|Home | About | Journals | Submit | Contact Us | Français|
Interferon regulatory factor 6 (IRF6) belongs to a family of nine transcription factors that share a highly conserved helix–turn–helix DNA-binding domain and a less conserved protein-binding domain. Most IRFs regulate the expression of interferon-α and -β after viral infection1, but the function of IRF6 is unknown. The gene encoding IRF6 is located in the critical region for the Van der Woude syndrome (VWS; OMIM 119300) locus at chromosome 1q32–q41 (refs 2,3). The disorder is an autosomal dominant form of cleft lip and palate with lip pits4, and is the most common syndromic form of cleft lip or palate. Popliteal pterygium syndrome (PPS; OMIM 119500) is a disorder with a similar orofacial phenotype that also includes skin and genital anomalies5. Phenotypic overlap6 and linkage data7 suggest that these two disorders are allelic. We found a nonsense mutation in IRF6 in the affected twin of a pair of monozygotic twins who were discordant for VWS. Subsequently, we identified mutations in IRF6 in 45 additional unrelated families affected with VWS and distinct mutations in 13 families affected with PPS. Expression analyses showed high levels of Irf6 mRNA along the medial edge of the fusing palate, tooth buds, hair follicles, genitalia and skin. Our observations demonstrate that haploinsufficiency of IRF6 disrupts orofacial development and are consistent with dominant-negative mutations disturbing development of the skin and genitalia.
To identify the locus associated with VWS, we carried out direct sequence analysis of genes and presumptive transcripts in the 350-kilobase (kb) critical region3. This approach is confounded by single-nucleotide polymorphisms (SNPs), normal DNA sequence variation that occurs about once every 1,900 base pairs8 (bp). To distinguish between putative disease-causing mutations and SNPs, we studied a pair of monozygotic twins discordant for the VWS phenotype and whose parents were unaffected. Monozygotic status was confirmed by showing complete concordance of genotype at 20 microsatellite loci. We proposed that the only sequence difference between the twins would result from a somatic mutation found only in the affected twin. We identified a nonsense mutation in exon 4 of IRF6 in the affected twin, which was absent in both parents and the unaffected twin (Fig. 1a). We subsequently identified mutations in 45 additional unrelated families affected with VWS and in 13 families affected with PPS (Fig. 1b; Table 1), demonstrating unequivocally that these two syndromes are allelic6,7. These mutations were not observed in a minimum of 180 control chromosomes.
Clefts of the lip with or without cleft palate and isolated cleft palate are developmentally and genetically distinct9, yet VWS is a single-gene disorder that encompasses both clefting phenotypes. To verify this, we analyzed pedigrees (n = 22) that had a single mutation in IRF6 and affected individuals with both phenotypes. Genotype analysis of family VWS25 demonstrated that affected individuals, regardless of their phenotype, shared the 18-bp deletion found in the proband (Fig. 1a). We observed similar results in the other families and conclude that a single mutation in IRF6 can cause both types of cleft.
To determine the effect of mutations on IRF6 gene activity, we compared the type and position of the mutation with the phenotype. Previous identification of deletions encompassing the VWS locus (including IRF6 in its entirety) had suggested that the phenotype is caused by haploinsufficiency10–12. In this study, we found protein-truncation (nonsense and frameshift) mutations in 22 families (Fig. 1b). Protein-truncation mutations were significantly more common in VWS than in PPS (P = 0.004) and were consistent with haploinsufficiency in the VWS pedigrees. The lone exception to this relationship was a nonsense mutation introducing a stop codon in place of a glutamine codon at position 393, found in pedigree PPS11, which may be a dominant-negative mutation (see below).
The position of the missense mutations provides insight into the structure and function of the IRF6 gene product. When we aligned the family of IRF proteins, we observed that IRF6 has two conserved domains (Fig. 1b), a winged-helix DNA-binding domain (amino acids 13–113) and a protein-binding domain (amino acids 226–394) termed SMIR (Smad-interferon regulatory factor–binding domain)13. Studies of IRF3 and IRF7 have shown that the SMIR domain is required to form homo- and heterodimers14,15. The dimers then translocate to the nucleus, associate with other transcription factors and ultimately bind to their DNA targets14. Of the missense mutations, 35 of 37 localized to regions encoding these two domains. This distribution is non-random (P < 0.001), and we conclude that the domains are critical for IRF6 function.
