RESEARCH ARTICLE 65

Development 138, 65-74 (2011) doi:10.1242/dev.058727 © 2011. Published by The Company of Biologists Ltd Cdx1 refines positional identity of the vertebrate hindbrain by directly repressing Mafb expression Kendra Sturgeon1,2, Tomomi Kaneko1,*, Melissa Biemann1, Andree Gauthier1, Kallayanee Chawengsaksophak1,† and Sabine P. Cordes1,2,‡

SUMMARY An interplay of transcription factors interprets signalling pathways to define anteroposterior positions along the vertebrate axis. In the hindbrain, these transcription factors prompt the position-appropriate appearance of seven to eight segmental structures, known as rhombomeres (r1-r8). The evolutionarily conserved caudal-type homeodomain transcription factors help specify the vertebrate trunk and tail but have not been shown to directly regulate hindbrain patterning . Mafb (Kreisler, Krml1, valentino), a basic domain , is required for development of r5 and r6 and is the first to show restricted expression within these two segments. The homeodomain vHnf1 (Hnf1b) directly activates Mafb expression. vHnf1 and Mafb share an anterior expression limit at the r4/r5 boundary but vHnf1 expression extends beyond the posterior limit of Mafb and, therefore, cannot establish the posterior Mafb expression boundary. Upon identifying regulatory sequences responsible for posterior Mafb repression, we have used in situ hybridization, immunofluorescence and chromatin immunoprecipitation (ChIP) analyses to determine that Cdx1 directly inhibits early Mafb expression in the neural tube posterior of the r6/r7 boundary, which is the anteriormost boundary of Cdx1 expression in the hindbrain. Cdx1 dependent repression of Mafb is transient. After the 10-somite stage, another mechanism acts to restrict Mafb expression in its normal r5 and r6 domain, even in the absence of Cdx1. Our findings identify Mafb as one of the earliest direct targets of Cdx1 and show that Cdx1 plays a direct role in early hindbrain patterning. Thus, just as Cdx2 and Cdx4 govern the trunk-to-tail transition, Cdx1 may regulate the hindbrain-to-spinal cord transition.

KEY WORDS: Caudal related (Cdx), Mafb (Kreisler, valentino), Axial patterning, Developmental gene regulation, Hindbrain patterning, Rhombomere

INTRODUCTION expression patterns, including those of Hox genes, the most studied Early anteroposterior (AP) patterning of vertebrates occurs through segmentation genes, respect rhombomere boundaries and are the interplay of transcription factors that interpret fibroblast growth limited to specific subsets of rhombomeres (Lumsden, 2004; factor (Fgfs), cysteine-rich secreted Wnt glycoprotein and retinoic Lumsden and Krumlauf, 1996; Schneider-Maunoury et al., 1998; acid (RA) signalling pathways. These transcription factors prompt Tumpel et al., 2009). Their physical compartmentalization and the position-appropriate appearance of segmental and, thereby, defined rhombomere-specific patterns facilitate segmentally derived structures: the rhombomeres in the hindbrain, identification of regulatory regions and, thereby, molecules which give rise to specific cranial motor nerves, and the somites, required for establishment, maintenance and refinement of the from which vertebra in the trunk and tail are derived. Hindbrain positional identity of individual rhombomeres. This provides an segmentation has been a particularly useful tool for understanding exceptional model for the study of vertebrate AP patterning. early AP neural patterning. The developing hindbrain is segmented RA, Fgfs and Wnts have all been implicated in AP patterning of early in embryogenesis along its AP axis into seven to eight the brain (Diez del Corral and Storey, 2004; Nordstrom et al., 2006). lineage-restricted compartments, the rhombomeres (Lumsden, Many studies have identified RA, a product of vitamin A (retinol) 2004; Lumsden and Krumlauf, 1996; Schneider-Maunoury et al., metabolism, as the dominant posteriorizing factor in hindbrain 1998; Tumpel et al., 2009). The first seven segments (r1-r7) form development (reviewed in Glover et al., 2006; McCaffery et al., from the midbrain/hindbrain boundary to the first somite pair, and 2003). It is thought that the RA gradient is set up by a posterior pseudo-rhombomere eight (r8) forms next to somites two to five source and an anterior sink. The source of RA is provided by (Lumsden, 2004; Schneider-Maunoury et al., 1998). Many gene retinaldehyde dehydrogenase 2 (Raldh2), which synthesizes RA in the presomitic posterior mesoderm that flanks the future r7, r8 and spinal cord. The sink is provided by RA-catabolizing enzymes, cytochrome P450s (Cyp26s), expressed in the floor plate region of Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, ON M5G 1X5, Canada. 2Department of Molecular Genetics, University of the forebrain and midbrain (Dupe and Lumsden, 2001; Toronto, 1 King’s Crescent, Toronto, ON M5S 1A8, Canada. Niederreither et al., 2000; Wendling et al., 2001). RA can act directly to induce transcription of some segmentation genes via RA *Present address: Department of Legal Medicine, School of Medicine, Tokyo Women’s Medical University, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan receptors (RARs) and the retinoid X receptors (RXRs), which are †Present address: Division of Genetics and Developmental Biology, Institute for ligand-dependent transcription factors that bind as homo- or Molecular Bioscience, University of Queensland, Brisbane, QLD 4072, Australia heterodimers to specific sequences, known as RA response elements ‡Author for correspondence ([email protected]) (RAREs). RAREs have been found to govern the hindbrain-specific

Accepted 25 October 2010 expression of some Hox genes, such as Hoxa1, Hoxb1, Hoxb4 and DEVELOPMENT 66 RESEARCH ARTICLE Development 138 (1)

