ARTICLE

Zebrafish Tshz3b Negatively Regulates HoxFunction in the Developing Hindbrain

Timothy Erickson,1 Laura M. Pillay,1 and Andrew J. Waskiewicz1,2,3*

1Department of Biological Sciences, University of Alberta, Edmonton, Canada

2Centre for Neuroscience, University of Alberta, Edmonton, Canada

3Women and Children’s Health Research Institute, University of Alberta, Edmonton, Canada

Received 15 August 2010; Revised 13 June 2011; Accepted 19 June 2011

Summary: In flies, the zinc-finger protein Teashirt pro-motes trunk segmental identities, in part, by repressingthe expression and function of anterior hox paraloggroup (PG) 1–4 genes that specify head fates. Anterior-posterior patterning of the vertebrate hindbrain alsorequires Hox PG 1–4 function, but the role of vertebrateteashirt-related genes in this process has not beeninvestigated. In this work, we use overexpression andstructure-function analyses to show that zebrafishtshz3b antagonizes Hox-dependent hindbrain segmen-tation. Ectopic Tshz3b perturbs the specification ofrhombomere identities and leads to the caudal expan-sion of r1, the only rhombomere whose identity is speci-fied independently of Hox function. This overexpressionphenotype does not require the homeodomain andC-terminal zinc fingers that are unique to vertebrateTeashirt-related proteins, but does require that Tshz3bfunction as a repressor. Together, these results arguethat the negative regulation of Hox PG 1–4 function is aconserved characteristic of Teashirt-related proteins.genesis 49:725–742, 2011. VVC 2011 Wiley-Liss, Inc.

Key words: hindbrain; hox; patterning; pbx; rhombomere;segmentation; teashirt; transcription factor

INTRODUCTION

During vertebrate brain development, the hindbrain istransiently segmented into lineage-restricted compart-ments called rhombomeres (Fraser et al., 1990; Lums-den and Keynes, 1989; Moens and Prince, 2002; vonBaer, 1828). The specification of each rhombomere’sidentity is homologous to that of Drosophila anterior-posterior embryonic segmentation (Pearson et al.,

2005), with both developmental processes requiringtranscriptional regulation by Hox proteins complexedwith their TALE-class homeodomain cofactors Pbx (Dro-sophila Extradenticle) and Meinox (Meis/Pknox; Dro-sophila Homothorax) (Chan et al., 1994; Mann andChan, 1996; Moens and Selleri, 2006; van Dijk andMurre, 1994). The anteriorly-expressed Hox paraloggroups (PG) 1–4 exhibit very clear homeotic propertiesin segmental hindbrain patterning, as a gain or loss ofHox function can lead to transformations of rhombo-mere identity (Bell et al., 1999; Gavalas et al., 1997;McClintock et al., 2001; Studer et al., 1996; Vlachakiset al., 2001; Zhang et al., 1994). Perturbations in Pbx orMeis function also result in profound hindbrain segmen-tation defects (Dibner et al., 2001; Popperl et al., 2000;Waskiewicz et al., 2001). In zebrafish embryos lackingboth Pbx2 and Pbx4 function, the hindbrain losesits segmental character and adopts the fate of the most

Additional Supporting Information may be found in the online version

of this article.

A.J.W. is a Canada Research Chair.Current address for Timothy Erickson: Howard Hughes Medical Insti-

tute, Oregon Hearing Research Center and Vollum Institute, Oregon Health

and Science University, 3181 SW Sam Jackson Park Road, Portland, OR

97239, USA.

* Correspondence to: Andrew J. Waskiewicz, CW405 Biological Sciences

Bldg., University of Alberta, Edmonton, Canada T6G 2E9.

E-mail: aw@ualberta.ca

Contract grant sponsor: NSERC, Contract grant number: 298371-2010

(to T.E., L.M.P., A.J.W.), Contract grant sponsor: Alberta Ingenuity Fund (to

T.E., L.M.P.), Contract grant sponsor: Foundation Fighting Blindness (to

A.J.W.), Contract grant sponsor: Capital Health (to A.J.W.)

Published online 28 June 2011 in

Wiley Online Library (wileyonlinelibrary.com).

DOI: 10.1002/dvg.20781

' 2011 Wiley-Liss, Inc. genesis 49:725–742 (2011)

anterior rhombomere r1, whose specification does notdepend on Hox function (Waskiewicz et al., 2002). Sim-ilar results are achieved through a triple knockdown ofHox1 paralogues in Xenopus (McNulty et al., 2005).These data suggest that Hox, Pbx, and Meis proteins actas homeotic factors during hindbrain segmentation andthat, in the absence of Hox function, the entire hind-brain adopts a Hox-independent r1-like fate.

In Drosophila, the zinc-finger protein Teashirt hasalso been identified as a homeotic transcription factor.Loss-of-function analyses reveal that Tsh can promotetrunk identity and repress head characteristics throughboth Hox-dependent and Hox-independent mecha-nisms (Fasano et al., 1991; Roder et al., 1992). Comple-menting the loss of function experiments, overexpres-sion studies of hox genes Antp, Ubx, and Abd in a tsh2/2

background reveal less severe head-to-trunk transforma-tions than when Tsh is functional (Alexandre et al.,1996; Andrew et al., 1994; Coiffier et al., 2008).Conversely, tsh overexpression before embryonic stage11 results in a partial head-to-trunk homeotic transfor-mation (de Zulueta et al., 1994). Ectopic Tsh is lesseffective at driving a head-to-trunk transformation inembryos lacking Scr, Antp, and BX-C function, althoughit can partially rescue the trunk-to-head transformationobserved in posterior Hox compound-mutant embryos(de Zulueta et al., 1994). Taken together, these studiessuggest that Tsh defines a trunk ground state uponwhich posterior Hox proteins can act to specify theidentity of the thoracic segments, and that Tsh can actindependently of Hox proteins to promote trunk andrepress head identity.

One of the ways in which Tsh promotes trunk iden-tity is by repressing anterior hox gene expression in thetrunk. tsh mutants express transcripts for the anteriorHox 1, 2, and 4 paralogues labial (lab), proboscipedia(pb), and deformed (dfd) in ectopic posterior positions(Roder et al., 1992; Rusch and Kaufman, 2000). Tsh canalso regulate the activity of anterior Hox proteins inde-pendently of their transcription. For example, thetrunk-to-head transformation induced by ectopicexpression of Dfd is much more effective in the absenceof Tsh function (Robertson et al., 2004). These dataindicate that Tsh can antagonize anterior hox genes byrepressing both their transcription and protein func-tion. Overall, Tsh participates in a complex transcrip-tion factor network that establishes functional domainsfor Hox proteins, thereby regionalizing the fly embryoalong the anterior-posterior axis.

Vertebrate teashirt-related genes are co-expressedwith hox, pbx, and meis genes throughout the develop-ing nervous system (Koebernick et al., 2006; Santoset al., 2010; Wang et al., 2007), raising the possibilitythat vertebrate Hox function is subject to regulation byTeashirt proteins as well. The ability of mouse Tsh

orthologues to substitute for the endogenous fly tsh

gene (Manfroid et al., 2004), and studies in Xenopus

and mice showing that vertebrate teashirt-related genesperform important functions in hindbrain development(Caubit et al., 2010; Koebernick et al., 2006) supportthis idea. However, the possibility that vertebrate tea-

shirt genes might regulate Hox function during hind-brain development has not been directly investigated.

In this work, we describe a role for zebrafish teashirt

zinc finger homeobox 3b (tshz3b) in hindbrain devel-opment. tshz3b is expressed in rhombomeres 4–7, andis regulated by a combination of Hox-Pbx-Meis tran-scriptional input and retinoic acid signaling. Overex-pression of tshz3b causes a loss of hindbrain segmenta-tion similar to that observed in Hox or Pbx-depletedembryos. Consistent with this loss of segmentation,tshz3b overexpression perturbs the specification of thehindbrain cranial motor neurons and reticulospinaltract. This effect of tshz3b overexpression is likelyachieved by antagonizing Hox function, as tshz3b over-expression synergizes with the pbx4

2/2 phenotype,and can block the r2-to-r4 homeotic transformationcaused by hoxb1b overexpression. Lastly, structure-function assays show that Tshz3b likely functions as atranscriptional repressor, and that the overexpressionphenotype does not require the vertebrate-specific C-terminal zinc fingers or homeodomain. However, incontrast to the ability of mouse tshz genes to function-ally rescue Drosophila tsh mutants, overexpression ofDrosophila tsh in zebrafish does not produce the samehindbrain patterning defects as tshz3b overexpression.In summary, this work suggests that tshz3b contributesto hindbrain patterning by modulating Hox function.Furthermore, it supports the idea that the regulation ofHox function is an evolutionarily conserved function ofTeashirt proteins, although some of the mechanismsthrough which this occurs may differ between flies andvertebrates.

