Greig cephalopolysyndactyly syndrome (GCPS, MIM 175700) is a rare autosomal dominant developmental disorder characterized by craniofacial abnormalities and post-axial and pre-axial polydactyly as well as syndactyly of hands and feet. Human GLI3, located on chromosome 7p13, is a candidate gene for the syndrome because it is interrupted by translocation breakpoints associated with GCPS. Since hemizygosity of 7p13 resulting in complete loss of one copy of GLI3 causes GCPS as well, haploinsufficiency of this gene was implicated as a mechanism to cause this developmental malformation. To determine if point mutations within GLI3 could be responsible for GCPS we describe the genomic sequences at the boundaries of the 15 exons and primer pair sequences for mutation analysis with polymerase chain reaction-based assays of the entire GLI3 coding sequences. In two GCPS cases, both of which did not exhibit obvious cytogenetic rearrangements, point mutations were identified in different domains of the protein, showing for the first time that Greig syndrome can be caused by GLI3point mutations. In one case a nonsense mutation in exon X generates a stop codon truncating the protein in the C-H link of the first zinc finger. In the second case a missense mutation in exon XIV causes a Pro→Ser replacement at a position that is conserved among GLI genes from several species altering a potential phosphorylation site.
Greig cephalopolysyndactyly syndrome (GCPS, MIM 175700) is a rare dominantly inherited developmental disorder. Prominent features of GCPS include post-axial and pre-axial polydactyly as well as syndactyly of hands and feet and mild craniofacial abnormalities, such as a slight hypertelorism and a high prominent forehead (1). The human GLI3 gene, located on chromosome 7p13, is considered as a candidate gene for the syndrome, because it is interrupted by translocation breakpoints associated with GCPS (2,3). Since 7p13 hemizygosity results in GCPS it was postulated that haploinsufficiency of GLI3 evokes the developmental malformations characteristic of this syndrome (4). This notion was corroborated by the finding that the phenotypically similar mouse mutant extra toes (Xt) is caused by deletions in Gli3 (5–7). GLI3 is a member of a zinc finger gene family related to Krüppel, a gene that is known to regulate development in Drosophila. Previously, we had shown that the 1596 amino acid zinc finger protein corresponding to the published cDNA (8) is encoded within a genomic segment of nearly 240 kb (9). The gene is flanked by a CpG island on the 5′-side and is in close proximity to the first exon detected by the cloned GLI3 cDNA (8,9). GLI3 and another family member, GLI2, had been isolated by homology to the zinc finger gene GLI, which is amplified in gliomas (8,10). Multiple regions of sequence similarity within the protein, aside from the zinc finger region, suggest that other aspects of function are shared among the members of this gene family of DNA binding proteins. The gene is expressed in a wide variety of human tissues, as are other family members (10). Here we report for the first time that Greig syndrome can be caused, as well as by the known translocation events, by GLI3 point mutations. Based on mutation analysis in part of the coding sequence, GLI3 was recently implicated as a candidate for Pallister-Hall syndrome (PHS), which exhibits some phenotypic overlap with GCPS (11). Since another GLI3 mutation is associated with post-axial polydactyly type A (PAP-A) (12; U.Radhakrishna, A.Wild, K.-H.Grzeschik and S.Antonarakis, in preparation) it emerges that different mutations of this gene evoke clinically distinct entities.
