2. Materials and Methods
2.1. Study participants and baseline study
In the Hokkaido Birth Cohort Study on Environment and Children’s Health, the participants of 20,296 pregnant women were enrolled from 2003 to 2012 in Hokkaido, Japan. During the first trimester, participants completed the baseline questionnaire on maternal and paternal characteristics. Maternal peripheral blood samples were taken during pregnancy without fasting. Medical birth records from the delivery hospital were collected for birth weight, height, sex, and other medical conditions (Kishi et al., 2013; 2017).
The institutional ethics board for epidemiological studies at Hokkaido University Graduate School of Medicine and Hokkaido University Center for Environmental and Health Sciences approved the study protocol (approval number 69). In this study, informed consent was obtained from all study participants before enrollment.
2.2. Follow-up study until age 7 and saliva samples collection.
We have collected data regarding the wheezing symptoms of children at 1, 2, 4, and 7 years of age using a modified section of the Japanese version of the International Study of Asthma and Allergies in Childhood (ISAAC) Phase Three questionnaire (Asher et al., 1995). We defined childhood wheeze as a maternal positive answer to the question ”Has your child had wheezing or whistling in the chest in the past 12 months?,” based on each ISAAC questionnaire administered when the child was 1, 2, 4, and 7 years of age (Figure 1). Moreover, we collected information child’s history of infectious diseases (Goudarzi et al., 2018).
In a case-control study, we selected all 314 children with wheezing and randomly selected 374 control children without any allergic symptoms (extraction rate 24%) from the ISAAC questionnaire at 7 years age. Salivary samples (2 ml) were collected from children of case and control using The Oragene® OG-300 DNA Self-Collection kit (DNA Genotek Inc., Ottawa, Ontario, Canada). Eventually, we used 275 salivary samples to analyze DNA methylation (Figure 1).
2.3. Quantification of DNA Methylation
Genomic DNA was extracted from saliva using a Maxwell 16 DNA Purification Kit (Promega, Madison, WI, USA). The DNA (500 ng) was then subjected to bisulfite conversion using Epitect Plus Bisulfite Kit (Qiagen, Venlo, The Netherlands). Bisulfite pyro sequence was performed as described previously (Murphy et al., 2012; Bollati et al., 2007). We analyzed DNA methylation in genes related to asthma for five of the CpG sites of IKZF3 (cg16293631, cg13432737), ORMDL3(cg02305874, cg14647739), and GSDMB (cg12360886). Pyrosequencing was performed using Pyromark Q24 system (Qiagen) and data were analyzed using the Pyro Q-CpG Software (Qiagen). Eventually, we completed the quantification of DNA methylation only three CpG sites of ORMDL3(cg02305874), and IKZF3 (cg16293631, cg13432737). Other PCR primers of two CpG sites including GSDMB (cg12360886) andORMDL3 (cg14647739) did not work. Conditions of primers were described in Supplementary Table 1. Average methylation levels of each CpG site that were analyzed in duplicate were used in statistical analyses.
2.4. Maternal folate and cotinine measurements
Folate is quantified by direct chemiluminescent acridinium ester technology. This technique has an acceptable imprecision of less than 10.0%, with an advanced Quality Control package. It has an analytical sensitivity of 0.91 nmol/l. Specimen preparations, shipping, and assays were performed in batches, depending on new recruitments. We have previously reported the significance of the inverse association between maternal folate and cotinine levels (Kishi et al., 2013; Yila et al., 2012). The details of serum folate and plasma cotinine measurements are described in our previous report (Sasaki et al., 2011).
2.5. Statistical analysis
We analyzed the association between participant’s characteristics and maternal folate level using Mann–Whitney’s U-test and rank correlation test. In logistic regression analysis, we evaluated odds ratios (ORs) and 95% confidence interval (CI) of childhood wheeze, according to maternal folate level, respectively. Because a biologically relevant threshold of folate on wheezing is not elucidated, we used two types of maternal folate levels as follows; first, ORs and 95% CI for the risk of wheezing were calculated for three categorical maternal folate levels in the folate suboptimal (6.80–13.59 nmol/l) or optimal (≥ 13.60 nmol/l) and compared to those in folate deficiency (< 6.80 nmol/l) with reference to the WHO guideline (Sauberlich, 1999). Second, ORs and 95% CI for the risk of wheezing were calculated for 4 categorical maternal folate levels in the second, third, and the fourth quartiles and compared to those in the lowest quartiles as reference by the distribution of their quartiles. For the calculation of P for trend, we handled categorical values of maternal folates as ordinal variables.
In a case-control study, we used two types of maternal folate levels, including categorical of quartile, and two categorical with reference to the WHO guideline, which were optimal (≥ 13.60 nmol/l) or under suboptimal category (< 13.60 nmol/l) (Supplementary Table 2). In liner regression analysis, we evaluated β and 95% CI for DNA methylation, including, three CpG sites of IKZF3 (cg16293631, cg13432737) and ORMDL3 (cg02305874), according to maternal folate levels. In logistic regression analysis, we evaluated odds ratios (ORs) and 95% CI of childhood wheezing at 7 years age according to DNA methylation.
In both logistic and liner regression analysis, we used adjusted factors as potential confounders, including maternal age, parity, delivery year, alcohol drinking during pregnancy, log10-transformed cotinine levels, maternal allergic history, paternal allergic history, annual household income, and sex of child (model 1). Furthermore, to confounding variables of the model 1, we used adjusted factors which were collected after birth, including current maternal and paternal smoking status, breast feeding, and day care attendance (yes, no) (model 2). Pvalue of less than 0.05 was considered statistically significant. All statistical analyses were performed with SPSS software for Windows (version 21.0J; IBM, Armonk, NY, USA).