A cluster of three long-range enhancers directs regional Shhexpression in the epithelial linings

The Shh pathway plays a key role in multiple aspects of morphogenesis and development in vertebrate embryos(Chiang et al., 1996; Ingham and McMahon, 2001; Varjosalo and Taipale, 2008),including dorsoventral patterning of the central nervous system (CNS)(Echelard et al., 1993; Roelink et al., 1994) and anterorposterior patterning of the limb(Riddle et al., 1993). It also plays an important role in the morphogenesis in the epithelial flat sheet from the oral cavity to the hindgut, where a vast array of mature morphologies are exhibited (Chuong et al.,2000). Targeted deletion of Shh has shown a role for Shh signaling in the development of feathers(Yu et al., 2002), hair(St-Jacques et al., 1998),teeth (Dassule et al., 2000),lingual papillae (Hall et al.,2003), pharyngeal arches and pouches(Moore-Scott and Manley,2005), foregut (Litingtung et al., 1998), and urogenital tracts(Haraguchi et al., 2001). However, the role of the Shh pathway in later stages of epithelial morphogenesis is poorly understood, because these tissues are severely malformed in Shh knockout (KO) mice.

A key step towards understanding the regulatory networks that control gene expression during morphogenesis and organogenesis is the identification of cis-regulatory elements of developmentally important genes. With the increasing availability of vertebrate genomic sequences, comparison of sequences across evolutionarily distant species has permitted the identification of numerous conserved non-coding sequences (CNCSs)(Abbasi et al., 2007; Boffelli et al., 2004; Frazer et al., 2004; Ghanem et al., 2003; Goode et al., 2005; Santagati et al., 2003; Woolfe and Elgar, 2007; Woolfe et al., 2007; Woolfe et al., 2005). Functional analysis of these elements has been carried out using BAC reporter transgenesis in mice (Gong et al.,2002; Jeong et al.,2006) and GFP reporter assays in zebrafish(Goode et al., 2005; McEwen et al., 2006; Woolfe et al., 2005).

We have previously reported that deletion of a CNCS located 840 kb upstream of the transcriptional start site of Shh results in a marked phenotype (Sagai et al.,2005). This CNCS is conserved among all tetrapods species examined, as well as in teleost fishes(Lettice et al., 2003; Sagai et al., 2004). Although the CNCS KO mice are viable, endogenous limb bud expression of Shh is completely lost, resulting in severe distal limb truncation indistinguishable from that observed in the KO mutant of the Shh-coding sequence(Chiang et al., 1996). Subsequent cis-trans tests verified that the CNCS contains a limb-specific Shh enhancer (Lettice et al.,2002; Sagai et al.,2004).

The 1 Mb genomic region spanning from the Shh coding region to the upstream limb-specific enhancer is unique in its low gene density and shows exceptionally long-range synteny between mammals and teleost fishes, with a number of CNCSs lining up in the same order and orientation in different species (Goode et al., 2005; Woolfe et al., 2005). Comparative sequence analysis and transgenic mouse reporter assays have uncovered three forebrain enhancers located 300 to 450 kb upstream of the Shh-coding sequence (Jeong and Epstein, 2003; Jeong et al.,2006). Displacement of these regulatory elements from the Shh promoter by chromosomal translocation is a likely cause of holoprosencephaly (HPE) in humans(Roessler et al., 1997),underscoring the importance of long-range enhancer elements in key development processes. Together with floor-plate enhancers near the transcriptional start site, these enhancers recapitulate Shh expression in the mouse embryonic central nervous system (Epstein et al., 1999; Jeong and Epstein, 2003; Jeong et al.,2006; Jeong et al.,2008). However, to date, epithelial linings-specific Shhenhancers have not been reported.

