Epithelial ovarian carcinomas are particularly deadly due to intratumoral heterogeneity, resistance to standard-of-care therapies, and poor response to alternative treatments such as immunotherapy. Targeting the ovarian carcinoma epigenome with DNA methyltransferase inhibitors (DNMTi) or histone deacetylase inhibitors (HDACi) increases immune signaling and recruits CD8+ T cells and natural killer cells to fight ovarian carcinoma in murine models. This increased immune activity is caused by increased transcription of repetitive elements (RE) that form double-stranded RNA (dsRNA) and trigger an IFN response. To understand which REs are affected by epigenetic therapies in ovarian carcinoma, we assessed the effect of DNMTi and HDACi on ovarian carcinoma cell lines and patient samples. Subfamily-level (TEtranscripts) and individual locus-level (Telescope) analysis of REs showed that DNMTi treatment upregulated more REs than HDACi treatment. Upregulated REs were predominantly LTR and SINE subfamilies, and SINEs exhibited the greatest loss of DNA methylation upon DNMTi treatment. Cell lines with TP53 mutations exhibited significantly fewer upregulated REs with epigenetic therapy than wild-type TP53 cell lines. This observation was validated using isogenic cell lines; the TP53-mutant cell line had significantly higher baseline expression of REs but upregulated fewer upon epigenetic treatment. In addition, p53 activation increased expression of REs in wild-type but not mutant cell lines. These data give a comprehensive, genome-wide picture of RE chromatin and transcription-related changes in ovarian carcinoma after epigenetic treatment and implicate p53 in RE transcriptional regulation.

Significance:

This study identifies the repetitive element targets of epigenetic therapies in ovarian carcinoma and indicates a role for p53 in this process.

See interview with Katherine B. Chiappinelli, PhD, recipient of the 2022 Cancer Research Early Career Award: https://vimeo.com/720726570

About 43% of the human genome is composed of repetitive elements (RE; ref. 1). As REs contain transcription factor–binding sites and other regulatory sequences, transcription of these elements is tightly regulated. REs may be expressed in embryonic stem cells but are mostly silenced by DNA methylation and repressive histone modifications in terminally differentiated cells. As part of the global epigenetic dysregulation that normal cells undergo during transformation, REs can lose repressive epigenetic marks, promoting aberrant transcription. Recent work has shown how aberrant RE transcription in cancer, especially from retrotransposons and transposable elements (TE), can lead to an immune response that promotes antitumor immunity (2–6).

There are three main classes of retrotransposons: long interspersed nuclear elements (LINE), short interspersed nuclear elements (SINE), and long terminal repeats (LTR, also known as endogenous retroviruses or ERVs). About 90% of LTRs have completely lost internal ORFs, leaving only the LTR sequences that cannot transpose (1). Only about 100 of 1.5 million LINEs remain intact enough to retrotranspose (1). Nevertheless, LTR and LINE promoters can alter gene expression (7) and destabilize the genome. As a result, retrotransposon sequences are silenced by epigenetic modifications including DNA methylation and repressive histone modifications (8).

Epigenetic regulation of transcription is disrupted in almost all cancers. This results in genome-wide loss of methylation and local hypermethylation at promoter regions that are normally unmethylated (9). 5-azacytidine (AZA) is a cytosine analogue that inhibits DNA methyltransferases (DNMTi). AZA can reactivate tumor suppressor genes silenced by DNA methylation at their promoters (10). AZA and another DNMTi, 5-aza-2′-deoxycytidine, are approved by the FDA for treatment of myelodysplastic syndrome (10) and AZA is approved for acute myeloid leukemia (11). We (4, 12) and others (13–16) have shown that low doses of DNMTis upregulate immune signaling, including the IFN response, cancer/testis antigens (CTAs), and antigen processing and presentation in breast, colon, lung, and ovarian carcinoma cell lines (12–16). The activation of the IFN response by DNMTi is caused by upregulation of double-stranded RNA (dsRNA), specifically LTR transcripts that activate the dsRNA sensors TLR3 and MDA5 (4, 15). Recent work has demonstrated the role of inverted repeat Alu (IR-Alu, a subgroup of SINEs) elements upregulated by DNMTi treatment (17) that bind to MDA5, triggering IFN signaling (2, 17). IFN signaling due to retrotransposon transcription can also be triggered by inhibitors of histone deacetylases (HDACi; ref. 18) or H3K9 methyltransferases (19). Combining DNMTi and HDACi increased REs in a mouse model of ovarian carcinoma (ID8; ref. 20), activating IFN signaling, and recruiting CD8+ T cells to kill the cancer cells (21). This work outlines a clear mechanism by which epigenetic therapies alter the immune microenvironment of cancer through transcriptional regulation of REs. However, it remains unclear which REs are activated by demethylation versus chromatin changes and whether tumor mutational background affects their transcription.

TP53 is a key tumor suppressor and the most commonly mutated gene in human cancers. About 30% of known p53-binding sites in the human genome are in LTRs, and p53 activates transcription of specific LTRs in the human colorectal cancer cell line HCT116 (22). In contrast, p53 can transcriptionally repress REs in fruit flies and zebrafish (23, 24). In human cell lines, LINE-1 is silenced by wild-type p53 and transcribed in TP53-null cells. Transcription in TP53-null cells is correlated with the loss of the H3K9me3 and H3K27me3 repressive histone marks (25). Ninety percent of TP53 mutations in human cancers cluster in the DNA-binding domain and are termed “hotspot mutations.” These hotspot mutations have diverse effects and may prevent p53 binding to canonical targets, instead promoting oncogenic transcription by binding to other loci through interaction with different binding partners (26). It remains unclear how mutant p53 affects retrotransposon expression in cancer, especially in ovarian carcinoma where the majority of cases contain a TP53 mutation (27).

To gain an understanding of which REs are affected by specific epigenetic therapies in ovarian carcinoma, we assessed the effect of AZA (DNMTi) and ITF-2357 (ITF, an HDACi) on ovarian carcinoma cell lines and ovarian carcinoma patient samples from a DNMTi clinical trial. Subfamily-level (TEtranscripts) and individual locus-level (Telescope) analysis of REs showed that DNMTi treatment upregulated significantly more REs than HDACi treatment, while the combination of DNMTi/HDACi increased RE transcription. Upregulated REs were dominated by the LTR and SINE families and SINEs showed the biggest differences in DNA methylation upon DNMTi treatment. Interestingly, cell lines with TP53 mutations exhibited significantly lower upregulation of REs with epigenetic therapy than wild-type TP53 cell lines. We validated this using isogenic cell lines (wild-type and mutant TP53) and found that the TP53-mutant cell line had significantly higher baseline expression of REs but upregulated fewer upon epigenetic treatment. Activation of p53 increases expression of REs in wild-type but not mutant cell lines, and wild-type p53 binds to genomic loci of specific RE families. These data give a comprehensive, genome-wide picture of RE chromatin and transcription changes in ovarian carcinoma after epigenetic treatment and implicate p53, a protein mutated in the majority of ovarian cancers, in RE transcriptional regulation.

Cell lines

Human ovarian carcinoma cell lines (A2780, Hey, Kuramochi, SKOV3, and TykNu) were kindly given to us by Dr. Stephen Baylin (Johns Hopkins University, Baltimore, MD) and have been verified by short tandem repeat analysis. The A2780, Hey, SKOV3, and TykNu cell lines were cultured in RPMI1640 (Corning, 10–104-CV) with 10% FBS (X&Y Cell Culture, FBS-500-HI), and 1% penicillin and streptomycin solution (Gibco, 15070063). The Kuramochi cell line was cultured in RPMI1640 media (Corning, 10–104-CV) with 10% FBS (X&Y Cell Culture, FBS-500-HI), 1% penicillin and streptomycin solution (Gibco, 15070063), and 1% nonessential amino acids (Gibco, 11140050). Cell lines were periodically tested for Mycoplasma via the Lonza MycoAlert kit.

Drugs and treatments

Cells were cultured in T75 dishes (Greiner, 658170) and treated with 500 nmol/L AZA (Sigma-Aldrich) or PBS. Media and AZA were replaced each day for five days (Fig. 1B). The cells were then split and allowed to reattach. ITF-2357 was added at a concentration of 100 nmol/L and treatment was continued for two days. The sequential treatment of these cells by these drugs was optimized by Topper and colleagues (28). The A2780 p53 chromatin immunoprecipitation sequencing (ChIP-seq) samples were treated with 500 nmol/L AZA (Sigma-Aldrich) or PBS for three days and 10 μmol/L Nutlin-3A (Cayman Chemical 18585) or DMSO for 6 hours.

Figure 1.

Epigenetic therapies upregulate transcription of REs. A, Primary mechanism by which epigenetic therapies upregulate transcription of repetitive elements, especially retrotransposons, to activate the IFN response. B, Treatment scheme for the ovarian carcinoma cell lines used. C, Counts of the repetitive element subfamilies upregulated by epigenetic therapies. Each bar represents data from one sample (n = 1) for each cell line and treatment combination. Left, raw counts. Right, proportion of counts. Color indicates class of repetitive element. Asterisks indicate enrichment by GSEA (FDR q < 0.05). D, qRT-PCR validation of the upregulation of two LTR loci by epigenetic treatments in the Hey and Kuramochi cell lines. Each qRT-PCR was performed twice in triplicate. Error bars, SEM. E, TEtranscripts data volcano plot showing log2 (fold change) and -log10 (DESeq2 Padj) for an independent set of three TykNu RNA-seq library replicates. HDACi, ITF; DNMTi, AZA.

Figure 1.

Epigenetic therapies upregulate transcription of REs. A, Primary mechanism by which epigenetic therapies upregulate transcription of repetitive elements, especially retrotransposons, to activate the IFN response. B, Treatment scheme for the ovarian carcinoma cell lines used. C, Counts of the repetitive element subfamilies upregulated by epigenetic therapies. Each bar represents data from one sample (n = 1) for each cell line and treatment combination. Left, raw counts. Right, proportion of counts. Color indicates class of repetitive element. Asterisks indicate enrichment by GSEA (FDR q < 0.05). D, qRT-PCR validation of the upregulation of two LTR loci by epigenetic treatments in the Hey and Kuramochi cell lines. Each qRT-PCR was performed twice in triplicate. Error bars, SEM. E, TEtranscripts data volcano plot showing log2 (fold change) and -log10 (DESeq2 Padj) for an independent set of three TykNu RNA-seq library replicates. HDACi, ITF; DNMTi, AZA.

