Phylogenetic conservation of a limb-specific, cis-acting regulator of Sonic hedgehog (Shh)
Polarized expression of the Sonic hedgehog (Shh) gene in the posterior mesenchyme is essential for pattern formation in the appendages of higher vertebrates, from teleost fins to tetrapod limb buds. We report on a sequence in intron 5 of the Lmbr1 gene, which resides approximately 1 Mb from the Shh coding region in the mouse genome and is highly conserved among teleost fishes and throughout the tetrapod lineage. Positional cloning revealed that two mouse mutations, Hx and M100081, characterized by mirror-image digit duplication and ectopic anterior Shh expression, have base substitutions in this sequence. Absence of the conserved sequence in limbless reptiles and amphibians and a cis-trans test using the Hx and Shh KO alleles suggest that the sequence is a cis-acting regulator that controls the polarized expression of Shh.
Acquisition of tetrapod limbs is among the most striking events in vertebrate evolution. It is currently thought that paired fins of teleost fishes and tetrapod limbs evolved from a common ancestral appendage and that the emergence of polarized Sonic hedgehog (Shh) expression in the posterior margin of fin buds played a crucial role in freeing the fins from the body axis (Tanaka et al. 2002). This polarized Shh expression in the posterior mesenchyme is conserved evolutionarily throughout teleost fins and tetrapod limbs (Echelard et al. 1993; Endo et al. 1997; Riddle et al. 1993; Torok et al. 1999), but the mechanism of its regulation is poorly understood.
Shh encodes a signaling protein expressed in the zone of polarizing activity (ZPA), which is located in the posterior mesenchyme of the limb buds and is known to control the anteroposterior (A/P) patterning of limbs (Johnson and Tabin 1997; Riddle et al. 1993; Saunders and Gasseling 1968). Its function is exerted either directly, by dose-dependent activation of target gene expression (Drossopoulou et al. 2000; Yang et al. 1997) or indirectly, via the regulation of the relative balance of Gli3 transcriptional activation, and repression of the Shh downstream genes (Litingtung et al. 2002; te Welsher et al. 2002b; Wang et al. 2000).
Preaxial polydactyly with mirror-image digit duplication is a major heritable abnormality of limb development, both in humans and in the mouse. A locus for the major form of human preaxial polydactyly (PPD) has been mapped to Chromosome (Chr) 7q36 (Heutink et al. 1994; Tsukurov et al. 1994). In the mouse, two preaxial polydactyly mutations, hemimelic extratoes (Hx) and Sasquatch (Ssq), have been mapped to the syntenic region that is closely linked to Shh in Chr 5 (Martin et al. 1990; Sharpe et al. 1999). Recently, a similar mutation, M100081, generated by large-scale ENU mutagenesis in RIKEN GSC, was mapped to the same region (data not shown). The phenotypes of these mutants are confined to the distal limb field of the zeugopods and autopods (Knudsen and Kochhar 1981; Sharpe et al. 1999) (Fig. 1), and ectopic expression of Shh is observed in the anterior limb margin (Masuya et al. 1995; Sharpe et al. 1999). The phenotype is probably caused by the disruption of genetic pathways involved in the polarized expression of Shh.
