Wnt1 and Wnt10b Function Redundantly at the Zebrafish Midbrain

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Available online at www.sciencedirect.com

Developmental Biology 254 (2003) 172Ð187 www.elsevier.com/locate/ydbio

wnt1 and wnt10b function redundantly at the zebraﬁsh midbrainÐhindbrain boundary

Arne C. Lekven,a,* Gerri R. Buckles,a Nicholas Kostakis,b and Randall T. Moonb

a Department of Biology, Texas A&M University, 3258 TAMU, College Station, TX 77843-3258, USA b Howard Hughes Medical Institute, Department of Pharmacology and Center for Developmental Biology, Box 357750, University of Washington School of Medicine, Seattle, WA 98195, USA Received for publication 28 May 2002, revised 8 October 2002, accepted 1 November 2002

Abstract Wnt signals have been shown to be involved in multiple steps of vertebrate neural patterning, yet the relative contributions of individual Wnts to the process of brain regionalization is poorly understood. Wnt1 has been shown in the mouse to be required for the formation of the midbrain and the anterior hindbrain, but this function of wnt1 has not been explored in other model systems. Further, wnt1 is part of a Wnt cluster conserved in all vertebrates comprising wnt1 and wnt10b, yet the function of wnt10b during embryogenesis has not been explored. Here, we report that in zebrafish wnt10b is expressed in a pattern overlapping extensively with that of wnt1. We have generated a deficiency allele for these closely linked loci and performed morpholino antisense oligo knockdown to show that wnt1 and wnt10b provide partially redundant functions in the formation of the midbrainÐhindbrain boundary (MHB). When both loci are deleted, the expression of pax2.1, en2, and her5 is lost in the ventral portion of the MHB beginning at the 8-somite stage. However, wnt1 and wnt10b are not required for the maintenance of fgf8, en3, wnt8b, or wnt3a expression. Embryos homozygous for the wnt1–wnt10b deficiency display a mild MHB phenotype, but are sensitized to reductions in either Pax2.1 or Fgf8; that is, in combination with mutant alleles of either of these loci, the morphological MHB is lost. Thus, wnt1 and wnt10b are required to maintain threshold levels of Pax2.1 and Fgf8 at the MHB. © 2003 Elsevier Science (USA). All rights reserved.

Keywords: Zebraﬁsh; wnt1; wnt10b; MidbrainÐhindbrain boundary; Acerebellar; No isthmus

Introduction Subsequent steps of neural A/P patterning appear to involve the refinement of initial A/P domains into smaller Patterning of the vertebrate nervous system requires the subdivisions. These steps of vertebrate neural patterning input from members of the Wnt family of secreted glyco- and morphogenesis are dependent on at least three Wnts. proteins at many points. For example, during neural induc- For example, in the mouse, the wnt3a locus is required for tion, a Wnt signal from the paraxial mesoderm is required to the formation of the hippocampus (Lee et al., 2000). Studies send a “posteriorizing” signal that establishes anteriorÐpos- of the regulation of the mouse emx2 promoter suggest that terior (A/P) polarity and induces the expression of posterior inputs of wnt3a in addition to wnt8b (Richardson et al., neural markers (i.e., posterior to forebrain; Bang et al., 1999) are involved in dorsal telencephalon formation (Theil 1999; McGrew et al., 1995). In zebrafish, the expression of et al., 2002). Further, in the zebrafish, the wnt8b locus wnt8 in the paraxial mesoderm has been shown to be the (Kelly et al., 1995) has been shown to be involved in earliest part of this “posteriorizing” signal (Erter et al., establishing subdivisions within the anterior neural plate 2001; Lekven et al., 2001). (Houart et al., 2002; Kim et al., 2002). In this case, regional inhibition of wnt8b activity permits the expression of telencephalic fates, but the mechanism that establishes telence- * Corresponding author. Fax: ϩ1-979-845-2891. phalic identity downstream of wnt8b inhibition is not E-mail address: [email protected] (A.C. Lekven). known.

Genetic studies of wnt1 in the mouse have shown that the et al., 1992a; Postlethwait et al., 1998). (Note: zebrafish formation of a portion of the midbrain and anterior hind- wnt10b has been referred to as wnt10br; because this locus brain, a region that encompasses a structure called the isth- represents the ortholog of other vertebrate wnt10b loci, we mus or midbrainÐhindbrain boundary (MHB), is dependent will refer to this locus as zebrafish wnt10b.) The linkage of on this locus (McMahon and Bradley, 1990; Thomas and wnt1 and wnt10b paralogs has been conserved in the verte- Capecchi, 1990; Thomas et al., 1991). This structure, a brate lineage, and this relationship may reflect the existence constriction in the neural tube at the junction between the of an ancient cluster of three wnt paralogs present before the midbrain and hindbrain, serves as an important signaling divergence of protostomes and deuterostomes; wnt1, wnt6, center for patterning the midbrain and anterior hindbrain and wnt10 (Nusse, 2001). The conservation of the spatial (Rhinn and Brand, 2001; Wassef and Joyner, 1997). In this arrangement of wnt1 and wnt10b raises the possibility that case, the function of wnt1 may, at least in part, be to regulate their expression may be regulated in a coordinated fashion, the expression of engrailed homologs, as expression of for example, similar to the vox and vent loci in Xenopus mouse en1 under the control of a wnt1 enhancer element can (Rastegar et al., 1999) and zebrafish (Melby et al., 2000). In rescue the wnt1 knockout phenotype (Danielian and Mc- support of this assertion, a fragment of the zebrafish wnt10b Mahon, 1996). Thus, Wnt signaling is required to pattern gene showed an expression pattern that may be similar to the early neurectoderm and also to establish a signaling that shown by wnt1 (Krauss et al., 1992a). If zebrafish center that will further pattern brain subdivisions. wnt10b and wnt1 are expressed in similar patterns and are at In zebrafish, mutagenesis screens have identified three least partially functionally redundant, this could explain the loci required for the formation of the MHB (Brand et al., lack of an identified wnt1 point-mutant recovered from two 1996). acerebellar (ace) mutants, which carry mutations in large-scale mutagenesis screens (Driever et al., 1996; Haff- the fgf8 gene (Reifers et al., 1998), fail to develop a MHB ter et al., 1996). and cerebellum and lose the expression of MHB markers We have analyzed the relationship of wnt1 and wnt10b during early somite stages (Reifers et al., 1998), no isthmus embryonic expression in zebrafish in addition to addressing (noi) mutants, which carry mutations in the pax2.1 gene their functions through a genetic loss-of-function approach (Brand et al., 1996; Lun and Brand, 1998), also fail to form (Lekven et al., 2000). We have found that zebrafish wnt10b the MHB, but also lack a portion of the midbrain, the optic is expressed in a pattern substantially overlapping that of tectum (Brand et al., 1996). In both of these mutants, the wnt1, including the time of onset of expression. We gener- expression of several markers for the prospective MHB is ated a zebrafish line carrying a deficiency allele for wnt1, initiated properly but quickly disappears, indicating that wnt10b, and neighboring loci and show that, in contrast to these genes are required for the early maintenance of gene mouse wnt1, zebrafish wnt1 and wnt10b are not required for expression at the MHB (Lun and Brand, 1998). spiel ohne the formation of the midbrain and cerebellum. However, gretzen (spg) mutants also fail to form the MHB, and similar these loci are required for the maintenance of the expression to ace mutants, have enlarged optic tecta (Brand et al., of several genes at the MHB, with the notable exceptions of 1996). The spg locus has recently been shown to encode fgf8, en3, wnt8b, and wnt3a. Through a combination of zebrafish Pou2 (Belting et al., 2001; Burgess et al., 2002), rescue experiments and morpholino antisense oligo knock- and analysis of early markers of the MHB domain show that down, we also show that wnt1 and wnt10b are at least spg is also required for the early maintenance of gene partially functionally redundant in their capacity to regulate expression (Reim and Brand, 2002). The function of spg gene expression at the MHB. Finally, we show through may be to mediate regional competence in the neurectoderm double mutant analyses that zebrafish lacking wnt1 and to Fgf8 signals (Reim and Brand, 2002). Thus, a complex wnt10b, though they retain expression of pax2.1 and fgf8, genetic network exists that includes at least Pax2.1, Fgf8, are sensitized to reductions in the levels of these genes. Pou2, and Wnt1, to perform several aspects of MHB formation: initiation of the MHB program, maintenance of gene expression, and morphogenesis (Rhinn and Brand, Materials and methods 2001). During two large-scale mutagenesis screens in the ze- Fish care and maintenance brafish, only the ace, noi, and spg loci were identified as being required for MHB formation (Driever et al., 1996; Routine fish care and maintenance were performed as Haffter et al., 1996); no wnt1 mutants were apparently described (Westerfield, 1995). Wild type fish used were of recovered. However, wnt1 has been isolated in zebrafish and the AB strain, or were a derivative of the AB strain we refer was shown to be expressed in a conserved pattern similar to to as KWT (Lekven et al., 2001). The alleles used in this other vertebrate wnt1 homologs (Molven et al., 1991). In- study were: noitu29a and aceti282a. To suppress pigmentation, terestingly, zebrafish wnt1 maps to linkage group (LG) 23, embryos were raised in medium containing 0.2 mM PTU in close proximity to another Wnt paralog, wnt10b (Krauss (Sigma). 174 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187