Close modal

qRT-PCR

RNA (1 μg) was treated with DNaseI (Thermo #EN0525) for 30 minutes at 37°C. One microliter of 50 mmol/L EDTA (4.6 mmol/L final concentration) was added to quench the reaction and the DNase was denatured for 10 minutes at 65°C. This RNA was used in the qRT-PCR and the remainder was stored at −80°C. The RNA was reverse transcribed using the Applied Biosystems High-Capacity cDNA Reverse Transcription Kit (catalog no. 4368814) random primers and incubated at 25°C for 10 minutes, 27°C for 120 minutes, 85°C for 5 minutes, followed by a 4°C hold. qRT-PCR was performed with Applied Biosystems SYBR Green on the QuantStudio3 Quantitative reverse transcriptase PCR system. All qRT-PCR primer sequences are listed in Supplementary Table S1.

CRISPR/Cas9 genome engineering

Hey (TP53 wild-type) cells were electroporated with a nontargeting control gRNA or a gRNA targeting TP53 at amino acid 175 previously screened to have the highest editing efficiency. RNPs were formed by complexing crRNA and tracrRNA and subsequently adding recombinant Cas9 V3 (IDT) according to the manufacturer's recommendations. For generating the R175H point mutation, HDR templates were designed and synthesized as ssODNs ranging from 75–150 bp in length containing the desired base pair change together with a silent point mutation at amino acid 175 as well as a mutated PAM site and an additional silent point mutation to allow for analysis using restriction fragment length polymorphism (RFLP). Multiple polyclonal lines were generated from single-cell clones, which were subsequently expanded and tested for TP53 mutation via CRISPResso (29) and functional analyses (Nutlin-3A treatment followed by Western blot and qRT-PCR of target genes; Fig. 7A and B).

Western blot analysis

Cells were lysed in RIPA buffer (Pierce, 89900) with 1X protease and phosphatase inhibitor (Pierce, A32961). Lysates were sonicated at 4°C in a water bath Bioruptor (Diagenode) for 8 minutes (8 cycles of 30 seconds on, 30 seconds off). Next, samples were centrifuged at 4°C, 10,000 × g for 10 minutes to remove cellular debris. Protein concentration was determined according to the Pierce BCA Protein Assay Kit protocol (Thermo Fisher Scientific, 23225). Samples were mixed with NuPAGE LDS 4x loading gel (NP0007) and NuPAGE 10x reducing agent (NP0009) and then placed on a heating block at 100°C. Samples were loaded into a 4%–20% (Bio-Rad, 4561093) and transferred to LF PVDF (Bio-Rad, 170–4274). Membranes were blocked with LI-COR Biosciences Odyssey Blocking Buffer (927–40100) for 2 hours at room temperature then incubated overnight at 4°C with the primary antibody [1:1,000 p53 (rabbit, Bethyl A300–247)]. After the overnight incubation, β-actin (mouse, Sigma A5441) was added, 1:3,000, at room temperature for 20 minutes. Proteins were detected using the Azure Biosystems Imaging System c600. Processing of images was performed using the LI-COR Biosciences Image Studio software. The secondary antibodies used were AzureSpectra700 AC2128 and AzureSpectra800 AC2135.

p53 ChIP-seq library preparation

ChIP-seq library preparation was adapted from the Active Motif ChIP-IT High Sensitivity Kit protocol (Active Motif, 53040). A2780 cells were grown and treated in 150-mm dishes until 80% confluency, aiming for 15 million cells per treatment prior to crosslinking. Cells were cross-linked for 10 minutes at room temperature. The samples were sheared using 2 μL MNase (NEB, M0247S) for 10 minutes, then briefly sonicated 4 cycles (30 seconds on/30 seconds off) at 4°C. Samples were then centrifuged and the supernatant was used as the input and for the downstream immunoprecipitation. The immunoprecipitation was performed following the manufacturer's protocol of the Active Motif Kit, using 4 μL of the p53 antibody (Bethyl Laboratories, A300–247A-M). DNA was purified and used for qPCR analysis to verify enrichment prior to ChIP-seq. ChIP-seq was completed using the NEBNext Ultra II DNA Library Prep (NEB, E7103S) with multiplex oligos for barcoding (NEB, E7335S).

RNA-seq library preparation

Following treatment, total RNA was extracted using TRIzol (Thermo Scientific 15596026). Ribosomal RNA was depleted using Ribozero (Illumina). RNA sequencing libraries were prepared using the ScriptSeq v2 RNA-seq Library Preparation kit (SSV21106). Indexed libraries were sent to the Washington University in St. Louis Center for Genome Sciences for 75 bp, paired-end sequencing. The triplicate TykNu libraries were prepared for sequencing by depleting rRNA from 1 μg of total RNA with the NEBNext rRNA Depletion Kit for Human (New England BioLabs, E6310). Libraries were then prepared by following the manufacturer's protocol for the Illumina TruSeq Stranded Total RNA Library Prep Human/Mouse/Rat kit (P/N 20020596). Custom 10 bp UDI TruSeq-Compatible Duplex Y Adapters from IDT were to index the libraries (IDT10_UDI_1 through IDT10_UDI_15). Final libraries were sent to the Washington University in St. Louis Center for Genome Sciences and combined into an equimolar pool for 2 × 150 bp sequencing on an Illumina NovaSeq S4 (target read depth of 100 million). TP53 wild-type (isogenic Hey-derived, CRISPR-edited control line designated HC2), and TP53-mutant (isogenic Hey-derived, CRISPR-edited R175H mutant line designated HH23) libraries were produced from 750 ng of total RNA with a minimum RIN score of 7.0. This RNA was prepared for sequencing following the manufacturer's protocol for the Illumina TruSeq Stranded Total RNA Library Prep Human/Mouse/Rat kit (P/N 20020596) and TruSeq RNA Single Indexes Set A and B (P/N 20020492 and 20020493). Final libraries were separated into two sets of 12 and each set was combined into an equimolar pool for sequencing on an Illumina NextSeq 500 using two High Output 150 cycle v2.5 kits (P/N 20024907) and PhiX Control v3 (P/N FC-110–3001) spike-in of 1%.

Assay for transposase-accessible chromatin using sequencing library preparation

Assay for transposase-accessible chromatin using sequencing (ATAC-seq) was performed according to Corces and colleagues (30). Following treatment, DNA was extracted using the Nextera DNA Library Prep Kit (FC-121–1030). The libraries were prepared according to Corces and colleagues (30). Indexed libraries were sent to the Washington University in St. Louis Center for Genome Sciences for 75 bp, paired-end sequencing on an Illumina NextSeq500.

MeDIP-seq and MRE-seq library preparation

Following treatment and DNA extraction with phenol/chloroform, DNA was sent to the Ting Wang lab at Washington University in St. Louis (St. Louis, MO). Library preparation occurred as described by Xing and colleagues (31). For MRE-seq, only the HpaII, SsiI, Hin6I, and HpyCH4IV restriction enzymes were used. The finished libraries were sequenced at the Center for Genome Sciences at Washington University on an Illumina NextSeq500.

Annotation files

The hg38 reference sequence corresponded to the initial release without patches, GCA_000001405.15. The sequences were downloaded from ftp://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/analysisSet/* on April 5, 2018.

The GENCODEv21 genomic feature annotations downloaded from ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_21/gencode.v21.chr_patch_hapl_scaff.annotation.gtf.gz on June 4, 2018.

The RepeatMasker annotation (32) used here is: hg38 - Dec 2013 - RepeatMasker open-4.0.5 - Repeat Library 20140131 and can be accessed here http://www.repeatmasker.org/species/hg.html. The RepeatMasker table was reformatted into GTF and BED files for use in subsequent analysis.

TEtranscripts RNA-seq analysis

The sequence quality of the FASTQ files was assessed with FastQC. Reads were trimmed and adapters were removed using cutadapt using the –minimum-length 1 and -q 20 flags. The adapter sequences were -a AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC and -A AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT. Trimmed reads were aligned to hg38 using the following flags: –sjdbOverhang 100 –winAnchorMultimapNmax 200 –outFilterMultimapNmax 100. For libraries prepared by the ScriptSeq v2 RNA-seq, TEtranscripts was run on the STAR output with the following flags: –mode multi –stranded yes. For later, triplicate libraries prepared by the Illumina TruSeq Stranded Total RNA Library Prep Human/Mouse/Rat kit, TEtranscripts was run on the STAR output with the following flags: –mode multi –stranded reverse. The GENCODEv21 and RepeatMasker GTF annotation files were supplied. The log2(fold change) values output by DESeq were used in subsequent analyses. See Jin and colleagues for further reference (33). Library quality metrics are in Supplementary Table S2.

Telescope RNA-seq analysis

Telescope analysis was performed on the same aligned reads files as TEtranscripts. Telescope was installed and used according to the guidelines here: https://github.com/mlbendall/telescope. Before starting, miniconda was installed with the bioconda and conda-forge channels. Then a conda environment specifically for this task was created: conda create -n telescope_env python = 3.6 future pyyaml cython = 0.29.7 numpy = 1.16.3 scipy = 1.2.1 pysam = 0.15.2 htslib = 1.9 intervaltree = 3.0.2. In this environment, the “telescope assign” command was used to quantify RE expression. The output *report.tsv files were combined for processing with DESeq2, and the log2(fold change) values were used in subsequent analysis. For further information, see Bendall and colleagues (34). Library quality metrics are in Supplementary Table S2.

The Cancer Genome Atlas RNA-seq analysis

The Cancer Genome Atlas (TCGA) RNA-seq counts from HTSeq were downloaded from the GDC Data Portal (https://portal.gdc.cancer.gov/). Differential expression was calculated by DESeq2.

