Translational Toxicology in Zebrafish

Abstract

A major goal of translational toxicology is to identify adverse chemical effects and determine whether they are conserved or divergent across experimental systems. Translational toxicology encompasses assessment of chemical toxicity across multiple life stages, determination of toxic mode-of-action, computational prediction modeling, and identification of interventions that protect or restore health following toxic chemical exposures. The zebrafish is increasingly used in translational toxicology because it combines the genetic and physiological advantages of mammalian models with the higher-throughput capabilities and genetic manipulability of invertebrate models. Here, we review recent literature demonstrating the power of the zebrafish as a model for addressing all four activities of translational toxicology. Important data gaps and challenges associated with using zebrafish for translational toxicology are also discussed.

1.0. Introduction

A main goal of toxicology is to determine the potential for and the mechanisms by which xenobiotic agents cause harm to biological systems. While the human is a predominant target species of interest, for most xenobiotics there is limited human data. To address this data gap, it is often necessary to extrapolate data integrated from diverse species across varying levels of organization, including computational, biochemical, in vitro, and in vivo systems. In vivo models offer a distinct advantage by enabling assessment of integrative effects across organ systems, and across different life stages. However, all experimental models fail to recapitulate some aspects of human biology so it is important to understand and account for the limitations of any given model (Figure 1A).

Figure 1: Comparisons of model organisms to humans.

(A) Strengths and weaknesses of widely used translational toxicology animal models. (B) Comparative timelines for zebrafish and human development (https://zfin.org/zf_info/zfbook/stages/). The timing of all developmental events in zebrafish is influenced by temperature. Abbreviations: dpf = days post-fertilization; hpf = hours post-fertilization; wk = weeks. ^aKeane et al. 2011 Nature 477:289–294; ^bFritz et al. 2013 Microbiome 1:14; ^cKostic et al. 2013 Genes Dev 27:701–718; ^dNagpal et al. 2018 Front Microbiol 9:2897; ^eBedell et al. 1997 Genes Dev 11:11–43; ^fSeok et al. 2013 Proc Natl Acad Sci U S A 110:3507–3512; ^gHowe et al. 2013 Nature 496:498–503; ^hMacRae and Peterson 2015 Nat Rev Drug Discov 14:721–731; ⁱCayuela et al. 2018 Front Cell Dev Biol 6(178); ^jAluru et al. 2018 Environ Epigenet 4:dvy005; ^kGoldsmith and Jobin (2012) J Biomed Biotechnol 2012: 817341; ^lBargmann 2006 Worm Book 1–29; ^mClark and Walker 2018 Cell Mol Life Sci 75:93–101; ⁿKim et al. 2018 Genetics 210:445–461; ^oKimmel et al. 1995 Dev Dyn 203:253–310; ^pO’Rahilly et al. 1979 Anat Embryol (Berl) 157:167–176; ^qPhelps et al. 2017 Sci Rep 7:11244; ^rRawls et al. 2007 Proc Natl Acad Sci U S A 104:7622–7627.

Due to genetic and physiologic conservation between zebrafish and humans (Box 1) and the relevance of this small aquatic vertebrate to translational toxicology (Box 2), the zebrafish has become a widely used model for toxicological research [1,2] that is increasingly being used to address long-standing challenges in toxicology. For example, zebrafish are a powerful model for studying the toxicity of chemical mixtures, as exemplified by a recent study in which gene expression changes and lethality were quantified in embryonic zebrafish exposed to multiple concentrations of three different pesticides, either individually or as binary or tertiary mixtures [3]. The authors concluded that the quantitative and qualitative effects of the mixtures would not have been predicted based on changes elicited by exposure to individual chemicals. Another long-standing challenge – sex differences in toxic outcomes – was recently studied in zebrafish exposed to perfluorooctane sulfonate (PFOS). Transcriptomic analysis of multiple organs revealed that PFOS altered expression of genes associated with fatty acid metabolism and neural function in a manner that varied not only according to the target organ, and concentration and duration of PFOS exposure, but also sex [4]. In a separate study, wildtype female zebrafish were found to be significantly more sensitive to the behavioral effects of chronic ethanol exposure than long fin striped females or males [5], suggesting that sex and genetic background interact to determine toxic outcome. These studies suggest the potential for using zebrafish to identify specific gene × environment interactions that influence individual susceptibility for adverse outcomes [6].

