Drug Discovery for Myocardial Diseases and Cardiac Arrhythmias Using Zebrafish

Challenges in Cardiovascular Drug Discovery

Despite the significant decline in cardiovascular disease mortality over the past several decades due to more effective medications and surgical procedures, cardiovascular diseases remain one of the leading causes of death and a major economic burden to the health care system globally. ^, Surprisingly, investment in cardiovascular drug development, particularly for myocardial diseases and arrhythmias, has stagnated, and approvals for new cardiovascular drug therapies have also declined substantially. Although multiple cardiovascular drug targets and novel molecules with potential cardiovascular applications have been identified through the application of genomic technologies and systems biology approaches, these scientific advances have not stimulated an increase in cardiovascular drug development. For example, fewer drug candidates are found in the cardiovascular research pipeline compared to other therapeutic areas, especially antineoplastic agents. ^,

Although there may be many reasons for this downward trend in cardiovascular drug discovery, two reasons are particularly important. The first and fundamental reason is that despite major scientific advances, our understanding of complex cardiovascular pathophysiology is still limited. This is in contrast to oncology where, as a result of clonality, there is often a more direct relationship between genomic features and short-term cellular proliferation or tumor growth, which are also key end points in clinical trials. In cardiovascular diseases, even when the primary molecular cause is known, the mechanistic links with the end phenotypes are extremely complex, with developmental, genetic, and environmental modifiers at play on a timeline of decades. For example, atrial fibrillation (AF) has been demonstrated to be, at least in part, an aggregate of inherited susceptibilities, environmental and acquired modifiers, and short-term triggers. ^, To date, genome-wide association studies have identified over 100 genetic loci associated with AF, yet these are able to explain only a modest proportion of the heritability of the disease. ^, The majority of the drugs developed to treat arrhythmias such as AF have been identified in assays of cellular excitability, or in screens for inhibitors of specific membrane currents implicated in excitability. In addition, the traditional preclinical animal models for AF remain dependent on sustained rapid atrial pacing, which typically models the biology downstream of AF onset rather than the primary substrate. As a result, the identified drug leads have exhibited only modest therapeutic indices, limited efficacy, or both in subsequent preclinical and clinical testing. There is also evidence that the pathophysiology underlying many forms of human AF may include significant extracardiac contributions such as obesity and subclinical thyroid abnormalities. Therefore, it remains a daunting challenge to discover a single (or even a few) drug target(s) that can address disease biology of such complexity. A second major reason for the downward trend in cardiovascular drug discovery is that the end points in cardiovascular diseases are necessarily stringent, given the natural history of the disorders themselves and the insights from prior clinical trials. Simple short-term increases in contractility or suppression of arrhythmias are not sufficient to define success, especially given the risks of increased mortality seen in prior studies such as the CAST trial. The substantial thresholds for safety as well as thresholds for robust and sustained clinical efficacy also make the cardiovascular drug discovery process more costly, risky, and time consuming.

Drug discovery over the past few decades has largely focused on target-based strategies. Target-based approaches are designed to look for small molecules with biological activity on a specific target with therapeutic potential, which is usually a single protein or a specific molecular mechanism that has been identified via human genetic studies or basic biological research. Once a robust link is identified between a specific human molecular pathway and a disease signature, molecules that target this gene product or pathway become lead candidates as potential therapeutic drugs. While this approach is robust in theory, it is constrained by the rigor with which the target has been mechanistically implicated in the disease and the extent to which the disease can be modeled in simple reductionist systems, since screening is usually executed in tightly controlled in vitro assays (e.g., biochemical outputs or simple cell culture systems). The advantages of this approach are that the initial screens are rapid and cost-effective, evaluating many hundreds of thousands of small molecules in parallel, a task that cannot be accomplished in traditional in vivo preclinical models. ^, However, many biological processes cannot be faithfully reproduced in target-focused biochemical assays or even in cultured cells. Multiple interactions between different cells, tissues, organs, and systemic factors are major contributors to hierarchical disease pathogenesis and/or drug responses and cannot be replicated in vitro, although advances in three-dimensional culture technologies promise to overcome some of these challenges.

