Nutrients, one-carbon metabolism and DNA methylation

DNA methylation is the most studied epigenetic mechanism which entails the addition of a methyl group at cytosine–guanine dinucleotides (CpG)⁽ Reference Zilberman and Henikoff ^{9 )}. The reactions involved in DNA methylation are a part of one-carbon metabolism, which regulates the transfer of the one-carbon moiety (methyl group) into biological methylation reactions⁽ Reference Choi and Friso ^{4 )}. B-vitamins are coenzymes in one-carbon metabolism (Fig. 1), supporting that nutrients regulate epigenetic reactions⁽ Reference Friso and Choi ^{10 )}. The amino acid methionine is converted to S-adenosylmethionine (SAM), the unique methyl donor for many biological methylation reactions including DNA methylation⁽ Reference Choi and Friso ^{4 )}. Folate-derived methyl groups are utilised for remethylation of homocysteine to produce methionine⁽ Reference Friso and Choi ^{10 )}. Choline also provides methyl groups for the folate-independent homocysteine remethylation reaction; together these demonstrate how nutrients can serve as the source of epigenetic modifications⁽ Reference Tibbetts and Appling ^{11 )}. After transferring the methyl group, SAM is converted to S-adenosylhomocysteine (SAH), an inhibitor of methyltransferases⁽ Reference Tibbetts and Appling ^{11 )}. Deficiency of B-vitamins, methionine and/or choline can significantly affect DNA methylation by altering the levels of SAM and SAH⁽ Reference Choi and Friso ^{4 ,} Reference Taban-Shomal, Kilter and Wagner ^{12 ,} Reference Kim, Hong and Lee ^{13 )}.

Fig. 1. One-carbon metabolism. S-adenosylmethionine (SAM) is the unique methyl donor for many biological methylation reactions including DNA and histone methylation. S-adenosylhomocysteine (SAH) is an inhibitor of methyltransferases such as DNA methyltransferases (DNMT) and histone methyltransferases. In one-carbon metabolism vitamins B₂, B₆, B₁₂ and folate are coenzymes, while methionine, choline, betaine and serine are methyl donors. THF, tetrahydrofolate; 5-mTHF, 5-methyltetrahydrofolate; MT, methyltransferases; HMT, histone methyltransferases; MTHFR, methylenetetrahydrofolate reductase; MS, methionine synthase; SHMT, serine hydroxymethyltransferase; GNMT, glycine N-methyltransferase; CBS, cystathionine-β-synthase; MAT, methionine adenosyltransferase; SAHH, S-adenosylhomocysteine hydrolase; BHMT, Betaine homocysteine methyltransferase; B₂, vitamin B₂; B₆, vitamin B₆; B₁₂, vitamin B₁₂.

Modifying the intake of these nutrients alters DNA methylation⁽ Reference Niculescu and Zeisel ^{14 )}. In animal models, folate deficiency along with aging⁽ Reference Keyes, Mason Joel and Liu ^{15 )} and multiple B-vitamins deficiency⁽ Reference Liu, Choi and Crott ^{16 )} induced genomic DNA hypomethylation in the colon. It is also reported that choline deficiency can change DNA methylation independently⁽ Reference Niculescu, Craciunescu and Zeisel ^{17 )} or in conjunction with deficiency of other methyl donors⁽ Reference Niculescu and Zeisel ^{14 )}. Prolonged intake of diets deficient in sources of methyl groups such as methionine, choline, folate and vitamin B₁₂ induce profound genomic hepatic DNA hypomethylation in a rodent model⁽ Reference Pogribny, Ross and Wise ^{18 )}. This model has also demonstrated that a methyl-deficient diet changes histone methylation⁽ Reference Pogribny, Ross and Tryndyak ^{19 )} and microRNA production⁽ Reference Tryndyak, Ross and Beland ^{20 )}, which may potentiate the development of liver cancer.

Two previous human studies conducted in a metabolic unit demonstrated that marginal folate deficiency can change blood genomic DNA methylation⁽ Reference Jacob, Gretz and Taylor ^{21 ,} Reference Rampersaud, Kauwell and Hutson ^{22 )}. Compared to animal studies, however, evidence demonstrating significant effects of folate supplementation on DNA methylation in free living human subjects is limited. Recently, Pizzolo et al. ⁽ Reference Pizzolo, Blom and Choi ^{23 )} reported that 8-week daily supplementation of 5 mg folic acid did not change DNA methylation in peripheral mononuclear cells despite increases in blood concentrations of folate and SAM and decreases in SAH. This observation suggests that changes in the levels of SAM and SAH may not always induce changes in DNA methylation. Consistently, two previous cystathionine β-synthase-deficient mouse studies demonstrated that increased levels of SAM and SAH are not always correlated with DNA methylation status in a tissue-specific manner⁽ Reference Choumenkovitch, Selhub and Bagley ^{24 ,} Reference Caudill, Wang and Melnyk ^{25 )}. It is important to note that different tissues have different susceptibilities to methyl deficiency and therefore this lack of response may not be consistent across all tissues⁽ Reference Liu, Pickford and Meagher ^{26 )}.

