Variable Resistance of RMS to Interferon γ Signaling

Aims. Chimeric T cells directed to the γ-subunit of the fetal acetylcholine receptor (fAChR) produce large amounts of interferon-γ (IFNγ) on coculture with fAChR-expressing rhabdomyosarcoma (RMS) cells prior to RMS cell death. The aim of this study was to elucidate whether IFNγ blocks proliferation and survival of RMS cells and modulates expression of genes with relevance for cytotoxicity of chimeric T cells. Methods. Expression levels of IFNγ receptor (IFNGR), AChR, MHCI, MHCII, and CIITA (class II transactivator) by RMS were checked by flow cytometry, qRT-PCR, and western blot. Proliferation and cell survival were investigated by annexin V and propidium iodide staining and MTT (thiazolyl-blue-tetrazolium-bromide) assay. Key phosphorylation and binding sites of IFNGRs were checked by DNA sequencing. Results. IFNγ treatment blocked proliferation in 3 of 6 RMS cell lines, but reduced survival in only one. IFNGR was expressed at levels comparable to controls and binding sites for JAK and STAT1 were intact. Induction of several target genes (e.g., AChR, MHCI, and MHCII) by IFNγ was detected on the RNA level but not protein level. Conclusions. IFNγ does not significantly contribute to the killing of RMS cells by fAChR directed chimeric T cells. Signalling downstream of the IFNR receptor, including the posttranscriptional level, is impaired in most RMS cell lines.


Introduction
Interferon gamma (IFNγ) plays a crucial role in tumor formation and protects host against growth of spontaneous or transplanted tumors [1,2]. Besides its immunomodulatory effects, IFNγ has an influence on proliferation and induces apoptosis in vitro in many primary tumor cells and established tumor cell lines [3][4][5][6].
IFNγ is the only member of the type II interferon family and is mainly produced by activated NK-cells and NKT cells [7], as well as CD4 + T-cells and cytotoxic CD8 + lymphocytes [8]. The active form of the cytokine is a dimer which binds to a heterodimeric receptor complex that consists of IFNGR1 and IFNGR2 subunits and is associated with two Janus kinase family members, Jak1 and Jak2. Changes in confirmation of receptor subunits after IFNγ binding activate Jak1 and Jak2, which in turn phosphorylate IFNGR1 and generate a binding site for recruitment, phosphorylation, and dimerization of signal transducer and activator of transcription 1 (STAT1). After translocation of STAT1 homodimers to the nucleus and binding to GAS (IFNγ activated sites) promotor elements, transcription of target genes is initiated [9][10][11], including MHC class I and II genes with immunomodulation function. Other genes affected by IFNγ are the cyclin-dependent kinase inhibitors p21 WAF1/CIP1 and p27 KIP [12], which mediate growth arrest, as well as PI3K, PKC, and different MAPK involved in STAT1 function [13][14][15]; recently genes such as Bik/Blk/Nbk with an importance for apoptotic pathways have been linked to IFNγ response [16].
In the current study, we focus on Rhabdomyosarcoma (RMS), the most common form of soft tissue sarcoma, which mainly affects children and adolescents [17,18]. RMS are subdivided in alveolar RMS (ARMS) and embryonal RMS (ERMS). While overall survival of patients with localized and resectable RMS improved significantly during the last decades, with an overall survival rate of 65%, survival has remained poor in metastatic disease [17,19,20]. As a new treatment strategy for RMS, we have used chimeric T cells with a specificity against the fetal acetylcholine receptor (AChR) which is expressed on the surface of RMS 2 ISRN Oncology [21]. Chimeric T cells are generated by transduction with expression vectors that code for a fully humanized chimeric antigen receptor (CAR) against the AchRγ subunit [22]. Binding to target antigen results in strong IFNγ secretion by chimeric T cells that exert specific cytotoxicity against RMS cell lines in vitro [22,23]. One of the previous studies suggested that IFNγ might significantly contribute to the proapoptotic effects of RMS-directed chimeric T cells [23]. Furthermore, work by Poëa-Guyon et al. revealed that proinflammatory cytokines such as IFNγ induce overexpression of AChR, that is, the target of chimeric T cells, on the cell surface of RMS-like transformed thymic myoid cells [24]. Therefore, we studied the influence of IFNγ on ARMS and ERMS cell lines, showing that most of them are resistant to even high concentrations of IFNγ in terms of induction of apoptosis and AChR overexpression. Recombinant IFNγ was purchased from R&D Systems. The demethylation reagent 5 -Aza-2 -deoxycytidine was obtained from Sigma Aldrich (St. Louis, MO, USA). Mouse anti-human AChR antibodies against alpha and gamma subunit were obtained from GeneTex (Irvine, CA, USA); rat anti-human antibodies against alpha and gamma subunit of the AChR were a kind gift from S. Tzartos (Department of Biochemistry, Helenic Pasteur Institute, Athens, Greece); to detect human MHC class II (HLA-DRA), we used a mouse anti-human antibody (clone L243; kind gift from H. Kalbacher; Interfaculty Institute of Biochemistry, University of Thübingen); mouse anti-human IFNGR1 and goat antihuman IFNGR2 antibody were purchased from R&D Systems (Minneapolis, MN, USA). To detect CIITA we used a goat anti-human antibody from Santa Cruz Biotechnology (Santa Cruz, CA, USA), Caspase analysis was done with a mouse anti-human caspase 8 antibody from cell signalling (Danvers, MA, USA).

