Expression of the Sonic hedgehog (SHH) Gene during Early Human Development and Phenotypic Expression of New Mutations Causing Holoprosencephaly

Abstract

Holoprosencephaly (HPE), the most common developmental defect of the forebrain and the face, is genetically heterogeneous. One of the genes involved, Sonic hedgehog (SHH), on 7q36, has been identified as the first HPE-causing gene both in mouse and humans. In order to delineate the phenotype of specific SHH mutations, we described the expression of the SHH gene during early human embryogenesis and investigated the phenotype of novel SHH mutations. In situ hybridization studies were performed on paraffin-embedded human embryo sections at three different development stages. These studies show that SHH is expressed in the notochord, the floorplate, the brain, the zone of polarizing activity and the gut. We also report on the phenotype of four novel mutations identified in 40 HPE families (two in isolated HPE and two in familial HPE). Expressivity ranged from alobar HPE to microcephaly and hypoplasia of the pituitary gland in one family, and from HPE to an asymptomatic form in another family. No SHH mutation was found in six polymalformed cases combining HPE with other defects, such as skeletal, limb, cardiac, anal and/or renal anomalies. This study confirms the genetic heterogeneity of HPE, and further demonstrates that SHH mutations are associated with a broad spectrum of cerebral midline defects.

Introduction

Holoprosencephaly (HPE; 1:16 000 live births; 1:250 conceptuses) is a common developmental defect affecting both the forebrain and the face (1,2). Clinical expressivity is variable, ranging from a single cerebral ventricle and cyclopia to clinically unaffected carriers in familial HPE. The disease is genetically heterogeneous but additional environmental agents also contribute to the aetiology of HPE (3–5). A previous segregation analysis led to the conclusion that the transmission of isolated HPE was compatible with autosomal dominant inheritance with incomplete penetrance (82%) of the disease (3). Also, a careful analysis of cytogenetic rearrangements in HPE patients led to the identification of at least 12 chromosome regions implicated in the pathogenesis of HPE (5). Among the putative loci for the disease, HPE3 has been shown by linkage analysis of families segregating for the disorder to be localized on chromosome 7q36, and Sonic hedgehog (SHH) has been identified as the first HPE-causing gene both in mouse and human (6–9). Both nonsense mutations and cytogenetic deletions suggest that a loss of function of the SHH gene is causing HPE. More recently, mutations in other genes have been shown to be associated with HPE (10–13; i.e. ZIC2 on 13q32; SIX3 on 2p21; PATCHED on 9q22.3; TGIF on 18p11.3).

SHH belongs to a vertebrate evolutionarily conserved family of genes (including Desert hedgehog and Indian hedgehog) which are related to the Drosophila segment polarity gene hedgehog (14,15). In vertebrates, shh is expressed in the midline central nervous system (CNS), the notochord and the limb bud zone of polarizing activity (ZPA), and the secreted shh protein is an inductive signal in the patterning of the ventral neural tube, the anterior-posterior limb axis and the ventral somites (16–19). However, whether SHH expression is the same in early human development is unknown. The human SHH has three exons (Fig. 1A) and encodes a 462 amino acid polypeptide (7). The protein is synthesized as a precursor molecule that undergoes cleavage of a signal peptide and then autoproteolytic cleavage. This reaction mediated by cholesterol leads to a 19 kDa N-terminal product (SHH-N) with the signalling domain and a C-terminal product of 25 kDa (SHH-C) possessing the cleavage domain closely associated with cholesterol transferase activity (20; Fig. 1B). Mutations of the SHH gene in human HPE are scattered along the whole protein (8,21; Fig. 1A).

In order to understand better the function of SHH and its relationships to the isolated HPE, we studied the expression of the SHH gene during early human embryogenesis. This study of SHH expression in human embryos strongly suggests that, as in other vertebrates, SHH plays an important role in human embryonic patterning. Moreover, we describe the clinical expression of four novel SHH mutations in 40 HPE families. This study, which represents only the second analysis of a large series, helps to establish the range of mutation types causing human HPE and allows SHH mutations to be correlated with a large spectrum of HPE phenotypes.