RESULTS

Embryonic Expression of Four Zebrafishteashirt-Related Genes: tshz1, tshz2,tshz3a, and tshz3b

To find out which tshz genes are expressed in thehindbrain during segmentation, we performed mRNA insitu hybridizations for all four zebrafish tsh-relatedgenes. tshz3b expression is first detectable by in situhybridization between 9 and 11 h postfertilization (hpf)when it is enriched in the presumptive forebrain, in thehindbrain posterior to r3, and in the trunk midline(Fig. 1a,l). Over the first 2 days of development, this pat-tern becomes refined where the early broad forebrainand hindbrain expression domains resolve to specificcells in those regions (Fig. 1b–h,m,n,p). Additionally,tshz3b is expressed in the developing pectoral fins

726 ERICKSON ET AL.

FIG. 1. Zebrafish teashirt-related gene expression. (a–h)mRNA in situ hybridizations (ISH) for tshz3b from 11 to 48 hpf showing expressionin the forebrain, midbrain, hindbrain, and anterior spinal cord. (i–k) tshz3b ISH in the developing pectoral fins (black arrows) from 24 to 48hpf. (l–n) mRNA ISH for tshz3b (purple) costained with egr2b (red) in r3 (l) and r3 and r5 (m, n) detailing tshz3b expression in the hindbrain.(o)Whole mount immunostain for Tshz3b showing protein accumulation in the hindbrain at 20 hpf. (p)mRNA ISH showing tshz3b expressionin the forebrain at 20 hpf. (q) Whole mount immunostain for Tshz3b showing protein accumulation in the forebrain at 20 hpf. (r–t) mRNA ISHfor tshz1 (purple) showing expression in the posterior hindbrain/spinal cord and forebrain between 16 and 26 hpf. (u–w) mRNA ISH for tshz2showing weak, diffuse expression between 14 and 24 hpf, and enriched hindbrain expression at 50 hpf. (x–z) mRNA ISH for tshz3a (purple)between 12 and 20 hpf showing enriched expression in the forebrain, midbrain, and hindbrain. All views are dorsal with anterior to the left,except for (i–k; anterior at the top), and (h, m, u-w; lateral). Embryos in l–n, r, and y are costained with the r3/r5 marker egr2b (red). ISH, insitu hybridization; hpf, hours postfertilization; ov, otic vesicle; r, rhombomere.

(arrows in Fig. 1i–k). We also raised a peptide polyclo-nal antibody against Tshz3b, and in whole mount immu-nostains on 20 hpf embryos, it recognizes a nuclear-enriched protein in a pattern that is consistent withtshz3b mRNA expression at that stage (Fig. 1o,q). Takentogether, tshz3b is expressed up to the r3/4 boundaryduring the initial stages of hindbrain segmentation, con-sistent with what has been previously found in zebra-fish, mouse, chick, and frog (Caubit et al., 2010; Manf-roid et al., 2006; Onai et al., 2007; Santos et al., 2010).

The other zebrafish teashirt-related genes also exhibittissue-specific embryonic expression patterns. Between16 and 26 hpf, tshz1 is expressed in the eyes and fore-brain at low levels, and at high levels in the posteriorhindbrain and spinal cord up to the r6/7 boundary (Fig.1r–t) (Wang et al., 2007). tshz2 expression is barely de-tectable by in situ hybridization before 24 hpf, but by50 hpf it is broadly expressed in the head with enrichedexpression in the hindbrain and anterior spinal cord(Fig. 1u–w) (Santos et al., 2010). Lastly, tshz3a, theparalogue of tshz3b (Santos et al., 2010), also exhibits arhombomere-restricted expression pattern during hind-brain segmentation. At 12 hpf, tshz3a is expressed in r2and r4, and although it becomes expressed throughoutr2–6 by 20 hpf, its expression remains enriched in even-numbered rhombomeres (Fig. 1x–z). As a whole, thezebrafish tshz genes are expressed in unique, but partiallyoverlapping, regions of the neural tube, from the eyes andforebrain, to the midbrain, hindbrain, and spinal cord.

tshz3b Hindbrain Expression is Regulated byHox/TALE-Class Homeodomain TranscriptionFactors and Retinoic Acid Signaling

In Drosophila, tsh expression is directly activated byHox proteins and their TALE-class co-factors Exd (verte-brate Pbx) and Hth (vertebrate Meis/Pknox) (Mathieset al., 1994; McCormick et al., 1995; Merabet et al.,2007; Rauskolb and Wieschaus, 1994; Roder et al.,1992). In the zebrafish hindbrain at 10.5 hpf, tshz3b isco-expressed with hox genes, including hoxb1a,

hoxa2, and hoxb2, suggesting that tshz3b may be regu-lated in a Hox-dependent fashion. To test this, we exam-ined tshz3b expression in embryos lacking the Hoxcofactors Pbx4 and Meis1. In 20 hpf pbx4/lazarusmutants, tshz3b expression is decreased in the mid-brain and hindbrain (Fig. 2a,b). Similarly, the hindbrainexpression of tshz3b is downregulated in 16 hpf meis1

morphants (Fig. 2c,d). These data suggest that tshz3b isregulated in a Hox-dependent manner. To test this moredirectly, we knocked down Hoxb1 function using acombination of hoxb1a and hoxb1b translation-block-ing morpholinos (McClintock et al., 2002). In 10.5 hpfhoxb1 morphants, tshz3b expression is not initiatedcorrectly (n 5 8/8; Fig. 2e,f). At 20 hpf, hoxb1 mor-phants lack tshz3b expression specifically in the r4

region, but expression in r5–7 and the spinal cord isnormal (n5 18/18; Fig. 2g,h). To see if Hoxb1 can drivetshz3b expression, we injected one-cell zebrafishembryos with mRNA coding for Hoxb1b fused to an N-terminal HA tag (Vlachakis et al., 2000). Overexpressedhoxb1b can transform r2 to an r4-like identity, as shownby the ectopic expression of hoxb1a in the r2 region(compare the red stain in Fig. 2i,j) (McClintock et al.,2001). Consistent with the results of the Hoxb1 knock-down experiment, hoxb1b overexpression drives ec-topic tshz3b expression in the r4-like region (n 5 38/38; blue stain in Fig. 2j). These data suggest that hoxb1genes play a critical role in regulating the r4 domain oftshz3b expression.

Retinoic acid (RA) is a signaling molecule that isessential for anterior-posterior hindbrain patterning.Pharmacological inhibition of RA synthesis causes a lossof r5 and r6 hindbrain identity, while exogenous RA cantransform the entire anterior neural tube into posteriorhindbrain tissue (Dupe and Lumsden, 2001; Gavalas,2002; Gavalas and Krumlauf, 2000; Hernandez et al.,2007; Maves and Kimmel, 2005). To see if RA contrib-utes to tshz3b expression, we treated embryos with 10and 100 nM solutions of RA from 3 hpf until 15 hpf, andassayed for tshz3b expression by in situ hybridization.Embryos treated with 10 nM RA show an increase intshz3b expression, especially in the dorsal neural tube(n 5 39/39; Fig. 2k,l). Embryos treated with 100 nM RAexhibit severe morphological defects as well as a dra-matic upregulation of tshz3b expression (n 5 40/40;Fig. 2m). To determine what effect the inhibition of RAsignaling has on tshz3b expression, we treated embryoswith 10 lM diethylaminobenzaldehyde (DEAB), a com-petitive inhibitor of the aldehyde dehydrogenase (Aldh/Raldh) enzymes that synthesize retinoic acid. DEAB-treated embryos lack r5 and r6 identity, and r4 isexpanded posteriorly (Fig. 2n,o) (Maves and Kimmel,2005). Likewise, the r4 domain of tshz3b is expandedposteriorly in DEAB-treated embryos (n 5 31/31; Fig.2o). Taken together, tshz3b expression is positivelyregulated by RA signaling. Given that tshz3b is stillexpressed in r4 and the spinal cord following DEAB-treatment, it is likely that the r5/6 domain of tshz3b

expression is specifically RA-responsive.

tshz3b Overexpression Produces HoxLoss-of-Function Hindbrain Patterning Defects

In flies, Tsh promotes trunk identity, in part, by antag-onizing the expression and function of anterior HoxPG1, 2, and 4 proteins that specify head segments (Rob-ertson et al., 2004; Roder et al., 1992; Rusch and Kauf-man, 2000). Moreover, Drosophila Tsh can repress an-terior hox expression in a posterior Hox-independentfashion, as tsh overexpression has no effect on posteriorhox gene expression (de Zulueta et al., 1994). To test

728 ERICKSON ET AL.

FIG. 2. Regulation of tshz3b hindbrain expression by Hox/Pbx/Meis proteins and retinoic acid signaling. (a, b) mRNA in situ hybridizations(ISH) for tshz3b in wild-type and pbx42/2 mutant embryos at 20 hpf. tshz3b is reduced in the midbrain and r4-7 of the hindbrain in pbx4mutants. (c, d) tshz3b ISH in wild-type and meis1 morphant embryos at 16 hpf. tshz3b expression is reduced in the hindbrain of Meis1-depleted embryos. Embryos in a–d are costained with the r3/r5 marker egr2b (red) and shown in dorsal view with anterior to the left.(e–h) tshz3b ISH in wild-type and hoxb1a/b1b (hoxb1) morphant embryos. (e) At 10.5 hpf, tshz3b is expressed in the presumptive r4 domain,and in the midline. This expression pattern is not initiated correctly in hoxb1 morphants (f; black arrow). At 20 hpf, the r4 domain of tshz3bexpression is lost in hoxb1 morphants (g, h; black arrow). Embryos in (e, f) are shown in dorsal view with anterior up; embryos in g and h areshown in dorsal view with anterior to the left. (i, j) mRNA ISH for tshz3b (purple) and hoxb1a (red) in wild-type and hoxb1b overexpressingembryos at 20 hpf. Ectopic Hoxb1b drives hoxb1a expression in the r2 region, where tshz3b is also ectopically expressed. Embryos are shownin dorsal view with anterior to the left. (k–m)mRNA ISH for tshz3b expression in 15 hpf embryos treated with DMSO, 10 nM retinoic acid (RA) or100 nM RA. Increasing concentrations of RA lead to increased tshz3b expression. (n, o)mRNA ISH for tshz3b (purple) and hoxb1a (red) expres-sion in 15 hpf embryos treated with DMSO or 10 lM DEAB (an inhibitor of Aldh1 enzymes that synthesize RA). DEAB-treatment leads to a cau-dal expansion of r4 identity, and a concomitant expansion of r4 tshz3b expression. Embryos in k–o are lateral views with anterior to the upperleft. DEAB, diethylaminobenzaldehyde; hpf, hours postfertilization; ISH, in situ hybridization; ov, otic vesicle; r, rhombomere; RA, retinoic acid.