Molecular biology and histology MHB. Embryos were photographed, and portions of their tails were excised to provide DNA for PCR genotyping. For histology, 48-h embryos were embedded in Araldite Recombination of PAC 176A22 was performed according 502 (Polysciences), and 3- to 5-␮m sections were cut with to a previously published report (Jessen et al., 1998). We a glass knife. Sections were stained with methylene blue/ generated the wnt1 targeting construct in the following toluidine blue and mounted in Permount. Acridine orange steps: a 2-kb EcoRVÐXmnI fragment containing 5ЈUTR just (Sigma) staining was performed by adding 1 ␮l saturated upstream of the wnt1 initiation codon was ligated into acridine orange solution to dechorionated embryos in ϳ10 EcoRV digested pBluescript SKϩ (Stratagene). A clone ml fish water. Embryos were incubated for2hatRT,rinsed with the insert in the proper orientation was selected; then, two to three times, then fluorescence was observed with a the insert was removed by digestion with HindIII ϩ BamHI FITC filter. In situ hybridizations were performed essen- and ligated into pBE.GFPLT CS2ϩ (J. Miller and R.T.M., tially as described (Oxtoby and Jowett, 1993). cDNA probes unpublished observations). The resulting 5Ј wnt1-GFP mod- used in this study are: wnt1 (Kelly and Moon, 1995), pax2.1 ule was excised with HindIII ϩ XbaI and ligated into (Krauss et al., 1992b), fgf8 (Reifers et al., 1998), her5 HindIII/XbaI- digested pRM4N (Jessen et al., 1998). An (Mu¬ller et al., 1996), en2 (Fjose et al., 1992), en3 (Ekker et EcoRIÐMscI fragment from pBR322 containing the tetracy- al., 1992), wnt8b (Kelly et al., 1995), and wnt3a (Krauss et cline resistance gene was then ligated into the pRM4N-5Ј al., 1992a). To perform PCR mapping of Df w5, DNA was wnt1-GFP construct that had been digested to generate an prepared from individual 24- to 48-h mutant embryos or EcoRI end and a blunted XhoI end. The targeting construct wild type siblings as described (Westerfield, 1995). Primer (p5BRT3) was completed by ligation of a PstIÐHindIII sequences for mapped loci are available from the Zebrafish fragment containing wnt1 exon3 into the pRM4N-5Ј wnt1- Information Network (ZFIN), the Zebrafish International GFP-tetr construct digested with KpnI ϩ XbaI. Resource Center, University of Oregon, Eugene, OR (URL: http://zfin.org/). cDNA for 5ЈRACE was prepared from 12- PAC recombination to 24-hpf embryo poly (A)-selected RNA according to the specifications of the kit used (Marathon cDNA Amplifica- The targeting vector p5BRT3 was digested with NotIto tion Kit; Clontech). Primers for RACE were derived from remove the origin of replication and was gel purified. PAC the PCR product previously used for mapping the wnt10b 176A22 was electroporated into Escherichia coli strain locus to LG23 (Postlethwait et al., 1998). PCR products MC1061 (recϩ), and a colony was selected that showed were ligated into pGEM-Teasy (Promega) and sequenced intact PAC DNA as judged by comparison of EcoRI restric- by using Big Dye (Applied Biosystems). Sequence assem- tion digests. CaCl2-competent cells were prepared from this bly and analysis was performed with the Wisconsin GCG colony and were transformed with 100Ð400 ng linearized package or Sequencher (Gene Codes Corp.). ClustalW targeting DNA as described (Dabert and Smith, 1997). Col- alignment was performed via the PBIL Web site: http:// onies that were kanr, tetr, and amps were then tested by pbil.ibcp.fr/. Southern blotting and PCR to confirm proper homologous recombination. PAC identification, rescue, and recombination vector construction Morpholinos

Clones from a PAC library (Amemiya and Zon, 1999) The wnt1 morpholino (gift of D. Raible) had the following positive for wnt1 or wnt10b were identified by PCR. The sequence: 5Ј-AAGCAACGCGAGAACCCGCATGATA-3Ј, genomic organization of wnt10b was determined by se- and the sequence of the wnt10b morpholino was: 5Ј-GGT quencing from PAC 66D18 (sequencing of PAC DNA was AACTCCATTGCGTCGAACGCTT-3Ј, where bases in performed with the BigDye kit; Applied Biosystems, ac- bold correspond to the initiation codon. Morpholinos were cording to manufacturer’s instructions), by sequencing of dissolved in 1ϫ Danieau’s buffer as recommended (Gene plasmid subclones from PAC 176A22, and also by virtual Tools, LLC) to concentrations of 1 mg/ml. Then, 1Ð5nlof walking: cDNA sequences were blasted against genomic morpholino was injected per embryo. DNA sequences in the Ensembl Trace Server database (Sanger Institute; URL: http://trace.ensembl.org/perl/ ssahaview) to generate genomic contigs. For rescue exper- Results iments, PAC DNA was diluted in water to 40Ð50 ng/␮l. Then, 1Ð5 nl diluted DNA was injected into one- to two- The zebrafish wnt10b expression domain mirrors cell-stage embryos collected from crosses between Df w5 that of wnt1 carriers. Embryos were fixed at 24 hpf and processed for in situ hybridization with a pax2.1 probe. After hybridization, In order to examine the sequence and expression pattern embryos were scored for pax2.1 expression, with attention of zebrafish wnt10b, we used 5Ј RACE to amplify the being paid to the presence of expressing cells in the ventral wnt10b coding region from 12- to 24-h embryonic cDNA. A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 175

Fig. 1. ClustalW alignment of Wnt10b peptide sequences from zebraﬁsh (GenBank accession number AY182171), pufferﬁsh (Fugu), human, and mouse. Asterisks below sequence indicate identities; colons indicate conservative substitutions. Dashes in sequence are introduced gaps; arrow indicates Wnt10b truncation point in PAC 66D18.

We were able to design primers based on the wnt10br 3Ј tial expression pattern of zebrafish wnt10b in comparison to untranslated region (UTR) sequence that had been used to that of wnt1, and we found a high degree of similarity map the locus to Linkage Group (LG) 23 (Postlethwait et between the two (Fig. 2). Beginning from the onset of al., 1998). We amplified a 1.5-kb product (not shown) that expression at approximately 80% epiboly, wnt1 and wnt10b proved to contain the entire open reading frame of zebrafish are expressed in similar domains within the prospective wnt10b in addition to both 5Ј and 3Ј UTR sequences. Com- MHB. We have not been able to detect a difference in the parison of the zebrafish Wnt10b protein sequence to that onset of their expression by in situ or by RT-PCR (data not from other vertebrates shows a high degree of conservation shown). By 100% epiboly, wnt1 (Fig. 2A) and wnt10b (Fig. (Fig. 1), with the zebrafish and pufferfish orthologs showing 2F) are expressed in two stripes that converge at the dorsal ϳ80% amino acid identity. Structurally, zebrafish wnt10b midline. In comparison to other wnts expressed at this stage, comprises four exons in comparison to five for the puffer- double labeling for both wnt1 and wnt8b shows that the fish. Genomic sequence corresponding to the position of wnt8b domain does not completely overlap that of wnt1 at intron 3 in the pufferfish wnt10b is not spliced out of the the midline (Fig. 2A, inset). Also at this stage, wnt3a ex- mature zebrafish transcript, resulting in a polypeptide con- pression is found in two domains on either side of the taining a unique stretch of ϳ40 amino acids (Fig. 1). In midline, but with a greater gap between the medial edge of comparison, zebrafish and Fugu wnt1 homologs are 92% each expression domain and the midline than seen for wnt8b identical, they have the same number of amino acids, and (data not shown). Thus, wnt1 and wnt10b are expressed in the gene exonÐintron structure is conserved (not shown). the prospective MHB at early stages with a unique expres- We performed in situ hybridizations to examine the spa- sion domain at the midline. 176 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187

Fig. 2. Comparison of wnt1 and wnt10b expression patterns. In situ hybridizations to detect wnt1 (AÐE) or wnt10b (FÐJ) transcripts. In all images except (E) and (J), anterior is to the left. (A, F) Expression at 100% epiboly, dorsal view. Expression of both wnt1 and wnt10b forms a chevron pattern marking the prospective MHB. Inset in (A), double in situ for wnt8b (blue) and wnt1 (red). Arrow indicates region of nonoverlap at the midline. (B, G) Lateral views of embryos at the 16- somite stage. Expression of both genes marks the MHB and dorsal hindbrain/spinal cord (arrows), and both have anterior limits at the prospective epiphysis (asterisks). wnt10b expression is visible in the prospective cerebellum, but wnt1 is not (white arrows). (CÐE, HÐJ) Embryos at approximately 30 hpf. (C, H) Lateral views of heads. wnt1 and wnt10b are both expressed from the epiphysis (asterisks) to the MHB. wnt10b is more strongly expressed at the posterior edge of the cerebellum (arrowheads), but both genes show enrichment in the rhombomeres (arrows). (D, I) Dorsal views of the same embryos as in (C, H). Brackets denote the MHB, and asterisks indicate the hindbrain ventricle. Note that wnt10b expression in the MHB appears slightly weaker than wnt1, but both are expressed on the anterior medial edge of the fold. (E, J) Posterior views of the MHBs of the same embryos as in (C, H). wnt10b expression appears generally weaker, but clear differences are seen in the ventral MHB where both genes are expressed in a stripe above the ventral midline (arrows), but wnt10b appears to be absent immediately above (asterisks).