Here, we explore new CNCSs in the mouse 300 kb genomic region 600 to 900 kb upstream of the Shh-coding sequence by comparing the mouse genome with the genomes of other mammalian species and teleost fishes. We identify a cluster of three CNCSs that drive lacZ reporter expression in the epithelia of the oral apparatus, the pharyngeal apparatus, and the lung and gut. Interestingly, the co-linear genomic arrangement of the three CNCSs mirrors the anteroposterior order of their expression domains, partitioning the continuous epithelial lining into three Shh expression domains:the ectoderm-derived oral cavity, the anterior endoderm-derived pharynx, and posterior endoderm-derived respiratory and digestive tubes. We also generate KO mouse mutants that lack the CNCS that drives expression in the pharyngeal apparatus. In these animals, endogenous Shh expression is lost specifically in the pharyngeal epithelia, resulting in postnatal lethality owing to hypoplasia of the soft palate, epiglottis and arytenoid that are essential for respiratory and swallowing functions. These results demonstrate that the Shh pathway is essential for the morphogenesis and the development of the pharyngeal region in the mouse.

The Shh-coding sequence KO mouse(Shh^–/–) was kindly provided by Dr P. Beachy,and is maintained at the National Institute of Genetics (NIG), Mishima, Japan. C57BL/6J and (C57BL/6×DBA/2)F1 mice were purchased from Japan Crea(Tokyo, Japan). The animal experiments in this study were approved by the Animal Care and Use Committee of the NIG.

For sequence alignment and homology comparisons, we used the ClustalW system(http://www.ddbj.nig.ac.jp/search/clustalw-j.html),VISTA program(http://genome.lbl.gov/vista/index.shtml),UCSC genome Browser Home(http://genome.ucsc.edu/)and Ensemble database(http://www.ensembl.org/index.html). The draft genome sequences of humans, mice and medaka fish have been previously described. For homology comparison, we used the following sequences from the UCSC database: mouse Chr5, 28,783,380-29,704,930; human Chr7,155,294,520-156,378,6670; chicken Chr2, 8,031,550-8,431,420; Xenopus Chr28,4,064,830-4,399,140; medaka Chr20, 17,738,740-17,852,820. The medaka genome sequence is also referenced from the NIG DNA sequence center(http://dolphin.lab.nig.ac.jp/medaka/index.php). The sequence data of the CNCSs MRCS1, MFCS3, MFCS4, MACS1 and MFCS2 have been submitted to DDBJ under Accession Numbers AB453051, AB453050, AB258402,AB453049 and AB453052, respectively. MFCS3 and MFCS4 are mouse orthologs of the previously described fugu or human sequences,SHH2 (Accession Number CR847489) and SHH1 (Accession Number CR847488)(Woolfe et al., 2005). The genomic positions of CNCSs in different species, which were used for the Vista analysis, are listed in Table S1 in the supplementary material.

Mouse genomic DNA fragments, including the CNCSs were amplified from RP23-284A9 or RP23-428P20 BAC DNA. After sequencing, the amplified fragments were inserted into the HindIII or SalI site of the hsp68/LacZ expression vector(Shashikant et al., 1995). Details of the primer pairs used for amplification of the inserts can be provided on request. To obtain the MFCS4 fragment lacking a 217 bp ultra-conserved sequence, inverted tail-to-tail primer pairs were used to amplify a basic whole vector, except for the 217 bp ultra-conserved sequence of MFCS4. The primer pair used was: F, 5′-AGATTGGGTTCACTGTGTGC-3′;R, 5′-CACAAGCCTCTTTAGTCAGG-3′. Then, the deleted form of MFCS4(ΔMFCS4) was subcloned into the lacZ reporter construct. The XhoI and NotI double-digested fragments were cut out from an 0.8% low melting agarose gel and digested with GELase enzyme (Epicentre Technologies, Madison, WI, USA) at 43°C overnight. After phenol and chloroform extraction, DNA was precipitated with ethanol and dissolved in a small volume of injection buffer [5 mM Tris-HCl (pH 7.5); 0.1 mM EDTA (pH 8.0)]. DNA (1-4 ng/μl) was purified using a filter unit and used in injection experiments. Transient transgenic embryos and stable transgenic mouse lines were generated by pronuclear injection into fertilized eggs derived from the (C57BL/6×DBA/2)F1 or C57BL/6 strain. Transgenic animals were selected using the following primer pairs for the lacZ gene: F,5′-TCACCCTGCCATAAAGAAACT-3′; R,5′-CTGTCGTCGTCCCCTCAAACT-3′. Whole-mount lacZ staining was carried out as previously described(Maas and Fallon, 2005). For histological analysis of transgenic embryos, embryos were fixed overnight in 4% paraformaldehyde, dehydrated in an ethanol series and embedded in paraffin. Sections were cut at 5 μm and counterstained with acidic Fast Red.

