Convergence of distinct pathways to heart patterning revealed by the small molecule concentramide and the mutation heart-and-soul

Research Paper 1481

Convergence of distinct pathways to heart patterningrevealed by the small molecule concentramideand the mutation heart-and-soulRandall T. Peterson, John D. Mably, Jau-Nian Chen*and Mark C. Fishman

Background: One of the earliest steps in heart formation is the generation Address: Cardiovascular Research Center,Massachusetts General Hospital, Harvardof two chambers, as cardiogenic cells deployed in the epithelial sheet ofMedical School, 149 13th Street, Charlestown,mesoderm converge to form the nascent heart tube. What guides thisMassachusetts 02129.

transformation to organotypic form is not known.

Present address: *Department of Molecular, CellResults: We have identified a small molecule, concentramide, and a genetic and Developmental Biology, University of

California, Los Angeles, Life Sciences Building,mutation called heart-and-soul (has) that disrupt heart patterning. BothRoom 5109, 621 Charles E. Young Drive South,cause the ventricle to form within the atrium. Here, we show that the hasLos Angeles, California 90095.gene encodes PKC�. The effect of the has mutation is to disrupt epithelial

cell-cell interactions in a broad range of tissues. Concentramide does notCorrespondence: Mark C. Fishman

disrupt epithelial interactions, but rather shifts the converging heart field E-mail: [email protected]. What is shared between the concentramide and has effects is areversal of the order of fusion of the anterior and posterior ends of the Received: 15 June 2001

Revised: 19 July 2001heart field.Accepted: 10 August 2001

Conclusions: The polarity of cardiac tube assembly is a critical determinantPublished: 2 October 2001of chamber orientation and is controlled by at least two distinct molecular

pathways. Combined chemical/genetic dissection can identify nodal points Current Biology 2001, 11:1481–1491in development, of especial importance in understanding the complexpatterning events of organogenesis. 0960-9822/01/$ – see front matter

2001 Elsevier Science Ltd. All rights reserved.

Background make them complementary to mutational analysis [6, 7].What pathways impose global structure and pattern to an For example, their application can be limited to shortorgan? For the vertebrate heart, the key visible step is windows of time, and their targets extend beyond thosechamber morphogenesis, the fashioning of atrium and accessible to gene disruption alone. These abilities haveventricle. The proper orientation of these two functionally proved particularly useful in elucidating pathways in sin-different contractile units guarantees unidirectional flow gle cells. Colchicine, cycloheximide, and phobol ester,of blood beginning with the first beat of the heart. Cham- among others, have proven the power of this approach.ber form thereafter becomes the substrate upon which However, unlike mutagenesis, in which the functionalthe rest of heart development is superimposed. deficit may be ascribed to a specific gene, it is conceivable

that, especially in vivo, a small molecule may perturb alarge set of processes, with widespread biological deficitsOver recent years, much has been learned about the mo-reflecting this concatenation. Thus, we felt that we wouldlecular mechanisms that drive the acquisition of character-initially focus upon chemical effects that disrupt attributesistic atrial and ventricular cell fates [1, 2]. However, steps,of organ patterning in a manner identical to mutations. Ifboth embryological and molecular, that fashion theirthese two very different perturbants (chemical and ge-higher order structure have proved to be more elusive, innetic) do not affect the same pathway, they could revealpart because, unlike cell fate decisions, they can be stud-a key step in development at their point of biologicalied meaningfully only in the living organism.convergence.

In the zebrafish, one successful approach for identifyingIn this report, we demonstrate that the small moleculesteps of cardiogenesis is genetic screening [3–5]. A secondconcentramide, identified by a chemical screen in zebra-approach, which we explore here, is the use of large-scalefish, alters the global organization of the zebrafish heart.screens for small molecule modulators of development.We find that, after exposure to concentramide, both atrialSmall molecules that permeate the zebrafish and disruptand ventricular cells are generated, and chambers form,specific developmental pathways can be identified. Such

reagents have certain advantages that, in principle, might but the ventricle is generated within the atrium. Here, we

1482 Current Biology Vol 11 No 19

Figure 1analyze the morphology of the heart in zebrafish embryosharboring the heart-and-soul (has) mutation, previously iso-lated from a genetic screen because it causes small hearts,and we find that it causes a nearly identical phenotype.

The phenotypic similarity between the hearts of cencen-tramide-treated and has mutant embryos suggests thatthere is an early step of chamber morphogenesis perturbedboth by concentramide and has. We demonstrate thathas mutants harbor a point mutation in an atypical PKC,PKC�, which is known to be involved in cell polarity andthe integrity of epithelial sheets in Caenorhabditis elegansand Drosophila [8, 9]. Consistent with the role of PKC�in C. elegans and Drosophila, we show that its mutation inzebrafish disrupts cell polarity, epithelial layer integrityin many tissues, and ultimately heart patterning. In con-trast, the small molecule concentramide does not appearto affect cell polarity or epithelial layer integrity but ratheralters anterior-posterior (AP) patterning in the anteriorend of the developing zebrafish, including the heart field.

A small molecule that alters heart patterning. (a) The structure ofWhat is shared between the effects of the small molecule concentramide. (b) A lateral view of the mushroom-shaped heart ofconcentramide and mutation of PKC� is a reversal in a live, concentramide-treated embryo 30 hpf. The atrium is indicated

with an “A”, and the ventricle is indicated with a “V”. (c) A timethe order of fusion of the anterior and posterior heartcourse of concentramide effectiveness. Black bars indicate theprimordia, suggesting that this step is critical to properdevelopmental time periods during which groups of embryos were

chamber formation and alignment. The complementarity immersed in water containing concentramide. An “x” indicates thatof genetic and chemical approaches can therefore facilitate treatment during the indicated time period alters the wild-type brain

or heart phenotypes. An “o” indicates that the wild-type phenotypedissection of even relatively late and higher order organo-was observed. Blue and pink boxes mark the critical periods fortypic decisions.development of the brain and heart phenotypes, respectively.

