Mutations in the C-Terminal Domain of Sonic Hedgehog Cause Holoprosencephaly

Abstract

Holoprosencephaly (HPE) is the most common brain anomaly in humans, involving abnormal formation and septation of the developing central nervous system. Among the heterogeneous causes of HPE, mutations in the Sonic Hedgehog (SHH) gene have been shown to result in an autosomal dominant form of the disorder. Here we describe a total of five different mutations in the processing domain encoded by exon 3 of SHH in familial and sporadic HPE. This is the first instance in humans where SHH mutations in the domain responsible for autocatalytic cleavage and cholesterol modification of the N-terminal signaling domain of the protein have been observed.

Introduction

Holoprosencephaly (HPE) has a prevalence of 1:16 000 live births and 1:250 during embryogenesis. HPE occurs when an altered gene product or teratogen interferes with critical early steps in normal development of the brain and midface. The severity of the brain malformation is extremely variable even within families with a single defined mutation in the Sonic Hedgehog (SHH) gene (1,2). The clinical variability ranges from cyclopia with a proboscis and perinatal lethality, at one extreme, to minimal manifestations including midfacial clefting, hypotelorism, microcephaly and a single central incisor (3–5). The disorder can be associated with environmental teratogens such as those cases related to maternal diabetes or, as recently described, with hypocholesterolemia.

Familial HPE occurs significantly less frequently than sporadic HPE; however, these families have been indispensable in dissecting the genetic factors that contribute to HPE. Based on specific cytogenetic anomalies non-randomly associated with HPE; at least four genetic regions are implicated in the pathogenesis of HPE (6–10). Autosomal dominant HPE has been linked to markers in the vicinity of a gene, designated HPE3, on chromosome 7q36 (11). The identity of HPE3 with the SHH gene was suggested by its location with respect to four different translocation breakpoints associated with HPE at 7q36 (2) and proven by the identification of five different mutations in families with autosomal dominant HPE (1).

Sonic hedgehog (Shh) (12) is the best described of three vertebrate genes (the others being Indian hedgehog and Desert hedgehog) that are highly homologous to the Drosophila segment polarity gene hedgehog (13,14). Shh is the only one of these vertebrate genes that is known to be expressed in the midline developing central nervous system (15). Shh is expressed early during development of the vertebrate notochord and floor plate (16,17) and controls patterning of ventral cell fates of the anterolateral mesoderm, somites (18) and motor neuron outgrowth (19) and is expressed along the entire rostrocaudal extent of the nervous system, including the ventral forebrain (20) and eye (21,22). In addition, Shh is expressed in the developing limb and contributes to specification of the antero-posterior axes (23,24).

The Shh protein is a secreted molecule with patterning activity in a variety of tissues. An understanding of the complex processing of Shh is essential to an understanding of the significance of the mutations that we describe in this report. The Shh protein is synthesized as a precursor that trafficks to microsomes, undergoes cleavage of a signal peptide and within the endoplasmic reticulum undergoes autoproteolytic cleavage into a 19 kDa N-terminal product (Shh-N) and a C-terminal product of 25 kDa (Shh-C). All of the known biological patterning activity resides in Shh-N (25–28). However, intrinsic to the autocatalytic cleavage reaction is hydrolysis of the reaction intermediate by cholesterol, which becomes covalently attached to the C-terminus of Shh-N (29–31). This unprecedented cholesterol modification is thought to play an important role in spatial restriction of the zone within which Shh acts in developing systems through the propensity for cholesterol to adhere to cell membranes (29). The view that these processing activities are also important in humans is strengthened by the illustration of HPE as a manifestation of Smith-Lemli-Opitz syndrome, a condition with biochemically proven defects in cholesterol metabolism (32–35). It is the C-terminal domain of Shh, encoded by exon 3, that possesses the sites for both cleavage and cholesterol modification activities.

Our initial search for mutations in the SHH gene focused on 30 families with autosomal dominant HPE and was limited to the first two exons (1). We now report analysis of the entire SHH gene in 41 HPE families (including the 30 families originally studied) and 184 sporadic HPE cases. In this report we demonstrate that mutations in the C-terminal domain of SHH are at least as common as those involving the signaling domain of the protein in human disease.

Results

We used the published sequence for exon 3 of the human SHH gene (2) to design primers for mutational analysis of 41 families with familial HPE and 184 patients with sporadic HPE using single-strand conformational polymorphism (SSCP) analysis. Exon 3 was divided into three overlapping amplicons for SSCP analysis. Previously we screened 30 HPE families for mutations in exons 1 and 2 (1). The 11 additional families were also screened for exons 1 and 2. Any band shifts detected in the screen were analyzed further by direct sequence analysis on either two affected individuals from each family or the same proband in two or more independent amplification reactions (1).

