Spondylocostal dysostosis (SCD) is an inherited disorder that is characterized by the presence of extensive hemivertebrae, truncal shortening and abnormally aligned ribs. It arises during embryonic development by a disruption of formation of somites (the precursor tissue of the vertebrae, ribs and associated tendons and muscles). Previously, three genes causing a subset of autosomal recessive forms of this disease have been identified: DLL3 (SCDO1: MIM 277300), MESP2 (SCDO2: MIM 608681) and LFNG (SCDO3: MIM609813). These genes are all important components of the Notch signaling pathway, which has multiple roles in development and disease. Here we have used autozygosity mapping to identify a mutation in a fourth Notch pathway gene, HAIRY-AND-ENHANCER-OF-SPLIT-7 (HES7), in an autosomal recessive SCD family. HES7 encodes a bHLH-Orange domain transcriptional repressor protein that is both a direct target of the Notch signaling pathway, and part of a negative feedback mechanism required to attenuate Notch signaling. A missense mutation was identified in the DNA-binding domain of the HES7 protein. Functional analysis revealed that the mutant HES7 was not able to repress gene expression by DNA binding or protein heterodimerization. This is the first report of mutation in the human HES7 gene, and provides further evidence for the importance of the Notch signaling pathway in the correct patterning of the axial skeleton.
The vertebral column relies on the linear alignment of regularly shaped vertebrae. The vertebrae, and associated tendons and muscles, are derived from somites, which are mesodermal condensations that arise in a reiterative manner during embryonic development. Abnormal vertebral segmentation (AVS) is a common congenital abnormality (1). Many manifestations of AVS exist resulting in uneven or fused vertebrae that can be present as a single defect in the vertebral column or in multiples. The term spondylocostal dysostosis (SCD) is applied to a wide variety of radiological features that include multiple AVS. These include contiguous involvement of more than 10 segments, often affecting all spinal regions, with mal-alignment of at least some ribs. Although there is some asymmetry in the rib alignment and the extent of rib fusions, there is a basic overall symmetry to the shape of the thorax. Genetic mapping and candidate gene sequencing approaches in recent years have identified causative mutations in three genes. SCDO1 (OMIM 277300) (2) represents around 20% of all cases, and is due to mutation of the DLL3 gene, with 25 distinct causative mutations identified to date (1–3 and unpublished data D Sparrow). Two rarer forms, each detected in only a single family, are due to mutation of MESP2 (SCDO2: OMIM 608681) (4) and LFNG (SCDO3: OMIM 609813) (5), respectively. Strikingly, all three genes are part of the Notch signaling pathway. In addition, some cases of the related Klippel–Feil syndrome (OMIM 214300, 118100, 148900), characterized by fusion of vertebrae within the cervical spine only, have been shown to be caused by mutation of the GDF6 gene, a member of the Bone Morphogenetic Protein family of signaling molecules (6).
Somites are paired blocks of mesoderm located on either side of the neural tube that form in a rostral to caudal direction by regular segmentation of newly arising paraxial mesoderm. The tissue that undergoes this segmentation is the presomitic mesoderm (PSM). Over the last decade research using mouse, chick, zebrafish and Xenopus laevis experimental models has revealed that this process is finely tuned at a molecular level by the interaction of several signal transduction pathways including FGF, Wnt and Notch (reviewed in 7). Of particular interest is the observation that the Notch pathway is activated in the unsegmented PSM in regular pulses, with a period corresponding to the time taken for a single somite to form. Other genes (perhaps as many as 100) (8), including some direct targets of Notch signaling such as Lfng and Hes7, are expressed in a similar oscillatory manner. Notch signaling is a highly evolutionarily conserved juxtacrine signal transduction pathway that is required for a wide range of cell fate decisions during embryonic development (9). Signaling occurs between single-pass transmembrane ligands and receptors present on adjacent cells. Ligand–receptor interaction triggers receptor cleavage, and the intracellular domain (ICD) of the receptor on the responding cell is transported to the nucleus, where it binds a transcription factor and transcriptionally activates target genes. This signal is terminated through ubiquitylation of the Notch ICD, followed by proteasome-dependent degradation. The basic pathway is modulated at a wide variety of levels including negative transcriptional feedback, post-translation modification and transcript and protein stability.