We used a previously described basic targeting vector to build the MFCS4 targeting construct (Sagai et al.,2005). The long arm (5478 bp) was amplified from BAC RP23-284A9 DNA with the primer pair 5′-ATGGTACCAGGAGATATGCTGCATCCTC-3′ and 5′-TACTCGAGAGAACTGCGGTTTAACCTGC-3′, and the short arm (1824 bp)was amplified with the primer pair 5′-CCGGAATTCGCATTAGAAGCTGGGATGGA-3′ and 5′-CGCGAATTCGGACCTTACATACGTGAAGC-3′. The 999 bp genomic sequence,including mouse MFCS4, was replaced with the Neo cassette (see Fig. S1 in the supplementary material). The targeting vector was electroporated into TT2 ES cells, which originated from a (C57BL/6×CBA)F₁ mouse(Yagi et al., 1993). ES cells were screened with the following PCR primer pair: p1,5′-AGTGCTGTCCCAGAGATAAG-3′ and p2,5′-CATCGCATTGTCTGAGTAGG-3′. Positive clones were aggregated with eight-cell embryos from (DBA/2×C57BL/6)F₁ mice and transplanted into surrogate females. Male chimeras were mated with C57BL/6 females. Segregation of the targeted allele was determined using three PCR primers: p2, p3, 5′-TCTCAATCTGAACACTGGGC-3′, and p4,5′-TCTCAATCTGAACACTGGGC-3′. Skeletal analysis of newborn mice was performed as previously described(Trokovic et al., 2003). For the histological analysis of the ΔMFCS4/ΔMFCS4 mutants, embryos at E18.5 were preserved in Bouin's fixative and embedded in paraffin. Serial sections of 5 μm were collected and stained with Hematoxylin-Eosin. We referred to an anatomical atlas (Kaufman,1992) throughout the experiments.

For cryosectioning, embryos were fixed in 4% paraformaldehyde and immersed in 30% sucrose/phosphate-buffered saline overnight, embedded and frozen in OCT, and sectioned at 15 μm. For paraffin sections, embryos were fixed with 4% paraformaldehyde and dehydrated in a methanol series, then embedded in paraffin and sectioned at 8 μm. In situ hybridization was performed using digoxigenin-UTP-labeled riboprobes, as previously described(Makino et al., 2001). Whole-mount in situ hybridization was performed as previously described(Wilkinson, 1992).

Total RNA was extracted from the anterior tongue, epiglottis-arytenoid swelling and lung tissues of C57BL/6 embryos at E13.5. One μg of each RNA sample was reverse transcribed into cDNA with SuperScript III Transcriptase(Invitrogen). One μl of the cDNA solution was used for PCR amplification. Details of primer pairs used can be provided on request.

