An exposure:activity profiling method for interpreting high ... - Core

Regulatory Toxicology and Pharmacology 71 (2015) 398–408

Contents lists available at ScienceDirect

Regulatory Toxicology and Pharmacology journal homepage: www.elsevier.com/locate/yrtph

An exposure:activity profiling method for interpreting high-throughput screening data for estrogenic activity—Proof of concept Richard A. Becker a,⇑, Katie Paul Friedman b, Ted W. Simon c, M. Sue Marty d, Grace Patlewicz e,1, J. Craig Rowlands d a

American Chemistry Council, 700 2nd St., NE, Washington, DC, United States Bayer CropScience LP, Research Triangle Park, NC, United States Ted Simon LLC, Winston, GA, United States d The Dow Chemical Company, Midland, MI, United States e DuPont Haskell Global Centers for Health and Environmental Sciences, Newark, DE, United States b c

a r t i c l e

i n f o

Article history: Received 8 September 2014 Available online 3 February 2015 Keywords: Endocrine disruption Estrogen assays Exposure:activity ratio Relative estrogenic exposure:activity quotient Genistein ToxCast™ Tox21 Exposure Potency

a b s t r a c t Rapid high throughput in vitro screening (HTS) assays are now available for characterizing dose– responses in assays that have been selected for their sensitivity in detecting estrogen-related endpoints. For example, EPA’s ToxCast™ program recently released endocrine assay results for more than 1800 substances and the interagency Tox21 consortium is in the process of releasing data for approximately 10,000 chemicals. But such activity measurements alone fall short for the purposes of priority setting or screening because the relevant exposure context is not considered. Here, we extend the method of exposure:activity profiling by calculating the exposure:activity ratios (EARs) using human exposure estimates and AC50 values for a range of chemicals tested in a suite of seven estrogenic assays in ToxCast™ and Tox21. To provide additional context, relative estrogenic exposure:activity quotients (REEAQ) were derived by comparing chemical-specific EARs to the EAR of the ubiquitous dietary phytoestrogen, genistein (GEN). Although the activity of a substance in HTS-endocrine assays is not a measure of health hazard or risk, understanding how such a dose compares to human exposures provides a valuable additional metric that can be used in decision-making; substances with small EARs and REEAQs would indicate low priority for further endocrine screening or testing. Ó 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction The intent to reduce, refine, and in some instances, replace the use of animals in toxicity testing has spurred the development of high-throughput screening (HTS) assays that eventually may be used for chemical risk assessment (Judson et al., 2011, 2014; Thomas et al., 2013). However, to date, many of these in vitro approaches have demonstrated limited applicability for predicting in vivo hazard using statistical classification methods (Thomas

Abbreviations: HTS, high throughput screening; EDSP, EPA’s Endocrine Disruptor Screening Program; ER, estrogen receptor; EAR, exposure:activity ratio; AC50, concentration at half maximal activity; REEAQ, relative endocrine exposure:activity quotients; GEN, genistein; BMD, bench mark dose; BPA, Bisphenol-A. ⇑ Corresponding author. Fax: +1 202 478 2503. E-mail address: [email protected] (R.A. Becker). 1 Current address: National Center for Computational Toxicology, U.S. Environmental Protection Agency, 109 T.W. Alexander Drive, Research Triangle Park, NC 27711, United States.

et al., 2012). Initially, it was proposed that the use of classification-based prediction models derived from HTS assays results were capable of matching the in vivo estrogen and androgen results of the Endocrine Disruptor Screening Program Tier 1 battery with high balanced accuracies approaching 100%, with notable data gaps for thyroid and steroidogenesis pathways (Rotroff et al., 2013a, 2014; Cox et al., 2014). However, when predictive models were developed from the Rotroff et al. (2013a) dataset using a cross-validation approach, the balanced accuracies for predicting in vivo endpoints fell to 85% for estrogen and 79% for androgen pathways (Cox et al., 2014). Recently, an approach that combines results from different estrogen pathway-based HTS assays into an ER Interaction Score has been suggested as a potential tool for prioritizing chemicals for further endocrine screening (Rotroff et al., 2014). Using a limited set of known positives and known negatives, the ER Interactions Score methodology was shown to have 91% sensitivity and 65% specificity (Ibid.). It is evident from these early efforts to prioritize chemicals for further estrogen pathway-related

http://dx.doi.org/10.1016/j.yrtph.2015.01.008 0273-2300/Ó 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

effects that more exposure context needs to be provided to permit a transparent, science-based approach for identifying priority chemicals. There are even larger barriers, including a lack of appropriate HTS assays, for the use of HTS data to predict and prioritize substances for potential activity in thyroid and steroidogenesis pathways, (Rotroff et al., 2013a, Paul et al., 2014). Despite these limitations, such HTS methods hold great promise for improving the biological basis for priority setting and chemical screening, and may critically support targeted testing with a tiered, riskbased framework (Thomas et al., 2013; Pastoor et al., 2014). While application of HTS results to predict adverse effects for use in risk assessment may be futuristic, scientific support is growing for more immediate applications in priority setting, particularly for endocrine screening, such as within EPA’s tiered testing and assessment approach, the Endocrine Disruptor Screening Program (EDSP). EPA has articulated its projected use of HTS methods in its ‘‘EDSP21’’ vision, which projects utilization of HTS approaches initially for priority setting, and then as experience is gained, and commensurate with achieving the requisite degree of scientific confidence, as replacements for specific EDSP screens and tests (USEPA, 2011, 2014b). Early efforts have been made to characterize the prediction landscape provided by HTS assay data from the ToxCast™ and Tox21 programs; these efforts were aimed at providing methods for ranking chemicals by various metrics. Reif et al. (2010) presented results of ToxCast™ endocrine screening using the ToxPi visualization and ranking tool. This endocrine ToxPi assembled the results of related assays into specific sections of a pie chart, and then for each compound, the area reflected its activity in a specific set of related ToxCast™ assays, with larger areas suggesting increasing hazard and/or potency. However, ToxPi does not employ standard reference compounds for normalizing responses, and instead normalizes each analysis in relation to the compound with the highest level of activity among the set of substances evaluated (Patlewicz et al., 2013). This relativistic approach falls short by not providing potencies benchmarked to reference substances, and therefore the predicted responses cannot be prioritized within the context of a well-characterized biological effect. In the absence of this context for priority setting using endocrinerelated HTS data, the point at which a potential biological effect falls below a regulatory level of concern becomes unclear, and the relative priorities may even be inconsistent with the large database of available toxicology data from animal studies. The strengths of the recent ER Interaction Score method of Rotroff et al. (2014) include incorporation of data from 13 assays indicative of ER signaling, but a primary weakness is that it uses an approach which compresses the relative potency information for screened chemicals, thereby limiting the utility of this method for priority setting based on integrated consideration of relative potency, dosimetry, and exposure information. The USEPA presented the integrated bioactivity and exposure ranking (IBER) in December 2014 (USEPA, 2014b), a ratio based on a model score similar to the ER Interaction Score and predicted human exposure, for priority-setting; implementation of this approach will likely require determination of IBER values that fall below a level of concern for further testing. The need to consider exposure information along with measures of biologic effects in the initial stages of a risk assessment was articulated in the RISK21 roadmap (Pastoor et al., 2014). Using an approach such as the RISK21 roadmap that is problem formulation based, exposure driven, and expresses the intersection of exposure and effect is an effective means for prioritizing chemicals for further assessment (Pastoor et al., 2014). Thus, we propose a profiling approach that incorporates standardization to a reference chemical such that the level of concern is rooted in transparent methodology and becomes obvious and evident. This proposed exposure:activity profiling approach reflects standardization tools used historically in toxicology, such

399

as dioxin toxicity equivalence factors and environmental estrogen equivalents (Giesy et al., 2002; Van den Berg et al., 1998, 2006), but goes beyond toxicity equivalence by incorporating exposure information rather than relying solely on a relative measure of potency or toxicity. Accordingly, the priority status or score of a chemical would reflect both exposure and potency (Borgert et al., 2012, 2013). This score would be presented relative to that of a chosen reference chemical to provide appropriate context. In summary, the exposure:activity profiling approach we demonstrate herein supports the use of HTS data in prioritization tasks.

