Heart and soul/PRKCi and nagie oko/Mpp5 regulate myocardial coherence and remodeling during cardiac morphogenesis

Vertebrate heart morphogenesis involves the generation of a simple tube that is shaped into a complex functional organ. The types of cellular and tissue rearrangements driving this process are largely unknown. In zebrafish,heart morphogenesis involves the fusion of two bilateral populations of cardiac progenitor cells at the midline of the embryo followed by the formation of the three-dimensional heart cone, a transitional structure that consists of a dorsal ventricular and a ventral atrial part(Fig. 1A)(Stainier et al., 1993; Yelon et al., 1999). Subsequently, the heart cone moves from the dorsoventral into the anterior-posterior plane, a process referred to as tilting. As a consequence,the ventricular portion of the heart cone comes to lie more posteriorly. Finally, the heart cone elongates along the anterior-posterior axis into the heart tube, which is converted into a looped, two-chambered organ.

The heart and soul (has) mutation causes severe defects in cardiac morphogenesis, in which tilting of the heart cone is blocked and heart tube elongation is impaired (Yelon et al., 1999; Horne-Badovinac et al., 2001; Peterson et al.,2001). The has gene encodes PRKCi, which is required for the establishment of apicobasal polarity of epithelial cells. PRKCs are components of the apical Par3 protein complex, which has been linked to the apical Crumbs-Pals1/Mpp5 protein complex(Hurd et al., 2003; Nam and Choi, 2003; Wang et al., 2004). Consistent with a function in apicobasal cell polarity, has mutants show defective formation and maintenance of several embryonic epithelia. This observation led us to investigate whether an epithelial tissue is involved in cardiac morphogenesis. As heart cone tilting and heart tube elongation occur in cloche, a zebrafish mutant that lacks endocardial cells,morphogenesis may largely depend on the myocardium or on tissues external to the heart (Stainier et al.,1995). During heart fusion and cone formation, myocardial cells are organized into two bilateral cell populations that have some characteristics of polarized epithelia(Trinh and Stainier, 2004; Trinh et al., 2005). At these early stages, PRKCs are localized to the apical junctions of myocardial cells.

In this study, we expand on the roles of nagie oko(nok)/mpp5 and has/prkci during polarization of the myocardial layer and cardiac morphogenesis. We show that has/prkciand nok/mpp5 function within myocardial cells during heart cone formation and that Has/PRKCi function requires its catalytic activity. Loss of has/prkci disrupts the epithelial organization of myocardial cells and blocks heart cone tilting. Using nok/mpp5 antisense morpholino oligonucleotides (MO), we have identified an early function of this gene in epithelial polarity of myocardial cells before heart cone formation, whereby heart cone fusion is severely impaired. Moreover, zygotic nok/mpp5mutants display defects in myocardial remodeling that contribute to the subsequent elongation of the heart tube.

Zebrafish were maintained at standard conditions(Westerfield, 1994). Embryos were staged by hours post-fertilization (hpf) at 28.5°C(Kimmel et al., 1995) and according to somite number. The nok^m520 allele analyzed in this study results in a premature stop codon at residue Arg 546 within the GUK domain (Wei and Malicki,2002), whereas nok^s305 causes the loss of the entire GUK domain, resulting in an even larger truncation (Wei, personal communication). The nok^m520 mutation has been shown to cause a complete, or almost complete, loss of function(Wei and Malicki, 2002). Moreover, the nok^s305 mutation placed in trans to the nok^m520 mutation produced a phenotype that was indistinguishable from nok^s305 homozygotes. Therefore, the nok^s305 mutation causes a complete, or almost complete,loss of function.

The Tg(cmlc2:prkci) and Tg(cmlc2:nok)transgenic lines were generated by injection of I-SceI cmlc2:prkci and I-SceI cmlc2:nok transformation vectors into 1-cell stage embryos (Thermes et al., 2002). F1 founders were identified by PCR genotyping and used for the establishment of stable lines. For rescue experiments using the transgenic lines in has or nok mutant backgrounds,respectively, embryos were PCR genotyped for the presence of the transgene insertion using transgene-specific primers (primer sequence information available upon request).

Clonal analysis of mRFP-positive cells was performed by DNA injection using an I-SceI vector containing an cmlc2:mRFP insert. The statistical significance of the data was evaluated using Student's t-test function on Excel software.

