DEVELOPMENTAL DYNAMICS 236:1891–1904, 2007

RESEARCH ARTICLE

REREa/Atrophin-2 Interacts With Histone Deacetylase and Fgf8 Signaling to Regulate Multiple Processes of Zebrafish Development Nikki Plaster,1,2* Carmen Sonntag,1 Thomas F. Schilling,2 and Matthias Hammerschmidt1

The transcriptional regulator RERE/Atrophin-2 (RERE) is required for the normal patterning of the early vertebrate embryo, including the central nervous system, pharyngeal arches, and limbs. Consistent with a role as a transcriptional corepressor, RERE binds histone deacetylase 1 and 2 (HDAC1/2), and orphan nuclear receptors such as Tlx. Here, we identify the zebrafish babyface (bab) as a mutant in rerea and show that it interacts genetically with fibroblast growth factor 8 (fgf8). We suggest that this finding is largely due to its interactions with HDAC, because genetic or pharmacological disruptions of HDAC phenocopy many features of the bab mutant. Furthermore, removing the functions of either REREa or HDAC synergizes with loss of Fgf8 function to disrupt posterior mesoderm formation during somitogenesis, midbrain– hindbrain boundary maintenance, and pharyngeal cartilage development. Together, these results reveal novel in vivo roles for REREa in HDAC-mediated regulation of Fgf signaling. We present a model for RERE-dependent patterning in which tissue-specific transcriptional repression, by means of an REREa-HDAC complex, modulates growth factor signaling during embryogenesis. Developmental Dynamics 236:1891–1904, 2007. © 2007 Wiley-Liss, Inc. Key words: RERE; Atrophin-2; HDAC; Fgf8; zebrafish; craniofacial development; mesoderm formation Accepted 12 April 2007

INTRODUCTION The Atrophins are an elusive family of proteins whose biological functions are very poorly understood. Mammals have two family members, Atrophin1/DRPLA (Atr1) and RERE/Atrophin-2 (RERE) defined by a highly conserved, carboxy-terminal atrophin domain. In addition to an atr1 gene, zebrafish have two rere genes designated rerea and rereb. Mutations in the founding human member, Atr1, cause the progressive neurodegenera-

tive disorder dentatorubral-pallidoluysian atrophy (DRPLA) whose clinical features include epilepsy, cerebellar ataxia, and dementia (OMIM: 125370). An expanded polyglutamine (polyQ) tract within the Atrophin domain of Atr1 causes DRPLA, similar to other neurodegenerative disorders caused by polyQ expansion, such as Huntington’s, spinobulbar muscular dystrophy (SBMA), and spinocerebellar ataxias (SCA 1, 2, 3, 6, and 7). The affinity of Atr1 for RERE increases

when it contains an expanded polyQ tract, suggesting that it sequesters RERE away from its normal function contributing to disease pathogenesis (Yanagisawa et al., 2000). Requirements for RERE in neural development and function, however, remain unclear. RERE was first identified by its Atrophin domain, which resembles that of Atr1, including multiple stretches of arginine– glutamic acid repeats (RERE; Yanagisawa et al., 2000).

The Supplementary Material referred to in this article can be found at http://www.interscience.wiley.com/jpages/1058-8388/suppmat 1 Max-Planck Institute of Immunobiology, Freiburg, Germany 2 Department of Developmental and Cell Biology, University of California, Irvine, Irvine, California Grant sponsor: Human Frontier Science Program; Grant number: RGP9/2003; Grant sponsor: National Institutes of Health; Grant number: DE-13828; Grant number: NS-41353; Grant sponsor: March of Dimes; Grant number: 1-FY01-198; Grant sponsor: Epilepsy Research Training Program: NIH; Grant number: T32 NS045540-03. *Correspondence to: Nikki Plaster, Department of Developmental and Cell Biology, University of California, Irvine, 4462 Natural Sciences II, Irvine, CA 92697. E-mail: [email protected] DOI 10.1002/dvdy.21196 Published online 20 June 2007 in Wiley InterScience (www.interscience.wiley.com).

© 2007 Wiley-Liss, Inc.

1892 PLASTER ET AL.

However, RERE lacks a polyQ tract. Several other conserved domains in the amino-terminal half of RERE give clues as to its function, including BAH (Bromo adjacent homology), ELM2 (Egl-27 and MTA1 homology 2), SANT (SWI3/ADA2/N-CoR/TFIIIB), and Gata domains. The combination of SANT and RERE domains in RERE resembles the SMRT/N-CoR/ SMRTER family of transcriptional corepressors that bind orphan nuclear receptors (Wang et al., 2006). Evidence that RERE acts as a novel nuclear receptor corepressor include (1) binding to orphan nuclear receptors, Tlx and human chicken ovalbumin upstream promoter-transcription factor (COUP-TF) through a short conserved sequence in its Atrophin domain (Wang et al., 2006) and (2) histone deacetylase (HDAC) activity mediated by binding HDAC1/2 through its ELM2 and SANT domains (Zoltewicz et al., 2004; Wang et al., 2006). These data agree with studies of Drosophila’s only Atrophin family member, Atro (Grunge), which binds to the Even-skipped (Eve) and Huckebein (Hkb) transcription factors and acts as a corepressor (Zhang et al., 2002). Acetylation or deacetylation of histones usually leads to gene activation or repression, respectively, and can have important consequences for embryogenesis. Thus, the action of histone acetyltransferases (HATs) and HDACs are developmentally relevant. Studies in zebrafish have demonstrated a role of HDAC1 in the promotion of neural fates in the hindbrain and spinal cord by repressing Notch targets and SoxB1-class transcription factors (Cunliffe, 2004; Cunliffe and Casaccia-Bonnefil, 2006). Likewise, HDAC1 represses Wnt and Notch pathways to allow for the switch from proliferation to differentiation in the zebrafish retina (Yamaguchi et al., 2005). In addition to its role in neurogenesis, HDAC activity has been implicated in Fgf-mediated mesoderm induction (Weinstein et al., 1998; Xu et al., 2000), interactions with muscle differentiation factors such as MyoD (for review, see McKinsey et al., 2002), development of craniofacial cartilage and pectoral fins in zebrafish (Pillai et al., 2004) and regulation of homeobox gene expression (Chang et al., 2001;

Miller et al., 2004). The relatively ubiquitous expression of hdac in the embryo suggests that other factors lend specificity to its function. These include several HDAC-binding proteins and related repression complexes, such as HDAC1-RERE-Tlx (Wang et al., 2006), but relatively few such factors have been identified. Analysis of RERE mutant phenotypes in vertebrates has revealed requirements in multiple tissues during embryonic development. Defects in the openmind (om) mouse mutant in RERE/Atr2 include an open neural tube and malformations of the floor plate, telencephalon, pharyngeal arches, limb buds, and heart. Floor plate cells in the brains of om mutants initially fail to express Shh, and expression of Fgf8 in the anterior neural ridge is reduced and diffuse. Studies in om mutants, however, have been limited to early stages due to lethality by cardiac failure at approximately E9.5 (Zoltewicz et al., 2004). Recent analysis of a zebrafish rerea mutant that survives beyond embryogenesis has shown that it is also required to regulate Fgf signaling in the developing ear (Asai et al., 2006). Here, we show that rerea/atrophin-2 (rerea) is the gene disrupted in the zebrafish craniofacial mutant babyface (bab). Our analyses of the bab mutant phenotype reveals novel roles for REREa in cartilage, pectoral fin, midbrain– hindbrain, and posterior mesoderm development, and synergistic interactions between REREa and Fgf8 signaling. We find that bab mutants disrupt a subset of cell types that are affected in hdac-1⫺/⫺ mutants or HDAC inhibitor-treated embryos and that interactions with Fgf are likely to be mediated at least in part through association with HDAC. These studies provide evidence that REREa modulates the Fgf signaling pathway as a cofactor that imparts tissue specificity to histone deacetylation and chromatin silencing during development.

