Evidence for Cross Species Extrapolation of Mammalian-BasedHigh-Throughput Screening Assay ResultsCarlie A. LaLone,*,† Daniel L. Villeneuve,† Jon A. Doering,‡ Brett R. Blackwell,†

Thomas R. Transue,§ Cody W. Simmons,§ Joe Swintek,¶ Sigmund J. Degitz,† Antony J. Williams,||

and Gerald T. Ankley†

†Office of Research and Development, National Health and Environmental Effects Research Laboratory, Mid-Continent EcologyDivision, US Environmental Protection Agency, 6201 Congdon Blvd., Duluth, Minnesota 55804, United States‡National Research Council, 6201 Congdon Blvd., Duluth, Minnesota 55804, United States§CSRA Inc., 109 T.W. Alexander Drive, Research Triangle Park, North Carolina 27711, United States¶Badger Technical Services, 6201 Congdon Blvd., Duluth, Minnesota 55804, United States||Office of Research and Development, National Center for Computational Toxicology, US Environmental Protection Agency,109 T.W. Alexander Drive, Research Triangle Park, North Carolina 27711, United States

*S Supporting Information

ABSTRACT: High-throughput screening (HTS) and computational technologies haveemerged as important tools for chemical hazard identification. The US EnvironmentalProtection Agency (EPA) launched the Toxicity ForeCaster (ToxCast) Program, whichhas screened thousands of chemicals in hundreds of mammalian-based HTS assays forbiological activity. The data are being used to prioritize toxicity testing on those chemi-cals likely to lead to adverse effects. To use HTS assays in predicting hazard to both humansand wildlife, it is necessary to understand how broadly these data may be extrapolated acrossspecies. The US EPA Sequence Alignment to Predict Across Species Susceptibility(SeqAPASS; https://seqapass.epa.gov/seqapass/) tool was used to assess conservation ofthe 484 protein targets represented in the suite of ToxCast assays and other HTS assays.To demonstrate the utility of the SeqAPASS data for guiding extrapolation, case studieswere developed which focused on targets of interest to the US Endocrine DisruptorScreening Program and the Organisation for Economic Cooperation and Development.These case studies provide a line of evidence for conservation of endocrine targets acrossvertebrate species, with few exceptions, and demonstrate the utility of SeqAPASS fordefining the taxonomic domain of applicability for HTS results and identifying organismsfor suitable follow-up toxicity tests.

■ INTRODUCTIONThe international adoption of 21st century toxicity testingpractices, aimed at increasing efficiencies in chemical safetyassessments and reducing reliance on whole animal studies, hasled to significant advancements in cell-based and computa-tional approaches.1 High-throughput screening (HTS) assaysallow for rapid screening of hundreds to thousands of chemi-cals, using cell-free assays, primary cells, immortalized cell lines,and even small scale in vivo models to prioritize which chemicalsare of most concern and therefore may require in vivo toxicitytesting. For example, the United States Environmental Protec-tion Agency (US EPA) Endocrine Disruptor Screening Program(EDSP) and, more broadly, the Organization of EconomicCooperation and Development (OECD; http://oe.cd/endocrine-disrupters) have both recently described frameworksto screen for endocrine active chemicals using computationalmodels informed by HTS data for prioritization of thousandsof chemicals for follow-up testing.2−5 When used forprioritization, HTS assays can significantly reduce the number

of animals used for testing and reduce the costs associated withwhole organism tests. Further, with advances in technology,many of these assays can be conducted using robotics and multi-well plates, of 96-wells or greater, which increase efficiencies inthe chemical testing pipeline.One example of such an effort includes the US EPA Toxicity

ForeCaster (ToxCast) screening and prioritization program,along with the collaborative multiagency Tox21 effort, whichhas led the development and employment of rapid HTS assaysfor the assessment of large numbers of chemicals and theutilization of generated data to characterize bioactivity to focusfuture testing.6−8 Since initiation of the ToxCast program in2007, over 4000 chemicals have been screened in at least asubset of the bioassay platforms.7 The suite of ToxCast

Received: August 15, 2018Revised: October 1, 2018Accepted: October 10, 2018Published: October 10, 2018

Article

pubs.acs.org/estCite This: Environ. Sci. Technol. 2018, 52, 13960−13971

© 2018 American Chemical Society 13960 DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

Dow

nloa

ded

via

US

EPA

MID

-CO

NT

INE

NT

EC

O D

IV L

AB

on

Febr

uary

4, 2

019

at 1

6:47

:24

(UT

C).

Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

bioassays were developed to have extensive biological coveragewith molecular targets spanning all major protein superfamilygroups, including many known to be chemical targets.6

■ SPECIES COVERAGETraditionally, HTS assays were developed for screeningchemicals for adverse impacts to human health and thereforeprimarily utilize mammalian-based cell lines, primary cells, andgene sequences.6 The available HTS assays with mammalianmodels were incorporated in the suite of ToxCast assays includ-ing human (Homo sapiens), cattle (Bos taurus), chimpanzee (Pantroglodytes), domestic guinea pig (Cavia porcellus), rabbit(Oryctolagus cuniculus), rat (Rattus norvegicus), house mouse(Mus musculus), pig (Sus scrofa), and sheep (Ovis aries). Morerecently, HTS assays have been developed and some incor-porated into ToxCast to evaluate chemical effects on alterna-tive vertebrate species using an African clawed frog (Xenopuslaevis)-based thyroid pathway screening assay and a zebrafish(Danio rerio) development assay.9,10 With the original inten-tion of ToxCast/Tox21 assays to identify and prioritize chemi-cals that have the greatest likelihood to produce adverse effectsin humans and the subsequent realization that such data maybe useful for protecting wildlife, a challenge emerged to under-stand whether the predominantly mammalian-based prioritiza-tion approach reasonably reflects potential impacts on otherspecies. Therefore, understanding the domain of applicabilityacross species for the HTS data is important to taking fulladvantage of both the existing and new data not only insupport of the EDSP and OECD efforts but also in support ofother hazard identification efforts for nonmammalian species.A tool that can aid in addressing this species-extrapolation

challenge is the US EPA Sequence Alignment to Predict AcrossSpecies Susceptibility (SeqAPASS v3.0; https://seqapass.epa.gov/seqapass/) tool.11 The SeqAPASS tool can be used toevaluate protein sequence similarity to predict chemical sus-ceptibility across species. The underlying assumption for suchprotein-based species comparisons is that, the greater the simi-larity between a chemical’s protein target in a known sensitivespecies to a protein in other species, the greater likelihood thatthe chemical would also interact with this protein in the otherspecies. This concept of linking evolutionary conservation ofprotein targets to cross species chemical sensitivity was firstdescribed in 2008.12 The publicly available SeqAPASS tool hassince expanded upon and advanced this concept into practiceallowing for the comparison of any species to all other specieswith sequence information available to computationally predictchemical susceptibility. Therefore, the SeqAPASS output pro-vides evidence as to whether a known protein target is presentor absent in another species based on similarity, which is usefulfor laying the initial foundation for species extrapolation. Becausethe SeqAPASS tool takes advantage of the National Center forBiotechnology Information (NCBI) protein database, programoutputs include similarity evaluations and conservation predic-tions for hundreds or even thousands of vertebrates, inverte-brates, plants, bacteria, and viruses.A tiered framework was proposed by Ankley et al.13 to evalu-

ate biological pathway conservation across taxa as a means tounderstand whether current HTS approaches would beconsidered protective of nonmammalian species in the contextof the EDSP. The tiered approach was previously applied toexplore the taxonomic applicability domain of human estrogenreceptor α (ERα)-based HTS assays.13 Since perturbation ormodulation of estrogenic pathways are relatively well studied in

a variety of species (e.g., refs 14−16), a robust evaluation ofconservation at all tiers (i.e., computational, in vitro, and in vivostudies) of the taxonomic relevance framework was feasible forERα. However, cross-species data related to perturbation ofother molecular targets are often much more limited. There-fore, for a majority of chemical targets, the most practical andrapid means to evaluate the potential for chemical−proteintarget interactions across species will be through computationalevaluation of protein sequence and structural conservation usingSeqAPASS.To initiate a SeqAPASS query in the context of HTS assays,

