Death by HDAC Inhibition in Synovial Sarcoma Cells

This article is featured in Highlights of This Issue, p. 2639

Synovial sarcoma is an aggressive soft tissue malignancy that primarily affects adolescents and young adults (1). It is characterized by the driving chimeric oncoprotein SS18-SSX, derived from the t(X;18)(p11.2;q11.2) translocation (2). The SS18-SSX fusion oncoprotein has been proposed to interact in place of native SS18, leading to aberrant SWI/SNF–mediated transcription (3), and to act as a nidus for abnormal assembly of polycomb group members (PcG) and transcription factors that lead to transcriptional repression by EZH2-mediated H3K27 trimethylation (4). Expression of tumor suppressor genes EGR1 and CDKN2A has been shown to be directly repressed by SS18-SSX, whereas the antiapoptotic protein BCL2 is characteristically upregulated (5) leading to a proliferative and antiapoptotic phenotype.

Current cytotoxic therapies, including doxorubicin and ifosphamide, provide limited benefit for synovial sarcoma patients. Following surgery and radiation, patients remain at high risk for both early and late metastases, and despite the use of multimodal therapies the mortality rate remains approximately 50% within 10 years of diagnosis (6). Thus, there is a significant need for targeted therapies against synovial sarcoma.

Histone deacetylase (HDAC) inhibition has been shown to elicit apoptosis in synovial sarcoma models, as well as to disrupt SS18-SSX–mediated repressive complexes (4, 7). This inhibition was observed to elicit reactivation of repressed gene targets, including CDKN2A (4, 7, 8). As a result, a phase II study of a pan-HDAC inhibitor, pracinostat, was initiated in translocation-associated sarcomas, from which stable disease was achieved in three of three assessable SS18-SSX–positive patients (9). While this study had to be terminated prematurely due to unexpected interruptions in the supply of the studied compound, initial results did warrant further investigation.

Histone acetylation status is an important element in the regulation of gene expression. HDAC inhibition has been shown to induce cell-cycle arrest (10), senescence (11), differentiation (12), and apoptosis (13) in cancer cells, generally leading to a decrease in tumor burden (14). While maintaining some pathway commonalities, the distinct mechanism of HDAC inhibitor–induced cell death may be specific to the underlying oncogenic context. With efficacy observed in reversing the abnormal epigenetic status of several cancer subtypes, a variety of HDAC inhibitors are currently under clinical investigation. Since 2006, five clinical drugs have been approved as treatments for advanced T-cell lymphoma or multiple myeloma, while several additional HDAC-inhibiting compounds are currently under investigation in solid tumor clinical trials (14). Quisinostat (15), used in this study, is a second-generation HDAC inhibitor showing promising efficacy against synovial sarcoma in preclinical experiments (16).

Although strong and specific responses to HDAC inhibition have been consistently observed in synovial sarcoma, the mechanism of apoptosis induction remains unclear. Advancing understanding of this relationship may lead to improved targeted therapy development and improved trial design, as well as aid in further elucidating the important pathways implicated in synovial sarcoma oncogenesis. In this study, we used next-generation sequencing analyses of six human synovial sarcoma cell lines to investigate the mechanisms of induced cell-cycle arrest and cell death in synovial sarcoma following HDAC inhibition by quisinostat, a relatively new, orally available HDAC inhibitor that was the top hit in a synovial sarcoma drug screen (16).

The following human synovial sarcoma cell lines were kindly provided: SYO-1 (ref. 17; Dr. Akira Kawai, National Cancer Centre Hospital, Tokyo, Japan, 2004), FUJI (ref. 18; Dr. Kazuo Nagashima, Hokkaido University School of Medicine, Sapporo, Japan, 2003), YaFuss (19), and HS-SY-II (ref.20; Dr. Marc Ladanyi, Memorial Sloan Kettering Cancer Centre, New York, NY, 2014), MoJo (ref. 21; Dr. K. Jones, University of Utah, Salt Lake City, UT, 2014), and Yamato-SS (ref. 22; Dr. K. Itoh, Osaka Medical Center for Cancer and Cardiovascular Diseases, Japan, 2013). Sarcoma cell lines were maintained in RPMI1640 medium supplemented with 10% FBS (Life Technologies) and the presence of a disease-defining SS18-SSX fusion oncogene was confirmed by RT-PCR analysis. Mycoplasma testing was undertaken every six weeks by RT-PCR analysis and any cell lines found positive were discarded. All cells were grown at 37°C, 95% humidity, and 5% CO₂. Pharmacologic compounds were purchased from Selleck Chemicals.

