The vHNF1 homeodomain protein establishes early rhombomere identity by direct regulation of Kreisler expression

Mechanisms of Development 122 (2005) 1300–1309 www.elsevier.com/locate/modo

The vHNF1 homeodomain protein establishes early rhombomere identity by direct regulation of Kreisler expression Florence A. Kim a,1, Angela Sing1 a,b, Tomomi Kaneko a, Melissa Bieman a, Nicole Stallwood a, Virginia S. Sadl a, Sabine P. Cordes a,b,* b

a Samuel Lunenfeld Research Institute, Mt Sinai Hospital, 600 University Avenue, Toronto, Ont., Canada M5G 1X5 Department of Medical and Molecular Genetics and Microbiology, University of Toronto, 1 King’s Circle, Toronto, Ont., Canada

Received 28 July 2005; accepted 2 August 2005

Abstract The early transcriptional hierarchy that subdivides the vertebrate hindbrain into seven to eight segments, the rhombomeres (r1–r8), is largely unknown. The Kreisler (MafB, Krml1, Val) gene is earliest gene expressed in an r5/r6-restricted manner and is essential for r5 and r6 development. We have identified the S5 regulatory element that directs early Kreisler expression in the future r5/r6 domain in 0–10 somite stage embryos. variant Hepatocyte Nuclear Factor 1 (vHNF1/HNF1b/LF-3B) is transiently expressed in the r5/r6 domain of 0–10 somite stage embryos and a vHNF1binding site within this element is essential but not sufficient for r5/r6-specific expression. Thus, early inductive events that initiate Kreisler expression are clearly distinct from later-acting ones that modulate its expression levels. This site and some of the surrounding sequences are evolutionarily conserved in the genomic DNA upstream of the Kreisler gene among species as divergent as mouse, humans, and chickens. This provides the first evidence of a direct requirement for vHNF1 in initiation of Kreisler expression, suggests that the role of vHNF1 is evolutionarily conserved, and indicates that vHNF1 collaborates with other transcription factors, which independently bind to the S5 regulatory region, to establish the r5/r6 domain. q 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Kreisler (MafB/Krml1/valentino); Hindbrain segmentation; Rhombomere specification; vHNF1/HNF1 beta/LF-3B; Hox genes; Embryonic patterning

1. Introduction Its visible physical compartmentalization and segmentspecific gene expression patterns make the vertebrate hindbrain an ideal model for studying how positional information is laid down in development. The embryonic hindbrain is transiently subdivided into seven to eight physical compartments, known as rhombomeres (r1–r8), along its anterior–posterior axis. Cell lineage, morphologic, and genetic criteria all indicate that rhombomeres represent true embryonic compartments (Fraser et al., 1990; Lumsden and Krumlauf, 1996). Many gene expression patterns, including those of Hox genes, the most studied segmentation genes, are limited to specific subsets of rhombomeres (Murphy et al., 1989; Wilkinson et al., 1989). The Kreisler segmentation gene encodes a basic domain

* Corresponding author. Tel.: C1 416 586 8891; fax: C1 416 586 8588. E-mail address: [email protected] (S.P. Cordes). 1 These authors contributed equally to this work and share first co-authorship.

0925-4773/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.mod.2005.08.001

leucine zipper (bZIP) transcription factor, which is expressed in and is essential for formation of the fifth and sixth rhombomeres (r5 and r6) (Cordes and Barsh, 1994). Initial expression of Kreisler precedes the appearance of rhombomeric boundaries and other r5/r6-restricted gene expression, and is required directly for the regulation of Hoxa3 and Hoxb3 in r5/r6 and r5, respectively (Cordes and Barsh, 1994; Manzanares et al., 1997, 1999). What determines the expression pattern of Kreisler itself is still not clear, but its early segment-specific expression is likely to occur via complex integration of anterior and posterior signals. Studies in the chick first showed that Kreisler expression, like that of Hox genes, was sensitive to posteriorizing signals such as retinoic acid (RA) and fibroblast growth factors (FGFs) (Grapin-Botton et al., 1998; Marin and Charnay, 2000; Papalopulu et al., 1991). Later analyses of the role of FGFs in zebrafish indicated that signals emanating from the future r4 domain also affected Kreisler expression either directly or indirectly. For example, in zebrafish transplantation studies, r4 appeared to induce Kreisler expression in

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

a Fibroblast growth factor dependent manner (Maves et al., 2002; Walshe et al., 2002). In order to identify upstream regulatory mechanisms that establish Kreisler expression domains, we have searched for rhombomere-specific regulatory elements in the Kreisler genomic region. Here, we have isolated an early acting rhombomere-specific enhancer of Kreisler. We have begun to dissect the regulatory elements required for Kreisler expression and have identified a single variant Hepatocyte Nuclear Factor 1 (vHNF1) binding site that is essential, but not sufficient alone for r5/r6-specific expression. Recent data in zebrafish show that loss of function of vHNF1 leads to hindbrain defects similar to those observed in animals lacking Kreisler/Valentino and that Kreisler/Valentino expression is missing in vHNF1 mutants (Sun and Hopkins, 2001). In the zebrafish neural tube, vHNF1 is expressed coincidentally with that of Kreisler in the future r5 and r6 domains (Barbacci et al., 1999; Coffinier et al., 1999a; Ott et al., 1991; Sun and Hopkins, 2001). Here, we show that vHNF1 is transiently expressed in the r5/r6 domain of the mouse in 0–10 somite stage embryos. Furthermore, vHNF1 protein is highly conserved between zebrafish and mouse (Sun and Hopkins, 2001), and the vHNF1 regulatory site in the Kreisler enhancer is also highly conserved across species as divergent as chickens, mouse, and humans. Together these observations suggest a conserved role for vHNF1 in