ATAC-seq analysis

The sequence quality of the FASTQ files was assessed with FastQC (0.11.5). Reads were trimmed and adapters were removed using cutadapt (1.16) using the –minimum-length 1 and -q 20 flags. The adapter sequences were -a CTGTCTCTTATACACATCT and CTGTCTCTTATACACATCT. Trimmed reads were aligned to hg38 with Bowtie2 (2.2.9) using the -X 2000 flag. To create input for HMMRATAC, the Bowtie2 output was converted to a BAM file, sorted, and indexed with samtools (1.3.1) view, sort, and index commands on the default settings. MACS2 callpeak (2.1.1.20160309) was used to create a peaks file on the standard settings. The bedGraph file output by MACS2 was sorted using the bash “sort -k1,1 -k2,2n” command then converted to a bigWig with bedGraphToBigWig. The HMMRATAC_v1.2.1 Java executable was run with the following flags: -m 75,200,400,600 –window 1250000 –bedgraph true -u 20 -l 10 -z 100. The peaks were filtered for a minimum score of 10 (i.e., at least 10 reads map to the open region). Then bedtools intersect (2.26.0) was used to remove peaks that had an 85% reciprocal overlap with peaks from the mock-treated sample. These treatment-specific peaks were used in subsequent analyses. For further reference, see Tarbell and Liu (35). Library quality metrics are in Supplementary Table S2.

methylMnM and methylCRF analysis

The sequence quality of the FASTQ files was assessed with FastQC (0.11.5). Reads were trimmed and adapters were removed using cutadapt (1.16) using the -q 20 flags. The adapter sequences were -a AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC and -A AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT. The reads were aligned to hg38 using bwa mem (0.7.12) on default settings. MeDIP and MRE data quality was further assessed using methylQA (0.2.1). Reference files for hg19 are publicly available. We made the necessary hg38 reference files for the methylMnM analysis using custom scripts provided by the Wang lab. Rscripts (R version 3.4.2) were used to further process the files and output DMRs (methylMnM) as well as predict methylation levels (methylCRF). For details, follow the protocol described in Xing and colleagues 2018 (31). Library quality metrics are in Supplementary Table S2.

ChIP-seq analysis

ChIP-seq reads were examined for gross problems with FastQC. Adapter removal and trimming were performed with cutadapt using this adapter sequence: -a AGATCGGAAGAGCACACGTCTGAACTCCAGT as well as the -q 20 and –minimum-length 1 flags. Reads were aligned to the genome with Bowtie2, and the alignments were quality checked with ChIPQC. Duplicate reads were marked with GATK MarkDuplicates (v4.0.8.0), and peaks were called with MACS2. Peaks consistent across replicates were identified with the ENCODE IDR pipeline (https://github.com/kundajelab/idr) using an IDR cutoff of 0.05. Library quality metrics are in Supplementary Table S2.

Mehdipour and colleagues' IR-Alu analysis

The list of IR-Alus reported by Mehdipour and colleagues (17) was shared with us by the Daniel de Carvalho laboratory at the University of Toronto and the liftOver tool was used to convert the hg19 coordinates to hg38. The methylCRF data in bigWig format was used via the deepTools computeMatrix and plotProfile tools to create metaplots of the methylation values over these IR-Alus.

Statistical analysis

Unless otherwise specified, all statistics were performed in R. The qRT-PCR statistics were run in Prism. When not performed as part of TEtranscripts or Telescope, RNA-seq expression values were called by DESeq2 (36). Venn diagrams were created using Vennerable, except for the nested Venn diagrams in Fig. 3 that were created with the eulerr package. Multi-set significance was calculated with the SuperExactTest package (37). For enrichment calculations involving lists of differentially expressed REs, we used gene set enrichment analysis preranked analysis on the RE log2(fold change) values (38). For enrichment calculations involving overlap of DMR, ATAC-seq peak, or ChIP-seq peak regions with REs, we used the regioneR package, running 1,000 permutations (39).

Availability of data and materials

The ovarian carcinoma cell line RNA-seq, methylMnM, methylCRF, ChIP-seq, and ATAC-seq data are available in GEO, accession GSE182430. The trial data from the Fang and colleagues study (40) are available in GEO, accession GSE102120. TCGA data was accessed via the NIH GDC Data Portal (https://portal.gdc.cancer.gov/). Custom scripts from our laboratory used in the analyses presented here can be found on the Chiappinelli Lab Github page (https://github.com/Chiappinelli-Lab/Ovarian-Cancer-RE-Analysis-Scripts).

Inhibiting DNA methylation drives global RE expression

Epigenetic therapy upregulates expression of a subset of LTRs to activate an IFN response in ovarian carcinoma (Fig. 1A; ref. 4). To identify other REs that change expression following epigenetic therapy, we treated the A2780, Hey, Kuramochi, and TykNu ovarian carcinoma cell lines with 5-azacytidine (AZA, a DNMTi), ITF-2357 (ITF, an HDACi), or a combination of the two (Fig. 1B). RNA-seq libraries from these samples were analyzed by TEtranscripts (33). This tool analyzes subfamily-level repetitive element (RE) expression by combining counts from all loci of a given RE subfamily. Both DNMTi and DNMTi/HDACi treatments increased interferon-stimulated gene (ISG) expression (Supplementary Fig. S1), as expected on the basis of previous analysis by qRT-PCR in lung cancer and ovarian carcinoma (28, 41).

409 repetitive element subfamilies of all major classes (LTR, LINE, SINE, DNA, and satellite repeats) were differentially expressed following treatment by DNMTi and/or HDACi (Fig. 1C; Supplementary Fig. S2A–S2C). The expression levels of several individual, upregulated retrotransposons, where primer design was possible, were validated by RT-qPCR (Fig. 1D). LINE and SINE transposon classes were significantly enriched in gene set enrichment analysis as represented by the asterisks in Fig. 1C. Although not statistically enriched, LTR retrotransposons were the most frequently upregulated transposon type and have been highlighted as key initiators of IFN signaling downstream of DNMTi treatment (4, 15) as have inverted repeat Alu (IR-Alu, SINE) elements (17). DNMTi upregulated more REs than HDACi and the combination of HDACi with DNMTi treatment upregulated a greater number of SINE elements than either treatment alone (Fig. 1C).

While REs were upregulated in all ovarian carcinoma cell lines by DNMTi and DNMTi/HDACi treatment, few RE subfamilies were commonly differentially expressed in any three or more of the four ovarian carcinoma cell lines (Supplementary Table S3). HERV9-int was the only LTR subfamily upregulated in all four cell lines by the DNMTi/HDACi combination treatment (Supplementary Table S3). Several other LTR, LINE, and satellite repetitive elements were upregulated in any three cell lines (Supplementary Tables S3–S5). Although few REs were upregulated across multiple lines, the overlap of these REs was nevertheless statistically significant and unlikely to occur by chance (Supplementary Fig. S2D). Further analysis of RE families after epigenetic treatment in the TykNu cell line confirmed that HDACi alone had very little effect while DNMTi drove RE transcription increases both alone and in combination with HDACi. Notably, many of the RE subfamilies that underwent DNMTi-induced transcription were LTRs (Fig. 1E).

Epigenetic therapies induce correlated epigenetic and transcriptomic changes

We next explored whether differential methylation or chromatin accessibility were responsible for the differential expression of RE subfamilies. We utilized MeDIP-seq and MRE-seq (31) followed by analysis with the methylCRF algorithm (31) to determine methylation at each CpG. This allowed us to calculate the average methylation change of each RE element subfamily or class (Fig. 2A). As expected, HDACi had little effect on DNA methylation while DNMTi alone or the DNMTi/HDACi combination drove methylation loss in all four cell lines. We observed the most significant demethylation at SINE elements (Fig. 2A and B). We validated DNMTi-induced methylation loss at the HERV-Fc2 5′ LTR by pyrosequencing analysis (Fig. 2C). This subfamily was one of the most significantly upregulated by DNMTi and DNMTi/HDACi in Fig. 1E.

Figure 2.

Retrotransposons, especially SINE elements, are demethylated by DNMTi treatment. A, Distribution in the average methylation difference for each repetitive element class. Methylation values are averaged across the entire repetitive element. The data for each cell line and treatment combination represent a single methylCRF library. Symbols indicate significance in a Kruskal–Wallis test and subsequent pairwise Wilcoxon test with Benjamini and Hochberg correction (P < 0.05). Individual symbols indicate a pairwise difference from a specific class (see legend). B, Counts of each repetitive element subfamily that overlap a differentially methylated region. Asterisks indicate significant enrichment of an RE class as determined by the regioneR package. C, Pyrosequencing percent methylation measurements at five CpGs in the HERV-Fc2 5′ LTR from three different cell lines. Asterisks indicate significant difference between the mock and AZA-treated samples by t test (P < 0.05). Error bars, SEM. D, Same as B, except the regions used in the enrichment calculations were the ATAC-seq peaks called by HMMRATAC. HDACi, ITF; DNMTi, AZA.

Figure 2.

Retrotransposons, especially SINE elements, are demethylated by DNMTi treatment. A, Distribution in the average methylation difference for each repetitive element class. Methylation values are averaged across the entire repetitive element. The data for each cell line and treatment combination represent a single methylCRF library. Symbols indicate significance in a Kruskal–Wallis test and subsequent pairwise Wilcoxon test with Benjamini and Hochberg correction (P < 0.05). Individual symbols indicate a pairwise difference from a specific class (see legend). B, Counts of each repetitive element subfamily that overlap a differentially methylated region. Asterisks indicate significant enrichment of an RE class as determined by the regioneR package. C, Pyrosequencing percent methylation measurements at five CpGs in the HERV-Fc2 5′ LTR from three different cell lines. Asterisks indicate significant difference between the mock and AZA-treated samples by t test (P < 0.05). Error bars, SEM. D, Same as B, except the regions used in the enrichment calculations were the ATAC-seq peaks called by HMMRATAC. HDACi, ITF; DNMTi, AZA.

Close modal

To better understand which RE classes were most affected by DNMTi and HDACi treatments, we calculated the enrichment of these REs within the regions of epigenetic change (Fig. 2B and D). For DNA methylation, we called differentially methylated regions (DMR) between the control and treated samples using methylMnM (31). We called regions of open chromatin from our ATAC-seq data using HMMRATAC (35). Although SINE elements had greater loss of methylation (Fig. 2A) and were enriched in both DMRs and ATAC-seq peaks (Fig. 2B and D), they were less affected at the transcriptional level (Fig. 1). From these data, we infer that SINE elements are the primary targets for the loss of methylation and gain of chromatin accessibility upon DNMTi treatment but that these changes are not strongly correlated with transcription from SINE elements.