Using the RNeasy Mini kit (Qiagen), total RNA was isolated from six human synovial sarcoma cell lines treated for 16 hours with the HDAC inhibitor quisinostat at 0.025 μmol/L or with vehicle (0.1% DMSO) control. Quality control was performed using the Agilent 2100 Bioanalyzer to confirm RNA Integrity Numbers over 8. Qualifying samples were further prepared using the TruSeq stranded mRNA library kit (Illumina) on the Neoprep Library System (Illumina). Paired-end sequencing was performed on a NextSeq 500 sequencer to a read length of 81, with a median of 26.3 million reads per sample. Sequence quality metrics were evaluated using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/): across cell line specimens (n = 12), the average quality of sequenced reads at any position was 33.6, and 90% of base calls were of quality 30 or greater. Alignment statistics were assessed using Picard tools (http://broadinstitute.github.io/picard/): across 12 specimens, the average value of median insert size was 226 bp, and the average percentage of high quality alignments was 95% (map quality of 20 or greater). Paired reads were aligned to the human genome version hg38 (v24) using STAR (23) with default parameter settings. Sequence data have been deposited at the European Genome-phenome Archive (EGA, http://www.ebi.ac.uk/ega/), which is hosted by the EBI, under accession number EGAS00001002637.

Reads of quality 10 or greater were enumerated on the sense strand for the exonic regions of each gene. Fragments Per Kilobase of exon per Million mapped reads (FPKM) values were calculated using the robust normalization method of DESeq2, and variance stabilized log fold changes were calculated using the regularized log transformation of DESeq2 (24). DESeq2 was used to determine significance values for differential expression between treated and untreated cells, specifying pairwise comparison of drug effects across cell lines. A significance threshold of 0.01 was selected for adjusted P values. Gene enrichment analyses were performed for gene ontologies using PANTHER with Bonferroni correction (25), and for the MsigDB (molecular signatures database) v5.2 (26) using the hypergeometric test.

Raw count data for the TCGA sarcoma dataset were obtained from Genomic Data Commons (https://gdc.cancer.gov/node/31). FPKM, log₂ fold changes, and significance values were determined using DESeq2, comparing synovial sarcoma to all other sarcoma types.

The gene expression ominibus was surveyed for studies of HDAC inhibition at lethal doses in cancer cell lines, and five studies on more recent microarray platforms were selected for comparison [prostatic adenocarcinoma: GSE74418, neuroblastoma: GSE49158, NUT midline carcinoma: GSE18668, ovarian carcinoma GSE53603, atypical teratoid rhabdoid tumor (AT/RT): GSE37373]. Expression data were downloaded from the processed data files supplied to Gene Expression Omnibus (GEO; series matrix files), and differential expression was determined using uniform criteria for each study using Limma (27). Gene IDs were mapped to the RNA-seq dataset using the supplied platform specifications. Expression data for GEO studies was ranked according to log₂ fold change of treatment versus control, and Gene Set Enrichment Analysis (GSEA version 2.2.3) was used to gauge the extent of over-representation of significant fold changes from RNA-seq data.

Protein was collected from indicated cell lines grown in 10-cm dishes or following 24-hour treatments with indicated compounds. Samples were separated by 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Bio-Rad Laboratories). Blots were incubated with indicated antibodies: SS18 sc-28698 1:200, GAPDH sc-25778 1:1500, BIK (NBK) sc-305625 1:500, BIM sc-374358 1:500, BCL2 sc-492 1:250, p-BCL2 sc-101762 1:250 (Santa Cruz Biotechnology); p16^INK4a ab108349 1:500, p14^ARF ab124282 1:500 (Abcam). Signals were visualized using the Odyssey Infrared System (LI-COR Biosciences).

