Facial clefts are common birth defects with a strong genetic component. To identify fetal genetic risk factors for clefting, 1536 SNPs in 357 candidate genes were genotyped in two population-based samples from Scandinavia (Norway: 562 case-parent and 592 control-parent triads; Denmark: 235 case-parent triads).
We used two complementary statistical methods, TRIMM and HAPLIN, to look for associations across these two national samples. TRIMM tests for association in each gene by using multi-SNP genotypes from case-parent triads directly without the need to infer haplotypes. HAPLIN on the other hand estimates the full haplotype distribution over a set of SNPs and estimates relative risks associated with each haplotype. For isolated cleft lip with or without cleft palate (I-CL/P), TRIMM and HAPLIN both identified significant associations with IRF6 and ADH1C in both populations, but only HAPLIN found an association with FGF12. For isolated cleft palate (I-CP), TRIMM found associations with ALX3, MKX, and PDGFC in both populations, but only the association with PDGFC was identified by HAPLIN. In addition, HAPLIN identified an association with ETV5 that was not detected by TRIMM.
Strong associations with seven genes were replicated in the Scandinavian samples and our approach effectively replicated the strongest previously known association in clefting—with IRF6. Based on two national cleft cohorts of similar ancestry, two robust statistical methods and a large panel of SNPs in the most promising cleft candidate genes to date, this study identified a previously unknown association with clefting for ADH1C and provides additional candidates and analytic approaches to advance the field.
Citation: Jugessur A, Shi M, Gjessing HK, Lie RT, Wilcox AJ, Weinberg CR, et al. (2009) Genetic Determinants of Facial Clefting: Analysis of 357 Candidate Genes Using Two National Cleft Studies from Scandinavia. PLoS ONE 4(4): e5385. https://doi.org/10.1371/journal.pone.0005385
Editor: Syed A. Aziz, Health Canada, Canada
Received: January 12, 2009; Accepted: March 20, 2009; Published: April 29, 2009
Copyright: © 2009 Jugessur et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: CIDR is fully funded through a federal contract from the National Institutes of Health (NIH) to The Johns Hopkins University, Contract Number N01-HG-65403. This research was supported in part by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences; by NIH grants DE08559, P60 DE13076, NIH P30 ES05605, and RO1 DE-11948-04; and by the Norwegian Research Council. We also thank the US National Institute of Dental and Craniofacial Research (NIDCR) for underwriting a significant proportion of the genotyping costs by CIDR. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
With an average worldwide prevalence of 1.2/1000 live births, facial clefts are the most common craniofacial birth defects and one of the most common major types of defect in humans . The extensive surgical, dental and speech involvement, as well as potential psychological sequelae, underscore the importance of elucidating the causes of these complex facial defects. Analysis of familial recurrence, segregation, and concordance in twins have provided compelling evidence for a very strong genetic component to clefting . In Norway, for example, the risk among first-degree relatives is approximately 40-fold higher than in the general population . However, the joint contribution of genetic variants has to date accounted for only a modest fraction of the recognized etiologies .
The past few years have witnessed major strides in the mapping of facial cleft loci, with the list of genes rapidly expanding from the first reported association of TGFA [MIM 190170] with isolated cleft lip with or without cleft palate (I-CL/P) in 1989  to now include IRF6 [MIM 607199], MSX1 [MIM 142983], TGFB3 [MIM 190230], FOXE1 [MIM 602617], FGFR1 [MIM 136350], FGFR2 [MIM 176943], FGF8 [MIM 600483], PDGFC [MIM 608452], CRISPLD2 [MIM 612434], PVRL1 [MIM 600644], GABRB3 [MIM 137192], MSX2 [MIM 123101], SATB2 [MIM 608148], TBX10 [MIM 604648], TBX22 [MIM 300307], GLI2 [MIM 165230], JAG2 [MIM 602570], MTHFR [MIM 607093], RARA [MIM 180240], LHX8 [MIM 604425], SKI [MIM 164780] and SPRY2 [MIM 602466], among the many promising candidate genes for clefts , , , , , , , , . Although linkage studies have been successful in mapping a number of these key candidate genes , studies relying on linkage disequilibrium (LD) have grown in popularity because they can be more effective in detecting weaker associations from multiple common and low-penetrance alleles , . Further, analyses based on haplotypes can outperform single-point analyses in which multiple SNPs in a gene are interrogated one at a time, because haplotypes can increase the overall information content at a given locus  and potentially capture association signals from variants that have not been directly typed .
We adopted this LD-based approach in the current search for fetal genetic risk factors for I-CL/P and isolated cleft palate (I-CP). A total of 1536 SNPs were selected in 357 candidate genes for facial clefts and genotyped in two population-based samples from Scandinavia (Norway and Denmark). The multi-SNP genotype data were analyzed using two complementary statistical methods, TRIMM  and HAPLIN , to detect multi-marker transmission distortion and to look for consistency in genetic associations across the two national samples.
A nationwide case-control study of facial clefts in Norway (1996–2001) provided 562 case-parent triads and 592 control-parent triads for analysis. The overall study design and characteristics of the study participants have been described elsewhere . Of the 562 case-parent triads, 114 were I-CP and 311 were I-CL/P. An additional 69 I-CP and 166 I-CL/P triads were available from a population-based study of facial clefts in Denmark (1991–2001) .
Candidate genes and SNP selection
Candidate genes for facial clefts were selected from a variety of resources, including published linkage and association studies on clefts, genome-wide scans, gene-knockout experiments in mice, studies of chromosomal rearrangements in humans, and gene expression analyses in human and mouse embryonic tissues , , , , , . The Craniofacial and Oral Gene Expression Network (COGENE; http://hg.wustl.edu/COGENE) catalogs human gene expression changes during early embryonic development and contains expression profiles on 25 different tissues/stages in the craniofacial region of human embryos . We searched the Serial Analysis of Gene Expression (SAGE) libraries to see whether a particular gene of interest is expressed in the relevant embryonic tissues at the pertinent developmental stage (weeks 5–6 for fusion of the embryonic lip; weeks 7–10 for fusion of the palatal shelves ).
Candidate genes were also chosen from two other global approaches for studying gene function. The first is the mouse mutagen N-ethyl-N-nitrosourea (ENU) project, which aims to characterize the functions of genes on mouse chromosome 11 (syntenic to human chromosome 17) by saturating the chromosome with point mutations . The second approach is to query the Developmental Genome Anatomy Project (DGAP; http://www.bwhpathology.org/dgap/default.aspx) database using the search option ‘craniofacial: including clefts, cranial abnormalities or defects such as microcephaly’. DGAP identifies chromosomal rearrangements in patients with multiple congenital anomalies and uses these rearrangements to map and identify genes that are disrupted or dysregulated in critical stages of human development .
Finally, Mendelian forms of clefting provide another important avenue for identifying genes that may underlie the more common and isolated forms of clefts, especially if these syndromic forms can occasionally manifest as phenocopies of isolated clefts (e.g. Van der Woude syndrome, VWS [MIM 119300]). The Online Mendelian Inheritance in Man (OMIM; http://www.ncbi.nlm.nih.gov/Omim) maintains a catalog of human genes and genetic disorders, providing a rich resource for selecting candidate genes for clefts. Using the search query ‘cleft lip OR cleft palate’, 439 entries were retrieved from OMIM.
Using the above strategies, we compiled a list of 357 candidate genes that may contribute to facial cleft etiology (Table S1). To guide the selection of SNPs, the genome browser of the International HapMap Consortium (http://www.hapmap.org) was used to retrieve genotype data on these 357 candidate genes. We used genotypes from the CEPH samples of Northern and Western European ancestry for consistency with our white study populations. SNPs were also selected from additional databases: dbSNP (http://www.ncbi.nlm.nih.gov/projects/SNP); Japanese SNP database (JSNP; http://snp.ims.u-tokyo.ac.jp); genome browser at UCSC (http://genome.ucsc.edu); CHIP bioinformatics tools (http://snpper.chip.org/bio/); and Seattle SNPs (http://pga.mbt.washington.edu). The following criteria were used to prioritize SNP selection: prior evidence of an association with facial clefts; haplotype-tagging properties; minor allele frequency (MAF) of at least 10%; and a preference for coding SNPs and SNPs located in putative regulatory regions in the UTRs. SNPs with high MAFs are more likely to have arisen on ancestral chromosomes, making them more useful in assessing the degree of LD in the Norwegian and Danish populations. Intragenic and regulatory SNPs are more likely to be of functional importance, while haplotype-tagging SNPs (htSNPs) are valuable in that they use LD to extract the maximum amount of genetic information from a particular haplotype block and to possibly tag for SNPs with an unrecognized biological function.
