Autosomal Dominant Nonsyndromic Cleft Lip and Palate: Significant Evidence of Linkage at 18q21.1

Abstract

Nonsyndromic cleft lip with or without cleft palate (NSCL/P) is one of the most common congenital facial defects, with an incidence of 1 in 700–1,000 live births among individuals of European descent. Several linkage and association studies of NSCL/P have suggested numerous candidate genes and genomic regions. A genomewide linkage analysis of a large multigenerational family (UR410) with NSCL/P was performed using a single-nucleotide–polymorphism array. Nonparametric linkage (NPL) analysis provided significant evidence of linkage for marker rs728683 on chromosome 18q21.1 (NPL=43.33 and P=.000061; nonparametric LOD=3.97 and P=.00001). Parametric linkage analysis with a dominant mode of inheritance and reduced penetrance resulted in a maximum LOD score of 3.61 at position 47.4 Mb on chromosome 18q21.1. Haplotype analysis with informative crossovers defined a 5.7-Mb genomic region spanned by proximal marker rs1824683 (42,403,918 bp) and distal marker rs768206 (48,132,862 bp). Thus, a novel genomic region on 18q21.1 was identified that most likely harbors a high-risk variant for NSCL/P in this family; we propose to name this locus “OFC11” (orofacial cleft 11).

Cleft lip and palate are birth defects that affect the upper lip and the roof of the mouth. Nonsyndromic cleft lip with or without cleft palate (NSCL/P) is one of the most common congenital craniofacial birth defects, affecting 1 in 700–1,000 newborns in the United States each year.¹ Its highest prevalence rates are in Native Americans and Asians.^2–4 Nonsyndromic familial NSCL/P represents almost half of facial malformations; the cases can be sporadic or inherited as an autosomal dominant trait. On the other hand, >400 syndromes, including numerous chromosomal anomalies, may include a facial cleft as one of the manifestations. Some of the common syndromes and/or anomalies associated with such clefting include van der Woude (MIM 119300), Apert (MIM 101200), Meckel (MIM 249000), Treacher Collins (MIM 154500), and Zlotogora-Ogur (MIM 225060) syndromes; however, the genes mutated in these disorders are not responsible for NSCL/P.⁵ Recent reports suggest that several genes on various chromosomal regions have pathogenic mutations associated with NSCL/P, including LHX8 (MIM 604425) on 1p31.1, SKI (MIM *164780) on 1p36.3, MTHFR (MIM *607093) on 1p36.3, IRF6 (MIM *607199) on 1q32-q41, TGFB2 on 1q41, TGFA (MIM *190170) on 2p13.3, GLI2 (MIM *165230) on 2q14.2, SATB2 (MIM *608148) on 2q32-q33, SUMO1 (MIM *601912) on 2q33.1, P63 (MIM *603273) on 3q27-q29, MSX1 (MIM *142983) on 4p16.3-p16.1, SPRY1 (MIM *602465) on 4q28.1, MSX2 (MIM *123101) on 5q34-q35, F13A1 (MIM +134570) on 6p25.3-p24.3, TGFBR1 (MIM *190181) on 9q33-q34, FOXE1 (MIM *602617) on 9q22.33, PVRL1 (MIM *600644) on 11q23-q24, SPRY2 (MIM *602466) on 13q31.1, TGFB3 (MIM *190230) on 14q24.3, JAG2 (MIM *602570) on 14q32.33, GABRB3 (MIM *137192) on 15q11.2-q12, RARA (MIM *180240) on 17q21.1, BCL3 (MIM *109560) on 19q13.31, TGFB1 (MIM *190180) on 19q13.2, TBX1 (MIM *602054) on 22q11.21, and PHF8 (MIM *300560) on Xp11.2^6–13; however, none of these seem to be a common locus for the majority of sporadic or familial cases of NSCL/P.^14–16