Figure 1

SHH gene and protein structure and human mutations. The SHH gene is composed of three exons (A) encoding a polypeptide of 462 amino acids (B)which is synthesized as a precursor molecule that undergoes cleavage of a signal peptide and then autoproteolytic cleavage (mediated by cholesterol) into a 19 kDa N-terminal product (SHH-N) and a 25 kDa C-terminal product (13; SHH-C). Reported mutations in the SHH gene are located on the cDNA, including four novel mutations reported in this study (underlined): missense mutations are at the top and nonsense or truncated mutations at the bottom. The localization of the probe used for in situ hybridization studies is represented.

Results

SHH expression during early human development

To define the temporal and spatial expression of SHH during early human development, we performed in situ hybridization on paraffin-embedded human embryos ranging from Carnegie 12 to 16. The SHH riboprobe was generated by cloning an exon 3 amplification product. Primers and probe sequences (173 bp) were chosen to amplify specifically the SHH gene and hybridize to the SHH mRNAs, respectively (8,14; no homology with human Desert hedgehog, 60% homology with human Indian hegdehog). At Carnegie 12, SHH is strongly expressed in both the notochord and the CNS. Here it is expressed along the entire length of the floorplate, in the spinal cord and the hind-brain (data not shown). SHH signal is also seen in the foregut (Fig. 2A and B), while no expression is detected in the midgut (Fig. 2C and D). At Carnegie 14, SHH expression is still observed in the notochord, the floorplate (Fig. 2K and L) and the foregut. In addition, a strong signal is detected in the hindgut (cloaca, Fig. 2E and F). In the CNS, it extends rostrally to the diencephalon and the prosencephalon (Fig. 2M and N), except in a small region of the ventral diencephalon where no SHH expression is detected. At this stage, posterior mesenchymal cells of the forelimb buds also express SHH (Fig. 2K and L). At Carnegie16, SHH expression is observed all over the CNS midline (Fig. 2O and P), except in the rostral part of the diencephalon where SHH message is located laterally (Fig. 2Q and R). SHH is still strongly expressed along the entire length of the notochord, the floorplate, the foregut and the hindgut. A weak signal is also detected in the lung buds (data not shown). At that stage of limb development, SHH is expressed in the posterior mesenchyme of both fore- and hindlimb buds (Fig. 2I and J). No expression was detected in the developing heart, liver or kidney. Table 1 summarizes SHH expression during early human development.

Table 1

SHH expression in early human embryos

Table 2

SHH mutations detected in our series

DNA analysis and patient phenotypes

We identified four novel SHH mutations in four unrelated non-syndromic HPE cases (two familial and two isolated). These four mutations were not observed in 100 control chromosomes. One mutation leads to a premature termination of SHH translation, whereas the three other mutations involved invariant amino acids (Table 2). In addition, no SHH mutation was found in six cases of HPE associated with other malformations.

The second mutation was a 474C→G transversion resulting in a stop codon (Y158X). This mutation abolished an RsaI restriction site. It was identified in a child (Fig. 3, Family 2) presenting with severe microcephaly (−7SD), a nasal pyriform aperture stenosis, a pituitary gland dysgenesis and growth retardation, but no brain malformation. The mutation was inherited from his mother who had hypotelorism and microcephaly (−4SD), a single central incisor, dysgenesis of the corpus callosum and a mild developmental delay. In addition, on the maternal side of the family, two patients died from typical HPE with arrhinia (III₅) and alobar HPE (III₆).