A combination of software, including HAPLOVIEW (http://www.broad.mit.edu/haploview/haploview; ), Best Enumeration of SNP Tags (BEST; http://www.genomethods.org/best/index.htm; ), and SNP Browser™ (Applied Biosystems; Foster City, CA), was used to evaluate MAF, inter-marker distance, as well as LD patterns and haplotype block structures for the selection of htSNPs. SNP assays were designed by Illumina (http://www.illumina.com; San Diego, CA) and a ‘design score’ was computed for each SNP using an algorithm that rigorously tests the performance of that SNP on an Illumina GoldenGate™ platform (Illumina, San Diego, CA). Approximately 10% more SNPs than the full panel of 1536 SNPs were initially submitted to allow flexibility in choosing substitute SNPs for those with poor design scores. After multiple rounds of SNP evaluations, a custom panel of 1536 SNPs was finalized for the 357 genes (Table S2). Genotyping was done by the US Center for Inherited Disease Research (CIDR; http://www.cidr.jhmi.edu).
The within- and between-plate genotype reproducibility rates of 96 duplicate DNA samples and three additional samples that were genotyped multiple times were used to evaluate the quality of genotyping. A SNP was deemed to have failed if fewer than 95% of the samples generated a genotype at the locus. SNPs with low minor allele frequency (MAF<1%) were also excluded due to their lack of statistical power. Deviations from HWE in control samples may indicate the presence of systematic genotyping errors, latent population substructure, or a biological effect such as natural selection , , . For the Triad Multi-Marker test  (TRIMM; see below), deviation from HWE (p<0.05) was used to search for assay problems directly, since the validity of TRIMM does not require SNPs to be in HWE. SNPs that showed deviation from HWE in all of the following three sample sets: parents in Norwegian case triads, parents in Norwegian control triads, and parents in Danish triads, were deemed to be problematic and three additional SNPs were removed based on this criterion.
To screen for Mendelian inconsistencies within families, we used PedCheck . Families with excessive Mendelian inconsistencies are probably the result of sample switches or misidentified paternity. We retained families with fewer than 10 inconsistencies and removed eighteen families with 65 or more Mendelian inconsistencies from the analyses (there were no families with between 10 to 64 Mendelian inconsistencies).
After data cleaning, the final number of SNPs available for analysis was 1315 in 334 candidate genes on autosomal chromosomes, with one to twelve SNPs typed per gene, four SNPs on average (see Tables S1 and S2 for details).
In this study, we investigated genes on autosomal chromosomes only. Two complementary methods, TRIMM  and HAPLIN , were used to analyze the case-parent triad genotype data. TRIMM is a robust and intuitive statistical method for association testing using multi-SNP genotype data from case-parent triads. It uses the genotypes directly, obviating the need for either a HWE assumption or haplotype phase inference. TRIMM identifies transmission distortion for sets of SNPs by comparing the genotypes of the offspring with those of the hypothetical “complement” child who would have inherited the two parental alleles not transmitted to the observed affected offspring at each locus. The LD structure is preserved in this hypothetical offspring.
The difference between the two offspring genotype vectors has an expected value of zero at each locus, under the null hypothesis of no linkage or no association with the disease locus under study. Statistical significance of the test is evaluated by randomly permuting the case-versus-complement labels for the pair of offspring from each family, i.e. randomizing the sign of the difference vector. In our analysis, we used the sum_logP test, which is constructed to optimize over scenarios where risk depends on a single studied SNP and also over scenarios where risk instead depends on a multi-SNP susceptibility haplotype. TRIMM can make full use of individuals with sporadic missing SNP data and has no limitation on the number of SNPs it can handle. All genotyped SNPs in a gene were thus used in the analyses.
HAPLIN (version 2.5) infers haplotypes from the unphased family genotype data when possible, using the Expectation Maximization (EM) algorithm. It estimates the disease relative risk for haplotypes that have a frequency high enough to be detected. The overall significance of the haplotypes is evaluated by a likelihood ratio test, or alternatively, a score test. HAPLIN can distinguish between single and double dose effects of haplotypes, but to reduce the number of parameters needed to be estimated in a multi-SNP scan, haplotype effects were assumed multiplicative in our analyses. Also, haplotypes with population frequency less than 1% were removed from the analyses.
A typical observation when constructing haplotypes with more than four SNPs is that the population LD structure dictates a large number of rare haplotypes, complicating the estimation. In our data, many of the genes with more than four SNPs displayed low LD between the SNPs at the outer ends. Using window lengths of more than four SNPs would thus generate many rare and perhaps irrelevant haplotypes. For the I-CL/P cleft category, we restricted HAPLIN to use four SNPs at a time with a sliding, overlapping window approach to cover all SNPs in a gene. Because considerably fewer cases were available for the analysis of I-CP, we used only three SNPs in the sliding-window for this cleft category. Within a gene, this produces a score test p-value for each window, and the smallest p-value was chosen. To adjust this p-value for within-gene multiple testing, the principle of ‘seemingly unrelated estimation’  was used. The individual (family) score contributions for each window were saved and their between-window correlations were used to correct the p-value.
We plotted the p-values from TRIMM and HAPLIN analyses using a Schweder-Spjøtvoll plot , which is a simple graphical procedure for the simultaneous evaluations of many tests. If none of the genes are truly associated with facial clefts, the p-values would fall along the straight sloping line. Conversely, the p-values would deviate from this line in the presence of genes that are truly associated with the disorder. Statistical analyses were carried out on the Norwegian and Danish samples separately, and p-values from these two sources were combined using Fisher's method . Fisher-combined p-values can be further corrected for multiple testing using the Bonferroni correction.
To test for within-gene differences in genotype frequency across the two national samples, we used all the case-parents in each triad. The test is based on computing the chi-squared value for each SNP in a given gene, followed by aggregating these values to obtain the within-gene sum chi-squared value. These were then converted to p-values using a permutation test with 1000 permutations (implying that the smallest p-value estimable is 0.001). The permutations permute the nationality label of each individual over all SNPs in a gene simultaneously, thus compensating for LD.
All analyses by TRIMM and HAPLIN were restricted to isolated clefts only, and were performed separately for the two major categories of clefts—I-CL/P and I-CP. Software for implementing TRIMM and HAPLIN are available for the R  computing environment from our web sites (TRIMM: http://www.niehs.nih.gov/research/atniehs/labs/bb/staff/weinberg/index.cfm#downloads; HAPLIN: http://www.uib.no/smis/gjessing/genetics/software/haplin).
Approval for this study was obtained from the Norwegian Data Inspectorate, the Regional Committee on Research Ethics for Western Norway, and the respective Institutional Review Boards of the US National Institute of Environmental Health Sciences (NIH/NIEHS) and the University of Iowa. For the Danish facial clefts study, approval was obtained from The Danish National Committee on Biomedical Research Ethics. Clinicopathological information from all participating families and biologic specimens for DNA extraction were obtained with the written informed consent of the mothers and fathers, and all aspects of this research were in compliance with the tenets of the Declaration of Helsinki for human research (http://www.wma.net).
This study was designed with the specific aim of identifying consistent genetic associations across two population-based cleft studies in Scandinavia. We genotyped 1536 SNPs in 357 candidate genes that were selected a priori for their potential roles in clefting. We observed a high duplicate reproducibility rate (99.98%), high Mendelian consistency rate (99.93%), and a low rate of missing genotypes (0.55%) in the data. After removing SNPs with more than 10 Mendelian errors and those with significant deviations from HWE (p<0.05), the final number of SNPs analyzed was 1315, representing a total of 334 candidate genes on autosomal chromosomes (Tables S1 and S2).
Tables S3, S4, S5, S6 summarize the results of the TRIMM and HAPLIN analyses by cleft type (I-CL/P vs. I-CP) and study population (Norway vs. Denmark). The corresponding Schweder-Spjøtvoll plots of p-values are provided in Figures 1 and 2. For I-CL/P, HAPLIN and TRIMM identified strong associations with IRF6 and ADH1C [MIM 103730] in both the Norwegian and Danish sample. In addition to these, HAPLIN identified FGF12 [MIM 605802] in both populations. For I-CP, HAPLIN identified strong associations with PDGFC and ETV5 [MIM 601600] in both populations. While TRIMM confirmed the association with PDGFC, it also identified additional associations with ALX3 [MIM 606014] and MKX (MIM entry not available for this gene). The quantile-quantile (QQ) plots of Fisher-combined p-values from the TRIMM and HAPLIN analyses in the two samples are shown in Figure 3. Compared with the QQ plot for I-CP, the QQ plot for I-CL/P displays an excess of small p-values.
