The role of conserved GFX DNA motif and Kaiso in lung cancer

Abstract

Lung cancer is a common cancer worldwide with high mortality but results of systemic treatment are still suboptimal. Novel approaches are needed to uncover new treatment targets, e.g., by decoding conserved cis-regulatory elements (CREs) of cancer phenotypes. This study focused on regulatory factors acting through GFX (TCTCGCGAGA) which is the most conserved CRE on human promoters. To identify GFX binding factors, integrated analysis of ChIP-seq data from ENCODE and RNAseq data from TCGA database was first performed. Interestingly, Kaiso, a well-known transcriptional repressor reported to paradoxically bind on unmethylated and highly active GFX, as well as FOXM1 were found to be involved in more than 3 cancer cell lines, suggesting Kaiso and FOXM1 could contribute to oncogenesis through formation of co-regulatory elements on GFX. To investigate the transcriptional activity statuses of Kaiso and FOXM1 on GFX, ChIP-seq for FOXM1, Kaiso, H3K27ac and H3K27me3 were obtained from ENCODE project, or additionally generated using a locally raised lung adenocarcinoma cell line (PTC02). Results showed both Kaiso and FOXM1 were enriched on the GFX motif in GM12878, K562 and A549, but Kaiso was absent on GFX in PTC02 and MCF7. Gene knockdown and cell function assays of the lung cancer cells A549 and PTC02 showed Kaiso and FOXM1 reciprocally transregulated each other, and both promoted cell cycle progression, cell proliferation, invasion and migration. To identify GFX targets regulated by Kaiso and FOXM1, the genes were alternatively knocked down by shRNA in A549 and PTC02, respectively, followed by RNAseq. Results showed 67 upregulated (p<0.05 and log2FC>2) candidates including EGFR in both A549 and PTC02, and 94 repressed candidates including TGFBR3 in A549. Next, pathway characterization using enrichment analysis of functional terms (DAVID) identified clusters of cell growth terms (cell cycle, cell division, DNA replication, DNA repair) for the 67 upregulated-genes, and tumor-stroma interaction clusters (immunoglobulin I-set domain, extracellular matrix, leucine-rich repeats) for the 94 repressed-genes. Also, pathway analysis of multiple expression profiles from GEO datasets by GSEA showed repression of the upregulated-genes by EGFR tyrosine kinase inhibitors (TKI) treatment, and induction of the repressed-genes by TGFB1. Together, results suggested the 67 upregulated-genes mediated cell growth through EGFR, while the 94 repressed-genes were targets of TGFB-mediated tumor-stroma interaction. The potentials of GFX-targets as prognostic biomarkers were analyzed using public databases. Low levels of the 94-repressed genes and high expression of the 67-upregulated genes were associated with cisplatin resistance of 52 lung adenocarcinoma cell lines registered in the Sanger Genomics of Drug Sensitivity in Cancer (GDSC) database. Further, analysis using TCGA expression profiles showed patients with high levels of both the 67-upregulated and 94-repressed genes had the worst 5-year survival compared with those showing low levels of either or both groups of GFX-targets. Collectively, this study has shown Kaiso and FOXM1 are regulatory factors on the conserved GFX motif. Functionally, their targets contribute to cell growth and tumor-stroma interaction through EGFR and TGFB pathways. Clinically, they may be useful predictors of chemotherapy resistance, and a role on lung cancer patient survival is also implicated.