Seongwon Park 1, Chang Seon Ryu 2, Sangwoo Lee 3,4, Woo-Keun Kim 3,4, Soyoung Lee 1, Hwi-yeol Yun 1,5,6, Jung-woo Chae 1,5,6
1 College of Pharmacy, Chungnam National University (Yuseong-gu, Republic of Korea), 2 Advanced Biotechnology Cluster, KIST Europe Forschungsgesellschaft mbH, (Saarbrücken, Germany), 3 Center for Predictive Model Research, Korea Institute of Toxicology (Yuseong-gu, Republic of Korea), 4 Human and Environmental Toxicology, University of Science and Technology (Yuseong-gu, Republic of Korea), 5 Senior Health Convergence Research Center, Chungnam National University (Yuseong-gu, Republic of Korea), 6 Department of Bio-AI convergence, Chungnam National University (Yuseong-gu, Republic of Korea)
Objectives:
Genetic homology between zebrafish and human cytochrome P450 (CYP) enzymes has been extensively reported. However, genetic similarity does not necessarily guarantee functional equivalence. This discrepancy is exemplified in the human CYP2C subfamily, where CYP2C-related biotransformation has been observed in vivo, no clear zebrafish counterpart corresponding to human CYP2C has been established. We aimed to develop a mechanistic zebrafish embryo toxicokinetic (TK) modeling framework that quantifies pathway-specific “CYP-like” metabolism from immersion exposure data and enables functional homology mapping between zebrafish embryos/early larvae (≤120 hpf) and humans. The framework was designed to distinguish true metabolic formation from confounding processes, including embryo-medium exchange and non-specific losses, thereby supporting mechanistically interpretable and translationally relevant outputs in alignment with the New Approach Methodology (NAM) paradigm.
Methods:
The zebrafish embryo TK platform was first established in pilot studies using naphthalene and DEHP/MEHP, in which the structural model explicitly separated the culture medium, embryo body, and yolk compartments. To account for physicochemical-dependent processes, an optional container compartment was introduced for highly hydrophobic compounds with non-negligible adsorption to container surfaces, and an optional air compartment was defined for volatile compounds.
Building on this pilot framework, published quantitative data in zebrafish embryos/early larvae (≤120 hpf) reporting both parent compounds and major metabolites after immersion exposure were identified for CYP-focused functional mapping. The compounds included (i) caffeine → 1,7-dimethylxanthine (human CYP1A2-associated pathway), (ii) diclofenac → 4-hydroxydiclofenac and 5-hydroxydiclofenac (human CYP2C9- and CYP3A4/5-associated pathways), and (iii) carbamazepine → carbamazepine-10,11-epoxide (human CYP3A4/CYP2C8-associated pathway). Parent and metabolite time courses were described using mechanistic uptake and distribution between medium and embryo compartments, growth dilution, and metabolite-specific formation and elimination. Parameter uncertainty was then assessed by bootstrap analysis, and model performance was evaluated by comparison with the reported concentration-time data. Models were implemented in R (v4.4.3) with RStudio (v2026.01.0+392) using the ‘nlmixr2’, ‘dplyr’, and ‘ggplot2’ packages.
Results:
The pilot models for naphthalene and DEHP/MEHP showed that explicit separation of medium, embryo body, and yolk compartments was necessary to capture compound-specific kinetics within a single framework. For highly hydrophobic compounds, inclusion of the container compartment improved mechanistic interpretability by distinguishing non-specific losses from apparent clearance. These pilot analyses established a TK platform for embryo-stage immersion studies. The estimated embryo-specific naphthalene elimination rate constant was 2.4 × 10^2 h^-1, whereas the estimated metabolic rate constant for DEHP-to-MEHP metabolism was 0.67 h^-1 per embryo. The naphthalene parameter should be interpreted as parent depletion rather than metabolite-confirmed metabolic formation. Even after considering the volatile nature of naphthalene, its decline in the exposure medium was primarily explained by embryo-related depletion processes, which likely contributed to the relatively large estimated rate constant.
In the CYP-focused mapping models, inclusion of metabolite measurements substantially improved separation between embryo-medium exchange and true metabolic formation. For caffeine, measurable formation of 1,7-dimethylxanthine was captured with an estimated metabolic rate constant of 7.1 × 10^-5 h^-1, supporting a quantifiable CYP1A2-like pathway. For diclofenac, simultaneous fitting of parent and metabolite profiles separated the 4- and 5-hydroxylation pathways, with estimated metabolic rate constants of 0.2 and 0.03 h^-1, respectively, supporting concurrent CYP2C9-like and CYP3A-like activity in embryos/early larvae despite the absence of a clearly established zebrafish ortholog for the human CYP2C family. For carbamazepine, detectable formation of carbamazepine-10,11-epoxide was described with an estimated metabolic rate constant of 6.0 × 10^-4 h^-1, providing an additional CYP3A/2C-like marker within the same developmental window.
Rate constants estimated from parent depletion alone differed substantially from those obtained from models jointly describing parent and metabolite data. While canonical human CYP3A4/5 probe substrates such as midazolam and testosterone have been reported not to generate detectable major human metabolites in zebrafish embryos/larvae, 5-hydroxydiclofenac and carbamazepine-10,11-epoxide were detectable within the same developmental window. These findings suggest that diclofenac 5-hydroxylation and carbamazepine epoxidation may serve as candidate CYP3-like probe reactions for embryo-stage functional mapping.
Conclusion:
Mechanistic TK modeling provides added value beyond in vivo observation by separating transport/partitioning and system-related losses from true metabolic capacity, thereby yielding comparable quantitative indices of pathway activity with associated uncertainty. This embryo-stage TK framework supports a more defensible functional homology map between zebrafish and human CYP pathways and provides a NAM-aligned basis for selecting robust embryo-stage probe substrates for translation and PBPK-oriented applications.
References:
[1] Ioana Chelcea et al., Physiology-informed toxicokinetic model for the zebrafish embryo test developed for bisphenols, Chemosphere, 2023.
[2] Pierre-Andre Billat et al., A PBPK model to evaluate zebrafish eleutheroembryos’ actual exposure: bisphenol A and analogs’ (AF, F, and S) case studies, Environmental Science and Pollution Research, 2023.
[3] Christian I. Rude et al., A mixture parameterized biologically based dosimetry model to predict body burdens of polycyclic aromatic hydrocarbons in developmental zebrafish toxicity assays, Society of Toxicology, 2025.
[4] Tasuku Nawaji et al., Cytochrome P450 Expression and Chemical Metabolic Activity before Full Liver Development in Zebrafish, Pharmaceuticals, 2020.
[5] Chloe Bars et al., Developmental Toxicity and Biotransformation of Two Anti-Epileptics in Zebrafish Embryos and Early Larvae, International Journal of Molecular Sciences, 2021.
Reference: PAGE 34 (2026) Abstr 11968 [www.page-meeting.org/?abstract=11968]
Poster: Methodology - New Modelling Approaches