Potential Role for Endocrine Disruptor Expert Systems in Investigating ‘Endocrine System-Epigenome’ Interactions P. Schmieder EPA, ORD, NHEERL, MED-Duluth, MN 2nd McKim Cancer Workshop

Effects-based Expert System Automated rule-based decision trees to predict which chemicals have the potential to disrupt endocrine systems. This is done by: testing key chemicals within a chemical class to set boundaries on biological/toxicological activity, to predict activity of other members of the “class” Grouping chemicals by a common biological activity, then determining what is similar about the chemical structures and properties that explain their activity writing rules that help categorize similar but untested chemicals. The Program Offices use these tools to decide which, of the hundreds of chemicals on Agency chemical lists, should be evaluated first for likelihood of disrupting endocrine-mediated pathways the results in hormonal imbalances.

ER-mediated AOP Significant evidence exists linking ER binding chemicals to adverse outcomes related to reproductive effects (ER-mediated reproductive impairment AOP): hh - drug design of anti-estrogens for breast cancer research and treatment env - multiple studies linking chemical ER binders to adverse effects -potent pharmaceuticals - e.g., EE2 -weak affinity environmental chemicals - e.g., APs If there is evidence, or evidence found in future, for NR mediated AOPs linked to cancer, then the approach described here will be relevant for hypothesizing what chemical might do this. This presentation will focus on: using an ER-mediated AOP to form chemical categories for addressing a specific risk assessment application

ER-mediated AOP The risk context to which AOP is applied influences the approach: This example: Risk context – USEPA needs to evaluate large lists of data-limited chemicals for ED potential; how can predictive tools be used to identify which of these chemicals have the greatest potential to cause an adverse effect because of their estrogenic potential; Goal - Given limited testing resources, prioritize chemicals on targeted inventories, so that those with the highest likelihood of producing an adverse outcome are tested first Targeted chemical inventories: -inert ingredients in pesticides used on food crops -antimicrobial active ingredients -inert ingredients in pesticides not used on food crops

OECD Principles for QSAR Validation Well-Defined Endpoint Well-defined biological endpoint – Informing important risk endpoint – Adverse Outcome Pathway (AOP) ending in impaired reproduction; plausible linkage of measure (initiating event) to higher level adversity Well-defined chemistry Does assay allow testing of the types of chemicals (range of properties) found on regulatory inventories? Is the chemical form and concentration in the assay understood? Mechanistic interpretation Can estimates be explained mechanistically - chemistry & biology ? ER-mediated Reproductive Impairment Adverse Outcome Pathway Relationship of chemical parameters to activity

OECD Principles for QSAR Validation (cont.) Defined Model Applicability Domain Well-defined application Is the regulatory question well-defined – priority setting is different than risk assessment? Is the QSAR model domain coverage well-defined? Does the QSAR chemical domain adequately cover the regulatory chemical domain i.e., the regulatory question? Appropriate measures of goodness of fit, robustness, ability to predict Measures appropriate for a regression model likely not appropriate to evaluate an expert system Unambiguous algorithm Expert Systems – logic tree, rules/queries, supporting information

Application of OECD Principles to Forming Chemical Categories: Key Questions Transparency How reasonable is the estimate compared with data for similar chemicals ? Can the QSAR estimate be explained mechanistically? Usefulness Are the predictions applicable to all the chemicals of regulatory concern? Does the model/expert system answer the regulatory question?

Mechanistic Basis of an Expert System to Predict Potential for Chemical Binding to the Estrogen Receptor ES development based on a defined AOP ER-mediated reproductive impairment adverse outcome pathway ER-mediated liver cell proliferation in fish ER Binding Domain Knowledge/theories of chemical-receptor interactions How chemicals interact with ER protein – LBD sub-pockets The Regulatory Chemical Domain Characterizing the FI and AM inventory chemicals Building from existing information The Receptor Binding Assay Domain Optimizing assays considering properties of inventory chemicals

