Grantee Research Project Results
2017 Progress Report: System Toxicological Approaches to Define Flame Retardant Adverse Outcome Pathways
EPA Grant Number: R835796Title: System Toxicological Approaches to Define Flame Retardant Adverse Outcome Pathways
Investigators: Tanguay, Robyn L. , Reif, David , Simonich, Mike , Sullivan, Chris , Du, Jane La
Institution: Oregon State University , North Carolina State University at Raleigh
EPA Project Officer: Aja, Hayley
Project Period: June 1, 2015 through May 31, 2018 (Extended to May 31, 2019)
Project Period Covered by this Report: June 1, 2016 through August 31,2017
Project Amount: $798,661
RFA: Systems-Based Research for Evaluating Ecological Impacts of Manufactured Chemicals (2014) RFA Text | Recipients Lists
Research Category: Chemical Safety for Sustainability
Objective:
The objectives of this project are to:
- Expose embryonic zebrafish to a comprehensive list of flame retardant chemicals (FRCs) and observe their morphology and behavior for signs of toxicity.
- Grow exposed zebrafish to adulthood and measure their physiology and behavior for signs of persistent toxicity.
- Bin FRC outcomes using multiple levels of biological organization (i.e., chemical structure, similarity of gene expression profiles, early and adult life stage adverse outcomes), and thereby define adverse outcome pathways (AOP) for mechanistic FRC hazard prediction.
- Share our data with EPA and the broader research community.
The following were the focus of this year’s efforts:
- Expand the flame retardant library to a total of 61 FRCs
- Compare the zebrafish and ToxCast in vitro assay results for 45 FRCs
- Build a model to predict classification of FRCs
- Complete developmental exposures and grow outs to assess adverse behavioral effects for selected 10 FRCs.
Progress Summary:
We have completed each of these objectives with a summary to each below
(1) Expand the flame retardant library to a total of 61 flame retardant chemicals (FRCs)
Previously, we obtained 42 FRCs from the US. EPA National Center of Computational Toxicology ToxCast chemical library and have since procured an additional 19, resulting in a library of 61 FRCs. Using our high-throughput zebrafish toxicity model, we evaluated the developmental toxicity of the 61 FRCs by exposing embryos from 6-120 hours post fertilization (hpf) to 11 concentrations (0.1 – 80 µM). At 24 and 120 hpf, 22 morphological endpoints were evaluated. Additionally, exposed embryos underwent 2 behavioral assays. We computed the lowest effect level for each morphological endpoint in addition to two 24h and 5d behavioral endpoints.
Figure 1 . Dot-heatmap representation summary of the developmental toxicity for 61 FRCs
(2) Compare the zebrafish and ToxCast in vitro assay results for 45 FRCs
Forty-five FRCs tested in the developmental zebrafish assay were also assessed by the EPA-NCCT in over 1,000 assay endpoints. Rather than evaluate each assay endpoint independently, the high throughput in vitro data was aggregated at the level of technology and vendor and all assays with no variance (indicating no different from its respective assay components) were removed from the analysis. ToxPi was employed to rank the FRCs based on its bioactivity across all the assay vendors. Each slice was a vendor (Figure 2b) and the weight for each vendor was the same. For each vendor, the concentration that induced 50% activity (AC50) for each respective assay endpoint was scaled, and then input into the computation of a slice score. The length of the slice indicates the potency (a higher slice score). Over a 1/3 (22 FRCs) were bioactive in the ToxCast in vitro assay. The top four bioactive FRCs were TDCPP, 2,2',6,6'-Tetrachlorobisphenol A, Triphenyl phosphate and chloropyifos. The exhibited a high slice score in three vendor assays (Attagene, ATG; BSK, BioSeek; and Tox21, assays in the Tox21 consortium toolbox).
Figure 2. Clustering of 45 FRCs assessed in ToxCast assay.
(3) Build a model to predict classification of FRCs
Using 12 chemical properties of the FRCs and the three zebrafish assays, we built a model using randomForest algorithm to classify the FRCs into five classes: aryl phosphate ester (APE), brominated phenol (BP), chlorinated phosphate ester (CPE), polybrominated diphenyl ether (PBDE), other brominated and other. The classification model performance had a strong overall balance accuracy of 83.33% with a p value of 0.02. The model underwent 10-fold cross-validation and 3X bootstrapping. For each of the six classes, the model predicted with varying balance accuracy and sensitivity. The model was best at predicting FRCs in APE, Other, OtherBrominated, and PBDEs with balance accuracies of 91.7, 100, 100, and 95%, respectively.
Figure 3. Classification modeling performance. RandomForest was implement to build a classification mode for 6 FRC classes: Other, OtherBrominated, Polybrominated Diphenyl Ether (PBDE), aryl phosphate ester (APE), brominated phenol (BP), and chlorinated phosphate ester (CPE). (A) An illustration of which of the 15 parameters is considered the most important for each of the FRC classes. (B) The balance accuracy for each of the FRC classes, and their sensitivity, specificity and positive prediction value.
