Epigenetics mechanisms mediate the miR-125a/BRMS1 axis to regulate invasion and metastasis in gastric cancer

Abstract

Purpose: Altered expression of breast cancer metastasis suppressor 1 (BRMS1), is a tumor suppressor, which is found in many types of cancers, including gastric cancer (GC), but the mechanism by which BRMS1 inhibits invasion and metastasis in GC is unknown. The aim of the study was to investigate the molecular mechanisms of miR-125a/BRMS1 in GC.

Materials and methods: The expression of BRMS1 and miR-125a were detected by quantitative real-time PCR (qRT-PCR) and analyzed by bioinformatics. BSP and MSP were used to detecte the methylation status of miR-125a and BRMS1 which was treated by 5-Aza or not. Western Blot and qRT-PCR were used to analyze the expression of BRMS1 and EZH2. Transwell was performed to explore the invasion and metastasis ability of GC cells. The nude mice were used for the tumor formation assay.

Results: BRMS1 may be regulated by copy number variation (CNV), methylation and miR-125a-5p. As one of the essential components of PRC2, EZH2 is an important regulatory factor resulting in the low expression of miR-125a. An epigenetic mechanism mediates the miR-125a/BRMS1 axis to inhibit the invasion and metastasis of GC cells. In vivo experiments, it is also showed that BRMS1 is involved in invasion and metastasis but not the proliferation in GC.

Conclusion: These studies shed light on the mechanism of BRMS1 inhibition of GC invasion and metastasis and the development of new drugs targeting the miR-125a/BRMS1 axis, which will be a promising therapeutic strategy for GC and other human cancers.

Keywords: EZH2; epigenetic; gastric cancer; invasion and metastasis; miR-125a/BRMS1 axis.

Figure 1

Evaluation of the potential mechanism of miR-125a and BRMS1 low expression in GC. (A) The relationship between CNV and miR-125a expression is not significant (P=0.9442). The data were downloaded from TCGA stomach carcinoma dataset. (B) Treatment with 5-Aza significantly increased miR-125a expression compared with untreated cells of SGC-7901 and MKN-28. (C) Treatment with 5-Aza significantly increased BRMS1 expression compared with untreated cells of SGC-7901 and MKN-28. (D) BRMS1 mRNA expression was analyzed by qRT-PCR after treatment with 5-Aza and/or anti-miR-125a in SGC-7901 cells. (E and F) The relationship between BRMS1 expression and CNV/DNA methylation presented a significant correlation (BRMS1 vs CNV: r=0.52, P<0.0001; BRMS1 vs DNA methylation: r=−0.21, P<0.0001). The data were downloaded from TCGA stomach carcinoma dataset. *P<0.05.

Figure 2

Promoter methylation is not the primary mechanism leading to the low expression of miR-125a. (A) Bioinformatic prediction of one CpG island upstream of miR-125a. (B) The miR-125a methylation status of GES-1, SGC-7901 and MKN-28 cells was analyzed by BSP. (C) The methylation level between GES-1 and SGC-7901 or GES-1 and MKN-28 showed no significant difference (P>0.05). (D) The qRT-PCR results showed that the expression levels of miR-125a in the 3 of 4 primary tumors were reduced compared with the paracarcinoma tissues (P<0.05). (E) The miR-125a methylation status in 4 paired GC tissues and para-carcinoma tissues was analyzed by BSP. (F) The statistical analysis showed no significant difference in the number of methylation between the GC tissues and paracarcinoma tissues(P=0.1671). *P<0.05.

Figure 3

EZH2 is an important molecule affecting the low expression of miR-125a. (A) The expression of EZH2 and miR-125a in GC from TCGA Pan-Cancer, which was analyzed by ChIPbase (r=−0.216, P=4.06x10⁻⁶). (B) MiR-125a did no target EZH2, as shown by microRNA.org, TargetScan and PITA database analysis. (C) EZH2 expression did not present a significant change with/without miR-125a mimics or anti-miR-125a (P>0.05) (Figure 3C). (D and E) qRT-PCR and Western blot analysis of EZH2 expression in SGC-7901 which was stably transduced with lentiviral vectors carrying EZH2-specific small hairpin RNAs (shEZH2). (F) Decreasing the expression of EZH2 results in a markedly improved the expression of miR-125a compared with the NC group. (G) Bioinformatic prediction of two sites indicates possible enrichment of EZH2 in the promoter region of miR-125a, and ChIP assays found that the miR-125a promoter region was enriched for EZH2 in GC. (H) Bioinformatic analysis showed that there is no significant correlation between EZH2 and BRMS1 expression (P=0.7131). The data were downloaded from TCGA stomach carcinoma dataset. *P<0.05.

Figure 4

Promoter methylation affects the expression of BRMS1. (A) Bioinformatic prediction of one CpG island upstream of BRMS1, which is mainly concentrated 300 bp above the transcription start site. (B) The bisulfite sequencing of the BRMS1 upstream region in MKN-28, SGC-7901 and GES-1 cell lines analyzed by BiQ Analyzer. (C) The number of methylations was higher in MKN-28 and SGC-7901 than in GES-1. (D) Representative MSP analyses for BRMS1 methylation in GC tissues. U, unmethylated state; M, methylated state. (E) The bisulfite sequencing of the BRMS1 upstream region in 2 paired GC tissues and paracarcinoma tissues analyzed by BiQ Analyzer. Each square represents a CpG site. For the figure B and E, methylated CpG dinucleotides are represented by yellow squares whereas unmethylated CpG sites are represented by blue squares. If the methylation state of a CpG site could not be defined, it is represented as not present. (F) The number of methylations was higher in GC tissues than in para-carcinoma tissues. *P<0.05.

Figure 5

Epigenetic regulation of the miR-125a/BRMS 1 axis to promote GC invasion and metastasis in vitro and in vivo. (A) Working model for the regulation of the miR-125a/BRMS1 axis by EZH2 in GC. (B and C) Western blot analysis of the expression of BRMS1 treated with/without Sh2-EZH2/anti-miR-125a. (D) Invasion assay analysis of the invasive cells in SGC-7901 cells treated with/without 5-Aza/siBRMS1. (E and F) Representative images of metastatic tumors in nude mice inoculated with NC/shBRMS1. The volume of metastatic tumors between NC and shBRMS1 was no significantly different (P>0.05). (G and H) Representative images of metastatic tumors in the mesentery and intestine of nude mice inoculated with NC/shBRMS1. The number of tumor nodes was significantly lower in the siBRMS1 group than in the NC group (P<0.05). *P<0.05.