ResultsConcentramide specifically modulates a biologicalpathway involved in heart patterning hearts that do not sustain circulation. It appears that bothZebrafish embryos have recently been shown to be ame- the atrium and ventricle form and beat in a coordinatednable to high-throughput screening to identify small mol- manner in these fish, but the ventricle forms in the centerecules that perturb developmental processes [10]. In one of the atrium, as shown in Figures 2e and 2f. The resultsuch screen, we exposed developing zebrafish embryos is a heart in which the atrium and ventricle form twoto small molecules from a large, diverse chemical library. concentric rings, the inner ring composed of the ventricleVisual inspection of the transparent embryos was used to and the outer ring composed of the atrium. From theidentify small molecules that affect the global patterning dorsal view, the heart looks like a bullseye (Figure 2d),of the heart. One of these small molecules is a biaryl and from the lateral view, it looks like an inverted mush-compound containing an acrylamide moiety that we call room in which the ventricle forms the stalk of the mush-concentramide (Figure 1a, originally identified as library room and the atrium surrounds and covers the ventriclenumber 32P6 [10]). like the top of the mushroom (Figure 1b).

Normally, by 24 hr postfertilization (hpf), the heart tube Several observations suggest that concentramide is aassembles in the midline, with the atrium positioned ante- highly specific modulator of a particular molecular path-rior to the ventricle and slightly displaced toward the left way critical to heart patterning. Concentramide is very(Figure 2a), and blood flow is driven from the atrial to the potent, with an ED50 of about 2 nM. More importantly,ventricular end, first by persistalsis and then by sequential higher doses of concentramide do not appear to causechamber contractions. By 30 hpf, chambers are clearly additional side effects. Concentramide causes virtuallydemarcated (Figure 2b, using cardiac myosin light chain the same phenotype when used at a concentration of 62, cmlc2, to label both chambers) and express different �M as it does when used at a concentration of 6 nM,genes, as shown in Figure 2c (ventricle-specific myosin suggesting that it modulates a specific molecular target

at least 1000 times more potently than it modulates otherheavy chain and atrial-specific antibody S46).proteins affecting visible developmental processes. Theeffect of concentramide on cardiovascular developmentEmbryos exposed to concentramide develop compact

Research Paper Convergent heart patterning pathways in zebrafish Peterson et al. 1483

Figure 2 mental processes. To identify the developmental stage atwhich concentramide disrupts heart patterning decisions,we added concentramide to the water of developing em-bryos at various times. As shown in Figure 1c, embryostreated at any time prior to 14 hpf exhibit the concentricchamber morphology at 24 hpf, whereas embryos treatedafter 17 hpf exhibit the wild-type heart morphology at 24hpf. Repeating the experiment with more precise stagingrevealed that concentramide must be present before the14-somite stage (approximately 15 hpf) to induce the con-centric chamber morphology. Therefore, a developmentalevent occurring at the 14-somite stage is critical for heartpatterning and is disrupted by the small molecule con-centramide.

The hearts of concentramide-treated embryosphenocopy heart-and-soul mutantsheart-and-soul (has) is a mutation isolated in our large-scalegenetic screen [3]. The hearts of homozygous has mutantembryos are small. We find here that, like those of con-Hearts from has mutant embryos phenocopy hearts fromcentramide-treated embryos, the hearts of has mutant em-concentramide-treated embryos. In situ hybridization was performed

with (a–c) wild-type, (d–f) concentramide-treated, and (g–i) has bryos have ventricular tissue within the atrium (Figureembryos. The expression pattern of cardiac myosin light chain 2 2g–i). They manifest radial sequential contractions of the(cmlc2) is shown for embryos (a,d,g) 24 hpf and (b,e,h) 30 hpf. The

atrium, then the ventricle. The has mutant embryos, how-relative locations of the atrium (A) and the ventricle (V) wereconfirmed by 7-�m sagital sections of embryos in which the ventricle ever, also manifest defects in many tissues, including thewas prestained blue by in situ hybridization to ventricle-specific retina, kidney, gut, and brain. These defects are not pres-myosin heavy chain (vmhc), (c,f,i) followed by staining of the atrium ent in concentramide-treated embryos. The brain of con-brown with the atrium-specific antibody S46. In (a), (d), and (g), the view

centramide-treated embryos does develop abnormally,is dorsal and anterior is oriented up. In all other frames, the view islateral and anterior is oriented toward the left. but treating embryos between 9 and 14 hpf eliminates

this brain defect while preserving the concentric heartchamber phenotype (Figure 1c).Therefore, the heart phe-notypes of concentramide-treated and has mutant em-

does not appear to be a result of general cytotoxicity. The bryos are very similar, but concentramide-treated embryosdevelopment of concentramide-treated embryos is not appear to have fewer developmental defects elsewhere,delayed relative to untreated siblings, and no increase in and the cardiac specificity of the phenotype can be in-cell death is apparent. Concentramide also has no effect creased further by controlling the timing of concentramideon the rate of the proliferation of yeast (data not shown) treatment.or bromodeoxyuridine incorporation in mammalian cells(P. A. Clemons and S. L. Schreiber, personal communica-

Heart-and-soul encodes atypical PKC�tion). Given the potency of concentramide, its phenotypicreproducibility over a broad concentration range, and the Given the phenotypic similarities between hearts fromrarity of the phenotype it produces (none of the �2000 has and concentramide-treated embryos, we reasoned thatother small molecules screened generates a similar pheno- cloning the has gene might provide molecular insight intotype), it is likely that concentramide is a specific modula- the process of heart patterning. Furthermore, cloning oftor of a biological pathway responsible for heart pat- has might allow us to determine whether has and con-terning. centramide influence heart patterning through similar or

distinct mechanisms. We mapped has by linkage analysiswith zebrafish SSR markers [11–13] and AFLP [14] to anA time window for concentramide effectsinterval flanked by markers z8451 and z11023 of approxi-One advantage of small molecules over genetic mutationsmately 1.1 cM (see Figure 3a). These were used to initiatein studying a developmental process is that small mole-a walk using YACs and BACs, which proceeded by endcules allow the process to be modulated with much greatercloning, refined mapping, and ultimately sequencing.temporal control. Small molecules can be added or washedGenes identified as candidates for the mutation were as-away at any time during development, whereas geneticsayed by in situ analysis and were assayed for cDNAmutations are generally present throughout development.polymorphism by RT-PCR of wild-type and mutant RNAThis temporal control afforded by small molecules facili-

tates the identification of critical periods for develop- pools. The genes contained within the BACs are shown


Figure 3 formed by another group that they had identified PKC�as the likely protein encoded by the has gene (S. Horneand D. Stainier, personal communication). By sequencingPKC� from wild-type and mutant embryos, we confirmedthat both has alleles harbor mutations in the PKC�-codingsequence. The mutation in the m567 allele causes a pre-mature stop codon after amino acid 518, and the mutationin the m129 allele causes a premature stop codon afteramino acid 514 (Figure 3a). We determined the completegenomic structure of the zebrafish PKC� gene by shotgunsequencing of BAC 23c14. It is comprised of 18 exonsspanning approximately 45 kb. We find PKC� mRNA tobe expressed in a broad range of tissues (data not shown).

