Age-dependent electrical and morphological remodeling of the Drosophila heart caused by hERG/seizure mutations

Understanding the cellular-molecular substrates of heart disease is key to the development of cardiac specific therapies and to the prevention of off-target effects by non-cardiac targeted drugs. One of the primary targets for therapeutic intervention has been the human ether a go-go (hERG) K+ channel that, together with the KCNQ channel, controls the rate and efficiency of repolarization in human myocardial cells. Neither of these channels plays a major role in adult mouse heart function; however, we show here that the hERG homolog seizure (sei), along with KCNQ, both contribute significantly to adult heart function as they do in humans. In Drosophila, mutations in or cardiac knockdown of sei channels cause arrhythmias that become progressively more severe with age. Intracellular recordings of semi-intact heart preparations revealed that these perturbations also cause electrical remodeling that is reminiscent of the early afterdepolarizations seen in human myocardial cells defective in these channels. In contrast to KCNQ, however, mutations in sei also cause extensive structural remodeling of the myofibrillar organization, which suggests that hERG channel function has a novel link to sarcomeric and myofibrillar integrity. We conclude that deficiency of ion channels with similar electrical functions in cardiomyocytes can lead to different types or extents of electrical and/or structural remodeling impacting cardiac output.


Introduction
Cardiac performance significantly impacts quality of life and heart disease accounts for more deaths (Centers for Disease Control and Prevention, American Heart Association, World Health Organization) than any other disease. However, much remains to be learned concerning the cellular and molecular mechanisms involved in pathological cardiac remodeling, in order to design newer, more effective treatments. The vertebrate heart is a complex multilayer structure whose autonomous function can be modulated by neurohormonal factors. In addition, genetic, metabolic and environmental factors all contribute to heart disease, thus complicating elucidation of the underlying cellular mechanisms. Complex genetic studies in vertebrate models are difficult to conduct and are not easily interpretable, and studies of the effects of aging, even in mouse models, take years. Significantly, studies in mice are also complicated by the fact that important K + channels that regulate repolarization in the human heart do not make major contributions to heart function in murine adults (reviewed in [1,2]). These channels include the hERG K + channel that is a target for many anti-arrhythmia drugs, as well as the KCNQ K + channel, and mutations in both of these genes contribute to long QT syndrome (LQTS) which can trigger cardiac arrhythmias and are likely significant contributors to sudden cardiac death.
The Drosophila heart model has been successfully used to elucidate the molecular-genetic basis of cardiac development and pioneered our understanding of the origins and specification of cardiogenesis in the animal kingdom [3][4][5]. More recently the fly heart has also become a prototypical model to study the genetic causes of cardiac dysfunction and aging (reviewed in [6][7][8]). The heart of Drosophila is a linear tube composed of a single layer of myocardial cells [9], which forms in a homologous fashion to the early embryonic heart in vertebrates [10], but remains much less complex than the looped and chambered multilayered adult vertebrate heart. Despite the structural differences, a number of studies indicate that there are significant functional similarities between the fly and human heart. As in vertebrates, heart function in Drosophila is myogenic [11,12] and its rate can be modulated by neuronal and hormonal input [13,14]. Heart muscle protein composition, as well as muscle function and dysfunction share many similarities to human hearts [15][16][17][18]. Both dilated and restricted cardiomyopathies have been described by us and others in the fly heart and have been linked to mutations in homologous genes with similar effects in human heart [17,[19][20][21][22]. In addition, there is evidence of channelopathies in the fly heart that are reminiscent of what is observed in human patients [6,14,16,[23][24][25]. For example, we have previously shown that the KCNQ K + channel, which underlies a slow outward rectifying current in human myocardial cells (I Ks ), also functions in the fly heart and that mutations in this channel contribute to heart arrhythmias [16], as they do humans ([26-28]).
Here, we show that the hERG homolog seizure (sei), which is responsible for a rapid repolarizing current (I Kr ) in human heart, also functions similarly in the fly heart. Mutations in sei and adult cardiac-specific knockdown of this channel cause early after-depolarizations and cardiac arrhythmia. In addition, sei mutant hearts show considerable morphological remodeling that is not observed in hearts from flies with mutations in the KCNQ voltagedependent K + channel [16]. Expression analysis suggests that Wnt signaling is misregulated in hearts from sei mutants, and that misregulation of this pathway enhances the compromised sei function in generating cardiac pathologies. Our results suggest that alterations in sei channel function may play novel roles in cardiac remodeling that involves Wnt signaling.

sei expression in the Drosophila heart
In human hearts, a number of different K + channels contribute to the repolarization of cardiac action potentials (APs). We used PCR to examine expression of different K + channels in one week old Drosophila adults. Hearts and heads were isolated and real time PCR (rtPCR) analysis showed significant expression of a number of K + channels ( Fig 1A). These included the voltage-dependent K + channels Shaker (Sh) and Shaker-like (Shal). Also fly homologs for the human Ether-a-gogo Related Gene (hERG) were identified, including seizure (sei) and ether-agogo-like (elk). The Ca 2+ activated K + channel, encoded by slowpoke (slo), which is homologous to the big K + (BK) channel, was also expressed in the heart. Expression observed for all these channels was at levels somewhat lower to those found in neural tissue ('head' in Fig 1A). To confirm this result we used Fluidigm's Biomark nanofludic qPCR to also measure gene expression levels in isolated hearts ( Fig 1B). We used expression of the L-type Ca 2+ channel, Ca-alpha1D (CAD), which has been shown to underlie the cardiac action potential (AP) in flies, [29] as a positive control. Results show that KCNQ, sei, sh, and the sulfonylurea receptor channel (sur, previously shown to play a role in fly heart function) are expressed at roughly the same levels as is CAD, with lower expression of elk and Ir. Thus, including the previously documented KCNQ channel [16] and K ATP channel dSUR [25] the fly heart expresses at least eight different potassium channels all of which are likely to be involved in regulating cardiac repolarization, as they do in mammals and humans [1,30].

