Jorge M. Naciff

Toward good read-across practice (GRAP) guidance

Summary Grouping of substances and utilizing read-across of data within those groups represents an important data gap filling technique for chemical safety assessments. Categories/analogue groups are typically developed based on structural similarity and, increasingly often, also on mechanistic (biological) similarity. While read-across can play a key role in complying with legislation such as the European REACH regulation, the lack of consensus regarding the extent and type of evidence necessary to support it often hampers its successful application and acceptance by regulatory authorities. Despite a potentially broad user community, expertise is still concentrated across a handful of organizations and individuals. In order to facilitate the effective use of read-across, this document presents the state of the art, summarizes insights learned from reviewing ECHA published decisions regarding the relative successes/pitfalls surrounding read-across under REACH, and compiles the relevant activities and guidance documents. Special emphasis is given to the available existing tools and approaches, an analysis of ECHAs published final decisions associated with all levels of compliance checks and testing proposals, the consideration and expression of uncertainty, the use of biological support data, and the impact of the ECHA Read-Across Assessment Framework (RAAF) published in 2015.

Framework for identifying chemicals with structural features associated with the potential to act as developmental or reproductive toxicants.

Developmental and reproductive toxicity (DART) end points are important hazard end points that need to be addressed in the risk assessment of chemicals to determine whether or not they are the critical effects in the overall risk assessment. These hazard end points are difficult to predict using current in silico tools because of the diversity of mechanisms of action that elicit DART effects and the potential for narrow windows of vulnerability. DART end points have been projected to consume the majority of animals used for compliance with REACH; thus, additional nonanimal predictive tools are urgently needed. This article presents an empirically based decision tree for determining whether or not a chemical has receptor-binding properties and structural features that are consistent with chemical structures known to have toxicity for DART end points. The decision tree is based on a detailed review of 716 chemicals (664 positive, 16 negative, and 36 with insufficient data) that have DART end-point data and are grouped into defined receptor binding and chemical domains. When tested against a group of chemicals not included in the training set, the decision tree is shown to identify a high percentage of chemicals with known DART effects. It is proposed that this decision tree could be used both as a component of a screening system to identify chemicals of potential concern and as a component of weight-of-evidence decisions based on structure-activity relationships (SAR) to fill data gaps without generating additional test data. In addition, the chemical groupings generated could be used as a starting point for the development of hypotheses for in vitro testing to elucidate mode of action and ultimately in the development of refined SAR principles for DART that incorporate mode of action (adverse outcome pathways).

Toxicogenomic Approach to Endocrine Disrupters: Identification of a Transcript Profile Characteristic of Chemicals with Estrogenic Activity

Public concerns have been raised in recent years over the possible adverse effects that may result from exposure to chemicals in the environment that have the potential to interfere with the normal function of the endocrine system in wildlife and humans (“endocrine disrupters”). Regulations have been established that require the testing of pesticides used in food crops and drinking water contaminants, for estrogenicity and other hormonal activities. In the United States, the U.S. EPA proposed the Endocrine Disrupter Screening Program, which consists of a Tier 1 screening battery of tests that is designed to identify chemicals capable of interacting with various hormonal systems, and different Tier 2 testing assays that are designed to verify and broaden the Tier 1 results. We identify 2 main problems with this approach: (1) the fact that the developmental stages that are the most susceptible to endocrine disruption are not represented in the screening tier, mainly because developmental effects tend to be latent, and there is no way to economically screen in developing models; and (2) the expense to screen each chemical to be included in this program. Thus, the need arises for an accurate, rapid, and cost effective method for assessing the potential endocrine activity of multiple chemicals during development. We hypothesize that the largely latent developmental effects of some endocrine disruptors are preceded by immediate changes in gene expression in the embryo and fetus. Therefore, an approach to assess the potential estrogenic (and other steroid hormonal) activity of different compounds is to identify those patterns of gene expression elicited in a tissue/organ exposed to these particular classes of chemicals. In this paper, the potential utility of such an approach for screening and better understanding of mechanism of action for specific chemicals with endocrine disrupter activities is presented, using as an example chemicals with estrogenic activity.

Case studies to test: A framework for using structural, reactivity, metabolic and physicochemical similarity to evaluate the suitability of analogs for SAR-based toxicological assessments

A process for evaluating analogs for use in SAR (Structure-Activity Relationship) assessments was previously published (Wu et al. 2010). Subsequently, this process has been updated to include a decision tree for estrogen binding (from US EPA) and flags for developmental and reproductive toxicity (DART). This paper presents the results of blinded case studies designed to test this updated framework. The results of these case studies support the conclusion that the process outlined by Wu et al. (2010) can be successfully applied to develop surrogate values for risk assessment. The read across results generated by the process were shown to be protective when compared to the actual toxicity data. Successful application of the approach requires significant expertise as well as discipline to not overstep the boundaries of the defined analogs and the rating system. The end result of this rigor can be the inability to read across all endpoints for all chemicals resulting in data gaps that cannot be filled using read across, however, this reflects the current state of the science and is preferable to making non-protective decisions. Future work will be targeted towards expanding read across capabilities. Two examples of a broader category approach are also shown.