Box 1. Relevance of zebrafish to human biology and disease.

The zebrafish expresses gene orthologs for >70% of human genes. 82% of human disease-causing proteins, anti 85% of known human drug targets [71]. Zebrafish proteins resemble their human counterparts, particularly within functional domains. For example, while the zebrafish glucocorticoid receptor is only 54% identical to the human glucocorticoid receptor, the ligand binding domain is 74% identical and its pharmacologic properties closely resemble those of the human [72], There is also considerable anatomic and physiologic conservation between zebrafish and humans. The zebrafish possesses counterparts of most human organ systems, and zebrafish organs largely perform the same functions as their human analogs, Physiologic mechanisms are well conserved at the molecular and cellular levels, and some cases (e.g., cardiac electrophysiology), the zebrafish is a better model of the human than rodents (reviewed in [72]). The zebrafish has many of the same sensory modalities as humans, including vision, olfaction, taste, touch, balance and hearing, and it exhibits an extensive behavioral repertoire, ranging from simple stimulus-response behaviors to complex behaviors such as sleep, pain, affective and depressive-like behavior, locomotion, social interactions and cognitive behaviors [16]

Zebrafish are widely used to model diverse human diseases. Because targeted gene mutations can be generated and phenotyped more efficiently in zebrafish than in rodents, (here is significant interest in using zebrafish to investigate rare genetic disorders. For example, using a scnllab mutant zebrafish that recapitulates critical clinical features of Dravet syndrome, a chemical library screen identified the 5-HT2B receptor as a novel therapeutic target for this rare genetic seizure disorder [73]. In a second example, zebrafish expressing human type I collagen gene mutations were engineered to investigate human genetic skeletal dysplasias because unlike mouse models, zebrafish bone mutants survive into adulthood [74], Using micro-computed tomography (μCT) for detailed and rapid skeletal phenotyping of zebrafish mutants and systematic collagen analysis by SDS/PAGE and mass-spectrometry, the authors demonstrated that zebrafish and human type i collagen are compositionaliy and functionally related, and that expression in the zebrafish of select human mutations in type I collagen gives rise to phenotypic variability that mirrors the clinical variability associated with the human disease [74].

Box 2. Zebrafish is a powerful model for toxicology research.

The zebrafish model is particularly well suited for molecular and developmental toxicology studies. Key advantages include: (i) the zebrafish genome has been completely sequenced and is highly homologous to the human genome [71]; (ii) powerful gene-editing techniques continue to be developed and optimized for use in zebrafish [75]; and (iii) zebrafish embryo and larvae are optically transparent, which enables visualization of the dynamic changes in individual cells and organs in vivo across a broad range of developmental stages [76].

Another significant advantage of zebrafish is that toxic outcomes can be measured at the molecular (e.g., mRNA or protein expression), structural (e.g., cells and organs) and systems level, including Structural [76] and electrophysiological [77] parameters of neural circuitry and behavior, This enables molecular effects to be anchored to phenotypic outcomes.

In contrast to cell-based assays that provide limited toxicokinetic information, zebrafish can reveal cnticat insights about the absorption, distribution, metabolism, and excretion (ADME) of xenobiotics. Zebrafish have a functional liver, kidneys, and blood-brain barrier, expression conserved tissue-specific transporters, and exhibit both phase I and II metabolism (reviewed in [72]).

Zebrafish are easily exposed by direct addition of chemicals to the house media (referred to as water-borne exposures). This is a significant advantage for screening studies, especially with the availability of robotics that automate delivery of compound into multi-well plates. However, nominal media concentrations are not necessarily representative of internal dosimetry [78,79], and this remains an important challenge in the use of zebrafish for toxicity testing. Compounds that do not readily dissolve in aqueous solutions represent another challenge for using waterborne exposures; however, this can be overcome by injecting chemicals directly into zebrafish [80].

