Result
In total, 63 articles were included (details in appendix). Half of all studies (n=32) were cohort study of population. The number of studies increased with years. Included studies were limited to observational studies, and exploratory experiment studies were rare. The frequency of genes studied was counted in the literature, and the genes were listed with frequencies more than once(Fig.2). ABCG1, involved in the transportation of lipid elements, was ranked first with as many as 11 studies. These 11 articles were all population cohort studies, and some researched the same CpG sites. Individual factors were classified into six categories, which were listed below.
Mother during pregnancy
There were 13 studies showed that mother’s behavior would leave an epigenetic mark on offspring’s DNA(Table.1). Mothers’ smoking would lead to lighter baby birth weight and DNA methylation played an important role in this process. CYP1A1 CpG sites was significantly associated with birthweight (P= 4.76×10-5) and had significant mediation effect together with GFI1 and AHRR genes[13]. Correlation differed in gender and race. CYP1A1 was considered to detoxification of the components of tobacco smoke in phase I metabolism. CYP1A1 methylation may be one of the signal paths that how smoking affects babies. Trace metals exposure also affected fetal gene methylation levels. Cadmium (Cd), lead (Pb), total mercury (Hg), manganese (Mn) and selenium (Se) were associated with CpG sites[14]. Especially, Pd was associated with CYP24A1(cg01243877) (P<0.001) and Arsenic(As) was associated with CYP2A4 and CYP7B1[15]. CYP24A1 CpG sites, involved in vitamin D3 metabolism and cellular calcium homoeostasis, provided an evidence supporting Pb as a neurotoxicant. Maternal hormone affected the fetus’ ABCA1 and CYP11A1 methylation during pregnancy[16, 17]. Maternal gestational weight gain was related to offspring ABCA1 methylation (β = -1.1% per quartile; 95% CI: -2.0, -0.3)[18]. Poor nutrition or food deprivation during fetal development was related to PPARγ and ABCA1 methylation[19, 20]. ABCA1 may play a role in the efflux of intracellular cholesterol to apoliproteins and the formation of nascent high density lipoproteins (HDLs). These factors may influence baby HDL through DNA methylation. These studies suggested that parental generation affects DNA methylation in offspring, a reminder to be more careful during pregnancy to protect the fetus.
Environmental pollutants exposure
There were 11 studies exploring environment affects(Table.2). Environmental pollution has been attached great importance, and scientists are studying its impact on people. Smoking was one of well-known risk factors. Studies showed that smoking led to CYP1A1, CYP1A2, CYP2A6, CYP11B2 and PARβ methylation alteration[13, 21-26]. Polycyclic aromatic hydrocarbons (PAHs), one kind of carcinogens of cigarette, were metabolized by phase I (e.g. CYP1A1) and detoxified by phase II (e.g. GSTM1) before targeting DNA[27]. Epigenetic differences in CYP1A1 may explain individual metabolic differences and lung cancer risk in smokers. CYP2A6 was involved in 90% nicotine metabolism and its expression differed in gender and age. Men had lower nicotine clearance than women, and older people also had lower nicotine clearance than younger people[24]. DNA methylation explained some of the variation. Other common environmental pollutants, like PM2.5, dust mite and chemicals, were related to aberrant methylation. Shang Y et al. employed the mouse model to prove that the reprogramming of lung or airways by dust mite can be mediated through epigenetic[28]. Polycyclic aromatic hydrocarbons including dioxin affected CYP1A1 demethylation via aryl hydrocarbon receptor (Ahr)[29]. The change was tissue specific and may account for their carcinogenicity. These results suggested that our genes were sensitive to environmental pollutants, and we should minimize environmental pollution exposure.