RESULTS Disruptions of rerea Are Responsible for the bab Phenotype Two alleles of bab were identified during the 1996 Tuebingen screen,

babtb210 and babtw220c, both displaying nearly identical phenotypes (Schilling et al., 1996). Mutant characteristics included a small head and eyes, severely reduced pharyngeal arches and pectoral fins, and enlarged ear capsules (Fig. 1A,B). Our bulk segregation analysis of babtb210 placed the bab locus at approximately 32.2 cM on linkage group 23 between the markers z15422 and z42693 (Fig. 1C). These markers were used to identify a contig Zv5_scaffold1160, with which additional fine-mapping markers were designed (NP1, NP2, NP3, and NP4). Further fine mapping of 2,094 alleles defined the bab locus by the north markers z15422 and NP1 (1 recombination) and the south marker z42693 (3 recombinations). No recombinations were found for the markers NP2, NP3, and NP4. Zv5_scaffold1160 contains a predicted gene (ENSDARG00000037331) with similarity to mammalian RERE/ Atrophin-2 (RERE). To identify the entire coding sequence of this gene in zebrafish, we performed reverse transcriptase-polymerase chain reaction (RT-PCR) on wild-type and mutant cDNA with primers throughout the predicted exons of ENSDARG00000037331, including the 5⬘ and 3⬘ untranslated regions. Our analysis of this sequence showed that it encodes zebrafish REREa (1517 aa) with a predicted amino acid identity of 75% with both human and mouse RERE/Atr2. Zebrafish REREa contains several putative DNA binding domains in its amino terminus and an Atrophin domain (putative protein–protein interaction domain) in its carboxy terminus (Fig. 1E). Sequencing of mutant cDNA revealed that both alleles of bab contain nonsense mutations in the SANT domain (babtb210: Y361X, babtw220c: Y394X) resulting in truncations of REREa at this domain. From these data, we conclude that disruptions in rerea are responsible for the bab phenotype.

rerea Is Expressed Throughout Development To examine when and where rerea is expressed, we performed in situ analysis from 4 –72 hours postfertilization (hpf; Fig. 2). rerea is maternally sup-

REREA/ATROPHIN-2 IN ZEBRAFISH DEVELOPMENT 1893

plied (data not shown) and expressed ubiquitously until approximately 8 hpf (70% epiboly). At that time, expression decreases but persists in the mesodermal germ ring, predominantly in the embryonic shield on the dorsal side (Fig. 2A,B). At tail bud stage, expression begins in the head and spreads to become ubiquitous throughout somitogenesis (Fig. 2C,D). After 24 hpf, posterior mesodermal expression begins to decrease while strong expression remains throughout the head, spinal cord, gut, and pectoral fins (Fig. 2E–G). These areas of expression are maintained through 3 days postfertilization (dpf) with apparently uniform expression throughout the central nervous system (CNS; Fig. 2H). At 48 hpf, rerea is also expressed throughout the pectoral fin bud and in

Fig. 1. The bab phenotype is caused by a disruption in rerea. A,B: The wild-type (WT, A) and bab (B) larvae at 5 days postfertilization (dpf) are shown in lateral view. Ventral distension of the pharyngeal area in bab is caused by the inverted second arch cartilage. C: The bab locus is at approximately 32 cM on linkage group 23. Expanded view of the genetic map shows the number of recombinations in 2,094 alleles above the marker name. The genomic contigs that span the completely linked region, their Genbank accession numbers, and the position of rerea are shown below. D: Sequence chromatographs of both the babtb210 and babtw220c in comparison with wild-type sequence are shown with the mutation underlined. E: The REREa schematic displays the positions of identified protein domains with colored boxes and nuclear localization signals (NLS) with stars. Arrowheads mark the location and direction of primers used to amplify cDNA. Nonsense mutations in both alleles of bab truncate REREa in the SANT domain.

Fig. 1.

Fig. 2. rerea is expressed dynamically in all developmental stages examined. All panels show whole-mount in situ hybridizations of rerea for wild-type embryos or larvae of ages indicated in upper right corner. A–G: Lateral views with anterior to the left (A,C,D,G), dorsal view with animal pole up (B), lateral view on tail (E), and frontal view of fin bud with medial to the left (F). H,I: Frontal section though head (H) and horizontal sections with anterior to the left through the pharyngeal arches (I). d1, dorsal cartilages of arch 1; e, endoderm; gcl, ganglion cell layer; inl, inner nuclear layer; j, joint; m, somitic mesoderm; pa, pharyngeal arches; prcl, photoreceptor cell layer; sc, spinal cord; tb, tail bud; v1 and v2, ventral cartilages of arches 1 and 2, respectively. Fig. 2.

1894 PLASTER ET AL.

the ectoderm, neural crest, and endoderm of the pharyngeal arches (Fig. 2G). Soon afterward, cartilage begins to differentiate in the arches and pharyngeal rerea expression is limited to the endoderm surrounding the cartilages, with particularly high levels of mRNA localized adjacent to the developing jaw joint in the first arch (Fig. 2I). rerea expression at 72 hpf is strongest in the CNS, including the retina (Fig. 2H) and spinal cord, in the epithelial pillars and maculae of the ear, olfactory epithelium, neuromasts, pharyngeal endoderm, and intestine (data not shown).

bab/rerea Is Required in Multiple Embryonic Tissues bab mutants were originally identified by their small head and eyes, severely reduced pharyngeal arches and pectoral fins at larval stages. Further examination of bab has uncovered a complex phenotype, consistent with the dynamic expression of rerea throughout development. Homozygous bab mutants can first be identified at 23 hpf by defects in the forebrain, similar to RERE⫺/⫺ mice (Zoltewicz et al., 2004). Dorsal telencephalic markers such as emx1 (Fig. 3A,B) and eom (not shown) expand ventrally, with a corresponding reduction of ventral markers such as dlx2 (Fig. 3D,E) and nk2.1b (data not shown). Thus, REREa is necessary for dorsal–ventral (D-V) patterning of the telencephalon. The craniofacial skeleton in bab mutants is reduced to varying degrees (Fig. 4A,B; Schilling et al., 1996). Mutant chondrocytes remain rounded instead of flattening into stacks as in wild-type larvae (Fig. 4L,M). Additionally, dorsal (d1) and ventral (v1) cartilages fuse in the first pharyngeal arch (mandibular) in mutants, resulting in a loss of the joints (Fig. 4F middle). In severe cases, the second arch (hyoid) joint also fuses. We examined markers of joint specification and chondrocyte differentiation in bab mutants, such as bapx-1, which is necessary for joint development; loss of function results in the fusion of first arch elements along the D-V axis (Miller et al., 2003; Wilson and Tucker, 2004). Expression of bapx-1 is severely reduced in bab mutants (Fig.

4H). Subsequently, mutants fail to down-regulate the chondrocyte differentiation marker sox9a at the joint, and instead the cells mis-express type II collagen (col2a1; Fig. 4K). This appears to lead to fusion of d1 and v1 cartilages through chondrification of presumptive joint cells. Thus, we conclude that REREa is required for joint specification in the first arch, most likely in joint maturation. In addition to the CNS and pharyngeal cartilage, bab mutants exhibit defects in ear, eye, and pectoral fin development. In the ear, the epithelial pillars that form the semicircular canals fail to fuse and the ear capsule expands by 72 hpf (Supplementary Figure S1A,B). Small eyes in mutants correlate with a failure of the choroid fissure to fuse (coloboma; data not shown) and in severe cases retinal tissue merges with the optic nerve (Supplementary Figure S1C,D). Photoreceptor differentiation is delayed in the bab mutant retina (data not shown), and retinal lamination is disrupted (Supplementary Figure S1C,D). Finally, bab mutants have severe defects in their pectoral fins. The early mutant fin bud at 32 hpf expresses ptc1, indicative of Shh signaling, suggesting that fin bud initiation occurs (Supplementary Figure S1E,F). However, later maintenance of the Shh– Fgf feedback loop is disrupted, because by 40 hpf, the bud loses expression of ptc1 (data not shown) and the Fgf target mkp3 (Supplementary Figure S1G,H), and fin outgrowth fails.