2. Exposure:activity profiling Here, as a proof of concept, we illustrate how to apply the method of exposure:activity profiling (Becker et al., 2014) to HTS endocrine results by calculating exposure:activity ratios (EARs). This method is a variation of that presented previously (Wetmore et al., 2012, 2013). To demonstrate this method, we selected a set of results from seven HTS assays for estrogen receptor interaction in the ToxCast™ and Tox21 assay battery. Because this is a proof of concept exercise, substances were chosen based on the availability of (1) estimates of current human oral exposures and (2) calculated human oral equivalent doses corresponding to in vitro HTS AC50 values (these values were obtained from Supplemental Table 8 of Wetmore et al., 2012). Thus, EARs were developed using human exposure estimates expressed as oral doses in mg/kg/d in comparison to predicted oral equivalent doses, also in mg/kg/d, corresponding to in vitro activity in ToxCast™ assays (e.g., AC50 or LEC50 values) (Wetmore et al., 2012). If a chemical was negative in an assay the EAR was defaulted to zero. This approach enabled us to readily calculate EARs without having to independently derive oral equivalent values. The operation and source of the assays is described in detail elsewhere (Judson et al., 2010; Rotroff et al., 2013a,b, 2014), as are descriptions of the publicly available chemical library and quality control measures (Sipes et al., 2013; http://www.epa.gov/ncct/dsstox/). The derivation of assay AC50 values is reported on the ToxCast website (http://epa.gov/ncct/toxcast/data.html, December 2013 release). The in vitro HTS bioactivity data were not corrected using a cytotoxicity filter and reflect the publicly-available data from ToxCast™ Phase II at the time the database was accessed. In enterocytes, in liver, and in other tissues in vivo many chemicals with demonstrated estrogenic activity in vitro are inactivated by conjugation to the glucuronide or sulfate (Strassburg et al., 2002; Guillemette et al., 2004). Hence, the steady state concentrations in plasma used to determine the EAR will be the bioactive aglycone rather than the total that includes conjugated forms. When using in vitro data as the source of activity concentrations, the exposure estimate must be represented as either steady state plasma concentrations derived from blood or urinary concentrations, or from oral equivalent doses (see SF-1 in Supplemental Information for a schematic diagram of the three dosimetric levels—oral equivalents or external dose, steady state plasma concentrations, and urinary excretion values). Because the major portion of GEN, as well as that of many other estrogenic chemicals, exists in plasma as an inactive conjugate, it is necessary to express both the exposure and activity levels as steady state plasma levels of the active moiety. This approach also requires that urinary excretion values of the chemical of interest be expressed as steady state plasma levels of the bioactive aglycone. For estimating activity values from in vivo animal data, Becker et al. (2014) used a BMD value from the uterotrophic assay to obtain the corresponding biomonitoring equivalent (BE) value; thus, they expressed an external dose as a urinary concentration. This enabled comparison of the human exposure values (mg/L in

400

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

urine from human biomonitoring studies) to the concentration in urine corresponding to the uterotrophic BMD. For estimating activity values, Wetmore et al. (2012) used an equation that included plasma protein binding as well as renal and metabolic clearance to predict a steady state blood concentration and a unitary dose rate of 1 mg/kg/d; then, the ratio between the AC50 value from a ToxCast™ assay and the predicted steady state concentration at the unit dose provided the oral equivalent dose representing the AC50 value from the assay. Since the Wetmore et al. (2012) report did not include an oral equivalent dose calculation for GEN, the EAR values for GEN are based on comparison of steady state plasma levels. The EAR values for other chemicals will be based on comparison of oral-equivalent doses. Eqs. (1)–(3) below provide the details of the calculations.

EAR ðunitlessÞ ¼

Exposure ðoral equiv:dose in mg=kg=dÞ Activity ðAC50 oral equiv:dose in mg=kg=dÞ ð1Þ

EARGEN ðunitlessÞ ¼

GEN Exposure ðsteady state blood conc:in lMÞ GEN Activity ðAC50 in lMÞ ð2Þ

REEAQ Chem: ¼

EARChem: EARGEN

ð3Þ

For GEN, steady state blood concentrations of the active unconjugated compound were developed from measurements reported in the literature (Arai et al., 2000; Busby et al., 2002; Cimino et al., 1999; Grace et al., 2003, 2004; Liu et al., 2013; Richelle et al., 2002; Setchell et al., 2003, 2011; Shelnutt et al., 2000, 2002; Shimazu et al., 2011; Takashima et al., 2004; Teeguarden et al., 2011; Völkel et al., 2002; Yuan et al., 2012; Zhang et al., 2003). To obtain the EAR value for GEN, the estimated steady state plasma concentration GEN in human serum (see below) was divided by the AC50 value from each of the assays considered. 3. Derivation of steady state concentrations of GEN in human serum Because the major portion of GEN, as well as that of many other putative estrogenic chemicals, exists in plasma as inactive conjugates, expressing both the exposure and activity levels as steady state plasma levels of the active moiety is desirable. However, this approach requires that urinary excretion values of the chemical of

interest also be expressed as steady state plasma levels. This particular IVIVE calculation proved challenging for bisphenol-A (BPA) because urinary values grossly overestimate the unconjugated active moiety, and this may also be the case for other chemicals as well (Teeguarden and Hanson-Drury, 2013; Teeguarden et al., 2005, 2011, 2013). Phase II glucuronidation and sulfation metabolically inactivate GEN. Unfortunately, most analyses for GEN use a pre-treatment of b-glucuronidase/sulfatase H-2 and report only total aglycone concentrations following such enzymatic digestion (Zhang et al., 2003). Early methods of analysis of both the aglycone and conjugates involved splitting the sample and using enzymes on one split and not the other (Cimino et al., 1999; Shelnutt et al., 2000, 2002). More recent studies employ LC/MS with 13C labeling, or HPLCdiode array detection to perform simultaneous analysis of the aglycone and metabolites (Hosoda et al., 2011; Setchell et al., 2011). These studies provide percentage values for the aglycone. The results and calculation methods are shown in Supplementary Information Table 1. From these three studies, a rounded value of 0.3% represents the aglycone in plasma. Four studies provided measurements of the fraction of total GEN in plasma existing as the aglycone (Busby et al., 2002; Metzner et al., 2009; Setchell et al., 2011; Yuan et al., 2012). These studies measured both the aglycone and the conjugate. Eight studies provided data for total GEN in plasma (Arai et al., 2000; Grace et al., 2004; Heald et al., 2006; Iwasaki et al., 2008; Ritchie et al., 2004; Setchell et al., 2003; Shimazu et al., 2011; Takashima et al., 2004). The free GEN was estimated by multiplying the total GEN value by 0.3%. Calculation methods and values are shown in Supplementary Information. The distributions of free GEN in plasma from measurements of free GEN (left) and total GEN (right) are depicted in Fig. 1. The median values for measurements of free GEN range from 0.001 to about 0.02 lM, whereas those estimated from total GEN measurements range from 0.00001 to 0.001 lM. As noted, a value for the unconjugated fraction of 0.003 was estimated from two more recent papers (Hosoda et al., 2011; Setchell et al., 2011). An earlier paper measured both free GEN before enzymatic digestion, and sulfate and glucuronide conjugates following digestion, and observed a value of 12.5% (Shelnutt et al., 2002). There are three possible explanations for this approximately 30fold difference resulting in free GEN values greater than total GEN values. First, in the four studies that measured free GEN, some ex vivo hydrolysis of conjugates may have occurred. Second, Hosoda et al. (2011) and Setchell et al. (2011) failed to measure

Fig. 1. Box plots of the distributions of free GEN in plasma from measurements of free GEN (left) and total GEN (right). Plots are presented on the same scale. The median values for free GEN range from 0.001 lM to about 0.02 lM, whereas those estimated from total GEN measurements range from 1E05 to 0.001 lM.