The zebrafish is also proving to be a strong model for addressing emerging questions in toxicology, such as the influence of xenobiotics in the developmental origins of health and disease (DOHaD), epigenetic mechanisms of toxicity [7], and the role of the microbiome in modifying toxic effects of xenobiotics [8,9]. For example, the effects of bisphenol A (BPA) or the replacement chemicals BPAF, BPB, BPF, or BPS on developmental toxicity and microbiome community structure were recently studied in zebrafish [10]. Chemical potency was conserved in a zebrafish developmental toxicity assay when compared to previously reported estrogen receptor activity in both human in vitro and zebrafish reporter systems [10]. However, an inverse relationship between zebrafish developmental toxicity and chemical-dependent microbiome disruption was observed indicating that traditional toxicology tests fail to capture microbiome-dependent effects. Through the use of colonized, microbe-free axenic, and conventionalized zebrafish, recent work has shown that host-associated microbes biotransform xenobiotic agents into metabolites with unknown toxicity profiles [11,12].

2.0. Evaluating zebrafish for translational toxicology research

Translational toxicology broadly refers to the determination of toxicological effects as conserved or divergent across different experimental systems. Four activities are proposed to comprise translational toxicology [13]. First, assessing chemical toxicity across multiple life stages. Second, identifying chemical mode of action and relevance of key events across models. Third, using data from one model to predict chemical toxicity in other systems. Fourth, deploying models to develop and evaluate interventions to protect or restore a healthy status following chemical exposure. Here, we discuss recent evidence collected in zebrafish that encompasses these four activities (Figure 2).

Figure 2: Translational toxicology in zebrafish.

Zebrafish can be used to assess the four components of translational toxicological research, including assessment of chemical toxicity across life stages, delineation of chemical mode of action, development or testing of interventions that block chemical-dependent toxicity outcomes and restore health, or prediction of chemical toxicity in other systems, including humans.

2.1. Assessing chemical toxicity across the zebrafish lifespan

Zebrafish pass through four major life stages: embryonic, larval, juvenile, and adult. Zebrafish are considered embryos from fertilization until hatching, which can occur between 48–96 hours post-fertilization (hpf), at which point they are considered larvae. Zebrafish transition to the juvenile stage at ~30 days post-fertilization (dpf) (https://zfin.org/zf_info/zfbook/stages/), which corresponds to the age when many laboratory-bred strains have determined their sex. Sexual maturity and the ability to produce offspring signals the adult stage, which occurs by ~90–120 dpf. Compared to humans, the key molecular and cellular transitions that occur during the development and maturation of most major organ systems are similar in terms of sequence, but occur more rapidly in zebrafish (Figure 1B).

Zebrafish embryos and larvae are widely used for developmental toxicology studies for theoretical reasons – the molecular and cellular mechanisms of early development are among the most conserved between zebrafish and humans [14] – and practical considerations – zebrafish develop rapidly and external to the mother and for the first 7 dpf, obtain most of their nutrients from the yolk sac. Zebrafish are therefore readily adapted to higher throughput formats that deploy 96- or 384-well plates and automated tools for image acquisition, processing, and associated analyses. Because of its relatively short life cycle, the zebrafish offers significant advantages for assessing transgenerational (e.g., epigenetic) effects [7] and differential vulnerability to toxic effects across the lifespan. With regard to the latter, a recent evaluation of embryonic (3 hpf), larval (3 dpf), juvenile (30 dpf) and adult (3 month old) zebrafish exposed to varying concentrations of four different strobilurin fungicides revealed that the larval stage was the most susceptible [15]. Whether this reflects toxicokinetic or toxicodynamic mechanisms has yet to be determined.