The C-terminal truncation of PKC� does not appear todestabilize the protein, since truncated protein is detectedby Western blot analysis of mutant embryos (Figure 3b).However, truncation might be predicted to eliminate adomain essential for PKC� function, given that truncationremoves the predicted autophosphorylation site Thr556and the conserved hydrophobic site Glu575 (Figure 3a).Furthermore, C-terminal truncation of PKC� or PKC�renders these related kinases catalytically inactive [15,16]. In order to confirm the role of PKC� mutation in thephenotype, we injected antisense morpholino oligomerscomplementary to the PKC� translational start site. Theseinjections phenocopy the mutation entirely. The injectedembryos are indistinguishable at the gross morphologicallevel from the genetic mutants (Figure 3c), supportingthe idea that loss of the C-terminal 70 amino acids issufficient to eliminate gene function.

The heart-and-soul gene encodes PKC�. (a) A map of the has interval The integrity of epithelia is affected by PKC� mutation,with the genomic structure of the zebrafish PKC� gene. YAC and but not by treatment with concentramideBAC clones are indicated by addresses beginning with “y” and “b”. PKC� belongs to the large PKC family of kinases and,The BAC clone 23c14 was sequenced to determine the entire

with PKC�, is classified as an “atypical” PKC [17]. Thegenomic structure of the has gene. From the partial sequence of theBACs listed, a preliminary transcript map of the region was presumptive ortholog of PKC� in C. elegans, PKC-3, colo-determined (see Table 1). The zebrafish PKC� gene comprises 18 calizes with Par3 and Par6 at the anterior pole of the one-exons represented by vertical lines. The site of the mutations cell embryo [8, 18]. PKC-3 is necessary for the establish-associated with the m129 and m567 alleles is indicated with an

ment of embryonic polarity, and inactivation of PKC-3asterisk. The domain structure of PKC� is also shown, including theC2-like domain, pseudosubstrate (psub.) domain, C1 domain, and leads to mislocalization of the Par genes and a symmetricalcatalytic domain. The activation loop phosphorylation site (T404), first cell division. Drosophila possess only one atypicalautophosphorylation site (T556), and hydrophobic site (E575) are also

PKC (DaPKC) that also associates with a Par3-like proteinshown. A vertical line marks the truncation site for the has m129 and(Bazooka) and is implicated in the control of cell polaritym567 alleles. (b) An anti-PKC� Western blot of protein extracts from

wild-type embryos (WT), has mutant embryos (m567 �/�), and [9]. DaPKC mutants exhibit disordered epithelial lay-siblings of has mutant embryos (m567 �/� & �/�). (c–e) Antisense ering, irregular cell shapes, and loss of epithelial cell polar-disruption of PKC� expression phenocopies the has mutation. (c)

ity, believed to be due to defects in cell adhesion. InWild-type embryos, (d) has embryos, and (e) wild-type embryosvertebrate cells, PKC� and PKC� both localize to epithe-injected with a PKC� antisense morpholino oligomer were

photographed live 2 days postfertilization. lial tight junctions and associate with a Par3-like protein(ASIP) [19–22]. We therefore examined whether the hasmutation and concentramide treatment perturb epithelialpatterning and tight junctions, focusing upon the retina

in Table 1. The gene assignments are based on BLASTX and the kidney. The neural retina arises from an epithelialalignments. sheet that is bordered by the lens on the basal surface

and by a second epithelial sheet (the retinal pigmentedepithelium, RPE) on the apical surface [23]. Prior to cellDuring the screening of candidate genes, we were in-


Table 1

Candidate genes identified within the heart-and-soul interval.

BAC address Identified genes (GenBank accession number)

109f10/122n17 KIAA0670 protein/acinus (NP_055792)membane-type 1 metalloproteinase precursor (AAD13803)adaptin, gamma (NP_001119)KIAA1416 protein, novel Helicase C-terminal domain and SNF2 N-terminal domain-

containing protein, similar to KIAA0308 (CAB57836)ZPC domain-containing protein 2 (AAD38907)zinc finger protein sal (AAB51127)cerebellin 1 precursor (NP_004343)RING finger protein (AAB05873)

152p21 unknown (NP_056541)89i15 precerebellin-like protein (AAF04305)23c14 PKC�

transforming protein sno-N; chicken (I51298)53c17 no genes detected by BLASTX (mostly repetitive)

differentiation, the nuclei of the neuroepithelial cells mi- of wild-type, has, and concentramide-treated embryos. Asin the retina, cell polarity appeared to be largely conservedgrate between the apical and basal surfaces of the epithe-

lium. During M-phase, cell nuclei localize to the apical in has kidneys (Figure 5a–c). The has kidneys did, how-ever, exhibit irregularities in the shapes of epithelial cellssurface, adjacent to the neighboring RPE [24]. Beginning

at about 30 hpf, these neuroepithelial cells exit the cell and occasional gaps between cells (data not shown), con-sistent with a defect in epithelial cell adhesion. We didcycle and differentiate into one of seven distinct cell types

[25, 26]. Each cell type then migrates to a specific layer not observe these defects in embryos exposed to con-centramide. Similarly, concentramide does not cause ain the retina, resulting in a highly organized, laminar pat-

tern (see Figure 4a). loss of cell asymmetry or Par2 localization during the firstcell division in nematodes (data not shown). Therefore,

