Functional Modulation of Cardiac Form through Regionally Confined Cell Shape Changes

nppa Expression Distinguishes Chamber Curvatures in Zebrafish

Chamber emergence occurs rapidly in the zebrafish embryo: the linear heart tube (LHT), which contracts and drives circulation by 24 h post-fertilization (hpf), transforms into two morphologically evident chambers, the ventricle and the atrium, by 48 hpf (Figure 1) [30]. As in other vertebrate hearts [10–12], each chamber is bean-shaped, with a distinct OC and IC (Figure 1A). In chick and mouse, expression of the natriuretic peptide precursor a (nppa) gene, also known as atrial natriuretic factor (anf), is restricted to the OC of the ventricle and atrium [10,31]. Similarly, we find that zebrafish nppa [28] marks OC myocardium as chambers emerge. Expression of nppa is first detectable by 24–28 hpf in a domain on the right side of the ventricular portion of the LHT (Figure 1C–1E, arrow) and a domain on the left side of the atrial portion of the LHT (Figure 1E, arrowhead). These domains are positioned to become the future ventricular and atrial OCs. As ballooning proceeds, nppa expression remains regionally restricted, being localized to the OC myocardium of both chambers while absent from the IC and atrioventricular canal regions (Figure 1G–1I). Thus, in parallel to the process observed in higher vertebrates, chamber emergence in zebrafish features the development of morphological and molecular differences between OC and IC myocardium.

Regionally Confined Cell Shape Changes Accompany Chamber Emergence

Despite the rapid emergence of the OC and IC during zebrafish heart development, it is not known which cellular mechanisms are responsible for establishing chamber curvatures during this time frame. We chose to test the hypothesis that the morphologically and molecularly distinct OC and IC regions also possess distinct cellular characteristics. Analyses of chick hearts have indicated that OC cells increase their apical surface area and volume during looping and ballooning [17,23]. To determine whether cell morphology changes contribute to the formation of chamber curvatures in zebrafish, we assessed the shape and size of cardiomyocytes as chambers emerge, focusing our analysis on the ventricle (Figure 2). We examined cell shape and surface area by using phalloidin to outline cells in embryos carrying the transgene Tg(cmlc2:egfp), which is expressed throughout the myocardium [32] (Figure 2A and 2B). Additionally, in order to view multiple cell surfaces and to examine cell morphology without fixation, we imaged live cells expressing Tg(cmlc2:dsredt4) mosaically within Tg(cmlc2:egfp)-expressing hearts (Figure 2E–2L; see Materials and Methods).

First, we examined cardiomyocyte morphology within the ventricular portion of the LHT (24–28 hpf; Figure 2A, 2C–2E, and 2I). Surface area measurements and shape assessments demonstrate that all cells in this region exhibit small and rounded morphologies (Figure 2A, 2C–2E, and 2I). Thus, even though nppa expression within the LHT anticipates formation of the OC (Figure 1D), cell morphologies are relatively uniform among the ventricular precursors at this stage. As development proceeds, ventricular cells increase their surface area (Figure 2C). Notably, however, only the cells in the OC of the expanded ventricle (48–58 hpf) significantly change their shape (Figure 2D). OC cells become flattened and elongated (Figure 2B, 2D, 2F, 2G, 2J, and 2K), and they are typically aligned with one another and oriented with their long axes perpendicular to the arterial–venous axis (Figure 2B, 2F, and 2J; Video S1), although occasionally they lack an obvious orientation (Figure 2B, 2G, and 2K; Video S2). Meanwhile, IC cells fail to flatten or elongate and instead maintain a cuboidal morphology (Figure 2B, 2D, 2H, and 2L; Video S3). We note that elongation is already apparent in some cells positioned to become OC at an intermediate stage of chamber emergence (36 hpf; unpublished data), suggesting that elongation of OC cells occurs gradually, concurrent with the gradual process of chamber expansion. We therefore propose that regionally confined cell shape changes underlie the acquisition of chamber morphology: the elongation and orientation of OC cells, coupled with the maintenance of cuboidal shape by IC cells, could account for the characteristic curvatures of the expanded ventricle.

