Larval zebrafish heart rate assay for early cardiac safety liability detection
By 72-96 hours post-fertilization, the zebrafish larval heart is a fully functional two-chambered pump — one atrium, one ventricle, connected by an atrioventricular canal — beating at 120-180 beats per minute and clearly visible through the translucent body wall without any surgery, staining, or instrumentation.
Despite having two chambers instead of four, the zebrafish heart shares striking electrophysiological similarity with the human heart. Critically, the ratio of the rapid delayed rectifier potassium current (IKr, carried by the hERG/KCNH2 channel) to other repolarizing currents is closer to human cardiomyocytes than it is in mouse or rat, whose heart rates (400-600 bpm) are dominated by different repolarizing currents. This makes zebrafish unusually predictive for detecting QT-interval-prolonging drugs — a leading cause of late-stage drug attrition and post-market withdrawal.
The larval heart at 72-96 hpf beats spontaneously and rhythmically without neural input required (myogenic automaticity from the sinoatrial ring), so the assay measures intrinsic cardiac electrophysiology and contractility directly, uncomplicated by systemic autonomic regulation.
Roughly 45 human cardiac ion channel and structural genes have direct zebrafish orthologs, including kcnh2 (hERG), scn5a (Nav1.5), and cacna1c (Cav1.2) — the same three channels responsible for the majority of clinically observed drug-induced arrhythmias.
The zebrafish heart forms remarkably fast: cardiac progenitor cells migrate to the midline and fuse into a linear heart tube by ~24 hpf; the tube loops into an S-shape positioning the future atrium and ventricle by ~36-48 hpf; chamber-specific gene expression (myl7 in both chambers, vmhc in ventricle, amhc in atrium) differentiates the two chambers by 48 hpf; and by 72 hpf the heart is pumping blood through a closed circulatory loop with visible red blood cell flow in the trunk and tail vasculature.
By the time the cardiotoxicity assay window opens (72-96 hpf), the heart has reached a stable rhythmic baseline, making this the standard timepoint for compound screening.
Larvae are typically arrayed one-per-well in 96- or 384-well plates, oriented laterally or dorsally under a fixed imaging stage. Because no surgery or dye injection is required — the heart is visible directly through the skin — an entire plate can be recorded in minutes, enabling hundreds of compounds to be screened per week per imaging station, positioning the zebrafish larval heart rate assay as an efficient bridge between low-throughput patch-clamp hERG assays and expensive in vivo mammalian telemetry studies.
Natural inter-individual variability in larval heart rate (typically ±15-20 bpm even among clutch-mates at identical developmental stage) means that comparing a treated group to a separate untreated group can obscure real drug effects. The gold-standard design instead uses each larva as its own control.
Each larva is recorded once before compound exposure (baseline) and again after a fixed exposure period (typically 1-24 hours depending on protocol), under identical temperature, lighting, and camera settings. Percent change from that individual's own baseline — rather than absolute bpm — is the primary readout, which controls for developmental-stage variability, clutch effects, and minor differences in larval size or orientation.
Temperature control is critical: zebrafish are ectothermic, and heart rate is strongly temperature-dependent (a Q10 effect similar to many ectotherm physiological rates), so recording chambers are thermostatted to 28.5°C ± 0.5°C, the standard zebrafish husbandry temperature.
Using within-animal baseline normalization typically reduces the effective variability of the assay by more than half compared to between-group comparison, which is what allows the assay to reliably detect heart rate changes as small as 10-15% — well within the range produced by clinically relevant hERG blockers at sub-lethal concentrations.
For consistent optical access to the heart, larvae are gently immobilized — either lightly anesthetized with low-dose tricaine (MS-222, at a concentration too low to itself suppress heart rate) or physically restrained in low-melting-point agarose or a microfluidic channel — and oriented laterally so both chambers are in the same focal plane. Anesthetic dose is carefully titrated and validated against unanesthetized controls, since excess tricaine itself measurably slows heart rate and can confound the assay.
Test compounds are added directly to the well medium at a dilution series spanning several orders of magnitude, and larvae are re-imaged after a standardized incubation window. Cardiotoxic compounds produce a spectrum of effects — from mild rate slowing to complete atrioventricular conduction block.
Cardiotoxic compounds in the zebrafish larval assay produce several recognizable, quantifiable phenotypes:
• Bradycardia — uniform slowing of both atrial and ventricular rate, often the first sign of general cardiac ion channel disruption • Atrioventricular (AV) block — the atrium continues beating at normal rate while the ventricle beats slower or intermittently (e.g., 2:1 block, where the ventricle contracts only once for every two atrial beats) — a hallmark of conduction-pathway-specific toxicants and a classic signature of hERG/IKr blockade • Arrhythmia / irregular rhythm — beat-to-beat interval variability increases, sometimes with skipped beats or fibrillation-like chaotic contraction • Pericardial edema secondary to cardiac dysfunction — reduced cardiac output leads to fluid accumulation, visible as the heart's mechanical output fails • Complete cardiac arrest — at high concentrations, terminal cessation of contraction, usually indicating severe or lethal cardiotoxicity
2:1 atrioventricular block — where the ventricle drops every other beat while the atrium continues normally — is one of the most specific and easily scored phenotypes in the assay, and closely mirrors the conduction abnormalities seen on human ECG with strong hERG blockers such as terfenadine and cisapride, both withdrawn from market for QT-prolongation risk.