Loss of seizure function causes arrhythmias and reduced contractility
Analysis of high speed movies of beating hearts combined with our Semi-automated Heartbeat Analysis (SOHA) software (www.sohasoftware.com) allows precise quantification of a number of cardiac function parameters [31,32]. Heart function in the fly, as in humans, is myogenic and modulated by neuronal and hormonal input (Dlucis & Levine 1997). Hearts from young (1 week old) wildtype (wt) flies, when denervated as in the semi-intact preparations [16], typically beat with a regular endogenous rhythm of 1-3 Hz; this preparation allows us to monitor effects on myocardial cells in the absence of CNS input. In the absence of innervation the heart contraction/relaxation events (systolic intervals, SI) in young flies are between 0.2-0.28 seconds and relaxation periods (diastolic intervals, DI) are between 0.5-0.8 seconds (e.g. [16,18]) with DIs tending to be much more variable than SIs.
Our analysis of heart function in both sei mutants, sei ts1 and sei ts2 , revealed that mutant hearts exhibit a significantly slowed heart rate, compared to that of the Canton S genetic background controls (Wt CS ). This bradycardia was due primarily to a lengthening of the DI that is apparent, even in heterozygotes (sei ts1 ;+ and sei ts2 ;+, Fig 1C-left), with little change in the SI length (Fig 1D-left). sei mutant flies also exhibited an increase in arrhythmic beating compared to controls, whereas this parameter was unaffected in heterozygotes (Fig 1E-left). In addition, we observed a significant reduction in diastolic diameters (Fig 1F-left), but not in systolic diameters (Fig 1G-left), when compared to hearts from controls. We also observed significant decreases in cardiac contractility, measured as fractional shortening (FS), in sei homozygous mutants whereas hearts from heterozygotes showed only mild or no reductions in fractional Fig 1. (A) qPCR of isolated hearts and heads from wildtype (w1118) Drosophila. Seven potassium channels, KCNQ, seizure (sei), the BK channel slowpoke (slo), shaker (sh), the inward rectifier (Ir), ether a go-go like (elk) and Shaker like (shal) exhibited significant expression in isolated hearts. Expression in the head (mostly nervous tissue) is shown for comparison. (B) Δ Ct values of different voltage-activated K + channels calculated from raw Ct values in the wild-type Canton-S (Wt CS ) background. Note that ΔCt values are inversely correlated with relative expression such that a channel like elk is weakly expressed while the Ca 2+ channel, Ca-alpha1D (CAD), as well as K + channels KCNQ and sei have relatively high levels of expression. (C-E) Functional analysis of sei mutants shows alterations in heart function. (C, left) Mean Diastolic Intervals (DI) were significantly increased in hearts from 1 week old sei heterozygotes and homozygote mutants compared to their genetic background control Wt CS . (C, right) The increase in mean DI observed for sei ts1 mutants was rescued by insertion of two genomic copies of wildtype sei (sei wt2 ). Overexpression of an extra copy of wt sei in the wt background (+;sei wt1 ) did not affect DI. (D, left) Mean Systolic Intervals (SI) did not vary among the different genotypes. (D, right) Mean SI did not vary among the genotypes. (E, left) The incidence of arrhythmia (quantified as the heart period standard deviation normalized to the median heart period) was significantly increased in hearts from homozygous sei ts1 homozygous mutants. (E, right) The increase in arrhythmia was partially rescued by insertion of two copies of the wt sei gene (sei wt2 ). (F-H) Contractility is impaired in sei mutants. (F, left) Heart diameters during diastole were significantly smaller in one heterozygote and in both homozygous sei mutants compared to their background control, Wt CS . (F, right) Heart diameters during diastole were significantly smaller in hearts from sei rescue and overexpression flies compared to sei ts1 mutant. (G, left) The diameters of the hearts during systole did not vary significantly between the sei mutant lines and their background control, Wt CS . (G, right) Heart diameters during systole were significantly smaller in hearts from sei rescue and overexpression flies compared to sei ts1 mutant. (H, left) Fractional shortening is reduced significantly in both of the homozygous sei mutant lines (H, right) but was not rescued in hearts from flies expressing genomic sei wt . For all figures significance was calculated using one-way ANOVA and Dunnett's multiple comparison post-hoc test.; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Numbers of hearts examined are shown in bars in G.
shortening (Fig 1H-left). Similar alterations in heart function parameters were also seen in hearts from 3 and 5 week old homozygous mutant flies (S1 Fig).
In order to unequivocally demonstrate that the heart defects observed in mutants are due to a dysfunctional sei locus, we attempted to rescue the cardiac abnormalities of the sei mutation with wt genomic copies of this locus (see Materials and methods) in a sei ts1 mutant background. Hearts from homozygous sei ts1 flies crossed into the rescue construct background (sei ts1 ;+) showed the expected bradycardia due to increased DI. This bradycardia was rescued by the introduction of two genomic copies of the wt gene (sei ts1 ;sei wt2 , Fig 1B-right). Similarly the increased arrhythmicity observed in response to mutation of the sei gene could be partially rescued by the introduction of two wt sei genes (Fig 1D-right). However, the reductions in heart diameters and contractility (FS) that we observed in hearts with two copies of the mutant gene were not rescued by the addition of genomic copies of sei (Fig 1E-1G, right); this may be due to the particular insertion site of the rescue construct causing suboptimal expression. An extra genomic copy of the sei gene in the wt rescue background (+,sei Wt1 ) by itself had little effect on heart function compared to controls (Fig 1C-1H).
Because the sei channel plays a role in nervous system function we wanted to confirm a heart autonomous role for this ion channel. We crossed a cardiac-specific driver (tinCΔ4-Gal4) with UAS-sei-RNAi flies to specifically knockdown (KD) sei in myocardial cells. We examined heart-specific sei KD at three weeks of age because our structural data suggested that the effects of sei mutations were age-dependent (see below); thus the cardiac phenotype would be expected to be more robust in KD hearts compared to wt hearts which do not exhibit many functional defects at 3 weeks. Cardiac sei KD caused increases in the DI as was seen in sei mutants and also caused significant increases in the SI (Fig 2A & 2B, solid bars). In addition, there was a significant increase in arrhythmia ( Fig 2C, solid bars). Hearts also showed a significant decrease in fractional shortening by three weeks of age ( Fig 2D, solid bars), as for both systemic sei mutants (S1 Fig).
We also used a heart-specific temperature-sensitive Gal4 driver (tubGal80 ts ; tinCΔ4-Gal4) to determine whether KD only during adult stages was sufficient to generate the observed heart defects, or whether developmental effects also played a role. Flies were cultured at the permissive temperature of 18˚C until eclosion, at which time the adult flies were collected and switched to 25˚C. At this non-permissive temperature Gal80 is no longer expressed and its inhibitory effects on Gal4 activity are removed. Hearts from 3 week old adults, maintained at 25˚C, showed significantly increased SI, DI and AI, compared to driver and UASsei RNAi only controls exposed to the same temperature regime (Fig 2A-2C; striped bars) and we again observed a significant decrease in fractional shortening ( Fig 2D; striped bars).
In sum, we found that that sei plays a role in heart function in the fly and compromised sei function in the adult heart causes bradycardia even at young ages. In contrast to KCNQ mutants, we find that impaired sei function causes reduced contractility and causes age-dependent morphological remodeling, suggesting that sei may have broader consequences than just affecting electrical properties of the heart.

