant NF2 (terminated by a splice mutation that causes a R477C; MSchw-SKY). We observed that reductions of RKIP, p53, and TbR2 by silica treatment were suppressed in the MSchw-WT cell line (Supplementary Fig. S4H), strongly supporting the hypothesis that NF2 is a critical factor for stability of RKIP and TβR2. In addition, from this experiment, we observed that silica suppressed p-Smad2/3 in regardless of NF2 status. We also checked the p-YAP S127 expression at the same samples. Although induction of p-YAP S127 (inactivation form) was obviously detected by NF2 transfection, treatment of silica did not alter the p-YAP S127 expression (Supplementary Fig. S4H). Taken together with our previous results that silica treatment suppressed p-Smad2/3 in A549 (Supplementary Fig. S3B), silica itself was not related with canonical TGFβ signaling as well as Hippo/Yap pathway.

Because reduction of RKIP induces Snail and promotes EMT (7, 33, 34) and numerous malignancies are known to harbor TβR2 mutation (28, 29, 35, 36), we evaluated the effect of silica on cancer cell invasion using the transwell assay. As expected, silica promoted cell invasion (Fig. 2F; Supplementary Fig. S5A). Interestingly, TβR2 expression blocked the cancer cell invasion by promoting the NF2 and RKIP overexpression. Indeed, silica and pressure suppressed E-cadherin (Supplementary Fig. S5B and S5C) and induced the mesenchymal marker vimentin (Fig. 2G; Supplementary Fig. S5B). We obtained similar results from the E-cadherin luciferase system, wherein the reduction of E-cadherin by silica was eliminated by TβR2, RKIP, and NF2 expression (Supplementary Fig. S5D). However, elimination of Snail did not suppress the cell viability of HEI-193 (Supplementary Fig. S5E), suggested that induction of Snail would be related only with EMT, but not cell proliferation at NF2-deficient condition. Because TβR1-mediated physical stress–induced signaling reduced RKIP and NF2 expression, we assumed that balance between TβR1 and TβR2 could block the RKIP and NF2 reduction. In fact, TβR2 transfection could block the silica-induced RKIP, NF2, and p53 reduction (Fig. 2H). These results suggest that TβR1-mediated signaling cascade is oncogenic, and TβR2 has the capability to block it.

Because reduction of RKIP is a critical event in the TβR1-mediated oncogenic pathway, we speculated how TβR1 alone suppresses RKIP. To address this, we assessed the interaction of RKIP and TGFβ signaling components. Among the tested proteins, TβR1 showed a strong binding affinity with RKIP (Fig. 3A), which was inhibited in the presence of NF2 (Supplementary Fig. S6A and S6B). Also, inhibitors of TβR1 kinase (37–41) induced the RKIP and NF2 as well as p53 expression (Fig. 3B) and block the silica-induced RKIP and NF2 reduction (Fig. 3C). Furthermore, TβR1 inhibitor blocked the interaction of TβR1 and RKIP (Fig. 3D). We also detected the affinity of RKIP for p-Smad antibody when recombinant RKIP was reacted with cell lysate obtained from silica-treated TβR1-transfectant (Fig. 3E). This implies that RKIP is a target of TβR1 kinase. Analysis of the RKIP amino acid sequence reveals well conserved SSXS/T motif, which is the TβR1 kinase consensus phosphorylation site (38). To confirm this, we two different RKIP mutants were generated (T101A and T101D), and we found that phospho-mimetic RKIP mutant had a very low protein stability, and the RKIP T101A showed resistance to the silica-mediated reduction (Fig. 3F; Supplementary Fig. S7B). In addition, this mutant blocked the NF2 reduction (Fig. 3F; Supplementary Fig.S7B). Essentially, RKIP T101A was resistant to TβR1-mediated reduction (Supplementary Fig. S7C and S7D) and increased NF2 expression (Supplementary Fig. S7E and S7F). We further observed an increase in the TβR2 expression by RKIP transfection, particularly by RKIP T101A (Supplementary Fig. S7C and S7D).