two key pieces of information are necessary. Specifically, themodel organism from which the cell, protein, or gene wasderived for a given assay is needed, and the gene for the pro-tein target measured in the assay must be known. Dependingon the objective of the analysis and how well the protein hasbeen characterized, as well as the level of detail known aboutthe chemical−protein interaction, the SeqAPASS tool allowsfor evaluation of protein similarity at three different levels ofsequence specificity.11 Level 1 facilitates comparative cross-species evaluation of the full primary amino acid sequence;Level 2 focuses the comparative analysis on the functionaldomain(s) in the protein target, and Level 3 compares indi-vidual amino acid residues across species.11,17 Level 3 queriesare typically based on information obtained from the publishedliterature regarding which specific amino acids are critical forthe interaction of a specific chemical, or group of chemicals,with the protein of interest. Level 3 evaluations may be appliedto compare known amino acid residues that are important forthe specific action of that chemical.11,17 Advancing througheach level of the SeqAPASS evaluation adds additional evi-dence for protein similarity, with Level 3 providing the highestdegree of taxonomic resolution for predictions of proteinconservation and chemical susceptibility. However, in con-ducting the SeqAPASS evaluation to determine whether theassay target is conserved in other species, Level 1 and 2 data aregenerally sufficient to provide an initial line of evidence that caninform further toxicity testing (e.g., selection of relevant speciesto test) and extrapolation of screening data when considering allchemicals screened (as opposed to an individual chemical).As a starting point in determining the taxonomic domain of

applicability for mammalian-based HTS screening data,SeqAPASS evaluations comparing primary amino acid sequences(Level 1) and functional domains (Level 2) were performed forall molecular targets associated with the ToxCast assays andmade publicly available. Since it was not reasonable to discussall SeqAPASS results for these targets, data for two ToxCastassay groups are summarized. In addition, case studies weredeveloped for hypothalamic−pituitary−gonadal/−thyroidmo-lecular targets relevant to the EDSP and OECD. These casestudies examined conservation of the androgen receptor (AR),enzymes involved in the sex steroid synthesis (i.e., steroido-genesis), and proteins found in the thyroid axis. These casestudies are used as examples to demonstrate the practical appli-cation of this approach to a current regulatory challenge.

■ MATERIALS AND METHODSIdentifying Protein Targets from ToxCast/Tox21 HTS

Assays. ToxCast data were downloaded from https://www.epa.gov/chemical-research/toxicity-forecaster-toxcasttm-dataand filtered to identify all unique protein identifications (i.e.,protein accessions) associated with HTS assay end points(Supplemental Data, Materials and Methods). This protein

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13961

information, which included the protein accession and modelspecies represented in the ToxCast assay, was used as a queryprotein sequence in the SeqAPASS analysis (SupplementalData, Workbook S1).Additional assay targets were identified as new HTS assays

have been, and are being, developed to support the EDSP andOECD to more thoroughly screen for chemicals that maydisrupt the thyroid axis and steroidogenesis. Since the sum-mary data that was made public in October 2015, six thyroidrelated HTS assays and the high-throughput H295R assay (HT-H295R; a steroidogenesis assay) were developed to supportEDSP and OECD efforts. The data for the HT-H295R assay andthe thyroid peroxidase (TPO) assay are available through theToxCast dashboards and available for public download. Chemi-cal screening results from the other thyroid-related HTS assayscan be obtained from the published literature but have yet tobe incorporated in the ToxCast dashboard.18,19,9,20,21

SeqAPASS Level 1: Primary Amino Acid SequenceComparisons of HTS Targets. A total of 484 unique assaytargets available in the ToxCast suite were evaluated usingSeqAPASS v3.0, Data Version 3 (Supplemental Data, Table S1).The output from SeqAPASS provides a prediction of relativeintrinsic susceptibility across species based on similarity of thehit proteins to the query protein (Supplemental Data, Table S2).In the context of evaluating HTS assays, a susceptibility pre-diction of “yes” in the SeqAPASS output indicates the proteintarget is conserved in that species whose sequence aligned withthe query species (Supplemental Data, Materials and Methods).SeqAPASS Level 2: Functional Domain Comparisons

of HTS Targets. For each protein accession submitted inSeqAPASS Level 1, functional domains (e.g., ligand bindingdomains, DNA binding domains, zinc finger domains) cate-gorized as “specific hits” in the NCBI Conserved Domain Data-base (CDD; https://www.ncbi.nlm.nih.gov/cdd/) were identified(0 to 17 domains) and submitted as SeqAPASS Level 2 queries(Supplemental Data, Materials and Methods).11,22,23 Overall,functional domains were evaluated for 452 of the 484 proteins.Processing SeqAPASS Data To Evaluate the Taxo-

nomic Domain of Applicability for HTS Data. All gen-erated SeqAPASS Level 1 and Level 2 data were saved in a .zipfile using the Save Reports function described in theSeqAPASSv3.0 user guide.22 Briefly, this save function allowsthe user to download and save multiple reports at one time(Supplemental Data, Materials and Methods). All Level 1 andLevel 2 data from the current evaluation can be accessed viaSeqAPASS and through the US EPA CompTox ChemistryDashboard (https://comptox.epa.gov/dashboard/) (Supple-mental Data, Tables S3−S4; Supplemental Data, Zip File S1).The CompTox Chemistry Dashboard is a resource that pro-vides access to chemical information, including HTS bioassaydata.24 Therefore, links from the dashboard to the SeqAPASStool, and specifically to SeqAPASS data for HTS assay targets,are available as a means to efficiently access chemical and effectsdata. As new “Data Versions” are released in the SeqAPASS tool(see SeqAPASS About page), the SeqAPASS data on theCompTox Chemistry Dashboard will be updated for all bio-assay targets, allowing access to the most current protein datafor cross-species extrapolation.To broadly describe the results of the SeqAPASS data,

Level 1 and 2 data outputs with default settings (i.e., primaryreports; E-value <0.01; common domains ≥1; default first localminimum was used to set the susceptibility cutoff)25,11 werecollected as a multidata set download (SeqAPASS User

Guide22) for each protein target. Protein targets were thengrouped according to their intended target family (SupplementalData, Materials and Methods). The target families included:(1) cell adhesion molecules;(2) cytochrome P450 enzymes (i.e., cyp);(3) cytokine; (4) DNA binding; (5) esterase; (6) G-protein-coupledreceptors (i.e., gpcr); (7) growth factor; (8) hydrolase; (9) ionchannel; (10) kinase; (11) lyase; (12) methyltransferase;(13) nuclear receptor; (14) oxidoreductase; (15) phosphatase;(16) protease; (17) protease inhibitor; (18) transporter. Of the484 proteins corresponding to ToxCast assays and evaluated inSeqAPASS, 346 were placed into these groups (SupplementalData, Table S1). The remaining proteins were not placed in agroup and were categorized as miscellaneous proteins. Con-servation was broadly described within a defined assay group(except for miscellaneous proteins), deeming protein groups eitherbroadly conserved (e.g., vertebrates and invertebrates, plants) ornarrowly conserved (e.g., vertebrates only) (Supplemental Data,Zip File S1). Synthesis of the SeqAPASS Level 1 results for twoof the ToxCast assay groups (G protein coupled receptor andgrowth factor) are described below to illustrate the informationavailable for all assay groups. Level 2 SeqAPASS data aredescribed for an individual protein, human transforming growthfactor beta 1 (found in assay group growth factor), althoughLevel 2 data were produced for all proteins evaluated (Supple-mental Data, Zip File S1).

Case Studies Focused on Endocrine Targets. Andro-gen Receptor. Eleven ToxCast assays have been developed todetect chemicals that perturb the AR, most of which utilize thehuman AR gene or protein. However, AR genes from the chim-panzee and Norway rat are also represented in the suite ofassays (Table 1). Therefore, to evaluate conservation of theAR across species, proteins for all three species were queriedusing SeqAPASS. Level 1 analysis comparing primary aminoacid sequences and Level 2 analyses focused on the ligandbinding domain were performed (Table 1; Supplemental Data,Figure S1a−d).

Steroidogenesis. The SeqAPASS tool was used to evaluatecross-species conservation of steroidogenic enzymes by com-paring primary amino acid sequences representing several steroido-genic genes, human cholesterol side-chain cleavage enzyme(CYP11A1), steroid 17-alpha-hydroxylase/17,20 lyase(CYP17A1), 3β-hydroxysteroid dehydrogenase (HSD3B2),aromatase (CYP19A1), 17β-hydroxysteroid dehydrogenasetype 1 (17βHSD1), 17β-hydroxysteroid dehydrogenasetype 3 (17βHSD3), 21-hydroxylase (CYP21A2), and steroid11-beta-hydroxylase (CYP11B1), to similar proteins in otherspecies (Table 1). Only Level 1 SeqAPASS evaluations werecarried out for the CYPs, as the domain identified as a specifichit in the NCBI conserved domain database (pfam00067) wasshown to encompass almost the entire protein sequence foreach of the CYPs. Therefore, Level 2 evaluations of functionaldomain(s) were not used to describe the data for steroidogenicCYPs as they did not add greater taxonomic resolution to theSeqAPASS prediction of conservation.