Hoxd4 (Frasch et al., 1995; Gould et al., 1998; Langston and Gudas, al., 2004). Thus, vHNF1 directly regulates RA-dependent, r5-r6- 1992; Marshall et al., 1994; Moroni et al., 1993; Nolte et al., 2003; specific Mafb activation. However, although the anterior expression Popperl and Featherstone, 1993; Studer et al., 1998; Whiting et al., limits of Hnf1b and Mafb coincide at the r4/r5 boundary, Hnf1b 1991). These, along with many other studies performed in the chick, expression extends more posteriorly in the neural tube than Mafb quail, mouse, fish and frog have identified RA as a posteriorizing and, therefore, does not establish the posterior boundary of Mafb factor in vertebrate hindbrain patterning. expression (Kim et al., 2005). The initial posterior boundary of In response to the RA gradient, initial segmentation genes are Mafb expression could be established by two simple mechanisms: expressed in subsets of prospective rhombomeres and induce a (1) An unknown activator, expressed in r5 and r6, but not molecular cascade that promotes further refinement of region- posteriorly of the r6/r7 boundary, collaborates with vHNF1 to specific expression patterns and segment identity (Lumsden and activate Mafb in r5-r6; or (2) a repressor with an anterior Krumlauf, 1996; Schneider-Maunoury et al., 1998). One of the expression limit at the r6/r7 boundary inhibits Mafb expression in earliest genes to be expressed segmentally in the posterior hindbrain the posterior neural tube (Kim et al., 2005). Because the earliest is Mafb [previously known as kreisler, Krml1, valentino (Val)], a genes expressed in the hindbrain have well-defined anterior basic leucine zipper transcription factor (Cordes and Barsh, 1994). expression boundaries and show diffuse expression posteriorly Mafb is expressed in the future rhombomere five and six (r5-r6) (Cordes, 2001; Deschamps et al., 1999; Hunt and Krumlauf, 1992) domain and is required for formation of these rhombomeres (Cordes and Mafb is the earliest gene confined to r5-r6, we hypothesized and Barsh, 1994; Frohman et al., 1993). The original X-ray-induced that a repressive mechanism might be likely. Given the absence of Mafb mouse mutation, kreisler (kr), is a near-perfect chromosomal RARE elements in the S5 enhancer, such a repressive mechanism inversion that abolishes rhombomere-specific Mafb expression might depend directly on an RA-inducible transcription factor(s). (Cordes and Barsh, 1994). In kr/kr mutants and in mouse embryos The RA-responsive caudal-type homeodomain transcription homozygous for the severely hypomorphic or null allele krenu, the factors, which have conserved roles in posterior axial patterning of r5-r6 domain does not develop properly, Hox genes expressed in the animals, are candidate Mafb repressors (Young and Deschamps, hindbrain are misregulated, and morphological segmentation in the 2009). In the early Drosophila embryo, Caudal protein is present posterior hindbrain (r4-r7) is lost (Cordes and Barsh, 1994; Frohman in a posterior to anterior gradient and initiates expression of et al., 1993; McKay et al., 1994; Sadl et al., 2003). Mafb has been posterior gap and pair rule genes that establish embryonic shown to directly activate rhombomere-specific expression of segmentation and induce expression (Hader et al., 1998; Hoxa3 and Hoxb3 (Manzanares et al., 1999; Manzanares et al., La Rosee et al., 1997; Ochoa-Espinosa et al., 2005; Rivera-Pomar 1997; Yau et al., 2002), but whether it directly regulates other Hox et al., 1995). In the mouse, there are three Cdx genes, Cdx1, Cdx2 genes is unknown. and Cdx4, that are expressed in a stepwise manner along the AP Expression of Mafb is regulated positively and negatively by RA axis of the embryo, with Cdx1 being the most anterior, followed by (Dupe and Lumsden, 2001; Gavalas and Krumlauf, 2000; Grapin- Cdx2, then by Cdx4 (Beck et al., 1995; Gamer and Wright, 1993; Botton et al., 1998; Hernandez et al., 2004). Mafb activation in the Meyer and Gruss, 1993). Although their anterior limits are hindbrain depends on RA. For example, application of an RAR staggered, they are all expressed into the posterior of the embryo, pan-agonist at the 0-somite stage, eliminates Mafb expression in essentially creating a posterior-to-anterior gradient of Cdx protein mouse and chick embryos. In mouse embryos that are Raldh2–/– (Gaunt et al., 2003; Gaunt et al., 2005). This nested pattern of or RAR–/–;RAR–/–, Mafb expression is lost completely expression is highly conserved across vertebrates (Charite et al., (Niederreither et al., 2000; Wendling et al., 2001). RA levels also 1998; Deschamps et al., 1999; Gaunt et al., 2005; Houle et al., control the position of the posterior boundary of Mafb expression. 2000; Lohnes, 2003; Savory et al., 2009a). Cdx genes are regulated So, reducing RA signalling in the posterior hindbrain and neural by signalling molecules involved in AP patterning, including RA, tube allows Mafb expression to expand further posteriorly beyond Fgfs and Wnts (Gaunt et al., 2005; Keenan et al., 2006; Lohnes, the r6/r7 boundary. For instance, in Raldh2 mutant zebrafish, RA 2003; Pilon et al., 2006). In the mouse, RA directly regulates Cdx1 signalling is decreased (Grandel et al., 2002), and embryos show through an upstream RARE (Houle et al., 2000). Once Cdx1 posterior expansion of Mafb (val) (Grandel et al., 2002; Maves and expression is initiated, Wnt3a, the expression of which overlaps Kimmel, 2005). These results suggest that Mafb induction by RA with that of Cdx1, acts to maintain Cdx1 expression through two is concentration dependent: higher RA levels in the spinal cord and response elements for the lymphoid enhancer-binding factor r7-r8 suppress it, whereas lower ones activate it in r5-r6. (LEF)/T-cell factor (TCF), the nuclear interpreters of canonical RA-dependent activation of Mafb depends, at least in part, on Wnt signaling (Ikeya and Takada, 2001; Prinos et al., 2001). Cdx1 the homeodomain protein, variant hepatic nuclear factor 1 binds as a heterodimer with LEF1 at the LREs to maintain its own (vHnf1/HNF1) (Aragon et al., 2005; Hernandez et al., 2004; Kim expression (Beland et al., 2004). Loss of the RARE and the LREs et al., 2005; Lecaudey et al., 2004; Maves and Kimmel, 2005; strongly reduces Cdx1 expression (Pilon et al., 2007), indicating Wiellette and Sive, 2003). At the 0-somite stage, Hnf1b expression, that RA and Wnt3a signalling act synergistically to regulate Cdx1 which is induced by RA, extends from the future r4/r5 boundary expression (Beland et al., 2004; Lohnes, 2003; Pilon et al., 2007; posteriorly into the posterior hindbrain and spinal cord of mouse, Prinos et al., 2001). Experiments in Xenopus and chick indicate that chick and zebrafish embryos (Aragon et al., 2005; Kim et al., 2005; Cdx1 is also induced by Fgfs (Bel-Vialar et al., 2002; Keenan et al., Lecaudey et al., 2004; Maves and Kimmel, 2005). In the mouse, 2006). Thus, Cdx1 responds to and integrates these signalling identification of the r5-r6 specific S5 enhancer from Mafb pathways during AP patterning (Young and Deschamps, 2009). illustrated that vHNF1 is essential, but not sufficient, for driving Cdx proteins specify the vertebrate trunk and tail by activating r5-r6 expression in 0- to 10-somite-stage embryos. In vHNF1–/– posteriorly expressed Hox genes. For example, loss of Cdx genes zebrafish, Mafb(val) is not expressed (Sun and Hopkins, 2001). causes a posterior shift in Hox gene expression and anterior Treatment of zebrafish with an RAR panagonist leads to loss of transformations of vertebra in Cdx1–/– and Cdx1–/–; Cdx2+/– mouse vHnfl and Mafb(val) expression, but Mafb(val) expression can be embryos (Allan et al., 2001; Subramanian et al., 1995; van den –/–

restored through the addition of exogenous Hnf1b (Hernandez et Akker et al., 2002). The homeotic transformations seen in Cdx1 DEVELOPMENT Cdx1 directly regulates hindbrain patterning RESEARCH ARTICLE 67

Cdx2+/– mice are more severe than either of the single mutants, independent transgenic lines. Embryos were stained for LacZ as previously indicating an overlapping role in vertebral patterning (van den described (Whiting et al., 1991). Transgenic lines were maintained by Akker et al., 2002). Conversely, ectopic or overexpression of Cdxs crossing with CD1 mice. –/– expands trunk regions at the expense of the posterior hindbrain in Cdx1 mice, first described by Subramanian et al. (Subramanian et al., Xenopus, zebrafish and mice (Epstein et al., 1997; Gaunt et al., 1995), were a generous gift from D. Lohnes (University of Ottawa) and P. Gruss, and were maintained by intercrossing. Cdx2–/– embryos were 2008; Isaacs et al., 1998; Shimizu et al., 2006; Skromne et al., 2007; generated by tetraploid aggregation of Cdx2–/– embryonic stem cells Young et al., 2009). For example, ectopic expression of Cdx1a or (Chawengsaksophak et al., 2004; Nagy et al., 2010) and harvested at E8.0- Cdx4 in the zebrafish eliminates expression of posterior hindbrain E8.5. All animal husbandry and embryo harvesting were performed specific markers, including Mafb (val) and the Mafb target Hoxb3 according to Canadian Council on Animal Care guidelines and approved (Shimizu et al., 2006; Skromne et al., 2007). These and other by the Samuel Lunenfeld Research Institute animal care committee. observations suggest that Cdx proteins might activate trunk-specific Hox genes and may also repress Mafb in the hindbrain. Of course, Immunofluorescence and RNA in situ hybridization analyses RNA in situ hybridization and LacZ staining were performed as previously such repression of hindbrain character could be a secondary described (Kim et al., 2005). Images were captured with QImaging consequence of misexpressing trunk-specific Hox genes. MicroPublisher 3.3 RTV. Cdx proteins might transduce AP positional information to Hox Immunofluorescence was performed on 10 mM frozen sections prepared and other segmentation genes in several ways. First, in the mouse, as previously described (Sadl et al., 2003). Imaging was performed on Cdx proteins have been shown to directly regulate some trunk- Nikon Eclipse 80i fluorescence microscope with Nikon DS-Ri1 Camera. specific Hox genes, including Hoxa5 (Tabaries et al., 2005), Hoxa7 Electrophoretic mobility shift assays (Gaunt, 2001), Hoxb7 (Charite et al., 1998; Subramanian et al., Mouse Cdx1 was in vitro transcribed and translated from the 1995), Hoxc8 (Shashikant et al., 1995) and Hoxc9 (Papenbrock et Cdx1/pRcCMV plasmid, provided by J. Rossant, using the TNT al., 1998). Alternatively, Cdxs might regulate another layer of reticulolysate system (Promega). DNA-binding reactions using in vitro transcription factor(s), such as vertebrate equivalents of the ‘Gap’ transcribed and translated Cdx1 protein and radiolabelled Cdx1-S5A, genes in the fly, which then govern Hox gene expression directly Cdx1-S5B or Cdx1-S5C double-stranded oligonucleotides were performed (Rivera-Pomar and Jackle, 1996). Finally, Cdxs can affect Hox and resolved by gel electrophoresis as described by Kim et al. (Kim et al., gene expression indirectly by regulating RA and Wnt3a signalling. 2005). The forward strand oligonucleotide sequences used in these Ј For example, in Cdx2–/–; Cdx4–/+ mouse embryos, the RA- experiments were: Cdx1-S5A: 5 -CCAATCATAAAGTTATTTTTAT - Ј Ј Ј degrading enzyme Cyp26a1 is downregulated, RA localization is TGC-3 , Cdx1-S5A-Mut1: 5 -CCAATCATGCCGTTATTTTTATTGC-3 , Cdx1-S5A-Mut2: 5Ј-CCAATCATAAAGTTATTGCGATTGC-3Ј, Cdx1- shifted posteriorly and Wnt3a expression is also decreased (Young S5A-Mut1+2: 5Ј-CCAATCATGCCGTTATTGCGATTGC-3Ј, Cdx1-S5B: et al., 2009). So, Cdx proteins might affect Mafb expression by 5Ј-GCTACTTCATAACTGTAATTTTGC-3 Ј, Cdx1-S5B-Mut1: 5Ј-GCTA- altering RA signalling. CTTCAGCCCTG TAATTTTGC-3 Ј, Cdx1-S5B-Mut2: 5Ј-GCTACTTCAT- Based on the role of Cdx1 in segmental skeletal development, AACTGTGCCTTTGC-3Ј, Cdx1-S5B-Mut1+2: 5Ј-GCTACTTCAGCCCT - its expression at the appropriate region of the hindbrain, its GT GCCTTTGC-3Ј, Cdx1-S5C: 5Ј-TTGCTTTTGTTATAAACC ATAA - proposed role as a transducer of RA signalling, and given that Mafb GGT-3Ј, Cdx1-S5C-Mut1: 5Ј-TTGCTTTTG TTGCGGCCATAAGGT-3Ј, expression is highly sensitive to positive and negative regulation Cdx1-S5C-Mut2: 5Ј-TTGCTTTTGTTATAA ACCAGCCGGT-3Ј, Cdx1- Ј Ј Ј via RA, we hypothesized that Cdx1 might repress Mafb, thereby S5C-Mut1+2: 5 -TTGCTTTTGTTGCGG CCAGCCGGT-3 , Cdx-Sif1: 5 - Ј Ј establishing the posterior boundary of Mafb expression in the TGAAGCAATAAACTTTATGGCCGGA-3 , Cdx-MutSif1: 5 -TGAAGC - AATGCCCTTGCGGGCCGGA-3 Ј. hindbrain. Here we show that Cdx1 is present at the right time and place to act as a posterior repressor of Mafb; that Cdx1-binding ChIP on embryos sites are present in a regulatory element required to repress reporter ChIP was performed on sets of ~150 embryos at the 0- to 10-somite stage gene expression in the posterior hindbrain and spinal cord. Mafb that had been fixed in 4% paraformaldehyde and stored in 100% methanol expression is shifted posteriorly in Cdx1–/– embryos. Chromatin according to a protocol provided by D. Lohnes and previously described immunoprecipitation (ChIP) experiments demonstrate that in by Pilon et al. (Pilon et al., 2006). To verify IP of Cdx1 protein, western blots were performed on a tenth of each IP using standard protocols. embryos at embryonic day 8.0-8.5 (E8.0-E8.5) Cdx1 binds within Remaining samples were eluted from agarose beads; formaldehyde cross- the Cdx1-binding site containing S5 enhancer. Together, these data links were reversed. Samples were purified using Qiagen QIAquick PCR reveal that Cdx1 acts as an early posterior repressor of Mafb and, Purification Kit. Real-time PCR was performed on the Applied Biosystems thus, plays a direct role in refining posterior hindbrain identity. 7900HT Real time PCR system. Samples were prepared using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) with a total MATERIALS AND METHODS volume of 20 l and were run for a total of 40 cycles with an annealing Mice temperature of 60°C. Each experiment was performed a minimum of twice Transgenic mice were generated by pronuclear injection into fertilized from the embryo stage and PCRs were performed in triplicate. ChIP oocytes from superovulated mice from the FVB/NJ strain. The 240 bp assays were analyzed by a two-way ANOVA using GraphPad Prism 4.0 containing the predicted Cdx1 sites (S5-LacZ) was removed from the S5 (GraphPad Software, La Jolla, CA). ChIP values with IgG and with Cdx1 regulatory element by digestion with XbaI and partial digestion with BglII antibody were entered as percent of input for two separate IP reactions. The and, after filling in, was cloned into the NotI site of the hsp68LacZ ANOVA test compared each group with every other group followed by a expression construct. The S5Hox mutation was introduced using Bonferroni post-hoc test. Primers used were: Cdx1-autoF 5Ј-GAGG - Quickchange Site directed mutagenesis (Stratagene, La Jolla, CA, USA) AGTCGGAACCCCAAGCTG-3Ј, Cdx1-autoR 5Ј-GCGGCTTTGC ATT - using the oligonucleotides: 5Ј-GTATCCCCCAATCATAAAGTTATC - TCAAAGCGG-3Ј, S5–2kb F 5Ј-CCACC GAAGGTCAGTCTCCAAA-3Ј, AGGCCTAG GACTTCATAACTGTAATTTTGC-3Ј and 5Ј-GCAAAAT - S5–2kbR 5Ј-AGAGGTCTGTCCTG TCCTT CAG-3Ј, S5F 5Ј-CTCCTT - TACAGTTATGAAGTCCTAG GCCTGATAACTTTATGATTGGGGG A - TCTCTTCCTGTTTCTCTG-3Ј, S5R 5Ј-CACTACCTC TGCTGTA GA - TAC-3Ј to generate the S5Hoxmut-LacZ reporter. For microinjection, SalI GTGCAT-3Ј, S5+2kbF 5Ј-AGCCTCATTGATA TGGGCTGGA-3Ј, digestion followed by DNA electrophoresis separated inserts from vector S5+2kbR 5Ј-ATCTCAGGGCAGGGTCATCACA-3Ј, krprF 5Ј-ATCG - sequences, and inserts were purified using GeneClean beads (Bio101, CTAAA AGCGAGGCTCAG-3Ј, krprR 5Ј-CTAGAGCCAAAGAGT -