rerea Genetically Interacts With fgf8 but Not fgf24 or fgf3 in Mesoderm Patterning of the telencephalon and the mandibular arch depends on Fgf8 (Tucker et al., 1999; Shigetani et al., 2000; Shinya et al., 2001). To test if rerea and fgf8 interact genetically, we crossed adults heterozygous for both bab and acerebellar (ace), the zebrafish fgf8 mutant. To our surprise, bab;ace double mutants (bab;ace) have severely truncated tails and lack posterior tail mesoderm, while no such tail defects are observed in either single mutant alone (Fig. 5A–C). In severe cases, the truncation occurs at the approximate level of somite 12. Fgf signaling is a

well-known regulator of somitic mesoderm and tail development (Kimelman and Kirschner, 1987; Amaya et al., 1991; Griffin et al., 1995), and fgf8, fgf24, and fgf3 are coexpressed in the marginal region of the gastrula, as is rerea. By 15 hpf (13-somite stage), phosphorylated ERK1/2 (indicative of Fgf signaling) is severely reduced in the tail bud of bab;ace but only weakly reduced in ace alone (Fig. 5A⬘–C⬘), whereas wild-type and bab embryos are indistinguishable. Analysis at earlier stages revealed that early mesodermal precursors in bab;ace are induced normally and express spt and ntl throughout gastrulation (data not shown). However, by 11 hpf (four-somite stage), presomitic mesoderm (marked by papc expression) is severely reduced in double mutants compared with ace alone (Fig. 5D–F), and by mid-somitogenesis, reductions in paraxial (myod) and axial (Ntl) mesoderm are observed in the posterior tail (Fig. 5G–I). These data suggest that rerea and fgf8 genetically interact in the posterior mesoderm, possibly to maintain a population of undifferentiated mesodermal precursors. fgf24, which is mutated in ikaros (ika), is proposed to act redundantly with fgf8 to pattern posterior mesoderm. ace;ika double mutants display synergistic tail defects and reductions in mesodermal expression of spt and ntl during gastrulation (Draper et al., 2003). As mentioned previously, fgf3, which is mutated in lia, is also expressed in the germ ring (future mesendoderm). However, ace;lia double mutants (Martinez-Morales et al., 2005) do not display tail defects (M.H. and J. Wittbrodt; unpublished observations). To determine whether rerea genetically interacts with either fgf24 or fgf3, we constructed bab;ika and bab;lia double mutants. Interestingly, the tails of both of these double mutants develop normally (data not shown), suggesting that genetic interactions with rerea in the posterior mesoderm are specific to fgf8 (see also the Discussion section).

Additional Genetic Interactions Between rerea and Fgf Signaling Maintenance of the midbrain– hindbrain boundary (MHB) depends on Fgf signaling; the zebrafish fgf8 mu-

tant ace does not form the MHB (Reifers et al., 1998). pax2.1 is also necessary for the maintenance but not the induction of the MHB and is dependent on Fgf8 during somitogenesis (Reifers et al., 1998; Heisenberg et al., 1999). In bab mutants, the MHB appears thin by 48 hpf, and fgf8 expression takes on a speckled appearance (Fig. 6A,B). This finding suggests a role for REREa in maintenance of the MHB. To test if the MHB in bab is sensitive to a reduction in Fgf8, we examined pax2.1 expression in

Fig. 3. Disruption of REREa or histone deacetylase (HDAC) activity causes a dorsalization of the telencephalon. A–F: All panels show lateral views of the telencephalon in whole-mount in situ hybridizations on wild-type (WT, A,D), bab (B,E) or Trichostatin A (TSA) -treated (C,F) embryos. Probes are denoted in the lower left corner with stages in the lower right. A–C: emx1 expression extends ventrally from the dorsal telencephalon to the arrowhead and is expanded in bab and TSA-treated embryos. The dot marks the ventral telencephalic boundary. D–F: This results in a reduction of dlx2 expression in the telencephalon. Arrows mark the dorsal limit of dlx2 expression. pax2.1 expression (white asterisk) in F overlaps the diencephalic expression of dlx2, but is not expressed in the telencephalon.

Fig. 3.

Fig. 4. rerea genetically interacts with fgf8 in pharyngeal arch development, and disruption of histone deacetylase (HDAC) activity results in bab-like fusions of the first arch. A–M: Alcian blue-stained cartilage flat-mounts at 5 dpf (A– F), whole-mount in situ hybridizations of embryos with bapx1 (G–I) and col2a1 (J,K), and frontal sections through the palatoquadrate (d1 element, L,M) are shown for wild-type (WT; A,F top, G,J,L), bab (B,F middle, H,K,M), Trichostatin A (TSA) -treated (C,F bottom, I), ace (D), and bab;ace (E) embryos or larvae. Anterior is to the left in all images except G–I,L,M, which are frontal views on the mouth. A–E: Ventral views of flat-mounted cartilages with posterior arches numbered 3–7, reveal the presence but reduction in all elements in bab (B), a missing posterior arch in both TSA-treated (C) and ace (D), and a complete loss of posterior arches in bab; ace (E). Disruption of REREa (H) or TSA treatment (I) results in reduction of joint specification marker, bapx1 compared with wild-type embryos (G). Later col2a1 is misexpressed in cells of the presumptive joint (white arrow) in bab (K) compared with WT (J) leading to a fusion of the first arch cartilages (F, lateral view of arches 1 and 2) in wild-type (top F), bab (middle F), and TSA-treated (bottom F) embryos. Dots mark the position of the joint in each preparation. Chondrocyte morphology at 4 hpf is round in bab (M) compared with the stacked structure in wildtypes (L). d1 and d2, dorsal cartilages of arches 1 and 2, respectively; v1 and v2, ventral cartilages of arches 1 and 2, respectively.

Fig. 4.

1896 PLASTER ET AL.

clutches from bab⫹/⫺;ace⫹/⫺ double heterozygous females mated with bab⫹/⫺ heterozygous males (no ace in background). By double labeling these embryos for dlx2 expression in the forebrain, we could independently confirm distinct mutant classes. We found that 50% of bab mutants had an MHB but severely reduced pax2.1 expression (Fig. 6E). By PCR genotyping, we confirmed that these embryos were heterozygous for ace. In contrast, ace heterozygotes and bab⫺/⫺; ace⫹/⫹ mutants displayed normal pax2.1 expression in the MHB (Fig. 6C,D). This finding suggests that Fgf signaling is compromised in the MHB of bab mutants and that further reduction of signaling by heterozygosity at the fgf8 locus can no longer maintain pax2.1 expression and development of the MHB. In addition to fgf8, the MHB expresses fgf3. However, pax2.1 expression is unaffected in lia null mutants, which lack a functional fgf3, as well as in bab⫺/⫺; lia⫹/⫺ and bab⫺/⫺; lia⫺/⫺ double mutants (data not shown), further confirming that rerea specifically interacts with fgf8. In contrast to the MHB, craniofacial development depends on both Fgf8 and Fgf3, with Fgf3 playing a predominant role (David et al., 2002; Walshe and Mason, 2003; Crump et al., 2004). Therefore, we examined bab;ace and bab;lia double mutants for genetic interactions in pharyngeal arch development. As in bab (Fig. 4B), pharyngeal cartilages in ace mutants are of rela-

Fig. 5. rerea interacts genetically with fgf8 in posterior mesoderm development. A–C: Lateral views of live 24 hpf embryos with genotypes in the upper right corner. Aⴕ–Cⴕ: Inset panels show the corresponding lateral views of diphosphorylated ERK1&2 (pERK) immunostained tail buds at 13 somites. Arrows denote the position of pERK staining. D–I: Wholemount in situ hybridizations of papc at 4 somites (D–F) or myod with Ntl immunostains at 16 somites (G–I) shown in dorsal view with genotypes shown in the upper right corner. The arrowhead in I marks a gap in Ntl staining. “WT(bab)” notes that there is no observable difference between wild-type (WT) and bab embryos. ace shows a slight reduction in the amount of posterior mesoderm (B) and presomitic mesoderm as seen by papc expression at four somites (E), while bab;ace (C,F) shows a severe reduction in comparison to wild-type or bab (A,D).

Fig. 5.

REREA/ATROPHIN-2 IN ZEBRAFISH DEVELOPMENT 1897

Fig. 6. rerea and fgf8 genetically interact in the midbrain–hindbrain boundary (MHB), and disruption of REREa or inhibition of histone deacetylase (HDAC) activity results in the inability to maintain the MHB. A–F: All panels show whole-mount in situ hybridizations in frontal view (A,B) or lateral view (C–G) of 48 hours postfertilization (hpf; A–E) or 24 hpf (F,G) embryos. Genotypes are shown in the upper right corner, whereas probes and stages are shown in the lower left and right corners, respectively. A,B: bab (B) displays degeneration of the MHB as seen by the patchy disappearance of fgf8-expressing cells compared with WT embryos (A). C,D: pax2.1 is expressed in the MHB (arrowheads) in wild-type (WT; ace⫹/⫺, C) and bab⫺/⫺;ace⫹/⫹ (D) embryos at 48 hpf. E: Reduction of Fgf8 in bab⫺/⫺;ace⫹/⫺ leads to a reduction of pax2.1 in the MHB. dlx2 expression was used for the purpose of identifying bab embryos and does not interfere with the MHB staining of pax2.1. F,G: The 350 nM Trichostatin A (TSA) treatment from dome stage to 24 hpf also leads to a loss of pax2.1 expression at the MHB (asterisk in G) compared with wild-type embryos treated with dimethylsulfoxide (DMSO) alone (F). C–G: pax2.1 expression in the eye (arrows) is up-regulated in bab (D,E) and TSA-treated embryos (G) compared with wild-types (C,F).