Blood Flow Promotes Ventricular Cell Enlargement and Elongation

Because chamber emergence takes place following the onset of embryonic circulation, the physical forces inherent to cardiac function could influence the cell shape changes that accompany curvature formation. For example, normal blood flow, which creates changes in pressure and shear forces, has the potential to contribute to the induction of cellular enlargement and elongation at the OC. The zebrafish mutation weak atrium (wea) causes a phenotype appropriate for testing this hypothesis (Figure 3). The wea locus encodes atrial myosin heavy chain (amhc, also known as myh6), an atrium-specific myosin isoform [28]. Thus, wea mutants lack atrial sarcomeres, resulting in impaired atrial function and decreased blood flow through the ventricle compared with wild type ([28]; Figure 3M, 3N, 3Q, and 3R, and Videos S4 and S5). Although amhc is not expressed in the ventricle, ventricular shape and size are altered in wea mutants compared with wild type ([28], Figure 3A, 3B, 3E, 3F, 3I, and 3J). The wea mutant ventricle is small and dysmorphic, with less-pronounced curvatures than in the OC and IC of the wild-type ventricle. This secondary consequence of atrial dysfunction presumably indicates an impact of blood flow on ventricular morphology [28]. Because the cardiac cell number in the wea mutant ventricle is normal [28], the ventricle's abnormal morphology does not reflect a defect in cardiomyocyte proliferation. Instead, its aberrant curvatures could be the result of aberrant cell shapes or cellular organization. To investigate this possibility, we examined cardiomyocyte morphologies in the wea mutant ventricle (Figure 4).

In the wea LHT, cell shapes are uniform, round, and indistinguishable from those of wild type, although measurements of surface area suggest a trend toward smaller cell size (Figure 4A, 4B, and 4E). By the time of chamber emergence, the surface area of individual cells throughout the ventricle is significantly smaller than in wild type (Figure 4C–4E). Although elongated cells can be found in the wea OC, those cells tend to be less elongated than cells in the wild-type OC (Figure 4C–4H). In addition, cuboidal cell morphologies, normally confined to the IC, are frequently found in the wea OC (Figure 4G and 4H). The persistence of small, cuboidal cells in the wea ventricle does not reflect a simple developmental delay, because we detect similar cell morphologies in wea mutants at 72 hpf (H. J. Auman and D. Yelon, unpublished data). Additionally, treatment with 5 mM 2,3-butanedione monoxime (BDM), another method for reducing blood flow in zebrafish embryos [27], has an effect on ventricular cardiomyocyte morphology similar to that observed in wea mutants: at 48–52 hpf, BDM-treated ventricular cells are smaller (83 ± 2 μm² surface area) and less elongated (0.82 ± 0.01 circularity) than wild-type ventricular cells (Figure 4E; 110 ± 2 μm² surface area, 0.77 ± 0.01 circularity; both significantly different from BDM-treated values, p < 0.0001). The reduced degree of elongation observed in the wea OC correlates with a reduction in nppa expression at this location (Figure 5). Regionalized nppa expression initiates normally in the wea LHT (Figure 5A, 5B, 5E, and 5F), but, by 52 hpf, ventricular nppa expression is weak and occupies a smaller proportion of the chamber in comparison to wild-type (Figure 5I, 5J, 5M, and 5N). Together, these data suggest that normal blood flow is needed to promote the cell enlargement, cell elongation, and regionalized gene expression that accompany chamber curvature formation.

Contractility Restricts Ventricular Cell Size and Elongation

As blood flow promotes cell enlargement, we hypothesized that an opposing mechanism might restrict or moderate cell size. We reasoned that a good candidate for such resistance would be a feature intrinsic to the myocardial cell, such as its contractility. To test this idea, we utilized a mutation in the zebrafish ventricular myosin heavy chain gene (vmhc, also known as myh7) that blocks contractility in ventricular cardiomyocytes (Figure 6). The ventricle-specific expression of vmhc complements the atrium-specific expression of amhc ([28,33] and Figure 6A and 6B). The half-hearted (haf) mutation (see Materials and Methods) creates a premature stop codon in vmhc that is predicted to result in a truncated and non-functional Vmhc protein (Figure 6C). As a result, haf mutants lack ventricular contractility and provide a chamber-specific functional complement to wea mutants (Figure 3O and 3S; Video S6).