Impact of the phytoestrogen content of laboratory animal feed on the gene expression profile of the reproductive system in the immature female rat

The effect of the dietary background of phytoestrogens on the outcome of rodent bioassays used to identify and assess the reproductive hazard of endocrine-disrupting chemicals is controversial. Phytoestrogens, including genistein, daidzein, and coumestrol, are fairly abundant in soybeans and alfalfa, common ingredients of laboratory animal diets. These compounds are weak agonists for the estrogen receptor (ER) and, when administered at sufficient doses, elicit an estrogenic response in vivo. In this study, we assessed the potential estrogenic effects of dietary phytoestrogens at the gene expression level, together with traditional biologic end points, using estrogen-responsive tissues of the immature female rat. We compared the gene expression profile of the uterus and ovaries, as a pool, obtained using a uterotrophic assay protocol, from intact prepubertal rats fed a casein-based diet (free from soy and alfalfa) or a regular rodent diet (Purina 5001) containing soy and alfalfa. Estrogenic potency of the phytoestrogen-containing diet was determined by analyzing uterine wet weight gain, luminal epithelial cell height, and gene expression profile in the uterus and ovaries. These were compared with the same parameters evaluated in animals exposed to a low dose of a potent ER agonist [0.1 μg/kg/day 17α-ethynyl estradiol (EE) for 4 days]. Exposure to dietary phytoestrogens or to a low dose of EE did not advance vaginal opening, increase uterine wet weight, or increase luminal epithelial cell height in animals fed either diet. Although there are genes whose expression differs in animals fed the soy/alfalfa-based diet versus the casein diet, those genes are not associated with estrogenic stimulation. The expression of genes well known to be estrogen regulated, such as progesterone receptor, intestinal calcium-binding protein, and complement component 3, is not affected by consumption of the soy/alfalfa-based diet when assessed by microarray or quantitative reverse transcriptase–polymerase chain reaction analysis. Our results indicate that although diet composition has an impact on gene expression in uterus and ovaries, it does not contribute to the effects of an ER agonist.

Predicting developmental toxicity through toxicogenomics

Global analysis of gene expression in target cells or tissues in response to a toxicant holds significant promise for predictive toxicology. Toxicants elicit a characteristic pattern of gene expression that is dependent on mechanism of action. These mechanism-specific transcript profiles can be used as the basis for predictive toxicology. Potential applications include prioritizing chemicals for testing and customizing testing approaches based on the chemical. Results that are useful in this predictive context can be obtained from animal or in vitro models. Gene expression analysis can also be used to elucidate the shape of the dose-response curve at exposure levels below the no observed adverse effect level, an important need in risk assessment. In this review, we will illustrate each of these points using our research on estrogen and an estrogenic mode of action as a model for how to use gene expression data in a predictive way. Although gene expression in response to estrogens is tissue, life stage, and sex specific, it is feasible to identify transcript profiles that are diagnostic of this mode of action.

Effects of Transplacental 17-α-Ethynyl Estradiol or Bisphenol A on the Developmental Profile of Steroidogenic Acute Regulatory Protein in the Rat Testis

Previous research from our laboratory has determined the transcript profiles for developing fetal rat female and male reproductive tracts following transplacental exposure to estrogens. Prenatal exposure to bisphenol A (BPA) or 17-α-ethynyl estradiol (EE) significantly affects steroidogenic acute regulatory (StAR) protein transcript levels in the developing male rat reproductive tract. The purpose of this study was to establish the intratesticular distribution and temporal expression pattern of StAR, a key gene involved in steroidogenesis. Beginning on gestation day (GD) 11, pregnant Sprague-Dawley rats were exposed daily to 10μg/kg/day EE and fetal testes were harvested at GD16, 18, or 20. Quantitative reverse transcriptase PCR (QRT-PCR) demonstrated no significant difference in StAR transcript levels present at GD16. However, at GD18, StAR transcripts were significantly decreased following exposure. Immunohistochemistry demonstrated similar StAR protein levels in interstitial region of GD16 testes and an obvious decrease in StAR protein levels in the interstitial region of GD18 testes. Moreover, starting at GD11 additional dams were dosed with 0.001 or 0.1 μg/kg/day EE or 0.02, 0.5, 400 mg/kg/day BPA via subcutaneous injections. QRT-PCR validated previous microarray dose-related decreases in StAR transcripts at GD20, whereas immunohistochemistry results demonstrated decreases in StAR protein levels in the interstitial region at the highest EE and BPA doses only. Neither EE nor BPA exposure caused morphological changes in the developing seminiferous cords, Sertoli cells, gonocytes, or the interstitial region or Leydig cells at GD16-20. High levels of estrogens decrease StAR expression in the fetal rat testis during late gestation.

