Excitation and Excitation-Contraction Coupling of the Zebrafish Heart: Implications for the Zebrafish Model in Drug Screening Excitation and Excitation-Contraction Coupling of the Zebrafish Heart: Implications for the Zebrafish Model in Drug Screening

There are several similarities and differences in electrical excitability between zebrafish and human ventricles. Major ion currents generating ventricular action potentials are largely the same in human and zebrafish hearts with some exceptions. A large T-type calcium current is unique to the zebrafish ventricle (absent in human ventricle), and two potassium currents (I Ks and I to ) may be absent in zebrafish ventricular myocytes. However, there are substantial differences among alpha subunit isoforms of the ion channel families or subfamilies (e.g. zebrafish Kv11.2 vs. human Kv11.1; zebrafish Kir2.4 vs. human Kir2.1) between human and zebrafish hearts. Contraction of zebrafish ventricle is strongly dependent on extracellular calcium, while human ventricle relies heavily on cal- cium stores of the sarcoplasmic reticulum. These differences may affect the use of zebrafish as a model in drug screening and safety pharmacology.


Introduction
Zebrafish (Danio rerio), a tropical freshwater fish species, is a popular vertebrate model and widely used to resolve diverse research questions in developmental biology and genetics, human diseases, environmental toxicology and several other disciplines. The advantages of the zebrafish model are research technical (e.g. well-annotated and easily modifiable genome, transparency of embryos), economical (cost and ease of maintenance, large number of offspring and short generation time) and ethical (replacement of mammalian models-3R principle) [1,2]. Those qualifications have made zebrafish an interesting object in studies, where new molecules are searched and selected for drug development programs [3][4][5]. Potentially zebrafish could be a high throughput and relatively inexpensive in vivo model for screening therapeutically effective and nontoxic candidate molecules for drug development programs. Indeed, great expectations are set on the zebrafish model, which is sometimes regarded as an ideal system for preclinical screening of cardiovascular drugs [6]. The expectations are based on the conserved properties of cardiac physiology between humans and zebrafish, such as the similarities in the shape of ventricular action potential (AP) and heart rate [7][8][9]. The documented responses of zebrafish hearts (e.g. bradycardia, atrioventricular block, prolongation QT interval of electrocardiogram) to the inhibitors of human ether-à-go-go-related (KCNH2) channel provide some credence to those expectations, even though sensitivity and specificity of the responses are not optimal [9][10][11][12][13]. In the screening of cardiovascular drugs, the in vivo zebrafish model has the advantage that all cardiac ion channels are simultaneously exposed to the compound, thereby allowing phenotype-based screening. However, in order to provide an accurate mode for the human heart, molecular composition, voltage-dependence and gating kinetics of ion channels of the zebrafish heart should closely match those of the human heart. Unfortunately, the ionic and molecular bases of electrical excitability of the zebrafish heart are still unsatisfactorily known. This is a significant shortcoming, since the requirements set for effective and safe drugs are extremely rigorous, and safety evaluation necessitates exact knowledge about the mode of drug action [14]. Those requirements are delineated in Comprehensive in vitro Proarrhythmia Assay (CiPA) initiative for cardiac safety evaluation of new drugs, which provides an accurate mechanistic-based assessment of proarrhythmic potential [14][15][16]. Rational evaluation of drug toxicity in the zebrafish model is not possible before ion currents and channels of the zebrafish cardiac myocytes are known in sufficient detail. The present overview compares ion current and ion channel compositions of zebrafish and human ventricles in order to indicate similarities and differences between the fish model and the human heart, and gaps in our knowledge of the zebrafish cardiac excitability and excitation-contraction (e-c) coupling. These issues have also been discussed in other recent reviews [17][18][19].

