Effects of RET, NRG1 and NRG3 Polymorphisms in a Chinese Population with Hirschsprung Disease

The RET proto-oncogene was identified as a major locus involved in Hirschsprung disease (HSCR). A genome-wide association study (GWAS) and whole exome sequencing identified NRG1 and NRG3 as additional HSCR susceptibility loci. We investigated the effects of RET (rs2506030 and rs2435357), NRG1 (rs2439302, rs16879552 and rs7835688) and NRG3 (rs10748842, rs10883866 and rs6584400) polymorphisms in a Chinese population with HSCR. We assessed single nucleotide polymorphisms (SNPs) in the RET, NRG1 and NRG3 genes in a cohort of 362 sporadic HSCR patients and 1,448 normal controls using a TaqMan genotyping assay. Significant associations were found between HSCR risk and rs2506030, rs2435357, rs2439302 and rs7835688 (odds ratio [OR] 1.64, P = 1.72E-06; 2.97, P = 5.15E-33; 1.84, P = 9.36E-11; and 1.93, P = 1.88E-12, respectively). Two locus analyses of SNPs indicated increased disease risks of HSCR between NRG1 rs2439302 and RET rs2435357 or rs2506030. RET rs2506030 (GG genotype) and rs2435357 (TT genotype), in combination with NRG1 rs2439302 (GG genotype), were strongly associated with the highest risk of HSCR (OR = 56.53, P = 4.50E-07) compared with the two loci or a single SNP of either RET or NRG1. Our results support the association between genetic variation of RET and NRG1 and susceptibility to HSCR in the Chinese population.

Hirschsprung disease (HSCR, OMIM 142623) or aganglionic megacolon is one of the most common developmental disorders of the enteric nervous system (ENS) in children. It is characterized by the absence of intramural ganglion cells in the myenteric and submucosal plexuses of the gastrointestinal tract, resulting in tonic contraction of the affected segment, massive distension of the proximal bowel and intestinal obstruction [1][2][3] . This common congenital disorder, which occurs in 1:5,000 live births, shows a significant difference among ethnic groups (10, 15, 21 and 28 per 100,000 live births among Hispanics, Caucasian-Americans, African-Americans, and Asians, respectively) 4 . Based on the location of aganglionosis, the patients can be classified into three anatomical types: short-segment HSCR (S-HSCR; accounting for 80% of cases), long-segment HSCR (L-HSCR; accounting for 15-20% of cases) and total colonic aganglionosis (TCA; accounting for less than 5% of cases) 5 . HSCR also shows sex-dependent penetrance 6 and appears to be a non-Mendelian malformation with low penetrance 7 .

Results
Characteristics of study subjects. The characteristics of the 362 HSCR patients and 1,448 healthy subjects are summarized in Table 1. There were 275 (75.97%) male and 87 (24.03%) female HSCR subjects, while 1,100 (75.97%) male subjects and 348 (24.03%) female subjects were included in the healthy control group. The cases and controls were well matched regarding gender distribution (P = 1.00). All the patients were diagnosed with S-HSCR.

