Cadherins are homotypic cell-cell adhesion molecules that regulate morphogenesis and differentiation (Marrs and Nelson, 1996). E-cadherin is essential for early embryogenesis in mouse (Larue et al., 1994; Riethmacher et al., 1995). Detailed analysis of E-cadherin expression in embryogenesis was described for mouse, chick, and Xenopus (Broders et al., 1993; Choi and Gumbiner, 1989; Hirai et al., 1989a,b; Levi et al., 1991; Shimamura and Takeichi, 1992; Shimamura et al., 1994; Takeichi, 1988; Thiery et al., 1984). E-cadherin expression in the ectoderm is downregulated during neurulation (Takeichi, 1988). However, discrete regions of E-cadherin expression are activated in the developing and adult brain (Choi and Gumbiner, 1989; Shimamura and Takeichi, 1992; Shimamura et al., 1994; Thiery et al., 1984). The functional significance of neural E-cadherin expression is unclear. Cadherin expression patterns implicate these molecules in regionalization processes during nervous system development (Redies, 1997; Redies and Takeichi, 1996; Yagi and Takeichi, 2000). A combinatorial code has been proposed that describes how differential cell–cell adhesion segregates neuron groups into functional circuits (Redies and Takeichi, 1996).
In the mouse and chick but not in Xenopus, E-cadherin is expressed in the MHB (Choi and Gumbiner, 1989; Shimamura and Takeichi, 1992; Shimamura et al., 1994; Thiery et al., 1984). In zebrafish, mutations in the Pax2.1 gene (no isthmus or noi) and in the Fgf8 gene (acerebellar or ace) disrupt MHB development (Brand et al., 1996). When formed, the MHB displays organizer activities (Marin and Puelles, 1994; Martinez et al., 1991), generating signals that induce the development of cerebellum, tegmentum, and optic tectum. wnt1 mutations in mouse disrupt MHB development (McMahon and Bradley, 1990; McMahon et al., 1992; Thomas and Capecchi, 1990; Thomas et al., 1991), and wnt1 distributions suggest a similar role in MHB formation in zebrafish (Kelly and Moon, 1995), but mutations in the zebrafish wnt1 gene have not been identified. wnt1 mutations in mice expand E-cadherin and reduce αN-catenin expression domains in the developing diencephalon and mesencephalon (Shimamura et al., 1994). E-cadherin expression domains were reduced in wnt1 mutant mice in a limited domain in the ventral midbrain (Shimamura et al., 1994).
In this study, zebrafish E-cadherin cDNAs were cloned and developmental expression patterns were characterized. At the pharyngula stage (24 hpf), cdh1 expression was found in the brain, including the MHB. Disrupting developmental signaling that controls MHB formation also blocked cdh1 expression in this brain region.
Zebrafish E-Cadherin cDNA Cloning and Expression in Early Development
To identify cadherins expressed during early neural development, a 15–19 hpf embryo cDNA library (generously provided by Dr. Bruce Appel) was screened using a pan-cadherin antibody (Marrs et al., 1993). Two cDNAs (clones 26 and 28) were recovered with significant sequence identity to E-cadherins from other species. Complete sequencing showed that these cDNAs overlap, encoding most of the cadherin, except the N-terminus. 5′-RACE was used to recover the remaining coding sequences for this cadherin (Fig. 1A). Sequence comparison, phylogenetic analysis (W.J. Gallin and J.A. Marrs, unpublished data) and developmental expression pattern (see below) indicate that we identified the zebrafish E-cadherin orthologue (cdh1 gene designation). Radiation hybrid analysis using the LN54 panel (Hukriede et al., 1999) localized cdh1 gene to linkage group 7 (lod score 13.1; 39.01 cR from Z20576).
cdh1 expression was detected in Northern blots of RNA from cleavage stage embryos and expression was maintained throughout development (data not shown). To determine whether zebrafish E-cadherin is a maternal message, total oocyte RNA was probed by RT-PCR using cdh1 specific primers. cdh1 signal was strongly detected (Fig. 1B). wnt1 is not maternally loaded and was not detected in oocyte RNA (Fig. 1B). As a positive control, cdh1 and wnt1 were detected in 16 hpf RNA (Fig. 1B).
All blastomeres expressed high levels of cdh1 (Fig. 2A,B), until gastrulation began. A subtle concentration of cdh1 expression was consistently detected in the shield (Fig. 2B), a region equivalent to the organizer in amphibian embryos and the node in avian and mammalian embryos. As gastrulation proceeded, cdh1 expression was detected at the midline and in the anterior mesendoderm (Fig. 2C,D). At 10.5 hpf, E-cadherin message was found in the forming polster (a mesendodermal derivative), concentrated in the anterior adaxial mesendoderm in a V-shaped pattern (Fig. 2E,F).
In 16 hpf embryos, cdh1 expression was seen in the anterior mesendoderm and in the polster, the most anterior portion of the prechordal plate (Fig. 2G). At this stage, cdh1 message was also detected in the forming otic placode (Fig. 2G,H) and the forming nephric duct (Fig. 2H). Staining was not observed in the developing nervous system at this stage.