We compared the mouse genomic sequence 600-900 kb upstream of the Shh transcriptional start site with the syntenic regions of various tetrapod species and teleost fishes, using public genome databases. Previously, we have identified three CNCSs in this region that are conserved between mammals and teleost fishes and that we called Mammal Fish Conserved Sequence 1 to 3 (MFCS1 to 3) (Fig. 1) (Sagai et al.,2005). Here, we describe the identification of three additional CNCSs in this region (Fig. 1A)that are evolutionarily conserved at different level(Fig. 1B). MFCS4 is conserved between mammals and teleost fishes, albeit with slightly lower sequence similarity than MFCS1-3 (Fig. 1A,B), and is an ortholog of a human-fugu conserved sequence called SHH1, which has been reported by another group(Goode et al., 2005; Woolfe et al., 2005). Mammal Reptile Conserved Sequence 1 (MRCS1) is conserved among mammals, chicken and lizard, but not in Xenopus and teleost fishes. Mammal Amphibian Conserved Sequence 1 (MACS1) is conserved among mammals, chicken, lizard and Xenopus, and has very weak similarity to a short sequence in the medaka genome (Fig. 1B). MACS1 is a partial cDNA fragment previously registered as AK043126 in the FANTOM data set (Carninci and Hayashizaki,1999) and is located in intron 8 of Rnf32, a gene with unknown function.

Thus, in total there are six CNCSs presently known to reside in the 600 to 900 kb region upstream of the Shh transcriptional start site(Fig. 1). Notably, the order and orientation of these CNCSs relative to direction of Shhtranscription are conserved across evolutionarily distant species.

To test the function of the newly identified CNCSs, we carried out transgenic assays by examining β-galactosidase activity in mouse embryos carrying a ∼1 kb mouse genomic sequence containing the five different CNCSs linked to the lacZ reporter gene. We first examined reporter gene expression in whole-mount transgenic embryos at embryonic day (E)9.0-15.5. As summarized in Table 1, MFCS3, MFCS4, MRCS1 and MACS1 directed reproducible tissue-specific reporter gene expression, whereas MFCS2 did not show tissue-specific expression. MFCS3 drives lacZ expression in the brain and motor neurons at E11.5 (see Fig. S2 in the supplementary material),whereas the other three CNCSs (MFCS4, MRCS1 and MACS1) drive expression in different epithelial linings.

MRCS1 drives strong lacZ expression in the epithelia of the incisor and molar teeth at E12.5 (Fig. 2A,B) and weak expression in the lingual papillae(Fig. 2B). At E13.5, lacZ signal appears in the whisker buds, tooth primordia, the rugae of the hard palate (Fig. 2C). Expression in the fungiform papilla of the anterior tongue also becomes more prominent at this stage (Fig. 2D). Expression in these domains continues at least until the palatal structure is established at E15.5(Fig. 2E-G). At E14.5-15.5, the reporter signals were also detected in the hair and nail buds(Fig. 2E).

In the MFCS4 transgenic mice, very weak signal was detected in the oral epithelium at E11.0 (data not shown). At E11.5, weak and transient signal was detected in the dental germ (Fig. 2H). Strong expression was observed in the pituitary fossa region and in the tympanic tube and recess (Fig. 2H). In the lower facial area, punctuate expression was detected on the tongue surface, but not in the fungiform papillae (inset in Fig. 2H). lacZexpression persists in these regions until E13.5(Fig. 2I,J). Notably, signal was detected in the palatal shelves of the soft palate, but not in the anterior half of the palate or in the hard palate(Fig. 2I), where MRCS1 drives reporter expression (Fig. 2C,F). Strong expression was detected in the epiglottis and arytenoid swelling at E13.5 (Fig. 2J), which are essential for respiratory and swallowing functions. At E15.0, palatal formation is completed and signal was observed in the midline of the soft palate (Fig. 2K), but was decreased in the epiglottis and arytenoid swelling(Fig. 2L). Histological analysis showed that MFCS4-driven expression is restricted to the epithelial lining (Fig. 2M).

MACS1 drives lacZ reporter expression in the gut endoderm and caudal region at E9.0 (Fig. 3A). At E9.5, intense signal marks the region where the future respiratory tube will evaginate from the primitive gut(Fig. 3B). As shown in the sagittal section, signal was detected throughout the continuous lining of the gut endoderm at E9.5 (Fig. 3C). At E10.5, signal was observed in the gut and cloaca(Fig. 3D), as well as the primary lung buds (Fig. 3E). At E10.5, intense signal was apparent in the epithelia of the laryngo-tracheal tube and cloaca. At E11.5, strong expression was observed in the laryngotracheal tube, lung, gut (Fig. 3G) and genital tubercle (Fig. 3H). A mid-sagittal section confirmed the intense signal in the epithelia of the laryngotracheal tube, lung, digestive tube and urogenital tract (Fig. 3I; data not shown)at E12.5. Thus, MACS1 strictly specifies the expression in the epithelia of the respiratory and digestive tubes, which are derived from the primitive gut endoderm.