Association analysis of candidate polymorphisms with HSCR risk. For the eight genetic markers
at RET (rs2506030 and rs2435357), NRG1 (rs2439302, rs7835688 and rs16879552), and NRG3 (rs10748842, rs10883866 and rs6584400), a Hardy-Weinberg equilibrium test was performed on HSCR cases and controls. Genotype distributions were in Hardy-Weinberg equilibrium for the polymorphisms in both the HSCR cases and the controls (Chi-squared test, P > 0.05), and the statistical test power was > 95.0%. The distribution of the eight SNPs in the HSCR cases and controls is shown in Table 2 and Supplementary Table S1. The SNPs rs2435357, rs2506030, rs2439302 and rs7835688 showed a positive correlation with HSCR susceptibility, with an OR greater than 1.0 (P < 0.05). The risk allele frequencies for these three SNPs were significantly higher in HSCR cases than in controls (rs2439302 risk allele G: 26.9% vs. 18.6%, P = 9.36E-11; rs7835688 risk allele C: 30.1% vs. 18 Under a multivariate logistic regression model, individuals with CG and GG genotypes of rs2439302 had a significantly increased risk of HSCR with an OR of 1.49 (95% CI = 1.16-1.91, P = 2.00E-03) and 5.01 (95% CI = 3.14-8.00, P = 1.57E-11), respectively, compared to those with the CC genotype. Thus, the CG and However, no genetic correlation was found between HSCR and the three SNPs at NRG3 (rs10748842, rs10883866 and rs6584400) or the SNP at NRG1 (rs16879552) when comparing the risk allele with the non-risk allele or in the additive, recessive and dominant models. In addition, no relevancy was observed in the heterozygote or the homozygote variant (see Supplementary Table S1).
Correlations between RET and NRG1 expression in HSCR ganglionic colon tissues and genetic variants of rs2435357, rs2506030 and rs2439302. Because rs2435357 and rs2506030 are located on the same chromosome and were identified to be associated with HSCR, we calculated the linkage disequilibrium (LD) of these two SNPs. The r 2 value of these two SNPs is 0.041, which suggest they are in low association and have independent genetic effects for HSCR. The two variants at the NRG1 locus, rs7835688 and rs2439302, are in high LD (r 2 = 0.95). This result indicated that rs7835688 and rs2439302 are strongly associated and in high allelic identity. In this way, one variant shows significant association with HSCR, the other variant is expected to also show disease association by the phenomenon of LD. Therefore, we report rs2439302 as representative of the association. To explore whether any of the three genotypes in rs2435357, rs2506030 and rs2439302 were associated with RET and NRG1 gene expression, we performed western blotting and real-time quantitative reverse transcription PCR (RT-qPCR) on ganglionic colon tissues obtained from HSCR patients. Compared with rs2435357-CC or rs2506030-AA, relative RET protein expression was decreased in rs2435357-CT and rs2435357-TT or rs2506030-AG and rs2506030-GG ( Fig. 1A and C). Image analysis of western blots demonstrated a significant decrease of RET protein expression (P < 0.05) ( Fig. 1A and C). RT-qPCR revealed that RET was down-regulated in HSCR gut tissues. The relative expression of RET was lower in heterozygous and homozygous carriers of the HSCR risk allele than that in homozygous wild type subjects (P < 0.05) ( Fig. 1B and D). For rs2439302, NRG1 mRNA and protein expression was lowest in individuals homozygous for the G risk allele and highest in those homozygous for the CC genotype (P < 0.05; Fig. 1E and F).