Vista, CA). LacZ expression was analyzed in embryos from two CGGAGAG-3Ј. DEVELOPMENT 68 RESEARCH ARTICLE Development 138 (1)

RESULTS A Cdx-binding-site-containing region restricts posterior S5-LacZ reporter expression To address the molecular mechanisms that act to exclude Mafb expression from the neural tube posterior of the r6/r7 boundary in the hindbrain, we analyzed the early-acting, rhombomere-specific S5 enhancer from the Mafb gene, which directs expression of a LacZ reporter in r5-r6 in 0- to 10-somite-stage embryos (Fig. 1A,B) (Kim et al., 2005). Upon dissection of this regulatory region, we found that removing a 240 bp sequence from S5 (S5-LacZ) resulted in expansion of LacZ expression past the r6/r7 boundary into the posterior neural tube in 0- to 12-somite-stage embryos from two out of two transgenic lines (Fig. 1C,D). These observations suggest that binding site(s) for a repressor may reside within this region. Analysis of the transcription-factor-binding sites within the deleted segment identified multiple candidate Cdx-binding sites and one candidate Hox/Pbx-complex-binding site, referred to as Fig. 1. Loss of a sequence containing predicted Cdx sites S5Hox (Fig. 1F). Hox proteins achieve high-affinity DNA binding abolishes the posterior r6/r7 boundary of S5 enhancer- by forming heterodimers with the Pbx family of transcription dependent LacZ expression. (A)Mafb expression is detected in r5-r6 factors (Chan et al., 1997; Monica et al., 1991). Such Hox/Pbx of 0- to 12-somite-stage embryos by whole-mount RNA in situ complexes have been shown to directly govern some rhombomere- hybridization. (B)The 2 kb S5 enhancer from the Mafb gene drives LacZ specific genes (Gould et al., 1997; Popperl et al., 1995; Saleh et al., expression in r5-r6. (C,D)Removal of a 240 bp sequence from the S5 enhancer ( Cdx-S5-LacZ) causes LacZ expression to expand throughout 2000). Hoxb4 and Hoxd4 genes show expression up to the r6/r7  the posterior neural tube in transgenic embryos from two out of two boundary in E8.5 embryos (Brend et al., 2003; Chan et al., 1997; transgenic founder lines. (C)High magnification view of a Cdx-S5-LacZ Folberg et al., 1999; Gould et al., 1998; Morrison et al., 1997; embryo; (D) full embryo view. (E)Embryos transgenic for the Zhang et al., 2000; Zhang et al., 1997), when the posterior S5Hoxmut-LacZ reporter show LacZ expression in r5-r6. (F)The 240 bp boundary of Mafb expression is first established, and thus, might sequence deleted from the S5 enhancer in Cdx-S5-LacZ mice contains be candidate Mafb repressors. To test possible involvement of the three predicted Cdx homodimer-binding sites, indicated in bold red S5Hox site in Mafb repression, we introduced mutations, which type and labelled Cdx1-S5A, Cdx1-S5B and Cdx1-S5C. A yellow box had been shown to eliminate Hox/Pbx binding in other studies, to denotes the predicted Hox/Pbx consensus site, and the point mutations generate the S5Hox-mut reporter (Di Rocco et al., 1997; Gould et introduced to eliminate possible Hox/Pbx binding and generate the al., 1997; Kutejova et al., 2005; Serpente et al., 2005). In mouse S5Hox-mut transgenic lines is shown. Consensus Cdx half-binding sites are indicated in bold type underneath the three CDX-S5 homodimer embryos, from three out of three lines carrying the S5Hox-mut sites. (G)The Cdx consensus site, as determined by Margalit et al. transgene, LacZ expression was still restricted to r5 and r6 (Fig. (Margalit et al., 1993) is shown. 1E). A few speckles of LacZ expression were observed posterior of the r6/r7 boundary, but an equivalent amount of staining was also seen in the S5-LacZ reporter (Fig. 1B) and in embryos carrying an enhancerless hsp86-LacZ reporter (data not shown). Thus, Hox/Pbx Cdx1 binding (Fig. 2A-C: lanes 7-10). However, simultaneously complexes do not play a significant role in directly establishing the mutating both single binding sites in each of the Cdx1-S5A, Cdx1- posterior boundary of Mafb expression via the S5Hox site. Of S5B or Cdx1-S5C oligonucleotides eliminated the ability of these course, these findings do not exclude the possibility that Hox oligonucleotides to effectively compete for Cdx1 binding (Fig. 2A- proteins could influence Mafb repression by unconventional sites. C: lanes 11, 12). Competition with oligonucleotides containing a Because the 240 bp deleted region contains multiple consensus mutated SIF1 site, that had previously been shown to no longer Cdx-binding sites (Margalit et al., 1993) (Fig. 1F,G) and such bind Cdxs, had no effect on Cdx1 binding to any of the S5-Cdx arrays of Cdx-binding sites have been shown to regulate Hox gene sites (data not shown). These three Cdx homodimer sites were also expression within the trunk (Tabaries et al., 2005; Gaunt, 2001; bound specifically by Cdx2 and Cdx4 in EMSA assays and testing Charite et al., 1998; Subramanian et al., 1995; Shashikant et al., overlapping fragments spanning the 240 bp region demonstrated 1995; Papenbrock et al., 1998), we tested whether any of these that Cdx binding localized only to the 70 bp containing these three could bind Cdx proteins in vitro in electrophoretic mobility shift Cdx homodimer sites (data not shown). Thus, Cdx proteins bind assays (EMSAs). In vitro Cdx proteins are thought to preferentially specifically to all three S5-Cdx homodimer sites in vitro and these bind as homo- or heterodimers to two closely juxtaposed Cdx sites. sites may be involved in restricting Mafb expression. Hence we refer to pairs of juxtaposed Cdx sites as homodimer sites. Out of four predicted Cdx homodimer binding sites, three The anterior boundary of Cdx1 and posterior homodimer sites, which we named Cdx1-S5A, Cdx1-S5B and boundary of Mafb expression coincide Cdx1-S5C, bound in vitro transcribed-translated Cdx1 protein and In order for Cdx1 to establish the posterior boundary of Mafb did so specifically. Cdx1 binding could be competed away by expression Cdx1 protein must be present in the hindbrain at the addition of a double-stranded oligonucleotide containing an r6/r7 boundary and more posteriorly early in embryogenesis. established Cdx1-binding site from the Sif1 gene (Suh et al., 1994) Previous reports determined that Cdx1 is present in the posterior or the Cdx1-S5A, Cdx1-S5B or Cdx1-S5C sites themselves (Fig. hindbrain from the 1- to 10-somite stages (E8.0-E8.75) (Houle et 2A-C: lanes 3-6). Mutating a single Cdx site within each al., 2000; Meyer and Gruss, 1993), a period that coincides with the