We observed endogenous expression of Shh and its downstream gene Ptch1, and compared it with lacZ reporter expression(Fig. 4). Endogenous Shh expression is observed in the epithelia of the tooth primordia,lingual papillae, palate, epiglottis, arytenoid swelling, lung, gut and urogenital tracts at E12.5-13.5 (Fig. 4A,C,E,G), whereas Ptch1 endogenous expression is mainly observed in the mesenchyme of the tooth primordia, tongue, epiglottis,arytenoid swelling, lung and gut at these stages(Fig. 4B,D,F,H). As described above, the reporter signals driven by MRCS1, MFCS4 and MACS1 are confined to the epithelial linings (Fig. 2M; Fig. 3C,F,I),and mirror the endogenous Shh expression domains. Even though the MFCS4 and MACS1 expression domains partially overlap at their borders in the arytenoid swelling (Fig. 4I-L),they are distinct, with MFCS4 driving expression in the epiglottis(Fig. 2M; Fig. 4I) and MACS1 driving expression in the laryngotracheal tube(Fig. 4J,L). The MFCS4 and MACS1 tandem construct drives reporter expression throughout the pharyngeal apparatus and the respiratory and digestive tubes(Fig. 2M; Fig. 3I; Fig. 4M). Thus, these three CNCSs appear to contain enhancers that together recapitulate the endogenous pattern of Shh expression in the epithelial linings.

To examine whether the CNCSs act as functional elements in vivo, we performed gene targeting, focusing on MFCS4. Following the same strategy used to generate MFCS1 knockout mice (Sagai et al., 2005), we replaced a ∼1 kb genomic region, including MFCS4 with a neo cassette. We successfully generated two lines of germline chimeras, then, heterozygotes (ΔMFCS4/+) were intercrossed to generate homozygous mutants (ΔMFCS4/ΔMFCS4)(Table 2).ΔMFCS4/ΔMFCS4 mice were present at E18.5 and early postnatal stages, but all of them died within a few days of birth(Table 2). AlthoughΔMFCS4/+ mice are viable and grossly indistinguishable from wild-type mice, many ΔMFCS4/ΔMFCS4 neonates have bloated bellies, which might be caused by air accumulation in the stomach and bowels(Fig. 5A), and their stomachs are always devoid of milk (Fig. 5B). The tongue, soft palate, epiglottis and arytenoid comprise the pharyngeal apparatus, which is essential for respiration and swallowing(Fig. 5C,D,F,G). In theΔMFCS4/ΔMFCS4 mice, truncation of the soft palate(Fig. 5E), loss or reduced size of the epiglottis and hypotrophy of the arytenoid(Fig. 5H,I), and tongue deformation (Fig. 5H) are observed. Moreover, deletion of MFCS4 appears to result in minor morphological abnormalities of the basisphenoid and basioccipital bones, as is visible from a ventral view of the skull (Fig. 5J,K). The cartilage between these bones, which is probably the pituitary fossa region that closes after formation of the pituitary, has a hole (marked with the broken white line in Fig. 5K). TheΔMFCS4/ΔMFCS4 mice display defects in the hyoid bone and thyroid cartilage (Fig. 5L,M). Although there is variability in the severity of the defects, the major phenotype was completely penetrant, with abnormalities observed in the aforementioned structures in all the ΔMFCS4/ΔMFCS4 mice examined(n=17).