Cardiac action potential
Contraction of cardiac myocytes is triggered by electrical excitation of myocyte sarcolemma in the form of cardiac AP. Propagation of AP through the heart can be recorded as electrocardiogram ( Figure 1A). Each functionally different cardiac tissue has a characteristic AP shape generated by the tissue-specific ion currents and ion channel compositions. The five different phases of the mammalian ventricular AP-with the exception of phase 1 fast repolarization-are readily discernible in the zebrafish ventricular AP (Figure 1B). Similar to the human ventricular AP (but unlike the murine AP), the zebrafish ventricular AP has a distinct plateau phase (phase 2) at positive voltages [8,9,20] (Figure 1B). Indeed, the only major difference between zebrafish and human ventricular APs is the absence of the fast phase 1 repolarization in the zebrafish AP [19]. This may be due to the absence of the transient outward K + current (I to ) in zebrafish ventricular myocytes [8].
Zebrafish are ectothermic vertebrates and therefore their AP characteristics may change depending on the rearing temperature of the fish, as has been reported for several teleost fish species [21,22]. In the adult zebrafish, reared at 28°C, the duration of ventricular AP (APD 50 ; at 28°C) is about 30% shorter than that of the human ventricular AP at 37°C (Table 1). At 36°C, the duration of zebrafish ventricular AP is only 80 ms, i.e. about 25% of the duration of human ventricular AP at 37°C. The shorter AP of the zebrafish heart may be associated with and ventricular action potential (B) of the zebrafish heart at 26°C. Electrocardiogram was recorded from surface of spontaneously beating heart in vitro. Ventricular action potential was recorded from an enzymatically isolated cardiac myocyte with patch-clamp technique. The main ion currents responsible for different phases of ventricular action potential are also shown. I Na , Na + current; I CaL , L-type Ca 2+ current; I CaT , T-type Ca 2+ current; I Kr , the fast component of delayed rectifier K + current; I Ks , the slow component of the delayed rectifier K + current; I K1 , the inward rectifier K + current. the higher heart rate of the fish, which at 28°C is about double (120-130 beats per minute) and at 37°C about quadruple (287 beats per minute) the human resting heart rate [19,23] (Table 1). Temperature is an important environmental factor for an ectothermic vertebrate, which modifies cardiac gene expression and ion channel function [24][25][26]. Therefore, rearing and experimental temperatures should be carefully controlled and reported in zebrafish studies.

Ion currents and ion channels
Density and kinetics of ion currents must be such that chamber-specific APs are generated and can be adjusted to heart rates according to the circulatory demands. This is reflected in the composition of ion channel assemblies and their abundances in different cardiac chambers and in a species-specific manner [21]. This overview is limited to ventricular myocytes, since atrial ion currents/channels of the zebrafish heart are still relatively poorly known.

Sodium currents and channels
Sodium influx through the voltage-gated Na + channels initiates the all-or-none action potential (AP) of atrial and ventricular myocytes, when the current flow from the upstream cell depolarizes the membrane of the downstream cell to the threshold level (about −55 mV). At the threshold voltage, the density of inward Na + current (I Na ) exceeds the total density of the outward K + currents (I K ). The rapid opening of Na + channels generates a fast upstroke (depolarization) of AP and an overshoot to the level of about +40 mV (phase 0) [19,20] (Figure 1B). Then I Na quickly inactivates due to the closing of the inactivation gate of the channel.

Reference Human Reference
Relative heart size (% of body mass) 0.1 [48] 0.64 [50] Diastolic/systolic blood pressure (mm Hg) 0.42/2.51 [70] 70/125 [71] Myocyte size (ventricle) (pF) 26-33 [8,39] 117-227 [72,73] T-tubuli No [20,48] Yes [67,74] Role of CICR 1 in e-c coupling (%) 15 [ The density of I Na is the main determinant for the rate of AP propagation over the heart. The rate of AP upstroke in zebrafish ventricular myocytes at 28°C is about 130 V s −1 (RMP ~ −84 mV) (Haverinen et al., submitted), which is less than half of the rate of AP upstroke in human ventricular myocytes at 37°C [27]. These findings suggest that the density of ventricular I Na is lower and the rate of AP propagation slower than in human ventricles at the speciesspecific temperatures (28°C vs. 37°C). However, a thorough analysis of the zebrafish I Na is needed to reveal to what extent these differences are due to RMP (availability of Na + channel for opening), Na + channel density and kinetic properties of the cardiac Na + channels.
The zebrafish heart expresses eight different Na + channel alpha subunits. The main isoforms are Na v 1.5Lb (83.1% of the transcripts) and Na v 1.4b (16.2%), which are orthologues to the human cardiac Na v 1.5 (71.1% in the right ventricle) and skeletal Na v 1.4 channels, respectively [28] ( Table 2). Na v 2.1 is abundantly expressed in the human right ventricle (27.8%), but seems to be absent in zebrafish ventricular myocytes. Unlike the mammalian cardiac I Na , which is tetrodotoxin-resistant (IC 50 about 1 μM), the zebrafish I Na is more than two orders of magnitude more sensitive to tetrodotoxin (IC 50 about 6 nM) (Haverinen et al., submitted), similar to the I Na of other fish species [29,30]. Thus, there is a remarkable difference in tetrodotoxin-sensitivity and some minor differences in Na + channel composition between zebrafish and human hearts.