In contrast, there are significantly fewer examples of juvenile and adult zebrafish being used for toxicology research. This may be because unlike embryonic and larval zebrafish, juvenile and adult zebrafish cannot be maintained in multi-well format plates for prolonged periods of time and they are not optically transparent. Despite these limitations, juvenile and larval zebrafish are advantageous for toxicological studies of phenotypes not exhibited at earlier life stages, such as sex, reproductive function, and adaptive immunity (Figure 1B), and behaviors that cannot be readily assessed in younger fish, including learning, memory, social, and anxiety-like behaviors [16]. For example, adult zebrafish were recently used to evaluate the therapeutic and toxic effects of the antidepressant amitriptyline [17]. Adult zebrafish are also gaining traction as models for studying chemical effects on phenotypes and diseases associated with aging, including various cancers [18]. An adult zebrafish model was used to screen novel small molecule therapeutics for liver cancer to identify compounds with a better therapeutic index than the standard of care, sorafenib [19]. Adult zebrafish were also recently validated as a model for evaluating drug-induced kidney injury [20]. Larval zebrafish have only one pair of nephrons whereas the adult zebrafish kidney has several hundred nephrons with similar histological structure and physiological function as the mammalian kidney (reviewed in [20]). The renal pathology observed in adult zebrafish exposed to nephrotoxic levels of gentamicin or doxorubicin was similar to that seen in mammals, and a screen of 28 chemicals with known nephrotoxicity and 14 with no known nephrotoxicity in humans demonstrated that 16 of the nephrotoxic chemicals and none of the negative controls caused drug-induced kidney injury in adult zebrafish.

2.2. Using zebrafish to define chemical mode-of-action

A significant strength of the zebrafish is that molecular insights into chemical mode-of-action can be obtained using diverse approaches ranging from chemical screens to elucidate structure-activity relationships (SARs) to genetic manipulation that identifies molecular targets of xenobiotics. Chemical screens in zebrafish have revealed novel mechanistic information about compounds with unknown modes-of-action via phenotypic mapping to compounds with known modes-of-action. For example, in a screen of 14,000 compounds, automated behavior testing of zebrafish coupled with a barcoding-based computational approach was used to identify novel neuroactive compounds that shared behavioral profiles with compounds with known modes-of-action [21]. This strategy has since been applied to identify chemicals that regulate zebrafish sleep/wake cycles [22], passive and active threat response [23], addiction [24], and psychosis [25]. Phenotypic SARs have been identified for the developmental toxicity of oxygenated, hydroxylated, or heterocyclic polycyclic aromatic hydrocarbon (PAH) derivatives [26]. In the same study, a relationship between behavior phenotypes and specific PAH substitutions was not apparent [26]. More recently, a smaller-scale comparison of alkyl sulphonic acid, alkyl carboxylic acid, or branched or ether containing per- and polyfluoroalkyl substances (PFAS) showed that exposure to alkyl sulfonic acid PFAS with more than four fluorinated carbons caused hyperactivity and the potency for this structural subclass of PFAS correlated with fluorinated carbon chain length [27]. Quantitative SARs (QSARs) have also recently been used to predict zebrafish acute toxicity for neutral [28] or ionizable [29] organic chemicals.

Medium-to-high-throughput zebrafish screens coupled with automated morphological and behavioral phenotyping represent a powerful strategy to identify SARs that illuminate phenotypic readouts particularly sensitive to chemical disruption. Subsequent unbiased pathway-level assessment and gene editing can then be used to solve mode-of-action in vivo. Unbiased [30–37] or targeted RNA sequencing [38] are routinely used to identify chemical-dependent perturbations in zebrafish at the level of genes and pathways. As an example of the translational potential of this approach, changes in gene expression following exposure to three hepatotoxic compounds were compared across whole zebrafish, mouse and rat livers, in vitro mouse and rat hepatocytes, and primary human hepatocytes [39]. While specific changes in gene expression were not generally conserved across models, shared pathway-level perturbations were identified demonstrating that the zebrafish has the capacity to identify pathway-level transcriptomic-based disruptions that indicate liver toxicity in a suite of mammalian models [39]. The observed lack of concordance on the gene level likely stems from comparing profiles obtained from whole zebrafish homogenates versus liver-specific human cells or tissues. Future studies should consider isolating specific cell types via cell sorting or microdissecting specific tissues from zebrafish to enable cross-species transcriptomic comparisons based on similar cell or tissue types.