The has mutation causes disruption of the layering of the although concentramide treatment and PKC� inactivationboth result in similar heart patterning phenotypes, con-neural retina and patchy loss of the RPE (Figure 4e).centramide does not appear to inactivate zebrafish PKC�These defects resemble those noted previously in zebra-or its nematode ortholog.fish bearing the mutations oko meduzy (ome) and mosaic

eyes (moe)[27, 28]. In has mutants, the severity of laminardisruption correlates with the position and degree of RPE The molecular target of concentramide is involveddiscontinuity, suggesting that the RPE epithelial defect in AP patterningcauses or exacerbates that of the neural retina. This would If the molecular target of concentramide does not affectbe concordant with the evidence that a normal RPE is the continuity of epithelial sheets as PKC� does, by whatcritical to lamination [29, 30] and the fact that the retinal sort of process might it influence heart patterning? Treat-epithelium of has mutants does manifest at least one attri- ment with concentramide appears to affect the relativebute of proper apical-basal polarity in that the majority positions of several anatomical structures along the anterior-of the mitotic nuclei localize correctly to the apical surface posterior (AP) axis. For example, the distance betweenof the neuroepithelium (Figure 4b,d,f; 89% of M-phase Pax2.1-expressing cells in the eyes and at the midbrain/nuclei from has embryos localize to the apical surface hindbrain boundary is reduced in concentramide-treated

embryos (Figure 6a–c). Perhaps more significantly, theversus 97% of nuclei from wild-type embryos). As acardiac myosin light chain 2 (cmlc2)-expressing cells ofmarker of tight junctions, we examined immunoreactivethe heart field are shifted rostrally in concentramide-zonula occludens (ZO-1), an integral tight junction pro-treated embryos at the 18-somite stage. This shift is eventein, and found it to be mislocalized (Figure 4g–h). There-more pronounced than the shift of the midbrain/hindbrainfore, loss of adhesion between RPE cells may be a causeboundary. The heart field lies completely posterior toof retinal mispatterning in has mutants. Notably, retinasthe midbrain/hindbrain boundary in 18-somite wild-typefrom concentramide-treated embryos do not exhibit de-embryos but is partially anterior to the midbrain/hindbrainfects in cell polarity (Figure 4d), RPE continuity, or lami-boundary of 18-somite concentramide-treated embryosnation (Figure 4c)(Figure 6d–e). The distance between the anterior edgeof the cmlc2-expressing field and the anterior extreme ofThe developing kidney is another structure composed of

highly polarized epithelial cells. We examined the distri- the embryo is about 40% greater in wild-type embryos(3.1 0.2 arbitrary units, n 8) than in concentramide-bution of apical and basolateral proteins in the kidneys


Figure 4 Figure 5

The effects of PKC� inactivation and concentramide treatment on thepolarity of the zebrafish kidney. An apical kidney marker (3G8) wasused to stain kidneys of (a) wild-type, (b) concentramide-treated, and(c) has embryos. Transverse 2-�m sections of the pronephric ductare shown.

of concentramide appears to play a role in AP patterning,PKC� is required for lamination, cell polarity, and epithelial cell-cell especially in the positioning of the heart field relative tointeraction in the retina. Transverse 5-�m sections of (a–b) wild- the midbrain/hindbrain boundary, the notochord, and thetype, (c–d) concentramide-treated, and (e–f) has embryos were

anterior extreme of the embryo.stained with (a,c,e) hematoxylin-eosin 5 days postfertilization or with(b,d,f) dapi 30 hpf. Arrowheads indicate mitotic nuclei. Zonulaoccludens-1 localization in the retina is shown by 5-�m transversesections following staining of (g) wild-type or (h) has embryos with PKC� and the target of concentramide both influencean anti-ZO-1 antibody. the fusion order of heart primordia

PKC� and the molecular target of concentramide appearto act via distinct cellular mechanisms, but modulation ofeither results in a very similar change in the patterning

treated embryos (2.2 0.3 arbitrary units, n 12, Figure of the heart. To identify the commonalities between the6f). The position of the heart field in has mutants (3.1 two mechanisms that allow such similar mispatterning of0.3 arbitrary units, n 12) does not differ significantly the heart, we took advantage of the temporal control withfrom the wild-type position. The concentramide-induced which small molecules can modulate biological processes.shift in the heart field alters the location of the field As described above, we determined that embryos mustrelative to signaling centers believed to participate in heart be treated with concentramide at or prior to the 14-somiteformation, including the notochord [31]. The posterior stage to cause formation of the ventricle within the atrium.end of the heart field extends to the tip of the notochord From this observation, we conclude that a critical heartor beyond in wild-type, 16-somite embryos (Figure 6g). In patterning process is initiated shortly after the 14-somiteconcentramide-treated, 16-somite embryos, the posterior stage, and perturbation of this process results in the con-end of the heart field remains anterior to the tip of the centric chamber phenotype observed in both has and con-

centramide-treated embryos. This allowed us to focus ournotochord (Figure 6h). Therefore, the molecular target


Figure 6 Figure 7

The order of anterior and posterior heart field fusion. Dorsal views ofcmlc2 expression at the (a–c) 18-somite and the (d–f) 20-somitestages. Expression patterns for (a,d) wild-type, (b,e) concentramide-treated, and (c,f) has embryos are shown. Anterior is oriented up.

search for commonalities between has and concentramide-treated embryos to this critical time period.

The generation of the primitive heart tube is accom-plished by midline coalescence of the bilateral cardiacprimordial sheets. In the zebrafish, this coalescence firstgenerates a single midline cone, with its base on the yolk[1, 32]. Subsequently, the cone tilts to assume a midlineAP orientation with the preventricular end oriented to-ward the posterior, later swinging anteriorly as yolk isresorbed.