Whereas the missense mutations that cause VWS were almost evenly divided between the two domains, most missense mutations that cause PPS were found in the DNA-binding domain (11 of 13, Fig. 1b). This distribution is significant (P = 0.03) and suggests that missense mutations in the DNA-binding domain associated with VWS and PPS affect IRF6 function differently. When we compared their positions with the crystal structure of the IRF1 DNA-binding domain16, we found that every amino-acid residue that was mutant in individuals with PPS directly contacts the DNA, whereas only one of seven of the residues mutant in the individuals with VWS contacts the DNA. Most notably, we observed missense mutations involving the same residue, Arg84, in seven unrelated PPS families (Fig. 1a,b). The Arg84 residue is comparable to the Arg82 residue of IRF1. It is one of four residues that make critical contacts with the core sequence, GAAA, and is essential for DNA binding16. The observed change of this residue to a cysteine or histidine caused a complete loss of that essential contact (Fig. 2). One possible explanation for this apparent genotype–phenotype relationship is that missense mutations that cause VWS are due to a complete loss of function of the mutated IRF6 protein, affecting both DNA and protein binding, whereas missense mutations causing PPS affect only IRF6's ability to bind DNA. The ability of the mutated IRF6 to bind to other proteins is unaffected, and it therefore forms inactive transcription complexes; thus, this is a dominant-negative mutation. Similarly, deletion of the DNA-binding domain of IRF3 or IRF7 exerts a dominant-negative effect on the virus-induced expression of the type I interferon genes and the RANTES gene15,17.
To correlate the expression of IRF6 with the phenotypes of VWS and PPS, we carried out RT–PCR, northern-blot analysis and whole-mount in situ hybridization. We found that Irf6 was broadly expressed in embryonic and adult mouse tissues (Fig. 3a,b), a pattern also seen in human fetal and adult tissues (data not shown). Greater expression of Irf6 seemed to occur in secondary palates dissected from day 14.5–15 mouse embryos and in adult skin. Whole-mount in situ hybridization demonstrated that Irf6 transcripts were highly expressed in the medial edges of the paired palatal shelves immediately before, and during, their fusion (Fig. 3d). Similarly high Irf6 expression was seen in the hair follicles and palatal rugae (Fig. 3d), tooth germs and thyroglossal duct (Fig. 3f) and external genitalia (Fig. 3h), and in skin throughout the body (data not shown). These observations are in accord with the VWS/PPS phenotype: notably, 20% of individuals with VWS exhibit agenesis of the second premolar teeth and 40% of individuals with PPS display genital anomalies.
Although we demonstrated that VWS and PPS are caused by mutations in a single gene, the phenotype for any given mutation varied in at least three ways even within the same family. Of the families with known mutations, we observed 32 families with multiple combinations of orofacial anomalies, 22 families with mixed clefting phenotypes (individuals with cleft lip and individuals with cleft palate only in the same family) and four families affected with PPS that included individuals who exhibit orofacial (VWS) features exclusively. The marked phenotypic variation in our cohort strongly implicates the action of stochastic factors or modifier genes on IRF6 function. In this context, we identified the sequence variant Val274Ile (Fig. 1b). This variant occurs at an absolutely conserved residue within the SMIR domain, is common in unaffected populations (3% in European-descended and 22% in Asian populations) and is an attractive candidate for a modifier of VWS, PPS, and other orofusial clefting disorders.
The mixed clefting phenotype is common in families affected with VWS, but very rare in families with non-syndromic orofacial clefts, and is not seen in most other syndromic forms of orofacial clefts. It is, however, also seen in clefting disorders caused by mutations in the genes MSX1 (ref. 18) and TP63 (ref. 19,20), suggesting that these may be involved in a common genetic pathway. In support of a common pathway, we found two IRF binding sites in the promoter of MSX1 and one in the intron, all of which are conserved between human mouse.
We are taking an integrated approach to dissecting the complex pathways that underlie development of the lip and palate, including genetic analysis to identify the mutations that cause orofacial clefts. The discordant monozygotic twins proved useful in this effort, and provided proof of principle21 that discordant monozygotic pairs can be used to search for modifiers or mutations, especially in regard to complex traits where mapping may be imprecise and mutation analysis may be confounded by SNPs. We also used a large number of samples from unrelated individuals to confirm that mutations in IRF6 are pathogenic for both VWS and PPS and to prove that IRF6 is essential for development of the lip and palate and is involved in development of the skin and external genitalia. The SMIR domain has been proposed to mediate an interaction between IRFs and Smads13, a family of transcription factors known to transduce TGF-β signals22. In addition, the expression of Irf6 along the medial edge of the palate seems to overlap with Tgfb3 (ref. 23), and Tgfb3, along with other members of this super-family such as Tgfb2 and Inhba, is required for palatal fusion24–27. Together with our data, these observations support a role for IRF6 in the transforming growth factor-β (TGF-β) signaling pathway, a developmental pathway of fundamental significance. The identification of IRF6 as a key determinant in orofacial development will help us to further delineate and integrate the molecular pathways underlying morphogenesis of the lip and palate.