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

all of the free GEN in plasma, possibly due to ex vivo conjugation from enzymes occurring in plasma. Third, Shelnutt et al. (2002) may have overestimated the amount of free GEN because of contamination of samples or glassware with sulfatase/glucuronidase reagent. If one assumes the measurement of Shelnutt et al. (2002) is correct and would use 12% as a measure of free GEN from total GEN, then the two sets of measurements become more consistent. Here, a value of 0.005 lM will be used as the steady state plasma concentration representing GEN exposure. It is representative of the central value of the free GEN measurements and of the upper end of the total GEN measurements. 4. Deriving and interpreting EAR and REEAQ values from ToxCastTM and Tox21 assay results To calculate a REEAQ value as in Becker et al. (2014), two EAR values are required: one for the chemical of interest and the other for the reference chemical, in this case, GEN. For each of these EAR values, estimates of exposure and activity are needed. All four of the values required (two for each EAR) should be expressed in the same units and represent a comparable biological quantity, e.g., dose or tissue concentration. For the set of substances evaluated, results from one of the ToxCast™ assays (Attagene: human ERa reporter gene construct) are depicted in Fig. 2. The AC50 results are presented in (a), the EARs based on human exposures in (b), and the REEAQs (normalized to the EARGEN) in (c). The assay ATG_ERa_TRANS had the most positive chemical responses for which exposure information also existed. Assay negatives resulted in a default EAR value of zero. The chemicals were ordered by increasing EAR from left to right. This order was maintained in all the other plots in Figs. 3 and 4. Numerical values for EARs and REEAQs are provided in Supplementary Tables 3 and 4. In addition, exposure information was not available in Wetmore et al. (2012) for all chemicals tested; chemicals without exposure data were not included in the analysis. Among the chemicals included in the analysis BPA, boscalid, cyprodinil, ethalfluralin and fenhexamid each exhibited an REEAQ greater than 1 in one of the seven assays; 2-phenylphenol had REEAQ values greater than 1 in three of the seven assays. 5. Uncertainty in exposures estimates and in vitro-to-in vivo extrapolation (IVIVE) For exposure, estimation of oral-equivalent doses relied on the methods used by Wetmore et al. (2012). For chemicals other than GEN, IVIVE was required because the estrogen receptor-related bioactivity of interest were expressed as in vitro concentrations; Supplementary Information SF-1 shows a schematic for comparison of the methodology in Becker et al. (2014), Wetmore et al. (2012), and also that used here. Hence, EAR values for GEN were derived directly from an estimate of exposure measured as an in vivo blood concentration and the AC50 with both values in units of concentration. However, the EAR values for chemicals were derived using a generic IVIVE dosimetry model for activity and existing chemical-specific dosimetric models of varying sophistication and accuracy (LaKind et al., 2008). Hays et al. (2008) discuss some of the issues of estimating oral-equivalent doses from measurements of chemicals or their metabolites in urine, including length of half-life, identification and quantification of all metabolites and hydration status. Although the calculation of the two EAR values used to calculate a REEAQ value have different starting points, the utility of REEAQ values for priority screening and contextualizing results of HTS assays remains clear.

401

In applying the EAR and REEAQ methodologies beyond the proof of concept stage, additional consideration may be warranted for application of a statistical methodology to estimate the uncertainty in the EAR and REEAQ values. Wetmore et al. (2012) presented the range of exposure estimates obtained for various age- and gender-based subpopulations and compared these to the distribution of the oral equivalent dose ranges required to achieve the upper 95th percentile steady-state blood concentration across the suite of in vitro assays. More recently, EPA illustrated uncertainty by presenting the range of human exposure predictions (from the median to upper 95-percentile values) compared to the range from the median and minimum in vitro oral equivalent activity doses. Such visualizations can help to identify those EARs that have the greatest degree of uncertainty or variability. One should also keep in mind that the level of effort devoted towards characterizing uncertainty should be commensurate with the impact and importance of the decision, and in employing the EAR and REAAQ methods for priority setting for endocrine screening, use of median values may provide sufficient precision to support prioritization decisions. 6. Discussion The concept of relative potency has previously informed prioritization processes for risk assessment. Toxicity equivalent factors (TEFs) for human health risk assessment of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and dioxin-like compounds are derived from a consensus relative potency value, using TCDD as the index chemical (Van den Berg et al., 1998, 2006) to standardize the interpretation of AhR ligand responses by single chemicals or aggregated mixtures. Indeed, one of the noted limitations of this approach is that there are endogenous AhR ligands in healthy human diets, and the TEF approach does not establish a safe threshold based on dietary exposure (Safe, 1998; Van den Berg et al., 2006). Environmental estrogen equivalents (EEQs) have largely been used to evaluate the potential estrogenicity of environmental matrices via standardization to 17b-estradiol potency, i.e., assuming the same slope of the dose–response curve, in in vitro bioassays (Avbersek et al., 2011; Bermudez et al., 2010, 2012). Recently, a revision to the EEQ approach to derive assay-specific thresholds of safety was proposed by defining the amount of 17b-estradiol needed to obtain an effect in each assay (Jarosova et al., 2014). The REEAQ approach advances this precedent by standardizing estrogenic activity to a well-understood dietary isoflavone, and also incorporating potential human exposure, such that both exposure and activity potential are used to rank chemicals relative to genistein. This REEAQ approach thus provides a means for establishing a threshold for chemicals that warrant further investigation based on consideration of putative mode of action, potency, and exposure. A chemical with high estrogenic potency predicted from in vitro results, but with limited potential exposure, would thereby be deprioritized in this comparison. This REEAQ approach fulfills a need for high-throughput prioritization of potentially estrogenic chemicals for further testing in models of greater biological complexity and reliability. Using a read-across approach, the consensus REEAQ values could be used to prioritize chemicals for further confirmatory testing. 6.1. The value of the REEAQ in priority decisions Calculation of EAR and REEAQ values is relatively straightforward, assuming the activity and exposure values are available and can be expressed in the same units. The utility of exposure:activity profiling is readily seen in Fig. 2; striking differences are apparent between evaluating activity only (Fig. 2(a)) versus

402

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

Fig. 2. Plots of (A) activity (AC50), (B) EAR and (C) REEAQ for ATG_ER_TRANS ToxCast™ assay. GEN is shown in red.

evaluating both exposure and activity (Fig. 2(b) and (c)). Considering just the ATG_ERa_TRANS ToxCast™ assay for priority setting purposes, reliance on activity profiling would identify 20 of the 26 substances as having AC50 values in excess of genistein. But, when EARs are used, only 1 of the 26 substances has an EAR value

exceeding EARGEN. And more than 50% have EAR values 100 times lower than the EARGEN. There is variation in AC50 values across ToxCast™ and Tox21 assays (Figs. 3 and 4), the same rank order of potencies was not observed as in the EAR values. However, irrespective of the particular assays, the EAR and REEAQ methods can

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

403

Fig. 3. Ordered vertical bar plots for EAR (left column) and REEAQ values for the assays OT_ER_ERaERb_48, ATG_ERE_CIS, and OT_ERbERb_48. In all plots, GEN is shown in red.