Thyroid Axis. The ToxCast/Tox21 suite contains sevenassays relevant to the thyroid axis, three of which were developedto detect chemicals that act on human thyroid hormone receptoralpha (THRα), one for human thyroid hormone receptor beta(THRβ), one each for Norway rat and pig thyroid peroxidase(TPO), and one for the human sodium/iodide cotransporter(NIS).21,19,18 In addition to the assays already incorporatedinto ToxCast, SeqAPASS analyses were also performed foradditional thyroid axis targets for which HTS assays are being

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13962

Table 1a

assay name assay target model organismSeqAPASS query

accessiondomain evaluated (NCBI

accession)

ATG_TRANS androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

NVS_NR_hAR androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

OT_AR_ARELUC_AG_1440 androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

OT_AR_ARSRC1_0480 androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

OT_AR_ARSRC1_0960 androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

TOX21_AR_BLA_Agonist androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

TOX21_AR_BLA_Antagonist androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

TOX21_AR_LUC_MDAKB2_Agonist androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

TOX21_AR_LUC_MDAKB2_Antagonist androgen receptor, AR human (Homo sapiens) P10275 ligand binding domain(cd07073)

NVS_NR_cAR androgen receptor, AR chimpanzee (Pantroglodytes)

O97775 ligand binding domain(cd07073)

NVS_NR_rAR androgen receptor, AR norway rat (Rattusnorvegicus)

P15207 ligand binding domain(cd07073)

HT-H295R cholesterol side-chain cleavageenzyme, CYP11A1

human (Homo sapiens) P05108 NA

HT-H295R steroid 17α-hydroxylase/17,20lyase, CYP17A1

human (Homo sapiens) P05093 NA

HT-H295R 3β-hydroxysteroiddehydrogenase, HSD3B2

human (Homo sapiens) P26439 NA

HT-H295R aromatase, CYP19A1 human (Homo sapiens) P11511 NANVS_ADME_hCYP19A1 aromatase, CYP19A1 human (Homo sapiens) P11511 NATOX21_Aromatase_Inhibition aromatase, CYP19A1 human (Homo sapiens) P11511 NAHT-H295R 17β-hydroxysteroid

dehydrogenase type1,17βHSD1

human (Homo sapiens) P14061 NA

HT-H295R 17β-hydroxysteroiddehydrogenase type3,17βHSD3

human (Homo sapiens) P37058 NA

HT-H295R steroid 21-hydroxylase, CYP21A2 human (Homo sapiens) AFK10138 NAHT-H295R steroid 11-beta-hydroxylase,

CYP11B1human (Homo sapiens) AAA35741 NA

ATG_THRa1_TRANS thyroid hormone receptor alpha,THRα

human (Homo sapiens) P10827 ligand binding domain(cd06935)

NVS_NR_hTRa_Antagonist thyroid hormone receptor alpha,THRα

human (Homo sapiens) P10827 ligand binding domain(cd06935)

TOX21_TR_LUC_GH3_Agonist thyroid hormone receptor alpha,THRα

human (Homo sapiens) P10827 Ligand binding domain(cd06935)

ATG_THRb_TRANS2 thyroid hormone receptor beta,THRβ

human (Homo sapiens) P10828 Ligand binding domain(cd06935)

NCCT_TPO_AUR thyroid peroxidase, TPO norway rat (Rattusnorvegicus)

P14650 thyroid _peroxidase(cd09825)

NCCT_TPO_GUA thyroid peroxidase, TPO pig (Sus scrofa) P09933 thyroid _peroxidase(cd09825)

hNIS-HEK293T-EPA sodium/iodide cotransporter,NIS

human (Homo sapiens) Q92911 NA

HTS assay in development type I iodothyronine deiodinase,DIO1

human (Homo sapiens) P49895 NA

HTS assay in development type II iodothyronine deiodinase,DIO2

human (Homo sapiens) Q92813 NA

HTS assay in development type III iodothyroninedeiodinase, DIO3

human (Homo sapiens) P55073 NA

HTS assay in development thyroid stimulating hormonereceptor, TSHR

human (Homo sapiens) P16473 peptide binding pocket(cd15964)

HTS assay in development iodotyrosine deiodinase, IYD human (Homo sapiens) Q6PHW0 iodotyrosine_dehaloge-nase domain (cd02144)

aAssay name key: Attagene = ATG; NovaScreen = NVS; Odyssey-Thera = OT; Tox21 platform = TOX21; Steroidogenesis Assay = HT-H295R;Radioactive iodide uptake assay (RAIU) assay = hNIS; high-throughput screening = HTS.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13963

developed (i.e., human type I (DIO1), II (DIO2), and III(DIO3) deiodinases, thyroid stimulating hormone receptor(TSHR), and iodotyrosine deiodinase (IYD);9 Personal com-munication S. Degitz, April 2018). Each target was evaluatedwith SeqAPASS using default settings (Table 1). Relevantdomains were either not identified or encompassed nearly thefull length of the full protein sequence for DIO1, DIO2, andDIO3 in NCBI. Because the primary reports for SeqAPASSLevel 1 identify hits with at least one common domain to thequery protein, very few protein hits were found for DIO1,DIO2, and DIO3. Therefore, for these proteins, the full reports(as opposed to the default primary reports) were used todescribe the SeqAPASS data (see SeqAPASS user guide fordetailed description of full report22).

■ RESULTS AND DISCUSSIONSeqAPASS results can be used as an initial line of evidence fordefining the taxonomic domain of applicability for HTS assays,with the recognition that the tool specifically identifies aprotein conservation across species. Such evidence of proteinconservation indicates that chemicals identified as modulatorsof a particular protein (or gene) in an assay developed with aspecific model organism are likely to act as such in otherorganisms that have the same protein. This information can beuseful for both extrapolation of HTS results to other taxa or inidentifying an organism for which assay development might bewarranted due to unique biology not covered by current HTSassays. The level of detail used in the SeqAPASS evaluation isdependent on the purpose. The intent of the present study wasto provide an initial understanding of how broadly the HTS datamay be reasonably extrapolated across species. All 484 proteinscorresponding to ToxCast assay gene targets were evaluated.Consequently, it is not feasible to discuss the results for everytarget in detail. Instead, two protein groups were selected fordiscussion, G protein coupled receptor and growth factor. Thedescription of these data is intended to provide an illustrationof the available data and methods for application.SeqAPASS To Evaluate Select ToxCast Assay Groups.

G protein-coupled receptors are transmembrane cell surfacereceptors that bind extracellular substances and generate intra-cellular signals.26 There are 79 GPCR-related HTS assays forthese receptors in the ToxCast suite, representing 71 uniquemammalian protein targets (Supplemental Data, Table S1).The SeqAPASS evaluations of several of these (e.g., human Beta-2 adrenergic receptor, Norway rat neurotensin receptor type 1A,human D(1A) dopamine receptor, cattle adenosine receptor)provided evidence of conservation across both vertebrates andinvertebrates. At least 62 of the 71 GPCRs were conservedamong vertebrates including Mammalia, Testudines, Aves,Crocodylia, Lepidosauria, Amphibia, Actinopteri, Coelacanthi-formes, and Chondrichthyes. Between 28 and 41 gpcr wereconserved across invertebrate taxon such as Insecta, Merosto-mata (horseshoe crab, sea scorpions), Arachnida, Branchiopoda(water fleas), Bivalvia, Malacostraca, and Gastropoda. Flower-ing plants were rather limited in their similarity to the mam-malian GPCRs with only the Liliopsida taxonomic groupshown to have a glutamate receptor aligning with the Norwayrat gamma-aminobutyric acid (GABA) B receptor 1.There is only one growth factor protein represented in

ToxCast, human transforming growth factor beta 1 (TGF-β),which is involved in intercellular signaling during develop-ment.27 On the basis of Level 1 SeqAPASS analysis, thisprotein was only found in vertebrates and Trematoda (i.e.,

parasitic flatworms). Two domains for human TGF-β wereidentified as specific hits in the NCBI conserved domaindatabase and evaluated using SeqAPASS Level 2. The firstdomain is from the protein family TGF-β propeptide, while thesecond is a disulfide-linked homo- or heterodimer. Both domainsare conserved across vertebrates and Trematoda, providing addi-tional evidence of conservation.On the basis of SeqAPASS evaluation of proteins in all other

assay groups, conservation was broadly identified acrossvertebrates, invertebrates, and even plants. As such, the initialSeqAPASS data (using default settings) provides a line ofevidence that screening results generated from mammalian-based ToxCast assays are more broadly applicable beyond themodel organism used in the assay itself. These data are nowavailable through both SeqAPASS and the CompTox ChemistryDashboard (Supplemental Data, Tables S3 and S4). Below, wepresent three illustrative case examples of how SeqAPASS data caninform the EDSP and OECD for understanding the taxonomicdomain of applicability for available and developing HTS assays.