Epigenetic therapy significantly upregulates individual RE loci

Correlating the broad epigenetic and expression changes at the subfamily level is problematic because data summarizing the effect at all loci mutes the signal from individual retrotransposon loci. To address this, we used the tool Telescope (34) to analyze LTR and LINE1 expression changes from individual loci. Telescope considers a curated set of 14,968 (approximately 1.9% of 771,683 total) LTR loci and 13,545 (approximately 0.8% of 1,609,790 total) LINE1 loci that slightly enrich for more intact, younger elements (42). We found that individual LTR and LINE1 loci are upregulated, especially by DNMTi treatment (Fig. 3A and B; Supplementary Fig. S3A) and several thousand LTR and LINE loci are upregulated by DNMTi and HDACi in ovarian carcinoma cell lines (Fig. 3C; Supplementary Tables S6 and S7). The number of upregulated elements common to all four cell lines is lowest for HDACi (26 LINEs, 142 LTRs) and higher for DNMTi (197 LINEs, 444 LTRs) and DNMTi/HDACi (259 LINEs, 175 LTRs; Fig. 3C; Supplementary Fig. S3B–S3E). Again, the overlap of all upregulated Telescope REs was small but statistically significant across cell lines (Supplementary Fig. S3F).

Figure 3.

Epigenetic therapies increase transcription of specific LTR and LINE loci. A, Distribution of the transcription fold change for individual LTR and LINE loci as determined by Telescope. B, Telescope data volcano plot showing log2 (fold change) and -log10 (DESeq2 Padj) for an independent set of three TykNu RNA-seq library replicates. C, Count of upregulated LTR and LINE loci. The innermost group was present in all four ovarian carcinoma lines and each circle outward indicates upregulation in one less line. These data are also presented in bar chart form in Supplementary Fig. S3. D, Cluster of repetitive element loci that lose methylation over the 5 kb upstream of the start coordinate and is upregulated. Each plot is centered on the LTR or LINE start coordinate, and each dot represents differential methylation at a single CpG (indicated by blue tick marks). Blue text indicates the locus ID. The number in red text indicates the upregulation fold change. Additional WIMSi clusters can be found in Supplementary Fig. S5. E, Metaplots of average methylation values covering the IR-Alus reported by Mehdipour and colleagues (17). Graphs present mean methylation and shaded areas indicate the SD. HDACi, ITF; DNMTi, AZA.

Figure 3.

Epigenetic therapies increase transcription of specific LTR and LINE loci. A, Distribution of the transcription fold change for individual LTR and LINE loci as determined by Telescope. B, Telescope data volcano plot showing log2 (fold change) and -log10 (DESeq2 Padj) for an independent set of three TykNu RNA-seq library replicates. C, Count of upregulated LTR and LINE loci. The innermost group was present in all four ovarian carcinoma lines and each circle outward indicates upregulation in one less line. These data are also presented in bar chart form in Supplementary Fig. S3. D, Cluster of repetitive element loci that lose methylation over the 5 kb upstream of the start coordinate and is upregulated. Each plot is centered on the LTR or LINE start coordinate, and each dot represents differential methylation at a single CpG (indicated by blue tick marks). Blue text indicates the locus ID. The number in red text indicates the upregulation fold change. Additional WIMSi clusters can be found in Supplementary Fig. S5. E, Metaplots of average methylation values covering the IR-Alus reported by Mehdipour and colleagues (17). Graphs present mean methylation and shaded areas indicate the SD. HDACi, ITF; DNMTi, AZA.

Close modal

To determine which REs lost methylation and gained expression, we used the WIMSi technique (43) with a list of LTR and LINE sites from Telescope. WIMSi calculates the difference in methylation between two samples at each CpG within 5 kb of a gene's transcription start site and interpolates a curve that represents the methylation pattern. These curves are clustered by similarity and a statistical test is applied to find clusters where more genes are coordinately up- or downregulated than expected by chance (43). We identified several clusters of REs that enriched for loci with similar methylation and expression changes (Fig. 3D; Supplementary Fig. S4A–S4E). The strongest cluster shows a correlation between upstream hypomethylation and upregulation of LTR and LINE loci upon treatment with DNMTi and HDACi (Fig. 3D). Each plot is centered on the LTR or LINE start coordinate, and each dot represents the difference in methylation at a single CpG (indicated by blue tick marks). Positive values indicate increased DNA methylation in the treated sample compared wth the control and negative values indicate decreased DNA methylation in the treated sample compared with the control. Blue text indicates the locus ID. The number indicates the expression fold change; red text represents RE upregulation, and green text indicates RE downregulation. The correlated changes in Fig. 3D are consistent with the known promoter activity of LTR and LINE sequences in general, although promoter activity at these specific loci is not guaranteed. The other clusters (Supplementary Fig. S4A–S4E) are weaker and suggest variability in the epigenetic changes induced by DNMTi and HDACi that hinders the clear identification of a good anchor point to use for this correlation (Supplementary Fig. S4F and S4G).

Besides decreases in methylation at specific RE loci, we observe a significant demethylation of inverted repeat Alu elements throughout the genome (Fig. 3E). It was recently shown that DNMTi demethylation of orphan, often intronic CpG islands led to upregulation of inverted repeat Alu elements in colorectal cancer. This study found that the large majority of REs that bind to MDA5 and trigger immune signaling are IR-Alus, though LTRs and other REs were also present (17). These elements, which we find are significantly upregulated by DNMTi treatment (Fig. 1E), are also demethylated by DNMTi treatment in ovarian carcinoma cell lines (Fig. 3E).

REs upregulated by DNMTis in ovarian carcinoma cell lines are also upregulated by DNMTis in ovarian carcinoma patient samples

Having identified specific REs upregulated by epigenetic therapies in ovarian carcinoma cell lines, we expanded our analysis to ovarian carcinoma patient samples and found similar results. We analyzed patient samples from a randomized controlled trial testing the combination of guadecitabine (DNMTi) and carboplatin in patients with ovarian carcinoma (40). Guadecitabine is a dinucleotide that resists cytosine deaminase degradation (44) and releases decitabine to inhibit DNMTs by the same mechanism as AZA. This trial demonstrated that DNMTi can reverse carboplatin resistance in patients with ovarian carcinoma by undoing methylation-induced gene silencing. Fang and colleagues included a total of 98 patients and produced RNA-seq libraries from 40 of them (40) with posttreatment RNA-seq libraries for eight patients. Our analysis contains an “all” category where all pretreatment (n = 40 patients, 75 samples) and all posttreatment (n = 8 of the 40 patients, 17 samples) samples are combined as well as analysis of the duplicate pre/post-treatment samples for eight patients individually. These latter samples are designated by a four-digit code (i.e., 0203). 980 RE subfamilies and 20,744 individual RE loci were differentially expressed as analyzed by TEtranscripts (Supplementary Fig. S5A and S5B; Supplementary Tables S8 and S9) and Telescope (Supplementary Fig. S5C and S5D; Supplementary Tables S10 and S11), respectively. The overall distribution of RE classes was similar to our cell line models: LTRs dominate the upregulated RE subfamilies (Supplementary Fig. S5A–S5D). The patients with ovarian carcinoma, like the cell lines, demonstrated variability in REs induced (Supplementary Fig. S5E and S5F). Only a single subfamily was upregulated in five or more of the eight paired samples when analyzed by TEtranscripts: 5S rRNA. However, there was significant overlap between the REs upregulated by DNMTi treatment in any ovarian carcinoma cell line and the REs upregulated by DNMTi treatment in any ovarian carcinoma patient sample, both at the subfamily (TEtranscripts) and individual locus (Telescope) levels (Supplementary Fig. S5G). These data show that (i) ovarian carcinoma cell line responses mimic patient responses to DNMTi, showing upregulation of LTRs, and (ii) any given RE may be targeted by epigenetic therapies and contribute to the immune response, but it is not common for the same RE subfamilies and/or loci be targeted in multiple patients and/or cell lines (Supplementary Tables S5, S7, S9 and S11).

Influence of TP53 on retrotransposon expression

To further understand the variability of RE upregulation in ovarian carcinoma lines and patient samples, we examined whether TP53 mutational status affected RE expression after epigenetic therapy. Approximately 95% of high-grade serous ovarian carcinoma tumors harbor TP53 mutations (27). Several studies have linked p53 with the regulation of retrotransposon transcription (22, 45). About 30% of p53-binding sites are located in LTR sequences, and p53 binding upregulates these LTRs (22). When we compare WT to mutant TP53 ovarian carcinoma cell lines, the mutant cell lines exhibit significantly less upregulation of REs with any epigenetic therapy compared with the wild-type cell lines at both the class/subfamily level TEtranscripts (Fig. 4AD; Supplementary Fig. S6A) and the locus-level Telescope data (Fig. 4E; Supplementary Fig. S6B and S6C). Although the differences shown in Fig. 4A are small, they are meant to show the role of wild-type p53 in promoting RE upregulation when all RE classes/subfamilies are considered in aggregate. This relationship is highlighted in subsequent plots for specific subfamilies/loci in Fig. 4B–E and is especially prominent for SINEs (Fig. 4B).

Figure 4.

TP53 mutation status and binding sites affect RE expression in ovarian carcinoma cell lines. A, Distribution of transcription fold change for RE subfamilies analyzed by TEtranscripts. Samples are subdivided by TP53 mutation status and individual plots are created for RE classes. WT cell lines are A2780 and Hey. Mutant cell lines are Kuramochi and TykNu. Asterisks indicate P < 0.05 by t test. B, Same as A, except only data from SINE subfamilies are presented. Asterisks indicate P < 0.05 by t test. C, Same as A, except only data for the HERV9-int subfamily are presented. D, Same as A, except only data for the LTR10B1 subfamily are presented. E, Distribution of Telescope fold change for the HERVIP10F_6q25.1 locus. F, Count of RE subfamilies (left) or individual LTR or LINE loci (right) that overlap a p53 ChIP-seq peak in A2780. Those significantly enriched (P < 0.05) by regioneR R package are indicated by asterisks. HDACi, ITF; DNMTi, AZA.

Figure 4.