Histological analysis of the sagittal sections showed that theΔMFCS4/ΔMFCS4 embryos have deformed posterior tongues, hypoplasia of the hyoid cartilage, loss of the epiglottis, hypoplasia of the arytenoid and shortening of the soft palate (Fig. 5O), which eventually resulted in atypical junctions between the nasopharynx and oropharynx. The posterior tongue edge, including hyoid bone and thyroid cartilage was deformed (Fig. 5M,O), and the base of tongue seemed to be affected as well, which eventually result in the atypical tongue(Fig. 5H). Though the transgenic MFCS4 reporter signal was detected in the tympanic tube and recess,visible defects were not obvious in the auditory organs of theΔMFCS4/ΔMFCS4 neonates. Most of the ΔMFCS4/ΔMFCS4 had normal hard palate, and a few neonates (5/37) exhibited cleft palate (data not shown). Notably, in the ΔMFCS4/ΔMFCS4 neonates, the tooth and whisker, in which the MFCS4-mediated lacZ expression was not observed, but the MRCS1-mediated lacZ expression was observed, were not affected (inset in Fig. 5O;data not shown). As described before, the reporter expression domains driven by MFCS4 and MACS1 are partially overlapped in the arytenoid swelling. It is consistent with the fact that the posterior border of the defects observed in the ΔMFCS4/ΔMFCS4 neonates corresponds to the posterior end of the pharynx. The ΔMFCS4/ΔMFCS4 neonates showed no visible defects in the cricoid cartilage, trachea, lung, esophagus, digestive tube and urogenital organs (see Table S2 in the supplementary material). Thus, MFCS4 is crucial for development specifically of the pharyngeal apparatus.

We next examined whether the Shh pathway is altered in theΔMFCS4/ΔMFCS4 embryos by whole-mount and section in situ hybridization using a Shh riboprobe. Strong Shh expression was detected in the epithelium of the epiglottis and arytenoid swelling of wild-type embryos at E13.0 and E13.5 (Fig. 4A,C; Fig. 6A,C),whereas the ΔMFCS4/ΔMFCS4 embryos lost almost all Shhexpression in the epithelia of the epiglottis and arytenoid swelling(Fig. 6B,D). However, Shh expression in the tooth primordia was not altered in the mutant embryos (Fig. 6D), in accordance with the normal tooth development in the ΔMFCS4/ΔMFCS4 embryos (Fig. 5O).

Finally, we carried out a cis-trans test by mating animals heterozygous for the Shh-coding sequence KO allele (Shh^+/–)to the ΔMFCS4/+ mouse. All compound heterozygotes exhibited hypoplasia of the epiglottis and arytenoids, indicating that MFCS4 is cis-acting to the Shh-coding sequence (Fig. 6E,F). Together, our results indicate that MFCS4 is a pharyngeal epithelium-specific enhancer of Shh, and is indispensable for morphogenesis of the pharyngeal structures in the mouse.

Alignment of the MFCS4 nucleotide sequences across distant species revealed a 217 bp ultra-conserved sequence, which is conserved from mammals to teleost fish medaka (Fig. 1A; see Fig. S3 in the supplementary material). To assess importance of the sequence for the MFCS4 enhancer activity, we carried out a transgenic assay for the lacZ reporter construct, in which the MFCS4 fragment lacks the 217 bp sequence (see Fig. S3 in the supplementary material). The result showed that this deletion form of MFCS4 drives no reporter expression in the relevant tissues of the transgenic embryos (0/5). It supports the fact that the 217 bp ultra-conserved sequence is indispensable for the MFCS4 enhancer activity. To explore transcription factors that bind to this sequence, we carried out the evolutionary rigidity assay (ERA) for the 217 bp sequence. We identified four nearly perfect match motifs (NPMMs) (see Fig. S3 in the supplementary material), which are potentially binding sites for Pbx1 (MFCS4-MF-A), Sox5(MFCS4-MF-B), TCFs (MFCS4-MF-C), inner-cell mediators of Wnt signaling and Hes1 (MFCS4-MF-D), a well known target of Notch signaling.

