MEK inhibitors enhance therapeutic response towards ATRA in NF1 associated malignant peripheral nerve sheath tumors (MPNST) in-vitro

Objective Neurofibromatosis type 1 (NF1) is a hereditary tumor syndrome characterized by an increased risk of malignant peripheral nerve sheath tumors (MPNST). Chemotherapy of MPNST is still insufficient. In this study, we investigated whether human tumor Schwann cells derived from NF1 associated MPNST respond to all-trans retinoic acid (ATRA). We analyzed effects of ATRA and MEK inhibitor (MEKi) combination therapy. Methods MPNST cell lines S462, T265, NSF1 were treated with ATRA and MEKi U0126 and PD0325901. We assessed cell viability, proliferation, migration, apoptosis and differentiation as well as mRNA expression of RAR and RXR subtypes and ATRA target genes such as CRABP2, CYP26A1, RARB and PDK1. We also analyzed CRABP2 methylation in cell lines and performed immunohistochemistry of human MPNST specimens. Results ATRA therapy reduced viability and proliferation in S462 and T265 cells, accompanied by differentiation, apoptosis and reduced migration. NSF1 cells which lacked RXRG expression did not respond to ATRA. We furthermore demonstrated that ATRA signaling was functional for common targets, and that mRNA expression of CRABP2 and its targets was raised by ATRA therapy, whereas alternative pathways via FABP5 were not induced. Finally, combination of ATRA and MEKi demonstrated additively reduced viability of T265 and S462 cells. Conclusions We observed therapeutic effects in two of three MPNST cell lines pronounced by combination therapy. These data point to a potentially successful treatment of MPNST by combined application of ATRA and MEK inhibitors such as U0126 or PD0325901.


Results
ATRA therapy reduced viability and proliferation in S462 and T265 cells, accompanied by differentiation, apoptosis and reduced migration. NSF1 cells which lacked RXRG expression did not respond to ATRA. We furthermore demonstrated that ATRA signaling was functional for common targets, and that mRNA expression of CRABP2 and its targets was raised by ATRA therapy, whereas alternative pathways via FABP5 were not induced. Finally, combination of ATRA and MEKi demonstrated additively reduced viability of T265 and S462 cells. PLOS  Introduction of retina [28]. Due to its anti-carcinogenic activities ATRA is used in therapy and prevention of cancer such as acute promyelocytic leukemia (APL) and acute myeloic leukemia (AML) [29,30]. In-vitro and in-vivo effects of ATRA therapy were also seen in cervical cancer, lung cancer, colon adenocarcinoma, breast cancer, kidney cancer, neuroblastoma, germ cell tumors, and glioblastoma [31][32][33][34][35][36]. However, clinical use of ATRA can be hampered by retinoic acid resistance. Hence, ATRA is used in combination therapy utilizing potential synergistic effects. Administration of 13-cis RA significantly improved overall survival after consolidation therapy of neuroblastoma in children, and patients with relapse of APL showed a lower relapse rate when ATRA was combined with anthracycline [37,38]. Benefits of combination therapy were also shown for metastatic renal cell carcinoma (13-cis-RA + IFNA2) and acute myeloic leukemia (ATRA + valproic acid) [39,40]. In neuroblastoma cells, being resistant to ATRA due to loss of the zinc finger protein ZNF423, the inhibition of the MAPK cascade restored ATRA responsiveness. ZNF423 loss was addressed to NF1 loss that in turn was supposed to cause the over activation of the Ras-MEK signaling pathway in these cells [41,42].
Here, we investigated if treatment of tumor cells derived from NF1 associated MPNST respond to ATRA therapy. We furthermore analyzed, if combination of ATRA and MEK inhibitors (MEKi) enhance therapeutic effects. Thus, we elucidated involved signaling pathways and molecular mechanisms of ATRA in MPNST cells. Finally, our in-vitro data point to a very promising therapeutic approach for human MPNST by combining ATRA and MEKi application.