The pharyngula stage in zebrafish begins at 24 hpf and represents the phylotypic stage, where the body plan has been laid down and the brain has been segmented into the five main divisions (telencephalon, diencephalon, mesencephalon, metencephalon, and myelencephalon). In 24 hpf embryos, cryostat sections and whole mount in situ hybridization showed that cdh1 was expressed in the ectoderm, developing pronephros, gut, otic vesicle, and developing brain (Fig. 3B–E). Expression was found in the telencephalon, diencephalon, and rhombencephalon and was prominent at the MHB (Fig. 3A–C), a structure that acts as an organizer for development of the adjacent optic tectum, tegmentum, and cerebellum (Martinez et al., 1991; Marin and Puelles, 1994; Brand et al., 1996).
cdh1 Expression Is Controlled by MHB Formation and Ventral Midline Signaling Pathways
A signaling pathway that includes Wnt1, Pax2.1, and Fgf8 controls zebrafish MHB formation (Brand et al., 1996; Kelly and Moon, 1995). In gastrulation stage embryos, wnt1 and pax2.1 expression codistributes in a chevron-shaped domain that represents the presumptive MHB (Kelly and Moon, 1995) (Fig. 4B). At this early time, cdh1 expression was not detected in this domain (Fig. 4A). At 22 hpf, cdh1 expression was detected in the hindbrain, but no cdh1 signal was detected at the MHB (Fig. 4E). At 24 hpf, cdh1 and wnt1 were coexpressed in the developing MHB (Fig. 4G,H).
To examine the effects of disrupting ventral midline signaling and pax2.1 expression on E-cadherin expression in the developing nervous system, cyclops mutant embryos (Hatta et al., 1991) were studied. In cyclops mutant embryos, ventral midline signaling is defective, producing a graded effect on ventral neural induction that is greatest in the forebrain, more moderate in the midbrain, and least in the hindbrain (Hatta et al., 1994). In the forebrain ventral midline of cyclops mutant embryos, the optic stalk is reduced, and development of retinal tissue expands along the ventral margin of the diencephalon, giving rise to an enlarged, single, fused eye field (Hatta et al., 1991, 1994). In cyclops mutant embryos, pax2.1 expression is decreased and pax6.1 expression is expanded in the eye/optic stalk and throughout the neural axis (Macdonald et al., 1995). In the most severe mutant phenotype, pax2.1 message was not detected in the MHB, and these embryos showed little MHB formation (Fig. 5F). In severe cyclops mutant embryos, cdh1 was also not detectable in the MHB (Fig. 5E). Forebrain cdh1 staining was lost in cyclops mutant embryos displaying both moderate and severe phenotypes, but hindbrain cdh1 staining was maintained (Fig. 5C,E). In embryos displaying a less severe cyclops mutant phenotype, pax2.1 staining was reduced but detectable in the MHB (Fig. 5D), and cdh1 staining in the MHB was correspondingly reduced (Fig. 5C).
cdh1 Expression During Early Embryogenesis Indicates a Role in Gastrulation, Morphogenetic Movements, and the Formation of Signaling Centers
cdh1 message was concentrated in the shield, maintained in the anterior mesendoderm during gastrulation and concentrated in a V-shaped, anterior adaxial region. Previous studies have implicated cadherin function in promoting cell migrations and in maintaining associations between cells during morphogenetic movements that accompany gastrulation (Levine et al., 1994; Yamamoto et al., 1998). cdh1 expression in anterior mesendodermal cells during gastrulation indicates a similar role of regulating cell movements and maintaining adhesion among this cell population. These anterior mesendodermal cells are implicated in neural induction processes and organizing the anterior-posterior axis (Bally-Cuif and Boncinelli, 1997). E- cadherin adhesion may sequester this cell population during these signaling events.
cdh1 was detectable in the developing brain, notably in the MHB, a signaling center that induces the formation of the optic tectum and tegmentum of the mesencephalon and the cerebellum of the metencephalon (Brand et al., 1996; Joyner, 1996; Marin and Puelles, 1994; Martinez et al., 1991). E-cadherin adhesion may help maintain the integrity of this organizing center.
Regulation of cdh1 During MHB Formation and by cyclops
Zebrafish cdh1 was expressed in the MHB at the pharyngula stage. Temporal analysis of wnt1 and cdh1 expression showed that cdh1 expression was delayed by over 12 hr from the onset of wnt1 (and pax2.1) expression (Kelly and Moon, 1995). Therefore, any actions of wnt1 and pax2.1 that lead to the activation of cdh1 neural expression must be indirect or have a significant temporal delay.
Analysis of cyclops mutant embryos showed that cdh1 expression in the MHB correlated with pax2.1 expression and MHB differentiation. In the forebrain, cdh1 expression was severely downregulated in embryos displaying moderate and severe cyclops mutant phenotypes. Previous analysis of cyclops mutants suggested that effects were limited to ventral brain structures (Hatta et al., 1994). However, cdh1 expression was noted in the epiphysis, a dorsal structure, and cdh1 expression in this domain was significantly affected in cyclops mutant embryos.