1301

establishing aspects of hindbrain patterning in collaboration with other transcription factors. 2. Results 2.1. Identification of an early-acting rhombomere-specific enhancer for Kreisler The classical kreisler (kr) mutation was caused by a submicroscopic chromosomal inversion that does not disrupt the Kreisler coding sequence or transcription unit, but abolishes Kreisler expression in the prospective r5 and r6 region and leads to ectopic expression of Kreisler in r3 (Cordes and Barsh, 1994; Sadl et al., 2003). A simple model that might account for the loss of r5/r6-specific expression would be that a rhombomere-specific enhancer required for early r5/r6-specific Kreisler expression is located 5 0 of the transcriptional start and is thus removed by the kr inversion. To determine whether rhombomere-specific enhancers for Kreisler were located 5 0 of the Kreisler transcriptional start site, we began to examine genomic P1 clones that spanned the kr centromere distal inversion breakpoint and the Kreisler transcription unit (Cordes and Barsh, 1994) (Fig 1A). We subcloned overlapping BamHI and SacI fragments from the P1-222 and P1-730 clones into a reporter construct containing the minimal hsp68

Fig. 1. Identification of a regulatory element required for early Kreisler expression. (A) Schematic diagram of the genomic region 5 0 of the Kreisler transcription unit. The position of the S5 element is shown. P1 clones, P1-222 and P1-730, span the centromere-distal kreisler inversion breakpoint (Cordes and Barsh, 1994). (B) A restriction map of the 3244 bp SacI fragment, S5, which directs LacZ expression in r5 and r6 of 0–10 somite stage embryos. The regions of highest identity between mouse and human is indicated by a light gray box, and the location of the vHNF1 site is indicated by a dark gray oval. The S5 element directs LacZ expression in r5 and r6 domain. (C) Kreisler expression as detected by whole mount RNA in situ hybridization in an eight somite stage embryo coincides with that of (D). LacZ directed by the S5-hspLacZ reporter in the future r5 and r6. (E) Kreisler protein is detected by immunoflourescence with an anti-Kreisler antibody in r5 and r6 of an eight somite stage transgenic S5-hspLacZ embryo, which exhibit (F) LacZ staining in r5 and r6 as well.

1302

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

promoter and a LacZ reporter gene, and began to test these for their ability to direct rhombomere-specific LacZ expression. In this manner, we identified a 3.2 kb SacI fragment from P1-222 referred to as S5, which directed LacZ expression in the prospective r5/r6 region in 0–10 somite stage embryos in one transgenic line and three transgenic founder animals (Fig. 1, Table 1). LacZ expression disappeared completely from the hindbrain by the 10 somite stage. To determine whether rhombomere-specific LacZ expression coincided with Kreisler expression we stained embryos lightly for LacZ and used immunofluorescence to localize Kreisler protein. In sections of 6–8 somite stage embryos, the domain containing light and slightly spotty LacZ expression coincided with the Kreisler expressing r5/r6 region (Fig. 1C–F). Some faint LacZ speckles were scattered at the dorsal edges of rhombomeres 7 and 8 and sometimes in the trunk mesoderm. Because this LacZ speckling was only seen in two lines containing the S5 element or subdomains of S5 and has been seen in animals transgenic for the hsp68 minimal promoter-LacZ reporter alone, it is likely not caused by the S5 element. Thus, the S5 enhancer directs LacZ expression to the prospective r5 and r6 region between the 0–10 somite stages and is probably involved in initiating rhombomere-specific Kreisler expression. Surprisingly, the S5 element is located between the inversion breakpoint and the transcriptional start site of Kreisler and is not removed by the kr inversion. Thus, a position effect caused by the kr inversion rather than the complete removal of rhombomere-specific enhancers may alter Kreisler expression in the hindbrains of kr/ kr and kr/C mice. Of course, it is possible that other rhombomere-specific regulatory elements that synergize with the S5 element to initiate Kreisler expression or act slightly later at the 10–20 somite stages to maintain and refine Kreisler expression may have been removed by the kr inversion. Table 1 Transgenic embryos with r5 and r6-specific LacZ expression Construct

No. with r5/r6 expression

Transgenic mice generated

S5

4/4

One transgenic line Three founder embryos Four founder embryos Two transgenic lines Three transgenic lines Two founder embryos Three transgenic lines Three transgenic lines Two founder embryos

S5-2 kb

4/4

S5-900 kb

0/2

S5mut

0/5

2xvHNF1

0/3

5xvHNF1

2/5

The regions of S5 used to drive expression of the hsp68LacZ reporter in transgenic lines are shown. The number of transgenic lines and founder embryos with expression in r5/r6 and the number of transgenic lines and transgenic founder embryos examined for each construct are indicated.

2.2. An evolutionarily conserved consensus site is present in the Kreisler enhancer By examining subdomains of the S5 region in transgenic mice we localized the rhombomere-specific enhancer further to a 2 kb SacI–XbaI fragment, S5-2 kb (Fig. 4A). We then used cross-species comparisons to search for conserved regions in this 2 kb segment (Ovcharenko et al., 2004). Standard Blastn searches initially identified a 121 bp stretch with 85% identity to regions upstream of the human Kreisler (MafB) gene on chromosome 20 (Wang et al., 1999). Subsequent sequence alignment extended the highly conserved domain to a 584 bp region which shared 70.5% identity between mouse and human genomic regions located 16.5 and 17.6 kb, respectively, upstream of the Kreisler gene. Even more compellingly, a 176 bp region 13 kb upstream of the Kreisler gene in chicken genomic DNA shared 40% identity with both the S5 mouse and homologous human region (Fig. 2A). Interestingly, the S5 region that shows high homology to chick is distinct from the 121 bp region with the highest homology to human genomic DNA. Overall, the conserved nature of this rhombomerespecific enhancer and the evolutionarily conserved role of Kreisler suggest that at least some regulatory mechanisms governing early Kreisler expression will be shared across species. Previous transplantation experiments in the chick had implicated retinoic acid in the regulation of early Kreisler expression (Grapin-Botton et al., 1998; Marin and Charnay, 2000). However, analysis of this 2 kb sequence for the presence of candidate hindbrain patterning transcription factor binding sites could not detect any predicted consensus binding sites for RAR/RAR or RAR/RXR dimers. However, one predicted binding site for vHNF1 was present and evolutionarily conserved between mouse S5 and the regions upstream of Kreisler in chicken, dog, and human genomic DNA (Fig. 2B) (Wingender et al., 1996). 2.3. vHNF1 is expressed transiently in r5 and r6 When we examined vHNF1 expression in the embryonic hindbrain, we found that vHNF1 expression coincides with Kreisler induction and mirrors S5-directed LacZ expression precisely both spatially and temporally. Previous studies in zebrafish showed that vHNF1 is coexpressed with Kreisler in r5 and r6. In the mouse hindbrain, vHNF1 expression has not been extensively described. In 0–10 somite stage mouse embryos (8.0–8.5 dpc), we found that vHNF1 mRNA expression in the neural tube extended from the future r5 to roughly the level of the heart, and abutted the Hoxb1expressing future r4 domain (Fig. 3). However, between the 10–14 somite stages, a gap between vHNF1 expression and the Hoxb1 expressing r4 domain becomes apparent as the anterior boundary of vHNF1 expression shifts posteriorly to the r6/r7 boundary. Such a posterior shift of expression has not been reported previously, and suggests that vHNF1 plays a role only in Kreisler induction and not in its later expression.