The coding region of zebrafish has/prkci was amplified by PCR from a full-length clone (Horne-Badovinac et al., 2001), introducing XhoI and XbaI restriction sites 5′ and 3′, respectively, and subcloned into the expression vector pCS2+. Site-directed mutagenesis was performed using the Quick Change Site Directed Mutagenesis Kit (Stratagene, CA, USA).

5′-CCCAATTACATTGCAGCAGCGATTCTGAGAGGAGAAG-3′. Further details are available upon request. pXT7NOK containing full-length nok was a gift from J. Malicki (Massachusetts General Hospital, Boston, MA). The cmlc2:mRFP was cloned by introducing the PCR-amplified mRFPfragment into the I SceI cmlc2 transformation vector.

Constructs were transcribed using the SP6 MessageMachine kit (Ambion). In vitro synthesized capped mRNA was dissolved in water and mixed with the MO prior to injection. Typically, 60 pg of RNA were injected into WIK or WIK/TL embryos. For overexpression, 120 pg of prkci^2A were injected into wild-type embryos. MOs (Gene Tools) were injected at a concentration of 100 μmol/l. MO sequences were: nokMO,5′-TGAGGTCAGCAGC-GGCTCCAAACAC-3′; hasMO,5′-TGTCCCGCAGCGTGGGCATTATGGA-3′.

Whole-mount in situ hybridization was performed as previously described(Jowett and Lettice, 1994). Digoxigenin-UTP-labeled riboprobes were synthesized according to the manufacturer's instructions (Boehringer Mannheim). The cmlc2 probe(AF114428) was amplified from cDNA and subcloned into pCS2+. Probes for amhc and vmhc were a gift from D. Yelon. Embryos were mounted in Permount (Fisher Scientific) and documented using an Axioplan 2 microscope (Zeiss). Images were processed with Adobe Photoshop software (Adobe Systems).

Antibody and phalloidin stainings were performed as previously described(Horne-Badovinac et al., 2001). Nuclear stainings were performed using SYTOX green nucleic acid stain(Molecular Probes) as described (Picker et al., 1999). The following antibodies were used: rabbit anti-aPKCζ (1:100, Santa Cruz Biotechnology, USA), mouse anti-ZO-1(1:200, Zymed), rabbit anti-α-catenin (1:200, Sigma), goat anti-rabbit RRX (1:200), anti-mouse Cy5 (1:100) (Jackson ImmunoResearch). Transverse sectioning was performed according to Trinh et al. (Trinh et al., 2004). Hearts were analyzed using a Leica TCS SP2 confocal microscope. For reconstruction of myocardial cell morphology single sections of recorded z-stacks were analyzed using Leica software. For quantification of myocardial cell size, individual scans of cmlc2:mRFP-positive cells were analyzed with TINA version 2.08e (Raytest Isotpenmeßgeräte GmbH)and absolute values recalculated as relative values. Standard deviation was calculated using Excel software (Microsoft).

Sixteen-somite stage embryos (nok^s305 and wild-type siblings) were yolk injected with 10 mmol/l BrdU in 0.2 mol/l KCl. Embryos were fixed at 32 hpf in 4% PFA for 2 hours at RT. After rinsing in 1×PBS, embryos were dehydrated in MeOH and stored overnight at -20°C. After rehydration, the samples were digested in 10 μg/ml proteinase K for 15 minutes and refixed in 4% PFA for 20 minutes. Incubation in 2N HCl for 1 hour was performed to relax the chromatin and facilitate the following immunolabeling using anti-BrdU antibody (Roche 1170376) diluted 1:100 in blocking solution. BrdU was visualized with Cy3-conjugated goat anti-mouse(Jackson ImmunoResearch; 1:500).

The fragmented DNA of apoptotic cells was detected in 32-hpf-old nok^s305 and wild-type siblings using the In Situ Cell Death Detection Kit TMR Red (Roche) following the protocol from Shepard et al.(Shepard et al., 2004), with the exception that the yolk was removed manually after fixation for improved accessibility to the heart.

Microangiographic images were generated by injection of 70 kD FD into the dorsal aorta of 36-hpf embryos. Embryos were imaged 5-10 minutes after injection with a CoolSnap ES camera (Photometrics). Data were collected and analyzed using Metamorph software version 6.1 (Visitron Systems).