Despite the thousands of potential therapeutic targets identified since the human genome was decoded, it remains quite difficult to predict which proteins, when later modulated in vivo in the whole organism, will reverse a disease phenotype or alter a poorly understood pathologic process, especially for complex chronic diseases. ^, In cardiac arrhythmias, for example, the fundamental disease mechanisms are still difficult to elucidate in spite of advances in genetics, genomics, and systems biology. Furthermore, the pharmacokinetic behavior of bioactive molecules, as described by parameters quantitating compound absorption, distribution, metabolism, and excretion, can only be determined at the whole organism level. In sum, in vitro target-based screens are limited in that they cannot model the complexity of disease pathophysiology or mimic the pharmacokinetic behaviors of bioactive molecules in a whole organism. These constraints call for alternative approaches to drug development for complex, life-long disorders.

Myocardial Disease and Arrhythmia Mechanisms

The concept of myocardial disease or cardiomyopathy dates back to 1899, when Fiedler described a series of fatal cases in young adults with cardiomegaly and heart failure. The term cardiomyopathy (noncoronary) was first used in 1950s by Brigden to describe patients with idiopathic myocardial disease, several of whom had familial disease. In 1980, a task force of the World Health Organization, chaired by John Goodwin, defined cardiomyopathies as heart muscle diseases of unknown cause and classified them into dilated (DCM), hypertrophic (HCM), and restrictive (RCM) cardiomyopathies based on the predominant structural and hemodynamic phenotype. After the discovery of the genetic causes of a number of the cardiomyopathies, the World Health Organization redefined cardiomyopathies as diseases of the myocardium associated with cardiac dysfunction, dropping the “unknown,” and also added arrhythmogenic right ventricular cardiomyopathy, which has since been renamed as arrhythmogenic cardiomyopathy (ACM). With the rapid advancement in our understanding of the genetic basis of many of these myocardial diseases, the American Heart Association presented an updated and “living” classification in 2006, dividing cardiomyopathies into (1) primary, when the disease is solely or predominantly of the heart muscle, and (2) secondary, when myocardial involvement is associated with a multisystem disorder (e.g., Anderson-Fabry disease, sarcoidosis, and amyloidosis).

Primary cardiomyopathies were classified as genetic (e.g., HCM, ACM), acquired (e.g., myocarditis, takotsubo, tachycardia induced), or mixed (e.g., both genetic and acquired DCM). To the surprise of some, ion channel diseases with grossly normal hearts (i.e., channelopathies) were also added under genetic primary cardiomyopathies because channelopathies and related electrophysiologic disturbances should also be considered as disorders of the cardiomyocyte. Therefore, channelopathies and related disorders, such as long QT and short QT syndrome, Brugada syndrome, and catecholaminergic polymorphic ventricular tachycardia, are now classified under primary genetic cardiomyopathies because they are diseases of the cardiomyocytes characterized by electrophysiologic dysfunction that is arrhythmogenic. Indeed, with the help of improved molecular diagnosis and phenotyping, we are only now beginning to understand the molecular pathogenesis of disease that is operating long before histologic, morphologic, and functional manifestations, which are traditionally used to diagnose and classify overt myocardial diseases. This chapter considers both myocardial diseases and cardiac arrhythmias as a spectrum of cardiovascular diseases whose molecular pathogenesis we are beginning to understand and use as a basis to develop disease-modifying treatments with the help of a powerful tool, the zebrafish.

Phenotype-Based Screens

In contrast to in vitro target-based screens, a phenotype-based screen is designed to study biologically active small molecules based on downstream functional whole organism or cellular end points rather than the activity of the target molecule itself. Before in vitro approaches were made possible by advances in molecular biology, many biologically active molecules were serendipitously or empirically discovered from effects they had on the phenotypes of whole organisms. For examples, the anticoagulant dicoumarol and its derivatives were first discovered when cattle that fed on rotting sweet clover died of internal bleeding, ^, and the Hedgehog signaling antagonist cyclopamine was discovered when a fetal deformity, cyclopia, was observed in offspring from sheep grazing on veratrum californicum. Attempts were also made to accelerate discovery of bioactive small molecules by prospective, systematic chemical screening in animals, albeit in costly screens with only modest numbers of rodents.

