A cross-sectional study was conducted in Wuzhi County of Henan Province in China by random sampling in 2013. Participants were excluded as the following: (1) people with type 1 diabetes; (2) secondary diabetes (drug-induced, chemical-induced, exocrine pancreatic insufficiency, and genetic defects); (3) gestational diabetes and rare forms of diabetes. Finally, a total of 1,143 local permanent residents were recruited in this study. The project was approved by the Institutional Review Board at Zhengzhou University. All participants were informed of the purpose of the study and provided written informed consent.

The diagnostic testing for T2DM was performed according to the criteria of World Health Organization (1999) (20) and the guidelines of American Diabetes Association (2002). All the nondiabetic people had normal glucose tolerance after an oral glucose tolerance test (OGTT).

Trained investigators conducted a face-to-face interview with each participant using a standard questionnaire for obtaining information of demographic characteristics including age, gender, economic status, educational level, dietary habits and lifestyle (smoking and drinking, salt intake, physical activity, et, al). Among participants, those who had smoked at least 100 cigarettes in their lifetime were defined as “smoking”; those consuming any drink containing alcohol more than 12 times during the past 12 months were defined as “drinking.” A total of 10 mL fasting blood samples (5 ml of anticoagulative and 5 ml of non-anticoagulative) were collected from each participant. Serum samples were isolated from non-anticoagulative blood after centrifugation (3,000 rpm for 15 min) at 4°C and frozen at −80°C for subsequent analyses.

The concentration (mmol/L) of HDL-C, LDL-C, TG, total cholesterol (TC) and FPG in serum samples were measured with direct method of catalase clearance, direct method of surfactant removal, glycerol phosphate oxidase-peroxidase (GPO-PAP), cholesterol oxidase-peroxidase (GHOD-PAP) and glucose oxidative method, respectively (21). Finally, 120 serum samples were randomly selected for repeated measurements. All analyses were run on an automatic biochemical analyzer (Kehua Bio-engineering Co., Ltd, Shanghai, China).

The genomic DNA was extracted from whole blood samples using a BioTeke Magnetic beads kit (Bioteke Crporation, Beijing, China). The concentration of DNA samples was measured using a Nanodrop ND-2000 spectrophotometer (Thermo, MA, USA). Subsequently, the genomic DNA were treated with sodium bisulfite using an EZ DNA Methylation-Gold kit (Zymo Research, CA, USA). The putative promoter sequences of ESRα the sequence of the gene promoter region assumed to be 2,000 bp upstream from the ESRα start codon (22)] were obtained from UCSC/Ensembl, and then primer sequences (methylated specific primers: L, 5′-CGT AGG TTT ACG GTT AGA TCGG-3′; R, 5′-ATA CAA TAA CAT CAA CGA ACT CGAA-3′; unmethylated specific primers: L, 5′-ATG GTT AGA TTG GTT TTT TTT TAGG-3′; R, 5′-ACA TCA ACA AAC TCA AAA ACA CACT-3′) were designed using the methylation primer design software (Methyl Primer Express v1.0). The ESRα methylation level was analyzed using quantitative methylation-specific PCR on a MX3000P real-time PCR system (Aglient, Santa Clara, CA, USA). PCR amplification was performed in a 15 μl reaction mixture contained 5.5 μl of diluted DNA template (100 ng), 7.5 μl of 2 × Power SYBR Green PCR Master Mixture (CWBIO, Beijing, China), 2 μl of primer with a concentration of 1.25 μmol/L each. The PCR cycling parameters were as follows: 95°C for 10 min; 40 cycles for degeneration at 94°C for 15 s, annealing at 54°C for 30 s, and extension at 72°C for 30 s. Two negative controls (replace the DNA template with ddH₂O) were set for each plate. The level of DNA methylation was calculated according to the formula: [1/(1+2^−ΔCt)]×100%, where ΔCt=Ct(unmethylated)–Ct(methylated) (23). Ct is the threshold of PCR cycle number at which the increase in fluorescent signal reaches a critical point. Each sample was analyzed in duplicate.