Primary cell cultures and cell lines
Normal primary human and rat Schwann cells, the MCF-7 cell line and three human NF1 associated MPNST cell lines (S462, T265, NSF1) were analyzed [43][44][45][46][47]. MPNST cell lines S462 and T265 were provided by VF Mautner (University of Hamburg). The NSF1 cell line was provided by D Kaufmann (University Hospital Ulm, Germany). The MCF-7 cell line was provided by V Senner (University Hospital Muenster, Germany). Experiments were performed in cooperation with VF Mautner and were approved for these purposes by ethics committee of "Ä rztekammer Hamburg" (19.04.2011, No. WF-018/11). Normal primary rat Schwann cells (nrSC) were provided by B Gess (University Hospital Muenster, Germany). Primary normal human Schwann Cells (nhSC) were purchased from ScienCell (HSC, California, USA). Primary Schwann cells were cultivated in standard Dulbeccos modified eagles medium (DMEM) containing 4.5 g/L glucose, 2 mM L-glutamine, 10% fetal bovine serum (FBS), 1 mM sodium pyruvate and 100 U/mL penicillin/streptomycin supplemented with 0.5 mM 3-isobutyl-1-methylxanthine (IBMX), 10 nM β1-heregulin, 2.5 μg/mL insulin and 2.5 mg/L fungizone. Culture vessels were coated with 1 mg/mL poly-L-lysine (10 min at room temperature) and 4 μg/mL laminin/PBS (2 hours at 37˚C). Enrichment of Schwann cells was achieved by mild trypsination and adhesion of fibroblasts to uncoated culture vessels. All primary cultures showed a proportion of Schwann cells (versus fibroblasts) of above 95% (S100 immunostaining). MPNST cell lines and the MCF-7 cell line were cultured using standard DMEM in uncoated vessels.

Treatment of cells with ATRA and MEK inhibitors (MEKi)
ATRA (Sigma-Aldrich), MEK inhibitors U0126 (NEB) and PD0325901 (Selleckchem) were dissolved in DMSO, aliquots were stored at -80˚C. Treatment was performed for 7 days unless otherwise stated (three to five replicate wells per sample) (

RNAi by transient transfection of siRNA
Transient transfection of siRNA (12 nM) was performed using HiPerFect reagent (Qiagen) according to the manufacturer's protocol. Simultaneously to transfection with specific siRNA, a non-specific siRNA (si-scrambled) was used as control. Cells were seeded the day before transfection.

Proliferation and migration assays
Cell proliferation and migration of cells was analyzed by xCELLigence DP system (Roche) [48,49]. Migration and proliferation of cells were detected by measurement of impedance of gold electrodes, that refers to "Cell Index". Migration assays were performed in CIM-plates 1 according to the manufacturer's instructions and monitored for 24 h. For analysis of migration after ATRA treatment, cells were pre-treated with ATRA (5 μM) / DMSO for two days before seeding for migration analysis under treatment conditions. Cell proliferation was monitored continuously in E-plates 1 for 120 h according to the manufacturer's instructions. All experiments were performed in triplicates with 4 replicates.

Viability assay
Viability of cells was determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT)-assay in 48-or 96-well plates. Cells were incubated with 0.5 g/L MTT labelling solution / medium for 2 or 3 hours, depending on cell line, Thereafter, medium was aspirated and cells were incubated with isopropanol for 5 min on an orbital shaker at 180 rpm. Measurement of absorbance was performed at 560 nm and at the reference wavelength of 750 nm in a GloMax 1 -Multi+ Detection System (Promega).

Apoptosis (TUNEL) assay and flow cytometry
Cells treated with 5 μM ATRA (7 d) were subjected to APO-BrdU TUNEL Assay kit (Invitrogen) according to the manufacturer's instructions for immune fluorescence and flow cytometry applications. For analysis of differentiation, cells were labeled with anti-S100 or anti-PMP22 antibodies and Alexa Fluor 488 secondary antibodies. Cells treated similarly but without primary antibody served as internal control. Cells were fixed in 1% ice-cold paraformaldehyde (PFA) and permeabilized in 70% ethanol before labeling with primary antibodies. Labeled cells were analyzed using a BD FACS Canto II. Each flow cytometry analysis was performed on 10.000 cells and experiments were performed 2-3 times.