Genomic organization of the GLI3 gene
The GLI3 gene is flanked by a 2 kb CpG island that spans exon I and extends 1.7 kb into intron sequences (the sequence of the CpG island will be submitted to the EMBL database). A microsatellite [(GGA)7 GAA (GGA)6 (GGC)9] was identified at the beginning of the CpG island. The complete intron-exon organization of the human GLI3 gene was determined by sequencing appropriate fragments of cosmid clones derived from the GLI3 contig (9). Comparison with the published cDNA sequence (accession no. M34366; 8) showed no differences between the sequences determined from the cosmids and the cDNA sequence. The sequences of the exon-intron boundaries and exon sizes are given in sizes are given in Table 1. All splice junctions exhibit a high similarity to consensus sequences of donor and acceptor sites. The gene consists of 15 exons and 14 introns spanning 240 kb of genomic DNA (Fig. 1A). Although the transcription start has not been determined, it is located at least 52 nt further upstream than the 5′-end of the published cDNA, as longer cDNAs were isolated (data not shown). Exon I and part of exon II are 5′-untranslated regions. The translation start codon is positioned at nt 43–45 of exon II. The zinc finger domain (ZFD) of GLI3 is contained within four exons (X–XIII). While exon X carries only part of finger I, exon XI and XII encode more than a single finger structure. The translation stop codon is located within exon XV, which spans >2.5 kb of continuously transcribed DNA. The contribution of the exons to the different regions within the protein is shown in Figure 1B.
From the genomic sequence or from the cDNA sequence we designed 39 primer pairs (Table 1) and amplified all exon segments translated into protein from DNA of Greig syndrome patients which did not show a deletion or another cytogenetically obvious rearrangement of 7p13. We developed a polymerase chain reaction (PCR)-based mutation assay for the human GLI3 gene. In the GCPS patients analyzed two SSCA variants were detected (Fig. 2A and B) which were not found in a sample of 100 controls (data not shown). SSCA variants in exons II, V1, V2 and IX also appeared in controls and were considered to represent polymorphisms (data not shown). Sequencing of the DNA of heterozygotes MR2110 and MR2111, a mother and her daughter, both affected with GCPS, indicated a nucleotide exchange at position nt 1485 in exon X. Sequencing of the deviant SSCA bands of both patients demonstrated a transition (C→T) resulting in a nonsense mutation (TAG, amber; Fig. 2C and E). The protein product is truncated at amino acid 496 (Fig. 2G). By sequencing the variant fragment of patient MR2122 a missense mutation was detected in exon XIV. At nucleotide 2119 a C→T transition results in the alteration of a Pro to Ser residue at amino acid 707 (Fig. 2D-F).
The mutations observed previously in Greig syndrome either completely eliminate one copy of GLI3 by deletion or they truncate the gene upstream of or within the ZFD (Fig. 1) or the gene is inactivated by a position effect following a translocation involving 7p13, 10 kb downstream of GLI3 (3,9). One of the mutations we describe now is predicted to result in a GLI3 protein truncated at the C-H link of the first zinc finger of the ZFD (Fig. 1), probably eliminating the DNA binding potential, as did two of the translocations.
The missense mutation in exon XIV, the first mutation not truncating the protein, is also associated with a typical GCPS phenotype. It changes a Pro to a Ser at amino acid 707, within a conserved domain 3′ of the ZFD. Ruppert and co-workers identified this domain (designated ‘region 3’) in a screen for similarity between members of the human GLI family (8). The similarity region extends beyond the human species to mouse, Xenopus and chicken Gli family members (U.Radhakrishna, A.Wild, K.-H.Grzeschik and S.Antonarakis, in preparation).
The amino acid exchange Pro→Ser affects a putative consensus phosphorylation site for serine/threonine kinases (13). This may unveil an intriguing explanation for the functional consequences of this mutation: lack of phosphorylation at this site might impair the function of GLI3 and result in the same phenotype as haploinsufficiency. This explanation is compatible with the role GLI3 is expected to play during limb development: in Drosophila the GLI3 homolog cubitus interruptus (Ci) is a critical component of the hedgehog (Hh) pathway of developmental control involved in midline signaling, which has a pivotal role in regulation of many aspects of larval and adult pattern formation, e.g. anterior-posterior orientation in limbs and wings (14–17). Hh signaling results in up-regulation of synthesis of the morphogen decapentaplegic (Dpp), a member of the TGF-β family of transcriptional regulators, most likely via Ci. Hh signal transduction allows stabilization (or modification) of the Ci protein by antagonizing the post-transcriptional regulation imposed by protein kinase a (PKa), a cAMP-dependent serine/threonine kinase (14), or by patched (Ptc), a transmembrane protein (17). Another serine/threonine kinase, fused (fu), directly regulates Ci (18). If this pathway also applies to humans extinction of a phosphorylation site might critically affect GLI3 function.