Adverse Outcome Pathway ER-Mediated Reproductive Impairment Chemical effects across levels of biological organization QSAR focus area In vitro Assay focus area In vivo Inerts; Antimicrobial Chemicals Chemicals CELLULAR Response POPULATION MOLECULAR Target TISSUE/ORGAN INDIVIDUAL Skewed Sex Ratios; Yr Class Liver Altered proteins(Vtg)& hormones; Gonad Ova-testis; Complete ovary in male Liver Cell Protein Expression Vitellogenin (egg protein transported to ovary) Sex reversal; Altered behavior; Repro. Receptor Binding ER Binding Series of events that can be measured across levels of biological organization Toxicity Pathway Adverse Outcome Pathway Greater Toxicological Understanding Greater Risk Relevance

Adverse Outcome Pathway ER-mediated Reproductive Impairment QSAR focus area In vitro Assay focus area Inerts; Antimicrobial Chemicals CELLULAR Response POPULATION MOLECULAR Target TISSUE/ORGAN INDIVIDUAL Skewed Sex Ratios; Yr Class Liver Altered proteins(Vtg) Liver Cell Protein Expression Vitellogenin (egg protein transported to ovary) Sex reversal; Altered behavior; Repro. Receptor Binding ER Binding Toxicity Pathway Adverse Outcome Pathway Greater Toxicological Understanding Greater Risk Relevance

Mechanistic Basis of the Expert System ER Binding Affinity: An Indicator of Potential Reproductive Effects ER-mediated reproductive impairment adverse outcome pathway ER Binding Domain Knowledge/theories of chemical-receptor interactions ER sub-pockets The Regulatory Chemical Domain Characterizing the FI and AM inventory chemicals Building from existing information The Receptor Binding Assay Domain Optimizing assays considering properties of inventory chemicals

ER Binding Domain - Bioassays for ER binding were available from drug-design, although extant methods were focused on potent anti-estrogens - Drug-design research provides mechanistic insights on multiple types of interaction within the ER binding site

Distance = 10.8 for 17-Estradiol A_B Interaction T 347 C QOxygen= -0.25 QOxygen= -0.32 E 353 H 524 H CH3 H A B HO OH R 394 H H Distance = 10.8 for 17-Estradiol J. Katzenellenbogen

Mechanistic Basis of the Expert System ER Binding Affinity: An Indicator of Potential Reproductive Effects ER-mediated reproductive impairment adverse outcome pathway ER Binding Domain Knowledge/theories of chemical-receptor interactions ER sub-pockets The Regulatory Chemical Domain Characterizing the FI and AM inventory chemicals Building from existing information The Receptor Binding Assay Domain Optimizing assays considering properties of inventory chemicals

Apply knowledge/theory of ER Binding Domain to The Regulatory Chemical Domain (continuing to seek MECHANISTIC understanding) Hypothesize ER interactions of Inventory Chemicals Inert ingredients and antimicrobial pesticides are non-steroidal and do not contain multiple H-bonding groups at distance needed for steroid-like interactions Hypotheses: - Any pesticide inert or antimicrobial that does bind ER will do so through an interaction mechanism that results in low affinity binding - Only a small % of these chemicals are likely to bind ER - A chemical group approach will facilitate regulatory application - Chemicals can be grouped based on how they interact with the ER (within specific ER sub-pockets)

C A B T 347 E 353 H 524 HO R 394 CH3 “A_site only” Interaction QOxygen= -0.25 E 353 H 524 A B HO R 394 CH3 J. Katzenellenbogen

C A B T 347 E 353 H 524 NH2 R 394 H3C “B_site only” Interaction QOxygen= -0.32 E 353 H 524 A B NH2 R 394 H3C J. Katzenellenbogen

HO OH CH3 H QOxygen= -0.32 B A QOxygen= -0.25 A_site B_site

ER Binding Site A Homologous Series 4-n-Alkylphenols C0 Log Kow = 1.50 Log Kow = 1.97 Log Kow = 2.47 Log Kow = 3.65 Log Kow = 3.20 C1 C2 C3 C4 C5 Log Kow = 4.06 Log Kow = 5.68 Log Kow = 5.76 Log Kow = 4.62 Log Kow = 4.15 C7 C6 C8 C9 Note: pink shading indicates RBA > 0.00001%; grey shading indicates RBA < 0.00001%