729ECTOPIC tshz3b PERTURBS HINDBRAIN PATTERNING

the hypothesis that zebrafish tshz3b antagonizes verte-brate Hox PG1–4-dependent hindbrain patterning, weassayed for the expression of hindbrain marker genes at10 and 20 hpf in embryos injected with tshz3b mRNA.The a-Tshz3b polyclonal antibody is able to recognize asingle band in lysates made from 4 hpf tshz3b-injectedembryos that is not present in lysates made from equiva-lently staged uninjected embryos (Supporting Informa-tion, Fig. 1a), confirming the production of ectopicTshz3b protein. At 10 hpf, hoxb1a marks the presump-tive hindbrain up to r3, while myod is expressed in theposterior paraxial mesoderm. In tshz3b mRNA-injectedembryos, hoxb1a expression is downregulated whilemyod expression is undisturbed (n 5 9/10; Fig. 3a,b).Similar results are observed at 20 hpf, when tshz3b

overexpression results in downregulated hoxa2b

expression in r2–5 (n 5 3/6; Fig. 3c,d). These resultssuggest that ectopic Tshz3b function can perturb theestablishment of segmental identities in the developingzebrafish hindbrain.

To determine if this loss of hox expression results in aposterior expansion of r1 identity as in Pbx-depletedembryos (Waskiewicz et al., 2002), we also examinedthe expression of epha4a, which marks r1, r3, and r5 at20 hpf. Consistent with the hypothesis that tshz3b

antagonizes hox function, tshz3b overexpression leadsto an expansion of r1-like identity at the expense ofmore posterior rhombomere identities (Fig. 3e,f). A sim-ilar result is observed when tshz3b mRNA is injectedinto Fgf-responsive Tg(dusp6:eGFP) transgenic embryos(Molina et al., 2007). In uninjected embryos, the trans-gene is expressed at high levels in the cerebellum andr1, as well as in r4 and r6 (Fig. 3g). Overexpression oftshz3b downregulates Fgf-responsiveness in r4 and r6,and expands the r1-like domain caudally (n 5 49/134;Fig. 3h). Taken together, these results suggest thattshz3b overexpression perturbs hindbrain segmenta-tion in a fashion similar to that caused by a loss of Hoxfunction.

To see if the ability to disrupt hindbrain segmentationwas a property shared by other zebrafish tshz homo-logs, we injected one cell stage embryos with mRNAencoding either Myc-tagged or untagged Tshz1 andassayed for the hindbrain markers hoxa2b and epha4a

by in situ hybridization at 16 hpf. Nuclear-localizedMyc-Tshz1 protein was detected by 9E10 immunostainsagainst the Myc epitope (Supporting Information, Fig.2a,b). The expression of both hoxa2b (n 5 14/15) andepha4a (n 5 11/11) is normal in embryos injected withtshz1 mRNA (Supporting Information, Fig. 2c–f). Con-sistent with these findings, the specification and posi-tioning of the branchiomotor neurons are unaffected byoverexpressed tshz1 (n 5 30/31; Supporting Informa-tion, Fig. 2g,h). These data suggest that tshz1 andtshz3b possess different abilities to regulate anteriorHox-dependent hindbrain patterning.

tshz3b Overexpression Causes Mispatterning ofthe Branchiomotor and Reticulospinal Neurons

One of the outputs of Hox-dependent anterior-poste-rior hindbrain patterning is the correct positioning,identity and function of the branchiomotor and reticulo-spinal neurons (Chandrasekhar, 2004; Chandrasekharet al., 1997; Cooper et al., 2003; Ferretti et al., 2000;Gavalas et al., 1998; Guthrie, 2007; Kimmel et al., 1985;Metcalfe et al., 1986; Moens and Prince, 2002; Waskie-wicz et al., 2001, 2002). To determine if ectopic Tshz3baffects the development of the hindbrain cranial nervenuclei, we overexpressed tshz3b in Tg(isl1:GFP)embryos, which express GFP in the motor nuclei of theV, VII, and X hindbrain cranial nerves under the controlof the isl1 promoter (Higashijima et al., 2000). Com-pared with wild-type Tg(isl1:GFP) embryos, tshz3b

mRNA-injected embryos have mispatterned trigeminal(V) and facial (VII) cranial motor neurons, (n 5 26/61;Fig. 4a,b). Conversely, the oculomotor (III), trochlear(IV), and vagal (X) nuclei are still present in tshz3b-injected embryos, suggesting that the effect of ectopicTshz3b is largely limited to the hindbrain. Additionally,as shown by zn5 antibody stain, tshz3b-injectedembryos exhibit disrupted commissural neuron devel-opment (white bracket in Fig. 4b) and fail to producevisible abducens (VI) motor neuron cell bodies (n 511/12; white brackets in Fig. 4c,d). Furthermore, theladder-like array of reticulospinal neurons is also dis-rupted by tshz3b overexpression (n 5 5/9; Fig. 4e,f).These phenotypes are consistent with the idea that ec-topic tshz3b antagonizes hox function and that the seg-mental patterning defects in tshz3b-overexpressingembryos also perturb the specification of neuronal iden-tity in the hindbrain.

The Effect of tshz3b Overexpressionon the Expression of tshz1 and tshz3a

In flies, teashirt is known to positively regulate itsown expression (de Zulueta et al., 1994). Conversely,Drosophila teashirt and its paralogue tiptop (tio) havebeen shown to negatively regulate one another’sexpression (Laugier et al., 2005). To see if overexpres-sion of zebrafish tshz3b modulates the expression ofother tshz genes, we compared the expression of tshz1and tshz3a between wild type and tshz3b mRNA-injected embryos at 18 hpf. Similar to hoxb4a andhoxd4a, tshz1 is strongly expressed in the spinal cordup to the r6/7 boundary (Wang et al., 2007), with addi-tional expression in the caudal forebrain and hindbrain(Fig. 5a). Ectopic Tshz3b leaves tshz1’s forebrain andspinal cord domains intact, although the r6/7 boundaryis somewhat disorganized (Fig. 5b). A diffuse r6/7boundary is also observed in Pbx-depleted embryos(Waskiewicz et al., 2002). This indicates that tshz3b

does not positively regulate tshz1 expression, and

730 ERICKSON ET AL.

FIG. 4. The hindbrain cranial and reticulospinal neurons are mispatterned in tshz3b overexpressing embryos. (a, b) Branchiomotor neurons(green) in Tg(isl1:GFP) embryos costained with the zn-5 antibody against Alcam a (red) at 48 hpf. Overexpressed tshz3b disrupts the specifi-cation of the trigeminal (V) and facial (VII) motor neurons. White brackets (top) and labels indicate the identities of the cranial motor neurons.Bottom white bracket in b indicates perturbed commissural neuron patterning in tshz3b-overexpressing embryos. (c, d) The abducensmotor nuclei of the VI cranial nerve, as visualized by zn-5 antibody stain. Embryos in c and d are the same as those shown in a and b, butwith the zn-5 stain shown in grayscale to highlight the abducens cell bodies (VI, white brackets in c). Zn-5-positive abducens neurons areabsent in tshz3b-overexpressing embryos (white brackets in d). The white arrowhead indicates incorrectly positioned zn-5-positive hind-brain tissue. (e, f) Reticulospinal neurons (RN) as detected by immunostaining with the rmo44 a-neurofilament-medium antibody. tshz3b-overexpressing embryos exhibit disorganized anterior RNs. White arrow indicates a missing Mauthner (Mth) cell in r4. All views are dorsalwith anterior to the left. hpf, hours postfertilization; r, rhombomere.

FIG. 3. tshz3b overexpression perturbs segmental patterning of the hindbrain. (a, b) hoxb1a (brackets) and myoD expression in 10 hpfwild-type and tshz3b-injected embryos. hoxb1a expression in the hindbrain is reduced while myoD expression is unaffected by over-expressed tshz3b. (c, d) hoxa2b expression in 20 hpf wild-type and tshz3b-injected embryos. hoxa2b expression in r2-5 is reduced by over-expressed tshz3b. (e, f) epha4a expression in 20 hpf wild-type and tshz3b-injected embryos. epha4a expression in r3 and r5 is reduced byoverexpressed tshz3b, and the r1 domain is expanded caudally. (g, h) Fgf-signaling domains in wild-type and tshz3b-injected Tg(dus-p6:eGFP) embryos. tshz3b overexpression perturbs the r4 and r6 Fgf-responsive domains, while expanding an r1-like level of Fgf-respon-siveness into the anterior hindbrain. All views are dorsal with anterior to the left, except in a and b where anterior is at the top. hpf, hourspostfertilization; ov, otic vesicle; r, rhombomere.

FIG. 6. (a–d) tshz3b overexpression synergizes with the pbx4 mutant hindbrain phenotype. mRNA in situ hybridization (ISH) for eng2a (MHB)and egr2b (r3 and r5) in wild-type (a), pbx42/2 (b), tshz3b-injected (c), and pbx42/2 embryos injected with tshz3b mRNA (d). The low dose oftshz3bmRNA produces little phenotype on its own, but, in a pbx4 mutant background, can further reduce r3 expression of egr2b. (e–h) tshz3bmRNA blocks the r2-to-r4 homeotic transformation caused by hoxb1b overexpression. mRNA ISH for hoxb1a in 19 hpf wild-type (e), hoxb1b-injected (f), tshz3b-injected (g), and hoxb1b 1 tshz3b-injected (h) embryos. Overexpressed hoxb1b is able to drive hoxb1a expression an ec-topic r4-like region (b), but is unable to do so when coexpressed with tshz3b (d). tshz3bmRNA alone reduces hoxb1a expression in r4 (c). hpf,hours postfertilization; ISH, in situ hybridization; MHB, midbrain-hindbrain boundary; ov, otic vesicle; r, rhombomere.