Calcium currents and channels
The vertebrate heart usually has two main types of Ca 2+ currents, a high-threshold or longlasting L-type current (I CaL ) and a low-threshold or transient T-type (I CaT ) current. The former is activated at voltages more positive than −40 mV and with the peak amplitude at about +10 mV, while the latter is generated already at −60 mV and with the peak current amplitude at about −30 mV [31,32].
I CaL is the main I Ca of atrial and ventricular myocytes. It has a significant physiological function in maintaining the long plateau (phase 2) of the cardiac AP and mediating Ca 2+ influx into the myocyte ( Figure 1B). I CaL contributes to the Ca 2+ transient, which sets cardiac contraction in motion, either directly by increasing cytosolic Ca 2+ concentration or triggering a further release of Ca 2+ from the sarcoplasmic reticulum (SR) (for more details see excitationcontraction coupling). The mean density of I CaL in ventricular myocytes of the zebrafish heart at 28°C is 6-8 pA pF −1 , which is about double the density of the human ventricular I CaL at 35°C (3-4 pA pF −1 ) [8,20,33,34]. This difference may signify a larger role of sarcolemmal Ca 2+ influx in e-c coupling of the zebrafish heart [33]. In the human ventricle, the main L-type Ca 2+ channel isoform is Ca v 1.2, which represents 98.4% of the total Ca 2+ channel transcripts in the right ventricle [35] ( Table 2). In the zebrafish ventricle, seven different L-type Ca 2+ alpha subunits are expressed, among them three paralogue pairs [36]. Similar to the human ventricle, Ca v 1.2 is the most abundant L-type Ca 2+ channel isoform in zebrafish ventricle consisting of 38.3% of the all Ca 2+ channels transcripts.
In vertebrate hearts I CaT is a sizeable current in nodal myocytes and it may be also present in atrial myocytes, but it is usually absent in ventricular myocytes. In this respect, the zebrafish is clearly different. A characteristic feature for zebrafish ventricular myocytes is a large I CaT with a   current density almost equal to that of I CaL [8,37]. T-type Ca 2+ channels pass Ca 2+ influx at more negative voltages than L-type Ca 2+ channels. Therefore, they may contribute to upstroke and early plateau of the ventricular AP. Although I CaT inactivates faster than I CaL , it allows a significant sarcolemmal Ca 2+ entry into zebrafish ventricular myocytes. During a 300 ms depolarizing pulse to −30 mV the Ca 2+ influx through T-type Ca 2+ channels is about 35% of the Ca 2+ influx of L-type Ca 2+ channels during 300 ms pulse to +10 mV [36]. Therefore, I CaT may have a significant role in e-c coupling of zebrafish ventricular myocytes. T-type Ca 2+ channels are abundantly expressed in the zebrafish ventricle constituting majority of the transcripts (about 55%) of the total Ca 2+ channel population (  Table 2). Taken together the prominent expression of I CaT in zebrafish ventricular myocytes is one of the most striking differences in ion channel composition between zebrafish and human hearts. The exact role of I CaT in excitation and e-c coupling of zebrafish ventricular myocytes needs to be examined in detail.
Overall, the diversity of Ca 2+ channels in zebrafish ventricle is larger than in the human ventricle. Most notably, T-type Ca 2+ channels are more abundant than L-type Ca 2+ channels.

Potassium currents and channels
Outward potassium currents (I K ) are repolarizing, i.e. they maintain negative resting membrane potential (RMP) and limit the duration of cardiac AP.