4-t-Alkylphenols were also assayed Log Kow = 1.50 Log Kow = 1.97 Log Kow = 2.47 Log Kow = 3.65 Log Kow = 3.20 Log Kow = 4.06 C0 C2 C1 C3 C4 C5 Log Kow = 2.90 C3 Log Kow = 3.83 Log Kow = 3.31 Log Kow = 3.32 C4 C5 Log Kow = 5.76 Log Kow = 4.62 Log Kow = 4.15 C7 C6 Log Kow = 5.68 C8 C9 Log Kow = 4.36 C6 Log Kow = 4.89 C7 Log Kow = 5.16 C8 C10 C10 Log Kow = 6.61 Log Kow = 7.91 + C12

Relationship between Log Kow and RBA for 4-alkylphenols Site A Relationship between Log Kow and RBA for 4-alkylphenols

ER Binding Site B Homologous Series 4-n-Alkylanilines Log Kow = 3.05 Log Kow = 5.12 Log Kow = 3.39 Log Kow = 4.06 Log Kow = 0.90 Log Kow = 1.96 Log Kow = 2.40 Log Kow = 1.39 Trout ER Knowledge base C0 C1 C2 C8 C3 C4 C5 C6 Note: pink shading indicates RBA > 0.00001%; grey shading indicates RBA < 0.00001%

Relationship between Log Kow and RBA for 4-alkylanilines Site B Relationship between Log Kow and RBA for 4-alkylanilines

RBA ranges of low ER affinity (Site A or Site B only) chemicals relative to high ER affinity Site A-B (estrogens) or Site A-C (anti-estrogens) Site A-B or A-C Estradiol Ethinyl Estradiol Site A Site B Note: high affinity estrogens indicated by black dots; high affinity anti-estrogens indicated by orange dots.

Chemicals with RBA > 0.00001% Site A & Site B Chemicals with RBA > 0.00001% Site A Site B Symbols - Site A chemical groups - diamonds Symbols - Site B chemical groups - triangles

Mechanistic Basis of the Expert System to Predict Relative Estrogen Receptor Binding Affinity ER Binding Affinity: An Indicator of Potential Reproductive Effects ER-mediated reproductive impairment adverse outcome pathway ER Binding Domain Knowledge/theories of chemical-receptor interactions ER sub-pockets The Regulatory Chemical Domain Characterizing the FI and AM inventory chemicals Building from existing information The Receptor Binding Assay Domain Optimizing assays considering properties of inventory chemicals

Food use Inert Ingredients Inert chemicals in pesticides used on food crops The 2004 List included: 893 entries = 393 discrete chemicals + 500 non-discrete substances (44% discrete : 56% non-discrete) 393 discrete chemicals include: 366 organic chemicals (93%) 24 inorganic chemicals (6%) 3 organometallic compounds (1%) 500 non-discrete substances include: 147 polymers of mixed chain length 170 mixtures 183 undefined substances

Antimicrobial Pesticides Antimicrobials and sanitizers list included: 299 = 211 discrete chemicals + 88 non-discrete substances (71% discrete : 29% non-discrete) 211 discrete chemicals include: 153 organic chemicals (72%) 52 inorganic chemicals (25%) 6 acyclic organometallic compounds (3%) 88 non-discrete substances include: 25 polymers of mixed chain length 35 mixtures 28 undefined substances

Mechanistic Basis of the Expert System to Predict Relative Estrogen Receptor Binding Affinity ER Binding Affinity: An Indicator of Potential Reproductive Effects ER-mediated reproductive impairment adverse outcome pathway ER Binding Domain Knowledge/theories of chemical-receptor interactions ER sub-pockets The Regulatory Chemical Domain Characterizing the FI and AM inventory chemicals Building on existing information The Receptor Binding Assay Domain Optimizing assays considering properties of inventory chemicals

Adverse Outcome Pathway ER-mediated Reproductive Impairment QSAR focus area In vitro Assay focus area Inerts; Antimicrobial Chemicals CELLULAR Response POPULATION MOLECULAR Target TISSUE/ORGAN INDIVIDUAL Skewed Sex Ratios; Yr Class Liver Altered proteins(Vtg) Liver Cell Protein Expression Vitellogenin (egg protein transported to ovary) Sex reversal; Altered behavior; Repro. Receptor Binding ER Binding Toxicity Pathway Adverse Outcome Pathway Greater Toxicological Understanding Greater Risk Relevance