Schweder-Spjøtvoll plot of p-values for (A) isolated cleft lip with or without cleft palate (I-CL/P) and (B) isolated cleft palate (I-CP). All genes with p-values ≤0.1 are shown on the X-axis and ordered according to observed p-values (Y-axis). Genes with p-values ≤0.05 are highlighted in red. The sloping line represents the expected uniform distribution under the null (of no effect). Genes with p-values ≤0.1 in both the Norwegian and Danish samples are indicated by vertical lines connecting the upper (Norway) and lower (Denmark) plots.
Schweder-Spjøtvoll plot of p-values for (A) isolated cleft lip with or without cleft palate (I-CL/P) and (B) isolated cleft palate (I-CP). All genes with p-values ≤0.1 are shown on the X-axis and ordered according to observed p-values (Y-axis). Genes with p-values ≤0.05 are highlighted in red. The sloping line represents the expected uniform distribution under the null (of no effect). Genes that had p-values ≤0.1 in both the Norwegian and Danish samples are indicated by vertical lines connecting the upper (Norway) and lower (Denmark) plots.
The QQ plot compares the distribution of the observed Fisher-combined p-values (−log scale) for both populations with an expected uniform distribution under the null (sloping line). The plots for I-CL/P and I-CP are provided separately for TRIMM in (A) and (B), respectively, and the corresponding plots for HAPLIN are shown in (C) and (D). Gene labels for the top six most significant genes are displayed in each plot.
Finally, Figure 4 provides a graphical summary of the main results, diagrams the genomic location of the seven genes showing associations, and details the type and position of the SNPs within each gene.
Fisher-combined p-values are shown for all genes with p-values ≤0.05 in both the Norwegian and Danish populations for either analysis. The distance between genes on the same chromosome is indicated in megabases (Mb). Gene structure is shown in red for coding regions and blue for untranslated regions, with exon boxes connected by intron lines. The seven identified genes are on the minus strand compared to the reference sequence and are scaled relative to one another based on the length shown in kilobases (kb). Each SNP is represented by a symbol for its functional status as indicated at the bottom of the figure. Two SNPs represented by a square with a star were considered in the region of IRF6, but actually fell within the neighboring gene TRAF3IP3.
Of the 334 autosomal cleft candidate genes analyzed in this study, associations with seven genes—IRF6, PDGFC, ADH1C, MKX, ALX3, FGF12 and ETV5—were replicated in the population-based samples from Norway and Denmark. As expected with this large number of statistical tests, there were many more associations with p-values <0.05 in each population. Of 334 candidate genes, 17 false positives (p<0.05) in each cleft category in each country would be expected by chance alone, and fewer than one gene would show replication across the two countries. As shown in Figures 1 and 2, significant associations were found with 27 genes for I-CL/P and 21 genes for I-CP in the Norwegian and Danish sample respectively using TRIMM and/or HAPLIN. Since all candidate genes in this study were chosen a priori for their potential roles in facial clefting, it is important to separate true associations from spurious ones. We will therefore focus this discussion on genetic associations that were replicated in both populations and that produced the smallest p-values.
TRIMM and HAPLIN analyses
I-CL/P—IRF6, ADH1C and FGF12.
We initially hypothesized similar genetic contributions to clefting in both populations because of the shared ancestry of the Norwegian and Danish populations. The genes for interferon regulatory factor 6 (IRF6), alcohol dehydrogenase 1C (ADH1C), and fibroblast growth factor 12 (FGF12) were strongly associated with I-CL/P in both populations using TRIMM and/or HAPLIN. An even larger number of genes were significantly associated in one population but not the other. Many are no doubt merely chance associations, although a few may reflect true risk genes with associations too weak to emerge in both Danish and Norwegian samples.
The role of IRF6 in facial clefting is now well established , , , , , , , , and our findings provide proof-of-principle that this study design can reliably detect gene/SNP combinations where the relative risks are relatively low. ADH1C is the third gene of the class I alcohol dehydrogenase family consisting of ADH1A, ADH1B, and ADH1C. These genes are tandemly organized as a gene-cluster on chromosome 4q21–q23 and encode the alpha, beta and gamma subunits responsible for most of the ethanol-oxidizing capacity in the liver . The gamma 2 enzyme was previously reported to oxidize ethanol to acetaldehyde at a lower rate than the major gamma 1 variant . Furthermore, the gamma 2 allele also appears to have a protective effect on the risk of facial clefts .
Alcohol consumption during the first trimester of pregnancy is a recognized risk factor for facial clefts . Our recent analysis of maternal binge drinking in the same Norwegian sample suggested that women who reported binge drinking were more likely to have an infant with either CL/P or CP . However, our preliminary analyses show no evidence of interaction between variants in ADH1C and maternal alcohol consumption (results not shown). Previous support for a role of ADH1C in facial clefting stems from an exploratory study from the US where 64 candidate genes were investigated in a combined sample of 58 I-CL/P and I-CP case-parent triads . Despite the small number of triads analyzed in that study, the strong association with ADH1C persisted even after correcting for multiple comparisons.
FGF12 was significantly associated with I-CL/P in both samples in our analyses. Although numerous studies have investigated several members of the FGF and FGFR family in clefting, there are no previously published association studies with FGF12 specifically. The FGF signaling pathway is known to play a crucial role in craniofacial development , and when dysregulated causes craniosynostosis and other facio-skeletal malformations in humans . Using a combination of direct sequencing, association studies and in silico protein modeling, our group recently showed that the FGF signaling pathway may contribute to as much as 3–5% of I-CL/P .
Our data also suggest population-specific differences in susceptibility to facial clefts. For example, both HAPLIN and TRIMM identified FOXE1 as being significantly associated with I-CL/P in the Danish sample, but a significant association did not emerge in the larger Norwegian sample. Conversely, both methods identified MSX1 among the genes most significantly associated with I-CL/P in the Norwegian sample, whereas no such association was seen in the Danish sample (Figures 1 and 2). Although these findings do not contradict our previously reported associations with these genes , , , this population-specific susceptibility nevertheless highlights the problem with lack of replication across diverse populations , .
Among other genes significantly associated with facial clefts in one population but not the other, zinc finger homeobox 1b (ZFHX1B [MIM 605802]; a.k.a. ZEB2) is of particular interest. Mutations in this gene underlie the autosomal dominant Mowat-Wilson syndrome (MWS [MIM 235730]), characterized by severe mental retardation and multiple congenital anomalies . Among the wide range of clinical features associated with MWS, a small subset of patients also present with facial clefts . In the mouse model, cranial neural crest cells fail to dissociate and migrate when Zfhx1b is knocked out . These cells are early progenitors of craniofacial cartilage, bone, and facial connective tissue. It is thus plausible that a failure of the cells to delaminate could underlie the distinctive facies and occasional clefting seen in MWS patients.
The paralog Zfhx1a (a.k.a. Zeb1) when knocked out in mice causes a cleft palate phenotype with 100% penetrance , . The defect appears to be largely due to a delay in palatal elevation. In addition, the Zfhx1a null embryos also show reduced expression of Jag2, Tgfb3 and Mmp13—genes known to be involved in palatogenesis. There also appears to be complex synergistic interactions between Zfhx1a and Zfhx1b during mouse embryogenesis . Notably, the double Zfhx1a/Zfhx1b mutants have midline defects at the fusion sites of the nasal and maxillary processes.
I-CP—ALX3, MKX, PDGFC and ETV5.
TRIMM and HAPLIN identified four genes strongly associated with I-CP in both populations. Aristaless (ALX)-related genes are a subset of the Paired-related homeobox genes involved in the development of craniofacial structures , . ALX3 encodes a nuclear protein containing a homeobox DNA-binding domain that functions as a transcriptional regulator in embryonic development. In mice, expression of Alx3 is highly localized and the characteristic expression pattern suggests an important role in patterning of neural-crest derived mesenchyme and shaping of craniofacial structures . Compound mutants of Alx3 and Alx4 exhibit severe craniofacial abnormalities that are otherwise absent in Alx4 single mutants . In addition, Alx3/Alx4 double mutant newborn mice have cleft nasal regions, most likely as a result of a failure of the medial nasal processes to fuse in the facial midline .