Obesity and Diet
There were 18 studies showed that body mass index (BMI) and blood lipids affected the DNA methylation(Table.3). Obesity or high BMI had wide effects on gene methylation, including ABCG1,ABCC1, CYP27B1, SLC45A3,SLC1A5,and SLCO3A1. ABCG1 was an important part of lipid metabolism and also the most researched and conclusive gene. ABCG1 was believed that responsible for macrophage cholesterol and phospholipid transport[30]. Downregulation of ABCG1 led to reduce cholesterol efflux, which was associated with cardiovascular disease risk, obesity and dyslipidemia. High-risk groups of these diseases were often found hypermethylation of ABCG1[31-33]. Exactly, the results correspond to its function. However, we were not sure whether the methylation changes occur before or after the disease. Studies on ABCG1 were population cohort studies or case-control studies, while none of the studies explored specific mechanisms. These studies suggested that changes in epigenetics was one of the reasons for individual differences in obesity.
The ratio of different nutrients in the diet affected LMAO2, MnSOD, GSTM1, GSTT1, CYP1A1 and CYP2E1 methylation[34-36]. High fat diet led to CYP2R1,CYP27A1,CYP27B1,CYP24A1 and PPAR-α methylation alteration, but had no significant effect on PPAR-γ methylation[37-39]. These genes were involved in the metabolism of adipocytes, and studies shown that the body’s response to the stimulation of high-fat diet. However, the specific mechanism was still unclear. Specially, lack of folic acid led to demethylation of ABCG2[40], lack of Vitamin D led to hypermethylation of CYP24A1 and CYP27B1[41, 42]. The folic acid excretion cells increased or decreased when folic acid was excessive or deficient. This process was transported by ABCG2 and the expression of ABCG2 was regulated by methylation. Finally, the total folic acid maintain stable[40]. 24-hydroxylase encoded by the CYP24A1 gene was a catabolic enzyme and both 25(OH)D and 125(OH)2D were catabolized by the 24-hydroxylase into inactive metabolites, thereby lowering the vitamin D levels[41].
Drugs
There were 7 studies showed that some drugs can also influence the DNA methylation (Table.4). Although most studies did not display the magnitude of the effect, the genes drugs affected were important metabolic genes. Berberine can lead to hypermethylation of CYP2B6 and CYP3A4[43]. Aspirin can induce hypermethylation of ABCB1[44]. Methadone can cause hypermethylation of ABCB1 and CYP2D6, and the influence can be transmitted to the fetus[45]. Therefore we need pay more attention to drug combination and may have deeper understandings of drug interactions. However, there were another results that attracted our attention. García-Calzón, S. et al. found that metformin can influence methylation degree of SLC22A1, SLC22A3 and SLC47A1[46]. At the same time, metformin was their metabolic substrate. Additionally, Wang, X. K. et al. found that afatinib can effectively resist to the multidrug resistance (MDR) by hypermethylation in promoter and downregulating the expression of ABCG2[47]. Meanwhile, afatinib was also the transporter substrate of BCRP, coded by ABCG2.These findings may help us better understand the process of drug metabolism in the human body.
Gender
There were 6 studies showed that DNA methylation alteration was different in gender groups (Table.5). Study showed that CYP1A1, CYP2E1 and CYP7B1 methylation was different among genders[48]. Moreover, the same exposure affected different genders differently. Smoking showed significant methylation alternation of CYP11B2 and ABCG1 in different gender[22, 26]. Lead exposure influenced GPX1, CYP1A1 and SOD3 methylation differently in gender groups[49]. Polycyclic aromatic hydrocarbons exposure caused different PPARγ methylation in gender groups[50]. LDL-C and TG had different association with ABCG1 DNA methylation in different gender groups[33]. These studies reminded us that gender needed to be considered in personalized medicine.
Other factors
Besides above factors, other studies reported some meaningful experiment (Table.6). Some factors that often taken into account, like race, age and inflammation, also been shown to affect gene methylation[51-53]. Nano-SiO2, not often mentioned, led to hypomethylation of PARP and decreased expression on mRNA and protein level[54]. ABCA1 promoter methylation level was an independent risk factors for premature coronary artery disease along with traditional risk factors, like high BMI and HbA1c[55]. Worthy of attention, 2 studies explored the correlation of methylation changes with time and periodicity[56, 57]. CYP27B1 methylation was weakly association with season and CYP17A1 promoter was hypomethylated after circadian rhythm was disrupted. These findings provided evidence for the new idea to explain the relationship between the affecting factors and individualized medicine that time affected methylation periodically.