During somitogenesis stages, wnt1 (Fig. 2B) and wnt10b mutagenesis screen to generate deficiency alleles (Lekven et (Fig. 2G) expression domains are seen at the dorsal midline al., 2000). In this screen, we recovered one allele of the midbrain extending to the prospective epiphysis (as- [Df(LG23)wnt1w5; hereafter referred to as Df w5] that deletes terisk in Fig. 2B and G) and also at the dorsal midline of the both wnt1 and wnt10b (Fig. 3A). PCR tests on homozygous hindbrain (arrows in Fig. 2B and G). At this stage, the first mutant DNA using flanking markers from the LN54 radia- clear difference in expression of wnt1 and wnt10b is visible tion hybrid panel (Hukriede et al., 1999) showed that the in that wnt10b transcripts can be detected in the prospective deletion spans the loci fb98d06 to fb02e09. These loci have cerebellum, while wnt1 cannot (Fig. 2B and G, white ar- centiRay (cR) values of 261.60 and 265.47, respectively, rows). These general expression domains persist, and at which sets the upper limit of the deletion size at 3.76 cR. On approximately 30 hours postfertilization (hpf; wnt1: Fig. this map, 1 cR corresponds to approximately 150 kb 2CÐE; wnt10b: Fig. 2HÐJ), the expression of both genes is (Hukriede et al., 1999), translating the size of the deletion to seen in the epiphysis, dorsal midline of the optic tectum, the potentially less than 500 kb. The number of genes in this anterior half of the MHB constriction, and in the hindbrain region is unknown, but this interval does include the walls (arrows in Fig. 2C and H). At this stage, differences taram-a locus, which encodes a TGFbeta-related type I between wnt1 and wnt10b expression patterns can be seen. receptor (Renucci et al., 1996) implicated in endoderm wnt10b appears to be more strongly expressed than wnt1 at development (Aoki et al., 2002). Thus, Df w5 represents a the posterior edge of the cerebellum (Fig. 2H, arrowhead; deficiency for both wnt1 and wnt10b in addition to other compare with wnt1 in Fig. 2C), but its expression is weaker loci. than that of wnt1 in the ventral region of the MHB (compare We examined the morphology of embryos derived from Fig. 2J with 2E). Thus, the expression domains of wnt1 and crosses between two Df w5 carriers (Fig. 3BÐG). All em- wnt10b are highly similar. This similarity in expression bryos from such crosses appear grossly normal until approx- patterns raises the possibility that they may be functionally imately 48 hpf, despite multiple loci being removed by the redundant, similar to other zebrafish gene pairs (for exam- deficiency, and by 5 days postfertilization (dpf), mutants ple, Imai et al., 2001). show localized defects (Fig. 3B and C). Df w5 is a recessive embryonic lethal allele since ϳ25% of embryos from such Df w5 isadeficiency for the wnt1 and wnt10b loci crosses show an obvious mutant phenotype at 48 hpf that shows a strict segregation with the deficiency (63 mutant/ In order to understand the function of zebrafish wnt1 and 237 total embryos ϭ 26.6%; genotypes confirmed by PCR; wnt10b during embryogenesis, we performed a gamma-ray data not shown). The most characteristic phenotype of mu- A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 177

Fig. 3. Df(LG23)wnt1w5 isadeficiency for wnt1 and wnt10b. (A) PCR analysis of wild type (ϩ/ϩ) and homozygous mutant DNA (Df/Df). Loci indicated are from the LN54 radiation hybrid panel (see http://zfish.uoregon.edu/ZFIN for mapping panel information). (B, C) Lateral views of 5-dpf embryos, anterior to the left. Melanocytes form normally in Df w5 embryos (arrows), somites appear normally shaped (asterisks), and the gut tube appears normal (arrowheads), indicating that tissues from all three germ layers are able to form normally. Defects appear most noticeably in the head, as Df w5 embryos have reduced eyes (e). (DÐG) Wild type and Df w5 embryos at the prim-25 stage (36 hpf). Viewed laterally (D, E), wild type and homozygous mutants are almost indistinguishable (arrows indicate the cerebellum). When viewed dorsally (F, G), an increased distance between the medial edges of the MHB fold is visible in the mutants (double arrows). Additionally, cell death is apparent in the optic tectum (arrow in G). (Insets) Acridine orange staining of 24-hpf embryos. (H, I) Sagittal sections of 48-hpf embryos. Arrow indicates abnormal structure in the Df w5 tegmentumÐ hindbrain interface. tc, tectum; tg, tegmentum; hb, hindbrain; tel, telencephalon; di, diencephalon; ov, otic vesicle; v, ventricle. tant embryos is an abnormal blood circulation pattern in et al., 2002). Mutant embryos that survive to 5 dpf display which blood flow through the trunk and tail is absent (not abnormalities likely resulting from the edema (Fig. 3B and shown). The circulation phenotype and resulting edema of C). Thus, with the exception of vbg, Df w5 does not appear Df w5 mutants is due to the loss of the violet beuaregarde to delete any zygotic genes that are required globally for locus (vbg), which also maps within the deficiency (Roman early embryogenesis. 178 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187

Fig. 4. Analysis of MHB markers in Df w5 embryos. (A, B, E, F, I, J, M, N, Q, R) 12-somite stage embryos. (C, D, O, P) 27 hpf. (G, H, K, L, S, T) 24 hpf. Embryo genotypes are indicated above each column, and probes assayed are indicated to the right of each row. (U, V) 27-hpf Df w5 embryos; probes are indicated. (All panels) Lateral views of heads, anterior to the left. Reduction in pax2.1 in Df w5 is ﬁrst noticeable as a weaker staining at the MHB at the 12-somite stage (B, arrow). (A, B) Insets: Posterior views of MHB pax2.1 expression. Arrow in (B) inset indicates dorsal pax2.1 expression. At 27 hpf, ventral expression of pax2.1 is absent (arrowheads in C, D) leaving only a spot of pax2.1 expression between the cerebellum (arrows in C, D, G, H, K, L) and optic tectum. The reduction in en2 is more clearly seen as a loss of ventral expression at 12 somites (compare arrows in E, F), but at 24 hpf, en2 expression mirrors that of pax2.1 (compare G, H with C, D). The reduction of her5 in 12-somite-stage Df w5 embryos is visible as a reduction in the entire expression domain (arrows in J indicate thinner band of her5 expression). At 24 hpf, loss of ventral her5 in Df w5 embryos is evident (L, arrowhead). In contrast, fgf8 expression frequently appears normal in Df w5 embryos at 12 somites (M, N) and 24 hpf (O, P; asterisk indicated MHB expression domain). Similarly, en3 expression in 12-somite Df w5 embryos is indistinguishable from wild type (Q, R), but an overall reduction in expression is clear at 24 hpf (S, T; compare width of band between arrows). (U) Expression of wnt3a is indistinguishable between wild type and Df w5. (V) Expression of wnt8b is indistinguishable between wild type and Df w5.

Because wnt1 and wnt10b are expressed in the central indicates that these are dying cells (Fig. 3F and G, insets). nervous system (CNS) and wnt1 mutant mice have abnor- Second, a very slight difference can be seen at the MHB in malities in the formation of the isthmus (McMahon and that the distance between the medial edges of the MHB Bradley, 1990; Thomas and Capecchi, 1990; Thomas et al., constriction is slightly greater in mutants than in wild type 1991), we examined more closely the brains of homozygous (Fig. 3F and G, double arrows). A third structural defect is Df w5 mutant embryos (Df w5 embryos) for subtle visible apparent in histological sections of 48-hpf embryos (Fig. 3H abnormalities (Fig. 3DÐG). From crosses between Df w5 and I). Sagittal sections in a plane near the midline highlight carriers, we were unable to visually detect any differences the fact that there is a fissure at the junction of the tegmen- among embryos younger than 24 hpf. At 24 hpf, however, tum and hindbrain (Fig. 3H, arrow). This fissure is not we were able to detect two subtle differences that segregate clearly visible in Df w5 mutants (Fig. 3I, arrow). These with the deficiency. First, with DIC optics, pronounced embryos also display enlarged ventricles, which is a conse- rounded cells are visible on the optic tectum (arrow in Fig. quence of the circulation defect from the loss of vbg (Ro- 3G). These cells stain strongly with acridine orange, which man et al., 2002). As described below, the junction of the A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 179

