Polymorphisms at 1 q 32 , 8 q 24 , and 17 q 22 loci are associated with nonsyndromic cleft lip with or without cleft palate risk in the Slovak population

Background. Nonsyndromic cleft lip with or without cleft palate (nsCL/P) is the most common orofacial birth defect with an aetiology involving both genetic and environmental factors. Genome-wide association studies (GWAS) have identified several genomic susceptibility regions for nsCL/P. In the present study, the three well established single nucleotide polymorphisms (SNPs) identified by GWAS (rs987525 at 8q24, rs7078160 at 10q25, and rs227731 at 17q22 loci) and one SNP identified by candidate gene study (rs642961 in IRF6 gene at 1q32 locus) were analysed for an association with nsCL/P in Slovak population. Methods. Nucleotide variants were genotyped in 165 nsCL/P patients and 326 unaffected controls. All variants of interest were genotyped using high-resolution melting analysis after real-time PCR. Results. We found significant differences between patient and control groups with respect to the allele and genotype frequencies for the SNPs at the 1q32, 8q24, and 17q22 loci. SNP at the 10q25 locus showed a trend toward association with nsCL/P risk. Conclusions. The results suggest that SNPs at the 1q32, 8q24 and 17q22 loci may contribute to the nsCL/P risk in Slovak population.


INTRODUCTION
Nonsyndromic cleft lip with or without cleft palate (nsCL/P), the most common type of orofacial clefting, is one of the most frequent birth defects.2][3][4] ).
NsCL/P impose a large burden on the health and health-related quality of life, and represent a challenge to patient care 5 .The aetiology of nsCL/P is multifactorial, including both genetic and environmental components 6 , though the specific genetic and environmental factors remain largely unclear.A variety of genetic approaches have been used to identify genes and pathways underlying nsCL/P.Before 2009, only one gene (interferon regulatory factor 6, IRF6) had been identified with a sufficient degree of consistency across studies, thus being considered a true nsCL/P-associated gene [7][8][9] .Association of nsCL/P with single nucleotide polymorphism (SNP) rs642961 in the IRF6 gene located at chromosomal region 1q32 was confirmed in candidate gene and genome-wide association studies (GWAS) of European-derived populations [10][11] .2][13] ), principally tagged by SNP rs987525.Additional significant GWAS signals in the European populations 14 were detected at loci 10q25 and 17q22, SNPs rs7078160 at 10q25 and rs227731 at 17q22 being the strongest susceptibility markers.
In the present case-control study, four SNPs (rs642961 at 1q32 (IRF6 gene), rs987525 at 8q24, rs7078160 at 10q25, and rs227731 at 17q22) that have been reported to be significantly associated with nsCL/P in European populations were analysed for an association with this birth defect in Slovak population.

Study subjects
Patients (165 Caucasians) were recruited at the Departments of Paediatrics and Department of Plastic, Reconstructive and Aesthetic Surgery in Košice and Bratislava.Cases with congenital anomalies or developmental delays that could reflect a malformation syndrome other than nsCL/P and patients with syndromic orofacial clefts were excluded.
The population control group was composed of 326 unselected and unrelated newborns (Caucasians) from the University Hospital in Košice.The newborns were free from congenital anomalies and had no family history of clefting or other major congenital anomalies.All parents signed an informed consent.The study was approved by the ethics committees of the contributing centers.
Probes (mismatch-tolerant probes capable of detecting allele-specific targets) were blocked at the 3'-end by phosphate to prevent extension.For all four SNPs the amplification was immediately followed by melting analysis starting with denaturation at 95 ºC for 30 s and renaturation at 40 ºC for 1 min.Data were acquired over 45-95 ºC range.Genotypes were identified using Eco TM Software 4.1 (Illumina, Inc., San Diego, CA, USA) by melting temperatures indicated by probe peaks on derivative plots.Genotyping success rate for all examined variants was 100%.The sequencing on ABI 3500 Genetic Analyzer (Applied Biosystems Inc., Forster City, CA, USA) was used to verify the results of genotyping, with 100% concordance reached (15 samples for each SNP; 5 samples of each genotype were used).For all SNPs the genotype frequencies were in agreement with those predicted under the conditions of Hardy-Weinberg equilibrium.

