Novel Activities of Mafb Underlie Its Dual Role in Hindbrain Segmentation and Regional Specification

Developmental Biology 253, 150 –162 (2003) doi:10.1006/dbio.2002.0864

Novel Activities of Mafb Underlie Its Dual Role in Hindbrain Segmentation and Regional Specification Franc¸ois Giudicelli,* ,1 Pascale Gilardi-Hebenstreit,* Fatima Mechta-Grigoriou,† Christophe Poquet,* and Patrick Charnay* ,2 *Unite´ 368 de l’Institut National de la Sante´ et de la Recherche Me´dicale, Ecole Normale Supe´rieure, 46 rue d’Ulm, 75230 Paris Cedex 05, France; and †Unite´ “Expression ge´ne´tique et maladies,” CNRS URA 1644, Institut Pasteur, 25 rue du Docteur Roux, 75724 Paris Cedex 15, France

The bZip transcription factor Mafb is expressed in two segments of the developing vertebrate hindbrain: the rhombomeres 5 and 6. Loss of Mafb expression in the mouse mutant kreisler leads to elimination of r5 and to alterations of r6 regional identity. Here, we further investigated the role of Mafb in hindbrain patterning using gain-of-function experiments in the chick embryo. Our work has revealed novel functions for Mafb, including a positive autoregulatory activity, the capacity to repress Hoxb1 expression, and the capacity to synergise with or antagonise Krox20 activity. These different activities appear to be spatially restricted in the hindbrain, presumably due to interactions with other factors. Reinvestigation of the kreisler mutation indicated that it also results in an ectopic activation of Mafb in rhombomere 3, accounting for the previously described molecular alterations of this rhombomere in the mutant. Together, these data allow us to refine our view of the dual function of Mafb in both segmentation and specification of anteroposterior identity in the hindbrain. © 2003 Elsevier Science (USA)

Key Words: hindbrain; rhombomere; segmentation; Mafb; Hox; Eph; kreisler; Krox20.

INTRODUCTION Following formation of the primary cephalic vesicles, the vertebrate embryonic hindbrain is transiently segmented along the anteroposterior (AP) axis into 7/8 metameric transverse units, termed rhombomeres (r) (reviewed in Lumsden, 1990). The rhombomeres act as compartments, characterized by specific gene expression patterns and restrictions of cell movements across rhombomere boundaries (Fraser et al., 1990; Wilkinson, 1995). The rhombomeric organisation underlies metameric patterns of neuronal differentiation within the hindbrain and segmental specification and migration of neurogenic and branchial neural crest (Keynes et al., 1990; Lumsden and Guthrie, 1991; Clarke and Lumsden, 1993; Graham and Lumsden, 1996; Trainor and Krumlauf, 2001). Hindbrain segmentation occurs concomitantly with the beginning of somito1 Current address: Vertebrate Development Laboratory, Cancer Research UK, 44 Lincoln’s Inn Fields, London, WC2A 3PX, UK. 2 To whom correspondence should be addressed. Fax: ⫹33 1 44 32 39 88. E-mail: [email protected].

150

genesis, but appears to proceed through a different mechanism, involving progressive subdivision of the neural plate rather than sequential generation of repeated units (Guthrie, 1996; Pourquie, 2001). To date, functional analyses have identified a number of genes involved in different aspects of hindbrain segmentation: subdivision and delimitation of territories, cell segregation between adjacent rhombomeres, and acquisition of a specific AP identity. Four genes, encoding transcription factors, Hoxa1, Krox20, Gbx2, and Mafb/ kreisler/Krml1, have been shown to be involved in establishment of rhombomeric territories, at different levels of the AP axis (Lufkin et al., 1991; Chisaka et al., 1992; Swiatek and Gridley, 1993; Schneider-Maunoury et al., 1993, 1997; Wassarman et al., 1997; Cordes and Barsh, 1994). The two-way signalling system consisting of the Eph family tyrosine kinase receptors and their ephrin-B transmembrane ligands has been implicated in cell segregation (Xu et al., 1999), and the anteriorly expressed Hox genes have been shown to play an essential role in the specification of AP rhombomeric identity (reviewed in Lumsden and Krumlauf, 1996; see also Gavalas et al., 1997; Bell et al., 1999; Davenne et al., 1999; Barrow et al., 2000; Gram0012-1606/03 $35.00 © 2003 Elsevier Science (USA) All rights reserved.

151

Mafb Functions in Hindbrain Segmentation

matopoulos et al., 2000). Interestingly, Hoxa1, Krox20, and Mafb appear to be involved in both hindbrain subdivision and the acquisition of positional identity (for a review, see Schneider-Maunoury et al., 1998). Mafb is among the first genes to be expressed with an AP regionalised pattern in the zebrafish (Moens et al., 1998), chick (Eichmann et al., 1997), and mouse hindbrain (Cordes and Barsh, 1994; F. Mechta-Grigoriou, F. Giudicelli, C. Pujades, P. Charnay, and M. Yassito, unpublished observations). In the chick, Mafb is expressed in the caudal hindbrain from the 5– 6 somite (s) stage in the prospective rhombomeres (pr) 5 and 6, and maintained in r5/6 until E3 (Eichmann et al., 1997). In addition, Mafb expression extends progressively to the neural folds of r7 and r8 between the 6 s and 10 s stages; it is observed from the 14 s stage in the neural crest migrating adjacent to r6 and extends progressively along the roof plate of the entire hindbrain. After E3, Mafb expression is down-regulated in r5/6 but maintained in the roof plate (Eichmann et al., 1997). Originally identified as a member of the Maf family of leucine zipper-containing transcription factors (Kataoka et al., 1994), Mafb was subsequently shown to be encoded by the mouse kreisler (kr) gene. The recessive kr mutation consists in a large X-ray irradiation-induced chromosomal inversion encompassing the Mafb locus (Cordes and Barsh, 1994). It is assumed that it represents a cis-acting regulatory mutation abolishing expression from the locus specifically in the hindbrain, with no effect on MafB expression or function in other tissues (Cordes and Barsh, 1994; Eichmann et al., 1997). kr/kr mice are deficient in hearing and exhibit inner ear hypoplasia, most likely as a result from an earlier segmentation defect in the developing hindbrain (Deol, 1964). Indeed, the mutant hindbrain lacks identifiable rhombomeres 5 and 6, the r4 –r7 region remains unsegmented, and the putative r5/r6 territory is replaced by a smaller, so-called rX territory, retaining a large part of, but not all, r6 properties (Frohman et al., 1993; McKay et al., 1994; Manzanares et al., 1999b). A similar hindbrain phenotype is observed in fish embryos homozygous for a null allele of the zebrafish Mafb ortholog, valentino (val) (Moens et al., 1996, 1998). Studies in kr and val mutants combined with cis-acting regulatory region analyses have shown that Mafb is involved in the direct transcriptional activation of Hoxa3 and Hoxb3 in r5/r6 (Prince et al., 1998; Manzanares et al., 1999b and references therein). Together, these data suggest the involvement of Mafb in both hindbrain segmentation and rhombomeric AP specification. Mafb is likely to be able to form heterodimers in vivo with a number of factors (reviewed in Blank and Andrews, 1997), including the large Maf family members (Mafb, c-Maf, Nrl, L-maf) and AP1 subunits (c-jun, junB, junD, c-fos, fosB, fra-1, fra-2). It has also been shown to directly interact with the helix–turn– helix factor Ets-1 in the myelomonocytic lineage (Sieweke et al., 1996). However, the exact molecular nature of the Mafb-containing complexes involved in r5/6 development and Hoxa3 and Hoxb3 activation, in particular whether they consist of Mafb homo- or

heterodimers, is not known. While Mafb homodimers have been shown to efficiently bind Hoxb3 and Hoxa3 enhancers in vitro (Manzanares et al., 1997, 1999a), circumstantial evidence suggests the involvement of other dimerisation partners in vivo (Manzanares et al., 2002; F. MechtaGrigoriou, F. Giudicelli, C. Pujades, P. Charnay, and M. Yassito, unpublished observations). The segmentation phenotype associated with Mafb lossof-function mutations impeded detailed analysis of its mode of action, leaving unanswered questions regarding which aspects of hindbrain patterning Mafb actually controls. To further investigate Mafb function, we have turned to ectopic expression experiments by electroporation in the chick neural tube, an approach that has been very useful in the analysis of Krox20 involvement in hindbrain segmentation (Giudicelli et al., 2001). The present work has revealed novel activities for Mafb, which are likely to be involved in both proper segmentation and specification of regional identity. In addition, it has also indicated that these activities are restricted along the AP axis and are likely to be controlled by interactions with other factors. These observations provide a better understanding of the dual role of Mafb in hindbrain patterning.