homodimer binding site did not abrogate its ability to compete for presence of Hnf1b expression and Mafb induction (Kim et al., DEVELOPMENT Cdx1 directly regulates hindbrain patterning RESEARCH ARTICLE 69

Fig. 2. Cdx1 protein binds to three predicted Cdx homodimer-binding sites from the 240 bp deleted S5 sequence in vitro. (A-C)EMSA using in vitro transcribed-translated Cdx1 protein detect binding to oligonucleotides containing each of three predicted Cdx1 homodimer-binding sites: (A) Cdx1-S5A, (B) Cdx1-S5B and (C) Cdx1-S5C. Adding 10- or 100-fold molar excess of the SIF1 oligonucleotide containing the known Cdx1- binding sites from SIF1 gene (lanes 3,4) abolished binding, as did competition with 10- or 100-fold excess of Cdx-S5A (A, lanes 5,6), Cdx1-S5B (B, lanes 5,6) and Cdx1-S5C (C, lanes, 5,6). Oligonucleotides containing mutations within in only one half binding site still effectively competed for binding (lanes 7-10). However, oligonucleotides, in which both Cdx1 half binding sites were mutated, could no longer compete for Cdx1 binding (lanes 11,12). (D)Sequences of the predicted normal and mutated Cdx1-S5A, Cdx1–S5B and Cdx1-S5C binding sites are shown.

2005). Expression is maintained throughout the posterior neural should be observed. To test this hypothesis, we performed in situ tube from the appearance of the first somite to the 35-somite stage hybridization analyses on Cdx1–/– embryos at the 0- to 22-somite (E8.0-E10.5), but the anterior boundary of expression regresses stages (E7.5-E9.5). In Cdx1–/– embryos, expansion of Mafb into the spinal cord during development (Meyer and Gruss, 1993). expression into the posterior neural tube was observed (Fig. 4). Cdx2 and Cdx4 expression does not extend into the hindbrain However, this expansion was seen only in 0- to 8-somite stage (Beck et al., 1995; Gamer and Wright, 1993). To localize the Cdx1 (E8.0-E8.5) embryos (Fig. 4A-D), after which, by the 10-somite expression domain relative to the Mafb-expressing r5-r6 region, we stage, ectopic posterior Mafb expression disappeared and the performed immunofluorescence experiments with Cdx1- and normal r5-r6 expression pattern was regained (Fig. 4E,F). This Mafb-specific antibodies on serial sections of embryos at the 0- to expansion was most pronounced at four somites (E8.0), and 12-somite stages (E8.0-E8.5). In these experiments we found that thereafter expression began to regress towards the anterior. the anterior boundary of Cdx1 expression coincided with the Expansion of Mafb protein expression was also seen in Cdx1–/– posterior boundary of Mafb expression at 0- to 8-somite stages embryos at the four-somite stage (E8.0) (Fig. 4G,H). We also (E8.0-E8.5) (Fig. 3A-D). At the 10-somite stage (E8.75), there was examined whether lowering overall Cdx dosage in the posterior a gap between the posterior boundary of Mafb and the anterior neural tube might affect Mafb expression by examining Cdx2–/– boundary of Cdx1 expression (data not shown). By the 12-somite embryos. Cdx2 is not expressed in the hindbrain, but is expressed stage, Cdx1 expression had receded from the hindbrain and anterior early in a domain that extends from the tailbud to the anterior spinal cord altogether (Fig. 3E,F). These findings are in line with spinal cord. Cdx2–/– embryos die at preimplantation because of other analyses of Cdx1 expression (Meyer and Gruss, 1993; Gaunt placental defects, which can be rescued in embryos generated by et al., 2003; Gaunt et al., 2005; Houle et al., 2000; Lohnes, 2003; tetraploid aggregation. When we examined Mafb expression in Béland et al., 2004; Pilon et al., 2007). Moreover, analyses of E9.5 Cdx2–/– embryos generated by tetraploid aggregation, Mafb was embryos doubly transgenic for the Z/AP reporter, in which LacZ expressed essentially normally in r5 and r6 in 0- to 18-somite stage expression can be permanently activated by Cre recombinase- embryos (see Fig. S1 in the supplementary material; data not mediated excision, and a Cre recombinase under the direction of shown). So, Cdx2 on its own is not required for repressing Mafb in the Cdx1 regulatory region that appears to recapitulate early the posterior neural tube. Thus, Cdx1 directly or indirectly endogenous regulation, but, of course, may not be entirely establishes the initial posterior boundary of Mafb expression and a complete, further supports that Cdx1 is unlikely to be expressed Cdx1-independent mechanism acts to maintain r5-r6-specific Mafb anteriorly of the r6/r7 boundary (Hierholzer and Kemler, 2009). expression in r5-r6 after the 10-somite stage. The presence of Cdx1 at the r6/r7 boundary at early somite stages supports its possible role as a repressor of Mafb in the posterior Cdx1 binds at the S5 enhancer in vivo hindbrain. To test whether Cdx1 might bind within the S5 regulatory element in embryos, we performed ChIP analyses with an anti-Cdx1 Posterior expansion of Mafb expression in Cdx1–/– antibody on 0- to 12-somite-stage embryos (E8.0-E8.5). The embryos embryo ChIP samples were analyzed with quantitative PCR If the posterior extension of LacZ staining seen in the S5-LacZ (qPCR) using primers within and flanking the S5 Cdx-binding sites reporter were due to the loss of Cdx1-binding sites then, in the as well as primers at the auto-regulatory region and the Mafb

absence of Cdx1 protein, posterior expansion of Mafb expression promoter as positive and negative controls, respectively. Cdx1 DEVELOPMENT 70 RESEARCH ARTICLE Development 138 (1)