be used to contextualize and ‘‘ground truth’’ ToxCast™ and Tox21 results. Since GEN is a component of a healthy diet, and we have used an estimate of the average GEN serum level in US population as the exposure value for determining EAR and REEAQ values, those substances with values substantially below the EARGEN would seem to be candidates for deprioritization with respect to the EDSP or similar endocrine screening/testing programs. Since risk is a function of both exposure and hazard, EAR and REEAQ metrics provide a scientifically sound approach for placing HTS results into context for use in setting priorities. It is important to note that, because GEN is only very weak or lacks estrogenic effects at typical exposures in humans (e.g., a high isoflavone diet induced only slight effects and no clinically relevant effects on vaginal epithelium or endometrium in postmenopausal women (Duncan et al., 1999)) the REEAQ methodology reflects a high degree of conservatism and is health protective for humans. To capture uncertainty in the REEAQ approach that uses oral exposure estimates derived from dosimetry models and activity

estimates derived from IVIVE, an arbitrary, policy-based threshold could be applied; for example, chemicals demonstrating REEAQ values greater than 0.1 could be prioritized for further testing. This REEAQ value could be adjusted based on obtaining greater confidence in the dosimetry and IVIVE models, resources available for continued toxicity testing, or other factors and could contribute to a value-of-information approach as part of EDSP21 prioritization (NRC, 2007, 2009). The challenges associated with IVIVE and dosimetry notwithstanding, the use of the REEAQ value provides a straightforward way to contextualize exposures to putative estrogenic substances. Because the EAR and REEAQ metrics incorporate indications of both potency and exposure, use of these metrics is preferred for priority setting decisions rather than the use of activity values by themselves (Morgan et al., 2013; Rotroff et al., 2014). This use of exposure context is especially important when considering results from sensitive in vitro HTS assays. The straightforward and scientifically logical method of linking activity of a compound in a HTS endocrine assay to the human exposure to that substance,

404

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

Fig. 4. Ordered vertical bar plots for EAR (left column) and REEAQ values for the assays ACEA_T47D_8hr_Negative, Tox21_ERA_LUC_BG1_Agonist, and Tox21_ERA_BLA_Agonist_ratio. In all plots, GEN is shown in red.

and benchmarking this to the potency and exposure of a normal dietary constituent, makes the REEAQ a useful communication tool in discussions with a wide range of audiences about potential concerns arising from detection of positive responses in HTS (or similar in vitro or in vivo) endocrine screening assays. 6.2. The AC50 is the preferred measure of activity Location and steepness are characteristics of every dose– response curve (Slob and Setzer, 2014). Receptor-based responses, such as those produced by endocrine-active substances, involve many steps (Ong et al., 2010). A general assumption about the dose–response curve for substances that act by receptor binding is that a first-order Hill function is generally able to provide a mathematical description of the dose–response (Chow et al., 2011). However, the assays considered here and other estrogen-

related ToxCast™ and Tox21 assays measure events at different points downstream of receptor binding. For example, ATG_ERa_TRANS, ATG_ERE_CIS, and Tox21_Era_LUC_BG1_Agonist measure changes in transcriptional activity subsequent to ER activation, though by orthogonal technologies, whereas the ACEA_T47D_80hr_Positive and ACEA_T47D_80hr_Negative assays measure estrogen-dependent cellular growth kinetics Rotroff et al. (2013b). These latter two assays measure cellular events much further downstream than the assays of ER transcriptional activity, and could be expected to have steeper dose–response curves (Chow et al., 2011; Ong et al., 2010). In fact, both a first- and a second-order Hill function fit the data for the Tox21_ERa_LUC_BG1_Agonist for 17b-estradiol equally well (Simon et al., 2014). Since a diverse set of assays based on measurement of different events along the ER activity pathway are available for determining potential estrogen receptor

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

interaction with xenobiotics, the most reliable measure of potency is the AC50. The importance of co-regulatory proteins in modulating estrogen-related gene expression is being recognized (Aarts et al., 2013; Houtman et al., 2012; Wang et al., 2013). Competition for coregulatory proteins may reduce the efficacy of gene expression responses at high ligand concentrations, a phenomenon known as squelching (Kraus et al., 1995; Zhang and Teng, 2001). In fact, squelching is apparent in the responses to GEN in measurements of cellular growth kinetics and responses at high doses had to be removed from the dose–response fit (Rotroff et al., 2013b). Coregulatory proteins can modulate both potency and efficacy in estrogen-induced gene expression (Jeyakumar et al., 2008, 2011; Simons, 2008; Simons and Kumar, 2013). Some have discussed using LEC, or lowest effective concentration, of a chemical as the measure of comparative activity for endocrine related endpoints assays (Rotroff et al., 2010, 2014), Because of the considerations of potency, efficacy and slope discussed above, an AC50 is the least variable dose–response metric, and thus affords greater certainty than other metrics along the dose–response curve. For all these reasons, the AC50 is the most appropriate metric to use as the activity component for EAR calculations. 6.3. Potential future uses of the REEAQ methodology for androgen, thyroid and steroidogenesis pathways For endocrine priority setting, the ToxCast™ and Tox21 assays have focused on evaluating estrogen, androgen, thyroid, and steroidogenesis pathways. The selection of substances for contextualizing androgen, thyroid and steroidogenesis assay results in a manner similar to GEN for estrogen assays results will require some additional investigation. Much continuing effort surrounds the integration of multiple assays of estrogenic activity into a single interpretation of the potential to drive estrogen receptor (ER) activity. The new approach of Rotroff et al. (2014) in which a composite ER Interaction Score is derived by integrating responses across ToxCast™ assays for estrogen receptor activity does just this. Though this model provides a ranked estrogen score using binding, agonist, antagonist, and ER-related growth signals, this score is unitless, and is not immediately available for utilization in exposure:activity profiling. Moreover, this method appears to compress measures of potency that differ by 1000–10,000-fold to a scale where scores differ only by a factor of 2–3-fold. Thus, a direct exposure:activity profiling approach would require incorporation of potencies and conversion of the ER interaction scores into concentrations that could be converted to oral equivalent doses using dosimetric approaches. While the ER Interaction Score solves the problem of using single assays with different strengths and vulnerabilities for false positives and negatives, it does not directly address the need to incorporate exposure information context for prioritization of further testing. A simpler, more straightforward approach would be to derive an EAR value from results of each assay and then calculate a mean EAR value for each substance across all the assays. Comparison could then be made amongst the mean EARs and the mean EARGEN. In this manner, strengths and limitations of the assays are normalized and exposure:activity profiling can be accomplished. While this proof of concept has shown the utility of the EAR and REEAQ approach for priority setting for estrogen activity, wider application poses a number of challenges. First and foremost, as discussed by Borgert et al. (2013), thresholds govern endocrine activity, and even though activities may be observed from in vitro estrogen assays, and EARs could theoretically be calculated, substances with low potencies are likely not to be biologically active in vivo. Thus, in addition to developing an REEAQ cutoff value for

405

deprioritization, consideration should also be given to developing activity cut off values for each hormonal system. Substances with activity values lower than these cutoff values could be deprioritized based on activity alone, and not require calculation of EAR values. To utilize the EAR and REEAQ approach for androgen and thyroid pathways, suitable reference agents must be identified. Previously, as an androgen reference material, we suggested investigating the possibility of using 3,3-diindolylmethane (formed during digestion of cruciferous vegetables), it is commercially available (e.g., SigmaAldrich) and has been described as a ‘‘naturally occurring pure androgen antagonist’’ (Le et al., 2003). To date, there are not enough HTS assay data to adequately classify putative thyrotoxicants, largely because not all modes of action for thyroid disruption are evaluated by current ToxCast™ or Tox21 assay tools (Rotroff et al., 2013a; Paul et al., 2014). For thyroid applications, a family of active compounds would likely be needed to address different molecular initiating events that result in thyroid perturbation, including thyroperoxidase inhibition, sodium-symporter inhibition, deiodinase inhibition, upregulation of hepatic catabolism, interference with serum binding proteins, and interactions with thyroid hormone receptors (Murk et al., 2013; Crofton, 2008). Glucosinolate metabolites, such as 5-vinyloxazolidine-2thione or various isothiocyanates/thiocyanates derived from numerous vegetables (e.g., broccoli, cabbage, cauliflower, radish, turnip) may be candidates for a thyroid iodine uptake inhibitor reference compound (Crozier et al., 2006). Soy isoflavones including genistein are known to inhibit thyroperoxidase in vivo and ex vivo, but only result in systemic thyroid disruption with coincidence of iodine deficiency (Doerge and Chang, 2002). It may be necessary to identify several reference chemicals in this manner in order to evaluate REEAQs for in vitro thyroid activity. Additional efforts to improve the available HTS assays for thyroid and steroidogenesis disruption, and to identify and evaluate potential reference substances for these pathways, should be a priority for EDSP21. Key to implementing exposure:activity profiling for priority setting and screening is overcoming the challenge of obtaining suitable exposure estimates. To address such challenges, considerable efforts have been devoted over the last few years in exploring high throughput exposure modeling approaches, including the ExpoCast (Wambaugh et al., 2013) and ExpoDat (Armitage et al., 2014) research programs. Integral to these approaches is research that translates bioactivity concentrations in HTS in vitro assays to concentrations of the chemical in the target tissue or blood in vivo (Yoon et al., 2014). Since development of exposure estimates is the rate-limiting step for expanding HTS exposure:activity profiling, research to develop and establish scientific confidence in rapid high throughput predictions of internal concentrations needs to be a priority. EPA is modifying the Stochastic Human Exposure and Dose Simulation model to enable coupling of high throughput modeling with toxicokinetic modeling to be used in a forward dosimetry process (USEPA, 2014a). This approach appears to have considerable promise. The approach could provide a useful basis for generating conservative, screening level, predictions of internal concentrations that can then be matched to HTS results from ToxCast™ and Tox21 assays for priority setting. While such screening level estimates of exposure may be suitable for many chemicals, particularly where exposure data is lacking, a more refined approach may be appropriate for other chemicals. Thus, even at the level of priority setting, consideration should be given to implementing a tiered exposure assessment framework, which encourages generation and use of refined exposure estimates/data as a substitute for modeled estimates, when such refined estimates/data can be shown to be relevant and reliable.