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

1303

Fig. 2. Evolutionary Sequence conservation within the rhombomere-specific S5 enhancer. (A) DNA homology searches identified evolutionary conserved region surrounding the vHNF1 site. The sequence alignments are shown here for mouse, human, and chick. The pink box marks the conserved vHNF1 binding site. Yellow boxes indicate identity shared between two species, while orange indicates shared identity between mouse, humans, and chick. (B) A highly conserved vHNF1 consensus binding site is found upstream of the Kreisler gene in mouse, Human, dog, and Chick genomic DNA. N denotes any nucleotide, while M denotes any nucleotide except for guanosine.

2.4. vHNF1 interacts functionally with a single site in the Kreisler enhancer We examined whether this potential vHNF1 site, S5vHNF1, could bind vHNF1 protein in electrophoretic mobility shift assays, and found that in vitro transcribed translated vHNF1 protein bound specifically to the S5-2 kb region (Fig. 4B). When we subdivided the 3.2 kb S5 region into 2.0 and 1.2 kb fragments by digestion with XbaI, only the 2.0 kb SacI–XbaI fragment, S5-2 kb, which contained the predicted vHNF1 binding site, could bind and compete effectively for vHNF1 protein binding. To examine whether this was the only vHNF1 binding site present within S5 we mutated the S5vHNF1 site into the mutS5-vHNF1 site (GCtaATCAttAAc to GCgtATCAggAAc). Introduction of similar mutations into the HNF1 binding site, that regulates hepatocyte-specific transthyretin expression, eliminates HNF1 dependent activation in hepatoma cells (Costa and Grayson, 1991; Tronche et al., 1990, 1999). vHNF1(HNF1b) is highly homologous to the closely related HNF1a protein in regions involved in DNA binding, i.e. the N-terminal dimerization domain, POU-like domain, and the divergent homeodomain, and in all studies so far their DNA

binding affinities have been identical (Baumhueter et al., 1988; Mendel et al., 1991; Rey-Campos et al., 1991; Tronche and Yaniv, 1992). Thus, these mutations in the S5-vHNF1 site were predicted to abolish vHNF1 binding. This mutated S5 fragment (mutS5) could not compete for vHNF1 binding to S5-2 kb. Addition of 10- or 100-fold molar excess of double-stranded oligonculeotides containing the predicted S5-vHNF1 binding site or a known vHNF1 binding site (PE-vHNF1) competed for vHNF1 binding to S5 (Costa and Grayson, 1991; Tronche et al., 1990, 1999). However, oligonucleotides, which contained either the mutated S5-vHNF1 site or a mutated PE-vHNF1 site (DS34vHNF1), which previously had been shown not to bind vHNF1, could not compete for vHNF1 binding. Finally, we demonstrated that vHNF1 could bind specifically to a S5-vHNF1 site containing oligonucleotide (Fig. 4C). In transfection studies performed in COS-7 cells, a dimer of the S5-vHNF1 binding site can direct vHNF1 dependent transcription as efficiently as a dimer of the previously described PE-vHNF1 binding site (Fig. 5). By contrast, the mutated S5-vHNF1 binding site failed to induce vHNF1 dependent transcription. Thus, the S5-vHNF1 binding site can

1304

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

Fig. 3. Whole mount RNA in situ hybridization detected vHNF1 expression in the hindbrains of 0-14 somite stage mouse embryos (8.0-8.5dpc). In 8 somite stage embryos, (A) vHNF1 expression extends from the posterior hindbrain into the future spinal cord, while (B) Hoxb1 is expressed specifically in r4. Here, double staining with vHNF1 and Hoxb1 probes detects continuous hybridization from the r3/r4 boundary posteriorly into the future spinal cord in 4-6 somite (C, E, and F) and 8-10 somite (G, H) stage embryos. This suggests that the anterior boundary of vHNF1 expression is located roughly at the future r4/r5 sulcus in 4-10 somite stage embryos. By contrast in 12-14 somite stage embryos (D, I, and J) a decided gap in expression between the Hoxb1 expressing r4 domain and the vHNF1 expression domain can be detected. Thus, presence of vHNF1 mRNA coincides with Kreisler expression only in 0-10 somite stage mouse embryos. Anterior arrows coincide with the r4/r5 boundary, while posterior arrows coincide with the morphologically identifiable r6/r7 boundary in 12-16 somite stage embryos.

