Homeobox gene Rhox5 is regulated by epigenetic mechanisms in cancer and stem cells and promotes cancer growth

Abstract

Background: Homeobox genes murine Rhox5 and human RHOXF1 are expressed in early embryonic stages and then mostly restricted to germline tissues in normal adult, yet they are aberrantly expressed in cancer cells in vitro and in vivo . Here we study the epigenetic regulation and potential functions of Rhox5 gene.

Findings: In Rhox5-silenced or extremely low expresser cells, we observed low levels of active histone epigenetic marks (H3ac, H4ac and H3K4me2) and high levels of repressive mark H3K9me2 along with DNA hypermethylation in the promoter. In Rhox5 low expresser cells, we typically observed modest levels of both active and repressive histone marks along with moderate DNA methylation. In Rhox5 highly expressed CT26 cancer cells, we observed DNA hypomethylation along with high levels of both active and repressive histone marks. Epigenetic drugs (retinoic acid and MS-275) induced F9 cell differentiation with enhanced Rhox5 expression and dynamic changes of epigenetic marks. Finally, Rhox5 knockdown by small hairpin RNA (shRNA) in CT26 colon cancer decreased cell proliferation and migration in vitro and tumor growth in vivo .

Conclusions: Both DNA methylation and histone methylation/acetylation play key roles in modulating Rhox5 expression in various cell types. The stem cell-like "bivalent domain", an epigenetic feature originally identified in key differentiation genes within stem cells, exists in the Rhox5 gene promoter in not only embryonic stem cells but also cancer cells, cancer stem cells, and differentiated Sertoli cells. As Ras signaling-dependent Rhox5 expression promotes tumor growth, Rhox5 may be an ideal target for therapeutic intervention in cancer.

Figure 1

Rhox5 expression in ES cells, somatic cells and cancer cells. (A). Schematic presentation of the Pd and Pp promoter regions of the Rhox5 gene. Vertical bars represent CpG dinucletides and open and filled boxes indicated non-coding and coding exons. Three segments (BS-1, BS-2, BS-3) covering CpG dinucleotides subject to bisulfite sequencing and corresponding regions for ChIP assay (ChIP-1 and ChIP-2) were also indicated. (B). Rhox5 mRNA in representative cancer cell lines was detected by semi-quantitative RT-PCR. M: PCR marker. (C). Relative Rhox5 mRNA levels as normalized to 1.0E4 copies of GAPDH mRNA. Real-time RT-PCR was performed in triplicates and data from multiple experiments are presented as mean +/- s.d. The data for ES cells were obtained from single RT-PCR in triplicates. (D). Transcripts from either Pd or Pp were determined by promoter-specific RT-PCR. Lane M: DNA MW Markers; T: testis; H₂ O: water instead of RNA was used in RT. (E). Western blot analysis of Rhox5 in germline tissues and representative cancer cell lines.

Figure 2

RhoxF1 mRNA expression in liver lesions of colorectal cancer and matched liver tissues from colorectal cancer patients. Total RNA from matched liver tissues (N) and metastatic colorectal cancer (T) was purified and RHOXF1 mRNA was quantified by RT-qPCR. Relative Rhox5 mRNA levels were expressed as normalized to 1.0E6 copies of β-actin mRNA. Bars represent means with s.d. In patient 4, RHOXF1 mRNA from tumor (T) was below the level of detection.

Figure 3

Rhox5 is bivalent marked in ES cells, somatic cells and cancer cells. (A). Histone methylation marks (H3K4me2; H3K27me3 and H3K9me2) on Rhox5 promoter regions were determined by ChIP assays. Locations of ChIP-1 and ChIP-2 regions were illustrated in the cartoon of Fig. 1A. (B). Histone acetylation and methylation marks on promoter regions in cancer cells with Rhox5 expression at high (CT26) and low (EMT6) levels. (C). Quantification of the data from CT26 and EMT6 cells. Bars are mean with s.d.

Figure 4

Promoter DNA methylation pattern and Rhox5 mRNA expression level. (A). DNA methylation status in Pd, Pp, and TSS regions in different cell types including ES, somatic cells, and cancer cells. Each column represented one CpG site and one row indicated separate clones picked for sequencing. Open and filled circles indicate individual unmethylated and methylated CpGs, respectively. (B). Correlation of Pd DNA methylation status versus gene transcription in 10 different cell types. Cell types are, NE: normal mammary epithelial cells MM3MG. Six cancer cell lines used are, EMT6, 4T1, P815, MOSEC, CT26 and MC38. -: non-detectable; +/-: extremely low; +: low; ++: moderate; +++: high.

Figure 5

Upregulation of Rhox5 mRNA was associated with dynamic changes of H3K4 and H3K27 methylation marks in Pd region. (A). Upregulation of Rhox5 in differentiating F9 cells as induced by MS-275, RA or RA plus cAMP. The mRNA was quantified by RT-qPCR, and the level of Rhox5 mRNA in mock-treated was arbitrarily set as 1.0. (B). Occupancy of H3 K4me2, K27me3, and K9me2 marks in Pd region in mock, MS-275, and RA-treated F9 cells as determined by real-time RT-PCR. (C). Upregulation of Rhox5 mRNA in DAC or MS-275-treated cancer cells. Fold change of Rhox5 mRNA (as normalized by Gapdh ) was compared to mock-treated cells. (D). ChIP analysis of K4me2, K27me3 and K9me2 marks at Rhox5 Pd region (ChIP-1) in mock and epigenetic drug-treated EMT6 and P815 cancer cells.

Figure 6

Rhox5 mRNA expression and K4 and K27 bivalent marks in mock or MS-275 treated SP and NSP cells. (A). Existence of SP and NSP in MOSEC cancer cells. (B). Regulation of Rhox5 mRNA expression in mock or MS-275-treated MOSEC, SP, and NSP cells. *: P < 0.05. (C). Occupancy of K4 and K27 marks on the chromatin of Pd ChIP-1 region in SP, NSP, and MS-275-treated SP cells. Experiments were performed in triplicate; error bars depict SEM.

Figure 7

ShRNA-mediated knockdown of Rhox5 reduced cell migration and attenuated growth of CT26 cancer cells. (A). Efficiency of Rhox5 knockdown of shRNA clones with different target sequences against Rhox5. Lentiviral particles with control empty vector (CTV) or 2 different target sets (clone 48, 49) of shRNA were used to infect CT26 cells. After 3^rd round of puromycin selection, total RNA was purified and analyzed with RT-qPCR and expression of relative Rhox5 mRNA was compared to the empty vector control (set as 1.0). (B). Reduced Rhox5 protein level in Rhox5 shRNA-treated clone 49 cells. Representative result of Western blot film exposed for a short time (10 seconds) was shown. Rhox5 protein in clone 49 was detected in a longer exposure. (C). Proliferation of CT26 cells stably-transfected with control or targeted shRNA at different time points in vitro. (D). Migration ability of CT26 cells stably-transfected with control or targeted shRNA. (E). Tumor volumes (left side) and tumor weight (right side) of control and targeted shRNA stably-transfected CT26 cells at day 19 after inoculation in athymic nude mice. *: p < 0.05 when compared with control groups.