Revealing nuclear receptor hub modules from Basal-like breast cancer expression networks.

Sharon Nienyun Hsu,James Smith,Di Wu,Mengzhen Liu,Thomas A Hughes,Erika Wong En Hui

doi:10.1371/journal.pone.0252901

Abstract

Nuclear receptors are a class of transcriptional factors. Together with their co-regulators, they regulate development, homeostasis, and metabolism in a ligand-dependent manner. Their ability to respond to environmental stimuli rapidly makes them versatile cellular components. Their coordinated activities regulate essential pathways in normal physiology and in disease. Due to their complexity, the challenge remains in understanding their direct associations in cancer development. Basal-like breast cancer is an aggressive form of breast cancer that often lacks ER, PR and Her2. The absence of these receptors limits the treatment for patients to the non-selective cytotoxic and cytostatic drugs. To identify potential drug targets it is essential to identify the most important nuclear receptor association network motifs in Basal-like subtype progression. This research aimed to reveal the transcriptional network patterns, in the hope to capture the underlying molecular state driving Basal-like oncogenesis. In this work, we illustrate a multidisciplinary approach of integrating an unsupervised machine learning clustering method with network modelling to reveal unique transcriptional patterns (network motifs) underlying Basal-like breast cancer. The unsupervised clustering method provides a natural stratification of breast cancer patients, revealing the underlying heterogeneity in Basal-like. Identification of gene correlation networks (GCNs) from Basal-like patients in both the TCGA and METABRIC databases revealed three critical transcriptional regulatory constellations that are enriched in Basal-like. These represent critical NR components implicated in Basal-like breast cancer transcription. This approach is easily adaptable and applicable to reveal critical signalling relationships in other diseases.

Highlights

Nuclear receptors (NRs) are ligand induced transcriptional factors that regulate essential pathways in normal physiology and in disease
We used an unsupervised clustering method to stratify breast cancer into classes with distinct NR expression patterns and identified three critical modules that are enriched in Basal correlation networks
This work employed a nonparametric multinomial Dirichlet process clustering algorithm, Bayesian Hierarchical Clustering (BHC) to stratify breast cancer patients and discriminate the Basal-like patients based on the mRNA expression signals of nuclear receptors and their immediate co-regulators

Summary

Introduction

Nuclear receptors (NRs) are ligand induced transcriptional factors that regulate essential pathways in normal physiology and in disease. Nuclear receptor hub modules from Basal-like breast cancer technology, researchers are beginning to comprehend the innate complexity of signalling pathways involving NRs and their involvement in tumour development and progression [1]. We used an unsupervised clustering method to stratify breast cancer into classes with distinct NR expression patterns and identified three critical modules that are enriched in Basal correlation networks. Network-based modules provide a robust description of transcriptional characteristics underlying Basal-like breast cancer, and take into account dynamic properties and collaborative behaviours of NRs

Objectives

Methods

Results

Conclusion