|Home | About | Journals | Submit | Contact Us | Français|
Van der Woude syndrome (VWS) is an autosomal dominant malformation syndrome characterized by orofacial clefting (OFC) and lower lip pits. The clinical presentation of VWS is variable and can present as an isolated OFC, making it difficult to distinguish VWS cases from individuals with nonsyndromic OFCs. About 70% of causal VWS mutations occur in IRF6, a gene that is also associated with nonsyndromic OFCs. Screening for IRF6 mutations in apparently nonsyndromic cases has been performed in several modestly sized cohorts with mixed results. In the current study we screened 1521 trios with presumed nonsyndromic OFCs to determine the frequency of causal IRF6 mutations. We identified seven likely causal IRF6 mutations, although a posteriori review identified two misdiagnosed VWS families based on the presence of lip pits. We found no evidence for association between rare IRF6 polymorphisms and nonsyndromic OFCs. We combined our results with other similar studies (totaling 2,472 families) and conclude that causal IRF6 mutations are found in 0.24%-0.44% of apparently nonsyndromic OFC families. We suggest that clinical mutation screening for IRF6 be considered for certain family patterns such as families with mixed types of OFCs and/or autosomal dominant transmission.
Orofacial clefts (OFCs) are collectively the most common craniofacial birth defect in humans, and include cleft lip with or without cleft palate (CL/P) and cleft palate alone (CP) . The majority, approximately 70%, of OFCs are nonsyndromic, while the other 30% are syndromic, i.e. they occur in combination with other significant structural or cognitive anomalies. Nonsyndromic OFCs are etiologically complex and are causaully influenced by genetic risk factors, the combined action of genetic susceptibility and environmental exposures or from other unknown events. In contrast, syndromic OFCs are often caused by structural chromosomal anomalies or by coding mutations in a single gene . The most common OFC syndrome is Van der Woude syndrome (VWS, MIM #119300), accounting for 2% of all OFCs with an overall prevalence of 1/34,000 live births . The characteristic features of VWS are OFC congenital lower lip pits. While VWS is highly penetrant, there is remarkably variable expressivity . Individuals with VWS mutations can present with CL, CLP (cleft lip with cleft palate), CP, or no OFC at all. Further CL/P and CP may both occur in VWS families; such “mixed” families are rare in nonsyndromic OFC. The range of lip pit phenotypes is also broad as they may be bilateral or unilateral, and can also appear as conical elevations or shallow openings [5, 6]. Significantly, approximately 15% of individuals diagnosed with VWS lack lip pits, making it difficult to distinguish some individuals with VWS from individuals with nonsyndromic OFC .
All causal VWS mutations identified to date are in either IRF6  or GRHL3 . Loss-of-function mutations in IRF6 are responsible for about 70% of cases of VWS (VWS1, MIM # 119300), gain-of-function mutations in GRHL3 account for another 5% (VWS2, MIM # 606713), with the remaining 25% of VWS cases having no known causal mutation(s) and no identified gene(s). As an autosomal dominant disorder, the expected recurrence risk for first degree relatives (FDRs) of individuals with mutations in these genes is 50%, although the observed risk is less (~35% ) due to non-penetrance and de novo mutations. In contrast, the empiric recurrence risk for FDRs of individuals with nonsyndromic OFC falls in the 3-5% range [9-11]. Thus the difficulty of distinguishing VWS from nonsyndromic OFC can lead to skewed estimates of FDR recurrence risk for some families of affected individuals.
A number of studies have screened the IRF6 gene in nonsyndromic OFC cases in an effort to determine if clinical mutation screening should become routine or if unrecognized IRF6 mutations underlie the well-known association between common variants in IRF6 and nonsyndromic OFCs (summarized in Table 1) [11-22] . In this study, we report the results of our own IRF6 mutation screen of 1,521 case-parent trios with apparently nonsyndromic OFCs and compare these findings with the existing literature.
Samples for the current study were drawn from a large cohort of families with OFCs assembled for the purpose of sequencing genomic regions previously identified through genome-wide association studies (GWAS) . A total of 1521 case-parent trios were included where the affected subjects were diagnosed as having cleft lip alone (CL), cleft lip with cleft palate (CLP), or cleft palate alone (CP). This study included 1129 trios of Asian ancestry from China (N=497) and the Philippines (N=634); 383 trios of European ancestry from the United States (N=273), Hungary (N=65), Spain (N=27), Denmark (N=10), and Turkey (N=9); and 9 trios from Guatemala. Counts of trios by cleft type and population is available in Supplementary Table 1. All research work was approved by the Institutional Review Boards of participating institutions and informed consent was obtained for all individuals.
