ASPSCR1::TFE3 Drives Alveolar Soft Part Sarcoma by Inducing Targetable Transcriptional Programs

Abstract

Alveolar soft part sarcoma (ASPS) is a rare mesenchymal malignancy driven by the ASPSCR1::TFE3 fusion. A better understanding of the mechanisms by which this oncogenic transcriptional regulator drives cancer growth is needed to help identify potential therapeutic targets. Here, we characterized the transcriptional and chromatin landscapes of ASPS tumors and preclinical models, identifying the essential role of ASPSCR1::TFE3 in tumor cell viability by regulating core transcriptional programs involved in cell proliferation, angiogenesis, and mitochondrial biology. ASPSCR1::TFE3 directly interacted with key epigenetic regulators at enhancers and promoters to support ASPS-associated transcription. Among the effector programs driven by ASPSCR1::TFE3, cell proliferation was driven by high levels of cyclin D1 expression. Disruption of cyclin D1/CDK4 signaling led to loss of ASPS proliferative capacity, and combined inhibition of CDK4/6 and angiogenesis halted tumor growth in xenografts. These results define the ASPS oncogenic program, reveal mechanisms by which ASPSCR1::TFE3 controls tumor biology, and identify a strategy for therapeutically targeting tumor cell-intrinsic vulnerabilities.

INTRODUCTION

ASPS is a rare mesenchymal neoplasm, accounting for under 1% of sarcoma diagnoses with an annual population incidence rate of 1.2 per 10⁷ (1). ASPS typically presents as a slow growing deep soft tissue mass of the trunk or extremities in young adults. In contrast to many other soft tissue sarcomas, ASPS commonly presents with metastatic disease and has a propensity to spread to the brain in addition to lung and bone; nevertheless, it is often slowly progressive with the majority of patients surviving five years or more following diagnosis (2,3). Histologically, eosinophilic epithelioid tumor cells are nested within highly vascularized septa, occasionally with central necrosis leading to a pseudo-alveolar pattern that gives this disease its name (4). The cell of origin of ASPS is not well understood, with transcriptional, histologic and mouse modeling suggesting diverse myogenic, neural or other mesenchymal progenitors (5–9).

Complete surgical resection is the standard treatment for localized disease (10). While ASPS is inherently resistant to cytotoxic chemotherapy, tyrosine kinase inhibitors active against the vascular endothelial growth factor receptors (VEGFRs) including sunitinib and pazopanib have shown activity (11,12), presumably related to a substantial reliance of this disease upon angiogenesis. Inhibition of MET may represent another targeted therapy in select tumors that express this protein (13). More recently, immune checkpoint inhibitors have unexpectedly shown activity in ASPS, either alone or in combination with TKIs (14–16), though the mechanism of sensitivity is poorly understood.

The characteristic t(X;17)(p11;q25) translocation involving the Alveolar Soft Part Sarcoma Chromosomal Region 1 (ASPSCR1) and Transcription Factor E3 (TFE3) genes produces the oncogenic ASPSCR1::TFE3 protein. Canonically, the first 7 exons of ASPSCR1 are fused to exon 6 (type 1) or exon 5 (type 2) of TFE3 (17). However, paralleling perivascular epithelioid cell tumor (PEComa) and MiT family translocation renal cell carcinomas (tRCC), alternate TFE3 fusion partners have also been described in ASPS (18). The oncogenic program originated by the ASPSCR1::TFE3 chimeric transcription factor is poorly understood. Previous research efforts have implicated angiogenesis in the absence of hypoxia, MET and p21 upregulation, and metabolic utilization of lactate as unique oncogenic features of ASPS (6,9,19–22). However, the epigenetic and transcriptional programming by ASPSCR1::TFE3 has remained poorly understood, and novel therapies targeting this program have remained elusive.

Here, using ASPS tumor samples and preclinical models, we performed RNA-seq, assay for transposase-accessible chromatin with sequencing (ATAC-seq), and chromatin immunoprecipitation with sequencing (ChIP-seq) of H3K27ac and the fusion protein to describe the transcriptional and epigenetic landscapes of this disease. Utilizing an ASPSCR1::TFE3 construct bearing a degradation tag, we define the genes acutely regulated by the fusion protein that drive cellular proliferation, angiogenesis, mitochondrial biogenesis, and suppression of differentiation. Utilizing proximity proteomics, we found that ASPSCR1::TFE3 associates with BRD4, the Mediator complex, and components of the transcription initiation complex, among others, placing the fusion protein at multiple key regulatory points of gene expression. We found that ASPS relies on high expression of Cyclin D1, whose expression was controlled by ASPSCR1::TFE3, that ASPS was sensitive to CDK4/6 inhibition in vitro and in vivo in a patient derived xenograft (PDX) model, and that treatment of ASPS xenografts with both CDK4/6 and VEGFR inhibitors had superior effects in restricting tumor growth than either therapy alone. Taken together, these mechanistic studies uncover the epigenetic and transcriptional programs driven by ASPSCR1::TFE3 and targeted strategies to disrupt this oncogenic transcriptional regulator.

MATERIALS AND METHODS

Tumor Samples.

Fresh and formalin fixed and paraffin embedded (FFPE) archival ASPS samples were obtained and evaluated under institutional review board (IRB) approved protocols at Boston Children’s Hospital, Dana-Farber Cancer Institute, and University of Massachusetts Chan Medical School. The ASPS-1 PDX was obtained from the National Cancer Institute (23).

Cell Culture and Virus Production.

All cell lines tested negative for mycoplasma infection on routine surveillance (MilliporeSigma Cat# MP0025–1KT). Human embryonic kidney (HEK) 293FT (Thermo Fisher Scientific Cat# R70007, RRID: CVCL_6911), ASPS-1 (RRID: CVCL_S738 (23)), ASPS-KY (RRID: CVCL_S737, (24)), FU-UR-1 (RRID: CVCL_6997, (25)), and GIST-T1 (Cosmo Bio Cat# PMC-GIST01-COS, RRID:CVCL_4976) were cultured in Dulbecco’s modified Eagle’s medium containing 10% FBS, 2 mM L-glutamine, 100 mg/ml penicillin, and 100 mg/ml streptomycin. The ASPSCR1::TFE3 translocation was amplified from cDNA libraries to confirm cell line identity. Cells were thawed from original or derived stocks and used in the described experiments within approximately 3 months. Transfections were performed with X-tremeGene (Roche, Cat# 6365809001). Lentiviral production was performed as previously described (26). Briefly, 293FT cells were cotransfected with pMD2.G (Addgene #12259), psPAX2 (Addgene #12260) and the lentiviral expression plasmid. Viral supernatant was collected at approximately 72 h and debris removed by centrifugation at 1,000g for 5 min. Cells were transduced with viral supernatant and polybrene at 8 μg/mL by spinoculation at 680g for 60 min. Cell lines stably expressing Cas9 (Addgene #73310) were selected using a blasticidin (Thermo Fisher Scientific Cat# A1113903) selection marker and propagated as a polyclonal population. Tagged ASPSCR1::TFE3 cell lines were generated by simultaneous transduction with sgRNAs targeting ASPSCR1 or TFE3 to knockout the endogenous fusion protein and rescue ASPSCR1::TFE3 constructs optimized to alter sgRNA binding sequences and with fusion to function tags (dTAG or BioID). Drugs were used at the indicated concentrations and included palbociclib (LC Laboratories Cat# P-7788), sunitinib (LC Laboratories Cat# S-8803), NU9056 (MedChemExpress Cat# HY-110127), JQ1 (Selleck Chemicals Cat# S7110), THZ1 (Selleck Chemicals Cat# S7549), staurosporine (TargetMol Chemicals Cat# T6680), and dTAG13 (Tocris Cat# 6605). For growth over time assays, 15 x 10³ cells were dispensed per well in a 96 well plate, transduced with virus or treated with drug, and cell count performed on a Guava easyCyte Flow Cytometer (Luminex Corporation), with normalization of cell count to the control condition. Images of cells and area calculations were obtained on an EVOS M5000 imaging system (Thermo Fischer Scientific).

Cell cycle and apoptosis.

Cell cycle analysis was performed following palbociclib treatment for 72 h. Cells were trypsinized, washed in PBS and fixed in 70% ethanol. Propidium iodide at 25 μg/mL (Life Technologies Cat# P1304MP) and RNAse A at 0.2 mg/mL (Thermo Fischer Scientific Cat# EN0531) were used to stain nuclear DNA. Analysis was performed on a Guava easyCyte Flow Cytometer (Luminex Corporation), and single cells were assessed for nuclear content using Guava InCyte software. Apoptosis and cell death were measured following 72 h of palbociclib treatment using Guava Nexin Reagent (Luminex Corporation Cat# 4500–0450) per manufacturer’s recommendations. Non-apoptotic cells stain negative for Annexin V and 7-AAD, early apoptotic cells stain positive for Annexin V but negative for 7-AAD and late apoptotic and dead cells stain positive for both Annexin V and 7-AAD. Staining was assayed on a Guava easyCyte Flow Cytometer and data analyzed using Guava InCyte software.