tively normal size, but individual arches are often missing on one side or branched (Fig. 4D), correlated with a failure of the endodermal pouches to properly subdivide the neural crest (Crump et al., 2004). In stark contrast to either of these single mutants, bab; ace double mutants have severely reduced anterior cartilages and completely lack arches 3–7 (Fig. 4E). Cranial neural crest migrates into the arches in double mutants as revealed by the expression of dlx2 at 24 hpf (data not shown) but the streams of neural crest do not subdivide properly, suggesting a disruption in the endodermal pouches. This suggestion was confirmed by the loss of expression of nkx2.3 at 48 hpf in the pharyngeal pouches in bab;ace mutants (Fig. 7A,B). This finding is not due to an overall reduction in pharyngeal endoderm, because expression of foxa2 in more medial domains of endoderm is largely unaffected, although slightly altered in shape at its lateral edges (Fig. 7C,D). Compared with fgf8 (ace) and rerea (bab), mutants in fgf3 (lia) display much more severe defects in arches 3–7, which are completely missing at 120 hpf, whereas arches 1 and 2 develop normally (Herzog et al., 2004). In contrast to its effect on ace (see

Fig. 7. rerea and fgf8 interact during formation of the endodermal pouches. A–D: All panels show dorsal views, with anterior to the left, of whole-mount in situ hybridizations of wild-type (A,C) and bab;ace (B,D) embryos. Probes are shown in the lower left corner, whereas stages are in the lower right. Combined disruption of rerea and fgf8 leads to loss of the posterior endodermal pouches as seen by nkx2.3 expression (A,B) while medial endoderm (foxa2) remains (C,D). Horizontal bars (C,D) mark the anterior–posterior extent of the pharyngeal endoderm. p1–5, endodermal pouches 1–5; pe, posterior endoderm.

1898 PLASTER ET AL.

above), the effect of loss of REREa function on lia is additive, rather than synergistic, bab;lia double mutants lack arches 3–7 like lia single mutants, whereas arches 1 and 2 resemble bab single mutants (data not shown), and heterozygosity for lia does not strengthen the arch defects in bab mutants. We also failed to observe any differences between bab⫺/⫺; lia⫹/⫹ and bab⫺/⫺;lia⫹/⫺ mutants in the expression of nkx2.3 in the endodermal pouches at 32 and 48 hpf, or in cartilage morphology of the arches at 120 hpf (data not shown). These results show that rerea genetically interacts with fgf8 but not fgf3 in endodermal development and suggest that Fgf8 and Fgf3 functions are not completely redundant in pharyngeal arch development.

bab Displays Both Up- and Down-regulation of Fgf Pathway Transcripts The mechanisms by which REREa interacts with Fgf signaling are not understood. It is known that Fgf signaling involves several levels of autoregulation. In addition to the negative feedback loops of Sef and Mkp3 (Tsang and Dawid, 2004), Fgf8 itself is thought to be autoregulated (Reifers et al., 1998; Shanmugalingam et al., 2000; Furthauer et al., 2001). We examined expression of several Fgfs and Fgf target genes in bab embryos at the 12- to 14-somite stage (data not shown) and 24 hpf (double labeled in red with dlx2 for genotyping purposes). We did not observe any differences in pattern or intensity of fgf8, fgf3, sef, and mkp3 expression in 12- to 14-somite stage bab mutants compared with their wild-type siblings (data not shown), but at 24 hpf there are subtle differences that can vary depending on the gene and structure in which it is expressed (Supplementary Figure S2A–H). Thus, compared with wild-type siblings (Supplementary Figure S2A), fgf8 expression in the posterior otic capsule of bab mutants increases, while expression in the telencephalon decreases (Supplementary Figure S2A,E). In contrast, fgf3 expression decreases in mutants at the MHB and increases in the ventral diencephalon (Supplementary Figure S2B,F). sef (Supplementary

Figure S2C,G) and mkp3 (Supplementary Figure S2D,H) expression levels in bab mutants show no differences from those in wild-type siblings. Together, these results highlight the complex nature of Fgf regulation and tissue specificity and suggest that REREa does not directly regulate the expression of these Fgf pathway molecules.

HDAC Inhibition Phenocopies bab Because RERE interacts with HDAC1 and HDAC2, we asked if disruption of HDAC activity reproduced any of the defects in bab mutants. Previous studies have shown that hdac1 mutants in zebrafish display defects in eyes, ears, pectoral fins, and craniofacial development similar to bab mutants (Pillai et al., 2004) and that treatment of live zebrafish embryos with Trichostatin A (TSA) leads to the inhibition of HDAC activity (Collas et al., 1999; Miller et al., 2004; Yamaguchi et al., 2005). We incubated embryos from late gastrula stages (80% epiboly/8.5 hpf) until 24 hpf in various amounts of TSA and examined D-V markers in the telencephalon. Embryos incubated in 500 nM TSA had a ventral expansion of emx1 and dorsal reduction of dlx2 in the telencephalon, mimicking the bab phenotype (compare Fig. 3C,F with Fig. 3B,E). Furthermore, treatment of embryos from 4.3 hpf to 24 hpf with 350 nM TSA led to a loss of fgf8 (data not shown) and pax2.1 expression in the MHB and an increase in pax2.1 expression in the eye (Fig. 6F,G) as is observed in bab⫺/⫺;ace⫹/⫺ mutants (see Fig. 6E for a slightly later stage). These studies provide evidence that HDAC activity is necessary for proper D-V patterning of the telencephalon and maintenance of the MHB, as is REREa function. In addition, we found similar effects of loss of HDAC and REREa function during joint formation in the first two pharyngeal arches. Sox9 regulates type II collagen expression by interacting with the histone acetyltransferase p300/CBP (Furumatsu et al., 2005), suggesting that transcriptional regulation by histone modification could be an important aspect of cartilage and joint formation. Indeed, after treatment with 500 nM of TSA from

19 hpf until 52 hpf, larvae at 120 hpf display a fusion of both the first and second arch joints similar to bab mutants (compare Fig. 4F, bottom, with Fig. 4F, middle), and the occasional loss of posterior cartilages as in ace mutants (compare Fig. 4C with Fig. 4D). Similarly, bapx1 expression is severely reduced in TSA-treated embryos at 46 hpf (treated with 500 nM TSA from 13.5 hpf [12 somites] through 46 hpf; Fig. 4I). These data indicate that histone deacetylation is necessary for proper joint specification and repression of chondrocyte differentiation factors at the presumptive joint. They support a model in which REREa regulates joint formation through its reported association with HDAC.

Fgf Signaling Is Sensitive to Reductions in HDAC Activity If the proposed interaction of REREa and HDAC also applies to REREa’s role during Fgf signaling, we would expect ace mutants to be particularly sensitive to inhibition of HDAC activity, similar to the genetic interaction between ace and bab described above. To test this prediction, we incubated embryos from ace/fgf8, ika/fgf24, or lia/fgf3 heterozygous parents in various concentrations of TSA at 30% epiboly (4.6 hpf) until 24 hpf and examined tail morphology. In 100 nM TSA, ace mutants at 24 hpf are indistinguishable from their wild-type siblings. In 200 nM TSA, tail defects in ace mutants resemble bab;ace double mutants, whereas TSA-treated wildtype controls are unaffected (Fig. 8A– D). The disruption of the posterior tail in TSA-treated ace embryos can be observed by the four-somite stage (11 hpf) as a severe reduction of presomitic mesoderm, similar to bab;ace double mutants (compare Fig. 8E–H with Fig. 5D–F). In agreement with our findings on bab;ika and bab;lia double mutants, neither ika nor lia mutants are sensitive to TSA treatments (data not shown). These results suggest that embryos with disrupted Fgf8 signaling are particularly sensitive to decreases in HDAC activity in the posterior mesoderm. To determine the effects of reducing HDAC activity on Fgf targets, we ex-

REREA/ATROPHIN-2 IN ZEBRAFISH DEVELOPMENT 1899

amined the expression of fgf8, fgf3, mkp3, and sef at the seven-somite stage (13 hpf) in embryos treated with 350 nM TSA from 4.5 hpf through fixation (see Fig. 9). TSA treatment decreased or abolished the expression of all four markers in the telencephalon and tail bud, and eliminated the expression of fgf8 and sef in the developing somites (Fig. 9). The MHB expression of the four markers was more differentially affected by the TSA treatment. The normally distinct MHB domains of fgf8 and mkp3 expression became diffuse (compare Fig. 9B,F with Fig. 9A,E), and fgf3 remained largely unaffected (compare Fig. 9D with Fig. 9C), whereas sef expression in the MHB was completely lost (compare Fig. 9H with Fig. 9G). The mild defects in MHB expression of fgf8 and mkp3 at this stage, compared with the complete loss of fgf8 and pax2.1 when embryos are TSA-treated until 24 hpf (see Fig. 6), are consistent with our findings that REREa-HDAC is required for maintenance but not induction of the MHB. Generally, Fgf target gene expression upon HDAC inactivation is most severely affected in those tissues that later display defects in bab or bab;ace double mutants (telencephalon, posterior mesoderm, MHB; see above). However, the effects of TSA treatment on Fgf target gene expression are more severe and develop earlier than in bab mutants (compare Fig. 9 with Supplementary Figure S2), suggesting that the function of HDAC is not restricted to its interaction with REREa, and that HDAC can regulate Fgf targets through additional REREa-independent pathways.