Protecting Human Health and Wildlife from Endo-crine Disruption. The EDSP and OECD are exploring theutility of HTS data for screening to identify endocrine activechemicals for additional, higher-tier, toxicity testing.3,28 Assaysincluded in the ToxCast HTS screening battery suitable forthis purpose capture aspects of estrogen, androgen, or thyroidhormone signaling pathways, as well as steroidogenesis. Empiri-cal studies that evaluate conservation of key protein componentsof these pathways across taxonomic groups are relativelylimited, so a simple pragmatic approach is needed to providean initial understanding of the taxonomic domain of applica-bility for the HTS results. We previously described use ofSeqAPASS to evaluate cross-species conservation of one keyendocrine target, the estrogen receptor.13 Herein, three addi-tional SeqAPASS case studies have been developed, focusingon the AR, thyroid axis, and steroidogenic enzymes. In addi-tion to defining the taxonomic domain of applicability of theHTS data, the results of these SeqAPASS analyses could beused to focus higher-tiered toxicity testing on those speciesthat are most likely susceptible to chemicals that can act onrelevant protein targets in these endocrine pathways. Addi-tionally, the results from SeqAPASS may inform the develop-ment of additional HTS assays to capture unique biology of aparticular taxonomic group.

Taxonomic Domain of Applicability for AR HTSAssays. Eleven HTS assays using human, chimpanzee, andNorway rat ARs are included in the ToxCast suite to screen forendocrine disrupting chemicals (Table 1).29 Because human,chimpanzee, and Norway rat SeqAPASS evaluations yieldedthe same susceptibility predictions across species, in the follow-ing analysis, we focus on the human AR as a comparative probesequence. However, these data are representative of the otherspecies (both chimpanzee and rat SeqAPASS data are providedin Supplemental Data, Figure S1a−d). The primary amino acidsequence comparison shows that the human AR and specifi-cally the ligand binding domain is conserved in vertebrate taxaincluding Mammalia, Testudines, Lepidosauria, Crocodylia,Amphibia, Aves, Coelacanthiformes, Actinopteri, Chondrich-thyes, Ceratodontimorpha (lungfish), Cladistia (eel shapedbony fish), Myxiniformes (hagfish), and Petromyzontiformes(lamprey) (Figure 1a,b; Supplemental Data, Workbook S2).However, upon further evaluation of the SeqAPASS data, itwas noted that the proteins aligning with the human AR ligandbinding domain from Myxiniformes and Petromyzontiformes

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13964

were annotated as nuclear receptors but not specifically AR(e.g., steroid receptor 2, ER1 or partial sequences for the

corticoid receptor, retinoic acid receptor 2, and retinoic acidreceptor 4). These results are consistent with previously published

Figure 1. Boxplots depicting SeqAPASS (v3.0) data illustrating the percent similarity across species compared to human (Homo sapiens) androgenreceptor (AR), examining the primary amino acid sequences (a) and the ligand binding domain (b). In each plot, the ○ represents the human ARand ● represents the species with the highest percent similarity within the specified taxonomic group. The top and bottom of each box representthe 75th and 25th percentiles, respectively. The top and bottom whiskers extend up to 1.5 times the interquartile range. The mean and medianvalues for each taxonomic group are represented by horizontal thick and thin black lines on the box, respectively. The dashed line indicates thecutoff for predictions of intrinsic susceptibility. Species from taxonomic groups that cross the susceptibility cutoff or those found above the cutoffare predicted as likely to be susceptible to chemicals that act on the query species protein target. The protein target is therefore conserved in speciespredicted to be susceptible.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13965

evolutionary evidence from genome mapping and phylogeneticanalyses demonstrating that orthologs of AR are present intetrapods, teleosts, and sharks, whereas AR is not present inlamprey.30−33 Overall, these data provide a line of evidence thatToxCast assay results with human, chimpanzee, or Norway ratARs are likely applicable to a majority of vertebrate species dueto the presence of a similar molecular (i.e., protein) target.Taxonomic Domain of Applicability for HTS Assays

Evaluating Steroidogenesis. Several enzymes are involvedin converting cholesterol to steroid hormones, including cortico-steroids and sex steroids. Sex hormones serve key functions inimmunosuppression, blood pressure, sexual differentiation, repro-duction, and fertility. Steroidogenic enzymes feature many struc-turally related CYPs. However, the coverage of all steroido-genic enzymes in the suite of ToxCast assays is minimal, withonly human CYP19A1 (i.e., aromatase) represented as a discretetarget in the HTS assays. To address this gap, the EDSPisevaluating the utility of a HTS assay that quantitativelymeasures steroid hormone production in the human adreno-cortical H295R cell line.34 This assay measures 11-deoxycortisol,deoxycorticosterone, cortisol, corticosterone, 17α-hydroxypro-gesterone, 17α-hydroxypregnenolone, progesterone, pregneno-lone, dehydroepiandrosterone (DHEA), androstenedione,testosterone, estrone, and estradiol in control and chemicallytreated cells (Supplemental Data, Figure S2). Therefore, datafrom the HT-H295R assay provides profiles of steroid synthe-sis that can serve as indirect measures of chemical perturbationof enzyme activity/function.As an initial step in determining the taxonomic domain of

applicability of the HT-H295R assay results, it is important tounderstand whether other species also have similar steroido-genic enzymes to humans. To this end, several key enzymeswere queried using the SeqAPASS tool (Table 1; SupplementalData, Workbook S2). The human CYP11B1 primary aminoacid sequence was conserved across vertebrate taxa, with theexception of Chondrichthyes, Aves, and Petromyzontiformes(Table 2; Supplemental Data, Figure S3a). From these results,it could be hypothesized that the biosynthesis of corticosteronemay be different for sharks, skates, rays, birds, and lamprey thanfor other vertebrates. Consistent with SeqAPASS results, CYP11Bhas not been identified in any species of Chondrichthyes.35

In fact, Chondrichthyes have been shown to convert 11-deoxy-corticosterone to a unique steroid, 1α-hydroxycorticosterone,which has been reported to function as a corticosteroid.36 Theenzyme responsible for the formation of 1α-hydroxycorticos-terone in Chondrichthyes has yet to be identified.35 Alsoconsistent with SeqAPASS results, CYP11B gene was reportedto be found only in the kiwi but not in other birds.37 TheBrown kiwi (Apteryx spp.) was present in the SeqAPASS data,having a predicted protein sequence for CYP27C isoform X1aligning with the human CYP11B1 with 19.7% similarity. It hasbeen reported that lamprey species do not have a CYP11Bortholog, which is consistent with their lack of serum cortisoland corticosterone.38,39

Steroid 21-hydroxylase catalyzes the conversion of proges-terone to 11-deoxycorticosterone and 17α-OH-progesterone to11-deoxycortisol. Comparative evaluation of human CYP21A2with SeqAPASS provided evidence that the enzyme wasconserved in a majority of vertebrate taxa evaluated with theexception of Crocodylia and Testudines (Table 2; SupplementalData, Figure S3b). In these taxa, a putative “PREDICTED:steroid 17-alpha-hydroxylase/17,20 lyase” aligned most closelywith the human enzyme.

There is a group of 14 human enzymes known as 17β-hydroxy-steroid dehydrogenases which perform various reactions withinthe steroidogenesis pathway. However, due to the well-established roles of 17βHSD1 in the interconversion of estroneand estradiol and 17βHSD3 in the reduction of DHEA,5α-androstanedione, and androsterone to precursors of testo-sterone and dihydrotestosterone, these two enzymes were thefocus of the SeqAPASS evaluation (Table 1).40 For 17βHSD1,conservation was limited to vertebrates with the exception ofLepidosauria and Chondrichthyes (both had retinol dehydrogen-ase 8-like align with human 17βHSD1) (Table 2; SupplementalData, Figure S3c). Results of a Baker et al.38 study of the evolu-tion of enzymes involved in adrenal and sex steroid biosyn-thesis support SeqAPASS results and found that Chondrich-thyes do not contain an ortholog to human17βHSD1.SeqAPASS analysis of human CYP11A1, CYP19A, and

17βHSD3 provided evidence for conservation of these enzymesacross all vertebrates but not invertebrates (Table 2;Supplemental Data, Figure S3d−f). However, humanCYP17A1, the enzyme responsible for converting pregnenoloneand progesterone to 17-alpha-hydroxylated products and furtherto dehydroepiandrosterone and androstenedione, was broadlyconserved across both vertebrate and invertebrate taxa(Table 2; Supplemental Data, Figure S3g). The same wastrue for HSD3B, the enzyme that catalyzes the conversion ofpregnenolone to progesterone, 17α-hydroxypregnenolone to17α-hydroxyprogesterone, and DHEA to androstenedione(Table 2; Supplemental Data, Figure S3h).Overall, the SeqAPASS analyses of enzymes involved in

steroidogenesis suggest that results from the human cell-basedHT-H295R assay may be broadly extrapolated to other vertebratesbut not invertebrates (Supplemental Data, Workbook S2). How-ever, caution should be applied in extrapolating screening resultsinvolving glucocorticoid synthesis to Chondrichthyes, Aves,and Petromyzontiformes, as enzymes involved in biosynthesismay differ from humans in these taxa. Additionally, theapparent lack of 17βHSD1 in Lepidosauria and Chondrich-thyes indicate that estrone and 17β-estradiol profiles measuredby the HT-H295R assay may not be reflective of potentialimpacts to species in these groups.