Seizure and KCNQ mutations have differential effects on cardiac contractility under loaded conditions
To explore the roles of sei and KCNQ in the heart further, we used a method we have recently developed for monitoring contractility under conditions that put a load on the fly's heart using artificial hemolymph of increasing viscosity [32]. This methodology allows us to automatically parse individual contractions into three distinct phases: shortening (SP), isometric (ISO) and lengthening phases (LP, Fig 3A). In addition, we can calculate shortening velocities in conjunction with measurements of heart wall movement distances. Assays done under unloaded conditions (normal artificial hemolymph) and then loaded conditions (plus 20% Ficoll) allow us to examine relative force production by the heart.
Under unloaded conditions, mean shortening velocities for KCNQ null mutants (KCNQ 370 , see  were not significantly different from that of Wt PE hearts (the precise excision genetic background control, Fig 3C). However, sei mutants showed a significantly reduced shortening velocity compared to Wt CS , the sei genetic background control, as well as compared to hearts from Wt PE and KCNQ mutants. This reduced velocity was due to in part to the decreased FS exhibited by the sei mutants. The length of time that hearts were able to maintain an isometric contraction under unloaded conditions did not vary between genotypes ( Fig  3D-"Pre"). However, under loaded conditions hearts from both control flies and KCNQ 370 mutants exhibited significantly increased isomeric contraction times ( Fig 3D-"Load", 20% Ficoll). In contrast, sei ts1 mutants appeared unable to increase isometric contraction times under a load, suggesting a defect in the ability of these hearts to generate tension. Hearts from Heart and adult-specific knockdown of sei affects heart rhythmicity and contractility. Heart-specific KD of sei was achieved by crossing flies with a UASsei RNAi construct to the heart-specific driver tinCΔ4 Gal4. Adult specific KD was achieved using a temperature-sensitive driver tub-gal80 ts in combination with tinCΔ4 Gal4 to knock down sei specifically in the adult heart (see Materials and methods). (A&B) Heart-specific (tin and adult specific KD of sei causes a significant increases in both systolic (A) and Diastolic Intervals (B) compared to controls (tinCΔ4 driver/+, tub gal80 ts ; tinCΔ4 driver/+, and UASsei RNAi /+) at 3 weeks of age. (C) Heart-specific and adult specific KD of sei causes a significant increases in arrhythmia compared to controls (tinCΔ4 driver/+, tub gal80 ts ; tinCΔ4 driver/+, and UASsei RNAi /+) at 3 weeks. (D) The fractional shortening, an indicator of heart contractility, was significantly reduced in 3 week sei KD hearts compared to controls. Significance was calculated using a twoway ANOVA and Tukey's post-hoc test; *p<0.05, **P<0.01,****p<0.0001; number of flies examined is shown in bars in D.
https://doi.org/10.1371/journal.pgen.1006786.g002 all four genotypes recovered to pre-load values 30 min after the load was removed ( Fig 3D-"Post"). These finding further support the notion that sustained sei deficiency affects contractility more than a loss of KCNQ.
Electrophysiological analysis of young sei and KCNQ mutants show early afterdepolarizations that is similar to electrical activity seen in aged wt flies To monitor the intrinsic electrical activity of myocytes in the cardiac tube, we used standard intracellular recording techniques with sharp electrodes. In all experiments (except where specifically indicated) we used a pretreatment with 10uM blebbistatin to inhibit muscle contraction [33], which under the conditions used does not significantly affect electrical activity [34]. Although reversible, this pretreatment blocked contractions for extended periods following removal of blebbistatin and permitted us to obtain recordings that were stable and could be maintained in some cases for as long as one hour. Inhibition of active contractions would be expected to minimize damage to the cell membrane at the point of electrode penetration and to prevent movement artifacts. All hearts examined under these conditions exhibited spontaneous electrical activity, as expected of myogenic heart tissue in a denervated preparation.  Table 1). Resting membrane potentials ranged between -50 and -35 mV; these values in adult hearts are significantly more hyperpolarized than previously reported for intracellular recordings from larval hearts [35,36]. Recordings from myocardial cells in sei mutant hearts often exhibited APs with two peaks (Fig 4B, Table 1), likely triggered by early afterdepolarizations (EADs) as has previously been described in vertebrate myocardial cells [37]. Hearts from both sei and KCNQ mutants typically exhibited bursts of APs but the frequency of this type of event was higher for KCNQ compared to sei mutants (Fig 4; Table 1 quantified as "Peaks/Burst" and "Event Duration"). Triggered APs were rarely observed in young wt hearts (Table 1, Fig 4A). In addition, single peak APs from sei mutant hearts exhibit dramatically broadened APs compared to both wt lines (Wt PE and Wt CS , Fig 4G) at points where repolarization is 10% (APD 10 ) and 50% (APD 50 ) complete (Fig 4I and 4H), and significantly broader than Wt PE APs at 90% repolarization (APD 90 ) ( Fig 4J, Table 1). We were unable to record single peak APs from KCNQ mutants so they were not included the AP duration analysis.
Interestingly, electrical activity in wt hearts changed with age, in that hearts from older (4-6 weeks) wt flies show triggered APs that are reminiscent of those seen in hearts from young sei mutants (Fig 4, compare B&D). This altered electrical activity in old wt hearts may reflect the age-related decline in repolarization reserve that we have previously documented [7,16]. Old sei and KCNQ mutant hearts almost always exhibited bursting APs, an exaggerated form of the electrical activity observed at younger ages (compare Fig 4B & We wished to determine the relationship between the bursting electrical activity and cardiac contractions. To accomplish this hearts were not exposed to blebbistatin and we recorded intracellular electrical activity in spontaneously beating hearts while simultaneously recording high speed movies. Resting membrane potentials and electrical activity recorded under these conditions were similar to those observed with blebbistatin pretreatment (Fig 4A-4F), reaffriming that exposure to this compound did not significantly affect the electrophysiological activity and that for these short periods of electrical recording the impaled and beating myocardial cells were functioning as expected. In order to synchronize the electrical recording with the optical recording we had the image capture software send a signal to the amplifier that could be visualized in a separate channel (Fig 5E). Under these conditions we again observed regular, single peak action potentials in young wt hearts. Alignment of M-modes from the movies with the electrical activity that occurred during the filming interval showed that the action potentials in young wt hearts was correlated with individual heart contractions ( Fig 5A). Electrical recordings from sei ( Fig 5B) and KCNQ mutant hearts (Fig 5C & 5D) showed the expected triggered AP bursting activity and, notably, the AP peaks correspond 1:1 with the movements of the heart tube shown in the synchronized M-mode. Thus these AP bursts underlie the fibrillatory contraction patterns that we have previously documented in KCNQ mutants using our optical methodology [16].
The selective hERG antagonist E4031 causes bradycardia in hearts from wt but not sei mutants We attempted to reproduce the effects of sei mutations using the selective hERG antagonist E-4031 [38]. Semi-intact heart preparations were recorded before and after a 15 m exposure to either AHL containing 1μM E-4031 or AHL alone (vehicle). Histograms showing the distribution of the individual DIs for all flies examined are shown in Fig 6. E-4031 caused a right shift in the DI peak from 0.6 to 0.7 sec as well as an increase in the percentage of DIs longer than 0.7 s in hearts from wt flies compared to baseline (T 0 , Fig 6A). Notably, there was a dramatic reduction in the total number of heart beats recorded in these 30sec movies in response to the drug exposure. In contrast the distribution of DIs in sei mutants is significantly broader at baseline than for Wt CS and there is no shift in the maximum peak following drug exposure (Fig 6B). There is an increase in the percentage of DIs that are longer than 1.0 sec. Overall averages for all flies are shown in Fig 6E; Wt CS hearts, but not sei mutants, exhibited a significant increase in DI following exposure to 1μM E-4031. A second, selective hERG antagonist Dofetilide [39] (at 1μM) produced similar bradycardia in hearts from wt flies, also characterized by significantly prolonged diastolic intervals and not observed in sei mutants (Fig 6F).