Thirteen regions with the best statistical and/or biological evidence for a role in nonsyndromic OFC  were selected for targeted sequencing. Full methodology including the exact coordinates was described in detail previously . Sequencing was performed by the Genome Institute at Washington University-St. Louis. Briefly, Illumina multiplexed libraries were constructed with 1ug of genomic DNA as previously described . NimbleGen (Roche NimbleGen, Madison, WI) custom target probes were designed to cover the target region, hybrid capture was performed on pools of 96 indexed samples, and each pool was sequenced on two lanes of Illumina HiSeq (Illumina Inc., San Diego, CA). Reads were mapped to the GRCh37-lite reference sequence using bwa v0.5.9, alignments were merged and duplicated marked with Picard v1.46, and germline and de novo variant calling was performed using Polymutt (v0.11). Candidate de novo mutations were further screened with VarScan trio and dbSNP 132 to remove likely germline variants that were under-called in parents.
Small indels were called in a three-step process. First, indels were called on a per-individual basis using VarScan v2.3.6 and filtered to remove likely artifacts . Next, the BAM files for all samples were realigned with DINDEL at all indel positions identified in the cohort. Finally, consensus indel genotypes were generated for all samples simultaneously using the realigned BAM files and VarScan 2.
The sequence data described here is available in dbGaP (accession number phs000625.v1.p1). Note, at the time the targeted regions were selected (2010), numerous candidate gene and GWAS studies had found statistically significant association between common variants in IRF6 and nonsyndromic OFC . GRHL3 had not yet been identified as the causative gene at the VWS2 locus  and had not shown evidence of association in GWAS studies of nonsyndromic OFC. Thus the complete IRF6 region, including all exons and introns, was sequenced but not the GRHL3 region.
After sequencing 1,521 OFC trios, we identified seven presumably pathogenic variants in the IRF6 gene (Table 1).
Three of the mutations occurred de novo. In the proband of family A (from Spain), we identified NM_006147.3:c.250C>T; p.Arg84Cys. The proband had left unilateral CLP and no reported other major medical conditions. In family B, from China, we identified a de novo mutation, c.1060+1G>T [p.?] in the exon 7 splice donor site in a boy with left unilateral CLP. Family C was a nuclear family from the Philippines. The proband had bilateral CLP and was the only affected individual in the immediate family although the family also reported two distant relatives (7th degree) with unknown cleft types. The de novo mutation, c.379delG; p.Gly127Valfs*43, resulted in deletion of the last nucleotide of exon 4 and likely alters the location of the splice donor site leading to a shift in the reading frame and truncation of the protein. In each of these families, lip pits were not explicitly reported. However, no photographs or medical records are available for review and these families were not examined for other signs of VWS at the time of recruitment (e.g. syndactyly or hypodontia).
Family D was a nuclear family from the Philippines with mixed OFC; i.e. each of CL, CLP, and CP was present in one of the three siblings. The mutation, c.39G>A; p.Trp13*, identified in the proband, was also found in his father. A posteriori examination of the gathered clinical data revealed the proband was missing his upper lateral incisors. No lip pits were reported or identified from pictures. Although the father was initially reported to be unaffected because he did not have an OFC, a posteriori review of facial photographs revealed bilateral lower lip pits. In addition, the father reported having an extra right maxillary molar. Other family members did not participate in the study; i.e. DNA and phenotypic data were not available. No other family history of OFC, lip pits, or dental anomalies was reported. This a posteriori review of collected phenotypic data is consistent with a diagnosis of VWS for this family.
Family E is a multiplex family from China with CL in the child and CLP in the mother (Figure 1). We identified a novel IRF6 mutation c.254G>T; p.Cys85Phe in both affected individuals. The proband’s brother also had CL but no DNA was available. The family was not evaluated for lip pits nor other signs of VWS, however no other malformations were reported at the time of recruitment. A similar story pertains to Family F, also from China. We identified a 1bp deletion (c.165delC; p.Ile56Phefs*7) in the proband who had left CL and her unaffected mother. A first-cousin on the paternal side was also reported to have an OFC. No photographs or medical records were available for review.