Total RNA was isolated from treated samples using the RNeasy Mini kit (Qiagen) and reverse transcribed to cDNA using Oligo(dT) (Invitrogen) and SuperScript III (Invitrogen). SYBR Green (Roche) reagent was used for qPCR expression analysis, using an Applied Biosystems ViiA7 qPCR system. All transcript levels were normalized to GAPDH RNA expression as well as to DMSO-treated conditions to calculate relative fold induction of expression using the comparative C_t (ΔΔC_t) method. BIK: sense: AGATGGACGTGAGCCTCAGG antisense: CTAATGTCCTCAGTCTGGTCGTAG; BIM: sense: GGTCCTCCAGTGGGTATTTCTCTT antisense: ACTGAGATAGTGGTTGAAGGCCTGG; BMF: sense: CCCTTGGGGAGCAGCCCCCTG antisense: GCCGATGGAACTGGTCTGCAA; CDKN2A: sense: CAACGCACCGAATAGTTACGG, antisense: AACTTCGTCCTCCAGAGTCGC

Cells were seeded in 96-well plates at 1 × 10⁴ cells/well and treated in triplicate at indicated doses of the tested compounds. IC₅₀ doses were determined in synovial sarcoma cell lines by dose curve studies. Cell viability was assessed in the cell lines as compared with the vehicle condition (0.1% DMSO) at 48 hours posttreatment using MTS reagent (Promega). Cell confluency and apoptosis induction were assessed over a 48-hour timeframe utilizing the IncuCyte Zoom live cell imaging software (Essen BioScience). Apoptosis was assessed using the IncuCyte Kinetic Caspase-3/7 Apoptosis Assay Reagent (Essen BioScience).

Reactive oxygen species (ROS) levels were assessed by dichlorofluorescein diacetate (DCFDA) assay (Abcam, ab113851). Cells were seeded in a 96-well plate at 1 × 10⁴ cells/well. The following day cells were treated with indicated compounds. At 24 hours posttreatment, wells were overlayed with 2× diluted DCFDA reagent, as suggested by the manufacturer's protocol. Fluorescence was read by microplate reader at Ex/Em = 485/535 nm. A ratio of untreated to treated wells was calculated to determine relative response. Wells were normalized to media-only wells to account for background signal.

Duplex oligo (sense, CAAGAAGCCAGCAGAGGAATT; antisense, UUCCUCUGCUGGCUUCUUGTT) siRNAs were designed to target the SSX portion of SS18-SSX using the Integrated DNA Technologies RNA interference (RNAi) design tool, and synthesized by Integrated DNA Technologies as described previously (28). siFOXO1 (sc-35382), siFOXO3 (sc-37887), and siFOXO4 (sc-29650) were purchased from Santa Cruz Biotechnology. siBIK (L-004388-00-0005), siBIM (L-004383-00-0005), and siBMF (L-004393-00-0005) were purchased from GE Healthcare Dharmacon. siFOXO1 (sc-35382), siFOXO3 (sc-37887), and siFOXO4 (sc-29650) were purchased from Santa Cruz Biotechnology. SYO-1 cells were seeded in 6-well plates. At 60% confluence, cells were transfected with 30 pmol pooled siRNA and 9 μL Lipofectamine RNAiMax transfection reagent (Invitrogen) in Opti-MEM serum-free media (Life Technologies), according to the manufacturer's protocol. Protein was harvested at 48 hours posttransfection, and knockdown was confirmed by Western blot (Supplementary Fig. S1).

