RESUMO
Metastasis is a multi-step process that leads to the dissemination of tumor cells to new sites and, consequently, to multi-organ neoplasia. Although most lethal breast cancer cases are related to metastasis occurrence, little is known about the dysregulation of each step, and clinicians still lack reliable therapeutic targets for metastasis impairment. To fill these gaps, we constructed and analyzed gene regulatory networks for each metastasis step (cell adhesion loss, epithelial-to-mesenchymal transition, and angiogenesis). Through topological analysis, we identified E2F1, EGR1, EZH2, JUN, TP63, and miR-200c-3p as general hub-regulators, FLI1 for cell-adhesion loss specifically, and TRIM28, TCF3, and miR-429 for angiogenesis. Applying the FANMOD algorithm, we identified 60 coherent feed-forward loops regulating metastasis-related genes associated with distant metastasis-free survival prediction. miR-139-5p, miR-200c-3p, miR-454-3p, and miR-1301-3p, among others, were the FFL's mediators. The expression of the regulators and mediators was observed to impact overall survival and to go along with metastasis occurrence. Lastly, we selected 12 key regulators and observed that they are potential therapeutic targets for canonical and candidate antineoplastics and immunomodulatory drugs, like trastuzumab, goserelin, and calcitriol. Our results highlight the relevance of miRNAs in mediating feed-forward loops and regulating the expression of metastasis-related genes. Altogether, our results contribute to understanding the multi-step metastasis complexity and identifying novel therapeutic targets and drugs for breast cancer management.
Assuntos
Neoplasias da Mama , Neoplasias da Mama/genética , Neoplasias da Mama/patologia , Metástase Neoplásica , Regulação Neoplásica da Expressão Gênica , Fatores de Transcrição/genética , MicroRNAs/genética , Redes Reguladoras de Genes , HumanosRESUMO
In this article we focus on how the hierarchical and single-path assumptions of epistasis analysis can bias the inference of gene regulatory networks. Here we emphasize the critical importance of dynamic analyses, and specifically illustrate the use of Boolean network models. Epistasis in a broad sense refers to gene interactions, however, as originally proposed by Bateson, epistasis is defined as the blocking of a particular allelic effect due to the effect of another allele at a different locus (herein, classical epistasis). Classical epistasis analysis has proven powerful and useful, allowing researchers to infer and assign directionality to gene interactions. As larger data sets are becoming available, the analysis of classical epistasis is being complemented with computer science tools and system biology approaches. We show that when the hierarchical and single-path assumptions are not met in classical epistasis analysis, the access to relevant information and the correct inference of gene interaction topologies is hindered, and it becomes necessary to consider the temporal dynamics of gene interactions. The use of dynamical networks can overcome these limitations. We particularly focus on the use of Boolean networks that, like classical epistasis analysis, relies on logical formalisms, and hence can complement classical epistasis analysis and relax its assumptions. We develop a couple of theoretical examples and analyze them from a dynamic Boolean network model perspective. Boolean networks could help to guide additional experiments and discern among alternative regulatory schemes that would be impossible or difficult to infer without the elimination of these assumption from the classical epistasis analysis. We also use examples from the literature to show how a Boolean network-based approach has resolved ambiguities and guided epistasis analysis. Our article complements previous accounts, not only by focusing on the implications of the hierarchical and single-path assumption, but also by demonstrating the importance of considering temporal dynamics, and specifically introducing the usefulness of Boolean network models and also reviewing some key properties of network approaches.