MATERIALS AND METHODS Expression Constructs and Reporter Constructs The Mafb expression plasmid, pAdRSVMafb, was constructed by cloning of a 1.56-kb cDNA fragment containing the mouse Mafb coding sequence plus 590 bp of 3⬘ flanking sequence (Cordes and Barsh, 1994), into the pAdRSVSp plasmid, downstream to the RSV promoter enhanced by a type 5 adenovirus inverted terminal region. The vector also contains an artificial 5⬘ intron and the SV40 late polyadenylation signal sequence. pAdRSVFMafb differs from pAdRSVMafb by the insertion of a FLAG epitope coding sequence, 5⬘-GATTACAAGGATGACGACGATAAGCTA-3⬘, immediately after the Mafb initiation codon. pAdRSVMafbR22E and pAdRSVMafbL2PL4P were constructed by directed mutagenesis with the Exsite PCR site-directed mutagenesis kit (Stratagene) using the following oligonucleotides (mutated residues are in bold): 5⬘-TTTCCCGCTTCTGCTTCAGGCGG-3⬘ and 5⬘-CGTTGAAGAACCGGGGCTACGCC-3⬘ for R22E; 5⬘-GCTGCTCCACCTGCTGAATGGGCTGCGTCTTCTCG-3⬘ and 5⬘-TTAAGCAGGAGGTGTCCCGGCGGGCCCGCGAGAGAG-3⬘ for L2PL4P. For all constructs, cloning junctions were checked by sequencing, and molecular weights of the encoded flagged proteins were verified by immunoblotting performed on extracts of transfected chick embryos. The two reporter constructs contain the ␤-galactosidase coding sequence driven by a minimal human ␤-globin promoter, associated with an enhancer sequence derived from the following mouse genes: Hoxb3-r5 is a 637-bp BamHI/StuI fragment carrying the Hoxb3-r5 enhancer (construct #5 of Manzanares et al., 1997); EphA4-r3r5 is a 470-bp SacI/BglII fragment carrying the EphA4 r3/r5 enhancer (construct #10 of Theil et al., 1998). Control electroporations of a ␤-globin–lacZ enhancerless construct in the presence or absence of the various expression vectors used here never led to detectable ␤-galactosidase activity in more than three to four cells in the neural tube (not shown).

© 2003 Elsevier Science (USA). All rights reserved.

152

Giudicelli et al.

In Ovo Electroporation and ␤-Galactosidase Detection Electroporation was performed as previously described (Giudicelli et al., 2001), using commercial fertilised hens eggs (Morizeau) and a BTX820 electroporator (Quantum). The following parameters were used: 4 pulses of 25 V and 50 ms at a frequency of 1 Hz, except for experiments involving lacZ reporter constructs, where we used 10 –12 pulses of 23 V and 40 ms. In coelectroporation experiments, the reporter construct concentration was 0.5 ␮g/␮l, and the empty vector pAdRSVS p was used to achieve a final DNA concentration of 1 ␮g/␮l. For ␤-galactosidase detection, embryos were fixed in 4% paraformaldehyde for 40 min at 4°C, and ␤-galactosidase activity was revealed for 1 h at 30°C, using Bluogal (Sigma) as a substrate. In the experiments presented in Figs. 2C and 3C, the ratio between Mafb and GFP expression plasmids was 2:1, the total DNA concentration being 1 ␮g/␮l. Embryos electroporated with these constructs were processed sequentially for in situ hybridisation and anti-GFP immunostaining.

Whole-Mount Immunohistochemistry and Immunofluorescence Immunochemical detection of proteins was performed on dissected neural tubes by using the following primary antibodies and dilutions: rabbit polyclonal antibodies directed against Krox-20 (Babco; 1/200), EphA4 (Becker et al., 1995; 1/20,000), GFP (Molecular Probes; 1/500) and ␤-galactosidase (Cappel; 1/500), mouse monoclonal antibodies directed against EphA4 (Hirano et al., 1998; 1/20) or the FLAG epitope (Sigma M2; 1/500). The following secondary antibodies were used at a 1/200 dilution: peroxidasecoupled goat antibody directed against mouse IgG (Sigma), alkaline phosphatase-coupled goat antibody directed against mouse IgG (Vector), biotinylated donkey antibody directed against rabbit IgG (Amersham), biotinylated horse antibody directed against mouse IgG (Vector), Cy3-coupled goat antibody directed against mouse IgG (Interchim), or FITC-coupled donkey antibody directed against rabbit IgG (Jackson Labs). The biotinylated antibodies were detected by using streptavidin– horseradish peroxidase (Amersham; 1/300). Peroxidase activity was revealed with diaminobenzidine (Sigma; brown staining). Occasionally, the staining was enhanced with nickel ammonium (dark blue staining; Adams, 1981). Alkaline phosphatase activity was detected by using 4-nitroblue tetrazolium chloride/5-bromo 4-chloro 3-indolyl phosphate (NBT/BCIP, Roche; purple staining). In some cases, 2-(4-) 5-(4-nitrophenyl) 3-phenyltetrazolium chloride (INT/BCIP, Roche; orange/red staining) was used instead.

Mouse Breeding and Whole-Mount in Situ Hybridisation kr mutant mice were produced by mating kr/kr males with kr/⫹ females, thus generating only homozygous and heterozygous mutant embryos. The presence of the wild-type allele was detected by a polymerase chain reaction performed on DNA extracted from embryonic tissues, according to Frohman et al. (1993). Wholemount in situ hybridisation was performed essentially as described (Wilkinson and Nieto, 1993), using digoxigenin-labelled riboprobes. For double detection, one of the probes was labelled with fluorescein–UTP. Digoxigenin and fluorescein were then detected sequentially by using alkaline phosphatase-coupled antibodies (Roche; 1/2000). NBT/BCIP (purple) staining was always carried out

first, and the antibody was stripped in 0.1 M glycine–HCl, pH 2.2. The embryos were then incubated with the other antibody and stained with INT/BCIP (orange/red). The riboprobes were as follows: chick Hoxb1 (Guthrie et al., 1992), chick Krox-20 (Giudicelli et al., 2001), chick EphA4 (Sajjadi and Pasquale, 1993), chick Hoxa3, chick Hoxb3 (Grapin-Botton et al., 1995), chick Hoxa2 (Prince and Lumsden, 1994), quail Mafb (Eichmann et al., 1997), chick Wnt1 (Bally-Cuif and Wassef, 1994), mouse Mafb (Cordes and Barsh, 1994), and chick FoxD3 (Kos et al., 2001). The chick Mafb probe was transcribed from a 480-bp XbaI/SspI fragment derived from the 3⬘ untranslated region of chick Mafb (Kataoka et al., 1994), presenting no detectable sequence similarity to the mouse Mafb cDNA present in pAdRSVMafb.