FIG. 5. The effect of overexpressed tshz3b on tshz1 and tshz3a expression at 18 hpf. (a, b) Although the faint hindbrain expression oftshz1 is perturbed in tshz3b-injected embryos, the forebrain and caudal hindbrain/spinal cord domains are unaffected by tshz3b mRNAoverexpression. (c, d) tshz3a expression in the midbrain and hindbrain is downregulated in tshz3b overexpressing embryos, and this pheno-type is similar to that observed in pbx2,4 morphant embryos (e, f). All views are dorsal with anterior to the left. hpf, hours postfertilization;MB, midbrain; ov, otic vesicle; r, rhombomere.

furthermore, it suggests that the loss of hindbrain seg-mental identity is not due to a rostral expansion of ante-rior spinal cord identity. With regard to tshz3a, itsexpression in the midbrain, r2, and r4 is downregulatedby tshz3b overexpression (Fig. 5c,d). This loss oftshz3a expression is also consistent with the tshz3b

overexpression phenotype being caused by a loss ofHox function, as pbx2,4 double morphants exhibit asimilar loss of tshz3a expression (Fig. 5e,f). This loss oftshz3a expression in Pbx-depleted embryos is similar tothat observed for tshz3b expression (Fig. 2a,b). In sum-mary, tshz3b is not a positive regulator of tshz1 ortshz3a expression, tshz3b overexpression does notcause a rostral expansion of spinal cord identity, andboth tshz3a and tshz3b are positively regulated by Pbx-Hox function.

tshz3b Overexpression SynergizesWith the pbx4 Mutant Phenotype

To further explore the hypothesis that tshz3b antago-nizes hox function, we looked for a synergistic interac-tion between tshz3b overexpression and a partial lossof Pbx function by injecting a suboptimal dose oftshz3b mRNA into the progeny of a pbx4

1/2 cross.Compared with wild-type embryos, pbx4 mutants havemarkedly reduced, but not eliminated, egr2b expres-sion in r3 (n 5 3/3, Fig. 6a,b). 200 pg of tshz3b RNAalone produces only a very mild reduction in rhombo-mere size (n 5 5/5; Fig. 6c). The combination ofectopic tshz3b and loss of pbx4 function causes asynergistic effect where r5 egr2b is reduced andthe r3 domain is eliminated (n 5 3/4; Fig. 6d). Theresults of this experiment support the idea thattshz3b is a negative regulator of hox-dependent hind-brain patterning.

tshz3b Overexpression Blocks the r2-to-r4Homeotic Transformation Caused byEctopic hoxb1b Function

As a final way to demonstrate that tshz3b can inhibithox function, we examined the interaction betweenoverexpressed hoxb1b and tshz3b. As demonstratedpreviously, overexpressed hoxb1b can transform r2 toan r4-like identity, as shown by hoxb1a expression (n 520/32; Fig. 6e,f), while overexpressed tshz3b reduceshoxb1a expression in r4 (n 5 23/24; Fig. 6g). Whenthe two mRNAs are injected simultaneously, ectopichoxb1a expression in r2 is never observed (n 5 47),and the r4 domain of hoxb1a is typically reduced (n 539/47; Fig. 6h). Thus, even when hoxb1b is ectopicallysupplied, tshz3b overexpression is able to block hox

function and perturb segmental patterning of the hind-brain. This lends further support to the idea that tshz3bnegatively regulates Hox function in the hindbrain.

The Tshz3b Homeodomain and C-Terminal ZincFingers are Dispensable for its OverexpressionPhenotype

To determine which domains of Tshz3b are requiredto antagonize hox function in vivo, we created a seriesof tshz3b deletion constructs. The three N-terminal zincfingers represent the main region of sequence homol-ogy between fly and vertebrate Tsh proteins (Caubitet al., 2000). To test whether these zinc fingers arerequired for the tshz3b overexpression phenotype, wecreated a construct that omits the first three zinc fin-gers, but includes a CtBP-interaction (PIDLT) motif asso-ciated with transcriptional repression (Manfroid et al.,2004), as well as the vertebrate-specific homeodomainand last two zinc-fingers (Tshz3bDZnF1-3). Expressingthis construct in zebrafish does not cause a hindbrainpatterning phenotype (n 5 16/16; compare Fig. 7a–c).Similarly, overexpression of mRNA encoding the firstthree zinc fingers only, without the PIDLT motif, home-odomain, or C-terminal zinc fingers (Tshz3bZnF1-3) alsofails to perturb hindbrain patterning (n 5 13/13; Fig.7d). However, when the PIDLT motif is included alongwith the N-terminal zinc fingers (Tshz3bDHD), this con-struct can perturb hindbrain patterning with a similarefficiency as the full-length protein (n 5 13/13; Fig. 7e).Ectopic Tshz3bDHD also perturbs hoxa2b expressionand causes a caudal expansion of r1 identity (data notshown), similar to that observed for full length Tshz3b(Fig. 3e,g). Thus, the homeodomain and C-terminal zincfingers do not contribute to the tshz3b overexpressionphenotype. In summary, the ability of ectopic Tshz3b toperturb Hox-dependent hindbrain segmentationrequires only the first three zinc finger domains and thePIDLT motif.

Tshz3b Functions as a Transcriptional Repressor

Evidence from studies done in Drosophila using bothfly and mouse tsh genes suggest that Teashirt-relatedproteins act as repressors of transcription (Manfroidet al., 2004; Saller et al., 2002). To confirm that this isalso true of Tshz3b, we created constructs that encodeFLAG-tagged Tshz3b fused to either a repressor domainfrom the Engrailed protein (EnR), or a transactivationdomain from the virion protein 16 (VP16) of the herpessimplex virus type 1. If Tshz3b acts as a repressor, thenTshz3b-EnR should produce the same phenotype asTshz3b alone, while Tshz3b-VP16 should have either noeffect or produce a new phenotype. Consistent withprevious findings, overexpressed tshz3b and tshz3b-

EnR cause the same hindbrain patterning phenotype(n 5 13/13 and 13/15 respectively; Fig. 7f), whileembryos injected with tshz3b-VP16 have no discern-able phenotype (n 5 11/11; Fig. 7g). Thus, it is likelythat Tshz3b functions as a repressor to antagonize hox

function.

733ECTOPIC tshz3b PERTURBS HINDBRAIN PATTERNING

FIG. 7. (a–e) Structure-function analysis for tshz3b. mRNA in situ hybridizations (ISH) for pax2a (MHB, ov) and egr2b (r3 and r5) in 19 hpfembryos injected with mRNAs coding for (b) full length Tshz3b, (c) Tshz3bDZn1-3, (d) Tshz3bZn1-3, or (e) Tshz3bDHD. Overexpression oftshz3bDZn1-3 or tshz3bZn1-3 produces little effect on hindbrain patterning, while tshz3bDHD has the same phenotype as full length tshz3b.(f, g) Tshz3b acts as a repressor to disrupt hindbrain patterning. mRNA ISH for pax2a (MHB, ov) and egr2b in embryos injected with mRNAcoding for FLAG-tagged (e) Tshz3b-EnR (fused to an engrailed repressor domain), or (f) Tshz3b-VP16 (fused to a Virion protein 16 activationdomain). Tshz3b-EnR produces the same hindbrain patterning defect as Tshz3b, while Tshz3b-VP16 has no effect. All views are dorsal withanterior to the left. ISH, in situ hybridization; ORF, open reading frame; MHB, midbrain-hindbrain boundary; ov, otic vesicle; r, rhombomere;ZnF, zinc finger.

734 ERICKSON ET AL.

Expressing the Fly Tsh Protein in Zebrafish Doesnot Perturb Hindbrain Patterning

The observation that only the evolutionarily con-served N-terminal zinc fingers and CtBP-binding domainare required for the tshz3b overexpression phenotypesuggests that the fly Tsh protein might produce a similarphenotype when expressed in fish. The ability of mouseTshz proteins to rescue the tsh mutant phenotype inflies also suggests that, although Tsh-related proteins arenot well conserved at the sequence level, they exhibit ahigh level of functional conservation (Manfroid et al.,2004). To test this hypothesis, we injected mRNAsencoding for either untagged or Myc-tagged Drosophila

Tsh, confirmed its translation by a-Myc immunostains(Supporting Information, Fig. 3a,b) and analyzedembryos for patterning defects. Injecting either low(300 pg) or high (400–600 pg) doses of untagged tsh

mRNA into one-cell embryos causes tail morphologydefects that are not observed in tshz3b-injectedembryos (low dose n 5 22/40; high dose n 5 20/25;

Fig. 8a–c). To determine if tsh overexpression also per-turbs hindbrain patterning, we analyzed the expressionof wnt1 (MHB), egr2b (r3/r5), and hoxd4a (r7/spinalcord). Embryos injected with a low dose of tsh mRNAdo not have any of the hindbrain segmentation defectsassociated with tshz3b overexpression or perturbationsin Hox function (n 5 32/32; Fig. 8d,e). Consistent withthe in situ hybridization results, a low dose of tsh mRNAdoes not disrupt branchimotor neuron patterning inTg(isl1:GFP) embryos (n 5 70/76; Fig. 8g,h). However,in higher doses, tsh mRNA can cause morphologicaldefects that, in addition to severe tail truncations, alsoincludes a dismorphic neural tube and disorganizedbranchiomotor neurons (Fig. 8f,i). However, thesedefects are not consistent with the hindbrain patterningphenotypes caused by tshz3b mRNA overexpression ora loss of hox function. Thus, despite previous evidencethat fly and vertebrate Tsh-related proteins are function-ally conserved, and that the tshz3b overexpression phe-notype does not require any of the characterized verte-