Cardiac dilation and myofibrillar remodeling in sei but not KCNQ mutants
To visualize muscle structure, hearts were exposed as for functional analyses and left in situ, attached to the dorsal cuticle. These hearts were then fixed and stained for muscle proteins as previously described [21,40]. Phalloidin staining of F-actin in hearts from wt flies reveals the distinctive circumferential organization of myofibrils within the myocardial cells that make up the heart tube. A similar circumferential arrangement was observed in the anterior-most region of sei mutant hearts, a structure called the conical chamber ( Fig 7A, upper portion). However, myofibrils became increasingly thinner and this arrangement was more disorganized in more posterior chambers in sei mutant hearts with myofibrils oriented primarily longitudinally ( Fig 7A, bottom set of double headed arrows).
We examined hearts at different ages and found that this cardiac remodeling was age-dependent. We quantified the number of hearts showing structural alterations ranging from relatively mild (presence of gaps) to more severe (reoriented myofibrils) (Fig 8A). A relatively small percentage (20% or less) of hearts from young (1-2 week) wt flies (Wt PE and Wt CS ) showed gaps between myofibrils and none showed reorganization to a longitudinal orientation (Fig 8B &  8C). While almost 50% of hearts from young sei flies showed gaps, this was not significantly different from the Wt CS control. However, sei mutant hearts did exhibit significant reorganization Duration (APD) was measured for single-peak APs at the point where the membrane voltage had repolarized to 10% (H), 50% (I) and 90% (J) of its maximum depolarized potential. (one-way ANOVA;*p<0.05; ** p<0.01). Electrical and morphological remodeling in potassium channel mutants of their myofibrillar structure at 1-2 weeks of age. By 5-7 weeks 80% of the hearts from sei mutant flies showed this more severe form of myofibrillar remodeling (Fig 8C). In contrast only the CS background controls showed evidence of structural rearrangement in older flies ( Fig 8C). The reorganization of circumferential to more longitudinally oriented disorganized myofibrils was most reliably observed in the heart chamber located in the fourth abdominal segment (compare Fig 7B & 7C). This correlates very well with our observations that many hearts from older flies showed abnormal contractions in this same region (see S1 Movie).
To establish that this remodeling was cardiac autonomous we used tinCΔ4-Gal4 driver to express either UASsei RNAi or UAS KCNQ RNAi specifically in myocardial cells during embryonic and adult (but not larval) stages. 3-week old adult hearts from outcrossed driver flies showed the expected circumferential arrangement (tinCΔ4-Gal4/+, Fig 9). However, cardiac sei KD showed evidence of gaps (tinCΔ4-Gal4>UASsei RNAi, Fig 9C 1 & 9C 2 ). Myofibrillar structure in the KCNQ KD hearts more closely resembled that of the controls (Fig 9B). These finding are consistent with data from our functional analysis of hearts with sei KD, which revealed a significant reduction in contractility (Fig 1H).
Since it is possible that the remodeling we observed was due to a developmental defect we examined heart structure in 3 rd instar larva just prior to pupation. The myofibrillar organization in wt wandering third instar larva revealed a more loosely organized arrangement compared to adults. However we observed no gross differences in size or in the overall arrangement of myofibrils between wt and sei mutants (Fig 10A-10D).

