Abstract:
Anthropogenic activities have drastically altered chemical exposure, with traces of
synthetic chemicals detected ubiquitously in the environment. Many of these chemicals
are thought to perturb endocrine function, leading to declines in reproductive health and
fertility, and increases in the incidence of cancer, metabolic disorders and diabetes.
There are over 90 million unique chemicals registered under the Chemical Abstracts
Service (CAS), of which only 308,000 were subject to inventory and/or regulation, in
September 2013. However, as a specific aim of the EU REACH regulations, the UK is
obliged to reduce the chemical safety initiatives reliance on in vivo apical endpoints,
promoting the development and validation of alternative mechanistic methods. The
human health cost of endocrine disrupting chemical (EDC) exposure in the EU, has
been estimated at €31 billion per annum. In light of the EU incentives, this study aims
to evaluate current in silico and in vitro tools for EDC screening and hazard
characterisation; testing the hypothesis that in silico virtual screening accurately predicts
in vitro mechanistic assays. Nuclear receptor binding interactions are the current focus
of in silico and in vitro tools to predict EDC mechanisms. To the author’s knowledge,
no single study has quantitatively assessed the relationship between in silico nuclear
receptor binding and in vitro mechanistic assays, in a comprehensive manner.
Tripos ® SYBYL software was used to develop 3D-molecular models of nuclear
receptor binding domains. The ligand binding pockets of estrogen (ERα and ERβ),
androgen (AR), progesterone (PR) and peroxisome proliferator activated (PPARγ)
receptors were successfully modelled from X-ray crystal structures. A database of
putative-EDC ligands (n= 378), were computationally ‘docked’ to the pseudo-molecular
targets, as a virtual screen for nuclear receptor activity. Relative to in vitro assays, the in
silico screen demonstrated a sensitivity of 94.5%. The SYBYL Surflex-Dock method
surpassed the OECD Toolbox ER-Profiler, DfW and binary classification models, in
correctly identifying endocrine active substances (EAS). Aiming to evaluate the current
in vitro tools for endocrine MoA, standardised ERα transactivation (HeLa9903), stably
transfected AR transactivation (HeLa4-11) assays in addition to novel transiently
transfected reporter gene assays, predicted the mechanism and potency of test
substances prioritised from the in silico results (n = 10 potential-EDCs and 10 hormone
controls). In conclusion, in silico SYBYL molecular modelling and Surflex-Dock
virtual screening sensitively predicted the binding of ERα/β, AR, PR and PPARγ
potential EDCs, and was identified as a potentially useful regulatory tool, to support
EAS hazard identification.