RESULTS In order to perform gain-of-function studies in the chick embryo, we constructed a series of expression plasmids in which mouse Mafb expression was driven by an RSV/ adenovirus promoter/enhancer. This vector has been shown previously to lead to efficient expression in a large proportion of neuroepithelial cells after electroporation of the chick hindbrain region (Giudicelli et al., 2001). Several versions of the mouse Mafb coding sequence were inserted into this vector, respectively encoding the wild-type protein, a tagged version carrying an N-terminal FLAG epitope, a mutant (MafbR22E) carrying a basic to acidic substitution that prevents binding of the Maf protein to DNA, and a mutant (MafbL2PL4P) carrying two leucine-to-proline substitutions in the leucine zipper domain preventing dimerization and DNA binding (Kataoka et al., 1993). There was no phenotypic difference between embryos electroporated with the wild-type Mafb plasmid and those electroporated with its flagged version. In consequence, most of the experiments shown below were performed by using the latter. In order to express the exogenous Mafb in a time window similar to that of the endogenous gene, the embryos were electroporated just before formation of the rhombomeres, i.e., between stages HH8 and HH10 (Hamburger and Hamilton, 1951) and analysed 18 –24 h later, at stages HH14 –HH16.

Ectopic Mafb Expression Affects the Dorsal Hindbrain but Does Not Prevent Segmentation Immunodetection of the FLAG epitope revealed that, as expected, the exogenous protein was efficiently expressed and restricted to neuroectodermal and epidermal tissues of the electroporated side of the embryos (Fig. 1A). In electroporated regions, Mafb expression led to a smoothening of the neural tube (Figs. 1A and 1B). This effect required a fully functional Mafb protein, since it was not observed with the MafbR22E and MafbL2PL4P mutants (Figs. 1C and 1D), although these were expressed at even higher levels, as determined by Western blotting (Fig. 1E) and immunohistochemistry (see Figs. 3F and 3G; and data not shown). This argues against artefactual consequences due to overexpression or squelching of partners for heterodimerisation, since

© 2003 Elsevier Science (USA). All rights reserved.

153

Mafb Functions in Hindbrain Segmentation

FIG. 1. Ectopic expression of Mafb leads to morphological modifications of the neural tube. (A0-D) Dorsal views of HH13–HH14 chick embryos electroporated at HH8 –HH9 with pAdRSVFMafb (A, B), pAdRSVMafbR22E (C), and pAdRSVMafbL2PL4P (D). The embryo in (A) has been treated by immunochemistry to follow the expression of the flagged exogenous protein. Note in (A) and (B) the reduction of neural constrictions. (E) Western blotting performed on protein extracts from electroporated embryos, showing the presence of the flagged proteins Mafb (lane 1), MafbR22E (lane 2), and MafbL2PL4P (lane 3). Lane 4 corresponds to nonelectroporated control embryos. The arrow indicates the expected position of the full-length proteins. Note that, although the protein is partially degraded in the case of the mutants, its full-length form is present in amounts higher than in the wild type situation. (F–H) Flat-mounted hindbrains from embryos electroporated with pAdRSVMafb (F, H) or pAdRSVMafbR22E (G) and in situ hybridised with Wnt1 (F, G) or Hoxa2 (H) probes. Note in (F) the thinner morphological appearance of the dorsal-most part of the electroporated side and the lower expression of Wnt1 in this region (arrow). Hoxa2 expression is not affected by Mafb ectopic expression, except in the most dorsal, thinner part of the alar plate, where it is reduced (arrow in H). Electroporation was on the left (A, F–H) or bottom (B–D) side.

MafbR22E could normally participate in the formation of such heterodimers. In addition to the shape modifications, a thinning of the neuroectoderm in the dorsal part of the neural tube was often observed with the functional Mafb protein (see below). Together, these observations could suggest an alteration of segmentation and an enlargement of the roof plate at the expense of the alar plate. Therefore, we analysed the expression of the roof plate marker Wnt1 (Bally-Cuif and Wassef, 1994) and examined morphological segmentation. We found that, despite the apparent thinning of the dorsal neuroectoderm (arrow), Wnt1 roof plate expression was not expanded, but rather repressed, upon Mafb ectopic expression (Fig. 1F). Similarly, the expression of FoxD3, a marker

of neural crest precursors (Kos et al., 2001), was also repressed upon Mafb ectopic expression (data not shown), suggesting that the reduction of dorsal neural tube is not due to increased crest production. This thinner dorsal region observed in electroporated embryos was found to be negative for the expression of all regional markers examined (Fig. 1H shows the case of Hoxa2; see also Figs. 2A, 3D, and 4H). Dorsal repression of Wnt1 and of other markers was not observed upon ectopic expression of the R22E mutant (Fig. 1G; and data not shown). Observation of flat-mounted embryos revealed the presence of a normal number of rhombomere boundaries, separating territories of the expected size (Figs. 1F and 1H). Therefore, segmentation of the hindbrain was not grossly affected by ectopic expres-

© 2003 Elsevier Science (USA). All rights reserved.

154

Giudicelli et al.

FIG. 2. Mafb ectopic expression affects the AP molecular identity in the hindbrain. (A–C, F) Flat-mounted hindbrains from embryos electroporated with pAdRSVMafb (A, F) or pAdRSVMafb together with a GFP expression vector (C) or pAdRSVMafbR22E (B) and in situ hybridised with Hoxa3 alone (A, B), or Hoxa3 (brown) and Krox20 (orange) (F) probes, or Hoxa3 (purple) combined with GFP immunostaining (orange) (C). Hoxa3 expression is up-regulated by Mafb mostly within r3, as indicated by the colabelling with Krox20 (F). The arrow in (C) indicates Hoxa3-expressing cells in r3. (D, E) Flat-mounted hindbrains from embryos electroporated with pAdRSVMafb (D) or pAdRSVMafbR22E (E) and in situ hybridised with a Hoxb1 probe. Expression of Hoxb1 is severely down-regulated by Mafb but not by MafbR22E. (G–I) Dorsal views of embryos coelectroporated with the Hoxb3 r5 enhancer-lacZ reporter, together with a control empty vector (G), a Mafb expression vector (H), or a MafbR22E expression vector (I). The reporter is up-regulated by Mafb in r3 and r5 (H), and repressed in r5 by MafbR22E (I). Electroporation is on the left (A–F) or bottom (G–I) side.

sion of Mafb. This was confirmed by the analysis of the expression pattern of several AP regional markers (see below), and particularly of Hoxa2, whose expression covers the entire hindbrain caudal to the r1/2 rhombomeric boundary (Prince and Lumsden, 1994): Mafb electroporation in the hindbrain did not affect Hoxa2 expression, except for its repression in the dorsalmost region (Fig. 1H); in particular, the rostral limit of expression of Hoxa2 at the r1/2 boundary and its upregulation in r3 appeared normal. In conclusion, although Mafb ectopic expression affects the formation of constrictions and alters the morphological character of the very dorsal region in the neural tube, it does not disrupt the formation of the rhombomeric territories.

Involvement of Mafb in the Specification of AP Regional Identity Since Mafb misexpression appeared not to affect the number and extent of rhombomeric territories, we investigated possible consequences on their molecular identity by following the expression of specific Hox genes. In mice, Mafb has been shown to be involved in the direct transcriptional activation of Hoxa3 in r5 and r6 and Hoxb3 in r5 (Manzanares et al., 1999b and references therein). Furthermore, ectopic expression of Mafb in r3 in transgenic mice

has been shown to lead to Hoxa3 induction in this rhombomere (Theil et al., 2002). We have therefore analysed the consequences of Mafb electroporation on Hoxa3 and Hoxb3 expression in the chick neural tube. Surprisingly, while the exogenous Mafb protein appeared widely expressed in the entire hindbrain (Figs. 1A and 3E), Mafb expression led to preferential ectopic activation of Hoxa3 in r3 (Figs. 2A and 2F; 21/37 embryos) and, much less frequently and to a lesser extent, in r4 or r2 (data not shown; 5/37 and 4/37 embryos, respectively). Coelectroporation of a green fluorescent protein (GFP)-expressing vector confirmed that Hoxa3 activation was mainly restricted to r3 even though other territories exhibited a large number of electroporated, GFPpositive cells (Fig. 2C). This suggested that another regionalized factor, not induced by Mafb, is required to restrict Hoxa3 activation to r3. Krox20 is a major determinant of r3 and r5 identity (Schneider-Maunoury et al., 1993; Giudicelli et al., 2001) and could constitute a candidate for such a factor. However, coelectroporation of Mafb and Krox20 expression vectors did not significantly extend the domain of expression of Hoxa3 as compared with Mafb alone (data not shown). Finally, electroporation of MafbR22E had no effect on Hoxa3 expression (Fig. 2B), indicating that the activation of Hoxa3 in r3 requires the DNA binding activity of Mafb.