Expression analysis of sei and KCNQ mutant hearts
We used Affymetrix microarrays to analyze gene expression in KCNQ (N = 3 replicates) and sei (N = 5 replicates) mutants as well as their respective background controls; CS (N = 5 replicates) and KCNQ 97 (N = 3 replicates). Interestingly, we found a dramatic difference between the two mutants in the numbers of misregulated genes, with sei mutants having significantly more genes with altered expression than KCNQ mutants. We identified 1019 upregulated and 712 downregulated probesets in sei mutants compared to controls, but only 91 upregulated and 68 down-regulated probesets in KCNQ mutants compared to controls (|Fold|>2 and P<0.05) (Fig 11A, S1 Dataset).
Gene set enrichment analysis of sei mutant data showed 6 GO function categories were significantly (Z score >2) altered in KCNQ mutant hearts compared to background controls ( Fig  11B) compared to significant 17 down regulated and 38 up regulated GO function categories in the sei mutant hearts (Fig 11C). The top down regulated functions included "repression of translation", "protein phosphatases", "receptor binding" and "wnt protein binding". The majority of the up regulated functional categories (32/38) were involved in some aspect of cell metabolism. Consistent with the morphological data, a number of sarcomere-related proteins were also significantly down regulated (p<0.05) in sei but not KCNQ mutant hearts, although most of the |Fold| changes were less than 2. Nevertheless we were able to confirm a significant downregulation of MHC das well as adenomatous polyposis coli (APC), a Wnt pathway gene with qPCR ( S2 Fig). sei interacts genetically with Wnt effector pygopus in the fly heart Our previous studies revealed a required role for the Wnt signaling gene pygopus (pygo), that codes for a β-catenin-associated nuclear receptor protein [41][42][43][44] in heart function [45,46]. These studies showed that pygo KD in the heart caused bradycardia and myofibrillar Electrical and morphological remodeling in potassium channel mutants  Fig 6B). Ages are in weeks; numbers of hearts examined are indicated in each bar in B. Significance was determined using multiple T-tests comparing each mutant with its genetic background control at each age; **p<0.01, ***p<0.001. remodeling associated with a reduction in contractility, phenotypes that are reminiscent of the effects shown above for sei mutant hearts. In addition, our microarray analysis suggests that pygo expression may be downregulated in sei, but not KCNQ, mutant hearts (S3 Dataset).
We used a double heterozygote approach to look for genetic interactions between pygo and sei. We examined cardiac function in single heterozygotes of sei and pygo mutants (sei ts1 /+ and pygo/+). For either single heterozygote, the mean diastolic and systolic intervals were not significantly different from each other ( Fig 12B) and were similar to those seen for controls (Figs  1 & 2). In contrast, we observed significant bradycardia for the double heterozygote combination (sei ts1 /pygo, Fig 12B). Fractional shortening was also significantly reduced in the sei ts1 /pygo double heterozygotes compared hearts from single heterozygotes (Fig 12D). Taken together these data suggest a genetic interaction between the sei K + channel gene and Wnt signaling.

