Zebrafish embryo teratogenicity screening (OECD TG 236 fish embryo acute toxicity)
The zebrafish (Danio rerio) embryo has become the workhorse of predictive developmental toxicology. Optically transparent, externally fertilized, and developing from a single cell to a free-swimming larva in 5 days, it allows every organ system to be watched forming in real time under a compound of interest.
Zebrafish embryos develop entirely outside the mother in a transparent chorion, so every stage of organogenesis — gastrulation, neurulation, somitogenesis, heart tube formation, fin bud outgrowth — can be observed non-invasively with a standard dissecting microscope. A clutch of 100-300 eggs is obtainable from a single mating pair on demand, enabling statistically powered replicate groups (typically n=20 embryos per concentration) at a fraction of the cost and time of rodent teratology studies.
Development is extremely rapid and highly stereotyped: by 24 hours post-fertilization (hpf) the primary body plan, neural tube, and beating heart are established; by 72 hpf, the larva has functional eyes, jaw, gut, and swim bladder primordium; by 120 hpf it feeds independently. This compressed timeline means an entire developmental toxicity screen — from fertilization to definitive endpoint scoring — completes within 5 days, versus 10-14 days gestation-plus-observation windows in rodent studies.
OECD Test Guideline 236 (adopted 2013) formalizes the Fish Embryo Acute Toxicity (FET) test using zebrafish as a validated, internationally recognized alternative to juvenile/adult fish acute toxicity testing, reducing animal use under the 3Rs principle since embryos below free-feeding stage are generally not classified as protected "animals" under most regulatory frameworks.
Adult zebrafish are set up in breeding tanks the evening before, separated by a divider that is removed at first light to trigger spawning within 15-30 minutes, synchronized with the light cycle. Eggs are collected by gentle sieving, rinsed of debris, and examined under a stereomicroscope.
Only embryos meeting strict inclusion criteria proceed: fertilized (visible cell division), at the 4-64 cell cleavage stage (within ~2.5 hpf), spherical with an intact chorion, and free of visible abnormalities. Unfertilized eggs (opaque, no cleavage) and asymmetric or fragmented embryos are discarded. This staging window is critical — compound exposure must begin before major developmental decisions (gastrulation at ~5.25 hpf) to capture the full teratogenic window.
Embryos are individually pipetted into wells of a 24- or 96-well plate containing embryo medium (E3: NaCl, KCl, CaCl2, MgSO4) with the test compound at one of 5-8 half-log dilution concentrations, plus solvent (vehicle) and negative controls. A reference toxicant (e.g., 3,4-dichloroaniline) run in parallel validates assay sensitivity on each plate per OECD TG 236 acceptance criteria (≥90% negative-control survival).
Standard group size is n=20 embryos per concentration, run in at least biological duplicate, giving the statistical power to detect a 10-20% shift in malformation incidence — the resolution needed to place a compound's teratogenic potency on a comparative ranking scale.
Exposure begins at the blastula/gastrula stage and continues uninterrupted through the most sensitive window of organogenesis. The chorion is permeable to most small molecules, but its pore structure (~0.5-0.7 μm) can restrict uptake of larger or highly lipophilic compounds, a key consideration in FET assay interpretation.
A preliminary range-finding study (limit test or 5-point log dilution) establishes the working concentration window before the definitive study. Concentrations typically span from a no-observed-effect level up to and beyond the LC50, in half-log (√10) increments to give even spacing on a log-dose axis for curve fitting.
Compounds are dissolved directly in embryo medium where solubility allows; poorly soluble compounds are carried in a minimal solvent (DMSO, ≤0.1% v/v) with a solvent-only control included to rule out carrier effects. Static-renewal dosing — replacing exposure medium every 24 h — maintains nominal concentration despite compound degradation, evaporation, or adsorption to the polystyrene plate, and is preferred over fully static exposure for compounds with limited stability.
Because the chorion is a semi-permeable barrier, nominal (dosed) concentration and internal (embryo tissue) concentration can diverge substantially for large, charged, or highly protein-reactive compounds — a source of false negatives that dechorionation (enzymatic pronase treatment) can resolve when internal dose is the parameter of interest.
Teratogenic risk is highest when exposure coincides with the formation of a given organ system. In zebrafish this maps onto a well-characterized timeline: neural tube closure and somitogenesis (10-24 hpf), heart tube formation and looping (24-48 hpf), fin bud outgrowth and craniofacial cartilage patterning (48-72 hpf), and swim bladder inflation (72-120 hpf).
A compound that disrupts retinoic acid signaling, for example, produces its most severe phenotype when exposure spans 10-24 hpf (fin and hindbrain patterning), whereas a cardiotoxicant's critical window centers on 24-48 hpf. Standard FET protocol exposes continuously from ~2-24 hpf through 96 hpf to capture the full spectrum regardless of mechanism.
Embryos/larvae are inspected daily (and at 24 hpf specifically per OECD TG 236) for four apical observations that together define lethality: coagulation of the embryo (opaque, necrotic mass), lack of somite formation, non-detachment of the tail from the yolk, and absence of heartbeat. Any one of these four signs scores the animal as dead, and dead animals are removed immediately to prevent water quality deterioration affecting surviving wellmates.
Between 48 and 72 hpf the zebrafish larva acquires nearly all of its recognizable vertebrate anatomy — a beating two-chambered heart, segmented axial musculature, pigmented eyes, and paired fin folds. Teratogenic compounds perturb this choreography in characteristic, scoreable ways well before lethal concentrations are reached.