DISCUSSION Here, we demonstrate that REREa is required for development of multiple tissues during zebrafish embryogenesis, including the eye, ear, craniofacial cartilages, telencephalon, MHB, and the pectoral fins. From our TSA studies and characterization of the hdac1 mutant by Pillai et al. (2004), we propose that REREa acts in an HDACdependent manner in each of these tissues. rerea genetically interacts with Fgf8 signaling in development of the mesendoderm and in the maintenance of the MHB, most likely

through its association with the transcriptional repressor HDAC. Our results lead us to propose a model in which REREa acts as a tissue-specific transcriptional corepressor that coordinates growth factor signaling. In particular, we provide evidence for an in vivo function of REREa and HDAC in promoting Fgf8 signaling, although the molecular details of this interaction remain largely unclear.

Conserved Requirements for RERE in Vertebrate Embryonic Development Atrophin-1 was identified over 12 years ago as the gene disrupted in the neurodegenerative disorder DRPLA (Koide et al., 1994; Nagafuchi et al., 1994). Both Atr1 and RERE/Atr2 have been implicated in neuronal maintenance and survival in adults. However, we show that zebrafish rerea is required in numerous other tissues during development including the retina, ear, and craniofacial skeleton. The bab mutant phenotype is consistent with the dynamic, widespread expression of rerea both maternally and zygotically in the embryo. RERE is highly conserved in zebrafish, mouse, and human both in structure and function. Early phenotypes caused by loss-of-function mutations in both fish and mouse include defects in D-V patterning of the telencephalon and limb/fin buds. The mouse openmind mutant dies at a much earlier stage than the zebrafish bab mutant, possibly due to partial genetic redundancy in zebrafish. We identified a second rere gene in the zebrafish genome, rereb, which is approximately 60% and 50% identical to zebrafish rerea and human RERE/ ATR2, respectively (data not shown). Both zebrafish genes display conserved synteny with their human homologue RERE/ATR2. Preliminary RT-PCR analysis suggests that rereb transcript is also maternally supplied and expressed throughout development, and we are currently investigating the role of this gene during development.

REREa Is Required for HDAC-Dependent Processes Mammalian RERE has been shown to bind HDACs and repress transcrip-

tion in vitro; the significance in vivo remains unclear. HDACs are widely expressed in zebrafish embryos, although inhibition of their functions leads to very specific developmental defects (Pillai et al., 2004; Cunliffe and Casaccia-Bonnefil, 2006). Pillai et al. (2004) described bab-like defects in the ear, retina, and pectoral fin of hdac1 mutants, and we found that inhibition of HDAC activity results in bab-like defects in joint formation in the mandibular arch, MHB maintenance, and D-V patterning of the telencephalon. The specificity of HDACs may be due in part to tissue-specific associations with cofactors such as REREa. Recently, it was shown that binding of RERE to nuclear receptors brings the RERE-HDAC complex into association with areas of the genome to be repressed (Wang et al., 2006). Whether providing nuclear receptors with a link to HDACs is RERE’s only function remains unclear. We can recreate all known bab phenotypes by inhibition of HDAC activity, providing compelling support for REREa acting mainly through its HDAC association. However, defects resulting from general inhibition of HDAC activity are more severe than defects observed in bab, most likely due to REREa-independent roles for HDAC. For instance, inhibition of HDAC activity during gastrulation and somitogenesis leads to a strong decrease in expression of fgf8, fgf3, sef, and mkp3, while bab mutant embryos do not show such early reductions. It is notable that in cell culture studies, RERE has been detected in both the nucleus and in the cytoplasm (Yanagisawa et al., 2000; Waerner et al., 2001), and Drosophila’s only Atrophin, Atro/Grunge, can bind a transmembrane atypical cadherin, Fat (Fanto et al., 2003). Therefore, it is possible that RERE has HDAC-independent activities that we have not yet uncovered. Of interest, it has been demonstrated previously that the other family member of the Atrophin family, Atr1, does not bind HDACs (Wang et al., 2006), but does bind to RERE. Investigation into the consequences of this interaction and its relationship to the HDAC activity of RERE could provide interesting insights into the regulation of transcriptional repression of this complex.

1900 PLASTER ET AL.

Fig. 8.

Fig. 9.

REREA/ATROPHIN-2 IN ZEBRAFISH DEVELOPMENT 1901

rerea Genetically Interacts With Fgf8, but Not With Fgf3 or Fgf24 Signaling A common feature of the different cell types disrupted in bab mutants is that their early development is regulated by Fgf signaling. Our double-mutant analysis indicates that bab mutants have reduced Fgf signaling in many of these tissues. rerea interacts genetically with fgf8 in maintenance of the posterior mesoderm. Thus, whereas mesoderm is normal in rerea/bab single mutants and only moderately affected in fgf8/ace single mutants, double mutants display a severe reduction in posterior mesoderm, very similar to that of mutants lacking Fgf8 and Fgf24 (Draper et al., 2003). Remarkably, rerea does not show this interaction with fgf3 or fgf24, despite the partially redundant roles of fgf8 and fgf24 in mesodermal development. Furthermore, in support of a rerea;fgf8 interaction during posterior mesoderm development, we have shown that the phenotype of fgf8 mutants, but not that of fgf3 or fgf24 mutants, is synergistically enhanced by

Fig. 8. Histone deacetylase (HDAC) activity is required for proper Fgf signaling during development of the posterior mesoderm. A–D: Lateral views of live 26 hours postfertilization (hpf) wild-type (A,B) or ace (C,D) embryos, treated with either dimethylsulfoxide (DMSO) control medium (A,C) or 200 nM Trichostatin A (TSA, B,D) are shown. TSA-treated embryos are slightly delayed. The black arrow in D marks severely disrupted tail mesoderm, resembling the bab;ace phenotype at 24 hpf. E–H: Wholemount in situ hybridizations (dorsal views on the tail bud) of papc at the four-somite stage in wild-type embryos (E,F) or ace (G,H) treated with DMSO control medium (E,G) or 200 nM TSA (F,H) from 30% epiboly. Fig. 9. The proper expression of fgf3 and fgf8 and fgf target genes, sef and mkp3, requires histone deacetylase (HDAC) activity. A–H: All panels show lateral views of whole-mount in situ hybridizations of seven-somite stage WT embryos treated with dimethylsulfoxide (DMSO, A,C,E,G) or 350 nM Trichostatin A (TSA, B,D,F,H) from dome to the seven-somite stage. Probes are shown in the bottom left corner of each wild-type panel. A,B,E,F: The forebrain (arrows) and tail bud (asterisk) expression of all markers is decreased or absent in TSAtreated embryos while the midbrain– hindbrain boundary (MHB) domains (dots) are decreased and diffuse. Both fgf8 (A,B) and mkp3 (E,F) expression in the developing somites (arrowheads) is also lost in TSA-treated embryos.

