Fermitins, the Orthologs of Mammalian Kindlins, Regulate the Development of a Functional Cardiac Syncytium in Drosophila melanogaster

The vertebrate Kindlins are an evolutionarily conserved family of proteins critical for integrin signalling and cell adhesion. Kindlin-2 (KIND2) is associated with intercalated discs in mice, suggesting a role in cardiac syncytium development; however, deficiency of Kind2 leads to embryonic lethality. Morpholino knock-down of Kind2 in zebrafish has a pleiotropic effect on development that includes the heart. It therefore remains unclear whether cardiomyocyte Kind2 expression is required for cardiomyocyte junction formation and the development of normal cardiac function. To address this question, the expression of Fermitin 1 and Fermitin 2 (Fit1, Fit2), the two Drosophila orthologs of Kind2, was silenced in Drosophila cardiomyocytes. Heart development was assessed in adult flies by immunological methods and videomicroscopy. Silencing both Fit1 and Fit2 led to a severe cardiomyopathy characterised by the failure of cardiomyocytes to develop as a functional syncytium and loss of synchrony between cardiomyocytes. A null allele of Fit1 was generated but this had no impact on the heart. Similarly, the silencing of Fit2 failed to affect heart function. In contrast, the silencing of Fit2 in the cardiomyocytes of Fit1 null flies disrupted syncytium development, leading to severe cardiomyopathy. The data definitively demonstrate a role for Fermitins in the development of a functional cardiac syncytium in Drosophila. The findings also show that the Fermitins can functionally compensate for each other in order to control syncytium development. These findings support the concept that abnormalities in cardiomyocyte KIND2 expression or function may contribute to cardiomyopathies in humans.


Introduction
The mammalian Kindlins are a family of evolutionarily conserved proteins that mediate cell-cell and cell-matrix adhesion by regulating integrin function [1][2][3]. Findings from several animal models suggest that the formation of junctions between neighbouring cardiomyocytes (intercalated discs) may require integrins and integrin binding proteins [4][5][6][7][8]. It is important to understand the role of these proteins because mutations and polymorphisms that affect intercalated disc formation may be associated with the development of cardiomyopathies in humans [6,9,10].
Murine KIND2 protein is detected at intercalated discs, however genetic ablation of Kind2 in mice causes early embryonic lethality at the peri-implantation stage, before cardiogenesis [11,12]. In zebrafish, the global knock-down of Kind2 causes cardiac hyperplasia, disrupts intercalated disc formation and significantly reduces cardiac contractility [11]. These findings link Kind2 with the development of a functional cardiac syncytium; however a cardiomyocyte-specific test of this hypothesis is yet to be conducted.
Drosophila melanogaster is an important model for the identification and study of genes that have relevance to mammalian cardiovascular development and function [13][14][15][16][17]. The abdominal heart in Drosophila consists of individual cardiomyocytes that form a contractile syncytium and pump haemolymph from the abdomen, through the thorax towards the head (see schematic in Figure 1A).
Drosophila expresses two orthologs of human KIND2, Fermitin1 (Fit1) and Fermitin2 (Fit2) that contain a highly conserved FERM (4.1, Ezrin, Radixin, Moesin) domain which is required for binding to the cytoplasmic tail of b-integrins [18] (see Fig. 1B and Fig. S1). Both Fit1 and Fit2 message and protein are detected in adult heart tissue, consistent with a role in cardiac function [19,20].
To explore the role of Fermitins in the Drosophila heart we silenced their expression in cardiomyocytes and analysed the function of the adult heart. It was found that cardiomyocyte expression of Fermitins is critical for the development of a functional syncytium in Drosophila and that the genes can functionally compensate for each other. The findings support the hypothesis that disruption of cardiomyocyte KIND2 may lead to abnormal intercalated discs and cardiomyopathies in humans.

