Using network and functional enrichment clustering analyses to find therapeutic targets for breast cancer: The role of cyclin‐dependent kinase 2

Abstract

MSC Classification: 62H30; 62F03Population aging is rapidly increasing in developing countries; thus, covering medical needs for breast cancer diagnosis and treatment is a priority in Latin America. We describe an approach for integrating differential expression analysis, biological pathway enrichment, in silico transcription-binding sites and network topology, to find key genes that may be used as biomarkers or therapeutic targets. This approach is based on publicly available data from microarrays of the MCF-7 breast cancer cell line treated with estrogen. We found significant estrogen-responsive genes, which were used as nodes to construct networks based on protein-protein interactions reported in the literature. Then, we conducted a topology analysis of the networks, revealing the most-connected nodes, ie, those responsible for maintaining the network structure corresponding to genes with well-acknowledged functions in G1/S cell cycle transition, such as cyclin-dependent kinase 2 (CDK2), which has been proposed as a therapeutic target in classical biochemical studies. In addition, analyses of biological pathway enrichment and in silico transcription-binding sites support the biological meaning and importance of these key genes and help to decide the best target genes. Therefore, we postulate that the integrative bioinformatics approach shown here, unlike the classical bioinformatics approach that only includes differentially expressed genes and enriched biological pathways, can be applied as an approach for finding novel biomarkers and/or therapeutic target genes for nonresponsive treatments. KEYWORDS breast cancer biomarkers, functional enrichment analysis, gene expression clustering, gene network analysis 8514 VALENZUELA AND ASSAR 8515to new cases, and the second-leading cancer with respect to mortality in women. According to BC data, 3 over half of the deaths occur in South America and Africa. 3 Studying BC is difficult because it is genetically, morphologically, and clinically heterogeneous. To obtain better diagnoses and treatments, many studies have tried to classify this cancer into subtypes, using morphological and molecular approaches. Gene expression analyses have identified at least 4 intrinsic subtypes of BC: Luminal A, Luminal B, HER2-enriched, and a triple-negative group with significant differences in terms of risk, response to therapies, and prognoses. 4 The histological characteristics of BC are currently used to determine subtype, size, lymph node status, and the cytoplasmic and nuclear grade.Worldwide, the use of the indicated markers is widely recommended to define subtypes of BC and to even rank them according to their severity. 5,6 These differences between tumor subtypes are often associated with clinical features, such as relapse-free and overall survival 7 and also treatments. The triple-negative subtype has the worst survival rates, and tamoxifen is used against ER+ cells. However, there is a controversy over whether ER, PR, HER2, and KI67 proteins can effectively capture the complex ...