Once key phenotypes and pathway-level perturbations have been identified, zebrafish can easily be used for mechanistic research, with the goal of identifying causative events that link chemical exposure to phenotypic outcomes. Alternatively, molecular toxicology approaches can be used to disprove dogma related to assumed or predicted modes-of-action. One recent example of the latter was the demonstration of the non-essentiality of PPARγ for ciglitazone-dependent dorsoventral patterning defects in early zebrafish development [40]. Injection of an anti-sense oligonucleotide morpholino into single cell stage zebrafish to transiently suppress the generation of PPARy protein revealed that defects in patterning elicited by ciglitazone exposure occurred via a PPARy-independent mechanism [40]. While some researchers have argued that morpholino knockdown phenotypes can be more severe than mutant phenotypes because of off-target effects [41], lack of concordance between morpholino knockdowns and stable gene knockouts may be more complex and involve genetic compensatory mechanisms specific to gene knockouts, but not knockdowns, at least at certain loci [42]. Nevertheless, gene editing approaches (e.g., CRISPR/Cas9) are now widely used to discover mutations that cause phenotypes and define toxicological modes-of-action. This endeavor is aided by a wide array of mutant zebrafish available via the Zebrafish Mutation Project [43] and the generation of cell type-specific mutant zebrafish lines [44]. While concerns regarding off-target effects of CRISPR/Cas9-based gene editing have been raised [45], recent whole exome sequencing evidence obtained across two generations of zebrafish derived from the same founding mutant pair failed to show evidence of off-target, de novo mutations [46], supporting the use of CRISPR/Cas9-based gene editing to uncover mechanisms by which toxicants elicit adverse outcomes.

There are several recent examples of using gene editing to solve toxicological mode-of-action in zebrafish. In an elegant study that integrated human hepatocellular carcinoma samples, human hepatocyte culture, and zebrafish, CRISPR/Cas9-dependent knockout of G protein-coupled receptor-1 (gper-1) was sufficient to block liver growth in 17-beta-estradiol exposed zebrafish [47]. This work identified gper-1 as a fundamental hepatic estrogen sensor. Given data from human studies demonstrating a causal link between gper signaling and atherosclerosis, heart failure, reproduction, metabolic disorders, cancer, and menopause (reviewed in [48]), this research broadly illustrates that zebrafish can be used to elucidate human-relevant toxicity mechanisms [47]. CRISPR/Cas9-dependent gene knockout was also effectively deployed to show that the efflux transporter multi-resistance-associated protein-1 (mrp1) functions to efflux both cadmium and benzo[a]pyrene [49]. Increased compound accumulation, mortality and, at lower concentrations, increased incidence of pericardial edema and failure to hatch, was observed in mrp1 mutant zebrafish exposed to either compound [49]. Perturbation of arylhydrocarbon receptor (ahr)-dependent signaling represents a well-studied molecular mechanism by which xenobiotic exposure triggers adverse outcomes. Mutant zebrafish lines have revealed the essentiality of ahr2 [50] and sox9b [51] in mediating TCDD-dependent effects on zebrafish heart development. Morpholino-mediated knockdown of the long non-coding RNA slincR was used to demonstrate that slincR repressed sox9b expression as part of the mechanism by which TCDD induced vascular hemorrhage in zebrafish [52]. Mutant ahr2 zebrafish have also been leveraged to reveal the essentiality of the receptor for mono-substituted isopropylated triaryl phosphate, a component of Firemaster 550, to cause a heart looping defect [53].

2.3. Retrofitting zebrafish toxicity data to build or evaluate predictive toxicity models

Historically, zebrafish toxicity data has been used for hazard identification and chemical prioritization [54–59]. To fully leverage available zebrafish toxicity data, its ability to predict toxicity in humans must be defined. An early key paper calculated overall concordances between developmental toxicants in zebrafish and rat (52%) or rabbit (47%) guideline studies [60]. Interestingly, the percentage of concordant chemicals identified between rat and rabbit studies was similar (58%), indicating that at least for the evaluated set of chemicals, zebrafish toxicity data was generally as predictive as the calculated concordance between two widely used mammalian models [60]. A subsequent meta-analysis compared zebrafish toxicity data on 443 chemicals, 19 aggregated toxicity phenotypes (e.g. cardiovascular), and 57 individual toxicity phenotypes (e.g. pericardial edema) to guideline toxicity data collected in rat, mouse, and rabbit [61]. Zebrafish LC₅₀ values were highly correlated with acute mammalian inhalation toxicity with zebrafish LC₅₀ values roughly 180% more sensitive than their mammalian counterparts [61]. From a developmental perspective, zebrafish hatching rate, pericardial edema, and decreased heart rate positively correlated with rabbit lowest effect levels (LELs) for prenatal loss [61]. In an interesting twist, the authors incorporated human exposure values to rank chemicals based on the integration of zebrafish toxicity data and human exposure estimates [61]. The resulting hazard index identified 14 chemicals where exposure levels in humans occur at concentrations that cause toxicity in zebrafish and therefore deserve further scrutiny [61].