Analysis of cardiac synchrony and generation of cardiac raster plots
A video of the beating heart was opened in ImageJ [22], converted to greyscale and a movement movie processed by subtracting a frame's pixel intensity from that of the preceding frame using the 'Image Calculator' function. The mean pixel intensity of two, non-overlapping regions (the upper left quartile and lower right quartile) of each frame in the movement movie was then measured using the 'Measure stack' function. Regression analysis was then performed on the mean pixel intensity data obtained from the two regions. When the two regions are in perfect synchrony, R 2 = 1; when regions are not in synchrony, R 2 approaches zero. The mean (6SEM) R 2 value of 7 flies for each genotype is presented. Cardiac raster plots were generated by concatamerising the image stills from the movement movies in ImageJ so that there were 12 consecutive images per row over 30 columns. The resulting raster plot corresponds to 10.8 seconds of cardiac motion. Under normal conditions a rhythmically occurring event produces a geometric pattern in a raster plot, whereas loss of rhythmicity leads to a more random pattern. This methodology of illustrating a physiological rhythm is based on the analysis of biological rhythm data in the circadian biology field.

Fluorescent labelling and imaging of adult heart structures
Fluorescent labelling of Drosophila heart structures was performed as described [23]. In brief, semi-intact heart preparations of four flies of each genotype were fixed in 4% formaldehyde (Fisher Scientific, Loughborough, UK) in PBS for 20 mins then 3610 min washes with PBS-Triton-X (PBST)(Sigma, Poole, Dorset, UK). Hearts were then incubated with primary antibodies and/or stained in PBST overnight at 4uC with constant agitation. Hearts were washed 3610 mins in PBST, incubated with secondary antibodies for 1-hour at room temp, washed 3610 mins in PBST plus final wash in PBS alone. Hearts were then mounted in 50% glycerol (Sigma, Poole, Dorset, UK) in PBS and visualised using a Zeiss Axioskop 2 mot plus fluorescent microscope and images were taken using Openlab imaging software. Anti-bPSintegrin (1:10 of hybridoma supernatant (100 mg/ml), Developmental Studies Hybridoma Bank) was used to stain cell junctions, AlexaFluor 546-conjugated phalloidin (13.2 nM) to stain F-actin (Invitrogen, Paisley, UK), and DAPI (1 ng/ml) to stain nuclei (Sigma, Poole, Dorset, UK). The hybridoma developed by [24] 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.

Measurement of distance between neighbouring cardiomyocytes
ImageJ (http://rsbweb.nih.gov/ij/) was used to measure the distance between neighbouring cardiomyocytes. In ImageJ, the 'straight line' drawing tool was selected and a line was drawn between neighbouring cardiomyocytes. The scale was calibrated (Analyse.Set Scale) to measure in microns, and distances were measured using the 'Measure' function (Analyse.Measure, or CTRL+M) in ImageJ. Four measurements were taken for each fly heart, and up to 7 flies were used for each genotype -resulting in 12-28 measurements from 4-7 independent flies for each genotype. Data was then grouped for each genotype and mean +/2 SEM was calculated.

Heart preparation and functional assays
It is necessary to surgically expose the fly heart for detailed analysis of cardiac contractions -making them accessible for video microscopy [25]. We used the previously published heart assay and analysis methodology [26], and used between 6 and 11 female flies, aged 5-7 days old for all analyses. In brief, the ventral abdomen was dissected in artificial haemolymph (108 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 8 mM MgCl 2 , 1 mM NaH 2 PO 4 , 4 mM NaHCO 3 , 5 mM HEPES, 10 mM sucrose, 5 mM trehalose) to expose the fly's heart, which lines the dorsal cuticle, and digital movies of the heart over body segments A2 and A3 were taken (video acquisition by Pinnacle Software (Pinnacle Systems, Iver Heath, UK)). Video footage was analysed using a MatLabbased image analysis program, M-modes were generated and cardiac parameters including heart rate, arrhythmias, heart periods, diastolic and systolic diameters, and fractional shortening were quantified [27]. Data was then grouped for each genotype and mean +/2 SEM was calculated.

