|Home | About | Journals | Submit | Contact Us | Français|
Cleft lip/palate comprises a large fraction of all human birth defects, and is notable for its significant lifelong morbidity and complex etiology. Several studies have shown that genetic factors appear to play a significant role in the etiology of cleft lip/palate. Human chromosomal region 9q21 has been suggested in previous reports to contain putative cleft loci. Moreover, a specific region (9q22.3-34.1) was suggested to present a ~45% probability of harboring a cleft susceptibility gene. Fine mapping of fifty SNPs across the 9q22.3-34.11 region was performed to test for association with cleft lip/palate in families from United States, Spain, Turkey, Guatemala, and China. We performed family-based analysis and found evidence of association of cleft lip/palate with STOM (rs306796) in Guatemalan families (P=0.004) and in all multiplex families pooled together (P=0.002). This same SNP also showed borderline association in the US families (P=0.04). Under a nominal value of 0.05, other SNPs also showed association with cleft lip/palate and cleft subgroups. SNPs in STOM and PTCH genes and nearby FOXE1 were further associated with cleft phenotypes in Guatemalan and Chinese families. Gene prioritization analysis revealed PTCH and STOM ranking among the top fourteen candidates for cleft lip/palate among 339 genes present in the region. Our results support the hypothesis that the 9q22.32-34.1 region harbors cleft susceptibility genes. Additional studies with other populations should focus on these loci to further investigate the participation of these genes in human clefting.
Oral-facial clefts arise as a failure of facial embryonic processes to completely merge and/or fuse. They comprise a large fraction of all human birth defects, and may occur as part of single-gene Mendelian syndromes, as part of chromosomal abnormalities, or due to teratogen exposure [Murray, 2002]. Clefts of the lip and clefts of the palate are notable for their significant lifelong morbidity and complex etiology, and are considered different entities: cleft lip can occur with or without cleft palate (cleft lip/palate) while cleft palate may occur as an isolated defect [Fogh-Andersen, 1942]. Cleft lip/palate affects about 1/700 births with wide variability related to geographic origin and socioeconomic status. In general, Native American and Asian populations present the highest frequencies, sometimes at 1/500 or higher, followed by Caucasian, and African-derived populations showing the lowest frequencies around 1/2500 births [Mossey and Little, 2002].
Several studies have shown that genetic factors appear to play a significant role in the etiology of cleft lip/palate. In recent years, it has become evident that cleft lip/palate is heterogeneous, and different chromosome regions such as 1q, 2p, 4q, 6p, 14q, 17q, and 19q have been claimed to contain a clefting locus [Vieira, 2008]. Association of specific genes such as MSX1 [Lidral et al., 1998; Beaty et al., 2001; Jezewski et al., 2003; Vieira et al., 2003], IRF6 [Zucchero et al., 2004; Ghassibe et al., 2005; Park et al., 2007; Vieira et al., 2007; Jia et al., 2009; Jugessur et al., 2009]; PVRL1 [Sozen et al., 2001; Avila et al., 2006], and genes of the FGF signaling pathway [Riley et al., 2007; Menezes et al., 2008] have been reported as well. Recently, genome-wide association studies with German, American and European populations have revealed a new major susceptibility locus for cleft lip/palate located in a gene desert region on chromosome 8q24 [Birnbaum et al., 2009; Grant et al., 2009; Nikopensus et al., 2009].
Human chromosomal region 9q21 has also been suggested in previous reports [Marazita et al., 2002; 2004; 2009], and in a mouse model [Juriloff et al., 2004] to contain putative cleft loci. A marker (D9S1122) located at 9q21.13 showed positive association with cleft lip/palate in Chinese multiplex families [Marazita et al., 2002]. Furthermore, a meta-analysis of a 10-cM genome scan of 388 extended multiplex families with cleft lip/palate from multiple populations revealed potential candidate genes in six chromosomal regions, including 9q21 (heterogeneity LOD score [HLOD] = 6.6). Positive association results (indicating close proximity to a clefting locus) were also found with four of the five selected candidate genes in the region [Marazita et al., 2004]. Moreover, the PPL (posterior probability of linkage) statistical test pointed towards a nearby region, from bands 9q22.3 to q34.1 (between 95Mb and 128Mb), as presenting ~45% probability of harboring a causative gene for clefting (Govil et al., unpublished observations). The FOXE1 (forkhead box E1) gene, located in this region and implicated in syndromic cleft palate (Bamforth-Lazarus syndrome, OMIM 241850) has recently been reported to be associated with both isolated cleft lip with or without cleft palate and isolated cleft palate, and suggested as a key player in primary palatogenesis [Moreno et al., 2009].
