br Fig Low SUV H
Fig. 2. Low SUV39H1 Oxaliplatin correlates with features of enhanced migration in human cervical cancers. a. Survival in cervical SCC patients with low v/s high tumoural SUV39H1 (median group split, n = 94 for each group). Quartile split approach yielded similar results (Fig. S2a). b, c. Survival in patients with low v/s high tumoural SUV39H1 in Skin Cutaneous Melanoma (SKCM) and Kidney Renal Clear Cell Carcinoma (KIRC) datasets respectively (quartile split). d. Volcano plot depicting differential gene expression in SUV39H1-low tumours v/s SUV39H1-high tumours (n = 47 for each group). Blue points indicate genes significant at cutoff fold change (1.5 fold) and FDR-corrected q-value (0.05). e. Gene set enrichment analysis (GSEA) plot examining enrichment for a cell migration gene set (M6281, from the Msigdb) in SUV39H1 low tumours (p-value = 0.0163). f. Enriched KEGG pathway terms corresponding to genes upregulated in SUV39H1-low tumours. Arrows indicate adhesion and migration associated terms. Results shown in this figure are based on data obtained from The Cancer Genome Atlas Project (TCGA) Research Network (http://cancergenome.nih.gov/).
overall survival over time, compared to the SUV39H1-high group (p = 0.012) (Fig. 2a, Fig. S2a). We also examined whether SUV39H1 expression correlated with survival in other human cancers. Interest-ingly, among 21 TCGA cancers examined, cervical cancer was the only cancer in which low SUV39H1 correlated with poor survival (Table S1, Fig. 2a-c, Fig. S2b, c). 3 other cancers (Cutaneous melanoma, Renal clear cell carcinoma, and Renal papillary cell carcinoma) showed sig-nificant correlations (p < 0.05) of SUV39H1 expression with survival, but here high SUV39H1 expression correlated with poor survival in-stead. Correlations with survival in SCC cases were not observed with SUV39H2, and HP1a homolog CBX5, which are known to synergise
with SUV39H1 in some contexts to regulate H3K9me3 (Fig. S2d, e). Thus, low tumoural SUV39H1 abundance correlates with poor clinical outcomes in cervical cancers, and this correlation may be unique to cervical cancers.
2.3. Low SUV39H1 expression correlates with features of enhanced migration in cervical cancers
Poor clinical outcomes associated with SUV39H1-low cervical tu-mours could be due to various aspects of disease progression, such as metastasis, recurrence, and drug resistance. As CD66high populations
C. Rodrigues, et al.
also show SUV39H1 depletion, along with enhanced migration and associations with metastasis, we sought to examine whether SUV39H1-low tumours showed features of enhanced migration and invasion. We examined genes differentially expressed in cervical SUV39H1-low tu-mours. 1710 genes were enriched and 721 genes were depleted in SUV39H1-low tumours (Table S2). These SUV39H1-low tumours showed an enrichment (> 1.5 fold) of several migration and matrix remodelling associated genes, including TWIST1, ZEB1 and several MMPs (Fig. 2d). Gene Set Enrichment Analysis (GSEA) indicated an enrichment for a cell migration gene set in SUV39H1-low tumours (p-value=0.0163) (Fig. 2e). Gene Ontology (GO) and KEGG pathway analysis revealed a significant (p < 0.01) enrichment for terms in-cluding cell migration, vasculature development, TGF-β signalling, and cell adhesion in SUV39H1-low tumours (Fig. 2f, Fig. S3a). The SUV39H1-low group also showed a significant (p < 0.01) depletion of DNA replication and cell cycle signatures (Fig. 2d, Fig. S3b). Finally, we
noted that genes enriched in SUV39H1-low tumours show a statistically significant overlap with a CD66high cell profile  (Fig. S3c). Thus, low
SUV39H1 abundance in cervical tumours correlates with poor clinical outcomes, and SUV39H1-low tumours show gene expression signatures of cell migration.