Generation of Fit1 D161 deletion line
The P-element inserted in Fit1 KG05576 was mobilized by following standard crosses. Mobilization events were screened by PCR using different primer combinations in the genomic DNA flanking the P-element -Fit1 D161 (59-CATCCGTTCCGA-TAAGTTCG-39 and 59-CCAACAGCTCCTCCTTCTTG-39) resulted from the imprecise excision of the P-element and creates an out-of-frame deletion which removes 1685 bp (629 bp from 59 UTR region and 1055 bp from coding sequence)(deletion corresponds to nucleotides 67-1752) and retains a 12 bp fragment (CATGATGAAATA) from the start of the 11,467 bp P-element.

Statistics
One-way analysis of variance (ANOVA) followed by Tukey's post-hoc test was used to identify differences between three or more means derived from uneven sample sizes. The student's unpaired t-test was used to identify significant differences between two means of uneven sample size. A statistical difference of P,0.01 was regarded as significant.

Fermitins regulate cardiac syncytium development
To assess the effect of Fermitin knock-down on adult heart morphology, the cardiac Hand-GAL4 driver was used to disrupt gene expression. In control flies the cardiomyocytes display a characteristic spiral myofibrillar arrangement that can be visualised by phalloidin staining, with b-integrin staining junctions between adjacent cardiomyocytes (Figure 2, Hand-Gal4; w 1118 ). Silencing Fit1 using the RNAi line from the Vienna Drosophila RNAi centre (VDRC [28]) led to loss of the cardiac syncytium and loss of contiguous b-integrin staining between adjacent cardiomyocytes) (Fig. 2, Hand-Gal4; UAS.Fit1 VDRC ). When both Fit1 and Fit2 were silenced, a more severe phenotype developed (Fig. 2, Hand-Gal4; UAS.Fit1/Fit2 VDRC ), with cardiomyocytes exhibiting a 'rounded-up' phenotype, comparable to that resulting from the silencing of the b-integrin myospheroid in cardiomyocytes (Fig. 2, Hand-Gal4; UAS.Mys VDRC ). In contrast, when Fit1 was silenced using the Harvard TRiP (Transgenic RNAi Project) RNAi line (Fig. 2, Hand-Gal4; UAS.Fit1 TRiP ), the cardiac syncytium developed normally and the cardiomyocytes exhibited a cytoskeletal morphology similar to that seen in controls. Similarly, when Fit2 was silenced on its own, the cardiac syncytium developed normally (Fig. 2, Hand-Gal4; UAS.Fit2 VDRC ). The discrepancy between the phenotype caused by the Fit1 VDRC and Fit1 TRiP RNAi lines can be explained by the VDRC line targeting both Fit1 and Fit2 message (see Fig. S2), whereas the TRiP line targets only Fit1.
To confirm the effect of Fermitin knock-down on the heart the distance between neighbouring cardiomyocytes was quantified (Fig. 3A). In control flies and flies where no effect of an RNAi was observed (w 1118 flies crossed with the UAS.Fit1 VDRC line in the absence of the Hand-Gal4 driver), the distance between adjacent cardiomyocytes were never more than 3 mm. In contrast, the distance between cardiomyocytes when Fit1 was silenced using the VDRC line, was ten-fold greater (34.263.7 mm, (P,0.001)). When Fit2 was silenced there appeared to be a small effect on distances between cardiomyocytes, however this did not reach statistical significance.