To follow up on this data, we performed association studies with densely spaced markers spanning the 9q22.3-34.1 region to test for association with cleft lip/palate.
Details of the population data sets are presented in Table I. Briefly, we assessed 291 multiplex cleft families (one or more members affected with an oral cleft) from the United States, Spain, Turkey, Guatemala and China. All cases had nonsyndromic cleft lip with or without cleft palate. Families were ascertained through probands, and additional relatives were recruited. Individuals presenting syndromic clefts, cleft palate only, or unknown cleft types, as well as control individuals presenting positive family history of clefting were excluded. This study was approved by local and the University of Pittsburgh institutional review boards. A signed informed consent sheet was obtained from all study participants after being explained the objectives and procedures of the study.
Blood or saliva samples were collected to obtain genomic DNA. DNA samples from the populations from China, Turkey, and United States recruited until 2005 were obtained from blood. DNA samples from the population from Guatemala, Spain, and United States recruited after 2005 were obtained from saliva. DNA extraction was performed according to previously published protocols [Trevillato and Line, 2000]. There was no difference in the performance of the samples obtained from blood or saliva in regards to our genotyping approach described below.
We used the International HapMap Project Database to guide our selection of single nucleotide polymorphisms (SNPs) to be genotyped. We based our selection using the approach devised by Carlson et al. [Carlson et al., 2004], selecting from a set of SNPs which maximally represented the linkage disequilibrium structure of the 9q22.3-34.1 region. Fifty SNPs spanning the 9q22.3-34.1 region were selected (Figure I). Preference was given to tag SNPs with a minor allele frequency of at least 10%, taking into consideration the various ethnicities represented in this study.
Genotypes were generated using Taqman chemistry [Ranade et al., 2001] on an automatic sequence-detection instrument (ABI Prism 7900HT, Applied Biosystems, Foster City, CA). Assays and reagents were supplied by Applied Biosystems (Applied Biosystems, Foster City, CA). Details of the studied polymorphisms are presented in Table II.
The transmission disequilibrium test was used to assess association of alleles at each marker with cleft lip with or without cleft palate with the use of the Family Based Association Test (FBAT) software [Horvath et al., 2001]. We also performed haplotype analyses using the “hbat” function of the FBAT software to identify if sets of markers were in linkage disequilibrium with cleft lip/palate. We also analyzed the entire dataset according to the following cleft subgroups: CLO + CLCLP (families where all affecteds have cleft lip only plus families where at least one affected has cleft lip only and one affected has cleft lip and palate, excluding any family where an affected has cleft palate only); and CLO +CLP +CLCLP (families where all affecteds have cleft lip only plus families where all affecteds have cleft lip and palate plus families where at least one affected has cleft lip only and one affected has cleft lip and palate, excluding any family where an affected has cleft palate only). Populations were analyzed individually and as a pooled dataset. P-values below 0.001 (0.05 divided by 50 markers) were considered statistically significant.
We attempted to identify candidate genes present in the 9q22.3-34.1 region using the gene prioritization software Endeavour [Aerts et al., 2006; Tranchevent et al., 2008]. Endeavour is a web server that allows users to prioritize candidate genes with respect to their biological processes or diseases of interest. It relies on the similarity between candidate genes and models built with a set of manually inputted reference genes. We used 10 genes to compose the reference set: IRF6, MSX1, PVRL1, TGFB3, CLPTM1, TGFA, FGFR1, RARA, BCL3, and TBX22. We did not use PTCH or FOXE1 as reference genes in the target region as they would be automatically removed by the program due to redundancy between reference and test sets of genes.