Cells were seeded in culture-treated chamber slides at 3 × 10⁴ cells/well. The following day, wells were treated with indicated compounds. Following indicated incubation times, cells were then washed twice with PBS, fixed with methanol at −20°C, and permeabilized with 0.1% Triton X-100. Wells were blocked with blocking buffer and incubated overnight at 4°C with primary antibodies at a 1:250 dilution: FOXO1 (sc-374427), FOXO3 (sc-11351), FOXO4 (sc-5221), Ac-FOXO1 (sc-49437; Santa Cruz Biotechnology), P-FOXO1 (Cell Signaling Technology 9464S). Wells were washed three times with PBST then incubated for 45 minutes with AlexaFluor secondary antibody (Life Technologies). Fluorescence was detected using a Zeiss Axioplan 2 microscope at 40×.

Pten^fl/fl;hSS2 mice (as previously described in ref. 29) received a 10-μL injection of 42 μmol/L TATCre in the hindlimb at 4 weeks of age, to induce expression of SS18-SSX2. A cohort of 8 mice with tumor volumes ranging from 600 to 1,500 mm³ were randomly assigned treatment of quisinostat (2 mg/kg; n = 4) or vehicle control (10% hydroxyl-propyl-β-cyclodextrin/25 mg/mL mannitol/H₂O) (n = 4). Mice received daily intraperitoneal injections and tumor volumes were measured three times weekly for 7 days. Tumor volumes (mm³) were measured using digital calipers and calculated with the formula (length × width²)/2 (30). Mice were monitored daily and tumors monitored at least three times weekly during treatment. Tumors were excised and RNA was isolated with the RNeasy Mini kit (Qiagen)

No toxicity was observed in the treatment regimens of these mice. Mice demonstrating poor mobility, poor feeding/watering behaviors, or poor interactions with cage mates, suggestive of ill-health or discomfort, were humanely euthanized. Any tumor burden greater than 10% body mass, that impeded mobility, feeding, or watering or that caused apparent discomfort, or any ulcerated mass or weight loss of more than 20% was also managed by humane euthanasia. Mice were euthanized by exposure to compressed CO₂ followed by a bilateral thoracotomy. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH (Bethesda, MD). The protocol was approved by the Institutional Animal Care and Use Committee of the University of Utah (permit number: 14-01016).

All IHC analysis was performed on 4-μm formalin-fixed paraffin-embedded whole sections. Unstained slides were processed with the Ventana Discovery XT automated system (Ventana Medical Systems), according to the manufacturer's protocol with proprietary reagents. The heat-induced antigen retrieval method was used in Cell Conditioning solution (CC1-Tris-based EDTA buffer, pH 8.0; Ventana Medical Systems). Rabbit monoclonal Ki-67 (clone SP6, CRM 325 A, Biocare Medical) was used for primary incubation, for 60 minutes (37°C) at a 1:50 dilution. Biotinylated goat anti-rabbit IgG antibody (BA-1000, Vector Laboratories) was incubated for 32 minutes at 37°C. The detection systems used was the DABMap kit (760-124, Ventana Medical Systems).

Quisinostat significantly decreases cell viability in synovial sarcoma cell lines (Supplementary Fig. S2). The complete expression profiles from six synovial sarcoma cell lines following 16-hour treatment with the HDAC inhibitor quisinostat at 25 nmol/L or vehicle control (DMSO) was assembled by RNA-seq. The time point and dose were selected on the basis of previous studies demonstrating these variables to elicit the most significant transcriptional changes (8, 16). To focus on larger transcriptional changes, genes showing significant differential expression in paired comparisons of treatment to control were analyzed, as defined by an adjusted P value of less than 0.01 and a log₂ fold change greater than 1, in combination with expression above the 25th percentile observed in quisinostat-treated conditions across a minimum of 4 of the 6 cell lines (Fig. 1A). In this manner, lists of 1,967 upregulated genes and 609 downregulated protein-coding genes were compiled. Analysis of expression patterns with gene ontology (GO) revealed at least six highly enriched categories when using the genes observed to be up- and downregulated. These core effects of HDAC inhibitors in synovial sarcoma are summarized with representative genes in Fig. 1B.