Discussion
Channel dysfunction, or channelopathies, underlie a number of cardiac disorders such as LQTS and are thought to contribute to sudden cardiac arrest, infant sudden death syndrome [47] and increased risk of cardiac arrhythmias. The human ether-a-go-go related K + channel (hERG) along with the KCNQ K + channel (I Kr , and I Ks respectively) are the major contributors to cardiac repolarization (Phase 2 & 3) in humans. Defects in KCNQ and hERG channels have been shown to cause LQT1 and LQT2, respectively and cardiac arrhythmias in humans. The hERG channel in particular has been a major target for the development of anti-arrhythmia drugs and can be inhibited by a variety of drugs that do not specifically target the heart [48,49]. We have previously shown that the KCNQ K + channel is functional in the adult fly heart and that mutations in this channel contribute to cardiac arrhythmias [16]. In the fly, seizure and KCNQ mutations do not significantly affect cardiac development (Fig 6), although in the mouse some hERG mutations cause looping defects and embryonic lethality [50]. Importantly for heart function studies, survivorship of adult flies is not acutely affected by badly functioning hearts that would quickly result in death in vertebrates, likely because the fly does not rely on the heart for oxygen distribution, which is carried out by a separate system of tracheoles.
The current data demonstrate that the sei gene, the fly homolog of hERG, along with a number of other genes encoding K + channels, are also expressed in fly myocardial cells (Fig 1A) and likely contribute to the repolarization capacity of the heart. This set of K + channels is reminiscent of what is observed in vertebrate hearts [1] [2]. The primary effect of sei dysfunction in mutants was bradycardia and this phenotype could be replicated by cardiac-specific and adult, cardiac-specific sei KD (Fig 2) as well as acute application of selective hERG antagonists ( Fig  6). However, the increases in SI seen in both KD experiments (Fig 2) are different from the mutant phenotype and possibly reflect some compensatory genetic alterations in the systemic sei mutants. Our data also show that, as in humans, mutations in sei cause arrhythmias ( Fig  1E) and electrical remodeling in the form of AP bursts that are likely triggered by EADs (Figs 4  and 5, quantified in 6). In addition we show that systemic mutants or cardiac-specific KD of  sei, but not of KCNQ, reduced heart contractility (Figs 1G and 2D), and were associated with structural remodeling (Figs 7 and 8).
These effects appear to be specific to alterations in sei as most of the sei cardiac function phenotypes can be rescued by over-expression of the wt channel (Fig 1). In addition, the morphological/functional effects of channel dysfunction appear not to be due to developmental defects (Fig 10) but are adult stage-and cardiac-specific (Figs 2 and 9).
The intracellular electrical activity in adult fly cardiomyocytes appears more nodal or atriallike [1,2] and is similar to what has previously been observed in larva [35,36,51] although more robust in terms of the resting potential and AP amplitude ( Table 1, Figs 4 & 5). We observed frequent AP bursts in both sei and KCNQ mutants and in hearts from old flies compared to young Wt fly hearts (Figs 4 and 5) reminiscent of the increase in electrical remodeling with disease and age in human myocardial cells [52,53]. However, there appears to be a significant difference between mutations in KCNQ and sei in terms of their effects on electrical and morphological remodeling, with mutations in KCNQ resulting in significantly more electrical arrhythmia at younger ages compared to sei mutations (Figs 4, 5 and 6). In addition, although 100% of APs recorded from both old KCNQ and sei mutants are arrhythmogenic, the severity of the events appears to be worse in KCNQ mutants. Notably, our ability to simultaneously record both intracellular APs while optically monitoring intact heart function demonstrate that the arrhythmogenic AP's we observe correlate directly with unsustained fibrillatory contractions of the heart wall revealed in M-modes (Fig 5).
Although both sei and KCNQ channel mutations cause electrical remodeling and arrhythmia in the fly heart model they appear to have differing effects on muscle structure (Fig 7). Mutations in sei appear to be linked to an increase in myofibrillar disorganization (Fig 7) and this effect on myocardial structure is manifest by reductions in fractional shortening (Fig 1H). In addition, the observed reductions in shortening velocities under loaded conditions suggests that the ability to generate tension is significantly lower in hearts from sei mutants than for controls or KCNQ mutants and are consistent with the reduced ability to sustain an isometric contraction in the sei mutant hearts (Fig 3). Together these results suggest differential roles for these channels on structural and electrical integrity of the adult heart.
Our microarray results suggest that the different effects of the two K + channel mutants are the result of underlying differences in gene expression. Our observation that many pathways involved in metabolism are significantly upregulated in sei mutants is consistent with previous observations in rabbits, although in that study both LQT1 and LQT2 [54] models exhibited similar effects. In particular, in the adult fly heart the Wnt signaling pathway appears to be selectively perturbed in the sei mutant hearts, as evidenced by a downregulation of many of its pathway components. This is consistent with our previous results demonstrating that mutation or cardiac KD of the TCF co-factor encoded by pygo [45] cause bradycardia and morphological remodeling similar to that observed for sei mutants. Importantly, we now demonstrate an interaction between sei channel mutations and mutations in pygo (Fig 11B-11D). Wnt signaling has previously been shown to play significant roles in cardiac development in flies [55][56][57][58] and vertebrates [59,60] and in cardiac disease [59]. We have shown in the fly that pygo likely plays a role in the maintenance of cardiac function and structure in the adult [45,46] and our current data suggest that Pygo and Sei channel dysfunction are likely linked genetically. Many components of both the canonical and non-canonical pathways exhibit reduced expression, which would suggest an overall reduction in Wnt signaling. However, we also observe downregulation of potential negative effectors of Wnt signaling, such as APC, the core of the destruction complex, and CtBP, which has been shown to mediate both activation and repression of transcription [61]. Both effects could be expected to result in increased stabilization of β-catenin. Thus, the exact role of different Wnt signaling components in maintaining cardiac structure and function remains to be determined.
Electrical and morphological remodeling have previously been shown to be linked; for example, knockout of the heart development transcription factor Ptx1 was shown to affect both electrical and morphological remodeling in mouse and humans [62]. A number of studies have documented pro-fibrotic and apoptotic effects of atrial fibrillation [63][64][65][66][67] as well as tachypacing in cardiomyocytes [68], in dog models [69] and in response to sustained atrial fibrillation in patients [63]. Further, a genetic variant situated close to the long QT syndrome (LQTS) type 2 gene KCNH2 has been shown to be associated with early onset AF [70]. Thus, cellular/molecular links between channel function, electrical activity and morphological remodeling and ultimately heart failure have been suggested but have yet to be clearly elucidated [66]. An understanding of how hERG channels interact with cellular pathways involved in electrical and structural remodeling and with other repolarizing currents such as I Ks , mediated by KCNQ channels, will be important to the development of novel anti-arrhythmia therapies. Our data using the fly heart model now provides the first clear genetic and physiological evidence that some channelopathies may be contributing to cardiac remodeling during disease progression via Wnt signaling.

