Division of Biostatistics and Bioinformatics, Institute of Population Health Sciences, National Health Research Institutes, Zhunan, Taiwan.
The Ph.D. Program for Translational Medicine, College of Medical Science and Technology, Taipei Medical University, Taipei, Taiwan.
Diabetologia. 2021 Jul;64(7):1613-1625. doi: 10.1007/s00125-021-05449-9. Epub 2021 Apr 12.
AIMS/HYPOTHESIS: An elevated fasting glucose level in non-diabetic individuals is a key predictor of type 2 diabetes. Genome-wide association studies (GWAS) have identified hundreds of SNPs for fasting glucose but most of their functional roles in influencing the trait are unclear. This study aimed to identify the mediation effects of DNA methylation between SNPs identified as significant from GWAS and fasting glucose using Mendelian randomisation (MR) analyses.
We first performed GWAS analyses for three cohorts (Taiwan Biobank with 18,122 individuals, the Healthy Aging Longitudinal Study in Taiwan with 1989 individuals and the Stanford Asia-Pacific Program for Hypertension and Insulin Resistance with 416 individuals) with individuals of Han Chinese ancestry in Taiwan, followed by a meta-analysis for combining the three GWAS analysis results to identify significant and independent SNPs for fasting glucose. We determined whether these SNPs were methylation quantitative trait loci (meQTLs) by testing their associations with DNA methylation levels at nearby CpG sites using a subsample of 1775 individuals from the Taiwan Biobank. The MR analysis was performed to identify DNA methylation with causal effects on fasting glucose using meQTLs as instrumental variables based on the 1775 individuals. We also used a two-sample MR strategy to perform replication analysis for CpG sites with significant MR effects based on literature data.
Our meta-analysis identified 18 significant (p < 5 × 10) and independent SNPs for fasting glucose. Interestingly, all 18 SNPs were meQTLs. The MR analysis identified seven CpGs near the G6PC2 gene that mediated the effects of a significant SNP (rs2232326) in the gene on fasting glucose. The MR effects for two CpGs were replicated using summary data based on the European population, using an exonic SNP rs2232328 in G6PC2 as the instrument.
CONCLUSIONS/INTERPRETATION: Our analysis results suggest that rs2232326 and rs2232328 in G6PC2 may affect DNA methylation at CpGs near the gene and that the methylation may have downstream effects on fasting glucose. Therefore, SNPs in G6PC2 and CpGs near G6PC2 may reside along the pathway that influences fasting glucose levels. This is the first study to report CpGs near G6PC2, an important gene for regulating insulin secretion, mediating the effects of GWAS-significant SNPs on fasting glucose.
目的/假设：非糖尿病个体的空腹血糖升高是 2 型糖尿病的一个关键预测因子。全基因组关联研究（GWAS）已经鉴定出数百个与空腹血糖相关的 SNP，但它们在影响该特征方面的大多数功能作用仍不清楚。本研究旨在使用孟德尔随机化（MR）分析，鉴定 GWAS 中鉴定为显著的 SNP 与空腹血糖之间的 DNA 甲基化的中介效应。
我们首先对来自台湾的三个队列（台湾生物银行 18122 人、台湾健康老龄化纵向研究 1989 人和斯坦福亚太高血压和胰岛素抵抗计划 416 人）的个体进行 GWAS 分析，这些个体均具有汉族血统，然后对三个 GWAS 分析结果进行荟萃分析，以鉴定与空腹血糖显著且独立的 SNP。我们通过测试这些 SNP 与台湾生物银行 1775 名个体附近 CpG 位点的 DNA 甲基化水平之间的关联，来确定这些 SNP 是否为甲基化定量性状基因座（meQTLs）。基于 1775 名个体，我们使用作为工具变量的 meQTLs 进行 MR 分析，以鉴定对空腹血糖具有因果效应的 DNA 甲基化。我们还使用两样本 MR 策略，根据文献数据，对具有显著 MR 效应的 CpG 位点进行复制分析。
我们的荟萃分析确定了 18 个与空腹血糖显著相关（p < 5×10）且独立的 SNP。有趣的是，这 18 个 SNP 都是 meQTLs。MR 分析确定了位于 G6PC2 基因附近的 7 个 CpG，这些 CpG 介导了该基因中的一个显著 SNP（rs2232326）对空腹血糖的影响。使用 G6PC2 基因中的外显子 SNP rs2232328 作为工具，基于欧洲人群的汇总数据，对两个 CpG 的 MR 效应进行了复制。
结论/解释：我们的分析结果表明，G6PC2 中的 rs2232326 和 rs2232328 可能影响基因附近 CpG 的 DNA 甲基化，而这种甲基化可能对空腹血糖有下游影响。因此，G6PC2 中的 SNP 和 G6PC2 附近的 CpG 可能位于影响空腹血糖水平的途径上。这是第一项报道位于调节胰岛素分泌的重要基因 G6PC2 附近的 CpG 介导 GWAS 显著 SNP 对空腹血糖影响的研究。
