Association between ABHD1 and DOK6 polymorphisms and susceptibility to Hirschsprung disease in Southern Chinese children

Abstract Hirschsprung disease (HSCR) is an infrequent congenital intestinal dysplasia. The known genetic variations are unable to fully explain the pathogenesis of HSCR. The α/β‐hydratase domain 1 (ABHD1) interferes with the proliferation and migration of intestinal stem cells. Docking protein 6 (DOK6) is involved in neurodevelopment through RET signalling pathway. We examined the association of ABHD1 and DOK6 genetic variants with HSCR using 1470 controls and 1473 HSCR patients from Southern Chinese children. The results clarified that DOK6 rs12968648 G allele significantly increased HSCR susceptibility, in the allelic model (p = 0.034; OR = 1.12, 95%CI = 1.01~1.24) and the dominant model (p = 0.038; OR = 1.12, 95%CI = 1.01~1.25). Clinical stratification analysis showed that rs12968648 G allele was associated with increased risk of short‐segment HSCR (S‐HSCR), in the allelic model (p = 0.028; OR = 1.14, 95%CI = 1.01~1.28) and the additive model (p = 0.030; OR = 1.14, 95%CI = 1.01~1.28). ABHD1 rs2304678 C allele had higher risk to develop total colonic aganglionosis (TCA) in the allelic model (p = 7.04E‐03; OR = 1.67, 95%CI = 1.15~2.43) and the dominant model (p = 4.12E‐03; OR = 1.93, 95%CI = 1.23~3.04). DOK6 rs12968648 and ABHD1 rs2304678 had significant intergenic synergistic effect according to logical regression (p = 0.0081; OR = 0.76, 95%CI = 0.63~0.93) and multifactor dimensionality reduction (MDR, p = 0.0045; OR = 1.25, 95%CI = 1.07~1.46). This study verified two susceptible variations of HSCR on ABHD1 and DOK6. Their roles in HSCR should be conducted in further studies.

HSCR susceptibility, in the allelic model (p = 0.034; OR = 1.12, 95%CI = 1.01~1. 24) and the dominant model (p = 0.038; OR = 1.12, 95%CI = 1.01~1. 25). Clinical stratification analysis showed that rs12968648 G allele was associated with increased risk of short-segment HSCR (S-HSCR), in the allelic model (p = 0.028; OR = 1.14, 95%CI = 1.01~1. 28) and the additive model (p = 0.030; OR = 1.14, 95%CI = 1.01~1. 28). ages of approximately 80%, 15% and 5%, respectively. 2,3 There is an ethnic difference in the incidence of HSCR, with an average of one case per 15,000 European live births; however, the incidence of HSCR is three times higher in Asian populations (approximately 1 in 5000 live births). 4 In addition, HSCR is divided into familial HSCR and sporadic HSCR, according to familial genetics. There are family aggregation and more sporadic cases in more than one-fifth of HSCR patients. 5 Mainly caused by genetic factors, the occurrence of HSCR is related to the heterogeneity of alleles. 6 There are more than 24 HSCR-related genes, most of which are related to ENS signalling pathways, including RET, PHOX2B and NRG1. 7,8,9 However, the pathogenesis of HSCR cannot be fully explained by these genetic variations.
As a neuronal adaptor protein molecule, docking protein 6 (DOK6) is involved in multiple neurotrophic factor-mediated neurite outgrowth and serves as a substrate for multiple tyrosine kinase receptors. 10 In addition to being critical for neurodevelopment, DOK6 is highly expressed in the developing nervous system. DOK6 is reported to be involved in the RET signalling pathway and plays an important role in cell proliferation, survival, migration and invasion. 11 Previous genome-wide association (GWAS) studies on HSCR showed that rs12968648 had a moderate significance level (p = 1.12 × 10−4; OR = 1.733, 95%CI = 1.31~2.27). In addition, it also has high biological relevance. 12 The single nucleotide polymorphism (SNP) rs12968648 in DOK6 was replicated and has been used to calculate the risk value of HSCR.
α/β-Hydrolase domain 1 (ABHD1) is the human homolog of and has the closest similarity to AlkB. AlkB family proteins are enzymes that can repair alkylated RNA and DNA through oxidative demethylation. 13 By mediating the transcription of methionine (Met), ABHD1 exerts influences on the synthesis of normal intestinal proteins and the proliferation and migration of intestinal stem cells (ISCs). In ABHD1 (−/−) knockout mice, it was confirmed that the only phenotype reported relates to deficiencies in placental trophoblast lineage differentiation. 14 The SNP rs2304678 in ABHD1 is a missense mutation according to HaploReg database (https:// pubs.broad insti tute.org/mamma ls/haplo reg/haplo reg.php), which suggests rs2304678 is likely to be functional. The SNP rs2304678 in ABHD1 was replicated to explore the correlation of HSCR in this study.
There are uncertain risks of the candidate genes, especially the interactions of these susceptible genes. This study was designed to decide whether the genetic polymorphisms of DOK6 (rs12968648) and ABHD1 (rs2304678) were linked to HSCR in 1473 Chinese HSCR patients and 1470 controls. In our study, there is the first manifestation of two susceptibility genes in the epistatic association of ABHD1 and DOK6 with HSCR.