The mohawk homeobox (MKX) gene product is another important regulator of vertebrate development . In mice, Mkx is expressed in somite-derived cell lineages that give rise to skeletal muscle, among other cell types . Mkx was formerly described as Irxl1 (iroquois homeobox protein-like 1) in Liu et al. . Localized in the middle of the 4.3 Mb critical region in the spontaneous Twirler mouse mutant, Mkx is a likely candidate for the Twirler (Tw) gene. Mice heterozygous for the Twirler mutation have inner ear defects, whereas those homozygous have cleft lip and cleft palate. Mkx is also highly expressed in the palatal mesenchyme during palatal growth and fusion . Interestingly, both MKX and ZFHX1A (a paralog of ZFHX1B described above) are located within the 10p15-11 region which is associated with cleft lip and cleft palate when duplicated , but not when deleted . Finally, Blanton and co-workers have reported strong associations between markers from the 10p13 region and I-CL/P .
Compared with ALX and MKX, the gene for platelet-derived growth factor C (PDGFC) has a well-substantiated role in palatogenesis. Linkage, association and cytogenetic studies have supported the existence of a CL/P locus in the chromosome 4q31-ter region containing the PDGFC locus , , , . More recently, Choi and colleagues showed that a SNP in the PDGFC regulatory region decreased promoter transcriptional activity and was associated with CL/P in multiple populations . Animal studies have also shown a specific role for PDGFC in palatal development. For example, Pdgfc−/− mice die in the perinatal period, presumably due to feeding and respiratory difficulties from having a complete cleft of the secondary palate . In these mice, the palatal bones fail to extend across the roof of the oral-nasal cavity, suggesting that hypoplasia of palate tissues combined with fusion defects of the medial edge epithelia may contribute to the cleft palate seen in the Pdgfc−/− mice . In vitro PDGFC, a potent stimulator of mitosis, is downregulated by retinoic acid in mouse embryonic palatal mesenchymal cells . Reduced PDGFC by retinoic acid may cause inhibition of proliferation in palatal shelves, resulting in the pathogenesis of cleft palate in Pdgfc−/− mice or retinoic acid-treated mice. Consistent with the absence of cleft lip in Pdgfc−/− mice, the association with PDGFC was found exclusively in the I-CP group in our data.
Lastly, the gene encoding transcription factor ETV5 (ets variant 5) is located on chromosome 3q28 (as FGF12 above). In the E13.5 mouse palatal epithelium, expression of Etv5 is strong on the oral surface of the palatal shelf and overlaps to some extent with that of the FGF signaling antagonist Spry2 . Mice carrying a deletion that disrupts Spry2 have cleft palate, and point mutations in SPRY2 may be rare causes of I-CL/P . The link between ETV5, SPRY2 and members of the FGF family offers an exciting opportunity to further explore these interacting developmental pathways in the pathogenesis of cleft palate.
Both TRIMM and HAPLIN correct for within-gene multiple testing, ensuring the validity of the individual p-values for each gene. When assessing the overall results, no single gene remained significant after a full Bonferroni correction. However, it is generally agreed that the Bonferroni requirement of ensuring an overall type 1 error rate of below 5% is too strict and may result in too many false negatives , particularly in this study where the selected genes already had an a priori connection to clefting. As an alternative, we focused here on genes showing significance in both populations. As proof-of-principle, our approach consistently replicated the previously known strong associations with IRF6 and PDGFC, while proposing several new candidate genes including ETV5, FGF12, MKX, ADH1C and ALX3 for further analyses.
Quantile-quantile (QQ) plot.
A QQ plot compares the observed p-values with the expected uniform distribution under the null when multiple tests are performed. If the plot reveals a marked difference between the expected and observed p-value distribution, this may indicate that a number of genes are significantly associated with disease, although individual genes may not have a marginal effect that is strong enough to withstand Bonferroni correction. We used the plot to test for within-gene differences in SNP genotype distribution across the two national samples, as well as to evaluate the results of the genetic association tests by TRIMM and HAPLIN.
Our data revealed systematic differences in genotypes for a subset of the SNPs across the Norwegian and Danish samples. These differences cannot be simply ascribed to random genotyping errors, as these had already been filtered out by our initial HWE check and data-cleaning steps. Biases in genotype call rates may arise if DNA comes from different sources (e.g. blood versus buccal swab). However, the genotype call rates were 99.6% for DNA extracted from blood (Norwegian sample) and 99.1% for DNA extracted from buccal swabs (Danish sample). Even if the marked heterogeneity is a true indication of population differences, different genotype frequencies do not imply differences in gene-effects on phenotype. Moreover, our methods, based on analysis of case-parent triads, insulate against population-based differences of this kind.
QQ plots of the combined p-values from the TRIMM and HAPLIN analyses for I-CL/P and I-CP are depicted in Figure 3. Of the two cleft categories, the QQ plot for I-CP showed the least evidence of association with the selected SNPs, yet this is the cleft subtype with the stronger recurrence risk in families . If there are real but weak genetic effects, these would still be expected to deviate from the uniform distribution. The lack of associated genes in I-CP may be a direct consequence of insufficient SNP coverage rather than a lack of statistical power. Also, there may be important gene-gene/gene-environment interactions that were not considered in the current analyses.
Our selection of htSNPs was based primarily on genotype information publicly available from the International HapMap Consortium. Although the use of haplotypes may guard against the loss of power in detecting an association signal, inadequate SNP resolution might still be a concern in this study, as pointed out for the lack of associated genes in I-CP. Adequate representation of the genetic architecture of each candidate gene is feasible only if htSNPs are selected from complete re-sequencing of the genes in the two populations examined. Furthermore, the effect of SNP resolution on inferred haplotype block structure is not fully understood. Studies using lower marker density identify progressively longer blocks, and vice-versa . The small set of reference samples (30 CEPH trios) of [Western] European descent in HapMap may not be sufficiently representative of the Danish or Norwegian population, especially if common alleles are overrepresented and rare alleles more strongly linked to facial clefting are underrepresented. Hence, blocks detected at low marker density may not adequately reflect the true genetic diversity in our data.
HAPLIN and TRIMM.
These two analytic methods are both designed to detect multi-marker transmission distortion, but they accomplish this in different ways. TRIMM is non-parametric and can accommodate population structure, deviation from HWE, multiple SNPs, missing SNPs and non-negligible recombination rates. TRIMM does not attempt to infer haplotypes and, consequently, is computationally efficient. TRIMM performs best when there is one important risk haplotype or one SNP associated with risk. When applied to a set of SNPs within a gene, TRIMM accounts for within-gene SNP correlations by permuting alleles at all SNPs simultaneously. HAPLIN on the other hand is better suited to handle more complicated scenarios, for example, when there is more than one risk-associated haplotype. Being parametric, HAPLIN estimates the full haplotype distribution over a set of SNPs and also estimates relative risks associated with each haplotype. Through the use of a full maximum likelihood model, it produces a complete description of the “risk structure” over the set of haplotypes in a region. HAPLIN requires HWE, assumes no recombination, and is computationally more demanding.
Of particular relevance to our analyses here is that if one of the genotyped SNPs has a direct effect on disease risk, independently of the other SNPs, TRIMM would be most likely to detect it. Alternatively, if the specific risk locus has not been genotyped directly, it is still likely to be associated with a haplotype composed of SNPs measured in the surrounding region. Since HAPLIN scans window-lengths of 3 and 4 SNPs, it will typically surround such a locus with 1–2 SNPs on either side, provided the disease locus resides within the genotyped region. In such cases, the sliding-window approach of HAPLIN should have a better chance of detecting the true association.
Replication of findings.
Replication has a central role in the evaluation of association findings, particularly in studies involving large numbers of statistical tests . To help separate true associations from spurious ones, we looked for associations that were consistent across the two populations of shared ancestry, which are more likely to represent true associations as opposed to spurious associations that show up randomly—even in populations of similar ancestral background. We opted not to combine the Norwegian and Danish samples (although this would have boosted power) because we had no prior data to assume that the same haplotype(s) contribute to the risk of clefts in both populations. Although our data do not speak to that issue, our test for heterogeneity between these two populations showed marked and systematic differences in genotypes for a subset of the SNPs (data not shown), suggesting ancestral differences between the two populations.