We find that, normally, the generation of the midline conedoes not occur uniformly around the cone’s circumferencebut rather progresses from posterior to anterior, with pos-terior regions merging at the 18-somite stage and anteriorregions merging at the 20- to 22-somite stage. This step

Alterations in anterior-posterior patterning after treatment with is perturbed by both concentramide and the has mutation.concentramide. (a–c) In situ hybridization was used to show Pax2.1 In both has mutant embryos and concentramide-treatedexpression in (a) untreated and (b) concentramide-treated 18-somite

embryos, there is a failure to merge the posterior endsembryos. The expression patterns have been false-colored blue for(Figure 7a–c). Even by the 20-somite stage, when theuntreated embryos and red for concentramide-treated embryos. (c)

An overlay of images (a) and (b). Arrowheads indicate areas of anterior ends of the primordia begin to fuse normally, thePax2.1 expression at the midbrain-hindbrain boundary and in the otic posterior ends remain separated in the has and con-placodes. In (a)–(c), the view is lateral and anterior is oriented toward

centramide-treated embryos (Figure 7d–f). Eventually,the left. (d–e) Double in situ staining of the heart field (F, cmlc2 probe)and Pax2.1-expressing cells (arrowheads) in (d) untreated and (e) the posterior ends do fuse in has and concentramide-concentramide-treated 18-somite embryos. Lateral view; anterior is treated embryos, just before the emergence of the concen-oriented toward the left. (f) The distance between the anterior edge tric-chambered heart. Thus, a critical patterning decisionof the heart field, as defined by cmlc2 in situ staining, and the rostral

occurs at about the 16- to 18-somite stage that regulatesextreme of the zebrafish embryo were measured in wild-type (WT),concentramide-treated (conc.), and has embryos at the 18-somitestage. Error bars represent the standard error. (g–h) Double in situstaining of the heart field (cmlc2) and the notochord (no tail, ntl probe)

expressing cells (arrowheads) in (i) untreated and (j) concentramide-in (g) wild-type and (h) concentramide-treated 16-somite embryos.treated 20-somite embryos. Lateral view; anterior is oriented towardThe view is dorsolateral; anterior is oriented toward the left. Arrowheadsthe left.mark the anterior extreme of the notochord. (i–j) Double in situ

staining of the heart field (F, cmlc2 probe) and �-tropomyosin-


the fusion order of the anterior and posterior ends of epithelium, and the kidney and fail to generate properlamination of the neural retina.the heart field. This process can be blocked either by

inactivation of PKC� or by modulating the target of con-How PKC� deficiency causes epithelial disruption is de-centramide.bated. In C. elegans and Drosophila, PKC� is important tothe establishment of polar cell division in the one-celledDiscussionembryo. In epithelia, PKC� and its Drosophila orthologLike other complex biological processes, organogenesislocalize to tight junctions and adherens junctions, in mam-likely involves the cooperation of several biological path-mals and flies, respectively [9, 22]. These sites of epithe-ways. To understand how organs form, experimental en-lial cell-cell contact are required for maintaining cohesivetrance points to critical positions in those biological path-epithelial sheets [20] and also for controlling the planeways are needed [33]. In the case of the vertebrate heart,of epithelial cell division [37]. Loss-of-function DaPKCwe have been interested in identifying the pathways thatmutants in Drosophila exhibit multilayering of epitheliaorganize differentiated heart tissues into a precisely pat-and gaps in epithelia [9], similar to the effects noted hereterned, functional structure. We have sought informativein the RPE, neural retina, and renal epithelium. However,entrance points into these pathways via two complemen-analysis of the has phenotype does not clarify which, iftary approaches, screening for genetic mutations and,any, of these widespread epithelial deficits contribute tomore recently, screening for small molecules that affectthe cardiac phenotype, especially since neighboring epi-heart patterning in zebrafish.thelia of the neural plate and gut are known to transmitimportant signals to the developing heart [1].

There is no question that genetic screens can reveal boththe overall biological logic and specific key genes of devel- The small molecule concentramide causes the concentricopmental design. However, gene mutation may be unin- heart chamber phenotype by a mechanism that does notformative due to the redundancy of systems supporting involve widespread disruptions of epithelia. Epithelialcritical developmental events, or widespread deformity cell morphology appears normal in concentramide-treatedand early lethality due to effects distant from those of embryos, and no discontinuities in cell-cell contacts areparticular interest. In principle, some of these lacunae evident in the heart, RPE, or kidneys. Tight junctionsmight be filled by a complementary chemical screen. localize properly in concentramide-treated retinas (dataThese have additional benefits in that targets of small not shown), and lamination occurs normally. What con-molecules need not even be proteins, and exposure to centramide appears to alter is the AP patterning of thethe compound can be temporally controlled with great anterior end of the embryo. Perhaps most significantly,precision. the heart field is shifted rostrally in concentramide-treated

embryos, as are the otic placodes and the midbrain-hind-In this report, we define a key step in heart formation by brain boundary. The developmental event affecting heartits perturbation with a small molecule and a mutation. patterning that is modulated by concentramide occursThis step is the proper alignment of the two cardiac cham- shortly after the 14-somite stage; treatment of embryosbers just as the primitive heart tube assembles. Two per- after this stage does not affect heart patterning. This stageturbants, the small molecule concentramide and the has of development coincides with the migration of the bilat-mutation, elicit a previously undescribed chamber malal- eral heart primordia toward each other, prior to their fusionignment, in which the ventricle forms inside of the atrium. to form the heart tube.This means that establishment of the cardiocyte cell fatesis largely accomplished, but the higher order assembly of It is at the critical stage of midline fusion of the bilateralchamber structure is disrupted. heart primordia that the effects of has and concentramide

appear to converge. In both, the normal order of fusionof the anterior and posterior ends of the heart field isWe find the has mutation to be in PKC�. This gene has

been found to play roles in many processes including disrupted. In wild-type embryos, the posterior end of theinsulin-induced glucose uptake [34], the establishment heart field fuses first. Posterior fusion is delayed in hasof embryonic asymmetry in worms [8], and Ras-mediated mutants and concentramide-treated embryos, and the an-reorganization of actin stress fibers [35, 36]. One role of terior end fuses first.possible relevance here is its importance in the mainte-nance and patterning of certain epithelia. This has been This commonality of organotypic mispatterning, arrived

at by what appear to be quite distinctive cellular pathways,noted in Drosophila mutants [9] and in mammalian cells, inwhich expression of kinase inactive atypical PKC disrupts strongly suggests that proper order of fusion of the heart

primordia is a key element to the generation of the cham-tight junction structure [20]. Here, we note that embryoswith the has mutation have structural discontinuities in bers and to their assembly and alignment. In situ expres-

sion analysis using ventricular and whole-heart markersthe heart field (see Figure 2g), the retinal pigmented


Figure 8 [31], and therefore could deprive the migrating cells ofimportant information needed for the proper assembly atthe posterior end of the ring.