In contrast to the small ventricle of wea mutants, the haf mutant ventricle is large and distended, with abnormally spherical contours (Figure 3C). As for wea mutants, we found that cardiac cell number in haf mutants is normal (at 48 hpf, 231 ± 3 cells in wild type vs. 229 ± 13 in haf, n = 4), suggesting that aberrant cell shapes, rather than aberrant proliferation, might underlie the development of abnormal ventricular curvatures in haf. Indeed, examination of cell morphology in the haf ventricle revealed a significant increase in cell surface area, which is already apparent in the LHT and becomes more pronounced during chamber emergence (Figure 7A–7E). Many cells within the expanded haf ventricle are also highly elongated (Figure 7C–7G), although not all are elongated to the same extent (Figure 7H). As in wild type, the orientation of elongated cells in haf is typically perpendicular to the arterial–venous axis. However, rather than being restricted to the OC, elongated cells are often found throughout the chamber. The overextension and overrepresentation of elongated cells in the haf ventricle correlate with the increased and widespread expression of nppa throughout the chamber (Figure 5K). In contrast, nppa expression is correctly regionalized in the haf LHT, despite a slight delay in initiation of expression (Figure 5C). Overall, these data indicate that Vmhc function is required to restrict cell size, to limit the extent of cellular elongation, and to confine the distribution of elongated cell types during chamber curvature formation.

Intrinsic Contractility Independently Regulates Cardiomyocyte Morphology

Our analysis of the haf mutant phenotype indicates that ventricular cell shapes are significantly affected by the loss of Vmhc function, raising the possibility that cardiomyocyte contractility directly influences cardiomyocyte morphology. However, it is also possible that the particular hemodynamic environment in haf mutants could have an indirect effect on ventricular cell shape. Although the haf ventricle is noncontractile, the haf atrium contracts normally (Video S6). As a result, blood fills the haf ventricle without subsequent ventricular response. Thus, haf ventricular cells are likely subjected to an elevation in preload that could potentially lead to secondary effects on cell shape. We therefore wished to uncouple the influences of blood flow and contractility in the establishment of cell shapes, and proceeded to test whether the morphology of a noncontractile ventricular cell changes when placed in different hemodynamic conditions (Figure 8).

The wea;haf double mutant, which lacks both atrial and ventricular contractility (Figure 3L, 3P, and 3T; Video S7), enabled us to test the role of vmhc in the acquisition of ventricular cell shape without the complicating factor of atrial contraction and elevated preload. Although the wea;haf ventricle is typically not as distended as the haf ventricle, both ventricles have a relatively spherical appearance (Figure 3C and 3D). Cell shape and size are normal in the wea;haf LHT; the double mutants exhibit neither the large cells of the haf LHT nor the trend toward small cells seen in the wea LHT (Figure 8A, 8B, and 8E). However, cells in the wea;haf expanded ventricle are clearly abnormal: they are irregularly shaped and lack apparent alignment, consistent orientation, or regional organization (Figure 8C and 8D). The lack of a particular area of cellular elongation in the wea;haf ventricle correlates with its diffuse, weak nppa expression (Figure 5I and 5L). Strikingly, unlike wea cells, which maintain small size and cuboidal morphology within the expanded chamber, wea;haf cells do enlarge and elongate, despite an absence of blood flow. Indeed, although they exhibit irregular morphologies, wea;haf cells exhibit an average surface area and circularity similar to that of wild-type cells (Figure 8E). Therefore, wea;haf ventricular cells seem to lack an intrinsic mechanism for maintenance of cell morphology.

