Identification and Functional Characterization of Cardiac Pacemaker Cells in Zebrafish

In the mammalian heart a conduction system of nodes and conducting cells generates and transduces the electrical signals evoking myocardial contractions. Specialized pacemaker cells initiating and controlling cardiac contraction rhythmicity are localized in an anatomically identifiable structure of myocardial origin, the sinus node. We previously showed that in mammalian embryos sinus node cells originate from cardiac progenitors expressing the transcription factors T-box transcription factor 3 (Tbx3) and Islet-1 (Isl1). Although cardiac development and function are strikingly conserved amongst animal classes, in lower vertebrates neither structural nor molecular distinguishable components of a conduction system have been identified, questioning its evolutionary origin. Here we show that zebrafish embryos lacking the LIM/homeodomain-containing transcription factor Isl1 display heart rate defects related to pacemaker dysfunction. Moreover, 3D reconstructions of gene expression patterns in the embryonic and adult zebrafish heart led us to uncover a previously unidentified, Isl1-positive and Tbx2b-positive region in the myocardium at the junction of the sinus venosus and atrium. Through their long interconnecting cellular protrusions the identified Isl1-positive cells form a ring-shaped structure. In vivo labeling of the Isl1-positive cells by transgenic technology allowed their isolation and electrophysiological characterization, revealing their unique pacemaker activity. In conclusion we demonstrate that Isl1-expressing cells, organized as a ring-shaped structure around the venous pole, hold the pacemaker function in the adult zebrafish heart. We have thereby identified an evolutionary conserved, structural and molecular distinguishable component of the cardiac conduction system in a lower vertebrate.


Introduction
The cardiac conduction system comprises several components, amongst which the sinus node, the site of electrical impulse generation, and the Purkinje fibers transducing the impulse rapidly through the myocardium [1]. The sinus node harbors specialized pacemaker cells which, due to regular and spontaneous membrane depolarization, generate the electrical signal necessary to induce cardiomyocyte contractions [2]. Although the sinus node was described histologically and functionally more than a century ago [3], the molecular regulators required for pacemaker cell differentiation and function are not fully understood. Nonetheless, several recent developments have provided new insights. These include identification of the embryonic origins of the sinus-and atrioventricular nodes [4,5] and of several transcriptional regulators involved in their development (reviewed in [1]). A major advance for the field was the identification of T-box transcription factor 3 (Tbx3) in pacemaker cells, and the subsequent demonstration that it is required for sinus-and atrioventricular node development and postnatal homeostasis [6,7]. Other transcriptional regulators that have been identified for their role in sinus node development are Nkx2.5, Tbx5, Pitx2 and Shox2 [8][9][10][11][12][13]. The molecular signature of the mouse sinus node primordium has been confirmed in human embryonic hearts, indicating evolutionary conservation of the developmental mechanism [14].
The LIM domain transcription factor Isl1 is expressed in the mammalian cardiac progenitor cells [15,16]. Isl1 expression gradually decreases during differentiation and is eventually lost in mature cardiomyocytes [15,17,18], except for myocytes pertaining to the sinus node [19,20]. Due to structural heart defects and early lethality of mouse embryos deficient for Isl1, its putative role in the developing and mature sinus node has remained elusive.
Although the presence of specialized conduction system components in the heart of lower vertebrates has been suggested by functional analysis [21][22][23], their identification has remained elusive due to the lack of morphologically distinctive structures and molecular markers.
Our research presented in this manuscript resolved this issue by describing the first molecular and structural identification of specialized cardiac pacemaker cells in the embryonic and adult zebrafish heart, utilizing a combination of in vivo microscopic examination, 3D gene expression pattern reconstruction, reporter transgenics and ex vivo electrophysiology. Our findings establish that Islet-1 is required for pacemaker activity in the embryonic heart and that Islet-1 marks the pacemaker cells in the adult heart, which represents a previously unappreciated role for Isl1 in the cardiac conduction system.

