Reciprocal epigenetic remodeling controls testicular cancer hypersensitivity to hypomethylating agents and chemotherapy

Abstract

Testicular germ cell tumors (TGCTs) are aggressive but sensitive to cisplatin-based chemotherapy. Alternative therapies are needed for tumors refractory to cisplatin with hypomethylating agents providing one possibility. The mechanisms of cisplatin hypersensitivity and resistance in TGCTs remain poorly understood. Recently, it has been shown that TGCTs, even those resistant to cisplatin, are hypersensitive to very low doses of hypomethylating agents including 5-aza deoxy-cytosine (5-aza) and guadecitabine. We undertook a pharmacogenomic approach in order to better understand mechanisms of TGCT hypomethylating agent hypersensitivity by generating a panel of acquired 5-aza-resistant TGCT cells and contrasting these to previously generated acquired isogenic cisplatin-resistant cells from the same parent. Interestingly, there was a reciprocal relationship between cisplatin and 5-aza sensitivity, with cisplatin resistance associated with increased sensitivity to 5-aza and 5-aza resistance associated with increased sensitivity to cisplatin. Unbiased transcriptome analysis revealed 5-aza-resistant cells strongly downregulated polycomb target gene expression, the exact opposite of the finding for cisplatin-resistant cells, which upregulated polycomb target genes. This was associated with a dramatic increase in H3K27me3 and decrease in DNMT3B levels in 5-aza-resistant cells, the exact opposite changes seen in cisplatin-resistant cells. Evidence is presented that reciprocal regulation of polycomb and DNMT3B may be initiated by changes in DNMT3B levels as DNMT3B knockdown alone in parental cells resulted in increased expression of H3K27me3, EZH2, and BMI1, conferred 5-aza resistance and cisplatin sensitization, and mediated genome-wide repression of polycomb target gene expression. Finally, genome-wide analysis revealed that 5-aza-resistant, cisplatin-resistant, and DNMT3B-knockdown cells alter the expression of a common set of polycomb target genes. This study highlights that reciprocal epigenetic changes mediated by DNMT3B and polycomb may be a key driver of the unique cisplatin and 5-aza hypersensitivity of TGCTs and suggests that distinct epigenetic vulnerabilities may exist for pharmacological targeting of TGCTs.

Keywords: 5-aza deoxycytidine; DNA methylation; H3K27me3; cisplatin; epigenetics; polycomb repressive complex.

Fig. 1

Derivation of 5‐aza‐resistant testicular cancer cells reveals reciprocal drug sensitivities between 5‐aza‐ and cisplatin‐resistant cells. (A) Human testicular cancer derived EC cell line 2102EP was exposed to stepwise dosages of 5‐aza starting at 1 nm and then allowed to recover for 1–2 weeks between doses. After selection in 100 nm 5‐aza, cells were cloned. (B) Schematic of 5‐aza and cisplatin treatment protocols for cell viability assays. CTG, CellTiter‐Glo. (C) 5‐aza‐selected cell lines are stably resistant to 5‐aza. Cell survival and viability were measured using CellTiter‐Glo in parental or acquired 5‐aza‐resistant TGCT cells treated with indicated 5‐aza doses for 3 days. (D) 5‐aza IC50 values for parental and 5‐aza‐resistant cells were estimated from a best‐fit dose–response model. (E) Cisplatin‐resistant cell lines 2102EP‐B3, 2102EP‐C1, and 2102EP‐C4 have increased sensitivity to 5‐aza compared to parental cells. Cell survival and viability were measured using CellTiter‐Glo in parental, 5‐aza‐resistant, and cisplatin‐resistant TGCT cells treated with indicated 5‐aza doses for 3 days. (F) 5‐aza‐resistant cell lines 2102EP‐AH2A5, 2102EPAH2‐B9, and 2102EP‐AH2A2 have increased sensitivity to cisplatin compared to parental cells. Cell survival and viability were measured using CellTiter‐Glo in parental, 5‐aza‐resistant, and cisplatin‐resistant TGCT cells treated with cisplatin for 6 h and assayed 24 h later. (G) 5‐aza IC50 values for parental, 5‐aza‐resistant, and cisplatin‐resistant 2102EP cells. (H) Cisplatin IC50 values for parental, 5‐aza‐resistant, and cisplatin‐resistant 2102EP cells. (I) 5‐aza‐resistant and cisplatin‐resistant 2102EP cells have similar basal doubling times compared to parental cells. All data represent mean ± standard error of the mean of four biological determinations. Two‐tailed Student's t‐tests were performed for statistical analysis. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.005, ****P ≤ 0.001 compared to parental cells. All experiments were repeated at least twice with similar results.

