Fig 2: (A) Heatmap shows differentially expressed microRNA (miRNAs) in VP- vs. VN-MCC cell lines from microarray data analysis (false discovery rate (FDR) <0.05). (B) miRNA pathway analysis using mirPath v.3 in the top differentially expressed miRNAs, showing enriched pathways (i.e., viral carcinogenesis and miRNAs involved in cancer). (C) Intersection of top upregulated miRNAs in VP-MCC with top miRNAs that could potentially target PRDM8 (scored using miRWalk) indicates miR-20a-5p as the only miRNA that can target PRDM8 and at the same time is downregulated in VN-MCCs. (D) Overexpressing miR-20a-5p using miRNA mimic decreases PRDM8 expression level significantly in MCC13 cells. (E) Immunoblot shows that PRDM8 protein levels are decreased upon miR-20a-5p overexpression. Ectopic overexpression of miR-20a-5p also led to the decrease in H3K9me3 levels. Histone H3 is used as the control. (F) Colony formation assay performed in MCC13 cells compared to miR-20a-5p-overexpressed MCC13 cells indicated lower clonogenicity capability upon overexpression of miR-20a-5p. Representative figures of colonies in MCC13 cells are shown. * p-value < 0.05.

Fig 3: (A) UpSetR plot showing intersection of transcription factors (TFs) with high binding affinity to PRDM8 from the chromatin immunoprecipitation sequencing (ChIP-seq) data of Encyclopedia of DNA Elements (ENCODE) database. Sample IDs including both fibroblasts and keratinocytes as cells of origin in MCC are indicated in lower part of the plot. Blue circles show samples with high binding affinity of TF-PRDM8 and white circles are samples with no TF-PRDM8 binding. Gray bars show the number of top-scored TFs that bind to PRDM8. Aggregate number of samples for each row is shown in the green horizontal bar next to each TF. (B) Integrative Gene Viewer (IGV) browser snippet of H3K4me3 ChIP-seq track of the MCC cells, showing promoter binding site around the transcriptional start site (TSS) of early growth response protein 1 (EGR1). Chromosome ideogram and coordinates are shown in the upper panel. Binding site DNA sequence was split into ˜400bp sites (S1-S10) for further enrichment analysis using primers specific to each site. (C) PRDM8-depleted and control cells (MCC13_sgPRDM8 and MCC13_CTRL) underwent PRDM8 chromatin immunoprecipitation followed by qPCR (ChIP-qPCR) for enrichment at each binding site. PRDM8-depleted cells showed significantly lower binding enrichment in bins S3 to S8 compared to control cells. (D) Plotting ChIP peak signals using sequential ChIP-qPCR data from EGR1 binding site (S1-S10). Signal intensity was normalized to 1 in all the PRDM8-depleted bins. Projected peak height for control cells was calculated for every bin and smoothed spline fit was plotted. This plot showed significant decrease in normalized binding enrichment in PRDM8-depleted cells compared to the control cells. (E) IGV snapshot of the PRDM8 ChIP-seq performed using KO and control MCC-13 cells (two replicates) illustrating PRDM8 binding at EGR1 region. (F) EGR1 and PRDM8 expression levels in VN- vs. VP-MCC indicated higher levels of EGR1 and PRDM8 co-expression in VN-MCC compared to VP-MCC cells.

Fig 4: (A) Analysis of immunohistochemistry (IHC) performed in VN- and VP-MCC patient tumors, indicating higher overall score in VN-MCC compared to VP-MCC (p < 0.05). (B) Representative PRDM8 IHC in a VN-MCC vs. a VP-MCC tumor (20× magnification). (C) Knocking down PRDM8 using CRISPR-Cas9 in two VN-MCC cell lines resulted in lower expression levels of this gene in both cell lines. (D) Image of the agarose gel from the PCR experiment targeting PRDM8, confirming depletion of this gene in the cell lines. (E) Western blot performed on PRDM8-KO and control VN-MCC cell lines. ß-actin was used as an internal positive control. (F) Cell proliferation experiment performed in two VN-MCC cell lines, indicating a decrease in proliferation rate upon PRDM8 knockdown. Asterisks show p-value < 0.05. (G) Clonogenicity assay performed in VN-MCC cells, indicating a lower number of colonies in the PRDM8-depleted cells. Asterisks show p-value < 0.05. Representative figures of MCC13 cells are shown.

Fig 5: (A) H3K9 immunohistochemistry performed in a stepwise manner (H3K9me1, H3K9me2, and H3K9me3) comparing global expression of these repressive histone marks in VN-MCC and VP-MCC tumors. Significantly higher expression levels of methylated H3K9 were observed in VN-MCC tumors. Asterisks show p-value < 0.05. (B) Representative methylated H3K9 (me1, me2, and me3) IHC in a VN-MCC and a VP-MCC tumor (20× magnification). (C) Immunocytochemistry (ICC) performed on MCC-13 and MCC-26 cell lines and their corresponding PRDM8-KO cells, indicating significantly lower levels of H3K9me3 upon PRDM8 depletion (* p < 0.05).