Introduction

Nonsyndromic cleft lip with or without cleft palate (CL/P) is the most common craniofacial anomaly. It has a prevalence of approximately 1 in 600 among Thai newborns (Shotelersuk et al. 2003). Environmental and genetic factors have been implicated in CL/P and several different loci and genes have been associated with them (Jugessur and Murray 2005).

In 1994, MSX1 first emerged as a candidate based on the CL/P and foreshortened maxilla phenotype in the knockout mouse (Satokata and Maas 1994). Several association studies of the gene with CL/P and cleft palate only (CPO) further supported the role of MSX1 in nonsyndromic clefting in different populations (Lidral et al. 1998; Blanco et al. 2001; Jugessur et al. 2003). Later on, a study of a Dutch family with tooth agenesis and various combinations of CL/P and CPO showed a nonsense mutation in MSX1, suggesting that disease-causing mutations in MSX1 might be etiological in a portion on nonsyndromic CL/P cases (van den Boogaard et al. 2000). Two previous studies, reporting that 2% of cases of clefting had MSX1 mutations, supported this hypothesis (Jezewski et al. 2003; Suzuki et al. 2004). Of the proposed pathogenic mutations, the P147Q variant was found in approximately 2% of Vietnamese (Suzuki et al. 2004) and 0.15% of Filipino cases (Vieira et al. 2005). In the current study, we used direct sequencing covering all the coding sequences of MSX1 to determine whether MSX1 mutations might be etiological in some cases of Thai patients with nonsyndromic CL/P.

Subjects and methods

The subjects of this study were 88 sporadic cases of nonsyndromic CL/P and 12 additional cases with a positive family history. Details of their characteristics and recruitment have been previously reported (Leoyklang et al. 2006). The study was approved by the institutional review board of the Faculty of Medicine of Chulalongkorn University, and written informed consent was obtained from each person included in the study. The control samples were Thai blood donors with no oral clefts who denied history of oral clefts in other family members.

Genomic DNA was isolated from peripheral blood, according to established protocols. Primers in noncoding regions were used to specifically amplify fragments encompassing coding regions in both exons (primers 1F and 1R for exon 1 and primers 2F and 2R for exon 2; Table 1) of the MSX1 gene. Polymerase chain reactions (PCR) were carried out in a 20-μl volume containing 50 ng genomic DNA, 1× PCR buffer, 1.9–2.0 mM MgCl2, 0.2 mM dNTPs, 0.2 μM of each primer, and 0.5 U Taq polymerase, using the following parameters: 40 s at 94°C, 40 s at the annealing temperature (Table 1), and 40 s at 72°C for 35 cycles. PCR products were treated with ExoSAP-IT (USP, Cleveland, OH, USA) according to the manufacturer’s recommendations, and sent for direct sequencing to Macrogen, Seoul, Korea. Primers used for sequencing were the same as those for PCR reactions, except the primer 1RS (Table 1), which was used for sequencing exon 1 in the 3′–5′ direction. Analyses were performed using Sequencher 4.2. When the results indicated a possible new variant, the sample was resequenced. The position of variants corresponds to the coding sequence for the nucleotide position within the Genbank entry AF426432.

Table 1 Oligonucleotides and polymerase chain reaction (PCR) conditions for MSX1 mutation analysis. SNP single nucleotide polymorphism
Full size table

The P147Q variant was verified in the patients by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) using a mutagenesis primer P147Q-F and the primer 1RS (Table 1) and restriction endonuclease, DdeI. In addition, to determine whether the variant was associated with cleft lip in our Thai population, we performed an association study by genotyping 50 more patients with cleft lip with or without CL/P (bringing the total number of participants to 150) and 100 Thai controls.

Standard Chi-squared and p values were calculated by a program available at http://www.unc.edu/∼preacher/chisq/chisq.htm. Odds ratio and 95% confidence intervals (95%CI) were calculated from the Epi Info 2000 program downloaded from http://www.cdc.gov/epiinfo/.

We determined whether the P147Q variant in the Thai population was on the same haplotype as the Vietnamese by typing three single nucleotide polymorphisms (SNPs), snp1, snp4, and snp5 (SNPs were designated in accordance with a previous study, Suzuki et al. 2004) using primers SNP1-F and SNP1-R (Table 1) and BstBI for snp1; primers SNP4/5-F and SNP4/5-R (Table 1) and BseRI for snp4; and primers SNP4/5-F and SNP4/5-R and MboII for snp5, in 50 Thai controls. Haplotype frequencies were estimated by the EH program, which tested and estimated linkage disequilibrium between different markers, downloaded from http://www.linkage.rockefeller.edu/ott/eh.htm.

Both of the novel nonsynonymous coding variants, 799G > T and 832C > T, were verified by PCR-RFLP, using the primer 2F and a mutagenesis primer G267C-R (Table 1) and DdeI for the 799G > T, and primers 2F and 2R and MwoI for the 832C > T. One hundred and sixty-two Thai control individuals were also examined for the variants by restriction enzyme analysis.

