Disruption of a long-range cis-acting regulator for Shh causes preaxial polydactyly

The translocation patient was clinically examined, and a member of her family was interviewed for family history at the Niigata University Hospital. All studies were approved by the local ethics committee. A member of the family gave written informed consent on behalf of the patient. The PPD families used in this study are unrelated, as described (1, 4).

The end sequences of PAC 123K10, BAC 111G19, and cosmid 171G2 were generated to enable alignment with the existing genomic sequence. FISH analysis using YAC, BAC, PAC, and cosmid clones as probes was performed as described (10). Epstein–Barr-virus-mediated lymphoblastoid cell lines established from peripheral blood of the patients were treated with colcemid and harvested according to routine methods. Probes were labeled with digoxigenin-11-dUTP by using a nick translation kit (Roche).

STS probes used for Southern blot analysis were PCR-amplified from human genomic DNA of normal control, using primers specific to the genomic sequence of a BAC clone Rp11-332E22. Probes were labeled with [α-³²P]dCTP in a random priming reaction using the RediPrime II DNA labeling system (Amersham Pharmacia Biotech). Southern blot analyses were performed as described (11).

Primers were designed complementary to intronic DNA around all exons of LMBR1 to cover at least 60 bp on both sides of the exon. PCR products were generated from either genomic DNA cleaned with the Amicon Microcon PCR purification kit (Millipore) or treated with a PCR product presequencing kit (Amersham Pharmacia). Larger fragment sequencing used a transposon-based system (GPS–1 Genome Priming system, New England Biolabs) and plasmid DNA was made using an ABI Prism Miniprep kit (Perkin–Elmer/Applied Biosystems). DNA was sequenced on an ABI-377 automated fluorescence dye sequencer, using Bigdye or dRhodamine chemistry (Perkin–Elmer/Applied Biosystems).

Sequences conserved between mouse and human were also screened for mutations in familial PPD patients. These regions were: 3.7 kb adjacent to exon 5 (downstream); 100 bp at 1 kb upstream of exon 6; 230 bp adjacent to exon 6 (upstream); intron 7; intron 14; 2 kb adjacent to exon 16 (downstream). Database screening and sequence analysis was done using programs from the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov) and Human Genome Mapping Project (www.hgmp.mrc.ac.uk).

Ssq/Ssq genomic libraries were made in Lambda FixII and in SuperCos (Stratagene) following the manufacturer's instructions. λ clones (1.6 × 10⁶) were screened using PCR product generated from the transgene (primers GCTTCAGCTCTGTGACATACT and CAGTTTGTCCTTCTTCTGCCC). Subsequent screening identified one λ as containing only one end of the transgene. DNA was made from this (Qiagen Lambda kit) and end sequence obtained. The genomic end fragment was used as a probe on the RPCI21 library (HGMP Resource Centre, Cambridge, U.K.) and PACs identified (RCPI21-542n10 and RCPI21-576b18). FISH and exon trapping (12) were conducted by standard techniques. Complete intron 5 sequence was determine by assembling HTGS draft sequence (GenBank accession no. AC058788) in combination with sequence generated from the integration sites and directed sequence to fill gaps.

RNA was extracted from E11.5 mouse embryos by using RNAzol (Biogenesis) and cDNA was made using a First-strand cDNA Synthesis Kit (Amersham Pharmacia Biotech). PCR primers used are specific to exons of mouse Lmbr1, except that the exon 6 primer crossreacts with the glycosylphosphatidylinositol-anchored protein homologue and serves as an internal quantification control. PCR was also performed with primers to the mouse HPRT gene as a positive control. Samples were taken after 25, 30, and 35 cycles of PCR, and quantified either by Southern blotting of the gel and probing with an internal oligonucleotide or by addition of [α-³²P]dCTP to the reaction and subsequent scanning on a Storm phosphoimager (imagequant package).