Fig. 2

5‐aza resistance in testicular cancer cells is associated with decreased expression of polycomb target genes. (A) Volcano plots indicating the number of significantly upregulated (blue) and downregulated (red) genes in 5‐aza‐resistant 2102EP‐AHA2A, 2102EP‐AH2B9, and 2102EP‐AH1A5 cells compared to parental 2102EP cells. Number of genes upregulated and downregulated with a 1.5‐fold cutoff and FDR ≤ 0.01 for each resistant line is indicated. Volcano plots are based on RNA‐seq analysis from biological triplicates. (B) Downregulated genes from 5‐aza‐resistant cells are enriched for gene sets associated with H3K27 methylation and PRC2 targets. GSEA indicating all polycomb‐related negatively enriched gene sets within the top 50 ranked by NES from 5529 curated gene gets from the MSigDB C2 collection in 5‐aza‐resistant cells compared to respective parental cells. (C) Representative gene set enrichment plots. NES, normalized enrichment score.

Fig. 3

5‐aza‐resistant and cisplatin‐resistant testicular cancer cells have reciprocally altered levels of H3K27me3 and DNMT3B. (A) Immunoblot analysis of indicated cell lines (2102EP parent is loaded twice) with antibodies recognizing DNMT3B, DNMT1, H3K27me3, H2AUbK119, and actin. (B–D) Real‐time PCR analysis of mRNA expression of DNMT1, DNMT3A, DNMT3B, BMI1, EZH2, KDM6A, and KDM6B. Data are the mean of triplicate determinations, and error bars are standard error of the mean. Two‐tailed Student's t‐tests were performed for statistical analysis. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.005, ****P ≤ 0.001. Each experiment was repeated at least twice with similar results.

Fig. 4

DNMT3B‐knockdown results in induction of H3K27me3, BMI1 and EZH2, and confers 5‐aza resistance and cisplatin sensitization to TGCT cells. (A) Real‐time PCR and immunoblot confirmation of stable shRNA knockdown of DNMT3B in 2102EP cells. (B) Knockdown of DNMT3B in 2102EP cells results in increased protein levels of BMI1, EZH2, and H3K27me3. (C) Real‐time PCR indicates an induction of BMI1 mRNA after DNMT3B knockdown in 2102EP cells. Data are the mean of triplicate determinations, and error bars are standard error of the mean. (D) DNMT3B knockdown in 2102EP cells confers 5‐aza sensitivity and cisplatin sensitization. Data represent mean ± standard error of the mean of four biological determinations. Two‐tailed Student's t‐tests were performed for statistical analysis. *P ≤ 0.05, ****P ≤ 0.001. Each experiment was repeated at least twice with similar results.

Fig. 5

Similar to 5‐aza‐resistant cells, DNMT3B knockdown in TGCT cells results in a genome‐wide decrease in polycomb target gene expression. (A) Volcano plot indicating the number of significantly upregulated (blue) and downregulated (red) genes in 2102EP‐DNMT3B knockdown cells compared to PLK0.1 control cells. Volcano plot of 5‐aza‐resistant 2102EP‐AH2 vs parental and cisplatin‐resistant 2102EP‐C1 cells vs parent is provided for comparison. Number of genes upregulated and downregulated with a 1.5‐fold cutoff and FDR ≤ 0.01 for each resistant line is indicated. Volcano plots are based on RNA‐seq analysis from biological triplicates. The DNMT3B gene is indicated by circle. (B) Downregulated genes from 2102EP‐DNMT3BKD cells are enriched for gene sets associated with H3K27 methylation and polycomb targets. GSEA indicating all polycomb related negatively enriched gene sets within the top 50 ranked by NES from 5529 curated gene gets from the MSigDB C2 collection in 2102EP‐DNMT3BKD cells compared to PLK0.1 control cells. (C) Representative gene set enrichment plots. NES, normalized enrichment score.

Fig. 6

Genome‐wide overlap analysis between 5‐aza‐resistant, cisplatin‐resistant, and DNMT3B knockdown TGCT cells reveals mutual enrichments in the same polycomb target genes. (A) Venn diagram indicating overlap in differentially expressed genes between genes downregulated in 5‐aza‐resistant 2102EP‐AH2A2 cells, upregulated in cisplatin‐resistant 2102EP‐C1 cells, and downregulated in 2102EP‐DNMT3B knockdown cells. Differentially expressed genes are defined as changed 1.5‐fold or greater and an FDR of ≤ 0.01 compared to parental cells or shRNA control cells, respectively. (B) Fisher exact tests against the 5529 curated gene sets of the Broad MSigDB C2 collection performed on the overlap genes between downregulated in 5‐aza‐resistant cells and upregulated in cisplatin‐resistant cells (green), downregulated in 5‐aza‐resistant cells and downregulated in DNMT3B‐KD cells (red), and upregulated in cisplatin‐resistant cells and downregulated in DNMT3B‐KD cells (blue). The top 10 gene sets for each comparison based on P value are presented.