TP53 mutation status and binding sites affect RE expression in ovarian carcinoma cell lines. A, Distribution of transcription fold change for RE subfamilies analyzed by TEtranscripts. Samples are subdivided by TP53 mutation status and individual plots are created for RE classes. WT cell lines are A2780 and Hey. Mutant cell lines are Kuramochi and TykNu. Asterisks indicate P < 0.05 by t test. B, Same as A, except only data from SINE subfamilies are presented. Asterisks indicate P < 0.05 by t test. C, Same as A, except only data for the HERV9-int subfamily are presented. D, Same as A, except only data for the LTR10B1 subfamily are presented. E, Distribution of Telescope fold change for the HERVIP10F_6q25.1 locus. F, Count of RE subfamilies (left) or individual LTR or LINE loci (right) that overlap a p53 ChIP-seq peak in A2780. Those significantly enriched (P < 0.05) by regioneR R package are indicated by asterisks. HDACi, ITF; DNMTi, AZA.

Close modal

To learn more about which REs are regulated by direct wild-type p53 binding, we performed p53 ChIP-seq on the TP53 wild-type A2780 cell line (Fig. 4F; Supplementary Fig. S7A–S7F). We treated the cells with Nutlin-3A (N3A) or AZA (DNMTi). Nutlin-3A activates wild-type p53 by binding to MDM2, a negative regulator of p53, to inhibit its repressive activity (46). Nutlin-3A induced p53 peaks at RE families, significantly enriched for LTRs and SINEs (Fig. 4F; Supplementary Table S12). Surprisingly, DNMTi treatment had a similar effect to Nutlin-3A, inducing p53 binding at all major classes of REs including the LTR, LINE, and SINE retrotransposons, although not to the same extent, and only significantly enriching for SINEs (Fig. 4F). At the individual locus level, LTRs were significantly enriched in p53 ChIP-seq data following Nutlin-3A or DNMTi treatment (Fig. 4F). From these data, we conclude that p53 binding to REs follows DNMTi treatment and affects RE transcription.

Having observed an effect of p53 on RE transcription, we expanded our analysis to analyze RE expression and TP53 mutations in The Cancer Genome Atlas ovarian carcinoma RNA-seq data (27). TCGA includes 372 total RNA-seq data sets with matching mutation data. We compared TP53 wild-type and mutant samples, all untreated. Interestingly, TP53-mutant cancers tended to express more individual LTR and LINE elements than their WT counterparts (Fig. 5A–C). Mutant TP53 status was also associated with an increase in lymphoid infiltration (although nonsignificant) based on xCell analysis of RNA-seq data (Supplementary Fig. S8A and S8B; ref. 47) as well as ISG upregulation (Supplementary Fig. S8C). Of the 308 RE loci that were upregulated in mutant TCGA ovarian carcinoma samples (compared with wild-type samples), 115 were also upregulated in the mock-treated TP53-mutant lines (Kuramochi and TykNu) compared with the mock-treated TP53 wild-type lines (A2780 and Hey; Supplementary Fig. S8D). Interestingly, only 14 of the 308 were also upregulated by epigenetic therapies. These data suggest that TP53-mutant ovarian carcinomas have baseline higher levels of specific REs, separate from those targeted by epigenetic therapies.

Figure 5.

TP53 mutation status affects RE expression in TCGA samples. A, Distribution of the fold changes for individual LTR and LINE loci. For A–C, the data are TCGA ovarian carcinoma samples: WT n = 17, Mutant n = 224. B, Count of LTR and LINE loci that are upregulated or downregulated in samples with mutant TP53. C, Volcano plots showing significance and magnitude of the expression changes in TCGA patient samples.

Figure 5.

TP53 mutation status affects RE expression in TCGA samples. A, Distribution of the fold changes for individual LTR and LINE loci. For A–C, the data are TCGA ovarian carcinoma samples: WT n = 17, Mutant n = 224. B, Count of LTR and LINE loci that are upregulated or downregulated in samples with mutant TP53. C, Volcano plots showing significance and magnitude of the expression changes in TCGA patient samples.

Close modal

To validate our observed induction of RE expression by p53, we performed targeted experiments in ovarian carcinoma cell lines. These included the TP53 wild-type A2780 and Hey lines, the TP53-mutant Kuramochi (p.D281Y, c.841G>T) and TykNu (pR175H, c.524G>A) lines, and SKOV3, which contains a nonsense mutation that prevents expression of the p53 protein (48). As expected, Nutlin-3A treatment increased p53 protein levels in the A2780 and Hey wild-type cell lines, but not the TykNu and Kuramochi mutant cell lines (Fig. 6A) and increased transcript levels of CDKN1A (P21; Fig. 6B), an important cell-cycle regulator and downstream transcriptional target of p53, in only the WT cell lines. Nutlin-3A treatment increased LTR (ERV-FC2 env, ERV-K env, Syncytin-1 (ERV-W1) env, and Syncytin-3 (ERV-P(b) env) expression in TP53 wild-type cell lines (Fig. 6C and D) but not in TP53-null or -mutant cell lines (Fig. 6E–G). ERV-FC2, Syncytin-1, and Syncytin-3 were previously shown to be increased by DNMTi in ovarian carcinoma cell lines (4) and the FC2 subfamily was one of the most significantly upregulated by DNMTi and DNMTi/HDACi the ovarian carcinoma cell lines profiled here (Fig. 1E). DNMTi upregulates LTR loci in a pattern similar to Nutlin-3A (Fig. 6H–L). However, the p53 hotspot mutant cell lines maintain upregulation of REs with DNMTi (Fig. 6J and K).

Figure 6.

Effects of TP53 on RE expression. A, Immunoblot of p53 in ovarian carcinoma cell lines treated with Nutlin-3a or control, 24 hours after Nutlin-3a treatment. β-Actin was used as a loading control. Cell lines were treated with Nutlin-3a and RNA was extracted 24 hours after treatment. B, qRT-PCR was performed for the CDKN1A (P21) transcriptional target of p53 in RNA from the A2780, Hey, and SKOV3 cell lines. C–G, qRT-PCR was performed for RE transcription in A2780 TP53 wild-type (C), Hey TP53 wild-type (D), Kuramochi TP53 mutant (E), TykNu TP53 R175H mutant (F), and SKOV3 TP53-null cell lines (G). H–L,TP53 WT, R175H, or null cell lines were treated with AZA and LTRs measured by qRT-PCR. A2780 TP53 wild-type (H), Hey TP53 wild-type (I), Kuramochi TP53 mutant (J), TykNu TP53 R175H mutant (K), and SKOV3 TP53-null cell lines (L). Asterisks indicate P < 0.05 by t test. Error bars, SEM.

Figure 6.

Effects of TP53 on RE expression. A, Immunoblot of p53 in ovarian carcinoma cell lines treated with Nutlin-3a or control, 24 hours after Nutlin-3a treatment. β-Actin was used as a loading control. Cell lines were treated with Nutlin-3a and RNA was extracted 24 hours after treatment. B, qRT-PCR was performed for the CDKN1A (P21) transcriptional target of p53 in RNA from the A2780, Hey, and SKOV3 cell lines. C–G, qRT-PCR was performed for RE transcription in A2780 TP53 wild-type (C), Hey TP53 wild-type (D), Kuramochi TP53 mutant (E), TykNu TP53 R175H mutant (F), and SKOV3 TP53-null cell lines (G). H–L,TP53 WT, R175H, or null cell lines were treated with AZA and LTRs measured by qRT-PCR. A2780 TP53 wild-type (H), Hey TP53 wild-type (I), Kuramochi TP53 mutant (J), TykNu TP53 R175H mutant (K), and SKOV3 TP53-null cell lines (L). Asterisks indicate P < 0.05 by t test. Error bars, SEM.

Close modal

We treated ID8 murine ovarian carcinoma cell lines modified by CRISPR/Cas9 (ID8 Trp53+/+ and ID8 Trp53−/−) with Nutlin-3A to activate p53 and assessed transcription of murine REs. P21 (CDKN2A) was induced by Nutlin-3A in the Trp53+/+ but not the Trp53−/− cell line (Supplementary Fig. S9A). Murine REs showed a similar pattern as in the human ovarian carcinoma lines, as Nutlin-3A induced their expression in the Trp53+/+ but not Trp53−/− ID8 cells (Supplementary Fig. S9B and S9C). Collectively, these data show that wild-type p53 induces transcription of LTR loci in response to DNMTi treatment in both human and murine ovarian carcinoma cells. This is consistent with the presence of p53-binding sites in many LTR elements (22).

The four ovarian carcinoma cell lines profiled in Figs. 16 differ in their TP53 mutation status and have other significant differences that include mutations in other tumor suppressors or oncogenes as well as cell doubling time. Thus, we used CRISPR/Cas9 genome editing to introduce the R175H TP53 hotspot mutation into the TP53 WT Hey cell line (Fig. 7A and B; Supplementary Fig. S10A–S10C). The R175H hotspot mutant Hey cell line exhibits baseline p53 protein expression, similar to the TykNu cell line (Figs. 6A and 7A). Nutlin-3A treatment induces expression of p53 target genes in the p53 WT Hey cell line, but not the p53 R175H–mutant cell line (Fig. 7B). We compared RE expression between the wild-type (Hey WT, designated HC2) and mutant (Hey R175H, designated HH23) cell lines baseline and after DNMTi or Nutlin-3A (N3A) treatment. Baseline expression of RE families, including LTRs and SINEs (Fig. 7C), and individual loci (Fig. 7D) was greater in the Hey R175H compared with the Hey WT cell line. Upon Nutlin-3A activation of p53, only the WT cell line showed significant upregulation of REs, mostly LTRs (Fig. 7E and F). Upon DNMTi treatment, the Hey R175H cell line upregulated fewer REs compared with the Hey WT cell line (Fig. 7E and F). Combining DNMTi with Nutlin-3A p53 activation upregulated many more RE families and individual loci in the Hey WT cell line than either AZA or Nutlin-3A alone, but had minimal effects in the Hey R175H cell line (Fig. 7E and F). These data show that the hotspot R175H TP53 mutation causes an increase in RE expression baseline and that epigenetic therapy has less of an effect on RE transcription in the mutant line.

Figure 7.

