Streamlining drug discovery assays for cardiovascular disease using zebrafish

ABSTRACT

Introduction: In the last decade, our armamentarium of cardiovascular drug therapy has expanded significantly. Using innovative functional genomics strategies such as genome editing by CRISPR/Cas9 as well as high-throughput assays to identify bioactive small chemical compounds has significantly facilitated elaboration of the underlying pathomechanism in various cardiovascular diseases. However, despite scientific progress approvals for cardiovascular drugs has stagnated significantly compared to other fields of drug discovery and therapy during the past years.

Areas covered: In this review, the authors discuss the aspects and pitfalls during the early phase of cardiovascular drug discovery and describe the advantages of zebrafish as an in vivo organism to model human cardiovascular diseases (CVD) as well as an in vivo platform for high-throughput chemical compound screening. They also highlight the emerging, promising techniques of automated read-out systems during high-throughput screening (HTS) for the evaluation of important cardiac functional parameters in zebrafish with the potential to streamline CVD drug discovery.

Expert opinion: The successful identification of novel drugs to treat CVD is a major challenge in modern biomedical and clinical research. In this context, the definition of the etiologic fundamentals of human cardiovascular diseases is the prerequisite for an efficient and straightforward drug discovery.

KEYWORDS:

• Zebrafish
• cardiovascular disease
• drug discovery
• small compound screen

1. Introduction

During the past 10 years, our armamentarium of medical drug therapy for various cardiovascular diseases (CVD) such as acute myocardial infraction (AMI), heart failure (HF), stroke or diabetes (DM) has expanded significantly.

For instance, application of novel adenosine diphosphate receptor (P2Y12) inhibitors in AMI patients undergoing percutaneous coronary artery stent implantation (PCI) reduces significantly rate of death from myocardial infarction in comparison to standard drug therapy with clopidogrel [1,2]. The recently introduced therapeutic drug LCZ696, an angiotensin-receptor-neprilysin-inhibitor (ARNI), leads to an impressive reduction in heart failure-related death and rehospitalization in patients with severely reduced left ventricular function [3]. Also, with the introduction of the sodium-dependent-glucose-transporter (SGLT-2) inhibitor empagliflozin, patients suffering from type-II-diabetes can reduce their risk dying from severe cardiovascular events [4].

To enable successful drug discovery, identification of druggable molecular targets, which are involved in the underlying disease pathomechanism, is obligatory. Development of novel disease modeling systems such as reprogramming adult somatic cells into induced pluripotent stem cells and subsequently the desired cell type such as iPS-derived cardiomyocytes or innovative genomic technologies such as genome/gene editing by CRISPR/Cas9 have facilitated tremendously biological evaluation of candidate drugs or are even used for therapeutic approaches in the recent past [5,6]. Despite these substantial forward steps in CVD drug discovery, cardiovascular disorders are still the number one cause of death worldwide with more than 17 million estimated death in 2020 [7–9].

However, even though innovative cardiovascular drug discovery is more important and necessary than ever, approvals for cardiovascular drugs have stagnated compared to other fields of drug therapy, e.g. antineoplastic agents for the treatment of cancer diseases [10,11]. Several reasons are responsible for this dilemma ranging from a still limited understanding of CVD pathophysiology due to a lack of appropriate disease models to high drop-out rates of candidate drugs already in the early phase of drug discovery. Hence, making the challenging run of CVD drug discovery more efficient is one of the major challenges in modern biomedical research. Especially, timely in vivo assessment of essential cardiac functional parameters such as the heartbeat or systolic ventricular pump function are needed to predict both biological impact and severe side effects of a candidate therapeutic agent in the early preclinical stage of drug discovery.

In our article, we primarily discuss aspects and pitfalls during early phase drug development for CVD and highlight the advantages of the zebrafish as a cardiovascular disease modeling organism as well as an in vivo platform for high-throughput screening (HTS). Furthermore, we outline current and emerging techniques of automated HTS read-out systems for evaluation of important cardiac functional parameters in zebrafish with the potential to significantly streamline CVD drug discovery.