The Student's t-test, Wilcoxon test and Chi-square test were used to analyze the differences in normal/near-normal characteristics, other continuous variables and categorical variables between participants in T2DM group and non-diabetic group. We then utilized linear regression model to exam the association between ESRα methylation level and lipid metabolism. Besides, the association between T2DM and lipid metabolism (as well as ESRα methylation level) were estimated using binary logistic regression model. The linear trends across increasing quartiles of ESRα methylation level and lipid metabolism were estimated by treating the median of each quartile as a continuous variable. The quartiles of ESRα methylation and lipid metabolism are shown in Supplementary Table 1. After stratifying the participants according to the median of ESRα promoter methylation level, we analyzed the interactive effect of ESRα methylation and lipid metabolism on the risk of T2DM by adding an interaction term “ESRα methylation ^* lipid metabolism” to the logistic regression model. In addition, we performed other interactive analyses by adding an interaction term (“ESRα methylation^*gender” or “lipid metabolism^*gender”) to each models. According to the characteristics of the participants and the previous reports, we adjusted a variety of potential confounding variables (including age, gender, BMI, educational level, smoking, drinking and household income) in this study.

All statistical analyses were performed by SPSS 22.0 (IBM Corp, Armonk, NY, USA). The P-values < 0.05 were considered statistically significant.

The characteristics of the 1,143 residents are summarized in Table 1. As compared to the non-diabetic group, the T2DM group have a lower proportion of men/women, drinking and vegetables intake (≥500 g/day), and a higher proportion of illiteracy and family diabetes history (P < 0.05). Besides, the average age, BMI, TG and TC in T2DM group were higher than those in non-diabetic group (P < 0.05). The distributions of household incomes, ESRα promoter methylation, HDL-C and LDL-C are comparable between the two groups.

After stratifying the participants according to the tertiles of age, we found that found that the levels of ESRα methylation, FPG and LDL-C in the elderly were elevated (all P < 0.05). There is no statistically significant difference in the levels of TG, TC, HDL-C among different groups (Supplementary Table 2).

As shown in Table 2, the HDL-C level showed a downward trend with the increase of the ESRα promoter methylation level regardless of gender stratification (P for trend < 0.05). Besides, we found that in the continuous analysis, the ESRα promoter methylation level was negatively associated with HDL-C levels whether gender stratification was performed (all P < 0.05). For each unit standard deviation (SD) increment in ESRα promoter methylation, the level of HDL-C decreased by 0.03 mmol/L in all participants, 0.04 mmol/L in men and 0.03 mmol/L in women. Besides, we observed an increase of 0.08 mmol/L of LDL-C in men with each unit SD increment in ESRα promoter methylation level (P=0.037).

The association between lipid metabolism and T2DM is presented in Table 3. The risk of T2DM shown an upward trend with increasing quartiles of TG whether gender stratification was performed (all P for trend < 0.05). In continuous analysis, after adjusting for potential confounding factors, each unit SD increase in TG was associated with an increase of 43%, 36% and 49% in the risk of T2DM in all population, men and women respectively (all P < 0.05).

In addition, the risk of T2DM shown a downward trend with increasing levels of HDL-C in men (P for trend <0.05). In continuous analysis, we observed a decrement of 25% in the risk of T2DM with each unit SD increase of HDL-C levels in men. Furthermore, the interactive effect between HDL-C and gender on the risk of T2DM was evaluated, and significant association was observed.

We did not find any association between ESRα promoter methylation and FPG (Supplementary Table 3) or T2DM (Table 4). After stratifying the population into two groups by the ESRα promoter methylation level, we found that the risk of T2DM in participants with lower methylation (≤4.07%) were more susceptible to HDL-C (P for interaction = 0.030), as manifested by a decreased of 22% in the risk of T2DM with the increment of each unit SD in HDL-C concentration (Table 5).