Quantitative RT-PCR.

Cells were trypsinized and washed in PBS for RNA extraction using the Rneasy Mini Kit (Qiagen Cat# 74106). Libraries of cDNA were made using SuperScript IV VILO cDNA Synthesis Kit (Invitrogen Cat# 11766050). RT-PCR was performed using Power SYBR Green PCR Master Mix (Life Technologies Cat# 4367659) on a QuantStudio 6 Flex Real-Time PCR System (Thermo Fischer Scientific). Relative mRNA levels were calculated by the ΔΔCt method using GAPDH expression as reference. Primers are listed in Table S1.

RNA-seq.

Total RNA was isolated using an Rneasy Plus Kit (Qiagen Cat# 74136), and concentration measured by Nanodrop (Thermo Fisher Scientific) and quality by TapeStation 4200 (Agilent). Library preparation was performed using the NEBNext Ultra II DNA Library Prep Kit (New England Biolabs Cat# E7645S). Paired-end 150 bp sequencing was performed on a NovaSeq 6000 (Illumina). RNA-seq data were aligned to hg19 using STAR (27) with expression quantification using Cufflinks (28) to generate gene expression values in fragments per kilobase of transcript per million mapped reads (FPKM) units. Gene set enrichment analysis (GSEA, RRID:SCR_003199 (29)) was performed using custom gene lists or those in the Molecular Signatures Database (software.broadinstitute.org/gsea/).

ChIP-seq and ATAC-seq.

For ChIP-seq, approximately 20 x 10⁶ cells or 75 mg minced tumor tissue were incubated in 1% formaldehyde for 10 min. Following fixation, excess formaldehyde was quenched with glycine at 0.125 M for 5 min. Samples were washed with PBS, and intact nuclei suspended in SDS Buffer (0.5% SDS, 50 mM Tris, 100 mM NaCl, 5 mM EDTA with protease inhibitor cocktail (Roche Cat# 11873580001)) and sonicated in a E220 Focused-ultrasonicator (Covaris, Inc.). Sonicated samples were spun 20,000g for clarification and supernatant diluted to <0.1% SDS then incubated with Dynabeads Protein A (Life Technologies Cat# 10002D) pre-bound with antibody (H3K27ac, Abcam ab4729, RRID: AB_2118291; HA, Cell Signaling Technology Cat# 3724, RRID:AB_1549585) overnight. Samples were washed serially with Buffer A (150 mM NaCl, 5 mM EDTA, 5% sucrose, 1% Triton X-100, 0.2% SDS, 20 mM Tris), Buffer B (5 mM EDTA, 1% Triton X-100, 0.1% Deoxycholate, 20 mM Tris), Buffer C (250 mM LiCl, 1 mM EDTA, 0.5% NP40, 0.5% Deoxycholate, 10 mM Tris) and TE following resuspension of beads in Elution Buffer (200 mM NaCl, 100 mM NaHCO3, 1% SDS) and incubation at 65°C to reverse crosslinks for 12–15 h. DNA was purified using AMPureXP beads (Beckman Coulter Cat# A63881) per manufacturer recommendation, and quality assessed by Qubit dsDNA HS Assay Kit (Life Technologies Cat# Q32854) and TapeStation 4200 (Agilent). Sequencing libraries were prepared using a ThruPLEX DNA-seq Kit (Takara Bio Cat# R400675) and sequenced on a NextSeq 500 or NovaSeq X Plus (Illumina). All ChIP and input sequencing data (Table S2) were aligned to the human reference genome assembly hg19 using Bowtie2 (30). Normalized read density was calculated using Bamliquidator (version 1.0) read density calculator. Aligned reads were extended by 200 bp and the density of reads per base pair was calculated. In each region, the density of reads was normalized to the total number of million mapped reads, generating read density in units of reads per million mapped reads per bp (rpm/bp). ChIP-seq data sets were not downsized for comparisons. Peak finding was performed by Model-based Analysis for ChIP-seq (MACS, version 1.4.2 (31), and ROSE2 (32) was used to identify regions of signal enrichment. Individual ChIP-seq track displays were generated using bamplot (github.com/linlabbcm). Heat map visualizations of ChIP-seq data were generated using ChAsE (33). Gene ontology analysis was performed using Metascape (34). For ATAC-seq, preparation of nuclei from 50,000 viable cells, tagmentation, and library preparation were performed according to manufacturer’s instructions (Zymo Research Cat# D5458). Sequencing data were aligned, peaks called, and analysis performed as described for ChIP-seq. For ATAC-seq data set comparisons, individual samples were normalized to total aligned read counts.

Immunoblotting.

Cells were lysed in RIPA buffer containing protease inhibitor cocktail (Roche Cat# 11873580001) and centrifuged at 14,000g for 10 min to remove genomic DNA and debris. Protein concentrations were determined using a bicinchoninic acid-based assay (Pierce Biotechnology Cat# 23225). Protein samples were subjected to SDS-PAGE and Western blotting with the following antibodies: HA (1:1,000, Cell Signaling Technology Cat# 2367, RRID:AB_10691311), TFE3 (1:1,000, Cell Signaling Technology Cat# 14779, RRID: AB_2687582), GAPDH (1:1,000, Cell Signaling Technology Cat# 5174, RRID: AB_10622025), Rb (1:1,000, Cell Signaling Technology Cat# 9309, RRID:AB_823629), pRb (1:1,000, Cell Signaling Technology Cat# 8516, RRID:AB_11178658), E2F1 (1:1,000, Cell Signaling Technology Cat# 3742, RRID:AB_2096936), Cyclin D1 (1:1,000, Cell Signaling Technology Cat# 55506, RRID:AB_2827374), CDK4 (1:1,000, Cell Signaling Technology Cat# 12790, RRID:AB_2631166), CDK1 (1:1,000, Cell Signaling Technology Cat# 9116, RRID:AB_2074795), HIF1A (1:1,000, Cell Signaling Technology Cat# 36169, RRID:AB_2799095), Cas9 (1:1,000, Cell Signaling Technology Cat# 19526, RRID:AB_2798820), or streptavidin-HRP (1:40,000, Abcam Cat# ab7403). Western blots were probed with anti-mouse or anti-rabbit secondary antibodies and detected using the Odyssey CLx infrared imaging system (LI-COR Biosciences), or streptavidin-HRP by chemiluminescence (MilliporeSigma Cat# WBKLS0500). Immunoblots shown are representative of at least three independent experiments.

Mass Spectrometry and BioID.

ASPS cell lines were generated which stably expressed control or experimental mutant biotin ligase (BirA* R118G)-tagged fusion proteins. 24 h biotin-labeled whole cell lysate was subject to affinity pulldown overnight at 4°C using streptavidin-sepharose beads (GE Healthcare Cat# 17–5113-01). Beads were washed three times in 2% SDS in 50 mM Tris, twice in BioID buffer (50 mM Tris, 500 mM NaCl, 0.4% SDS), six times in 50 mM Tris and resuspended in 100 µL of ammonium bicarbonate. Samples were subject to tryptic digestion, and beads and salts removed in a reverse-phase cleanup step. Extracts were dried on a speed-vac, and later reconstituted in 5–10 µl of 2.5% acetonitrile and 0.1% formic acid. A nano-scale reverse-phase HPLC capillary column was created by packing 2.6 µm C18 spherical silica beads into a fused silica capillary (100 µm inner diameter x ~30 cm length) with a flame-drawn tip. After equilibrating the column each sample was loaded via a Famos Auto Sampler (LC Packings). A gradient was formed and peptides were eluted with increasing concentrations of 97.5% acetonitrile and 0.1% formic acid. As peptides eluted, they were subjected to electrospray ionization and then entered into an LTQ Orbitrap Velos Pro ion-trap mass spectrometer (Thermo Fisher Scientific). Peptides were detected, isolated, and fragmented to produce a tandem mass spectrum of specific fragment ions for each peptide. Peptide sequences (and hence protein identity) were determined by matching protein databases with the acquired fragmentation pattern by Sequest (Thermo Fisher Scientific). All databases include a reversed version of all the sequences and the data was filtered to between a one and two percent peptide false discovery rate. Label-free quantification of signal intensity was used in replicate samples for quantitative comparisons. Enriched proteins were defined as exhibiting >2-fold intensity enrichment in ASPS-KY^BioID compared to ASPS-KY^IKΔCT control with at least 10 PSM identified.

Xenograft Models and Immunohistochemistry.