Material and Methods
FITC-conjugated anti-mouse antibody was purchased from R&D and TRI-conjugated antibody from CALTAG Laboratories. The PE-conjugated donkey anti-rat antibody and a FITC-conjugated donkey anti-goat antibody were from Jackson ImmunoResearch. Isotype-matched antibodies or secondary antibodies of irrelevant specificities were used as staining controls (Sigma Aldrich, St. Louis, MO, USA).
Horse-radish-peroxidases-(HRP-) conjugated antibodies (Santa Cruz) with specific specificity to primary antibodies were used as secondary antibody for western blot analyses.

Sequencing.
Sequencing of phosphorylation sites in IFNGR1 was done using ABI BigDye Terminator sequencing kit according to the manufacturer's instructions. The following primers were used for JAK binding site: IFNGR1 JAK fwd: CTGACTGATTGATGGCAGGT, IFNGR1 JAK rev: AGAATTGCAGAGCTGGGAAG and STAT1 phosphorylation site: IFNGR1 STAT fwd: GGAGGTGGTCTGTGAA-GAGC and IFNGR1 STAT rev: TCTTTACCGCTATCATC-CACAA.

Western Blot
Analysis. Cells were washed three times with ice-cold PBS and incubated 30 min in 2% (w/v) SDS, 60 mM Tris pH 6.8, phosphatase, and protease inhibitor cocktail (ProteoBlock, Fermentas, St. Leon-Roth, Germany) on ice. Cellular debris was removed by centrifugation and proteins (20 μg) were separated by 12% (w/v) SDS polyacrylamid electrophoresis, followed by protein transfer to PVDF membranes (GE Healthcare, Fairfield, CT, USA). Membranes were blocked with 5% (w/v) low-fat milk or BSA (PAA, Pasching, Austria) for 30 min, incubated with primary antibody for 2 h at room temperature or over night at 4 • C, washed in TBS, 0,05% (w/v) Tween and incubated with the HRP-conjugated secondary antibody. Binding of antibodies was visualized with the "ECL detection reagent" (GE Healthcare) and documented using the Chemi-smart 5100 (PEQLAB, Erlangen, Germany).

Cytotoxicity Assay.
To analyze apoptotic effects towards target cells after different incubation periods with IFNγ 1 × 10 4 , tumor cells per well were seeded out in 96 well plates and incubated with 1% FCS 24 h before IFNγ treatment, followed by addition of 100 ng/mL IFNγ and incubation for 0, 24, 48, 72, and 96 h. Cells were then incubated for 4 h with 20 μL MTT (5 mg/mL). MTT salt was solved in 200 μL DMSO and reduction of MTT by viable tumour cells was colorimetrically determined at an adsorbance wavelength of 560 nm and a reference wavelength of 670 nm. The viability of tumour cells was calculated as the mean of three wells containing tumour cells, the background as the mean of three       temperature, and analyzed by flow cytometry after addition of 400 μL Annexin V binding buffer.

Statistical Analysis.
For statistical analysis an unpaired t-test was applied using the GraphPad Software (San Diego, CA, USA).