The Hh signaling pathway is conserved from flies to mammals (18); however, the mechanism by which Hh signals are transduced in vertebrates is less well understood. GLI3, as its homolog Ci, appears to function at a central position in this pathway, since mutations in human homologs of the hedgehog pathway genes, potentially influencing the critical level of GLI3 activity or representing a target of GLI3 action, result in phenotypes which are mostly clinically different but share a disturbed development of the fingers, in particular the thumbs and halluces. This phenotypic peculiarity is observed for human patched (PTC) mutated in Gorlin syndrome (19,20), sonic hedgehog (SHH) mutated in holoprosencephaly 3 (21), the transcription factor CBP, a possible co-activator of Gli (22), mutated in Rubinstein-Taybi syndrome (23) and also for one of the potential morphogens induced by GLI3, the Dpp homologous bone morphogenetic protein 4 (BMP4), overexpressed in fibrodysplasia ossificans progressiva (24). From these observations it is likely that GLI3 serves a similar role in the hedgehog pathway in humans as in Drosophila.
The domain containing this putative phosphorylation site can also be predicted to be of particular functional importance based on the striking phenotypic differences between the dominant traits PHS (11), Greig syndrome and PAP-A (12). Protein chain termination 5′ of this domain, leaving the ZFD intact, induces hypothalamic hamartoma, hypopituitarism, imperforate anus and central or post-axial polydactyly (PHS) (11). We now find a C→T transition in a critical position within the domain associated with GCPS. Both PHS and GCPS show polysyndactyly and abnormal craniofacial features, but no GCPS case has hypothalamic hamartoma and PHS does not cause hypertelorism, broadening of the nasal root or a prominent forehead (11). Truncation of the protein 3′ of this domain leads to PAP-A only (U.Radhakrishna, A.Wild, K.-H.Grzeschik and S.Antonarakis, in preparation). The ectopic induction of additional fingers and toes varies in the three syndromes associated with GLI3 mutations. The polydactyly of GCPS is commonly pre-axial and, in addition, a post-axial postminimus is frequently seen. The polydactyly of PHS is typically central or post-axial, PAP-A usually shows a functioning extra post-axial digit (1,11,12). Since different phenotypes emerge when this region is present, deleted or altered it can be assumed that this domain is important for GLI3 function during development at different locations and time periods.
Considering the large size of the GLI3 transcript (Fig. 1) and the number of different domains well conserved during evolution in the homologous genes of the Gli-Krüppel family throughout the animal kingdom, as well as in paralogous GLI genes in man, and the almost ubiquitous expression of the gene in tissues during development and in adults (8,10), it can be anticipated that other
GLI3 mutations will be detected with the test system described here, in particular among the multitude of syndromes showing phenotypic overlap with the disorders caused by mutations of the hedgehog pathway genes.
Materials and Methods
The patients with GCPS analyzed here were clinically examined at the referring institutions and included into the study after informed consent was obtained. Cytogenetic analysis including FISH with a probe from within the GLI3 gene showed that they had normal karyotypes. The probands showed the typical phenotype associated with this syndrome, i.e. pre-axial polydactyly of the feet, broad or duplicated thumbs and halluces, post-axial postminimi on the hands, syndactylies of hands and feet and mild craniofacial abnormalities. Six affected members from three unrelated families and two sporadic cases were subjected to SSCA analysis of the entire GLI3 coding region. DNA samples from GCPS patients and a sample of 100 control individuals from the German population were purified using standard methods.