To test our hypothesis that Has/PRKCi regulates cardiac morphogenesis intrinsically within the myocardium, we generated a line of transgenic zebrafish that expresses wild-type PRKCi under the control of the cardiac myosin light chain 2 (cmlc2) promoter region within all myocardial cells [Tg(cmlc2:prkci)]. Translation of the transgene expression is insensitive to the has/prkci MO used in this study. We used cmlc2, a marker for cardiac differentiation(Yelon et al., 1999), to assess heart morphogenesis. Within Tg(cmlc2:prkci)transgenic embryos, injection of the has/prkci MO produced the complete range of epithelial defects characteristic of has/prkci but failed to affect cardiac morphogenesis(Fig. 1D). has/prkcimorphants that were Tg(cmlc2:prkci) transgenic showed correct heart cone tilting followed by heart tube elongation. In total, we examined cardiac morphogenesis in 61 transgenic embryos, 41 of which produced an elongated heart tube (65.1%). By contrast, almost all of the control wild-type siblings injected with has/prkci MO produced the has/prkci cardiac morphogenesis phenotype with an arrested heart cone(Fig. 1C). The variability of heart morphogenesis rescue among Tg(cmlc2:prkci) embryos may be due to the critical period of has/prkci function, which coincides with the onset of cmlc2 transgene expression. Previously, the promoter region used for the production of the transgenic line has been shown to display a variable onset of transgene expression(Huang et al., 2003). Therefore, Has/PRKCi acts tissue-autonomously within myocardial cells and is sufficient for correct heart cone tilting and heart tube elongation.

To test whether Has/PRKCi activity within myocardial cells is sufficient for heart function in an otherwise morphant embryo, we injected 70 kD fluorescein dextran into the dorsal aorta of 36-hpf embryos and found that peripheral circulation was indeed present, which is never seen in has/prkci morphants (Fig. 2B,C). A similar result was obtained after outcross of the transgene into the has mutant background(Fig. 2E,F). Therefore,Has/PRKCi activity within myocardial cells is sufficient for heart function.

To further assess whether Has/PRKCi function during heart morphogenesis depends on its kinase activity, we generated a mutation of has/prkcithat encodes a kinase-dead version of the protein(prkci^2A; residues Pro 408 and Glu 409 within the catalytic center exchanged for Ala) (Hanks and Hunter, 1995). Whereas co-injection of 60 pg of wild-type prkci mRNA together with has/prkci MO gave a complete rescue(data not shown), co-injection of prkci^2A mRNA with has/prkci MO resulted in the has/prkci external phenotype. We refer to these animals as prkci^2A mutants. The 32-hpf wild-type heart is an elongated two-chambered tube with posterior ventricle and anterior atrium, one-third of prkci^2A mutants had the characteristic dysmorphic cone-like heart of has/prkci mutants. Thus,normal morphogenetic movements of heart cone tilting and regular elongation into the two-chambered heart tube had failed. Moreover, 54% of prkci^2A mutants displayed an even more severe phenotype in which cardiac fusion was not completed, resulting in bilateral wings of cells(Fig. 1E). Therefore, the prkci^2A mutation affects earlier steps of heart morphogenesis than the zygotic has/prkci mutation, which we attribute either to an interference with maternal Has/PRKCi protein or redundant function of PRKCζ. Thus, the biological activity of Has/PRKCi during heart morphogenesis crucially depends on its catalytic kinase activity within myocardial cells.

To test whether has/prkci cardiac morphogenesis defects are attributed to defective epithelial cell polarity, we investigated whether other regulators of epithelial cell polarity cause similar defects. As overexpression of prkci^2A mRNA in wild-type embryos produced a phenotype similar to that of nok mutants(Fig. 1N,O), we first characterized the role of this gene in cardiac morphogenesis. The nokgene encodes a membrane-associated guanylate kinase (MAGUK) with 68% identity to mammalian Pals1/Mpp5 and is required for the maintenance of epithelial cell polarity (Wei and Malicki,2002; Horne-Badovinac et al.,2003). In mammalian cells, Pals1/Mpp5 functions as a scaffolding protein that links the transmembrane receptor Crumbs with Par6-PRKC(Hurd et al., 2003; Wang et al., 2004). We tested whether Nok is a functional homolog of mammalian Pals1/Mpp5 by injecting murine Pals1/MPP5 mRNA into nok mutants and morphants. Murine Pals1/MPP5 mRNA largely prevented ventral curvature of morphants (n=29/30 morphants rescued; Fig. 3A-C). To further evaluate the rescue efficiency of Pals1/MPP5 mRNA, we assayed the organization of the retinal pigmented epithelium (RPE), which is almost completely lost in nok mutants but was significantly recovered upon injection of murine Pals1/MPP5 mRNA (Fig. 3D-F). Finally, we assessed the recovery of heart function by quantifying the presence of circulation within peripheral blood vessels in nok morphants at 36 hpf. Injection of Pals1/MPP5 mRNA caused a significant increase in the number of nok/mpp5 morphants with functional circulation (Fig. 3G). The failure to completely rescue the nok morphant phenotype may be attributed to sequence variations between Nok and murine Pals1. Therefore, Nok is a functional homolog of murine Pals1/Mpp5.