Because the zebrafish is a whole organism, a slowed heart rate can arise indirectly — from general systemic toxicity, sedation, or reduced metabolic rate — rather than a direct cardiac ion channel or contractility effect. Well-designed screens address this by including a viability/activity readout in parallel (e.g., touch-evoked response, general locomotion) so that a compound producing bradycardia together with clear general malaise is flagged differently from one producing selective, isolated cardiac rate/rhythm effects at concentrations where the larva otherwise behaves normally.
Manual beat-counting by eye is slow and subjective. Modern cardiotoxicity screens use automated video analysis pipelines that track the periodic optical density oscillation of each chamber as it fills and empties, converting raw pixel intensity time series into precise, chamber-resolved heart rate and rhythm metrics.
A region of interest (ROI) is drawn (manually or via automated chamber detection) around the atrium and separately around the ventricle in each video frame. As each chamber contracts and relaxes, its optical density changes cyclically — the chamber appears darker when full of blood and lighter when contracted and empty (or vice versa depending on illumination). Plotting mean pixel intensity within each ROI across all video frames produces a clean periodic waveform whose peak-to-peak interval directly gives the beat period for that chamber.
Fast Fourier Transform (FFT) or peak-detection algorithms applied to this waveform extract instantaneous heart rate, beat-to-beat interval variability (a rhythm regularity metric), and — critically — the time delay between the atrial and ventricular waveform peaks, which is the AV conduction interval.
Comparing the independently-derived atrial rate and ventricular rate automatically reveals AV block: if the atrial FFT peak frequency differs from the ventricular FFT peak frequency by a simple integer ratio (2:1, 3:1), the software flags conduction block without any manual interpretation. Beat-to-beat interval coefficient of variation above a threshold flags arrhythmia. This automation is what allows a full 96- or 384-well plate to be scored in minutes rather than the hours required for manual visual beat counting, and removes observer bias entirely — a critical feature for regulatory-quality screening data.
Fully automated video pipelines can process an entire 384-well plate — over 300 larvae with pre/post dosing videos — in under 20 minutes of compute time, versus multiple technician-days for manual scoring, making heart rate one of the highest-throughput in vivo toxicity endpoints available.
The final analytical step converts raw per-larva heart rate measurements into a population dose-response curve and benchmarks the resulting potency against a reference library of known hERG channel blockers and clinically safe comparators, translating a fish phenotype into an actionable human cardiac safety signal.
Percent change in heart rate from individual baseline is averaged across replicate larvae (typically n=8-16 per concentration) at each dose and plotted against log(concentration). A four-parameter logistic fit yields the EC50 for bradycardia — the concentration producing a 50% reduction in heart rate — along with the concentration at which AV block or arrhythmia first appears, which is often, but not always, at a lower dose than outright bradycardia.
This EC50, along with the compound's free plasma concentration at the intended human therapeutic dose, allows calculation of a safety margin analogous to the "IC50 margin" used in traditional hERG patch-clamp risk assessment.
A screening lab validates and calibrates its zebrafish assay by running a reference panel of compounds with well-characterized human hERG channel activity and clinical QT outcomes — for example strong blockers such as terfenadine, cisapride, and dofetilide (withdrawn or black-box-labeled for QT risk) alongside weak/non-blockers such as loratadine or metoprolol. Plotting zebrafish bradycardia/AV-block EC50 against literature hERG patch-clamp IC50 for this reference set establishes the assay's predictive concordance and its useful potency range, so that a novel compound's zebrafish EC50 can be interpreted against this calibration curve.
Because the zebrafish assay captures the integrated output of the whole cardiac action potential — not just the hERG channel in isolation — it can also detect risk from off-target mechanisms (calcium or sodium channel effects, mitochondrial cardiotoxicity, structural effects) that a single-channel patch-clamp hERG assay would miss entirely, making it a valuable complementary, whole-organism tier in the cardiac safety testing cascade.
The zebrafish larval heart rate assay is typically deployed as a mid-throughput, whole-organism tier: after high-throughput single-channel hERG patch-clamp or fluorescence-based ion-flux assays (which screen thousands of compounds but miss whole-organism and off-target effects) and before resource-intensive mammalian in vivo telemetry or isolated perfused heart (Langendorff) studies, which are reserved for a small number of lead candidates. This positioning lets a screening program triage and de-risk hundreds of compounds per month at modest cost before committing to the most expensive downstream mammalian cardiac safety studies.