A Shh null heterozygous mouse was crossed to a Ssq homozygote mouse to generate Shh null heterozygous mice carrying the Ssq insertion, the F₁ generation. The F₁s were crossed to CBA wild-type mice to generate the G₂ generation. DNA from all mice was analyzed by PCR for the Ssq insertion and for both the Shh wild-type and null alleles. Primers for the Ssq insertion were CTCTGTTTCCTTTTCCTCTATC and GTATGGGATTAATTAAATCTTGTGTC, which generate a 180-bp product. The Shh PCR primers were described by Chiang et al. (13) and generated a 550-bp product for the null allele and 230-bp product for the wild-type allele.

A 3-year-old girl presented with bilateral duplication of the triphalangeal thumb and triplication of the great toe without any associated abnormality (Fig. 1A). She was found to carry a de novo reciprocal translocation t(5,7)(q11,q36). Fine mapping of the translocation breakpoint by use of FISH analysis in the lymphoblastoid cell line derived from the patient (Fig. 1C) identified a BAC clone Rp11-332E22 (GenBank accession no. AC007097) that spanned the breakpoint (Fig. 1B). Using fragments from the BAC, a cosmid clone 171g2 was identified that spanned the breakpoint (Fig. 1B). Sequencing the ends of the cosmid insert identified the genomic fragment as lying inside one of the genes (originally given the designation C7orf2 (5) within the genetically defined critical region for PPD. This gene is the orthologue of the mouse Lmbr1 gene (14). Southern blot analysis of DNA from a lymphoblastoid cell line further refined the site of the breakpoint (Fig. 1D). A number of repeat-free STSs including 13–293 and 13–479 were PCR-amplified and used as probes. This analysis showed that the translocation breakpoint is located in a 714-bp BamHI/XbaI fragment residing between exons 5 and 6 of LMBR1 without an accompanying large deletion (Fig. 1E). This DNA fragment consists of 95% LINES and 5% simple repeated sequence, and seems to neither encode an ORF nor contain known regulatory sequences.

Subsequently, we performed mutation analysis on the LMBR1 gene in patients. Analysis of the exons and the intron/exon boundaries of the LMBR1 gene in five unrelated PPD families (1, 4) uncovered no pathogenic mutations (data not shown). Therefore, in the PPD families mutations interrupting either the coding region or the structure of the transcript are not apparent.

To isolate the transgenic insertion site from the Ssq mutant mouse, a λ genomic library was made from Ssq/Ssq mice. Clones were isolated using the transgene as a probe, one of which contained a junction of the transgene insertion site (Fig. 2B). The genomic end fragment of this λ clone (Fig. 2B) was used to identify mouse PAC clones which, by chromosomal FISH analysis, were shown to co-localize with the transgene (Fig. 2A). Subsequent exon trapping (12) of the PAC RPCI21-542N10 identified a surrounding gene. The complete sequence of the corresponding mouse cDNA was determined and is identical to mouse Lmbr1 (14).

Subsequently, a cosmid library was made from Ssq genomic DNA and additional clones were isolated to more fully elucidate the integration site. The transgene/genomic junction from the other end of transgenic integration was identified (Fig. 3B). Comparison of sequence from this cosmid clone with the complete intron 5 sequence we additionally assembled defined the position of this integration. Multiple copies of the transgene have integrated, lying wholly within intron 5 of the Lmbr1 gene (genomic sites of insertion are marked by asterisks in Fig. 2B). The process of integration has resulted in a duplication of approximately 20 kb of the intron with the transgene residing between the duplication endpoints (lower line in Fig. 2B). The duplication was confirmed by Southern blot analysis using unique sequence probes (probes numbered 1–4 in Fig. 2B). The surrounding DNA including exons 5 and 6 was apparently unaffected by the integration.

The consequence of the Ssq transgenic insertion on Lmbr1 transcription was analyzed. The transgene containing Lmbr1 gene encodes full-length transcripts, further supporting the proposal that arrangement of exons is unaffected by the insertional mutation. However, Lmbr1 transcription in Ssq embryos does show a high level of premature termination. The 5′ end of the mutant Lmbr1 transcript (Fig. 2C Top) is produced at wild-type levels, but transcription of exons 3′ of the insertion site is detectably lower, representing about 10–30% of wild-type levels (Fig. 2C Middle). The transgene insertion apparently leads to termination of transcription between exons 5 and 6.