bind vHNF1 in vitro and direct vHNF1 dependent transcription in vivo. 2.5. The vHNF1 binding site is essential, but not sufficient for early Kreisler expression To determine whether this S5-vHNF1 binding site was required for rhombomere-specific LacZ expression, we made transgenic mice using the entire S5 region with the vHNF1 binding site mutated. No LacZ activity could be observed in the prospective r5 or r6 domains in any of these lines (Fig. 6B). Thus, this vHNF1 site is essential for rhombomere-specific LacZ expression. However, this vHNF1 site alone cannot direct high level rhombomere-specific LacZ expression. We could not detect any r5/r6-specific LacZ expression in embryos transgenic for constructs containing two copies of the S5-vHNF1 binding sites linked directly to the hsp68–lacZ reporter (Fig. 6C,D). However, in two transgenic founder lines containing five copies of the S5-vHNF1 site, we detected very faint LacZ staining in the neural tube extending from r5 posterior to roughly the level of the heart in 0–10 somite stage embryos. This expression coincides with that of vHNF1 itself. These results indicate that vHNF1 is essential but not sufficient for rhombomere-specific Kreisler expression. Therefore, other factors must collaborate with vHNF1 to allow for rhombomerespecific transcriptional activation of Kreisler. Because a 934 bp SacI–EarI fragment that contains the highly homologous region and the vHNF1 site is not sufficient for r5/r6-specific LacZ expression (Table 1, data not shown), some of these additional factors do not bind S5 near the vHNF1 site, are unlikely to associate directly with vHNF1, and have binding

sites located outside the region with the highest homology between mouse, human, and chick genomic sequences. 3. Discussion These studies strongly suggest that vHNF1 acts as a key regulator of early hindbrain patterning by directly regulating initiation of Kreisler expression during early development. vHNF1 likely activates initiation of Kreisler expression directly via a single binding site that is present in the rhombomere-specific S5 enhancer located 16.5 kb upstream of the Kreisler transcriptional start site in the mouse. Portions of the S5 enhancer are evolutionarily conserved 17.5 kb upstream of the Human Kreisler gene and 13 kb 5 0 of the chicken Kreisler gene. S5 homologous regions are also present in equivalent positions in all other mammalian genomes, such as dog and rat, for which the sequence surrounding the Kreisler gene is currently available. Strikingly, the single vHNF1 binding site, for which the consensus sequence is NM(T/A) AATN(A/T)TTAAC(C/T)(Tronche and Yaniv, 1992), is present in the S5 homologous region in all of these instances, and in transcription factor binding site searches of the mouse and human genomic sequences surrounding the Kreisler gene this was the only vHNF1 binding site detected. As yet Xenopus tropicales, Fugu, and zebrafish genomes are incomplete and the sequence surrounding their Kreisler genes is not yet available in the public domain. Thus, we can not draw any conclusions regarding the conservation of the S5 region and vHNF1 site for amphibians or teleosts. However, zebrafish and mouse vHNF1 proteins and, in particular, their DNA binding regions are highly conserved (Sun and Hopkins, 2001). Thus, it seems likely that a similar vHNF1 binding site will eventually be

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

1305

Fig. 4. vHNF1 binds to a single site within the S5 element. (A) The sequences of the vHNF1 sites used in these studies. (B) In vitro transcribed/translated vHNF1 protein binds specifically to a single site within the radiolabeled 2 kb SacI–XbaI region of S5, shown in Fig. 1, in electrophoretic mobility shift assays. Lanes 2–20 contain 50 ng of vHNF1 protein. Lane 1 contains only the radiolabeled probe. Competitor DNA present is shown above the lanes. (Lane 3) 10-fold and (Lane 4) 100fold molar excess of the SacI–XbaI 1.3 kb S5 fragment does not compete for vHNF1 binding, while effectively competes for vHNF1 binding, at 10-fold (Lane 5) and 100-fold (Lane 6) molar excess. 10! (Lane 7), 100! (Lane 8), and 1000! (Lane 9) molar excess of the entire S5 fragment can compete for vHNF1 binding, but the S5mut fragment (Lanes 10–12; 10!, 100!, and 100! molar excess, respectively), which contains the mutated vHNF1 site shown in (A), cannot. 10! and 100! molar excess of double-stranded oligonucleotides of the S5-vHNF1 site (Lanes 13,14) and the PE-vHNF1 site from the rat albumin enhancer (Lanes 15,16) bind vHNF1. The mutated S5 (Lanes 15,16) and DS34 (Lanes 19,20) vHNF1 sites cannot compete for vHNF1 binding. (C) A radiolabeled S5-vHNF1 site containing oligonculeotide binds vHNF1 protein (Lane 2) specifically. Lane 1 does not contain vHNF1 protein. All other lanes contain 50 ng of vHNF1 protein. Addition of either 10-, 100-, or 1000-fold molar excess of S5-vHNF1 (Lanes 3–5, respectively) or PE-vHNF1 (Lanes 9–11) oligonucleotides compete for vHNF1 binding, but addition of 10-, 100-, or 1000-fold molar excess of S5mutvHNF1 (Lanes 6–8, respectively) or DS34-vHNF1 (Lanes 1–14) does not affect vHNF1 binding to the S5vHNF1 site.

identified and shown to govern initiation of Kreisler(Val) expression directly in zebrafish, just as in the mouse. An interaction between vHNF1 and Kreisler was first strongly suggested by studies in zebrafish showing that no Kreisler (Val) expression was detected in vHNF1K/K zebrafish and that ectopic expression of vHNF1 extends Kreisler (Val) expression ectopically (Sun and Hopkins, 2001). In mice, knock-out of vHNF1 is lethal prior to hindbrain formation, because vHNF1 is required for visceral endoderm development. In the mouse, an unusually small and rounded otic vesicle similar to that seen in the absence of kr/kr mutants is observed in vHNF1K/K embryos generated by tetraploid rescue, but Kreisler expression has so far not been examined in vHNF1K/K mouse embryos (Barbacci et al., 1999; Coffinier et al., 1999b). These observations taken together with the evolutionary conservation of vHNF1 structure and Kreisler function in mouse and zebrafish, strongly suggest that vHNF1

acts upstream of Kreisler in the mouse as well. Furthermore, the expression pattern of vHNF1 in the mouse is consistent with its putative role in hindbrain patterning, and, specifically, in Kreisler induction. In the mouse neural tube, vHNF1 expression extends from the future r4/r5 boundary posteriorly to roughly the level of the heart starting at the 0 somite stage, and recedes posteriorly out of the r5/r6 domain by the 10–12 somite stages. S5 directed LacZ expression is only transient and mirrors the appearance and disappearance of vHNF1 expression in the future r5/r6 domain precisely. Thus, the present studies provide the first evidence of a direct requirement for vHNF1 in regulating initiation of Kreisler expression. However, vHNF1 alone is not sufficient to drive r5/6 expression nor is the enhancer containing vHNF1 sufficient for later maintenance of Kreisler expression patterns. vHNF1 contains a dimerization domain, a POU-specific box, and a divergent homeodomain, but lacks the activation domains that