We genotyped fifty SNPs spanning 339 genes present in chromosomal region 9q22.3-q34.1. Detailed information on allele frequencies per population and additional replication analyses of specific markers are available as Supplemental Material. We performed family-based analysis to assess genotype and allelic associations between the investigated SNPs and cleft lip/palate. After correcting for multiple testing, we found borderline association of cleft lip/palate with a SNP in STOM (rs306796, 9q33.2) in Guatemalan families (P=0.004) and all families pooled together (P=0.002). This same SNP (rs306796) also showed a trend for association in the US population (P=0.04) (Table III). Of note, SNPs in or nearby STOM also showed modest association with cleft lip/palate in four of the five populations studied (P=0.04). Under a nominal value of 0.05, other SNPs also showed association with cleft lip/palate and cleft subgroups (Tables III and andIVIV).
When analyzing families according to cleft subgroups, we found association of markers in PTCH (rs2234607, P=0.004; rs2297088, P=0.01; and rs10512248, P=0.01) and in/nearby STOM (rs306796, P=0.01; and rs2300934, P=0.03) with the CLO+CLCLP subgroup in Guatemalan families. We also found borderline association of markers nearby FOXE1 in Chinese (rs2900463; P=0.03) and Guatemalan (rs1031111; P=0.04) families (Table IV). Results for the CLO +CLP +CLCLP groups did not differ from the association results in individual populations (data not shown).
We also performed haplotype analysis of the investigated SNPs by population (Table V) and by population according to cleft subgroup (Table VI). Altered transmission of haplotypes involving markers in/nearby STOM was evident in families from the US (rs1928623-rs2251495-rs2808416; P=0.02), Turkey (rs1928623-rs2251495-rs2808416-rs1928482; P=0.02), Guatemala (rs2300934-rs306796; P=0.02 and rs306796-rs10739600; P=0.03), and China (rs2300934-rs306796; P=0.04) (Table V). Altered transmission was also observed for PTCH rs357654-rs2236407-rs2297088 haplotype (P=0.03) and for were PTCH rs357654-rs2236407-rs2297088-rs10512248 haplotype (P=0.02) in Guatemalan families (Table V). When analyzed by cleft subgroup, interesting results were observed for haplotype markers involving PTCH (rs357654-rs2236407; P=0.01) and PTCH and FOXE1 (rs10512248-rs894673) in families from Guatemala (P=0.01) and China (P=0.03) presenting CLO + CLCLP (Table VI).
We found 339 genes within the 9q22.3-34.1 region with Endeavour (Supplementary Table I). The algorithm prioritized all genes within the 9q22.3-34.1 region in relation to cleft lip/palate. The highest priority gene according to Endeavour was TGFBR1. The associated genes, PTCH and STOM, ranked in tenth and fourteenth places in the priority output, respectively. FOXE1 ranked in 39th place. Detailed results are available as Supplemental Material.
We investigated a region on chromosome 9q identified in previous studies as to harbor cleft susceptibility loci. Among the genes present in the region are PTCH and FOXE1, in which missense mutations and polymorphic variants have been described as having a role in cleft lip/palate [Marazita et al., 2004; 2009; Mansilla et al., 2006; Vieira et al., 2005; Moreno et al., 2009].
We performed fine mapping association studies with fifty SNPs across a 32.65 Mb region and multiple populations. Although using 50 SNPs to investigate 339 genes in a region may not seem a dense enough strategy, we used information on linkage disequilibrium of the markers to avoid choosing redundant markers. We found association of a polymorphism in STOM (rs306796) in the Guatemalan families (P=0.004) and in all families pooled together (P=0.002). STOM encodes stomatin, a membrane protein that was first isolated from human red blood cells [Zhang et al., 1999]. In mouse, the STOM homolog presents a wide pattern of expression, with high levels of mRNA in heart, liver, skeletal muscle, and testis but low levels in lung, brain, and spleen [Gallagher et al., 1995b]. Despite being associated with a variety of diseases such as cancer, kidney failure and anemia, precise functions of this protein remain unclear. In humans, the absence of stomatin is associated with a form of hemolytic anemia known as hereditary stomatocytosis [Yokoyama et al., 2006]. Lower expression of stomatin results in an increase in the basal rate of glucose transport [Zhang et al., 2001]. Interestingly, diabetic pregnant women are at an increased risk for having offspring with neural tube defects and oral-facial clefts [Spilson et al., 2001] and excessive exposure to glucose has been postulated to play a role in the pathogenesis of nonsyndromic cleft lip/palate [Krapels et al., 2004].