Small molecules have proven to be valuable tools for cellbiological studies, but their use as probes in studies ofdevelopment has been limited to a few notable examples(e.g., retinoic acid). The limited use of small moleculesin developmental studies has been due to two factors.First, some of the developmental model systems possessmechanisms that limit their susceptibility to small mole-cules. For example, the mammalian placenta and the C.elegans cuticle prevent many small molecules from reach-ing the developing embryo. This appears to be less of animpediment using zebrafish embryos. Second, many ofthe small molecules used in cell biological studies areA model for chamber patterning in the zebrafish heart based on innatural products. These small molecules are often pro-situ expression analysis (see the Supplementary material available

with this article online). Normally, the bilateral primordia of the heart duced by plants and microorganisms to inhibit the growthfield converge and fuse first at the posterior end and then at the anterior of competing microorganisms, so they target fundamentalend to form a cone. The cone then rotates to orient atrial precursors

cellular processes. Producing small molecule modulatorstoward the anterior and ventricular precursors toward the posterior inof complex processes like organogenesis may be less ad-an extended heart tube. In concentramide-treated and has mutant

embryos, the fusion order of the ends of the heart field is reversed, vantageous to a microorganism, and fewer natural prod-proceeding from the anterior to the posterior end. Rotation of the cone ucts of this sort exist. Recent technological advances areis blocked, preventing formation of the heart tube and causing the

expanding the armamentarium of small molecules for useconcentric heart chamber phenotype. Presumptive atrial precursor cellsare colored red, ventricular precursor cells are colored blue. Views in developmental studies. The techniques of modern syn-are dorsal; anterior is oriented up. thetic chemistry, including combinatorial chemistry, have

increased the number of potential developmental modula-tors available, decreasing reliance on functionally biasednatural products (reviewed in [7, 39]).

suggests that medial cells of the bilateral heart field aredestined to become ventricle, while lateral cells are des-tined to become atrium ([32] and data not shown). In Obviously, disruptions by small molecules may speak im-wild-type embryos, fusion of the heart field proceeds from portantly to toxicological effects upon development. Theyposterior to anterior, transiently forming a concentric might help to categorize and collate the growing compen-structure with ventricular cells along the inside of the ring dium of biologically active molecules, based upon concat-(Figure 8). This structure rapidly undergoes a 90� rotation, enations of in vivo phenotypes. Yet it has been unclearforming the heart tube. In both concentramide-treated how informative such screens might be about normal de-and has mutant embryos, heart field fusion proceeds from velopment. We believe that, as shown here for con-anterior to posterior, and rotation of the heart ring to centramide, small molecule modulators are likely to beform a tube does not occur (Figure 8). We assume that most informative when they combine great potency andventricular cells need to assemble at the posterior end of specificity and are used to complement mutational analy-the ring at the right time and, if blocked in this midline sis. We are developing techniques to identify the molecu-convergence, can later maintain their cell fates and fashion lar target of concentramide, and we expect that technolog-a chamber, but one disoriented with regard to the atrium. ical improvements will continue to simplify the processThe anterior and posterior ends of the heart field are of identifying other small molecule targets [40]. As smallquite different in patterns of genes expressed, in part molecules and their targets become easier to identify,attributable to the proximity to the notochord, so the they will become useful conditional probes for dissectingprocess of chamber morphogenesis could certainly require organogenesis and other complex integrative biologicalexposure to exquisitely timed signals from neighboring processes as well.structures [1, 38]. This could be deficient in has due tointrinsic heart primordial defects, as demonstrated by the

Materials and methodspatchy expression of cmlc2 and concordant with moreSmall molecule treatmentwidespread epithelial dysfunction, or due to aberrant sig-Zebrafish were maintained at 28.5�C as described [41]. Unless specifiednals from the overlying defective neural plate. Con- otherwise, embryos were treated prior to gastrulation by adding con-

centramide causes anterior displacement of the heart field centramide to the egg water at a final concentration of 34 nM from a34-�M stock solution in DMSO.relative to other signaling centers, including the notochord


(ventricle) probes is available at http://images.cellpress.com/supmat/Whole-mount in situ hybridization and immunohistochemistrysupmatin.htm.Digoxigenin-labeled antisense RNA probes were generated by in vitro

transcription for cmlc2 [32], vmhc [32], ntl [42], �-tropomyosin [43],Acknowledgementsand pax2.1 [44]. In situ hybridization was carried out as described [45].We thank J. Miller and S. Tsukita for atrium-specific and ZO-1 antibodies,For whole-mount immunohistochemistry, embryos were fixed in 4% para-S. Horne and D. Stainier for sharing cloning data prior to publication, W.formaldehyde in phosphate-buffered saline (S46 and 3G8) or 80% meth-Winston, A. Kay, and C. Hunter for assistance with C. elegans embryology,anol, 20% dimethyl sulfoxide (�-ZO-1), permeablized in acetone for 30M. Mohideen for help with has mapping, P. Clemons for performing bromo-

min at –20�C (3G8), blocked with 5% fetal bovine serum, and incubated deoxyuridine assays, and S. Schreiber, J. Tallarico, T. Huxford, B. Link, andwith the antibodies S46 (a gift from Dr. J. Miller), 3G8 [46], or �-ZO-1 members of the Fishman lab for advice and helpful discussions. This work(a gift from Dr. S. Tsukita). An anti-mouse horseradish peroxidase conju- was supported by the National Institutes of Health grants HL49579,

HL63206, and DK55383 to M.C.F.gate was used as a secondary antibody for S46 and 3G8, and anAlexa 488-labeled anti-mouse secondary antibody was used for �-ZO-1staining. References

1. Fishman MC, Chien KR: Fashioning the vertebrate heart:earliest embryonic decisions. Development 1997, 124:2099-Histology2117.Fixed embryos were dehydrated, embedded in plastic (JB-4, Poly-

2. Srivastava D, Olson EN: A genetic blueprint for cardiacsciences), and sectioned at 2–7 �m. Retinal sections were stained with development. Nature 2000, 407:221-226.hematoxylin-eosin or dapi. 3. Stainier DY, Fouquet B, Chen JN, Warren KS, Weinstein BM, Meiler