Results and Discussion
Cardiac pacemaker activity is affected in Islet-1 mutant hearts Zebrafish, embryos lacking functional Isl1 protein are immobile and display reduced heart rate (bradycardia) at 2 days post fertilization (dpf) [24]. Unlike the mouse Isl1 mutant hearts that fail to loop and lack recognizable chambers [15], zebrafish isl1 mutant hearts loop normally and are morphologically indistinguishable from their wild-type siblings at 2 days post fertilization (dpf), allowing their functional characterization. To identify the primary defect responsible for the previously reported bradycardia phenotype we further investigated the heart rhythm in isl1 mutant embryos by high-speed video imaging combined with functional image analysis. Both the isl1 mutant and wild-type sibling embryos showed initiation of contraction in the inflow region (venous pole) continuing in the atrium, followed by rapid contraction of the ventricle (Movies S1 and S2). To quantify heart rhythm, we drew kymographs for atrium and ventricle of sibling and mutant hearts using a 1 pixel-thick collection of lines positioned perpendicularly to the blood flow (Fig. 1A, white dotted lines). The resulting kymographs (Fig.1B) readily confirmed the bradycardia of the isl1 mutant, as the distance between two dotted lines encompassing a full cardiac cycle (Fig.1B, white double arrows) is much longer in the mutant than in the sibling. Moreover, the kymograph-based quantification of the cardiac cycle (interval necessary for one complete contraction) illustrates the slow and irregular heart rhythm in isl1 mutant embryos (Fig. 1C). Cardiac cycles of the imaged isl1 mutants oscillated between about 450 ms and 800 ms, showing not only high variability within a single embryo (Fig. 1C, box-whisker plots for mutants; maximal and minimal values vary of about 200 ms for all mutants analyzed), but also between different individuals (Fig.1C, mutants 1,2 and 3 display average cardiac cycle period times of 586 ms, 661 ms and 494 ms respectively over the recording time). In comparison, siblings displayed a maximal variation in cardiac cycle measured at the atrium of about 20 ms and average cardiac cycle periods of 325 ms (Fig. 1C, Movie S1). We never observed uncoupling of the atrial and ventricular contraction, fibrillation or atrioventricular block, as every atrial contraction was followed by a ventricular contraction ( Fig. 1B; Movie S2).
As development proceeded, the severity of the heart beat phenotype increased. We frequently observed a sinus block in isl1 mutant hearts at 3-4 dpf resulting in the absence of atrial and ventricular contraction for 10-20 seconds (Fig. 1D). Altogether, the combination of phenotypes displayed by the isl1 mutant is compatible with defective initiation of contraction, suggesting faulty pacemaker activity.
Isl1 is expressed at the putative sinoatrial boundary of the embryonic and adult heart In mammals, including humans, the primary pacemaker function is held by the sinus node, which resides at the junction of the superior caval vein and right atrium [25][26][27][28]. For example, in mouse, the right sinus venosus was shown to form the sinus node, which includes the venous lining of the right venous valve [4,13]. Isl1-expressing cells are found in the developing and mature mammalian sinus node [14,19,20].
We hypothesized that Isl1 expression marks the pacemaker tissue in the zebrafish heart, as no molecular markers for the zebrafish sinus node or the pacemaker cells within it have been identified to date. Using an antibody recognizing both zebrafish Isl1 and Isl2 [29] we observed few Isl-positive cells in both the dorsal and ventral regions of the proposed sinoatrial junction at 2 dpf ( Fig. 2A, B, C, D), consistent with the proposed pacemaker function at the sinoatrial junction in the zebrafish heart [21]. The Isl-expressing cells expressed Tg(myl7:eGFP), which marks cardiomyoctes. Isl expression at the sinoatrial junction was detected continuously during development and was maintained in the same region in the adult zebrafish heart (Fig. 2E, F, G, H). This indicates that Isl expression in the zebrafish proposed sinoatrial junction is constant between embryonic stages and adulthood. We continued by molecular characterization of the proposed sinoatrial junction region in the adult heart using in situ hybridization (ISH). Contrary to what has been observed in mammals, the zebrafish sinus venosus did not express the myocardial marker myl7. Consequently, the upstream border of myocardium muscle was at the venous valves and atrium (Fig. 3A, B). We observed that isl1 expression was confined to the myocardium located at the base of the venous valves (Fig. 3C). The zebrafish ortologue of hcn4 encodes a member of the family of ion channels responsible for the hyperpolarization-activated current, I f , in pacemaker cells. Its expression is enriched in pacemaker tissue of mammalian hearts [28,[30][31][32][33]. In the zebrafish adult heart hcn4 had a broader expression pattern than isl1 (Fig. 3D), similar to observations made in the mammalian embryonic heart [13,14]. Expression of the two genes overlapped at the sinoatrial junction (Fig. 3D). A 3D reconstruction based on a series of ISH on sagittal adult heart sections revealed that isl1 expression is confined to a ring-like structure within the myocardial tissue at the base of the venous valves, around the proposed sinoatrial junction (Fig. 3F). In the mammalian heart expression of Nppa is specific for the fast conducting working myocardium of the atrium and ventricle but is not expressed in the slow conducting primitive myocardium of the pacemaker tissue at the sinoatrial junction [13]. We observed a very similar mutually exclusive expression pattern of nppa and isl1 at the proposed sinoatrial junction of the zebrafish heart (Fig. 3E). Furthermore, since bmp4 and tbx2b are expressed at the venous pole of the embryonic heart ( [24] and Figure S1) and since the bmp4-tbx2b regulatory axis suppresses chamber differentiation and nppa expression, we analyzed their expression in the adult heart. Interestingly, expression of bmp4 and tbx2b was maintained in the myocardium of the proposed sinoatrial junction of the adult heart and restricted to the base of the venous valves or the entire venous valve tissue, respectively (Fig. 3G, H). In summary, we identified an isl1+cell population in the hcn4+tbx2b+nppa-negative myocardium that is organized in a ring-shape at the proposed sinoatrial junction. Together with the above-described observation that isl1 mutants display irregular heart rhythms, this suggests that isl1 expression identifies cardiac pacemaker cells in the zebrafish heart.
Using an Isl1-LacZ knock-in model, Isl1/LacZ activity was observed in the sinus node of the adult mouse heart [20]. Corroborating these findings we detected endogenous Isl1 expressing cells in the adult sinus node of the mouse heart using an anti-Isl antibody ( Figure S2). Using a similar approach we also detected Isl1 expressing cells in the sinus node of the adult human heart ( Figure S3). Interestingly, we observed that only a subpopulation of sinus node cells expresses Isl1, suggesting that the Isl1 expressing cells have different properties compared to the isl1negative sinus node cells.