Furthermore, we investigated for possible premature transcriptional termination in the familial PPD patients. We amplified LMBR1 transcripts by reverse transcription (RT)-PCR from RNA from lymphoblastoid cell lines of Dutch and Cuban patients, and sequenced the products. Polymorphisms in a Cuban (in exons 13 and 17) and two related Dutch (one in exon 5 and the other in exon 14) patients indicated that both LMBR1 alleles, at the 5′ end and the 3′ end, were capable of being expressed. Furthermore, an overlapping series of three PCR products covering the whole LMBR1 transcript showed no detectable difference in band size between patients and controls (data not shown), indicating that no exons are skipped in patients. These data argue that mutations disrupting LMBR1 transcriptional capacity are not commonly associated with the PPD phenotype.

We previously showed (9) an intriguing relationship between the Ssq mutation and Shh expression. The reporter gene contained within the Ssq transgene insertion has acquired a limb-specific, Shh-like expression pattern; not only in the appropriate posterior ZPA (15), but also at an ectopic site at the anterior margin of the limb bud. It is unclear from the present data whether the mechanism for Shh misexpression involves disruption of the Lmbr1 gene and is therefore indirect or, as suggested (9), direct because of disruption of a long range Shh regulatory element. Long range regulation of Shh has been previously postulated for a translocation identified in a holoprosencephaly patient in which the breakpoint resides 250 kb upstream (16).

To examine the basis for preaxial polydactyly we devised a cis–trans genetic test. We predicted that an Shh regulator would function in a cis-acting manner—i.e., that the Ssq mutation would affect only chromosomally linked Shh. In contrast, disruption of a gene that secondarily affected Shh expression would operate on both Shh alleles and, therefore, in trans. A mouse cross was devised to derive a recombinant chromosome 5 in which the Ssq mutation was located in cis to an easily distinguishable Shh allele (Fig. 3A). Toward this end, we crossed Ssq/Ssq male mice with females carrying the mutant Shh null allele (Shh^null) (13) to produce the F₁ generation. The Shh^null allele provided a straightforward assay for both the genotype (Fig. 3B) and the phenotype of each offspring. In the analysis of the phenotype, mice carrying the recombinant chromosome would exhibit extra preaxial toes if Ssq is acting on Shh in trans and wild-type feet—i.e., suppression of the Ssq phenotype—if acting in cis.

The F₁ generation (n = 14) carrying Ssq and Shh^null on opposing chromosomes showed complete penetrance of the Ssq mutant phenotype. (Shh^null shows no phenotype in the heterozygous state.) The second generation (the G₂) was produced by mating the F₁ mice to wild type with the aim of generating the recombinant chromosome 5. Each G₂ mouse was assessed by PCR for the presence of the Ssq transgene insertion (Fig. 3B Upper) and the neo^r gene associated with Shh^null (Fig. 3B Lower). In this cross we produced 446 G₂ offspring of which eight were recombinants, representing a recombination frequency of 1.8 ± 0.6%. Three of the recombinants were uninformative because they contained the wild-type alleles (Fig. 3A). The other five recombinants carried both the Ssq insertion and Shh^null allele situated in cis. Analysis of the limb phenotype in these five recombinants showed no preaxial polydactyly or other detectable limb phenotypes. In addition, we bred two males carrying the recombinant chromosome 5 (Ssq and Shh^null) with wild-type females and showed that the next generation had no limb abnormalities (21 mice carrying the recombinant chromosome of 37 offspring). On some genetic backgrounds the Ssq heterozygous mutation is not fully penetrant. In this cross 7% of the Ssq/+ mice generated showed no detectable limb phenotype. Two of these Ssq/+ males, which displayed no limb phenotype, were bred further and were shown to transmit the phenotype (7 with additional preaxial digits of 19 offspring). Thus the data demonstrate that the Shh null allele inactivates the effects of the Ssq mutation when located in cis (but not in trans) on chromosome 5. It follows that the Ssq insertion is a dominant acting mutation that interferes with the limb-specific expression of Shh. The data are consistent with a long range limb-specific regulator of Shh residing within or near the Lmbr1 gene.