FIG. 8. Overexpressed fly tsh does not perturb Hox-dependent hindbrain patterning. (a–c) Dose-dependent effects on the morphology oftsh mRNA-injected embryos. Embryos injected with 300 pg of tsh mRNA exhibit defects in the ventral tail and yolk-extension region (blackarrows), while embryos injected with 600 pg of tsh mRNA exhibit more severe tail defects. (d–f) The expression of wnt1 (MHB, HB), egr2b(r3 and r5), and hoxd4a (r7 and spinal cord) in wild-type and tsh-overexpressing embryos at 24 hpf. Although a high dose of tsh mRNA canperturb the morphology of the midbrain and hindbrain tissues, ectopic fly tsh does not produce the same hindbrain patterning defects asoverexpressed tshz3b. (g–i) Branchiomotor neurons as visualized in Tg(isl1:GFP) embryos in 48 hpf. Low doses of tsh mRNA do notaffect branchiomotor neuron patterning (h), while higher doses lead to variable dismorphic phenotypes that do not resemble the tshz3b-overexpression phenotype (i). Embryos in a–c are lateral views of live embryos with anterior to the left, while views in d–i are dorsal. MHB,midbrain-hindbrain boundary; r, rhombomere.

735ECTOPIC tshz3b PERTURBS HINDBRAIN PATTERNING

brate-specific domains, overexpression of the fly tsh

mRNA in fish does not specifically antagonize Hox-dependent hindbrain patterning.

DISCUSSION

In this article, we explore the role of the zebrafishteashirt-related gene tshz3b in hindbrain segmenta-tion by detailing the regulation of its mRNA expres-sion pattern and examining its overexpression pheno-types. We find that zebrafish tshz3b expression inthe hindbrain is positively regulated by Pbx, Meis,and Hox function, with additional positive inputfrom retinoic acid signaling (see Fig. 2). Furthermore,we observe that Tshz3b acts as a transcriptionalrepressor to antagonize Hox-dependent hindbrain pat-terning (Figs. 3–7). Taken together, tshz3b is both atranscriptional target and a negative regulator of thehox genes that are responsible for anterior-posteriorpatterning of the vertebrate hindbrain.

The Regulation of tshz3b Expressionin the Hindbrain

The results of the Pbx, Meis1, and Hoxb1 loss of func-tion experiments, as well as the hoxb1b overexpressionassay, suggest that both tshz3b and tshz3a expression ispositively regulated by Hox and TALE-class transcriptionfactors (Figs. 2a–j and 5e,f). This is consistent with thesituation in flies, where tsh expression is directly acti-vated by Antp-Exd-Hth trimeric complexes (McCormicket al., 1995; Merabet et al., 2007). It is possible thatHox-TALE complexes directly regulate tshz3b as well,as we have identified putative Pbx-Meis binding sites inevolutionarily conserved blocks of tshz3b promotersequence (T.E., unpublished observations). Definitivelydemonstrating the direct transcriptional regulation ofthe tshz3 paralogues by Hox-Pbx-Meis complexeswould add to the small but growing list of direct Hoxtargets in vertebrates.

Retinoic acid signaling also positively regulatestshz3b transcription (Fig. 2k–o). It is likely that ther5 and r6 domains of tshz3b expression require RAinput, as these domains of tshz3b expression are lostin DEAB-treated embryos. tshz3b expression in r4and the spinal cord is not reduced in DEAB-treatedembryos, suggesting that these domains of tshz3b

expression are not dependent on RA signaling. Previ-ous research has demonstrated that RA also positivelyregulates tshz1 expression (Wang et al., 2007). How-ever, not all tshz genes are RA-responsive, sincetshz3a expression in the hindbrain does not respondto exogenous RA treatment (data not shown). Takentogether, tshz3b expression in the hindbrain ispositively regulated by both Hox and RA transcrip-tional input.

tshz3b Overexpression Inhibits Hox-DependentHindbrain Segmentation

Ectopic tshz3b produces hindbrain segmentationdefects that closely resemble those caused by a Hox-Pbx-Meis loss of function. These phenotypes includethe downregulation of hoxb1a, hoxa2b, and egr2b

expression (Figs. 3, 6, and 7), a posterior expansion ofHox-independent r1 identity (see Fig. 3), and a mispat-terning of the hindbrain cranial and reticulospinal neu-rons (see Fig. 4). Furthermore, ectopic tshz3b can blockthe r2-to-r4 homeotic transformation caused by hoxb1b

overexpression (Fig. 6e–h). This tshz3b-mediated lossof segmental identity is not caused by ectopic activity ofposterior hox genes, since r1 identity is expanded cau-dally (Fig. 3g), and the genetic and neuronal markers ofspinal cord identity are not expanded rostrally (Figs. 4band 5b). All of these phenotypes are consistent withoverexpressed tshz3b functioning as a negative regula-tor of Hox PG 1–4 function.

The question now becomes one of how Tshz3b isable to inhibit Hox function in the hindbrain. Structure-function analyses suggest that Tshz3b functions as arepressor and requires the evolutionarily conserved N-terminal zinc fingers and CtBP-binding motif in order toantagonize Hox function (see Fig. 7). Addition of aVP16 activation domain, or removal of the CtBP-interac-tion motif abrogates the Hox-antagonizing activity of ec-topic Tshz3b. This suggests that Tshz3b represses thetranscription of Hox target genes. Of course, some ofthese targets could be hox genes themselves. In Dro-sophila, the Wnt-mediated repression of Ultrabithoraxtranscription (Ubx/Hox7) is accomplished by a com-plex of Tsh and the co-repressors CtBP and Brinker(Saller et al., 2002; Waltzer et al., 2001). In this exam-ple, it is thought that Tsh does not directly bind DNA,but rather is recruited to its target gene via its interac-tion with the DNA-binding protein Brinker. CouldTshz3b use a similar piggybacking mechanism forrecruitment to Hox target sequences? Fly Tsh is able tobind Hox proteins such as Scr and its vertebrate ortho-logue Hoxa5 via its N-terminal acidic domain (Taghli-Lamallem et al., 2007). Additionally, Tsh can bind to theHox6 orthologue Antp, though the interaction domainhas not been mapped. Similar to the fly acidic domain,Tshz3b contains an N-terminal stretch of 23 amino acidswith 12 acidic residues, hinting at the possibility thatTshz3b can bind Hox proteins using a similar mecha-nism. Additionally, fly Tsh has been shown to interactdirectly with TALE-class homeodomain proteins Exd(Pbx) and Hth (Meis) in vitro, and this interactionbetween Hth and Tsh may be important to repress thetranscription of eyes absent (eya) during eye develop-ment (Bessa et al., 2002). These data suggest that zebra-fish Tshz3b may be recruited to Hox-regulated pro-moters through a direct interaction with Hox proteins

736 ERICKSON ET AL.

and/or their TALE-class cofactors. As such, it is possiblethat the tshz3b-overexpression phenotype is the resultof perturbing either the normal regulation of Hox targetgenes, or by hampering the normal biochemical interac-tions between Hox proteins and their TALE-class part-ners. Alternatively, tshz3b may function independentlyof hox genes to repress shared target genes, as has beenalso been observed in Drosophila (Alexandre et al.,1996; de Zulueta et al., 1994; Taghli-Lamallem et al.,2007). Although these hypotheses remain to be tested,it is possible that Tshz3b antagonizes Hox function viaall three mechanisms.

Given the ability of mouse Tshz genes to rescue tshmutant flies, and the observation that the tshz3b over-expression phenotype does not require the vertebrate-specific homeodomain or C-terminal zinc fingers, it wassurprising to find that overexpressing fly tsh in zebrafishdid not produce similar hindbrain patterning defects(see Fig. 8). However, fly and vertebrate Teashirt pro-teins have not been well conserved over evolution(Caubit et al., 2000). Thus, while both zebrafish Tshz3band fly Tsh can repress anterior Hox function in theirrespective organisms, they may do so via distinctmechanisms that reflect their evolutionary divergence.Alternatively, they may function by homologous mecha-nisms, albeit one where the fly Tsh protein is incompati-ble with the vertebrate system.

The Role of Vertebrate teashirt-Related Genes inHindbrain Patterning and Development

Other vertebrate teashirt-related genes can also nega-tively regulate Hox function, as overexpression of tsh1in Xenopus perturbs Hox expression in the hindbrainand cranial neural crest (Koebernick et al., 2006). How-ever, the results from that study differ from ours in thattheir overexpression phenotype was not limited to per-turbations in hindbrain segmentation. Whereas tshz3b

overexpression in zebrafish primarily affects hindbrainpatterning, ectopic tsh1 in frogs also eliminatesEngrailed-2 expression at the midbrain-hindbrainboundary (MHB) and mildly downregulates Otx-2 in theforebrain, both of which are regulated by hox-independ-ent mechanisms. Although Hox proteins are notinvolved in anterior neural patterning, maintenance ofthe MHB does require a functional complex of Pbx andEngrailed proteins (Erickson et al., 2007). However,other than downregulating tshz3a expression in themidbrain (see Fig. 8), ectopic tshz3b does not pheno-copy a loss of Pbx function in the midbrain (see eng2a

expression in Fig. 6a–d and pax2a expression in Fig. 7).Thus, while both fish tshz3b and frog tsh1 can perturbhindbrain segmentation, it appears that the effects oftshz3b are largely limited to the hindbrain, while Xeno-

pus tsh1 provokes more widespread neural patterningdefects. Notably, no hindbrain-patterning defects are

observed in tshz1-overexpressing embryos (SupportingInformation, Fig. 2c–h). Combined, these data suggestthat the ability to antagonize anterior hindbrain Hoxfunction is not shared by all zebrafish tshz homologs.Whether this reflects functional differences betweenTshz1 and Tshz3 proteins, or differences betweenmodel organisms remains to be seen.