Statistical analysis
Differences in allele distributions between the case and the control groups (Table 1) were analyzed using the two by two table test (http://www.openepi.com;ref. 15 ).All P values were two-tailed with a significance level of 0.05.For allele distribution Bonferroni correction (multiple comparisons), a corrected P<0.0125 (0.05/4) was considered statistically significant.Odds Ratio (OR) calculations and 95% confidence intervals (CIs) were used to estimate the relative risk of nsCL/P by comparing allele and genotype frequencies between cases and controls.
The HWE test, genotype frequencies, and association analysis under different inheritance models (Table 2) were conducted using the web-based association study application program SNPStats (http://bioinfo.iconcologia.net/SNPstats) (ref. 16).The genetic models tested were as follows: codominant model Dd vs DD and dd vs. DD (d is the minor allele), dominant model dd + Dd vs. DD, recessive model dd vs. Dd + DD and additive model: effect of allele dosage.The aim was to choose the best covariate model for the gene effect.The Bayes information criterion (BIC) (ref. 17) and the Akaike information criterion (AIC) (ref. 18) were used; the smaller AIC and BIC, the better was the model.For evaluation of different genetic models a Bonferroni corrected P<0.003125 (0.05/16) was considered statistically significant.

RESULTS
Table 1 shows the results of our association analysis for the allelic distributions and Table 2 for genotype distributions with respect to nsCL/P risk.
The allelic frequencies for all the tested SNPs in control group (Table 1) did not differ significantly from an- other publicly available data on Caucasian populations (http://www.ensembl.org)and were also similar to the frequencies reported from other case-control studies conducted on Central European populations 10,12,19,20 .With respect to allele distributions, the significant associations with nsCL/P risk, persisting even after Bonferroni correc-tion for multiple comparisons, were found for SNPs at the 1q32, 8q24 and 17q22 loci.Regarding genotype distributions, significant association was similarly observed for SNPs at the 1q32, 8q24 and 17q22 loci for all the tested genetic models.We found that in the additive model, the association of rs642961 at 1q32 with nsCL/P risk sur-vived the correction for multiple comparisons (Table 2).SNP rs7078160 at the 10q25 locus reached borderline significance, manifesting a trend toward association with nsCL/P susceptibility in the Slovak population (Table 1  and 2).The alleles associated with an increased risk of nsCL/P in this study (allele A in rs642961, A in rs987525, A in rs7078160, and C in rs227731) corresponded to the risk alleles identified in previous investigations.

