Epigenetic signatures differentiate uterine and soft tissue leiomyosarcoma

Abstract

Leiomyosarcomas (LMS) are diverse, rare, and aggressive mesenchymal soft tissue sarcomas. Epigenetic alterations influence multiple aspects of cancer, however epigenetic profiling of LMS has been limited. The goal of this study was to delineate the molecular landscape of LMS for subtype-specific differences (uterine LMS (ULMS) vs soft tissue LMS (STLMS)) based on integrated analysis of DNA methylation and gene expression to identify potential targets for therapeutic intervention and diagnosis. We identified differentially methylated and differentially expressed genes associated with ULMS and STLMS using DNA methylation and RNA-seq data from primary tumors. Two main clusters were identified through unsupervised hierarchical clustering: ULMS-enriched cluster and STLMS-enriched cluster. The integrated analysis demonstrated 34 genes associated with hypermethylation of the promoter CpG islands and downregulation of gene expression in ULMS or STLMS. In summary, these results indicate that differential DNA methylation and gene expression patterns are associated with ULMS and STLMS. Further studies are needed to delineate the contribution of epigenetic regulation to LMS subtype-specific gene expression and determine the roles of the differentially methylated and differentially expressed genes as potential therapeutic targets or biomarkers.

Keywords: DNA methylation; epigenetics; gene expression; leiomyosarcoma; uterine leiomyosarcoma.

Figure 1. Overview of the analysis.

Leiomyosarcoma (LMS) samples from the TCGA-SARC dataset were compared to identify the differentially methylated regions (DMRs), differentially methylated genes (DMGs) and differentially expressed genes (DEGs) in ULMS and STLMS.

Figure 2. DNA methylation landscape of LMS.

(A) Principal Component Analysis of STLMS (orange) and ULMS (blue) samples based on DMRs (n = 8,502). x- and y-axis show the principal components (PCs). Numbers in brackets indicate the percentage variance for each PC. (B) Unsupervised hierarchical clustering analysis of DMRs in ULMS and STLMS. Samples are presented in columns and the DMRs (n = 8,502) are presented in rows. Variables were sorted by hierarchical clustering on both the x-axis and the y-axis. The heatmap scale for methylation is based on β values (ranging from 0 (unmethylated) to 1 (methylated)). (A, B) Analysis was performed using the following filters in Qlucore: 5% variance, q-value < 0.05 (t-test) and β difference (Δβ (β^ULMS − β^STLMS) > |0.2|). (C) Distribution of DMRs based on methylation status. (D) Distribution of DMRs in different genomic regions relative to CpG islands including CpG islands, CpG shores and CpG shelves. (E) Distribution of DMRs relative to gene region features (TSS1500, TSS200, 5’UTR, 1st exon, gene body, and 3’UTR). Probes binding to multiple different gene regions and intergenic probes are not included in this graph. Please see Supplementary Table 2 for detailed distribution of the DMRs. (F) Venn diagram of the DMRs (n = 8,502), DMRs mapped to the island region (n = 1,449) and DMRs mapped to TSS200-TSS1500 regions (including probes binding to multiple regions that include TSS 200 and/or TSS 1500) (n = 1,454)). (G) Top 10 canonical pathways associated with ULMS-hypermethylated DMGs (n = 77). (H) Top 10 canonical pathways associated with STLMS-hypermethylated DMGs (n = 73). (G, H) Analysis was performed in Ingenuity Pathway Analysis using the following cutoffs: −log (p-value) >1.3. A detailed list of the canonical pathways is shown in Supplementary Table 4 and Supplementary Table 6.

Figure 3. Profiling of differentially expressed genes in LMS.

(A) Volcano plot of the global transcriptional changes in ULMS relative to STLMS. Each circle represents one gene. Statistical cut-off values are indicated by the grey lines: p ≤ 0.01 (corresponding to q-value < 0.05) and log2 fold change (FC) > |1.5|. The colored circles are the DEGs that pass the statistical filtering step (n = 2,196) (red-higher expression in ULMS (lower expression in STLMS), green-lower expression in ULMS (higher expression in STLMS)). y-axis indicates the minus log10 of p-value for gene, and x-axis shows the log2 FC between ULMS and STLMS. (B) Principal Component Analysis of STLMS (orange) and ULMS (blue) samples. x- and y-axis show the PCs. Numbers in brackets indicate the percentage variance for each PC. (C) Unsupervised heatmap representing color-coded expression levels of DEGs in ULMS relative to STLMS. Variables were sorted by hierarchical clustering on both the x-axis and the y-axis. The heatmap colors are based on gene expression, with red being upregulated and green being downregulated. Analysis in B–C was performed using the following filters in Qlucore: 5% variance, q-value < 0.05 (t-test), log2 FC > |1.5|. (D) Top 10 canonical pathways associated with STLMS-downregulated DEGs (n = 843). (E) Top 10 canonical pathways associated with ULMS-downregulated DEGs (n = 1,343). (G, H) Analysis was performed in Ingenuity Pathway Analysis using the following cutoffs: q-value < 0.05, log2 FC > |1.5|, −log (p-value) > 1.3. A detailed list of the canonical pathways is shown in Supplementary Table 8 and Supplementary Table 9.

Figure 4. Integrated analysis of differentially methylated and expressed genes in ULMS and STLMS.

(A) Summary of DMRs mapped to CpG island regions and TSS200/TSS1500 regions (q-value < 0.05, Δβ > |0.2|). (B) Summary of DEGs corresponding to DMGs in panel A (q < 0.05, log2 FC > |1.5|). (C) Genes grouped according to the methylation and expression changes in ULMS and STLMS. (D) Supervised clustering of hypermethylated and downregulated genes in STLMS. (E) Supervised clustering of hypermethylated and downregulated genes in ULMS.

Figure 5. Correlation between gene expression and DNA methylation.

RNA expression and DNA methylation values for the genes with moderate correlation ((|0.5| < r < |0.7|), p < 0.01) for ULMS-Hypermethylated-Downregulated (A) and STLMS-Hypermethylated-Downregulated group (B). Shown are the top 3 genes in each group. RNA expression values are plotted as log2 RSEM and DNA methylation are plotted as β values. Detailed analysis results are shown in Supplementary Table 10.

Figure 6. Network analysis of the DM-DEGs associated with STLMS-Hypermethylated-Downregulated and ULMS-Hypermethylated-Downregulated groups.

Ingenuity network analysis was used to plot the gene relationships. The colored shapes indicate the downregulated genes in the STLMS-Hypermethylated-Downregulated group (orange color) (A, B) and the ULMS-Hypermethylated-Downregulated groups (blue color) (C). Genes that do not have corresponding colors, were not identified as differentially expressed in our analysis, and were integrated based on the Ingenuity Pathway Analysis evidence indicating a relevance to this network.