Mutant TP53 changes baseline expression and DNMTi induction of repetitive elements in ovarian carcinoma cell lines. A, p53 Western blots from CRISPR/Cas9-edited TP53 wild-type (HC2) and R175H mutant (HH23) cell lines expanded from single clones. B, RNA was isolated from the cells in A and qRT-PCR performed for p53 target genes. C, TEtranscripts volcano plot showing the comparison between the TP53 mutant and TP53 wild-type Hey–derived lines. Bar plot at right indicates the class composition of the REs that are significantly upregulated in the TP53 R175H cell line with a greater than two-fold change in expression. D, Telescope volcano plot showing the comparison between the TP53 mutant and TP53 wild-type Hey–derived lines. Bar plot at right indicates the class composition of the REs that are significantly differentially expressed with a greater than two-fold change in expression. E, TEtranscripts volcano plots showing REs upregulated by epigenetic therapy in the TP53 wild-type line (HC2, left) and the R175H mutant line (HH23, right). Bar plots show the count of elements that are significantly upregulated with greater than two-fold change in expression to emphasize the differences between the TP53 wild-type and mutant lines. F, Same as in E, except the data are from Telescope analysis of TP53 wild-type and mutant cell lines. DNMTi, AZA.

Figure 7.

Mutant TP53 changes baseline expression and DNMTi induction of repetitive elements in ovarian carcinoma cell lines. A, p53 Western blots from CRISPR/Cas9-edited TP53 wild-type (HC2) and R175H mutant (HH23) cell lines expanded from single clones. B, RNA was isolated from the cells in A and qRT-PCR performed for p53 target genes. C, TEtranscripts volcano plot showing the comparison between the TP53 mutant and TP53 wild-type Hey–derived lines. Bar plot at right indicates the class composition of the REs that are significantly upregulated in the TP53 R175H cell line with a greater than two-fold change in expression. D, Telescope volcano plot showing the comparison between the TP53 mutant and TP53 wild-type Hey–derived lines. Bar plot at right indicates the class composition of the REs that are significantly differentially expressed with a greater than two-fold change in expression. E, TEtranscripts volcano plots showing REs upregulated by epigenetic therapy in the TP53 wild-type line (HC2, left) and the R175H mutant line (HH23, right). Bar plots show the count of elements that are significantly upregulated with greater than two-fold change in expression to emphasize the differences between the TP53 wild-type and mutant lines. F, Same as in E, except the data are from Telescope analysis of TP53 wild-type and mutant cell lines. DNMTi, AZA.

Close modal

Overall, these data demonstrate that epigenetic therapies upregulate diverse RE subfamilies and individual loci in ovarian carcinoma cell lines and patient samples, and this upregulation is affected by TP53 status. HDACi alone had minimal effect on RE expression while DNMTi treatment significantly increased RE expression, notably at LTR, LINE, and SINE elements, and the combination of DNMTi/HDACi upregulated more REs than either treatment alone. These results are similar to regulation of genic sequences (49). Both LTR and SINE transcripts are protected from RNase digestion by MDA5 (17), and thereby transcripts from the RE subfamilies listed in Supplementary Table S3 are potential triggers of IFN signaling and the subsequent antitumor response.

While there was significant variability in RE induction between cell lines, several RE subfamilies and individual transposable elements were upregulated in multiple cell lines following treatment with DNMTi and/or HDACi. Importantly, these REs were also found in ovarian carcinoma patient samples. The RE subfamilies that are upregulated in multiple samples include HERV9-int and HERV-Fc1-int. HERV9 had previously been identified by Brocks and colleagues as commonly upregulated by DNMTi/HDACi in lung cancer cell lines (18) and we confirm this in ovarian carcinoma cell lines. HERV9 elements are relatively recent integrations into the human genome are closely related to the HERVW elements that gave rise to Syncytin-1 (50). Syncytin-1 expression can occur due to loss of methylation in ovarian carcinoma and precancerous legions (51). SINEs, including Alu elements, were significantly demethylated and upregulated by DNMTi treatment in ovarian carcinoma cell lines and showed less upregulation in TP53-mutant cell lines.

Mehdipour and colleagues (17) recently showed that DNMTi demethylation of orphan, often intronic CpG islands led to upregulation of inverted repeat Alu elements in colorectal cancer. The IFN signaling that resulted from the upregulation of Alu transcription decreased tumor size in treated mice (17). Their study found that the large majority of REs that bind to MDA5 and trigger immune signaling were Alus, although LTRs and other REs were also present. This is consistent with earlier work that showed Alu binding to MDA5 can be regulated by RNA editing (2, 5). Because a growing body of evidence supports the role of Alus in the IFN response, it is interesting that we observe significant Alu demethylation by DNMTi in ovarian carcinoma. We also found noticeable upregulation of Alus by DNMTi/HDACi treatment while DNMTi alone did not robustly upregulate these elements. This hints at a slightly different mechanism of action where histone acetylation may be more important for Alu regulation than in the Mehdipour and colleagues' study, which could be caused by the differences between cancer types.

RE expression can be driven by nonepigenetic events including transcription factor binding. Our data show that p53 can transcriptionally activate REs in cancer cells. Hotspot TP53 mutants show higher baseline levels of RE expression and less RE upregulation upon DNMTi treatment (Figs. 5 and 7). This may occur because TP53-mutant cell lines express p53 even in the absence of Nutlin-3A induction (Figs. 6A and 7A) and thus the p53 protein can bind and activate RE expression without treatment in TP53-mutant cell lines. Interestingly, combining DNMTi (AZA) treatment with Nutlin-3A induction of p53 induced many more RE families and individual loci than either treatment alone, but only in the p53 WT cell line (Fig. 7). This confirms transcriptional activation of REs by wild-type p53, which is further enhanced by loss of methylation genome-wide. It is also interesting to note that HDACi can downregulate p53 expression (52). Although we did not examine the effects of ITF treatment on p53 expression in our cell lines, perhaps this effect contributes to the low levels of RE upregulation with ITF treatment alone. Thus, p53 plays a crucial role in regulation of RE transcription in ovarian carcinoma. This role may be direct, as we infer at loci that exhibit p53 binding by ChIP-seq (Fig. 4F), or indirect, through downstream signaling after p53 activation. p53 transcriptionally enhances interferon signaling (53) and recent work has shown that type I IFN signaling can induce LTR expression (54). When used at high doses, the DNMTi, AZA, can induce DNA damage and p53 activation, but at the low dose used in these experiments (500 nmol/L) does not induce dsDNA breaks (55). Further mechanistic work is thus clearly needed to determine how wild-type and mutant p53 regulate RE expression in the context of epigenetic therapies.

Data from primary ovarian carcinomas also correlates the presence of hotspot mutant p53 with RE expression. In ovarian cancers, STIC lesions—the carcinoma in situ precursors to high-grade serous ovarian cancer—are characterized by mutant TP53 along with demethylation and increased expression of L1 elements (56). LINE-1 retrotransposition is limited by the p53 DNA damage response as well as IFN signaling and p53 directly represses human LINE1 (25, 56). We observed p53 binding to LINE1 elements after Nutlin-3A or DNMTi treatment (Fig. 4F) but did not observe significant downregulation of LINE1 elements after these treatments (Fig. 7E and F). Instead, we observe p53 enrichment at and upregulation of the LTR and SINE RE classes (Figs. 4F, 6, and 7). The ChIP-seq peaks we observed at REs were often at or near LTR10 elements (Supplementary Fig. S11A–S11C). This is consistent with previous work showing that LTR10 elements contain p53-binding sites (22).

Limitations of our study include the etiology of the ovarian carcinoma cell lines used. We recognize that the A2780 and Hey cell lines likely do not model high-grade serous ovarian carcinoma, while Kuramochi and TykNu are much better models of high grade serous ovarian carcinoma (57, 58) A2780 and Hey exhibit wild-type TP53 and mutations found in other ovarian carcinoma subtypes, including ARID1A. Differences in histologic subtype between the four ovarian carcinoma lines profiled likely contribute to the significant variability we observe in which RE families and individual loci are induced by DNMTi/HDACi treatment. As high-grade serous ovarian carcinoma is characterized by nearly 100% TP53 mutations (59), there is no TP53 wild-type high-grade serous cell line that we can use for our studies. To isolate the effects of p53, we thus generated isogenic TP53 wild-type and -mutant cell lines and analyzed the effects of epigenetic therapies and p53 activation in this controlled setting, showing a significant effect of p53 status on RE baseline transcription and upregulation by DNMTi (Fig. 7).

The importance of determining how and which REs are regulated by epigenetic factors and p53 is emphasized by ongoing clinical trials for combined epigenetic and immune therapy. A phase Ib trial combining DNMTi treatment with anti-CTLA-4 to fight melanoma showed promising results, including improved immune activation and antitumor activity (60). This combination therapy is currently being tested in clinical trials for melanoma, colorectal cancer, ovarian cancer, and kidney cancer, among others (NCT01928576, NCT02961101, NCT03019003, NCT02811497, NCT02546986, NCT02397720, NCT02530463; ref. 60). A recent study on TP53 hotspot mutations in acute myeloid leukemia showed that cancers with mutant TP53 exhibited higher IFN signaling, more infiltrating immune cells, and stronger response to immunotherapy, emphasizing the need for further study of how this transcription factor cooperates with epigenetic regulators to shape RE transcription and subsequent interferon signaling (61). Finally, REs including ERVs are being actively investigated as potential cancer-specific antigen targets for immunotherapy (6, 62–65). Understanding how the most commonly mutated protein in cancer affects RE expression, both baseline and with epigenetic therapy, will impact future immunotherapies for ovarian carcinoma and other solid tumors.

J.I. McDonald reports grants from NCI during the conduct of the study. S. Gomez reports grants from NCI during the conduct of the study. P.L. Strissel reports grants and personal fees from NIH during the conduct of the study. R. Strick reports grants from NIH-R21 during the conduct of the study. K.B. Chiappinelli reports personal fees from Rome Therapeutics outside the submitted work. No disclosures were reported by the other authors.