A number of the highly enriched gene ontology categories relate to neuron differentiation, including pronounced upregulation of genes encoding a spectrum of synapse components, neurotransmitter systems, and BDNF/NGFR/NTRK1 signaling. Expression levels of homeodomain transcription factors characteristic of neural differentiation were also observed to be upregulated upon treatment with quisinostat, including PAX6, SOX2, POU3F1, POU3F2, POU4F1, and POU4F2 (Fig. 1B and C).

HDAC inhibition in synovial sarcoma cells was associated with upregulation of several components of MHC class I and class II antigen presentation, the immunoproteasome component PSMB9, and natural killer ligands MICA, MICB, and ULBP1-3 (Fig. 1B and C).

Upregulation of factors important for response to oxygen-containing species occurred following HDAC inhibition (Fig. 1B and C), including FOXO1, FOXO4, JAK3, and SESN3.

Significant enrichments in the MSigDB dataset included upregulation to targets of polycomb components SUZ12, EZH2, and EED in hESC, which are known to regulate proximal promoters with high CpG content (refs. 31–33; Fig. 1D). Significant overlap with gene targets downregulated with EZH2 depletion was also observed (ref. 34; Fig. 1F).

The primary therapeutic goal of HDAC inhibition is the induction of cell death in cancer cells, and is reflected in the increased expression of apoptosis genes in the synovial sarcoma cell lines treated with 16-hour quisinostat, including activated inducers CDKN2A (p14^ARF), BIK, BCL2L11, and BMF (Fig. 1B).

A major class effect of HDAC inhibition is the negative regulation of proliferation, evident in the downregulation of E2F target genes including MCM and PCNA, concomitant with upregulation of several cyclin-dependent kinase inhibitors. Cell-cycle progression GO terms were significantly decreased with HDAC inhibition (Fig. 1E and F). Expression of cell-cycle effectors and macromolecule biogenesis globally decreased following HDAC inhibition in synovial sarcoma, further suggesting the cells to be coming out of cycle (Supplementary Fig. S3).

The derived average transcript expression values from the six synovial sarcoma cell lines were compared to five available HDAC inhibitor treatment microarray human cell lines studies: neuroblastoma (GSE49158), NUT midline carcinoma (GSE18668), atypical teratoid rhabdoid tumor (AT/RT): (GSE37373), prostatic adenocarcinoma (GSE74418), and ovarian carcinoma (GSE53603). Expression profiles in synovial sarcoma correlated significantly with the previous studies as measured by GSEA (Fig. 2A). A common core gene expression set was revealed by comparison analysis of significantly changed genes when comparing the synovial sarcoma study to the five additional studies; 501 genes were upregulated and 678 genes were downregulated in at least five studies (Fig. 2B). Gene ontology analysis revealed the most significantly increased biological terms included cell development processes, nervous system development and response to ROS (Fig. 2C). The most significantly downregulated terms included cell-cycle progression, macromolecule biogenesis, and chromatin organization (Fig. 2D). A summary of the core effects of HDAC inhibition in cancer cell lines is presented in Fig. 2E, including representative genes involved in nervous system development, response to ROS, cytostasis, and developmental processes (Fig. 2E).

To relate cell line–based findings to the expression profile of synovial sarcoma patient tumor tissue, published RNA-seq expression profiles of patient specimens from The Cancer Genome Atlas (TCGA) were examined in comparison with the HDAC inhibitor–treated expression data revealing 1,323 significantly upregulated and 3,368 significantly downregulated genes (Fig. 3A). In comparing average expression across the (untreated) synovial sarcoma TCGA dataset to HDAC inhibition–mediated changes from the cell line data, it was observed that globally, genes with low relative expression in the synovial sarcoma tumor data were reactivated in the quisinostat-treated cell line data, while many highly expressed targets in the untreated tumor data decreased following HDAC inhibition in the cell lines (Fig. 3B). Gene ontology analysis confirmed core effects of HDAC inhibition in synovial sarcoma to include upregulation of mesenchymal cell differentiation, nervous system development, response to stress and apoptosis, as well as downregulation of proliferation, macromolecule biogenesis, and transcription (Fig. 3C).