Recent studies of another human autosomal recessive disorder, acheiropodia, which is characterized by bilateral congenital amputation of the upper and lower extremities, have suggested that mutation of the LMBR1 (C7orf2) gene is involved in disease pathogenesis (Ianakiev et al. 2001). The LMBR1 gene and its mouse homolog, Lmbr1, reside approximately 1 Mb away from the Shh coding region and encode a transmembrane receptor-like protein (Clark et al. 2000; Heus et al. 1999). In acheiropodia, a 4 to 6-kb genomic deletion has been found in the region that includes exon 4 of LMBR1. More recently, the mouse Ssq mutation was reported to be caused by the insertion of a transgene into intron 5 of Lmbr1, and a human PPD was associated with a translocation involving intron 5 of LMBR1 (Lettice et al. 2002). The Ssq phenotype was abrogated when the Ssq allele was placed in a cis position relative to the Shh knockout allele. More recently, Lettice et al. reported that there is an evolutionary conserved sequence in intron 5 of the Lmbr1 gene, and that mouse Hx and four unrelated families of human PPD have base substitutions in the conserved sequence. In addition, when the conserved sequence was incorporated into transgenic construct containing the heterologous beta-globin promoter and lacZ reporter gene, the reporter gene was expressed in the posterior margin of developing limb buds (Lettice et al. 2003). These findings suggest that genetic linkage of Lmbr1 is merely incidental to the Hx phenotype and that the Ssq and PPD mutations instead interrupt a cis-acting regulator of Shh.
In the past several years, we have also carried out positional cloning of the Hx causative gene. In this study, a large-scale linkage analysis indicated that Hx is confined to a short DNA segment including intron 5 of the Lmbr1 gene, and two preaxial polydactyly mutations, Hx and M100081, are caused by base substitutions in the highly conserved sequence in intron 5 of Lmbr1. Phylogenetic study revealed that this sequence is conserved throughout the tetrapod lineage and among teleost fishes, but not in limbless species of reptiles and amphibians. A cis-trans test using the Hx mutant allele and the Shh knockout allele demonstrated that the Hx phenotype is completely abrogated when Hx is placed in a cis position relative to the Shh knockout allele. All the data suggest that the conserved sequence found in intron 5 of Lmbr1 is a cis-acting regulator of Shh.
Materials and methods
A mutant mouse, Hemimelic-extratoes (Hx), and C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). The Hx heterozygotes were maintained at the National Institute of Genetics (NIG) (Mishima, Japan). An Hx congenic line, JF1-Hx, was recently established at NIG by breeding 12 generations of backcrosses of the Hx heterozygotes to an inbred strain, JF1/Ms (Koide et al. 1998). The JF1-Hx mice exhibit a largely restored phenotype in the zeugopods. MSM and MAL are inbred strains derived from the wild mouse Mus musculus molossinus and Mus musculus castaneus, respectively, and were established after 20 generations of brother-sister mating at NIG. BLG is an inbred strain derived from the wild mouse Mus musculus musculus and was established by F. Bonhomme (University Montpellier). The Shh knockout mouse was a gift from Dr. P. Beachy (HHMI, Johns Hopkins).
Mouse skeletons were stained by alizarin red and alcian blue as described elsewhere (Wallin et al. 1994).
Whole-mount in situ hybridization
In situ hybridization of whole-mount embryos was performed according to the method of Wilkinson (Wilkinson 1992). Briefly, digoxygenin-labeled riboprobes were transcribed in vitro according to the manufacturer’s protocol (Roche). The following sequences were used as probes: a 642-bp EcoRI fragment for Shh (Echelard et al. 1993), the entire coding region for murine dHand (Srivastava et al. 1995), a 648-bp fragment, including the paired-tail domain for Alx4, an 800-bp partial cDNA fragment for Gli3 (Hui and Joyner 1993), a 620-bp BamHI fragment for Fgf4 (Niswander and Martin 1992), and an 841-bp EcoRI fragment for Ptc (M. Scott).
Genetic and physical mapping
Linkage analysis for the Hx mutation was first carried out using the cross with the MSM strain. MIT microsatellite markers (Research Genetics) and RFLPs detected with an Shh cDNA probe (Echelard et al. 1993) were used for genotyping of the progeny. In parallel, the second linkage analysis was carried out with crosses with BLG and MAL strains derived from wild mice. For physical mapping, YAC libraries from the ICRF (Oxford, UK) and Research Genetics, and a BAC library from Research Genetics were screened with MIT and STS markers. STS markers were prepared from the BAC ends and the shotgun-sequencing fragments of BAC141C6. These DNA markers were used for identification of the recombination breakpoints in the linkage analyses. The Hx critical region was narrowed down in the 82-kb region by two primer pairs. For the proximal limit: PF, 5′-TGAGGCTCTTGGCTCATAAG-3′, and PR, 5′-CTGAATGCAGCATCTGTATG-3′, and for the distal limit: DF, 5′-CATGTGCTTGATACCGACTG-3′, and DR, 5′-ATGCTGAGGACCAAATGCAG-3′. A cosmid library was prepared from DNA of BAC311J12 by using a packaging kit, Gigapack III gold (Stratagene), and the Hx critical region was covered with several cosmid clones.