Loss-of-function studies of vertebrate teashirt-relatedgenes also suggest that they play a role in hindbraindevelopment. Morpholino-mediated knockdown ofXenopus Tsh1 protein causes severe hindbrain segmen-tation defects that, like Xenopus tsh1 overexpression,resemble a Hox loss-of-function phenotype (Koebernicket al., 2006). The exact mechanisms by which both again and loss of tsh1 function produce similar hindbrainphenotypes in frogs have not been determined. In con-trast, morpholino-mediated knockdown of zebrafishTshz3b does not alter hindbrain segmental identity, andproduces only mild defects in hindbrain morphology(Supporting Information, Fig. 4a–e). Because these phe-notypes are not consistent with a loss or gain of Hoxfunction, our data suggest that endogenous Tshz3bdoes not play a major role in regulating Hox function atthe developmental stage when hindbrain segmentalidentity is specified. In support of this idea, Tshz3 mu-tant mice also fail to exhibit defects in hindbrain seg-mental identity (Caubit et al., 2010). However, it is pos-sible that Tshz3 proteins function redundantly withother factors in regulating Hox activity. In zebrafish,tshz3a is a possible candidate in this regard. Addition-ally, it is also possible that the modulation of Hox func-tion by endogenous Tshz3 proteins is spatially and tem-porally restricted, such as in specific cell types or at dis-crete genetic loci. In support of this idea, Tshz3inactivation does lead to defects in the hindbrain motorneurons that control respiratory rhythm, causing Tshz3

mutant mice to suffocate at birth (Caubit et al., 2010).Hox proteins are known regulators of the developmen-tal pathways involved in breathing behaviour (Cham-pagnat et al., 2009; Chatonnet et al., 2003). WhetherTshz3 inactivation directly perturbs Hox-dependentdevelopment of the respiratory system remains to beelucidated. However, the ability of ectopic Tshz3b toinhibit Hox function in the zebrafish hindbrainpresents the interesting possibility that the Tshz3-nullphenotype in mice is caused by a failure to properlyregulate Hox function during the development of respi-ratory neural circuitry.

The Involvement of tshz3b With Non-HoxDevelopmental Pathways

Teashirt-related proteins are multifunctional tran-scription factors, able to directly interact with transcrip-tional repressor complexes (Manfroid et al., 2004; Salleret al., 2002), with Hox and TALE-class homeodomain

737ECTOPIC tshz3b PERTURBS HINDBRAIN PATTERNING

transcription factors (Bessa et al., 2002; Taghli-Lamal-lem et al., 2007), and with transcriptional effectors ofthe Wnt pathway such as b-catenin and Tcf3 (Galletet al., 1998, 1999; Onai et al., 2007). Although ectopicTshz3b could potentially disrupt numerous develop-mental processes, the most obvious phenotype is a dis-ruption of Hox-dependent hindbrain segmentation.Our results stand in contrast to studies in frogs, whereectopic tsh3 dorsalizes the embryonic axis by enhanc-ing canonical Wnt signaling (Onai et al., 2007). How-ever, this discrepancy does not eliminate the possibil-ity that endogenous zebrafish tshz3b participates onthe Wnt pathway. Genetic interaction experimentsbetween tshz3b and members of the Wnt pathway,along with biochemical evidence of a functionally rel-evant interaction between Tshz3b and b-catenin orTcf3 in the hindbrain would help to clarify the rela-tionship between tshz3b and the Wnt pathway. Defini-tive evidence that tshz3b acted on both the Wnt andHox pathways could define tshz3b as point ofintegration between these two pathways during hind-brain development.

METHODS AND MATERIALS

Animal care, Fish Lines, and General Procedures

Embryonic and adult fish were cared for according tostandard protocols (Westerfield, 2000) in accordancewith the Canadian Council for Animal Care (CCAC)guidelines. Embryos were grown at either 25.58C,28.58C, or 338C in embryo media (EM) and stagedaccording to standardized morphological milestones(Kimmel et al., 1995). Embryos that were analyzed pastthe stage of 24 hpf were grown in embryo media sup-plemented with 0.003% 1-phenyl 2-thiourea (PTU)(Sigma) to prevent pigment formation. The AB strain ofwild type fish was used for all experiments exceptwhere noted. We also used the wild type Tubingen (TU)strain, the lazarus mutant (lzr/pbx4b557/b557) (Popperl

et al., 2000), and the transgenic lines Tg(isl1:GFP)(Higashijima et al., 2000), and Tg(dusp6:EGFP) (Molinaet al., 2007). All fish lines were acquired from ZIRC,except the Tg(dusp6:EGFP) strain, which was a kindgift from Michael Tsang.

Riboprobe Synthesis and mRNA In SituHybridization

Digoxigenin- or fluoroscein-labeled riboprobes weresynthesized from either linearized plasmid templates orPCR product containing either a T3 or T7 RNA polymer-ase site (Thisse and Thisse, 2008), and one or two-colour mRNA in situ hybridizations were performed aspreviously described (Erickson et al., 2010).

Rapid Amplification of cDNA Ends (RACE), tshz3bcDNA Cloning, and In Vitro mRNA Synthesis

Total RNA from zebrafish was isolated by TRIzol (Invi-trogen) extraction followed by standard phenol/chloro-form purification. To determine the full open readingframe for the zebrafish tshz3b gene, we performed 50

RACE using the BD SMART RACE kit (BD Biosciences)from total RNA isolated from 18 hpf zebrafish embryos.The first reaction was performed using a 50-CTGGATTCACTCAGGTGGGACTCGCTGTC-30 tshz3b-specificreverse primer, while the nested PCR reaction wasdone using a 50-GGCAGGTTCCTCCTCCAAAGCAGAATCC-30 tshz3b reverse primer. In this way, we estab-lished that the zebrafish tshz3b (GenBank HQ116415)gene is organized as a two-exon gene containing a 3,444bp ORF coding for a putative 1,147 amino acid proteinthat contains the same domains as previously describedfor vertebrate Teashirt-related proteins (Koebernicket al., 2006; Onai et al., 2007). The Drosophila tsh-A

(NM_078891) ORF was amplified from total RNA iso-lated from second instar larvae (a kind gift from KirstKing-Jones). To make mRNA expression constructs,cDNA was first synthesized using the SuperScriptIIIFirst-Strand Synthesis System for RT-PCR (Invitrogen),

Table 1Primers Used to Create the tshz3b, tsh, and tshz1 mRNA Expression Constructs

Construct Primer sequence 50-30

tshz3b Fwd: CACA(GAATTC)CACCATGCCGCGGAGGAAACAGRev: CACA(CTCGAG)CTAAGGTTTCTCAAGTTCACTAACRev2: CACA(CTCGAG)GAGGTTTCTCAAGTTCACTAAC

tshz3bZnF1–3 Rev: CACA(CTCGAG)TCATTTGCCTTTCTTGATGGCAGAGtshz3bDHD Rev: CACA(CTCGAG)TCAAGCAGGAGAGATCTCCTCTGATTtshz3bDZnF1–3 Fwd: CACA(GAATTC)CACCATGGAGTCAATGTCCACAACtsh-RA Fwd: CACA(AGATCT)ACATGTTACACGAGGCTCTGATGCTCGAAATCTACAG

Rev: CACA(CTCGAG)AGGCGGTCTTCTCCTTCTTCACGCtshz1 Fwd: CACA(GAATTC)CACCATGCCGCGGAGAAAGCAGC

Rev: CACA(CTCGAG)TTAGGCTTTTTCTAACTTTTCCAGCTCAGTAAC

The zebrafish tshz3b forward primer was used to generate inserts for all pCS21, pCS3 1 MT, and pCS3 1 FLAG expression constructs,except for tshz3bDZnF1–3, which required a unique forward primer. The tshz3b Rev2 primer lacks a stop codon and was used to createEnR and VP16 fusions. Primer sequences annotation: CACA—leader sequence; brackets indicate a restriction enzyme site incorporatedinto the primer for cloning purposes; gene-specific sequence is underlined. ZnF, zinc-finger; Fwd, forward primer; Rev, reverse primer.

738 ERICKSON ET AL.

followed by a PCR reaction using Phusion (NEB) poly-merase. Both steps were performed according to themanufacturers’ suggested protocols. The primers usedto make the zebrafish tshz3b, tshz1, and fly tsh con-structs are listed in Table 1. Inserts were cloned intopCS21, pCS3 1 MT, pCS3 1 FLAG, pCS3 1 FLAG-EnR,or pCS3 1 FLAG-VP16 expression vectors. Sequences ofall constructs were confirmed by Dyenamic ET sequenc-ing (Amersham). Preparation of linearized templateDNA, in vitro synthesis of capped mRNA, and reactionpurification was done as previously described (Gongaland Waskiewicz, 2008). The in vivo translation of themRNAs coding for the various Tshz3b deletion/EnR/VP16 constructs was confirmed by Western blot (datanot shown). Unless otherwise noted, the followingquantities of mRNA were injected into one cell stageembryos: tshz3b—600 pg; tshz1—600 pg; tsh—lowdose 300 pg, high dose—400 to 600 pg.