The third mutation (562G→C), a transversion of the last base of exon 2, resulted in a glutamic acid to glutamine substitution (E188Q) in an invariant amino acid in all of the vertebrate Hedgehog proteins, yet is divergent from the Drosophila sequence. This mutation created a BsaAI restriction site. The male proband (Figs 3 and 4B; Family 3) presented with lobar HPE, microcephaly (−5SD), diabetes insipidus and mental retardation. He was the only affected individual in the family. Surprisingly, the mutation was inherited from the mother who had a normal phenotype. A retrospective family history revealed the existence of several neonatal deaths on the maternal side of the family, but no information on brain or facial malformation was noted. Therefore, a familial form of HPE cannot be excluded. In addition, the causal nature of this mutation should be demonstrated at the functional level.

The fourth mutation was a 6647G→A transition changing an aspartic acid to an asparagine (D222N), 26 amino acids from the cleavage site between SHH-N and SHH-C. This mutation segregated over four generations. The female proband (Fig. 3, Family 4) had a semilobar HPE, whereas the great-grandmother had hypotelorism with unilateral cleft lip and palate. The proband's mother (III₉) and brother (IV₅) both had microcephaly and hypotelorism. In addition, many severe forms of HPE (II₃,II₄, III₃, III₆, III₇, III₁₁,IV₁) and at least one clinically unaffected carrier (III₁₀) were observed in this family.

Figure 2

SHH expression in human embryos. Bright-field (A, C, E, G, K, M, O and Q) and dark-field (B, D, F, H, J, L, N, P and R) illumination. (A-D) Transverse sections through a Carnegie 12 embryo showing SHH expression in the stomach (St) and duodenum (Duod) but not in the midgut (MG). (E and F) Transverse section through the caudal region of a Carnegie 14 embryo showing SHH expression in the cloaca. SHH is also expressed in the forelimb buds by Carnegie 14 (G and H) and in the hindlimb buds by Carnegie 16 (I and J). (K-N) Sagittal sections through a Carnegie 14 embryo showing a strong SHH signal in the notochord (NC), the floorplate (Ventral NT) and the ventral midline of the brain. (O-R) Transverse sections through the head of a Carnegie 16 embryo showing the midline expression of SHH in the CNS in the mesencephalon (Mes, O-R), and lateral expression in the diencephalon (Di, Q and R). Ao, aorta; H, heart; NC, notochord; NT, neural tube. Magnification: (A-D, H and J) ×80; (E and F) ×60; (I) ×35; (O and P) ×25; (G and K-N) ×20.

Discussion

HPE is an aetiologically heterogeneous developmental disoder involving both environmental and genetic factors. Among teratogens, the increased risk of HPE associated with maternal diabetes has been clearly established (4). HPE is also observed in several multiple malformation syndromes and chromosomal anomalies (5). Until recently, SHH was the first gene where mutations causing HPE were described, and the present study confirms and extends Roessler et al.'s data (8,21). Among 40 HPE families examined (including six cases of HPE associated with multiple malformations), we identified four novel SHH mutations in four unrelated HPE families (Table 2). As expected, the percentage of mutations was higher in familial cases (21). Presumably, several factors including ascertainment bias and our mutation detection strategy (direct sequencing) could explain why a higher rate of mutations was observed, especially in non-familial HPE.

Figure 3

Pedigrees of families with an SHH mutation. Filled symbols indicate typical cases of HPE (alobar, semilobar or lobar); half-filled symbols indicate minor manifestation of the HPE spectrum; open circle containing small closed circle indicates SHH mutation carrier with normal phenotype. M, SHH mutation carrier; N, no SHH mutation.

Figure 4

Variable phenotypes in patients with SHH mutations. (A) Family 1, II₃; (B) Family 3, II₂.

The Q100H mutation observed here is the first de novo SHH mutation described so far. A similar case was reported previously, but the parents of the proband were unavailable for analysis (21). In contrast, in one other isolated case, a clinically normal mother was found to be an SHH mutation carrier.