2. Aspects and pitfalls during the early phase of cardiovascular drug discovery

Efficient drug discovery comprises several systematic steps primarily including in vitro assessment of chemical small compound libraries of bioactive molecules, followed by application of candidate bioactive agents in vertebrate disease modeling organisms and finally clinical application of a promising drug candidate in humans. It is estimated that in this long-lasting and expensive process thousands of bioactive small compounds must be tested to achieve a single therapeutic drug approved for clinical practice [12,13]. Especially, in the early phase, most of the initially tested small compounds drop out due to limited biological impact or severe ‘on- or off-target’ side effects. Hence, optimizing small compound screening during the early preclinical phase of drug discovery has the potential to increase the number of candidate bioactive molecules in cardiovascular drug discovery research pipeline.

Traditionally, initial small compound screens are using cell lines or invertebrate model systems such as Drosophila melanogaster or Nematodes as a high-throughput assay allowing to study the molecular function of plenty of potential druggable targets at the same time. This target-based strategy implies a comprehensive understanding of the pathophysiological background to determine involved key molecules. However, most of the CVDs in humans are highly sophisticated and caused by multiple factors, thereby complicating identification of a single target. Furthermore, cell-based reporter assays or invertebrate-based model organisms can never simulate the orchestrated physiology of cardiac myocytes or even the entire heart not to mention the molecular underpinnings of specific cardiovascular diseases.

Using vertebrate in vivo model organisms in the early stage of drug discovery allows drug screening by analyzing the influence of a chemical agent on a certain phenotype without knowing the causal mechanism. This phenotype-based approach aims to investigate compounds rescuing the disease-associated phenotypes in disease models for human disorders [14–16]. Classical mammalian model organism such as mice, dogs, and pigs are mandatory for preclinical evaluation of candidate therapeutic agents, since they replicate many features of the human cardiovascular system [17]. However, due to their high demand of time and resources, these models are not suitable for high-throughput drug screening in the early preclinical stage of CVD drug discovery as it is possible in lower non-mammalian vertebrate models such as the zebrafish [14,18].

During the last two decades, the zebrafish has emerged as a valuable and efficient vertebrate model organism enabling large-scale in vivo small compound screening during the early phase of drug discovery. It combines the advantages of in vitro screens such as easy handling, cost-effectiveness and high-throughput with the holism of an entire and living organism allowing to easily assess important cardiovascular parameters such as heartbeat, blood flow, ventricular systolic pump function and nevertheless electrophysiological status of the myocardium as early as at the embryonic stage. Zebrafish are small (2-4 cm) and show a high fecundity allowing the weekly production of up to 300 eggs per breeding pair. The externally fertilized eggs develop rapidly, the developing embryos are optically translucent and their development is not completely dependent on the regular function of the cardiovascular system since adequate oxygen supply of all tissues and organs is guaranteed by passive diffusion from the surrounding medium for up to 120 hours post-fertilization (hpf) [19,20]. This fact allows the assessment of the role of the loss of known and novel cardiovascular disease genes in vivo that usually results in early embryonic lethality in other animal model systems such as mice. As in all vertebrate models, the zebrafish heart is the first organ that forms and functions during development. Already by 24 hpf, first cardiac contractions are observable and blood is pumped through the vascular bed. Heart structure is much simpler in zebrafish compared to the human heart, for instance, the zebrafish heart is composed of an inflow tract, sinus venosus, only two cardiac chambers (single atrium and single ventricle) instead of four heart chambers in humans and an outflow tract, bulbus arteriosus. Furthermore, the ventricular myocardial wall in zebrafish is composed of an outer compact cell layer and an inner layer that is much more trabeculated. By contrast, the myocardial wall of the human left ventricle is entirely compact [21]. Nevertheless, despite its simplicity, genes, molecular signaling cascades and cellular processes that orchestrate the development of the heart are grossly conserved between zebrafish and mammals including humans. Additionally, cardiac physiology including heart rate, cardiac contractility, mechano-transduction as well as electrophysiological properties is highly similar between zebrafish and humans [18,22–24]. For example, 3 days post-fertilization, the zebrafish heartbeats 150 times per minute which is much closer to the human situation than the mouse heart that beats appr. Five hundred times per minute. Interestingly, comparing human and zebrafish ECG recordings shape and duration of the QT interval is highly comparable. Drug interferences with the QT interval are one of the most common reasons for premature developmental breakup in drug discovery in general. Hence, industrial standard practice for pre-clinical cardiac safety evaluation is to analyze the impact of novel drug candidates on myocardial repolarization duration, measured by the length of the QT interval in ECG recordings. Due to its pronounced electrophysiological similarities to humans, the zebrafish is an ideal model system to estimate QT prolonging side effects of novel drug candidates already in the early phase of cardiovascular drug discovery [25].