mild TSA treatments that partially inhibit HDAC function (Fig. 8). The rerea;fgf8 genetic interaction is important in other tissues as well. We believe that reduced Fgf signaling in the bab mutant MHB underlies the observed degeneration of this structure after 48 hpf. This idea is supported by the fact that further reduction of Fgf8 signal in bab with heterozygosity for the ace allele leads to loss of pax2.1 expression in the MHB, a phenotype also caused by inhibition of HDAC activity. Furthermore, our analysis uncovers a synergistic effect of REREa and Fgf8 in endodermal development, as bab;ace double mutants lack all but the first pharyngeal pouch. Previous studies have shown that these endodermal pouches are necessary for pharyngeal cartilage differentiation (David et al., 2002). Thus, the cartilage defect observed in the double mutants is likely to be secondary. The phenotype of bab; ace double mutants is more severe than that seen in fgf3⫺/⫺ embryos (Herzog et al., 2004), but less severe than in homozygous ace embryos in which Fgf3 has been depleted by means of morpholino injection (Crump et al., 2004). Of interest, we did not observe any genetic interaction between rerea and fgf3 in endodermal pouch or cartilage development. This specific genetic interaction with Fgf8 does not rule out a role for REREa in signaling through other Fgfs. If REREa acted exclusively upor downstream of Fgf8 signaling, simultaneous reduction of REREa and Fgf8 activities should have a synergistic effect, whereas defects caused by complete loss of both activities should resemble loss of either of them alone. This is the case for the MHB, in which rerea⫺/⫺; fgf8⫹/⫺ embryos are more strongly affected than rerea⫺/⫺ single mutants, whereas rerea⫺/⫺;fgf8⫺/⫺ double mutants resemble fgf8⫺/⫺ single mutants (Fig. 6; and data not shown). However, things are different for the posterior mesoderm and pharyngeal arches. In the posterior mesoderm, Fgf24 is also involved. Loss of Fgf8 activity does not eliminate phosphorylated ERK1/2 expression or tail formation, whereas both traits are strongly enhanced in rerea⫺/⫺;fgf8⫺/⫺ double mutants. Of interest, the tail defects in these

double mutants are identical to fgf8⫺/⫺;fgf24⫺/⫺ embryos, suggesting that it is Fgf24 activity that requires REREa, whereas Fgf8 itself is sufficient to allow proper posterior development even in the absence of REREa or Fgf24. Whether this dominance and REREa-independence of Fgf8 versus Fgf24 is due to quantitative differences in the relative contributions of the two signals, or to qualitative differences (e.g., time points of action; activation of different signaling pathways), remains unclear. In the pharyngeal endoderm, Fgf3 is the predominant Fgf signal. However, loss of REREa function does not enhance the phenotype caused by loss of Fgf3, but again only interacts genetically with Fgf8. In addition, fgf3 heterozygosity fails to enhance the pharyngeal defects of rerea mutants. Together, this finding suggests that, in contrast to Fgf8 and Fgf24, REREa does not influence Fgf3. The mechanisms underlying this selectivity are unclear. However, it is interesting to note that Fgf8 and Fgf24 belong to the same Fgf subgroup, which is believed to signal through the IIIb isoform of Fgf receptor 2 (Fgfr2IIIB), whereas Fgf3 signals through Fgfr2IIIC and possibly Fgfr1 (Dailey et al., 2005), suggesting that REREa might selectively promote signaling through particular receptors. Thus, our data further point to the existence of a thus far unidentified member of the Fgf8/24 subgroup involved in pharyngeal endoderm development to explain the strong genetic interaction between rerea and fgf8.

Possible Mechanisms of FgfREREa-HDAC Interaction and REREa Specificity The synergistic enhancement of the phenotype of fgf8 mutants by loss of REREa or by HDAC inhibition provides compelling evidence for an in vivo role of REREa/HDAC-mediated transcriptional repression in promoting Fgf signaling. This finding is consistent with the recently reported rescue of an REREa morphant ear phenotype by simultaneous morpholino-based inactivation of the Fgf signaling inhibitor Sef, to restore stronger Fgf signaling (Asai et al., 2006). However, the exact molecular mecha-

1902 PLASTER ET AL.

nism by which REREa and HDAC might promote Fgf signaling remains largely unclear. Negative regulation of the Fgf pathway by Fgf-induced antagonists, such as Sef, Spry4, and Mkp3 (for review, see Tsang and Dawid, 2004), Fgf8’s proposed autoregulation (Reifers et al., 1998; Shanmugalingam et al., 2000; Furthauer et al., 2001), and the redundancy of Fgf signaling make it difficult to carry out conclusive epistasis analyses. Asai et al. (2006) reported a strong up-regulation of fgf8, sef, and spry4 in REREa morphant embryos and suggested that REREa functions in the fgf8 autoregulatory loop, possibly repressing expression of Fgf inhibitor genes. However, this model should be taken with caution. First, REREa morphant embryos, for unknown reasons, display a much stronger phenotype than bab mutants, yet based on the nature of the lesions the mutants are predicted to be rerea nulls. For this reason, we limit our analyses in this study to bab mutants. Second, we find that fgf8 and fgf3 expression is both up- and down-regulated in bab mutants, depending on the tissue, while sef and mkp3 remain relatively unchanged (Supplementary Figure S2, and unpublished data). Third, HDAC inhibition during gastrulation and early somitogenesis leads to a loss of fgf8, fgf3, mkp3, and sef expression in the tail bud and telencephalon, a loss of fgf8 and mkp3 in the developing somites, and a loss of sef in the MHB (Fig. 9), in contrast to the reported up-regulation of these genes in REREa morphants (Asai et al., 2006). Together, these results strongly suggest that the transcriptional effect of REREa/HDAC on these FGF pathway genes is not direct. Still, we cannot rule out that these seemingly contradictory results are due to timing differences in the response to a “runaway” Fgf regulatory loop. For instance, if disruption of REREa leads to a perturbation of the fgf8 autoregulatory loop, one might expect a transient up-regulation of fgf8 and downstream targets before the pathway shuts down and transcript levels drop by the negative regulation of Sef and Mkp3. Thus, increased levels of expression of Fgf targets may not yet have been downregulated by Sef/Mkp3, while areas of

decreased expression may have already shut down. Arguing against this model, differences in expression of fgfs and fgf target genes upon TSA treatment were first observed only after 2 hr of treatment, and during TSA timecourse studies, up-regulation of these markers was never observed (data not shown). The presence of additional tissue-specific Fgf regulators could also explain some of the differences. Additionally, the functions of REREa as a transcriptional repressor may extend well beyond the regulation of Fgf signaling to include other growth factor signaling pathways. Zoltewicz et al. (2004) have shown that a disruption of RERE in the mouse mutant openmind leads to a disruption of Shh signaling, and our preliminary studies in the fin buds and elsewhere in bab mutant zebrafish support this idea (data not shown). Investigation into the area of REREa-dependent growth factor regulation will undoubtedly lead to our better understanding of the functions of the Atrophin family both during development and in adult tissues, and the mechanics of specificity in transcriptional repression downstream from growth factor signaling.

EXPERIMENTAL PROCEDURES Maintenance of Fish Lines and Identification of Mutants Zebrafish were maintained and embryos treated using standard techniques (Westerfield, 1994). For genetic mapping, babtb210 was outcrossed to the WIK line, and mutants were identified live at 5 dpf by their reduced or absent pectoral fins, enlarged ear capsules, and inverted ceratohyal cartilage in the second pharyngeal arch (Schilling et al., 1996). All phenotypic analysis was performed on the babtb210 allele. babtb210 embryos younger than 24 hpf were genotyped by sequencing the genomic PCR product from the primer set rerea_mut (Supplementary Table S1, which can be viewed at http:// www.interscience.wiley.com/jpages/ 1058-8388/suppmat) that spans the mutation. aceti282 embryos were genotyped as previously described

(Mathieu et al., 2004). ika mutants were genotyped by PCR amplification of primer set ika_mut followed by an AccI restriction digest that cuts the mutant allele. fgf3t24149 embryos were genotyped as previously described (Herzog et al., 2004).

Tissue Labeling Procedures and Microscopy Cartilage was visualized by Alcian blue staining and prepared as previously described (Javidan and Schilling, 2004). Whole-mount in situ hybridization was carried out as previously described (Thisse et al., 1993). Antisense probe for rerea was prepared from EcoRI digested fv03c06 plasmid (GenBank accession no. BM071136), transcribed with SP6 RNA polymerase. The following riboprobes were used as previously described: bapx1 (Miller et al., 2003), col2a1 (Yan et al., 1995), dlx2 (Akimenko et al., 1994), emx1 (Kawahara and Dawid, 2002), fgf3 (Herzog et al., 2004), fgf8 (Reifers et al., 1998), foxa2 (Strahle et al., 1993), mkp3 (Kawakami et al., 2003), nkx2.3 (Lee et al., 1996), papc (Yamamoto et al., 1998), pax2.1 (Krauss et al., 1991), ptc1 (Concordet et al., 1996), and sef (Furthauer et al., 2002). Whole-mount immunostainings were performed with anti-diphosphorylated ERK1/2 (Sigma) antibody using the Vectastain Elite ABC kit (Vector Laboratories) as described previously (Schulte-Merker et al., 1992). For sectioning, larvae were embedded in JB-4 plastic as recommended by the manufacturer (Polysciences, Inc.) and cut in 3- to 5-␮m sections. Retinal and chondrocyte morphology was visualized with methylene blue–azure II stained JB-4 sections as previously described (Malicki et al., 1996).