© 2003 Elsevier Science (USA). All rights reserved.

155

Mafb Functions in Hindbrain Segmentation

FIG. 3. Mafb autoregulation and repression of EphA4. (A–I) Flat-mounted hindbrains from embryos electroporated with pAdRSVMafb (A, D, F, H, I), or pAdRSVMafb together with a GFP expression vector (C) or pAdRSVMafbR22E (B, E, G). The expression of the indicated markers was analysed by in situ hybridisation or immunochemistry [chick Mafb alone (A, B) or combined with GFP immunostaining (orange) (C), Krox20 (D, E), EphA4 and flagged-Mafb (F, G), EphA4 (H), and EphA4 and Krox20 (I)]. Endogenous chick Mafb is up-regulated by exogenous Mafb in a territory strictly limited rostrally by the r2/r3 boundary (A, C). EphA4 expression is severely repressed by Mafb, whereas only minor modifications of the Krox20 pattern are observed (D, F, H, I); arrowhead in (D) indicates the few Krox20-positive cells in r6. (J–L) Dorsal views of embryos coelectroporated with the EphA4 r3/r5 enhancer-lacZ reporter and the empty vector (J), the Mafb expression vector (K), or the MafbR22E expression vector (L). Mafb represses the activity of the enhancer, whereas MafbR22E has no effect. Electroporation was on the left (A–I) or bottom (J–L) side. The lower level of EphA4 staining at the electroporated side of r5 in (F, G) was often observed following electroporation of any expression vector and is therefore likely to represent an electroporation artefact. The reddish/brown colour in (I) denotes a double staining for EphA4 (orange) and Krox20 (purple) (the resulting colour is the same in r3/r5 on the control side).

In contrast to Hoxa3, the expression of the endogenous Hoxb3 gene was not affected by either ectopic expression of Mafb or coelectroporation of Mafb and Krox20 (data not shown). However, Mafb can activate a mouse Hoxb3 r5 enhancer in Krox20-positive territories in coelectroporated embryos (Manzanares et al., 2002; Figs. 2G and 2H). This discrepancy may be related to the very low level of endogenous Hoxb3 expression in r5 in the chick (Grapin-Botton et al., 1995), which may make ectopic up-regulation at an r5-like level almost undetectable. Interestingly, coelectroporation of the mouse Hoxb3 r5 enhancer with the MafbR22E expression vector, not only did not lead to ectopic activation of the reporter in r3, but repressed its basal expression in r5 (Fig. 2I). This effect was not observed with an empty vector and suggests that the DNA bindingdeficient Mafb protein can sequester endogenous factors required for r5-specific reporter gene expression.

Hoxb1 is a major determinant of r4 identity (Studer et al., 1996). It is initially expressed in the neural plate at prerhombomeric stages in a domain extending rostrally up to the prospective r3/r4 boundary, subsequently upregulated in pr4 and down-regulated in pr5/6, and maintained in r4 throughout segmentation (Sundin and Eichele, 1990). Mafb expression in pr5/6 in chick and mouse slightly precedes the extinction of Hoxb1 expression in these territories (Eichmann et al., 1997). In addition, in kreisler and valentino mutants, a caudal extension of the territory of Hoxb1 expression was observed (Frohman et al., 1993; Prince et al., 1998). These data raise the possibility that Mafb could be involved in the transcriptional repression of Hoxb1. Indeed, ectopic expression of Mafb, but not of MafbR22E, led to dramatic repression of Hoxb1 expression, both in r4 and in more caudal territories (Figs. 2D and 2E).

© 2003 Elsevier Science (USA). All rights reserved.

156

Giudicelli et al.

In conclusion, we have shown that Mafb is likely to play an important role in the specification of AP regional identity in the caudal hindbrain, being capable of inducing the expression of r5/6-specific genes (Hoxa3, possibly Hoxb3) while repressing at least one r4-specific gene (Hoxb1). In contrast, it seems not capable of modifying the expression of Hoxa2, the most rostrally expressed Hox gene. Mafb function is likely to involve regionalized coregulators as suggested by the partial restriction of Hoxa3 activation to r3.

Mafb Affects the Regulation and the Activity of Segmentation Genes Despite the absence of obvious effects of Mafb ectopic expression on hindbrain segmentation, we investigated the consequences on the expression of two segmentation genes, Mafb itself and Krox20. To analyse the expression of the endogenous Mafb gene, we generated a probe complementary to the 3⬘-most part of the chicken Mafb transcript, which does not cross-hybridize with the exogenous mouse mRNA. In most embryos electroporated with the mouse Mafb expression vector, ectopic expression of chicken Mafb was observed in r3 and r4 and caudally to r6 (Fig. 3A). In contrast, no ectopic expression of the endogenous gene was ever observed in r2 or more rostrally (Fig. 3A), although these territories could be electroporated very efficiently (GFP-positive cells in Fig. 3C). The R22E mutation abolished this autoregulatory activity (Fig. 3B). These data indicate that Mafb can positively autoregulate its own expression and suggest that this autoregulation is dependent on other factors which are differentially distributed on both sides of the r2/r3 boundary. Krox20 is expressed in r3 and r5 and plays a dual function in both segmentation of the hindbrain and specification of odd- vs even-numbered rhombomere identity (SchneiderMaunoury et al., 1993, 1997; Giudicelli et al., 2001; Voiculescu et al., 2001). The segmental pattern of expression of Krox20 was not grossly affected by ectopic expression of Mafb (Fig. 3D). However, Krox20-positive cells appeared sparser than normal in r3 and r5, and the boundaries of these rhombomeres were not straight. In addition, a few Krox20-expressing cells were observed in even-numbered rhombomeres (Fig. 3D, arrowhead). These effects were not observed with the mutant versions of Mafb (Fig. 3E; and data not shown). Because of the role of Krox20 in the extension of its own expression and in the establishment of a segregation between even- and odd-numbered rhombomere cells (Giudicelli et al., 2001; Voiculescu et al., 2001), the consequences of Mafb misexpression on the Krox20 pattern might be explained by an antagonistic action on Krox20 activity. To investigate this possibility, we examined the expression of EphA4, which constitutes a direct target of Krox20 (Theil et al., 1998) and is involved in the restriction of cell movement between adjacent rhombomeres (Xu et al., 1999). Mafb ectopic expression led to a dramatic repression of EphA4 in r3 and r5, observed both at the protein (Fig. 3F) and mRNA (Fig. 3H) levels. Further-

more, when Krox20 and Mafb expression vectors were coelectroporated, no ectopic activation of EphA4 was observed in even-numbered rhombomeres (data not shown), in contrast to the electroporation of Krox20 alone (Giudicelli et al., 2001). Double-staining experiments indicated that the repression of EphA4 was not a secondary consequence of a disappearance of Krox20, since numerous EphA4negative cells were Krox20-positive (Fig. 3I). Control electroporations with the R22E or L2PL4P mutant versions of Mafb did not affect the EphA4 expression pattern (Fig. 3G; and data not shown). A minimal enhancer responsible for the expression of mouse EphA4 in r3 and r5 has been identified (Theil et al., 1998). We tested the activity of this enhancer driving the lacZ gene by electroporation in the chick neural tube. This led to strong expression in r3 and much weaker expression in r5 (Fig. 3J; 12/17 embryos), a pattern similar to its activity in transgenic mice, although in the mouse the difference between r3 and r5 was less dramatic (Theil et al., 1998). Coelectroporation of the EphA4 r3/5 enhancer reporter with the Mafb expression vector led to a marked reduction of the expression of the reporter (Fig. 3K; no lacZ expression in 8/20 embryos, strongly reduced expression in 9/20 embryos). In contrast, coelectroporation with the MafbR22E mutant left the reporter expression pattern unchanged (Fig. 3L; 7/10 embryos). In conclusion, these data indicate that Mafb, at least in conditions of misexpression, is able to repress EphA4 expression, correlating with defects observed in cell segregation and in the establishment of straight rhombomere boundaries.