RMS Cells Are Highly Resistance against IFNγ-Induced
Cell Death. As shown before [23], killing of RMS cells following coculture with fAChR-specific chimeric T cells is preceded by the production of large amounts of IFNγ. To examine whether IFNγ contributes to RMS cell death, we treated various RMS cell lines with 100 ng/mL IFNγ and determined survival at different time points. The IFNγsensitive colon carcinoma cell line HT29 served as positive control [25]. HT29 cells started to undergo apoptosis 24 h after the beginning of IFNγ treatment. Their proliferation decreased in parallel, resulting in significantly reduced numbers of viable cells after 48 h (69%) and 72 h (10%) (Figure 1(a)). As opposed to HT29 cells, ERMS cell lines RD6 and TE671 and the translocation negative alveolar ARMS cell line FLOH maintained proliferation and survival during IFNγ incubation periods up to 96 h (Figure 1(b)) with only minor effects on cell growth. By contrast, IFNγ elicited reduced proliferation and growth arrest without cell death in the translocation-positive ARMS cell lines CRL2061 and RH41 (Figure 1(a)), while only RH30 cells showed a decline in viability after 72 h (Figure 1(a)). Apoptosis was checked in RH30, FLOH1, TE671, and HT29 cells by Annexin V/Propidium iodide (PI) double staining and caspase 8 cleavage assay. Percentage of PI positive cells after 96 h of treatment approached 100% in HT29 cells, 60% in RH30 cells and <20% in the other, IFNγ-resistant cell lines (Figures 2(a) and 2(b)). Surprisingly, caspase 8 cleavage after 24 h (Figure 2(c)), 48 h, and 96 h (not shown) was only observed in HT29 cells but not in any RMS cell line tested, including apoptosis-prone RH30 cells (Figure 2(c) and data not shown).

RMS Cells Show Intact IFNRs and STAT1 Phosphorylation
In Vitro. Since IFNγ resistance could be due to diminished expression of IFNGR subunits, we next analyzed expression of the IFNGR1 and IFNR2 subunits on RMS cell lines. Apart from CRL2061 cells, that showed barely detectable IFNGR2 expression levels by FACS, both subunits were expressed on the surface of the other RMS cell lines (Figures 3(a) and  3(b)). IFNγ treatment (48 h) induced normal [26] decline of IFNGR1 by receptor internalization (not shown) in all tested cell lines. Sequencing of essential phosphorylation sites for JAK binding and STAT1 phosphorylation revealed wild-type sequences (Figure 3(d)). Furthermore, we found that RMS cell lines express high levels of pStat after different incubation periods with IFNγ (Figure 3(e)).

IFNγ Treatment Does Not Alter Protein Expression of FAchR and MHCII by RMS cells.
To check whether resistance of most RMS cell lines against IFNγ-mediated killing reflects a facet of a broader block of IFNγ-driven gene expression, we analyzed AChR and MHC expression on RMS cell lines after incubation with IFNγ for up to 72 h. In contrast to a previous report about IFNγ-driven AChR induction in RMS-like transformed myoid cells [24], AChR expression on RMS cell was not altered either by IFNγ treatment alone (Figure 4) or when combined with TNFα (data not shown). As to bona fide IFNγ targets, expression of MHC class II and its upstream regulator, CIITA, was not inducible in any RMS cell line (Figures 5(a) and 5(c)), while MHC class I expression was slightly inducible in RH41, RD6, and TE671 but only marginally in CRL2061, RH30, and FLOH1 cells ( Figure 5(b)). Of note, IFNγ-susceptible, apoptosis-prone HT29 cells exhibited strong induction of MHCI, MHCII ( Figure 5(d)), and CIITA ( Figure 5(c)) expression on IFNγ treatment. MHC protein expression occurs on the transcriptional or posttranscriptional level, we analyzed expression of MHCI, MHCII, and two AChR subunit genes (α and γ) by qRT-PCR ( Figure 6). IFNγ increased MHCI (2-to 7-fold) and MHCII (3-to 8.000-fold) mRNA levels in RMS cells. These increases were much lower than in HT29 cells (17-fold for MHCI and 12.700.000-fold for MHCII). Transcripts of AChR subunits were significantly increased only in FLOH1 and TE671 cells, but neither in the other RMS cell lines nor HT29 cells.