Family G was a multigenerational consanguineous family from Turkey (Figure 1). The proband had bilateral CLP and her father, grandfather, and great-grandmother all had bilateral CL. An aunt on her mother’s side had a unilateral CL. Only the proband and her parents were sequenced, identifying an in-frame deletion (c.1289_1297del; p.Asp430_Ile432del) in the proband and her father. A posteriori review of the medical notes indicated the presence of a single pit on the lower lip in the proband. Although no other history of lip pits was reported, these data are consistent with a diagnosis of VWS.
In the 1521 trios, we identified 10 other rare variants (minor allele frequencies less than 1%) in IRF6 that are unlikely to be causal for VWS (Table 2). Using the rare variant association extension of the Family Based Association Test (FBAT)  we found no evidence that these 10 rare IRF6 variants were associated with nonsyndromic OFCs (p=0.85). Three of these rare variants in IRF6 were present in a parent but not transmitted to the affected child, including a p.Arg45Trp variant that was previously reported as potentially pathogenic in a VWS family  despite the fact that the unaffected mother and unaffected twin sibling of the proband in the previous study also carried the Arg45Trp variant . In addition, while Arg45 is absolutely conserved  and is predicted to contact the DNA, the DNA contact is limited to the phosphate backbone outside of the core binding site . Finally, another missense mutation at this residue (Arg45Asn; rs121434229), was found previously in a control population. Overall, these data do not clearly support Arg45Trp as a causal variant for CL/P or VWS.
Previous estimates suggest VWS accounts for approximately 2% of all cases of OFC [29, 30]. Lip pits are absent in 15% of recognized VWS cases, resulting in a phenotype mimicking nonsyndromic OFC. Therefore, we estimate 0.3% of individuals with an OFC and no lip pits (i.e. those likely to be recruited for studies such as ours that target nonsyndromic OFC) will have a mutation in IRF6. In this current screen of 1,521 nonsyndromic CL/P trios, we identified seven likely causal IRF6 mutations (0.46%), slightly higher than the expected frequency. To date, screening of IRF6 in nonsyndromic cases has been undertaken in 12 studies, including this one (Table 3). Of the 2,472 families or singleton cases screened, only 17 carried causal mutations. In 6 of these, the proband was later determined to have lip pits, bulges, or other lip anomalies. Considering only those cases in which the proband did not have lip pits, the overall frequency of IRF6 mutations was 0.44% (11 of 2,466). We note photos were unavailable for five of our probands found to carry a detrimental IRF6 mutation so one or more of these individuals may have had lip pits. If we also exclude these families, the frequency of IRF6 mutations is 0.24%. Therefore, in all of the families sequenced to date (Table 1, ,3),3), IRF6 mutations have been found in 0.24-0.44% of probands who otherwise appear nonsyndromic.
Variants in IRF6 are associated with multiple phenotypes, e.g. common variants are associated with nonsyndromic OFC, while structural mutations cause VWS1 or popliteal pterygium syndrome (PPS, MIM #119500). Due to the incomplete penetrance observed for VWS, the least severe end of the VWS spectrum can include isolated thus apparently “nonsyndromic” OFCs, or lip pits alone, dental anomalies, or even no discernable phenotype. Because of this variability, whether IRF6 mutations cause truly “nonsyndromic” OFC remains a matter of some controversy. It is clear that common variants in IRF6 are strongly associated with nonsyndromic OFCs (including our own analysis of common variants from this same cohort ) and, due to linkage disequiibrium, may be tagging etiologic variants in regulatory regions. Lately, focus has shifted to rare variants as a potential source of the “missing” heritability in complex traits that remains undetected by GWAS. While rare variants in other genes have been suggested as contributing factors for OFC [31, 32], the current study suggests that rare coding mutations in IRF6 are unlikely to have a major role in risk for nonsyndromic OFCs.
