Suppression of Enhancer Overactivation by a RACK7-Histone Demethylase Complex

Abstract

Regulation of enhancer activity is important for controlling gene expression programs. Here, we report that a biochemical complex containing a potential chromatin reader, RACK7, and the histone lysine 4 tri-methyl (H3K4me3)-specific demethylase KDM5C occupies many active enhancers, including almost all super-enhancers. Loss of RACK7 or KDM5C results in overactivation of enhancers, characterized by the deposition of H3K4me3 and H3K27Ac, together with increased transcription of eRNAs and nearby genes. Furthermore, loss of RACK7 or KDM5C leads to de-repression of S100A oncogenes and various cancer-related phenotypes. Our findings reveal a RACK7/KDM5C-regulated, dynamic interchange between histone H3K4me1 and H3K4me3 at active enhancers, representing an additional layer of regulation of enhancer activity. We propose that RACK7/KDM5C functions as an enhancer "brake" to ensure appropriate enhancer activity, which, when compromised, could contribute to tumorigenesis.

Figure 1. Identification of RACK7 and KDM5C interaction and their binding events at active enhancers and super-enhancers

(A) Schematic representation of the domain architecture of RACK7 protein. (B) Venn diagram analysis of ChIP-seq peaks of RACK7, H3K4me1 and H3K27Ac. P value by Pearson’s Chi-squared test. (C) Total numbers of super-enhancers and RACK7 bound super-enhancers in ZR-75-30, MCF-7 and mES cells. (D) Tandem affinity purified FLAG-HA-RACK7 (F.H.RACK7) protein complex was resolved and visualized by silver staining. (E) Reciprocal immunoprecipitation between endogenous RACK7 and KDM5C. (F) In vitro pull-down between recombinant RACK7 and KDM5C proteins. (G) Venn diagram analysis shows the overlap between RACK7 and KDM5C co-bound regions and active enhancers defined by H3K4me1 and H3K27Ac, P value by Pearson’s Chi-squared test. (H) UCSC tracks showing RACK7 and KDM5C ChIP-seq signals in parental ZR-75-30 and KDM5C ChIP-seq in RACK7 KO1 cells at a select genomic location. (I) KDM5C recruitment to chromatin in the parental and the RACK7 KO1 cells examined by the genome-wide analyses of ChIP-seq signals (J) KDM5C recruitment to chromatin in parental ZR-75-30 and the RACK7 KO cells confirmed by ChIP-qPCR. q-PCR Data are represented as mean ± SD from three biological replicates, * P< 0.05; ** P < 0.01, T test. See also Figure S1, Table S1 and S2

Figure 2. RACK7 and KDM5C suppress H3K4me3 at active enhancers and super-enhancers

(A) Heatmap analyses of ChIP-seq signals of RACK7, KDM5C and select histone modifications in the parental and RACK7 KO1 ZR-75-30 cells, ranked by RACK7 ChIP-seq signals in parental ZR-75-30 cells. All ChIP-seq signals are displayed from −10 kb to +10 kb surrounding the center of each annotated RACK7 peak. (B) Normalized levels of H3K4me3 and H3K4me1 at the RACK7 bound super-enhancers in the parental and the RACK7 KO1 ZR-75-30 cells. P values by ANOVA test. (C–D) H3K4 methylation states at three representative enhancers in parental ZR-75-30 and RACK7 KO cell lines, shown by ChIP-seq snapshots (C) and confirmed by ChIP-qPCR (D). (E–F) RACK7 binding and H3K4me3 level at three select enhancers examined by ChIP-qPCR in the parental, RACK7 KO1 and RACK7 KO1 containing a rescuing, wildtype RACK7 transgene. (G) H3K4me3 levels at three selected enhancers in parental ZR-75-30 and KDM5C KO cells. In all panels, q-PCR Data are represented as mean ± SD from three biological replicates. * P < 0.05; ** P < 0.01, T test. See also Figure S2, Table S1 and S2