A refinement of toxicity concordances that considers SARs is critical to understanding chemical blind-spots in the zebrafish test system. For example, several compounds identified as reference compounds for developmental neurotoxicity because of documented human developmental toxicity were negative hits in a screen of 91 compounds for teratological and behavioral effects in larval zebrafish [54]. These included lead acetate trihydrate, valproic acid sodium salt, and toluene. Whether these compounds are true negatives or the lack of toxic effect is due to toxicokinetic (e.g. reduced bioavailability due to minimal uptake of the compound, lack of metabolic activation, or photoinactivation of the compound) or toxicodynamic (e.g. deficient target expression) differences in developing zebrafish versus mammalian models remains to be determined.

These observations are relevant to a second challenge in establishing concordance between zebrafish and mammalian toxicity data, which is that most of the zebrafish data used in these analyses compare nominal media concentrations to phenotypic outcomes. Tissue dose in zebrafish as a result of waterborne exposure is affected by diverse physicochemical properties [28,29,56]. If a compound fails to provoke phenotypic effects in zebrafish, paired analytical chemistry data is necessary to demonstrate chemical uptake and confirm the assumption that a chemical is negative for the measured toxicity outcome. For example, a recent study showed that GenX, an emerging PFAS compound of public health concern, was unstable in dimethylsulfoxide (DMSO), a solvent widely used in zebrafish chemical screening studies. Without tissue dose measurements, this compound would have been assumed to be negative for a number of developmental toxicity and developmental neurotoxicity endpoints [27].

The zebrafish also contributes to translational toxicology as a tool for evaluating computational models developed using in vitro and biochemical data generated with human cells or receptors. A computational model predicting xenobiotic disruption of blood vessel development [62] was subsequently validated using a transgenic zebrafish assay for evaluating chemical-dependent effects on vessel development [63]. Comparison of human amino acid sequence similarities for members of the predictive signature in the computational model to the zebrafish analog demonstrated biological domain-specific differences in protein sequence conservation [64], and the zebrafish assay proved zebrafish are particularly adept at detecting vascular disruptors associated with chemokine and/or extracellular matrix disruption in human in vitro assays [64].

2.4. Testing interventions in zebrafish

Zebrafish are increasingly used in phenotypic screens to identify compounds that reverse or suppress adverse effects of genetic mutations. For example, a zebrafish expressing a germline mutation in the vhl gene was used to identify pharmacologic approaches for reversing the loss of vision associated with von-Hippel Lindau (VHL) syndrome [65], a rare disease characterized by vision loss associated with retinal capillary hemangioblastomas (tumors of retinal blood vessels). Zebrafish nullizygous for vhl, which were developed because Vhl knockout is embryolethal in mice, exhibit ectopic ocular blood vessels and aberrant eye development associated with an absent optokinetic response and significantly reduced visual motor response [65]. Sunitinib malate, an anti-angiogenic compound approved for cancer treatment, was found to reverse the ocular behavioral and morphological phenotypes in the vhl knockout zebrafish [65]. The methods used in this study were not high throughput; however, an automated system for histological analyses in zebrafish was recently described in which a commercially available platform that automates the transfer of zebrafish larvae from multi-well plates was combined with a customized spinning disk confocal microscope interfaced to software for high resolution image acquisition and analysis [66]. Using this system to screen 175 chemicals in Tg(mbp:eGFP) larvae, a transgenic zebrafish line that expressed enhanced green fluorescent protein (eGFP) in myelinating oligodendrocytes, three novel compounds that significantly altered myelination were identified [66].