Estrogen Receptor Binding Displacement Assay Data Example - primary In vitro assay used : Estrogen Receptor Binding Displacement Assay Cyto rtER rtER Test Chemicals: Positive response Test Chemicals: Negative response Positive Control: Estradiol Positive Control: Estradiol RBA = relative binding affinity; (a ratio of measured chemical affinity for the ER relative to 17-beta-Estradiol = 100%) Log Kow = Log of octanol/water partition coefficient ); indicator of lipophilicity

Data example – Confirmatory in vitro Assay: Gene Activation Positive Control: Estradiol Positive Control: Estradiol Test Chemicals: Positive response Test Chemical: Negative response

Chemicals with RBA > 0.00001% Site A & Site B Chemicals with RBA > 0.00001% Site A Site B Symbols - Site A chemical groups - diamonds Symbols - Site B chemical groups - triangles

393 Food Use Inerts I II III IV V VI VII Start Here: Chemical List No IV Meets specific distance criteria for: Sites “A-B”; Sites “A-C”; Sites “A-B-C” Contains some attenuating feature, steric or other High Affinity “A-B”;“A-C” or “A-B-C” RBA > 0.1% RBA < 0.00001% Low Affinity “A-B”,“A-C” or “A-B-C” 0.00001% < RBA < 0.1% Strength of attenuation factors Yes No Strong Weak Contains a Cycle Contains a Charge Belongs to known Inactive Sub-class Contains a phenolic OH, and additional OH and/or =0 Yes Yes No No No Yes Yes “Special Rule” Applies V Yes Alkylaromatic Sulfonic Acids Log Kow (-0.6 - 5.7) Sulfonic Acid Dyes Log Kow (-0.4 - 6.0) 4-Alkylbenzthiols Log Kow ( 2.9 - 4.2) Miscellaneous Functional Groups w/ Charge Log Kow Range 4-n-Alkylfluorobenzenes ( 2.3 - 4.9) Alkylphenol (2,6-subst) ( 4.2) Benzamides ( 1.7 - 2.1) Borate Esters (-0.5 - 1.0) Benzoates (non-ring subst) ( 3.6 - 4.2) Bis-anilines ( 1.6 - 6.2) Hydrofurans (alcohol & ketone) ( 0.0 - 1.2) Imidazolidines (-0.9 - 0.6) Isothiazolines ( 0.6 - 2.5) Mono-cyclic Hydrocarbons ( 2.7 - 4.6) Oxazoles (<0.3) Phenones (n-alkyl) ( 1.6 - 4.1) Pyrrolidiones (-0.4 - 3.3) Sorbitans ( 3.2 - 5.9) Triazines (-0.4 - 3.4) RBA < 0.00001% No DDT-Like Tamoxifen-Like Multicyclic hydrocarbons Alkylchlorobenzenes Thiophosphate Esters Possible Low Affinity Site A-Type, Site B-Type Yes RBA > 0.00001% Log KOW <1.3 Exact match to Group Training Set Structure and measured RBA Yes RBA < 0.00001% No RBA < 0.00001% Unknown Binding Potential No Site “A” Contains Phenol Fragment VI Belongs to known Active Sub-class Log Kow Range Alkylphenols (1.5 - 8.0) Alkoxy phenols (2.3 - 4.3) Parabens (1.9 - 5.4) Salicylates (2.5 - 5.1) Yes Yes RBA > 0.00001% Yes RBA > 0.00001% No No Exact match to Mixed Phenols Training Set Structure and measured RBA RBA < 0.00001% Mixed Phenols No Unknown Binding Potential Log Kow Range 4-Alkylanilines ( 1.4 - 5.1) 4-Alkoxy Anilines ( 2.6 - 3.7) Phthalates ( 2.5 - 6.8) Phenones-(branched) ( 2.7 - 4.8) 4-Alkyl cyclohexanols ( 2.4 - 3.8) 4-Alkyl cyclohexanones ( 2.4 - 3.8) * 2-; 4-; or 2,4,6-Benzoates (substitution-dependent Log Kows) Site “B” Contains “Specified” Fragment Belongs to known Active Sub-class VII Yes Yes RBA > 0.00001% No No RBA > 0.00001% Yes Belong to untested class Exact match to Mixed Organics Training Set Structure and measured RBA RBA < 0.00001% Mixed Organics No Unknown Binding Potential Unknown Binding Potential

RBA ranges of low ER affinity (Site A or Site B only) chemicals relative to high ER affinity Site A-B (estrogens) or Site A-C (anti-estrogens) Site A-B or A-C Estradiol Ethinyl Estradiol Site A Site B Note: high affinity estrogens indicated by black dots; high affinity anti-estrogens indicated by orange dots.