One notable class of mammalian Notch target genes are the Hairy-and-enhancer-of-split (HES) and Hairy-and-enhancer-of-split-related (HEY) families (10,11). These proteins belong to the basic helix–loop–helix (bHLH) superfamily of more than 125 DNA-binding transcription factors that regulate many biological processes in vertebrates, invertebrates and plants. The basic portion of the domain is required for DNA binding to E-box sequences (5′-CANNTG-3′), and the HLH portion is required for homo- and heterodimerization between family members. HES and HEY proteins are characterized by an additional conserved domain immediately C-terminal to the bHLH domain (the Orange domain), and a C-terminal tetrapeptide (WRPW in HES and YRPW in HEY) that interacts with transcriptional co-repressors of the Groucho family. In addition, HES proteins have a conserved proline residue in the middle of the basic domain that changes DNA-binding site specificity to N-box sequences (5′-CACNAG-3′). HES (and HEY) genes are direct targets of Notch signaling and some are key components of the somitogenesis clock. In mouse, genes such as Hes1, Hes5 and Hes7 are expressed in the PSM in an oscillatory pattern (12–14). This is achieved by an autoregulatory loop: once translated, HES proteins act on their own promoters to repress transcription. Due to the short half-life of HES proteins, autorepression is relieved allowing a new wave of transcription and translation every 90–120 min (which is also the time taken for a single mouse somite to form, compare humans in which one somite forms in ∼6 h) (7). In addition to somitogenesis, HES and HEY genes are involved in the developmental regulation of neurogenesis, vasculogenesis, cardiogenesis and cancer (reviewed in 10).
Here we have investigated a consanguineous family of Caucasian Mediterranean origin with a single offspring affected with SCD (Fig. 1). The proband was born at term, to a G3, P2, A0 35-year-old healthy mother, by caesarean section because of prenatal diagnosis of hydrocephalus and myelomeningocele at 8 months of gestation. At birth, weight was 3000 g (50th centile) and occipito-frontal circumference was 34 cm (50th centile). Bell-shaped, symmetrical and shortened thorax, lumbo-sacral myelomeningocele, ectopic stenotic anus and talipes were noted. Radiological examination showed shortening of the spine, with multiple and contiguous vertebral segmentation defects involving all spinal regions, but mainly the thoracic spine (Fig. 1). The ribs had very crowded origins on the left side, and were irregularly aligned with variable points of fusion along their length on the right side. Myelomeningocele and anal stenosis were repaired and ventriculoperitoneal shunt was placed. Neurogenic bladder was present. Cerebral CT scan showed Chiari II malformation. Abdominal-pelvic ultrasound was normal. Genitourinary malformations have not been identified. Only motor development has been mildly retarded (the patient was able to walk independently at 2 years of age). High-resolution karyotype was normal. At 4 years of age the patient weighed 12.5 kg (3rd centile), 90 cm in height (<3rd centile) and occipito-frontal circumference was 53 cm (98th centile). Analysis of the entire coding region and splice sites for the proband of the three genes (DLL3, MESP2 and LFNG), previously shown to cause SCD, revealed no deviations from the published sequence.
Autozygosity mapping of the disease locus
Since the parents were consanguineous (second cousins), we used autozygosity mapping to search for regions ‘identical-by-descent’ in the proband (15). Such regions are characterized by long runs of consecutive homozygous single-nucleotide polymorphisms (SNPs), referred to as regions-of-homozygosity (ROH), present in the affected individual, but that are heterozygous in the parents and unaffected siblings. SNP genotypes for 238,000 polymorphic markers were determined using the Affymetrix GeneChip® Human Mapping 250 K Sty Array for the proband, parents and two unaffected siblings. Genotype data were analyzed using Genotyping Console 2.0 software (Affymetrix), revealing only a single ROH >2 Mb in size. This was a 10.1 Mb run of 1081 homozygous SNPs on chromosome 17 (between SNPs rs8064630 and rs9893391) that was present in the proband, but absent from parents and unaffected siblings (Fig. 2A). This region contains 201 genes in NCBI MapViewer build 36.2.