Isolation and quantification of RNA and DNA
Total RNA from cultured cells was isolated using Gene Elute Mammalian total RNA Miniprep Kit (Sigma) according to the provided protocol including DNase digestion with the RNasefree DNase Set (Qiagen). Isolation of genomic DNA from cultured cells was performed using Quick-gDNA MiniPrep Kit (Zymo Research) according to the provided protocol. Concentration of RNA and DNA was determined by using a ND-1000 spectrophotometer (NanoDrop Technologies) using ND-1000 V3-8.4 software.

Polymerase Chain Reaction (PCR)
CDNA was prepared from 100-500 ng of total RNA with oligo (dT) primers and High Capacity cDNA-RT Kit including RNase inhibitor (Applied Biosystems) according to the manufacturer's instructions. PCR was performed in a PTC-200 Peltier Thermal Cycler (MJ Research) using the HotStarTaq DNA Polymerase Kit (Qiagen). Each preparation contained 10 x PCR Buffer, 2 mM MgCl 2 , 200 μM of each dNTP (dNTP Mix, PCR Grade, Qiagen), 1 U Taq Polymerase, 0.4 μM of each primer ( Table B in S8 Fig) and 1 μL cDNA template corresponding to 1-25 ng total RNA applied in cDNA-synthesis. Amplified cDNA fragments were separated in agarose (1.5%, 10 μg/mL ethidium bromide) and visualized in a Gel Doc EZ Imager (BioRad). Quantitative real-time-PCR (qRT-PCR) was performed in a StepOnePlus real time PCR system (Applied Biosystems). Endogenous GAPDH (glyceraldehyde 3-phosphate dehydrogenase) served as internal control. Expression levels of specific mRNA sequences were calculated relative to GAPDH levels using the ΔΔCT method described previously [50]. QRT-PCR using TaqMan Universal Master Mix (Applied Biosystems) was performed in a total volume of 20 μL, containing 1 μL Gene Expression Assay (Applied Biosystems) or 0.4 μM of each primer (

Methylation analysis of the CRABP2 promoter
Sodium bisulfite treatment of DNA was performed using the EpiTect 1 Fast DNA Bisulfite Kit (Qiagen) according to manufacturer's instructions. Primers (Tables C and D in S8 Fig) were designed using the PyroMark Assay Design 2.0 software (Qiagen). PCR was performed using the Maxima Probe qPCR Master Mix including Maxima Hot Start Taq DNA Polymerase (Thermo Fisher Scientific) in a total volume of 20 μL, containing 0.5 μM of each primer. Pyrosequencing was performed in a PyroMark Q24 system (Qiagen) according to the manufacturer's instructions, using the PyroMark Gold Q24 Reagents Kit (Qiagen). Fully methylated DNA and un-methylated DNA were used as controls. In order to visualize successful bisulfite conversion a T was added into the dispensation order. Quantification of CpG sites was performed with PyroMark Q24 software (Qiagen).

Western blot analysis
To obtain whole cell protein extracts, cells were washed twice with cold PBS and lysed with RIPA buffer containing protease inhibitors. Determination of protein concentrations was performed by using the Lowry method and the DCTM Protein Assay Kit (BioRad) according to the manufacturer's instructions. The absorbance at 750 nm was detected in a GloMax 1 -Multi+ Detection System [51]. Protein samples were separated by SDS-PAGE according to the Laemmli method at 120 V until the desired separation had been reached (1-1.5 h). To determine molecular weight of detected protein bands a prestained protein marker (Precision Plus Protein™ Dual Color Standards, BioRad) was loaded [52]. A tank-blotting chamber (Mini Trans Blot Cell, BioRad) was used to transfer proteins to a PVDF membrane (Roche) at 250 mA for 45 min. After transfer, the PVDF membrane was incubated with blocking solution followed by incubation with primary antibodies (