Silencing Fermitin expression disrupts heart contractility
The cardiac syncytium in vertebrates is a critical determinant of the heart's contractility. It was therefore reasoned that loss of the syncytium caused by Fermitin knock-down in Drosophila would dramatically impact cardiac function and this hypothesis was tested in adult flies using videomicroscopy. It was found that the cardiomyocytes of Fit1 VDRC silenced flies (in which both Fit1 & Fit2 are silenced) beat at the same rate as controls (w 1118 and w 1118 crossed with Fit1 VDRC ) and flies in which Fit2 had been knocked down ( Figure 3C, P.0.05). However the distance the heart moved between diastole and systole (fractional shortening) was greatly reduced in Fit1 VDRC silenced flies, compared to all other genotypes (see the M-mode in Figure 3B and the quantified data in Figure 3D; P,0.001). Consistent with this finding, the end systolic diameter was greater in the Fit1 VDRC silenced flies than the other genotypes ( Figure 3E, P,0.01); whereas diastolic diameter was similar to the controls (P.0.05). Examples of the beating hearts from the different genotypes and the disruption caused by the UAS.Fit1 VDRC line can be seen in videos S1 to S4.

Synchrony between cardiomyocytes is reduced when Fermitin expression is silenced
It was clear from the functional studies that cardiomyocytes in the Fit1 VDRC silenced flies were capable of beating at a frequency similar to that of controls, but that fractional shortening was Fermitins regulate coupling between cardiomyocytes. The monochrome micrographs show the adult Drosophila heart (segments A2 to A5) stained with phalloidin, oriented with the anterior region to the left. Genes were silenced using the Hand-Gal4 enhancer which drives expression in cardiomyocytes (as well as pericardial nephrocytes and enterocytes of the gut) and abnormal cardiomyocyte/heart morphology is highlighted by the arrows. The colour panels show higher magnification micrographs of at least two cardiomyocytes stained with phalloidin (red) and antibodies to the Drosophila b-integrin, myospheroid (green); arrowheads indicate normal integrin staining between cardiomyocytes; asterisks indicate an abnormal staining pattern associated with loss of cardiomyocyte junction integrity. A wild type phenotype (Hand-Gal4; w 1118 ), is characterised by contiguous cardiomyocytes and b-integrin staining between cardiomyocytes. doi:10.1371/journal.pone.0062958.g002 dramatically reduced. It was also clear that cardiomyocytes were no longer in close contact with each other and it was therefore reasoned that synchrony between cardiomyocytes had been lost. This loss of synchrony is a novel phenotype in Drosophila and was quantified by analysing movement movies from videomicroscopy footage that depict the motion of the heart during the cardiac cycle. To provide a 2-dimensional example of heart movement we explored the use of 'actograms', which are used to depict cyclical events over a given time period, especially in the field of biological rhythm research. In these images, a time-series of a measured or recorded parameter (e.g. behaviour, movement or gene expression) is presented as a raster plot (i.e. plotted left to right, top to bottom). A rhythmic system will produce a geometric pattern in a raster plot, whereas a random pattern is produced when a system is asynchronous. To illustrate the phenotype of the abnormal hearts in this way we concatamerised stills from the movement movies to generate raster plots of 10.8 seconds of heart activity and compared the plots from control flies (w 1118 , UAS.Fit1 VDRC ) with those from the Fermitin silenced flies (Hand-Gal4; UAS. Fit1 VDRC , Fig. 4A, B). Using this method it was easy to identify the loss of rhythmicity as a transition from the geometric pattern of the control heart to the disordered plot from the Fermitin silenced fly. In order to quantify the degree of synchrony in different areas of the heart, two independent regions of the movement movie were compared to see if changes in pixel intensity (i.e. a measurement of cardiac motion) were in synchrony or not. Using simple regression analysis of two independent regions of the contracting heart in the movies, it was confirmed that synchrony was significantly reduced in Hand-Gal4; UAS.Fit1 VDRC flies compared to control flies (w 1118 , UAS.Fit1 VDRC ; mean R 2 values, of 0.3460.11 and 0.8160.03, respectively (P,0.001, Figure 4C & D). Examples of the beating hearts and movement movies from the different genotypes can be seen in videos S5 to S8.