Despite the promise of phenotype-based approaches demonstrated by the examples above, cost and lack of scalability led to the dominance of target-based approaches during the past few decades. In recent years, there has been a resurgence of interest in phenotype-based screening because of the growing number of early discovery successes from such efforts, including the use of fruit flies ( Drosophila melanogaster ), worms ( Caenorhabditis elegans ), and zebrafish ( Danio rerio ) disease models. An analysis of first-in-class drugs that were approved by the US Food and Drug Administration (FDA) between 1999 and 2008 revealed that 62% were discovered by phenotype-based screens, despite the fact that such screens represented only a small subset of drug discovery efforts. The advent of machine learning–based end points for cellular and whole organism phenotypes is driving further investment in these approaches to drug discovery.

Several advantages of phenotype-based approaches may contribute to their success. First, phenotype-based drug discovery does not require prior knowledge of the underlying biological processes because it is based on modulation of a downstream phenotypic readout. As such, a phenotype-based strategy can be used to identify chemical modifiers of virtually any biological process, while target-based approaches typically can only discover modifiers of a specific target and usually only in a cell-autonomous context. This also means that phenotype-based approaches also provide an opportunity to reveal novel targets and their functions and to obtain fundamental insights into poorly understood biological processes. This is in contrast to target-based approaches in which a robust prior understanding of the relevant biology is a prerequisite for success. Second, phenotype-based discovery can identify chemical modifiers that produce a therapeutic effect through activity at multiple targets simultaneously. For instance, the serendipitously discovered antiarrhythmic agent amiodarone exhibits activity on multiple molecular targets, including ion channels, adrenergic receptors, and possibly the nuclear thyroid receptor. On the other hand, many molecules that were designed to exclusively target the conductance of individual ion channels as antiarrhythmic agents have proven to be unsuccessful. One possible explanation for this discrepancy is our incomplete understanding of the roles of the various types of ion channels and their regulatory networks. Finally, phenotype-based approaches can identify chemical modifiers in the context of a whole organism, which allows for simultaneous evaluation of their pharmacokinetic/pharmacodynamic profile and rejection of those with undesirable qualities, including obvious toxicities. Consequently, hit compounds coming from phenotype-based screens have a higher probability of passing further tests in other models for effectiveness, toxicity, and pharmacokinetic profile compared to compounds identified in target-based screens.

Given the emerging interest in phenotype-based screens, efforts are now turning to finding the most appropriate model organism for cardiovascular drug discovery. In vertebrate models, the effects of individual human drugs are typically representative of human complexity, including most known drug-drug interactions. In addition, there is increasing evidence that drug distribution across active physiologic boundaries such as the blood-brain barrier can also be faithfully observed in animal models, including zebrafish. The following sections outline how zebrafish may be one of the most suitable model organisms for early cardiovascular drug discovery, particularly in the areas of myocardial diseases and cardiac arrhythmias. To date, few drugs have made it all the way to regulatory approval from in vivo screening in zebrafish, but several examples are in early clinical trials. The combination of new technological advances and the tractability of the zebrafish model is bringing new enthusiasm and investment and paving the way for more efficient cardiovascular drug discovery.

Using Zebrafish for Cardiovascular Drug Discovery

The zebrafish ( Danio rerio ) was initially used as a model organism for the study of developmental biology but increasingly became an organism of choice for the scalable study of human disease mechanisms, including cardiovascular diseases and cancers. ^, In recent years, the zebrafish has also emerged as a powerful tool for phenotype-driven genetic or chemical screens because it is more representative of human physiology and pathophysiology than other model organisms such as flies, worms, and yeast.