To demonstrate the specificity of this expression pattern reversal, published primary tumor microarray expression profiles of patient specimens from the TCGA were examined in comparison with additional sarcoma subtypes (liposarcoma and leiomyosarcoma). Human synovial sarcoma tumor expression data is characterized by a distinctive repression of several cyclin-dependent kinase inhibitors: CDKN2A (p16^INK4a, p14^ARF), CDKN2B (p15^INK4b), CDKN2D (p19^INK4d), CDKN1A (p21^CIP1), and tumor suppressor EGR1, as well as upregulation of cell cycle (CDK6, CCND1) and antiapoptotic effectors (BCL2; Fig. 3D). Transcriptome analysis in the six synovial sarcoma cell lines comparing HDAC inhibitor–treated to vehicle-treated conditions demonstrated a quisinostat-induced reversal of this proliferative phenotype concomitant with induction of proapoptotic effector expression [BIK, BCL2L11 (BIM), BMF, FOXO1, JUN, FOS; Fig. 3E). Expression reversal was specific to synovial sarcoma most significantly in CDKN2A and BIK when compared with the five published transcriptome studies of HDAC inhibition in cell lines representing other cancer subtypes (Fig. 3F).

FOXO-regulated targets involved in apoptotic and cell-cycle regulation (BIM, BMF, GADD45) were upregulated in the transcriptome data following HDAC inhibition, whereas FOXO-regulated antioxidant factors (SOD1 and SOD2) were reduced, albeit not significantly changed, suggesting a systemic switch toward apoptosis in response to ROS accumulation (Fig. 4A). The transcriptome analysis further revealed an HDAC inhibitor–mediated significant increase in FOXO transcription factor levels in the six synovial sarcoma cell lines, particularly in FOXO1 (Fig. 4A).

On the basis of the observed modulation of factors related to the regulation of the cellular oxidative environment upon quisinostat treatment, we further investigated ROS levels. This revealed that quisinostat treatment induces ROS accumulation in a dose-dependent manner (Fig. 4B) that can be attenuated by N-acetylcysteine (N-AC) treatment (Fig. 4C). Recovering the ROS accumulation induced by HDAC inhibition with N-AC rescues synovial sarcoma cell lines by approximately 50%, as measured by increased cell viability (Fig. 4D).

Consistent with this observation, FOXO1, FOXO3, and FOXO4 translocate into the nucleus following quisinostat treatment, aligning with their activation in response to HDAC inhibition (Fig. 4E). Under vehicle-treated conditions FOXO1 exists phosphorylated (as measured at Thr24) in the cytoplasm (Fig. 4F). Following knockdown of FOXO1 and FOXO3 transcription factors, the quisinostat-induced activation of proapoptotic factors BIM and BMF was significantly attenuated (Fig. 4G), whereas FOXO4 depletion had no observable effect. FOXO knockdown rescues HDAC inhibitor–induced apoptosis in synovial sarcoma cell lines, by approximately 50% for FOXO1 (Fig. 4H).

BCL2 levels are known to be elevated in synovial sarcoma tumors resulting in an antiapoptotic phenotype (5). Following HDAC inhibition, mRNA levels of the proapoptotic factor BIK increased by an average of approximately 23-fold, significantly more than other proapoptotic factors (Fig. 5A and B). This resulted in a shift in the average ratio of BCL2:BIK from approximately 58:1 pretreatment down to almost 1:1 following HDAC inhibitor treatment, favoring apoptosis (Fig. 5C). Cleaved caspase-3/7 could be detected at increasing levels with increased doses of quisinostat over time, indicating apoptosis was occurring at much higher levels than with standard doxorubicin treatment assayed in this manner (Fig. 6D). At the protein level, increases in BIK and BIM as well as the phosphorylation of BCL2 occurred in a dose-dependent manner with increasing levels of quisinostat treatment in synovial sarcoma cell lines (Fig. 5E). Cleaved caspase-3/7 levels were decreased with the depletion of BIK, BIM, and BMF suggesting each was important for apoptosis induction following HDAC inhibition, with the most significant contribution by BIK (Fig. 5F).