Although the effect of the E188Q and the D222N mutations on the SHH protein should be studied at the functional level, these observations underline the presence of incomplete penetrance. Therefore, in isolated or familial cases of HPE, phenotypically normal individuals could still be at risk of having offspring with severe forms of HPE. In addition, the variability of expression does not allow prediction of the severity of the phenotype, making genetic counselling difficult. The identification of a mutation within a family, however, permits exclusion of the risk of affected children for the non-carrier subjects, and prenatal diagnosis should be offered in other cases.

Table 3

SHH mutations in non-syndromic HPE

There is good evidence to believe that haploinsufficiency for SHH mutations might be responsible for HPE in humans: the presence of a premature truncated protein as predicted by the Y158X mutation, the 7q36 deletions observed in some HPE patients (22) and finally the existence of different missense mutations scattered along the coding sequence. To date, the functions of only a handful of mutations have been described in Drosophila hedgehog (23). These include missense mutations either affecting the structure of the N product (N-terminal domain, class I mutations) or blocking its release from the precursor (C-terminal domain, class II mutations). The three missense mutations reported here involved invariant amino acids of the SHH protein. Presumably these mutations affect the protein function, but in vitro functional studies will be worthwhile to confirm this point. Strikingly, in both series, all the inherited mutations were of maternal origin (21; 11/12 cases). This observation, along with a previous segregation analysis showing a significant excess of affected males with severe forms of HPE, suggest the possible involvement of a sex-dependent or an X-linked modifier gene (3).

Several studies conducted in chicken, mouse and zebrafish showed that the shh gene is expressed in the same embryonic spatial and temporal pattern (16–18). To date, however, no formal study of SHH expression had been done in humans. Here we report that, as in other vertebrates, SHH expression is observed in the notochord, the floorplate of the neural tube and the posterior limb buds (Table 1), where shh is suggested to be a key inductive signal. In addition, the shh gene is highly conserved among vertebrates. For example, there is 92.4% identity between human and mouse shh protein (14). This high level of interspecies conservation of both shh sequence and expression strongly suggests conserved functional properties in patterning of the ventral neural tube and the anterio-posterior limb axis. In the zebrafish, notochord shh expression disappears in a rostro-caudal fashion as it is activated in the floorplate, and it is restricted finally to the caudal region by stage 22 somites (16). Interestingly, in both mouse (16) and human, the expression in the notochord in observed for a longer period. Also in humans, SHH was expressed at the two opposite sites of the developing gut, in the foregut and hindgut, but not in the midgut. In the mouse, shh is restricted to the foregut endoderm at the 5-somite stage, and it is expressed in the caudal endoderm at the 8- to 11-somite stage (16). In the zebrafish, however, expression is described only in the foregut (18). Recent analyses of mice with a targeted deletion of the shh gene have demonstrated that shh is important in foregut development (24). Moreover, Gli2 and Gli3 are also essential in foregut development (25). This study shows in addition a strong SHH expression in the developing hindgut. Therefore, there is good evidence to suggest that genes of the SHH signalling pathway might be involved in foregut, but could also be responsible for hindgut defects in humans.

With respect to phenotype-genotype correlation, it is worthwhile to note that despite both the SHH expression in human and the phenotype of the shh^−/−mouse showing cerebral but also heart, limb, lung, vertebral and kidney anomalies (9), no SHH mutation has been identified in six cases of HPE associated with other malformations including other midline CNS, skeletal, limb, renal, anal and/or cardiac defects. In addition, those HPE patients with an SHH mutation reported so far do not show limb, vertebral or gut anomalies. This suggests that in humans, the midline CNS might be particularly sensitive to SHH gene dosage, which might be related to the short-range effect of the protein. Finally, in Family 2, the phenotypes of the mother (III₂) and her son (IV₁) were remarkable since they both presented a microcephaly and a mild midline defect without a brain malformation. These findings strongly suggest that SHH mutations are involved in some cases with dysgenesis/agenesis of the corpus callosum or of the pituitary gland, but the frequency of this event remains to be determined.

