SETD5 expression was positively related to BC progression
Firstly, we analyzed the clinical pathological diagnosis of SETD5 expression in normal breast (n = 18) and BC tissues (n = 125) in the Clinical Proteomic Tumor Analysis Consortium (CPTAC) database through the UALCAN web portal. As shown in Fig. 1a, SETD5 protein expression significantly increased in BC tissues relative to normal tissues (p < 0.001). CPTAC data also showed that SETD5 was over-expressed in BC tissues with major subclass (Luminal vs Normal, p < 0.001; HER2 vs Normal, p < 0.01; TNBC vs Normal, p < 0.05), patient’s age (21–40 years vs Normal, p < 0.01; 41–60 years vs Normal, p < 0.001; 61–80 Years vs Normal, p < 0.001; 81–100 years vs Normal, p < 0.05), and individual cancer stages (stage 1 vs Normal, p < 0.001; stage 2 vs Normal, p < 0.01; stage 3 vs Normal, p < 0.001) (Fig. 1b-d). Post-progression survival (PPS) was considered as the median overall survival minus the median progression-free survival for each trial arm . Furthermore, the Kaplan-Meier Plotter online tool analysis showed that, SETD5 expression was positively related to a poor PPS in BC patients using all Affymetrix ID (Fig. 1e). These data suggest that overexpression of SETD5 in BC tissues could be used as a potential predictor of survival.
SETD5 is a positive regulator of BCSC
Our data provide evidence that SETD5 expression was predictive of poor prognosis and contributed to BC progression. Next, we explored the effect of SETD5 on the properties of CSCs. The protein levels of SETD5 in adherent monolayers and spheroid body-forming cells were determined by western blot analysis. SETD5 were obviously higher in BC spheroid body-forming cells than in adherent monolayer cells (Fig. 2a). The role of SETD5 in BC spheroid body-forming cells was then further investigated. Knockdown of SETD5 expression significantly decreased the number and size of spheroid body-forming cells of BC (Fig. 2b). Immunofluorescent (IF) analysis demonstrated that knockdown of SETD5 resulted in a significant reduction in LGR5 staining in BC spheroid body-forming cells (Fig. 2c).
To further explore how SETD5 acted as a crucial regulator of BCSC maintenance in vivo, subcutaneous injections of MDA-MB-231 cells dissociated from the spheroid body were administered in nude mice for 6 weeks. Then, we measured the xenograft tumor size and tumor volume per week by visual analysis (n = 4 tumors/group). Xenograft tumor growth indicated that inhibition of SETD5 expression could significantly reduce the weight and volume in vivo (Fig. 3a). Immunohistochemical (IHC) staining of CSC markers was performed in BC xenograft tumor samples. esi-SETD5 downregulated the IHC score of CSC markers, CD133, and LGR5 (Fig. 3b). Additionally, we used IF to examine the correlation between SETD5 expression and CSC markers. Figure 3c showed that SETD5 was co-located with CD133 or LGR5 in BC patient tissues. These findings support a positive functional role of SETD5 in the maintenance of BCSCs.
SETD5 activated BCSC glycolysis
To test the effect of SETD5 on the cellular metabolic reprogramming of BCSC, glucose uptake and lactate production of BC spheroid body-forming cells were assessed. SETD5 depletion inhibited glucose uptake and lactate production in BC spheroid body-forming cells (Fig. 4a, b). Consistently, blocking SETD5 expression significantly reduced glycolytic genes HK2 and PFKFB3 expression in BC spheroid body-forming cells, which was assessed using IF (Fig. 4c). A similar result was obtained in vivo by IHC examination (Fig. 4d). SETD5 depletion also downregulated HK2 and PFKFB3 protein expression in xenograft tumor samples. Thus, SETD5 may be a potential regulator of aerobic glycolysis in BCSCs.