For drug treatment studies, ASPS PDX or ASPS-1/Cas9 engrafted mice were enrolled into treatment groups when tumors reached approximately 100–200 mm³ in size, as measured by calipers and determined by the tumor volume equation: volume = short diameter² × long diameter × 0.5. Mice were randomly assigned to treatment groups administered palbociclib (100 mg/kg gavage daily, 5 days per week), sunitinib (40 mg/kg gavage daily, 5 days per week), or combination treatments. No statistical methods were used to predetermine sample size, and no animals died from drug treatment. Tumors were dissected and fixed in 10% formalin for corollary studies including H&E staining and immunohistochemistry of sectioned tumors. Immunohistochemical staining was performed on 4 µm sections prepared from formalin-fixed, paraffin-embedded tissue blocks after antigen retrieval using a citrated buffer pressure cooker protocol with following antibody: Ki-67 (1:400, Cell Signaling Technology Cat# 9027, RRID:AB_2636984). A secondary antibody conjugated to HRP was used (Cell Signaling Technology Cat# 8114). Reactions were developed using DAB (Cell Signaling Technology Cat# 8059) or NovaRed (Vector Laboratories Cat# SK-4800) substrate kits per manufacturer recommendations. For Ki-67 and cell count quantification, the total nucleated tumor cells or percent of Ki-67-positive cells in two separate areas of a stained section were divided into quadrants and individually counted by blinded reviewers (up to 8 regions per tumor). The resulting cell counts or percent Ki-67 positive cells were averaged across quadrants and then the entire section to generate a single value for each sample. For assessment of xenograft regions of viable, necrotic or fibrotic tumor, a pathologist with expertise in sarcoma scored sections in blinded fashion. All procedures were conducted under protocols approved by the Institutional Animal Care and Use Committee at Dana-Farber Cancer Institute.

Statistical analysis.

Center values, error bars, P-value cutoffs, number of replicates and statistical tests are identified in the corresponding figure legends. For box plots, the box extends from the 25^th to 75^th percentiles, with the center line indicating the median and whiskers drawn to the 10^th and 90^th percentile. Samples sizes were not predetermined for statistical calculations.

Data and Materials Availability.

Novel sequencing data are available through the Gene Expression Omnibus (GEO) and are available under GEO Publication Reference ID GSE235739. Additional sequencing data was derived from GSE95864, GSE107447, GSE150474, GSE188885, the Cancer Cell Line Encyclopedia (35),and the NCI Genomic Data Commons. The results here are in part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga. The Genotype-Tissue Expression (GTEx) Project was supported by the Common Fund of the Office of the Director of the National Institutes of Health, and by NCI, NHGRI, NHLBI, NIDA, NIMH, and NINDS. The data used for the analyses described in this manuscript were obtained from the GTEx Portal. All other raw data are available upon request from the corresponding author.

RESULTS

Transcriptional profiling of ASPS identifies enrichment in angiogenic, cell cycle, developmental, metabolic, and immune signatures.

Previous studies have evaluated microarray and RNA-seq of ASPS tumor and mouse models, implicating processes including angiogenesis, myogenic and stem cell gene expression programs, MET upregulation, and lactate metabolism as key to oncogenesis, with diverse proposals of the cell of origin including muscle, neural, pericyte or other mesenchymal progenitors (6,8,9,19,22). To further detail the transcriptional program of ASPS and evaluate the fidelity of laboratory models, we sought to perform a comparative analysis of ASPS clinical samples, PDX, and cell lines to other sarcomas to identify pathways enriched in this disease. We focused initial comparisons on GIST, as we have previously extensively characterized the transcriptional and epigenetic landscapes of GIST clinical samples and preclinical models (26,36). Unsupervised hierarchical clustering of RNA-seq data derived from ASPS and GIST tumors, xenografts, and cell lines showed clustering by disease subtype (Fig. 1A), indicating that the oncogenic programs driven by a mutant receptor tyrosine kinase or TFE3-translocation are conserved in laboratory models of GIST and ASPS, respectively. For ASPS cell lines, we utilized ASPS-1 and ASPS-KY; additionally, we included FU-UR-1, which is derived from tRCC but similarly driven by the ASPSCR1::TFE3 fusion oncogene. To identify transcriptional programs enriched in ASPS, we next performed gene set enrichment analysis (GSEA). ASPS showed enrichment in gene sets involved in oxidative phosphorylation, hypoxia, P53 and mesenchymal tissue development (Fig. 1B-E). Myogenic transcription factors including BHLHE40, BHLHE41, MYF5, and MYF6 all showed enrichment in ASPS compared to other sarcomas, including GIST, leiomyosarcoma (LMS), liposarcoma (LPS), malignant peripheral nerve sheath tumor (MPNST), myxofibrosarcoma (MFS), synovial sarcoma (SS) and undifferentiated pleiomorphic sarcoma (UPS); similarly, genes associated with angiogenesis and hypoxia, including HIF1A, VEGFA, ANGPTL2 and MDK showed unique enrichment in ASPS (Fig. 1F). Ultrastructurally, ASPS is characterized by abundant mitochondria (7), and we observed enrichment in ASPS for the oxidative phosphorylation gene signature (Fig. 1G). We also identified an increase in expression of genes localized to mitochondria in comparison to other sarcomas, as well as a distinctive enrichment in PPARGC1A (PGC-1𝛂, Fig. 1H-I), which functions as a master regulator of mitochondrial biogenesis and oxidative phosphorylation (37) and may account for the observed abundant mitochondrial gene expression and associated ultrastructural findings in ASPS. Compared to cancer types profiled by The Cancer Genome Atlas (TCGA) and cell lines in the Cancer Cell Line Encyclopedia (CCLE), ASPS expressed relatively higher levels of PPARGC1A with the exception of chromophobe renal cell carcinoma (Fig. S1A), which is known to highly express this gene and have correspondingly high mitochondrial content (38).

Figure 1. ASPS transcriptional profiling reveals enrichment in genes related to hypoxia, myogenesis, and mitochondrial function.

A, Unsupervised hierarchical clustering of RNA-seq data comparing the top 10,000 expressed genes across ASPS tumors (n = 5), ASPS PDX (2 passages of a single PDX model; indicated with ‘X’), ASPS cell lines (n = 4, including an ASPS-1 cell line xenograft indicated with ‘*’), RTK-mutant GIST tumors (n = 13), KIT mutant GIST PDX (n = 1) and KIT mutant GIST cell lines (n = 4). B, Butterfly plot of all Hallmark gene sets indicating the normalized enrichment score (NES) and FDR q-value comparing GIST and ASPS tumors (red) and cell lines (yellow). The Oxidative Phosphorylation (OxPhos), P53 Pathway, Hypoxia, and Myogenesis gene sets are labeled for each comparison. C-E, GSEA Hallmark P53 Pathway, Hypoxia, Myogenesis gene sets comparing ASPS and GIST tumors. F, Relative expression across ASPS tumors, ASPS PDX, ASPS cell lines and other sarcoma subtypes for myogenic transcription factors and hypoxia/angiogenesis genes. Sarcoma subtypes include GIST (n = 23, including SDH-deficient and RTK mutant GIST), leiomyosarcoma (LMS, n = 88), liposarcoma (LPS, n = 50), malignant peripheral nerve sheath tumor (MPNST, n = 8), myxofibrosarcoma (MFS, n = 15), synovial sarcoma (SS, n = 10), and undifferentiated pleiomorphic sarcoma (UPS, n = 40). G, Hallmark OxPhos gene set comparing ASPS and GIST tumors. H, Mean normalized FPKM in sarcoma tumors and ASPS models of all expressed genes with corresponding proteins localized to the mitochondria (n = 650). I, FPKM of PPARGC1A across ASPS and other sarcomas. Data were analyzed by one-way ANOVA with Tukey’s multiple comparison test (compared to ASPS tumors; ***,P<0.001; ASPS cells compared to other sarcomas; #,P<0.001). J, Unsupervised hierarchical clustering of CIBERSORT-determined leukocyte fraction comparing immune subsets between ASPS and other sarcoma subtypes. K, Expression of select targets of immune-oncology therapies and immune-related genes across ASPS and other sarcoma subtypes. L, Expression of MHC class I and B2M genes in ASPS and other sarcoma subtypes, with asterisk indicating genes with highest expression in ASPS. M, Tumor mutational burden (TMB, in mutations per megabase) in ASPS tumors (n = 5).