Drosophila strains
Adult flies were collected upon eclosion, segregated by sex and raised at 25˚C with food changes every 3 days. Sei mutants, sei ts1 and sei ts2 , were previously isolated in a screen for ethylmethane sulfonate (EMS)-induced temperature-sensitive paralytic mutations on the second chromosome [71,72] and were a gift from Dr. Barry Ganetzky. Sei ts1 mutation introduces a stop codon before the membrane spanning region and sei ts2 has a Glu-to-Lys substitution at a critically conserved site at the channel pore [70]. The KCNQ 370 mutants were created by imprecise P-element excision that also introduces a stop codon before the membrane spanning region as previously described [16]. We used the genetic background strains of these mutations as Wt controls: Canton S (designated Wt CS , Bloomington Stock Center) for the sei mutants and a line with a precisely excised P element from the KCNQ locus of w 1118 flies (designated Wt PE , see [15]) for the KCNQ mutants. For knockdown experiment we used UAS-seiRNAi line (VDRC #190504), UAS-KCNQ RNAi (VDRC #106655) and UAS-APC2 RNAi (VDRC #100104) from the Vienna Drosophila Resource Center (Vienna, Austria). Heart-specific driver line used was tinCΔ4-Gal4 [73] that express in the myocardial cells of the heart. A temperature sensitive tubulin-Gal80 ts Gal4 driver that expresses Gal4 only when shifted to permissive temperatures above 18˚C was used in combination with tinCΔ4-Gal4 and UAS-sei RNAi to specifically KD sei in the adult heart [74,75].

Semi-intact preparations and Semi-automated Optical Heartbeat Analysis (SOHA)
Hearts from adult flies were exposed by dissection in room temperature bubbled artificial hemolymph (AHL, [77]). Heart movements were filmed with a Hamamatsu EM-CCD camera and recorded using HC Image software (Hamamatsu Corp). Heartbeat movements were analyzed as previously described using SOHA software ( [16,31,77,78]; http://sohasoftware.com/). For some experiments heart function was examined under loaded conditions as previously described (Cammarato et al, 2015). Semi-intact preparations were first filmed in artificial hemolymph which was then replaced with AHL containing 20% Ficoll 400 (w:v, Sigma Aldrich). Hearts were incubated with aeration for 30 min in this solution and filmed; following filming the Ficoll solution was replaced by AHL and hearts were allowed 30 min to recover before being filmed a final time. For these latter experiments frame rates were 200-250 fps.

Adult Drosophila cardiomyocyte electrophysiology
The semi-intact heart dissections and optical recording was performed as described above. The semi-intact preparation was incubated in artificial hemolymph containing 10 μM blebbistatin (Sigma Aldrich) and equilibrated with oxygenation in the dark for at 45-60 minutes until the hearts stopped beating [33,34]. The preparation was then supplied with fresh saline without blebbistatin and electrical potentials were recorded from the conical chamber using sharp glass electrodes (20-50 MΩ) filled with 3M KCl and standard intracellular electrophysiological techniques. Data were acquired using an Axon-700B Multiclamp amplifier, signals were digitized using the DIGIDATA 1322A and data were captured and analyzed using PClamp 9.0 and Clampfit 10.0 software respectively (all from Molecular Devices). Data was quantified from representative 30s recordings where the resting membrane potential had remained stable for at least 30s. APD10, 50 and 90 data was determined for 20-50 individual APs per fly and averaged.

Simultaneous optical and electrophysiological recordings
For simultaneous recordings we again used the semi-intact heart dissection as described above, however, we did not treat the prep with blebbistatin, thus it continued to actively beat during impalement and electrical recording. To coordinate the optical and electrical recordings a TTL pulse was sent by the image capture software to the Digitizer. The pulse duration lasted for the entire period of optical recording and was recorded in a separate channel by the PClamp software allowing us to delineate the beginning and the end of the optically recording directly within the electrical record (Fig 4).