To test this idea further, we challenged the shape of haf cells by placing them into a hemodynamic environment in which they would be subjected to normal flow, as is found within the wild-type ventricle. We transplanted haf mutant cells into wild-type hosts at midblastula stages and analyzed the morphology of resultant haf ventricular clones in the expanded chamber. In this context, haf cells adopt an entirely different and unusual morphology: whether in the OC or IC, the noncontractile haf donor cells round up and extrude from the contractile ventricular wall (Figure 8H and 8K). This protrusion phenotype, seen in four out of four embryos, was never observed when wild-type (n > 10 embryos) or wea (n > 12 embryos) cells were transplanted in control experiments (Figure 8F, 8G, 8I, and 8J). These results demonstrate that Vmhc is required cell-autonomously for maintenance of cardiomyocyte morphology and chamber wall integrity. Additionally, our haf, wea;haf, and transplantation data argue that the particular irregular shape adopted by a cell lacking Vmhc depends upon the contractility of its neighbors and the nature of its hemodynamic environment. Together, our data indicate that maintenance of appropriate cardiomyocyte size and shape requires cell-autonomous traits that are dependent on contractility and independent of the influence of blood flow.

Cell Shape Changes Underlying Chamber Emergence Require a Balance of Biomechanical Forces

Together, our analyses provide a new model for the cellular processes creating chamber shape. During ventricular chamber emergence, regionally confined cell shape changes correlate with the locations of curvature formation. We therefore propose that cardiomyocyte elongation plays a substantial role in creating the characteristic convex shape of the OC; on the opposing side of the chamber, maintenance of cuboidal cardiomyocyte morphology contributes to creating the concave IC. Blood flow through the developing ventricle promotes cardiomyocyte elongation and thereby encourages establishment of the OC. The limits of cardiomyocyte elongation are constrained by the cell's intrinsic contractility, and these constraints are critical for maintaining the IC and avoiding chamber dilation. Thus, a balance between extrinsic and intrinsic functional inputs is important for the control of cardiomyocyte shape and, consequently, chamber shape.

The proposed balance between the functional inputs of flow and contractility is likely to reflect a balance between opposing biomechanical forces. Although our experiments do not address the specific physical forces impacting chamber curvature, our data are consistent with a model in which equilibrium between fluid forces and cellular tension influences cardiomyocyte morphology. Fluid forces could affect cardiomyocyte shape in several ways. Sensation of shear forces by the endocardium could indirectly trigger cellular deformation in the myocardium. Endothelial cells are known to respond to shear forces via changes in gene expression in vitro [34,35] and in vivo [36] and, in zebrafish, shear forces are predicted to exist at levels detectable by endocardium at 37 hpf, when chamber emergence is underway [26]. Alternatively, hemodynamic pressure could play a more direct role in stretching cardiomyocytes. We favor this idea, since the flattened, elongated, and aligned morphologies of OC cells are reminiscent of the shapes assumed by stretched cardiomyocytes in culture [37,38].

Although fluid forces seem to encourage distortion of cardiomyocyte dimensions, cardiomyocyte contractility provides resistance to cell shape change, perhaps by producing inherent cellular tension. This tension could result from the elastic nature of actively contracting sarcomeres. Alternatively, sarcomere assembly could contribute to the organization of actomyosin cytoskeletal structures that create cellular stiffness. In either case, our data indicate that sarcomere integrity is critical for the maintenance of cardiomyocyte morphology. This conclusion is compatible with a previous study in zebrafish demonstrating that transplanted cells lacking the giant protein Titin have difficulty integrating into the wild-type chamber wall [39]. Titin is a multidomain protein with several proposed functions [40]; our results suggest that the protrusion phenotype observed in cells lacking Titin is related to the role of Titin in sarcomere organization.

Although both IC and OC cells respond to functional inputs, it is clear that they respond differently, such that elongation occurs only in the region of the OC. We propose that intrinsic patterning programs evident in the LHT create differences among future IC and OC cells, perhaps conferring differential vulnerability to functional inputs. These initial differences could be the predecessors of the distinct mechanical properties of each curvature, such as the greater stiffness of the IC observed in the chick heart [41]. Our examination of nppa expression indicates a molecular distinction between the future IC and OC cells within the LHT, before cell shape changes begin. This initial regionalization of nppa expression, apparent in the LHT of wea mutants, haf mutants, and wea;haf mutants, is evidently not dependent on either blood flow or contractility (Figure 5A–5D). In contrast, later nppa expression in the expanded chamber appears to relate to functional inputs, responding as cells do to the forces promoting elongation. As stretch is a primary inducer of nppa expression [42,43], expression of nppa in the OC may correlate with increased levels of stretch in those cells. Together, our data suggest a model in which organ shape is the product of cell shape derived from the integration of patterning programs with balanced physical forces. The interplay between extrinsic forces and intrinsic cellular properties has the potential to influence organ shape acquisition even in noncontractile tissues, particularly those that experience biomechanical inputs generated by circulating fluids (e.g., kidney tubules [44] or brain ventricles [45]).