The Genomic Response of a Human Uterine Endometrial Adenocarcinoma Cell Line to 17α-Ethynyl Estradiol

We have determined the gene expression profile induced by 17 alpha-ethynyl estradiol (EE) in Ishikawa cells, a human uterine-derived estrogen-sensitive cell line, at various doses (1 pM, 100 pM, 10 nM, and 1 microM) and time points (8, 24, and 48 h). The transcript profiles were compared between treatment groups and controls (vehicle-treated) using high-density oligonucleotide arrays to determine the expression level of approximately 38,500 human genes. By trend analysis, we determined that the expression of 2560 genes was modified by exposure to EE in a dose- and time-dependent manner (p </= 0.0001). The annotation available for the genes affected indicates that EE exposure results in changes in multiple molecular pathways affecting various biological processes, particularly associated with development, morphogenesis, organogenesis, cell proliferation, cell organization, and biogenesis. All of these processes are also affected by estrogen exposure in the uterus of the rat. Comparison of the response to EE in both the rat uterus and the Ishikawa cells showed that 71 genes are regulated in a similar manner in vivo as well as in vitro. Further, some of the genes that show a robust response to estrogen exposure in Ishikawa cells are well known to be estrogen responsive, in various in vivo studies, such as PGR, MMP7, IGFBP3, IGFBP5, SOX4, MYC, EGR1, FOS, CKB, and CCND2, among others. These results indicate that transcript profiling can serve as a viable tool to select reliable in vitro systems to evaluate potential estrogenic activities of target chemicals and to identify genes that are relevant for the estrogen response.

Design of a Microsphere-Based High-Throughput Gene Expression Assay to Determine Estrogenic Potential

Recently gene expression studies have been multiplied at an accelerated rate by the use of high-density microarrays. By assaying thousands of transcripts at a time, microarrays have led to the discovery of dozens of genes involved in particular biochemical processes, for example, the response of a tissue/organ to a given chemical with therapeutic or toxic properties. The next step in these studies is to focus on the response of a subset of relevant genes to verify or refine potential therapeutic or toxic properties. We have developed a sensitive, high-throughput gene expression assay for this purpose. In this assay, based on the Luminex xMAP system, carefully selected oligonucleotides were covalently linked to fluorescently coded microspheres that are hybridized to biotinylated cRNA followed by amplification of the signal, which results in a rapid, sensitive, multiplexed assay platform. Using this system, we have developed an RNA expression profiling assay specific for 17 estrogen-responsive transcripts and three controls. This assay can evaluate up to 100 distinct analytes simultaneously in a single sample, in a 96-well plate format. This system has improved sensitivity versus existing microsphere-based assays and has sensitivity and precision comparable with or better than microarray technology. We have achieved detection levels down to 1 amol, detecting rare messages in complex cRNA samples, using as little as 2.5 μg starting cRNA. This assay offers increased throughput with decreased costs compared with existing microarray technologies, with the trade-off being in the total number of transcripts that can be analyzed.

Grouping 34 Chemicals Based on Mode of Action Using Connectivity Mapping

Connectivity mapping is a method used in the pharmaceutical industry to find connections between small molecules, disease states, and genes. The concept can be applied to a predictive toxicology paradigm to find connections between chemicals, adverse events, and genes. In order to assess the applicability of the technique for predictive toxicology purposes, we performed gene array experiments on 34 different chemicals: bisphenol A, genistein, ethinyl-estradiol, tamoxifen, clofibrate, dehydorepiandrosterone, troglitazone, diethylhexyl phthalate, flutamide, trenbolone, phenobarbital, retinoic acid, thyroxine, 1α,25-dihydroxyvitamin D3, clobetasol, farnesol, chenodeoxycholic acid, progesterone, RU486, ketoconazole, valproic acid, desferrioxamine, amoxicillin, 6-aminonicotinamide, metformin, phenformin, methotrexate, vinblastine, ANIT (1-naphthyl isothiocyanate), griseofulvin, nicotine, imidacloprid, vorinostat, 2,3,7,8-tetrachloro-dibenzo-p-dioxin (TCDD) at the 6-, 24-, and 48-hour time points for 3 different concentrations in the 4 cell lines: MCF7, Ishikawa, HepaRG, and HepG2 GEO (super series accession no.: GSE69851). The 34 chemicals were grouped in to predefined mode of action (MOA)-based chemical classes based on current literature. Connectivity mapping was used to find linkages between each chemical and between chemical classes. Cell line-specific linkages were compared with each other and to test whether the method was platform and user independent, a similar analysis was performed against publicly available data. The study showed that the method can group chemicals based on MOAs and the inter-chemical class comparison alluded to connections between MOAs that were not predefined. Comparison to the publicly available data showed that the method is user and platform independent. The results provide an example of an alternate data analysis process for high-content data, beneficial for predictive toxicology, especially when grouping chemicals for read across purposes.