This experiment reveals that inhibition of TβR1 kinase blocks the RKIP and NF2 reduction, implying that TβR1 inhibitor is useful for NF2-deficient tumors, such as NF2 syndrome Schwannoma and mesothelioma as well as TβR2-mutated cancers. Indeed, treatment of TEW7197 dose-dependently induced RKIP and p53 in Schwannoma (HEI-193), mesothelioma (H28), and HCT116 (Fig. 4A). In addition, TEW7197 blocked the TβR1-induced RKIP and NF2 reduction (Fig. 4B; Supplementary Fig. S8A); and TEW7197 and SB431542 induced the NF2, RKIP, and p53 expression in HCT116 (Supplementary Fig. S8B). In mesothelioma, cotreatment of TEW7197 and adriamycin exerted a synergic effect on p53 expression (Fig. 4C) and growth suppression (Fig. 4D). However, TEW7197 did not show growth suppression on restoring the NF2 in ACHN (Supplementary Fig. S8C) or normal MEFs (Supplementary Fig. S8D). Comparing with Rad001 that is recently employed in clinical trials for NF2 syndrome (42, 43), the TβR1 inhibitor exerted an antiproliferative effect on Schwannoma cells (Fig. 4E; Supplementary Fig. S8E), with obvious induction of p53 and RKIP in HEI-193 (Fig. 4F).

Because of the recent developments of numerous selective TβR1 inhibitors, we evaluated the effect to choose the ideal chemical candidate for NF2 syndrome and mesothelioma. First, we assessed the efficacy of 6 chemicals, including TEW7197 and SB431532 (Supplementary Fig. S9A), on the cell viability of HEI-193 (Fig. 4G), and H28 (Supplementary Fig. S9B). In both cell lines, certinib and crizotinib induced obvious effects, whereas alectinib and PF-06463922 were less effective than TEW7197 (Fig. 4G; Supplementary Fig. S9B). Although all tested chemicals, except crizotinib, could block the RKIP reduction by silica (Supplementary Fig. S9C), two chemicals (certinib and crizotinib) showed toxicity toward normal MEF, together with their obvious cytotoxic activity on NF2-null cells (Supplementary Fig. S9D). In addition, TEW7197 was most effective on RKIP induction in HEI-193 (Fig. 4H). Thus, we our results determined TEW7197 as an ideal drug candidate for further studies.

To know the global effect of TEW7197, we performed the microarray with HEI-193 (Supplementary Fig. S10A). Compared to the control, TEW7197 treatment altered the expression of about 5,000 genes, which differed on the basis of the treatment time. NF2 transfection also altered 4229 gene expressions (Fig. 5A). Of these, 1,183 genes were commonly altered by TEW7197 and NF2 transfection. Many of these were grouped as cholesterol biosynthesis or adipose tissue formation (Fig. 5A; Supplementary Fig. S10B–S10E). We also obtained similar results by Pathway & Biocarta analysis (Supplementary Fig. S10F–S10H), and confirmed the induction of several genes by RT-PCR (Fig. 5B). Because Schwann cells accumulated cholesterol and other lipids for myelin sheet formation, the gene expression profile indicates that TEW7197 and NF2 transfection induce Schwann cell differentiation (44, 45). Indeed, HEI-193 showed a differentiated phenotype after treatment with TEW7197 (Fig. 5C). To confirm this, we checked other Schwann cell differentiation markers such as sox2 and GFAP (46). As expected, TEW7197 suppressed the sox2 expression and GFAP induction (Supplementary Fig. S11A). In addition, MPZ, a marker of myelinated Schwann cells (46), was increased after exposure to TEW7197 (Supplementary Fig. S11B). We further confirmed the reduction of sox2 and induction of GFAP by treatment of TEW7197 in HEI-193 through IF staining (Supplementary Fig. S11C and S11D). These results strongly suggest that TEW-7197 promotes the differentiation of Schwannoma into Schwann cells. Next, we evaluated the expression of TβR1 and TβR2 in the NF2 syndrome mouse model for a more concrete and logical analysis. We observed a reduction of TβR2 in the dorsal root ganglion of Postn-Cre;NF2^f/f mice (Fig. 5D; Supplementary Fig. S12).

Because TEW7197 is a specific TβR1 kinase inhibitor that suppresses the viability of NF2-deficient cells, we tested the in vivo efficacy of TEW7197 in the NF2 syndrome mouse model (14, 20). TEW7197 is now in the phase II clinical trial, implying that there is severe toxicity. To do this, we injected 10 mg/kg of TEW7197 intraperitoneally (3 times/week) for 6–8 weeks (Supplementary Fig. S13A). Through the PET/CT analysis, we observed that TEW7197 suppresses the tumor formation (Fig. 5E; Supplementary Fig. S13B). Histologic analysis confirmed that TEW7197 diminishes the tumor formation (Fig. 5F). Indeed, large tumors presented in the dorsal root ganglion of Postn-Cre; NF2^f/f were not detected in TEW7197-treated mice (Fig. 5F; Supplementary Fig. S13C). Moreover, this chemical increased the body weight (Supplementary Fig. S13D). The vehicle-treated mouse presented with very large tumors (Supplementary Fig. S14A), from which we isolated and cultured the primary tumor cells. These cells also revealed the differentiated phenotype in response to TEW7197 (Supplementary Fig. S14B). Resistance to pressure-induced growth suppression was also overcome by TEW7197 (Supplementary Fig. S14C). These results strongly suggest that this specific inhibitor of TβR1 is a strong anticancer candidate for NF2- or RKIP-deregulated cancers, including mesothelioma and NF2 syndrome–related tumors.