Inward rectifier K + currents and channels
The inward rectifier K + current (I K1 ) maintains RMP of atrial and ventricular myocytes and provides K + efflux for the final phase 3 repolarization of AP [38] (Figure 1B). This current is generated by the Kir2 subfamily channels in vertebrate hearts. Characteristic for the inward rectifier K + channels of the Kir2 family is that they pass outward I K1 in the voltage range from about −80 to 0 mV with the peak current at −59 mV in zebrafish ventricular myocytes [39]. At RMP, the net K + flux at the RMP is almost zero, but the outward I K1 activates instantaneously on depolarization of sarcolemma and generates a fast outward surge of I K1 at the rising phase of AP. At the plateau voltage, K + efflux is small due to the voltage-dependent block of the channel pore by intracellular polyamines and Mg 2+ ions. When membrane potential starts to repolarize (due to the activation of I Kr and I Ks and inactivation of I Ca ), the polyamine block of Kir2 channels relaxes and K + efflux through Kir2 channel accelerates repolarization at phase 3. Different members of the Kir2 family differ in the ease with which they allow K + efflux through the sarcolemma. Large difference in Ba 2+ -sensitivity between Kir2 isoforms suggests that their interaction with potential medicinal drug molecules might also differ [39].
The main Kir2 channel isoform of the zebrafish ventricle is Kir2.4, as in the most teleost fishes studied thus far [39]. In the zebrafish atrium Kir2.2a channels are abundantly expressed in addition to Kir2.4 channels. There are striking differences in isoform composition between zebrafish and human ventricles. The main isoform in the human ventricles is Kir2.1 (46.5% of the transcripts in the right ventricle), which appears only in trace amounts (0.8%) in the zebrafish ventricle. Kir2.3 is abundantly expressed in human ventricle (24.5%), but almost totally absent in the zebrafish ventricle ( Table 2). Since the Kir2 isoforms differ in their rectification properties those differences in isoform composition are likely to have functional consequences for repolarization of AP.

Voltage-gated K + currents and channels
Voltage-gated K + currents provide sarcolemmal K + efflux for repolarization of AP. Several different voltage-gated K + currents are expressed in human ventricle including fast and slow components of the delayed rectifier current (I Kr and I Ks , respectively) and transient outward current (I to ) [27,40,41]. Transcripts for the ultra-rapid component of the delayed rectifier channel (K v 1.5) are expressed in the human ventricle, but the current seems to be specific for atrial myocytes and not expressed in the ventricles [42]. Much less is known about the voltage-gated K + currents of the zebrafish heart. Similar to the human ventricle, I Kr seems to be the main repolarizing current also in zebrafish ventricle [8,37,43]. Until now, I Ks has not been recorded in zebrafish cardiac myocytes, even though transcripts of the K v 7.1 (KCNQ1) channel are expressed in the zebrafish ventricle (Hassinen et al., unpublished) ( Table 2). Neither has been I to found in zebrafish cardiac myocytes [8]. These findings suggest that repolarization of zebrafish ventricular myocytes might be more strongly dependent on single voltage-gated K + current, I Kr , than human ventricular myocytes. However, a closer examination of the K + ion composition of zebrafish ventricular myocytes is needed to verify/falsify these assumptions.
I Kr is the dominant repolarizing current in human and zebrafish ventricular myocytes.
Notably the current is generated by different isoforms in human and zebrafish heart. In the human cardiac myocytes, I Kr flows through the erg1 (Herg or KCNH2 or K v 11.1) channels ( Table 2). In zebrafish myocytes I Kr is generated almost exclusively by KCNH6 (K v 11.2a,b or zebrafish erg) channels [19,44], although it is often referred to as an orthologue to human KCNH2 channels [9,11,45]. Indeed, the expression of K v 11.1a,b transcripts in the zebrafish ventricle is only 0.2%, while K v 11.2a,b channel transcripts constitute 84.9% of the total voltage-gated K + channel alpha subunit population. Transcripts of the K v 7.1 channels form about 15% of the total transcripts of the voltage-gated K + channels in both human and zebrafish ventricles ( Table 2). K v 1.5 channels are expressed in the human ventricle, but no reports exist about any zebrafish orthologues.