A number of reports suggest that Pbx, Sox, Wnt and Notch signaling pathways are indispensable for normal development of epithelial tissues such as tooth,hair follicle, taste papillae, lung and gastrointestinal tract(Ito et al., 2000; Iwatsuki et al., 2007; Li et al., 2005; Okubo et al., 2006; Schnabel et al., 2001). We examined expression of many genes involving these signaling pathways in the tongue, epiglottis and lung. As shown in Fig. S4 in the supplementary material, most of the genes are activated in the epiglottis and arytenoid swelling at E13.5. The results suggested that they are good candidates for the MFCS4 activity.

In this study, we have identified three novel CNCSs in the region 620 to 740 kb upstream of the Shh transcriptional start site. Their sequences, order and orientation are highly conserved between evolutionarily distant species (Fig. 1A; Fig. 7A). These CNCSs direct regional expression of Shh in the epithelial linings from the oral cavity to the hindgut along the anteroposterior body axis(Fig. 7B). At the most anterior extent, MRCS1 directs Shh expression in the epithelia of the hair and whisker buds, dental placode, rugae of the hard palate and fungiform papillae of the anterior tongue, all of which are presumptive derivatives of the oral ectoderm. MFCS4 drives expression in the soft palate, epiglottis, arytenoid swelling and other pharyngeal tissues, forming a border with the more anterior MRCS1-directed expression domain. Finally, MACS1 drives expression in the respiratory and digestive tubes, with the anterior border of expression partially overlapping the posterior border of MFCS4-driven expression in the pharyngeal structures. Thus, this study clearly shows that three distinct enhancers regulate Shh expression along the long continuous epithelial linings from the oral cavity to the hindgut. Interestingly, the genomic order of these three enhancers is co-linear with the regional control of Shh expression along the anteroposterior body axis(Fig. 7). This suggests that,at least with regard to the Shh pathway, morphogenesis of the epithelial linings is roughly partitioned into three components: the oral ectoderm, pharyngeal endoderm and gut endoderm. At present, the biological implications of the co-linearity of these enhancers are uncertain, but it is possible that clustering of the three enhancers is required for the proper spatiotemporal regulation of Shh expression.

Endogenous Shh expression is initiated progressively along the continuous epithelial linings spanning from the oral cavity to the hindgut(Bitgood and McMahon, 1995; Iseki et al., 1996; Varjosalo and Taipale, 2008). In the anterior ectoderm-derived epithelial lining, Shh expression starts around E11.5-12.5 and is implicated in development of the oral cavity,the tooth and the tongue (Cobourne et al.,2004; Hall et al.,1999). At E11.5-12.5, the pharyngeal arches become broadened and flattened externally, and form the neck of the embryo. Around this stage, Shh expression is observed in the pharyngeal epithelial lining(Fig. 4E,G)(Rice et al., 2006). In the primary gut, the earliest Shh expression is detected around E8.0 and is necessary for the formation of the lung, gut and urogenital organs(Haraguchi et al., 2007; Litingtung et al., 1998). Our transgenic experiments showed that timing of reporter expression onset driven by MRCS1, MFCS4 and MACS1 in each lining is consistent with the endogenous temporal expression pattern of Shh, suggesting that the three enhancers identified in this study regulate endogenous Shhexpression. However, it is notable that the co-linearity in the temporal expression patterns driven by the three enhancers is not as obvious as that observed for the regional expression patterns.

At present we have no compelling interpretation for the biological implication of the long-range enhancers that regulate the Shhexpression. Our very recent study (Amano et al., 2009) demonstrated that chromosome dynamics is involved in Shh expression in the mouse developing limb bud. A long-range limb bud-specific enhancer, MFCS1, specifically interacts with the Shhpromoter via change of the chromosome conformation. The Shhexpression regulated by the epithelial lining-specific enhancers identified in this study is probably exerted by the similar mechanism. Therefore, it is of interest to investigate how the co-linear Shh expression in the epithelial linings along the anteroposteior body axis is regulated by orchestrated chromosome dynamics.