Assessment of heart morphology in nok/mpp5 mutant hearts(nok^s305 and nok^m520 alleles) revealed that they largely developed beyond the cone stage. However, expression of cmlc2 showed that the heart failed to completely elongate by 36 hpf. The defects observed ranged from severe (the heart tube failed to elongate)(Fig. 1F) to moderate (the heart tube was significantly shorter than wild type)(Fig. 1G) to mild (the heart tube elongated but was narrow) (not shown). Expression of ventricular myosin heavy chain (vmhc) and atrial myosin heavy chain(amhc) revealed that chamber-specific differentiation occurred but that both the atrium and the ventricle were defective in size and form(Fig. 1J-M)(Berdougo et al., 2003). Therefore, we conclude that nok/mpp5 is required for correct heart tube elongation.

We tested whether nok/mpp5 has a function prior to heart tube elongation by injecting a 5′-UTR-directed MO that has previously been shown to completely eliminate maternal and zygotic protein expression by the second day of development (Wei and Malicki, 2002). Similarly, for has/prkci, we injected an ATG-directed MO that has been shown to eliminate protein expression(Horne-Badovinac et al., 2001). Indeed, myocardial morphogenesis in these morphants resembled the has/prkci mutant heart phenotype (see below). The co-injection of nok/mpp5 MO with wild-type nok/mpp5 mRNA completely rescued myocardial morphogenesis, thereby demonstrating that the MO causes a specific developmental defect (Fig. 1I).

To assess the effects of nok/mpp5 and has/prkci on the polarization of the myocardial epithelium, we used a transgenic line of zebrafish that expresses green fluorescent protein (GFP) under the control of the cmlc2 promoter region [Tg(cmlc2:GFP)] to visualize heart morphology (Huang et al.,2003). We utilized this transgenic line in conjunction with antibodies against the junctional protein Zonula occludens-1 (ZO-1), against PRKCi and z (anti-aPKC) at the apical junction of myocardial cells andα-catenin at the adherens junction and lateral membranes, and rhodamine phalloidin to detect filamentous actin. In wild type, before heart cone formation (16-somite stage), myocardial cells were positioned in two bilateral sheets that converged toward the midline(Fig. 4A). Within each bilateral sheet, myocardial cells were highly polarized and displayed junctional belts positive for PRKC and ZO-1(Fig. 4J,K)(Trinh and Stainier, 2004; Trinh et al., 2005). Within the next 2 hours, myocardial cells converged onto the midline and fused to form the heart cone (20-somite stage; Fig. 4B). At this stage, the myocardium was a monolayered epithelial sheet that lay symmetrically near the midline and had the appearance of an evenly shaped cone structure with PRKC- and ZO-1-positive junctional belts. In transverse sections, lateral portions of the myocardium were organized as a monolayered sheet of mostly cuboidal cells. There was asymmetrical localization of ZO-1 at the apical side and PRKC just basal of ZO-1 and more diffusely along basolateral membranes (Fig. 5B). The adherens junctional protein, α-catenin, largely co-localized with ZO-1 at the apical membrane(Fig. 5D).

Before heart cone tilting, myocardial cells further converged and the heart cone elongated along the dorsoventral axis(Fig. 4C). During tilting(Fig. 4D), the wild-type heart cone became highly asymmetrical and revealed the presence of several morphologically elongated myocardial cells that were attached to surrounding tissues (Fig. 4S). These cells were located at the anterior/ventral edge of the heart cone.