Morpholinos

All morpholinos were purchased from Gene Tools,LLC. The pbx2/pbx4 (Erickson et al., 2007; Pillay et al.,2010),meis1 (French et al., 2007), and hoxb1a/hoxb1b

(McClintock et al., 2002) morpholinos have been previ-ously described. The following morpholinos againsttshz3b were used: ATG translation blocker—50-CGCGGCATGTTTCTCTTTCAGGGTT, 3 to 6 ng perembryo; Exon1/Intron1 splice blocker—50-AAGAAGAAGAGCCGTACCTGCCGAG, 2 to 4 ng per embryo.

Pharmacological Treatments

All chemicals were dissolved in DMSO and diluted totheir appropriate concentrations in EM, with equivalentdilutions of DMSO alone used as solvent controls.Embryos were protected from light during the period ofdrug exposure and grown at either 25.58C or 28.58C. Toantagonize retinoic acid signaling, a 10 lM solution ofdiethylaminobenzaldehyde (DEAB) (Sigma) was appliedto embryos in their chorions at 3 hpf and removedupon fixation at 15 hpf. The retinoic acid signaling path-way was activated by the application of exogenous all-trans-retinoic acid (RA) (Sigma) at a concentration 10 to100 nM to embryos in their chorions at 3 hpf andremoved upon fixation at 15 hpf.

Zebrafish Cell Lysate and Western Analysis

The protocol to remove the embryonic yolk proteinsand make cell lysates suitable for Western analysis isbased on a protocol and recipes described previously(Link et al., 2006). The embryos were dechorionatedand immersed in 1 ml of cold deyolking buffer plusRoche Complete Mini, EDTA-free protease inhibitorcocktail. To disrupt and solubilize the yolk, embryoswere pipetted repeatedly and lightly vortexed for 30 s.The cells were pelleted by centrifugation at 2,000 rpm

for 30 s at 48C and washed twice with 1 ml of colddeyolking wash buffer. Cells were lysed with 3 ll perstarting embryo of 13 sample loading buffer (Invi-trogen) with 2.5% b-mercaptoethanol. Protein sampleswere separated using NuPAGE Novex Bis-Tris gels andXCell Sure Lock Mini-Cell system (Invitrogen) accordingto manufacturer’s protocols. We raised polyclonal anti-sera against zebrafish Tshz3b (Covance). Two Tshz3bKLH-conjugated peptides were synthesized based ontheir high potential antigenicity: peptide 1—acetyl-CSSDAGESARGESPKERR-amide representing aminoacids 682 to 699; peptide 2—[C]-SKTHGKSPEDHLMYV-SELEKP-acid representing amino acids 1,127 to 1,147.Antibodies were affinity purified using the original anti-genic peptides. Nonaffinity purified antibody was usedat a 1:1,000 dilution for Western analysis. The second-ary antibody (donkey anti-rabbit HRP FAB fragmentsfrom Amersham) was used at a 1:7,500 dilution. ECLreaction was performed using PicoSignal (Pierce).

Whole Mount Immunohistochemistry

Immunohistochemical stains were performed essen-tially as previously described (French et al., 2009; Was-kiewicz et al., 2001) using the following primary anti-bodies: a-Tshz3b affinity purified peptide 1—1:200;9E10 a-Myc—1:250; rmo44 a-NF-M—1:250. Embryoswere imaged on Leica TCS-SP2, Nikon Eclipse C1, orZeiss LSM700 confocal microscopes. Z-projections weremade in ImageJ and figures assembled in Photoshop.

ACKNOWLEDGMENTS

The authors thank Mattea Bujold and Kirst King-Jones forproviding the Drosophila RNA used to clone teashirt.

LITERATURE CITED

Alexandre E, Graba Y, Fasano L, Gallet A, Perrin L, DeZulueta P, Pradel J, Kerridge S, Jacq B. 1996. TheDrosophila teashirt homeotic protein is a DNA-binding protein and modulo, a HOM-C regulatedmodifier of variegation, is a likely candidate forbeing a direct target gene. Mech Dev 59:191–204.

Andrew DJ, Horner MA, Petitt MG, Smolik SM, Scott MP.1994. Setting limits on homeotic gene function:Restraint of Sex combs reduced activity by teashirtand other homeotic genes. EMBO J 13:1132–1144.

Bell E, Wingate RJ, Lumsden A. 1999. Homeotic transfor-mation of rhombomere identity after localizedHoxb1 misexpression. Science 284:2168–2171.

Bessa J, Gebelein B, Pichaud F, Casares F, Mann RS.2002. Combinatorial control of Drosophila eyedevelopment by eyeless, homothorax, and teashirt.Genes Dev 16:2415–2427.

Caubit X, Core N, Boned A, Kerridge S, Djabali M,Fasano L. 2000. Vertebrate orthologues of the

739ECTOPIC tshz3b PERTURBS HINDBRAIN PATTERNING

Drosophila region-specific patterning gene teashirt.Mech Dev 91:445–448.

Caubit X, Thoby-Brisson M, Voituron N, Filippi P, Beven-gut M, Faralli H, Zanella S, Fortin G, Hilaire G,Fasano L. 2010. Teashirt 3 regulates development ofneurons involved in both respiratory rhythm and air-flow control. J Neurosci 30:9465–9476.

Champagnat J, Morin-Surun MP, Fortin G, Thoby-BrissonM. 2009. Developmental basis of the rostro-caudalorganization of the brainstem respiratory rhythmgenerator. Philos Trans R Soc Lond B Biol Sci364:2469–2476.

Chan SK, Jaffe L, Capovilla M, Botas J, Mann RS. 1994.The DNA binding specificity of Ultrabithorax ismodulated by cooperative interactions with extra-denticle, another homeoprotein. Cell 78:603–615.

Chandrasekhar A. 2004. Turning heads: Developmentof vertebrate branchiomotor neurons. Dev Dyn229:143–161.

Chandrasekhar A, Moens CB, Warren JT Jr, Kimmel CB,Kuwada JY. 1997. Development of branchiomotorneurons in zebrafish. Development 124:2633–2644.

Chatonnet F, Dominguez del Toro E, Thoby-Brisson M,Champagnat J, Fortin G, Rijli FM, Thaeron-Antono C.2003. From hindbrain segmentation to breathing af-ter birth: Developmental patterning in rhombo-meres 3 and 4. Mol Neurobiol 28:277–294.

Coiffier D, Charroux B, Kerridge S. 2008. Common func-tions of central and posterior Hox genes for therepression of head in the trunk of Drosophila. De-velopment 135:291–300.

Cooper KL, Leisenring WM, Moens CB. 2003. Autono-mous and nonautonomous functions for Hox/Pbx inbranchiomotor neuron development. Dev Biol253:200–213.

de Zulueta P, Alexandre E, Jacq B, Kerridge S. 1994.Homeotic complex and teashirt genes co-operate toestablish trunk segmental identities in Drosophila.Development 120:2287–2296.

Dibner C, Elias S, Frank D. 2001. XMeis3 protein activityis required for proper hindbrain patterning in Xeno-

pus laevis embryos. Development 128:3415–3426.Dupe V, Lumsden A. 2001. Hindbrain patterning

involves graded responses to retinoic acid signal-ling. Development 128:2199–2208.

Erickson T, French CR, Waskiewicz AJ. 2010. Meis1specifies positional information in the retina andtectum to organize the zebrafish visual system. Neu-ral Dev 5:22.

Erickson T, Scholpp S, Brand M, Moens CB, WaskiewiczAJ. 2007. Pbx proteins cooperate with Engrailed topattern the midbrain-hindbrain and diencephalic-mesencephalic boundaries. Dev Biol 301:504–517.

Fasano L, Roder L, Core N, Alexandre E, Vola C, Jacq B,Kerridge S. 1991. The gene teashirt is required forthe development of Drosophila embryonic trunk

segments and encodes a protein with widely spacedzinc finger motifs. Cell 64:63–79.

Ferretti E, Marshall H, Popperl H, Maconochie M, Krum-lauf R, Blasi F. 2000. Segmental expression of Hoxb2in r4 requires two separate sites that integrate coop-erative interactions between Prep1, Pbx and Hoxproteins. Development 127:155–166.

Fraser S, Keynes R, Lumsden A. 1990. Segmentation inthe chick embryo hindbrain is defined by cell line-age restrictions. Nature 344:431–435.

French CR, Erickson T, Callander D, Berry KM, Koss R,Hagey DW, Stout J, Wuennenberg-Stapleton K, NgaiJ, Moens CB, Waskiewicz AJ. 2007. Pbx homeodo-main proteins pattern both the zebrafish retina andtectum. BMC Dev Biol 7:85.

French CR, Erickson T, French DV, Pilgrim DB, Waskie-wicz AJ. 2009. Gdf6a is required for the initiation ofdorsal-ventral retinal patterning and lens develop-ment. Dev Biol 333:37–47.

Gallet A, Angelats C, Erkner A, Charroux B, Fasano L,Kerridge S. 1999. The C-terminal domain ofarmadillo binds to hypophosphorylated teashirt tomodulate wingless signalling in Drosophila. EMBO J18:2208–2217.

Gallet A, Erkner A, Charroux B, Fasano L, Kerridge S.1998. Trunk-specific modulation of wingless signal-ling in Drosophila by teashirt binding to armadillo.Curr Biol 8:893–902.

Gavalas A. 2002. ArRAnging the hindbrain. Trends Neu-rosci 25:61–64.

Gavalas A, Davenne M, Lumsden A, Chambon P, Rijli FM.1997. Role of Hoxa-2 in axon pathfinding and rostralhindbrain patterning. Development 124:3693–3702.

Gavalas A, Krumlauf R. 2000. Retinoid signallingand hindbrain patterning. Curr Opin Genet Dev10:380–386.