In Drosophila the autocatalytic site (29) lies between Gly257 and Cys258, which corresponds to Gly197 and Cys198 in the human protein (1,2), because the Drosophila signal peptide is considerably longer than in vertebrates. Therefore, all five of the disease-associated exon 3 mutations shown in Table 1 reside in the non-signaling domain of SHH-C and presumably affect the cleavage and/or cholesterol transferase activities of this portion of the protein.

Figure 1

SSCP analysis of the segregation of band shifts with HPE in the respective families. In (a) the SSCP pattern for family 11 shows that affected individuals are heterozygous for a novel allele (upper band) which contains a Val224Glu mutation and a normal allele (lower band). This novel band shift is present in all clinically affected individuals. Note that a DNA sample of the affected female in the first generation was not available for testing. In (b) the family with the Ala226Thr mutation both the proband and his clinically normal father have a normal and mutant allele by SSCP and by sequencing. In (c) the 263–269 deletion is detectable as the faster migrating allele which segregates with the affected members of the pedigree. The putative polymorphisms Ser190Ser and Gly292Gly were also detected in the SSCP screen for mutations on the basis of band shifts and were confirmed by sequencing (data not shown).

The Val224Glu mutation

Family 11 is a multi-generational family with autosomal dominant HPE (36) that has been shown to be linked to HPE3 (11). We detected a GTG (Val224)→GAG (Glu224) sequence change that predicts an amino acid substitution at an invariant position in the hedgehog gene family that is demonstrated by SSCP in all affected individuals and none of the unaffected family members (Figs 1a and 2a). This sequence variation is detectable by generation of a novel AluI restriction cleavage site. This codon is 28 amino acids from the cleavage site between SHH-N and SHH-C.

The Ala226Thr mutation

The second disease-associated sequence change (Figs 1b and 2b) was identified in a familial case of HPE not previously tested for linkage to 7q36 and occurs two codons 3′; of the Val224Glu mutation, implicating this region of the protein in processing. It is of note that the clinically unaffected father is also a mutation carrier, a phenomenon which has been previously described in familial HPE (11). The GCG (Ala226)→ACG (Thr226) sequence change predicts an amino acid substitution in an invariant amino acid in all of the vertebrate Hedgehog proteins, yet is divergent from the Drosophila sequence (Fig. 3). The C-terminus of the hedgehog gene products is far less conserved than the N-terminus. This mutation can be detected by the loss of a FnuIVH restriction site.

Figure 2

Representative chromatograms of the sequence alterations detected in HPE patients are illustrated: (a) Val224Glu; (b) Ala226Thr; (c) 263–269del; (d) Glu284(amber); (e) Ala384Thr. Note that the sequence in (c) was derived from the subcloned deleted allele and that a normal allele was also identified (data not shown). Two individuals were sequenced from each family (where applicable) or alternatively the same individual was analyzed in two separate amplification reactions. Note that the pre-peak sequencing artifact at position 223 immediately preceding the Val224Glu mutation was absent on the reverse strand sequence (data not shown).

The 263–269 deletion

The third sequence change was detected as a downward shift in the mobility of an allele which segregated in autosomal dominant HPE family 4 linked to 7q36 (Figs 1c and 2c). Sequence analysis suggested a deletion and this was confirmed by subcloning the amplicon from two affected individuals and the detection of two insert fragments that differed by exactly 21 bp. The smaller of these fragments contained a precise deletion of seven amino acids (RLLLTAA) immediately preceding a key His residue that is thought to be involved in processing of the Drosophila hedgehog protein (27; Fig. 3).

The Glu284(amber) mutation

The fourth sequence variation predicts premature termination of the protein at position 284 based on a GAG→TAG change. This mutation in autosomal dominant HPE family 3 was detected by sequence analysis of the entire SHH gene in families known to be linked to 7q36 (11). A subtle SSCP band shift was detectable in the initial screen (data not shown).

The Ala384Thr mutation

This is the only sequence variation that predicts a change in the primary coding sequence of the SHH protein that was detected in a screen of 184 sporadic HPE cases. The GCG→ACG sequence change predicts the substitution of Thr for Ala at position 384. The parents of this child were unavailable for analysis.

