Histone-acetylated control of fibroblast growth factor receptor 2 intron 2 polymorphisms and isoform splicing in breast cancer

Abstract

Recent genome-wide association studies have identified fibroblast growth factor receptor (FGFR)2 as one of a few candidate genes linked with breast cancer susceptibility. In particular, the disease-predisposing allele of FGFR2 is inherited as a 7.5-kb region within intron 2 that harbors eight single nucleotide polymorphisms. The relationship between these single nucleotide polymorphisms and FGFR2 gene expression remains unclear. Here we show the common occurrence of polymorphisms within the intron 2 region in a panel of 10 breast cancer cell lines. High FGFR2-expressing cell lines such as MCF-7 cells displayed polymorphic sequences with constitutive histone acetylation at multiple intron 2 sequences harboring putative transcription binding sites. Knockdown of Runx2 or CCAAT enhancer binding protein beta in these cells resulted in diminished endogenous FGFR2 gene expression. In contrast FGFR2-negative MDA-231 cells were wild type and showed evidence of histone 3/4 deacetylation at the rs2981578, rs10736303, and rs7895676 disease-associated alleles that harbor binding sites for Runx2, estrogen receptor, and CCAAT enhancer binding protein beta, respectively. Histone deacetylation inhibition with trichostatin A resulted in enhanced acetylation at these intron 2 sites, an effect associated with robust FGFR2 reexpression. Isoform analysis proved reexpression of the FGFR2-IIIc variant the splicing of which was positively influenced by trichostatin A-mediated recruitment of the Fas-activated serine/threonine phosphoprotein survival protein. Our findings highlight the potential role of histone acetylation in modulating access to selected polymorphic sites within intron 2 as well as downstream splicing sites in generating variable FGFR2 levels and isoforms in breast cancer.

Figure 1

The FGFR2 intron 2 polymorphisms are frequent in human breast carcinoma cell lines. Shown is the FGFR2 gene locus, and the intron 2 SNPs numbered as described in genome-wide studies (3,4). The SNP positions associated with increased breast cancer risk reported in Ref. are denoted by a single asterisk; those reported in Ref. are denoted by a double asterisk.

Figure 2

FGFR2 mRNA expression, isoform splice variants, and protein expression in human breast carcinoma cell lines. A, Total RNA was examined by RT-PCR using primers that cover the FGFR2 coding region (FGFR2). Primers specific to the cytoplasmic C-terminal splice variants (C1, C2, C3) are shown immediately below. RNA integrity was verified by amplification of the phosphoglycerate kinase housekeeping gene. B, Unique restriction site permits distinction of the IIIb isoform which contains an AvaI restriction site (b) from the IIIc isoform that contains an EcoRV digestion site (c). Longer exposures of MCF-10A digests show two restricted products following AvaI or EcoRV enzymatic treatment consistent with dual isoform expression in this cell line. C, Western blotting of total cell lysates using an antibody that recognizes the C terminus of FGFR2. D, Two low-FGFR2-expressing cell lines as indicated show gene up-regulation upon xenografting into SCID mice as demonstrated by immunocytochemistry. Inset photomicrographs represent higher magnifications of corresponding images. PGK, Phosphoglycerate kinase.

Figure 3

Histone acetylation modulates access to binding sites within intron 2 of FGFR2. A, Schematic representation of the four intron-2 SNPs that were targeted for ChIP analysis. The corresponding putative transcription factor binding sites are shown immediately above. B, Sites rs10736303 (303) (the ER binding site) and rs7895676 (676) (the C/EBPβ binding site) show relative deacetylation in FGFR2-deficient MDA-231 cells. Site rs2981578 (578) harboring a potential Oct/Runx2 binding site displays H4 deacetylation. C, Consistent with these features endogenous gene expression at the mRNA (upper panel) and protein levels (lower panel) are induced by the HDAC inhibitor TSA but not by the DNA methylation inhibitor AZC at the indicated doses in MDA-231 cells. D, Quantitative real-time PCR confirms the positive effect of TSA treatment (+) on FGFR2 mRNA expression in different cell lines as indicated. E, Down-regulation through siRNA directed at Runx2 or C/EBPβ in the FGFR2-expressing MCF-7 cells results in reduction of the endogenous receptor’s expression. Respective scrambled controls are represented by an “S” and the phosphoglycerate kinase housekeeping gene is represented by PGK. AcH3, Antiacetylhistone 3.

Figure 4

HDAC inhibition results in selective reexpression of the FGFR2-IIIc isoform. A, MDA-231 breast cells with very low FGFR2 expression (Fig. 2) were treated with AZC or TSA at the indicated doses. To detect the FGFR2-IIIb and IIIc isoforms before and after treatment in these cells, 10 μg of PCR product from five separate reactions was pooled for restriction analysis as shown. Only TSA treatment enhances reexpression mostly of the IIIc but not the IIIb isoform. B, Schematic representation of the intron/exon junctions that were examined by ChIP assays. C, TSA treatment enhances 5′-promoter region histone acetylation as well as the exon IIIc region. Little impact is observed surrounding the IIIb exon. AcH3, antiacetylhistone 3.

Figure 5

HDAC inhibition promotes FAST binding to FGFR2 mRNA splicing. A, Schematic representation of the overlapping regions including exon IIIb/IIIc junctions that were examined by ChIP. B, ChIP analysis demonstrates constitutive FAST binding to the IIIb regions in BT-20 cells, which express the receptor isoform, compared with BT-549, which recognize FAST binding at the IIIc isoform, that they exclusively express (Fig. 2). C, TSA treatment enhances FAST binding selectively to the IIIc region favoring its splicing inclusion in MDA-231 cells. D, FAST down-regulation using an siRNA (si-FAST) in MDA-231 cells abrogates the effect of TSA on FGFR2 induction. PGK, Phosphoglycerate kinase.