Automated high-throughput heart rate measurement in medaka and zebrafish embryos under physiological conditions

Rationale Accurate and efficient quantification of heartbeats in small fish models is an important readout to study cardiovascular biology, disease states and pharmacology at large scale. However, dependence on anesthesia, laborious sample orientation or requirement for fluorescent reporters have hampered the establishment of high-throughput heartbeat analysis. Objective To overcome these limitations, we aimed to develop a high-throughput assay with automated heart rate scoring in medaka (Oryzias latipes) and zebrafish (Danio rerio) embryos under physiological conditions designed for genetic screens and drug discovery and validation. Methods and Results We established an efficient screening assay employing automated label-free heart rate determination of randomly oriented, non-anesthetized specimen in microtiter plates. Automatically acquired bright-field data feeds into an easy-to-use HeartBeat software, a MATLAB algorithm with graphical user interface developed for automated quantification of heart rate and rhythm. Sensitivity of the assay and algorithm was demonstrated by profiling heart rates during entire embryonic development. Our analysis pipeline revealed acute temperature changes triggering rapid adaption of heart rates, which has implications for standardization of experimental layout. The approach is scalable and allows scoring of multiple embryos per well resulting in a throughput of >500 embryos per 96-well plate. In a proof of principle screen for compound testing, our assay captured concentration-dependent effects of nifedipine and terfenadine over time. Conclusion A novel workflow and HeartBeat software provide efficient means for reliable and direct quantification of heart rate and rhythm of small fish in a physiological environment. Importantly, confounding factors such as anesthetics or laborious mounting are eliminated. We provide detailed profiles of embryonic heart rate dynamics in medaka and zebrafish as reference for future assay development. Ease of sample handling, automated imaging, physiological conditions and software-assisted analysis now facilitate various large-scale applications ranging from phenotypic screening, interrogation of gene functions to cardiovascular drug development pipelines.


Non-standard abbreviations and acronyms
Transposon Oryzias latipes 2

Introduction
Quantification of heartbeats is an important readout to study physiology and disease states of the heart. The resting heart rate (beats per minute, bpm) is a strong predictive risk factor of overall mortality 1 and associated with a growing catalogue of genetic variants and environmental-sensitive alleles 2,3 . However, geneticists are facing an enormous challenge to establish causality for associated candidate variants. Small fish models provide efficient means of functional assessment in the context of a vertebrate.
In fish models heart rate is examined in biomedical research 4 , including the analysis of inherited or acquired arrhythmia 5,6 , in phenotypic drug discovery and safety pipelines 7 or in toxicological studies 8 . Moreover, the recent establishment of a vertebrate panel of isogenic strains in medaka fish enables to study variations of quantitative traits such as heart rate in genome-wide association studies 9 . However, there is a lack of efficient screening workflows that allow large-scale quantitative scoring of cardiac phenotypes. Therefore, novel and efficient protocols were needed encompassing sample preparation, automated imaging and image analysis.
The fish species medaka (Oryzias latipes) and zebrafish (Danio rerio) emerged as major vertebrate models for phenotypic screening, follow-up functional studies, reverse genetics and for studying human pathologies in mechanistical details 10,11 .
Medaka and zebrafish have tractable diploid genomes, feature a short generation time (2-3 months), external embryonic development and established protocols for transgenesis and manipulation of gene function including CRISPR/Cas-based approaches 12,13 . Economic husbandry and small transparent embryos are particularly suited for large-scale screening and detailed imaging 14,15 . In both fish species, rapid cardiovascular development leads to a functional cardiovascular system emerging by around 24 hours post fertilization (hpf) in zebrafish 16 and approximately 48 hpf in medaka 17 . Although the fish heart is two-chambered, its development and function are very similar to the mammalian heart 18 . Strikingly, components of the electrocardiogram of zebrafish and medaka have more similarities to those in humans than have those of rodents, making them ideal models to investigate heart rate and rhythm phenotypes and leverage these properties for drug screens [19][20][21] . In contrast to mammalian models, early zebrafish embryos are sufficiently supplied by oxygen diffusion allowing the study of severe cardiovascular and dysfunctional hemoglobin phenotypes that result in embryonic lethality in mammals 22 .
Readouts of heart rate and rhythm have been employed increasingly in zebrafish-and medaka-based screens [23][24][25][26][27][28][29][30] . Challenges of current methods to capture heart rate in high-throughput include: (i) immobilization of dechorionated embryos or hatchlings, which in the majority of cases is achieved by anesthesia with tricaine leading to severe adverse cardiovascular effects 31 , (ii) imaging of larvae in standardized orientations, involving labor-intensive manual mounting on dedicated sample carriers or embedding in low-melting agarose, (iii) fluorescent reporter readouts, which preclude studies of specific genetic backgrounds and (iv) lack of user-friendly software for efficient analysis of multiple hearts simultaneously.
We therefore developed a high-throughput screening pipeline with simple sample handling and fully automated imaging of unhatched medaka and zebrafish embryos.  35 .