Implications for the Origins of Cardiomyopathy

Our proposed model is likely to be relevant to the mechanisms underlying cardiomyopathy. Mutation of the human cardiac myosin heavy chain genes MYH6 or MYH7 can cause hypertrophic cardiomyopathy [46,47], and mutation of MYH7 can cause dilated cardiomyopathy [38]. It is not yet understood how the various mutations identified lead to the diversity of phenotypes observed in patients [48]. Here, we demonstrate two distinct cellular mechanisms by which mutation of myosin heavy chain genes can create abnormal ventricular morphology: a cell-autonomous mechanism via mutation of vmhc and a cell non-autonomous mechanism via mutation of amhc. Furthermore, our model suggests that even subtle modification of circulation or contractility could lead to abnormalities in cell morphology with significant consequences for chamber morphology. Thus, our findings broaden the array of mechanisms to consider as cellular etiologies of cardiomyopathy.

Immunofluorescence and F-actin staining.

Whole-mount immunostaining was performed as previously described [49], using primary antibodies against sarcomeric myosin heavy chain (MF20) and tropomyosin (CH1). MF20 and CH1 were obtained from the Developmental Studies Hybridoma Bank (Iowa City, Iowa, United States), maintained by the Department of Biological Sciences, University of Iowa, under contract NO1-HD-2–3144 from the National Institute of Child Health and Human Development (NICHD). Secondary antibodies were obtained from Southern Biotech (Birmingham, Alabama, United States). We visualized cell outlines with rhodamine-conjugated phalloidin (Molecular Probes, Eugene, Oregon, United States) at a dilution of 1:75, using a protocol that preserves GFP fluorescence in fixed embryos [50].

In situ hybridization.

Probes for cmlc2, vmhc, amhc, and nppa [28,51] were used as previously described for standard in situ hybridization. For fluorescent in situ hybridization, probes were labeled with digoxygenin (Roche, Basel, Switzerland), fluorescein (Roche), or dinitrophenol (Mirus, Madison, Wisconsin, United States) and were detected by deposition of fluorescein or Cy3 tyramides (PerkinElmer, Wellesley, Massachusetts, United States). A detailed protocol for fluorescent in situ hybridization will be published separately (J. Schoenebeck, B. Keegan, and D. Yelon, unpublished data).

Mosaic labeling of cardiomyocytes.

Embryos homozygous for Tg(cmlc2:egfp) were injected at the one-cell stage with 5 pg of Tg(cmlc2:dsredt4) plasmid DNA. The Tg(cmlc2:dsredt4) transgene drives expression of dsredt4 [52] with an approximately 0.8-kb fragment found upstream of the 5′ UTR of cmlc2 (−870/+3; [32]). Injected embryos typically had zero to ten cardiomyocytes expressing both dsredt4 and egfp. Embryos of interest were mounted in viewing chambers with 0.2% tricaine to anesthetize the embryo and minimize cardiac contraction during confocal imaging.

Imaging.

Embryos were examined with Zeiss M2Bio and Axioplan microscopes and photographed with a Zeiss Axiocam digital camera (Carl Zeiss, Oberkochen, Germany). Images were processed using Zeiss Axiovision and Adobe Photoshop software (Adobe Systems, San Jose, California, United States). Live embryo videos were recorded and processed using an Optronics DEI750 video camera, a Zeiss M2Bio microscope, and iMovie software. Transmission electron microscopy procedures were conducted as previously described [28]. Confocal images were obtained using a Zeiss LSM510 confocal laser-scanning microscope and LSM software. Z-stacks were rendered in 3D and analyzed with Volocity software (Improvision, Coventry, United Kingdom).

Morphometrics.