Mapping of the bab Locus and Cloning of Atrophin-2 The babtb210 mutation was mapped to linkage group 23 by PCR analysis of simple sequence length polymorphism (SSLP) markers (Tuebingen version 4) on genomic DNA pools of 75 wild-type siblings and 25 mutant larvae (Geisler, 2002). Genomic sequence was identified in the bab locus by sequence similarity searches

REREA/ATROPHIN-2 IN ZEBRAFISH DEVELOPMENT 1903

of markers near z15422 and z42693 (http://134.174.23.167/zonrhmapper/ Maps.htm) against the Ensembl Zv5 genome database (http://www. ensembl.org). SSLP markers NP1, NP2, NP3, and NP4 (see Supplementary Table S1) were designed from this genomic sequence for fine mapping of single mutant larvae. All PCR reactions were performed with a 55°C annealing temperature. To identify the putative complete coding sequence of zebrafish REREa, sequence similarity searches (tblastn) were performed with mouse RERE (accession no. Q80TZ9) or human RERE (accession no. Q9P2R6) peptide sequences against the zebrafish genomic contig Zv5_scaffold1160 using two sequence BLAST (http://www.ncbi.nlm.nih.gov/ blast/bl2seq/wblast2.cgi). cDNA of mutant larvae and wild-type siblings was synthesized, and RT-PCR was performed as described previously (Plaster et al., 2006) using primer sets rereaA, rereaB, rereaC (Fig. 1 and Supplementary Table S1). RERE protein similarity searches were performed with Fasta3 (http://www.ebi. ac.uk/fasta33/) into the UniProt database.

Trichostatin A (TSA) Assay TSA (Sigma) was dissolved to a stock concentration of 1 mM in dimethylsulfoxide (DMSO) and diluted in embryo medium for final working concentrations of between 100 and 500 nM. Embryos were treated in 500 ␮l of diluted TSA or DMSO (control) in a 24-well plate (15–20 per well) at appropriate times. Embryos were dechorionated for treatments starting later than 24 hpf.

ACKNOWLEDGMENTS We thank Michael Brand (fgf8, pax2.1), Igor Dawid (sef, mkp3), Eddy De Robertis (papc), Marc Ekker (dlx2), Corinne Houart (emx1), Philip Ingham (ptc1), Charles Kimmel (bapx1), Jo¨rg Odenthal (foxa2), Yilin Yan (col2a1), and Quiling Xu (mkp3) for reagents; and Hans Georg Frohnho¨fer from the Tu¨bingen stock center for bab, Carl Neumann for ika, Tatjana Piotrowski for lia, and Michael Brandt for ace mutant fish. N.P. is grateful to the Alexander von Hum-

boldt Foundation and to EMBO for fellowships during her time in M.H.’s laboratory, and to the Epilepsy Research Training Program while in T.F.S.’s laboratory. Work in M.H.’s laboratory was supported by the MaxPlanck Society and the Human Frontier Science Program, and work in T.F.S.’s laboratory by grants from the NIH and March of Dimes.

REFERENCES Akimenko MA, Ekker M, Wegner J, Lin W, Westerfield M. 1994. Combinatorial expression of three zebrafish genes related to distal-less: part of a homeobox gene code for the head. J Neurosci 14:3475– 3486. Amaya E, Musci TJ, Kirschner MW. 1991. Expression of a dominant negative mutant of the FGF receptor disrupts mesoderm formation in Xenopus embryos. Cell 66:257–270. Asai Y, Chan DK, Starr CJ, Kappler JA, Kollmar R, Hudspeth AJ. 2006. Mutation of the atrophin2 gene in the zebrafish disrupts signaling by fibroblast growth factor during development of the inner ear. Proc Natl Acad Sci U S A 103:9069 –9074. Chang YL, Peng YH, Pan IC, Sun DS, King B, Huang DH. 2001. Essential role of Drosophila Hdac1 in homeotic gene silencing. Proc Natl Acad Sci U S A 98: 9730 –9735. Collas P, Liang MR, Vincent M, Alestrom P. 1999. Active transgenes in zebrafish are enriched in acetylated histone H4 and dynamically associate with RNA Pol II and splicing complexes. J Cell Sci 112(Pt 7):1045–1054. Concordet JP, Lewis KE, Moore JW, Goodrich LV, Johnson RL, Scott MP, Ingham PW. 1996. Spatial regulation of a zebrafish patched homologue reflects the roles of sonic hedgehog and protein kinase A in neural tube and somite patterning. Development 122:2835–2846. Crump JG, Maves L, Lawson ND, Weinstein BM, Kimmel CB. 2004. An essential role for Fgfs in endodermal pouch formation influences later craniofacial skeletal patterning. Development 131: 5703–5716. Cunliffe VT. 2004. Histone deacetylase 1 is required to repress Notch target gene expression during zebrafish neurogenesis and to maintain the production of motoneurones in response to hedgehog signalling. Development 131:2983–2995. Cunliffe VT, Casaccia-Bonnefil P. 2006. Histone deacetylase 1 is essential for oligodendrocyte specification in the zebrafish CNS. Mech Dev 123:24 –30. Dailey L, Abrosett D, Mansukhani A, Basilico C. 2005. Mechanisms underlying differential responses to FGF signaling. Cytokine Growth Factor Rev 16:233– 247. David NB, Saint-Etienne L, Tsang M, Schilling TF, Rosa FM. 2002. Require-

ment for endoderm and FGF3 in ventral head skeleton formation. Development 129:4457– 4468. Draper BW, Stock DW, Kimmel CB. 2003. Zebrafish fgf24 functions with fgf8 to promote posterior mesodermal development. Development 130:4639 –4654. Fanto M, Clayton L, Meredith J, Hardiman K, Charroux B, Kerridge S, McNeill H. 2003. The tumor-suppressor and cell adhesion molecule fat controls planar polarity via physical interactions with Atrophin, a transcriptional co-repressor. Development 130:763–774. Furthauer M, Reifers F, Brand M, Thisse B, Thisse C. 2001. sprouty4 acts in vivo as a feedback-induced antagonist of FGF signaling in zebrafish. Development 128: 2175–2186. Furthauer M, Lin W, Ang SL, Thisse B, Thisse C. 2002. Sef is a feedback-induced antagonist of Ras/MAPK-mediated FGF signalling. Nat Cell Biol 4:170 –174. Furumatsu T, Tsuda M, Yoshida K, Taniguchi N, Ito T, Hashimoto M, Asahara H. 2005. Sox9 and p300 cooperatively regulate chromatin-mediated transcription. J Biol Chem 280:35203–35208. Geisler R. 2002. Mapping and cloning. In: Nuesslein-Volhard CDR, editor. Zebrafish: a practical approach. Oxford: Oxford University Press. p 175–212. Griffin K, Patient R, Holder N. 1995. Analysis of FGF function in normal and no tail zebrafish embryos reveals separate mechanisms for formation of the trunk and the tail. Development 121:2983– 2994. Heisenberg CP, Brennan C, Wilson SW. 1999. Zebrafish aussicht mutant embryos exhibit widespread overexpression of ace (fgf8) and coincident defects in CNS development. Development 126: 2129 –2140. Herzog W, Sonntag C, von der Hardt S, Roehl HH, Varga ZM, Hammerschmidt M. 2004. Fgf3 signaling from the ventral diencephalon is required for early specification and subsequent survival of the zebrafish adenohypophysis. Development 131:3681–3692. Javidan Y, Schilling TF. 2004. Development of cartilage and bone. Methods Cell Biol 76:415–436. Kawahara A, Dawid IB. 2002. Developmental expression of zebrafish emx1 during early embryogenesis. Gene Expr Patterns 2:201–206. Kawakami Y, Rodriguez-Leon J, Koth CM, Buscher D, Itoh T, Raya A, Ng JK, Esteban CR, Takahashi S, Henrique D, Schwarz MF, Asahara H, Izpisua Belmonte JC. 2003. MKP3 mediates the cellular response to FGF8 signalling in the vertebrate limb. Nat Cell Biol 5:513–519. Kimelman D, Kirschner M. 1987. Synergistic induction of mesoderm by FGF and TGF-beta and the identification of an mRNA coding for FGF in the early Xenopus embryo. Cell 51:869 –877. Koide R, Ikeuchi T, Onodera O, Tanaka H, Igarashi S, Endo K, Takahashi H, Kondo R, Ishikawa A, Hayashi T. 1994. Unstable expansion of CAG repeat in heredi-