Although most drug screening approaches and functional analyses are conducted in zebrafish embryos, novel zebrafish lines and techniques also allow the in vivo measurement of functional cardiac parameter in the juvenile and adult zebrafish. For instance, the transparent casper zebrafish line lacking melanocytes and iridophores due to mutations in the genes mitfa and mpv17, in combination with transgenic cardiac reporter lines expressing cardiomyocyte-specific fluorophores like GFP or dsRed or Ca²⁺ sensors like GCaMP6f, are used to assess functional and electrophysiological cardiac parameters such as heart rate, fractional shortening or cardiac action potential also in juvenile or adult zebrafish [26,27].

Taken together, as timely assessment of these essential functional parameters of the heart is one of the critical steps in drug discovery [28,29], the zebrafish has the potential to fill in the gap between cellular expression systems or invertebrates and mammalian models facilitating the long run of CVD drug discovery [14,18,30,31].

2.1. Drug absorption and drug metabolism in the zebrafish

Whether findings from drug discovery screens derived from non-human model systems can be translated into clinical application depends among others on drug absorption as well as on pharmacological metabolism profile of the underlying model organism. Whereas intravenous injection is a standard approach for drug testing in mammals, vascular access in zebrafish is a challenging approach [32]. Hence, standard drug absorption in the zebrafish is realized via incubation and in parts via diffusion of the chemical substance into the animal organism limiting the usefulness of the zebrafish as a human drug absorption model. Next to drug absorption, also drug distribution is of essential interest in drug discovery. Especially, permeability of the blood-brain-barrier (BBB) determines whether a cardiovascular drug candidate has a neurological impact even if the brain is not the therapeutic target. Interestingly, BBB in zebrafish has rather low permeability and includes strong tight junctions, drug transporters and protein expression compared to the BBB in higher vertebrates [33].

Also, in xenobiotic and consequently in drug metabolism zebrafish has its similarities compared to higher mammals. Evolutionary, genes coding for metabolism-regulating proteins are highly conserved between vertebrates and mammals. Hence, it is not astonishing that 70% of metabolism genes in humans have an orthologue in the zebrafish genome [34,35]. Especially, the hepatic metabolism with its complex interplay of various cytochrome P450 (CYP) enzymes regulating human biotransformation is very similar to the zebrafish. Actually, essential zebrafish CYP enzymes have direct orthologs in humans including enzymes of the CYP3A family, which is one of the most important regulators of human biotransformation [36,37].

3. Modeling human cardiovascular disorders using the zebrafish

Since several years, the zebrafish is used as an attractive model organism to dissect the genetic and molecular underpinnings of cardiovascular development and (patho-)physiology [20,38,39].

Compared to other small teleost fish models, e.g. the medaka, the genetic and molecular mechanisms contributing to the development and the function of the cardiovascular system in zebrafish is well studied [40,41]. Nevertheless, standard experimental methods and techniques such as microinjections, whole-mount in situ hybridization, immunostainings or CRISPR/Cas9 approaches are applicable in both fish species. Additionally, embryonic development is fast in zebrafish and medaka, both fish species are highly fertile and also housing is comparable. One major difference between zebrafish and medaka embryogenesis is that the medaka embryo develops within a tough chorion protecting the embryo for up to 7 days post fertilization thereby significantly interfering with imaging approaches and experimental manipulations and thereby also hindering easy and efficient high-throughput screening and phenotyping approaches [42]. By contrast, zebrafish embryos hatch 2 days post fertilization from their highly transparent chorion allowing the unrestricted application of imaging and analyses procedures. These facts make the zebrafish the teleost fish model of choice for high-throughput screening and imaging approaches.