SE, et al.: Mutations affecting the formation and function ofthe cardiovascular system in the zebrafish embryo.Cloning of hasDevelopment 1996, 123:285-292.Embryos were separated into mutant and wild-type pools based on

4. Fishman MC, Olson EN: Parsing the heart: genetic modules forphenotypic analysis. Genomic DNA was isolated from individual embryos organ assembly. Cell 1997, 91:153-156.by incubation in DNA isolation buffer overnight at 50�C (DNA isolation 5. Chen JN, Haffter P, Odenthal J, Vogelsang E, Brand M, van Eedenbuffer: 10 mM Tris-HCl [pH 8.3], 50 mM KCl, 0.3% Tween 20, 0.3% FJ, et al.: Mutations affecting the cardiovascular system andNonidet P40, 0.5 mg/ml proteinase K). Proteinase K was inactivated other internal organs in zebrafish. Development 1996, 123:293-

302.prior to PCR setup by heating samples to 98�C for 10 min. PCR reactions6. Mitchison TJ: Towards a pharmacological genetics. Chem Biolwere performed using diluted genomic DNA as described [47].

1994, 1:3-6.7. Schreiber SL: Chemical genetics resulting from a passion forRNA was isolated (RNeasy columns, Qiagen) from pools of wild-type synthetic organic chemistry. Bioorg Med Chem 1998, 6:1127-

and mutant embryos to generate cDNA for RT-PCR analysis (SMART 1152.RACE cDNA amplification kit, Clontech). Fragments were then sub- 8. Tabuse Y, Izumi Y, Piano F, Kemphues KJ, Miwa J, Ohno S, et al.:cloned into PCRII-TOPO (Invitrogen). PCR primers were synthesized Atypical protein kinase C cooperates with PAR-3 to

establish embryonic polarity in Caenorhabditis elegans.based on sequence from an EST for PKC� (fc69h04, GenBank acces-Development 1998, 125:3607-3614.sion number AI883774) and genomic sequence (Genome Systems,

9. Wodarz A, Ramrath A, Grimm A, Knust E: Drosophila atypicalBAC clone address 23c14) and were used to sequence the entireprotein kinase C associates with Bazooka and controlsPKC�-coding region and 3� UTR.polarity of epithelia and neuroblasts. J Cell Biol 2000,150:1361-1374.

Genomic clones were isolated by PCR analysis of DNA pools from BAC 10. Peterson RT, Link BA, Dowling JE, Schreiber SL: Small molecule(Genome Systems) and YAC (Research Genetics) libraries using primer developmental screens reveal the logic and timing ofsets for the linked markers z11023 and z8451. YAC end sequence was vertebrate development. Proc Natl Acad Sci USA 2000,

97:12965-12969.determined as described [48]. BAC ends were sequenced directly using11. Michelmore RW, Paran I, Kesseli RV: Identification of markersSP6 and T7 primers, and BACs 53c17, 89i15, and 152p21 were sub-

linked to disease-resistance genes by bulked segregantcloned by shotgun cloning of partial AluI-digested fragments into pBlues-analysis: a rapid method to detect markers in specificcript. For the complete sequencing of BACs, a hydroshear was used togenomic regions by using segregating populations. Proc

produce fragments of 2–3 kb. These fragments were then blunt-end Natl Acad Sci USA 1991, 88:9828-9832.ligated into pGEM5 (Promega) and sequenced using an ABI3700 to 12. Knapik EW, Goodman A, Ekker M, Chevrette M, Delgado J, Neuhaussgenerate approximately 5-fold coverage. The sequence was assembled S, et al.: A microsatellite genetic linkage map for zebrafishusing Phred/Phrap/Consed [49–51]. (Danio rerio). Nat Genet 1998, 18:338-343.

13. Shimoda N, Knapik EW, Ziniti J, Sim C, Yamada E, Kaplan S, et al.:Zebrafish genetic map with 2000 microsatellite markers.Western blotting Genomics 1999, 58:219-232.

Groups of 25 embryos were lysed in 0.5% Triton X-100 in phosphate- 14. Vos P, Hogers R, Bleeker M, Reijans M, van de Lee T, Hornes M, etbuffered saline. Lysates were clarified by centrifugation and separted al.: AFLP: a new technique for DNA fingerprinting. Nucleicby 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis. Acids Res 1995, 23:4407-4414.Western blotting was performed using an �-PKC rabbit polyclonal anti- 15. Riedel H, Hansen H, Parissenti AM, Su L, Shieh HL, Zhu J, et al.:

Phorbol ester activation of functional rat protein kinase Cbody (Santa Cruz Biotechnology).beta-1 causes phenotype in yeast. J Cell Biochem 1993, 52:320-329.

Morpholino injection 16. Riedel H, Su L, Hansen H: Yeast phenotype classifiesAn antisense morpholino oligonucleotide of sequence 5�-CTGTCC mammalian protein kinase C cDNA mutants. Mol Cell Biol 1993,CGCAGCGTGGGCATTATGG-3� (GeneTools, LLC) was dissolved at 13:4728-4735.

17. Mellor H, Parker PJ: The extended protein kinase C superfamily.a concentration of 100 �M in 1� Danieau’s buffer (5 mM HEPES [pHBiochem J 1998, 332:281-292.7.6], 58 mM NaCl, 0.7 mM KCl, 0.6 mM Ca(NO3)2, 0.4 mM MgSO4).

18. Hung TJ, Kemphues KJ: PAR-6 is a conserved PDZ domain-A total of 1 nL of this solution or 1� Danieau’s buffer was injected intocontaining protein that colocalizes with PAR-3 ineach 1- to 4-cell embryo before allowing the embryos to develop atCaenorhabditis elegans embryos. Development 1999, 126:127-28.5�C. 135.

19. Joberty G, Petersen C, Gao L, Macara IG: The cell-polarity proteinSupplementary material Par6 links Par3 and atypical protein kinase C to Cdc42. NatSupplementary material including a figure showing in situ expression Cell Biol 2000, 2:531-539.