An inducible mouse model of synovial sarcoma was treated by HDAC inhibition to assess efficacy of apoptosis induction in vivo. At 7 days posttreatment, the quisinostat-treated tumors had decreased in volume by approximately 40% while vehicle-treated tumors grew by approximately 30% (Fig. 6G and H). In comparison with other proapoptotic factors, tumor BIK mRNA levels were found to be significantly higher following treatment than the vehicle-treated group, again resulting in a proapoptotic shift in the ratio of BCL2:BIK (Fig. 6I). Tumor cell proliferation decreases with quisinostat treatment, as measured by Ki-67 staining (Fig. 6J and K)

CDKN2A is a known target of SS18-SSX–mediated repression. In depleting the oncoprotein or when treating with quisinostat, a reactivation of p14^ARF and p16^INK4A protein was observed, and a loss of SS18-SSX protein in response to HDAC inhibition was confirmed (Fig. 6A). CDKN2A depletion attenuated BIK activation at the RNA (Fig. 6B) and protein levels (Fig. 6C), resulting in a decrease in cleaved caspase-3/7 induction following quisinostat treatment (Fig. 6D), suggesting CDKN2A is necessary for BIK activity in response to HDAC inhibition.

Current cytotoxic therapies offer limited benefit against synovial sarcoma; more targeted therapeutic approaches are needed to improve outcome when local control measures are not curative. The synovial sarcoma fusion oncoprotein SS18-SSX disrupts gene expression in a characteristic manner leading to a distinctive phenotype. Previous studies have shown HDAC inhibition disrupts the driving complex in synovial sarcoma, resulting in reactivated expression of tumor suppressors otherwise repressed by SS18-SSX (4, 16). In this study, we delineate the common transcriptional patterns present across studies of HDAC inhibition in six cancer subtypes to uncover core class effects, investigate the effects of HDAC inhibition specific to synovial sarcoma and uncover mechanisms of apoptosis induction in this cancer type.

Quisinostat treatment was found to bring about mesenchymal and neuronal differentiation transcriptional patterns in six cell line studies of different cancer subtypes as well as to promote cytostasis and responses to oxidative stress. HDAC inhibition is thought to further counteract the proliferative and antiapoptotic phenotype of synovial sarcoma by activating repressed gene targets, such as CDKN2A which, in turn, facilitate expression of cell-cycle regulators and proapoptotic proteins. ROS accumulation in response to HDAC inhibition and proapoptotic protein-mediated mitochondrial permeabilization further drives apoptotic pathways.

Expression of cell-cycle regulator CCND1 (cyclin D1) and cyclin-dependent kinases CDK4 and CDK6 decreased with the upregulation of CDKN2A and CDKN2B in response to quisinostat treatment. The cyclin D1/CDK4/CDK6 axis is required for cell-cycle progression through G₁–S phase and is known to be highly active in synovial sarcoma (35). Under normal conditions, CDK4 and CDK6 are inhibited by p16^INK4A (CDKN2A) and p15^INK4B (CDKN2B), negatively regulating cell division. Hampering the cyclin D1/CDK4/CDK6 axis by reactivation of CDKN2A and CDKN2B may therefore contribute to the observed cell-cycle arrest in response to HDAC inhibition.

The CDKN2A locus has been previously shown to be directly repressed by SS18-SSX in synovial sarcoma (4). CDKN2A is normally regulated by polycomb group proteins, repressing its expression in the recruitment of EZH2 (36), and is activated by SWI/SNF-mediated polycomb eviction (37). In this study, we validate the reactivation of CDKN2A-encoded proteins p16^INK4A and p14^ARF in response to SS18-SSX knockdown and HDAC inhibition, and demonstrate an epistatic relationship between CDKN2A expression and BIK activation in synovial sarcoma. The transcriptome data reveal proapoptotic factor BIK activation to be significant in all studied synovial sarcoma cell lines following HDAC inhibition. BIK has been shown in previous studies to be inhibited by BCL2, and when activated to function in initiating the mitochondrial apoptosis pathway (38). p14^ARF has been observed to increase BIK levels by inhibition of transcriptional repressor CtBP (39) as well as to activate p53 by inhibiting MDM2, inducing mitochondrial-mediated apoptosis (40). HDAC inhibition may therefore impede SS18-SSX–mediated repression of the CDKN2A locus, allowing for reactivation of normal cell-cycle regulation and induction of proapoptotic pathways.