In has/prkci and nok/mpp5 morphants, the epithelial organization of the myocardium was affected already before heart cone formation (16-somite stage). The localization of ZO-1 was in a spotted distribution rather than in a contiguous junctional belt(Fig. 4N,P). Moreover, within nok/mpp5 morphant myocardial cells, PRKC was mislocalized in a diffuse pattern (Fig. 4M). Transverse sections revealed that nok/mpp5 and has/prkcimorphant myocardial layers appeared less coherent and were, in part,multilayered (Fig. 5C,E). Both morphants lacked the asymmetric apical distribution of ZO-1, which was,instead, found to be diffusely distributed. Moreover, PRKC was mislocalized in a diffuse pattern in nok/mpp5 morphants(Fig. 5C). In has/prkci morphants, there was no co-localization of ZO-1 andα-catenin. Instead, α-catenin was diffusely distributed along the membrane (Fig. 5E). Therefore,Nok/Mpp5 functions to maintain PRKC localization basal of ZO-1. The coherence of the myocardial sheets was disrupted and heart cone fusion was delayed in nok/mpp5 morphants (Fig. 4E,G). Indeed, nok/mpp5 morphants failed to undergo heart cone fusion at the 28-somite stage (Fig. 4G), a stage at which the wild-type heart cone has already tilted into the anterior-posterior plane of the embryo(Fig. 4D). In has/prkci morphants, the coherence of the myocardial layer was affected as indicated by holes that disrupt the integrity of this layer(Fig. 4H). Moreover, tilting of the heart cone did not occur. Myocardial cells that connect the myocardium with surrounding tissues were not seen in has/prkci mutants. We conclude that the polarized epithelial organization and coherence of the myocardium is affected in has/prkci and nok/mpp5 morphants. Therefore, nok/mpp5 morphants reveal an early gene function that is not apparent in the two zygotic mutants characterized.

To further investigate the zygotic role of nok/mpp5 in myocardial morphogenesis, we characterized the heart tube elongation defects. Defective heart tube elongation in nok/mpp5 mutants may result because of an altered cellular architecture of the myocardial layer, fewer cells, or a combination of the two. To distinguish among these possibilities, we first counted myocardial cells at 36 hpf in confocal sections of wild-type and nok^s305 mutant embryos that were also Tg(cmlc2:GFP) transgenic. Individual myocardial cells are identifiable by the presence of a strong nuclear GFP expression(Fig. 6A,B). Wild-type hearts had, on average, 28% more myocardial cells than nok^s305mutant hearts [wild type, 269±13 s.d. (n=3 hearts); nok^s305, 211±19 s.d. (n=6 hearts)]. This difference could account for part of the heart tube elongation defects in nok/mpp5.

A smaller cell count may be caused by a reduced rate of cell proliferation or by an increased rate of cell death. We first compared the proliferation rates of wild type and nok/mpp5 mutants that were also Tg(cmlc2:GFP) transgenic using a BrdU assay. The mitotic index of myocardial cells (mitotic cells/100 myocardial cells) that divided between the 16-somite stage and 32 hpf was similar between wild type (10.9%) and nok/mpp5 mutants (10.3%) [wild type, 29.3±8.3 s.d. BrdU-positive cells/heart (n=10 hearts); nok/mpp5,21.8±15.1 s.d. BrdU-positive cells/heart (n=6 hearts)](Fig. 6H,I,L). Therefore, cell proliferation contributes only slightly to heart tube elongation between the 16-somite stage and 32 hpf, and the comparable mitotic index does not explain the differences in cell numbers of wild-type and nok/mpp5 mutant myocardia. We next tested whether an increased rate of cell death might cause the reduced myocardial cell count in nok/mpp5 mutants. TUNEL analysis of wild type and nok/mpp5 mutants that were also Tg(cmlc2:GFP) transgenic indicated a low rate of apoptotic cell death that was slightly increased in mutants [wild type, 0.5±1.0 s.d. TUNEL-positive cells/heart (n=10 hearts); nok/mpp5,2.0±1.2 s.d. TUNEL-positive cells/heart (n=10 hearts)](Fig. 6J,K,L). Similar results were obtained with Acridine Orange stainings (not shown). Therefore, the smaller cell count in nok/mpp5 mutants is not caused by a reduced cell proliferation rate or increased rates of apoptotic cell death during heart tube elongation. We suggest that a smaller heart is the result of altered proliferation and/or cell survival rates at earlier stages of cardiogenesis, or a smaller pool of progenitor cells.