DISCUSSION
In the present study, a selection representing the four significant SNPs identified by previous GWAS was examined for an association with nsCL/P in an independent sample of the Slovak population.
The first conclusively identified genetic risk factor for orofacial clefts was the interferon regulatory factor 6 (IRF6) gene located at chromosomal region 1q32 (ref. 7).Subsequent analyses identified a causative variant rs642961 in the IRF6 promoter region which disrupts the binding site of the transcription factor AP-2a involved in craniofacial development 8 .In this report, we found that rs642961 was associated with nsCL/P risk in all of the models of inheritance tested, with a 1.6 to almost a threefold increase in risk.The most significant results, persisting after Bonferroni correction, were observed when the additive model was used, confirming dosage risk effect of minor allele A. Positive association between the IRF6 rs642961 variant and nsCL/P has been confirmed by multiple studies involving various ethnicities [21][22][23][24] .Also, our results are in accordance with the findings from two case-control studies on Central European populations which showed similar magnitudes of odds ratios 10,19 .6][27].
Another finding of our study was an association of the SNP located at the 8q24 locus with nsCL/P.The rs987525 showed nominal significance in all of the models of inheritance tested, with a 1.5 to more than twofold increase in risk.Similar results for rs987525 were obtained in a study on the Central European sample which showed a 1.7fold increase in the risk of nsCL/P under the dominant inheritance model 19 .Although the association between rs987525 and the risk of nsCL/P has been replicated in numerous studies on various ethnicities [28][29][30][31] , the underlying functional mechanism remains unknown as rs987525 resides in a region of the genome containing no known genes 12 .It is possible that rs987525 or a linked causal SNP affects tissue-specific enhancers, altering the expression of one or more unspecified genes.Known genes mapped to the area near the 8q24 gene desert include MYC and GSDMC, encoding myc proto-oncogene protein, and melanoma-derived leucine zipper-containing extranuclear factor, respectively.A recent study of murine embryos found that the orthologous 280-kb region adjacent to human rs987525 harboured potential craniofacial enhancer elements, suggesting the presence of a regulatory effect 32 .
For the variant rs227731 at 17q22 locus, in the present sample we found a significant association with nsCL/P in all of the genetic models tested, with the OR values ranging from 1.4 to 2.1.Similar values were reported from a study by Mostowska et al. 20 on another Central European (Polish) population.Evidence for the involvement of this locus in the nsCL/P aetiology was originally reported by GWAS of Mangold et al. 14 in the European-derived population.However, the subsequent replication studies conducted in the Mesoamerican, Italian, Chinese Han and Kenyan populations did not confirm these positive results 21,31,33,34 .The nucleotide variant rs227731 is located 100kb centromeric of the NOG gene.NOG acts as antagonist of members of the TGF-b superfamily, which have been implicated in orofacial clefting 35 .Recently, Leslie et al. 36 identified a common causative variant rs227727 near the NOG gene, which leads to disruption of its enhancer activity.This SNP was in complete linkage disequilibrium with rs227731.
Finally, the remaining tested SNP rs7078160 at the 10q25 locus approached significance, manifesting only a trend toward association with nsCL/P susceptibility in the Slovak population.SNP rs7078160 is intergenic, located near the VAX1 gene encoding a transcriptional regulator with a DNA-binding homeobox domain.An association between rs7078160 and nsCL/P has been replicated in several studies, which included European and Mesoamerican samples 20,21,37 , but Chinese and Kenyan studies 33,34 have reported contradictory results.A meta-analysis of 7 eligible studies 38 showed only a modest association of rs7078160 with nsCL/P risk.This also revealed that the magnitudes of the associations varied among the white, the Asian, and mixed populations, which may suggest that ethnic heterogeneity and environmental factors may have different impacts with respect to the association between 10q25 rs7078160 and the risk of nsCL/P.

CONCLUSION
Genetic studies of complex traits in populations are important, both to define common and population-specific risk alleles, and also to understand the underlying biology 6 .Recent studies have reported association of nsCL/P with many loci throughout the genome [11][12][13][14][39][40][41][42] . Repliction of those results on various populations and the investigation of mechanisms by which these variants influence cleft susceptibility are crucial for better understanding of genetic architecture of nsCL/P.
The present study replicated nsCL/P risk loci 1q32, 8q24, and 17q22 in a new population.It also suggested that the 10q25 locus might be implicated in the risk of nsCL/P in the Slovak population, although it failed to reach significance.The study, however, has some limitations such as a relatively small sample size in the case group resulting in limited statistical power, and a lack of the data related to potential environmental risk factors.Further studies on various populations are needed to replicate GWAS results to identify population-specific nsCL/P genes/loci, as well as to find etiologic nucleotide variants and biologic mechanisms conferring susceptibility to orofacial clefts.

Table 1 .
Association of SNP allele distribution with nsCL/P and OR estimation in the Slovak population.Odds Ratio for the minor allele as the target one and the major allele as the reference nsCL/P, nonsyndromic cleft lip with or without cleft palate; SNP, single nucleotide polymorphism; MA, minor allele; MAF, minor allele frequency; CI, confidence interval *Significant P-values after Bonferroni correction for multiple testing (P < 0.0125).a

Table 2 .
Association of SNP genotype distribution with nsCL/P and OR estimation in the Slovak population.