For protein sequence comparisons, MSX1 orthologs were first identified through a BLAST search of the nonredundant database using Homo sapiens MSX1, accession NP_002439, as the reference sequence. All known and complete MSX1 sequences were included from the vertebrate lineage. These files in FASTA format were then analyzed by ClustalX 1.81 program. The human MSX1 was aligned with cow (accession NP_777223), Norway rat (accession NP_112321), house mouse (accession NP_034965), red jungle fowl (accession XP_444660), African clawed frog (accession AAH81101), and zebrafish (accession NP_571348). The program classified amino acids by the variation in polarity, assessing both amino acid class conservation and evolutionary conservation at any given site.

PolyPhen (http://www.bork.embl-heidelberg.de/PolyPhen/) was used to predict the effect of the nonsynonymous mutations. ESEfinder software (http://www.rulai.cshl.edu/tools/ESE/) was used to predict potential exonic splicing enhancers (ESEs; Shotelersuk et al. 2004).

Results

The sequencing effort concentrated on the coding regions of the MSX1 gene. In 100 DNA samples from subjects with nonsyndromic CL/P, 8 variant sites were identified. Seven were single nucleotide changes, comprising 3 transitions (2 in coding regions) and 4 transversions (all 4 in a coding region). The other variant was a single nucleotide deletion in intron 1 (Table 2).

Table 2 Variant sites of MSX1 found in 100 Thai patients with nonsyndromic cleft palate (CL/P)
Full size table

The coding regions of MSX1 contained 6 different variants, 2 synonymous and 4 nonsynonymous. Two nonsynonymous variants, 101C > G (A34G) and 440C > A (P147Q), were in exon 1 and have been previously reported. The A34G variant, previously reported as a nonpathogenic polymorphism (Suzuki et al. 2004), was found in 8 of our patients, 4 were heterozygous and the other 4 were either homozygous or hemizygous. Individuals who were homozygous/hemizygous for A34G have never been reported previously. The P147Q variant, previously proposed to be a pathogenic missense mutation in Vietnamese and Filipino cases (Suzuki et al. 2004; Vieira et al. 2005), was found in 3 of our 100 patients, all in the heterozygous state. Because of its relatively high prevalence in our patient group, we determined its frequencies in 50 more patients with nonsyndromic CL/P and 100 normal Thai controls. The observed frequencies of the 440C and 440A alleles, and the CC and CA genotypes in affected subjects and controls are shown in Table 3. The observed distribution of genotypes among controls was compared with that expected according to the Hardy–Weinberg equilibrium: no difference was found (χ2 = 0.007, df = 1, P = 0.93). Genotype frequencies of the patients also followed the Hardy–Weinberg equilibrium (χ2 = 0.002, df = 1, P = 0.96). The distributions of alleles and genotypes among patients were compared with those among controls: no differences of either allelic (P = 0.285) or genotypic (P = 0.277) distributions between patients and controls were found. In addition, no association was found with the CA genotype compared with the CC (odds ratio 0.56, 95% CI 0.18–1.78; Table 3). In contrast to a previous study in a Vietnamese population (Suzuki et al. 2004), the P147Q variant was not associated with cleft lip in the Thai population. Next, haplotype analysis was performed in 50 unrelated control Thai. The results are shown in Table 4. Of the 8 440A alleles found in Thai controls, 7 were on the haplotype #5 (included in the haplotype #4 of a previous study, Suzuki et al. 2004), the same haplotype as in the Vietnamese. Due to the unavailability of samples from the relatives of our CL/P patients, we did not determine the haplotype of their P147Q variants.

Table 3 Genotypic and allelic distributions and comparisons of the MSX1 440C > A in patients with CL/P and controls
Full size table
Table 4 Top 92% of haplotypes from 50 unrelated control Thai subjects
Full size table

The two nonsynonymous variants in exon 2, 799G > T (G267C) and 832C > T (P278S), have not been previously reported (Fig. 1). One variant was found in each of 2 patients; both were sporadic cases, with normal development, no anomalies besides the oral clefts, no consanguinity and no mutations in the coding region of p63 (Leoyklang et al. 2006). Clinical and molecular features of these two patients with potential mutations are shown in Table 5.