Remarkably, about 70% of genes identified in the human reference genome have at least one obvious zebrafish orthologue and CVD associated genes in humans are conserved to a great extent compared to orthologous zebrafish genes [34,43–45]. Hence, it comes as no surprise that a broad range of standard cardiovascular drugs for humans used in daily clinical practice has analogous effects on zebrafish heart and vessels underlining the usage of the zebrafish as an important animal model in cardiovascular research (Table 1) [26,46–54].

Table 1. Pharmacological impact of common human cardiovascular drugs on zebrafish.

drug	Targeted disease in humans	Pharmacological effect in humans	Major finding in zebrafish	Reference
Amiodarone	Heart rhythm disorder (tachycardia)	Inhibition of sodium ion currents	Reduced heart rate and prolonged QT interval in wild-type zebrafish	[52]
Propranolol	Heart rhythm disorder (tachycardia)	β-receptor antagonist	Reduced calcium currents and decreased heart rate in wild-type zebrafish	[26]
Isoproterenol	Heart rhythm disorder (bradycardia)	β-receptor agonist	Increased heart rate and increased ventricular systolic pump function in wild-type zebrafish	[26]
Dobutamine	Heart failure, bradycardia	β-receptor agonist	Increased heart rate in zebrafish mutant line hiphop with severe bradycardia	[51]
Spironolactone	Heart failure	Mineralocorticoid receptor antagonist	Increased systolic pump function in aristolochic acid induced heart failure model	[47]
Empagliflozin	Heart failure	Inhibition of sodium-glucose co-transport	Restoration of structural and functional defects in aristolochic acid induced heart failure model	[48]
Nifedipinie	Arterial hypertension	L-type calcium channel blocker	Reduction of calcium currents in both atrium and ventricle in embryonic zebrafish heart	[26,49]
Warfarin	Ischemic stroke	Vitamin K antagonist	Antagonims of vitamin K dependent gamma-carboxylation of hemostatic proteins leading to severe bleeding in wild-type zebrafish	[53,55]
Rivaroxaban	Ischemic stroke	Factor Xa antagonist	Inhibition of thrombin generation in wild-type zebrafish	[50]
Clopidogrel	Vascular thrombosis	Irreversible inhibition of platelet P2Y12-receptor	Reduced thrombus formation in an arachidonic acid induced thrombosis zebrafish model	[54]

Whereas in the beginnings of zebrafish research, developmental biology studies had priority [40,56–60], more and more, the ability to model essential features of human cardiac disorders moved into the center of attraction [51,61–65]. Since early development of zebrafish embryos does not require proper blood circulation, this model organism is amenable to a broad range of functional genomic approaches including forward genetics by e.g. ENU-mutagenesis [57] and reverse genetics using innovative genome editing tools such as clustered regularly interspaced short palindromic repeats CRISPR/Cas9 and transcription activator-like effector nuclease (TALEN) or Morpholino-oligonucleotide-mediated gene knockdown [18,20,66–69]. Based on zebrafish transparency during embryogenesis functional impact on zebrafish cardiovascular system after genome/gene editing can be easily evaluated via simple light microscopy. Using innovative functional genomic approaches several important key molecules regulating cardiac pump function and rhythm were identified leading to a more in-depth understanding of important human cardiac disorders such as cardiomyopathies, arrhythmias or heart failure [38,51,63,64,70]

Large forward genetic screens led to the identification of numerous zebrafish mutants with heart defects resembling severe human cardiac pathologies such as cardiomyopathies, valvular defects, arrhythmias and heart failure [57]. Genetic analyses by positional cloning of these mutants identified known and novel genes and molecular pathways of cardiovascular disease and demonstrated the validity of the zebrafish to simulate and model human cardiovascular diseases as well as to dissect the molecular pathomechanisms of human CVD (selected zebrafish mutants see Table 2).

Table 2. Selected zebrafish mutants as models for human CVD.