(4) Complete developmental exposures and grow outs to assess adverse behavioral effects for 10 FRCs
Based on the composite developmental toxicity data, we selected 10 FRCs for grow out studies. For these studies, we performed the exact exposure protocols used in the screening detailed above, but at the calculated no observable adverse effect level (NOAEL). Embryos are exposed from 6-120 hpf, then raised into adulthood in chemical-free water. The selected FRCS were: TBBPA (Tetrabromobisphenol A), TBBPA-DBPE (Tetrabromobisphenol A bis(2,3-dibromopropyl ether), TDCPP (Tris(1,3-dichloro-2-propyl)phosphate, TBEP (Tris(2-butoxyethyl) phosphate, TCPP (Tris(1,3-dichloro-2-propyl)phosphate, IPP-1 (Tris(1,3-dichloro-2-propyl)phosphate, TPP (Triphenyl phosphate), BPBP (tert-Butylphenyl diphenyl phosphate), TBPH (Bis(2-ethylhexyl) tetrabromophthalate), TCEP (Tris(2-chloroethyl) phosphate). When the fish reach adulthood (4 months of age), we evaluated them in custom-built multi-level behavioral units to evaluate: (1) schooling behavior, (2) predator avoidance response, (3) startle response, and (4) conditioned place preference learning in a shuttlebox. Based on this data set, FRCs that impacted larval photomotor response, also exhibited deficiencies in adulthood.
Table 1. FRC behavioral profiles
FRC | Conc(µM) | Predator Avoidance | Social Behavior | Startle | Larval Response |
---|---|---|---|---|---|
BPDP | 2 | Enhanced Avoidance | More Gregarious | Habituate; NS | Hypo; bioactive |
IPP-1 | 0.5 | Enhanced Avoidance | Less Gregarious | Habituate; NS | Hypo; bioactive |
TBBPA | 1 | Enhanced Avoidance | More Gregarious | Habituate; NS | Not bioactive |
EBBPA-DBPE | 1 | Enhanced Avoidance | More Gregarious | Habituate; hypo | Not bioactive |
TBEP | 0.5 | Enhanced Avoidance | Less Gregarious | Habituate; hyper | Hypo; bioactive |
TBPH | 1 | NS | More Gregarious | Habituate; NS | Hyper; bioactive |
TCEP | 0.5 | NS | More Gregarious | Habituate; NS | Hyperactive |
TCCP | 0.5 | NS | NS | Habituate; NS | Hyperactive |
TDCPP | .05 | NS | Less Gregarious | Habituate; NS | Hypo; bioactive |
TPP | 2 | NS | More Gregarious | Habituate; hyper | Hypo; bioactive |
NS=not statistically significant; All other entries are tested at p<0.05; Predator avoidance and social anxiety statistics consist of 2 way ANOVA; Startle response significance was determined by One way ANOVA
Future Activities:
- Complete the validation studies to calculate EC80 for the selected FRCs
- Author two manuscripts: (1) adult behavior responses based on the developmental response bin for the FRCs and (2) genome-wide transcriptomics manuscript
- Conduct genome-wide transcriptomics on the FRCs.
Journal Articles on this Report : 4 Displayed | Download in RIS Format
Other project views: | All 32 publications | 7 publications in selected types | All 7 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Truong L, Bugel SM, Chlebowski A, Usenko CY, Simonich MT, Simonich SLM, Tanguay RL. Optimizing multi-dimensional high throughput screening using zebrafish. Reproductive Toxicology 2016;65:139-147. |
R835796 (2017) R835168 (Final) |
Exit Exit Exit |
|
Zhang G, Truong L, Tanguay RL, Reif DM. A new statistical approach to characterize chemical-elicited behavioral effects in high-throughput studies using zebrafish. PLoS One 2017;12(1):e0169408 (16 pp.). |
R835796 (2017) R835168 (Final) |
Exit Exit Exit |
|
Zhang G, Marvel S, Truong L, Tanguay RL, Reif DM. Aggregate entropy scoring for quantifying activity across endpoints with irregular correlation structure. Reproductive Toxicology 2016;62:92-99. |
R835796 (2017) R835168 (Final) R835802 (2015) R835802 (2016) R835802 (2017) R835802 (2018) R835802C003 (2015) |
Exit Exit Exit |
|
Zhang G, Roell KR, Truong L, Tanguay RL, Reif DM. A data-driven weighting scheme for multivariate phenotypic endpoints recapitulates zebrafish developmental cascades. Toxicology and Applied Pharmacology 2017;314:109-117. |
R835796 (2017) R835802 (2016) R835802 (2017) R835802 (2018) |
Exit Exit Exit |
Supplemental Keywords:
Dose-response, teratogen, animal, stressor, toxics, chemical mixtures, adverse outcome pathwaysProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.