1904 PLASTER ET AL.

tary dentatorubral-pallidoluysian atrophy (DRPLA). Nat Genet 6:9 –13. Krauss S, Johansen T, Korzh V, Fjose A. 1991. Expression of the zebrafish paired box gene pax[zf-b] during early neurogenesis. Development 113:1193–1206. Lee KH, Xu Q, Breitbart RE. 1996. A new tinman-related gene, nkx2.7, anticipates the expression of nkx2.5 and nkx2.3 in zebrafish heart and pharyngeal endoderm. Dev Biol 180:722–731. Malicki J, Neuhauss SC, Schier AF, Solnica-Krezel L, Stemple DL, Stainier DY, Abdelilah S, Zwartkruis F, Rangini Z, Driever W. 1996. Mutations affecting development of the zebrafish retina. Development 123:263–273. Martinez-Morales JR, Del Bene F, Nica G, Hammerschmidt M, Bovolenta P, Wittbrodt J. 2005. Differentiation of the vertebrate retina is coordinated by an FGF signaling center. Dev Cell 8:565–574. Mathieu J, Griffin K, Herbomel P, Dickmeis T, Strahle U, Kimelman D, Rosa FM, Peyrieras N. 2004. Nodal and Fgf pathways interact through a positive regulatory loop and synergize to maintain mesodermal cell populations. Development 131:629 –641. McKinsey TA, Zhang CL, Olson EN. 2002. Signaling chromatin to make muscle. Curr Opin Cell Biol 14:763–772. Miller CT, Yelon D, Stainier DY, Kimmel CB. 2003. Two endothelin 1 effectors, hand2 and bapx1, pattern ventral pharyngeal cartilage and the jaw joint. Development 130:1353–1365. Miller CT, Maves L, Kimmel CB. 2004. moz regulates Hox expression and pharyngeal segmental identity in zebrafish. Development 131:2443–2461. Nagafuchi S, Yanagisawa H, Sato K, Shirayama T, Ohsaki E, Bundo M, Takeda T, Tadokoro K, Kondo I, Murayama N. 1994. Dentatorubral and pallidoluysian atrophy expansion of an unstable CAG trinucleotide on chromosome 12p. Nat Genet 6:14 –18. Pillai R, Coverdale LE, Dubey G, Martin CC. 2004. Histone deacetylase 1 (HDAC-1) required for the normal formation of craniofacial cartilage and pectoral fins of the zebrafish. Dev Dyn 231: 647–654. Plaster N, Sonntag C, Busse CE, Hammerschmidt M. 2006. p53 deficiency rescues apoptosis and differentiation of multiple cell types in zebrafish flathead mutants deficient for zygotic DNA polymerase delta1. Cell Death Differ 13:223–235. Reifers F, Bohli H, Walsh EC, Crossley PH, Stainier DY, Brand M. 1998. Fgf8 is mu-

tated in zebrafish acerebellar (ace) mutants and is required for maintenance of midbrain-hindbrain boundary development and somitogenesis. Development 125:2381–2395. Schilling TF, Piotrowski T, Grandel H, Brand M, Heisenberg CP, Jiang YJ, Beuchle D, Hammerschmidt M, Kane DA, Mullins MC, van Eeden FJ, Kelsh RN, Furutani-Seiki M, Granato M, Haffter P, Odenthal J, Warga RM, Trowe T, Nusslein-Volhard C. 1996. Jaw and branchial arch mutants in zebrafish I: branchial arches. Development 123:329 – 344. Schulte-Merker S, Ho RK, Herrmann BG, Nusslein-Volhard C. 1992. The protein product of the zebrafish homologue of the mouse T gene is expressed in nuclei of the germ ring and the notochord of the early embryo. Development 116:1021– 1032. Shanmugalingam S, Houart C, Picker A, Reifers F, Macdonald R, Barth A, Griffin K, Brand M, Wilson SW. 2000. Ace/Fgf8 is required for forebrain commissure formation and patterning of the telencephalon. Development 127:2549 –2561. Shigetani Y, Nobusada Y, Kuratani S. 2000. Ectodermally derived FGF8 defines the maxillomandibular region in the early chick embryo: epithelial-mesenchymal interactions in the specification of the craniofacial ectomesenchyme. Dev Biol 228:73–85. Shinya M, Koshida S, Sawada A, Kuroiwa A, Takeda H. 2001. Fgf signalling through MAPK cascade is required for development of the subpallial telencephalon in zebrafish embryos. Development 128:4153–4164. Strahle U, Blader P, Henrique D, Ingham PW. 1993. Axial, a zebrafish gene expressed along the developing body axis, shows altered expression in cyclops mutant embryos. Genes Dev 7:1436 –1446. Thisse C, Thisse B, Schilling TF, Postlethwait JH. 1993. Structure of the zebrafish snail1 gene and its expression in wildtype, spadetail and no tail mutant embryos. Development 119:1203–1215. Tsang M, Dawid IB. 2004. Promotion and attenuation of FGF signaling through the Ras-MAPK pathway. Sci STKE 2004: pe17. Tucker AS, Yamada G, Grigoriou M, Pachnis V, Sharpe PT. 1999. Fgf-8 determines rostral-caudal polarity in the first branchial arch. Development 126:51–61. Waerner T, Gardellin P, Pfizenmaier K, Weith A, Kraut N. 2001. Human RERE is localized to nuclear promyelocytic leu-

kemia oncogenic domains and enhances apoptosis. Cell Growth Differ 12:201– 210. Walshe J, Mason I. 2003. Fgf signaling is required for formation of cartilage in the head. Dev Biol 264:522–536. Wang L, Rajan H, Pitman JL, McKeown M, Tsai CC. 2006. Histone deacetylase-associating Atrophin proteins are nuclear receptor corepressors. Genes Dev 20:525– 530. Weinstein DC, Marden J, Carnevali F, Hemmati-Brivanlou A. 1998. FGF-mediated mesoderm induction involves the Src-family kinase Laloo. Nature 394:904 – 908. Westerfield M. 1994. The zebrafish book: a guide for the laboratory use of zebrafish. Eugene: University of Oregon Press. Wilson J, Tucker AS. 2004. Fgf and Bmp signals repress the expression of Bapx1 in the mandibular mesenchyme and control the position of the developing jaw joint. Dev Biol 266:138 –150. Xu RH, Peng Y, Fan J, Yan D, Yamagoe S, Princler G, Sredni D, Ozato K, Kung HF. 2000. Histone acetylation is a checkpoint in FGF-stimulated mesoderm induction. Dev Dyn 218:628 –635. Yamaguchi M, Tonou-Fujimori N, Komori A, Maeda R, Nojima Y, Li H, Okamoto H, Masai I. 2005. Histone deacetylase 1 regulates retinal neurogenesis in zebrafish by suppressing Wnt and Notch signaling pathways. Development 132:3027–3043. Yamamoto A, Amacher SL, Kim SH, Geissert D, Kimmel CB, De Robertis EM. 1998. Zebrafish paraxial protocadherin is a downstream target of spadetail involved in morphogenesis of gastrula mesoderm. Development 125:3389 –3397. Yan YL, Hatta K, Riggleman B, Postlethwait JH. 1995. Expression of a type II collagen gene in the zebrafish embryonic axis. Dev Dyn 203:363–376. Yanagisawa H, Bundo M, Miyashita T, Okamura-Oho Y, Tadokoro K, Tokunaga K, Yamada M. 2000. Protein binding of a DRPLA family through arginine-glutamic acid dipeptide repeats is enhanced by extended polyglutamine. Hum Mol Genet 9:1433–1442. Zhang S, Xu L, Lee J, Xu T. 2002. Drosophila atrophin homolog functions as a transcriptional corepressor in multiple developmental processes. Cell 108:45–56. Zoltewicz JS, Stewart NJ, Leung R, Peterson AS. 2004. Atrophin 2 recruits histone deacetylase and is required for the function of multiple signaling centers during mouse embryogenesis. Development 131: 3–14.