Discussion
HSCR is a multifactorial genetic disorder with high heritability (> 80%, depending on the sex of the proband and affected sibling), high sibling recurrence risk (200-fold greater than the population), significant gender bias, and complex inheritance patterns 38 . The molecular and cellular events during ENS development in the embryonic stage may shed light on the pathogenesis of HSCR 39 . The RET gene, a key gene in type 2 multiple endocrine neoplasia 4 and familial medullary thyroid carcinoma, has been considered a major gene in mediating susceptibility to HSCR. However, loss-of-function germ-line mutations in RET do not account for all of the HSCR cases. Therefore, genetic variations in other susceptible genes may be associated with HSCR. Neuregulins (NRGs) comprise a large family of EGF-like signaling molecules that are involved in cell-cell communication. NRGs transmit their signals to target cells by interacting with transmembrane tyrosine kinase receptors of the ErbB family. NRG-ErbB interactions activate intracellular signaling cascades, which induce cellular responses such as migration, proliferation, differentiation, and apoptosis. Some studies have demonstrated that the ErbB family plays an important role in the development of ENCCs 40 . Among the four family members (NRG1, NRG2, NRG3 and NRG4), NRG1 is expressed throughout the developing nervous system. NRG3 shares the same receptor with NRG1, and these genes may work synergistically to affect the development of neuroblasts 41 .
In this case-controlled study of the Chinese population, we assayed eight polymorphisms at the RET, NRG1, and NRG3 gene loci using whole blood specimens from 362 unrelated S-HSCR patients and 1,448 controls. We confirmed the association of the genetic variants in the RET proto-oncogene (rs2506030 and rs2435357) and NRG1 (rs2439302 and rs7835688) with sporadic HSCR. Furthermore, this study provided insights into how these four SNPs interact with each other and jointly and/or synergistically increase the risk of HSCR development.
Intronic variation in rs2435357 appears to have a significant genetic association with HSCR in different ethnicities 35,36,42 , and this represents a functional site 43 . In our study, rs2435357 and rs2506030 were significantly associated with HSCR risk at the allele level and in different models (Table 2). Rs2435357 had a higher relative risk of HSCR in the allelic model (OR = 2.97). The prevalence of the rs2506030 susceptibility allele in the Chinese population is ~70% 38 , which is obviously higher than that in Africans (~8%) and Europeans (~40%). The different susceptibility allele frequencies may mean that the Chinese have a higher incidence of HSCR. The common NRG1 genetic variant rs2439302 was initially reported by Gudmundsson et al. 33 . Here, we present its correlation with HSCR for the first time. NRG1 encodes the signaling protein neuregulin 1 and is involved in the development of the nervous system and organs. It has been reported that each G allele confers a 40% decrease in NRG1 expression compared to the C allele. Therefore, relative NRG1 expression is significantly lower in GG carriers compared with CC carriers. It is reasonable to consider that rs2439302 may affect NRG1 expression. This result may provide a basis for future studies on HSCR. In our study, logistic regression analysis revealed an increased risk associated with CG and GG carriers compared with CC carriers, and a significant association of rs2439302 and HSCR was found in dominant, recessive, additive and allelic models. Our results demonstrated that the rs7835688 SNP increased the risk for HSCR in our subjects. These associations do not agree with those in previous reports 38,44 , perhaps due to different races or anatomical types. Although some results support the hypothesis that the NRG1 rs16879552 variant is a genetic risk factor for HSCR, the results in our study are inconsistent with these observations 26,32 . However, our findings correspond with those from studies in a Chinese population 37 and Asian ancestry subjects 45 . The contradictory results may be due to the use of different samples from different population. We could not find any significant influence of variations in the three NRG3 polymorphisms tested on the incidence of HSCR. In addition, the risk of HSCR was higher among patients with the heterozygote genotype and in the additive and dominant models (Table 2). However, none of these results reached significance in the present study. These findings suggest that NRG3 polymorphisms do not confer a significant risk for HSCR. However, the results need to be confirmed with a larger patient population or a multicenter study.
The two variants rs2435357 and rs2506030 were identified as disrupting enhancers of RET, and they may play a similar role in affecting RET expression. The polymorphism rs2439302 at NRG1 was shown to influence NRG1 expression. However, with respect to the variations in these polymorphisms, there was a lesser influence on gene expression. Our western blotting and RT-qPCR results demonstrated that individuals carrying the risk genotypes had the lowest protein and mRNA expression levels of RET and NRG1 in HSCR ganglionic colon tissues (Fig. 1). These results indicated that RET and NRG1 protein and mRNA levels correlated with the different genotypes of the SNPs. It has been suggested that NRG1 and RET play important roles in the pathogenesis of HSCR, and some data support the hypothesis that rs2506030 and rs2435357 variants disrupt RET enhancer activity 38 . However, further research is needed to elucidate how deficient expression results in aganglionosis.
In some previous studies, the cross-gene or cross-variant (common variants/rare variants) associations suggested joint effects or interactions with each other when certain variants coexist 10,23 . In the present study, we explored the interaction between RET and NRG1 by analyzing combinations of SNPs involving these two genes. We found that these genetic interactions significantly enhanced the risk of HSCR in the study population. More importantly, compared with individuals whose genotype contained neither risk allele, the OR for the NRG1 rs2439302 risk homozygote (GG) was increased by 25.69-fold in presence of the RET rs2506030 risk genotype (GG) and by 25.57-fold in the presence of the RET rs2435357 risk genotype (TT). RET and NRG1 variants also showed genetic interactions with genotypes at three loci, revealing enhanced HSCR risk (OR = 56.53). Because of the high LD between the two variants rs7835688 and rs2439302 at the NRG1 locus, the joint effects between rs7835688 and rs2435357 and/or rs2506030 were basically the same as those described above. Although functional assays supported genetic interactions between NRG1 and RET that could be translated into functional interactions, the underlying molecular mechanism is still not well understood, and therefore, more sophisticated biological analyses are necessary to explain the complex interactions between these genes.
Despite the sufficient power of the comprehensive analysis in our study, several limitations should be acknowledged. First, the sample size of our case-controlled study was relatively small. Second, HSCR is a complex disease,  affected by interactions between genetic and environmental factors. The unavailability of adequate clinical information limited our ability to further investigate the interactions.
In conclusion, our study showed for the first time that the functional common genetic variant rs2439302 in NRG1 might confer altered risk to HSCR in the Chinese population and provided genotype-phenotype correlations regarding RET and NRG1 expression levels in HSCR. We also demonstrated the existence of cumulative genetic risk for HSCR between RET and NRG1 polymorphisms in the Chinese population. These results may provide a basis for further biological and/or molecular studies of HSCR.