Suggestive evidence of association was also observed for several other markers among the populations implicating that the chromosome 9q region might contain cleft susceptibility genes (P<0.05). When analyzing the dataset according to cleft subgroups, we observed borderline association of PTCH variants (rs2236407, P=0.004; rs2297088, P=0.01; and rs10512248, P=0.01) in Guatemalan families with CLO + CLCLP (families where all affecteds have cleft lip only plus families where at least one affected has cleft lip only and one affected has cleft lip and palate). PTCH was initially suggested as a candidate for human clefting because of its mouse homolog located in the candidate region clf2 [Juriloff et al., 2001]. Although no causal mutation in Ptch coding sequence was found in mice, alterations in the regulatory sequence are not ruled out. Given the “normal” phenotype of noncleft A/~ strain mice, a mild regulatory defect seems likely if Ptch is clf2 [Juriloff et al., 2004]. In humans, mutations in PTCH cause Gorlin syndrome with diverse developmental anomalies, often including rib and craniofacial abnormalities (and cleft palate) and a mixture of tumor types [Cohen, 1999]. In their study with PTCH and isolated cleft lip/palate, Mansilla et al. sequenced all 23 exons of the gene and found seven new normal variants spread along the entire gene and three missense mutations in cases with cleft lip/palate, one of which was not found in 1,188 control samples [Mansilla et al., 2006]. Although the authors did not find statistically significant evidence of linkage (multipoint HLOD peak=2.36), they reported over-transmission of the PTCH rs2297088–rs2236407 haplotype with borderline statistical significance (P=0.08) in Filipino families with two or more affected members. They further concluded that missense mutations in PTCH may be rare causes of isolated cleft lip/palate and yet unidentified variants near PTCH may act as modifiers of the cleft phenotype. Corroborating with these findings, we observed significant evidence of transmission distortion for the PTCH rs357654-rs2236407-rs2297088 (P=0.03) and rs357654-rs2236407-rs2297088-rs10512248 (P=0.02) haplotypes in Guatemalan families.
It is unknown whether the associated PTCH polymorphisms (rs2236407 and rs2297088) located in introns and therefore not in coding regions or splice sites might cause a functional change in the final protein. Notwithstanding, intronic polymorphisms have been demonstrated in association with other complex diseases, including the association of IRF6 with cleft lip/palate [Zucchero et al., 2004; Ghassibe et al., 2005; Park et al., 2007; Vieira et al., 2007; Jia et al., 2009; Jugessur et al., 2009] and thus should not be disregarded as potentially damaging. Of note, in a recent association study with genes in chromosome 9q, PTCH SNP rs2297088 showed the strongest signal for Filipino families, based on the lowest p-value (P-value=6.49E-03), indicating that PTCH may have a possible role in the etiology of cleft lip with or without cleft palate in some populations.
