The two canonical Apert mutations occur in a 16 amino acid linker segment, conserved in sequence between all four FGFRs. Strikingly, mutations exactly equivalent to the Pro253Arg mutation occur in both FGFR1 and FGFR3. These are uniquely and specifically associated with mild Pfeiffer (Pro252Arg in FGFR1) and Muenke (Pro250Arg in FGFR3) craniosynostosis syndromes respectively. In the case of FGFR3, a further genotype-phenotype correlation is that mutations of the adjacent amino acids in the linker (Arg248Cys and Ser249Cys) cause an entirely distinct disorder, type 1 thanatophoric dysplasia (Fig. 2). These differences in phenotype point to differences in the pathophysiological mechanism of the FGFR3 mutations. Whereas the Arg248Cys and Ser249Cys mutations create unpaired cysteines that predispose to ligand independent, constitutive activation via covalent cross-linking of receptor monomers (Table 2 and Naski et al 1996), the mechanism of the Pro250Arg mutation in FGFR3 has not been reported. By analogy with the Pro253Arg mutation in FGFR2, it is likely that this mutation acts by a ligand dependent mechanism (see previous section). In the study by helix 1 helix 2 helix 3
craniosynostosis parietal foramina
FIG. 3. Structure and amino acid sequence of the MSX2 homeobox. Amino acids that are conserved across all known msh-class homeoboxes are shown in bold. The positions of mutations mentioned in the text are indicated. The major DNA contacts are made by helix 3 and the flexible N-terminal arm, whereas helix 1 and 2 stabilize the domain folding.
Anderson et al (1998), FGF2 exhibited the greatest enhancement of binding affinity amongst those ligands tested (FGF1, -2, -4 and -6). However it seems unlikely that FGF2 is normally limiting for signalling in the cranial suture because Fgf2 knockout mice have normal skulls (Zhou et al 1998). An intriguing possibility is that there exists another FGF ligand, relatively specific to the cranial sutures, the binding affinity for which is enhanced by corresponding linker Pro ^ Arg mutations in the three FGFRs. The restriction of phenotypic expression of these FGFR linker mutations largely to the skull would then be explained by ligand limitation. To date, however, no FGF has been shown to have the requisite combination of properties.
Contrasting phenotypes of gain- and loss-of-function mutations of MSX2
As described above, a specific MSX2 mutation (Pro148His in the homeodomain) was the first molecular mutation in craniosynostosis. Homeodomains are highly conserved DNA binding motifs (Fig. 3), and it was subsequently demonstrated that this mutation enhanced binding, compared with wild-type, to an optimal target DNA sequence (Ma et al 1996). This remains the solitary MSX2 mutation causing craniosynostosis.
Rare patients with three wild-type copies of TWIST (owing to a chromosome duplication) have wide gaps between the skull bones, a phenotype termed cranium bifidum (Stankiewicz et al 1998). This suggests that craniosynostosis and cranium bifidum may represent opposite ends of a developmental spectrum of cranial suture anomalies, corresponding respectively to increased and decreased differentiation. By analogy we speculated that parietal foramina, a mild variant of cranium bifidum, might in some cases be caused by reduced signalling by MSX2. We have recently confirmed this idea by demonstrating three independent heterozygous MSX2 mutations associated with congenital parietal foramina (Wilkie et al 2000). The mutations identified are a complete deletion of MSX2 and two mutations of the homeodomain, an in-frame deletion of two amino acids (Arg159 and Lys1óO) and an Arg172His substitution. The LyslóO and Argl72 residues occur at positions 19 and 31, respectively of the homeodomain protein, which are identical (Fig. 3) across all known msh-class sequences (including those of several invertebrate phyla). Comparison of the structure of the closely related protein MSX1 to the homeodomain of engrailed, indicates that the residues play important roles in stabilising the protein or contacting the phosphate backbone of the bound DNA target (Li et al 1997).
For the MSX2 homeodomain mutants, we demonstrated reduction in binding to an optimal DNA target sequence of 97% (ArgLys159-1óOdel) and 85% (Arg172His) respectively (Wilkie et al 2OOO). We conclude that the phenotype of MSX2 haploinsufficiency in humans is parietal foramina. Most mutations of the homeodomain are likely to reduce, rather than increase, the binding of the domain to a target DNA sequence. Hence, parietal foramina are expected to be a more frequent phenotype caused by MSX2 mutation than craniosynostosis. Of note, the anatomical distribution of the skull defects in humans mirrors two lacunae in the calvarial expression of an Msx2 promoter—LacZ transgene in the mouse (Liu et al 1999).
This work confirms the hypothesis that cranium bifidum/parietal foramina and craniosynostosis may be viewed as phenotypes at opposite ends of a spectrum of abnormal calvarial development. We and others (Wuyts et al 1999) have demonstrated genetic heterogeneity in the aetiology of parietal foramina, which suggests new avenues to identify disease genes in craniosynostosis through the isolation of disease loci for parietal foramina.
A.O.M.W. thanks Yvonne Jones, John Heath and Gillian Morriss-Kay for discussions and the Wellcome Trust for financial support (Senior Research Fellowship in Clinical Science). He is honoured to record that he is the third member of his family to have contributed to the Ciba/ Novartis series of symposia. His mother, June Hill, participated in one and his father Douglas in two of these meetings (Hill 1961, Wilkie 1975,1981). This chapter is dedicated to their memory.
Was this article helpful?