1306

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

Fig. 5. vHNF1 can activate transcription of reporter plasmids containing dimers of the PE-vHNF1 site or the S5-vHNF1 site to an equivalent extent upon transfection into COS-7 cells. By contrast, vHNF1 fails to activate transcription of a reporter construct containing a dimer of the mutated S5mut-vHNF1 site above the level seen for the empty reporter vector.

are present in the closely related HNF1a protein (Mendel et al., 1991). vHNF1 can form extremely stable homo- and heterodimers that bind DNA with HNF1a (Mendel et al., 1991; Rey-Campos et al., 1991). HNF1a expression overlaps to some degree with that of vHNF1 in liver, kidney, pancreas and intestine, but not in the hindbrain (Barbacci et al., 1999; Cereghini et al., 1992; Coffinier et al., 1999b; Ott et al., 1991). Furthermore, mice lacking HNF1a are viable and have normal hindbrains (Lee et al., 1998; Pontoglio et al., 1996). Thus, vHNF1 and HNF1a may interact directly during these later developmental processes, but HNF1a is not likely the partner for vHNF1 in hindbrain development. In Kreisler regulation, as in many other cases, vHNF1 relies on additional factors, which have their own distinct binding sites, to activate high level region- or tissue-specific gene expression. Fibroblast growth factors (FGFs) have been implicated in posterior hindbrain patterning and play an as yet not well understood role in Kreisler regulation. Studies in the chick first

demonstrated that exogenous FGF2, FGF4, and FGF8 could increase the r3 and r5 domains at the expense of r4 (Marin and Charnay, 2000). In addition, local insertion of FGF4 or 8 loaded beads could induce Kreisler expression in r7 and r8, but only in regions in close proximity to the FGF source. Furthermore, inhibition of FGF signaling abolished Kreisler expression in r5 and r6. Later, Wiellette and Sive showed that FGF signaling acted similarly in patterning the posterior zebrafish hindbrain and that FGF signaling acted with vHNF1 to induce Kreisler expression(Wiellette and Sive, 2003). While this paper was under review, the above results were confirmed yet again in the fish by others (Hernandez et al., 2004). As yet it is unclear whether FGF induces factors specific to the posterior hindbrain or activates more generally expressed FGF responsive transcription factors. However, a clue as to where vHNF1 collaborators and FGF inducible factors important for Kreisler regulation may normally reside may be gleaned from ectopic expression studies performed in zebrafish. Sun and Hopkins found that ectopic vHNF1 expression in zebrafish could only extend the Kreisler (Val) expression domain into r4, but not more anteriorly or posteriorly(Sun and Hopkins, 2001). By contrast, others reported that ectopic vHNF1 expression in zebrafish lead to additional low level Kreisler (Val) expression in r2 and r3 (Hernandez et al., 2004). Even in this latter case, Kreisler (Val) expression appeared significantly higher in r4 than in r2 and 3. A simple way to reconcile these differences is that intermediate to low levels of ectopic vHNF1 would readily activate expression of Kreisler (Val) in the presence of the proper collaborators in r4. Much higher vHNF1 levels are required for Kreisler (Val) activation in r2 and r3, because normal collaborators are absent or present at lower levels in these domains and, thus, expression depends on vHNF1 to a greater degree. Some support for this view comes from our findings that S5-vHNF1 binding site dimers cannot activate LacZ reporter expression, but quintuple binding sites lead to low level reporter gene expression. So too, we would expect unusually high levels of vHNF1 to eventually drive low levels of Kreisler (Val) expression in ectopic domains even in

Fig. 6. vHNF1 binding site is essential, but not sufficient, for r5- and r6-specific expression of Kreisler in 0–10 somite stage embryos. (A) LacZ expression can be seen in r5 and r6 in a six somite stage embryo transgenic for the S5-2 kb construct. (B) In a six somite embryos transgenic for the S5mut construct, in which the vHNF1 site has been mutated to that shown in Fig. 5, no rhombomere-specific LacZ expression can be seen. (C) A dimer (2xvHNF1) is are not sufficient to direct rhombomere-specific expression of the hsp68LacZ reporter. However, in embryos transgenic for (D). A quintuplet (5xvHNF1) of the S5vHNF1 site faint LacZ staining can be detected in the neural tube extending from r5 throughout the posterior hindbrain.

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

Fig. 7. (A) The simplest molecular model for vHNF1 dependent activation of Kreisler predicts that a vHNF1 collaborator is expressed in the future r4-r6 domain. (B) However, other observations support a model, in which at least one vHNF1 collaborator is expressed throughout the posterior hindbrain up to the r3/r4 boundary, and a repressor is expressed throughout the posterior hindbrain up to the r6/r7 boundary. (C) A schematic diagram shows the regulatory network that governs early r5 and r6 development. Retinoic acid (RA) and Fibroblast growth factors (FGFs) may regulate vHNF1 and/or Kreisler directly. vHNF1 directly regulates Kreisler expression. Kreisler in turn activates expression of r5/r6-specific Hoxa3 expression and in collabration with Krox20 activates r5-specific Hoxb3 expression.

the absence of the ideal or normal collaborator(s). Once Kreisler (Val) expression is activated, autoregulatory activity of Kreisler (Val) can act to maintain and increase Kreisler (Val) expression in these ectopic domains. So, we postulate that the lower level Kreisler (Val) expression in r2 and r3 is due to very high ectopic vHNF1 levels, and that under normal conditions the molecule(s) that collaborate with vHNF1 to activate Kreisler expression are found at high levels or in an activated state in r4–r6. In the simplest case a single collaborator or set of collaborators expressed in r4–r6 would be involved (Fig. 7A). Of course, r4-derived FGF signals could activate molecules in r4–r6 that are present more ubiquitously in the neural tube. However, most early gene expression patterns in the developing hindbrain show sharp anterior expression boundaries and more diffuse posterior expression boundaries, like vHNF1 itself, and no r4–r6 specific early acting transcription factor has yet been identified. Thus, aside from a vHNF1 co-activator, another molecule or set of molecules would be required to repress Kreisler expression posterior of the r6/r7 boundary. The observation that the posterior boundary of Kreisler expression extends caudally into r7 and r8 in RARa/RARb compound null mice lends support to the existence of such a repressor (Dupe et al., 1999). Thus, we favor this second model (Fig. 7B). Further dissection of the regulatory sites within S5 will reveal the identity of these vHNF1 co-activators and repressors. The analysis of mouse and zebrafish mutants has lead to the identification of individual transcription factors that determine specific anterior–posterior domains, but understanding links