Families affected with VWS (n = 107) and PPS (n = 15) were identified and examined by one or more geneticists or clinicians as previously described12. Nearly all families are of northern European descent. Sample collection and inclusion criteria for VWS and PPS were described previously3. We obtained written informed consent from all subjects and approval for all protocols from the Institutional Review Boards at the University of Iowa and at the University of Manchester.
We amplified exons 1–8 and part of exons 9 and 10 by standard PCR. The primer sequences are available on request. The amplified products were purified (Qiagen) and directly sequenced with an ABI Prism 3700. The sequence was analyzed using the computer program PolyPhred.
The IRF6 protein structure was predicted from its amino-acid sequence using Expasy, and aligned with the known crystalline structure of the DNA-binding domain of IRF1 using the UNIX-based computer software package Quanta (Accelrys). To model the mutations found at position Arg84 in the IRF6 DNA-binding domain, the residue was manually altered to a cysteine or a histidine. The package predicts all possible orientations of the altered side chain and displays the position with the highest probability.
We extracted total RNA using a standard guanidinium isothiocyanate, acid–phenol protocol. RT–PCR analyses were performed and analyzed as detailed previously28 using a forward primer designed in exon 4 and a reverse primer designed in exon 6 of Irf6. These primers generate a single product of 212 bp from cDNA.
A multiple-tissue northern blot (Seegene) was hybridized with a probe generated by PCR using primers derived from the distal end of the 3′ untranslated region of Irf6 and labeled as recommended by the manufacturer with the StripE-Z system (Ambion). We hybridized the blot in Express Hyb (Clontech), washed it as recommended and exposed it to X-ray film for 72 h at −80 °C.
Sense and anti-sense riboprobes were 1,600 bp in length, derived from the 3′ untranslated region of Irf6 and generated with Sp6 and T7 promoters, respectively. We fixed embryos dissected from time-mated MF1 mice in 4% paraformaldehyde overnight, processed them and subjected them to hybridization with sense or anti-sense probes as described previously29.
Statistical significance of mutation location was calculated with the Fisher's exact test using the assumption of equal probability for a mutation at each residue.
We thank our many clinical colleagues and their patients for contributing samples for this study (N. Akarsu, M. Aldred, Z. Ali-Khan, W.P. Allen, L. Bartoshesky, B. Bernhard, E. Bijlsma, E. Breslau-Siderius, C. Brewer, L. Brueton, B. Burton, J. Canady, A. Chakravarti, K. Chen, J. Clayton-Smith, M. Cunningham, A. David, B.B.A. de Vries, F.R. Desposito, K. Devriendt, R. Falk, J.-P. Fryns, R.J.M. Gardner, M. Golahi, J. Graham, M. Greenstein, M. Hannibal, E. Hauselman, R. Hennekam, G. Hoganson, L. Holmes, J. Hoogeboom, E. Hoyme, S. Kirkpatrick, J. Klein, T.C. Matise, L. Meisner, Z. Miedzybrodzka, J. Mulliken, A. Newlin, R. Pauli, W. Reardon, S. Roberts, H. Saal, A. Schinzel, J. Siegel-Bartelt, D. Sternen, V. Sybert, D. Tiziani, M.-P. Vazquez, L. Williamson-Kruse, F. Wilt, C. Yardin and K. Yoshiura). We appreciate the advice of K. Buetow, J. Dixon and C. Baldock; technical assistance from S. Hoper, M. Malik, J. Allaman, C. Hamm, N. Rorick, C. Nishimura, B. Ludwig, M. Fang, P. Hemerson, A. Westphalen and S. Lilly; administrative support from K. Krahn, D. Benton and L. Muilenburg; and sharing of unpublished results by P. Jezewski, A. Grossman and T.W. Mak. This work was supported by grants from the US National Institutes of Health and by grants to M.J.D. from Wellcome Trust, Action Research, Biotechnology and Biological Sciences Research Council, The European Union and the Fundação Lucentis (R.L.L.F.L. & D.M.F.).
Competing interests statement: The authors declare that they have no competing financial interests.