Fit1 is dispensable for syncytium development
The discrepancy between the phenotypes caused by the Fit1 VDRC and TRiP lines was likely due to the VDRC also targeting Fit2. This 'double whammy' would effectively knock-down both Fit genes and inhibit any functional compensation occurring between the two. To test this hypothesis, a Fit1 mutant was generated via D2-3 mutagenesis. The P-element in Fit1 KG05576 allele was excised along with 1.7 kb of the Fit1 gene in the 39 direction to create the Fit1 D161 line ( Figure 5A,B). Fit1 D161 flies were viable and had normal heart morphology ( Figure 5). However, abnormal heart morphology was observed when Fit2 was silenced (Hand-Gal4; UAS.Fit2 VDRC ) in Fit1 D161 mutant cardiomyocytes ( Figure 5C), despite the same Fit2 RNAi line, when used on its own, having no effect on heart function (compare Figs. 2 & 3 with Fig. 5C). These findings were confirmed when the distance between neighbouring cardiomyocytes was quantified ( Figure 5D). Thus, loss of Fit1 or reduced expression of Fit2 is not sufficient to cause a heart phenotype, indicating that functional redundancy exists between Fit1 and Fit2. Only when the expression of both genes is affected does a severe cardiomyopathy develop in Drosophila.   Synchrony and contractility are disrupted when Fit2 expression is silenced in Fit1 D161 mutant hearts Synchronicity between discrete regions of the Drosophila heart was quantified and control (w 1118 ) flies displayed regular synchronous cardiac contractility. Fit1 D161 , and UAS.Fit2 VDRC ; Fit1 D161 flies without Hand-Gal4 displayed comparable synchronicity to w 1118 controls. In contrast, when Fit2 was silenced (Hand-Gal4; UAS.Fit2 VDRC ) in Fit1 D161 mutant cardiomyocytes there was a significant loss of synchronicity between these regions of the heart ( Figure 6A,B). There was no significant difference in heart rate between genotypes, whereas fractional shortening was significantly reduced in the Hand-Gal4; UAS.Fit2 VDRC Fit1 D161 mutants ( Figure 6C,D). The Hand-Gal4; UAS.Fit2 VDRC Fit1 D161 mutants also had significantly increased end systolic and end diastolic diameters relative to the control and other genotypes ( Figure 6E, P,0.01). Examples of the beating hearts and movement movies from the different genotypes can be seen in videos S9 to S16.