Various independent association and linkage studies of different populations have identified 10 loci with evidence of linkage for syndromic and/or nonsyndromic orofacial cleft (OFC): on 6p24.3 (OFC1 [MIM 119530]), 2p13 (OFC2 [MIM 602966]), 19q13 (OFC3 [MIM 600757]), 4q21-q31 (OFC4 [MIM 608371), 4p16.1 (OFC5 [MIM 608874]), 1q32-q41 (OFC6 [MIM 608864]), 11q23-q24 (OFC7 [MIM 600644]), 3q27 (OFC8 [MIM 129400]), 13q33.1-34 (OFC9 [MIM 610361]), and 2q32.2-q33 (OFC10 [MIM 601912]). Pathogenic mutations have been identified in 4 of these 10 loci: MSX1 on 4p16.1, IRF6 on 1q32-q41, PVRL1 on 11q23-q24, and TP73L on 3q27.^17–20 Increased incidence of clefts associated with chromosomal aberrations involving chromosomes 13 and 18 have also been reported.^21,22

A genomewide linkage analysis was performed on an extended American family (UR410) of self-reported European origin with NSCL/P and likely autosomal dominant mode of inheritance. The family initially was tested for linkage to loci reported elsewhere to be candidates for autosomal dominant forms of NSCL/P (OFC4). Markers analyzed by Beiraghi et al.²³ for the OFC4 locus on chromosome 4q provided weak evidence of linkage (LOD score <2.3). The present analysis with use of SNP markers provides significant evidence of a susceptibility locus on a genomic region of ∼5.7 Mb on chromosome 18q21.1.

The six-generation NSCL/P-affected family (UR410) with probable autosomal dominant inheritance was first ascertained in Switzerland; however, a branch of 30 individuals residing in the United States has extended the pedigree reported by Ward et al.²⁴ Of the 52 individuals in this family, 11 (7 males and 4 females) showed NSCL/P phenotypes, ranging from unilateral to bilateral NSCL/P. There were no other associated anomalies observed in the individuals of this family. Informed consent was obtained from all subjects who participated in the study, and blood samples were collected from all cooperating family members. Samples from a total of 27 individuals were used for the linkage analysis: 9 affected individuals, 4 obligate carriers, and 14 apparently unaffected individuals (fig. 1).

Figure 1. .

Pedigree of family UR410 with NSCL/P. Affected individuals are shown with blackened symbols, and unaffected individuals are shown with unblackened symbols. Samples included in the analysis are numbered below their pedigree symbols. A dot in the center of a symbol indicates an individual who is an obligate carrier and produced affected children with NSCL/P. BCLP=bilateral cleft lip and cleft palate; UCLP=unilateral cleft lip and cleft palate; BCL=bilateral cleft lip; UCL=unilateral cleft lip.

DNA from peripheral-blood samples was isolated using DNA-extraction kits. A genomewide genotyping was undertaken using GeneChip Mapping 10K XbaI Array containing 10,555 SNPs. These SNP markers are equally distributed in the genome, with mean intermarker distances of 250 kb and an average heterozygosity of 0.38 (Affymetrix). The assay was done using 250 ng of genomic DNA, and >99% of the SNPs were determined unequivocally for each sample. Scanned images were processed with Affymetrix Micro Array Suite Software, and data were analyzed with GDAS v2 software. PedCheck was used for the detection of any Mendelian incompatibility of genotypes.²⁵ SNP genotype data were imported into the linkage-analysis programs GENEHUNTER²⁶ and MERLIN.²⁷ In the initial genome scan, evidence of linkage was assessed with a nonparametric, penetrance-independent, affected-only, and allele-sharing analysis (NPL score). With use of MERLIN, one can convert this into a nonparametric LOD score (LOD*) by maximizing the likelihood with respect to a scalar parameter, δ, that measures the amount of excess sharing of identical-by-descent alleles among affected relatives (with δ=0 corresponding to the null hypothesis of no linkage²⁸). We used the S_all scoring function^26,29 and the exponential allele-sharing model to generate the relevant linkage statistic.