As immune checkpoint inhibitors have shown clinical activity in ASPS, we sought to evaluate immune cell populations and relevant markers in bulk RNA-seq data from ASPS and draw comparisons with other sarcoma subtypes. Using CIBERSORT (39), we found relative enrichment in M0 macrophages, mast cells, and γδ T-cells in ASPS compared to other sarcomas (Fig. 1J). Unexpectedly, across sarcomas ASPS had among the lowest expression of genes that modulate the immune response (CD274, PDCD1, TNFSF4, LAG3, HAVCR2, CTLA4), and the lowest expression of select B- and T-cell markers; in contrast, macrophage-associated CD68 was more highly expressed in ASPS than other sarcomas (Fig. 1K, Fig. S1B). Whereas many immune modulatory genes exhibited low expression, ASPS had the highest expression of several MHC class I molecules including HLA-A, HLA-F, HLA-G and HLA-H, as well as B2M (Fig. 1L, Fig. S1C-E). We next evaluated tumor mutational burden (TMB) in 5 clinical samples using a targeted next-generation sequencing (NGS) assay. As expected, an overall low mutational burden was observed in this translocation-associated sarcoma (Fig. 1M). Taken together, these data demonstrate the unique transcriptional signature of ASPS, confirm that ASPS models retain the oncogenic program of their tumors of origin, and that there is enrichment in the expression of myogenic, angiogenic, and mitochondrial programmatic gene sets compared to other sarcomas. Though bulk RNA-seq indicates less inflammatory infiltrates than other sarcomas and low TMB, ASPS exhibits a higher expression of MHC class I genes suggesting a greater capacity for antigen presentation which may underlie the sensitivity of this disease to immune checkpoint blockade.

ASPS has as recurrent chromatin landscape bound by ASPSCR1::TFE3.

As ASPSCR1::TFE3 is an oncogenic transcription factor that putatively orchestrates the epigenome, we next sought to evaluate the chromatin landscape of ASPS tumors, PDX, and cell lines using H3K27ac ChIP-seq. We evaluated the H3K27ac signal at super enhancers (SEs), which represent highly regulated regions of chromatin. ASPS tumors, PDX, and cell lines clustered together and separately from GIST tumors and cell lines (Fig. 2A), mirroring the transcriptomic data. Similar clustering by tumor histology was observed when considering all H3K27ac regions (Fig. S2A-B). Recurrent ASPS SE regions, generated by merging H3K27ac signal enriched above input chromatin in all ASPS tumors, exhibited higher signal in ASPS compared to GIST samples (Fig. 2B, Fig. S2C-E, Fig. S3A-H). Evaluation of ASPS SEs showed enrichment for genes regulating hypoxia, angiogenesis, cell cycle, mitochondrial biogenesis, development, and myogenic transcription factors (Fig. 2C), among others. These processes were also enriched in ASPS cell lines, and expression of ASPS SE-associated genes was higher in ASPS compared to GIST in both tumor samples and cell lines (Fig. S4A-E).

Figure 2. The ASPSCR1::TFE3 fusion protein invades sites of active chromatin in ASPS.

A, Unsupervised hierarchical clustering of H3K27ac ChIP-seq SEs from ASPS tumors (n = 4), ASPS PDX (n = 1, indicated with ‘X’), ASPS cell lines (n = 3), GIST tumor (n = 9), and GIST cell lines (n = 4). B, Ranked enhancer plot of merged H3K27ac ChIP-seq of ASPS tumors, with select enriched genes highlighted with their rank indicated in parentheses. C, Plot of the top significantly enriched gene ontology terms among SE-associated genes. D, Ranked enhancer plot of HA-tagged ASPSCR1::TFE3 ChIP-seq in ASPS-KY cells, with enriched genes labeled in B indicated on the plot. E, Tornado plot heatmaps of ASPS-KY H3K27ac ChIP-seq and ASPSCR1::TFE3 ChIP-seq centered around all TSS (upper panels) or H3K27ac-defined enhancers (bottom panels). F-J, Tracks from H3K27ac ChIP-seq of ASPS tumors and cell lines and ChIP-seq of the ASPSCR1::TFE3 fusion protein at the ASPSCR1, BHLHE40, BHLHE41, VEGFA, and CCND1 loci. K, Expression of CCND1 in FPKM comparing ASPS (n = 5), ASPS PDX (n = 2), ASPS cell lines (n = 3), GIST (n = 23), LMS (n = 88), LPS (n = 50), MPNST (n = 8), MFS (n = 15), SS (n = 10), and UPS (n = 40). Data were analyzed by one-way ANOVA with Tukey’s multiple comparison test (compared to ASPS tumors; ***,P<0.001; ASPS cells compared to other sarcomas; #,P<0.05).

To determine where in the genome ASPSCR1::TFE3 binds, we performed ChIP-seq of an HA-tagged fusion protein in ASPS-KY, identifying locations of genome-wide occupancy with enrichment at many regions with established SEs (Fig. 2D). Evaluating genome-wide binding of ASPSCR1::TFE3, the protein was found to be present at virtually all locations of active chromatin, with analogous enrichment near transcriptional start sites (TSSs) and enhancers across the genome (Fig. 2E), suggesting a model of global chromatin invasion and transcriptional amplification similar to that seen with oncogenic Myc (40–42). Enhancer regions bound by ASPSCR1::TFE3 showed recurrent enrichment in ASPS tumors and cell lines compared to GIST, and regions bound by ASPSCR1::TFE3 had significantly higher H3K27ac signal in ASPS compared to GIST (Fig. S5A-C).

Sites of H3K27ac and ASPSCR1::TFE3 enrichment may be particularly relevant for disease biology. We found three upstream and downstream enhancers from the ASPSCR1 TSS (Fig. 2F, Fig. S6A-D), which constitutes one of the largest regions of H3K27ac and ASPSCR1::TFE3 enrichment across the genome. This finding implies a positive regulatory feedback loop with the fusion protein controlling its own production. Enrichment was found in transcriptional regulators BHLHE40 and BHLHE41 (Fig. 2G-H, Fig. S7A-C, Fig. S8A-C), which have described roles in myogenic differentiation and hypoxia response (43,44), in genes involved in angiogenesis including VEGFA (Fig. 2I, Fig. S9A-C), and in B2M and across the HLA locus (Fig. S9D-E). We also found enrichment at CCND1, encoding Cyclin D1, which binds to cyclin-dependent kinase 4 (CDK4) to drive cell cycle progression from G1 to S phase (Fig. 2J, Fig. S10A-B) (45). At the level of gene expression, CCND1 was more highly expressed in ASPS patient samples, PDX, and cell lines in comparison to other sarcomas (Fig. 2K). Only liposarcoma, which commonly amplifies the locus containing CDK4, showed higher expression of CDK4 than ASPS; CDK6 was not expressed in ASPS (Fig. S10C-D). Taken together, these transcriptional and chromatin profiling studies highlight the conserved epigenetic program driving ASPS, implicate the ASPSCR1::TFE3 protein’s global involvement in transcriptional regulation, and demonstrate that ASPS PDX and cell lines represent high-fidelity models suitable for further mechanistic studies.

ASPSCR1::TFE3 acutely controls oncogenic transcription and suppression of myogenic differentiation.

All ASPS tumors and cell lines expressed the ASPSCR1::TFE3 fusion transcript, and of the cell lines only ASPS-KY expressed wild-type TFE3 (Fig. S11A-B). When we applied CRISPR/Cas9-mediated disruption of ASPSCR1::TFE3 with sgRNAs targeting exons encoding the fusion protein, proliferation of all ASPS cell lines was acutely halted; notably, the ASPS-1 cell line, following selection for stable Cas9 expression, demonstrated a higher proliferative rate compared to its parental cell line. (Fig. S11C-F). This result demonstrates the crucial role of the fusion protein to ASPS biology and its vulnerability to targeted disruption. To better understand the role of the fusion protein in oncogenic transcriptional regulation, we exploited this genetic dependency to knockout the endogenous fusion protein and rescue cells with an engineered ASPSCR1::TFE3 construct bearing dTAG (46), which allows for rapid and selective proteasome-mediated degradation of tagged proteins upon treatment with a heterobifunctional molecule (e.g. dTAG13) that binds to dTAG and recruits the E3 ligase cereblon. Using ASPS-KY cells, this system allowed for CRISPR/Cas9-mediated disruption of endogenous TFE3 or ASPSCR1 while ablating native ASPSCR1::TFE3, which is functionally replaced by the higher molecular weight tagged rescue construct (ASPS-KY^dTAG); treatment with dTAG13 led to rapid degradation of the HA-tagged fusion protein (Fig. 3A, Fig. S12A). That the tagged construct rescues viability indicates both the functional replacement of native ASPSCR1::TFE3 and that wild-type TFE3 and ASPSCR1 are dispensable for cellular function. As expected, treatment of ASPS-KY^dTAG cells with dTAG13 led to complete loss of proliferative capacity and G0/G1 cell cycle arrest (Fig. 3B-C), analogous to CRISPR/Cas9-mediated deletion of the fusion oncogene. Together with loss of proliferative capacity, dTAG13 treatment produced morphologic changes in cells including increased cell size and elongated shape, suggestive of cellular differentiation and/or senescence (Fig. 3D-E).

Figure 3. ASPSCR1::TFE3 orchestrates chromatin organization in ASPS.