Sample preparation and time lapse experiments
One day prior to imaging medaka eggs were optically cleared by rolling the eggs on sandpaper (P1000). One hour before imaging, medaka and zebrafish embryos were transferred from methylene blue-containing medium into plain ERM and E3,

Data analysis and statistics
Statistical analyses were computed in R language 36 and data visualized using the ggplot2 package 37 . Differences between samples were tested with Student's t-test and statistical significance accepted at a threshold of P<0.05. Multiple comparisons were tested with one-way ANOVA and significant results (P<0.05) were analyzed with pairwise comparisons using Student's t-test applying significance levels adjusted with the Bonferroni method. Significant P-values are indicated with asterisks (*) with *P<0.05, **P<0.01 and ***P<0.001. Correlation analysis was performed using Pearson's correlation.

High-throughput assay and Heartbeat software
We implemented a rapid workflow for automated imaging of medaka and zebrafish embryos and a new Heartbeat software for extraction of heart rate and rhythm. The multi-step pipeline is schematically summarized in Figure 1.

Heart rate dynamics in medaka and zebrafish development
We applied the assay to score heart rates during the entire development of medaka and zebrafish embryos. Onset of heartbeat varied in the two fish species with earliest heartbeats appearing after 22 hpf in zebrafish and 40 hpf in medaka. From this time points onwards, image data was captured every 4 h (Figure 4). In medaka, embryonic heart rate was rapidly increasing over developmental time and started to oscillate in a day-night rhythm from 72 hpf onwards and reached approximately 175 bpm (at 28°C) around hatching stage (7 dpf, Figure 4A). Similarly, after onset of heartbeat in zebrafish, heart rate increased and plateaued at around 210 bpm (28°C, 72 hpf; Figure   4B). Zebrafish heart rate did not exhibit a day-night dependent fluctuation during the period of analysis (22-62 hpf).

Temperature-dependent modulation of heart rate
Ambient temperature is a key environmental factor affecting heart rate in most teleosts.
To quantify this effect, we studied the influence of ambient temperature changes on heart rate between 21-35°C ( Figure 5). Heart rates, both in medaka and zebrafish, accelerate linearly with temperature. Zebrafish exhibited a higher heart rate baseline at all temperatures and a slightly steeper slope of increase ( Figure 5B). These results underline the importance of tight temperature control when assessing any heartbeat related cardiac phenotypes.

Compound effects on heart rate of unhatched medaka and zebrafish embryos
Heart rate is frequently scored in preclinical screening assays of chemical compounds employing small fish models. To demonstrate the compatibility of our platform with high-throughput chemical screening, we trialed two pharmaceuticals, nifedipine and terfenadine, with known inhibitory effects on heart rate in fish 7,38 . In one 96-well plate, groups of 12 embryos were exposed to terfenadine and nifedipine each at 10, 30 and  Figure 7A). In zebrafish embryos, heart rates were decreased by nifedipine in a concentration-and time-dependent manner resulting in asystole for the majority of embryos treated with 100 µmol/L (µM) nifedipine for 120 min ( Figure 7B). To provide a reference for embryonic applications using DMSO carrier, heart rates were assessed for a concentration range of 0.2-1%. In medaka, heart rates were not significantly affected by different DMSO concentrations within 2 h of incubation, whereas minor heart rate fluctuations occurred at lower DMSO concentrations in zebrafish (Supplemental Figure 2). As a control for stage-dependent changes of heart rate, all measurements (drugs and DMSO) were normalized to a negative control group of age-matched untreated embryos (Figure 7 and Supplemental Figure 2).