Zebrafish mutant	Mutated zebrafish gene	Phenotype	Human CVD gene	Human disease	Reference
pickwick	titin	Contractile dysfunction	TITIN	DCM, HCM, ARVC	[71,109–111]
silent heart	tnnt2	Contractile dysfunction	TNNT2	DCM, HCM	[72,73]
tell tale heart	myl7	Contractile dysfunction	MYL7	DCM, HCM	[74–76]
lazy susan	cmlc1	Contractile dysfunction	cMLC	HCM	[77,78]
heart strings	tbx5a	Heart development defect	TBX5	Holt-Oram syndrome	[79,80]
breakdance	kcnh6a/kcnh2	Arrhythmia, long QT syndrome	KCNH6	long QT syndrome	[81–83]
reggae	kcnh6a/kcnh2	Arrhythmia, short QT syndrome	KCNH6	short QT syndrome	[62,84]
hiphop	atp1a1a.1	Arrhythmia, bradycardia	ATP1A1	QT interval modulation	[51,118]
island beat	cacna1c	Atrial fibrillation	CACNA1C	Brugada syndrome, long/short QT syndrome	[85–87]
cardiofunk	acta1b	AVC defect	ACTA1	nemaline myopathy, DCM	[88–90]
jekyll	ugdh	AVC defect	UGDH	AVS defect	[91–93]
lost-a-fin	acvr1l	Heart looping defects	ACVRL1	AVS defect	[94,95]
apc	apc	AVC defect, EC hyperplasia	APC	AVS defect	[96,97]
weak atrium	myh6	atrial dysfunction	MYH6	DCM, HCM, atrial septal defect	[98,99]
weiches herz	tbx20	Contractile dysfunction, hypoplasia	TBX20	atrial septal defect, DCM	[70,100–103]
main squeeze/lost contact	ilk	Contractile dysfunction	ILK	DCM	[104,105]
T2EGEZ8	tpm4	Contractile dysfunction	TPM4	HCM	[106–108]

Abbreviations: DCM, Dilated Cardiomyopathy; HCM, Hypertrophic Cardiomyopathy; ARVC, Arrhythmogenic Right Ventricular Cardiomyopathy; AVC, Atrioventricular Canal; AVS, Atrioventricular Valve and Septum; EC, Endocardial Cushions

In 2002, Xu and coworkers found a titin mutation in the zebrafish mutant line pickwick to cause severe contractile deficiency due to the blockade of regular sarcomere assembly [109]. Titin is a giant sarcomeric protein that spans the half-sarcomere from Z-disc to M-line in both, cardiomyocytes and skeletal muscle cells. Recently, titin mutations were found in patients suffering from dilated cardiomyopathy (DCM), hypertrophic cardiomyopathy (HCM) and arrhythmogenic right ventricular cardiomyopathy (ARVC) [110,111], highlighting the relevance of the zebrafish model to study the pathomechanisms of human cardiomyopathies.

Using a reverse genetic approach by morpholino-mediated gene knockdown of the Z-disk protein Nexilin in zebrafish cardiomyocytes, which leads to severely perturbed Z-disk stability and heart failure in zebrafish, the disease-causing nature of Nexilin mutations found in human heart failure patients was elucidated on molecular level [61]. Using CRISPR/Cas9 technology, Zou and coworkers recently elaborated genetic underpinnings of divergent phenotypical severity between N- and C-terminal Titin truncation mutation in zebrafish. As in humans, C-terminal mutations caused severe cardiomyopathy whereas N-terminal mutations resulted in less intensive phenotypes. An internal Titin promotor site was found to compensate N-terminal Titin protein expression explaining variation in phenotypical manifestation of cardiomyopathy [112].

Next to its ability to model human cardiomyopathies, zebrafish mutants also emerged as valuable animal models for the dissection of human cardiac arrhythmia syndromes. Since several main aspects of cardiac electrophysiology in zebrafish such as electrical impulse formation and propagation, myocardial de- and repolarization (Figure 1 A-F) as well as electro-mechanical coupling process are very similar to humans, genetic and molecular underpinnings of several life-threatening human arrhythmia, especially short- and long-QT syndrome, could be elucidated in more detail [14,62,113–115]. In both, short- and long-QT syndrome, the human ether-a-go-go-related-gene (hERG, KCNH6), coding for the rapidly activating delayed rectifier potassium channel, plays a fundamental role by regulating myocardial repolarization. Hence, the two by chemical mutagenesis screen identified zebrafish hERG-mutants reggae (reg) and breakdance (bre) present with prolonged (bre) or shortened (reg) cardiac repolarization reflecting typical electrocardiographic findings from patients with inherited repolarization disorders [116,117].