Figure 3. RACK7 or KDM5C loss leads to an increase of eRNA production

(A) Heatmap analyses of H3K4me3 ChIP-seq and nascent RNA-seq data from the parental and RACK7 KO1 ZR-75-30 cells ranked by RACK7 ChIP-seq signals at all enhancers in the parental ZR-75-30 cells. Nascent RNA-seq signals are displayed from −2 kb to +2 kb and ChIP-seq signals are displayed from −10 kb to +10 kb surrounding the centers of the annotated RACK7 peaks. Sense and antisense stands are indicated by “+” and “−”. (B) Comparison of eRNA levels from RACK7 bound super-enhancers between the parental and RACK7 KO1 ZR-75-30 cells, as assessed by nascent RNA-seq. P value by ANOVA test. (C) Snapshots showing eRNA increases at all three select RACK7 bound enhancers in RACK7 KO1 cells. (D–E) Nascent RNA RT-qPCR confirmation of eRNA increase in the RACK7 KO (D) and KDM5C KO (E) cells from the same three select enhancers shown in 3C. All q-PCR data are represented as mean ± SD from three biological replicates. * P < 0.05; **P < 0.01, T test. See also Figure S2, Table S1 and S2

Figure 4. Loss of RACK7 or KDM5C leads to hyper-activated enhancers

(A) Expression comparison (using mRNA-seq data) of the adjacent genes to the RACK7 bound and unbound enhancers (Left: top 2,000; and right: bottom 2,000) between the parental and RACK7 KO1 ZR-75-30 cells. P values by ANOVA test. (B) Expression comparison (using mRNA-seq data) of the adjacent genes to the RACK7 bound super-enhancers between the parental and RACK7 KO1 ZR-75-30 cell lines. P value by ANOVA test. (C) Snapshots showing mRNA increases from the nearest genes of the three select RACK7 bound enhancers in RACK7 KO1 cells. (D–E) RT-qPCR confirmation of the expression changes of the three select target genes in the parental, RACK7 KO (C) and KDM5C KO (D) cells. Total mRNA samples were used. All q-PCR data are represented as mean ± SD from three biological replicates. * P < 0.05; ** P < 0.01, T test. See also Figure S2, S3 and Table S1, S2

Figure 5. Loss of RACK7 or KDM5C promotes tumorigenic potential of ZR-75-30 cells

(A–C) In vitro soft agar, invasion and migration assays examining the anchorage independent growth, invasion and migration abilities of the parental, RACK7 KO1 and the RACK7 KO1 cells with a rescuing, wildtype RACK7 transgene. (D) Xenograft growth (24 days, n=10) analysis of the parental, RACK7 KO1 and the RACK7 KO1 cells with a rescuing, wildtype RACK7 transgene. Quantifications of tumor volume (Upper) and representative images of the tumors (Lower) are shown. Error bars represent SEM of the mean, **P < 0.01, T test. (E–F) In vitro invasion and migration assays examining the invasion and migration abilities of the parental and KDM5C KO ZR-75-30 cells. (G) RACK7 expression is significantly lower in IDC than the paired DCIS tumors from 6 breast cancer patients (GDS2046) (Schuetz et al., 2006). In panel (A–C, E and F), all data are represented as mean ± SD from three biological replicates, ** P < 0.01, T test. See also Figure S4

Figure 6. RACK7 suppresses tumorigenesis in part through repressing the S100A family of oncogenes

(A) ChIP-seq profiles of RACK7, KDM5C, H3K4me1 and H3K27Ac in ZR-75-30, and nascent RNA-seq and mRNA-seq profiles in the parental and RACK7 KO1 cells at S100A oncogene cluster. S100A4 and S100A16 are selected as examples to show increased eRNA and mRNA expression in the RACK7 KO1 cells (Shadow). (B) S100A4 protein levels in the parental and the RACK7 KO1 cells examined by Western blotting. (C) RT-qPCR showing the expression of multiple S100A oncogenes in the RACK7 KO1 cells compared to the parental ZR-75-30 cells. (D) S100A4 expression is significantly higher in IDC than the paired DCIS tumors from the same dataset shown in Figure 5G (GDS2046). (E) S100A4 protein levels were examined by Western blotting in the RACK7 KO1 cells with indicated treatment. (F) The tumorigenic abilities of the RACK7 KO1 cells with the indicated treatments were examined and showed by the numbers of the invaded and migrated cells. In panel C, all data are represented as mean ± SD from three biological replicates, * P < 0.05 and ** P < 0.01, T test.