In this search among 357 candidate genes for facial clefting, we found a relatively small number of genes to be significantly associated with either I-CL/P or I-CP. The modest number of genes associated with this strongly genetic birth defect has several possible interpretations. It may suggest the presence of etiological variants not genotyped in the current study (copy number variants for example), or cleft candidate genes overlooked in our gene selection. It may also reflect the effects of non-additive interactions with other genetic variants and/or with environmental factors that have been and will be explored in other studies . Furthermore, it is possible that the remaining genetic risks arise in the context of allelic heterogeneity, in which case they would not be detectable by the current LD-based association approach. We have analyzed only individual genes; epistasis between these may play an important role as well. There is also the possibility that, regardless of apparent similarities between the Norwegian and Danish populations, these two populations may have inherently distinct genetic susceptibilities to facial clefting.
The use of two powerful and complementary statistical methods for haplotype analysis of genetic data from offspring-parent triads, two well-defined cleft phenotypes, two national cleft cohorts of similar ancestry, and the use of a very large number of SNPs in many genes (10 times the number of SNPs previously used to look for cleft associations), provided a powerful opportunity to look for associations that are most likely to be genuine. Our approach effectively replicated the strongest previously known association with cleft lip—the one with IRF6—in both populations and by both methods, and suggested new associations with ADH1C and PDGFC. Finally, the novel genes detected here (MKX, ALX3, ETV5, and FGF12) provide new insights for future analyses of this common and complex birth defect.
Overview of the 357 candidate genes selected for facial clefts.
(0.77 MB DOC)
We are grateful to the families for their participation in this study. The following individuals provided valuable input in candidate gene selection and sample preparation: Kathy Frees, Ecaterina Dragan, Adela Mansilla, Bridget Riley, Brian Schutte, Edward Lammer, Temis Felix, Michael Lovett, Satoshi Suzuki, Andrew Olshan, Rulang Jiang, Diana Caprau, Richard Finnell, Christine Jewell, John Cidlowski, Fedik Rahimov, Jack Taylor and Alexandre Vieira. We are grateful to Susie McConnell for outstanding secretarial support, Dorthe Grosen for help with the Danish samples, and Dr. Dana Hancock for her comments on an earlier draft of this manuscript. Genotyping services were provided by the Center for Inherited Disease Research (CIDR) and we thank Ivy McMullen, Corinne Boehm, Kim Doheny, and other CIDR staff involved in this project.
Conceived and designed the experiments: AJ MS RTL AW CRW KC ALB JCM. Performed the experiments: AJ MS SDH. Analyzed the data: AJ MS HG CRW ALB JCM. Contributed reagents/materials/analysis tools: AJ MS HG RTL AW TNT CB JCM. Wrote the paper: AJ MS HG RTL AW CRW KC ALB CB ACL JCM.
- 1. Mossey PA, Little J (2002) Epidemiology of oral clefts: an international perspective. In: Wyszynski DFE, editor. Cleft lip and palate: from origin to treatment. New York: Oxford University Press. PA MosseyJ. Little2002Epidemiology of oral clefts: an international perspective.DFE WyszynskiCleft lip and palate: from origin to treatmentNew YorkOxford University Press
- 2. Lidral AC, Moreno LM (2005) Progress toward discerning the genetics of cleft lip. Curr Opin Pediatr 17: 731–739.AC LidralLM Moreno2005Progress toward discerning the genetics of cleft lip.Curr Opin Pediatr17731739
- 3. Lie RT, Wilcox AJ, Skjaerven R (1994) A population-based study of the risk of recurrence of birth defects. N Engl J Med 331: 1–4.RT LieAJ WilcoxR. Skjaerven1994A population-based study of the risk of recurrence of birth defects.N Engl J Med33114
- 4. Jugessur A, Murray JC (2005) Orofacial clefting: recent insights into a complex trait. Curr Opin Genet Dev 15: 270–278.A. JugessurJC Murray2005Orofacial clefting: recent insights into a complex trait.Curr Opin Genet Dev15270278
- 5. Ardinger HH, Buetow KH, Bell GI, Bardach J, VanDemark DR, et al. (1989) Association of genetic variation of the transforming growth factor- alpha gene with cleft lip and palate. Am J Hum Genet 45: 348–353.HH ArdingerKH BuetowGI BellJ. BardachDR VanDemark1989Association of genetic variation of the transforming growth factor- alpha gene with cleft lip and palate.Am J Hum Genet45348353
- 6. Ding H, Wu X, Bostrom H, Kim I, Wong N, et al. (2004) A specific requirement for PDGF-C in palate formation and PDGFR-alpha signaling. Nat Genet 36: 1111–1116.H. DingX. WuH. BostromI. KimN. Wong2004A specific requirement for PDGF-C in palate formation and PDGFR-alpha signaling.Nat Genet3611111116
- 7. Chiquet BT, Lidral AC, Stal S, Mulliken JB, Moreno LM, et al. (2007) CRISPLD2: a novel NSCLP candidate gene. Hum Mol Genet 16: 2241–2248.BT ChiquetAC LidralS. StalJB MullikenLM Moreno2007CRISPLD2: a novel NSCLP candidate gene.Hum Mol Genet1622412248
- 8. Marazita ML, Murray JC, Lidral AC, Arcos-Burgos M, Cooper ME, et al. (2004) Meta-analysis of 13 genome scans reveals multiple cleft lip/palate genes with novel loci on 9q21 and 2q32-35. Am J Hum Genet 75: 161–173.ML MarazitaJC MurrayAC LidralM. Arcos-BurgosME Cooper2004Meta-analysis of 13 genome scans reveals multiple cleft lip/palate genes with novel loci on 9q21 and 2q32-35.Am J Hum Genet75161173
- 9. Vieira AR, Avila JR, Daack-Hirsch S, Dragan E, Felix TM, et al. (2005) Medical sequencing of candidate genes for nonsyndromic cleft lip and palate. PLoS Genet 1: e64.AR VieiraJR AvilaS. Daack-HirschE. DraganTM Felix2005Medical sequencing of candidate genes for nonsyndromic cleft lip and palate.PLoS Genet1e64
- 10. Lidral AC, Murray JC (2004) Genetic approaches to identify disease genes for birth defects with cleft lip/palate as a model. Birth Defects Res A Clin Mol Teratol 70: 893–901.AC LidralJC Murray2004Genetic approaches to identify disease genes for birth defects with cleft lip/palate as a model.Birth Defects Res A Clin Mol Teratol70893901
- 11. Riley BM, Mansilla MA, Ma J, Daack-Hirsch S, Maher BS, et al. (2007) Impaired FGF signaling contributes to cleft lip and palate. Proc Natl Acad Sci U S A 104: 4512–4517.BM RileyMA MansillaJ. MaS. Daack-HirschBS Maher2007Impaired FGF signaling contributes to cleft lip and palate.Proc Natl Acad Sci U S A10445124517
- 12. Riley BM, Murray JC (2007) Sequence evaluation of FGF and FGFR gene conserved non-coding elements in non-syndromic cleft lip and palate cases. Am J Med Genet A 143: 3228–3234.BM RileyJC Murray2007Sequence evaluation of FGF and FGFR gene conserved non-coding elements in non-syndromic cleft lip and palate cases.Am J Med Genet A14332283234
- 13. Risch N, Merikangas K (1996) The future of genetic studies of complex human diseases. Science 273: 1516–1517.N. RischK. Merikangas1996The future of genetic studies of complex human diseases.Science27315161517
- 14. Marazita ML, Neiswanger K (2002) Association studies. In: Wyszynski DFE, editor. Cleft lip and palate: from origin to treatment. New York: Oxford University Press. ML MarazitaK. Neiswanger2002Association studies.DFE WyszynskiCleft lip and palate: from origin to treatmentNew YorkOxford University Press
- 15. Shi M, Umbach DM, Weinberg CR (2007) Identification of Risk-Related Haplotypes with the Use of Multiple SNPs from Nuclear Families. Am J Hum Genet 81: 53–66.M. ShiDM UmbachCR Weinberg2007Identification of Risk-Related Haplotypes with the Use of Multiple SNPs from Nuclear Families.Am J Hum Genet815366