Gavalas A, Studer M, Lumsden A, Rijli FM, Krumlauf R,Chambon P. 1998. Hoxa1 and Hoxb1 synergize inpatterning the hindbrain, cranial nerves and secondpharyngeal arch. Development 125:1123–1136.

Gongal PA, Waskiewicz AJ. 2008. Zebrafish model ofholoprosencephaly demonstrates a key role forTGIF in regulating retinoic acid metabolism. HumMol Genet 17:525–538.

Guthrie S. 2007. Patterning and axon guidanceof cranial motor neurons. Nat Rev Neurosci 8:859–871.

Hernandez RE, Putzke AP, Myers JP, Margaretha L,Moens CB. 2007. Cyp26 enzymes generate the reti-noic acid response pattern necessary for hindbraindevelopment. Development 134:177–187.

Higashijima S, Hotta Y, Okamoto H. 2000. Visualizationof cranial motor neurons in live transgenic zebrafishexpressing green fluorescent protein under the con-trol of the islet-1 promoter/enhancer. J Neurosci20:206–218.

740 ERICKSON ET AL.

Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schil-ling TF. 1995. Stages of embryonic development ofthe zebrafish. Dev Dyn 203:253–310.

Kimmel CB, Metcalfe WK, Schabtach E. 1985. T reticu-lar interneurons: A class of serially repeating cellsin the zebrafish hindbrain. J Comp Neurol 233:365–376.

Koebernick K, Kashef J, Pieler T, Wedlich D. 2006.Xenopus Teashirt1 regulates posterior identity inbrain and cranial neural crest. Dev Biol 298:312–326.

Laugier E, Yang Z, Fasano L, Kerridge S, Vola C. 2005. Acritical role of teashirt for patterning the ventralepidermis is masked by ectopic expression of tip-top, a paralog of teashirt in Drosophila. Dev Biol283:446–458.

Link V, Shevchenko A, Heisenberg CP. 2006. Proteomicsof early zebrafish embryos. BMC Dev Biol 6:1.

Lumsden A, Keynes R. 1989. Segmental patterns of neu-ronal development in the chick hindbrain. Nature337:424–428.

Manfroid I, Caubit X, Kerridge S, Fasano L. 2004. Threeputative murine Teashirt orthologues specify trunkstructures in Drosophila in the same way as the Dro-sophila teashirt gene. Development 131:1065–1073.

Manfroid I, Caubit X, Marcelle C, Fasano L. 2006. Tea-shirt 3 expression in the chick embryo reveals a re-markable association with tendon development.Gene Expr Patterns 6:908–912.

Mann RS, Chan SK. 1996. Extra specificity from extra-denticle: The partnership between HOX and PBX/EXD homeodomain proteins. Trends Genet 12:258–262.

Mathies LD, Kerridge S, Scott MP. 1994. Role of the tea-shirt gene in Drosophila midgut morphogenesis:Secreted proteins mediate the action of homeoticgenes. Development 120:2799–2809.

Maves L, Kimmel CB. 2005. Dynamic and sequential pat-terning of the zebrafish posterior hindbrain by reti-noic acid. Dev Biol 285:593–605.

McClintock JM, Carlson R, Mann DM, Prince VE. 2001.Consequences of Hox gene duplication in thevertebrates: An investigation of the zebrafish Hoxparalogue group 1 genes. Development 128:2471–2484.

McClintock JM, Kheirbek MA, Prince VE. 2002. Knock-down of duplicated zebrafish hoxb1 genes revealsdistinct roles in hindbrain patterning and a novelmechanism of duplicate gene retention. Develop-ment 129:2339–2354.

McCormick A, Core N, Kerridge S, Scott MP. 1995. Home-otic response elements are tightly linked to tissue-specific elements in a transcriptional enhancer of theteashirt gene. Development 121:2799–2812.

McNulty CL, Peres JN, Bardine N, van den Akker WM,Durston AJ. 2005. Knockdown of the complete Hoxparalogous group 1 leads to dramatic hindbrain andneural crest defects. Development 132:2861–2871.

Merabet S, Saadaoui M, Sambrani N, Hudry B, Pradel J,Affolter M, Graba Y. 2007. A unique Extradenticlerecruitment mode in the Drosophila Hox proteinUltrabithorax. Proc Natl Acad Sci USA 104:16946–16951.

Metcalfe WK, Mendelson B, Kimmel CB. 1986. Segmen-tal homologies among reticulospinal neurons in thehindbrain of the zebrafish larva. J Comp Neurol251:147–159.

Moens CB, Prince VE. 2002. Constructing thehindbrain: Insights from the zebrafish. Dev Dyn224:1–17.

Moens CB, Selleri L. 2006. Hox cofactors in vertebratedevelopment. Dev Biol 291:193–206.

Molina GA, Watkins SC, Tsang M. 2007. Generation ofFGF reporter transgenic zebrafish and their utility inchemical screens. BMC Dev Biol 7:62.

Onai T, Matsuo-Takasaki M, Inomata H, Aramaki T, Mat-sumura M, Yakura R, Sasai N, Sasai Y. 2007. XTsh3 isan essential enhancing factor of canonical Wntsignaling in Xenopus axial determination. EMBO J26:2350–2360.

Pearson JC, Lemons D, McGinnis W. 2005. ModulatingHox gene functions during animal body patterning.Nat Rev Genet 6:893–904.

Pillay LM, Forrester AM, Erickson T, Berman JN, Waskie-wicz AJ. 2010. The Hox cofactors Meis1 and Pbx actupstream of gata1 to regulate primitive hematopoie-sis. Dev Biol 340:306–317.

Popperl H, Rikhof H, Chang H, Haffter P, Kimmel CB,Moens CB. 2000. Lazarus is a novel pbx gene thatglobally mediates hox gene function in zebrafish.Mol Cell 6:255–267.

Rauskolb C, Wieschaus E. 1994. Coordinate regulationof downstream genes by extradenticle and thehomeotic selector proteins. EMBO J 13:3561–3569.

Robertson LK, Bowling DB, Mahaffey JP, Imiolczyk B,Mahaffey JW. 2004. An interactive network of zinc-finger proteins contributes to regionalization of theDrosophila embryo and establishes the domainsof HOM-C protein function. Development 131:2781–2789.

Roder L, Vola C, Kerridge S. 1992. The role of the tea-shirt gene in trunk segmental identity in Drosophila.Development 115:1017–1033.

Rusch DB, Kaufman TC. 2000. Regulation of probosci-pedia in Drosophila by homeotic selector genes.Genetics 156:183–194.

Saller E, Kelley A, Bienz M. 2002. The transcriptionalrepressor Brinker antagonizes Wingless signaling.Genes Dev 16:1828–1838.

Santos JS, Fonseca NA, Vieira CP, Vieira J, Casares F.2010. Phylogeny of the teashirt-related zinc finger(tshz) gene family and analysis of the developmentalexpression of tshz2 and tshz3b in the zebrafish. DevDyn 239:1010–1018.

741ECTOPIC tshz3b PERTURBS HINDBRAIN PATTERNING

Studer M, Lumsden A, Ariza-McNaughton L, Bradley A,Krumlauf R. 1996. Altered segmental identity andabnormal migration of motor neurons in mice lack-ing Hoxb-1. Nature 384:630–634.

Taghli-Lamallem O, Gallet A, Leroy F, Malapert P, Vola C,Kerridge S, Fasano L. 2007. Direct interaction betweenTeashirt and Sex combs reduced proteins, via Tsh’sacidic domain, is essential for specifying the identity ofthe prothorax in Drosophila. Dev Biol 307:142–151.

Thisse C, Thisse B. 2008. High-resolution in situ hybrid-ization to whole-mount zebrafish embryos. Nat Pro-toc 3:59–69.

van Dijk MA, Murre C. 1994. extradenticle raises theDNA binding specificity of homeotic selector geneproducts. Cell 78:617–624.

Vlachakis N, Choe SK, Sagerstrom CG. 2001. Meis3synergizes with Pbx4 and Hoxb1b in promotinghindbrain fates in the zebrafish. Development128:1299–1312.

Vlachakis N, Ellstrom DR, Sagerstrom CG. 2000. A novelpbx family member expressed during early zebrafishembryogenesis forms trimeric complexes withMeis3 and Hoxb1b. Dev Dyn 217:109–119.

von Baer KE. 1828. Entwicklungsgeschichte der Thiere:Beobachtung und Reflexion. Konigsberg: Borntrager.

Waltzer L, Vandel L, Bienz M. 2001. Teashirt is requiredfor transcriptional repression mediated by highWingless levels. EMBO J 20:137–145.

Wang H, Lee EM, Sperber SM, Lin S, Ekker M, Long Q. 2007.Isolation and expression of zebrafish zinc-finger transcrip-tion factor gene tsh1. Gene Expr Patterns 7:318–322.

Waskiewicz AJ, Rikhof HA, Hernandez RE, Moens CB.2001. Zebrafish Meis functions to stabilize Pbx pro-teins and regulate hindbrain patterning. Develop-ment 128:4139–4151.

Waskiewicz AJ, Rikhof HA, Moens CB. 2002. Eliminatingzebrafish pbx proteins reveals a hindbrain groundstate. Dev Cell 3:723–733.

Westerfield M. 2000. The zebrafish book. A guide forthe laboratory use of zebrafish (Danio rerio), 4th ed.Eugene, OR: University of Oregon Press.

Zhang M, Kim HJ, Marshall H, Gendron-Maguire M,Lucas DA, Baron A, Gudas LJ, Gridley T, Krumlauf R,Grippo JF. 1994. Ectopic Hoxa-1 induces rhombo-mere transformation in mouse hindbrain. Develop-ment 120:2431–2442.

742 ERICKSON ET AL.