406

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

Conflict of interest and bias statement The authors had complete control over the design, conduct, interpretation, and reporting of the analyses included in this manuscript. The contents of this manuscript are solely the responsibility of the authors and do not necessarily reflect the views or policies of their employers. Ted Simon received funding to support this research from the American Chemistry Council (ACC). Katie Paul Friedman, Sue Marty, and Craig Rowlands are employed by companies engaged in chemical product manufacturing. Richard Becker is employed by ACC, a trade association of U.S. chemical manufacturers. Note: during development of this manuscript Grace Patlewicz was employed by a company engaged in chemical product manufacturing; subsequent to acceptance of this manuscript, she became an employee of the National Center for Computational Toxicology, U.S. Environmental Protection Agency. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.yrtph.2015.01. 008.

References Aarts, J.M.M.J.G., Wang, S., Houtman, R., van Beuningen, R.M.G.J., Westerink, W.M.A., Van De Waart, B.J., Rietjens, I.M.C.M., Bovee, T.F.H., 2013. Robust array-based coregulator binding assay predicting ERa-agonist potency and generating binding profiles reflecting ligand structure. Chem. Res. Toxicol. 26, 336–346. http://dx.doi.org/10.1021/tx300463b. Avbersek, M., Somen, J., Heath, E., 2011. Dynamics of steroid estrogen daily concentrations in hospital effluent and connected waste water treatment plant. J. Environ. Monitor.: JEM 13 (8), 2221–2226. http://dx.doi.org/10.1039/ c1em10147a. Arai, Y., Uehara, M., Sato, Y., Kimira, M., Eboshida, A., Adlercreutz, H., Watanabe, S., 2000. Comparison of isoflavones among dietary intake, plasma concentration and urinary excretion for accurate estimation of phytoestrogen intake. J. Epidemiol. 10, 127–135. Armitage, J.M., Wania, F., Arnot, J.A., 2014. Application of mass balance models and the chemical activity concept to facilitate the use of in vitro toxicity data for risk assessment. Environ. Sci. Technol. [Epub ahead of print] Becker, R.A., Hays, S.M., Kirman, C.R., Aylward, L.L., Wise, K., 2014. Interpreting estrogen screening assays in the context of potency and human exposure relative to natural exposures to phytoestrogens. Birth. Defects. Res. B. Dev. Reprod. Toxicol. http://dx.doi.org/10.1002/bdrb.21085. Bermudez, D.S., Gray Jr., L.E., Wilson, V.S., 2010. Modeling the interaction of binary and ternary mixtures of estradiol with Bisphenol A and Bisphenol AF in an in vitro estrogen-mediated transcriptional activation assay (T47D-KBluc). Toxicol. Sci.: off. J. Soc. Toxicol. 116 (2), 477–487. http://dx.doi.org/ 10.1093/toxsci/kfq156. Bermudez, D.S., Gray, L.E., Wilson, V.S., 2012. Modelling defined mixtures of environmental oestrogens found in domestic animal and sewage treatment effluents using an in vitro oestrogen-mediated transcriptional activation assay (T47D-KBluc). Int. J. Androl. 35, 397–406. http://dx.doi.org/10.1111/j.13652605.2012.01278.x. Borgert, C.J., Sargent, E.V., Casella, G., Dietrich, D.R., McCarty, L.S., Golden, R.J., 2012. The human relevant potency threshold: reducing uncertainty by human calibration of cumulative risk assessments. Regul. Toxicol. Pharmacol. 62, 313–328. http://dx.doi.org/10.1016/j.yrtph.2011.10.012. Borgert, C.J., Baker, S.P., Matthews, J.C., 2013. Potency matters: thresholds govern endocrine activity. Regul. Toxicol. Pharmacol. 67, 83–88. http://dx.doi.org/ 10.1016/j.yrtph.2013.06.007. Busby, M.G., Jeffcoat, A.R., Bloedon, L.T., Koch, M.A., Black, T., Dix, K.J., Heizer, W.D., Thomas, B.F., Hill, J.M., Crowell, J.A., Zeisel, S.H., 2002. Clinical characteristics and pharmacokinetics of purified soy isoflavones: single-dose administration to healthy men. Am. J. Clin. Nutr. 75, 126–136. Chow, C.C., Ong, K.M., Dougherty, E.J., Simons, S.S., 2011. Inferring mechanisms from dose-response curves. Methods Enzymol. 487, 465–483. http://dx.doi.org/ 10.1016/B978-0-12-381270-4.00016-0. Cimino, C.O., Shelnutt, S.R., Ronis, M.J., Badger, T.M., 1999. An LC–MS method to determine concentrations of isoflavones and their sulfate and glucuronide conjugates in urine. Clin. Chim. Acta 287, 69–82. Cox, L.A.T., Popken, D., Marty, M.S., Rowlands, J.C., Patlewicz, G., Goyak, K.O., Becker, R.A., 2014. Developing scientific confidence in HTS-derived prediction models: lessons learned from an endocrine case study. Regul. Toxicol. Pharmacol. 69, 443–450. http://dx.doi.org/10.1016/j.yrtph.2014.05.010.