Genetic heterogeneity is now well demonstrated in HPE, where at least five genes when mutated are associated with this malformation. To date, in a large panel of unrelated HPE patients, there are no more than 5% of SHH mutations (21), and the frequency is even lower (<2%) for other genes: three mutations in S1X3 (11) and four mutations in Z1C2 (10), PATCHED (12) and TGIF (13), respectively. Interestingly, all HPE patients with a ZIC2 mutation had only minor facial dysmorphic features.

Conclusion

This study confirms the genetic heterogeneity of HPE. In addition, we show that SHH mutations are associated with a broad spectrum of cerebral midline defects and that apparently asymptomatic carriers are at risk of having affected children. Finally, our preliminary study of SHH gene expression in human embryos corroborates the results observed in model organisms and emphasizes the important role of SHH during human embryogenesis.

Materials and Methods

Patients and families

A total of 40 probands were studied for SHH mutations (fetuses and children). This population was comprised of 34 isolated HPE cases (20 sporadic and 14 familial cases) and six cases of HPE with other midline defects, skeletal, limb, renal, anal and/or cardiac anomalies. Patients' chromosomes were examined and were found to be normal.

DNA analysis

Standard DNA extraction protocols were followed for the processing of blood samples or established lymphoblastoid cell lines. All samples were obtained after informed consent according to the guidelines of our institutional review boards. PCR, single strand conformation polymorphism (SSCP) and sequencing were performed as described elsewhere (8,21,26). Amplification of exon 3 (except 3a) was performed by using Master Amp (Ampli Therm DNAPol 1.5 U) from Epicentre Technologies. SSCP was performed only for exon 3c. All the other amplicons were analysed by direct sequencing. The entire coding region was sequenced directly for each affected patient using the ABI prism dye terminator cycle sequencing ready reaction kit and the ABI 373A DNA sequencer (Perkin Elmer-Roche, CA). When a mutation was found in a proband, the segregation of this mutation in the family and in 100 control chromosomes was analysed. When possible, mutations were confirmed by enzymatic digestion on PCR amplification. Base pair changes were indicated by the nucleotide of the sequence (6) and according to the nomenclature system for human gene mutations (27).

In situ hybridization

Human embryos were collected from legally terminated pregnancies in agreement with the French law and the Ethics Committee recommendations. Tissues were fixed with 4% para formaldehyde, embedded in paraffin blocks and 5 µm sections were cut. A 173 bp PCR amplification of human genomic DNA encoding amino acids 210–267 was cloned into pCR-Script Amp SK(+) using a cloning kit (Stratagene). SHH exon 3 primers were: F5′-GCGGCACCAAGCTGGTGAAG-3′ and R5′-GGTGAGCAGCAGGCGCTCGC-3′. Sense and antisense riboprobes were generated using either T7 or T3 RNA polymerase in the presence of [α-³⁵S]UTP (1200 Ci/mmol; NEN). Labelled probes were purified on Sephadex G50 columns. Hybridization and post-hybridization washes were carried out according to standard protocols (28). Slides were dehydrated, exposed to BIOMAX X-ray films (Amersham) for 3 days, dipped in Kodak NTB2 emulsion for 3 weeks at +4°C, then developed and counterstained with toluidine blue, coverslipped with Eukitt, and analysed with dark and bright field illumination. No hybridization signal was detected with the α-³⁵S-labelled sense probes, confirming that the in situ hybridization patterns of the α-³⁵S-labelled antisense probes were specific.

Acknowledgements

We are very grateful to the families and for the support of clinical geneticists, obstetricians and paediatricians who sent us the blood samples (especially P. Talon, H. Balloul, M. Mathieu, J. Poinsot, V. Layet and P. Landrieu). We thank Martine Le Goudives, Brigitte Martinais, Valérie Perrault and Françoise Bessiere for technical assistance, and I. Kirillova for helpful discussions. This work was supported by the COREC (CHR Rennes).