Table 1 Rescue of Dfw5 by injection of LG23 PACs

Injected DNA wnt loci present concentration n No. Rescued % Rescue on PAC (ng/␮l) Df/Df Df/Df a Uninjected 69 11 0 0 66D18 wnt1 40 120 29 8 27 176A22 wnt1, wnt10b 40 55 18 10 55 176A22 wnt1, wnt10b 50 48 12 6 50 176⌬G16 wnt10b 70 60 17 8 47

a Rescue was defined as restoration of ventral pax2.1 expression. tegmentum and hindbrain corresponds to the region of gene PCR; data not shown). The first observable change in Df w5 misregulation due to the loss of wnt1 and wnt10b. In all embryos is a reduction in pax2.1 expression beginning at the other respects, Df w5 embryos at this stage appear grossly 8-somite stage (not shown). The reduction in pax2.1 expres- normal. sion is clearer at the 12-somite stage (Fig. 4A and B) and is visible as a reduction in the ventralmost portion of the The expression of MHB markers is not maintained neural tube (Fig. 4A and B, insets). The reduction in pax2.1 in Df w5 embryos expression is followed by similar reductions in en2 expression (Fig. 4E and F) and a slight but consistent reduction in We next examined the expression of several MHB mark- her5 (Fig. 4I and J). By 24 hpf, pax2.1 (Fig. 4C and D), en2 ers in Df w5 embryos at multiple stages of embryogenesis to (Fig. 4G and H), and her5 (Fig. 4K and L) expression is not determine whether changes could be observed at the MHB detectable in the ventral portion of the MHB (arrowheads), (Fig. 4). In all cases, no clear differences could be seen in the junction of the tegmentum and rhombomere 1, but spots embryos younger than 8 somites (genotypes confirmed by of expression are still visible in the dorsal half of the MHB

Fig. 5. Functional redundancy of wnt1 and wnt10b. (A) Confocal image of GFP expression from PAC 176DG16. Lateral view of head of 24-hpf embryo to illustrate strong expression in the tectum (arrow). Expressing cells are also found in the MHB and hindbrain (arrowheads). (BÐI) Probe used is indicated in upper right of each panel; genotype is in lower right. (BÐD) Rescue of pax2.1 expression in Df w5 embryos by PAC injection. (B) Wild type embryo. (C) Df w5 homozygote to show loss of pax2.1 in ventral MHB (arrow). (D) Df w5 homozygote with restored ventral pax2.1 expression (arrow) due to injection of PAC176⌬G16. (EÐI) Phenotypes induced by wnt1 and wnt10b morpholinos. (E) Control uninjected embryo. pax2.1 expression observed in embryos injected with either wnt1-MO or wnt10b-MO is indistinguishable from this example. Note that MHB pax2.1 expression appears normal (arrow). (F) Embryo coinjected with both wnt1-MO and wnt10b-MO. Note loss of ventral pax2.1 expression as seen in Df w5 homozygotes (compare with embryo in C). (G) en2 expression in control uninjected 24-hpf embryo. Embryos injected with wnt10b-MO are indistinguishable from this example. (H) Weak reduction in en2 expression seen in ϳ30% of embryos injected with wnt1-MO. Note reduction localized to the ventral MHB (arrow). (I) Stronger reduction in en2 expression upon wnt1-MO or wnt1 ϩ wnt10b- MO injection. Compare this phenotype with the Df w5 embryo in Fig. 4H. 180 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187

Table 2 Assays of wnt1 and wnt10b morpholino injections

Morpholino injected Assay n Normal, % Weak phenotype, % Df phenocopy, % wnt1-MO pax2.1 84 100 0 0 en2 84 65 29 5 wnt10b-MO pax2.1 54 100 0 0 en2 Ͼ50 100 0 0 wnt1 ϩ wnt10b-MO pax2.1 35 69 0 31 en2 48 42 42 16

between the cerebellum and the tectum. This phenotype is somitogenesis stages (Krauss et al., 1995). At 24 hpf, no specific to the MHB, since the other pax2.1 expression differences in either wnt8b or wnt3a expression can be seen domains are unaffected (Fig. 4C and D). Thus, the expres- between wild type and Df w5 embryos (Fig. 4U and V; Ͼ50 sion of pax2.1, en2, and her5 is initiated properly but is not embryos examined for each marker). Interestingly, both of maintained in portions of their expression domains begin- these wnt genes are expressed in only the dorsal portion of ning from the 8-somite stage. These phenotypes are 100% the MHB at 24 hpf. Therefore, the ventral MHB domain penetrant, but do display some variations in expressivity. represents a unique expression domain for wnt1 and Thus, while all mutant embryos display reductions in wnt10b. We can conclude from these expression studies that pax2.1, en2, and her5 expression, the degree of reduction loci within Df w5 are required in the ventral MHB for the can vary between embryos from little detectable expression maintenance of pax2.1, en2, and her5 expression but not for to slightly more staining than shown in Fig. 4 (n Ͼ 100). fgf8, en3, wnt8b, or wnt3a. Because wnt1 and wnt10b are In contrast to the previous three examples, the mainte- both deleted by Df w5, neither wnt1 nor wnt10b is required nance of fgf8, en3, wnt8b, and wnt3a is not strictly depen- for the maintenance of fgf8, en3, wnt8b, or wnt3a expres- dent on loci within Df w5 (Fig. 4MÐV). For example, when sion. assayed for the expression of fgf8, Df w5 embryos are frequently indistinguishable from wild type embryos at the The Df w5 pax2.1 expression phenotype can be rescued by 12-somite stage (Fig. 4M and N), 18-somite stage (not genomic DNA containing either the wnt1 or wnt10b loci shown), and 24 hpf (Fig. 4O and P), although some mutant embryos do have reduced ventral MHB staining (Fig. 4P, In order to determine whether the changes in gene ex- inset; genotypes confirmed by PCR). Thus, all embryos that pression at the MHB in Df w5 embryos can be attributed to have reduced levels of fgf8 expression prove to be Df w5 the loss of either wnt1 or wnt10b or both, we attempted homozygotes, but not all Df w5 homozygotes have reduced rescue experiments by injecting genomic DNA clones en- fgf8 expression. In one cross, 25% of mutant embryos had compassing these loci into embryos from Df w5/ϩ inter- reduced fgf8 expression, while 75% of mutant embryos crosses and subsequently assaying for pax2.1 expression. appeared wild type (total embryos ϭ 68, no. mutants ϭ 16). Because the Wnt pathway is involved in many early steps of In another test, 49% of mutant embryos appeared wild type, embryogenesis, overexpression of Wnt pathway compo- 43% had strongly reduced fgf8 expression, and 8% had nents affects development at many time points. By injecting slightly reduced fgf8 expression (total embryos ϭ 163; no. large insert genomic clones, each gene’s endogenous en- mutants ϭ 39). This result indicates that fgf8 expression is hancer elements most likely are intact and will result in the not dependent on wnt1 or wnt10b. In another example, en3 expression of the genes on each PAC according to their wild expression in Df w5 embryos is indistinguishable from wild type patterns (Jessen et al., 1999; Long et al., 1997; Yan et type at the 14-somite stage (Fig. 4Q and R), but at 24 hpf, al., 1998). We focused on two large-insert clones from a slight reductions in the anteriorÐposterior extent of expres- zebrafish PAC library (Amemiya and Zon, 1999) that we sion can be observed (Fig. 4S and T). No loss of en3 identified by PCR using wnt1 and wnt10b primers (Table 1). expression specifically in the ventral portion of the MHB is PAC 66D18 spans wnt1 and extends to the 5Ј portion of the observed. wnt10b locus but terminates between exons 3 and 4 (break- Because of the mild phenotype of Df w5 embryos in point of the PAC determined by sequencing; see Fig. 1). If comparison to the mouse wnt1 knockout, we reasoned that the wnt10b gene can be transcribed from this clone, it will other Wnts may still be present in the MHB in Df w5 mu- result in a Wnt10b peptide that lacks 164 amino acids from tants. wnt8b is expressed in the prospective MHB beginning the carboxy terminus, would be 90 amino acids shorter than at ϳ75% epiboly (prior to the onset of wnt1 and wnt10b the comparable truncation point in dominant-negative expression; unpublished observations) and subsequently is Xwnt8, and will be nonfunctional (Hoppler et al., 1996). expressed in other brain territories (Kelly et al., 1995). Thus, PAC 66D18 contains an intact wnt1 gene but not a wnt3a is expressed at the MHB, dorsal midline of the functional wnt10b. PAC 176A22 spans both loci com- midbrain, and at the midbrainÐforebrain junction from mid- pletely. The third PAC we utilized for rescue experiments A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 181

