Genomic screening of fibroblast growth-factor receptor 2 reveals a wide spectrum of mutations in patients with syndromic craniosynostosis

Abstract

It has been known for several years that heterozygous mutations of three members of the fibroblast growth-factor-receptor family of signal-transduction molecules-namely, FGFR1, FGFR2, and FGFR3-contribute significantly to disorders of bone patterning and growth. FGFR3 mutations, which predominantly cause short-limbed bone dysplasia, occur in all three major regions (i.e., extracellular, transmembrane, and intracellular) of the protein. By contrast, most mutations described in FGFR2 localize to just two exons (IIIa and IIIc), encoding the IgIII domain in the extracellular region, resulting in syndromic craniosynostosis including Apert, Crouzon, or Pfeiffer syndromes. Interpretation of this apparent clustering of mutations in FGFR2 has been hampered by the absence of any complete FGFR2-mutation screen. We have now undertaken such a screen in 259 patients with craniosynostosis in whom mutations in other genes (e.g., FGFR1, FGFR3, and TWIST) had been excluded; part of this screen was a cohort-based study, enabling unbiased estimates of the mutation distribution to be obtained. Although the majority (61/62 in the cohort sample) of FGFR2 mutations localized to the IIIa and IIIc exons, we identified mutations in seven additional exons-including six distinct mutations of the tyrosine kinase region and a single mutation of the IgII domain. The majority of patients with atypical mutations had diagnoses of Pfeiffer syndrome or Crouzon syndrome. Overall, FGFR2 mutations were present in 9.8% of patients with craniosynostosis who were included in a prospectively ascertained sample, but no mutations were found in association with isolated fusion of the metopic or sagittal sutures. We conclude that the spectrum of FGFR2 mutations causing craniosynostosis is wider than previously recognized but that, nevertheless, the IgIIIa/IIIc region represents a genuine mutation hotspot.

Figure 1

Identification of novel FGFR2 mutations, in the IgII domain (A) and in the TK1 and TK2 domains (B), in patients with craniosynostosis. The left side of each panel shows the DNA-sequence electropherogram of an individual with the heterozygous mutation (above), compared with that of a normal control (below), except that individually cloned alleles were sequenced in the case of the double-nucleotide substitution 514_515GC→TT; the right side of each panel shows the independent confirmation, by restriction digestion or ASO blotting, in family and control samples.

Figure 2

Amino acid sequence alignment around the sites of mutation, in the IgII domain (A) and in the TK1 and TK2 domains (B). In each case the FGFR2 sequence is shown at the top, with the substituted amino acids immediately above (red). Unfilled and hatched rectangles delimit regions of α-helix and β-sheet secondary structure, respectively, according to Plotnikov et al. (1999) (in the case of panel A) and Mohammadi et al. (1996) (in the case of panel B). Sequence identities of aligned proteins with FGFR2 are indicated by a dot. In panel A, Ig domains from human FGFRs, together with those from Drosophila melanogaster heartless (HTL) and Caenorhabditis elegans egg-laying defective 15 (EGL-15), are compared with MYLK, which is synonymous with telokin and provides a prototype for I-set Ig domains (Harpaz and Chothia 1994). In panel B, the positions of mutations in human FGFR3, D. melanogaster HTL, C. elegans EGL-15, and human KIT, RET, and MET are indicated by boxes, according to whether functional studies have shown the mutations to be activating (green) or inactivating (blue) or have not been undertaken (gray). Data are from DeVore et al. (1995); Gisselbrecht et al. (1996); Jeffers et al. (1997); Pelet et al. (1998); Iwashita et al. (1999); Longley et al. (1999); Stenberg et al. (; also see the KinMutBase protein-alignment web site); Bellus et al. (2000); Mortier et al. (2000); Iwashita et al. (2001); and the Human Gene Mutation Database.

Figure 3

Phenotype of patients heterozygous for FGFR2 mutations in the TK1 domain (A) and in the TK2 domain (B and C). A, Patient BL2622, with the N549H mutation, age 11 years. Note the crouzonoid appearance (left) and that thumbs and halluces are not significantly broadened (right). B, Patient OX2066, with the K641R mutation, age 8 mo. The facial appearance is not characteristic (left); note the tracheostomy. Viewed from above, the head has a markedly scaphocephalic contour; the thumb is broad with radial angulation (right). C, Patient OX1732, with the K659N mutation, age <1 year (left) and 3.4 years (right). The initial diagnosis was unclassified syndromic craniosynostosis. Note the marked turricephaly at the earlier age; the crouzonoid appearance is more evident with age.

Figure 4

Distribution of mutations identified in the complete screen of FGFR2. A, Distribution of de novo mutations in the Oxford sample (n=54), shown above a cartoon of the FGFR2 domain structure (drawn to scale) and a summary of the genomic organization (only exons are drawn to scale). Exons are shown as boxes, with coding regions in black, except for the alternatively spliced exons IIIb and IIIc, which are hatched. B, Distribution of inherited mutations in the Oxford sample (n=8). C, Distribution of all other mutations (n=23) described in the present report. Phenotypes shown are AS (blue), CS (green), PS (red), and other (syndromic or nonsyndromic) (black).

Figure 5

Position of residues mutated in the TK1 (upper lobe) and TK2 (lower lobe) domains of FGFR2, superimposed on the structure of the tyrosine kinase domain of FGFR1, as determined by Mohammadi et al. (1996), drawn by means of BobScript (Esnouf 1997), based on Raster3D software (Merritt and Murphy 1994). The catalytic (magenta) and activation (yellow) loops, as well as the side chains at which mutations occur (red), are shown.