Both EZH2 and JMJD6 regulate cell cycle genes in breast cancer

Abstract

Background: Strong evidences support the critical role of Jumonji domain containing 6 (JMJD6) in progression of breast cancer. Here we explore potential partners that coregulate gene expression, to understand additional pathways that are activated by higher amounts of JMJD6.

Methods: We used Gene Set Enrichment Analysis (GSEA) data to identify factors that display gene expression similar to cells treated with JMJD6 siRNA. Using chromatin immunoprecipitations (ChIP) against genomic regions that bind JMJD6 identified by in house and public database Encyclopaedia of DNA Elements (ENCODE), we confirmed JMJD6 occupancy by ChIP PCR. We tested the association of co-regulated genes with patient prognosis using The Cancer Genome Atlas (TCGA) datasets.

Results: JMJD6 profiles overlapped with those of Enhancer of Zeste homolog 2 (EZH2) and together they appear to co-regulate a unique cassette of genes in both ER+ and ER- cells. 496 genes including aurora kinases, which are currently being tested as novel therapeutic targets in breast cancer were co-regulated in MDA MB 231 cells. JMJD6 and EZH2 neither inter-regulated nor physically interacted with one another. Since both proteins are chromatin modulators, we performed ChIP linked PCR analysis and show that JMJD6 bound in the neighbourhood of co-regulated genes, though EZH2 data did not show any peaks within 100 kb of these sites. Alignment of binding site sequences suggested that atleast two types of binding partners could offer their DNA binding properties to enrich JMJD6 at regulatory sites. In clinical samples, JMJD6 and EZH2 expression significantly correlated in both normal and tumor samples, however the strongest correlation was observed in triple-negative breast cancer (TNBC) subtype. Co-expression of JMJD6 and EZH2 imposed poorer prognosis in breast cancer.

Conclusions: JMJD6 and EZH2 regulate the same crucial cell cycle regulatory and therapeutic targets but their mechanisms appear to be independent of each other. Blocking of a single molecule may not axe cell proliferation completely and blocking both JMJD6 and EZH2 simultaneously may be more effective in breast cancer patients.

Keywords: DREAM; Microarray; Prognosis; TNBC.

Fig. 1

Expression Pattern of genes commonly regulated by JMJD6 and EZH2 siRNAs in TCGA breast cancer dataset. Expression levels of the representative genes (Z-scores) identified by expression array in MDA MB 231 cells (n = 496 genes) and MCF7 (n = 155) are represented. Red symbolizes up-regulated and green down-regulated genes. Hierarchical clustering was performed using CLuster and Treeview software

Fig. 2

Linear regression analysis of the expression of JMJD6 and EZH2 in patient derived RNA. Graphs show the best fit regression line in Normal (a) and Tumor (b) samples for JMJD6-EZH2 where X-axis and Y-axis show normalized expression (ΔCt) of respective genes and in JMJD6 and EZH2 altered samples from TCGA RNA-Seq and mass spectrometry data for mRNA (c) and protein (d) expression respectively

Fig. 3

Inter-regulation of JMJD6 and EZH2. a) Expression of JMJD6 and EZH2 in different JMJD6 overexpressing clones as quantified by real-time PCR, JMJD6 and EZH2 levels following JMJD6 siRNAs and EZH2 siRNAs is shown in b) for MDA MB 231 and c) for HEK 293 cells. Lower panels show JMJD6 and EZH2 proteins detected by immunoblot analysis. Endogenous JMJD6 is marked by filled squares and empty squares indicate V5-tagged protein (right hand side of the immunoblot panel). Actin beta (ACTB) was used as an internal control

Fig. 4

Recruitment of JMJD6 in the regulatory regions of co-regulated genes. Percentage input enrichment for ChIP is shown in a) J1-C6 cells using V5-antibody, b) MDA MB 231 cells using JMJD6 antibody for representative co-regulated genes

Fig. 5

Multiple sequence alignment of binding sites of JMJD6. Two types of clusters are shown with a) IL6, SIRT4, BRIX1, and b) IGF2BP3, ADAM17, AURKA, Histones, IDE sites. The conserved sequences are marked by “*”

Fig. 6

Graphical representation of JMJD6 and EZH2 action