Crofton, K.M., 2008. Thyroid disrupting chemicals: mechanisms and mixtures. Int. J. Androl. 31, 209–223. Crozier, A., Clifford, M.N., Ashihara, H., 2006. Plant Secondary Metabolites, Occurrence, Structure and Role in the Human Diet. Blackwell Publishing Ltd., Oxford, U.K.. Doerge, D.R., Chang, H.C., 2002. Inactivation of thyroid peroxidase by soy isoflavones, in vitro and in vivo. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 777, 269–279. Duncan, A.M., Underhill, K.E., Xu, X., Lavalleur, J., Phipps, W.R., Kurzer, M.S., 1999. Modest hormonal effects of soy isoflavones in postmenopausal women. J. Clin. Endocrinol. Metab. 84, 3479–3484. Giesy, J.P., Hilscherova, K., Jones, P.D., Kannan, K., Machala, M., 2002. Cell bioassays for detection of aryl hydrocarbon (AhR) and estrogen receptor (ER) mediated activity in environmental samples. Mar. Pollut. Bull. 45, 3–16. Grace, P.B., Taylor, J.I., Botting, N.P., Fryatt, T., Oldfield, M.F., Al-Maharik, N., Bingham, S.A., 2003. Quantification of isoflavones and lignans in serum using isotope dilution liquid chromatography/tandem mass spectrometry. Rapid Commun. Mass Spectrom. 17, 1350–1357. http://dx.doi.org/10.1002/rcm.1059. Grace, P.B., Taylor, J.I., Low, Y.-L., Luben, R.N., Mulligan, A.A., Botting, N.P., Dowsett, M., Welch, A.A., Khaw, K.-T., Wareham, N.J., Day, N.E., Bingham, S.A., 2004. Phytoestrogen concentrations in serum and spot urine as biomarkers for dietary phytoestrogen intake and their relation to breast cancer risk in European prospective investigation of cancer and nutrition-norfolk. Cancer Epidemiol. Biomarkers Prev. 13, 698–708. Guillemette, C., Bélanger, A., Lépine, J., 2004. Metabolic inactivation of estrogens in breast tissue by UDP–glucuronosyltransferase enzymes: an overview. Breast Cancer Res. 2004 (6), 246–254. Hays, S.M., Aylward, L.L., LaKind, J.S., Bartels, M.J., Barton, H.A., Boogaard, P.J., Brunk, C., DiZio, S., Dourson, M., Goldstein, D.A., Lipscomb, J., Kilpatrick, M.E., Krewski, D., Krishnan, K., Nordberg, M., Okino, M., Tan, Y.-M., Viau, C., Yager, J.W., Biomonitoring Equivalents Expert Workshop, 2008. Guidelines for the derivation of biomonitoring equivalents: report from the biomonitoring equivalents expert workshop. Regul. Toxicol. Pharmacol. 51, S4–S15. http:// dx.doi.org/10.1016/j.yrtph.2008.05.004. Heald, C.L., Bolton-Smith, C., Ritchie, M.R., Morton, M.S., Alexander, F.E., 2006. Phyto-oestrogen intake in Scottish men: use of serum to validate a selfadministered food-frequency questionnaire in older men. Eur. J. Clin. Nutr. 60, 129–135. http://dx.doi.org/10.1038/sj.ejcn.1602277. Hosoda, K., Furuta, T., Ishii, K., 2011. Metabolism and disposition of isoflavone conjugated metabolites in humans after ingestion of kinako. Drug Metab. Dispos. 39, 1762–1767. http://dx.doi.org/10.1124/dmd.111.038281. Houtman, R., de Leeuw, R., Rondaij, M., Melchers, D., Verwoerd, D., Ruijtenbeek, R., Martens, J.W.M., Neefjes, J., Michalides, R., 2012. Serine-305 phosphorylation modulates estrogen receptor alpha binding to a coregulator peptide array, with potential application in predicting responses to tamoxifen. Mol. Cancer Ther. 11, 805–816. http://dx.doi.org/10.1158/1535-7163.MCT-11-0855. Iwasaki, M., Inoue, M., Otani, T., Sasazuki, S., Kurahashi, N., Miura, T., Yamamoto, S., Tsugane, S., Japan Public Health Center-based prospective study group, 2008. Plasma isoflavone level and subsequent risk of breast cancer among Japanese women: a nested case-control study from the Japan Public Health Center-based prospective study group. J. Clin. Oncol. 26, 1677–1683. http://dx.doi.org/ 10.1200/JCO.2007.13.9964. Jarosova, B., Blaha, L., Giesy, J.P., Hilscherova, K., 2014. What level of estrogenic activity determined by in vitro assays in municipal waste waters can be considered as safe? Environ. Int. 64, 98–109. http://dx.doi.org/10.1016/ j.envint.2013.12.009. Jeyakumar, M., Webb, P., Baxter, J.D., Scanlan, T.S., Katzenellenbogen, J.A., 2008. Quantification of ligand-regulated nuclear receptor corepressor and coactivator binding, key interactions determining ligand potency and efficacy for the thyroid hormone receptor. Biochemistry 47, 7465–7476. http://dx.doi.org/ 10.1021/bi800393u. Jeyakumar, M., Carlson, K.E., Gunther, J.R., Katzenellenbogen, J.A., 2011. Exploration of dimensions of estrogen potency: parsing ligand binding and coactivator binding affinities. J. Biol. Chem. 286, 12971–12982. http://dx.doi.org/10.1074/ jbc.M110.205112. Judson, R., Houck, K., Martin, M., Knudsen, T., Thomas, R.S., Sipes, N., Shah, I., Wambaugh, J., Crofton, K., 2014. In vitro and modelling approaches to risk assessment from the U.S. Environmental Protection Agency ToxCast Programme. Basic Clin. Pharmacol. Toxicol. 115, 69–76. http://dx.doi.org/ 10.1111/bcpt.12239. Judson, R.S., Houck, K.A., Kavlock, R.J., Knudsen, T.B., Martin, M.T., Mortensen, H.M., Reif, D.M., Rotroff, D.M., Shah, I., Richard, A.M., Dix, D.J., 2010. In vitro screening of environmental chemicals for targeted testing prioritization: the ToxCast project. Environ. Health Perspect. 118, 485–492. http://dx.doi.org/10.1289/ ehp.0901392. Judson, R.S., Kavlock, R.J., Setzer, R.W., Hubal, E.A.C., Martin, M.T., Knudsen, T.B., Houck, K.A., Thomas, R.S., Wetmore, B.A., Dix, D.J., 2011. Estimating toxicityrelated biological pathway altering doses for high-throughput chemical risk assessment. Chem. Res. Toxicol. 24, 451–462. http://dx.doi.org/10.1021/ tx100428e. Kraus, W.L., Weis, K.E., Katzenellenbogen, B.S., 1995. Inhibitory cross-talk between steroid hormone receptors: differential targeting of estrogen receptor in the repression of its transcriptional activity by agonist- and antagonist-occupied progestin receptors. Mol. Cell. Biol. 15, 1847–1857. LaKind, J.S., Aylward, L.L., Brunk, C., DiZio, S., Dourson, M., Goldstein, D.A., Kilpatrick, M.E., Krewski, D., Bartels, M.J., Barton, H.A., Boogaard, P.J., Lipscomb, J., Krishnan,