Discussion
The current findings unequivocally demonstrate that cardiomyocytes must express members of the Fermitin/Kindlin family in order to develop as a functional syncytium. When Drosophila cardiomyocytes fail to couple together to form a cardiac syncytium, synchronous contractions and fractional shortening of the adult heart are significantly reduced, despite individual cardiomyocytes remaining myogenic. Studies of vertebrate hearts suggest a role for Kind2 in cardiac development and function, however these studies are limited by the embryonic lethality of Kind2 knock-out in the mouse model, and the lack of tissue specific Kind2 silencing in the fish model [11,12]. The aberrant phenotype caused by silencing orthologs of Kind2 in the cardiomyocytes of an invertebrate demonstrates that the protein's role in cardiac development has been evolutionarily conserved and also reiterates the validity of using Drosophila to study genes relevant to mammalian cardiac physiology.
The aberrant heart phenotype caused by Fermitin knock-down in Drosophila reproduces many aspects of the cardiomyopathy caused by morpholino knock-down of Kind2 in zebrafish [11], and supports the authors' conclusion that Kind2 regulates heart development. Morpholino-knockdown in zebrafish disrupted angiogenesis but also affected intercalated disc formation in cardiomyocytes and significantly reduced fractional shortening. Previous studies in Drosophila have identified the Fermitins as mediators of muscle assembly [29], and the current findings extend this observation to include the assembly of a functional cardiac syncytium. Integrins and related adhesion proteins (ILK, Talin, Laminin A, Syndecan, Robo, and Slit) are necessary for heart assembly in Drosophila [30][31][32]. The current work is the first to demonstrate the functional consequences for the Drosophila heart when assembly of the cardiac syncytium is abnormal.
Human heart diseases attributed to intercalated disc dysfunction include arrhythmogenic right ventricular cardiomyopathy caused by mutations in D proteins (e.g. Desmoplakin, Plakophilin-2, and Desmoglein-2) [9], and hypertrophic and dilated cardiomyopathy caused by a missense mutation in the integrin-binding adhesion protein, Vinculin [10]. Integrins and related binding proteins (e.g. Vinculin and Talin) were also identified at intercalated discs in adult human cardiac tissue [33,34]. Thus, there is a need to develop genetically tractable models with which to study the role of these proteins in cardiac development and function.
Intercalated disc-like structures have been identified between Drosophila cardiomyocytes, however there have been no empirical studies examining the molecular components nor the functional role of this structure in the heart [35]. The current data provide evidence that the b-integrin encoded by myospheroid is localised at the boundaries between adjacent cardiomyocytes, and also indicate that the development of a functional cardiac syncytium is strictly dependent on myospheroid as well as the expression of Fermitins. It would be informative to further assess the integrity of the cardiomyocyte junctions in both wild type and mutant flies using antibodies to cadherin and b-catenin [36]. The cardiomyopathy caused by silencing Fit1 and Fit2 phenocopied myospheroid knock-down and is therefore consistent with the Fermitins regulating b-integrin signalling and promoting cardiomyocyte coupling during development [31,35]. There is evidence from mammalian models that Kindlins, via the FERM F3 subdomain, interact directly with the cytoplasmic tail of b-integrins [12,18]. The FERM domain of Drosophila Fermitins is conserved from flies to humans and it is therefore speculated that Fermitins directly interact with the b-integrin myospheroid to form a signalling complex required for cardiomyocyte coupling. No study has yet manipulated the expression of mammalian Kind2 solely in the cardiomyocytes, however it is predicted that a phenotype would develop that is similar to that seen in ventricular cardiomyocytetargeted b1-integrin knockout mice [7]. Histological studies of the b1-integrin KO hearts revealed significant disruption of myofibrils and intercalated discs, with subsequent onset of dilated cardiomyopathy.
There was evidence that Fit2 expression could compensate for Fit1 in the Drosophila cardiomyocyte model. A Fit1-null mutant (Fit1 D161 ) was generated that had normal heart morphology and function, indicating that Fit1 is not required for the development of a functional syncytium. This finding also suggested that Fit2 might be compensating for the loss of Fit1. When Fit2 was silenced in flies expressing Fit1, there was little impact on the heart. However, when the same Fit2 RNAi line was expressed in the Fit1null cardiomyocytes, it caused a severe cardiomyopathy. It can therefore be concluded that Fit2 was responsible for syncytium development in the absence of Fit1.
In summary, the findings identify Fermitins as important mediators of cardiomyocyte coupling and heart function in Drosophila. Given the evolutionary conservation between invertebrate Fermitins and mammalian Kindlins, the evidence supports the hypothesis that disruption of cardiomyocyte KIND2 may lead to abnormal intercalated discs and cardiomyopathies in humans. Figure S1 The Kindlin-2 FERM F3 subdomain, important for integrin-binding, is highly conserved in Drosophila Fermitins.

Supporting Information
(DOCX) Figure S2 The Fit1 VDRC RNAi targets both Fit1 and Fit2 gene expression in the Drosophila heart. (DOCX) Video S1 Heart of w 1118 control. Note: All videos show a representative 10 second sequence of the beating heart or movement movie derived from the beating heart. Videos S1 to S4, correspond to the data presented in Figure 3; videos S5 to S8 correspond to the data presented in Figure 4, and videos S9 to S16 correspond to the data in Figure 6. Video S10 Movement movie of w 1118 heart. (AVI) Video S11 Heart of Fit1 D161 .