393 Food Use Inerts I II III IV V VI VII Start Here: Chemical List No IV Meets specific distance criteria for: Sites “A-B”; Sites “A-C”; Sites “A-B-C” Contains some attenuating feature, steric or other High Affinity “A-B”;“A-C” or “A-B-C” RBA > 0.1% RBA < 0.00001% Low Affinity “A-B”,“A-C” or “A-B-C” 0.00001% < RBA < 0.1% Strength of attenuation factors Yes No Strong Weak Contains a Cycle Contains a Charge Belongs to known Inactive Sub-class Contains a phenolic OH, and additional OH and/or =0 Yes Yes No No No Yes Yes “Special Rule” Applies V Yes Alkylaromatic Sulfonic Acids Log Kow (-0.6 - 5.7) Sulfonic Acid Dyes Log Kow (-0.4 - 6.0) 4-Alkylbenzthiols Log Kow ( 2.9 - 4.2) Miscellaneous Functional Groups w/ Charge Log Kow Range 4-n-Alkylfluorobenzenes ( 2.3 - 4.9) Alkylphenol (2,6-subst) ( 4.2) Benzamides ( 1.7 - 2.1) Borate Esters (-0.5 - 1.0) Benzoates (non-ring subst) ( 3.6 - 4.2) Bis-anilines ( 1.6 - 6.2) Hydrofurans (alcohol & ketone) ( 0.0 - 1.2) Imidazolidines (-0.9 - 0.6) Isothiazolines ( 0.6 - 2.5) Mono-cyclic Hydrocarbons ( 2.7 - 4.6) Oxazoles (<0.3) Phenones (n-alkyl) ( 1.6 - 4.1) Pyrrolidiones (-0.4 - 3.3) Sorbitans ( 3.2 - 5.9) Triazines (-0.4 - 3.4) RBA < 0.00001% No DDT-Like Tamoxifen-Like Multicyclic hydrocarbons Alkylchlorobenzenes Thiophosphate Esters Possible Low Affinity Site A-Type, Site B-Type Yes RBA > 0.00001% Log KOW <1.3 Exact match to Group Training Set Structure and measured RBA Yes RBA < 0.00001% No RBA < 0.00001% Unknown Binding Potential No Site “A” Contains Phenol Fragment VI Belongs to known Active Sub-class Log Kow Range Alkylphenols (1.5 - 8.0) Alkoxy phenols (2.3 - 4.3) Parabens (1.9 - 5.4) Salicylates (2.5 - 5.1) Yes Yes RBA > 0.00001% Yes RBA > 0.00001% No No Exact match to Mixed Phenols Training Set Structure and measured RBA RBA < 0.00001% Mixed Phenols No Unknown Binding Potential Log Kow Range 4-Alkylanilines ( 1.4 - 5.1) 4-Alkoxy Anilines ( 2.6 - 3.7) Phthalates ( 2.5 - 6.8) Phenones-(branched) ( 2.7 - 4.8) 4-Alkyl cyclohexanols ( 2.4 - 3.8) 4-Alkyl cyclohexanones ( 2.4 - 3.8) * 2-; 4-; or 2,4,6-Benzoates (substitution-dependent Log Kows) Site “B” Contains “Specified” Fragment Belongs to known Active Sub-class VII Yes Yes RBA > 0.00001% No No RBA > 0.00001% Yes Belong to untested class Exact match to Mixed Organics Training Set Structure and measured RBA RBA < 0.00001% Mixed Organics No Unknown Binding Potential Unknown Binding Potential

Expert System Predictions for Food use Inerts and Antimicrobials Food use Inerts Antimicrobials Total Chemicals (%) Total Chemicals (%) 393 (100%) 211 (100%) Predicted RBA < 0.00001 378 (96%) 196 (93%) Predicted RBA > 0.00001 15 (4%) 15 (7%)

Expert System Profiler Series of nodes with possible yes/no logic applied. Structure categorized as alkylphenol w/ 1.5<LogKow <8.0 Chemical categorized as high EDC Potential – prioritization model.