When significant evidence of linkage was found by exceeding the predetermined threshold (P<.01), two-point as well as multipoint LOD scores maximized over various plausible genetic model parameters (MOD-score analysis) of the entire pedigree were performed, using a GENEHUNTER_MODSCORE analysis package. The map order and intermarker distances between SNPs were based on data from the National Center for Biotechnology Information (build 35.1). To evaluate the false-positive rate of linkage, simulations were performed to evaluate the results with use of empirical P values. The simulations were designed to match the observed data in marker density and informativeness, pedigree structure, and individual phenotypes. We generated 10,000 replicate data sets under the null hypothesis of no linkage, to estimate the empirical P value. On the basis of our simulations, the empirical threshold for genomewide significance at the 5% level was set at NPL=42.12. Putative haplotypes containing the disease-causing loci were determined by using the critical recombinants across the affected family members, with use of SIMWALK and GENEHUNTER.

Linkage analysis of tightly linked loci can lead to an excess of false-positive results if the markers are in strong linkage disequilibrium (LD) and parents are not available for genotyping.^30,31 Therefore, the impact of LD on LOD scores was reassessed at the linked regions, with use of two approaches. First, MERLIN was used to accommodate marker-marker LD in both parametric and nonparametric analyses, by organizing closely located adjacent markers into clusters. Although many studies have shown that the extent and distribution of LD are variable throughout the genome, in most cases, significant LD does not influence markers separated by >0.1 cM in outbred populations.^32–34 However, to be conservative, markers within 0.2 cM in a cluster were used. Second, SNPLINK³⁵ was used to calculate LD between markers, and the LD was then removed by considering each set of markers in LD (defined as “sets in which each consecutive marker pair in the set is found to be in LD”) and by retaining one SNP from each set, chosen as the middle SNP from the set. Linkage was then recalculated using the new LD-free set of markers. We used LD values >0.7 and >0.4 for D′ and r², respectively, to define significant LD, as suggested elsewhere.^36,37

Before this genomewide analysis, 19 polymorphic microsatellite markers on chromosome 4q (OFC4) had been genotyped in the region from 117.06 cM to 161.04 cM (GDB Human Genome Database). It was suggested that chromosome 4q was involved in NSCL/P, because a previous limited study²³ of a small branch of the UR410 family had detected LOD scores of 2.27 and 1.93 for markers D4S175 and D4S192, respectively. Linkage analysis was performed using LINKAGE, under the assumption of an autosomal dominant inheritance with reduced penetrance.

The linkage analysis of and haplotype data about chromosome 4q determined using samples from the extended pedigree excluded the published candidate genomic region²³ (table 1). The genomewide linkage scan with use of an NPL analysis provided significant evidence of linkage for an NSCL/P locus (OFC11) at and around marker rs728683 on chromosome 18q21.1 (NPL=43.33; P=.000061). With use of simulations (see above), this value is equivalent to a genomewide empirical P value of .012. We also converted these NPL values to nonparametric LOD* scores,²⁸ maximizing over δ. The peak linkage corresponds to the Kong and Cox²⁸ LOD* score of 3.97, with P=.00001. These data were also supported by subsequent parametric linkage analysis, with a dominant mode of inheritance and reduced penetrance. The maximum multipoint LOD score of 3.61 at position 47.4 Mb on chromosome 18 was observed. Twenty-three SNP markers, spanning a region of ∼8.2 cM (rs1824683–rs724349), showed parametric LOD scores from 3.61 (NPL=43.30; P=.000061) to 2.92 (NPL=17.33; P=.000244) (table 2). The best-fitted parametric model that was used for obtaining the optimum linkage results was incomplete dominance with 65% penetrance and disease-allele frequency of 0.0001. No additional peak with a statistically significant LOD score was found in the genome (fig. 2). Haplotype analysis for the 18q21.1-linked region was performed using 27 informative SNP markers. Informative crossovers in affected individuals IV-10 (4021) and V-3 (4009), between rs1824683/rs328149 and rs768207/rs768206, respectively, defined the NSCL/P candidate region of 5.7 Mb, bordered by proximal marker rs1824683 (physical map position 42,403,918 bp) and distal marker rs768206 (physical map position 48,132,862 bp) (fig. 3).