Fig. 4. Mafb expression in Cdx1+/+ and Cdx1–/– embryos. Whole- mount RNA in situ hybridization detects (A,C,E) Mafb expression in rhombomeres 5 and 6 in +/+ embryos. At the 4- (A,B) and 8- (C,D) somite stage Mafb expression extends beyond r6 into the posterior neural tubes of Cdx1–/– embryos (B,D). Posterior limits of expression are indicated by arrows. Dashed arrows mark the r8/spinal cord boundary. At the 10-somite stage (E,F), Mafb expression in Cdx1–/– embryos (F) is restricted to rhombomeres 5 and 6 and is comparable to normal Fig. 3. Cdx1 is present at the posterior limits of Mafb expression. expression (E). (G,H)Immunofluorescence with a Mafb-specific antibody Immunofluorescence on sequential sections of embryos using Mafb- in normal four-somite stage embryos (G) detects Mafb protein in r5-r6. (A,C,E) and Cdx1- (B,D,F) specific antibodies. r5-r6 localization of Mafb Lines demark the boundaries of these rhombomeres. In Cdx1–/– protein is seen at the 4- (A), 8- (C) and 12-somite stages (E). Cdx1 embryos (H), cells positive for Mafb appear beyond the posterior protein is localized at the caudal limit of Mafb at the 4- (B) and 8- boundary of r6. somite stages (D). By the 12-somite stage, no Cdx1 staining is seen in the hindbrain (F). bound at its auto-regulatory region as expected (P<0.001), but not showed that simultaneous reduction of both Cdx1a and Cdx4 leads at the Mafb promoter, or 2 kb upstream or downstream of the S5 to cell-autonomous Mafb derepression in the posterior neural tube Cdx1-binding sites (Fig. 5). Cdx1 binding was detected within the and, thus, suggested that either Cdxs or Cdx targets might normally S5 enhancer in embryos confirming an in vivo interaction repress Mafb in the posterior neural tube (Skromne et al., 2007). (P<0.001) (Fig. 5). This is the first evidence for in vivo binding of All past observations are consistent with our discovery of direct a transcription factor to the endogenous S5 rhombomere-specific Mafb repression by Cdx1. Thus, Mafb is the earliest AP patterning Mafb enhancer and is consistent with a direct role of Cdx1 in gene known to be directly regulated by Cdx1 – aside from Cdx1 regulating early Mafb expression. All of these data point towards a autoregulation. Cdx1 and Cdx2 are known to interpret the RA direct, but transient, role for Cdx1 in repressing Mafb expression signal to Hox genes expressed along the spinal cord (Allan et al., in the posterior hindbrain and spinal cord and, thereby, preventing 2001; Gaunt et al., 2008; Tabaries et al., 2005; van den Akker et these from assuming more anterior r5-r6 like identity. al., 2002). Similarly, Cdx1 appears to be responsible, at least in part, for the early RA-dependent repression of Mafb expression DISCUSSION posterior of the r6/r7 boundary. Here we have uncovered a role for Cdx1 in the direct repression of A redundancy in the actions of Cdx1 and Cdx2 could explain the the r5-r6 specific hindbrain segmentation gene Mafb. Ectopic or differences in expansion observed for S5-LacZ in wild-type overexpression of Cdx genes in Xenopus, mouse and zebrafish embryos when compared with endogenous Mafb expression in eliminates expression of posterior hindbrain patterning genes, Cdx1–/– embryos. Although there is obvious posterior expansion of including Mafb in some cases (Epstein et al., 1997; Gaunt et al., Mafb expression in Cdx1–/– embryos, it is not as dramatic as that 2008; Isaacs et al., 1998; Shimizu et al., 2006; Skromne et al., observed for the S5-LacZ reporter in which all Cdx sites were 2007; Young et al., 2009). Conversely, reducing Cdx gene dosage deleted. A simple explanation is that, when all Cdx binding is leads to expansion of the posterior hindbrain and Mafb expression eliminated, the LacZ reporter is free to be expressed throughout the into the posterior neural tube. However, these effects on Mafb posterior neural tube. By contrast, when only binding of Cdx1 is expression could have been mediated indirectly by Cdx target eliminated, Mafb expansion is limited to the posterior neural tube genes or by effects of Cdx dosage on overall Wnt and RA and anterior spinal cord, where Cdx1 is the only Cdx gene

signalling. Notably, transplantation experiments in zebrafish expressed. In the posterior neural tube, levels of Cdx2 and Cdx4 DEVELOPMENT Cdx1 directly regulates hindbrain patterning RESEARCH ARTICLE 71

from redundancy among Cdx genes, the number of Cdx sites within the S5 enhancer are likely to make it exquisitely responsive to even low levels of Cdx. Thus, Cdx1, Cdx2 and Cdx4 may all collaborate to repress Mafb in the posterior neural tube, and the three Cdx homodimer sites present in S5 may require only low levels of Cdx; for example in the hindbrain only Cdx1 is required to mediate repression. Although our findings are all in line with a pivotal repressive role for Cdx1 in the posterior hindbrain and possibly Cdx1, Cdx2 and Cdx4 in the spinal cord, we cannot exclude the possibility that other sites within the 240 bp region and/or additional regulators other than Cdxs may contribute to the posterior repression of Mafb. Nonetheless, our data do ascertain direct involvement of Cdx1 in establishing the early r6/r7 boundary of Mafb expression. Although Mafb is now the earliest known axial Cdx target, it is not the only segmentation gene repressed by Cdxs. Previously, Cdxs have been implicated in defining the posterior boundary of Hoxa5 expression in the spinal cord (Tabaries et al., 2005). Hoxa5 Fig. 5. Cdx1 binds at the S5 enhancer in vivo. (A)Representative expression extends from the caudal end of the posterior qPCR results for ChIP experiments performed with an anti-Cdx1 myelencephalon (the r8/spinal cord boundary) in the neural tube to antibody on 0- to 12-somite-stage embryos detect Cdx1 binding at the prevertebra 3 in the preaxial skeleton. In the mouse, two Cdx- Cdx1 autoregulatory region (Cdx1 auto) of the Cdx1 gene, as a positive binding sites in a Hoxa5 enhancer, essential for mesodermal control, and at the Cdx1-site-containing region of the S5 enhancer (*P<0.001). No Cdx1 binding can be detected in regions 2 kb 5Ј (S5- expression in cervical and upper thoracic regions of E12.5 2kb) or 2 kb 3Ј (S5+2kb) of the Cdx1 sites within S5 or at the Mafb embryos, were required for establishing the posterior boundary of promoter (Mafb prom). qPCR products are expressed as percent of LacZ reporter gene expression in transgenic mice. In contrast to our input (sonicated embryo DNA before IP) and compared with findings, in vivo data did not corroborate a possible repressive role background levels (+IgG) using a two-way Anova and post-hoc t-test. for Cdxs in defining the Hoxa5 expression domain, as endogenous Error bars indicate the standard error of the mean (s.e.m.). (B)A Hoxa5 expression was not affected in Cdx1–/– or Cdx2–/– embryos. schematic diagram shows the relative locations of the primer pairs used Because Cdx1, Cdx2 and Cdx4 are co-expressed posterior of the in ChIP-qPCR analyses. normal Hoxa5 expression domain, redundancy among Cdxs may have obscured any effects of reducing Cdx dosage in these experiments. In the case of Mafb, the in vivo binding of Cdx1 suffice to suppress endogenous Mafb in Cdx1–/– animals. Although detected by ChIP analyses and the posterior expansion of Mafb Cdx1, Cdx2 and Cdx4 differ in their expression patterns and some expression in Cdx1–/– embryos strongly support Cdx1 involvement structural details, at least in the case of Cdx1 and Cdx2 they have in direct Mafb repression. been shown to act redundantly in vertebral patterning (Savory et Cdxs may play a more global role in establishing posterior al., 2009b; van den Akker et al., 2002). In transgenic mice, where expression boundaries of segmentation genes. Previously, HoxB Cdx1 was replaced by Cdx2 at the endogenous locus, Cdx2 could genes were placed into two groups based on their reciprocal effectively replace Cdx1 function (Savory et al., 2009b). Aside sensitivity to RA or Fgf during early axial patterning (Bel-Vialar et