Taxonomic Domain of Applicability for HTS AssaysEvaluating Thyroid Axis Disruption. The thyroid axis isessential in vertebrate development, growth, and neurogenesis.41

There are several proteins that are critical to the normal functionof the human thyroid axis (Supplemental Data, Figure S4).Accordingly, several HTS assays are under development oralready being used in the EDSP to screen for chemicals that acton key proteins in the thyroid axis that could, if disrupted by achemical stressor, lead to adverse health effects.The current ToxCast suite includes assays for three thyroid

axis targets: THRα, THRβ, and TPO. The THRα and THRβbind triiodotyrosine (T3) and regulate subsequent thyroidhormone responsive gene transcription.42 Thyroid peroxidaseis an enzyme present at the apical cell surface of thyrocytes thatoxidizes iodide for incorporation into thyroglobulin andproduction of thyroid hormones, thyroxine (T4) and T3.43

In addition to these relatively well-established potential thyroidtargets, there are other proteins in the axis that could besubject to chemical perturbation and, hence, would bedesirable to capture with HTS assays. For example, a 96-wellplate, stable human cell-based (sodium/iodide symporter-expressing HEK293T cell line), radioactive iodide uptake(RAIU) assay was recently developed to screen for inhibitors

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13966

Table

2.SeqA

PASS

Level1Predictions

ofCon

servationof

Mam

malianSteroido

genicTargets

AcrossTaxaa

queryprotein

CYP1

1A1

CYP1

7A1

HSD

3B2

CYP1

9A1

17βH

SD1

17βH

SD3

CYP2

1A2

CYP1

1B1

SeqA

PASS

cutoffc

32.92

30.95

16.16

46.79

38.91

38.40

33.59

37.60

vertebrates

Mam

malia

yes(115

of127)

yes(113

of153)

yes(117

of118)

yes(111

of121)

yes(110

of116)

yes(112

of113)

yes(96of

142)

yes(108

of127)

Actinopteri

yes(58of

74)

yes(63of118)

yes(59of59)

yes(137

of147)

yes(19of49)

yes(42of

49)

yes(20of

118)

yes(38of

76)

Amphibia

yes(8

of8)

yes(5

of6)

yes(4

of4)

yes(8

of10)

yes(3

of3)

yes(3

of4)

yes(3

of6)

yes(6

of8)

Aves

yes(45of

65)

yes(61of

71)

yes(73of

73)

yes(65of67)

yes(16of66)

yes(45of

66)

yes(4

of71)

no(0

of70)

Chondrichthyes

yes(4

of5)

yes(3

of4)

yes(4

of4)

yes(4

of5)

no(0

of2)

yes(2

of2)

yes(1

of4)

no(0

of5)

Coelacanthiform

esyes(1

of1)

yes(1

of1)

yes(1

of1)

−yes(1

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

Crocodylia

yes(2

of4)

yes(4

of4)

yes(4

of4)

yes(4

of4)

yes(3

of3)

yes(4

of4)

no(0

of4)

yes(1

of3)

Lepidosauria

yes(4

of8)

yes(7

of8)

yes(7

of7)

yes(6

of6)

no(0

of7)

yes(7

of7)

yes(4

of8)

yes(5

of8)

Myxiniform

esyes(1

of1)

−−

−−

−−

no(0

of1)

Testudines

yes(3

of3)

yes(4

of6)

yes(3

of3)

yes(7

of7)

yes(2

of3)

yes(3

of3)

no(0

of6)

yes(1

of4)

invertebratesb

Anthozoa

no(0

of5)

yes(3

of5)

yes(3

of4)

no(0

of5)

no(0

of5)

no(0

of5)

no(0

of5)

no(0

of5)

Arachnidia

no(0

of15)

no(0

of11)

yes(8

of9)

no(0

of15)

no(0

of2)

no(0

of14)

no(0

of10)

no(0

of15)

Bivalvia

no(0

of8)

no(0

of7)

yes(4

of4)

no(0

of10)

no(0

of1)

no(0

of4)

no(0

of7)

no(0

of5)

Branchiopoda

no(0

of2)

no(0

of2)

yes(1

of2)

no(0

of2)

−no

(0of

2)no

(0of

2)no

(0of

2)Branchiostomidae

no(0

of2)

yes(2

of2)

yes(2

of2)

no(0

of2)

−no

(0of

2)no

(0of

2)no

(0of

2)Gastropoda

no(0

of5)

no(0

of3)

yes(2

of3)

no(0

of6)

no(0

of1)

no(0

of5)

no(0

of3)

no(0

of8)

Insecta

no(0

of176)

no(0

of137)

yes(8

of56)

no(0

of172)

no(0

of4)

no(0

of117)

no(0

of135)

no(0

of189)

aTable

cells

indicate

whether

theproteinisconservedcomparedto

thehuman

queryprotein,

“yes”,or

less

likelyto

beconserved,

“no.”In

parenthesesarethenumberof

speciesfoundabovethe

SeqA

PASS

cutoffof

thetotaln

umberof

speciesaligning

with

themam

maliantargetforthattaxonomicgroup.Dashedlineindicatesthatno

proteinsequencesalignedwith

thequerysequence

forthat

taxonomicgroup.

Abbreviations

andqueryproteinaccessions:CYP1

1A1=human

cholesterolside-chain

cleavage

enzyme(P05108.2);CYP1

7A1=human

steroid17-alpha-hydroxylase/17,20

lyase

(P05093.1);HSD

3B2=

human

3beta-hydroxysteroiddehydrogenase/Delta

5→

4-isom

erasetype

2(P26439.2);CYP1

9A1=

arom

atase(P11511.3);17βH

SD1=

human

estradiol17-beta-

dehydrogenase1(P14061.3);17βH

SD3=

human

testosterone

17-beta-dehydrogenase3(P37058.2);CYP2

1A2=

human

steroid21-hydroxylase

(AFK

10138.1);CYP1

1B1=

human

11-beta-

hydroxylase(A

AA35741.1).bInvertebratetaxonrepresenta

subsetof

thosefoundto

beconservedforCYP1

7A1andHSD

3B2.Those

invertebratetaxa

show

nrepresentcom

mon

taxonomicgroups

(full

taxonlistinSupplementalD

ata,WorkbookS2).Invertebrateswereshow

nto

beconservedforonlyCYP1

7A1andHSD

3B2.c The

SeqA

PASS

cutoffsdefine

thosespecieswith

conservedproteins

found

abovethecutoffandthosespeciesless

likelyto

beconservedbelowthecutoff(detailsin

SupplementalData,TableS2).

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13967

Table

3.SeqA

PASS

Level

1and2Predictions

ofCon

servationof

Mam

malianThyroid

Targets

AcrossTaxaa

queryprotein

THRα

THRβ

TPO

NIS

DIO

1DIO

2DIO

3TSH

RIYD

SeqA

PASS

level

12

12

12

11

11

12

12

SeqA

PASS

cutoffc

34.53

70.03

33.15

28.04

38.75

54.65

44.54

25.83

40.00

16.77

48.42

82.57

24.53

33.62

vertebrates

Mam

malia

yes (113

of-

121)

yes (115

of-

162)

yes (115

of-

167)

yes (113

of-

152)

yes (94of-

117)

yes (84of-

141)

yes (100

of-

110)

yes (126

of-

127)

yes (123

of-

127)

yes (127

of-

127)

yes (114

of-

157)

yes (113

of-

181)

yes (106

of-

106)

yes (111

of-

111)

Actinopteri

yes (68of121)

yes (71of127)

yes (71of120)

yes (78of117)

yes (17of53)

yes (19of63)

yes (48of63)

yes (62of

70)

yes (52of

70)

yes (69of

70)

yes (52of-

1409)

yes (1of2169)

yes (45of45)

yes (45of

45)