- 16. de Bakker PI, Yelensky R, Pe'er I, Gabriel SB, Daly MJ, et al. (2005) Efficiency and power in genetic association studies. Nat Genet 37: 1217–1223.PI de BakkerR. YelenskyI. Pe'erSB GabrielMJ Daly2005Efficiency and power in genetic association studies.Nat Genet3712171223
- 17. Gjessing HK, Lie RT (2006) Case-parent triads: estimating single- and double-dose effects of fetal and maternal disease gene haplotypes. Annals of human genetics 1–15.HK GjessingRT Lie2006Case-parent triads: estimating single- and double-dose effects of fetal and maternal disease gene haplotypes.Annals of human genetics115
- 18. Wilcox AJ, Lie RT, Solvoll K, Taylor J, McConnaughey DR, et al. (2007) Folic acid supplements and risk of facial clefts: national population based case-control study. Bmj 334: 464.AJ WilcoxRT LieK. SolvollJ. TaylorDR McConnaughey2007Folic acid supplements and risk of facial clefts: national population based case-control study.Bmj334464
- 19. Bille C, Knudsen LB, Christensen K (2005) Changing lifestyles and oral clefts occurrence in Denmark. Cleft Palate Craniofac J 42: 255–259.C. BilleLB KnudsenK. Christensen2005Changing lifestyles and oral clefts occurrence in Denmark.Cleft Palate Craniofac J42255259
- 20. Brown NL, Knott L, Halligan E, Yarram SJ, Mansell JP, et al. (2003) Microarray analysis of murine palatogenesis: temporal expression of genes during normal palate development. Dev Growth Differ 45: 153–165.NL BrownL. KnottE. HalliganSJ YarramJP Mansell2003Microarray analysis of murine palatogenesis: temporal expression of genes during normal palate development.Dev Growth Differ45153165
- 21. Cai J, Ash D, Kotch LE, Jabs EW, Attie-Bitach T, et al. (2005) Gene expression in pharyngeal arch 1 during human embryonic development. Hum Mol Genet 14: 903–912.J. CaiD. AshLE KotchEW JabsT. Attie-Bitach2005Gene expression in pharyngeal arch 1 during human embryonic development.Hum Mol Genet14903912
- 22. Gong SG, Gong TW, Shum L (2005) Identification of markers of the midface. J Dent Res 84: 69–72.SG GongTW GongL. Shum2005Identification of markers of the midface.J Dent Res846972
- 23. Mukhopadhyay P, Greene RM, Zacharias W, Weinrich MC, Singh S, et al. (2004) Developmental gene expression profiling of mammalian, fetal orofacial tissue. Birth Defects Res A Clin Mol Teratol 70: 912–926.P. MukhopadhyayRM GreeneW. ZachariasMC WeinrichS. Singh2004Developmental gene expression profiling of mammalian, fetal orofacial tissue.Birth Defects Res A Clin Mol Teratol70912926
- 24. Diewert VM (1985) Development of human craniofacial morphology during the late embryonic and early fetal periods. Am J Orthod 88: 64–76.VM Diewert1985Development of human craniofacial morphology during the late embryonic and early fetal periods.Am J Orthod886476
- 25. Beier DR, Herron BJ (2004) Genetic mapping and ENU mutagenesis. Genetica 122: 65–69.DR BeierBJ Herron2004Genetic mapping and ENU mutagenesis.Genetica1226569
- 26. Higgins AW, Alkuraya FS, Bosco AF, Brown KK, Bruns GA, et al. (2008) Characterization of apparently balanced chromosomal rearrangements from the developmental genome anatomy project. Am J Hum Genet 82: 712–722.AW HigginsFS AlkurayaAF BoscoKK BrownGA Bruns2008Characterization of apparently balanced chromosomal rearrangements from the developmental genome anatomy project.Am J Hum Genet82712722
- 27. Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21: 263–265.JC BarrettB. FryJ. MallerMJ Daly2005Haploview: analysis and visualization of LD and haplotype maps.Bioinformatics21263265
- 28. Sebastiani P, Lazarus R, Weiss ST, Kunkel LM, Kohane IS, et al. (2003) Minimal haplotype tagging. Proc Natl Acad Sci U S A 100: 9900–9905.P. SebastianiR. LazarusST WeissLM KunkelIS Kohane2003Minimal haplotype tagging.Proc Natl Acad Sci U S A10099009905
- 29. Lee WC (2003) Detecting population stratification using a panel of single nucleotide polymorphisms. Int J Epidemiol 32: 1120.WC Lee2003Detecting population stratification using a panel of single nucleotide polymorphisms.Int J Epidemiol321120
- 30. Nielsen DM, Ehm MG, Weir BS (1998) Detecting marker-disease association by testing for Hardy-Weinberg disequilibrium at a marker locus. Am J Hum Genet 63: 1531–1540.DM NielsenMG EhmBS Weir1998Detecting marker-disease association by testing for Hardy-Weinberg disequilibrium at a marker locus.Am J Hum Genet6315311540
- 31. Wittke-Thompson JK, Pluzhnikov A, Cox NJ (2005) Rational inferences about departures from Hardy-Weinberg equilibrium. Am J Hum Genet 76: 967–986.JK Wittke-ThompsonA. PluzhnikovNJ Cox2005Rational inferences about departures from Hardy-Weinberg equilibrium.Am J Hum Genet76967986
- 32. O'Connell JR, Weeks DE (1998) PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 63: 259–266.JR O'ConnellDE Weeks1998PedCheck: a program for identification of genotype incompatibilities in linkage analysis.Am J Hum Genet63259266
- 33. StataCorp (2003) Stata Statistical Software: Release 8.0. Texas, USA: Stata Corporation. StataCorp2003Stata Statistical Software: Release 8.0Texas, USAStata Corporation
- 34. Schweder T, Spjøtvoll E (1982) Plots of P-values to evaluate many tests simultaneously. Biometrika 69: 493–502.T. SchwederE. Spjøtvoll1982Plots of P-values to evaluate many tests simultaneously.Biometrika69493502
- 35. Fisher R (1958) Statistical Methods for Research Workers. New York: Hafner. R. Fisher1958Statistical Methods for Research WorkersNew YorkHafner
- 36. R Development Core Team (2006) R: A Language and Environment for Statistical ComputingR Development Core Team2006R: A Language and Environment for Statistical Computing
- 37. Ben J, Jabs EW, Chong SS (2005) Genomic, cDNA and embryonic expression analysis of zebrafish IRF6, the gene mutated in the human oral clefting disorders Van der Woude and popliteal pterygium syndromes. Gene Expr Patterns 5: 629–638.J. BenEW JabsSS Chong2005Genomic, cDNA and embryonic expression analysis of zebrafish IRF6, the gene mutated in the human oral clefting disorders Van der Woude and popliteal pterygium syndromes.Gene Expr Patterns5629638
- 38. Ingraham CR, Kinoshita A, Kondo S, Yang B, Sajan S, et al. (2006) Abnormal skin, limb and craniofacial morphogenesis in mice deficient for interferon regulatory factor 6 (Irf6). Nat Genet 38: 1335–1340.CR IngrahamA. KinoshitaS. KondoB. YangS. Sajan2006Abnormal skin, limb and craniofacial morphogenesis in mice deficient for interferon regulatory factor 6 (Irf6).Nat Genet3813351340
- 39. Jugessur A, Rahimov F, Lie RT, Wilcox AJ, Gjessing HK, et al. (2008) Genetic variants in IRF6 and the risk of facial clefts: single-marker and haplotype-based analyses in a population-based case-control study of facial clefts in Norway. Genet Epidemiol 32: 413–424.A. JugessurF. RahimovRT LieAJ WilcoxHK Gjessing2008Genetic variants in IRF6 and the risk of facial clefts: single-marker and haplotype-based analyses in a population-based case-control study of facial clefts in Norway.Genet Epidemiol32413424
- 40. Knight AS, Schutte BC, Jiang R, Dixon MJ (2006) Developmental expression analysis of the mouse and chick orthologues of IRF6: the gene mutated in Van der Woude syndrome. Dev Dyn 235: 1441–1447.AS KnightBC SchutteR. JiangMJ Dixon2006Developmental expression analysis of the mouse and chick orthologues of IRF6: the gene mutated in Van der Woude syndrome.Dev Dyn23514411447
- 41. Kondo S, Schutte BC, Richardson RJ, Bjork BC, Knight AS, et al. (2002) Mutations in IRF6 cause Van der Woude and popliteal pterygium syndromes. Nat Genet 32: 285–289.S. KondoBC SchutteRJ RichardsonBC BjorkAS Knight2002Mutations in IRF6 cause Van der Woude and popliteal pterygium syndromes.Nat Genet32285289
- 42. Rahimov F, Marazita ML, Visel A, Cooper ME, Hitchler MJ, et al. (2008) Disruption of an AP-2α binding site in an IRF6 enhancer is strongly associated with cleft lip. Nat Genet. F. RahimovML MarazitaA. ViselME CooperMJ Hitchler2008Disruption of an AP-2α binding site in an IRF6 enhancer is strongly associated with cleft lip.Nat GenetIn Press. In Press.