Alkylphenol node – training set

Expansion of an Expert System Non-Food Use Inert Ingredients Non-Food Use Inerts list included: 2888 = 1423 discrete chemicals + 1465 non-discrete substances (49% discrete : 51% non-discrete) 1423 discrete chemicals include: 1192 organic chemicals (84%) 205 inorganic chemicals (14%) 26 organometallic compounds (2%) 1465 non-discrete substances include: 596 polymers of mixed chain length 174 mixtures 695 undefined substances

Expansion of an Expert System Food Use Inerts Groups to build the mechanism model Alkylphenols Alkoxyphenols Parabens Gallates Salicylates Phthalates 2,6-Substituted Alkylphenols Thiophosphate Esters Mixed Organics Mixed Phenols Acyclic Alkylaromatic Sulfonic Acids Cyclohexanols Mono-Cyclic Hydrocarbons Multi-Cyclic Hydrocarbons Cyclohexanones Hydrofurans Chlorobenzenes Phenones – n-chain Pyrrolidiones Sorbitans Sulfonic Acid Dyes Fluorobenzenes Alkylbenzthiols Alkylanilines Alkoxyanilines DDT-like Tamoxifen-like Benzamides Phenones-branched-chain Phenones – Cyclic Ring-subst Benzoates Non-ring subst Benzoates Acetanilides High Affinity Binders Non-food Use Inerts Antimicrobials Aminoanthracenedione Naphthalenol Azophenyl Benzotriazole Phenol 2-Hydroxy Benzophenone Dimethylamino Phenol Phenolsulfonphthaleins Fluorosceins Iso-phthalates Tribenzoates Phosphorous Acid Esters Phosphoric Acid Esters Sugars Oxazoles Isothiazolines Borate Esters Imidazolidines Triazines

Human ER Binding Affinity and Gene Activation

Relationship between Log Kow and RBA demonstrated for Site A chemicals binding human ER. Chemical ER binding assayed using recombinant human ERα (rec hERα). Recombinant rainbow trout ERα (rec rtERα) shown for species comparison with same chemicals in assays with chemical bioavailability (i.e., assay total protein) comparable to rec hERα. Cytosolic rainbow trout ER (cyto rtER) also shown. 4-alkylphenols RBA more comparable when assay chemical bioavailability is the same regardless of species (rec hERα vs. rec rtERα), than for same species if assay bioavailability is different (rec rtERα vs. cyto rtER).

Relationship between Log Kow and RBA demonstrated for Site B chemicals binding human ER. Chemical ER binding assayed using recombinant human ERα (rec hERα). Recombinant rainbow trout ERα (rec rtERα) shown for species comparison with same chemicals in assays with chemical bioavailability (i.e., assay total protein) comparable to rec hERα. Cytosolic rainbow trout ER (cyto rtER) also shown. 4-alkylanilines RBA more comparable when assay chemical bioavailability is the same regardless of species (rec hERα vs. rec rtERα), than for same species if assay bioavailability is different (rec rtERα vs. cyto rtER).

rat uterine cytosol ER and trout liver cytosol ER for 55 diverse chemicals. (similar chemical bioavailability in both cytosol receptor assays with similar total protein concentration) Rat data from the literature

Working with EPA NCCT to determine how mammalian model-based ER HTS assays correlate with the in vitro assay used to build ES focus was on selection of chemicals to cover chemical classes used in development of ER Expert System The NCCT ER assays are: Human ER, Bovine ER, Mouse ER-alpha competitive binding assays Human ER-alpha reporter gene assays: agonist & antagonist mode (2 vendors) ERE interaction assay Future: plan to extend to AR activation in the future with goal to build class-based expert system Goal: use hER data (low-, medium-, or high-throughput) to build ES with expanded chemical classes.