Table 1. .

Two-Point LOD Scores between NSCL/P in family UR410 and Polymorphic Markers from Known Candidate Genomic Region on Chromosome 4q^[Note]

	LOD at θ=
Marker	.00	.01	.05	.10	.20	.30	.40	Mapping Position (cM)
D4S193	−9.86	−3.12	−.59	.28	.76	.71	.43	117.06
D4S1612	−10.37	−3.98	−1.33	−.31	.44	.59	.42	124.45
D4S194	−4.73	−1.85	−.70	−.28	−.01	.02	.00	126.15
D4S2975	−3.14	−.46	.24	.47	.51	.34	.12	126.71
D4S1615	−1.69	.77	1.27	1.31	1.08	.69	.26	128.31
D4S2307	−3.21	−.45	.76	1.12	1.19	.96	.55	129.92
D4S429	−10.72	−3.30	−1.35	−.62	−.09	.00	−.01	131.00
D4S3039	−6.96	−1.90	−.62	−.18	.10	.10	.01	132.72
D4S175	−3.36	.45	1.64	1.97	1.93	1.52	.87	134.74
D4S2939	−3.58	−1.94	−.72	−.17	.20	.20	.09	142.24
D4S192	−3.11	−.55	.16	.43	.60	.54	.34	143.31
D4S424	−1.22	.45	.99	1.10	1.00	.76	.42	144.56
D4S2998	−9.60	−3.87	−1.23	−.23	.50	.63	.44	145.98
D4S1586	−5.77	−1.98	−.66	−.16	.19	.25	.17	147.06
D4S3008	.63	.62	.59	.54	.42	.30	.16	152.98
D4S3021	−7.44	−5.00	−2.94	−1.99	−1.03	−.51	−.19	154.63
D4S2631	−7.21	−3.25	−1.31	−.59	.07	.06	.05	155.75
D4S413	−11.86	−3.41	−1.46	−.73	−.19	−.03	.00	157.99
D4S3033	−9.32	−2.55	−.09	.70	1.04	.84	.42	161.04

Note.— Scores were calculated under a dominant model with 75% penetrance and .0001 disease-allele frequency.

Table 2. .

Parametric-Linkage and NPL Scores for the Peak Region at Chromosome 18q21.1^[Note]

	Position		Genome Scan of theNSCL/P-Affected Family
				Nonparametric LOD
SNP	Genetic (cM)	Physical (bp)	Parametric LOD	NPL	P
rs726131	65.641	42,118,566	−.55	10.53	.000656
rs1824683	66.057	42,403,918	3.48	34.40	.000244
rs328149	66.087	42,424,469	3.51	36.10	.000244
rs645631	66.203	42,505,993	3.61	42.82	.000153
rs1073744	66.204	42,506,272	3.61	42.81	.000153
rs953291	67.077	43,124,497	3.61	42.78	.000153
rs1944574	68.548	43,979,261	3.61	42.82	.000153
rs953570	68.567	43,988,463	3.61	42.82	.000153
rs1944584	68.667	44,035,980	3.61	42.82	.000153
rs1398193	68.885	44,241,137	3.61	42.83	.000153
rs1877412	68.885	44,241,673	3.61	42.83	.000153
rs1943984	70.873	45,477,909	3.61	42.76	.000153
rs1985467	71.431	45,928,259	3.61	42.82	.000153
rs1822466	71.519	45,999,664	3.61	42.83	.000153
rs1373185	71.519	45,999,849	3.61	42.83	.000153
rs768360	71.542	46,018,071	3.61	42.83	.000153
rs2006747	71.795	46,222,356	3.61	42.83	.000153
rs728682	71.966	46,349,352	3.61	42.83	.000153
rs2969972	71.989	46,362,761	3.61	42.84	.000061
rs1822458	72.264	46,528,403	3.61	42.94	.000061
rs1941962	72.892	46,999,467	3.61	43.17	.000061
rs1369766	73.228	47,272,878	3.61	43.30	.000061
rs728683	73.339	47,362,995	3.61	43.33	.000061
rs724349	74.247	48,019,014	2.92	17.33	.000244
rs959655	74.477	48,132,862	.23	10.74	.000610
rs768207	74.477	48,132,968	−.04	10.73	.000610