Fig. 6. ace and noi are dominant enhancers of Df w5. (A) Schematic diagram of the cross performed to generate embryos homozygous for Df w5 and heterozygous for ace or noi. (BÐF) Oblique views of heads of live 27-hpf embryos. Genotypes are indicated in the lower right corners. (B) In wild type or Df w5 homozygotes, the fold is visible and creates a clear separation between the posterior edge of the tectum (arrowhead) and the cerebellum (small arrow). The midline of the fold (large arrow) is visible as an indentation in the lateral hindbrain wall. Asterisks in all panels: dorsal midline of the midbrain. (C) ace mutants fail to form a MHB fold and do not form a cerebellum. Rather, the posterior edge of the optic tectum is in a position corresponding to the position of the cerebellum in wild types (arrowhead). (D) noi mutants also fail to form the MHB fold and cerebellum, but also have defects in the formation of the midbrain. Thus, the tectum appears much smaller than in ace mutants and also has a turbid appearance (not visible). Df w5 mutants enhanced by ace (E) or noi (F) alleles appear similar in that the separation between the tectum and cerebellum is less distinct due to the absence of a definitive MHB fold. Arrowheads indicate the posterior edge of the tectum, and small arrows indicate the position of the cerebellum. A very slight bump in the hindbrain wall is visible (large arrow). (GÐJ) Loss of MHB gene expression in Df w5 embryos enhanced by ace (G, I) or noi (H, J). Probe used is indicated in the upper right corner. Note the absence of expression of each gene in the MHB region (arrows) in comparison to Df w5 homozygotes (Fig. 4). we generated through chi-site stimulated homologous re- (PAC 176⌬G16) which has exons one and two from the combination (Dabert and Smith, 1997; Jessen et al., 1998) wnt1 locus replaced with GFP coding sequence in frame within PAC 176A22. We generated a recombined PAC with the Wnt1 initiation codon (see Materials and methods; 182 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 data not shown). Thus, this PAC should function as a 5E). The loss of pax2.1 expression was specific to the MHB, reporter for wnt1 expression domains (assuming that no as other pax2.1 expression domains, for example, in the otic critical regulatory regions are contained within the first two vesicle or optic stalk, were unaltered. Thus, interference exons or the first intron), and will express functional with the activity of both wnt1 and wnt10b is required to Wnt10b but not Wnt1. Injection of PAC 176⌬G16 into wild affect the expression of pax2.1 at the MHB. type embryos results in detectable GFP expression at 24 hpf The expression of en2, on the other hand, appears to be in the optic tectum and hindbrain (Fig. 5A). wnt1 transcripts at least partly sensitive to the reduction of wnt1 (Table 2). are not detected in the optic tectum, so it is not clear whether At 24 hpf, wnt10b-MO-injected embryos all show normal the GFP expressed there is a result of perdurance of earlier en2 expression (n Ͼ 50). In contrast, 29% of wnt1-MO- expression or reflects the loss of negative regulatory ele- injected embryos had staining that was reduced slightly in ments. Nonetheless, Because GFP is expressed from this the ventral MHB region, corresponding to the region that PAC, it must be under the regulatory control of at least some has much higher levels of wnt1 expression (Table 2; Fig. wnt1 control elements. Whereas Df w5 embryos never dis- 5H; “weak” reduction). Five percent of wnt1-MO embryos play ventral MHB pax2.1 expression at 24 hpf (Fig. 5C, had only a spot of en2 expression remaining at the dorsal arrow; compare with wild type in Fig. 5B), injection of any MHB (Fig. 5I; “strong” reduction). In contrast, when both of the three PACs into Df w5 embryos is able to restore morpholinos are injected, the percentage of embryos that ventral pax2.1 expression (Table 1 and Fig. 5D, arrow; show a strong reduction in en2 expression increases (Table compare with Fig. 5C), which is never seen in uninjected 2). Thus, knockdown of wnt1 can lead to a reduction in the Df w5 embryos (n Ͼ 50). Thus, an element(s) common to all expression of en2 in the ventral MHB, but this effect is three PACs is sufficient to rescue pax2.1 expression in Df w5 enhanced by the knockdown of both wnt1 and wnt10b. embryos. If the MHB phenotype in Df w5 embryos is due to From the morpholino injection results, we conclude that the loss of the wnt1 and wnt10b loci, our rescue result raises the function of both Wnt1 and Wnt10b must be reduced the possibility that both wnt1 and wnt10b have the capacity simultaneously to completely recapitulate the Df w5 pheno- to restore normal MHB gene expression in Df w5 embryos. type. In conjunction with the results from our Df w5 rescue experiments, we further conclude that the MHB phenotype Morpholino knockdown of both Wnt1 and Wnt10b is in Df w5 embryos is a result of the loss of Wnt1 and Wnt10b required to recapitulate the Df w5 MHB phenotype activity, and that Wnt1 and Wnt10b are functionally redundant with respect to pax2.1 expression but not completely Because our previous rescue experiments do not exclude with respect to en2 expression. the possibility of loci other than wnt1 and wnt10b within the interval defined by PACs 66D18 and 176A22 being respon- noi and ace are dominant enhancers sible for the MHB phenotype of Df w5 embryos, we per- of the Df w5 phenotype formed antisense oligo knockdown experiments with morpholinos (Summerton, 1999) directed against wnt1, wnt10b, Df w5 mutants have a relatively mild MHB phenotype, or both (Fig. 5EÐI). We injected approximately 1Ð5ngof yet have altered levels of expression of pax2.1 and, variably, each morpholino per embryo at the one- to four-cell stage fgf8, genes both known to be required for the proper for- either alone or together, then assayed injected embryos for mation of the MHB (Brand et al., 1996; Reifers et al., 1998). the expression of pax2.1 (Fig. 5E and F) and en2 (Fig. 5GÐI) Because of this fact, we questioned whether Df w5 embryos since these genes exhibited the most significant changes in may be sensitive to reductions in either Pax2.1 or Fgf8 Df w5 embryos. The wnt1 morpholino (wnt1-MO) induced a levels. To test this hypothesis, we performed crosses to mild head necrosis that likely represents a nonspecific ef- generate embryos homozygous for Df w5 and heterozygous fect, since we do not observe a comparable phenotype in for either ace or noi null alleles (Fig. 6A). Thus, this ma- Df w5 embryos (not shown); therefore, in experiments with nipulation should result in groups of Df w5 embryos, half of the wnt1-MO, we selected for injected embryos that either which have their dosage of pax2.1 and fgf8 reduced by half had no head necrosis or were only mildly affected. The (Fig. 6A). background necrosis phenotype precluded examining the At 27 hpf, Df w5 homozygotes are discernible only by morphology of wnt1-MO injected embryos, but those in- visualization of the scattered cell death in the tectum and of jected with the wnt10b morpholino (wnt10b-MO) all ap- the slightly increased distance between the medial edges of peared morphologically normal at 24 hpf (not shown; n Ͼ the MHB fold (see Fig. 3). However, in crosses where half 100). of Df w5 homozygotes also inherit one mutant allele of either At 24 hpf, none of the embryos injected with either noi or ace, a new phenotypic class emerges that is charac- individual morpholino showed any change in pax2.1 ex- terized by the absence of a defined MHB fold (Fig. 6BÐF). pression (Table 2; Fig. 5E). However, when injected with This phenotypic class is never observed in Df w5 homozy- both morpholinos, a reduction in pax2.1 expression that gotes (n Ͼ 1000), indicating that it arises because of the loss appeared identical to that seen in Df w5 embryos was ob- of a functional copy of noi or ace. As indicated in Table 3, served (Table 2 and Fig. 5F; compare with control in Fig. the proportion of embryos displaying the new phenotype A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 183