This study revealed that physical stresses such as pressure or attachment of macromaterials activate the TβR1-mediated unusual oncogenic signaling (Fig. 5G). Activated TβR1 by physical stimuli induced phosphorylation-induced RKIP degradation. In contrast, NF2 induced TβR2 expression and maintains a balance between TβR1 and TβR2 (Fig. 5G). Thus, loss of NF2 induces a disturbance in the balance of TβR, and facilitates TβR1-mediated oncogenic signaling. Because we previously found that RKIP reduction by NF2 loss promotes Snail stabilization and p53 suppression (7), TβR1-mediated RKIP-NF2 reduction would suppress p53 activity. Indeed, we observed the induction of p53 by specific TβR1 inhibitor, accompanied with RKIP/NF induction (Fig. 4A). In addition, we found that TβR1-mediated oncogenic signaling promotes EMT and cell migration (Fig. 2F; refs. 47–49). We also observed the inhibitory effect on cell migration by transfection of TβR2 as well as NF2. These results indicate that rebalancing between TβR1 and TβR2 can block the TβR1-mediated oncogenic signaling, and that NF2-mediated stabilization of RKIP also blocks the TβR1-mediated oncogenic property. Because the final step of TβR1-mediated oncogenic signaling is achieved by reduction of RKIP, wild-type NF2 exerts an antioncogenic feature in TβR2-deficient conditions, such as in HCT116. Hence, HCT116 is moderately resistant to pressure-induced apoptosis (Supplementary Fig. S1C). These results also imply why cancer cells show frequent mutations in TβR2 (35, 50, 51).

We also demonstrate the association of FeO₂ with TβR1 (Supplementary Fig. S2B). This result indicates that any other heavy material is capable of activating the TβR1-mediated oncogenic signaling. Thus, this new oncogenic pathway may provide the answer for wood dust–induced nasal cavity carcinoma (52, 53) or asbestos-induced mesothelioma (54). Considering that micro-dust is significantly increased globally, an increase of related cancers induced by air pollution is therefore predicted. Thus, our mechanism would be useful for providing a potential therapy to cancers such as mesothelioma and lung cancer. Moreover, we have shown the TEW7197 suppresses tumor progression in the NF2 syndrome mouse model. In contact with neuro-fibers, Schwann cell precursor is differentiated into Schwann cell. On the basis of our hypothesis, this contact would be the physical stimuli. Under conditions of intact NF2, the Schwann cell precursor will differentiate into Schwann cell. However, under NF2-deficient conditions, this physical contact would trigger the TβR1-mediated oncogenic signaling and reduction of RKIP. Hence, inhibition of TβR1 kinase is one of the plausible targets for NF2 syndrome. This hypothesis is strongly supported by the outcome of our in vivo results.

Indeed, similar effects could be obtained after treatment with other TβR1 inhibitors (Supplementary Fig S9). The inhibitors selected for our experiments provided representative chemicals for similar backbone and clinical usage (all have approval for clinical trials from FDA). All the tested chemicals, except crizotinib, block the RKIP reduction and suppress the cell viability (except PF-06463922). However, alectinib and certinib also showed strong cytotoxicity in normal cells. Considering these results, TEW7197 would be the most plausible candidate as a therapeutic drug for NF2 syndrome. Alectinib and certinib would be more suitable for aggressive cancer treatments, such as mesothelioma.

Because clinical trial of TEW7197 has been approved, it would be applied to NF2 syndrome after adjustment of proper dosage with in near future. In addition, this chemical can also be applied to mesothelioma because this malignancy shows very low RKIP expression and/or NF2 deletion (7).

No potential conflicts of interest were disclosed.

Conception and design: J.-H. Cho, B.-J. Park

Development of methodology: J.-H. Cho, M.-H. Yoon, B.-J. Park

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.-D. Hong, H.-Y. Lee

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H.-Y. Lee, B.-J. Park

Writing, review, and/or revision of the manuscript: A.-Y. Oh, S. Park, B.-J. Park

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T.-G. Woo, J. Hwang, N.-C. Ha

Study supervision: S.-M. Kang, B.-J. Park

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Korea government (MSIP; NRF-2017R1A2B2007355; to B.-J. Park) and by the Bio & Medical Technology Development Program of the NRF funded by the Korean government (MSIP; NRF-2016M3A9D9945477).

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.