Immunohistochemistry and immunocytochemistry
Human MPNST specimens from NF1 patients and non-NF1 patients were kindly provided by Prof. W. Paulus (Institute of Neuropathology, University Hospital Muenster, Germany). These experiments were approved by ethics committee at "Ä rztekammer Westfalen-Lippe Münster"(2007-261-f-S). Written informed consent was provided for participation. MPNST cell pellets were frozen at -80˚C, fixed in 4% PFA and stored in 70% ethanol until paraffin embedding. Hematoxylin and eosin staining (HE) of FFPE samples was performed in an automated staining instrument Tissue Tek Glas and Prisma system (Sakura). For immunohistochemistry of FFPE samples, slices were deparaffinised and pre-treated with target retrieval solution (pH 6.1). Immunohistochemistry was performed using the DAKO REAL™ Detection Kit (K5001) according to the provided protocol and the automated staining instrument AutostainerLink 48 (Dako). Deparaffinised sections were incubated with primary anti-CRABP2 antibody (Abcam ab181255, monoclonal, from rabbit, 1:100). Staining was visualized by DAB staining followed by nuclear counterstaining with hematoxylin.

Statistical analysis
We used unpaired t-test, one-sided t-test, Wilcoxon test, Mann-Witney-U test, Kruskal-Wallis test and Students t-test for analyses. A p-value of < 0.05 was considered to be significant.

MPNST cell lines S462 and T265 are sensitive to ATRA treatment
We investigated if MPNST cells (NSF1, S462, T265) were sensitive to ATRA treatment. We first determined expression of RAR and RXR subtypes in MPNST cells by RT-PCR, normal human Schwann cells and fibroblasts, since intact expression of nuclear receptors is an essential requirement for mediation of ATRA effects: All receptor subtypes were expressed in normal Schwann and MPNST cells, except of the RXRG subtype which was absent in NSF1 cells (S1 Fig). We then investigated effects of ATRA therapy on viability at day 7 of treatment: A dose-dependent decrease of viability was seen in T265 and S462 cells, whereas NSF1 cells did not respond (Fig 1). Normal primary rat Schwann cells as well as ATRA sensitive MCF-7 cells were used as controls. MCF-7 cells were sensitive to treatment with ATRA, whereas normal rat Schwann cells were not (Fig 1). By pyrosequencing, we analyzed, if methylation status of the CRABP2 promoter region influenced response to ATRA treatment. In MPNST as well as in control cells methylation analysis of three different promoter regions of CRABP2 (S2A In conclusion, a markedly reduced viability and proliferation at concentrations > 10 μM was seen in 2/3 MPNST cell lines under ATRA treatment, indicating ATRA sensitivity in a subset of malignant tumor derived Schwann cells.

ATRA treatment induces morphological changes and differentiation
FACS analysis was carried out to analyze potential ATRA mediated differentiation of cells. The proportion of PMP22-and S100-positive Schwann cells after treatment was assessed ( Fig  2): Expression of S100 protein increased in all MPNST cells. A reduction of PMP22 expression was seen in 2/3 MPNST cell lines (T265, S462). Next, morphological changes resulting from ATRA treatment were assessed by FACS: all MPNST cell lines revealed significantly higher mean values concerning relative cell size (forward scatter) as well as granularity (sideward scatter) in treated versus untreated cells (S3 Fig). To summarize, anti-proliferative properties of ATRA in MPNST cell lines are accompanied by morphological changes and Schwann cell differentiation.

