Weighted Gene Correlation Network Analysis (WGCNA) Reveals Novel Transcription Factors Associated With Bisphenol A Dose-Response.

Alexandra Maertens,Andre Kleensang,Vy Tran,Thomas Hartung

doi:10.3389/fgene.2018.00508

Alexandra Maertens, Andre Kleensang + Show 2 more

Open Access

https://doi.org/10.3389/fgene.2018.00508

Copy DOI

Abstract

Despite Bisphenol-A (BPA) being subject to extensive study, a thorough understanding of molecular mechanism remains elusive. Here we show that using weighted gene correlation network analysis (WGCNA), which takes advantage of a graph theoretical approach to understanding correlations amongst genes and grouping genes into modules that typically have co-ordinated biological functions and regulatory mechanisms, that despite some commonality in altered genes, there is minimal overlap between BPA and estrogen in terms of network topology. We confirmed previous findings that ZNF217 and TFAP2C are involved in the estrogen pathway, and are implicated in BPA as well, although for BPA they appear to be active in the absence of canonical estrogen-receptor driven gene expression. Furthermore, our study suggested that PADI4 and RACK7/ZMYNDB8 may be involved in the overlap in gene expression between estradiol and BPA. Lastly, we demonstrated that even at low doses there are unique transcription factors that appear to be driving the biology of BPA, such as SREBF1. Overall, our data is consistent with other reports that BPA leads to subtle gene changes rather than profound aberrations of a conserved estrogen signaling (or other) pathways.

Highlights

Bisphenol A (BPA) is an industrial chemical used in the manufacture of polycarbonate plastic found in a number of consumer products such as thermal paper, canned foods and epoxy resins (Rubin, 2011) – many of these uses are being phased out (Zimmerman and Anastas, 2015)
Our analysis shows that while there is substantial overlap between genes altered by BPA and estrogen, which might imply that BPA is “estrogenic,” there are important differences in network topology as well as biological function, and that the overlap appears to be driven by transcription factors such as ZNF217, TFAP2C, PADI4, and RACK7/ZMYND8 rather than the estrogen receptor per se
We began by analyzing the dose-response curve of the BPA and estrogen dataset combined using weighted gene correlation network analysis (WGCNA) to look for a “consensus network”-a common pattern of genes that are correlated in all conditions

Summary

Introduction

Bisphenol A (BPA) is an industrial chemical used in the manufacture of polycarbonate plastic found in a number of consumer products such as thermal paper, canned foods and epoxy resins (Rubin, 2011) – many of these uses are being phased out (Zimmerman and Anastas, 2015). Exposure to BPA is widespread, with very low levels of BPA present in the majority of urinary samples taken in the general population (Calafat et al, 2008), serum levels are estimated to be lower (Teeguarden et al, 2013). Release of BPA to the environment exceeds one million pounds per year (Rubin, 2011). The CLARITY study, a three generation chronic study with low-levels of BPA used to mimic population exposures, involved 3,500 rats: while it resulted in no revision of safety standards by the FDA, it still failed to bring about a consensus as to a safe level (Academics Urge Caution in Interpreting Clarity-Bpa Results, n.d.). Despite the overwhelming amount of data, the mechanism(s) by which BPA may exert adverse effects remains unclear, nor is there a widely agreed upon endpoint on which to base a safe dose

Methods

Results

Conclusion