1307

between these transcription factors and thus the precise transcriptional networks involved has required the dissection of regulatory elements. We have begun to elucidate the transcriptional network required to build r5 and r6. We have shown that in the mouse vHNF1 is likely to collaborate with other transcription factors to initiate Kreisler transcription directly in the future r5/r6 domain, and Kreisler in turn directly activates the expression of Hoxa3 and Hoxb3 (Fig. 7C) (Manzanares et al., 1997, 1999). So, it is quite likely that the early signals, such as retinoic acid and FGFs, that establish anterior–posterior position govern vHNF1 expression directly. Indeed early studies in F9 cells demonstrate that vHNF1 can be induced by retinoic acid treatment and more recent analyses in zebrafish suggest that Retinoic acid induction of Kreisler occurs, at least in part, through vHNF1 (Hernandez et al., 2004; Kuo et al., 1991). Thus, further analyses of vHNF1 collaborators and regulators will provide us with a more complete picture of the molecular events that transform early positional signals into two precisely defined compartments. 4. Materials and methods 4.1. Mice Transgenic mice were generated by pronuclear injection into fertilized oocytes from superovulated FVB mice. For all constructs expression was analyzed in multiple founder embryos and/or in independent transgenic lines, as detailed in Table 1. Embryos were stained for LacZ as previously described (Whiting et al., 1991). Transgenic lines were maintained by crossing with CD1 mice. RNA in situ analyses were performed as described (Conlon and Rossant, 1992). 4.2. Transgenic DNA constructs Overlapping BamHI and SacI digested fragments from the upstream region of Kreisler were isolated from genomic P1 and BAC clones obtained from Research Genetics, Inc. All fragments were cloned into the filled in NotI site of the hsp68LacZ expression construct, in which a minimal hsp68 promoter can direct LacZ expression in the presence of additional enhancers. For microinjection, Sal I digestion followed by DNA electrophoresis separated inserts from vector sequences, and inserts were purified using GeneClean beads (Bio101, Vista, CA). The specific mutation in the vHNF1 site of the S5 fragment was generated using the QuickChange mutagenesis system (Stratagene, LaJolla, CA). Multimerized binding sites were generated from doublestranded oligonucleotides of the vHNF1 binding site (5 0 -GCTATCCCGCTAATCATTAACCCAATGAGCATGTC-3 0 ). The number of vHNF1 binding sites inserted into the 2xvHNF1 and 5xvHNF1 constructs was determined by DNA sequencing. 4.3. Immunofluorescence An affinity purified anti-full-length Kreisler antibody was used in immunofluorescence experiments (Sadl et al., 2002;

1308

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309

Sadl et al., 2003). Immunohistochemistry using this antiKreisler antibody was performed on 12 mm cryosections, which had been postfixed in 4% paraformaldehyde, rinsed in PBS, and blocked in 10% goat serum. Sections were incubated overnight with affinity purified primary antibody at a 1:5000 dilution at 4 8C, rinsed three times in PBS, and incubated at room temperature for 1 h at room temperature with Cy3donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA). Slides were rinsed three times in PBS, mounted with VectaShield mounting medium (Vector Laboratories, Inc., Burlingame, CA), and examined microscopically. 4.4. Whole mount in situ RNA hybridizations Whole mount in situ RNA hybridizations were performed as previously described in Conlon and Rossant (1992). Hoxb1 and Kreisler probes were generated as previously described (Frohman et al., 1993; Cordes and Barsh, 1994). The vHNF1 probe, generously provided by Dr Silvia Cereghini, was prepared as described (Cereghini et al., 1992). 4.5. Electrophoretic mobility shift assays vHNF from zebrafish, generously provided by Drs Sun and Hopkins, was in vitro transcribed and translated using the TNT reticulolysate system. DNA binding reactions using in vitro transcribed and translated vHNF1 and either the radiolabeled S52 kb SacI to XbaI fragment or S5-vHNF1 doublestranded oligonucleotide (5 0 -GCTATCCCGCTAATCATTAACCCAATG) performed at room temperature for 20 min in a 20 ml reaction containing 20 mM Hepes, pH 7.9, 1 mM EDTA, 20 mM KCl, 5 mM DTT, 4 mM MgCl, 5% glycerol, 200 mg/ml poly(dIdC). Reactions were resolved by DNA electrophoresis on 4% non-denaturing polyacrylamide gels and analyzed by autoradiography. Mock TNT reactions served as negative controls and caused no mobility shifts. For competition experiments, double stranded oligonucleotides or gel purified DNA fragments were added in 3–100-fold molar excess of the radiolabeled probes at the start of the binding reactions. The oligonucleotide sequences used in these experiments were: S5vHNF1 (5 0 -GCTATCCCGCTAATCATTAACCCAATGAGCATGTC-3 0 ), S5-mutvHNF1 (5 0 -GCTATCCCGCGTATCAGGAACCCAATG-3 0 ), PE-vHNF1 (5 0 -TGTGGTTAATGA TCTACAGTTA-3 0 ), and DS34vHNF1 (5 0 -TGTGGTGTATGAGGTACAGTTA-3 0 ) (Costa and Grayson, 1991; Tronche et al., 1990, 1999). DNA fragments were obtained by digestion of the S5 region with XbaI and SacI, and purification of 1.3 and 2.0 kb fragments by gel electrophoresis and the Qiagen gel extraction kits.