ATRA treatment reduces migration and induces apoptosis
To evaluate migration as a parameter for tumor invasion, we monitored MPNST cell lines under ATRA therapy in real-time (xCELLigence system). Whereas NSF1 cells did not show reduced migration due to ATRA therapy, a strong inhibition of migration was observed in cell line T265 and a minor effect was seen in cell line S462 (Fig 3). To evaluate induction of apoptosis, we performed TUNEL staining and a FACS based TUNEL assay: Increase of apoptosis was seen in T265 MPNST cells as the proportion of TUNEL positive cell nuclei increased in ATRA treated cultures compared to untreated controls (S4 Fig). A FACS based TUNEL assay demonstrated pro-apoptotic effects in 2/3 MPNST cell lines treated with ATRA: The strongest effect was seen in S462 cells showing an increase to 374% ± 35% in fluorescence intensity compared to untreated cells (Fig 4).
In summary, 2/3 MPNST cell lines showed reduction of migration due to ATRA treatment. Anti-proliferative properties of ATRA on tumor Schwann cells are mediated by induction of apoptosis.

ATRA therapy induces expression of target genes
To identify if ATRA signaling was functional in MPNST cell lines treated with ATRA, we analyzed expression of bona fide CRABP2 targets such as CRABP2 itself, CYP26A1 and RARB by qRT-PCR. Promoters of these target genes contain a RARE element and are therefore inducible by ATRA [53]. We found all targets to be induced in cells treated with ATRA confirming a transcriptional activation (Fig 5). Since ATRA is known to mediate effects not only via CRABP2 and RXR/RAR but also via FABP5 and PPAR/RXR (both reflecting opposing signaling pathways), we analyzed expression of FABP5 and its signaling target PDK1.     As CRABP2 is the major mediator of cellular ATRA response, its expression in a series of tumor Schwann cells had been analyzed in previous studies [21]. Here we demonstrated that Schwann cells in human MPNST (FFPE) as well as MPNST cell lines expressed CRABP2. To conclude, ATRA signaling was shown to be functional for targets of CRABP2 mediated ATRA signaling pathways. During ATRA therapy, mRNA expression of CRABP2 raised, whereas alternative pathways (via FABP5 and subsequent PDK1 up-regulation) were not induced.

MEK inhibitor treatment reduces viability
We first analyzed exclusive MEKi administration. MEKi treatment is known to decrease NF1 associated overactive MAPK signaling and to restore retinoic acid sensitivity in NF1 lacking neuroblastoma cells [41,42]. Therefore, we determined the effects of MEKi U0126 and PD0325901 treatment on viability of cells (MTT-assay): A dose dependent response was achieved by treatment with MEKi U0126 in MPNST cell lines T265, S462, and NSF1 (Fig 6A), although effectiveness was much lower in NSF1 cells. Treatment of tumor cells with MEKi . T265 cells and S462 cells highly responded to MEKi treatment with reduced viability, whereas NSF1 cells did not show enhanced sensitivity to MEKi treatment, compared to untreated control cells (black bars). Maximum reduction of viability was achieved at 10 μM of U0126 and 1 μM of PD0325901 (reduction down to 53% ± 7% and 2% ± 1% in T265 cells, and 59% ± 7% and 19% ± 11% in S462 cells) (mean value of 3-5 measurements in each cell line (A). Relative ERK-P expression in T265 cells after MEKi treatment was analyzed. Western blots of ERK-P, ERK and GAPDH antibodies are demonstrated exemplarily for T265 cells after treatment with 10 nM PD0325901 (+) compared to untreated cells (-). T265 cells treated with MEKi demonstrated strongly reduced phosphorylated ERK level (grey bar, 0.32-fold ± 0.11-fold) compared to untreated control cells (black bar, 1.00-fold ± 0.25-fold). PD0325901 treatment did not affect total ERK level in treated (grey bar, 1.04-fold ± 0.07-fold) versus untreated T265 cells (black bar, 1.00-fold ± 0.11-fold) (mean value of 3 repeated measurements). PD0325901 revealed a reduction of viability in T265 and S462 cells at lower doses, whereas NSF1 cells did not show marked sensitivity (Fig 6A). At 1000 nM viability of T265 cells was reduced to 2% ± 1% and viability of S462 cells was reduced to 19% ± 11%. NSF1 cells showed a slight reduction to 67% ± 10%.
To substantiate downstream effects of MEKi, we analyzed expression of phosphorylated ERK (ERK-P). Elevated levels of Ras-GTP and activation of ERK is typical for NF1 associated MPNST. In principal, ERK phosphorylation is mediated by MEK1/2, which can be inhibited by MEKi such as PD0325901. Since reduced ERK-P expression has already been shown for human S462 MPNST cells after treatment with PD0325901 [54] we here tested ERK-P in T265 cells as a proof of principle for our experiments: ERK1/2 phosphorylation was significantly reduced in treated versus untreated T265 MPNST cells (Fig 6B). Additionally, we analyzed a transcriptional co-activator for retinoic acid receptors, ZNF423, which has been described to be lost in human NF1 lacking neuroblastoma cells. Absence of ZNF423 was demonstrated to explain decreased retinoic acid response which was restored by MEKi treatment [41,42]. In contrast to these findings, we did not observe any induction of ZNF423 after PD0325901 treatment in all NF1 deficient MPNST cell lines. Unexpectedly, ZNF423 levels decreased under PD0325901 administration in T265 and NSF1 cells (S6 Fig). To summarize, MEKi treatment reduced viability in 2/3 MPNST cell lines and was shown to be functional since ERK1/2 phosphorylation was reduced. Contrarily to previous reports, ZNF423 levels decreased or were not affected at all by MEKi treatment.