Zebrafish are also being leveraged to screen for compounds that mitigate the adverse effects of xenobiotics. A screen of 2,271 small molecules identified 120 compounds that prevented cardiotoxicity in doxorubicin-exposed zebrafish, and subsequent SAR and target enrichment analyses of the seven most effective compounds identified CYP1A1 as a putative target [67]. This was corroborated by showing that cyp1a knockout zebrafish larvae were resistant to doxorubicin-induced cardiotoxicity [67]. In a separate study, the age-dependent sensitivity of zebrafish to cyanide was leveraged to identify novel therapeutic targets for cyanide poisoning [68]. Initial studies revealed that zebrafish embryos are highly resistant to cyanide during the first 3 dpf but become progressively more sensitive as the larvae mature. Unbiased transcriptomic and metabolomic analyses revealed age-dependent differences in energy metabolism during cyanide exposure [68]. This observation led to the identification of compounds that modulate the pyruvate dehydrogenase complex and the small molecule sodium glyoxylate as potential prophylactic treatments for modulating sensitivity to cyanide poisoning [68].

3.0. Major data gaps and summary

Zebrafish are widely used for hazard identification and chemical prioritization [27,32,33,54,56,58,59,61,64]. To improve the use of zebrafish toxicity data in human risk assessment, several data gaps need to be addressed. First, harmonization of common toxicity assays and assessments are necessary to overcome variability in zebrafish data that are due to differences in testing protocols [69]. Additionally, recent advances in developmental toxicity SARs [26,56] and chemical uptake [28,29] need to be expanded, and large-scale SAR analysis of more sensitive behavior endpoints, such as hyperactivity [27], are needed.

A major hindrance to the identification of relevant target pathways in zebrafish is the widespread use of whole animal transcriptomic data, due in part to the technical difficulties associated with obtaining organ- or cell type-specific expression data. Future work should capitalize on the ability to sort specific populations of cells from transgenic zebrafish to increase the ability to detect xenobiotic-dependent transcriptional effects on sensitive, but low abundance cell types. In addition, while there are a growing number of examples in zebrafish [47,49,51,70], more studies should consider using gene editing to characterize toxicity mechanisms.

Perhaps the area ripest for gains is the development of computational models to predict human toxicity from zebrafish toxicity data. Here, toxicokinetic data must be more routinely gathered in phenotypic zebrafish studies, both to identify true negative compounds [27] and to serve as the basis for dose extrapolation to human-relevant exposure scenarios.

In summary, the zebrafish is an exceptional model for the illumination of chemical-dependent toxic effects that are conserved or divergent across different experimental systems. Because of the inherent power of the system for medium-to-high throughput chemical-genetic screens, the zebrafish represents a powerful experimental system for assessing chemical toxicity across lifespan, identification of chemical mode-of-action, generation of datasets for the prediction of chemical toxicity in humans, and rapid assessment of interventions to prevent chemical toxicity in exposed organisms. Researchers who focus on translational research using zebrafish may ultimately have a deep impact on the protection of both human health and the environment.

Abbreviations:

AhR

arylhydrocarbon receptor

BPA

bisphenol A

BPAF

bisphenol AF

BPB

bisphenol B

BPF

bisphenol F

BPS

bisphenol S

CYP1A1

cytochrome P450, family 1, subfamily A, polypeptide1

DOHaD

developmental origins of health and disease

dpf

days post-fertilization

DMSO

dimethylsulfoxide

GenX

ammonium salt of hexafluoropropylene oxide dimer acid fluoride

eGFP

enhanced green fluorescent protein

gper-1

G protein-coupled receptor 1

hpf

hours post fertilization

LELs

lowest effect levels

mrp1

multi-resistance-associated protein-1

PFAS

per- and polyfluoroalkyl substances

PFOS

perfluorooctane sulfonate

SARs

structure activity relationships

slincR

sox9b long intergenic noncoding RNA

sox9b

SRY-box transcription factor 9b

QSARs

Quantitative SARs

TCDD

2,3,7,8-Tetrachlorodibenzo-p-dioxin

VHL

von-Hippel Lindau syndrome