Identification of candidate genes within the ROH
We first sought to identify SCD causative candidate genes within this interval by looking for genes that, when deleted in mouse embryos, cause somite dysmorphology similar to that shown by Dll3, Mesp2 and Lfng mutant null alleles. Two genes in the interval filled this criterion: HAIRY-AND-ENHANCER-OF-SPLIT 7 (HES7) and DISHEVELLED 2 (DVL2). In Hes7-null mice, somites are not properly segmented and their anterior–posterior polarity is disrupted (12). As a result, the somite derivatives such as the vertebrae and ribs are severely disorganized. When null mutant embryos were examined at the molecular level, it was found that although Notch1, Dll1 and Dll3 expression was not affected significantly, Lfng was expressed continuously throughout the mutant PSM (instead of oscillating). These results suggest that Hes7 controls the cyclic expression of Lfng and is essential for coordinated somite segmentation. Ninety percent of mice containing null mutations in Dvl2 have vertebral and rib malformations that affect the proximal as well as the distal parts of the ribs (16). In addition, around half of null embryos die prior to birth due to severe cardiovascular outflow tract defects, and a few embryos (2–3%) display thoracic spina bifida. Dvl2 is a member of the Dishevelled family of proteins that are essential for signaling by the Wnt family of secreted glycoproteins (17). Functionally, Dishevelled proteins act in the cytoplasm as a scaffold to help assemble a multiprotein complex responsible for establishing the intracellular concentration of beta-catenin (the ultimate effector molecule of canonical Wnt signaling that enters the nucleus to activate target gene transcription). They are also required for the other two Wnt signaling pathways (planar cell polarity, and Wnt-Ca2+/cyclic guanosine monophosphate). Wnt signaling has a poorly defined role in somitogenesis, though several pathway components are expressed in an oscillatory manner in the PSM, and several null mutations cause somitogenesis defects in mouse embryos. We complemented this approach by analyzing the interval using the Gentrepid candidate disease gene prediction tool (18). This uses two algorithms to prioritize candidates: Common Pathway Scanning (CPS) and Common Module Profiling (CMP). CPS is based on the assumption that common phenotypes are associated with mutations in proteins that are in the same complex or pathway. It applies data derived from protein–protein interaction and pathway databases to identify relationships between genes. Genes involved in the three signaling pathways known to be crucial for somite formation were our first priority for examination. CMP identifies likely candidates using domain-dependent sequence similarity, based on the hypothesis that disruption of genes of similar function will lead to the same phenotype. This analysis also flagged HES7 and DVL2 as being likely SCD disease gene candidates.
Candidate gene sequence analysis
We sequenced the entire coding region and splice sites of the HES7 and DVL2 genes in the proband (Fig. 2B) using primers listed in Table 1. DVL2 sequence did not deviate from the normal published sequence. In contrast, a homozygous missense mutation (c.73C>T) in exon 2 was detected in HES7, resulting in substitution of tryptophan for arginine (R25W). Sequencing of DNA from the proband’s parents and two siblings showed that both parents and one sibling were heterozygous for the mutant allele, and the other sibling was homozygous for the normal allele. The mutation deleted an SfoI restriction enzyme site, and the resultant restriction fragment length polymorphism (RFLP) was used to confirm the sequencing results in the pedigree (Fig. 2C). To demonstrate that this base change was not a common polymorphism unassociated with the SCD phenotype, 110 racially matched control subjects (220 chromosomes) were sequenced. No control chromosome contained a mutation at this position giving ∼80% power to distinguish a normal sequence variant from a mutation (19). In addition, the underlying base substitution was not present in the NCBI SNP database. In this analysis, two novel SNPs (one coding but synonymous and one non-coding) were identified that were also absent from the database (details submitted to dbSNP: rs61731639 and rs61758318). The R25W mutation alters a basic residue in the DNA-binding domain that is completely conserved through evolution (Fig. 2D). Crystal structures of distantly related E47, MyoD and NeuroD proteins binding to DNA in homo- and heterodimers (20,21) indicates that this arginine residue (R346 in E47) directly inserts into the major groove of the DNA backbone and contacts the base at position 3 of the recognition sequence, and is thus vitally important for DNA binding. Remarkably, a causative mutation for Saethre–Chotzen syndrome has been reported in the equivalent residue in the distantly related bHLH family protein TWIST1 that also causes an arginine-to-tryptophan substitution (22). This adds weight to our conclusion that this is a critical residue required for DNA binding, and that this particular amino acid substitution is deleterious to protein function.