- 43. Richardson RJ, Dixon J, Malhotra S, Hardman MJ, Knowles L, et al. (2006) Irf6 is a key determinant of the keratinocyte proliferation-differentiation switch. Nat Genet 38: 1329–1334.RJ RichardsonJ. DixonS. MalhotraMJ HardmanL. Knowles2006Irf6 is a key determinant of the keratinocyte proliferation-differentiation switch.Nat Genet3813291334
- 44. Washbourne BJ, Cox TC (2006) Expression profiles of cIRF6, cLHX6 and cLHX7 in the facial primordia suggest specific roles during primary palatogenesis. BMC Dev Biol 6: 18.BJ WashbourneTC Cox2006Expression profiles of cIRF6, cLHX6 and cLHX7 in the facial primordia suggest specific roles during primary palatogenesis.BMC Dev Biol618
- 45. Edenberg HJ (2007) The genetics of alcohol metabolism: role of alcohol dehydrogenase and aldehyde dehydrogenase variants. Alcohol Res Health 30: 5–13.HJ Edenberg2007The genetics of alcohol metabolism: role of alcohol dehydrogenase and aldehyde dehydrogenase variants.Alcohol Res Health30513
- 46. Hoog JO, Heden LO, Larsson K, Jornvall H, von Bahr-Lindstrom H (1986) The gamma 1 and gamma 2 subunits of human liver alcohol dehydrogenase. cDNA structures, two amino acid replacements, and compatibility with changes in the enzymatic properties. Eur J Biochem 159: 215–218.JO HoogLO HedenK. LarssonH. JornvallH. von Bahr-Lindstrom1986The gamma 1 and gamma 2 subunits of human liver alcohol dehydrogenase. cDNA structures, two amino acid replacements, and compatibility with changes in the enzymatic properties.Eur J Biochem159215218
- 47. Chevrier C, Perret C, Bahuau M, Nelva A, Herman C, et al. (2005) Interaction between the ADH1C polymorphism and maternal alcohol intake in the risk of nonsyndromic oral clefts: an evaluation of the contribution of child and maternal genotypes. Birth Defects Res A Clin Mol Teratol 73: 114–122.C. ChevrierC. PerretM. BahuauA. NelvaC. Herman2005Interaction between the ADH1C polymorphism and maternal alcohol intake in the risk of nonsyndromic oral clefts: an evaluation of the contribution of child and maternal genotypes.Birth Defects Res A Clin Mol Teratol73114122
- 48. Deroo LA, Wilcox AJ, Drevon CA, Lie RT (2008) First-Trimester Maternal Alcohol Consumption and the Risk of Infant Oral Clefts in Norway: A Population-based Case-Control Study. Am J Epidemiol. LA DerooAJ WilcoxCA DrevonRT Lie2008First-Trimester Maternal Alcohol Consumption and the Risk of Infant Oral Clefts in Norway: A Population-based Case-Control Study.Am J Epidemiol
- 49. Park JW, Cai J, McIntosh I, Jabs EW, Fallin MD, et al. (2006) High throughput SNP and expression analyses of candidate genes for non-syndromic oral clefts. J Med Genet 43: 598–608.JW ParkJ. CaiI. McIntoshEW JabsMD Fallin2006High throughput SNP and expression analyses of candidate genes for non-syndromic oral clefts.J Med Genet43598608
- 50. Nie X, Luukko K, Kettunen P (2006) FGF signalling in craniofacial development and developmental disorders. Oral Dis 12: 102–111.X. NieK. LuukkoP. Kettunen2006FGF signalling in craniofacial development and developmental disorders.Oral Dis12102111
- 51. Pauws E, Stanier P (2007) FGF signalling and SUMO modification: new players in the aetiology of cleft lip and/or palate. Trends Genet 23: 631–640.E. PauwsP. Stanier2007FGF signalling and SUMO modification: new players in the aetiology of cleft lip and/or palate.Trends Genet23631640
- 52. Jezewski PA, Vieira AR, Nishimura C, Ludwig B, Johnson M, et al. (2003) Complete sequencing shows a role for MSX1 in non-syndromic cleft lip and palate. J Med Genet 40: 399–407.PA JezewskiAR VieiraC. NishimuraB. LudwigM. Johnson2003Complete sequencing shows a role for MSX1 in non-syndromic cleft lip and palate.J Med Genet40399407
- 53. Marazita ML, Lidral AC, Murray JC, Field LL, Maher BS, et al. (2009) Genome scan, fine-mapping, and candidate gene analysis of non-syndromic cleft lip with or without cleft palate reveals phenotype specific differences in linkage and association results. Human Heredity. ML MarazitaAC LidralJC MurrayLL FieldBS Maher2009Genome scan, fine-mapping, and candidate gene analysis of non-syndromic cleft lip with or without cleft palate reveals phenotype specific differences in linkage and association results.Human HeredityIn press. In press.
- 54. Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG (2001) Replication validity of genetic association studies. Nat Genet 29: 306–309.JP IoannidisEE NtzaniTA TrikalinosDG Contopoulos-Ioannidis2001Replication validity of genetic association studies.Nat Genet29306309
- 55. McCarthy MI, Abecasis GR, Cardon LR, Goldstein DB, Little J, et al. (2008) Genome-wide association studies for complex traits: consensus, uncertainty and challenges. Nat Rev Genet 9: 356–369.MI McCarthyGR AbecasisLR CardonDB GoldsteinJ. Little2008Genome-wide association studies for complex traits: consensus, uncertainty and challenges.Nat Rev Genet9356369
- 56. Mowat DR, Wilson MJ, Goossens M (2003) Mowat-Wilson syndrome. J Med Genet 40: 305–310.DR MowatMJ WilsonM. Goossens2003Mowat-Wilson syndrome.J Med Genet40305310
- 57. Garavelli L, Mainardi PC (2007) Mowat-Wilson syndrome. Orphanet J Rare Dis 2: 42.L. GaravelliPC Mainardi2007Mowat-Wilson syndrome.Orphanet J Rare Dis242
- 58. Van de Putte T, Maruhashi M, Francis A, Nelles L, Kondoh H, et al. (2003) Mice lacking ZFHX1B, the gene that codes for Smad-interacting protein-1, reveal a role for multiple neural crest cell defects in the etiology of Hirschsprung disease-mental retardation syndrome. Am J Hum Genet 72: 465–470.T. Van de PutteM. MaruhashiA. FrancisL. NellesH. Kondoh2003Mice lacking ZFHX1B, the gene that codes for Smad-interacting protein-1, reveal a role for multiple neural crest cell defects in the etiology of Hirschsprung disease-mental retardation syndrome.Am J Hum Genet72465470
- 59. Jin JZ, Li Q, Higashi Y, Darling DS, Ding J (2008) Analysis of Zfhx1a mutant mice reveals palatal shelf contact-independent medial edge epithelial differentiation during palate fusion. Cell Tissue Res 333: 29–38.JZ JinQ. LiY. HigashiDS DarlingJ. Ding2008Analysis of Zfhx1a mutant mice reveals palatal shelf contact-independent medial edge epithelial differentiation during palate fusion.Cell Tissue Res3332938
- 60. Takagi T, Moribe H, Kondoh H, Higashi Y (1998) DeltaEF1, a zinc finger and homeodomain transcription factor, is required for skeleton patterning in multiple lineages. Development 125: 21–31.T. TakagiH. MoribeH. KondohY. Higashi1998DeltaEF1, a zinc finger and homeodomain transcription factor, is required for skeleton patterning in multiple lineages.Development1252131
- 61. Miyoshi T, Maruhashi M, Van De Putte T, Kondoh H, Huylebroeck D, et al. (2006) Complementary expression pattern of Zfhx1 genes Sip1 and deltaEF1 in the mouse embryo and their genetic interaction revealed by compound mutants. Dev Dyn 235: 1941–1952.T. MiyoshiM. MaruhashiT. Van De PutteH. KondohD. Huylebroeck2006Complementary expression pattern of Zfhx1 genes Sip1 and deltaEF1 in the mouse embryo and their genetic interaction revealed by compound mutants.Dev Dyn23519411952