Adult zebrafish were obtained from a local pet store and maintained under standard laboratory conditions (Westerfield, 1995). Developmental ages were given as hours postfertilization (hpf). cyclops (cyc) mutant embryos were obtained by breeding adult zebrafish heterozygous for cyclopsb16 (provided to Q.L. and P.A.R. by Dr. John Kuwada with permission of the University of Oregon Zebrafish Mutant Stock Center).
Library Screening, DNA Sequencing, and Radiation Hybrid Analysis
A 15 to 19 hpf zebrafish embryo cDNA expression library in lZAP (generously provided by Dr. Bruce Appel, Vanderbilt University) was screened with anti-E-cadherin cytoplasmic domain library antibody B.5 (Marrs et al., 1993) as described (Liu et al., 1999). All clones were fully sequenced within the Biochemistry Biotechnology Facility at Indiana University School of Medicine. Sequence analysis was performed using Vector NTI Suite (InforMax, North Bethesda, MD) and Basic Local Search Alignment Tool (BLAST) (Altschul et al., 1990) at the National Center for Biotechnology Information web site (http://www.ncbi.nlm.nih.gov/BLAST/). Radiation hybrid analysis was performed on the LN54 panel (Hukriede et al., 1999) using primers ctgcacctcagtacagacctcg and ctttccactgagccatcgcaac.
5′-Rapid Amplification of cDNA Ends (5′-RACE)
Total RNA was isolated from 24 hpf embryos using Trizol Reagent (Life Technologies Inc., Rockville, MD) using a kit (Roche Molecular Biochemicals, Indianapolis, IN) and primers: GSP1 (160 nM) (ttaccagtgttcctgtgcca) for first strand synthesis; GSP2 (ggaataaactcaggtggttc) for PCR; and GSP2 (tgaccaggctgttctcgtg) for nested PCR. A band of approximately 1.7 kb was excised, purified with Geneclean III (Bio 101, Vista, CA), and this fragment was cloned into pCR2.1 vector using the Original TA Cloning Kit (Invitrogen, Carlsbad, CA).
Northern Blotting and Analysis of Maternal RNA by RT-PCR
Total RNA from embryos was isolated using Trizol Reagent (Life Technologies Inc.). Northern blotting was performed as described (Ausubel et al., 1995). For RT-PCR, first strand cDNA was prepared from 2 μg total RNA using SuperScript II reverse transcriptase (Life Technologies, Inc.) and a cdh1 specific antisense primer (cccttgtcaccagcaatgat) or wnt1 specific antisense primer (cctgctccattcgagtcttg). In negative controls, reverse transcriptase was excluded. Samples were then RNase treated (Sigma), purified using the High Pure PCR Kit (Roche Molecular Biochemicals), and amplified by 35 cycles of PCR using cdh1 specific primers (ccaagatccaccatctccaa and cccttgtcaccagcaatgat) or wnt1 specific primers (actgtcgcatcagagatgca and cctgctccattcgagtcttg).
Probe Synthesis and In Situ Hybridization
Digoxigenin-labeled or fluorescein-labeled riboprobes were synthesized from cDNA as run-off transcripts from linearized templates using the Genius System DIG RNA Labeling Kit (Roche) or the Genius System DIG RNA Labeling Kit and fluorescein RNA Labeling Mix (Roche). Template cDNA of cdh1 clone 26 in plasmid BlueScript was linearized with EcoRI and transcribed with T7 RNA polymerase to generate antisense probe, or linearized with XhoI and then transcribed with T3 RNA polymerase to generate sense probe. wnt1 cDNA in plasmid pGEM1 (provided by Dr. Randall Moon, Howard Hughes Medical Institute, University of Washington) was linearized with EcoRI and transcribed with SP6 RNA polymerase to generate antisense probe, or linearized with HindIII and transcribed with T7 RNA polymerase to generate sense probe. Full-length zebrafish pax2.1 cDNA was obtained by RT-PCR, and subcloned into pCRII-TOPO (Invitrogen, Carlsbad, CA). HindIII was used to linearize the pax2.1 plasmid and T7 polymerase was used for the transcription of antisense probe. In situ hybridization of wholemount and thin sectioned zebrafish embryos was then performed as described (Liu et al., 1999). Images were digitally acquired with a color SPOT camera (Diagnostic Instruments Inc, Sterling Heights, MI) mounted on either a Leica MZ12 dissecting stereomicroscope (Leica Microsystems, Inc., Deerfield IL) or a Nikon Diaphot equipped with DIC optics (Nikon, Inc., Melville, NY).
The authors thank Dr. Bruce Appel (Vanderbilt University) for providing the cDNA library, and we thank Dr. Randall Moon (Howard Hughes Medical Institute, University of Washington) for providing wnt1 cDNA. We also thank Dr. John Kuwada for cyclops mutant strain. This work was supported by National Institute of Health grants to James Marrs (EY11365) and Pam Raymond (EY04318). Sherry Babb was supported by a postdoctoral fellowship from Prevent Blindness America/Fight For Sight. Qin Liu was supported by a NRSA award (EY06892) from the National Institute of Health.