Figure 1. Essential electrocardiographic parameters including p-wave, QRS complex, and T-wave, between (a) humans and (b) zebrafish are similar. Also, de- and repolarizing ion currents during action potential of human and zebrafish cardiomyocyte share common characteristics. (c) QT interval measurement by electrocardiogram (ECG) in embryonic wild-type zebrafish with normal QT interval in comparison to (D) prolonged QT interval in a zebrafish mutant with prolonged myocardial repolarization. (adapted from [14] and [51] with permission of Bentham Science Publishers and Elsevier, respectively).

Very recently, a candidate gene identified from a human genome-wide association study (GWAS) evaluating the genetic background of myocardial repolarization was mechanistically elucidated on genetic and molecular level by using the zebrafish mutant hiphop (hip), containing a missense mutation in the Na⁺/K⁺-ATPase gene [51,118]. Despite Na⁺/K⁺-ATPase is a well-known transmembrane ion pump crucial for cellular electrophysiology its impact on myocardial repolarization remained largely unclear [119]. However, by using hip zebrafish it was shown that the hip mutation leads to partial loss-of transmembrane Na⁺/K⁺-ATPase ionic pump function, which finally results in prolonged myocardial refractoriness as well as prolonged QT interval indicating an essential role of Na⁺/K⁺-ATPase in regulating myocardial repolarization for the first time.

Even though the zebrafish has striking similarities in cardiovascular development and function, there are important physiological and morphological differences between zebrafish and humans, which must be considered using the zebrafish for cardiovascular disease modeling. Although, the zebrafish is a vertebrate organism the heart consists of only two chambers and the ventricular myocardium in zebrafish is composed of trabeculae surrounded only by a small compact stratum of cardiomyocytes [120]. Hence, hemodynamic properties of the zebrafish myocardium might differ from higher mammals. In addition, zebrafish cardiomyocytes seem to have not-tubules, which in humans are necessary for the intracellular calcium release from the sarcoplasmic reticulum and finally for the electro-mechanical coupling process [121]. Furthermore, the zebrafish exhibit essential differences in cardiac electrophysiology as several electrogenic ion currents are absent or functional properties differ from the myocardial electrophysiology in humans. For instance, the ultra-rapid potassium current (I_Kur) and the transient outward rectifier current (I_to) have not been detected in the zebrafish heart so far [114]. Also, initiation of myocardial depolarization by cellular sodium inflow (I_na) in zebrafish heart is slower, since sodium channel expression in cardiomyocytes of the zebrafish is significantly decreased compared to human cardiomyocytes [122]. Hence, modeling human disorders of myocardial depolarization might be limited in the zebrafish [123].

4. Using automated screening strategies for streamlining drug discovery in zebrafish

The development, establishment, and validation of zebrafish lines that reliably model human cardiovascular disease is the fundamental prerequisite for their use as a phenotype-based screening tool (see also Table 2). During the last decade, several of these established zebrafish cardiovascular disease models were successfully utilized to screen for bioactive and therapeutically relevant substances [14,18]. For instance, already in 2004, Peterson and coworkers used the zebrafish mutant line gridlock that shows phenotypic characteristics similar to aortic coarctation in human patients as a whole-organism and phenotype-based screening tool [124]. In their initial screen and by testing and investigating 5000 different chemical compounds, they identified two structurally related substances, GS4012 and GS3999, both able to rescue the blood circulation defects in gridlock mutant embryos due to the activation of VEGF signaling. Only 2 years later, Hong and colleagues again used gridlock mutants to screen 7000 small chemical compounds to identify a distinct class of substances showing therapeutic effects on the aortic coarctation phenotype. Interestingly, Hong et al. found GS4898, a compound structurally distinct from GS4012 to be capable of restoring blood circulation in the trunk in gridlock zebrafish embryos. In this context, GS4898 seems to act as an inhibitor PI3K signaling thereby suppressing the gridlock phenotype [125].