ATRA administration boosts impaired viability resulting from MEKi therapy
Wecombined ATRA and MEKi treatment. Compared to monotherapy, combined therapy revealed a significant reduction of viability of T265 and S462 cells in an additive manner using both U0126 and PD0325901 (Fig 7). For example, administration of 10 nM PD0325901 and 10 μM ATRA reduced viability of T265 cells to 53% ± 8%, compared to 79% ± 7% under ATRA monotherapy or 81% ± 8% under MEKi monotherapy. NSF1 cells were not sensitive to mono-or combination therapy. Finally, we examined CRABP2, CYP26A1 and RARB gene expression after mono-and combination therapy by qRT-PCR. Bona fide targets of ATRAmediated signaling were induced in all MPNST cell lines, both under mono-and under combination therapy. Here, additive effects due to combination treatment were not seen (S7 Fig). In conclusion, combined treatment of ATRA and MEKi induced expression of bona fide targets of ATRA mediated signaling such as CRABP2, CYP21A1, and RARB. The combination of ATRA and MEKi U0126 or PD0325901 provokes additive effects on reduced viability of T265 and S462 cells, suggesting a new and potentially successful therapeutic option for MPNST.

Discussion
MPNST display aggressive and invasive growth and an early onset of metastasis [42]. Patients have a poor prognosis with 5 year survival rates between 26-60% [43][44][45][46]. About 40% of MPNST are associated with NF1, warranting regular checkup in NF1 patients with a high tumor load [47]. In our study we could demonstrate that in-vitro combination of ATRA and MEKi reduces proliferation of tumor Schwann cells derived from NF1 associated MPNST. Reduced proliferation by ATRA administration alone, seen in MPNST cell lines T265 and S462, was markedly intensified by combination with MEKi U0126 or PD0325901. MEKi treatment alone reduced viability in T265 and S462 MPNST cells accompanied by reduced ERK1/2 phosphorylation. Further studies should address if these combined effects are also achieved at doses of ATRA that are used for clinical therapy such as 1-2 μM. Compared to benign neurofibromas MAPK signaling is further elevated in MPNST cells which contributes to their aggressive malignancy. MAPK inhibition seems to further repress tumor cell survival combined with ATRA treatment. Our results indicate a therapeutic potential of ATRA. Combination with MEKi constitutes a novel approach for NF1 associated malignant peripheral nerve sheath tumors.
Interestingly, ATRA was shown to display anti-proliferative properties by induction of apoptosis, differentiation and morphologic changes and by reduction of cell migration (Table 1). ATRA sensitivity remarkably differed between the cell lines with most pronounced effects in S462 and T265 cells, which warrant further investigations. We demonstrated that ATRA induces moderate anti-proliferative effects in MPNST cells T265 and S462 via apoptosis. These findings are in concordance with other studies describing induction of apoptosis via increased expression of pro-apoptotic genes such as CASP7 and CASP9 in MCF-7 cells. Interestingly, CASP9 contains a functional RARE which implicates this gene to be a direct target of CRABP2-mediated ATRA signaling [33]. Reduced viability of MPNST cells treated with ATRA was accompanied by reduced migration properties and signs of differentiation. The latter was also recently shown for neuroblastoma cells, underlining that induction of differentiation and pro-apoptotic