In this study, we further find that ROS accumulation secondary to HDAC inhibition results in an initiation of FOXO transcription factor activation in synovial sarcoma. FOXO class forkhead transcription factors are known to act as cellular redox sensors that become modified posttranslationally in response to oxidative stress (41). FOXOs are negatively regulated by PI3K/AKT signaling (42) and translocate to the nucleus when activated by ROS accumulation where they can initiate transcription of antioxidant effectors such as superoxide dismutase 2 (SOD2) to promote survival (41). FOXO transcription factors are also important in initiating expression of proapoptotic effectors in response to ROS (42). As described herein, the induced FOXO program favors expression of proapoptotic factors BIM and BMF and is important for apoptotic induction in response to quisinostat. PTEN-mediated repression of the PI3K/AKT pathway is important for FOXO activation (42), which may be promoted in synovial sarcoma by reactivation of repressed EGR1 in response to HDAC inhibition (8, 28). PTEN deficiency in synovial sarcoma has further been shown to promote tumor angiogenesis and metastasis (43).

In synovial sarcoma, HDAC inhibition acts at multiple sites resulting in blocked cell-cycle progression, ROS accumulation, and initiation of proapoptotic programs regulated by FOXO transcription factors. Through HDAC-mediated reactivation of repressed CDKN2A, proapoptotic factors are able to counteract elevated BCL2 levels. Quisinostat also increases antigen presentation pathways, which may have implications for immune-oncology approaches in a tumor type that characteristically lacks an immunogenic phenotype. Our study provides mechanistic support for the use of HDAC inhibitors in the treatment of synovial sarcoma. While single-agent studies of HDAC inhibition in solid tumors have yielded mixed results, combinatorial drug protocols are the subject of active investigation (44). Rational drug combinations incorporating HDAC inhibitors may enhance tumor penetration and synergistically affect multiple pathways leading to apoptosis (45). In synovial sarcoma, low-dose combinations of HDAC and proteasome inhibitors are particularly efficacious in reactivating proapoptotic factors, inducing apoptosis and decreasing tumor burden in vivo (16). As shown here at the transcriptomic level, HDAC inhibition specifically reverses the proliferative and antiapoptotic phenotype brought about by SS18-SSX.

No potential conflicts of interest were disclosed.

Conception and design: A.N. Laporte, N.M. Poulin, T.M. Underhill, T.O. Nielsen

Development of methodology: A.N. Laporte, N.M. Poulin, K.B. Jones, T.M. Underhill, T.O. Nielsen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.N. Laporte, J.J. Barrott, X.Q. Wang, R. Vander Werff

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.N. Laporte, N.M. Poulin, A. Lorzadeh

Writing, review and/or revision of the manuscript: A.N. Laporte, N.M. Poulin, J.J. Barrott, T.M. Underhill, T.O. Nielsen

Administrative, technical, or material support: A.N. Laporte, N.M. Poulin, A. Lorzadeh

Study supervision: T.O. Nielsen

The authors would like to thank Angela Goytain for technical assistance and Drs. Martin Hirst, Gregg Morin, Chris Hughes, and Bertha Brodin for helpful discussion.

This work was supported by grants from the Canadian Cancer Society Research Institute (grant no. 701582; to T.O. Nielsen and T.M. Underhill), the Terry Fox Research Institute (TFF 105265 New Frontiers in Cancer; to T.M. Underhill and T.O. Nielsen), the Liddy Shriver Sarcoma Initiative (to T.O. Nielsen and K.B. Jones), and the Sarcoma Cancer Foundation of Canada (Beth England's Sarcoma Research Fund; to T.O. Nielsen and K.B. Jones).

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.