Note.— The model used for parametric analysis was incomplete dominance with 65% penetrance and a disease-allele frequency of .0001.

Figure 2. .

Multipoint linkage analysis with use of NPL in the genomewide scan of multigenerational family UR410 with NSCL/P. In the upper panel, the X-axis represents the chromosome location for the 22 autosomes, and the Y-axis represents the NPL score. The highest peak is on chromosome 18q21.1. In the lower panel, the X-axis shows chromosome location for the 22 chromosomes, and the Y-axis shows the location score, calculated using GENEHUNTER software. The maximal LOD score of 3.61 is observed around marker rs728683.

Figure 3. .

Genotypes and haplotypes of chromosome 18q21.1 SNP markers are shown below selected individuals of family UR410. Haplotypes associated with affected status are shown in red. Haplotype analysis indicated that the cosegregating segment of the NSCL/P locus is flanked proximally by marker rs1824682 (position 42403918) and distally by marker rs959655 (position 48132862) on chromosome 18q21.1.

To assess the impact of LD on linkage, we first used markers within 0.1–0.2 cM of each other in a cluster and estimated the linkage scores. Several clusters of 2–6 SNPs demonstrated LD. With the assumption of complete LD within the cluster, MERLIN uses population haplotype frequencies in the calculation of linkage. Interestingly, at chromosome 18, the LOD and NPL scores were virtually the same, especially for the significant markers. We also recalculated the linkage scores by removing the redundant SNPs (D^′>0.7 or r²>0.4), using SNPLINK, and arrived at almost similar results.

The etiology of OFCs is complex, related to both genetic and environmental factors.^2,38 Although there have been a number of studies undertaken to identify the genetic variations responsible for NSCL/P, the results have not been consistent among populations, and no single susceptibility gene has been identified to date that plays a major role in this disorder.^39–44 The majority of these reports used small nuclear families; however, there are a few exceptional large multigenerational families with both an autosomal dominant and autosomal recessive inheritance of NSCL/P that gave significant evidence of linkage to chromosomes 13q,⁴⁵ 17p,⁴⁶ and multiple loci.⁴⁷ Previous studies of a subset of pedigree UR410 suggested evidence of linkage on 4q; however, the present study, which includes 19 additional genotyped individuals (2 affected), fails to confirm this region (LOD score <2.3). The previous data were also considered to be a false-positive result by two independent studies.^42,48 We report here linkage analysis, with use of an SNP array containing ∼10,000 SNP markers, that provided significant evidence of a novel susceptibility locus on a genomic region of 5.7 Mb on chromosome 18q21.1.

Several genomewide linkage studies conducted using multiplex families of various ethnic origins showed evidence of linkage at chromosome 18; however, these suggested regions either do not include the present locus or do not have statistically significant results.^39,40,44,49 The genome-scan meta-analysis conducted by Marazita et al.⁴³ of 388 extended multiplex families with NSCL/P from seven diverse populations yielded positive evidence at multiple loci, including weak evidence at the chromosome 18q21.1 region. Furthermore, the study suggested SMAD2 as a candidate in the mapped genetic interval.