Amphibia

yes (12of

24)

yes (13of

29)

yes (15of

23)

yes (28of

28)

yes(3

of3)

yes(3

of3)

yes(3

of3)

yes(4

of10)

yes(4

of10)

yes(7

of10)

yes(2

of16)

yes(3

of16)

yes(3

of3)

yes(3

of3)

Aves

yes (68of

80)

yes (68of

75)

yes (69of

80)

yes (75of

80)

yes (58of68)

yes (58of68)

yes (21of67)

yes (70of

70)

yes (63of

70)

yes (70of

70)

yes (67of108)

yes (68of137)

yes (66of137)

yes (66of

68)

Ceratodontim

orpha

no(0

of2)

no(0

of2)

no(0

of2)

no(0

of2)

−−

−yes(1

of1)

yes(1

of1)

yes(1

of1)

no(0

of1)

no(0

of4)

−−

Chondrichthyes

yes(3

of6)

yes(3

of6)

yes(3

of7)

yes(5

of7)

yes(2

of2)

yes(2

of3)

yes(2

of4)

yes(3

of3)

yes(3

of3)

yes(3

of3)

yes(1

of9)

yes(1

of11)

yes(2

of2)

yes(2

of2)

Coelacanthiform

esyes(1

of2)

yes(1

of2)

yes(1

of2)

yes(1

of1)

no(0

of1)

no(0

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

yes(1

of1)

Crocodylia

yes(4

of6)

yes(4

of5)

yes(4

of6)

yes(4

of5)

yes(4

of4)

yes(4

of4)

yes(3

of4)

yes(4

of4)

yes(4

of4)

yes(4

of4)

yes(4

of4)

yes(4

of4)

yes(3

of3)

yes(4

of4)

Lepidosauria

yes(8

of15)

yes(8

of13)

yes(8

of15)

yes(9

of11)

yes(6

of6)

yes(6

of7)

yes(4

of6)

yes(9

of9)

yes(7

of9)

yes(8

of9)

yes (6

of341)

yes (7

of422)

yes(6

of6)

yes(6

of6)

Myxiniform

esyesd

(0of1)

yesd

(0of

2)no

(0of

1)yes(1

of2)

no(0

of1)

no(0

of1)

−−

−−

no(0

of1)

no(0

of1)

−−

Petrom

yzontiformes

yes(1

of4)

yes(1

of4)

yes(1

of4)

yes(4

of4)

−no

(0of1)

−yes(1

of1)

yes(1

of1)

yes(1

of1)

no(0

of5)

no(0

of5)

−−

Testudines

yes(3

of9)

yes(3

of5)

yes(3

of9)

yes(3

of4)

yes(3

of3)

yes(3

of3)

yes(2

of3)

yes(3

of3)

yes(3

of3)

yes(3

of3)

yes(3

of4)

yes(3

of26)

yes(3

of3)

yes(3

of3)

invertebratesb

Ascidiacea

no(0

of5)

no(0

of5)

yes(1

of5)

yes(2

of5)

no(0

of2)

yes(1

of2)

no(0

of1)

yes(2

of2)

no(0

of2)

yes(2

of2)

no(0

of1)

no(0

of1)

−−

Asteroidea

no(0

of1)

no(0

of1)

yes(1

of1)

yes(1

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of2)

no(0

of3)

yes(1

of1)

yes(1

of1)

Bivalvia

no(0

of10)

no(0

of10)

yes(2

of10)

yes(3

of10)

no(0

of6)

no(0

of8)

no(0

of4)

yes(4

of5)

no(0

of5)

yes(4

of5)

no(0

of4)

no(0

of5)

−−

Branchiostomidae

no(0

of4)

no(0

of4)

yes(3

of4)

yes(4

of4)

no(0

of3)

no(0

of4)

no(0

of2)

yes(2

of2)

no(0

of2)

yes(2

of2)

no(0

of4)

no(0

of4)

yes(2

of2)

yes(2

of2)

Dictyosteliida

−−

−−

−−

−no

(0of6)

no(0

of6)

yes(1

of5)

−−

−−

Enteropneusta

no(0

of1)

no(0

of1)

yes(1

of1)

yes(1

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

−yes(1

of1)

Gastropoda

no(0

of9)

no(0

of9)

yes(2

of2)

yes(8

of9)

no(0

of3)

no(0

of4)

no(0

of3)

no(0

of2)

no(0

of2)

no(0

of2)

no(0

of5)

no(0

of5)

−−

Lingulata

no(0

of1)

no(0

of1)

yes(1

of1)

yes(1

of1)

no(0

of1)

no(0

of1)

no(0

of1)

yes(1

of1)

no(0

of1)

yes(1

of1)

yesd

(0of1)

yesd

(0of1)

yes(1

of1)

yes(1

of1)

Polychaeta

no(0

of2)

no(0

of1)

yes(1

of2)

yes(1

of1)

no(0

of5)

no(0

of10)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of1)

no(0

of2)

no(0

of2)

yes(1

of1)

yes(1

of1)

aTablecells

indicate

whether

theproteinisconservedcomparedto

thehuman

queryprotein,

“yes”,or

less

likelyto

beconserved,

“no.”In

parenthesesarethenumberof

speciesfoundabovethe

SeqA

PASS

cutoffof

thetotaln

umberof

speciesaligning

with

themam

maliantargetforthattaxonomicgroup.Dashedlineindicatesthatno

proteinsequencesalignedwith

thequerysequence

forthat

taxonomicgroup.Abbreviations:T

HRα=human

thyroidhorm

onereceptor

alpha(P10827.1;domaincd06935);T

HRβ=human

thyroidhorm

onereceptor

beta(P10828.2;domaincd06935);T

PO=

ratthyroidperoxidase

(P14650.1;

domaincd09825);NIS

=human

sodium

/iodidecotransporter(Q

92911.1);DIO

1=human

type

Iiodothyroninedeiodinase

(P49895.3);DIO

2=human

type

IIiodothyroninedeiodinase

(Q92813.4);DIO

3=human

type

IIIiodothyroninedeiodinase

(P55073.4);TSH

R=human

thyroid-stimulatinghorm

onereceptor

(P16473.2;

domaincd15964);IYD

=iodotyrosine

deiodinase

1(Q

6PHW0.2;domaincd02144).bInvertebratetaxonrepresenta

subsetof

thosefoundto

beconservedforTHRβ,TPO

,NIS,D

IO1,DIO

3,andIYD.T

hose

invertebratetaxa

show

nrepresentcom

mon

taxonomicgroups

(fulltaxon

listinSupplementalD

ata,WorkbookS2).c The

SeqA

PASS

cutoffsdefine

thosespecieswith

conservedproteins

foundabovethecutoffandthose

specieslesslikelyto

beconservedbelowthecutoff(detailsin

SupplementalD

ata,TableS2).dPredictedas

conserved“yes,”becausespeciesin

Myxiniform

esforTHRαandin

LingulataforTSH

Rare

ortholog

candidates,alth

ough

thetaxonomicgroupisbelowthecutoff.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13968

of sodium/iodide symporter (NIS)-mediated iodide uptake.18

The NIS utilizes a Na+ gradient to allow for iodide uptake intothe thyroid cell and is the first rate-limiting step for synthesis ofT3 and T4 (Supplemental Data, Figure S4).18

Additional HTS assays are being developed to screen forchemicals that disrupt human DIO1, DIO2, DIO3, TSHR, andIYD.20,9 To provide insights as to how broadly results fromcurrently available and developing HTS assays may reasonablybe extrapolated across species, SeqAPASS was used to evaluatesequence similarity of all assay targets used or being developedto screen the thyroid axis (Table 1; Supplemental Data,Workbook S2).Levels 1 and 2 evaluations of human THRα and THRβ and

their respective ligand binding domains showed that thesereceptors are well conserved across vertebrates, with the excep-tion of Ceratodontimorpha (lungfish). Conservation of THRβbut not THRα is also found for several invertebrate taxa, includ-ing Polychaeta (sandworms), Gastropoda, Lingulata (lamp-shells),Bivalvia, Enteropneusta (Acorn worms), Asteroidea (star-fish), Branchiostomidae (lancelet), and Ascidiacea (sea squirts,tunicate) (Table 3; Supplemental Data, Figure S5a−d). TheseSeqAPASS results are concordant with results presented byHolzer and Laudet44 describing the evolution of thyroid hor-mone signaling across species.Norway rat and pig TPO sequences and their peroxidase

functional domains, which contain the heme binding site andputative substrate binding site, were conserved across vertebrates,as well as Ascidiacea (sea squirts, tunicate) taxa (Table 3; Supple-mental Data, Figure S5e−h). Conversely, probable orthologs ofhuman NIS were restricted to vertebrates only (Table 3;Supplemental Data, Figure S5i).The deiodinase enzymes activate and deactivate thyroid

hormones. The SeqAPASS results for DIO1, DIO2, and DIO3demonstrate conservation of these enzymes across vertebratespecies, including Ceratodontimorpha, Coelacanthiformes, andPetromyzontiformes (Table 3; Supplemental Data, Figure S5j−l).DIO1 and DIO3 are also found in invertebrate species,whereas DIO2 iodothyronine deiodinase is not. DIO3 has thebroadest conservation across invertebrates and is even found inDictyosteliida (slime molds) taxa.Through binding of thyroid stimulating hormone, the TSHR

signaling modulates thyroid function by regulating genescoding for enzymes involved in iodide metabolism, such as TPOand NIS. The human TSHR is well conserved across vertebrateswith the exception of Ceratodontimorpha, Myxiniformes, andPetromyzontiformes (Table 3; Supplemental Data, Figure S5m).Further, SeqAPASS results yielded the same conclusion uponevaluation of the TSHR functional domain containing theputative peptide ligand binding pocket (Table 3; SupplementalData, Figure S5n).The final protein evaluated using SeqAPASS was the human