| Study subjects
This study has been authorized by the institutional review committee  Table   S1). Blood samples of 1470 people without a history of neurological 2.2 | Single nucleotide polymorphisms genotyping and quality control Based on previous extensive studies, ABHD1 rs2304678 and DOK6 rs12968648 were selected. Two SNPs were genotyped on all samples using the iPLEX Gold MassARRAY system (Sequenom). Hardy-Weinberg equilibrium experiment was carried out to exclude SNPs (p < 0.05). The methods of SNP quality control included that: firstly, the patient/control was excluded from the final analysis if there was over 10% of the missing data of SNPs; secondly, all subjects with 10% of missing follow-up calls were deleted. At last, four SNPs were retained for further analysis.

| Subphenotype analysis and association analysis
The correlation between SNPs and disease was analysed by comparing the hazard of allele frequency (allele testing) between controls and patients, and other measurements were determined using an additive measurement of logistic regression, dominant and recessive model testing in PLINK 1.9. 15,16 The association of subphenotype stratification was analysed by comparing certain subphenotypes in the patients and controls.

| Genetic epistasis
The genetic association was estimated using the univariate logistic regression analysis. Multifactor dimensionality reduction (MDR) was applied for the statistical analysis of gene-gene interactions. The open-source MDR package 2.0 was used to implement MDR analysis, and this software can be obtained online for free (http://www. Epist asis.org). 17 Each cell in the 3 × 3 interaction table was assigned as having a low or high risk, and the genotype dimension was decreased from N to 1 through assembling these genotype combinations into two-level variables. Permutation tests and cross-validation (CV) were used to assess the ability of one-dimensional genotype variable quantities to predict and classify the disease state. 18 The Bold values indicate p < 0.05 and are considered significant.
ALLELIC, DOM, REC and ADD indicate the association test following allelic genetic, dominant, recessive and additive models. The p value indicates the significance based on allelic association tests. P adj : p value adjusted by gender and age; the calculation of odds ratio (OR) is also based on the risk allele of each SNP. OR adj , OR adjusted by gender and age.
(IS). The significance level of the Bonferroni correction (p < 0.008) was employed to carry out single-locus analysis correction for multiple tests. Significantly statistical differences were found at p = 0.05 in other two-tailed tests.
The epistasis test of logistic regression analysis (case-control) was used to analyse the parameters of genetic interaction via PLINK 1.9. 19,20 In PLINK, a model was used based on the allele dose of 0 to 2 representing the number of risk alleles of each SNP, SNP A and SNP B, and it was matched in the layout of Y

| Association of DOK6 and ABHD1 SNPs with HSCR
On DOK6 and ABHD1, one SNP was chosen for replication. The method shows the selection criteria. In Table 1, there is information on two SNPs genotyped with 1473 HSCR patients and 1470 controls from South China. The association between rs2304678 in ABHD1 and HSCR disease has not be replicated in our population (Table 1).
However, the SNP rs12968648 in the DOK6 gene was increased HSCR susceptibility (

| Intergenic SNPs show an epistatic effect on HSCR
Studies have shown that the associated SNPs increase the risk of HSCR disease by epistatic interactions. Epistatic test by logistic regression was adopted for parametric analysis using PLINK in current study. In Table 3 Table 3 (right top). In Figure 1, generated by MDR anal-

| DISCUSS ION
Hirschsprung disease is a complex congenital disease whose pathogenesis is related to several genes. An attempt was made to find extra loci associated with HSCR by designing an association study for the case-control or trio study. In this study, two SNPs were identified on ABHD1 and DOK6 and related to HSCR using the largest population-based study of HSCR with 1470 patients and 1473 controls. The DOK6 rs12968648 G allele significantly increased HSCR susceptibility ( Table 1)  increased risk of S-HSCR. In addition, the ABHD1 rs2304678 C allele was further replicated had higher risk to develop TCA (Table 2). It is worth noting that the interaction between SNPs confirmed by logistic regression analysis and MDR analysis indicated a significant interaction between the genotypes of the SNPS of ABHD1 rs2304678 and DOK6 rs12968648 (Table 3).  the standard AUG (Met) codon but also recognizes AUA (Ile) in the process of translation initiation and elongation. 34  Additionally, two SNPs are limited to elucidate the relationship between genes and disease. Therefore, more polymorphisms should be involved.
In summary, a systematic analysis of two candidate genes identified that the DOK6 rs12968648 G allele significantly increased the susceptibility in HSCR, especially S-HSCR, and the ABHD1 rs2304678 C allele increased the risk of severe TCA. This study indicated that the risk of HSCR was increased through the genetic interaction of relevant variants between DOK6 and ABHD1. This finding may determine DOK6 and ABHD1 as genes related to HSCR, improve the understanding of the aetiology of HSCR, but further studies are needed to verify their roles in HSCR.

ACK N OWLED G EM ENTS
We thank Yanlu Tong and Hezhen Wang for their assistance in

CO N FLI C T O F I NTE R E S T S
The authors declare that the study was performed without any financial or commercial relations that could be interpreted as poten-

DATA AVA I L A B I L I T Y S TAT E M E N T
All data included in this study are available upon request by contact with the corresponding author.