Total RNA was extracted from whole C57BL/6J mouse embryos at E10.5–11.5 and from limb buds of ICR mice at E11.5 with ISOGEN (Nippon Gene Ltd. Toyama, Japan). The mRNA was purified with a Quickprep Micro mRNA purification kit (Amersham Pharmacia Biotech Ltd.). First- and second-strand DNA was synthesized by a ZAP-cDNA Synthesis Kit (Stratagene), and after blunting the cDNA ends, an ERI adapter (Takara Ltd., Japan) was ligated to the PCR products. In parallel, BAC311 J12 DNA was digested partially with SauIIIA and biotinized. For prehybridization, we used mouse COT1 DNA (Gibco, BRL). Hybridization and subsequent steps were performed as described previously (Sasaki 1995).
Mutation analysis of the coding region of the Lmbr1 gene
For mutation analysis of the coding sequence of Lmbr1, RT-PCR was carried out with the Access RT-PCR System designed to reverse-transcribe (RT) and PC amplify a specific target RNA (Promega Corp.). The following primer pairs were used for amplification of exons 1 and 2 of Lmbr1: 1F, 5′-TCTAGGTGGCGTTTCTGTCC-3′ and 1R, 5′-TGACAGAGGTGAGAAGATGG-3′; and 2F, 5′-CA AAGATGTATCTCTTGTTGG-3′ and 2R, TCAGCCATCGGCATATTCAG-3′. The following primer pairs were used for the amplification of exon 3 to the 3′ terminus of the coding region: 3F, 5′-CTTGCCATTCTCTACATCG-3′ and 3R, 5′-AGTCTTCTCT-GGAGTGCTTC-3′; and 4F, 5′-CAGCGATTCTTGAAGACCTG-3′ and 4R, 5′-CTCACAGTGCTTTCTGATGC-3′.
Hx and Shh knockout heterozygotes were crossed. The double heterozygotes harboring the two mutant alleles, which had the Hx phenotype, were crossed to wild-type C57BL/6J mice. The obtained progeny were screened for recombinants between the Shh coding and the Hx critical region. To genotype a proximal marker closely linked to Shh, we used a length polymorphism of a DNA fragment amplified with the PCR primers 5′-AGTTCACCAAGGCAGGACAC-3′ and 5′-GCCCTAATAGCACCATCAAC-3′. As a distal marker linked to the Hx critical region, we used a length polymorphism of a DNA fragment amplified with the PCR primers 5′-TTGTGGCTGAAAGGTCATGG-3′ and 5′-AGTGTGGATTGCCATCTTGG-3′. The Shh knockout allele was typed for the presence of the PgkNeo cassette in the targeting vector, which was PCR-amplified with the primer pairs 5′-GGCTATTCGGCTATGACTGG-3′ and 5′-GAGATGACAGGAGATCCTGC-3′. The presence of the Hx mutant allele was confirmed by direct sequencing of the conserved sequence.