Isl1-expressing cells display electrical pacemaker activity
To functionally characterize the Isl1+ cells in the zebrafish heart, we generated an Isl1-GFP reporter line using the binary Gal4/UAS expression system ( Figure S4). To validate the reporter line we confirmed that all GFP+ cells were co-labeled by Isl1 immunostaining (Fig. 3I, J, K' arrowheads). In vibratome sections GFP+ cells visualized by fluorescent antibody labeling were located in bilateral cell populations at the proposed sinoatrial junction (Fig. 3I, J, K, arrows), conform the ISH staining. Neighboring GFP+/Isl1+ cells displayed cytoplasmic protrusions, which may connect them between each other (Fig. 3K', arrowheads). 3D reconstruction of confocal image stacks revealed that Isl1+ pacemaker cells are interspersed with Isl-negative cells. However, they form a coherent structure (Fig. 3L). It is known that isolated groups of cells with residual pacemaker activity, such as remnant embryonic nodal atrioventricular canal myocytes, may cause cardiac arrhythmias (reviewed in [34,35]). The cell-to-cell interconnection could therefore be an essential feature for proper pacemaker function, likely coordinating a synchronous activation of the myocardium. Moreover, the expression of myosin is low in Isl+ cells, which is supportive for a primitive myocardial identity of these cells, typical for pacemaker cells ( Figure S5) [27,36,37].
To elucidate whether the GFP+ cells localize to the region in which the electrical activation is initiated, we performed optical mapping of epicardial activation patterns on adult zebrafish hearts. First moment of atrial activation corresponded with the localization of the GFP+ cells (Fig. 4A, B, C), which is compatible with pacemaker function for isl1 expressing cells.
To unequivocally discriminate whether indeed isl1 expression marks cells with pacemaker activity, dissociated single GFP+ and GFP-cells from micro-dissected sinoatrial tissue were patch-clamped (Fig.4D). All measured GFP+ cells (n = 6) were spontaneously active, while all measured GFP-cells (n = 8) were typically quiescent. In GFP-cells, action potentials could be elicited by current pulses through the patch pipette. Figure 4D shows typical spontaneous action potentials of a GFP+ cell as well as typical action potentials recorded from a GFP-cell that was stimulated at 3 Hz, i.e. with a cycle length similar close to that of the GFP+ cell ( Table 1). All parameters, except for action potential duration at 90% repolarization (APD 90 ), differed significantly between both cell types (Table 1). GFP-cells had a stable resting membrane potential of -79.061.6 mV, while GFP+ cells showed a spontaneous diastolic depolarization rate (64617 mV/s) with a maximum diastolic potential of 265.063.0 mV. In GFP+ cells, this diastolic depolarization resulted in pacemaker activity with an intrinsic cycle length of 378658 ms. In GFP+ cells, the maximum upstroke velocity was typically low (7.462.6 V/s) as opposed to GFP-cells (112614 V/s). In both cell types, action potentials overshot the zero potential value, but the action potential amplitude was higher in GFP-cells. Action potentials of GFPcells repolarized earlier and faster, resulting in shorter APD 20 and APD 50 . Thus, the results obtained on the optical mapping and patch-clamp experiments revealed that while GFP-cells display characteristics specific to cardiac chamber myocytes, pacemaker activity resides in Isl+/GFP+ cells.

