Epithelial integrity requires the adhesion of cells to each other as well as to an underlying basement membrane. The modulation of adherence properties is crucial to morphogenesis and wound healing, and deregulated adhesion has been implicated in skin diseases and cancer metastasis. Here, we describe zebrafish that are mutant in the serine protease inhibitor Hai1a (Spint1la),which display disrupted epidermal integrity. These defects are further enhanced upon combined loss of hai1a and its paralog hai1b . By applying in vivo imaging, we demonstrate that Hai1-deficient keratinocytes acquire mesenchymal-like characteristics, lose contact with each other, and become mobile and more susceptible to apoptosis. In addition, inflammation of the mutant skin is evident, although not causative of the epidermal defects. Only later, the epidermis exhibits enhanced cell proliferation. The defects of hai1 mutants can be phenocopied by overexpression and can be fully rescued by simultaneous inactivation of the serine protease Matriptase1a(St14a), indicating that Hai1 promotes epithelial integrity by inhibiting Matriptase1a. By contrast, Hepatocyte growth factor (Hgf), a well-known promoter of epithelial-mesenchymal transitions and a prime target of Matriptase1 activity, plays no major role. Our work provides direct genetic evidence for antagonistic in vivo roles of Hai1 and Matriptase1a to regulate skin homeostasis and remodeling.

Hepatomegaly is a sign of many liver disorders. To identify zebrafish mutants to serve as models for hepatic pathologies, we screened for hepatomegaly at day 5 of embryogenesis in 297 zebrafish lines bearing mutations in genes that are essential for embryonic development. Seven mutants were identified, and three have phenotypes resembling different liver diseases. Mutation of the class C vacuolar protein sorting gene vps18 results in hepatomegaly associated with large, vesicle-filled hepatocytes,which we attribute to the failure of endosomal-lysosomal trafficking. Additionally, these mutants develop defects in the bile canaliculi and have marked biliary paucity, suggesting that vps18 also functions to traffic vesicles to the hepatocyte apical membrane and may play a role in the development of the intrahepatic biliary tree. Similar findings have been reported for individuals with arthrogryposis-renal dysfunction-cholestasis(ARC) syndrome, which is due to mutation of another class C vps gene. A second mutant, resulting from disruption of the tumor suppressor gene nf2 ,develops extrahepatic choledochal cysts in the common bile duct, suggesting that this gene regulates division of biliary cells during development and that nf2 may play a role in the hyperplastic tendencies observed in biliary cells in individuals with choledochal cysts. The third mutant is in the novel gene foie gras , which develops large, lipid-filled hepatocytes, resembling those in individuals with fatty liver disease. These mutants illustrate the utility of zebrafish as a model for studying liver development and disease, and provide valuable tools for investigating the molecular pathogenesis of congenital biliary disorders and fatty liver disease.

Polycystic kidney disease (PKD) is a common human genetic illness. It is characterized by the formation of multiple kidney cysts that are thought to result from over-proliferation of epithelial cells. Zebrafish larvae can also develop kidney cysts. In an insertional mutagenesis screen in zebrafish, we identified 12 genes that can cause cysts in the glomerular-tubular region when mutated and we cloned 10 of these genes. Two of these genes, vhnf1 ( tcf2 ) and pkd2 , are already associated with human cystic kidney diseases. Recently, defects in primary cilia have been linked to PKD. Strikingly, three out of the 10 genes cloned in this screen are homologues of Chlamydomonas genes that encode components of intraflagellar transport (IFT) particles involved in cilia formation. Mutation in a fourth blocks ciliary assembly by an unknown mechanism. These results provide compelling support for the connection between cilia and cystogenesis. Our results also suggest that lesions in genes involved in cilia formation and function are the predominant cause of cystic kidney disease, and that the genes identified here are excellent candidates for novel human PKD genes.

We identified four insertional alleles of foxi one ( foo ),an embryonic lethal mutation in zebrafish that displays defects in both otic placode and the jaw. In foo/foo embryos the otic placode is split into two smaller placodes and mutant embryos show a dorsoventral (DV)cartilage defect manifested as a reduced hyomandibular and reduced third and fourth branchial arches. We identified foxi one ( foo ), the zebrafish ortholog of Foxi1 (FREAC6, FKHL10, HFH-3, Fkh10) and a member of the forkhead domain transcriptional regulator family, as the gene mutated in foo/foo embryos. foo is expressed in otic placode precursor cells, and foo/foo embryos lack placodal pax8 expression and have disorganized otic expression of pax2.1 and dlx3 . Third stream neural crest cell migration, detected by dlx2 and krox20 expression, is aberrant in that it invades the otic placode territory. foo is expressed in pharyngeal pouch endoderm and is required for pouch expression of pax8 and proper patterning of other markers in the pouch such as nkx2.3 . In foo/foo embryos, we observed a failure to maintain fgf3 expression in the pouches,followed by apoptosis of neural crest cells in adjacent arches. We conclude that foo expression is essential for pax8 expression probably downstream of Fgf signaling in a conserved pathway jointly required for integrity of patterning in the otic placode and pharyngeal pouches. We propose that correct placement of survival/proliferation cues is essential for shaping the pharyngeal cartilages and that evolutionary links between jaw and ear formation can be traced to Fgf-Foxi1-Pax8 pathways.