A, Western blots for TFE3, HA and GAPDH in ASPS-KY parental cells or modified cells expressing an sgRNA targeting TFE3 and rescued with an ASPSCR::TFE3 construct fused with FKBP12^F36V including an HA tag (dTAG). The resulting ASPS-KY^dTAG/sgTFE3 cells were treated with dTAG13 at 500 nM for the indicated number of hours. B, Growth over time assay in ASPS-KY^dTAG/sgTFE3 cells treated with dTAG13 at 500 nM or DMSO as control. C, Effects of 500 nM dTAG13 treatment for 72 h on cell cycle in ASPS-KY^dTAG/sgTFE3 cells (n = 5 per condition). Data were analyzed using a two-way ANOVA with Tukey post hoc test, compared to DMSO; ***,P<0.001. D, Fluorescent images of GFP-positive ASPS-KY^dTAG/sgTFE3 cells treated with dTAG13 for the indicated time points (scale bar = 100 μm). E, Quantification of cell area in ASPS-KY^dTAG/sgTFE3 cells treated with dTAG13 for the indicated time points (n = 40 representative cells per condition). Data were analyzed by one-way ANOVA with Dunnett’s multiple comparison test; compared to 0 h; *,P<0.05; ***,P<0.001. F, Unsupervised hierarchical clustering of read count normalized ATAC-seq data comparing ASPS-KY^dTAG/sgTFE3 cells treated with DMSO (n = 3) or dTAG13 for 6 h (n = 3), 24 h (n = 2), and 72 h (n = 2). G, DMSO-normalized ATAC-seq read counts at all ASPS enhancers (n = 22,170), ASPS SE (n = 1,526), and regions bound by ASPSCR1::TFE3 (n = 17,905). Only regions with >50 aligned reads in control samples were considered. Absolute counts in each group were analyzed by one-way ANOVA with Dunnett’s multiple comparison test; compared to DMSO; ***,P<0.001. H, Volcano plot comparing ATAC-seq data from DMSO or 6 h dTAG13 treated ASPS-KY^dTAG/sgTFE3 cells at all ASPS SE (n = 1,526). Select regions are labeled, and the percentage of regions significantly up- or down-regulated is shown. I-K, ATAC-seq tracks from DMSO or dTAG13 treated ASPS-KY^dTAG/sgTFE3 cells at the CCND1, VEGF, and TRIM63 loci.

To evaluate for changes in chromatin arising from acute loss of ASPSCR1::TFE3, we treated ASPS-KY^dTAG/sgTFE3 cells with DMSO or dTAG13 for 6, 24, or 72 hours and performed ATAC-seq. While 6 hour dTAG13 treated samples were highly similar to control cells, 24 and 72 hour dTAG13 treatment showed increasing divergence (Fig. 3F). Globally, there was no significant change in read-count normalized signal at ASPS enhancers, ASPS SE or ASPSCR1::TFE3 bindings sites following 6 hours of dTAG13 treatment, whereas longer treatment periods led to a significant loss of open chromatin in these regions (Fig. 3G, Fig. S12B-E). To identify critical regions of accessible chromatin acutely altered by ASPSCR1::TFE3 loss, we compared ATAC-seq signal within ASPS SE regions between DMSO and 6 hour dTAG13 treated cells. While total read counts are not different between these treatment groups, 11% of these highly regulated regions showed significant changes in open chromatin (Fig. 3H).

Many chromatin regions identified as highly relevant to ASPS biology based on tumor, cell line, and ASPSCR1::TFE3 ChIP-seq and transcriptional profiling were acutely lost following dTAG13 treatment, including CCND1 and VEGFA (Fig. 3 I-K, Fig. S13A-F). While loci such as CCND1 and VEGFA showed preferential loss of accessible chromatin at enhancers following ASPSCR1::TFE3 degradation, other regions such as TRIM63 showed simultaneous loss of signal at the TSS, gene body, and enhancer. In contrast, several loci showed increased open chromatin following 6 hour dTAG13 treatment (Fig. S14A-C), including cytoskeletal genes such as PLEC and long non-coding RNAs. These results demonstrate the essential function of ASPSCR1::TFE3 in acutely controlling chromatin state in highly regulated genomic regions, and global effects on chromatin accessibility following prolonged loss of the fusion protein.

To assess the transcriptional changes arising from acute ASPSCR1::TFE3 degradation, we performed RNA-seq after 6 or 24 hours of dTAG13 treatment. Changes in global transcription were identified at both timepoints, with 8 and 16% of genes showing significant up- or down-regulation after 6 or 24 hours, respectively (Fig. 4A-C). At 6 hours, dTAG13 treatment produced significant downregulation of gene sets associated with oxidative phosphorylation and a subset of MYC transcriptional targets; by 24 hours, additional MYC and E2F transcriptional signatures were significantly lost, as were genes associated with the G2M checkpoint, which is a surrogate for cell cycle proliferation (Fig. 4D). Gene sets associated with epithelial-mesenchymal transition (EMT) and myogenesis were significantly enriched following dTAG13 treatment (Fig. 4D), suggesting myogenic differentiation of cells in the absence of oncogenic ASPSCR1::TFE3. Compared to global gene expression, all genes with protein products localized to mitochondria were selectively downregulated at both time points, while genes associated with myogenesis and myogenic differentiation were altered with tendency towards upregulation (Fig. 4E), reinforcing the essential role of ASPSCR1::TFE3 in mitochondrial biogenesis and suppressing differentiation down a myogenic lineage. Loss of ASPSCR1::TFE3 notably reduced key genes associated with mitochondrial biology, transcriptional regulation, angiogenesis, cell proliferation, and cell cycle regulation while altering markers of muscle differentiation (Fig. 4F-K, Fig. S14D-E). Particularly notable are the acute and complete loss of PPARGC1A (PGC-1𝛂), CCND1, and CDK4 transcripts, indicating that the fusion protein sustains the observed mitochondria-enriched phenotype of ASPS and drives cell cycle progression. A significant increase in CDKN2B expression with dTAG13 treatment also indicates the active suppression of this endogenous CDK4 inhibitor by ASPSCR1::TFE3. There was a strong concordance between genes with acute alterations in open chromatin and their resultant change in gene expression following ASPSCR1::TFE3 degradation (Fig. S14F-G). In keeping with ASPS originating from a myogenic lineage, expression of several of these highly regulated genes are exclusively found or highly enriched in muscle tissues (Fig. S15). TRIM63, for example, is a highly expressed and SE-associated gene in ASPS, is acutely sensitive to ASPSCR1::TFE3 disruption, and is exclusively expressed in skeletal muscle (Fig. 3K, Fig. S14E-F, Fig. S15). Loss of protein abundance was confirmed with select targets, including Cyclin D1 and CDK4, and associated loss of Rb phosphorylation was observed (Fig. 4L).

Figure 4. ASPSCR1::TFE3 supports cell cycle, mitochondrial, and angiogenic gene expression while suppressing myogenic differentiation.

A, Unsupervised hierarchical clustering of RNA-seq data (n = 10,000 genes) resulting from treatment of ASPS-KY^dTAG/sgTFE3 cells with dTAG13 at 500 nM for 6 or 24 h or DMSO as control (n = 4 per group). B-C, Volcano plots of RNA-seq data highlighting significantly changed transcripts arising from treatment with dTAG13 at 6 (B) or 24 h (C) (n = 5,190 expressed genes). The percentage of transcripts significantly up- or down-regulated is shown. D, Plot of Hallmark GSEA NES comparing ASPS-KY^dTAG/sgTFE3 cells treated with dTAG13 or DMSO as control. FDR compared to DMSO; **,P<0.01; ***,P<0.001. E, Log₂ transformed and DMSO-normalized values of all expressed genes (gray), expressed genes in the Hallmark Myogenesis gene set (orange, n = 64), or those with protein localized to the mitochondria (red, n = 624). F-K, FPKM from ASPS-KY^dTAG/sgTFE3 cells treated with dTAG13 or DMSO of exemplary mitochondrial genes (F), Hallmark Myogenesis genes (G), transcription factors (H), genes involved in hypoxia/angiogenesis (I), genes involved in cell division (J), and select cyclins and CDKs (K). Data in E-K were analyzed by one-way ANOVA with Tukey’s multiple comparison test; compared to All Genes or DMSO; **,P<0.01; ***,P<0.001. L, Western blot for the indicated proteins in lysates from ASPS-KY^dTAG/sgTFE3 cells treated with dTAG13 at 500 nM for 6 or 24 h or DMSO as control. M, Heatmap and clustering of FFPE-derived RNA-seq data showing expression of ASPS-associated genes comparing ASPS, TFE3-rearranged (Xp11) RCC and PEComa, conventional PEComa, and normal kidney. N, Heatmap and clustering of RNA-seq data derived from 293T cells transiently transfected with the indicated four TFE3 fusions, wild-type TFE3 or empty vector control showing expression of the indicated ASPS-associated genes.