Conclusions
We present here the molecular and functional identification of cardiac pacemaker cells in the embryonic and adult zebrafish heart. The embryonic cardiac expression pattern (Fig.2) and knockout phenotype of Isl1 in zebrafish (Fig.1) hinted at a role for this gene in pacemaker function. Indeed, analysis of 3D reconstructions of expression pattern and reporter transgenics in adult fish showed a previously unidentified ring-shaped region of Isl1 expression within the myocardium at the proposed sinoatrial junction of the mature zebrafish heart (Fig.3). Optical mapping of  (Fig. 4). Altogether, our data allow us to establish that (1) Isl1 is the first identified molecular marker for pacemaker cells in the zebrafish heart, (2) the functional pacemaker of the adult zebrafish heart is organized as a ring around the venous pole and (3) expression of Isl1 in the pacemaker cells of the adult heart is conserved from fish to human.
Cells pertaining to the primary pacemaker structure in zebrafish and mammals share a number of molecular markers, relative position in the heart, and functional identity. However, while in zebrafish the Isl1-expressing pacemaker cells are few and organized in a ring-shaped structure of interconnected cells in the venous valves, the mammalian sinus node is a more compact, spindle-shaped (also referred to as comma-shaped) clearly demarcated structure. Although at this stage we do not know whether the ring-shaped pacemaker can be extended to other lower vertebrates, the diverging and unpronounced structure of the zebrafish sinus node and the absence of molecular markers until now may explain why a pacemaker structure in lower vertebrates has not previously been identified.
Finally, the 'primitive' myocardial identity of Isl1+ cells in adult zebrafish is intriguingly accompanied by a capacity to spontaneously depolarize, which is absent in working cardiomyocytes. Future work focusing on Isl1 and its gene targets will help to elucidate whether Isl1 expression in the pacemaker cells is necessary to maintain the primitive myocardial fate of these cells, similarly to its suggested role in cardiac progenitor cells [15,17,18,38]. Alternatively, Isl1 could be required to drive a pacemaker gene program, in a similar fashion to what was shown for Tbx3 [7].

Ethics statement
All animal and human work conformed to ethical guidelines and was approved by the relevant local animal ethics committees.