Table 3 servable at 24 hpf in the ventral portion of the MHB and the w5 Dominant enhancement of Df by ace and noi cerebellum. This finding raises the question: does the coex- Cross No. No. lacking %of P value pression of wnt1 and wnt10b have functional significance? embryos constrictiona total In both the pufferfish and human genomes, wnt1 and Dfw5/ϩϫnoi/ϩ Ͼ100 0 0 wnt10b are arranged in a tail-to-tail orientation (Gellner and Dfw5/ϩϫace/ϩ Ͼ100 0 0 Brenner, 1999; Nusse, 2001), supporting the idea that this Dfw5/ϩϫDfw5/ϩ Ͼ100 0 0 genomic organization reflects an arrangement created upon Dfw5/ϩ; ace/ϩϫDfw5/ϩ 266 30 11.3 Ͻ0.0005 the duplication event in the vertebrate lineage that created w5 ϩ ϩϫ w5 ϩ Ͻ Df / ; noi/ Df / 168 24 14.3 0.0005 the wnt1Ðwnt10b paralog cluster (Nusse, 2001). Thus, the a Embryos were scored at 27 hpf for the presence of a MHB constriction. fact that zebrafish wnt1 and wnt10b share expression patterns could reflect either transcriptional control by a common regulatory element or control by duplicated elements. corresponds with that expected for this genetic class (P Ͻ Morpholino knockdown of wnt1 does not reduce expression 0.005). Wild type 27-hpf embryos have a clearly defined of wnt10b (data not shown), indicating that there is not a separation between the posterior edge of the optic tectum linear relationship between the two in zebrafish. Little data (arrowhead in Fig. 6B) and the cerebellum (small arrow in have been reported regarding the embryonic expression of Fig. 6B). The clear separation of these tissues is due to the wnt10b in other systems, with the exception that Chris- optical properties of the midline of the MHB fold. In en- tiansen et al (1995) showed that murine wnt10b (previously w5 hanced Df embryos (for either noi or ace enhancement), referred to as wnt12; Christiansen et al., 1995) is not ex- the loss of the midline of the fold results in an unclear pressed in the same pattern as murine wnt1. Thus, based on separation between the tectum and the cerebellum (Fig. 6E a comparison of their expression patterns, the zebrafish and F). ortholog of wnt10b appears to have maintained a greater We examined the expression of several MHB markers in function in early neural development than its murine coun- w5 enhanced Df embryos at 24 hpf to determine whether terpart. w5 molecular differences were observable. Strikingly, in Df / In the mouse, the expression of wnt1 has been found to w5 Df ; ace/ϩ embryos, the expression of pax2.1 at 24 hpf is require a 5.5-kb enhancer element located 3Ј of the wnt1 w5 absent at the MHB (Fig. 6G; compare with Df homozy- coding region (Danielian et al., 1997; Echelard et al., 1994). w5 w5 gote in Fig. 4D), and Df /Df , noi/ϩ embryos show no The 5.5-kb contains a 110-bp regulatory sequence that can expression of fgf8 at the MHB (Fig. 6H; compare with Fig. drive the expression of a lacZ reporter gene in an approxi- 4P). Furthermore, in both enhanced genetic combinations, mately wild type wnt1 pattern (Rowitch et al., 1998), and a the expression of en3 is also completely absent (Fig. 6I and comparison of genomic sequences flanking the mouse and w5 J; compare with Fig. 4T). Thus, in Df homozygotes, the pufferfish wnt1 coding regions found that the 110-bp regu- cross-regulatory network involving pax2.1, fgf8, and their latory element has been conserved, although the elements downstream targets (i.e., en3) can sufficiently maintain it- are found in opposite orientations in the pufferfish and self. However, when levels of pax2.1 or fgf8 are reduced, mouse genomes (Gellner and Brenner, 1999; Rowitch et al., the network is not adequate to maintain itself. We conclude 1998). Interestingly, our modified PAC (176⌬G16), which w5 that Df embryos are able to form their MHB, but are has a portion of the wnt1 coding region replaced with GFP, sensitized to variations in the levels of critical MHB genes. produces GFP expression in the tectum and hindbrain. The regions where we see GFP expression from the modified PAC do not appear to be regions in which wnt1 transcripts Discussion can be detected at the same time point, but they do match regions that display expression of a Lef/␤-catenin-depen- Zebrafish wnt1 and wnt10b: coordinated expression dent reporter gene (Dorsky et al., 2002). This result could and redundant function reflect either that the modification removed critical negative regulatory elements located within the first two exons and Wnt signaling is involved in many aspects of vertebrate first intron, or that early wnt1 expression is in a broader neural development, including patterning and proliferation domain than later wnt1 expression, and the GFP seen in the (Megason and McMahon, 2002), yet how individual mem- tectum represents perdurance. Clearly, our information re- bers of the Wnt family contribute to these processes is often garding the regulatory control of wnt1 and wnt10b expres- unclear. In the zebrafish, at least five members of the Wnt sion is incomplete, and the completion of the mouse and family (wnt1, wnt10b, wnt8b, wnt3a, and wnt8-ORF2; Lek- zebrafish genome sequences will be useful for identifying ven et al., 2001) are expressed in the neural tube from late important regulatory elements. gastrulation stages, but with the exception of wnt8b, the Do the comparative expression patterns of zebrafish wnt1 functions of these have not been explored. In this study, we and wnt10b to their mouse orthologs reveal their probable have shown that zebrafish wnt1 and wnt10b expression function? Our results show that, in zebrafish, wnt1 and patterns significantly overlap, with differences being ob- wnt10b are required only for the maintenance of gene ex- 184 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 pression in a portion of the MHB, and these two genes are 2001; Simeone, 2000). These three genes form a cross- at least partially functionally redundant in this process. regulatory network to maintain each other’s expression (Liu Thus, a deficiency that removes both loci results in a mor- et al., 1997; Lun and Brand, 1998; Ye et al., 2001). The phologically mild phenotype visible only after 24 hpf and importance of Fgf8 for the organizing function of the MHB characterized by cell death in the tectum and a slightly is illustrated by the fact that beads soaked in Fgf8 protein increased distance between the medial edges of the MHB can mimic the polarizing properties of the MHB when constriction. The expression of several MHB markers in implanted into avian embryonic midbrain tissue (Crossley et homozygous deficient embryos is initiated properly but not al., 1996), and expression of fgf8 isoforms under the control maintained in the ventral MHB domain beginning from the of a wnt1 enhancer results in patterning defects of the 8-somite stage. However, not all MHB markers are affected midbrain that can result in the induction of wnt1, en, and in homozygous deficient embryos as wnt3a and wnt8b do other MHB genes (Lee et al., 1997; Liu et al., 1999). not have significantly altered expression patterns and fgf8 In the zebrafish, a suite of genes have been identified that expression is only variably affected. The fact that wnt1 and are required not for the positioning or initiation of gene wnt10b function redundantly is supported by both our res- expression at the prospective MHB, but rather for the main- cue and morpholino antisense oligo knockdown experi- tenance of their expression. In noi mutants, expression of ments, since rescue of deficiency mutants can be achieved wnt1, her5, and en2 is initiated but very quickly disappears, with genomic DNA containing either wnt1 or wnt10b, and and en3 expression is not properly initiated, possibly indi- morpholino knockdown of both is required to phenocopy cating a direct role for pax2.1 in its regulation (Lun and the deficiency. This relationship in zebrafish is in contrast to Brand, 1998). Pou2, the product of the spg locus, also the mouse, where knockout of wnt1 results in the loss of appears to be required for early steps in maintenance of the midbrain and cerebellum (McMahon and Bradley, 1990; MBH regional identity. Similar to noi mutants, the expres- Thomas and Capecchi, 1990; Thomas et al., 1991), a much sion of MHB markers, like wnt1, pax2.1, and en2, in spg more severe phenotype in comparison to the zebrafish. It mutants is initiated but very quickly disappears (Belting et appears from this result that, in the mouse, not only do wnt1 al., 2001; Burgess et al., 2002; Reim and Brand, 2002). and wnt10b not overlap in expression or function, but wnt1 Pou2 appears to play a permissive role in patterning the appears to play a much more predominant role in brain MHB; however, since overexpression does not lead to ec- formation than seen in zebrafish. Thus, one could not make topic expression of either wnt1 or pax2.1 (Belting et al., conclusions regarding the function of zebrafish wnt1 or 2001; Burgess et al., 2002; Reim and Brand, 2002), leading wnt10b based on a comparison to their murine counterparts, to the conclusion that the role of Pou2 is perhaps to provide and our studies demonstrate that there are significant differ- regional competence in the neurectoderm for reception of ences between the relative functions of zebrafish and mouse Fgf signals, in this case Fgf8 (Belting et al., 2001; Reim and wnt1 and wnt10b in early nervous system development. Brand, 2002). These mutants also point to the establishment What is the relationship between the multiple Wnts that of an Fgf8 gradient as being the ultimate output of the are expressed in the developing neural tube? Genetic studies MHB. ace mutants lose gene expression later than noi of mouse wnt1 and wnt3a mutants have shown that these mutants, and they fail to form a cerebellum but do form the genes function redundantly in the development of the dorsal optic tectum (Reifers et al., 1998). Thus, a complex genetic neural tube (Ikeya et al., 1997). Also in the mouse, analysis network must exist which involves cross- regulatory inter- of the control of emx2 expression suggests that either wnt3a actions, but few studies have been reported to date regarding or wnt8b could be involved in establishing telencephalic cis-regulatory sequences for any of these genes. One excep- identity (Theil et al., 2002). Zebrafish wnt3a and wnt8b are tion to this is the finding that several pax2/5/8 binding sites also expressed during early neural patterning, and studies of are present in the mouse en2 promoter, perhaps indicating wnt8b have shown a requirement for its inhibition in the that pax2.1 functions through regulation of en2 (Song et al., induction of telencephalic fates (Houart et al., 2002; Kim et 1996). Consistent with this, simultaneous morpholino al., 2002). Our results indicate that these wnts do not regu- knockdown of zebrafish en2 and en3 results in a reduction late each other’s expression, and they do have independent of the optic tectum and the lack of MHB formation, a functions. Clearly, understanding how these loci function phenotype very similar to that seen in noi mutants (Scholpp independently and in concert is necessary for understanding and Brand, 2001). vertebrate neural patterning and morphogenesis. Where do wnt1 and wnt10b fit into the MHB genetic hierarchy? In the mouse, the primary function of wnt1 may Genetic interactions at the zebrafish MHB be to maintain the expression of engrailed homologs, since mutant embryos can be rescued by the expression of en1 The MHB arises at the point in the neural tube where under the control of the wnt1 enhancer (Danielian and cells expressing otx2 form an interface with cells expressing McMahon, 1996). However, the expression of fgf8 is also gbx2 (or gbx1 in zebrafish; Rhinn and Brand, 2001), and this lost in mouse wnt1 mutants, and fgf8 expression can also interface is where pax2, fgf8, and wnt1 expression arises induce en2 (Lee et al., 1997; Liu et al., 1999). In the case of (for reviews, see Joyner et al., 2000; Rhinn and Brand, the zebrafish, our results indicate that a portion, but not A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 185 all, of engrailed and pax2.1 expression is dependent on References the activity of wnt1 and wnt10b. This does not rule out the possibility of engrailed and pax2.1 expression in all of Amemiya, C.T., Zon, L.I., 1999. Generation of a zebrafish P1 artificial the MHB being dependent on Wnt activity, as at least two chromosome library. Genomics 58, 211Ð213. w5 Aoki, T.O., Mathieu, J., Saint-Etienne, L., Rebagliati, M.R., Peyrieras, N., other Wnts, wnt8b and wnt3a, are still expressed in Df Rosa, F.M., 2002. Regulation of nodal signalling and mesendoderm w5 embryos. Further, our result that Df embryos are sensi- formation by TARAM-A, a TGFbeta-related type I receptor. Dev. Biol. tized to alterations in the levels of pax2.1 and fgf8 suggests 241, 273Ð288. that at least part of the function of wnt1 and wnt10b is to Bang, A.G., Papalopulu, N., Goulding, M.D., Kintner, C., 1999. Expres- participate in a gene regulatory network that maintains sion of Pax-3 in the lateral neural plate is dependent on a Wnt-mediated signal from posterior nonaxial mesoderm. Dev. Biol. 212, 366Ð380. threshold levels of these products at the MHB. Thus, in the Belting, H.-G., Hauptmann, G., Meyer, D., Abdelilah-Seyfried, S., Chitnis, absence of wnt1 and wnt10b, threshold levels of en2, pax2.1, A., Eschbach, C., Soll, I., Thisse, C., Thisse, B., Artinger, K.B., Lunde, fgf8, and other gene products are expressed in at least a K., Driever, W., 2001. Spiel ohne grenzen/pou2 is required during portion of the MHB, and the presence of an adequate reg- establishment of the zebrafish midbrainÐhindbrain boundary organizer. Development 128, 4165Ð4176. ulatory network is sufficient for the morphogenetic changes Brand, M., Heisenberg, C.P., Jiang, Y.J., Beuchle, D., Lun, K., Furutani- involved in the formation of the MHB. In contrast, when the Seiki, M., Granato, M., Haffter, P., Hammerschmidt, M., Kane, D.A., levels of Pax2.1 or Fgf8 are reduced in the absence of wnt1 Kelsh, R.N., Mullins, M.C., Odenthal, J., van Eeden, F.J., Nusslein- and wnt10b, the network is not maintained sufficiently to Volhard, C., 1996. Mutations in zebrafish genes affecting the formation support the morphogenesis of the MHB. We conclude that of the boundary between midbrain and hindbrain. Development 123, 179Ð190. zebrafish wnt1 and wnt10b may be components of a regu- Burgess, S., Reim, G., Chen, W., Hopkins, N., Brand, M., 2002. The lative mechanism that protects the embryo against abnormal zebrafish spiel-ohne-grenzen (spg) gene encodes the POU domain phenotypes that would result from variability in gene activ- protein Pou2 related to mammalian Oct4 and is essential for formation ity. of the midbrain and hindbrain, and for pre-gastrula morphogenesis. Development 129, 905Ð916. Finally, genetic redundancy is a common theme in the Christiansen, J.H., Dennis, C.L., Wicking, C.A., Monkley, S.J., Wilkinson, zebrafish genome, and many examples are now known of D.G., Wainwright, B.J., 1995. Murine Wnt-11 and Wnt-12 have tem- duplicate genes possessing functional overlap (for example, porally and spatially restricted expression patterns during embryonic Imai et al., 2001; Itoh et al., 2002; Lekven et al., 2001). development. Mech. Dev. 51, 341Ð350. Because of functional redundancy, identifying genes in- Crossley, P.H., Martinez, S., Martin, G.R., 1996. Midbrain development induced by FGF8 in the chick embryo. Nature 380, 66Ð68. volved in an embryological process through mutagenesis Dabert, P., Smith, G.R., 1997. Gene replacement with linear DNA frag- screens can be more difficult. On the other hand, performing ments in wild-type Escherichia coli: enhancement by Chi sites. Genet- a genetic screen in the context of a sensitized system can be ics 145, 877Ð889. an extremely effective screening method (for example, Danielian, P.S., Echelard, Y., Vassileva, G., McMahon, A.P., 1997. A w5 5.5-kb enhancer is both necessary and sufficient for regulation of wnt-1 Rogge et al., 1991). The fact that Df embryos show few transcription in vivo. Dev. Biol. 192, 300Ð309. morphological abnormalities during the first 48 h of devel- Danielian, P.S., McMahon, A.P., 1996. engrailed-1 as a target of the Wnt-1 opment, yet are sensitive to slight alterations in the levels of signalling pathway in vertebrate midbrain development. Nature 383, other gene products, suggests that this would be an effective 332Ð334. sensitized background in which to search for novel muta- Dorsky, R.I., Sheldahl, L.C., Moon, R.T., 2002. A transgenic Lef1/beta- catenin-dependent reporter is expressed in spatially restricted domains tions affecting the formation of the midbrain and MHB. throughout zebrafish development. Dev. Biol. 241, 229Ð237. Driever, W., Solnica-Krezel, L., Schier, A.F., Neuhauss, S.C., Malicki, J., Stemple, D.L., Stainier, D.Y., Zwartkruis, F., Abdelilah, S., Rangini, Z., Belak, J., Boggs, C., 1996. A genetic screen for mutations affecting Acknowledgments embryogenesis in zebrafish. Development 123, 37Ð46. Echelard, Y., Vassileva, G., McMahon, A.P., 1994. Cis-acting regulatory We thank the reviewers for insightful comments, Dr. sequences governing wnt-1 expression in the developing mouse CNS. Development 120, 2213Ð2224. Richard Dorsky for discussions and a critical reading of the Ekker, M., Wegner, J., Akimenko, M.A., Westerfield, M., 1992. Coordi- manuscript, Dr. Chris Thorpe and the rest of the Moon lab nate embryonic expression of three zebrafish engrailed genes. Devel- for invaluable input, Dr. Bruce Riley for many stimulating opment 116, 1001Ð1010. discussions about this project, and Min-Yung Chiang for Erter, C.E., Wilm, T.P., Basler, N., Wright, C.V., Solnica-Krezel, L., 2001. generating the Df w5/ϩ; noi/ϩ and Df w5/ϩ; ace/ϩ fish used Wnt8 is required in lateral mesendodermal precursors for neural pos- teriorization in vivo. Development 128, 3571Ð3583. here. We also thank Dr. David Raible, Dr. Cecilia Moens, Fjose, A., Njolstad, P.R., Nornes, S., Molven, A., Krauss, S., 1992. Struc- Dr. David Kimelman, and Dr. Kevin Griffin for in situ ture and early embryonic expression of the zebrafish engrailed-2 gene. probes and stimulating discussions, and Dr. Jose Campos- Mech. Dev. 39, 51Ð62. Ortega for the her5 cDNA. A.C.L. was supported in part by Gellner, K., Brenner, S., 1999. Analysis of 148 kb of genomic DNA around the wnt1 locus of Fugu rubripes. Genome Res. 9, 251Ð258. an NIH NRSA award (F32 GM 19256). N.K. is an under- Haffter, P., Granato, M., Brand, M., Mullins, M.C., Hammerschmidt, M., graduate, and R.T.M. is an Investigator of the Howard Kane, D.A., Odenthal, J., van Eeden, F.J., Jiang, Y.J., Heisenberg, C.P., Hughes Medical Institute. Kelsh, R.N., Furutani-Seiki, M., Vogelsang, E., Beuchle, D., Schach, 186 A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187