The three cosmid DNAs covering the Hx critical region were sequenced by the shotgun method or conducting in vitro transposition by using the Genome Priming System (New England Biolabs). For the shotgun sequencing, DNA was partially digested with SauIIIA1 and ligated into pBluescript KS (Stratagene). For the transposition system, the transposed cosmids were introduced into DH1 cells by electroporation (Bio-Rads). Plasmids and cosmids were sequenced on a 377 sequencer (Applied Biosystems) with vector primers or transposon primers. The deduced sequences were compared by using the BLAST algorithm (NCBI) with the downloaded sequence of the human NH0332E22 clone, which has been localized to 7q36. For amplification of the mouse conserved sequence in intron 5 of Lmbr1, the following primer pairs were used: 5F, 5′-GTACTGGCCAGTGTTAAATG-3′ and 5R, 5′-CCAGATGTGCAAAGTTCAAG-3′.
Phylogenetic analysis of the conserved sequence
To amplify the conserved sequence in intron 5 of Lmbr1 from various tetrapod species, the following PCR primer pair was used for mammals and reptiles: 5′-GACCAATTATCCAAACCATC-3′ and 5′-TAACACTAAGCAGCACTTCC-3′. For amphibians and medaka fish, the following PCR primer pair was used: 5′-CTATCCTGTGTCACAGTTTG-3′ and 5′-TAACACTAAGCAGCACTTCC-3′. The ClustalW system (DDBJ, NIG, Japan) was used for sequence alignment and to draw the phylogenetic tree. In the Southern blot analysis, fragments amplified from the genomic DNA of lizard and newt were used as probes. Hybridization was performed at 65°C in a solution containing 10% dextran sulfate, 100 g/mL, salmon-sperm DNA, 1 mM EDTA, 6× SSC (1× SSC, 150 mM NaCl, and 15 mM sodium citrate), 5× Denhardt’s solution, and 1% sodium dodecyl sulfate (SDS). After hybridization, blotted filters were washed three times at room temperature with a wash solution of 2× SSC and 0.1% SDS, and with increased stringency at 63°C with 1× SSC to 0.1× SSC.
Limb skeletal phenotype of Hx and M100081 mutants
Hx, and M100081 are autosomal dominant mutations on Chr 5. Hx is a spontaneous mutation that arose in the B10.D2 strain, and M100081 is a mutant newly generated in C57BL/6J mice by ENU mutagenesis at the RIKEN Genomic Sciences Center. The phenotypes of both mutants were confined to the limbs and characterized by preaxial polydactyly with supernumerary metacarpals, metatarsals, and digits, and hemimelia caused by shortening of the radius, tibia, and talus (Fig. 1; Knudsen and Kochhar 1981). Although the severity of the phenotype was largely dependent upon the specific genetic background of the affected mice, the hindlimbs were usually more severely affected than the forelimbs. The homozygotes of both mutants were viable and showed a phenotype similar to that of the heterozygotes. In general, the zeugopodial defects were more severe in the homozygotes.
Expression patterns of marker genes in the limb buds of the Hx and M100081 mutant
Ectopic expression of Shh is frequently observed in preaxial polydactyly mouse mutants (Blanc et al. 2002; Masuya et al. 1995). In the Hx and M100081 mutants, ectopic Shh expression was observed in the anterior margin of the mesenchyme of the forelimbs at E11.0, which was delayed 1–1.5 days as compared with the normal expression of Shh in the posterior margin of the limb buds (data not shown). In the mutants, we analyzed the expression patterns of several marker genes that are known to act in limb development. The results are shown in Fig. 2. In the earlier stages of limb development, a basic helix-loop-helix transcription factor, dHand, is expressed in the posterior margin prior to Shh expression and is known to activate Shh transcription directly (Charite et al. 2000). It is also known that Gli3, a zinc-finger transcription factor, and Alx4, a paired-type homeodomain-containing transcription factor, are expressed in the anterior mesenchyme, where they repress the ectopic expression of Shh (Masuya et al. 1997; Qu et al. 1997). We examined the expression patterns of the dHand, Gli3, and Alx4 genes in the limb buds of the Hx embryos at E10.5, at which the ectopic expression of Shh dose not begin in the Hx mutant. The expression levels and patterns of these genes were normal in the forelimb and hindlimb buds of Hx mice. However, at E11.5, the genes downstream of Shh, such as Ptc, and Fgf4 (data not shown) reflected the expression pattern of Shh and were ectopically expressed in the anterior domain of Hx limb buds.