Functional analysis of the R25W HES7 mutation
To provide evidence that this was indeed a SCD-causative mutation, we assessed the effect of the mutation on HES7 function using two previously described cell culture transcription repression assays (23). HES family proteins repress transcription through two distinct mechanisms. These proteins bind directly to DNA via an N-box (CACNAG) using a basic region immediately N-terminal to the helix–loop–helix domain (which is involved in homo- and heterodimerization with other bHLH family members). Co-repressors are then recruited to the promoter via interaction with a WRPW C-terminal motif. They can also form heterodimers with the bHLH protein E47, thus preventing it (and other bHLH proteins that normally heterodimerize with E47 such as MyoD) from binding to E-boxes (CANNTG) and activating transcription. Since no full-length human HES7 cDNA was available, we created the R25W mutation in the mouse gene. Expression of wild-type HES7 in mouse muscle satellite C2C12 cells represses transcription from a beta-actin reporter with upstream N-boxes (Fig. 3A), and represses E47-dependent transcription from a beta-actin reporter with upstream E-boxes (Fig. 3B). When the R25W mutant form of HES7 was used in the N-box assay, the levels of transcription were significantly increased above the control. This suggests that this mutant lacks normal repression activity. In the E-box assay, the mutant HES7 did not repress the reporter significantly over the control (without HES7 protein). This suggests that the R25W mutation in the DNA-binding domain of HES7 also impairs the ability of HES7 to heterodimerize with E47.
Heterozygous individuals have normal spinal and costal anatomy
Frequently, human genetic diseases caused by mutation of transcription factors show a dominant mode of inheritance. In addition, it has been observed that 43% of mice heterozygous for the Hes7-null allele manifest tail kinks (12), which can be caused by isolated somitogenesis defects. Thus we investigated whether any of the family members heterozygous for the mutant allele (parents and one sibling) showed any phenotypic abnormalities. Heights for all three individuals were in the normal range: the father was 174.5 cm, the mother was 163.5 cm and the heterozygous sibling was 168 cm at 12 years (>97 centile). Radiological examination showed that the anatomy of the spine and ribs were normal.
Our data strongly supports the hypothesis that the R25W missense mutation of HES7 is causative of SCD in the proband because it is not functional in two in vitro assays of HES7 protein function. We propose that this form of SCD be referred to as ‘SCDO4’ because mutation in HES7 shows that this case is genetically distinct from SCDO1-3.
We find that the R25W mutation is indistinguishable from the control (no HES7 present) in the heterodimerization (E-box) assay, and is thus completely non-functional in this assay. In contrast, in the N-box assay, where HES7 binds the N-box and represses transcription, the R25W HES7 mutant caused an unexpected increase in reporter activity over the control. This may be because C2C12 cells contain some level of endogenous HES family proteins that partially repress the N-box reporter in the absence of exogenously introduced HES7. The mutant HES7 might then act in a dominant negative manner to block the action of these endogenous HES proteins, and allow higher levels of activation of the N-box reporter. However, this dominant negative effect may also be an artifact of overexpression of the mutant protein in cell culture, since the three individuals in this family that are heterozygous for this mutation demonstrate no overt phenotype.