J.I. McDonald: Conceptualization, data curation, software, formal analysis, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. N. Diab: Data curation, formal analysis, validation, investigation. E. Arthofer: Data curation, formal analysis, validation, investigation. M. Hadley: Data curation, software, formal analysis, validation, investigation, writing–original draft. T. Kanholm: Data curation, software, formal analysis, validation, investigation, writing–original draft. U. Rentia: Data curation, software, formal analysis, validation, investigation, visualization, writing–original draft. S. Gomez: Data curation, software, formal analysis, validation, investigation. A. Yu: Data curation, validation, investigation. E.E. Grundy: Data curation, software, formal analysis, validation, investigation, methodology, writing–original draft. O. Cox: Formal analysis, investigation. M.J. Topper: Conceptualization, resources, investigation, methodology. X. Xing: Conceptualization, resources, investigation. P.L. Strissel: Conceptualization, resources, software, writing–original draft. R. Strick: Conceptualization, resources. T. Wang: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S.B. Baylin: Conceptualization, resources, data curation, software, formal analysis, visualization. K.B. Chiappinelli: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The authors acknowledge the Wang Lab for their assistance in training on methylMnM and methylCRF and their kind provision of scripts to update the authors' annotation files to hg38. They acknowledge Dr. Keith Crandall and Dr. Matthew Bendall for their assistance using the Telescope pipeline and the GW Pegasus Performance Computing core that manages the server on which they can store data and run analyses. The authors also acknowledge Evan Tarbell and Dr. Tao Liu for granting the authors early access to their tool HMMRATAC and for their aid as the authors established the pipeline in their lab. The ID8 Trp53−/− cells and ID8 CRISPR control cells were a kind gift from Dr. Iain McNeish, Imperial College London. The authors thank Castle Raley and the George Washington University Genomics Core for providing library construction and sequencing services. Research reported in this publication was supported by the NCI under Awards R00CA204592 (to K.B. Chiappinelli), R21CA227259 (to K.B. Chiappinelli with R. Strick as MPI), and R37CA251270 (to K.B. Chiappinelli) as well as by The Marlene and Michael Berman Endowed Fund for Ovarian Cancer Research. J.I. McDonald was supported by an NCI NRSA Institutional Research Training Program grant (T32 CA 247756). S. Gomez was supported by an NRSA Predoctoral Fellowship (NIH/NCI 1F31CA254315-01). The authors acknowledge the Institute for Biomedical Sciences at the George Washington University for graduate student support and training.

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.