IYD, involved in the recycling of iodide. This protein hadortholog candidates across vertebrates and invertebrates, displayingconservation in all but one taxon, Trichomonadida (anaerobicprotists), based on the primary amino acid sequence that alignedwith human (Table 3; Supplemental Data, Figure S5o). However,upon evaluation of the functional domain (positions 93−285),which contains the putative dimer interface, additional taxa wereidentified that were not conserved compared to the humaniodotyrosine deiodinase (Supplemental Data, Figure S5p).Overall, the protein targets associated with ToxCast assays

intended to screen for thyroid axis disruption were conservedacross all vertebrate species, in some cases identifying

conservation in more ancient fish species such as hagfish,Coelacanth, lamprey, and lungfish (Supplemental Data, Work-book S2). Therefore, it would be anticipated that screeningresults identifying chemical disruption via HTS assays for thehuman thyroid axis targets could be reasonably extrapolatedacross vertebrate species. Additionally, THRβ, TPO, IYD,DIO1, and DIO3 are present in certain invertebrates, suggest-ing perhaps applicability of these assays to an even broaderarray of taxa. However, additional research is needed to under-stand the actual role of these proteins in invertebrate species.The overall objectives of this work were to evaluate mammalian-

based HTS assay targets for protein similarity using the SeqAPASStool, make the resultant data available for other researchers andregulators (via SeqAPASS and the CompTox Chemistry Dash-board), and demonstrate the utility of the analysis as applied to acurrent regulatory challenge, screening of endocrine-activechemicals. This study illustrates that HTS targets relevant toEDSP and OECD are likely to be broadly applicable acrossmost vertebrate taxa and some targets may be applicable tocertain invertebrates. Ultimately, application of the SeqAPASStool can lay the foundation for future work integrating in vitroand in silico approaches to support more rapid evaluations ofchemical safety for the protection of human health and wildlife.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.est.8b04587.

Supplemental data, including detailed information onthe SeqAPASS analyses for all ToxCast targets and datavisualization to accompany the SeqAPASS results (PDF)Supplemental data, workbook (XLSX)Raw reports from SeqAPASS Level 1 and Level 2 for allprotein targets and summary reports for Level 1 andLevel 2 (ZIP)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: LaLone.Carlie@epa.gov.ORCIDCarlie A. LaLone: 0000-0003-3174-1314Daniel L. Villeneuve: 0000-0003-2801-0203Brett R. Blackwell: 0000-0003-1296-4539Gerald T. Ankley: 0000-0002-9937-615XNotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe thank Dr. M. Hornung for providing thoughtful reviewcomments on an earlier version of the paper. This manuscripthas been reviewed in accordance with the requirements of theUS EPA Office of Research and Development; however, therecommendations made herein do not represent US EPApolicy. Mention of products or trade names does not indicateendorsement by the US EPA.

■ REFERENCES(1) National Research Council. Toxicity Testing in the 21st Century:A Vision and Strategy; The National Academies Press: Washington,DC, 2007.(2) Judson, R. S.; Magpantay, F. M.; Chickarmane, V.; Haskell, C.;Tania, N.; Taylor, J.; Xia, M.; Huang, R.; Rotroff, D. M.; Filer, D. L.;

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13969

Houck, K. A.; et al. Integrated model of chemical perturbations of abiological pathway using 18 in vitro high-throughput screening assaysfor the estrogen receptor. Toxicol. Sci. 2015, 148 (1), 137−154.(3) Browne, P.; Noyes, P. D.; Casey, W. M.; Dix, D. J. Application ofAdverse Outcome Pathways to US EPA’s Endocrine DisruptorScreening Program. Environ. Health Perspect. 2017, 125, 096001.(4) Pinto, C. L.; Markey, K.; Dix, D.; Browne, P. Identification ofcandidate reference chemicals for in vitro steroidogenesis assays.Toxicol. In Vitro 2018, 47, 103−119.(5) US Environmental Protection Agency. Endocrine disruptorscreening program; [accessed 2018 February 15]. Available fromhttp://www.epa.gov/endo/.(6) Kavlock, R.; Chandler, K.; Houck, K.; Hunter, S.; Judson, R.;Kleinstreuer, N.; Knudsen, T.; Martin, M.; Padilla, S.; Reif, D.;Richard, A.; Rotroff, D.; Sipes, N.; Dix, D. Update on EPA’s ToxCastprogram: providing high throughput decision support tools forchemical risk management. Chem. Res. Toxicol. 2012, 25 (7), 1287−1302.(7) Richard, A. M.; Judson, R. S.; Houck, K. A.; Grulke, C. M.;Volarath, P.; Thillainadarajah, I.; Yang, C.; Rathman, J.; Martin, M.T.; Wambaugh, J. F.; Knudsen, T. B.; Kancherla, J.; Mansouri, K.;Patlewicz, G.; Williams, A. J.; Little, S. B.; Crofton, K. M.; Thomas, R.S. ToxCast Chemical Landscape: Paving the Road to 21st CenturyToxicology. Chem. Res. Toxicol. 2016, 29 (8), 1225−1251.(8) Tice, R. R.; Austin, C. P.; Kavlock, R. J.; Bucher, J. R. Improvingthe human hazard characterization of chemicals: a Tox21 update.Environ. Health Perspect. 2013, 121 (7), 756.(9) Hornung, M. W.; Korte, J. J.; Olker, J. H.; Denny, J. S.; Knutsen,C.; Hartig, P. C.; Cardon, M. C.; Degitz, S. J. Screening the ToxcastPhase 1 Chemical Library For Inhibition of Deiodinase Type 1Activity. Toxicol. Sci. 2018, 162 (2), 570−581.(10) Truong, L.; Reif, D. M.; St Mary, L.; Geier, M. C.; Truong, H.D.; Tanguay, R. L. Multidimensional in vivo hazard assessment usingzebrafish. Toxicol. Sci. 2014, 137 (1), 212−33.(11) LaLone, C. A.; Villeneuve, D. L.; Lyons, D.; Helgen, H. W.;Robinson, S. L.; Swintek, J. A.; Saari, T. W.; Ankley, G. T. Editor’sHighlight: Sequence Alignment to Predict Across Species Suscepti-bility (SeqAPASS): A Web-Based Tool for Addressing the Challengesof Cross-Species Extrapolation of Chemical Toxicity. Toxicol. Sci.2016, 153 (2), 228−45.(12) Gunnarsson, L.; Jauhiainen, A.; Kristiansson, E.; Nerman, O.;Larsson, D.G. J. Evolutionary Conservation of Human Drug Targetsin Organisms use dfor Environmental Risk Assessments. Environ. Sci.Technol. 2008, 42 (15), 5807−5813.(13) Ankley, G. T.; LaLone, C. A.; Gray, L. E.; Villeneuve, D. L.;Hornung, M. W. Evaluation of the scientific underpinnings foridentifying estrogenic chemicals in nonmammalian taxa usingmammalian test systems. Environ. Toxicol. Chem. 2016, 35 (11),2806−2816.(14) Kohno, S.; Bernhard, M. C.; Katsu, Y.; Zhu, J.; Bryan, T. A.;Doheny, B. M.; Iguchi, T.; Guillette, L. J., Jr Estrogen receptor 1(ESR1; ERα), not ESR2 (ERβ), modulates estrogen-induced sexreversal in the American alligator, a species with temperature-dependent sex determination. Endocrinology 2015, 156 (5), 1887−1899.(15) Tohyama, S.; Miyagawa, S.; Lange, A.; Ogino, Y.; Mizutani, T.;Ihara, M.; Tanaka, H.; Tatarazako, N.; Kobayashi, T.; Tyler, C. R.;Iguchi, T. Evolution of estrogen receptors in ray-finned fish and theircomparative responses to estrogenic substances. J. Steroid Biochem.Mol. Biol. 2016, 158, 189−197.(16) Ogino, Y.; Tohyama, S.; Kohno, S.; Toyota, K.; Yamada, G.;Yatsu, R.; Kobayashi, T.; Tatarazako, N.; Sato, T.; Matsubara, H.;Lange, A.; Tyler, C. R.; Katsu, Y.; Iguchi, T.; Miyagawa, S. Functionaldistinctions associated with the diversity of sex steroid hormonereceptors ESR and AR. J. Steroid Biochem. Mol. Biol., in press;DOI: 10.1016/j.jsbmb.2018.06.002.(17) Doering, J.; Lee, S.; Kristiansen, K.; Evenseth, L.; Barron, M.;Sylte, I.; LaLone, C. In silico site-directed mutagenesis informsspecies-specific predictions of chemical susceptibility derived from the