A standardized semi-quantitative rubric captures teratogenic phenotype severity across independent organ systems, typically scored 0 (normal) to 2-3 (severe) per category:
• Pericardial and yolk sac edema — fluid accumulation from impaired osmoregulation or cardiac output, visible as a translucent swelling around the heart or yolk • Body axis curvature — dorsal, ventral, or lateral bending of the notochord/spine, often from notochord vacuolation defects or muscular dysfunction • Craniofacial malformation — underdeveloped jaw, small eyes (microphthalmia), or absent otoliths (inner-ear crystals needed for balance) • Fin fold and tail malformation — truncated, blistered, or fin-fold degeneration • Pigmentation deficits — reduced or absent melanophore patterning, a sensitive endpoint for compounds disrupting neural crest development • Circulatory defects — reduced or absent blood flow, visible directly in the transparent tail vasculature under brightfield or video microscopy
Pericardial edema is among the most sensitive and reproducible teratogenic endpoints in the zebrafish assay because cardiac output failure cascades rapidly into fluid retention — many cardiotoxic and several unrelated systemically toxic compounds converge on this single visible phenotype.
Standardized lateral-view images are captured for each larva at fixed magnification, and morphometric software (or trained scorers under blinded conditions) measures body axis angle, pericardial area, and yolk sac area against reference templates. Increasingly, deep-learning image classifiers trained on thousands of scored larvae automate this step at screening scale, reducing inter-observer variability from ~15% (manual) to under 5%.
A composite Developmental Toxicity Score is derived by summing individual category scores, allowing a single dose-dependent severity metric to be plotted alongside the binary mortality curve.
Hatching requires the coordinated action of hatching gland enzymes (chorionase) that digest the chorion, combined with physical thrashing movements of the larva. Because this process integrates enzymatic, muscular, and neurological function, delayed hatching is one of the most sensitive and mechanistically broad indicators of developmental toxicity.
Hatching gland cells (the polster, a cluster located on the anterior yolk) secrete zebrafish hatching enzyme 1 (ZHE1), a chorionase that proteolytically softens the chorion beginning around 48 hpf. Combined with vigorous tail-flexion movements driven by the developing spinal motor circuit, this ruptures the chorion and releases the free-swimming larva, normally by 72 hpf.
Because hatching requires normal enzyme secretion, normal muscle development, and normal neuromuscular signaling simultaneously, a delay or failure to hatch is a highly sensitive, mechanistically non-specific flag: compounds that inhibit protease activity, disrupt myogenesis, or block neurotransmission can all delay hatching, even at concentrations below those producing overt morphological malformation.
At the study's definitive timepoint (96 hpf under OECD TG 236, sometimes extended to 120-144 hpf for extended teratogenicity protocols), every surviving larva — hatched or manually dechorionated if still trapped — is individually imaged and scored against the full morphological rubric plus the four lethality criteria. Data are compiled per concentration group into incidence tables: percent mortality, percent delayed/failed hatching, percent with any malformation, and mean composite severity score.
A larva that fails to hatch by 96 hpf despite a beating heart and blood flow is classified as a sub-lethal developmental effect, not mortality — this distinction matters because pooling delayed hatchers into the mortality count would artificially inflate the LC50 and mask a real teratogenic signal.
Pooled incidence data across the concentration series are fit to a sigmoidal (typically four-parameter log-logistic) dose-response model, yielding two anchor potency values — LC50 for lethality and EC50 for teratogenicity — whose ratio quantifies a compound's selective developmental hazard relative to its general toxicity.
At each tested concentration, the fraction of embryos that died (mortality) and the fraction of survivors displaying any scored malformation (teratogenicity) are computed. These proportions are fit against log(concentration) using a four-parameter logistic (Hill) equation:
Response(C) = bottom + (top - bottom) / (1 + 10^((logEC50 - logC) × slope))
Nonlinear regression (least-squares or maximum-likelihood probit/logit) yields LC50 (concentration killing 50% of embryos) and EC50 (concentration producing malformation in 50% of survivors), each with 95% confidence intervals from the fit.
The Teratogenic Index (TI = LC50 / EC50) expresses how much lower the teratogenic threshold sits relative to the lethal threshold. A TI near 1 means malformation and death occur at essentially the same concentrations — the compound is generally toxic but not selectively teratogenic. A TI substantially greater than 1 (commonly TI > 3 used as a screening flag) indicates the compound perturbs development at doses well below those causing outright death — the hallmark signature of a developmental toxicant such as retinoic acid, valproic acid, or thalidomide analogs, all of which show strong positive control responses in the zebrafish FET assay.
This TI framework directly parallels mammalian teratology practice, where a similar ratio (maternal toxic dose vs. fetal effect dose) is used to flag selective developmental hazard, allowing zebrafish data to feed into early-stage compound triage before committing to costly rodent studies.
Thalidomide — historically missed by rodent screens due to species-specific metabolism — produces clear fin and pectoral appendage malformations in zebrafish at sub-lethal concentrations, illustrating both the assay's value and the importance of interpreting any single-species negative result cautiously.
The FET assay is accepted under OECD TG 236 as a surrogate for acute fish toxicity testing and is increasingly incorporated into ECHA REACH data requirements and pharmaceutical developmental toxicity screening cascades ahead of definitive mammalian studies. A single technician can run a 96-well definitive study (7 concentrations × triplicate × 20 embryos, plus controls) in under a week from egg collection to final data analysis — throughput that makes early teratogenicity triage of hundreds of candidate compounds per year feasible, dramatically reducing the number of compounds advanced into resource-intensive mammalian teratology studies.