Blockade of IFNγ Response Genes in RMS Cannot
Be Abrogated by Demethylation. Chen et al. [27] showed hypermethylation of p21 WAF promoter regions in RMS and demethylation with 5 aza 2 deoxycytidine (5 aza) reactivates p21 WAF expression. We found similar effects following demethylation that was paralleled by cell cycle arrest in all RMS cell lines. By contrast, demethylation rendered CRL2061, RH30, and RH41 susceptible to IFNγ-induced cell death ( Figure 7). Furthermore, pretreatment of RMS cell lines with 5 aza had no impact on the defective induction of MHCII or AChR expression by IFNγ (data not shown).

Discussion
In search of novel treatment options for otherwise refractory RMS we generated an immunoreceptor against the RMSspecific fAChR and used chimeric T cells (cTCs) to target RMS cells. However, RMS cell death on coculture with cTCs was rather protracted although cTCs exhibited strong IFNγ secretion on antigen recognition [22,23]. To explain the delayed death response of RMS cells the hypothesis has been put forward that granzyme B-driven apoptotic pathways may be attenuated and that locally secreted IFNγ may contribute to RMS cell death [23]. Furthermore, an inductive effect of IFNγ on the expression of fAChR, that is, the chimeric T-cell target, has been suggested in RMS-like cells [24]. To address these hypotheses, we here investigated the impact of IFNγ on proliferation, apoptosis, and fAChR expression in RMS cell lines. Our major finding was that IFNγ has antiproliferative effects on CRL2061 and RH41 and apoptotic effects on RH30 while other lines (FLOH1, RD6, and TE671) appeared refractory ( Figure 2). However, apoptotic effects even in RH30 cells were smaller than in highly IFNγ-sensitive HT29 colon carcinoma cells that served as positive control. In addition experiments with IFNγ target genes like MHCI, MHCII, and AChR illustrated a diminished alteration in gene expression after IFNγ treatment. Lack of IFNGR2 expression-one of the limiting factors in IFNγ signalling [28]-could be excluded ( Figure 3). Furthermore, mutations in two essential binding sites in IFNGR1, which are required for receptor function-the JAK binding motive LPKS and Stat1 binding site YDKPH with the essential phosphorylation site Y 30 440 -were also excluded by sequencing (Figure 3(d)). Indeed, phosphorylation of Stat1 that is necessary for successful IFNγ signalling [29,30], was comparable in RMS cells to phosphorylation in a highly IFNγ-sensitive control cell line (Figure 3(e)).
Since it is known that a broad spectrum of tumor cells lack MHC presentation and show hypermethylation of IFNγ target genes such as CIITA [31], we treated RMS cells with the demethylation reagent 5 aza 2 deoxycytidine. Further addition of IFNγ resulted in growth arrest and induced cell death in some but not all cell lines (Figure 7). However, induction of MHCII and AChR expression was not achieved. Our results fit in part to those of Chen et al., who described inhibition of cyclin-dependent kinase inhibitor p21 WAF1 by methylation of SIE-1 promotor elements that resulted in reduced cell cycle control [27] and increased growth. Taken together, hypermethylation of IFNγ target genes may be operative in defective cycle control, but may not explain diminished IFNγ responses of other target genes. Indeed, the study of Londhe et al. shows that CIITA induction is possible by the combined treatment of RMS cell lines with histone deacetylase (HDAC) inhibitors and demethylation agents, indicating a complex block of accessibility to some promoters in RMS cell lines [32]. However, even this mechanism may not apply to all promoters, considering our finding that some IFNγ response genes showed upregulation of transcription that did not translate into protein expression. Therefore, we hypothesize that (a) higher levels of mRNA of IFNγ target genes may be required for effective translation, which can be achieved by changes in epigenetic modifications and-not  mutually exclusive-(b) there could be a posttranscriptional block, for example, by miRNAs, with influence on IFNγdependent protein expression [33]. The current findings have therapeutic perspectives. In vivo, defective responsiveness to IFNγ is associated with more aggressive tumor behaviour, while IFNγ-responsive tumors have a better chance to be kept in check by the immune system [34][35][36]. Overcoming tumor escape by breaking IFNγ resistance in RMS is, therefore, worth to be tested as an adjunct to immunotherapies based on vaccination or adoptive transfer of tumor-reactive cytotoxic effector cells.