U., Fabian, C., Nusslein-Volhard, C., 1996. The identification of genes Lun, K., Brand, M., 1998. A series of no isthmus (noi) alleles of the with unique and essential functions in the development of the zebrafish, zebrafish pax2.1 gene reveals multiple signaling events in development Danio rerio. Development 123, 1Ð36. of the midbrain- hindbrain boundary. Development 125, 3049Ð3062. Hoppler, S., Brown, J.D., Moon, R.T., 1996. Expression of a dominant- McGrew, L.L., Lai, C.J., Moon, R.T., 1995. Specification of the antero- negative Wnt blocks induction of myoD in Xenopus embryos. Genes posterior neural axis through synergistic interaction of the Wnt signal- Dev. 10, 2805Ð2817. ing cascade with noggin and follistatin. Dev. Biol. 172, 337Ð342. Houart, C., Caneparo, L., Heisenberg, C.-P., Barth, K., Take-Uchi, M., McMahon, A.P., Bradley, A., 1990. The wnt-1 (int-1) proto-oncogene is Wilson, S.W., 2002. Establishment of the telencephalon during gastru- required for development of a large region of the mouse brain. Cell 62, lation by local antagonism of Wnt signaling. Neuron 35, 255Ð265. 1073Ð1085. Hukriede, N.A., Joly, L., Tsang, M., Miles, J., Tellis, P., Epstein, J.A., Megason, S.G., McMahon, A.P., 2002. A mitogen gradient of dorsal Barbazuk, W.B., Li, F.N., Paw, B., Postlethwait, J.H., Hudson, T.J., midline Wnts organizes growth in the CNS. Development 129, 2087Ð Zon, L.I., McPherson, J.D., Chevrette, M., Dawid, I.B., Johnson, S.L., 2098. Ekker, M., 1999. Radiation hybrid mapping of the zebrafish genome. Melby, A.E., Beach, C., Mullins, M., Kimelman, D., 2000. Patterning the Proc. Natl. Acad. Sci. USA 96, 9745Ð9750. early zebrafish by the opposing actions of bozozok and vox/vent. Dev. Ikeya, M., Lee, S.M., Johnson, J.E., McMahon, A.P., Takada, S., 1997. Biol. 24, 275Ð285. Wnt signalling required for expansion of neural crest and CNS pro- Molven, A., Njolstad, P.R., Fjose, A., 1991. Genomic structure and re- genitors. Nature 389, 966Ð970. stricted neural expression of the zebrafish wnt-1 (int-1) gene. EMBO J. Imai, Y., Gates, M.A., Melby, A.E., Kimelman, D., Schier, A.F., Talbot, 10, 799Ð807. W.S., 2001. The homeobox genes vox and vent are redundant repres- Mu¬ller, M., v. Weizsa¬cker, E., Campos-Ortega, J.A., 1996. Transcription sors of dorsal fates in zebrafish. Development 128, 2407Ð2420. of a zebrafish gene of the hairy-Enhancer of split family delineates the Itoh, M., Kudoh, T., Dedekian, M., Kim, C.H., Chitnis, A.B., 2002. A role midbrain anlage in the neural plate. Dev. Genes Evol. 206, 153Ð160. for iro1 and iro7 in the establishment of an anteroposterior compart- Nusse, R., 2001. An ancient cluster of Wnt paralogues. Trends Genet 17, ment of the ectoderm adjacent to the midbrainÐhindbrain boundary. 443. Development 129, 2317Ð2327. Oxtoby, E., Jowett, T., 1993. Cloning of the zebrafish krox-20 gene (krx- Jessen, J.R., Meng, A., McFarlane, R.J., Paw, B.H., Zon, L.I., Smith, G.R., 20) and its expression during hindbrain development. Nucleic Acids Lin, S., 1998. Modification of bacterial artificial chromosomes through Res. 21, 1087Ð1095. chi-stimulated homologous recombination and its application in ze- Postlethwait, J.H., Yan, Y.L., Gates, M.A., Horne, S., Amores, A., Brown- brafish transgenesis. Proc. Natl. Acad. Sci. USA 95, 5121Ð5126. lie, A., Donovan, A., Egan, E.S., Force, A., Gong, Z., Goutel, C., Fritz, Jessen, J.R., Willett, C.E., Lin, S., 1999. Artificial chromosome transgen- A., Kelsh, R., Knapik, E., Liao, E., Paw, B., Ransom, D., Singer, A., esis reveals long- distance negative regulation of rag1 in zebrafish. Nat. Thomson, M., Abduljabbar, T.S., Yelick, P., Beier, D., Joly, J.S., Genet. 23, 15Ð16. Larhammar, D., Rosa, F., et al., 1998. Vertebrate genome evolution and Joyner, A.L., Liu, A., Millet, S., 2000. Otx2, Gbx2 and Fgf8 interact to the zebrafish gene map. Nat. Genet. 18, 345Ð349. position and maintain a midÐhindbrain organizer. Curr. Opin. Cell Rastegar, S., Friedle, H., Frommer, G., Knochel, W., 1999. Transcriptional Biol. 12, 736Ð741. regulation of Xvent homeobox genes. Mech. Dev. 1, 139Ð149. Kelly, G.M., Greenstein, P., Erezyilmaz, D.F., Moon, R.T., 1995. Zebrafish Reifers, F., Bohli, H., Walsh, E.C., Crossley, P.H., Stainier, D.Y., Brand, wnt8 and wnt8b share a common activity but are involved in distinct M., 1998. Fgf8 is mutated in zebrafish acerebellar (ace) mutants and is developmental pathways. Development 121, 1787Ð1799. required for maintenance of midbrainÐhindbrain boundary develop- Kelly, G.M., Moon, R.T., 1995. Involvement of wnt1 and pax2 in the ment and somitogenesis. Development 125, 2381Ð2395. formation of the midbrain-hindbrain boundary in the zebrafish gastrula. Reim, G., Brand, M., 2002. Spiel-ohne-grenzen/pou2 mediates regional Dev. Genet. 17, 129Ð140. competence to respond to Fgf8 during zebrafish early neural develop- Kim, S.-H., Shin, J., Park, H.-C., Yeo, S.-Y., Hong, S.-K., Han, S., Rhee, ment. Development 129, 917Ð933. M., Kim, C.-H., Chitnis, A.B., Huh, T.-L., 2002. Specification of an Renucci, A., Lemarchandel, V., Rosa, F., 1996. An activated form of type anterior neuroectoderm patterning by Frizzled8a-mediated Wnt8b sig- I serine/threonine kinase receptor TARAM-A reveals a specific signal- nalling during late gastrulation in zebrafish. Development 129, 4443Ð ling pathway involved in fish head organiser formation. Development 4455. 122, 3735Ð3743. Krauss, S., Korzh, V., Fjose, A., Johansen, T., 1992a. Expression of four Rhinn, M., Brand, M., 2001. The midbrainÐhindbrain boundary organizer. zebrafish wnt-related genes during embryogenesis. Development 116, Curr. Opin. Neurobiol. 11, 34Ð42. 249Ð259. Richardson, M., Redmond, D., Watson, C.J., Mason, J.O., 1999. Mouse Krauss, S., Maden, M., Holder, N., Wilson, S.W., 1992b. Zebrafish pax[b] Wnt8B is expressed in the developing forebrain and maps to Chromo- is involved in the formation of the midbrainÐhindbrain boundary. some 19. Mamm. Genome 10, 923Ð925. Nature 360, 87Ð89. Rogge, R.D., Karlovich, C.A., Banerjee, U., 1991. Genetic dissection of a Lee, S.M., Danielian, P.S., Fritzsch, B., McMahon, A.P., 1997. Evidence neurodevelopmental pathway: son of sevenless functions downstream that FGF8 signalling from the midbrainÐhindbrain junction regulates of the sevenless and EGF receptor tyrosine kinases. Cell 64, 39Ð48. growth and polarity in the developing midbrain. Development 124, Roman, B.L., Pham, V.N., Lawson, N.D., Kulik, M., Childs, S., Lekven, 959Ð969. A.C., Garrity, D.M., Moon, R.T., Fishman, M.C., Lechleider, R.J., Lekven, A.C., Helde, K.A., Thorpe, C.J., Rooke, R., Moon, R.T., 2000. Weinstein, B.M., 2002. Disruption of acvrl1 increases endothelial cell Reverse genetics in zebrafish. Physiol. Genomics 2, 37Ð48. number in zebrafish cranial vessels. Development 129, 3009Ð3019. Lekven, A.C., Thorpe, C.J., Waxman, J.S., Moon, R.T., 2001. Zebrafish Rowitch, D.H., Echelard, Y., Danielian, P.S., Gellner, K., Brenner, S., wnt8 encodes two Wnt8 proteins on a bicistronic transcript and is McMahon, A.P., 1998. Identification of an evolutionarily conserved required for mesoderm and neurectoderm patterning. Dev. Cell 1, 110 base-pair cis-acting regulatory sequence that governs wnt-1 ex- 103Ð114. pression in the murine neural plate. Development 125, 2735Ð2746. Liu, A., Losos, K., Joyner, A.L., 1999. FGF8 can activate Gbx2 and Scholpp, S., Brand, M., 2001. Morpholino-induced knockdown of ze- transform regions of the rostral mouse brain into a hindbrain fate. brafish engrailed genes eng2 and eng3 reveals redundant and unique Development 126, 4827Ð4838. functions in midbrainÐhindbrain boundary development. Genesis 30, Long, Q., Meng, A., Wang, H., Jessen, J.R., Farrell, M.J., Lin, S., 1997. 129Ð133. GATA-1 expression pattern can be recapitulated in living transgenic Simeone, A., 2001. Positioning the isthmic organizer where Otx2 and zebrafish using GFP reporter gene. Development 124, 4105Ð4111. Gbx2meet. Trends Genet. 16, 237Ð240. A.C. Lekven et al. / Developmental Biology 254 (2003) 172Ð187 187