Although families are often recruited for research studies based only on the phenotype of the proband, a “nonsyndromic” proband does not necessarily come from a nonsyndromic family, highlighting the impreciseness of the traditional contrast of syndromic versus nonsyndromic birth defects. Often “syndromic” is used as short hand for the presence of other anomalies, and “nonsyndromic” when only a single anomaly is observed. In the future more predictive distinctions may be drawn based on a gene-based framework . The current study also provides support for deep phenotyping of study families: it is noteworthy that in the 9 families where unaffected family members were reexamined, lip pits were found in a first-degree relative in 8 families, thus meeting the diagnostic criteria for VWS . Only one family has been reported to have nonsyndromic CL/P and a confirmed absence of lip pits . This highlights the importance of evaluating all family members for lip and/or other VWS associated phenotypes (e.g. dental anomalies). This can be difficult as lip anomalies may be small or irregular, removed surgically early in life , or not recognized as a sign of the disease. Researchers face additional challenges when the family is ascertained based on the proband’s phenotype alone and photographs of family members are unavailable.
Of the seven mutations identified in families A-G only Arg84Cys and the in-frame deletion of Asp430-Ile432 have been previously reported in other VWS families . Arg84Cys is thought to act in a dominant negative manner and often arises de novo. Arg84Cys is strongly associated with PPS, but is sometimes reported in VWS [30, 35], for example a Japanese VWS family in which the mutation was carried by an unaffected father indicating that genetic modifiers or stochastic effects could contribute to the penetrance of the PPS phenotype .. We also found a de novo occurrence of this mutation, in an individual reported to have nonsyndromic CLP (Family A). While the mutations in families B-F have not been reported previously, nonsense (family D), splicing (family B), and frameshift (family C,F) mutations are commonly found in VWS . Cys85 is one of four critical residues in the IRF6 DNA-binding domain predicted to contact the DNA in the core binding site . Given the similarity of these mutations to those documented previously in VWS families, we conclude these are likely causative mutations.
Based on the results of their previous IRF6 screens, multiple groups have suggested screening the IRF6 gene for nonsyndromic OFC families with at least one affected parent-offspring pair, those segregating both CL/P and CP, and those where the presence or absence of a lip phenotype is unclear [12, 13, 15]. Others have suggested IRF6 testing should be considered for families with apparently dominant transmission of OFC, or those with mixed types of OFC (ie both CL/P and CP in the family) to maximize the accuracy of genetic counseling . In our seven families, only families D, E, and G would have been recommended for IRF6 testing under the latter criteria because families A, B, C, and F had no reported history of OFC. We would recommend genetic counseling for each of these families, as the empiric risk for OFC in FDRs of VWS probands has been estimated as about 35%  in marked contrast to the empiric FDR risk of 4-5% for nonsyndromic CL/P and 3% for nonsyndromic CP.
In summary, we have shown that pathogenic IRF6 mutations are likely to occur in a small fraction (~0.3%) of individuals with apparent nonsyndromic CL/P, across multiple ethnic groups. As others have shown, a posteriori review of phenotypic data or photographs will often reveal lip pits in first degree relatives. Therefore, we strongly recommend that research participants recruited for studies of nonsyndromic OFCs be photographed when possible and asked about personal and family histories of lip anomalies and other minor VWS signs, such as missing or extra teeth. Our data indicate that universal screening of IRF6 in nonsyndromic OFCs would have low diagnostic yield. However, at this point it is difficult to propose a definite strategy for routine clinical screening of the IRF6 gene, as none of the previously proposed strategies would have recommended that screening occur in all seven of our families with pathological IRF6 variants. We agree with others who have suggested that IRF6 testing should be considered for a subset of OFC families, such as those with mixed clefting or dominant inheritance patterns as such families are enriched with rare variants in IRF6.
We are grateful to the individuals who participated in this study and to the staff at each recruitment site around the world, without whose efforts this study would not have been possible. We would like to thank Drs. Hong Wang (Peking University), Xiaoqian Ye (Wuhan University) and Bing Shi (Sichuan University) who directed the recruitment sites in China. We would also like to acknowledge the contributions of the CleftSeq consortium, in particular Margaret A. Taub, Karyn Meltz Steinberg, Jacqueline B. Hetmanski, David E. Larson, and George M. Weinstock, for generation, quality control, and analysis of the targeted sequencing data. The authors would like to thank the Exome Aggregation Consortium and the groups that provided exome variant data for comparison. A full list of contributing groups can be found at http://exac.broadinstitute.org/about. This work was supported by grants from the National Institutes of Health DE025060 [EJL], HG005925 [JCM, MLM], DE008559 [JCM, MLM], DE009886 [MLM], DE016930 [MLM], DE016148 [MLM], DE014581 [THB], DE018993 [THB], DE011931 [JTH].
The authors have no conflicts to disclose.