Peal and coworkers exposed breakdance mutant zebrafish embryos that exhibit prolonged ventricular action potential duration and spontaneous early after-depolarization similar to patients suffering from long-QT syndrome to 1200 small chemical compounds. Here, they found that the loss of function of KCNH2 (HERG, human ether-a-go-go potassium channel) leading to the long-QT syndrome phenotype in breakdance embryos was effectively rescued by flurandrenolide and 2-methoxy-N-(4-methylphenyl) benzamide (2-MMB) due to a beneficial shortening of action potential duration [126].

All three elegant studies nicely demonstrated the utility and feasibility of phenotype-based small compound screens using established zebrafish cardiovascular disease models. Nevertheless, the screens mentioned were at most medium-throughput and not automated at all, hindering fast and effective screening on a high-throughput scale. In these studies, zebrafish embryos were placed manually on 96-well plates, compounds were manually added to the medium of each well, and each embryo was visually scored after compound incubation for the presence or absence of the respective phenotypic characteristics. Although finally successful, this ‘manual approach’ is extremely laborious and therefore unsuitable for the integration into straightforward modern drug discovery pipelines. More recently, high-throughput screening platforms were engineered allowing embryo dispensation into 96-well plates, delivery of the chemical compound, incubation and subsequent acquisition of pictures or videos for the analysis of cardiac parameters in an automated fashion [127–129]. In this context, by using transgenic reporter zebrafish that express eGFP under control of the cardiac Bmp4 promoter (Tg(bmp4::eGFP), Lin et al. developed an automated screening platform enabling the measurement of functional cardiac parameters such as heart rate, ventricular stroke volume, ejection fraction, diastolic filling and ventricular mass in zebrafish embryos [129]. By contrast, Pylatiuk and coworkers developed a screening platform that enables the automated detection of the beating zebrafish heart in each well and the prompt and automated acquisition of images and videos in zebrafish embryos without fluorescently labeled cardiac structures [130,131]. By using this automated small compound screening platform, we recently identified two small chemical compounds, calyculin A and okadaic acid, both known phosphatase inhibitors, to be able to efficiently restore cardiac contractile force (ventricular fractional shortening) in Integrin-linked kinase (ILK)-deficient main squeeze heart failure zebrafish embryos [16]. Both compounds were found to rescue myocardial function by significantly reconstituting protein kinase B (PKB/Akt) phosphorylation in main squeeze hearts.

Although significant developments and improvements regarding the automation of whole-organism screening pipelines ranging from robotics-assisted, microscopic screening platforms to online and real-time parameter acquisition and analysis were made during the recent past [130–136], it will still take time and efforts to be able to use the zebrafish in fully automated and high-throughput preclinical drug discovery pipelines. Nevertheless, novel approaches including artificial intelligence (AI) and neural network advancements are promising strategies to fill the current gaps thereby enabling streamlined cardiovascular drug discovery [137].

5. Expert opinion

As already mentioned, cardiovascular diseases are still the predominant cause of morbidity and mortality in all western world countries. Despite their enormous clinical significance, the identification of novel effective drugs to treat cardiovascular diseases is still lagging behind. One major hurdle in modern cardiovascular drug discovery is the high complexity of pathobiology underlying the disease. At present, the discovery of novel cardiovascular drugs mainly focuses on a number of well-known molecular targets instead of significantly investing into the scientific definition of the etiologic fundamentals and thereby promising new drug targets of each disease, which would finally enable tailoring, individualization, and personalization of therapy. In this context, dissection of the complex and highly orchestrated and interconnected mechanisms of human cardiovascular disease by just concentrating on single genes, molecules or pathways are highly ineffective. In the face of this biological disease complexity novel approaches such as systems biology or systems medicine has to be integrated into research pipelines to be able to model complexity and to define promising novel molecular targets [17]. Systems medicine, as an interdisciplinary area of research involving clinicians, biologists, bioinformaticians and mathematicians, aims on the integration of multifaceted and often heterogeneous ‘omics data’ (e.g. genomics, transcriptomics, proteomics, metabolomics) to be able to manage the high complexity of the data sets and thereby the underlying disease in a quantitative, modeled fashion. In the cardiovascular field, we are just at the beginning of the implementation of systems medicine approaches into our research routines. In the future, these systematic approaches will significantly foster or even enable the definition of novel and highly specific molecular druggable targets. Nevertheless, the definition of novel promising drug targets of the disease is just the initial step of a streamlined drug discovery pipeline. Fundamental for the efficient and straight-forward identification of novel bioactive therapeutic drugs are (1) the generation of animal disease models that accurately simulate pathology and pathobiology of the disease allowing the reliable translation of experimental findings into the human system and (2) the utilization of these disease models in highly efficient, fully automated high-throughput small compound screening platforms using defined and highly specific disease characteristics as endpoint readouts (e.g. heart rate/rhythm, systolic/diastolic volume, ejection fraction, stroke volumes, and fractional shortening) to estimate drug performance.