SETD5 binding to EP300/HIF1α in BC cells
Hypoxia is a pervasive environmental stimulus of glucose metabolic reprogramming. Recently, Huang et al. reported that histone methyltransferases SETD8 contribute to HIF1α regulation and mediate the metabolism process . Thus, the central (hypoxic) and peripheral (normoxic) regions from xenograft tumor tissues were analyzed using hematoxylin-eosin staining (Fig. 5a). IHC staining results showed that SETD5-positive cells mostly gathered centrally with highly hypoxic cells as revealed by HIF1α, when compared with the peripheral regions (Fig. 5b). HK2 and PFKFB3 were highly expressed in the central positions compared with the peripheral positions (Fig. 5c). Gene Set Enrichment Analysis (GSEA) and used the GEO database to find an enriched expression of SETD5 gene in BC tissues. We found a positive correlation between RNA levels of SETD5 and the hypoxic gene (NES = 2.07, FDR q = 0.14, p < 0.001) (Fig. 5d), as well as positive regulatory effects of SETD5 on glycolysis (NES = 1.39, FDR q = 0.23, p = 0.04) (Fig. 5e). Thus, these results indicate that SETD5 activates BCSC glycolysis which is a hypoxia stimulus.
To explore the functional connection of SETD5-HIF1α and glycolysis, we constructed the PPI network using the STRING database. Nine of these genes comprised a PPI network (PPI enrichment p-value < 0.001) (Fig. 6a). Co-expression analysis revealed tight co-expression across the tissue types between SETD5 and EP300 genes (Fig. S1). Confocal microscopy IF experiments also measured co-expression of SETD5 and EP300 in BC tissues (Fig. 6b).
Furthermore, Tissue Microarray (TMA) analysis was performed for the association between SETD5 and EP300/HIF1α protein expression in 117 cases of human BC specimens by IHC staining (Fig. S2). In 90 cases of the EP300-positive group, 76 cases (76/90, 84.4%) were positive for the SETD5 gene (p = 0.008) (Fig. 6c). In 76 cases of the HIF1α-positive group, 63 cases (63/76, 82.9%) were positive for the SETD5 gene (p = 0.018) (Fig. 6c). There was a positive correlation between SETD5 expression and EP300 and HIF1α detected in BC samples by Gene Expression Profiling Interactive Analysis (GEPIA) (Fig. 6d). Importantly, knocking down SETD5 expression using SETD5 specific esiRNA dramatically decreased the expression of EP300 and HIF1α in BC cells (Fig. 6e). Therefore, SETD5 expression strongly correlated with EP300/HIF1α and worked as an upstream effector in BC cells.
SETD5 facilitates glycolysis via HIF1α in the nucleus
The above-mentioned results prompted us to test the hypothesis that SETD5 regulates HIF1α protein stability to activate glycolysis. Firstly, we exposed BC cells to CoCl2 (100, 200, and 400 µmol/L) for 6 and 12 hours each. CoCl2 significantly induced the expression of HIF1α protein in a time- and dose-dependent manner (Fig. S3). HIF1α protein expression was most effectively upregulated after 200 µmol/L of CoCl2-induced hypoxia treatment for 12 hours. This treatment condition was then used in the ensuing hypoxia mimetic experiments. To get insight into the molecular mechanism of SETD5 in contributing to glycolysis in BC cells, the subcellular localization of SETD5 and its putative target genes HIF1α involved in hypoxia-induced glycolysis were investigated using cytoplasmic and nuclear extract assays. The nuclear expression of HIF1α was stronger than the cytoplasmic expression after treating with CoCl2 (Fig. 7a). SETD5 expression levels increased under chemically-induced hypoxia than under normoxic conditions in the total cell lysate. However, there was no obvious change in SETD5 nuclear and cytoplasmic distribution (Fig. 7a). Immunoprecipitation confirmed that SETD5 could bind HIF1α in the nuclear extracts of BC cells under chemically-induced hypoxia conditions (Fig. 7b). Interestingly, IF staining showed that SETD5 knockdown reduced HIF1 protein expression in the nuclear extracts of BC cells (Fig. 7c). Glycolytic enzymes were further examined in the nucleus of BC cells transfected with esi-SETD5 under chemically-induced hypoxia conditions. Figure 7d showed that SETD5 silencing decreased the expression of HK2 and PFKFB3 proteins.