signaling by retinoic acid are the reasons for its general usage in oncology and dermatology [55,56]. We think that apoptosis that we measured at day 7 after treatment may not largely influence migration of cells in our assay which was performed within the first 24h after treatment. To completely rule out an influence an apoptosis on migration behavior an apoptosis inhibitor might be applied when measuring migration of MPNST cells in further experiments. In fact, in agreement with our results, an earlier study described Combined therapy of MEK inhibitors and ATRA in NF1 associated MPNST induction of morphological changes in MPNST cells from a spindle-cell to an epitheloid-cell phenotype using ATRA. [57] In our hands, NSF1 MPNST cells demonstrated resistance to ATRA therapy since no major effects were seen regarding viability, proliferation, apoptosis and migration under ATRA treatment. However, loss of RXRG expression identified in NSF1 cells seems not to explain resistance to ATRA entirely, due to the fact that we demonstrated these cells to highly upregulate CRABP2 and its downstream targets after ATRA administration (Table 1). In contrast, embryonic stem cells lacking RARG fail transcriptional activation of ATRA target genes such as CRABP2 and CYP26A1 after ATRA treatment [58]. Although little is known about the role of RXRG in Schwann cells, absent expression of RXRG in NSF1 cells may contribute to resistance to ATRA therapy through other mechanisms. Similarly, a defective RARB promoter was demonstrated to lead to partial insensitivity of thyroid cancer cells [59,60]. A general pre-therapeutic testing of elements necessary for a functional retinoic acid transcription complex may uncover ATRA resistant MPNST subtypes prior to therapy. Another explanation for resistance of MPNST to ATRA treatment would include CRABP2 promoter methylation as a determinant of RA effects [56,61]. We expected CRABP2 methylation being causal in those cells that exhibited worse response to ATRA. Although we covered three different promoter regions between -445 up to +370 relative to TSS, we failed to determine a correlation between ATRA sensitivity and CRABP2 promoter methylation as it was shown for other cancer cells such as for gliomas, medulloblastomas, Wilms tumors and pancreatic adenocarcinomas [56,[61][62][63].
Since CRABP2 mediates functions of ATRA, we investigated CRABP2 expression and response to RA signaling in-vitro in detail. Bona fide targets of CRABP2-mediated ATRA signaling such as CRABP2, RARB, and CYP21A1 were demonstrated to be up-regulated due to ATRA treatment in all MPNST cell lines (Table 1) as it was shown in other studies for mouse embryonic carcinoma cells, for COS-7 monkey fibroblast cells as well as for head and neck squamous cell carcinoma cells [64][65][66]. Although ATRA treatment induced common targets in MPNST cell lines, functional effects of treatment were not seen in NSF1 cells, pointing to other resistance mechanisms as discussed above. Since ATRA is able to induce also opposing effects via FABP5 [27], we furthermore investigated mRNA of its target PDK1 in MPNST cells. Table 1. Therapeutic pattern and gene expression profiles in MPNST cells. Relative expression, functional effects and promoter methylation is depicted in a semi-quantitative manner applying the following score: three symbols-highest, two symbols-medium, one symbol-low, dash-no, induction/reduction/expression among the three MPNST cell lines.

