Epigenome-wide association meta-analysis of DNA methylation with coffee and tea consumption

Abstract

Coffee and tea are extensively consumed beverages worldwide which have received considerable attention regarding health. Intake of these beverages is consistently linked to, among others, reduced risk of diabetes and liver diseases; however, the mechanisms of action remain elusive. Epigenetics is suggested as a mechanism mediating the effects of dietary and lifestyle factors on disease onset. Here we report the results from epigenome-wide association studies (EWAS) on coffee and tea consumption in 15,789 participants of European and African-American ancestries from 15 cohorts. EWAS meta-analysis of coffee consumption reveals 11 CpGs surpassing the epigenome-wide significance threshold (P-value <1.1×10^-7), which annotated to the AHRR, F2RL3, FLJ43663, HDAC4, GFI1 and PHGDH genes. Among them, cg14476101 is significantly associated with expression of the PHGDH and risk of fatty liver disease. Knockdown of PHGDH expression in liver cells shows a correlation with expression levels of genes associated with circulating lipids, suggesting a role of PHGDH in hepatic-lipid metabolism. EWAS meta-analysis on tea consumption reveals no significant association, only two CpGs annotated to CACNA1A and PRDM16 genes show suggestive association (P-value <5.0×10^-6). These findings indicate that coffee-associated changes in DNA methylation levels may explain the mechanism of action of coffee consumption in conferring risk of diseases.

Fig. 1. Overview of the study flow.

The flowchart summarizes our study design including EWAS meta-analysis to identify DNA methylation sites associated with coffee and tea consumption, and post-EWAS in silico and in vitro experiments. QQ quantile–quantile, eQTM cis-expression quantitative trait methylation, meQTL methylation quantitative trait loci, RS Rotterdam Study, FHS Framingham Heart Study, ALSPAC The Avon Longitudinal Study of Parents and Children, CHS Cardiovascular Health Study, ARIC The Atherosclerosis Risk in Communities Study, EPIC Prospective Investigation into Cancer and Nutrition, KORA Cooperative Health Research in the Augsburg Region Study.

Fig. 2. Epigenome-wide association study Manhattan plots for coffee and tea consumption.

The plots depict the results of EWAS fixed-effects inverse-variance meta-analysis with the overall sample for coffee (A) (n = 15,789) and tea (B) consumption (n = 15,069) in the fully adjusted model. Each dot corresponds to a single CpG site plotted as the negative logarithm of the p-value (−log(p-value) (y-axis)) against the genomic position of the CpG site (x-axis). The red line indicates the Bonferroni adjusted threshold at epigenome-wide significance p-value of 1.1 × 10⁻⁷ (0.05/450.000).

Fig. 3. The CoMET plots depicting genomic regions where the CpGs annotated to AHRR (A) and PHGDH (B) are located.

The x-axis indicates the position in base pair (bp) (hg19) for the region, while y-axis indicates the strength of association from EWAS with coffee consumption. The red line indicates the Bonferroni threshold for epigenome-wide significance (P = 1.1 × 10⁻⁷). The figure was computed using the R-based package CoMET, while the Ensembl is a genome database resource (http://ensemblgenomes.org/). The correlation of the surrounding CpGs was computed using methylation measures in the Rotterdam Study.

Fig. 4. PHGDH gene expression levels in liver cell lines and relative to expression levels of lipid-associated genes.

a Relative expression levels of PHGDH against a reference gene (GAPDH) in 7 human liver cell lines. Gene expression levels were quantified by qRT-PCR. Data were normalized to the PLC cell line (PLC, set as 1). b Relative expression levels of 9 lipid-associated genes in SNU499 cell line (with the lowest level of PHGDH expression) and SNU398 cell line (with the highest level of PHGDH expression) are shown. Relative gene expression levels were quantified by qRT-PCR. GAPDH serves as a reference gene, and gene expression levels in SNU449 cell line set as 1. This figure shows that, compared with SNU449 cells, SNU398 cells differentially express five of the lipid-associated genes (FDFT1, HMGCR, LDLR, LPL, and ABCA1). c Established PHGHD knockdown cell lines (shPHGHD-1 and -2), PLC cells transduced with lentiviral shRNA vectors targeting PHGDH or scramble control. qRT-PCR analysis of PHGDH expression were performed in stable knockdown or scramble control PLC cells. Data are normalized to the scramble control (scramble, set as 1). d Expression levels of five lipid-associated genes in stable PHGDH knockdown or scramble control PLC cells. Data were normalized to the scramble control (scramble, set as 1). The figure demonstrates that knockdown of PHGDH gene expression by lentiviral shRNA vectors resulted in significant decrease in the expression level of LPL and significant increase in the expression levels of LDLR and ABCA1 in both knockdown cells. Data in the figures are presented as mean values ± SEM of n = 3 biologically independent experiments. The Mann–Whitney U-test (two-sided) was used to compare differences between two independent groups. Differences were considered significant at P < 0.05, which indicated by * (**P < 0.01 and ***P < 0.001).