Mafb Acts in a Cell-Autonomous Manner We have previously shown that Krox20 exerts part of its activity in a non cell-autonomous manner (Giudicelli et al., 2001). This raised the possibility that this could also be the case for Mafb. To address this issue, we constructed a vector allowing the expression of a bicistronic mRNA, the first cistron encoding ␤-galactosidase and the second the flagged Mafb, separated by an internal ribosome entry site (IRES). This design was chosen to allow the identification of electroporated cells which may have lost or failed to express the exogenous Mafb at a detectable level, through the sensitive detection of the relatively stable ␤-galactosidase protein. Accordingly, double immunofluorescence detection of ␤-galactosidase and flagged Mafb in embryos electroporated with the lacZ-IRES-Mafb construct showed that essentially all flag-positive cells were also ␤-galactosidasepositive, while the reverse was not true (Figs. 4A– 4C), although every ␤-galactosidase-positive cell necessarily has expressed the transcript encoding Mafb. This confirmed that this method allows the detection of cells expressing even very low levels of the protein of interest. Subsequently, hindbrains electroporated with the same expression vector were analysed by double immunofluorescence for ␤-galactosidase and EphA4 (Figs. 4D– 4F). If non-

© 2003 Elsevier Science (USA). All rights reserved.

157

Mafb Functions in Hindbrain Segmentation

FIG. 4. Cell autonomy of Mafb activity. (A–F) Confocal images of rhombomere 3 from flat-mounted hindbrains from embryos electroporated with pAdRSVlacZIRESMafb (A–F) and treated for immunofluorescence with antibodies directed against the FLAG epitope of Mafb (red, A, C), EphA4 (red, D, F), and ␤-galactosidase (green, B, C, E, F). (C) and (F) present a superposition of red and green labellings. The ␤-galactosidase staining allows detection of essentially all cells expressing the flagged Mafb (A–C) and EphA4 repression is observed only in electroporated cells (␤-galactosidase-positive, D–F). (G, H) Flat-mounted hindbrains from embryos electroporated with pAdRSVMafb, and analysed by double in situ hybridisation with EphA4 and quail Mafb (G), and Hoxa3 and EphA4 (H) probes. The quail Mafb probe recognizes both endogenous chick and exogenous mouse transcripts. Electroporation was on the left side.

cell-autonomous repression had occurred, EphA4- and ␤-galactosidase-negative cells would have been expected within odd-numbered rhombomeres. In contrast, all EphA4-negative cells in r3 were found to be positive for ␤-galactosidase (Figs. 4D– 4F; and data not shown). Control experiments showed that, when a ␤-galactosidase expression vector (Giudicelli et al., 2001) was electroporated, no effects were detected on the EphA4 staining (data not shown). Due to the absence of appropriate antibodies, it was not possible to perform a similar experiment to directly address the coincidence between electroporated cells and those ectopically expressing endogenous Mafb or Hoxa3. However, we could demonstrate that, in r3, ectopic activation of endogenous Mafb or Hoxa3 was restricted to cells that had down-regulated EphA4 (Figs. 4G and 4H). Taken together, these data suggest that Mafb acts in a cell-autonomous manner.

Mafb Expression in the kr Mutant The induction of Hoxa3 expression in r3 following Mafb ectopic expression is unexpectedly reminiscent of its expression in the mouse kreisler mutant (McKay et al., 1994). Actually, r3 is modified in several aspects in this mutant, since besides Hoxa3, FGF3 expression is also induced and the period of expression of Krox20 is extended (Frohman et al., 1993; McKay et al., 1994). Furthermore, a Mafb-

controlled r5/r6 enhancer element from the Hoxa3 gene is also active in r3 in kr mutant embryos (Manzanares et al., 1999b). These data are consistent with the acquisition of r5 characteristics by r3. We have therefore reinvestigated the expression of Mafb in kr mice. Previous studies have shown that Mafb is not expressed rostrally to r5 in wild-type embryos (Cordes and Barsh, 1994; Mechta-Grigoriou et al., unpublished observations). In contrast, detailed in situ hybridisation analyses performed on kr/⫹ (Figs. 5A–5C) and kr/kr (Figs. 5D–5H) embryos at various stages between 8.0 and 9.0 dpc, while confirming the absence of Mafb expression in the r5/r6 region in homozygous mutant embryos, revealed the existence of a more rostral, ectopic domain of expression in both heterozygous and homozygous mutants. This latter domain is likely to correspond to pr3 and then to r3, according to its relative position vs the Mafb-positive r5/r6 domain in heterozygotes and to rhombomere boundaries at later stages (the r3/r4 boundary is the caudal-most boundary in kr/kr mutants: dashed line in Figs. 5F and 5G). Expression in pr3 in homozygous mutants appeared at approximately the same stage as in r5/r6 in wild-type or heterozygous animals, but the level was lower (Figs. 5A and 5D) and was down-regulated earlier, by the 20 s stage (Figs. 5G and 5H). The level of expression in the r3 domain was even lower in heterozygous animals and could only be detected in some embryos around the 10 s stage (Fig. 5B). After the 20 s stage, weak Mafb expression was detected in the ventral rX domain in kr/kr embryos (Fig. 5H), resem-

© 2003 Elsevier Science (USA). All rights reserved.

158

Giudicelli et al.

FIG. 5. Mafb expression in r3 in the kr mutant. Flat-mounted hindbrains from mouse embryos heterozygous (A–C) or homozygous (D–H) for the kr mutation, analysed by in situ hybridisation for Mafb expression at the indicated somite (s) stage. The positions of prospective rhombomeres (pr), rhombomeres (r), and rhombomere boundaries (dashed lines) are indicated. From 4 to 18 s, all kr/kr embryos show expression of Mafb in the pr3-r3 region (D–G). A similar domain of expression is also observed, although much fainter, in some heterozygous embryos (arrowhead in B).

bling the expression pattern in wild-type animals at this stage (Mechta-Grigoriou et al., unpublished observations), suggesting that this late phase of expression is not abolished in kr/kr mutants. These data indicate that, in contrast to the conclusions from previous studies, Mafb is expressed in r3 from the kr allele. This accounts for the observed ectopic expression of several r5 markers in r3 in kr mutant embryos and could also be related to the late phenotype of both kr/⫹ and kr/kr embryos in r3-derived nuclei of the brainstem (Chatonnet et al., 2002).

DISCUSSION In this study, we have performed gain-of-function experiments designed to refine our view of the role of Mafb in hindbrain patterning. Our analyses have revealed the complexity of the consequences of Mafb expression: while bringing additional support to data derived from different approaches, they identify novel Mafb activities that are likely to be controlled by spatially restricted interacting factors.

© 2003 Elsevier Science (USA). All rights reserved.