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408 K., Nordberg, M., Okino, M., Tan, Y.-M., Viau, C., Yager, J.W., Hays, S.M., Biomonitoring Equivalents Expert Workshop, 2008. Guidelines for the communication of Biomonitoring Equivalents: report from the Biomonitoring Equivalents Expert Workshop. Regul. Toxicol. Pharmacol. 51, S16–S26. http:// dx.doi.org/10.1016/j.yrtph.2008.05.007. Le, H.T., Schaldach, C.M., Firestone, G.L., Bjeldanes, L.F., 2003. Plant-derived 3,30 Diindolylmethane is a strong androgen antagonist in human prostate cancer cells. J. Biol. Chem. 278 (23), 21136–21145. Liu, K.-C., Wu, R.S.S., Ge, W., 2013. Luteinizing hormone receptor (lhcgr) as a marker gene for characterizing estrogenic endocrine-disrupting chemicals in zebrafish ovarian follicle cells. Gen. Comp. Endocrinol. 192, 89–94. http://dx.doi.org/ 10.1016/j.ygcen.2013.06.023. Metzner, J.E., Frank, T., Kunz, I., Burger, D., Riegger, C., 2009. Study on the pharmacokinetics of synthetic genistein after multiple oral intake in postmenopausal women. Arzneimittelforschung 59, 513–520. Morgan, R.E., van Staden, C.J., Chen, Y., Kalyanaraman, N., Kalanzi, J., Dunn, R.T., Afshari, C.A., Hamadeh, H.K., 2013. A multifactorial approach to hepatobiliary transporter assessment enables improved therapeutic compound development. Toxicol. Sci. 136, 216–241. http://dx.doi.org/10.1093/toxsci/ kft176. Murk, A.J., Rijntjes, E., Blaauboer, B.J., Clewell, R., Crofton, K.M., Dingemans, M.M., Furlow, J.D., Kavlock, R., Kohrle, J., Opitz, R., Traas, T., Visser, T.J., Xia, M., Gutleb, A.C., 2013. Mechanism-based testing strategy using in vitro approaches for identification of thyroid hormone disrupting chemicals. Toxicol. In Vitro 27, 1320–1346 (24). National Research Council (NRC), 2007. Toxicity Testing in the 21st Century: A Vision and a Strategy. National Academies Press, Washington, DC. National Research Council (NRC), 2009. Science and Decisions: Advancing Risk Assessment. National Academies Press, Washington, DC. Ong, K.M., Blackford, J.A., Kagan, B.L., Simons, S.S., Chow, C.C., 2010. A theoretical framework for gene induction and experimental comparisons. Proc. Natl. Acad. Sci. USA 107, 7107–7112. http://dx.doi.org/10.1073/pnas.0911095107. Pastoor, T.P., Bachman, A.N., Bell, D.R., Cohen, S.M., Dellarco, M., Dewhurst, I.C., Doe, J.E., Doerrer, N.G., Embry, M.R., Hines, R.N., Moretto, A., Phillips, R.D., Rowlands, J.C., Tanir, J.Y., Wolf, D.C., Boobis, A.R., 2014. A 21st century roadmap for human health risk assessment. Crit. Rev. Toxicol. 44 (Suppl. 3), 1–5. Patlewicz, G., Simon, T., Goyak, K., Phillips, R.D., Rowlands, J.C., Seidel, S., Becker, R.A., 2013. Use and validation of HT/HC assays to support 21st century toxicity evaluations. Regul. Toxicol. Pharmacol. 65, 259–268. http://dx.doi.org/10.1016/ j.yrtph.2012.12.008. Paul, K.B., Hedge, J.M., Rotroff, D.M., Hornung, M.W., Crofton, K.M., Simmons, S.O., 2014. Development of a thyroperoxidase inhibition assay for high-throughput screening. Chem. Res. Toxicol. 27 (3), 387–399. Reif, D.M., Martin, M.T., Tan, S.W., Houck, K.A., Judson, R.S., Richard, A.M., Knudsen, T.B., Dix, D.J., Kavlock, R.J., 2010. Endocrine profiling and prioritization of environmental chemicals using ToxCast data. Environ. Health Perspect. 118, 1714–1720. http://dx.doi.org/10.1289/ehp.1002180. Richelle, M., Pridmore-Merten, S., Bodenstab, S., Enslen, M., Offord, E.A., 2002. Hydrolysis of isoflavone glycosides to aglycones by beta-glycosidase does not alter plasma and urine isoflavone pharmacokinetics in postmenopausal women. J. Nutr. 132, 2587–2592. Ritchie, M.R., Morton, M.S., Deighton, N., Blake, A., Cummings, J.H., 2004. Plasma and urinary phyto-oestrogens as biomarkers of intake: validation by duplicate diet analysis. Br. J. Nutr. 91, 447–457. http://dx.doi.org/10.1079/BJN20031062. Rotroff, D.M., Wetmore, B.A., Dix, D.J., Ferguson, S.S., Clewell, H.J., Houck, K.A., Lecluyse, E.L., Andersen, M.E., Judson, R.S., Smith, C.M., Sochaski, M.A., Kavlock, R.J., Boellmann, F., Martin, M.T., Reif, D.M., Wambaugh, J.F., Thomas, R.S., 2010. Incorporating human dosimetry and exposure into high-throughput in vitro toxicity screening. Toxicol Sci. 117 (2), 348–358. http://dx.doi.org/ 10.1093/toxsci/kfq22. Rotroff, D.M., Dix, D.J., Houck, K.A., Knudsen, T.B., Martin, M.T., McLaurin, K.W., Reif, D.M., Crofton, K.M., Singh, A.V., Xia, M., Huang, R., Judson, R.S., 2013. Using in vitro high throughput screening assays to identify potential endocrinedisrupting chemicals. Environ. Health Perspect. 121, 7–14. http://dx.doi.org/ 10.1289/ehp.1205065. Rotroff, D.M., Dix, D.J., Houck, K.A., Kavlock, R.J., Knudsen, T.B., Martin, M.T., Reif, D.M., Richard, A.M., Sipes, N.S., Abassi, Y.A., Jin, C., Stampfl, M., Judson, R.S., 2013. Real-time growth kinetics measuring hormone mimicry for ToxCast chemicals in T-47D human ductal carcinoma cells. Chem. Res. Toxicol. 26, 1097–1107. http://dx.doi.org/10.1021/tx400117y. Rotroff, D.M., Martin, M.T., Dix, D.J., Filer, D.L., Houck, K.A., Knudsen, T.B., Sipes, N.S., Reif, D.M., Xia, M., Huang, R., Judson, R.S., 2014. Predictive endocrine testing in the 21st century using in vitro assays of estrogen receptor signaling responses. Environ. Sci. Technol. http://dx.doi.org/10.1021/es502676e. Safe, S.H., 1998. Development validation and problems with the toxic equivalency factor approach for risk assessment of dioxins and related compounds. J. Anim. Sci. 76 (1), 134–141. Setchell, K.D.R., Faughnan, M.S., Avades, T., Zimmer-Nechemias, L., Brown, N.M., Wolfe, B.E., Brashear, W.T., Desai, P., Oldfield, M.F., Botting, N.P., Cassidy, A., 2003. Comparing the pharmacokinetics of daidzein and genistein with the use of 13C-labeled tracers in premenopausal women. Am. J. Clin. Nutr. 77, 411–419. Setchell, K.D.R., Brown, N.M., Zhao, X., Lindley, S.L., Heubi, J.E., King, E.C., Messina, M.J., 2011. Soy isoflavone phase II metabolism differs between rodents and humans: implications for the effect on breast cancer risk. Am. J. Clin. Nutr. 94, 1284–1294. http://dx.doi.org/10.3945/ajcn.111.019638.