The Ser190Ser and Gly292Gly sequence variants

Our mutational analysis of exon 3 of the human SHH gene identified two putative polymorphisms that predict no change in the primary sequence of the translated protein. The TCG (Ser190)→TCA (Ser190) sequence variant occurs in the wobble position and predicts no amino acid change. We detected this sequence change in two separate individuals with sporadic HPE and none of the familial cases or normal controls. No parents were available for analysis. This sequence variant predicts a novel DdeI site in the mutated allele.

The second putative polymorphism, GGG (Gly292)→GGA (Gly292), also occurs in the wobble position and predicts no change in the translated protein. Neither the Ser190Ser nor Gly292Gly mutations have homology to either the 5′ or 3′ splice consensus sequences, which might cause possible splicing anomalies as a result of these sequence changes. The protein sequence alignment of this region of the hedgehog genes demonstrates a high degree of divergence (Fig. 3). No parents were available for analysis. This sequence variant was observed only once in a sporadic case of HPE and not in the familial HPE or control samples. This sequence change predicts the creation of a novel AvaII restriction site.

Figure 3

A protein sequence alignment of the hedgehog homologs is illustrated to show the location of the mutations in the human gene (shown in blue) with respect to the known Drosophila mutations (shown in green). The protein alignment includes the known sequences for the human Sonic (hSHH), Indian (hIHH) and Desert Hedgehog (hDHH) proteins, with reference to the mouse (mShh), zebrafish (zShh), chicken (cShh) and Drosophila (dhh) proteins (12). The hDHH protein sequence was not completely available (12). Amino acids in common with two or more species are indicated by the color red. The position of the splice site that unites exon 2 with exon 3 is shown, as is the position of the autocatalytic cleavage between Gly197 and Ser198 of the human protein.

Table 1 summarizes the position and prevalence of the mutations in the human SHH gene that have been detected to date. A preliminary genotype-phenotype comparison did not reveal any correlation between the clinical findings and the position of these specific mutations. A more detailed analysis is in progress.

Discussion

HPE is a common brain malformation in humans that has serious medical consequences and diverse etiology. We have shown that mutations in the human SHH gene are among the causes for HPE. An analogous brain malformation (cyclopia) can be reproduced by targeted disruption of the mouse Shh gene (40), supporting the role of SHH in the human disease. Detailed analysis of the expression pattern and signaling pathway for hedgehog and its counterparts in lower vertebrates paves the way for a systematic search for additional HPE genes in humans. From the fruit fly to humans this hedgehog family of genes has fascinating properties (reviewed in 41,42), including multiple activities in different tissues (e.g. the central nervous system and limb) and multiple activities of the protein itself (N-terminus with signaling, C-terminus with autoprocessing).

We performed the most extensive analysis to date on mutations in the human SHH gene in order to begin to quantify the prevalence of these mutations in HPE. If this survey of mutations in our present sample is representative, then we estimate that mutations in SHH account for 9/41 (∼23%) of the familial cases of HPE and 1/184 sporadic cases. Since our SSCP screen for mutations accounts for seven of the eight familial cases that have been linked directly to HPE3 on human chromosome 7q36 (11), there is little evidence for an additional HPE locus at 7q36. However, there must be extensive genetic heterogeneity in HPE, since only a subset of autosomal dominant HPE families are linked to 7q36 markers (11).

In Drosophila the majority of the mutations identified in phenotypic screens are deletions with a predicted null expression pattern. In addition, there are at least 12 different point mutations that have been described in the hedgehog gene (27,36–39) and they are grouped into class I mutations in hh-N (the 19 kDa N-terminal domain) and class II in hh-C (the 25 kDa C-terminal domain). All three of the class II mutations in Drosophila appear to have a measurable effect on the autoprocessing reaction. One of these class II mutations in Drosophila affects a critical His329 residue that appears to be catalytically involved in the cleavage reaction (37). As shown in Figure 3, the 263–269del mutation deletes seven amino acids immediately preceding this codon and predicts a shift in the position of this amino acid.

Therefore, these exon 3 mutations in the human SHH gene are comparable with the Drosophila class II mutations, although it is not clear if either or both of the activities of the C-terminal domain, namely autoprocessing and/or cholesterol modification, are altered by these mutations. Unlike the signaling portion of Shh-N, whose structure has been solved in the mouse (43), far less information is available on the structure of the C-terminal domain of the protein. Identification of class II mutations in the human SHH gene should contribute to a better understanding of the functions of this domain of the protein.

All of the mutations described in the human SHH gene to date are consistent with a loss of function of the mutated allele (1) and a patterning activity, especially of the midline central nervous system, that is extremely sensitive to dose of the SHH protein (44). Supportive evidence for haploinsufficiency for the SHH gene as a cause for HPE comes from the analysis of patients with cytogenetic deletions of the SHH gene (45). The functional consequences of class I and II mutations that we have described can now be tested; for example, expression of the mutated forms of the gene in vitro and/or in vivo will allow analysis of their biological activity, molecular size, extent of processing and degree of cholesterol modification to examine these questions further.