159

Mafb Functions in Hindbrain Segmentation

Complexity of the Molecular Response to Mafb Expression We have shown that Mafb can affect the expression of a number of regulatory genes in the hindbrain. Mafb shares a positive autoregulatory capacity with other transcription factors involved in early hindbrain patterning [Hoxb1 (Studer et al., 1998) and Krox20 (Giudicelli et al., 2001)]. This activity is likely to be essential for the maintenance of Mafb expression during hindbrain development, as suggested by the analysis of the valentino zebrafish mutant (Moens et al., 1998). The occurrence of autoregulatory mechanisms precisely for the genes that play a role in the early delimitation of hindbrain segments suggests that the signals responsible for the initial induction of their expression are very transient, while their function is required during an extended period to allow the stable formation of the territories. We have also demonstrated that Mafb is able to participate in the induction of Hoxa3 and in the activation of the mouse Hoxb3 r5 enhancer. These data are consistent with studies of mouse Hoxa3 r5/6 and Hoxb3 r5 enhancers, which have shown that they contain Mafb binding sites required for their activities (Manzanares et al., 1997, 1999a) and with analyses of the kreisler mutant which have indicated that absence of Mafb in the rX territory prevents normal Hoxa3 and Hoxb3 expression (Frohman et al., 1993; McKay et al., 1994). However, in contrast to recent mouse transgenic experiments involving Mafb ectopic expression in r3 (Theil et al., 2002), Mafb electroporation in the chick did not lead to activation of endogenous Hoxb3 in this rhombomere. Nevertheless, the data obtained by electroporation of the mouse Hoxb3 r5 enhancer reporter construct suggest that the transacting factors required for enhancer activation are present in the chick hindbrain. The chicken homolog of the Hoxb3 r5 enhancer has been shown to be specifically active in r5 in mouse transgenic embryos (Manzanares et al., 1997). In contrast, we have shown that, in the chick electroporation system, this latter enhancer is not restricted to r5, although it is still responsive to Mafb (data not shown). We therefore propose that the particular behaviour of the endogenous chick Hoxb3 gene could be explained by the existence of specific negative regulatory elements, whose activity is not recapitulated by the electroporation procedure. This hypothesis is consistent with the very low level of expression of Hoxb3 in r5 in the chick, where, in contrast to the mouse, no up-regulation is observed relative to more caudal regions (Grapin-Botton et al., 1995). Besides its capacity for inducing the expression of postotic markers, this study also revealed Mafb repressive activities. Mafb ectopic expression led to a dramatic repression of Hoxb1 in both r4 and its more caudal domains of expression. This activity does not involve Krox20, another repressor of Hoxb1 (Giudicelli et al., 2001), since Krox20 expression is not induced by Mafb in these territories. The physiological significance of this regulatory link in the

establishment of the Hoxb1 expression pattern is supported by two types of observations: (1) it is consistent with the timing of Mafb induction and Hoxb1 repression in r5 and r6 (Eichmann et al., 1997; Sundin and Eichele, 1990); (2) the kreisler mutation leads to a caudal extension of the Hoxb1 expression domain (Frohman et al., 1993). However, since Hoxb1 repression still occurs in the rX domain in kr/kr mutants, mechanisms involving neither Mafb nor Krox20, which is absent from rX (Frohman et al., 1993), must be involved. Similar mechanisms may be responsible for the down-regulation of Hoxb1 expression caudally to r6 in the mouse neural tube (Murphy and Hill, 1991). We found that Mafb can also repress EphA4. This is surprising since the two genes are normally coexpressed in r5 and suggests the existence of a concentration threshold for the repression of EphA4 by Mafb. EphA4 repression requires Mafb DNA-binding activity and is exerted at least in part through the Krox20-controlled EphA4 r3/r5 enhancer element. These data raise the possibility that Mafb has the capacity to antagonise Krox20 activity on this enhancer. Krox20 and Mafb have been shown to bind to the Hoxb3 r5 enhancer and to synergistically cooperate for transcriptional activation (Manzanares et al., 2002). It will therefore be interesting to determine whether the EphA4 r3/r5 enhancer contains Mafb binding sites and whether, depending on the context, Mafb can cooperate with or antagonise Krox20. Repression of EphA4 raises the possibility of a more general role of Mafb in antagonizing other Krox20 activities. In this respect, the sparser aspect of Krox20-expressing cells in r3 and r5 following Mafb electroporation (Fig. 3D) might reflect not only the disruption of the Eph/ephrin-mediated mechanism of cell segregation, but also the counteraction of the Krox20 activity required for the homogenisation of its own expression domain (Giudicelli et al., 2001; Voiculescu et al., 2001). Such an antagonism between Mafb and Krox20 was unsuspected and might play a role in the determination of r3 vs r5 identity (see below).

Spatial Control of Mafb Activity Possibly related to the multiplicity of the functions of Mafb and to the necessity of strictly controlling them, our work has revealed the existence of several spatial restrictions placed on Mafb activities within the hindbrain. We have shown that expression of Mafb can lead to ectopic activation of Hoxa3, but that it is largely restricted to r3. Coexpression of Krox20 does not extend this domain of ectopic expression, indicating that Krox20 is not the limiting factor. These data therefore suggest that factor(s), distinct from Krox20 and present in r3 and r5/6, are required together with Mafb for Hoxa3 induction or that factor(s) excluded from these rhombomeres prevent such induction. Our data also indicate that, while analyses of the kreisler and valentino mutants have shown that functional Mafb is required for Krox20 induction in r5 (Frohman et al., 1993; Moens et al., 1996), Mafb misexpression is not sufficient to

© 2003 Elsevier Science (USA). All rights reserved.

160

Giudicelli et al.

lead to Krox20 ectopic expression. This demonstrates that, although Mafb is required for Krox20 expression in r5, it does not have an instructive role in the definition of its domain of expression. This role is likely to be played by other factors, presumably regionalized, and able to cooperate with Mafb for Krox20 induction. Finally, the capacity of autoregulation of Mafb is precisely regulated at the r2/r3 boundary, no induction being observed rostrally. Therefore, this capacity is likely to be controlled by other factors that are differentially distributed on both sides of the r2/r3 boundary. In conclusion, Mafb participates in either up- or downregulation of various target genes, and these activities appear to be spatially restricted. Considering the large variety of factors able to dimerise with Mafb, an attractive hypothesis is that Mafb takes part in heterodimeric complexes to regulate the expression of its target genes, the spatial distribution of the partners defining the domains of activity of Mafb.

Role of Mafb in Hindbrain Patterning Mafb is initially expressed in the hindbrain in a narrow transverse territory (at around 0 s and 5 s stages in the mouse and the chick, respectively), which then extends to correspond to prospective r5 and r6 (Mechta-Grigoriou et al., unpublished observations; Eichmann et al., 1997). Once initiated, Mafb expression is presumably maintained in a cell-autonomous manner by the autoregulatory activity identified in this study. At this stage, Mafb plays an essential role in the segmentation of the caudal hindbrain: (1) it is required together with other factor(s) to induce Krox20 and to restrict its expression to r5. In the absence of Krox20, r5 is not formed (Voiculescu et al., 2001), and therefore the elimination of r5 in the kreisler mutant could be a consequence of the inability to induce Krox20. Considering the involvement of Krox20 activity in the extension of its own domain of expression (Giudicelli et al., 2001; Voiculescu et al., 2001), the capacity of Mafb to antagonise Krox20 activity could also be involved in limiting the extension of the r5 Krox20 domain. (2) Both Mafb (this study) and Krox20 (Giudicelli et al., 2001) repress Hoxb1, while Hoxb1 and Hoxa1 have been proposed to repress Krox20 (Barrow et al., 2000). These mutually repressive activities, combined with the positive feedback loops associated with each of the three genes, may act as a switch involved in cell fate choice and therefore in the precise positioning of the r4/r5 boundary and possibly of the r6/r7 interface. In parallel to its involvement in the delimitation of r5 and r6, Mafb also plays a role in the acquisition of the AP identities of these rhombomeres, in part through the upregulation of group 3 paralogous Hox genes. In its absence, although r6 retains part of its characteristics, it is altered both in terms of expression of molecular markers and neurogenesis (Manzanares et al., 1999b). The case of r5 is more difficult to investigate since this rhombomere is

eliminated in the kreisler mutant. However, this work and other studies (Manzanares et al., 2002) indicate that the interplay between Krox20 and Mafb, involving both synergistic and antagonistic interactions, plays an essential role in the specification of r5 identity. In particular, Mafb may be responsible for specifying the differences between r5 and r3, in part by modulating the activity of Krox20. In conclusion, like Krox20, Mafb is involved in multiple steps in hindbrain patterning, both in proper segmentation and in acquisition of AP identity. This confirms the largely intertwined character of these two aspects of vertebrate hindbrain morphogenesis.