407

Shelnutt, S.R., Cimino, C.O., Wiggins, P.A., Badger, T.M., 2000. Urinary pharmacokinetics of the glucuronide and sulfate conjugates of genistein and daidzein. Cancer Epidemiol. Biomarkers Prev. 9, 413–419. Shelnutt, S.R., Cimino, C.O., Wiggins, P.A., Ronis, M.J.J., Badger, T.M., 2002. Pharmacokinetics of the glucuronide and sulfate conjugates of genistein and daidzein in men and women after consumption of a soy beverage. Am. J. Clin. Nutr. 76, 588–594. Shimazu, T., Inoue, M., Sasazuki, S., Iwasaki, M., Sawada, N., Yamaji, T., Tsugane, S., JPHC Study Group, 2011. Plasma isoflavones and the risk of lung cancer in women: a nested case-control study in Japan. Cancer Epidemiol. Biomarkers Prev. 20, 419–427. Simons, S.S., Kumar, R., 2013. Variable steroid receptor responses: intrinsically disordered AF1 is the key. Mol. Cell. Endocrinol. 376, 81–84. http://dx.doi.org/ 10.1016/j.mce.2013.06.007. Simon, T.W., Simons, S.S., Preston, R.J., Boobis, A.R., Cohen, S.M., Doerrer, N.G., Fenner-Crisp, P.A., McMullin, T.S., McQueen, C.A., Rowlands, J.C., RISK21 DoseResponse Subteam, 2014. The use of mode of action information in risk assessment: quantitative key events/dose-response framework for modeling the dose-response for key events. Crit. Rev. Toxicol. 44 (Suppl. 3), 17–43. http:// dx.doi.org/10.3109/10408444.2014.931925. Sipes, N.S., Martin, M.T., Kothiya, P., Reif, D.M., Judson, R.S., Richard, A.M., Houck, K.A., Dix, D.J., Kavlock, R.J., Knudsen, T.B., 2013. Profiling 976 ToxCast chemicals across 331 enzymatic and receptor signaling assays. Chem. Res. Toxicol. 26, 878–895. http://dx.doi.org/10.1021/tx400021f. Slob, W., Setzer, R.W., 2014. Shape and steepness of toxicological dose-response relationships of continuous endpoints. Crit. Rev. Toxicol. 44, 270–297. http:// dx.doi.org/10.3109/10408444.2013.853726. Simons, S.S., 2008. What goes on behind closed doors: physiological versus pharmacological steroid hormone actions. BioEssays 30, 744–756. http:// dx.doi.org/10.1002/bies.20792. Strassburg, C.P., Strassburg, A., Kneip, S., Barut, A., Tukey, R.H., Rodeck, B., Manns, M.P., 2002. Developmental aspects of human hepatic drug glucuronidation in young children and adults C P 1,1. Gut 50, 259–265. http://dx.doi.org/10.1136/ gut.50.2.259. Takashima, N., Miyanaga, N., Komiya, K., More, M., Akaza, H., 2004. Blood isoflavone levels during intake of a controlled hospital diet. J. Nutr. Sci. Vitaminol. (Tokyo) 50, 246–252. Teeguarden, J., Hanson-Drury, S., Fisher, J.W., Doerge, D.R., 2013. Are typical human serum BPA concentrations measurable and sufficient to be estrogenic in the general population? Food Chem. Toxicol. 62, 949–963. http://dx.doi.org/ 10.1016/j.fct.2013.08.001. Teeguarden, J.G., Waechter, J.M., Clewell, H.J., Covington, T.R., Barton, H.A., 2005. Evaluation of oral and intravenous route pharmacokinetics, plasma protein binding, and uterine tissue dose metrics of Bisphenol A: a physiologically based pharmacokinetic approach. Toxicol. Sci. 85, 823–838. http://dx.doi.org/ 10.1093/toxsci/kfi135. Teeguarden, J.G., Calafat, A.M., Ye, X., Doerge, D.R., Churchwell, M.I., Gunawan, R., Graham, M.K., 2011. Twenty-four hour human urine and serum profiles of Bisphenol A during high-dietary exposure. Toxicol. Sci. 123, 48–57. http:// dx.doi.org/10.1093/toxsci/kfr160. Teeguarden, J.G., Hanson-Drury, S., 2013. A systematic review of Bisphenol A ‘‘low dose’’ studies in the context of human exposure: a case for establishing standards for reporting ‘‘low-dose’’ effects of chemicals. Food Chem. Toxicol. 62, 935–948. http://dx.doi.org/10.1016/j.fct.2013.07.007. Thomas, R.S., Black, M., Li, L., Healy, E., Chu, T.-M., Bao, W., Andersen, M., Wolfinger, R., 2012. A comprehensive statistical analysis of predicting in vivo hazard using high-throughput in vitro screening. Toxicol. Sci. 128, 398–417. http:// dx.doi.org/10.1093/toxsci/kfs159. Thomas, R.S., Philbert, M.A., Auerbach, S.S., Wetmore, B.A., Devito, M.J., Cote, I., Rowlands, J.C., Whelan, M.P., Hays, S.M., Andersen, M.E., Meek, M.E.B., Reiter, L.W., Lambert, J.C., Clewell, H.J., Stephens, M.L., Zhao, Q.J., Wesselkamper, S.C., Flowers, L., Carney, E.W., Pastoor, T.P., Petersen, D.D., Yauk, C.L., Nong, A., 2013. Incorporating new technologies into toxicity testing and risk assessment: moving from 21st century vision to a data-driven framework. Toxicol. Sci. 136, 4–18. http://dx.doi.org/10.1093/toxsci/kft178. United States Environmental Protection Agency (USEPA). 2011. Endocrine Disruptor Screening Program for the 21st Century: (EDSP21 Workplan) The Incorporation of In Silico Models and In VitroHigh Throughput Assay in the Endocrine Disruptor Screening Program (EDSP) for Prioritization and Screening. United States Environmental Protection Agency (USEPA). 2014. Exposure SAP White Paper: New High-throughput Methods to Estimate Chemical Exposure. http:// www.regulations.gov/#!documentDetail;D=EPA-HQ-OPP-2014-0331-0005. United States Environmental Protection Agency (USEPA). 2014b. Integrated Bioactivity and Exposure Ranking: A Computational Approach for the Prioritization and Screening of Chemicals in the Endocrine Disruptor Screening Program. http://www.regulations.gov/#!documentDetail;D=EPAHQ-OPP-2014-0614. Van den Berg, M., Birnbaum, L., Bosveld, A.T., Brunstrom, B., Cook, P., Feeley, M., Giesy, J.P., Hanberg, A., Hasegawa, R., Kennedy, S.W., Kubiak, T., Larsen, J.C., van Leeuwen, F.X., Liem, A.K., Nolt, C., Peterson, R.E., Poellinger, L., Safe, S., Schrenk, D., Tillitt, D., Tysklind, M., Younes, M., Waern, F., Zacharewski, T., 1998. Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife Environ. Health. Perspect. 106, 775–792. Van den Berg, M., Birnbaum, L.S., Denison, M., De, V.M., Farland, W., Feeley, M., Fiedler, H., Hakansson, H., Hanberg, A., Haws, L., Rose, M., Safe, S., Schrenk, D., Tohyama, C., Tritscher, A., Tuomisto, J., Tysklind, M., Walker, N., Peterson, R.E.,

408

R.A. Becker et al. / Regulatory Toxicology and Pharmacology 71 (2015) 398–408

2006. The 2005 World Health Organization reevaluation of human and Mammalian toxic equivalency factors for dioxins and dioxin-like compounds. Toxicol. Sci. 93, 223–241. Völkel, W., Colnot, T., Csanády, G.A., Filser, J.G., Dekant, W., 2002. Metabolism and kinetics of bisphenol a in humans at low doses following oral administration. Chem. Res. Toxicol. 15, 1281–1287. Wambaugh, J.F., Setzer, R.W., Reif, D.M., Gangwal, S., Mitchell-Blackwood, J., Arnot, J.A., Joliet, O., Frame, A., Rabinowitz, J., Knudsen, T.B., Judson, R.S., Egeghy, P., Vallero, D., Cohen, Hubal E.A., 2013. High-throughput models for exposurebased chemical prioritization in the ExpoCast project. Environ. Sci. Technol. 47, 8479–8488. Wang, S., Houtman, R., Melchers, D., Aarts, J., Peijnenburg, A., van Beuningen, R., Rietjens, I., Bovee, T.F., 2013. A 155-plex high-throughput in vitro coregulator binding assay for (anti-)estrogenicity testing evaluated with 23 reference compounds. ALTEX 30, 145–157. Wetmore, B.A., Wambaugh, J.F., Ferguson, S.S., Sochaski, M.A., Rotroff, D.M., Freeman, K., Clewell, H.J., Dix, D.J., Andersen, M.E., Houck, K.A., Allen, B., Judson, R.S., Singh, R., Kavlock, R.J., Richard, A.M., Thomas, R.S., 2012. Integration of dosimetry, exposure and high-throughput screening data in chemical

toxicity assessment. Toxicol. Sci. 125, 157–174. http://dx.doi.org/10.1093/ toxsci/kfr254. Wetmore, B.A., Wambaugh, J.F., Ferguson, S.S., Li, L., Clewell, H.J., Judson, R.S., Freeman, K., Bao, W., Sochaski, M.A., Chu, T.-M., Black, M.B., Healy, E., Allen, B., Andersen, M.E., Wolfinger, R.D., Thomas, R.S., 2013. Relative impact of incorporating pharmacokinetics on predicting in vivo hazard and mode of action from high-throughput in vitro toxicity assays. Toxicol. Sci. 132, 327–346. http://dx.doi.org/10.1093/toxsci/kft012. Yoon, M., Efremenko, A., Blaauboer, B.J., Clewell, H.J., 2014. Evaluation of simple in vitro to in vivo extrapolation approaches for environmental compounds. Toxicol. In Vitro 28, 164–170. Yuan, B., Zhen, H., Jin, Y., Xu, L., Jiang, X., Sun, S., Li, C., Xu, H., 2012. Absorption and plasma disposition of genistin differ from those of genistein in healthy women. J. Agric. Food Chem. 60, 1428–1436. http://dx.doi.org/10.1021/jf204421c. Zhang, Y., Hendrich, S., Murphy, P.A., 2003. Glucuronides are the main isoflavone metabolites in women. J. Nutr. 133, 399–404. Zhang, Z., Teng, C.T., 2001. Estrogen receptor alpha and estrogen receptor-related receptor alpha1 compete for binding and coactivator. Mol. Cell. Endocrinol. 172, 223–233.