We next analyzed previously reported transcriptional data sets studying TFE3 translocations to evaluate the ASPS transcriptional program in comparison to other tumor subtypes and fusion partners. From a series of TFSE3-translocated tumors including ASPS, tRCC and PEComa (47), enrichment in ASPS tumors was observed in genes we have identified as being relevant to ASPS and/or regulated by ASPSCR1::TFE3; while genes related to angiogenesis (VEGFA, ANGPTL2), myogenesis, and cell cycle (CCND1, CDK4) were highest in ASPS, other TFE3-translocated tumors also showed enrichment in mitochondrial gene expression and HLA-A, though to a lesser degree than ASPS (Fig. 4M, Fig. S16A). In contrast, in a study utilizing exogenous expression of ASPSCR1::TFE3 in 293T cells in vitro (48), these key ASPS transcriptional features were poorly reproduced. Comparing exogeneous expression of ASPSCR1::TFE3 to other TFE3 translocations, wild-type TFE3, or mock control, modest to no enrichment of these genes was observed (Fig. 4N, Fig. S16B). Taken together, these results indicate that global ASPS chromatin regulation and gene expression are dependent upon ASPSCR1::TFE3 function, that key transcriptional features of this disease require native epigenetic foundations derived from the cell of origin (akin to Myc-driven cancers), and that the fusion protein actively suppresses differentiation while activating oncogenic metabolic and growth programs.

ASPSCR1::TFE3 interacts with key regulators of enhancer function and transcription initiation.

To better determine the mechanism by which ASPSCR1::TFE3 regulates transcription, we next explored proteins and protein complexes that interact with the fusion protein. We utilized the same genetic system in ASPS-KY, knocking out the endogenous TFE3 and fusion gene while rescuing with a fusion construct bearing an N-terminal BirA* tag (ASPS-KY^BioID) to enable proximal protein identification using BioID (Fig. 5A) (49,50). BirA*, as a promiscuous biotin ligase, covalently labels proteins localized within ~10 nm of the tag with a biotin moiety, facilitating discovery of interacting partners of ASPSCR1::TFE3, as can be detected by Western blot probing cellular protein lysates with streptavidin (Fig. 5B); as a nuclear background control for BioID, we used a construct that fused BirA* to the DNA binding domain of IKZF1 (ASPS-KY^IKΔCT). Following labeling of cells for 24 hours with biotin, we performed a streptavidin pulldown followed by mass spectrometry, identifying 114 proteins labeled by ASPS-KY^BioID enriched over ASPS-KY^IKΔCT control (Fig. 5C, Table S3). Gene ontology analysis of these ASPSCR1::TFE3-proximal proteins showed enrichment in processes related to chromatin modification, DNA repair, enhancer function, and transcription initiation, with key members of these terms showing significant enrichment compared to control (Fig. 5D-E).

Figure 5. ASPSCR1::TFE3 interacts with the NuA4 Histone Acetyltransferase complex, DNA repair proteins, BRD4/Mediator complex members, and the Transcriptional Initiation complex.

A, Western blots for TFE3, HA and GAPDH in ASPS-KY parental cells or modified cells expressing either an sgRNA targeting TFE3 and rescued with an ASPSCR1::TFE3 construct fused with BirA* (R118G) including an HA tag (ASPS-KY^BioID), or stably expressing the DNA binding domain of IKZF1 fused with BirA* including an HA tag as control (ASPS-KY^IKΔCT). The additional TFE3-positive band visualized in the ASPS-KY^BioID condition (indicated with an asterisk) represents an N-terminal degradation product of the rescue construct. B, Western blot of ASPS-KY parental cells, ASPS-KY^BioID, or ASPS-KY^IKΔCT cells labeled with biotin and probed with streptavidin (SA)-HRP. C, Plot of peptide spectral match (PSM) and log₂ signal intensity of all identified proteins following streptavidin enrichment and tandem mass spectrometry (n = 1,367). ASPS-KY^BioID-enriched proteins, indicated in blue, show >2-fold intensity enrichment compared to ASPS-KY^IKΔCT control with at least 10 PSM identified (n = 114). Select interacting proteins are labeled. D, GO term enrichment for ASPSCR1::TFE3 proximal proteins. E, Signal intensity of control (IKΔCT) and ASPSCR1::TFE3 BioID-identified proteins including interactors involved in histone modification, enhancer formation, transcription initiation, and DNA repair. Data were compared by two-tailed t-test; compared to IKΔCT; *,P<0.05; **,P<0.01; ***,P<0.001. F, Plot indicating members of the NuA4 HAT, Mediator Complex and BET Family, the Transcription Initiation Complex, and DNA Repair proteins that interact with the BirA*-tagged ASPSCR1::TFE3 fusion protein. Log₂ intensity is plotted, with the size of the circle indicating the PSM count. The top 3 interacting proteins in each complex is listed. G, Growth over time assay showing day-21 cell count following transduction with the indicated sgRNAs targeting KAT5, RPS19 or luciferase as negative control in ASPS cell lines FU-UR-1, ASPS-KY and ASPS-1 (n = 5 per sgRNA in each cell line). H, Growth over time assay showing day-21 cell count following treatment with the KAT5 inhibitor NU9056 in ASPS cell lines FU-UR-1, ASPS-KY, and ASPS-1 (n = 5 in each cell line). I-J, Cellular viability in ASPS cell lines or GIST-T1 as comparator in response to 72 h treatment with JQ1 or THZ1 (n = 6 per data point). K-L, Relative mRNA levels by qRT-PCR of select pro-angiogenic (K) and cell cycle (L) transcripts in ASPS cell lines following 6 h treatment with 500 nM JQ1, THZ1 or DMSO as control (n = 4 per condition for each cell line). Data were analyzed by one-way ANOVA with Dunnett’s multiple comparison test; compared to Luc or DMSO; *,P<0.05; **,P<0.01; ***,P<0.001.

Among the functional protein complexes with components found to be in proximity to BirA*-tagged ASPSCR1::TFE3, the NuA4 histone acetyltransferase (HAT) complex, the Mediator complex, the Transcription Initiation complex, and DNA repair complexes showed greatest enrichment (Fig. 5F). The NuA4 HAT complex has multiple described roles in regulating chromatin through acetylation of histones and other transcriptional modifiers, regulating gene expression, signal transduction, and DNA repair (51,52). Using sgRNAs targeting the catalytic component of the NuA4 HAT, KAT5/Tip60, we found genetic dependency upon this complex in ASPS (Fig. 5G). This result was corroborated using the selective but low potency KAT5 inhibitor, NU9056, which showed modest growth reduction at 1 μM in a growth-over-time assay (Fig. 5H). As both ChIP-seq and proximity proteomics for ASPSCR1::TFE3 disclosed occupancy at both enhancers and promotors, specifically interacting with BRD4 and CDK7, we sought to demonstrate activity of selective inhibitors of these interacting proteins on viability and ASPS-associated gene expression. Both BRD4 inhibition with JQ1 and CDK7 inhibition with THZ1 showed potent toxicity in ASPS cell lines (Fig. 5I-J), comparable to what we have previously observed in GIST (53). Both inhibitors disrupted ASPSCR1::TFE3-associated gene expression within 6 hours of drug treatment (Fig. 5K-L) with key ASPS-associated genes regulating angiogenesis, cell cycle, and myogenic transcription exhibiting reduced expression, and with greater effects seen following 24 hours of drug treatment (Fig. S17A-F). Taken together, these data demonstrate the diverse protein interactions and nuclear processes administered by ASPSCR1::TFE3, which regulates essential enhancer and promotor complexes to control oncogenic gene expression, disruption of which alters the ASPS-associated transcriptional program.

ASPSCR1::TFE3-driven Cyclin D1/CDK4 expression is a targetable dependency in ASPS.

Transcriptional and epigenetic profiling revealed elevated and unique activation of Cyclin D1 in ASPS compared to other sarcomas, with expression of CDK4 and Cyclin D1 being driven by ASPSCR1::TFE3. We hypothesized that disruption of Cyclin D1/CDK4 activity, as a main effector of the ASPSCR1::TFE3-driven growth program, represents a therapeutically tractable means of targeting a core oncogenic effector of the fusion protein. Administration of the selective CDK4/6 inhibitor palbociclib to ASPS cell lines showed significant reduction in proliferation at doses as low as 5 nM, with minimal effect on GIST-T1 by comparison (Fig. 6A-D). The ASPS-1/Cas9 cell line showed more profound growth-inhibitory effects than the relatively slowly proliferating parental ASPS-1 (Fig. S18A-B). To confirm genetic dependency upon Cyclin D1 and CDK4, we utilized a CRISPR/Cas9 system with sgRNAS targeting CCND1 and CDK4 in ASPS cell lines and GIST-T1. Whereas GIST-T1 had little or no reduction in cell proliferative capacity following genetic deletion of CCND1 or CDK4, all ASPS cell lines exhibited significant reductions in proliferative capacity (Fig. 6E), confirming Cyclin D1 and CDK4 as essential components of the ASPSCR1::TFE3 oncogenic program. Palbociclib treatment led to G0/G1 cell cycle arrest in all ASPS cell lines, but showed modest changes in ASPS cell apoptosis as compared with staurosporine, as anticipated with disruption of CDK4-driven cell cycle progression (Fig. 6F-G) (54). Reduction of Rb phosphorylation and loss of E2F1 was observed by Western blot in all ASPS cell lines, with stabilization of Cyclin D1 and CDK4 following 24 hour palbociclib treatment, consistent with anticipated effects of palbociclib on G0/G1 cell cycle arrest (Fig. 6H-J) (45). Unlike dTAG13, palbociclib treatment did not induce dramatic morphologic changes in ASPS-KY^dTAG/sgTFE3 cells (Fig. S18C), reflective of Cyclin D1/CDK4 disruption representing just one of the multiple oncogenic pathways driven by ASPSCR1::TFE3.