Fig. 1
figure 1

Mutation analysis. The left and right panels relate to patients 1 and 2 respectively. a Electropherograms of patients, showing 799G > T and 832C > T (arrows) in patients 1 and 2 respectively. b Electropherograms of controls showing normal genotypes at nucleotide 799 as GG and 832 CC (arrows). c Restriction enzyme digestion analysis. C1 and C2 denote controls, Mk 100-bp marker, and P patient. The arrowhead indicates the 200-bp marker. In the left panel, DdeI digested the 448-bp product of controls into 220-, 150-bp, and other smaller products (not shown). The 799G > T mutation in patient 1 creates another DdeI restriction site. Therefore, the 220-bp PCR product of the mutant allele of the patient is further digested into 181- and 39-bp (not shown) products, present with the 220-bp product of the normal allele, indicating that patient 1 is heterozygous for the 799G > T mutation. In the right panel, MwoI digested the 556-bp product of a control into 108-, 82-bp, and other smaller products (not shown). The 832C > T mutation in patient 2 eliminates a MwoI site, leaving the 120-bp product, present with the digested 108- and 12-bp (not shown) of the normal alleles, indicating that patient 2 is heterozygous for the 832C > T mutation.

Full size image
Table 5 Clinical and molecular features of patients with potential mutations
Full size table

Discussion

The sequencing analysis of the MSX1 gene found four nonsynonymous variants. Two of them, the 799G > T (G267C) and 832C > T (P278S) variants, were predicted to be “probably damaging” by PolyPhen, changed (either created or eliminated) themselves as potential ESEs for serine-arginine (SR) proteins, and not present in 162 control individuals of Thai ethnic background. Both were found in exon 2 and have not been previously reported. Exon 2 of the MSX1 was mostly conserved with significantly fewer sequence variations compared with exon 1 (Jezewski et al. 2003). These two potential mutations were found on the C-terminal side of the homeodomain protein while all of the previously reported potential missense mutations [E78V (Jezewski et al. 2003), G91D (Jezewski et al. 2003), G98E (Suzuki et al. 2004), V114G (Jezewski et al. 2003), G116E (Jezewski et al. 2003), and R151S (Jezewski et al. 2003)] were on the N-terminal side of the homeodomain protein. Moreover, for the P278S mutation, the proline 278 is conserved in MSX1 to the cow, rat, mouse, chicken, and frog protein sequence; and there is a substantial change in amino acid class, from nonpolar proline to polar serine.

Previous studies suggested that the P147Q variant is etiologically based on the strong conservation of the amino acid and the surrounding amino acids, the segregation analysis, and its absence in over 1,600 control individuals of various ethnic backgrounds (Suzuki et al. 2004; Vieira et al. 2005). Nevertheless, the fact that in some cleft families with the P147Q variant it was found in unaffected members while some affected did not carry it (Vieira et al. 2005) makes its role arguable. A previous observation suggested that it was a founder mutation in the Vietnamese population. In this study, the P147Q variant has been found in 8 out of 100 Thai controls, suggesting that it is not etiologic. In addition, an association between the variant and cleft lip in our population could not be detected (Table 3). Next, we determined whether the P147Q variant in our population was on the same haplotype as those in Vietnamese by genotyping three SNPs, which would be able to identify the four most common haplotypes in the Vietnamese (Suzuki et al. 2004). We did not genotype the snp2; therefore, haplotype #1 in our study included both haplotypes #1 and #5 of the previous study (Suzuki et al. 2004). We showed that the P147Q variant in our control Thai subjects (Table 4) was on the same haplotype as those in Vietnamese (Suzuki et al. 2004). This observation strongly suggests that this P147Q variant in both the Thai and Vietnamese populations is inherited by descent from a founder genetic change. PolyPhen predicted the P147Q variant to be benign. This evidence suggests that the P147Q is neither pathogenic by itself, nor associated with cleft lip in the Thai population.

The A34G variant was a change within amino acid class, previously reported in cases and controls, and previously proposed to be benign (Jezewski et al. 2003; Suzuki et al. 2004). The roles of the two synonymous variants and the two variants in noncoding regions need further investigation.

Cleft palate is a multifactorial disorder caused by a combination of genes and environmental interactions. These factors may contribute differently to CL/P in different populations. In the Thai population, we have shown that there were associations between cleft lip and the maternal MTHFR 677CT/1298AC genotype, with an odds ratio of 4.43 and a 95%CI of 1.33–15.10 (Shotelersuk et al. 2003), and the IRF6 820G > A SNP of the proband, with an odds ratio of 1.67 and a 95% CI of 1.13–2.47 (Srichomthong et al. 2005). We also demonstrated that the p63 mutation was responsible for approximately 1% of nonsyndromic cleft lips in the Thai population (Leoyklang et al. 2006). In this study, we have shown that MSX1 mutations cause 2% of Thai cases of nonsyndromic cleft lip. Other populations may have different percentages of contributions from these genes.

This report demonstrates that the MSX1 mutations appear to contribute about 2% of cases of nonsyndromic CL/P, consistent with two previous reports (Jezewski et al. 2003; Suzuki et al. 2004), but suggests that the P147Q variant is not pathogenic. Further studies of the full phenotypic spectrum and penetrance of MSX1 mutations may improve genetic counseling in families with mutations.