Materials and Methods

Patient cell lines and DNA extraction

For the majority of the samples from HPE patients analyzed, DNA was prepared from transformed lymphoblastoid cell lines and, where applicable, family members were also studied. Standard DNA extraction protocols were followed for processing of blood samples or lymphoblastoid cell lines. All samples were obtained by informed consent according to the guidelines of our institutional review board.

Single strand conformational polymorphism (SSCP) analysis

SSCP analysis was performed as described elsewhere (1,46) with minor modifications. The 10x dNTP stock mixture contained 2000 μM dATP, dGTP and dTTP, whereas dCTP was 625 μM for exons 3a and 3b, but 1250 μM for exon 3c. For a 15 μl reaction volume we used 60 ng genomic DNA template, 1.5 μl 10x PCR buffer with 1.75 mM MgCl₂ (Perkin-Elmer), 1.5 μl 10x dNTP (see above), 1.5 μl DMSO, 0.5 μl each forward and reverse primers, 0.1 μl Taq polymerase (Perkin-Elmer), 7.3 μl dH2O, 0.5 μl [α−³²P]dCTP (3000 Ci/mmol; NEN Dupont). Following amplification, 1 μl each sample was diluted with 10 μl stop solution [9.5 ml deionized formamide (Fluka), 0.4 ml 0.5 M EDTA, 0.5 ml 10% (w/v) bromophenol blue and 0.5 ml stock 10% (w/v) xylene cyanol] and the samples denatured for 10 min at 94°C in a thermal cycler and then immediately chilled on ice. The samples were analyzed on 2x MDE (FMC) sequencing gels followed by autoradiography.

PCR primers and sequencing

Exon 3 of the SHH gene was divided into three overlapping amplicons, 3a, 3b and 3c. Amplification of 3a yields a 305 bp product with forward primer 3F1a (5′-CCTCCCTCTCGGAACT-CAATGCCCTGTC-3′) and reverse primer SHH11 (5′ -TGCGC-GGCGGTGAGCAGCAGGCGCTCGCGC-3′). PCR reaction conditions for 3a were 94°C for 5 min, 94°C for 1 min, 60°C for 45 s, 72°C for 1 min for 35 cycles, followed by 72°C for 10 min. Amplification of 3b yields a 238 bp product with forward primer 3F2 (5′-CCAAGAAGGTCTTCTACGTGATCG-3′;) and reverse primer 3R1 (5′ AGCCGGCGGTCCCCGTCACGCTCG-3′). PCR reaction conditions for 3b were 94°C for 5 min, 94°C for 1 min, 60°C for 45 s, 72°C for 1 min for 35 cycles, followed by 72°C for 10 min. For amplification of 3c, because of its high GC content (75%), we could effectively analyze all but the last 221 bp of exon 3 using SSCP. Amplicon 3c uses forward primer 3F3 (5′-GCCCGGGCCAGCGCGTGTACGTGG-3′) and reverse primer 025 (5′-CCAGTGCAGCCAGGAGCGCGTGCG-3′) under PCR conditions 94°C for 5 min, 94°C for 1 min, 70°C for 45 s, 72°C for 1.5 min for 35 cycles, followed by 72°C for 10 min. This reaction yielded a 243 bp product.

Sequencing of amplicons from patients was performed by the Protein and DNA Core Facility of the Children's Hospital of Philadelphia on an ABI 3700 analyzer.

Acknowledgements

We are grateful for the participation of the families and support of clinical geneticists who facilitated these studies. We thank Jenny Chang for technical assistance and Charles Bevins for helpful comments. E.R. was supported by NIH training grant 5T32HD07107. F.V was partially supported by grant 201385/95-3 from CNP Brazil. E.B. was the recipient of a fellowship from Telethon, Italia. This work was supported in part by grants from the Canadian Genetic Disease Network, the Canadian Genome Analysis and Technology Program, a Howard Hughes International Scholarship and the Medical Research Council of Canada to L.-C.T. and by NIH grants HD28732 and HD29862 and by the Children's Hospital of Philadelphia Development Fund to M.M.

Author notes

+

Present address: Medical Genetics Branch, National Human Genome Research Institute, National Institutes of Health, 10 Center Drive-MSC 1852, Building 10, Room 10-C-101C, Bethesda, MD 20892-1852, USA