Fig. 6. Two-step regulation of rhombomere-specific Mafb expression. (A)During the 0- to 8-somite stages, an RA gradient establishes vHNF1 expression (green) up to the r4/r5 boundary and Cdx1 (orange) expression to the r6/r7 boundary. Mafb is expressed in r5-r6 and Hoxb4 in r7-r8 in Cdx1+/+ embryos. In Cdx1–/– embryos, Mafb expression extends from the r4/r5 boundary posteriorly into the future spinal cord and Hoxb4 expression is shifted posteriorly. (B)During the ~10- to 20-somite stages, Hnf1b and Cdx1 are absent and a mechanism that depends on Mafb autoregulation and probably an unidentified Fgf-responsive activator maintains Mafb expression in r5-r6 in both Cdx1+/+ and Cdx1–/– embryos. Hoxb4 expression, which is autoregulated, but not known to depend on Fgf signalling at these stages, remains shifted posteriorly in Cdx1–/– embryos (van den Acker et al., 2002). (C)Schematic outline of action of Mafb regulators on the S5 enhancer and later- acting, unidentified rhombomere-specific regulatory elements. DEVELOPMENT 72 RESEARCH ARTICLE Development 138 (1) al., 2002). Exogenous Fgf expands the expression domains of the 2007). Here the existence of unknown repressive interactor(s) has Cdx genes and 5Ј members from the HoxB complex (Hoxb6- been invoked as a possible explanation and provides a plausible Hoxb9), which are normally trunk specific, anteriorly in the neural mechanism for Cdx1-dependent repression of Mafb as well. Core tube up to the level of the otic vesicle. Conversely, the more promoter composition might further influence the choice between anteriorly expressed 3Ј Hoxb genes (Hoxb1 and Hoxb3-Hoxb5) are Cdx-dependent activation and repression. In flies, Caudal sensitive to RA, but not Fgf, treatment at these stages. These preferentially activates promoters containing a downstream core analyses focused on RA and Fgf sensitivity of Hox gene activation promoter element (DPE) (Juven-Gershon et al., 2008). The and anterior expression boundary establishment. Extrapolation promoters of Drosophila Hox genes are TATA-less, except for from our findings suggests that Fgfs and RA via Cdxs may also those from the abdominal-B (Abd-B) and Ultrabithorax (Ubx) selectively repress specific Hox genes and establish posterior genes, and can be activated by Caudal in preference to TATA- expression boundaries. Given the observations of direct Cdx- containing promoters. Whereas generally the focus in dependent repression of Hoxa5 and Mafb, perhaps the repressive developmental gene expression has been on enhancers, there has effects of Cdx proteins may be specific to ‘anterior’ segmentation been precedence, in studies on mouse Hoxb2 and Hoxb4 genes. Such repressive influences could either be masked or regulation, for promoter selectivity among Hox gene enhancers enhanced by subsequent regulatory mechanisms, e.g. inter-Hox (Gould et al., 1997). The functional composition of vertebrate Hox gene crossregulation or autoregulatory loops. Thus, as with Mafb and Mafb gene promoters have not been examined. Although Cdxs and Hoxa5, such mechanisms might only become apparent in have not been shown to block the use of specific promoters, it is, enhancer dissection experiments that reveal transient regulatory theoretically, conceivable that core promoter composition in events and developmental stages. combination with Cdx interactors acting via specific enhancers Our findings indicate that Cdx1-dependent repression is only might influence the choice between Cdx-responsive activation or temporary, suggesting that a later-acting mechanism stabilizes repression. Mafb expression in r5 and r6 after the 10-somite stage. Initial In conclusion, we have shown for the first time that Cdx1 rhombomere-specific Mafb expression occurs through the S5 promotes not just trunk specification, but plays a role in defining enhancer, which is active from the 0- to 10-somite stages. The the r7-r8 domain of the hindbrain by directly repressing Mafb activities of Cdx1 and Hnf1b in Mafb regulation are limited to the expression. Thus, the role of Cdx1 within the hindbrain is an early induction period, because these molecules regress towards the active but transient one. Our observations suggest that other posterior during embryonic development and are absent from the transient effects of Cdxs may have been missed. Do such transient hindbrain by the 10-somite stage (Fig. 6) (Houle et al., 2000; Kim perturbations subtly alter neuronal fate assignments or et al., 2005; Meyer and Gruss, 1993), long before Mafb expression distribution? Many important neuronal subtypes and neural is extinguished (Cordes and Barsh, 1994; Frohman et al., 1993; circuits are established during hindbrain patterning. Thus, it is not Theil et al., 2002). Regulatory elements governing Mafb expression difficult to imagine that subtle gene expression perturbations – between the 10- and 20-somite stages have not yet been described. even when apparently corrected later on – may have consequences However, post-initiation Mafb(val) expression depends on later in life. Thus, further analyses of Mafb and hindbrain functional Mafb(val) protein in both mice and zebrafish, indicative regulation will continue to offer scientific and possibly clinically of an autoregulatory loop (Hernandez et al., 2004; Huang et al., relevant insights. 2000; Moens et al., 1998; Sadl et al., 2003). For some Hox genes, such as Hoxb4 (Gould et al., 1998), an RA-responsive early Acknowledgements We thank D. Lohnes for providing the anti-Cdx1 antibody and Cdx1–/– mice, enhancer establishes rhombomere-specific expression and, originally generated by Subramanian et al. (Subramanian et al., 1995). We are subsequently, an autoregulatory enhancer helps maintain grateful to M. Gerstenstein, S. Tondat and the Samuel Lunenfeld transgenic expression. Consistent with this model, in Cdx1–/– mice, Hoxb4 core facility for generating transgenic mice and embryos using tetraploid expression is shifted posteriorly throughout the 0- to 20-somite aggregation. This research was funded by CIHR grants MOP 14312 and MOP stages (van den Acker et al., 2002) (Fig. 6). However, the Mafb 97966 to S.P.C. autoregulatory loop is unique in that it appears to depend not only Competing interests statement on Mafb but also on the presence of a regionally restrictive signal, The authors declare no competing financial interests. possibly an Fgf (Marin and Charnay, 2000; Maves et al., 2002; Walshe et al., 2002; Wiellette and Sive, 2003). This regionally Supplementary material Supplementary material for this article is available at restricted signal might affect either Mafb itself by perhaps a post- http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.058727/-/DC1 translational modification, such as phosphorylation, or may act via inducing another transcriptional activator. In any case, it is References probably the regionally dependent signal that promotes the post- Allan, D., Houle, M., Bouchard, N., Meyer, B. I., Gruss, P. and Lohnes, D. (2001). RARgamma and Cdx1 interactions in vertebral patterning. Dev. Biol. initiation retraction of Mafb expression into its normal r5-r6 240, 46-60. domain in Cdx1–/– mice. Aragon, F., Vazquez-Echeverria, C., Ulloa, E., Reber, M., Cereghini, S., The molecular mechanism by which Cdx1 acts to repress Mafb Alsina, B., Giraldez, F. and Pujades, C. (2005). vHnf1 regulates specification of caudal rhombomere identity in the chick hindbrain. Dev. Dyn. 234, 567- in the same regions in which it activates trunk-specific Hox genes, 576. is unknown. It seems probable that Cdx actions depend on the Beck, F., Erler, T., Russell, A. and James, R. (1995). Expression of Cdx-2 in the sequence-specific context of the Cdx-binding sites within the S5 mouse embryo and placenta: possible role in patterning of the extra-embryonic enhancer. In addition to its roles in embryonic patterning, Cdx2 membranes. Dev. Dyn. 204, 219-227. Béland, M., Pilon, N., Houle, M., Oh, K., Sylvestre, J. R., Prinos, P. and plays a role in tumorigenicity in a subset of human colorectal Lohnes, D. (2004). Cdx1 autoregulation is governed by a novel Cdx1-LEF1 cancer cell lines (Chawengsaksophak et al., 1997). In an transcription complex. Mol. Cell. Biol. 24, 5028-5038. established model for human colon cancer, the human colonic Bel-Vialar, S., Itasaki, N. and Krumlauf, R. (2002). Initiating Hox gene expression: in the early chick neural tube differential sensitivity to FGF and RA adenocancinoma cell line (LOVO), Cdx2 directly represses the signaling subdivides the HoxB genes in two distinct groups. Development 129,

insulin-like growth-factor-binding protein 3 gene (Chun et al., 5103-5115. DEVELOPMENT Cdx1 directly regulates hindbrain patterning RESEARCH ARTICLE 73