Figure 6. Palbociclib causes cell cycle arrest in ASPS.

A-C, Growth over time assay showing effects of the indicated concentrations of palbociclib on ASPS-1/Cas9, ASPS-KY and FU-UR-1 cell lines. Data were normalized to DMSO control (n = 5 per condition). D, Day-12 normalized cell count in GIST-T1 or ASPS cell lines in response to treatment with the indicated palbociclib dose. E, Growth over time assay showing passage-4 cell count (day-27 for ASPS-KY, day-21 for other lines) following transduction with the indicated sgRNAs targeting CCND1, CDK4, RPS19, or luciferase as negative control in Cas9-expressing ASPS cell lines FU-UR-1, ASPS-KY, and ASPS-1 or GIST-T1 as comparator (n = 5 per sgRNA in each cell line). F, Effects of 500 nM palbociclib treatment for 72 h on cell cycle in ASPS cell lines (n = 5 per condition). Data were analyzed using a two-way ANOVA with Tukey post hoc test, compared to DMSO; **,P<0.01; ***,P<0.001. G, Effects of 500 nM palbociclib or 1 μM staurosporine treatment for 72 h on cell viability and induction of early or late apoptosis in ASPS cell lines (n = 5 per condition). Data were analyzed in D-E and G by one-way ANOVA with Dunnett’s multiple comparison test, compared to control; *,P<0.05; **,P<0.01; ***,P<0.001; compared to GIST-T1 in the same condition; #,P<0.05. H-J, Western blot for the indicated proteins in lysates from ASPS cells lines treated with palbociclib at 500 nM for 6 or 24 h or DMSO as control. K, Butterfly plot of all Hallmark gene sets indicating the NES and FDR q-value of RNA-seq data derived from the ASPS-1 cell line (n = 4 per condition) or ASPS PDX (n = 2 per condition) treated with palbociclib. The E2F Targets, G2M Checkpoint, MYC Targets V1, and MTORC1 Signaling gene sets are indicated for each condition. ASPS-1 was treated for 24 h with 100 nM palbociclib, while the PDX was treated for 3 d with 100 mg/kg palbociclib. L-M, Hallmark GSEA plots showing E2F Targets and G2M Checkpoint in ASPS-1 and ASPS PDX. N-O, Expression of key genes involved in cell cycle progression (M) and ASPS-associated genes (N) in ASPS-1 cells (gray) and ASPS PDX (blue). Data were evaluated by two-tailed t-test, compared to Control for each condition; *,P<0.05; **,P<0.01; ***,P<0.001.

To characterize the effect of palbociclib on global gene transcription, we performed RNA-seq following 24 hours of palbociclib treatment at 100 nM in ASPS-1 cells or 3 days at 100 mg/kg daily by gavage in ASPS PDX. Palbociclib treatment both in vitro and in vivo led to downregulation of transcriptional programs associated with proliferation, including the G2M checkpoint and E2F targets gene sets (Fig. 6K-M). While genes associated with cell proliferation were reduced in expression, including CDC20, FOXM1 and PLK1, transcriptional abundance of CCND1 was increased in agreement with Western blot results; expression of other ASPS-associated genes including VEGFA and TFE3 were unchanged at these timepoints (Fig. 6N-O). Taken together, these results demonstrate that ASPS utilizes Cyclin D1/CDK4 as a major effector of the oncogenic program in this disease, and targeted disruption with a CDK4/6 inhibitor leads to loss of proliferation and global transcriptional changes associated with cell cycle arrest.

Effects of palbociclib alone or in combination with sunitinib in vivo in ASPS.

To further assess the effects of CDK4/6 inhibition on ASPS growth in vivo, we utilized the ASPS PDX and cell line xenograft models. We first evaluated effects of short-term palbociclib treatment on Ki-67 as a marker for tumor growth arrest. ASPS PDX were treated for 3 days with palbociclib at 100 mg/kg daily by gavage. Tumors were stained after the treatment period with Ki-67, with resulting significant reduction in Ki-67-positive cells following palbociclib treatment (Fig. 7A-B). We next performed a 21-day efficacy study in the slowly growing ASPS PDX, with palbociclib treatment (100 mg/kg 5 days per week by gavage) resulting in complete cessation of tumor growth without reduction in mouse weight (Fig. 7C, Fig. S19A). Following the treatment period, histologic review demonstrated that the palbociclib-treated xenograft tumor cells were larger, with significantly fewer cells per high power field (HPF) compared to control xenografts (Fig. 7D-E). Though the low rates of engraftment and growth were insufficient for longitudinal experimentation using combination therapies in the ASPS PDX and the ASPS-1 and ASPS-KY cell line xenografts, we unexpectedly found that ASPS-1/Cas9 cells, which grew faster than their parental cell line (Fig. S11C), were amenable to reliable xenograft development. We treated ASPS-1/Cas9 engrafted mice with vehicle control, palbociclib (100 mg/kg 5 days per week by gavage), sunitinib (40 mg/kg 5 days per week by gavage) or the drug combination for 21 days. Whereas palbociclib alone did not significantly reduce ASPS-1/Cas9 xenograft growth, sunitinib monotherapy and the combination of sunitinib and palbociclib led to significant tumor growth reduction compared to vehicle, with the addition of palbociclib to sunitinib significantly reducing tumor volume compared to sunitinib alone (Fig. 7F). Mice in all treatment groups tolerated therapy without reduction in weight (Fig. S19B).

Figure 7. Effects of palbociclib alone or in combination with sunitinib in ASPS xenografts.

A, Representative images of ASPS PDX treated with palbociclib at 100 mg/kg daily or vehicle for 3 days and stained for Ki-67 (scale bar = 50 μm). B, Quantification of Ki-67-positive cells in ASPS PDX treated with vehicle (n = 4) or palbociclib (n = 4). C, Tumor volume of ASPS PDX in response to treatment with palbociclib (n = 9) or vehicle (n = 9) for 21 days. D, Histologic images of H&E-stained ASPS PDX treated for 21 days with vehicle or palbociclib (scale bar = 20 μm). E, Quantification of ASPS tumor cell count per high power field (HPF) in ASPS PDX treated with vehicle (n = 5) or palbociclib (n = 5). Data in B and E were compared by two-tailed t-test; **,P<0.01; ***,P<0.001. F, Tumor volume of ASPS-1/Cas9 xenografts in response to treatment with vehicle (n = 5), palbociclib (100 mg/kg gavage, 5 days per week; n = 5), sunitinib (40 mg/kg gavage, 5 days per week; n = 5), or the combination treatment (n = 5). Data in C and F were analyzed by two-way ANOVA, compared to vehicle; ***,P<0.001; compared to color-indicated condition #,P<0.05. G, Histologic images of H&E-stained ASPS-1/Cas9 xenografts in each treatment group (upper panel scale bars = 200 μm, middle panel scale bars = 50 μm, lower panel scale bars = 20 μm). The upper panel shows global tumor architecture, middle panel areas of viable tumor, and lower panel areas of tumor replacement fibrosis and necrosis, where present. H, Percentage of tumor in each group constituted by viable or necrotic/fibrotic tissue.

Histologic evaluation of ASPS-1/Cas9 vehicle treated tumors demonstrated the characteristic appearance of ASPS, with epithelioid tumor cells nested within highly vascularized septa with occasional central necrosis and pseudo-alveolar pattern. In comparison to the vehicle condition, treatment with palbociclib, sunitinib, or the drug combination all showed a trend towards increased tumor necrosis and fibrosis, indicative of a treatment response (Fig. 7G-H). In drug treated xenografts, all viable areas of tumor appeared equivalent to vehicle control, though the area of viable tumor was markedly reduced in the xenografts treated with the combination of palbociclib and sunitinib. Whereas the areas of non-viable tumor were small and sporadic in the vehicle condition, areas of necrosis and fibrosis were more prominent and larger in drug-treated xenografts (Fig. 7G, lower panel). Confirming the mechanism of activity of sunitinib was on angiogenesis in vivo rather than by causing direct toxicity to ASPS cells, we found little effect of sunitinib on ASPS cell viability in vitro, in contrast to GIST-T1 cells where sunitinib inhibits oncogenic KIT (Fig. S19C). Taken together, these results demonstrate anti-proliferative effects of palbociclib on ASPS in vivo. Palbociclib monotherapy acutely decreased Ki-67 and transcription associated with cell cycle progression in ASPS PDX while halting tumor growth, whereas in ASPS-1/Cas9 xenografts palbociclib caused increased tumor necrosis and fibrosis but only decreased overall tumor size in combination with sunitinib. The combination of palbociclib and sunitinib was significantly more effective than either therapy alone, with palbociclib targeting cell-intrinsic Cyclin D1/CDK4 while sunitinib targets cell-extrinsic angiogenesis, two central effectors of the ASPSCR1::TFE3-driven oncogenic program.