Next, we analyzed the epithelial organization of myocardial cells at 36 hpf. At this stage, myocardial cells within the atrial/anterior portion of the heart tube had extended cell sizes and nuclei were distantly spaced, whereas myocardial cells within the posterior/ventricular portion of the heart tube were more narrowly spaced (Fig. 6A). Whereas atrial myocardial cells acquired elongated squamous epithelial shapes (Fig. 6A′′,C), ventricular myocardial cells had a cuboidal epithelial organization (Fig. 6A′).

Next, we compared the spatial arrangement of myocardial cells and found that nok/mpp5 mutant myocardial cells were more densely arranged than wild-type cells at 36 hpf. The differences were most striking within the atrium, where nok/mpp5 mutant myocardial cells were reminiscent of wild-type heart-cone-stage myocardial cells.

For visualization and measurement of individual myocardial cell shapes in wild type and nok/mpp5 mutants, we used an expression plasmid[cmlc2:membrane red fluorescent protein (cmlc2:mRFP)] to achieve transient and mosaic expression of mRFP within single myocardial cells. We compared single cells expressing mRFP between wild-type siblings and mutants within the Tg(cmlc2:GFP)/nok^s305and Tg(cmlc2:GFP)/nok^m520 transgenic backgrounds (Fig. 6E-G). The relative surface extension of mRFP-positive cells was quantified using TINA 2.08 software (Raytest Isotpenmeßgeräte GmbH). In summary, we found that wild-type myocardial cells were on average about 25-35% more extended than in nok/mpp5 [wild type, 100±25.4% s.d. (n=53 cells); nok^s305, 74.1±25.3% s.d. (n=58 cells); nok^m520, 64.1±22.6% s.d. (n=42 cells)]. The differences in cell surface extension are highly significant(P≤0.01, according to Student's t-test). Even more strikingly, wild-type atrial cells had an average surface extension that was about 40% larger than wild-type ventricle cells [wild-type atrium,136.01±19.48% s.d. (n=8 cells); wild-type ventricle,93.61±20.68% s.d. (n=45 cells)]. This demonstrates that the average wild-type ventricular cell is more extended than the average nok/mpp5 mutant myocardial cell (the average surface extension of mutant cells being derived from both atrial and ventricular myocardial cells). Due to similar cell surface extensions, we did not distinguish between atrial and ventricular cells in nok/mpp5 mutants. In summary, nok/mpp5 mutant myocardial cells do not fully expand in size. Both size reduction of myocardial cells and smaller cell number appear to be important factors contributing to heart tube elongation defects in nok/mpp5 mutants. We conclude that nok/mpp5 is required for the remodeling of myocardial cells during heart tube elongation.

To test whether nok/mpp5 is required tissue-autonomously within the myocardium for correct heart tube elongation, we generated a stable transgenic line of zebrafish that expresses wild-type Nok/Mpp5 under the control of the cmlc2 promoter region Tg(cmlc2:nok). This transgenic line was introduced into the nok^s305mutant background that also contained the Tg(cmlc2:GFP)insertion. Embryos that were homozygous mutant for nok^s305and displayed the full range of the nok/mpp5 mutant phenotypes were genotyped by PCR for the presence of the Tg(cmlc2:nok)insertion. Among 23 transgenic embryos, 15 produced an elongated and rescued heart tube (65.2%). Moreover, whereas in nok/mpp5 mutants, the inflow tract was mostly obstructed, this phenotype was also rescued in these transgenic animals. This rescue rate is comparable to the rescue efficiency with the Tg(cmlc2:prkci) transgene and may be attributed to the variable onset of transgene expression under the control of the cmlc2 promoter region (Huang et al., 2003) (Fig. 6M). Among the homozygous nok^s305 mutants that lacked the rescue transgene, most embryos displayed a shortened and malformed heart tube characteristic of the nok/mpp5 mutant phenotype(Fig. 6B). Therefore, similar to Has/PRKCi, Nok/Mpp5 functions tissue-autonomously within the myocardial layer.