Ethics statement. This research project has been approved by the Ethical Committee of the Medical
Association of Tongji Medical College, Huazhong University of Science and Technology. Informed consent was obtained from all participants or legal guardians.

Subjects.
We selected a total of 362 patients with sporadic S-HSCR from Wuhan Union Hospital and 1,448 healthy individuals as controls who were randomly recruited from the general population in the Wuhan region of Hubei Province, China. The diagnosis of HSCR was based on histological examination of either biopsy tissues or pathology specimens obtained during operations. A 2 ml sample of peripheral venous blood was collected from the patients and the controls, and the blood specimens were mixed with the anti-coagulant ethylene diamine tetraacetic acid (EDTA). Then, genomic DNA was extracted from the white blood cells using a QIAamp DNA Blood Midi Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocols. DNA concentrations were determined with a UV spectrophotometer, and the samples were stored at − 80 °C until use. Full-thickness ganglionic colon tissues were obtained from patients who were pathologically confirmed to have HSCR and were stored at − 80 °C.

RNA isolation and RT-qPCR.
Statistical analysis. The statistical analyses were performed with SPSS version 21.0 (SPSS Inc., Chicago, IL, USA). The Hardy-Weinberg equilibrium of the three SNPs was assessed with a goodness of fit Chi-squared test in the control subjects. To analyze the frequency of alleles and genotypes in all samples, the Chi-square test was used. The ORs and 95% CIs were determined for the association between disease status and genotype using a Chi-squared test, and two-tailed P-values less than 0.05 were considered significant. Dominant, recessive and additive models for the three SNPs in association with HSCR were analyzed to avoid the assumptions of genetic models. The wild type was considered as a reference. The genetic comparisons included a homozygous model, heterozygous model, dominant model and recessive model. The combined effects of NRG1 (rs2439302) and RET SNPs (rs2506030 and rs2435357) on HSCR risk were assessed. ORs and the frequency of the SNP alleles were calculated for disease markers. We used Haploview 4.2 for tests of LD determined by the r 2 value. For SNP-SNP interactions, we used a multiple logistic regression model to estimate the multiplicative interaction effect of the SNPs. All data are shown in Fig. 1 as the mean ± standard error of the mean (SEM).