T265
S462 NSF1 CRABP2 promoter methylation -+++ + A, B, G) and RXR (A, B, G) [63,[67][68][69]. In our study, we verified that FABP5 level did not interfere with ATRA sensitivity in MPNST, like in other human cancers. Thus, CRABP2 is not only necessary to enable ATRA signaling but is the prevalent mediator compared to opposing signaling pathways in MPNST. To underline these findings, we were able to detect expression of CRABP2 in a series of human MPNST FFPE specimens and in all three MPNST cell lines. Thus, expression of CRABP2 in tumor Schwann cells makes these cells principally suitable for ATRA treatment.

RAR (
In a previous study the transcriptional co-activator ZNF423 was suppressed in neuroblastoma cells lacking NF1 due to overactive MAPK cascade signaling, finally leading to ATRA resistance [41]. Inhibiting the cascade by MEKi treatment led to ATRA response [42]. Thus, a mutual factor of ATRA and Ras signaling was identified by this group. Although in our study ATRA and MEKi treatment induced enhanced therapeutic effects, this was not attributed to changes of ZNF423 mRNA levels. ZNF423 levels did not correlate at all with ATRA sensitivity of MPNST. This is in obvious contrast to previous studies and demonstrates other mechanisms involved in MPNST cells compared to neuroblastoma cells. Our data point to a potentially successful treatment of MPNST by combined application of ATRA and MEK inhibitors such as U0126 or PD0325901. In further experiments, effects other than reduced cellular proliferation of this combination therapy should be monitored to understand the biological impact. This therapeutic strategy utilizes two specific characteristics of NF1 associated MPNST-first, over active MAPK signaling and second, CRABP2 expression of these tumors. Additive effects on reduced viability were successfully demonstrated in two MPNST cells lines, whereas a probable resistance mechanism in ATRA/MEKi resistant NSF1 cells needs further investigations. Since NF1 deficient tumor cells are generally characterized by over-activation of Ras and, as we demonstrated, by CRABP2 expression, the combination of MEKi and ATRA seems to be a promising new approach, necessarily to be validated in animal experiments. . Methylation profiles differed highly between the MPNST cell lines. T265 cell line showed similar methylation pattern (max. methylation per CpG site < 16.9%) to nhSC control cells. S462 cells demonstrated high methylation status for all CpG sites (mean methylation of 71.0%). The NSF1 cell line showed a highly variable methylation pattern with methylation status ranging from 3.4% to 72.0% for single CpG sites (mean, n = 4). S2C: No differences were observed in relative methylation status (%) of all analyzed CpG sites in ATRA treated MPNST cells compared to untreated cells (mean ± SD, n = 4). (PDF) S3 Fig. Flow cytometry analysis of MPNST cell lines treated with ATRA. Relative increase of size (FSC, light grey) and granularity (SSC, dim grey) is given in % compared to untreated controls (0%). Relative cell size was increased by 16% in NSF1 cells, 14% in S462 cells and 6% in T265 cells. Granularity was increased by 14% in T265 cells, 22% in S462 cells and 39% in NSF1 cells (p<0.05, one-sided t-test, mean + SD, n = 3). (PDF) CRABP2 expression was found to be induced at all concentrations in T265 and S462 cells (grey bars) compared to untreated control cells (black line). NSF1 cells showed decreased CRABP2 level at 1 nM and 10 nM PD0325901, but increased level at 1000 nM. ZNF423 expression was reduced in T265 cells in a dose-dependent manner but was not affected in S462 cells at all concentrations. Reduced ZNF423 levels were also found in NSF1 cells. Relative mRNA level were not determined in T265 cells at 1000 nM PD0325901, since almost no alive cells were present (n.d. = not determined) (mean + SD, n = 3). (PDF)  (Table A). Primer sequences for RT-PCR (Table B). Primer sequences used for bisulfite-sequencing (Table C). Primer sequences used for amplification of bisulfite converted DNA (Table D). Specifications of antibodies used for western blot analysis (Table E). (PDF)