Mafb Regulation and the kr Mutation Our finding that Mafb is ectopically expressed in r3 in the kr mutant provides an explanation for the patterning and developmental defects previously observed in this rhombomere (Frohman et al., 1993; McKay et al., 1994; Chatonnet et al., 2002). In addition, it raises an interesting question about the molecular basis of this ectopic induction. The expression in r3 may in principle be a non cell-autonomous secondary consequence of the loss of expression in the r5/r6 region. However, this appears very unlikely considering the distance between the two territories and the observation of ectopic r3 expression, although weaker, in heterozygous mutants as well. We therefore favour the idea that both loss of expression in r5/r6 and induction in r3 are direct consequences of the chromosomal inversion in the Mafb locus. Furthermore, considering that Mafb expression is not modified by the kr mutation in any site other than the hindbrain (Eichmann et al., 1997), it may be that the two phenomena are linked and that the kr mutation actually results in a rostral shift of Mafb expression in the hindbrain. Since Mafb expression is responsive to retinoic acid (RA) in a dose-dependent manner, with low levels inducing it and high levels leading to repression (Grapin-Botton et al., 1998), it will be interesting to investigate whether the effect of the kr chromosomal rearrangement may be to modify the sensitivity of Mafb to RA.

ACKNOWLEDGMENTS We thank J. Ghislain, C. Goridis, S. Schneider-Maunoury, C. Vesque, and M. Wassef for critical reading of the manuscript. We are grateful to M. Nishizawa, M. Sieweke, D. Wilkinson, A. Grapin-Botton, H. Tanaka, V. Prince, R. Krumlauf, G. Barsh, C. Erickson, and M. Wassef for generous gifts of plasmids and antibodies and to J. Lewis for the kr mice. We acknowledge the expert contribution of C. Lefevre and her colleagues of the ENS animal house and the contribution of V. Jacob to the construction of the Mafb mutants. F.G. was supported by fellowships from MENRT, LNCC, and Institut Lilly. This work was supported by grants from INSERM, MENRT, EC, ARC, and AFM.

© 2003 Elsevier Science (USA). All rights reserved.

161

Mafb Functions in Hindbrain Segmentation

REFERENCES Adams, J. C. (1981). Heavy metal intensification of DAB-based HRP reaction product. J. Histochem. Cytochem. 29, 775. Bally-Cuif, L., and Wassef, M. (1994). Ectopic induction and reorganization of Wnt-1 expression in quail/chick chimeras. Development 120, 3379 –3394. Barrow, J. R., Stadler, H. S., and Capecchi, M. R. (2000). Roles of Hoxa1 and Hoxa2 in patterning the early hindbrain of the mouse. Development 127, 933–944. Becker, N., Gilardi-Hebenstreit, P., Seitanidou, T., Wilkinson, D., and Charnay, P. (1995). Characterisation of the Sek-1 receptor tyrosine kinase. FEBS Lett. 368, 353–357. Bell, E., Wingate, R. J., and Lumsden, A. (1999). Homeotic transformation of rhombomere identity after localized Hoxb1 misexpression. Science 284, 2168 –2171. Blank, V., and Andrews, N. C. (1997). The Maf transcription factors: Regulators of differentiation. Trends Biochem. Sci. 22, 437– 441. Chatonnet, F., del Toro, E. D., Voiculescu, O., Charnay, P., and Champagnat, J. (2002). Different respiratory control systems are affected in homozygous and heterozygous kreisler mutant mice. Eur. J. Neurosci. 15, 684 – 692. Chisaka, O., Musci, T. S., and Capecchi, M. R. (1992). Developmental defects of the ear, cranial nerves and hindbrain resulting from targeted disruption of the mouse homeobox gene Hox-1.6. Nature 355, 516 –520. Clarke, J. D., and Lumsden, A. (1993). Segmental repetition of neuronal phenotype sets in the chick embryo hindbrain. Development 118, 151–162. Cordes, S. P., and Barsh, G. S. (1994). The mouse segmentation gene kr encodes a novel basic domain-leucine zipper transcription factor. Cell 79, 1025–1034. Davenne, M., Maconochie, M. K., Neun, R., Pattyn, A., Chambon, P., Krumlauf, R., and Rijli, F. M. (1999). Hoxa2 and Hoxb2 control dorsoventral patterns of neuronal development in the rostral hindbrain. Neuron 22, 677– 691. Deol, M. S. (1964). The abnormalities of the inner ear in kreisler mice. J. Embryol. Exp. Morphol. 12, 475– 490. Dupe, V., and Lumsden, A. (2001). Hindbrain patterning involves graded responses to retinoic acid signalling. Development 128, 2199 –2208. Eichmann, A., Grapin-Botton, A., Kelly, L., Graf, T., Le Douarin, N. M., and Sieweke, M. (1997). The expression pattern of the mafB/kr gene in birds and mice reveals that the kreisler phenotype does not represent a null mutant. Mech. Dev. 65, 111–122. Fraser, S., Keynes, R., and Lumsden, A. (1990). Segmentation in the chick embryo hindbrain is defined by cell lineage restrictions. Nature 344, 431– 435. Frohman, M. A., Martin, G. R., Cordes, S. P., Halamek, L. P., and Barsh, G. S. (1993). Altered rhombomere-specific gene expression and hyoid bone differentiation in the mouse segmentation mutant, kreisler (kr). Development 117, 925–936. Gavalas, A., Davenne, M., Lumsden, A., Chambon, P., and Rijli, F. M. (1997). Role of Hoxa-2 in axon pathfinding and rostral hindbrain patterning. Development 124, 3693–3702. Giudicelli, F., Taillebourg, E., Charnay, P., and Gilardi-Hebenstreit, P. (2001). Krox-20 patterns the hindbrain through both cellautonomous and non cell-autonomous mechanisms. Genes Dev. 15, 567–580. Graham, A., and Lumsden, A. (1996). Patterning the cranial neural crest. Biochem. Soc. Symp. 62, 77– 83.

Grammatopoulos, G. A., Bell, E., Toole, L., Lumsden, A., and Tucker, A. S. (2000). Homeotic transformation of branchial arch identity after Hoxa2 overexpression. Development 127, 5355–5365. Grapin-Botton, A., Bonnin, M. A., McNaughton, L. A., Krumlauf, R., and Le Douarin, N. M. (1995). Plasticity of transposed rhombomeres: Hox gene induction is correlated with phenotypic modifications. Development 121, 2707–2721. Grapin-Botton, A., Bonnin, M. A., Sieweke, M., and Le Douarin, N. M. (1998). Defined concentrations of a posteriorizing signal are critical for Mafb/Kreisler segmental expression in the hindbrain. Development 125, 1173–1181. Guthrie, S., Muchamore, I., Kuroiwa, A., Marshall, H., Krumlauf, R., and Lumsden, A. (1992). Neuroectodermal autonomy of Hox-2.9 expression revealed by rhombomere transpositions. Nature 356, 157–159. Guthrie, S. (1996). Patterning the hindbrain. Curr. Opin. Neurobiol. 6, 41– 48. Hamburger, V., and Hamilton, H. L. (1951). A series of normal stages in the development of the chick embryo. J. Morphol. 88, 49 –92. Hirano, S., Tanaka, H., Ohta, K., Norita, M., Hoshino, K., Meguro, R., and Kase, M. (1998). Normal ontogenic observations on the expression of Eph receptor tyrosine kinase, Cek8, in chick embryos. Anat Embryol. 197, 187–197. Irving, C., and Mason, I. (2000). Signalling by FGF8 from the isthmus patterns anterior hindbrain and establishes the anterior limit of Hox gene expression. Development 127, 177–186. Kataoka, K., Fujiwara, K. T., Noda, M., and Nishizawa, M. (1994). Mafb, a new Maf family transcription activator that can associate with Maf and Fos but not with Jun. Mol. Cell. Biol. 14, 7581– 7591. Kataoka, K., Nishizawa, M., and Kawai, S. (1993). Structurefunction analysis of the maf oncogene product, a member of the b-Zip protein family. J. Virol. 67, 2133–2141. Keynes, R., Cook, G., Davies, J., Lumsden, A., Norris, W., and Stern, C. (1990). Segmentation and the development of the vertebrate nervous system. J. Physiol. 84, 27–32. Kos, R., Reedy, M. V., Johnson, R. L., and Erickson, C. A. (2001). The winged-helix transcription factor FoxD3 is important for establishing the neural crest lineage and repressing melanogenesis in avian embryos. Development 128, 1467–1479. Lufkin, T., Dierich, A., LeMeur, M., Mark, M., and Chambon, P. (1991). Disruption of the Hox-1.6 homeobox gene results in defects in a region corresponding to its rostral domain of expression. Cell 66, 1105–1119. Lumsden, A. (1990). The cellular basis of segmentation in the developing hindbrain. Trends Neurosci. 13, 329 –335. Lumsden, A., and Guthrie, S. (1991). Alternating patterns of cell surface properties and neural crest cell migration during segmentation of the chick hindbrain. Dev. Suppl., 9 –15. Lumsden, A., and Krumlauf, R. (1996). Patterning the vertebrate neuraxis. Science 274, 1109 –1115. Manzanares, M., Cordes, S., Kwan, C. T., Sham, M. H., Barsh, G. S., and Krumlauf, R. (1997). Segmental regulation of Hoxb-3 by kreisler. Nature 387, 191–195. Manzanares, M., Cordes, S., Ariza-McNaughton, L., Sadl, V., Maruthainar, K., Barsh, G., and Krumlauf, R. (1999a). Conserved and distinct roles of kreisler in regulation of the paralogous Hoxa3 and Hoxb3 genes. Development 126, 759 –769. Manzanares, M., Trainor, P. A., Nonchev, S., Ariza-McNaughton, L., Brodie, J., Gould, A., Marshall, H., Morrison, A., Kwan, C. T., Sham, M. H., Wilkinson, D. G., and Krumlauf, R. (1999b). The