The molecular genetic mechanisms of cartilage construction are incompletely understood. Zebrafish embryos homozygous for jellyfish ( jef )mutations show craniofacial defects and lack cartilage elements of the neurocranium, pharyngeal arches, and pectoral girdle similar to humans with campomelic dysplasia. We show that two alleles of jef contain mutations in sox9a , one of two zebrafish orthologs of the human transcription factor SOX9 . A mutation induced by ethyl nitrosourea changed a conserved nucleotide at a splice junction and severely reduced splicing of sox9a transcript. A retrovirus insertion into sox9a disrupted its DNA-binding domain. Inhibiting splicing of the sox9a transcript in wild-type embryos with splice site-directed morpholino antisense oligonucleotides produced a phenotype like jef mutant larvae, and caused sox9a transcript to accumulate in the nucleus; this accumulation can serve as an assay for the efficacy of a morpholino independent of phenotype. RNase-protection assays showed that in morpholino-injected animals, the percent of splicing inhibition decreased from 80% at 28 hours post fertilization to 45% by 4 days. Homozygous mutant embryos had greatly reduced quantities of col2a1 message, the major collagen of cartilage. Analysis of dlx2 expression showed that neural crest specification and migration was normal in jef ( sox9a )embryos. Confocal images of living embryos stained with BODIPY-ceramide revealed at single-cell resolution the formation of precartilage condensations in mutant embryos. Besides the lack of overt cartilage differentiation,pharyngeal arch condensations in jef ( sox9a ) mutants lacked three specific morphogenetic behaviors: the stacking of chondrocytes into orderly arrays, the individuation of pharyngeal cartilage organs and the proper shaping of individual cartilages. Despite the severe reduction of cartilages, analysis of titin expression showed normal muscle patterning in jef ( sox9a ) mutants. Likewise, calcein labeling revealed that early bone formation was largely unaffected in jef ( sox9a ) mutants. These studies show that jef ( sox9a ) is essential for both morphogenesis of condensations and overt cartilage differentiation.

In early embryonic development, the brain is divided into three main regions along the anteroposterior axis: the forebrain, midbrain and hindbrain. Through retroviral insertional mutagenesis and chemical mutagenesis experiments in zebrafish, we have isolated mutations that cause abnormal hindbrain organization and a failure of the midbrain-hindbrain boundary (MHB) to form, a region that acts as an organizer for the adjacent brain regions. The mutations fail to complement the spiel-ohne-grenzen ( spg ) mutation, which causes a similar phenotype, but for which the affected gene is unknown. We show through genetic mapping, cloning of the proviral insertion site and allele sequencing that spg mutations disrupt pou2, a gene encoding the Pou2 transcription factor. Based on chromosomal synteny, phylogenetic sequence comparison, and expression and functional data, we suggest that pou2 is the zebrafish ortholog of mouse Oct3/Oct4 and human POU5F1. For the mammalian genes, a function in brain development has so far not been described. In the absence of functional pou2 , expression of markers for the midbrain, MHB and the hindbrain primordium (pax2.1 , wnt1 , krox20 ) are severely reduced, correlating with the neuroectoderm-specific expression phase of pou2 . Injection of pou2 mRNA restores these defects in spg mutant embryos, but does not activate these markers ectopically, demonstrating a permissive role for pou2 . Injections of pou2 -morpholinos phenocopy the spg phenotype at low concentration, further proving that spg encodes pou2 . Two observations suggest that pou2 has an additional earlier function: higher pou2 -morpholino concentrations specifically cause a pre-gastrula arrest of cell division and morphogenesis, and expression of pou2 mRNA itself is reduced in spg -homozygous embryos at this stage. These experiments suggest two roles for pou2 . Initially, Pou2 functions during early proliferation and morphogenesis of the blastomeres, similar to Oct3/4 in mammals during formation of the inner cell mass. During zebrafish brain formation, Pou2 then functions a second time to activate gene expression in the midbrain and hindbrain primordium, which is reflected at later stages in the specific lack in spg embryos of the MHB and associated defects in the mid- and hindbrain.

Despite extensive studies, there are still many unanswered questions regarding the mechanism of hedgehog signaling and the phylogenic conservation of hedgehog function in vertebrates. For example, whether hedgehog signaling in vertebrates requires smoothened is unclear, and the role of hedgehog activity in zebrafish is controversial. We show that inactivation of smoothened by retroviral insertions in zebrafish results in defects that are characteristic of hedgehog deficiencies, including abnormalities in body size, the central nervous system, adaxial mesoderm, cartilage and pectoral fins. We demonstrate that, as in Drosophila , vertebrate smoothened is essential for hedgehog signaling, and functions upstream of protein kinase A. Further analysis of neural tube defects revealed the absence of lateral floor plate and secondary motoneurons, but the presence of medial floor plate and primary motoneurons in smoothened mutant embryos. Blocking maternal hedgehog signaling by cyclopamine eliminates primary motoneurons, but not medial floor plate. Interestingly, even after inhibition of maternal hedgehog activity, the midbrain dopaminergic neurons still form, and looping of the heart does not randomize in the mutants. We also found decreased proliferation and increased apoptosis in the mutants. Taken together, these data demonstrate the conserved role of vertebrate smoothened in the hedgehog signaling pathway, and reveal similarities and differences of hedgehog function between teleosts and amniotes.