Sequence Alignment to Predict Across Species Susceptibility(SeqAPASS) tool. Toxicol. Sci. 2018, DOI: 10.1093/toxsci/kfy186.(18) Hallinger, D. R.; Murr, A. S.; Buckalew, A. R.; Simmons, S. O.;Stoker, T. E.; Laws, S. C. Development of a screening approach todetect thyroid disrupting chemicals that inhibit the human sodiumiodide symporter (NIS). Toxicol. In Vitro 2017, 40, 66−78.(19) Paul, K. B.; Hedge, J. M.; Macherla, C.; Filer, D. L.; Burgess, E.;Simmons, S. O.; Crofton, K. M.; Hornung, M. W. Cross-speciesanalysis of thyroperoxidase inhibition by xenobiotics demonstratesconservation of response between pig and rat. Toxicology 2013, 312,97−107.(20) Titus, S.; Neumann, S.; Zheng, W.; Southall, N.; Michael, S.;Klumpp, C.; Yasgar, A.; Shinn, P.; Thomas, C. J.; Inglese, J.;Gershengorn, M. C.; et al. Quantitative high-throughput screeningusing a live-cell cAMP assay identifies small-molecule agonists of theTSH receptor. J. Biomol. Screening 2008, 13 (2), 120−127.(21) Freitas, J.; Miller, N.; Mengeling, B. J.; Xia, M.; Huang, R.;Houck, K.; Rietjens, I. M.; Furlow, J. D.; Murk, A. J. Identification ofthyroid hormone receptor active compounds using a quantitativehigh-throughput screening platform. Curr. Chem. Genomics Transl.Med. 2014, 8, 36.(22) LaLone, C. A. Sequence Alignment to Predict Across SpeciesSusceptibility (SeqAPASS), Version 3.0 User Guide; EPA/600/R-18/062; . US EPA Office of Research and Development: Washington,DC, 2018; https://cfpub.epa.gov/si/si_public_record_report.cfm?dirEntryId=339888.(23) Marchler-Bauer, A.; Derbyshire, M. K.; Gonzales, N. R.; Lu, S.;Chitsaz, F.; Geer, L. Y.; Geer, R. C.; He, J.; Gwadz, M.; Hurwitz, D. I.;Lanczycki, C. J.; Lu, F.; Marchler, G. H.; Song, J. S.; Thanki, N.;Wang, Z.; Yamashita, R. A.; Zhang, D.; Zheng, C.; Bryant, S. H. CDD:NCBI’s conserved domain database. Nucleic Acids Res. 2015, 43,D222−226.(24) Williams, A. J.; Grulke, C. M.; Edwards, J.; McEachran, A. D.;Mansouri, K.; Baker, N. C.; Patlewicz, G.; Shah, I.; Wambaugh, J. F.;Judson, R. S.; Richard, A. M. The CompTox Chemistry Dashboard: acommunity data resource for environmental chemistry. J. Cheminf.2017, 9 (1), 61.(25) LaLone, C. A.; Villeneuve, D. L.; Burgoon, L. D.; Russom, C.L.; Helgen, H. W.; Berninger, J. P.; Tietge, J. E.; Severson, M. N.;Cavallin, J. E.; Ankley, G. T. Molecular target sequence similarity as abasis for species extrapolation to assess the ecological risk of chemicalswith known modes of action. Aquat. Toxicol. 2013, 144-145, 141−154.(26) Marinissen, M. J.; Gutkind, J. S. G-protein-coupled receptorsand signaling networks: emerging paradigms. Trends Pharmacol. Sci.2001, 22 (7), 368−376.(27) Massague, J. The transforming growth factor-beta family. Annu.Rev. Cell Biol. 1990, 6 (1), 597−641.(28) Browne, P.; Judson, R. S.; Casey, W. M.; Kleinstreuer, N. C.;Thomas, R. S. Screening chemicals for estrogen receptor bioactivityusing a computational model. Environ. Sci. Technol. 2015, 49 (14),8804−8814.(29) Kleinstreuer, N. C.; Ceger, P.; Watt, E. D.; Martin, M.; Houck,K.; Browne, P.; Thomas, R. S.; Casey, W. M.; Dix, D. J.; Allen, D.;Sakamuru, S.; et al. Development and validation of a computationalmodel for androgen receptor activity. Chem. Res. Toxicol. 2017, 30(4), 946−964.(30) Colombe, L.; Fostier, A.; Bury, N.; Pakdel, F.; Guiguen, Y.Steroids 2000, 65, 319−328.(31) Escriva, H.; Safi, R.; Hanni, C.; Langlois, M. C.; Saumitou-Laprade, P.; Stehelin, D.; Capron, A.; Pierce, R.; Laudet, V. Proc. Natl.Acad. Sci. U. S. A. 1997, 94, 6803−6808.(32) Colbert, E. H.; Morales, M. Evolution of the Vertebrates: Ahistory of the Backboned Animals Through Time, 4th ed.; Wiley-Liss:New York, 1991.(33) Thornton, J. W. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 5671−5676.(34) Karmaus, A. L.; Toole, C. M.; Filer, D. L.; Lewis, K. C.; Martin,M. T. High-throughput screening of chemical effects on steroido-

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13970

genesis using H295R human adrenocortical carcinoma cells. Toxicol.Sci. 2016, 150 (2), 323−332.(35) Anderson, W. G. The endocrinology of 1α-hydroxycorticoster-one in elasmobranch fish: a review. Comp. Biochem. Physiol., Part A:Mol. Integr. Physiol. 2012, 162 (2), 73−80.(36) Nunez, S.; Trant, J. M. Regulation of interrenal glandsteroidogenesis in the Atlantic stingray (Dasyatis sabina). J. Exp.Zool. 1999, 284 (5), 517−525.(37) Nelson, D. R. Cytochrome P450 diversity in the tree of life.Biochim. Biophys. Acta, Proteins Proteomics 2018, 1866 (1), 141−154.(38) Baker, M. E.; Nelson, D. R.; Studer, R. A. Origin of theresponse to adrenal and sex steroids: Roles of promiscuity and co-evolution of enzymes and steroid receptors. J. Steroid Biochem. Mol.Biol. 2015, 151, 12−24.(39) Close, D. A.; Yun, S. S.; McCormick, S. D.; Wildbill, A. J.; Li,W. 11-Deoxycortisol is a corticosteroid hormone in the lamprey. Proc.Natl. Acad. Sci. U. S. A. 2010, 107 (31), 13942−13947.(40) Miller, W. L.; Auchus, R. J. The Molecular Biology,Biochemistry, and Physiology of Human Steroidogenesis and ItsDisorders. Endocr. Rev. 2011, 32 (1), 81−151.(41) Gauthier, K.; Chassande, O.; Plateroti, M.; Roux, J. P.; Legrand,C.; Pain, B.; Rousset, B.; Weiss, R.; Trouillas, J.; Samarut, J. Differentfunctions for the thyroid hormone receptors TRα and TRβ in thecontrol of thyroid hormone production and post-natal development.EMBO journal 1999, 18 (3), 623−631.(42) Bassett, J. D.; Harvey, C. B.; Williams, G. R. Mechanisms ofthyroid hormone receptor-specific nuclear and extra nuclear actions.Mol. Cell. Endocrinol. 2003, 213 (1), 1−11.(43) Ruf, J.; Carayon, P. Structural and functional aspects of thyroidperoxidase. Arch. Biochem. Biophys. 2006, 445 (2), 269−277.(44) Holzer, G.; Laudet, V. New Insights into Vertebrate ThyroidHormone Receptor Evolution. Nuclear Receptor Research 2017, 4, 1−14.

Environmental Science & Technology Article

DOI: 10.1021/acs.est.8b04587Environ. Sci. Technol. 2018, 52, 13960−13971

13971