A whole-genome scan, performed using a high-density SNP genotyping assay conducted by Radhakrishna et al.,⁴⁵ of two large Indian families provided suggestive evidence of linkage on chromosome 18q21.1 and strong evidence on chromosome 13q33.1-34. The SNP marker (rs959655) that gave the highest NPL peak (NPL=3.64; P=.0066) on chromosome 18 in the previous study⁴⁵ is also supported by parametric LOD score. This SNP marker is in the vicinity of the genetic interval mapped in the present study. Several studies have also found that genetic variation in IRF6 contributes to the development of NSCL/P^50–52; however, in the present study, we did not observe any positive linkage.

Numerous reports indicate that chromosome 18–related anomalies are associated with increased incidence of NSCL/P, including deletion of 18q^53–57 and full or partial trisomy 18 involving the region 18q21.^{21,22,58–69} Niedrist et al.²² analyzed 352 cases of trisomy 18 from northeastern Switzerland and found that 14% of the subjects had facial clefts. Schinzel et al.⁷⁰ reported three subjects with the 18q syndrome associated with various developmental anomalies and cleft palate. An isochromosome 18q in a fetus with congenital megacystis, growth retardation, and cleft lip and palate was reported by Chen et al.⁵⁹ Partial deletion of 18q in a patient with bilateral complete cleft lip was reported by Fujimoto et al.⁶⁸ Cody et al.⁵⁵ analyzed 42 individuals with deletions of 18q and observed several subjects with craniofacial abnormalities, including NSCL/P, associated with breakpoints at 18q21.1, 18q21, and 18q21.2.

The 5.7-Mb interval containing the locus for NSCL/P on chromosome 18q21.1 contains 36 annotated transcripts (Ensembl), including several genes implicated in developmental or pathological events in craniofacial birth defects. Candidate genes include protein inhibitor of activated STAT2 (PIAS2 [MIM 603567]), mothers against decapentaplegic and homologs of the Drosophila (MAD) genes (SMAD2 [MIM 601366], SMAD4 [MIM 600993], and SMAD7 [MIM 602932]). Recent reports suggested that the members of the transforming growth factor-β (TGF-β) superfamily that function in the embryonic palate are mediated through the SMAD-signaling system.⁷¹ SMAD proteins play a key role in intracellular TGF-β signaling, and mutations in the TGF-β pathway confer resistance to growth inhibition by TGF-β.⁷² The TGF-β3–knockout mouse resulted in cleft palate due to failure of palatal shelf fusion,^73–75 and polymorphisms in the gene for TGF-β3 have been linked to the development of cleft palate in humans.⁶

Haplotype analysis with critical recombination events allowed us to define a 5.7-Mb disease interval delimited by proximal SNP marker rs824683 (map position 42,403,918 bp) and distal SNP marker rs768206 (map position 48,133,056 bp). The critical interval of the 11th OCF locus (OFC11) could not be further reduced, since additional members of the family were not available. Haplotype analysis identified the risk haplotype shared by all affected individuals that was not found in any of the unrelated unaffected spouses. Since NSCL/P demonstrates reduced penetrance, all unaffected parents (subjects 4001, 4004, 4030, and 4032) who had affected children shared the affected haplotype. Therefore, we speculate that individuals 4018, 4020, 4024, and 4027 with normal phenotype and “at risk” haplotypes may also be at risk of having affected children. The haplotype data also support the autosomal dominant mode of inheritance in this family. Variation in the phenotypic expression—that is, the range from bilateral to unilateral NSCL/P—in members of the same pedigree (UR410) indicates that the genetic variation responsible for these abnormalities at a single locus might have variable phenotypic expression. Few of the positional candidate genes are currently being screened for a possible role in the pathogenesis of NSCL/P, but the results are inconclusive at present. We are also pursuing cytogenetic analysis and array comparative-genomic hybridization that may reveal microdeletions and/or duplications. In summary, this study demonstrates the power of using multigenerational families in the analysis of complex traits. Identification of the pathogenic genetic variation would potentially provide significant insights into the molecular mechanisms underlying the etiology of NSCL/P.