Positional cloning of the Hx mutation
Linkage analysis based on the 1,500 backcross progeny generated from the cross between the Hx mutant and MSM mice defined the location of the Hx critical region as a 0.6-cM interval between D5MIT387 and D5M1T44 on Chr 5. The coding region of Shh was excluded from the critical region by one recombinant obtained in the progeny. To narrow down further the critical region, we isolated YAC clones and generated a BAC contig that covers the approximately 1 Mb containing the Hx critical region. Linkage analysis based on the progeny generated from the cross of BLG- and MAL-strain mice narrowed the critical region to a 100-kb stretch of DNA contained within a single BAC clone, 311J12 (Fig. 3).
To identify candidate genes for Hx, we carried out direct cDNA selection using DNA from the 311J12 BAC clone and obtained 100 cDNA fragments from the cDNA libraries of E10-11.5 whole mouse embryos and the limb buds of E11.5 embryos. Among them, four cDNA fragments were assembled into Lmbr1, which was the mouse homolog of C7orf2, which had been identified as a candidate gene for limb deformities mapped in the human syntenic region, 7q36 (Heus et al. 1999). The protein product of C7orf2 contains nine transmembrane domains and a coiled-coil domain, and is likely to be a receptor protein. However, we failed to find in the Hx mutant any alteration in the Lmbr1 coding sequence, which was previously reported by another group (Clark et al. 2000). Moreover, Northern-blot and quantitative-PCR analysis indicated that Lmbr1 expression levels of the limb buds of the Hx mutant were similar with wild-type embryos (data not shown).
Base substitutions in intron 5 of the Lmbr1 gene in the Hx and M100081 mutant genomes
Because no other candidate gene was identified in the Hx critical region, we carried out genomic analysis of this region. Sequencing of about 100-kb genomic DNA fragment revealed that Lmbr1 encompassed most of the critical region, and that the proximal end of the Hx critical region was located in intron 5 of Lmbr1 (Fig. 3). Comparison of the genome sequence with the human genome sequence from the NH0332E22 clone identified a 1.2-kb highly conserved sequence in intron 5 of Lmbr1 (Figs. 3, 4). The human counterpart was located in intron 5 of LMBR1 as well. Sequencing of this fragment in the Hx and M100081 mutants identified single base substitutions in the conserved sequence in the parental strains, B10.D2 and C57BL/6J, from which Hx and M100081, respectively, arose (Fig. 3). We searched the TRANSFAC public database for known cis-elements in the conserved sequence, but failed to find any known motifs.
To test whether the conserved sequence acts as a cis regulator, we carried out a cis-trans test using the Shh knockout (KO) allele (Fig. 5). We screened the progeny generated from a cross of double heterozygotes of the Hx and Shh mutations in the C57BL/6J strain and obtained one informative recombinant that harbored both the Hx allele and the Shh KO allele on the same chromosome. Despite the dominance of the Hx phenotype (100% penetrance), the heterozygote containing the recombinant chromosome did not show the polydactylous phenotype. We confirmed this result in the progeny of the next generation (Table 1).
Phenotype of Hx mutation flanking Shh− allele
Shhr−−+/+−Hx × +−+/+−+
Shh−−Hxa/+−+ × +−+/+−+
Shh−−Hx/+−+ × Shh− −Hx/+−+
The phylogenetic conservation of the cis-acting regulator of Shh
If the conserved sequence in intron 5 of Lmbr1 is indeed a limb-specific, cis-acting regulator, it is conceivable that the sequence is conserved among all vertebrate tetrapods. To test this, we designed several PCR primer pairs based on the sequences conserved between human and mouse and attempted to amplify DNA fragments from different tetrapod species. Some of the primer pairs amplified a single band not only from mammalian genomic DNA, but also from that of reptiles and amphibians. A 500-bp fragment isolated from mammalian and reptile species and a 300-bp sequence in amphibians were highly conserved throughout all tetrapods tested. The average matching rate exceeded 80% among species. The most conserved blocks of the sequence are shown in Figs. 6A, B.