DISCUSSION

Approximately 20–30% of soft tissue sarcoma subtypes are driven by translocations, with the majority of these involving transcription factors or chromatin regulators (55). The mechanisms of oncogenesis of each of these fusions is diverse and often poorly understood. The ASPSCR1::TFE3 oncoprotein has previously been described to bind to limited regions genome-wide based on promotor microarray studies (22). However, our results using ChIP-seq demonstrate that the fusion protein is present at virtually all sites of active chromatin, with binding levels corresponding to global sites of H3K27ac enrichment. This manner of epigenetic engagement is analogous to oncogenic Myc, which functions primarily to bind at existing regions of active chromatin globally (TSS and enhancers) to augment the transcriptional program of the cell of origin and drive oncogenesis (40–42). In ASPS, the transcriptional program governed by ASPSCR1::TFE3 led to upregulation of specific pathways involved in mitochondrial biogenesis, myogenesis, angiogenesis, and cell cycle, with significant enrichment in ASPS compared to other sarcomas. These highly regulated genes and gene signatures were acutely dysregulated following loss of ASPSCR1::TFE3, leading to disruptions in chromatin state and gene expression, indicating the proximal control of these pathways by the fusion protein.

This mechanism of oncogenesis may provide additional insight into the cell of origin in ASPS. We found high expression of genes associated with myogenesis in ASPS tumor samples and preclinical models. Acute degradation of the fusion protein led to altered expression of genes related to more mature muscle structure and function, such as TRIM63, IP6K3, FHL1, SGCA, AGRN, PDLIM7 and COL6A2, which may be indicative of tumor cell differentiation down a myogenic lineage following loss of ASPSCR1::TFE3. Further suggestive of the influence of the cell of origin on tumor phenotype, neither TFE3 translocation-driven tumors of different histologies (PEComa, tRCC) nor epithelial cell lines transfected with ASPSCR1::TFE3 express many characteristic ASPS-associated genes. These findings further support the concept that both the epigenetic landscape of the cell of origin and activity of an oncogenic fusion protein are necessary for the ultimate cellular phenotype in individual cancers, even if identical oncogenic fusions are present. Alternatively, oncogenic fusions like ASPSCR1::TFE3 may reprogram the global chromatin state, leading to manifold changes in transcriptional programs that may not fully correlate with their cellular origins; more research is needed to understand this complex form of oncogenesis and the cellular origins of ASPS.

The activity of immune checkpoint inhibitors in the treatment ASPS patients was unexpected, and ASPS is currently the only sarcoma with such a therapeutic indication. In agreement with our bulk RNA-seq data, histologic analysis of ASPS has demonstrated the rarity of tumor infiltrating lymphocytes, with relatively abundant macrophages (56,57). Consistent with our assessment of TMB, others have found low mutational rates and few copy number alterations (58), as anticipated for this translocation-driven disease. Nevertheless, checkpoint blockade alone or in combination with TKIs has shown responses in prospective trials including ASPS patients (14,59–61). In agreement with our findings on the high expression of multiple MHC class I genes, a recent transcriptional profiling study of pediatric solid tumors reported high expression of HLA-A in ASPS (62). We further show that enrichment of specific immune subsets, such a γδ T-cells which have an established role in tumor immunobiology (63), may provide additional clues to the sensitivity of ASPS to checkpoint blockade. Further studies of antigen presentation and immune infiltration in ASPS are warranted. High capacity for antigen presentation and unique antigens, such as the fusion protein itself or components of the characteristic cytoplasmic granules in ASPS (64), may underlie immune recognition.

Using proximity proteomics, we identified interactions between ASPSCR1::TFE3 and the NuA4 HAT complex, Mediator/BET complex, and Transcription Initiation complex. These findings place the fusion protein at key regions of gene regulation within enhancers and the TSS, in full agreement with the observed genomic localization of ASPSCR1::TFE3 by ChIP-seq. Each of these complexes were essential for ASPS viability and/or ASPS-associated gene expression, as indicated by their genetic or pharmacologic disruption. These associations further detail the complex interaction network administered by an oncogenic fusion protein to drive oncogenesis and may indicate further opportunities to therapeutically disrupt its functions (e.g. KAT5, BRD4, or CDK7 inhibition).

Recently, a study from Tanaka and colleagues (65) reported that ASPSCR1::TFE3 was dispensable for ASPS cellular growth, and the oncogenic function of the fusion protein was principally through promoting angiogenesis in vivo by binding to SEs associated with angiogenic genes. In contrast, our results demonstrate the essential role of ASPSCR1::TFE3 in ASPS cellular viability, based on genetic (CRISPR/Cas9) and pharmacologic (dTAG13) experiments. Further, we confirm the ASPS chromatin landscape in human tumors aligns with the presented high-fidelity laboratory models, and demonstrate the much broader role of ASPSCR1::TFE3 in binding to and maintaining active chromatin, interacting with targetable chromatin regulatory complexes, and controlling the global oncogenic program in this disease. Additional research is needed to better understand the role of ASPSCR1::TFE3 and other TFE3-translocations across different model systems and cancer histologies.

We found robust evidence for mitochondrial enrichment in transcriptional data sets, in agreement with histologic features of ASPS (7), and genes encoding mitochondrial proteins were directly regulated by ASPSCR1::TFE3. Of particular importance, PGC-1𝛂, an established regulator of mitochondrial biogenesis in cancer (37), was highly expressed in ASPS compared to other sarcomas, and was acutely downregulated by loss of ASPSCR1::TFE3. Further work is needed to understand how ASPSCR1::TFE3 regulates mitochondrial biology and tumor metabolism, and whether this relationship represents a therapeutic vulnerability.

ASPS cellular proliferation driven by Cyclin D1/CDK4 represents a major effector and therapeutic liability of the ASPSCR1::TFE3 oncogenic program. We identified high levels of H3K27ac and ASPSCR1::TFE3 by ChIP-seq at the CCND1 locus together with high levels of CCND1 gene expression in ASPS. Expression of CCND1 and CDK4, and corresponding Cyclin D1 and CDK4 protein, were potently decreased by acute degradation of ASPSCR1::TFE3. The selective CDK4/6 inhibitor palbociclib potently induced cell cycle arrest, leading to decreased proliferation and accompanying transcriptional changes in vitro and in vivo. A limitation of our study was the availability of in vivo models for translational studies. While we could demonstrate short- and long-term therapeutic effects of palbociclib in the ASPS PDX, the low engraftment and growth rate of the ASPS PDX, ASPS-1, and ASPS-KY cell line xenografts prohibited combination studies. We were successful in developing a xenograft model using ASPS-1/Cas9, which for unclear reasons was more robust at engraftment and growth in vivo following stable expression of Cas9 and subsequent in vitro selection. This xenograft model showed insignificant tumor growth inhibition with palbociclib monotherapy, perhaps related to insufficient drug dose or exposure in the faster growing xenograft, though palbociclib-treated tumors did show increased tumor fibrosis and necrosis. In contrast, sunitinib monotherapy showed significantly reduced tumor growth rates in vivo, consistent with its known anti-angiogenic effect and the lack of direct toxicity on ASPS cells in vitro. The combination of sunitinib and palbociclib led to complete cessation of tumor growth, indicating the benefit of disrupting a cell-intrinsic growth program (Cyclin D1/CDK4) simultaneously with a cell-extrinsic growth program (angiogenesis). Taken together, these findings significantly advance our understanding of the mechanisms of oncogenesis of the ASPSCR1::TFE3 fusion, which may provide insight into how TFE3 translocations drive tumorigenesis in other cancer subtypes. We identify novel aspects of ASPS biology, notably with dependency upon Cyclin D1/CDK4 signaling, that may provide opportunity for targeted therapeutic intervention in this chemotherapy-refractory disease.

STATEMENT OF SIGNIFICANCE.

The ASPSCR1::TFE3 fusion propels the growth of alveolar soft part sarcoma by activating transcriptional programs that regulate proliferation, angiogenesis, mitochondrial biogenesis, and differentiation and can be therapeutically targeted to improve treatment.