Our data present the first evidence of the role of the apical Crumbs-Nok/Mpp5 complex in heart morphogenesis. We show that zebrafish Nok,which has previously been shown to be required for correct lamination of the neural retina and for the correct polarization of some lateral plate mesodermal cells is a functional homolog of mammalian Pals1/Mpp5 and that it is a regulator of vertebrate embryonic heart morphogenesis(Wei and Malicki, 2002; Horne-Badovinac et al., 2003). Within myocardial cells, Nok/Mpp5 is required for correct junctional localization of Has/PRKCi. We also show that Nok/Mpp5 and Has/PRKCi function autonomously within myocardial cells during heart morphogenesis and that Has/PRKCi function depends on the catalytic kinase activity. Hence, embryonic heart morphogenesis depends on localized Has/PRKCi catalytic activity at the apical junction, and Nok/Mpp5 is essential for its stabilization.

We have demonstrated that has/prkci and nok/mpp5 regulate heart cone tilting, and that nok/mpp5 also controls the subsequent step of heart tube elongation. The mechanism of heart cone tilting is currently not understood but it is conceivable that a directed motion that displaces the heart cone from the midline toward the left anterior side of the embryo may be involved (compare Fig. 4B,C). This myocardial motion could be caused by an intrinsic motive force and/or by external forces. We have shown that during the process of heart cone tilting myocardial cells are strongly converging to produce a narrow and extended cone that subsequently tilts into the anterior-posterior axis (Fig. 4C). This supports the possibility that myocardial cells are actively moving to cause heart cone tilting and that this movement requires the monolayered, polarized and cuboidal organization of myocardial cells at this stage(Fig. 7). Interestingly, loss of has/prkci or nok/mpp5 does not impede the convergence of myocardial cells toward the midline. Therefore, either heart progenitor cells do not require the integration into an epithelial sheet for migration, or the defective myocardium maintains some epithelial cohesion that is mediated by factors other than the ones tested in our study. We also hypothesize that the morphologically specialized population of myocardial cells that connects the heart cone with surrounding tissues may be necessary to convert external motive forces onto the heart cone or to generate an anchor point that may force the heart cone to always tilt in a stereotypical direction(Fig. 4S).

We have observed that prkci^2A mutants frequently displayed a myocardial phenotype that was more severe than in has/prkci morphants. In fact, overexpression of high levels of prkci^2A mRNA in wild-type embryos resulted in a phenotype similar to nok/mpp5 (Fig. 1N,O). Conceivably, PRKCi^2A interferes, in a dominant-negative fashion, with pools of maternal Has/PRKCi or another PRKC isoform that is within protein complexes associated with Nok/Mpp5. Immunohistochemistry performed on has/prkci mutants using an antibody against the 20 C-terminal amino acids of murine PRKCz demonstrated the presence of another PRKC isoform different from zygotic PRKCi(Horne-Badovinac et al.,2001).

As mammalian Pals1 physically interacts via an evolutionarily conserved region with Par6, and as physical association of the two apical protein complexes, Crumbs-Stardust/Mpp5 and Bazooka-DaPKC-DmPar6, has also been mapped in Drosophila, we propose that a similar protein complex is likewise present in zebrafish and that it plays an essential role in zebrafish epithelial maintenance (Wang et al.,2004; Nam and Choi,2003). The results presented in this study indicate a previously unknown function for these proteins in myocardial polarization and remodeling. The mechanisms by which heart tube elongation is regulated and the contribution of nok/mpp5 to the remodeling of atrial myocardial cells from a cuboidal shape (at heart cone stages; Fig. 7) to a squamous and elongated shape during heart tube elongation stages currently remain elusive. Moreover, it is possible that cellular rearrangements of myocardial cells during heart tube elongation that are reminiscent of convergence/extension type movements are essential for this process and that they could be regulated by nok/mpp5. Further studies into the roles of the apical Crumbs-Nok/Mpp5 and Par6-PRKC protein complexes in this process are warranted.

We thank the members of the Abdelilah-Seyfried laboratory for helpful discussions. Various constructs, antibodies and cDNAs were donations from Walter Birchmeier, Debbie Yelon, Didier Stainier, Klaus Rohr, Jochen Wittbrodt and Reinhard Koestner. Many thanks in particular to Herwig Baier and to members of his laboratory for helping to isolate an additional allele of nok, and to Xiangyun Wei for sharing unpublished results. We also thank Ulrike Ziebold, Thomas Willnow and Walter Birchmeier for critical reading of the manuscript. This work was supported in part by a presidential grant from the Helmholtz Society.