If the cis-acting regulator of Shh is involved in the transition from fins to limbs, the early form of the conserved sequence should be present in teleost fishes. We tested this possibility by amplifying a short conserved fragment from medaka fish (Oryzias latipes). As shown in Fig. 6C, medaka fish indeed contains the putative regulator sequence. We found the conserved sequence in the genome of another teleost, fugu (Takifugu), by searching public databases (Fig. 6C). In contrast, we failed to amplify a similar fragment from a cartilaginous fish, cephaloscyllium umbratile.
It is notable that the base changes identified in Hx and M100081 were localized to two highly conserved blocks and that the nucleotides at these positions are completely conserved in all tetrapod species examined. Conversely, the flanking sequences exhibited phylogenetic diversification, which allowed us to generate a phylogenetic tree (Fig. 7). The tree is consistent with a previous one constructed based on the sequences of an inserted retroposon (Shimamura et al. 1997).
Abrogation of the conserved sequence in limbless species
If a function of the conserved sequence in intron 5 of Lmbr1 is crucial for the polarized expression of Shh in the posterior mesenchyme of limb buds, this sequence may be lost in limbless species. We therefore examined whether several limbless species retain the conserved sequence. PCR amplification of DNA samples from four snake species and a limbless newt (Typhlonectes) (Fig. 8A) was carried out with the primer pairs applied for amplification of the lizard and amphibian sequences. No equivalent fragments were amplified from any of the limbless species. We confirmed the result by Southern blot analysis, using probes generated by PCR amplification from lizard and newt genomic DNA. As shown in Fig. 8B, each unique signal band was obtained in the tetrapod species, but no signal was detected in the limbless species.
A conserved sequence in intron 5 of Lmbr1/LMBR1 is a limb-specific, cis-acting regulator of Shh
The presence of a cis-acting regulator in intron 5 of Lmbr1/LMBR1 was proposed recently in the analysis of the mouse Ssq mutation and a human PPD (Lettice et al. 2002). Genetic analysis using a Shh knockout allele showed that the Ssq phenotype was abrogated when the Ssq allele was located together with the Shh knockout allele on the same chromosome.
This study demonstrated that the spontaneous mutation Hx and the ENU-induced mutation M10008 have single base substitutions in the conserved sequence located in intron 5 of Lmbr1. The mutation rate per base pair in ENU mutagenesis has been estimated empirically to be 10−5 (Coghill et al. 2002). The rate of spontaneous mutation has been estimated to be at least two orders of magnitude lower than that of ENU mutagenesis. Consequently, the probability that the two mutants coincidentally have base substitutions within the short 1.2-kb conserved sequence is extremely low. Hence, the phenotype of the two independent mutations is most likely to be caused directly by the base substitutions in the conserved sequence.
In conjunction with the data of Lettice et al. (2003), the result of fine linkage analysis of the two mouse mutations evidenced that the conserved sequence in the Lmbr1 intron 5 is directly involved in regulation of Shh expression in limb buds. In this study, a cis-trans test clearly showed that the phenotype of Hx mutation appeared only when the Hx allele was present in a cis position relative to the wild-type Shh allele. Taking into account all these data, it is most likely that the conserved sequence acts as a limb-specific, cis-acting regulator and is involved in polarizing Shh expression in posterior limb mesenchyme.