© 2003 Elsevier Science (USA). All rights reserved.

162

Giudicelli et al.

role of kreisler in segmentation during hindbrain development. Dev. Biol. 211, 220 –237. Manzanares, M., Nardelli, J., Gilardi-Hebenstreit, P., Marshall, H., Giudicelli, F., Martinez-Pastor, M. T., Krumlauf, R., and Charnay, P. (2002). Krox20 and kreisler co-operate in the transcriptional control of segmental expression of Hoxb3 in the developing hindbrain. Embo J. 21, 365–376. McKay, I. J., Muchamore, I., Krumlauf, R., Maden, M., Lumsden, A., and Lewis, J. (1994). The kreisler mouse: A hindbrain segmentation mutant that lacks two rhombomeres. Development 120, 2199 –2211. Moens, C. B., Yan, Y. L., Appel, B., Force, A. G., and Kimmel, C. B. (1996). valentino: A zebrafish gene required for normal hindbrain segmentation. Development 122, 3981–3990. Moens, C. B., Cordes, S. P., Giorgianni, M. W., Barsh, G. S., and Kimmel, C. B. (1998). Equivalence in the genetic control of hindbrain segmentation in fish and mouse. Development 125, 381–391. Murphy, P., and Hill, R. E. (1991). Expression of the mouse labial-like homeobox-containing genes, Hox 2.9 and Hox 1.6, during segmentation of the hindbrain. Development 111, 61–74. Niederreither, K., Vermot, J., Schuhbaur, B., Chambon, P., and Dolle´ , P. (2000). Retinoic acid synthesis and hindbrain patterning in the mouse embryo. Development 127, 75– 85. Pourquie, O. (2001). Vertebrate somitogenesis. Annu. Rev. Cell Dev. Biol. 17, 311–350. Prince, V., and Lumsden, A. (1994). Hoxa-2 expression in normal and transposed rhombomeres: Independent regulation in the neural tube and neural crest. Development 120, 911–923. Prince, V. E., Moens, C. B., Kimmel, C. B., and Ho, R. K. (1998). Zebrafish hox genes: Expression in the hindbrain region of wild-type and mutants of the segmentation gene, valentino. Development 125, 393– 406. Sajjadi, F. G., and Pasquale, E. B. (1993). Five novel avian Ephrelated tyrosine kinases are differentially expressed. Oncogene 8, 1807–1813. Schneider-Maunoury, S., Topilko, P., Seitandou, T., Levi, G., Cohen-Tannoudji, M., Pournin, S., Babinet, C., and Charnay, P. (1993). Disruption of Krox-20 results in alteration of rhombomeres 3 and 5 in the developing hindbrain. Cell 75, 1199 – 1214. Schneider-Maunoury, S., Seitanidou, T., Charnay, P., and Lumsden, A. (1997). Segmental and neuronal architecture of the hindbrain of Krox-20 mouse mutants. Development 124, 1215– 1226. Schneider-Maunoury, S., Gilardi-Hebenstreit, P., and Charnay, P. (1998). How to build a vertebrate hindbrain. Lessons from genetics. C. R. Acad. Sci. III 321, 819 – 834. Sieweke, M. H., Tekotte, H., Frampton, J., and Graf, T. (1996). Mafb is an interaction partner and repressor of Ets-1 that inhibits erythroid differentiation. Cell 85, 49 – 60.

Studer, M., Gavalas, A., Marshall, H., Ariza-McNaughton, L., Rijli, F. M., Chambon, P., and Krumlauf, R. (1998). Genetic interactions between Hoxa1 and Hoxb1 reveal new roles in regulation of early hindbrain patterning. Development 125, 1025–1036. Studer, M., Lumsden, A., Ariza-McNaughton, L., Bradley, A., and Krumlauf, R. (1996). Altered segmental identity and abnormal migration of motor neurons in mice lacking Hoxb-1. Nature 384, 630 – 634. Sundin, O. H., and Eichele, G. (1990). A homeo domain protein reveals the metameric nature of the developing chick hindbrain. Genes Dev. 4, 1267–1276. Swiatek, P. J., and Gridley, T. (1993). Perinatal lethality and defects in hindbrain development in mice homozygous for a targeted mutation of the zinc finger gene Krox20. Genes Dev. 7, 2071– 2084. Theil, T., Ariza-McNaughton, L., Manzanares, M., Brodie, J., Krumlauf, R., and Wilkinson, D. G. (2002). Requirement for downregulation of kreisler during late patterning of the hindbrain. Development 129, 1477–1485. Theil, T., Frain, M., Gilardi-Hebenstreit, P., Flenniken, A., Charnay, P., and Wilkinson, D. G. (1998). Segmental expression of the EphA4 (Sek-1) receptor tyrosine kinase in the hindbrain is under direct transcriptional control of Krox-20. Development 125, 443– 452. Trainor, P. A., and Krumlauf, R. (2001). Hox genes, neural crest cells and branchial arch patterning. Curr. Opin. Cell Biol. 13, 698 –705. Voiculescu, O., Taillebourg, E., Pujades, C., Kress, C., Buart, S., Charnay, P., and Schneider-Maunoury, S. (2001). Hindbrain patterning: Krox20 couples segmentation and specification of regional identity. Development 128, 4967– 4978. Wassarman, K. M., Lewandoski, M., Campbell, K., Joyner, A. L., Rubenstein, J. L., Martinez, S., and Martin, G. R. (1997). Specification of the anterior hindbrain and establishment of a normal mid/hindbrain organizer is dependent on Gbx2 gene function. Development 124, 2923–2934. Wilkinson, D. G., and Nieto, M. A. (1993). Detection of messenger RNA by in situ hybridization to tissue sections and whole mounts. Methods Enzymol. 225, 361–373. Wilkinson, D. G. (1995). Genetic control of segmentation in the vertebrate hindbrain. Perspect. Dev. Neurobiol. 3, 29 –38. Xu, Q., Mellitzer, G., Robinson, V., and Wilkinson, D. G. (1999). In vivo cell sorting in complementary segmental domains mediated by Eph receptors and ephrins. Nature 399, 267–271.

© 2003 Elsevier Science (USA). All rights reserved.

Received for publication April 29, Revised September 6, Accepted September 30, Published online November 21,

2002 2002 2002 2002