In addition to vertebral and costal malformations, the patient presented with hydrocephaly, myelomeningocele, ectopic stenotic anus and talipes. These phenotypes have not been noted in mice carrying two null Hes7 alleles, and could be a consequence of the R25W mutant protein having partial dominant negative effects in the patient. Although it was initially reported that in mouse Hes7 expression is exclusively limited to the PSM (12,24), recent high-throughput analysis of gene expression in the embryonic mouse brain suggests that low levels of Hes7 might also be present in the mid-brain and thalamus (24). Other members of the HES family (Hes1, 3, 5 and 6) are also present in the embryonic mouse brain (24), and null mutants for Hes1 show premature neurogenesis and severe neural tube defects (25). Thus, the R25W HES7 mutant may interfere with developmental processes controlled by other members of the HES family. However, the exact expression pattern of HES7 in the human embryo remains to be investigated, and it may differ from that determined for the mouse. Alternatively, the additional phenotypes seen in this patient could be a secondary consequence of the malsegmentation of the vertebrae in the developing human embryo, due to changes in mechanical stresses and alignments of tissues. However, it is also possible that the association of the additional phenotypes is purely coincidental. More patients with mutations in HES7 will be required to determine whether there is a correlation between mutation of HES7 and the other malformations present in the proband.
Various phenotypes associated with cytogenetic deletion of all or part of the short-arm of chromosome 17 are known, including Miller–Dieker syndrome (OMIM 247200) and various cancers (now shown to be due to deletion of the p53 tumor-supressor gene). However, the reported phenotypes of these deletions do not include AVS, and the likely causative genes have been identified, making it unlikely that loss of HES7 contributes to the observed phenotypes. In addition, partial or complete trisomy 17p has been reported in the literature (reviewed in 26), but is extremely rare. Nonetheless, it is interesting to note that in 2 out of 17 of such cases, scoliosis has been noted, suggesting that alterations in copy number of HES7 may have phenotypic consequences in such cytogenetic abnormalities.
MATERIALS AND METHODS
Appropriate informed consent was obtained from the proband and family for this study.
Affymetrix array processing and DNA sequencing
Genomic DNA was genotyped using the Affymetrix GeneChip® Human Mapping 250 K Sty Array at the Clive and Vera Ramaciotti Centre for Gene Function Analysis using standard protocols. Genomic DNA was sequenced as previously described (3), using the primers listed in Table 1.
The R25W mutation deletes an SfoI restriction enzyme site. A 273 bp fragment was amplified from genomic DNA using primers: 5′-CGGGCTCTCCCAGCGGCGGG-3′ (forward) and 5′-AACCAAGCTTGTGTCCCCAC-3′ (reverse), digested with SfoI (New England Biolabs) and separated on a 4% agarose gel. The wild-type sequence digests into 155 and 118 bp fragments, whereas the mutant sequence does not cut.
Mouse muscle satellite C2C12 cells were grown in DMEM (Gibco-BRL) containing 10% FCS (Sigma). Transfections were performed using LipofectAMINE/PLUS reagent (Invitrogen) in 12-well trays. Cells were co-transfected with plasmids encoding wild-type or mutant Hes7, E47 (for the E-box assay) and firefly (6× N-box beta-actin promoter in A or 7× E-box beta-actin promoter in B) and Renilla (SV40 promoter) luciferase reporters as previously described (23). Cells were harvested in 250 μl of Passive Lysis Buffer (Promega) 24 h after transfection. Firefly and Renilla luciferase activities were assayed using the Dual-Luciferase reporter system (Promega) and measured on a FLUOstar Optima Luminometer (BMG). Firefly luciferase counts were normalized against Renilla luciferase counts to account for differences in transfection efficiency. One-way ANOVA was performed on data from four independent experiments. Significance was determined using the Tukey’s post-hoc test.
National Health and Medical Research Council (Project Grant 404804 to S.L.D. and D.B.S.; and Senior Research Fellowship to S.L.D.). Westfield-Belconnen (Fellowship to D.B.S.). Pfizer Foundation Australia (Senior Research Fellowship to S.L.D.).
We thank the family for their cooperation, the Clive and Vera Ramaciotti Centre for Gene Function Analysis for Affymetrix array processing, and Wendy Chua for technical assistance. We are extremely grateful to Ryoichiro Kageyama for providing reagents.
Conflict of Interest statement. None declared.