Genomic sequence analysis of the exon-intron boundaries
The cosmids gc68, gc73, gc492, gc493 and gc550, which span all of the GLI3 coding sequences (9), were used as template for automated sequencing on a 373A DNA sequencing system (Applied Biosystems). Aliquots of 500 ng-1 μg cosmid DNA were sequenced with 3.2 pmol primer in the presence of one or more of the following reagents: 1% DMSO, 1 U AmpliTaq® DNA polymerase (Perkin Elmer) or 250–500 ng T4 gene 32 protein (Boehringer Mannheim) in a 20 μl reaction volume (ABI Prism DNA sequencing kit). The primers were derived from the GLI3 cDNA sequence (8). The splice sites were determined by comparison of the derived sequences with the EMBL:HSGLI3 sequence (8). Parts of the sequences, including the exon-intron boundaries, are shown in Table 1
Exon amplification and single-strand conformation analysis (PCR-SSCA)
Thirty nine primer pairs for exon amplification were designed from the genomic sequences close to the exon-intron boundaries. When exons together with the intron sequence between the primer pairs were longer than 200 bp, primers were chosen within the cDNA which produced small overlapping fragments to cover an exon completely and also to identify possible splice site mutations (Table 1). PCR conditions for exon amplification were established using standard methods or a PCR optimization kit (Stratagene). Exon amplifications were carried out in parallel with total human DNA from a control person, DNA from the corresponding cosmid and DNA from GCPS patients. The DNA fragments were labeled by [α-32P]dCTP incorporation (0.1 μl/sample, 3000 Ci/mmol; Amersham) (26). A master mix for up to 46 reactions (15 μl/reaction) was prepared containing the relevant reaction buffer (Amersham), 0.2 mM deoxynucleotide mix, 20 μCi [α-32P]dCTP, 50 ng each primer and 0.3 U thermostable Taq polymerase (Amersham). Reactions were assembled on ice by mixing 14 μl master mix with 1 μl DNA (50–100 ng). PCR was performed in a DNA thermocycler (Biometra, Trioblock) with denaturation at 95°C for 3 min, 30 cycles of 95°C for 30–45 s, 52–65°C for 30 s, 72°C for 30 s and a final extension at 72°C for 3 min. Aliquots of 2.5 μl PCR products were diluted in 20 μl loading buffer (95% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol) and denatured for 5 min at 95°C. The DNA samples were cooled on ice for at least 2 min. Aliquots of 4 μl were electrophoresed in two non-denaturating 5% polyacrylamide gels, one with 0.5x TBE and 10% glycerol at room temperature and the other with 0.5x TBE but without glycerol at 4°C. After electrophoresis for 4 or 7 h at 30 W depending on gel conditions, gels were dried onto filter paper and exposed to X-ray film at −70°C. Non-denatured PCR products were included as controls to identify weak signals from renaturated double-strand DNA. When a deviant fragment was detected, the presence of this band with altered mobility was tested in a population sample of 100 unrelated individuals from the German population.
Sequencing of variant SSCA bands
Sequencing analysis of PCR products obtained using DNA from heterozygotes as template was directly performed after purification with a PCR purification kit (Qiagen). Allele-specific sequencing analysis of the wild-type and aberrant PCR products was performed following PCR amplification of the appropriate gel fragments cut out of the dried SSCA gels. For extraction gel pieces were incubated in 100 μl HPLC H2O for 30 min at 60°C or overnight at 40°C.
We thank Dr S.Demuth for providing us with material from patients MR2110 and MR2111 and H.Engel, K.Schwarz, J.Pongratz, L.Karolyi, E.Gebhrai, M.Faber and J.Kunz for help and discussion. This work was supported by grants from the Deutsche Forschungsgemeinschaft to K.-H.G. and from the Kulemann-Stiftung to M.K.-S.