In addition to the three ECVAM-validated teratogenicity assays which have been available for some years now, many other in vitro assays and computational methods to predict developmental toxicity have been proposed or are the subject of active research. Of particular note is the trend toward using combinations of these assays in an integrated strategy so as to cover more of the potential mechanisms and adverse outcomes comprising developmental toxicity. Some of the major approaches currently being investigated are as follows:
Model organisms. The use of model lower organisms such as C. elegans (a nematode worm) and Drosophila melanogaster (the fruit fly) that contain well-defined and conserved genes shown to be involved in development has led to the use of these models as preliminary screens for developmental toxicants (NRC, 2000). The zebrafish in particular has become quite popular for this purpose as evidenced by its use in the pharmaceutical industry (Ball et al., 2014) as well as in the US EPA’s ToxCast™ high throughput chemical screening program (Padilla et al., 2012). The zebrafish embryo also has been used extensively as an investigative research tool due to the ease of generating transgenic mutants and gene knock-down models for a wide range of physiological and pathological conditions.
Although C. elegans is well established as a model for genetic studies, it is just beginning to emerge as a model for toxicity testing. Scientists at NIEHS have developed a high-throughput screen in C. elegans to assess the effects of chemical exposure on reproductive efficiency, feeding, and growth. However, several key differences between C. elegans and mammals need to be appreciated when considering this model, among them: 1) approximately one third of the organism is comprised of neurons, 2) the worms have a thick outer cuticle which forms a barrier for certain hydrophilic chemicals, and 3) they lack certain metabolizing enzymes, such as the CYP1 family.
In vitro functional assays in mammalian cells. A number of new assays of relevance to developmental toxicology were advanced as products of a European Union framework project called “ReproTect,” which ran from 2004-2009. The approach taken was to break down the reproductive cycle into discrete segments (fertility, implantation, prenatal development) and develop targeted assays to assess specific functions within these segments. The program resulted in new or refined assays involving placental trophoblast, uterine endometrium and oviductal epithelial cells, as well as embryonic stem cells.
Gene transcript/metabolite profiling. Several investigators have utilized gene expression (Pennings et al., 2011) or metabolomics (Kleinstreuer et al., 2011; Kameoka et al., 2014) as a means of increasing the data content derived from cell-based assays. These additional profiling data could be used to identify putative modes of action, to establish biomarkers for a particular type of effect or mode of action, or they might be used as fingerprints to compare with the profiles of reference chemicals to determine degree of similarity in biological response. The latter could be used to support the process of “read-across,” in which the toxicity of a data-poor chemical is inferred from that of a data-rich analog in order to avoid unnecessary animal testing.
Computational approaches. Quantitative structure-activity relationship (QSAR) models for developmental end points can work fairly well within certain chemistries, but at this time are generally limited in their ability to predict across diverse chemical structures (reviewed in Pipero and Worth, 2010). Another approach is qualitative SAR based on structural alerts in combination with expert toxicological judgement. The latter approach does not have fine discriminating power within a class of chemicals, but has been used for initial screening, prioritization, and grouping of analogs for read-across. An example of one such system is the developmental and reproductive toxicology decision tree framework created by Wu et al. (2010; 2013). The system is based on 25 major chemical categories, enriched for chemicals known to cause reproductive and developmental toxicity via known mechanisms, such as nuclear hormone receptor ligands, prostaglandin receptor ligands, mediators of sonic hedgehog signaling, etc. The system was recently evaluated in a series of read-across case studies and appeared to work well for this purpose (Blackburn et al., 2011).
The US EPA’s National Center for Computational Toxicology (NCCT) is building computational models of embryogenesis with the ultimate goal of using them to predict the effects of chemical exposure on the embryo. These multi-scale models draw on knowledge of molecular regulatory circuits which control basic processes in normal embryogenesis. Biological assay data on specific chemicals are fed into the models to predict potential adverse effects of exposure. Use of these models for regulatory purposes is likely many years away, but the concept is quite provocative. More information can be found at The Virtual Embryo Project (v-Embryo™).
Researchers at NCCT have also published predictive models for developmental toxicity based on correlations between high throughput in vitro and biochemical assays in the ToxCast™ program and in vivo prenatal developmental toxicity test guideline studies, the data from which are housed in the ToxRefDB database (Sipes et al., 2011).
Test methods for reproductive and developmental toxicity testing currently under development and/or validation at EURL ECVAM will be described on their Reproductive Toxicity page (currently under construction).
It has now become apparent that for highly complex end points such as developmental toxicity, integrated testing strategies which combine computational models, in vitro assays, and/or ‘omics methods will likely be needed. Considerations for the design of integrated testing strategies for developmental toxicity were recently reviewed by Basketter et al. (2012). An emerging concept is to design these testing strategies based on the knowledge housed in “Adverse Outcome Pathways (AOPs),” which describe a series of events linking initial interactions with a chemical and a biological target at the molecular level, to downstream changes at higher levels of biological organization, ultimately leading to adverse outcomes (see OECD AOPs programme). Assays can then be designed to evaluate the initial or intermediate events of the AOPs in order to predict downstream adverse outcomes. There currently is a strong need to develop integrated testing strategies for developmental toxicity, so that they can be used to reduce the heavy burden of animal testing for programs such as REACH, the Biocidal Products Directive, and many others.
The following invited commentaries provide the AltTox community with new perspectives on reproductive and developmental toxicity testing:
Edward W. Carney, PhD
Dow Chemical Company (deceased)
AltTox Editorial Board reviewer(s):
Martin L. Stephens, PhD
Johns Hopkins University, Center for Alternatives to Animal Testing