20. Suzuki A, Yamanaka T, Hirose T, Manabe N, Mizuno K, Shimizu M,analysis of the developing heart using cmlc2 (whole heart) and vmhc


et al.: Atypical protein kinase c is involved in the zebrafish paired box gene pax[zf-b] during earlyneurogenesis. Development 1991, 113:1193-1206.evolutionarily conserved par protein complex and plays a

45. Oxtoby E, Jowett T: Cloning of the zebrafish krox-20 gene (krx-critical role in establishing epithelia-specific junctional20) and its expression during hindbrain development.structures. J Cell Biol 2001, 152:1183-1196.Nucleic Acids Res 1993, 21:1087-1095.21. Lin D, Edwards AS, Fawcett JP, Mbamalu G, Scott JD, Pawson T,

46. Vize PD, Jones EA, Pfister R: Development of the Xenopuset al.: A mammalian PAR-3-PAR-6 complex implicated inpronephric system. Dev Biol 1995, 171:531-540.Cdc42/Rac1 and aPKC signalling and cell polarity. Nat Cell

47. Knapik EW, Goodman A, Atkinson OS, Roberts CT, Shiozawa M,Biol 2000, 2:540-547.Sim CU, et al.: A reference cross DNA panel for zebrafish22. Izumi Y, Hirose T, Tamai Y, Hirai S, Nagashima Y, Fujimoto T, et al.:(Danio rerio) anchored with simple sequence lengthAn atypical PKC directly associates and colocalizes at thepolymorphisms. Development 1996, 123:451-460.epithelial tight junction with ASIP, a mammalian homologue

48. Zhong TP, Kaphingst K, Akella U, Haldi M, Lander ES, Fishman MC,of Caenorhabditis elegans polarity protein PAR-3. J Cellet al.: Zebrafish genomic library in yeast artificialBiol 1998, 143:95-106.chromosomes. Genomics 1998, 48:136-138.23. Schmitt EA, Dowling JE: Early eye morphogenesis in the

49. Gordon D, Abajian C, Green P: Consed: a graphical tool forzebrafish, Brachydanio rerio. J Comp Neurol 1994, 344:532-542.sequence finishing. Genome Res 1998, 8:195-202.24. Sauer FC: Mitosis in the neural tube. J Comp Neurol 1935,

50. Ewing B, Green P: Base-calling of automated sequencer traces62:377-405.using phred. II. Error probabilities. Genome Res 1998, 8:186-25. Altshuler D, Turner D, Cepko C: Specification of cell type in the194.vertebrate retina. In Development of the Visual System. Edited

51. Ewing B, Hillier L, Wendl MC, Green P: Base-calling of automatedby Lam DM-K, Shatz C. Cambridge, MA: The MIT Press; 1991:37-sequencer traces using phred. I. Accuracy assessment.58.Genome Res 1998, 8:175-185.26. Dowling JE: The Retina. Cambridge, MA: Belknap Press; 1987.

27. Jensen AM, Walker C, Westerfield M: mosaic eyes: a zebrafishgene required in pigmented epithelium for apicallocalization of retinal cell division and lamination.Development 2001, 128:95-105.

28. Malicki J, Driever W: oko meduzy mutations affect neuronalpatterning in the zebrafish retina and reveal cell-cellinteractions of the retinal neuroepithelial sheet. Development1999, 126:1235-1246.

29. Raymond SM, Jackson IJ: The retinal pigmented epithelium isrequired for development and maintenance of the mouseneural retina. Curr Biol 1995, 5:1286-1295.

30. Vollmer G, Layer PG, Gierer A: Reaggregation of embryonicchick retina cells: pigment epithelial cells induce a highorder of stratification. Neurosci Lett 1984, 48:191-196.

31. Goldstein AM, Fishman MC: Notochord regulates cardiaclineage in zebrafish embryos. Dev Biol 1998, 201:247-252.

32. Yelon D, Horne SA, Stainier DY: Restricted expression of cardiacmyosin genes reveals regulated aspects of heart tubeassembly in zebrafish. Dev Biol 1999, 214:23-37.

33. Fishman MC: The genomic cosmos. Nature 2001, 410:1033.34. Kotani K, Ogawa W, Matsumoto M, Kitamura T, Sakaue H, Hino Y,

et al.: Requirement of atypical protein kinase clambda forinsulin stimulation of glucose uptake but not for Aktactivation in 3T3-L1 adipocytes. Mol Cell Biol 1998, 18:6971-6982.

35. Hellbert K, Kampfer S, Maly K, Hochholdinger F, Mwanjewe J, BaierG, et al.: Implication of atypical protein kinase C isozymeslambda and zeta in Ras mediated reorganization of the actincytoskeleton and cyclin D1-induction. Adv Enzyme Regul2000, 40:49-62.

36. Coghlan MP, Chou MM, Carpenter CL: Atypical protein kinasesClambda and -zeta associate with the GTP-binding proteinCdc42 and mediate stress fiber loss. Mol Cell Biol 2000,20:2880-2889.

37. Lu B, Roegiers F, Jan LY, Jan YN: Adherens junctions inhibitasymmetric division in the Drosophila epithelium. Nature2001, 409:522-525.

38. Serbedzija GN, Chen JN, Fishman MC: Regulation in the heartfield of zebrafish. Development 1998, 125:1095-1101.

39. Schreiber SL: Target-oriented and diversity-oriented organicsynthesis in drug discovery. Science 2000, 287:1964-1969.

40. King RW: Chemistry or biology: which comes first after thegenome is sequenced? Chem Biol 1999, 6:R327-333.

41. Westerfield M: The Zebrafish Book. Guide for the Laboratory Useof Zebrafish (Danio rerio). Eugene, OR: University of Oregon Press;1995.

42. Schulte-Merker S, Ho RK, Herrmann BG, Nusslein-Volhard C: Theprotein product of the zebrafish homologue of the mouseT gene is expressed in nuclei of the germ ring and thenotochord of the early embryo. Development 1992, 116:1021-1032.

43. Ohara O, Dorit RL, Gilbert W: One-sided polymerase chainreaction: the amplification of cDNA. Proc Natl Acad Sci USA 1989,86:5673-5677.

44. Krauss S, Johansen T, Korzh V, Fjose A: Expression of the

Documents

Convergence of distinct pathways to heart patterning revealed by the small molecule concentramide and the mutation heart-and-soul