The anterior endoderm-derived pharynx is the region of the digestive tube anterior to the point where the respiratory tube branches off. Loss-of-function and misexpression experiments have shown that the Shh pathway functions in the pharyngeal endoderm to generate multiple organs, including the pancreas, pituitary gland, parathyroid gland and jaw(Brito et al., 2006; Hebrok, 2003; Litingtung et al., 1998; Sbrogna et al., 2003; Treier et al., 2001). However,owing to severe abnormalities in the Shh KO mutant mouse, the function of Shh at later stages of the pharyngeal morphogenesis has been poorly understood. In this study, we examined a functional requirement for the pharyngeal epithelium-specific Shh enhancer, MFCS4, by targeted deletion of MFCS4 in mice. The result clearly showed that MFCS4 is essential for morphogenesis of the pharyngeal structures necessary for respiration and swallowing, including the soft palate, epiglottis and arytenoid. As shown in Table S2 in the supplementary material, relevance between domains of the reporter expression driven by the three epithelial Shh enhancers and the phenotype in the ΔMFCS4/ΔMFCS4 neonates shows a crucial role of the Shh pathway in the morphogenesis of the pharyngeal epithelium. Moreover, detailed characterization of the phenotype of the ΔMFCS4/ΔMFCS4 embryos suggested that the Shh pathway is involved in regional segmentation of the epithelial linings. This will be confirmed by future studies with knockout mutants of MRCS1 and MACS1.

We found that the 217 bp ultra-conserved sequence of MFCS4 has the four motifs for the transcription factors involved in the Pbx, Sox, Wnt and Notch signaling pathways. As we revealed that many genes encoding these transcription factors are expressed in the epiglottis and arytenoid swelling of the E13.5 embryos, in which we observed the highest level of the MFCS4-mediated lacZ reporter expression, these transcription factors most probably act as direct upstream regulators of the Shh expression in the pharyngeal epithelium.

Here, we need to pay attention to influence of the adjacent genes Lmbr1 and Rnf32 on the phenotype of theΔMFCS4/ΔMFCS4 embryos, because the deletion of MFCS4 may disrupt the functions of these genes. The Lmbr1 null mutation has been reported previously (Clark et al.,2000). The Lmbr1 KO homozygotes show limb defects, but they are viable and fertile. It suggests that disruption of the Lmbr1gene cannot cause crucial pharyngeal defects, as shown in theΔMFCS4/ΔMFCS4 embryos. The Rnf32 mutant mouse has not been reported thus far, and the function of this gene remains unclear. Thus,influence of the MFCS4 disruption on the pharyngeal phenotype is currently undeniable.

Morphological variation in the epithelial architecture is well exemplified in the evolutionary distant species(Botella et al., 2007; Brainerd and Owerkowicz, 2006; Delgado et al., 2005; Ichim et al., 2007; Iwasaki, 2002; Mitsiadis et al., 2003). We showed that the oral epithelium-specific enhancer MRCS1 is conserved in birds and reptiles at the almost same level as in mammals, whereas its homolog has not been identified in amphibians and teleost fishes. The lung-gut epithelium-specific enhancer MACS1 is conserved in amphibians, but not in teleost fishes. It is likely that the various epithelial derivatives evolved from a flat epithelial structure (Chuong and Edelman, 1985; Chuong et al., 2000), and that their shapes and sizes can be altered,depending on the timing and location of Shh signaling. Thus, regulation of Shh expression probably influences the architecture of the epithelial derivatives. At present, it is not clear how the oral and lung-gut epithelial architecture develop in the species without MRCS1 and MACS1. However, one possible explanation is that, in the past, MFCS4 specified Shhexpression in both the ectoderm-derived oral cavity and the anterior endoderm-derived pharynx. Indeed, we found that mouse MFCS4 drives transient reporter expression in the dental placode(Fig. 2H). Alternatively, there may be unidentified, species-specific enhancers that regulate Shhexpression in a tissue-specific manner in amphibians and teleost fishes. It would be of interest to explore whether such CNCSs exist within different teleost fishes or between amphibian and teleost fishes.