Brend, T., Gilthorpe, J., Summerbell, D. and Rigby, P. W. (2003). Multiple levels Grapin-Botton, A., Bonnin, M. A., Sieweke, M. and Le Douarin, N. M. (1998). of transcriptional and post-transcriptional regulation are required to define the Defined concentrations of a posteriorizing signal are critical for MafB/Kreisler domain of Hoxb4 expression. Development 130, 2717-2728. segmental expression in the hindbrain. Development 125, 1173-1181. Chan, S. K., Ryoo, H. D., Gould, A., Krumlauf, R. and Mann, R. S. (1997). Hader, T., La Rosee, A., Ziebold, U., Busch, M., Taubert, H., Jackle, H. and Switching the in vivo specificity of a minimal Hox-responsive element. Rivera-Pomar, R. (1998). Activation of posterior pair-rule stripe expression in Development 124, 2007-2014. response to maternal caudal and zygotic knirps activities. Mech. Dev. 71, 177- Charite, J., de Graaff, W., Consten, D., Reijnen, M. J., Korving, J. and 186. Deschamps, J. (1998). Transducing positional information to the Hox genes: Hernandez, R. E., Rikhof, H. A., Bachmann, R. and Moens, C. B. (2004). vhnf1 critical interaction of cdx gene products with position-sensitive regulatory integrates global RA patterning and local FGF signals to direct posterior elements. Development 125, 4349-4358. hindbrain development in zebrafish. Development 131, 4511-4520. Chawengsaksophak, K., James, R., Hammond, V. E., Kontgen, F. and Beck, F. Hierholzer, A. and Kemler, R. (2009). Cdx1::Cre allele for gene analysis in the (1997). Homeosis and intestinal tumours in Cdx2 mutant mice. Nature 386, 84- extraembryonic ectoderm and the three germ layers of mice at mid-gastrulation. 87. Genesis 47, 204-209. Chawengsaksophak, K., de Graaff, W., Rossant, J., Deschamps, J. and Beck, Houle, M., Prinos, P., Iulianella, A., Bouchard, N. and Lohnes, D. (2000). F. (2004). Cdx2 is essential for axial elongation in mouse development. Proc. Retinoic acid regulation of cdx1: an indirect mechanism for retinoids and Natl. Acad. Sci. USA 101, 7641-7645. vertebral specification [In Process Citation]. Mol. Cell. Biol. 20, 6579-6586. Chun, S. Y., Chen, F., Washburn, J. G., MacDonald, J. W., Innes, K. L., Zhao, Huang, K., Serria, M. S., Nakabayashi, H., Nishi, S. and Sakai, M. (2000). R., Cruz-Correa, M. R., Dang, L. H. and Dang, D. T. (2007). CDX2 promotes Molecular cloning and functional characterization of the mouse mafB gene. anchorage-independent growth by transcriptional repression of IGFBP-3. Gene 242, 419-426. Oncogene 26, 4725-4729. Hunt, P. and Krumlauf, R. (1992). Hox codes and positional specification in Cordes, S. P. (2001). Molecular genetics of cranial nerve development in mouse. vertebrate embryonic axes. Annu. Rev. Cell Biol. 8, 227-256. Nat. Rev. Neurosci. 2, 611-623. Ikeya, M. and Takada, S. (2001). Wnt-3a is required for somite specification Cordes, S. P. and Barsh, G. S. (1994). The mouse segmentation gene kr encodes along the anteroposterior axis of the mouse embryo and for regulation of cdx-1 a novel basic domain-leucine zipper transcription factor. Cell 79, 1025-1034. expression. Mech. Dev. 103, 27-33. Deschamps, J., van den Akker, E., Forlani, S., De Graaff, W., Oosterveen, T., Isaacs, H. V., Pownall, M. E. and Slack, J. M. (1998). Regulation of Hox gene Roelen, B. and Roelfsema, J. (1999). Initiation, establishment and expression and posterior development by the Xenopus caudal homologue maintenance of Hox gene expression patterns in the mouse. Int. J. Dev. Biol. 43, Xcad3. EMBO J. 17, 3413-3427. 635-650. Juven-Gershon, T., Hsu, J. Y. and Kadonaga, J. T. (2008). Caudal, a key Diez del Corral, R. and Storey, K. G. (2004). Opposing FGF and retinoid developmental regulator, is a DPE-specific transcriptional factor. Genes Dev. 22, pathways: a signalling switch that controls differentiation and patterning onset 2823-2830. in the extending vertebrate body axis. BioEssays 26, 857-869. Keenan, I. D., Sharrard, R. M. and Isaacs, H. V. (2006). FGF signal transduction Di Rocco, G., Mavilio, F. and Zappavigna, V. (1997). Functional dissection of a and the regulation of Cdx gene expression. Dev. Biol. 299, 478-488. transcriptionally active, target-specific Hox-Pbx complex. EMBO J. 16, 3644- Kim, F. A., Sing, A., Kaneko, T., Bieman, M., Stallwood, N., Sadl, V. S. and 3654. Cordes, S. P. (2005). The vHNF1 homeodomain protein establishes early rhombomere identity by direct regulation of Kreisler expression. Mech. Dev. 122, Dupe, V. and Lumsden, A. (2001). Hindbrain patterning involves graded 1300-1309. responses to retinoic acid signalling. Development 128, 2199-2208. Kutejova, E., Engist, B., Mallo, M., Kanzler, B. and Bobola, N. (2005). Hoxa2 Epstein, M., Pillemer, G., Yelin, R., Yisraeli, J. K. and Fainsod, A. (1997). downregulates Six2 in the neural crest-derived mesenchyme. Development 132, Patterning of the embryo along the anterior-posterior axis: the role of the caudal 469-478. genes. Development 124, 3805-3814. Langston, A. W. and Gudas, L. J. (1992). Identification of a retinoic acid Folberg, A., Nagy Kovacs, E., Luo, J., Giguere, V. and Featherstone, M. S. responsive enhancer 3Ј of the murine gene Hox-1.6. Mech. Dev. 38, (1999). RARbeta mediates the response of Hoxd4 and Hoxb4 to exogenous 217-227. retinoic acid. Dev. Dyn. 215, 96-107. La Rosee, A., Hader, T., Taubert, H., Rivera-Pomar, R. and Jackle, H. (1997). (1995). Evolutionary-conserved enhancers Frasch, M., Chen, X. and Lufkin, T. Mechanism and Bicoid-dependent control of hairy stripe 7 expression in the direct region-specific expression of the murine Hoxa-1 and Hoxa-2 loci in both posterior region of the Drosophila embryo. EMBO J. 16, 4403-4411. mice and Drosophila. Development 121, 957-974. Lecaudey, V., Anselme, I., Rosa, F. and Schneider-Maunoury, S. (2004). The Frohman, M. A., Martin, G. R., Cordes, S. P., Halamek, L. P. and Barsh, G. S. zebrafish Iroquois gene iro7 positions the r4/r5 boundary and controls (1993). Altered rhombomere-specific gene expression and hyoid bone neurogenesis in the rostral hindbrain. Development 131, 3121-3131. differentiation in the mouse segmentation mutant, kreisler (kr). Development Lohnes, D. (2003). The Cdx1 homeodomain protein: an integrator of posterior 117, 925-936. signaling in the mouse. BioEssays 25, 971-980. Gamer, L. W. and Wright, C. V. (1993). Murine Cdx-4 bears striking similarities to Lumsden, A. (2004). Segmentation and compartition in the early avian hindbrain. the Drosophila caudal gene in its homeodomain sequence and early expression Mech. Dev. 121, 1081-1088. pattern. Mech. Dev. 43, 71-81. Lumsden, A. and Krumlauf, R. (1996). Patterning the vertebrate neuraxis. Gaunt, S. J. (2001). Gradiants and forward spreading of vertebrate Hox gene Science 274, 1109-1115. expression detected by using a Hox/lacZ transgene. Dev. Dyn. 221, 26-36. Manzanares, M., Cordes, S. P., Kwan, C.-T., Sham, M.-H., Barsh, G. S. and Gaunt, S. J., Drage, D. and Cockley, A. (2003). Vertebrate caudal gene Krumlauf, R. (1997). Segmental regulation of Hoxb3 by Kreisler. Nature 387, expression gradients investigated by use of chick cdx-A/lacZ and mouse cdx- 191-195. 1/lacZ reporters in transgenic mouse embryos: evidence for an intron enhancer. Manzanares, M., Cordes, S., Ariza-McNaughton, L., Sadl, V., Maruthainar, Mech. Dev. 120, 573-586. K., Barsh, G. and Krumlauf, R. (1999). Conserved and distinct roles of kreisler Gaunt, S. J., Drage, D. and Trubshaw, R. C. (2005). /lacZ and /lacZ in regulation of the paralogous Hoxa3 and Hoxb3 genes. Development 126, protein gradients formed by decay during gastrulation in the mouse. Int. J. Dev. 759-769. Biol. 49, 901-908. Margalit, Y., Yarus, S., Shapira, E., Gruenbaum, Y. and Fainsod, A. (1993). Gaunt, S. J., Drage, D. and Trubshaw, R. C. (2008). Increased Cdx protein dose Isolation and characterization of target sequences of the chicken CdxA effects upon axial patterning in transgenic lines of mice. Development 135, homeobox gene. Nucleic Acids Res. 21, 4915-4922. 2511-2520. Marin, F. and Charnay, P. (2000). Hindbrain patterning: FGFs regulate Krox20 and Gavalas, A. and Krumlauf, R. (2000). Retinoid signalling and hindbrain mafB/kr expression in the otic/preotic region. Development 127, 4925-4935. patterning. Curr. Opin. Genet. Dev. 10, 380-386. Marshall, H., Studer, M., Popperl, H., Aparicio, S., Kuroiwa, A., Brenner, S. Glover, J. C., Renaud, J. S. and Rijli, F. M. (2006). Retinoic acid and hindbrain and Krumlauf, R. (1994). A conserved retinoic acid response element required patterning. J. Neurobiol. 66, 705-725. for early expression of the homeobox gene Hoxb-1. Nature 370, 567-571. Gould, A., Morrison, A., Sproat, G., White, R. A. and Krumlauf, R. (1997). Maves, L. and Kimmel, C. B. (2005). Dynamic and sequential patterning of the Positive cross-regulation and enhancer sharing: two mechanisms for specifying zebrafish posterior hindbrain by retinoic acid. Dev. Biol. 285, 593-605. overlapping Hox expression patterns. Genes Dev. 11, 900-913. Maves, L., Jackman, W. and Kimmel, C. B. (2002). FGF3 and FGF8 mediate a Gould, A., Itasaki, N. and Krumlauf, R. (1998). Initiation of rhombomeric Hoxb4 rhombomere 4 signaling activity in the zebrafish hindbrain. Development 129, expression requires induction by somites and a retinoid pathway. Neuron 21, 39- 3825-3837. 51. McCaffery, P. J., Adams, J., Maden, M. and Rosa-Molinar, E. (2003). Too much Grandel, H., Lun, K., Rauch, G. J., Rhinn, M., Piotrowski, T., Houart, C., of a good thing: retinoic acid as an endogenous regulator of neural Sordino, P., Kuchler, A. M., Schulte-Merker, S., Geisler, R. et al. (2002). differentiation and exogenous teratogen. Eur. J. Neurosci. 18, 457-472. Retinoic acid signalling in the zebrafish embryo is necessary during pre- McKay, I. J., Muchamore, I., Krumlauf, R., Maden, M., Lumsden, A. and segmentation stages to pattern the anterior-posterior axis of the CNS and to Lewis, J. (1994). The kreisler mouse: a hindbrain segmentation mutant that

induce a pectoral fin bud. Development 129, 2851-2865. lacks two rhombomeres. Development 120, 2199-2211. DEVELOPMENT 74 RESEARCH ARTICLE Development 138 (1)