were verified. The vHNF1 expression construct was generously provided by Gerald Crabtree. Cos-7 cells were maintained in Dulbecco’s modified Eagle medium (DMEM, high glucose) with 10% fetal bovine serum, penicillin (100 U/ml) and streptomycin (100 mg/ml) and transfected using the Lipofectamine Plus Reagent (Life Technologies, Inc., Rockville, MD), in keeping with manufacturer’s suggestions. Cells were seeded at 1.75!105 per 35 mm dish 18–24 h prior to transfection. Total DNA concentration per dish was 1.5 mg, with the reporter plasmid accounting for 425–550 ng, the internal control (pCMVb-gal) 1/ 20 (75 ng) of the total, and the expression plasmid or equivalent empty expression plasmids (pcDNA3). Each transfection mixture included 13 ml Plus Reagent and 5 ml Lipofectamine Reagent diluted in Opti-MEM I (Life Technologies, Inc., Rockville, MD). CAT (Roche Diagnostics, Laval, Que., Canada) and b-galactosidase assays were performed 36 h post-transfection. b-gal activity was used to normalize CAT activity. Sequence analyses. Sequence alignments were performed using standard Blastn searches and the ENSEMBL database initially, and later refined with the ECR browser (http://www. ECRbrowser.dcode.org) (Ovcharenko et al., 2004). Predicted vHNF1 transcription factor binding sites were identified via TRANSFAC database searches (http://www.gene-regulation. com) and rVISTA (http://rvista.dcode.org/) (Knuppel et al., 1994; Loots and Ovcharenko, 2004; Wingender et al., 1996). These analyses revealed homology between mouse chromosome 2: 160446648–160445202 (AL591665 nucleotides 119252– 119836), human chromosome 20: 38768899–38769453 (AL035665 nucleotides 111036–111620), and chick chromosome 20: 4452924–4453176 (contig 18.362 nucleotides 8492– 8745) of the current ENSEMBL database sequences.

Acknowledgements We would like to thank Drs Xhaoxia Sun and Nancy Hopkins for the generous gift of the pCS2-vHNF1 construct and for assistance, Dr Gerald Crabtree for the generous gift of the vHNF1 expression construct, and Dr Silvia Cereghini for the kind gift of the vHNF1 probe for RNA in situ hybridization, Dr Janet Rossant for critical reading of the manuscript and the hsp68LacZ construct, Dr Mark Featherstone for suggestions, and Dragana Vukasovic for excellent secretarial assistance. We are grateful to Sandra Tondat, Lois Schwartz, and the Samuel Lunenfeld transgenic core facility for generating transgenic mice for these studies. This research was funded by CIHR grant No. 14312 to S.P.C. S.P.C. is an MRC/CIHR scholar, a Canadian Foundation of Innovation young investigator, and an EJLB foundation scholar.

4.6. Cell culture and transfection

References

The (PE)2-CAT, (S5)2-CAT, and (S5mut)2-CATreporter plasmids were generated similarly by inserting dimers of the PE, S5, and S5mut-vHNF1 sites described above into the BamHI site of pBLCAT2 (Seto et al., 1990). Sequences of all constructs

Barbacci, E., Reber, M., Ott, M.O., Breillat, C., Huetz, F., Cereghini, S., 1999. Variant hepatocyte nuclear factor 1 is required for visceral endoderm specification. Development 126, 4795–4805. Baumhueter, S., Courtois, G., Crabtree, G.R., 1988. A variant nuclear protein in dedifferentiated hepatoma cells binds to the same functional sequences in

F.A. Kim et al. / Mechanisms of Development 122 (2005) 1300–1309 the beta fibrinogen gene promoter as HNF-1. Eur. Mol. Biol. Org. J. 7, 2485–2493. Cereghini, S., Ott, M.O., Power, S., Maury, M., 1992. Expression patterns of vHNF1 and HNF1 homeoproteins in early postimplantation embryos suggest distinct and sequential developmental roles. Development 116, 783–797. Coffinier, C., Barra, J., Babinet, C., Yaniv, M., 1999a. Expression of the vHNF1/HNF1beta homeoprotein gene during mouse organogenesis. Mech. Dev. 89, 211–213. Coffinier, C., Thepot, D., Babinet, C., Yaniv, M., Barra, J., 1999b. Essential role for the homeoprotein vHNF1/HNF1beta in visceral endoderm differentiation. Development 126, 4785–4794. Conlon, R.A., Rossant, J., 1992. Exogenous retinoic acid rapidly induces anterior ectopic expression of murine Hox-2 genes in vivo. Development 116, 357–368. Cordes, S.P., Barsh, G.S., 1994. The mouse segmentation gene kr encondes a novel basic domain-leucine zipper transcription factor. Cell 79, 1025–1034. Costa, R.H., Grayson, D.R., 1991. Site-directed mutagenesis of hepatocyte nuclear factor (HNF) binding sites in the mouse transthyretin (TTR) promoter reveal synergistic interactions with its enhancer region. Nucleic Acids Res. 19, 4139–4145. Dupe, V., Ghyselinck, N.B., Wendling, O., Chambon, P., Mark, M., 1999. Key roles of retinoic acid receptors alpha and beta in the patterning of the caudal hindbrain, pharyngeal arches and otocyst in the mouse. Development 126, 5051–5059. Fraser, S., Keynes, R., 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., Barsh, G.S., 1993. Altered rhombomere-specific gene expression and hyoid bone differentiation in the mouse segmentation mutant, kreisler (kr). Development 117, 925–936. Grapin-Botton, A., Bonnin, M.A., Sieweke, M., 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. Hernandez, R.E., Rikhof, H.A., Bachmann, R., Moens, C.B., 2004. vhnf1 integrates global RA patterning and local FGF signals to direct posterior hindbrain development in zebrafish. Development 131, 4511–4520. Knuppel, R., Dietze, P., Lehnberg, W., Frech, K., Wingender, E., 1994. TRANSFAC retrieval program: a network model database of eukaryotic transcription regulating sequences and proteins. J. Comput. Biol. 1, 191–198. Kuo, C.J., Mendel, D.B., Hansen, L.P., Crabtree, G.R., 1991. Independent regulation of HNF-1 alpha and HNF-1 beta by retinoic acid in F9 teratocarcinoma cells. Eur. Mol. Biol. Org. J. 10, 2231–2236. Lee, Y.H., Sauer, B., Gonzalez, F.J., 1998. Laron dwarfism and non-insulindependent diabetes mellitus in the Hnf-1alpha knockout mouse. Mol. Cell Biol. 18, 3059–3068. Loots, G.G., Ovcharenko, I., 2004. rVISTA 2.0: evolutionary analysis of transcription factor binding sites. Nucleic Acids Res. 32, W217–W221. Lumsden, A., Krumlauf, R., 1996. Patterning the vertebrate neuraxis. Science 274, 1109–1115. Manzanares, M., Cordes, S.P., Kwan, C.-T., Sham, M.-H., Barsh, G.S., Krumlauf, R., 1997. Segmental regulation of Hoxb3 by Kreisler. Nature 387, 191–195. Manzanares, M., Cordes, S., Ariza-McNaughton, L., Sadl, V., Maruthainar, K., Barsh, G., Krumlauf, R., 1999. Conserved and distinct roles of kreisler in regulation of the paralogous Hoxa3 and Hoxb3 genes. Development 126, 759–769. Marin, F., Charnay, P., 2000. Hindbrain patterning: FGFs regulate Krox20 and mafB/kr expression in the otic/preotic region. Development 127, 4925–4935. Maves, L., Jackman, W., Kimmel, C.B., 2002. FGF3 and FGF8 mediate a rhombomere 4 signaling activity in the zebrafish hindbrain. Development 129, 3825–3837.