1.
Kassiotis
G
,
Stoye
JP
.
Immune responses to endogenous retroelements: taking the bad with the good
.
Nat Rev Immunol
2016
;
16
:
207
.
2.
Ahmad
S
,
Mu
X
,
Yang
F
,
Greenwald
E
,
Park
JW
,
Jacob
E
, et al
.
Breaching self-tolerance to alu duplex RNA underlies MDA5-mediated inflammation
.
Cell
2018
;
172
:
797
810
.
3.
Cañadas
I
,
Thummalapalli
R
,
Kim
JW
,
Kitajima
S
,
Jenkins
RW
,
Christensen
CL
, et al
.
Tumor innate immunity primed by specific interferon-stimulated endogenous retroviruses
.
Nat Med
2018
;
24
:
1143
50
.
4.
Chiappinelli
KB
,
Strissel
PL
,
Desrichard
A
,
Li
H
,
Henke
C
,
Akman
B
, et al
.
Inhibiting DNA methylation causes an interferon response in cancer via dsRNA including endogenous retroviruses
.
Cell
2015
;
162
:
974
86
.
5.
Chung
H
,
Calis
JJA
,
Wu
X
,
Sun
T
,
Yu
Y
,
Sarbanes
SL
, et al
.
Human ADAR1 prevents endogenous RNA from triggering translational shutdown
.
Cell
2018
;
172
:
811
24
.
6.
Kong
Y
,
Rose
CM
,
Cass
AA
,
Williams
AG
,
Darwish
M
,
Lianoglou
S
, et al
.
Transposable element expression in tumors is associated with immune infiltration and increased antigenicity
.
Nat Commun
2019
;
10
:
1
14
.
7.
Jang
HS
,
Shah
NM
,
Du
AY
,
Dailey
ZZ
,
Pehrsson
EC
,
Godoy
PM
, et al
.
Transposable elements drive widespread expression of oncogenes in human cancers
.
Nat Genet
2019
;
51
:
611
.
8.
Goodier
JL
.
Restricting retrotransposons: a review
.
Mob DNA
2016
;
7
:
16
.
9.
Jones
PA
,
Issa
J-PJ
,
Baylin
S
.
Targeting the cancer epigenome for therapy
.
Nat Rev Genet
2016
;
17
:
630
41
.
10.
Kaminskas
E
,
Farrell
A
,
Abraham
S
,
Baird
A
,
Hsieh
L-S
,
Lee
S-L
, et al
.
Approval summary: azacitidine for treatment of myelodysplastic syndrome subtypes
.
Clin Cancer Res
2005
;
11
:
3604
8
.
11.
Center for Drug Evaluation and Research
.
FDA approves Onureg (azacitidine tablets) for acute myeloid leukemia
.
Silver Spring, MD
:
FDA
;
2020
.
Available from
: https://www.fda.gov/drugs/resources-information-approved-drugs/fda-approves-onureg-azacitidine-tablets-acute-myeloid-leukemia.
12.
Li
H
,
Chiappinelli
KB
,
Guzzetta
AA
,
Easwaran
H
,
Yen
R-WC
,
Vatapalli
R
, et al
.
Immune regulation by low doses of the DNA methyltransferase inhibitor 5-azacitidine in common human epithelial cancers
.
Oncotarget
2014
;
5
:
587
98
.
13.
Karpf
AR
,
Peterson
PW
,
Rawlins
JT
,
Dalley
BK
,
Yang
Q
,
Albertsen
H
, et al
.
Inhibition of DNA methyltransferase stimulates the expression of signal transducer and activator of transcription 1, 2, and 3 genes in colon tumor cells
.
Proc Natl Acad Sci U S A
1999
;
96
:
14007
12
.
14.
Karpf
AR
,
Lasek
AW
,
Ririe
TO
,
Hanks
AN
,
Grossman
D
,
Jones
DA
.
Limited gene activation in tumor and normal epithelial cells treated with the DNA methyltransferase inhibitor 5-Aza-2′-deoxycytidine
.
Mol Pharmacol
2004
;
65
:
18
27
.
15.
Roulois
D
,
Yau
HL
,
Singhania
R
,
Wang
Y
,
Danesh
A
,
Shen
SY
, et al
.
DNA-demethylating agents target colorectal cancer cells by inducing viral mimicry by endogenous transcripts
.
Cell
2015
;
162
:
961
73
.
16.
Siebenkäs
C
,
Chiappinelli
KB
,
Guzzetta
AA
,
Sharma
A
,
Jeschke
J
,
Vatapalli
R
, et al
.
Inhibiting DNA methylation activates cancer testis antigens and expression of the antigen processing and presentation machinery in colon and ovarian cancer cells
.
PLoS One
2017
;
12
:
e0179501
.
17.
Mehdipour
P
,
Marhon
SA
,
Ettayebi
I
,
Chakravarthy
A
,
Hosseini
A
,
Wang
Y
, et al
.
Epigenetic therapy induces transcription of inverted SINEs and ADAR1 dependency
.
Nature
2020
;
588
:
169
73
.
18.
Brocks
D
,
Schmidt
CR
,
Daskalakis
M
,
Jang
HS
,
Shah
NM
,
Li
D
, et al
.
DNMT and HDAC inhibitors induce cryptic transcription start sites encoded in long terminal repeats
.
Nat Genet
2017
;
49
:
1052
60
.
19.
Liu
M
,
Thomas
SL
,
DeWitt
AK
,
Zhou
W
,
Madaj
ZB
,
Ohtani
H
, et al
.
Dual inhibition of DNA and histone methyltransferases increases viral mimicry in ovarian cancer cells
.
Cancer Res
2018
;
78
:
5754
66
.
20.
Walton
J
,
Blagih
J
,
Ennis
D
,
Leung
E
,
Dowson
S
,
Farquharson
M
, et al
.
CRISPR/Cas9-mediated Trp53 and Brca2 knockout to generate improved murine models of ovarian high-grade serous carcinoma
.
Cancer Res
2016
;
76
:
6118
29
.
21.
Stone
ML
,
Chiappinelli
KB
,
Li
H
,
Murphy
LM
,
Travers
ME
,
Topper
MJ
, et al
.
Epigenetic therapy activates type I interferon signaling in murine ovarian cancer to reduce immunosuppression and tumor burden
.
Proc Natl Acad Sci U S A
2017
;
114
:
E10981
90
.
22.
Wang
T
,
Zeng
J
,
Lowe
CB
,
Sellers
RG
,
Salama
SR
,
Yang
M
, et al
.
Species-specific endogenous retroviruses shape the transcriptional network of the human tumor suppressor protein p53
.
Proc Natl Acad Sci U S A
2007
;
104
:
18613
8
.
23.
Wylie
A
,
Jones
AE
,
D'Brot
A
,
Lu
W-J
,
Kurtz
P
,
Moran
JV
, et al
.
p53 genes function to restrain mobile elements
.
Genes Dev
2016
;
30
:
64
77
.
24.
Lu
W-J
,
Chapo
J
,
Roig
I
,
Abrams
JM
.
Meiotic recombination provokes functional activation of the p53 regulatory network
.
Science
2010
;
328
:
1278
81
.
25.
Tiwari
B
,
Jones
AE
,
Caillet
CJ
,
Das
S
,
Royer
SK
,
Abrams
JM
.
p53 directly represses human LINE1 transposons
.
Genes Dev
2020
;
34
:
1439
51
.
26.
Baugh
EH
,
Ke
H
,
Levine
AJ
,
Bonneau
RA
,
Chan
CS
.
Why are there hotspot mutations in the TP53 gene in human cancers?
Cell Death Differ
2018
;
25
:
154
60
.
27.
The Cancer Genome Atlas
.
Integrated genomic analyses of ovarian carcinoma
.
Nature
2011
;
474
:
609
15
.
28.
Topper
MJ
,
Vaz
M
,
Chiappinelli
KB
,
Shields
CED
,
Niknafs
N
,
Yen
R-WC
, et al
.
Epigenetic therapy ties MYC depletion to reversing immune evasion and treating lung cancer
.
Cell
2017
;
171
:
1284
300
.
29.
Pinello
L
,
Canver
MC
,
Hoban
MD
,
Orkin
SH
,
Kohn
DB
,
Bauer
DE
, et al
.
Analyzing CRISPR genome editing experiments with CRISPResso
.
Nat Biotechnol
2016
;
34
:
695
7
.
30.
Corces
MR
,
Trevino
AE
,
Hamilton
EG
,
Greenside
PG
,
Sinnott-Armstrong
NA
,
Vesuna
S
, et al
.
An improved ATAC-seq protocol reduces background and enables interrogation of frozen tissues
.
Nat Methods
2017
;
14
:
959
62
.
31.
Xing
X
,
Zhang
B
,
Li
D
,
Wang
T
.
Comprehensive whole DNA methylome analysis by integrating MeDIP-seq and MRE-seq
.
DNA Methylation Protoc Methods Mol Biol
2018
;
1708
:
209
46
.
32.
Smit
A
,
Hubley
R
,
Green
P
.
RepeatMasker Open-4.0. 2013–2015 [Internet]
.
Available from
: http://www.repeatmasker.org.
33.
Jin
Y
,
Tam
OH
,
Paniagua
E
,
Hammell
M
.
TEtranscripts: a package for including transposable elements in differential expression analysis of RNA-seq datasets
.
Bioinformatics
2015
;
31
:
3593
9
.
34.
Bendall
ML
,
de Mulder
M
,
Iñiguez
LP
,
Lecanda-Sánchez
A
,
Pérez-Losada
M
,
Ostrowski
MA
, et al
.
Telescope: characterization of the retrotranscriptome by accurate estimation of transposable element expression
.
PLoS Comput Biol
2019
;
15
:
e1006453
.
35.
Tarbell
ED
,
Liu
T
.
HMMRATAC: a hidden Markov ModeleR for ATAC-seq
.
Nucleic Acids Res
2019
;
47
:
e91
.
36.
Love
MI
,
Huber
W
,
Anders
S
.
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
2014
;
15
:
550
.
37.
Wang
M
,
Zhao
Y
,
Zhang
B
.
Efficient test and visualization of multi-set intersections
.
Sci Rep
2015
;
5
:
16923
.
38.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
.
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
39.
Gel
B
,
Díez-Villanueva
A
,
Serra
E
,
Buschbeck
M
,
Peinado
MA
,
Malinverni
R
.
regioneR: an R/Bioconductor package for the association analysis of genomic regions based on permutation tests
.
Bioinformatics
2016
;
32
:
289
91
.
40.
Fang
F
,
Cardenas
H
,
Huang
H
,
Jiang
G
,
Perkins
SM
,
Zhang
C
, et al
.
Genomic and epigenomic signatures in ovarian cancer associated with resensitization to platinum drugs
.
Cancer Res
2018
;
78
:
631
44
.
41.
Moufarrij
S
,
Srivastava
A
,
Gomez
S
,
Hadley
M
,
Palmer
E
,
Austin
PT
, et al
.
Combining DNMT and HDAC6 inhibitors increases anti-tumor immune signaling and decreases tumor burden in ovarian cancer
.
Sci Rep
2020
;
10
:
1
12
.
42.
Steiner
MC
,
Marston
JL
,
Iñiguez
LP
,
Bendall
ML
,
Chiappinelli
KB
,
Nixon
DF
, et al
.
Locus-specific characterization of human endogenous retrovirus expression in prostate, breast, and colon cancers
.
Cancer Res
2021
;
81
:
1
12
.
43.
VanderKraats
ND
,
Hiken
JF
,
Decker
KF
,
Edwards
JR
.
Discovering high-resolution patterns of differential DNA methylation that correlate with gene expression changes
.
Nucleic Acids Res
2013
;
41
:
6816
27
.
44.
Chuang
JC
,
Warner
SL
,
Vollmer
D
,
Vankayalapati
H
,
Redkar
S
,
Bearss
DJ
, et al
.
S110, a 5-Aza-2′-deoxycytidine–containing dinucleotide, is an effective DNA methylation inhibitor in vivo and can reduce tumor growth
.
Mol Cancer Ther
2010
;
9
:
1443
50
.
45.
Levine
AJ
,
Ting
DT
,
Greenbaum
BD
.
P53 and the defenses against genome instability caused by transposons and repetitive elements
.
Bioessays
2016
;
38
:
508
13
.
46.
Vassilev
LT
,
Vu
BT
,
Graves
B
,
Carvajal
D
,
Podlaski
F
,
Filipovic
Z
, et al
.
In vivo activation of the p53 pathway by small-molecule antagonists of MDM2
.
Science
2004
;
303
:
844
8
.
47.
Aran
D
,
Hu
Z
,
Butte
AJ
.
xCell: digitally portraying the tissue cellular heterogeneity landscape
.
Genome Biol
2017
;
18
:
220
.
48.
Leroy
B
,
Girard
L
,
Hollestelle
A
,
Minna
JD
,
Gazdar
AF
,
Soussi
T
.
Analysis of TP53 mutation status in human cancer cell lines: a reassessment
.
Hum Mutat
2014
;
35
:
756
65
.
49.
Cameron
EE
,
Bachman
KE
,
Myöhänen
S
,
Herman
JG
.,
Baylin SB. Synergy of demethylation and histone deacetylase inhibition in the re-expression of genes silenced in cancer
.
Nat Genet
1999
;
21
:
103
7
.
50.
Vargiu
L
,
Rodriguez-Tomé
P
,
Sperber
GO
,
Cadeddu
M
,
Grandi
N
,
Blikstad
V
, et al
.
Classification and characterization of human endogenous retroviruses; mosaic forms are common
.
Retrovirology
2016
;
13
:
7
.
51.
Strissel
PL
,
Ruebner
M
,
Thiel
F
,
Wachter
D
,
Ekici
AB
,
Wolf
F
, et al
.
Reactivation of codogenic endogenous retroviral (ERV) envelope genes in human endometrial carcinoma and prestages: emergence of new molecular targets
.
Oncotarget
2012
;
3
:
1204
19
.
52.
Yan
W
,
Liu
S
,
Xu
E
,
Zhang
J
,
Zhang
Y
,
Chen
X
, et al
.
Histone deacetylase inhibitors suppress mutant p53 transcription via histone deacetylase 8
.
Oncogene
2013
;
32
:
599
609
.
53.
Muñoz-Fontela
C
,
Macip
S
,
Martínez-Sobrido
L
,
Brown
L
,
Ashour
J
,
García-Sastre
A
, et al
.
Transcriptional role of p53 in interferon-mediated antiviral immunity
.
J Exp Med
2008
;
205
:
1929
38
.
54.
Chuong
EB
,
Elde
NC
,
Feschotte
C
.
Regulatory evolution of innate immunity through co-option of endogenous retroviruses
.
Science
2016
;
351
:
1083
7
.
55.
Tsai
H-C
,
Li
H
,
Van Neste
L
,
Cai
Y
,
Robert
C
,
Rassool
FV
, et al
.
Transient low doses of DNA demethylating agents exert durable anti-tumor effects on hematological and epithelial tumor cells
.
Cancer Cell
2012
;
21
:
430
46
.
56.
Ardeljan
D
,
Steranka
JP
,
Liu
C
,
Li
Z
,
Taylor
MS
,
Payer
LM
, et al
.
Cell fitness screens reveal a conflict between LINE-1 retrotransposition and DNA replication
.
Nat Struct Mol Biol
2020
;
27
:
168
78
.
57.
Anglesio
MS
,
Wiegand
KC
,
Melnyk
N
,
Chow
C
,
Salamanca
C
,
Prentice
LM
, et al
.
Type-specific cell line models for type-specific ovarian cancer research
.
PLoS One
2013
;
8
:
e72162
.
58.
Domcke
S
,
Sinha
R
,
Levine
DA
,
Sander
C
,
Schultz
N
.
Evaluating cell lines as tumour models by comparison of genomic profiles
.
Nat Commun
2013
;
4
:
2126
.
59.
Vang
R
,
Levine
DA
,
Soslow
RA
,
Zaloudek
C
,
Shih
I-M
,
Kurman
RJ
.
Molecular alterations of TP53 are a defining feature of ovarian high-grade serous carcinoma: a rereview of cases lacking TP53 mutations in The Cancer Genome Atlas Ovarian Study
.
Int J Gynecol Pathol
2016
;
35
:
48
55
.
60.
Giacomo
AMD
,
Covre
A
,
Finotello
F
,
Rieder
D
,
Danielli
R
,
Sigalotti
L
, et al
.
Guadecitabine plus ipilimumab in unresectable melanoma: the NIBIT-M4 clinical trial
.
Clin Cancer Res
2019
;
25
:
7351
62
.
61.
Vadakekolathu
J
,
Lai
C
,
Reeder
S
,
Church
SE
,
Hood
T
,
Lourdusamy
A
, et al
.
TP53 abnormalities correlate with immune infiltration and associate with response to flotetuzumab immunotherapy in AML
.
Blood Adv
2020
;
4
:
5011
24
.
62.
Cherkasova
E
,
Scrivani
C
,
Doh
S
,
Weisman
Q
,
Takahashi
Y
,
Harashima
N
, et al
.
Detection of an immunogenic HERV-E envelope with selective expression in clear cell kidney cancer
.
Cancer Res
2016
;
76
:
2177
85
.
63.
Krishnamurthy
J
,
Rabinovich
BA
,
Mi
T
,
Switzer
KC
,
Olivares
S
,
Maiti
SN
, et al
.
Genetic engineering of T cells to target HERV-K, an ancient retrovirus on melanoma
.
Clin Cancer Res
2015
;
21
:
3241
51
.
64.
Saini
SK
,
Ørskov
AD
,
Bjerregaard
A-M
,
Unnikrishnan
A
,
Holmberg-Thydén
S
,
Borch
A
, et al
.
Human endogenous retroviruses form a reservoir of T cell targets in hematological cancers
.
Nat Commun
2020
;
11
:
5660
.
65.
Natoli
M
,
Gallon
J
,
Lu
H
,
Amgheib
A
,
Pinato
DJ
,
Mauri
FA
, et al
.
Transcriptional analysis of multiple ovarian cancer cohorts reveals prognostic and immunomodulatory consequences of ERV expression
.
J Immunother Cancer
2021
;
9
:
e001519
.

Supplementary data