Several recent reports demonstrated that an early event in A-P axis formation in limb development is controlled by an Shh-independent pathway involving the transcription factors dHand and Gli3 (te Welscher et al. 2002a). Whole-mount in situ hybridization in this study showed that in Hx mutant embryos at E10.5, a point prior to the observation of ectopic expression of Shh, dHand, and Gli3 is expressed normally in the posterior and anterior margins, respectively. The Alx4 expression pattern in the limb buds of Hx is also similar to that seen in wild-type embryos. On the other hand, ectopic expression of the downstream genes of Shh signaling, such as Ptc and Fgf4, was induced in the anterior domain, where ectopic Shh was observed. These results indicate that the Hx phenotype is induced directly by the ectopic expression of Shh, suggesting that there exists a genetic pathway that downregulates Shh in the anterior margin of limb buds independent of Gli3 and Alx4, or at least downstream of these genes. It is most likely that the cis-acting regulator found in intron 5 of Lmbr1 is a target element for transcription factors involved in this pathway. In this study, we could not identify any known cis-motif in the conserved sequence. Identification of the cis-motif and the transcription factor that binds to it will be a topic of future study.
Possible role of cis-acting regulators of Shh in the evolution of vertebrate limbs
Recent analyses of the fossil record and experimental studies have inferred the origin of tetrapod limbs (Clack 2002; Tabin 1992; Zhu and Yu 2002). One such experimental study showed that, in the dogfish, a primitive cartilaginous fish, dHand is detectable during fin development, but Shh is not (Tanaka et al. 2002). On the other hand, in a teleost, the zebrafish, Shh is expressed in the posterior margin of fin buds in a manner similar to its expression in tetrapod limb buds (Krauss et al. 1993). On the basis of these findings and the lateral-fin fold theory (Thacher 1877), it has been hypothesized that Shh played a crucial role in the evolution of tetrapod limbs, by freeing the fins from the body axis and establishing a separate limb axis (Tanaka et al. 2002). In this study, we found that the potential cis-acting regulatory sequence is extremely highly conserved throughout the tetrapod lineage, from mammals to amphibians. In addition, even in the teleosts, we found a partial equivalent sequence, but such a sequence could not be found in a cartilaginous fish. All these data suggest that the ancient conserved sequence was coupled to the polarization of Shh expression and thus the transition from fins to limbs. Further precise analysis of more primitive species, including the extant Sarcopterigians, coelacanths, and lungfish, may corroborate this proposed role of the conserved sequence in the evolution from fins to limbs.
In some vertebrate species, limb loss is often accompanied with trunk elongation and loss of regional differentiation in the axial skeleton (Carroll 1988). The mechanisms for limb loss in whales or snakes are controversial (Bejder and Hall 2002). In pythons, a group that evolved from the tetrapod lizards, forelimbs are absent, but hindlimb buds are formed during embryogenesis. However, in the python’s hindlimb buds, Shh expression is not detected, and no polarization is observed (Cohn and Tickle 1999). The current study revealed a loss of the conserved sequence in several limbless species of pythons and newt. Although it is less likely that the loss of the sequence is a direct cause of limb loss in these species, it is possible that limb loss in these organisms has obviated the need of this limb-specific, cis-acting regulator, leading eventually to the loss of the conserved sequence.
We are grateful to J. C. I. Belmonte and J. F. Fallon for reading the manuscript and giving useful comments; to H. Yamamoto, K. Tsuchiya, M. Sakaizumi, N. Miyashita, H. Suzuki, A. Watanabe, H. Tamate, M. Taira, Y. Kikkawa for their gifts of valuable materials for phylogenetic analysis; to S. Brown for the gift of YAC clone I116C10; and to P. Beachy for providing Shh knockout mice, A. McMahon for the Shh probe, G. Martin for the Fgf4 probe, C. C. Hui for the Gli3 probe, and M. Scott for the Ptc probe. We thank S. Makino, S. Tanaka, A. Oka, A. Mita, K. Ishihara, S. Sakamoto, N. Saito, and K. Sorimachi for their valuable advice. This study was supported in part by grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan. This paper is contribution No. 2489 from the National Institute of Genetics, Japan.