Epigenetic silencing of downstream genes mediated by tandem orientation in lung cancer

Abstract

Epigenetic deregulation is of importance in tumorigenesis. In particular CpG islands (CGI), are frequently hypermethylated. Here, genome-wide DNA-methylation profiles of 480,000 CpGs in lung cancer cells were generated. It was observed that intra- and intergenic CGI exhibited higher methylation compared to normal cells. The functional annotation of hypermethylated CGI revealed that the hypermethylation was associated with homeobox domain genes and targets marked by repressive histone modifications. The strongest methylation variation was observed in transitional areas of CGI, termed shores. 5'-shores of promoter-associated CGI in lung cancer cell lines were higher methylated than 3'-shores. Within two tandem-oriented genes, a significant hypermethylation of the downstream-located CGI promoters was revealed. Hypermethylation correlates with the length of the intergenic region between such tandem genes. As the RASSF1A tumor suppressor gene represents such a downstream tandem gene, its silencing was analyzed using an inducible system. It was determined that the induction of an upstream gene led to a repression of RASSF1A through a process involving histone deacetylases and CPSF1. A tumor-specific increase in expression of histone deacetylases and CPSF1 was detected in lung cancer. Our results suggest that the downstream gene could be susceptible to epigenetic silencing when organized in a tandem orientation.

Figure 1

Increased methylation of CpG island-associated regions (CGIR) occurs in lung cancer cell lines. (A) GC and CpG rich genomic elements, termed CpG islands are flanked by 2 kb shore and 2 kb shelf regions. Depending on their orientation on the chromosome from the p- to q-arm these regions are denoted as N- or S-shores and shelves respectively. (B) Methylation levels of CpGs (n = 485,577) in CGIR (CpG island and flanking regions) and in no CGIR in normal human bronchial epithelial cells (NHBEC) and in three lung cancer cell lines (A427, A549 and H322) as quantified by 450 K bisulfite bead chip arrays and depicted in violin plots. (C) Methylation levels of transcription start site (TSS ± 300 bp flanking regions) associated CpGs (n = 116,076) were analyzed according to their location in CGI regions (CGIR) and no CGIR. (D) Methylation of CpGs outside of TSS (±300 bp) were analyzed according to the position in CGIR and no CGIR.

Figure 2

Methylation of gene specific CpGs in lung cancer cell lines. (A) CpG sites are annotated depending on the gene specific orientation in TSS1500 (−1500 to −200 bp), TSS200 (−200 bp to TSS), 5′-UTR, 1^st exon, gene body and 3′-UTR. (B) Methylation levels of CpG islands and flanking N-and S-shores were analyzed in NHBEC and in the lung cancer cell line H322 by 450 K bisulfite bead chip arrays according to the gene specific annotation.

Figure 3

Increased 5′-shore methylation in promoter and coding regions. (A) Depending on the chromosome orientation genes are transcribed from the p- to the q arm (Watson strand genes) or in the opposite direction (Crick strand genes). For Watson strand genes the CGI flanking N-shores are the 5′-shores and for Crick strand genes S-shores are the 5′-shores. For further details are provided in Fig. 2. (B) Methylation levels of N- and S-shore gene specific regions in 69 NSCLC cell lines (ABC-1, SW1573, HOP18, HCC1171, LXFL529, H1703, H441, EBC-1, H322, A427, H1568, CHA-GO-K-1, H2087, H1975, H226, H2444, HCC515, H1299, H2030, H2405, HCC2935, H1373, H1666, RERF-LC-Ad1, Calu-6, H1155, H1651, H2347, H358, H838, H2228, HCC827, H2073, H1650, H23, H1993, Calu-3, H650, H460, H727, A549, H292, HCC4006, H2170, H1838, H820, H1355, RERF-LC-MS, H2122, H1793, H661, HOP62, HCC15, EKVX, H1792, H2110.1, Calu-1, H2110, HCC4017, H2009, HOP92, SK-MES-1, RERF-LC-KJ, H1437, H647, H2126, H2172, H1435 and H1755), six normal lung cell lines (NLC): gBEC1, gBEC1_UI, gSAC1, gSAC1_UI, gBEC and gSAC) and five normal lung tissues (NLT): GSM1264690, GSM1264711, GSM1264727, GSM1264764, GSM1264854). Methylation data were obtained from NLCBI-GEO-Accession: GSE36216 and GSE52401^, . Promoter regions are defined as the TSS1500 (−1500 to −200), TSS200 (−200 to TSS), 5′-UTR, 1^st exon and coding regions consisted of the gene body and 3′-UTR.