A previous study from our group has shown that point mutations in FOXE1 may be rare causes for isolated cleft lip/palate [Vieira et al., 2005]. Following studies have shown positive association and linkage between FOXE1 and cleft lip/palate [Marazita et al., 2004; 2009], including a recent publication which identified FOXE1 in strong association with both isolated cleft lip with or without cleft palate and isolated cleft palate [Moreno et al., 2009]. We did not find significant association of FOXE1 with cleft phenotypes in the populations tested although suggestive association was observed for variants in/nearby FOXE1 in Chinese and Guatemalan families with CLO + CLCLP (P=-0.03 and P=0.04, respectively). FOXE1 belongs to a family of fork-head domain-genes involved in embryonic pattern formation, which have been identified as factors that bind to regulatory elements in mammalian genes expressed in terminally differentiated cells [Kaestner et al., 1993]. Foxe1 null mice exhibit cleft palate and either a sublingual or completely absent thyroid gland [De Felice et al., 1998]. In humans, mutations in FOXE1 result in the Bamforth syndrome, characterized by thyroid agenesis, cleft palate, spiky hair, and choanal atresia [Clifton-Bligh et al., 1998]. It is possible that variants in this gene may be in linkage disequilibrium with variants in other genes that jointly increase the susceptibility to cleft lip/palate. For example, we found overtransmission of PTCH-FOXE1 haplotypes in families Guatemalan and Chinese families presenting CLO + CLCLP (P=0.01 and P=0.03, respectively).
Using gene prioritization software, we identified 339 genes in the 9q22.3-34.11 candidate region, which were ranked based on similar roles or participation in similar biological processes with genes known to be associated with cleft lip/palate. Other investigators have used this gene prioritization approach as well, in studies of obesity and type II diabetes genes [Elbers et al., 2007; Sookoian et al., 2009], idiopathic pulmonary fibrosis [Tzoulevekis et al., 2007] and even cleft lip/palate [Osoegawa et al., 2008]. This approach however has the limitation of not allowing any control on the quality of gene annotation and quantity and quality of information across genes in different databases. It is difficult to predict how much these aspects influenced our results and the results of others but at minimal, this information adds to the bulk of the results and help interpreting the association data. In comparison with a reference set of 11 established candidate genes for human clefting, PTCH and STOM ranked in tenth and fourteenth places, respectively, among the 339 genes present in the investigated region, thus at the top five percent ranking as candidates for cleft lip/palate. These observations taken together with the association findings, suggest that those genes may be considered plausible candidates for cleft lip/palate. Meanwhile, we checked for conservation of the associated SNPs, and found that the wild-type nucleotide is conserved in the following species: rhesus monkey, dog and horse for rs2297088 in PTCH, and dog and horse for rs306796 in STOM.
In summary, we performed fine mapping analysis of chromosome 9q22.32-34.1 region previously suggested to harbor cleft susceptibility genes. Our association results support a role for PTCH as contributor for cleft lip/palate and suggest STOM as a possible new candidate gene. Haplotype and gene prioritization analyses confirmed the individual association findings with PTCH and STOM, ranking in the top five percent of the highest priority candidate genes for cleft lip/palate. We failed to replicate previous findings suggesting FOXE1 contributes to nonsyndromic cleft lip and palate. Additional studies with other populations should focus on these loci to further investigate the participation of these genes in human clefting.
We gratefully acknowledge individuals and families for their valuable collaboration. Thanks to research coordinators and staff at each collection site. Wendy Carricato helped with sample organization. Rebecca DeSensi and Kathleen Deeley provided technical assistance. Maria Adela Mansilla and Lina Maria Moreno at the University of Iowa helped with SNP marker selection. This work was supported by the National Institutes of Health grants: K99-DE018954 (to AL); K99-DE018413 (to RM); K99-DE018085 (to MG); R01-DE016148 and P50-DE016215 (to MLM); R21-DE16718 (to ARV); FAPERJ grant E-26/152.831/2006, CNPq grants 308885/2006-0 and 401467/2004-0 (to IMO); and CAPES, Brazil (to RFF). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Dental and Craniofacial Research or the National Institutes of Health. This paper is partly based on a thesis submitted to the graduate faculty, Federal University of Rio de Janeiro, in partial fulfillment of the requirements for the PhD degree (for RFF).
Web Resources: Accession numbers and URLs for data presented herein are as follows:
Applied Biosystems, http://www.appliedbiosystems.com/index.cfm
Family Based Association Test, http://www.biostat.harvard.edu/~fbat
International HapMap Project, http://hapmap.org
Supplementary information is available at the American Journal of Medical Genetics website.