We used ImageJ software (National Institutes of Health [NIH] http://rsb.info.nih.gov/ij/) to trace cell outlines and measure each cell's surface area and perimeter. Cells were chosen for measurement when their outlines were clearly visible and sharply rendered in the xy plane of view. Circularity was calculated by ImageJ as a normalized ratio of area (A) to perimeter (P), with a ratio of 1 representing a perfect circle (circularity = 4πA/P²). In this way, the circularity value distinguishes cells with more circular surfaces from those with more elliptical or elongated surfaces.

Measurements of cells from multiple embryos at the LHT stage (24–28 hpf) and at the expanded chamber stage (48–58 hpf) were pooled into datasets. When analyzing the LHT, we measured cells throughout the outflow half of the tube, thereby including cells that will contribute to the future ventricular IC and OC. In the expanded wild-type ventricle (48–58 hpf), the IC and OC could easily be distinguished morphologically. Because the curvatures were distorted in mutant or BDM-treated embryos, we compared cardiomyocytes throughout the mutant or BDM-treated ventricle to pooled data from wild-type ventricular cardiomyocytes. Sample sizes were as follows: wild-type LHT, 173 cells from 11 embryos; wild-type chamber, 336 cells from 18 embryos; wild-type IC, 122 cells from 14 embryos; wild-type OC, 210 cells from 16 embryos; wea LHT, 99 cells from four embryos; wea chamber, 250 cells from 17 embryos; BDM-treated chamber, 99 cells from 14 embryos; haf LHT, 215 cells from 13 embryos; haf chamber, 149 cells from seven embryos; wea;haf LHT, 114 cells from six embryos; and wea;haf chamber, 161 cells from ten embryos. Area and circularity values were compared using unpaired t-tests.

Identification of the half-hearted mutation.

We isolated the half-hearted (haf^sk24) mutation in our laboratory's ongoing screen for ethylnitrosourea-induced mutations that disrupt chamber form and function (F. Olale, T. Bruno, A. Schier, and D. Yelon, unpublished data). In this screen, we analyzed the haploid progeny of F1 females at 30–48 hpf using established protocols for live imaging and immunofluorescent detection of sarcomeric myosin heavy chain [49]. haf mutants were evident during screening because of their noncontractile and dysmorphic ventricles. Corresponding with the lack of ventricular contractility, immunolocalization of sarcomeric myosin heavy chain was absent from the haf mutant ventricle (Figure 3K), and ultrastructural analyses via electron microscopy revealed an absence of ventricular sarcomeres (Figure 3O). Linkage analysis demonstrated that the haf mutation is located near the SSLP marker Z13475 on Chromosome 2 (one recombinant in 50 meioses), near the vmhc gene [28]. Sequencing of genomic DNA in haf mutants identified a C to T transition at position 3,094 of the vmhc open reading frame, creating a premature stop codon (Figure 6C). The wild-type Vmhc protein contains 1,937 amino acids, and the predicted haf Vmhc protein would terminate after amino acid 1,058, disrupting the myosin tail domain to result in a nonfunctional truncated product. Furthermore, in situ hybridization demonstrated a marked decrease of vmhc transcript in the haf mutant ventricle (Figure 6D and 6E), suggesting degradation of the mutant transcript by nonsense-mediated decay. Thus, protein production is unlikely, consistent with the loss of detectable ventricular myosin heavy chain (Figure 3K). Taken together, these analyses indicate that the haf locus encodes Vmhc.

Transplantation.

We used standard transplantation techniques at midblastula stages [53]. To create each chimeric embryo, we placed 10–15 blastomeres from a donor embryo near the embryonic margin of a host embryo. For transplant hosts, we used wild-type embryos homozygous for Tg(cmlc2:egfp). Transplant donors were generated from intercrosses of wea or haf heterozygotes carrying Tg(cmlc2:egfp) and injected with rhodamine dextran (Molecular Probes) as a lineage tracer. Donor embryos were retained for retrospective identification as mutant or wild-type based on phenotypic analysis at 48 hpf. We observed no differences in viability among transplanted cells for all genotypes.

Accession Numbers

The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession numbers for the genes discussed in this paper are myh6 (NM_198823) and vmhc (AF114427).