Figure 4

Epigenetic silencing of downstream genes located in tandem orientation. (A) Outline of the organization of tandem oriented genes with a CpG island at the transcriptional end site (TES) of the upstream gene and the transcriptional start site (TSS) of the downstream gene. (B) Methylation level of TSS-associated CGI in A549 cells was plotted depending on the distance between TES and TSS in kb (n = 258 to 297 CGI). Methylation as detected using 450 K bisulfite bead chip arrays. “All CpGs” and “all TSS” indicate the methylation level of CpGs located at CGI and TSS-associated CGI, respectively. (C) Methylation levels of tandem oriented downstream CGI in H322 cells. (D) RASSF1A expression levels were analyzed in lung cancer cell lines (LuCa: A427, A459, H322 and H358) and normal lung cells (NC: PAF and PASMC) by qRT-PCR. RASSF1A expression was normalized to the GAPDH level. (E) Methylation of RASSF1A was analyzed in 69 non small cell lung cancer (NSCLC) cell lines and 20 normal lung tissues (NL). For details see also Fig. 3. (F) ZMYND10 expression levels were analyzed in LuCa and NC as measured by qRT-PCR and normalized to GAPDH levels. (G) Methylation of ZMYND10 was analyzed in 69 NSCLC cell lines and 20 NL. *p < 0.05, **p < 0.01 and ***p < 0.001.

Figure 5

Induction of an upstream gene correlates with the downregulation of a tandem oriented downstream promoter. (A) Outline of the tandem oriented reporter gene construct. The ZMYND10 - RASSF1A promoter including the ZYMND10 transcriptional end site (TES) and the RASSF1A promoter and transcriptional start site were cloned in a tetracycline inducible vector system (Tet -On/TO). The EGFP-ZYMND fusion gene was under the control of the pCMV 2xTETO₂ promoter and the 500 bp RASSF1A promoter including 5′-UTR and 17 bp of Exon1α was ligated in frame to Myc tagged Renilla luciferase (RLuc). (B) UCSC genome browser view of the 2.3. kb ZMYND10 - RASSF1A promoter locus. Black boxes represent exons. Outline of the genetic structure of the tandem reporter TO-EGFP-2,3-RLuc. In the construct TO-EGFPpA-2,3-RLuc a 300 bp SV40 poly A site (pA) was inserted at the 3′-end of EGFP. To generate the tandem reporter TO-EGFP-0,5-RLuc a 1.8 kb fragment of ZYMND10 was deleted. (C) Luciferase assay of the TO-EGFP-2,3-RLuc construct. The tandem reporter was stably transfected into TREx293 cells and induced for four days with the indicated concentration of doxycycline (Dox). RLuc activity was analyzed as described in the methods section. (D) Time dependent downregulation of RLuc activity. TREx293 cells stably transfected with TO-EGFP-2,3-RLuc were induced with 4 mM Dox (+) or uninduced (−) for the indicated days (d) and RLuc activity was analyzed. (E) Analysis of different tandem constructs by western blot and luciferase assay. Two TREx clones (B7 and A7) of the stable transfected TO-EGFP-2,3-RLuc were induced with 4 mM Dox for four days and the protein levels were subsequently analyzed by western blot. Protein lysates were separated on SDS PAGE and blotted. Full-length blots are included in the supplementary information file. For detection of the indicated proteins primary antibody against myc, GFP and GAPDH were utilized. In parallel RLuc activity was analyzed with a luciferase assay. RLuc luciferase activity was normalized to transient transfected firefly luciferase. Additionally the expression of RLuc and EGFP were analyzed in two TO-EGFPpA-2,3-RLuc TREx clones (A1 and A2) and the C2 clone of TO-EGFP-0,5-RLuc by western blot and luciferase assay. *p < 0.05, **p < 0.01 and ***p < 0.001.

Figure 6

Epigenetic regulation of tandem genes in lung cancer. (A) Outline of the tandem oriented reporter gene construct analyzed by chromatin immuno-precipitation (ChIP). The analyzed TO/EGFP and RASSF1A/RLuc promoter regions are depicted. For details see Fig. 5. (B) TREx293 cells stably transfected with TO-EGFP-2.3-RLuc (clone B7) were induced with 4 mM Dox (+) or uninduced (−) for four days (d) and histone modifications were analyzed by ChIP. Chromatin was precipitated with anti histone H3K9ac, histone H4ac and IgG antibodies (negative control) antibodies. DNA was isolated and analyzed by qPCR. Levels are plotted relative to 1% of input sample. (C) Expression of HDAC1, HDAC2, HDAC5, HDAC6, DNMT1, DNMT3A, DNMT3B and DNMT3L was analyzed by microarray in 49 primary squamous cell lung cancer (SCCa), 41 primary lung adenocarcinoma (AdCa), 15 non malignant lung disease samples (LD) and 6 normal lung tissues (NL). Data are depicted as log2 transformed expression. (D) CPSF1 induced repression of RASSF1A. CPSF1 was cloned in pCMV-Tag1 and overexpressed in A427 cells. Endogenous RASSF1A level were analyzed by qRT-PCR and plotted relative to the empty control vector (=1). (E) Expression levels of CPSF1 in SCCa, AdCa, LD and NL. *p < 0.05, **p < 0.01 and ***p < 0.001.