Pharmacology
E-4031 and Dofetilide were obtained from Sigma-Aldrich. E-4031 stock solution (10 mM) was prepared by dissolving in distilled H 2 O; Dofetilide was dissolved in DMSO to make a 10mM stock. Hearts were dissected from both Wt CS and sei ts1 mutant flies, allowed to equilibrate for 30 min in oxygenated AHL and were then filmed for 30sec (T 0 ). The hemolymph was replaced with AHL containing either 1μM E-4031 or 1 μM Dofetilide. For E4032 experiments a second set of hearts were filmed using AHL as the "vehicle" and for Dofetilide an equivalent amount of DMSO without drug was added to the AHL. Hearts were filmed again for 30sec following a 15 min exposure to either drug or vehicle.

Immunostaining of adult Drosophila hearts
Semi-intact adult heart preps were prepared as above and larval hearts were dissected as previously described [79]. In all case hearts were stained as described in [40]. Briefly, hearts were fixed in 4% paraformaldehyde/phosphate-buffered saline (PBS) for 20 min, washed three times in PBT (PBS/0.1% Triton X-100) and incubated with primary antibodies in PBT for 2 hours. Hearts were then washed (3x 10') incubated with secondary antibodies and phalloidin for 1-2 hours at room temperature. After washing in PBT (3x 10'), they were mounted onto slides in Vectashield mounting medium for Florescence with DAPI (Vector Labs) and analyzed using ApoTome (Carl Zeiss). The monoclonal antibody against Pericardin was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52242. Antibodies against Drosophila Myosin were a gift from Dr. G. Melkani [80]. The secondary antibodies used were anti-rabbit and anti-mouse conjugated with Alexa Fluor 555 dye and Alexa Fluor 488 (Molecular Probes, Eugene, OR, USA).

RT-PCR
Total RNA was extracted from 10 adult hearts using TRIzol (Invitrogen, Carlsbad, CA) and purified with Mini RNA isolation kit (Zymo Research, Orange, CA) for heart RNA. After treatment with DNase I, first strand cDNA was transcribed by SuperScript III (Invitrogen) by oligo (dT)12-18primer, followed by second-strand synthesis. RT-PCR was carried out by using LightCycler FastStart DNA Master PLUS SYBR Green I kit (Roche, Basel, Switzerland). Semiquantitative PCR was performed for 32 cycles that consisted of denaturation at 94˚C for 30 s, annealing at 65˚C for 30 s, and extension at 72˚C for 30 s by using platinum Taq DNA polymerase (Invitrogen).
To verify the microarray results we performed qPCR analysis on 3 biological replicates each containing 10-12 adult hearts from both Wt CS and sei ts1 mutants. RNA was extracted using QIAzol (QIAGEN, Ref# 79306) and purified using miRNeasy Mini Kit (QIAGEN, Ref# 217004) following manufacturer's instructions for RNA purification of small biological sample sizes. Genomic DNA removal and cDNA synthesis was done using Quantitec Reverse Transcriptase kit (QIAGEN, Ref# 205311; see S1 Fluidigm analysis 5 fly hearts were isolated and snap frozen in 10μl of water. Each heart was lyophilized and lysed for 2 minutes at 98˚C in lysis buffer (LB) (0.25% NP40 in H 2 O). The lysed hearts were then briefly centrifuged, and 30μl supernatant transferred to a fresh tube. Reverse transcription was carried out by using 3.3μl of lysed heart (~8% of a single heart) in a 5μl final reaction volume using the VILO Reaction mix as per the manufacturer's instructions (Life Technologies). Genes were pre-amplified using a Fluidigm Master Mix (Fluidigm). For removal of single stranded DNA prior to nanofluidic cycling, 6 μL of Exosap solution (4.2 μL of water, 0.6 μL of Exonuclease 1 Rn Buffer, Exonuclease 1 (20 units μL −1 , New England Biolabs) was added to the 15 μL final reaction volume of the RT step. The resulting 21 μL final volume was then incubated at 37˚C for 30 min and then heat inactivated at 80˚C for 15 min. The pre-amplified volume (21 μL) was then diluted 10-fold in DNA suspension buffer (Teknova), and stored at −20˚C prior to running on the chip. Pre-amplified products for each individual fly heart was then assayed using Fluidigm's 48 nanofluidic qPCR arrays on a Biomark system (Fluidigm), according to their protocols. Biotium's EvaGreen DNA binding dye was used to detect amplified product according to Fluidigm's protocols. Values were calculated taking the experimental gene raw Ct values minus concomitant housekeeping control gene (minichromosome maintenance complex component 2, MCM2). Delta Ct values were calculated relative to raw Ct values from the wildtype Canton-S (Wt CS ) fly line. Ct values are inversely correlated with relative expression (see S1 Table for primer sequences).
We used GeneChip Drosophila Genome 2.0 Array to analyze gene expression for microarrays (30, 3-week old hearts per array) for each mutant line: sei ts1 (N = 5 replicates) and KCNQ 370 (N = 3 replicates) and for control flies representing each genetic background, Canton-S (CS; N = 5 replicates) for the sei mutants and the precise excision control (KCNQ 97; N = 3 replicates) for the KCNQ mutants. Affymetrix.cel files were quality controlled with the Bioconductor package affyQCReport [81]with R, an open-source software environment for statistical computing and graphics. RMA normalization for the dataset was then conducted with AltAnalyze [82,83]. For clustering analysis differentially regulated genes where considered those with an absolute fold change greater than 2 and a P<0.05 in compared to each mutants control line (1812 probesets). Clustering was conducted in AltAnalyze using the HOPACH algorithm with cosign similarity. For Gene Ontology (GO) enrichment analysis, GO-Elite [84] was used to find the enrichment of GO terms amongst the significantly up and downregulated genes (|Fold|>2, P<0.05) using each control stain as a base line. Raw data files and RMA normalized expression matrix have been deposited at GEO (GSE94589).