Song, D.L., Chalepakis, G., Gruss, P., Joyner, A.L., 1996. Two Pax- Swaying is a mutant allele of the proto-oncogene Wnt-1. Cell 67, binding sites are required for early embryonic brain expression of an 969Ð976. engrailed-2 transgene. Development 122, 627Ð635. Wassef, M., Joyner, A.L., 1997. Early mesencephalon/metencephalon pat- Summerton, J., 1999. Morpholino antisense oligomers: the case for an terning and development of the cerebellum. Perspect. Dev. Neurobiol. RNase H-independent structural type. Biochim. Biophys. Acta 1489, 5, 3Ð16. 141Ð158. Westerfield, M., 1995. The Zebrafish Book: A Guide for the Laboratory Use Theil, T., Aydin, S., Koch, S., Grotewold, L., Ruther, U., 2002. Wnt and of Zebrafish (Danio rerio). University of Oregon Press, Eugene, OR. Bmp signalling cooperatively regulate graded Emx2 expression in the Yan, Y.L., Talbot, W.S., Egan, E.S., Postlethwait, J.H., 1998. Mutant dorsal telencephalon. Development 129, 3045Ð3054. rescue by BAC clone injection in zebrafish. Genomics 50, 287Ð289. Thomas, K.R., Capecchi, M.R., 1990. Targeted disruption of the murine Ye, W., Bouchard, M., Stone, D., Liu, X., Vella, G., Lee, J., Nakamura, H., int-1 proto-oncogene resulting in severe abnormalities in midbrain and Ang, S.-L., Busslinger, M., Rosenthal, A., 2001. Distinct regulators cerebellar development. Nature 346, 847Ð850. control the expression of the mid-hindbrain organizer signal FGF8. Thomas, K.R., Musci, T.S., Neumann, P.E., Capecchi, M.R., 1991. Nat. Neurosci. 4, 1175Ð1181.