Conclusions ICPS MOA An Expert System based upon the ER-mediated AOP Knowledge of common initiating event across chemical classes facilitates development of QSARs and read-across methods to predict toxicity potential of untested chemicals OECD QSAR Validation Principles ICPS MOA

Relevance of ER model to predicting xenobiotic epigenetic chemicals?

endogenous estrogens stimulate substantial increases in Promotion by 17b-estradiol and b-hexachlorocyclohexane of hepatocellular tumors in medaka, Oryzias latipes. J.B. Cooke, D.E. Hinton. Aquatic Toxicology 45 (1999) 127–145 Hypothesis: Many laboratory and field studies with various fish species show a higher prevalence Of hepatocellular neoplasia in females than in males. During female sexual maturation, endogenous estrogens stimulate substantial increases in synthetic activity (e.g., vitellogenin, choriogenin productions) and hepatocytes proliferation. -tested hypothesis that estrogens promote growth of hepatic preneoplastic lesions and tumors. Medaka (Oryzias latipes) – -low dose of diethylnitrosamine (DEN; 200 mg /1 ,24 h) at 3 weeks of age then fed purified casein-based diet daily from 1 to 7 months of age with or w/o 17b-estradiol (E2; 0.01–10.0 mg g1 dry diet) With xenoestrogen, b-hexachlorocyclohexane (bHCH, 0.01–100.0 mg g1 dry diet). Livers examined for foci of cellular alteration (FCA) & and hepatocellular tumors. E2 increased prevalences of hepatocellular adenoma or carcinoma (26% in DEN plus 10ppm E2 group versus 4.6% in DEN only group, PB0.01). With incr E2, avg# basophilic FCA (BF) rose; #eosinophilic FCA (EF) sharply declined. DEN plus bHCH treatment groups > numbers of tumors in most, and greater numbers of BF in all

E2 increased prevalences of hepatocellular adenoma or carcinoma Promotion by 17b-estradiol and b-hexachlorocyclohexane of hepatocellular tumors in medaka, Oryzias latipes. J.B. Cooke, D.E. Hinton. Aquatic Toxicology 45 (1999) 127–145 Results: Livers examined for foci of cellular alteration (FCA) & and hepatocellular tumors. E2 increased prevalences of hepatocellular adenoma or carcinoma (26% in DEN plus 10ppm E2 group versus 4.6% in DEN only group, PB0.01). With incr E2, avg# basophilic FCA (BF) rose; #eosinophilic FCA (EF) sharply declined. DEN plus bHCH treatment groups > numbers of tumors in most, and greater numbers of BF in all DEN only: For all DEN-treated groups, BF were more common in female medaka, and EF more common in males. No tumors were found in fish fed E2 or bHCH without DEN exposure Liver wts -control medaka, significantly larger in females -E2 treatments (0.1, 1.0 or 10.0 ppm E2) elevated liver weights in males similar to that in females. -bHCH had no effect on liver weights.

-E2 is a tumor promoter in medaka. Promotion by 17b-estradiol and b-hexachlorocyclohexane of hepatocellular tumors in medaka, Oryzias latipes. J.B. Cooke, D.E. Hinton. Aquatic Toxicology 45 (1999) 127–145 Conclusions: -E2 is a tumor promoter in medaka. -Because tumor increases were not statistically significant, bHCH was considered a weakly positive modulator. -E2 particularly promoted tumor development in male medaka, indicating xenobiotics with mechanism of action like that of E2 may escalate growth in wild fish of previously initiated cells into tumors.

Epigenetic Adverse Outcome Pathway? ER-Mediated Hormonal Imbalance in fish DNA Binding DEN initation Chemical (parent or metabolite) DEN Chemical (parent or metabolite) ER agonist Individual Cell Organ CELLULAR Response POPULATION MOLECULAR Target MIE TISSUE/ORG INDIVIDUAL Liver Cell Replication Vitellogenin Hypertrophy Liver Tumors Cancer Receptor Binding ER Binding Cooke & Hinton, Aquatic Toxicology, 1999

ER model Relevance? Wildlife species: - reproductive impairment is more relevant than cancer from population perspective, so ER ES focused on repro impairment AOP Rodents: Rodent ER binding – good correlation with fishER - Estrogens and anabolic steroids can increase liver adenomas in rats/mice Mammals in general: Many of same chemicals bind ER and result in gene activation -possible role in promotion, progression? Liver is not considered one of the classic targets for hormonal carcinogenesis but liver cancer development is influenced by sex hormones (estrogen, testosterone).