Excitation-contraction coupling
There are prominent differences in size, shape and fine structure between zebrafish and human ventricular myocytes ( Table 1). Zebrafish ventricular myocytes are 5-10 times smaller (26.1-33.3 pF) than human ventricular myocytes (117-227 pF). Zebrafish ventricular myocytes are almost as long as human ventricular myocytes, but are much thinner [20,46]. In human ventricular myocytes, myofibrils are evenly distributed throughout the myocyte, while in zebrafish ventricular myocytes myofibrils locate immediately under the sarcolemma [47].
Due to the small diameter of ventricular myocytes and cortical location of myofibrils T-tubuli are unnecessary for cellular signaling and probably therefore completely absent in zebrafish myocytes [20,48].
The marked differences in myocyte size and structure appear as significant differences in the excitation-contraction (e-c) coupling between zebrafish and human (mammalian) cardiac myocytes (Figure 2). The contraction of human ventricular myocyte mainly relies on intracellular Ca 2+ stores of the sarcoplasmic reticulum (SR) in generating cytosolic Ca 2+ transients. A small Ca 2+ influx through L-type Ca 2+ channels triggers a large Ca 2+ release via ryanodine-sensitive Ca 2+ release channels of the SR so that about 77% of the cytosolic Ca 2+ transient originates from the SR [49,50]. In ventricular myocytes of the zebrafish heart Ca 2+ release from the SR makes only about 15% of the total Ca 2+ transient [48]. Voltage-dependence of cell shortening and Ca 2+ transients also suggest that sarcolemma Ca 2+ influx is the main source of cytosolic Ca 2+ in zebrafish cardiomyocytes [51]. In human cardiac myocytes, voltage dependence of the Ca 2+ transients is bell-shaped reflecting the voltage dependence of the trigger for Ca 2+ release from the SR, I CaL [52]. In zebrafish myocytes both I CaL and Na + -Ca 2+ -exchange directly contribute to cytosolic Ca 2+ transient resulting in monophasic voltage dependence of cell shortening and Ca 2+ transients [51]. Sarcolemmal Ca 2+ influx via I CaL and I CaT during a single twitch is almost 130 μM L −1 from which about 32% occurs through T-type Ca 2+ channels [36]. These differences in Ca 2+ handling are associated with 71% lower expression of ryanodine receptors in the zebrafish ventricle in comparison to mammalian (rabbit) ventricle, whereas little differences exist in the SR Ca 2+ content [48]. Ca 2+ sensitivity of ryanodine receptors of the fish heart is often low in comparison to that of mammalian cardiac ryanodine receptors [53], which might also contribute to the small SR Ca 2+ release in zebrafish cardiomyocytes [48]. Overall, the main differences in e-c coupling between zebrafish and human ventricular myocytes are the smaller role of intracellular Ca 2+ stores of the SR, the presence of large I CaT and the absence of T-tubuli in the zebrafish myocytes.