1309

Mendel, D.B., Hansen, L.P., Graves, M.K., Conley, P.B., Crabtree, G.R., 1991. HNF-1 alpha and HNF-1 beta (vHNF-1) share dimerization and homeo domains, but not activation domains, and form heterodimers in vitro. Genes Dev. 5, 1042–1056. Murphy, P., Davidson, D.R., Hill, R.E., 1989. Segment-specific expression of a homoeobox-containing gene in the mouse hindbrain. Nature 341, 156–159. Ott, M.O., Rey-Campos, J., Cereghini, S., Yaniv, M., 1991. vHNF1 is expressed in epithelial cells of distinct embryonic origin during development and precedes HNF1 expression. Mech. Dev. 36, 47–58. Ovcharenko, I., Nobrega, M.A., Loots, G.G., Stubbs, L., 2004. ECR Browser: a tool for visualizing and accessing data from comparisons of multiple vertebrate genomes. Nucleic Acids Res. 32, W280–W286. Papalopulu, N., Lovell-Badge, R., Krumlauf, R., 1991. The expression of murine Hox-2 genes is dependent on the differentiation pathway and displays a collinear sensitivity to retinoic acid in F9 cells and Xenopus embryos. Nucleic Acids Res. 19, 5497–5506. Pontoglio, M., Barra, J., Hadchouel, M., Doyen, A., Kress, C., Bach, J.P., et al., 1996. Hepatocyte nuclear factor 1 inactivation results in hepatic dysfunction, phenylketonuria, and renal Fanconi syndrome. Cell 84, 575–585. Rey-Campos, J., Chouard, T., Yaniv, M., Cereghini, S., 1991. vHNF1 is a homeoprotein that activates transcription and forms heterodimers with HNF1. Eur. Mol. Biol. Org. J. 10, 1445–1457. Sadl, V.S., Jin, F., Yu, J., Cui, S., Holmyard, D., Quaggin, S.E., et al., 2002. The mouse Kreisler (Krml1/MafB) segmentation gene is required for differentiation of glomerular visceral epithelial cells. Dev. Biol. 249, 16–29. Sadl, V.S., Sing, A., Mar, L., Jin, F., Cordes, S.P., 2003. Analysis of hindbrain patterning defects caused by the kreislerenu mutation reveals multiple roles of Kreisler in hindbrain segmentation. Dev. Dyn. 227 (1), 134–142. Seto, E., Mitchell, P.J., Yen, T.S., 1990. Transactivation by the hepatitis B virus X protein depends on AP-2 and other transcription factors. Nature 344, 72–74. Sun, Z., Hopkins, N., 2001. vhnf1, the MODY5 and familial GCKD-associated gene, regulates regional specification of the zebrafish gut, pronephros, and hindbrain. Genes Dev. 15, 3217–3229. Tronche, F., Yaniv, M., 1992. HNF1, a homeoprotein member of the hepatic transcription regulatory network. Bioessays 14, 579–587. Tronche, F., Rollier, A., Herbomel, P., Bach, I., Cereghini, S., Weiss, M., Yaniv, M., 1990. Anatomy of the rat albumin promoter. Mol. Biol. Med. 7, 173–185. Tronche, F., Kellendonk, C., Kretz, O., Gass, P., Anlag, K., Orban, P.C., et al., 1999. Disruption of the glucocorticoid receptor gene in the nervous system results in reduced anxiety. Nat. Genet. 23, 99–103. Walshe, J., Maroon, H., McGonnell, I.M., Dickson, C., Mason, I., 2002. Establishment of hindbrain segmental identity requires signaling by FGF3 and FGF8. Curr. Biol. 12, 1117–1123. Wang, P.W., Eisenbart, J.D., Cordes, S.P., Barsh, G.S., Stoffel, M., Le Beau, M.M., 1999. Human KRML (MAFB): cDNA cloning, genomic structure, and evaluation as a candidate tumor suppressor gene in myeloid leukemias [In Process Citation]. Genomics 59, 275–281. Whiting, J., Marshall, H., Cook, M., Krumlauf, R., Rigby, P.W., Stott, D., Allemann, R.K., 1991. Multiple spatially specific enhancers are required to reconstruct the pattern of Hox-2.6 gene expression. Genes Dev. 5, 20482059. Wiellette, E.L., Sive, H., 2003. vhnf1 and Fgf signals synergize to specify rhombomere identity in the zebrafish hindbrain. Development 130, 3821– 3829. Wilkinson, D.G., Bhatt, S., Cook, M., Boncinelli, E., Krumlauf, R., 1989. Segmental expression of Hox-2 homoeobox-containing genes in the developing mouse hindbrain. Nature 341, 405–409. Wingender, E., Dietze, P., Karas, H., Knuppel, R., 1996. TRANSFAC: a database on transcription factors and their DNA binding sites. Nucleic Acids Res. 24, 238–241.