Discussion
Heart rate is a key parameter in numerous small fish model applications including genetics and pharmacology and a reliable metric for environmental studies assessing cardiotoxic and adverse developmental effects of small molecules and pollutants.
Here, we present a physiological and low-cost assay and open-access HeartBeat software for rapid quantitative assessment of heart rate and rhythm of medaka and zebrafish embryos with high analytical power in terms of precision and throughput that permits large-scale phenotypic scoring. The performance of our pipeline allowed to capture concentration-dependent effects of cardiovascular active drugs with high temporal resolution.
Our assay overcomes laborious manual mounting and orientation of samples, motion artifacts, confounding effects of anesthesia and dependency on fluorescent reporters. To explore the maximal throughput, HeartBeat software was applied to readout multiple fish per well and reliably captured heart rates of up to 10 embryos per well.
Differences between mean heart rates extracted from wells with different embryo densities ( Figure 3) might be reflecting varying physiological environments. For comparative assays this needs to be controlled by using the same numbers of embryos per well for the entire plate. Automated capturing of multiple hearts was performed most efficiently up to 6 embryos per well resulting in >500 embryos processed per 96well plate, particularly suitable for drug screening pipelines.
To date, several methods have been used to assess heartbeat in zebrafish hatchlings, medaka embryos, fish species, Drosophila, cardiomyocytes or mouse embryos 23,24,26,27,29,30,[39][40][41][42][43] . Some of these methods require transgenic embryos with fluorescent cardiomyocytes or erythrocytes to measure heartbeats 28,39 . Other approaches depend on advanced microscopes such as confocal laser scanning microscope 28 or dual-beam optical reflectometer 41 , imaging modalities that may affect sample temperature and confounding effects on heart rate ( Figure 5). Another caveat is immobilization of fish larvae in low-melting agarose to achieve specific orientations, which is time-consuming and complicates genuine high-throughput assays. Some methods were restricted to a narrow time window of newly hatched zebrafish prior to swim bladder inflation to avoid movements of the larvae during sampling 30,39 . In addition, larvae were acclimatized to the illumination of the well scans. To achieve stable imaging condition many protocols rely on anesthesia 23,24,26,29,40 . The most widely used anesthetics is tricaine (MS-222, ethyl 3-aminobenzoate methanesulfonate), that affects cardiovascular functions including heart rate and contractility 44,45 . Our assay, on the other hand, relies on recording non-anesthetized unhatched medaka and zebrafish embryos in their natural environment within the chorion allowing fast sample distribution into microtiter plate, and is compatible with robotic handling of fish embryos in systematic screens 46,47 .
We demonstrate robust heart rate quantification of randomly positioned and oriented fish embryos and implemented options in the HeartBeat GUI to easily control artefacts, e.g. through excessive embryo movement during acquisition. Full 96-well plates are imaged in 20 min with 10 s videos at 13 fps for each well. High-confidence quantification can be performed within 45-60 min. In the meantime, our reliable and efficient assay permitted us to automatically quantify heart rate of more than 20 000 image sequences corresponding to approximately 20 TB of data.
While only scant data of embryonic medaka heart rate was available 17 , using the HeartBeat assay we now provide a systematic profile as reference for future studies ( Figure 4A). Interestingly, after 3 days of development heart rate in medaka showed day-night undulation with a nocturnal dipping resembling the circadian heart rate patterns in healthy humans 48 . Zebrafish has faster embryogenesis and morphological differentiation including looping of the heart is completed by 36 hpf 49 . Serial recordings from onset of heartbeat onwards reliably delineated the rapid increase of heart rate associated with major steps of cardiac development within the first 2.5 dpf ( Figure 4B).
As aerobic capacity and metabolic rate are directly interrelated with ambient temperature, incremental warming ( Figure 5) can be applied as physical stressor in ectothermic medaka and zebrafish embryos, which regulate cardiac output in response to temperature primarily through adaption of heart rate rather than stroke volume 50,51 .
Tight control of heart rate by temperature can be leveraged as readout of acute and chronic adaptive stress responses, temperature-sensitive strains 52 and of susceptibility to arrhythmia in strains with different genetic backgrounds. In summary, our detailed analysis stresses the importance to rigorously control temperature and normalize to stage-dependent changes of beat rates to ensure experimental standardization.
Heartbeat detection is a key metric for drug testing protocols assessing therapeutic and adverse actions of small molecules using fish models. To assess applicability of our assay for compound treatment, we used nifedipine and terfenadine. Concentrationdependent heart rate inhibition and dynamic changes of drug effects were recorded in a 2 h time lapse experiment, reliably correlating with published negative chronotropic effects of terfenadine and nifedipine in zebrafish 7,38 . In human, the antihistamine terfenadine can induce adverse prolongation of QT interval resulting in increased risk for torsades de pointes tachycardia. Detailed studies have demonstrated QT prolongation resulting in AV block in zebrafish 53 and medaka 52 . Probing effects of terfenadine on zebrafish and medaka embryonic heart rate, distinct time points of induction and time courses of AV block were uncovered by our assay (Figure 7). In zebrafish, nifedipine suppresses cardiac Ca 2+ channels resulting in a marked decrease of heart rate 38,54 . This drug action was reliably monitored with our assay showing significant effects ranging from sustained inhibition to asystole over time (Figure 7).
Taken together, this proof of concept pharmacological experiment in combination with the scalability of our assay raises the prospect of accelerating drug discovery pipelines and rapid pre-clinical prediction of adverse drug effects.
In summary, by combining ease and stability of imaging acquisition with softwareassisted analysis our method offers a comprehensive platform for forward and reverse genetics, gene x environment interaction studies, pharmaceutical screens and toxicological assessment, which can efficiently direct downstream studies on mechanistical details of heart rate related phenotypes.     96-well format for incubation of medaka and zebrafish embryos with terfenadine and nifedipine each at three different concentrations in µmol/L (µM). As a control for stagedependent changes of heart rate, all measurements (drugs and DMSO) were normalized to a negative control group of age-matched untreated embryos (row H).
The Heartbeat software directly provides a heatmap output for intuitive assessment of drug effects (heart rate response shown at 90 min; nd: not detected).