Implications for the use of zebrafish in drug screening
The use of animal models for studies of human cardiac electrophysiology is based on the similarity of animal and human hearts concerning ion current densities, ion channel compositions and mechanisms of ion channel regulation by rate changes and autonomic nervous system agonists [54,55]. However, electrophysiological properties of cardiac myocytes are species-specific and significantly different even between mammalian species (e.g. human vs. dog, rabbit and guinea pig) [54,55]. Therefore, it is necessary to consider, whether the noticed differences in electrophysiology between zebrafish and human ventricular myocyte might affect the status of zebrafish as a model for drug screening and safety pharmacology. In this respect, quantitative differences in repolarizing currents between model and human hearts are regarded significant [54,55]. Since ion channel isoforms often differ concerning activation and inactivation kinetics, voltage-dependence and drug affinity [39,43,[56][57][58], isoform-composition of the expressed ion channels is also expected to be important for drug screening.
The human erg (KCNH2) channel is known for its propensity of being blocked by wide variety of small molecules, which may lead to AP prolongation and lethal cardiac arrhythmias. Therefore, preclinical drug screening procedures aim to assess drug-induced inhibition of I Kr and prolongation of AP and QTc interval of electrocardiogram in order to remove proarrhythmic molecules from the drug development programs. However, drugs which inhibit I Kr do not always produce QT prolongation due to simultaneous inhibition of the depolarizing I Na and I Ca currents, or since other ion channels provide a repolarization reserve to compensate for I Kr inhibition. Proarrhythmic effects appear only when the drug changes the balance between inward and outward currents. To improve preclinical drug screening, the consortium of international stakeholders has recently launched an initiative called Comprehensive in-vitro Proarrhythmia Assay (CiPA). One of the central tenets of this initiative is that drug molecules are tested in vitro against multiple ion channels. Indeed, recent studies have indicated that the assessment of drug affinity toward multiple ionic targets improves the prediction of proarrhythmia risk in comparison to the sole I Kr analysis [59,60]. For example, if the outward I Kr and inward I CaL are inhibited with a similar IC 50 value (e.g. in the case of verapamil), the proarrhythmia risk is low.
Potential confounding factors in the applicability of zebrafish as a model is the presence of large I CaT and the putative absence of I Ks and I to . Since different channel isoforms of the same subfamily may have different electrophysiological properties and drug affinities, the erg (human KCNH2 vs. zebrafish KCNH6) and Kir2 (human Kir2.1 vs. zebrafish Kir2.4) channel isoform compositions may be also important. Zebrafish might be a useful high-throughput drug screening platform with the advantages of both phenotype screening, if it fulfils the qualifications of the CiPA procedure. Therefore, the calibration routine of the CiPA initiative should be conducted on zebrafish cardiac myocytes [61]. This routine involves key cardiac ion channels (I Kr , I CaL , I Na , I to , I Ks and I K1 ), which should be examined under standardized voltageclamp conditions for inhibition potency (IC 50 ) of 12 selected drugs of the "minimally acceptable" dataset [61]. These compounds are categorized into high, intermediate and low risk of torsades de pointes arrhythmia according to their currently known properties. The list of target channels should also include I CaT in the zebrafish.
One more factor that may be important in regard to proarrhythmia potency of zebrafish heart is the e-c coupling of ventricular myocytes. Factors recognized as significant causes of cardiac arrhythmias in mammals include APs that are too long or too short. If APs are abnormally long, early afterdepolarizations during the AP plateau (Phase 2 or 3) may be provoked by reactivation of Ca 2+ or Na + currents in the voltage "window," where all Ca 2+ and Na + channels have not yet been inactivated and can be reactivated. In addition, early afterdepolarizations are promoted by spontaneous Ca 2+ release from the SR that activates inward current via the reverse mode operation of Na + -Ca 2+ exchange [62]. APs that are too short can predispose the heart to delayed afterdepolarizations, which occur in early diastole (Phase 4), when spontaneous Ca 2+ releases from the SR activate the inward Na + -Ca 2+ -exchange current. Afterdepolarizations may depolarize membrane potential to the AP threshold and induce extra systoles (triggered activity). Generation of delayed afterdepolarizations requires spontaneous Ca 2+ release from the SR, which subsequently activates Ca 2+ efflux via Na + -Ca 2+ exchange and membrane depolarization to the AP threshold [63]. In zebrafish heart, SR makes only a minor contribution to cardiac e-c coupling, possibly due to the low Ca 2+ sensitivity of the ryanodine receptors. The relative independence of fish heart contraction from Ca 2+ -induced Ca 2+ release is expected to make the zebrafish heart relatively resistant against early and delayed afterdepolarizations and therefore less suitable as an arrhythmia model [18].
The similarities between human and zebrafish cardiac electrophysiology are often emphasized, while the differences are overlooked or neglected. However, it may be that those physiological functions and electrophysiological properties that are unique to the zebrafish heart are the most useful features for cardiac research. Perhaps the most spectacular example is the exceptional regeneration power of the zebrafish heart, which may reveal to us the molecular underpinnings needed to heal the damaged human heart [64]. Similarly, the exceptionally strong expression of the I CaT in the zebrafish ventricle may provide a test system to examine the role of T-type Ca 2+ channels in e-c coupling and its significance as a drug target. In the human heart, I CaT is re-expressed and T-tubuli are lost, when the heart is subjected to pathological stressors that induce hypertrophy and failure [65][66][67]. Thus, zebrafish heart could be a "natural" model for testing proarrhythmic propensity of the drugs in the diseased heart. Structural (small cell size, absence of T-tubuli) and functional characteristics (minor role of SR in e-c coupling, presence of I CaT ) of the zebrafish ventricle are more like those of fetal or neonatal mammalian heart than those of the adult human heart [68,69]. Therefore, the zebrafish heart might be a better model for fetal/neonatal and diseased human heart than for the adult human heart in drug screening.