Over the past 25 years, the zebrafish evolved as an excellent non-mammalian vertebrate animal model to define the genetic and molecular etiology of cardiovascular diseases [20,38,39]. Furthermore, the unique arsenal of forward and reverse genetic approaches that includes state-of-the-art genome editing tools such as the CRISPR/Cas9 technology in combination with the substantial homology of physiology and pathophysiology to the human cardiovascular system turns the zebrafish into a valuable and reliable model organism to simulate human cardiovascular diseases, even in an individualized or personalized manner. Excitingly, the large numbers of progeny, its fast and extrauterine development and its small size and transparency make the zebrafish also an ideal in vivo model for high-throughput small compound screening. Although significant progress in the automation of the screening process was made in recent years, further optimization of automation of the entire screening pipeline, but in particular in the context of the standardization and assessment automation of the measured morphological and functional cardiac parameters is fundamental. Whereas the development of microscopy-based, automated screening platforms refine consecutively, automation of image and video analysis and cardiac parameter measurement is still significantly lagging behind, currently hindering streamlined high-throughput screening in the zebrafish model. Only recently, artificial intelligence (AI), particularly machine learning approaches were developed and utilized to overcome limitations of the conventional assessment and quantification of morphological and functional readouts. Nevertheless, the applicability of these novel and promising approaches needs to be proved in future studies.

In conclusion, recent developments in the systematic definition of the molecular etiology of disease, in the establishment and validation of the zebrafish as an experimental system to model human cardiovascular diseases and the progress of engineering hardware and software solutions for the automation of high-throughput screening in the zebrafish model is raising the hope or even the expectation that current limitations within the pipeline of modern cardiac drug discovery will be resolved in the near future. We are convinced that zebrafish as a state-of-the-art preclinical non-mammalian vertebrate cardiovascular disease model will significantly contribute to the more efficient and reliable drug discovery, less early and late phase drop-out rates and higher success rates of therapeutically active drugs reaching the market.

Article highlights

• Innovative functional genomics strategies such as genome editing by CRISPR/Cas9 have significantly facilitated elaboration of the underlying pathomechanism in various cardiovascular diseases.

• The zebrafish is amenable for a broad range of forward and reverse functional genomic tools allowing to model various cardiovascular diseases.

• Due to its easy handling, short reproduction cycle and uncomplex phenotypical assessment the zebrafish has emerged as a suitable tool for high-throughput in vivo drug screening.

• The zebrafish combines the advantages of in vitro screens such as easy handling, cost-effectiveness and high-throughput with the holism of an entire and living organism

• Computer-/robotics-assisted automated high-throughput small compound screening on zebrafish disease models will further foster the identification of disease-modifying therapeutics.

This box summarizes key points contained in the article.

Declaration of interest

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer Disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

A Pott is supported by the Clinician-Scientist-Program 2018 of the Medical Faculty University Ulm. W Rottbauer is supported by the Deutsche Forschungsgemeinschaft (DFG) (RO2173/3-2). S Just is a member of the German Federal Ministry of Education and Research (BMBF)-funded e:Med Programme on Systems Medicine and is supported by the Symbol-HF (#01ZX1407A), coNfirm (#01ZX1708C), Cardio-HTS (#01ZX1907A) grants. S Just is also supported by the Deutsche Forschungsgemeinschaft (DFG) (JU2859/2-1 and JU2859/7-1).