- 62. Beverdam A, Meijlink F (2001) Expression patterns of group-I aristaless-related genes during craniofacial and limb development. Mech Dev 107: 163–167.A. BeverdamF. Meijlink2001Expression patterns of group-I aristaless-related genes during craniofacial and limb development.Mech Dev107163167
- 63. Meijlink F, Beverdam A, Brouwer A, Oosterveen TC, Berge DT (1999) Vertebrate aristaless-related genes. Int J Dev Biol 43: 651–663.F. MeijlinkA. BeverdamA. BrouwerTC OosterveenDT Berge1999Vertebrate aristaless-related genes.Int J Dev Biol43651663
- 64. ten Berge D, Brouwer A, el Bahi S, Guenet JL, Robert B, et al. (1998) Mouse Alx3: an aristaless-like homeobox gene expressed during embryogenesis in ectomesenchyme and lateral plate mesoderm. Dev Biol 199: 11–25.D. ten BergeA. BrouwerS. el BahiJL GuenetB. Robert1998Mouse Alx3: an aristaless-like homeobox gene expressed during embryogenesis in ectomesenchyme and lateral plate mesoderm.Dev Biol1991125
- 65. Beverdam A, Brouwer A, Reijnen M, Korving J, Meijlink F (2001) Severe nasal clefting and abnormal embryonic apoptosis in Alx3/Alx4 double mutant mice. Development 128: 3975–3986.A. BeverdamA. BrouwerM. ReijnenJ. KorvingF. Meijlink2001Severe nasal clefting and abnormal embryonic apoptosis in Alx3/Alx4 double mutant mice.Development12839753986
- 66. Mukherjee K, Burglin TR (2007) Comprehensive analysis of animal TALE homeobox genes: new conserved motifs and cases of accelerated evolution. J Mol Evol 65: 137–153.K. MukherjeeTR Burglin2007Comprehensive analysis of animal TALE homeobox genes: new conserved motifs and cases of accelerated evolution.J Mol Evol65137153
- 67. Anderson DM, Arredondo J, Hahn K, Valente G, Martin JF, et al. (2006) Mohawk is a novel homeobox gene expressed in the developing mouse embryo. Dev Dyn 235: 792–801.DM AndersonJ. ArredondoK. HahnG. ValenteJF Martin2006Mohawk is a novel homeobox gene expressed in the developing mouse embryo.Dev Dyn235792801
- 68. Liu H, Liu W, Maltby KM, Lan Y, Jiang R (2006) Identification and developmental expression analysis of a novel homeobox gene closely linked to the mouse Twirler mutation. Gene Expr Patterns 6: 632–636.H. LiuW. LiuKM MaltbyY. LanR. Jiang2006Identification and developmental expression analysis of a novel homeobox gene closely linked to the mouse Twirler mutation.Gene Expr Patterns6632636
- 69. Brewer C, Holloway S, Zawalnyski P, Schinzel A, FitzPatrick D (1999) A chromosomal duplication map of malformations: regions of suspected haplo- and triplolethality–and tolerance of segmental aneuploidy–in humans. Am J Hum Genet 64: 1702–1708.C. BrewerS. HollowayP. ZawalnyskiA. SchinzelD. FitzPatrick1999A chromosomal duplication map of malformations: regions of suspected haplo- and triplolethality–and tolerance of segmental aneuploidy–in humans.Am J Hum Genet6417021708
- 70. Brewer C, Holloway S, Zawalnyski P, Schinzel A, FitzPatrick D (1998) A chromosomal deletion map of human malformations. Am J Hum Genet 63: 1153–1159.C. BrewerS. HollowayP. ZawalnyskiA. SchinzelD. FitzPatrick1998A chromosomal deletion map of human malformations.Am J Hum Genet6311531159
- 71. Blanton SH, Bertin T, Serna ME, Stal S, Mulliken JB, et al. (2004) Association of chromosomal regions 3p21.2, 10p13, and 16p13.3 with nonsyndromic cleft lip and palate. Am J Med Genet A 125A: 23–27.SH BlantonT. BertinME SernaS. StalJB Mulliken2004Association of chromosomal regions 3p21.2, 10p13, and 16p13.3 with nonsyndromic cleft lip and palate.Am J Med Genet A125A2327
- 72. Beiraghi S, Foroud T, Diouhy S, Bixler D, Conneally PM, et al. (1994) Possible localization of a major gene for cleft lip and palate to 4q. Clin Genet 46: 255–256.S. BeiraghiT. ForoudS. DiouhyD. BixlerPM Conneally1994Possible localization of a major gene for cleft lip and palate to 4q.Clin Genet46255256
- 73. Marazita ML, Field LL, Cooper ME, Tobias R, Maher BS, et al. (2002) Genome scan for loci involved in cleft lip with or without cleft palate, in Chinese multiplex families. Am J Hum Genet 71: 349–364.ML MarazitaLL FieldME CooperR. TobiasBS Maher2002Genome scan for loci involved in cleft lip with or without cleft palate, in Chinese multiplex families.Am J Hum Genet71349364
- 74. Mitchell LE, Healey SC, Chenevix-Trench G (1995) Evidence for an association between nonsyndromic cleft lip with or without cleft palate and a gene located on the long arm of chromosome 4. Am J Hum Genet 57: 1130–1136.LE MitchellSC HealeyG. Chenevix-Trench1995Evidence for an association between nonsyndromic cleft lip with or without cleft palate and a gene located on the long arm of chromosome 4.Am J Hum Genet5711301136
- 75. Choi SJ, Marazita ML, Hart PS, Sulima PP, Field LL, et al. (2008) The PDGF-C regulatory region SNP rs28999109 decreases promoter transcriptional activity and is associated with CL/P. Eur J Hum Genet. SJ ChoiML MarazitaPS HartPP SulimaLL Field2008The PDGF-C regulatory region SNP rs28999109 decreases promoter transcriptional activity and is associated with CL/P.Eur J Hum Genet
- 76. Han J, Xiao Y, Lin J, Li Y (2006) PDGF-C controls proliferation and is down-regulated by retinoic acid in mouse embryonic palatal mesenchymal cells. Birth Defects Res B Dev Reprod Toxicol 77: 438–444.J. HanY. XiaoJ. LinY. Li2006PDGF-C controls proliferation and is down-regulated by retinoic acid in mouse embryonic palatal mesenchymal cells.Birth Defects Res B Dev Reprod Toxicol77438444
- 77. Welsh IC, Hagge-Greenberg A, O'Brien TP (2007) A dosage-dependent role for Spry2 in growth and patterning during palate development. Mech Dev 124: 746–761.IC WelshA. Hagge-GreenbergTP O'Brien2007A dosage-dependent role for Spry2 in growth and patterning during palate development.Mech Dev124746761
- 78. Rice TK, Schork NJ, Rao DC (2008) Methods for handling multiple testing. Adv Genet 60: 293–308.TK RiceNJ SchorkDC Rao2008Methods for handling multiple testing.Adv Genet60293308
- 79. Sivertsen A, Wilcox AJ, Skjaerven R, Vindenes HA, Abyholm F, et al. (2008) Familial risk of oral clefts by morphological type and severity: population based cohort study of first degree relatives. Bmj 336: 432–434.A. SivertsenAJ WilcoxR. SkjaervenHA VindenesF. Abyholm2008Familial risk of oral clefts by morphological type and severity: population based cohort study of first degree relatives.Bmj336432434
- 80. Cardon LR, Abecasis GR (2003) Using haplotype blocks to map human complex trait loci. Trends Genet 19: 135–140.LR CardonGR Abecasis2003Using haplotype blocks to map human complex trait loci.Trends Genet19135140
- 81. Boyles AL, Wilcox AJ, Taylor JA, Shi M, Weinberg CR, et al. (2008) Oral facial clefts and gene polymorphisms in metabolism of folate/one-carbon and vitamin A: a pathway-wide association study. Genet Epidemiol. AL BoylesAJ WilcoxJA TaylorM. ShiCR Weinberg2008Oral facial clefts and gene polymorphisms in metabolism of folate/one-carbon and vitamin A: a pathway-wide association study.Genet EpidemiolIn press. In press.