In silico discovery of transcription factors as breast cancer biomarkers

Abstract

Breast cancer is the most prevalent type of cancer affecting women. This disease is grouped into subtypes with different gene expression profiles, which affect the response to treatment and the prognosis of breast cancer patients. Estrogen receptor negative (ER-) breast cancer subtypes generally have a poor patient prognosis due to the lack of targeted treatment options and the high relapse rate after chemotherapy. The present study is aimed at computationally evaluating the differences in gene transcription between ER positive (ER+) and ER- breast cancer subtypes and identifying transcription factors (TFs) controlling these differences. RNAsequencing data was obtained from publically available databases for MCF7, MDA-MB-231 and MCF10A cell lines, representing ER+ breast cancer, ER- breast cancer and nontumorigenic breast cells respectively. Differentially expressing genes were selected by comparing the gene expression profiles of cancer cell lines to non-tumorigenic cells. Functional enrichment was performed using gene ontology and KEGG pathways to identify the biological roles the differentially expressing genes play in breast cancer. The promoters of differentially expressing genes were assessed for TF binding site (TFBS) enrichment to identify the transcriptional controllers of breast cancer-related gene expression. The expression of the TFs selected as key regulators was validated in breast cancer patient datasets. The prognostic value of the TFs upregulated in breast cancer patient data was evaluated using patient survival data. Potential biomarkers were selected based on prognostic value. E2F1, INSM1 and MYC were predicted as potential biomarkers from MCF7 expression data and FOXD1, TAL1, RUNX1 and MAX were predicted using MDA-MB-231 data. Finally, networks were constructed to visualise the interactions between potential TF biomarkers and the genes that they regulate. This preliminary prediction of TF biomarkers could provide a better understanding of the molecular mechanisms governing the characteristics of different breast cancer subtypes, and could be used as novel biomarkers for breast cancer with diagnostic and therapeutic potential after further validation using patient tumour samples.