Animal work
Animal Experimental Committee Secretariaat DEC/KNAW Koninklijke Nederlandse Akademie voor Wetenschappen Postbus 19121 1000 GC Amsterdam

Human Tissue Samples
The study met the criteria of the code of proper use human tissue that is used in the Netherlands for the use of human tissue. The study was approved by the scientific advisory board of the biobank of the University Medical Center Utrecht (RP2012-03).

Zebrafish lines
Fish used in this study were kept in standard conditions as previously described [39]. The Tg(myl7:gfp) and isl1 mutant line isl1K88X (isl1 SA0029 ) were described previously [24,40]. Generation of the Tg(Isl1BAC:GalFF; UAS:GFP) is described in more detail below. All animal and human work conformed to ethical guidelines and was approved by the relevant local animal ethics committees.

High-speed imaging and analysis
2 Dpf and 4 dpf isl1K88X mutant and sibling embryos were mounted in 0.25% agarose (Life Technologies BV) prepared in E3 medium embryonic medium with 16 mg/ml 3-amino benzoic acid ethylester. Embryonic hearts were imaged with a Hamamatsu C9300-221 high speed CCD camera (Hamamatsu Photonics, Hamamatsu City, Japan) at 150 fps mounted on a Leica AF7000 microscope (Leica Microsystems GmbH, Wetzlar, Germany) in a controlled temperature chamber (28.5uC) using Hokawo 2.1 imaging software (Hamamatsu Photonics GmbH, Herrsching am Ammersee, Germany). Image analysis was carried out with ImageJ (http://rsbweb.nih.gov/ij/). Statistical analysis and drawing of the box-whisker plot were carried out in Excel 2007 (Microsoft, Redmond, WA, USA).

In situ hybridisation and immunohistochemistry
ISH on embryos was carried out as previously described [39]. ISH on adult heart tissue was carried out as previously described [43] with minor modifications. 3D reconstructions of serial ISH-labeled sections were performed as described in [44]. Immunohistochemistry was carried out as previously described [24]. Embryos and adult hearts for immunocytochemistry were fixed in 2% paraformaldehyde, embedded in 3% agarose/1% gelatine and sectioned at 100 mm thickness. The primary antibodies used were mouse anti-Isl1 (Developmental Studies Hybridoma Bank, Iowa  City, IA, USA,

Patch clamp experiments
Action potentials were recorded by the amphotericin-perforated patch-clamp technique using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, USA). Action potentials of GFP+ cells were low-pass filtered (cut-off frequency 5 kHz) and digitized at 10 kHz; action potentials of GFP-cells at 5 kHz and 20 kHz, respectively. Potentials were corrected for the estimated liquid junction potential [46]. Data acquisition and analysis were accomplished using custom software. Patch pipettes (borosilicate glass; resistance 3-4 MV) were heat polished and filled with pipette solution containing (in mM): K-gluc 125, KCl 20, NaCl 10, amphotericin-B 0.22, and HEPES 10; pH was set to 7.2 with KOH. Action potentials in GFP-negative cells were elicited at 3 Hz by 3-ms, <1.2x threshold current pulses through the patch pipette. Parameter values obtained from 10 consecutive action potentials were averaged. Figure S1 Expression of tbx2b at the venous pole in embryonic heart. Expression patterns by mRNA in situ hybridization of tbx2b and nppa in 2 dpf embryos. Expression of tbx2b at the venous pole (blue staining indicated with arrowheads) does not overlap with nppa expression (red staining), which is confined to atrium and ventricle chamber myocardium. Pictures shown are ventral views with anterior to the top. Movie S1 Recording of the heart beating of an isl1 sibling embryo at 2 dpf. Ventral view (anterior to the left). Note the speed and regularity of the contractions (also see Fig.1B). Playback speed (150 fps) is set to match recording speed to render natural speed of the embryo's heart beat.

(AVI)
Movie S2 Recording of the heart beating of an isl1 2/2 embryo at 2 dpf. Ventral view (anterior to the left). Note the reduced speed of the contractions and their irregularity (also see Fig.1B). Playback speed (150 fps) is set to match recording speed to render natural speed of the embryo's heart beat. (AVI)