Meyer, B. I. and Gruss, P. (1993). Mouse Cdx-1 expression during gastrulation. Serpente, P., Tumpel, S., Ghyselinck, N. B., Niederreither, K., Wiedemann, L. Development 117, 191-203. M., Dolle, P., Chambon, P., Krumlauf, R. and Gould, A. P. (2005). Direct Moens, C., Cordes, S. P., Barsh, G. and Kimmel, C. (1998). Equivalence in the crossregulation between retinoic acid {beta} and Hox genes during genetic control of hindbrain segmentation in fish and mouse. Development 125, hindbrain segmentation. Development 132, 503-153. 381-391. Shashikant, C. S., Bieberich, C. J., Belting, H. G., Wang, J. C., Borbely, M. A. Monica, K., Galili, N., Nourse, J., Saltman, D. and Cleary, M. L. (1991). PBX2 and Ruddle, F. H. (1995). Regulation of Hoxc-8 during mouse embryonic and PBX3, new homeobox genes with extensive homology to the human proto- development: identification and characterization of critical elements involved in oncogene PBX1. Mol. Cell. Biol. 11, 6149-6157. early neural tube expression. Development 121, 4339-4347. Moroni, M. C., Vigano, M. A. and Mavilio, F. (1993). Regulation of the human Shimizu, T., Bae, Y. K. and Hibi, M. (2006). Cdx-Hox code controls competence HOXD4 gene by retinoids. Mech. Dev. 44, 139-154. for responding to Fgfs and retinoic acid in zebrafish neural tissue. Development Morrison, A., Ariza-McNaughton, L., Gould, A., Featherstone, M. and 133, 4709-4719. Krumlauf, R. (1997). HOXD4 and regulation of the group 4 paralog genes. Skromne, I., Thorsen, D., Hale, M., Prince, V. E. and Ho, R. K. (2007). Development 124, 3135-3146. Repression of the hindbrain developmental program by Cdx factors is required Nagy, A., Nagy, K. and Gertsenstein, M. (2010) Production of mouse chimeras for the specification of the vertebrate spinal cord. Development 134, 2147- by aggregating pluripotent stem cells with embryos. Methods Enzymol. 476, 2158. 123-149. Studer, M., Gavalas, A., Marshall, H., Ariza-McNaughton, L., Rijli, F. M., Niederreither, K., Vermot, J., Schuhbaur, B., Chambon, P. and Dolle, P. (2000). Chambon, P. and Krumlauf, R. (1998). Genetic interactions between Hoxa1 Retinoic acid synthesis and hindbrain patterning in the mouse embryo. and Hoxb1 reveal new roles in regulation of early hindbrain patterning. Development 127, 75-85. Development 125, 1025-1036. Nolte, C., Amores, A., Nagy Kovacs, E., Postlethwait, J. and Featherstone, M. Subramanian, V., Meyer, B. I. and Gruss, P. (1995). Disruption of the murine (2003). The role of a retinoic acid response element in establishing the anterior homeobox gene Cdx1 affects axial skeletal identities by altering the mesodermal neural expression border of Hoxd4 transgenes. Mech. Dev. 120, 325-335. expression domains of Hox genes. Cell 83, 641-653. Nordstrom, U., Maier, E., Jessell, T. M. and Edlund, T. (2006). An early role for Suh, E., Chen, L., Taylor, J. and Traber, P. G. (1994). A homeodomain protein WNT signaling in specifying neural patterns of Cdx and Hox gene expression related to caudal regulates intestine-specific gene transcription. Mol. Cell. Biol. and motor neuron subtype identity. PLoS Biol. 4, e252. 14, 7340-7351. Ochoa-Espinosa, A., Yucel, G., Kaplan, L., Pare, A., Pura, N., Oberstein, A., Sun, Z. and Hopkins, N. (2001). vhnf1, the MODY5 and familial GCKD- Papatsenko, D. and Small, S. (2005). The role of binding site cluster strength associated gene, regulates regional specification of the zebrafish gut, in Bicoid-dependent patterning in Drosophila. Proc. Natl. Acad. Sci. USA 102, pronephros, and hindbrain. Genes Dev. 15, 3217-3229. 4960-4965. Tabaries, S., Lapointe, J., Besch, T., Carter, M., Woollard, J., Tuggle, C. K. and Papenbrock, T., Peterson, R. L., Lee, R. S., Hsu, T., Kuroiwa, A. and Jeannotte, L. (2005). Cdx protein interaction with Hoxa5 regulatory sequences Awgulewitsch, A. (1998). Murine Hoxc-9 gene contains a structurally and contributes to Hoxa5 regional expression along the axial skeleton. Mol. Cell. Biol. functionally conserved enhancer. Dev. Dyn. 212, 540-547. 25, 1389-13401. Pilon, N., Oh, K., Sylvestre, J. R., Bouchard, N., Savory, J. and Lohnes, D. Theil, T., Ariza-McNaughton, L., Manzanares, M., Brodie, J., Krumlauf, R. (2006). Cdx4 is a direct target of the canonical Wnt pathway. Dev. Biol. 289, 55- and Wilkinson, D. G. (2002). Requirement for downregulation of kreisler 63. during late patterning of the hindbrain. Development 129, 1477-1485. Pilon, N., Oh, K., Sylvestre, J. R., Savory, J. G. and Lohnes, D. (2007). Wnt Tumpel, S., Wiedemann, L. M. and Krumlauf, R. (2009). Hox genes and signaling is a key mediator of Cdx1 expression in vivo. Development 134, 2315- segmentation of the vertebrate hindbrain. Curr. Top Dev Biol. 88, 103-137. 2323. van den Akker, E., Forlani, S., Chawengsaksophak, K., de Graaff, W., Beck, Popperl, H. and Featherstone, M. S. (1993). Identification of a retinoic acid F., Meyer, B. I. and Deschamps, J. (2002). Cdx1 and Cdx2 have overlapping response element upstream of the murine Hox-4.2 gene. Mol. Cell. Biol. 13, functions in anteroposterior patterning and posterior axis elongation. 257-265. Development 129, 2181-2193. Popperl, H., Bienz, M., Studer, M., Chan, S. K., Aparicio, S., Brenner, S., Walshe, J., Maroon, H., McGonnell, I. M., Dickson, C. and Mason, I. (2002). Mann, R. S. and Krumlauf, R. (1995). Segmental expression of Hoxb-1 is Establishment of hindbrain segmental identity requires signaling by FGF3 and controlled by a highly conserved autoregulatory loop dependent upon exd/pbx. FGF8. Curr. Biol. 12, 1117-1123. Cell 81, 1031-1042. Wendling, O., Ghyselinck, N. B., Chambon, P. and Mark, M. (2001). Roles of Prinos, P., Joseph, S., Oh, K., Meyer, B. I., Gruss, P. and Lohnes, D. (2001). retinoic acid receptors in early embryonic morphogenesis and hindbrain Multiple pathways governing Cdx1 expression during murine development. Dev. patterning. Development 128, 2031-2038. Biol. 239, 257-269. Whiting, J., Marshall, H., Cook, M., Krumlauf, R., Rigby, P. W., Stott, D. and Rivera-Pomar, R. and Jackle, H. (1996). From gradients to stripes in Drosophila Allemann, R. K. (1991). Multiple spatially specific enhancers are required to embryogenesis: filling in the gaps. Trends Genet. 12, 478-483. reconstruct the pattern of Hox-2.6 gene expression. Genes Dev. 5, 2048-2059. Rivera-Pomar, R., Lu, X., Perrimon, N., Taubert, H. and Jackle, H. (1995). Wiellette, E. L. and Sive, H. (2003). vhnf1 and Fgf signals synergize to specify Activation of posterior gap gene expression in the Drosophila blastoderm. rhombomere identity in the zebrafish hindbrain. Development 130, 3821- Nature 376, 253-256. 3829. Sadl, V. S., Sing, A., Mar, L., Jin, F. and Cordes, S. P. (2003). Analysis of Yau, T. O., Kwan, C. T., Jakt, L. M., Stallwood, N., Cordes, S. and Sham, M. H. hindbrain patterning defects caused by the kreislerenu mutation reveals multiple (2002). Auto/cross-regulation of Hoxb3 expression in posterior hindbrain and roles of Kreisler in hindbrain segmentation. Dev. Dyn. 227, 134-142. spinal cord. Dev. Biol. 252, 287-300. Saleh, M., Rambaldi, I., Yang, X. J. and Featherstone, M. S. (2000). Cell Young, T. and Deschamps, J. (2009). Hox, Cdx, and anteroposterior patterning in signaling switches HOX-PBX complexes from repressors to activators of the mouse embryo. Curr. Top. Dev. Biol. 88, 235-255. transcription mediated by histone deacetylases and histone acetyltransferases. Young, T., Rowland, J. E., van de Ven, C., Bialecka, M., Novoa, A., Carapuco, Mol. Cell. Biol. 20, 8623-8633. M., van Nes, J., de Graaff, W., Duluc, I., Freund, J. N. et al. (2009). Cdx and Savory, J. G., Bouchard, N., Pierre, V., Rijli, F. M., De Repentigny, Y., Kothary, Hox genes differentially regulate posterior axial growth in mammalian embryos. R. and Lohnes, D. (2009a). Cdx2 regulation of posterior development through Dev. Cell 17, 516-526. non-Hox targets. Development 136, 4099-4110. Zhang, F., Popperl, H., Morrison, A., Kovacs, E. N., Prideaux, V., Schwarz, L., Savory, J. G., Pilon, N., Grainger, S., Sylvestre, J. R., Beland, M., Houle, M., Krumlauf, R., Rossant, J. and Featherstone, M. S. (1997). Elements both 5Ј Oh, K. and Lohnes, D. (2009b). Cdx1 and Cdx2 are functionally equivalent in and 3Ј to the murine Hoxd4 gene establish anterior borders of expression in vertebral patterning. Dev. Biol. 330, 114-122. mesoderm and neurectoderm. Mech. Dev. 67, 49-58. Schneider-Maunoury, S., Gilardi-Hebenstreit, P. and Charnay, P. (1998). How Zhang, F., Nagy Kovacs, E. and Featherstone, M. S. (2000). Murine Hoxd4 to build a vertebrate hindbrain. Lessons from genetics. C R Acad Sci III 321, 819- expression in the CNS requires multiple elements including a retinoic acid 834. response element. Mech Dev. 96, 79-89. DEVELOPMENT