Other specific nutrients can modify one-carbon metabolism to alter DNA methylation⁽ Reference Choi and Friso ^{4 )}. Retinoic acid is known to affect glycine N-methyltransferase which catalyses the reaction from SAM to SAH⁽ Reference Ozias and Schalinske ^{27 )}. Genome-wide DNA methylation assays demonstrated 166 differentially methylated CpG sites between undifferentiated and retinoic acid-treated human embryonic stem cells⁽ Reference Cheong, Lee and Park ^{28 )}. Interestingly, a high-throughput DNA methylation array with neuroblastoma cells in vitro demonstrated that 402 gene promoters became demethylated and eighty-eight hypermethylated following retinoic acid treatment⁽ Reference Das, Foley and Bryan ^{29 )}. These studies indicate that nutrients can interact with the pathways regulating DNA methylation, though it remains to be determined how nutrients target specific genes for epigenetic modification or, alternatively, if these changes are stochastic in nature.

The trace element Se is an essential component of the antioxidant selenoproteins such as glutathione peroxidases⁽ Reference Jackson-Rosario and Self ^{30 )}. Se has been known to influence the trans-sulfuration pathway in one-carbon metabolism which converts homocysteine to cysteine and ultimately glutathione⁽ Reference Davis and Uthus ^{31 )}. In a rodent study, Uthus et al. ⁽ Reference Uthus, Yokoi and Davis ^{32 )} found that plasma total homocysteine and cysteine were significantly decreased and glutathione significantly increased by Se deficiency. They also found significantly decreased genomic DNA methylation by Se deficiency in the colon, with concomitant trend for decreased DNA methyltransferase (DNMT) activity (P<0·06), suggesting a relationship between Se metabolism and DNMT function⁽ Reference Uthus, Ross and Davis ^{33 )}. Most recently, Zeng et al. ⁽ Reference Zeng, Yan and Cheng ^{34 )} reported that dietary Se supplementation increases exon-specific DNA methylation of the tumour suppressor p53 in the rat liver and colon thereby suggesting the over-supplementation with Se may increase cancer risk. These observations indicate that Se alters one-carbon metabolism leading to both genomic and gene-specific changes in DNA methylation.

Research into the chemopreventive effects of vitamin D has taken the forefront due to the role of the vitamin D receptor in cell-cycle regulation and differentiation⁽ Reference Lamprecht and Lipkin ^{35 )}. Recent evidence demonstrates that incubation of MCF-7 breast cancer cells with vitamin D₃, which models oestrogen receptor+ non-invasive breast cancer, reduces the aberrant hypermethylation and restores gene expression of retinoic acid receptor β2⁽ Reference Stefanska, Rudnicka and Bednarek ^{36 )} and phosphatase and tensin homologue⁽ Reference Stefanska, Salamé and Bednarek ^{37 )}. Although this mechanism remains to be validated in vivo, this work indicates that vitamin D can modify gene-specific DNA methylation.

Bioactive food components and DNA methylation

Bioactive food components are compounds consumed in the diet that are not essential for life, though they may have beneficial health effects⁽ Reference Choi and Friso ^{4 )}. Numerous bioactive food components can alter epigenetic patterns through both direct and indirect interactions with the enzymes regulating the placement of epigenetic marks⁽ Reference Choi and Friso ^{4 )}. Research indicates that these interactions can be gene specific in nature, suggesting that the bioactive food compounds may target these enzymes to specific sites within the genome⁽ Reference Davis and Ross ^{38 )}.

Genistein is an isoflavone belonging to the flavonoids group of compounds derived from legumes that has demonstrated great potential to regulate the epigenome⁽ Reference Meeran, Ahmed and Tollefsbol ^{39 ,} Reference Zhang and Chen ^{40 )}. In the agouti mouse model, it is well established that maternal dietary genistein increases the level of DNA methylation at the agouti locus and produces more pseudo-agouti black coat-coloured offspring⁽ Reference Dolinoy, Weidman and Waterland ^{41 )}. In mouse embryonic stem cells, genistein does not affect de novo methylation occurring between day 0 and day 4, but interferes with subsequent regulatory processes leading to decreased methylation at the uncoupling protein 1 and synaptotagmin-like 1 promoters⁽ Reference Sato, Yamakawa and Masuda ^{42 )}. This indicates that genistein perturbed the methylation pattern of differentiated embryonic stem cells after de novo methylation in a time-dependent manner. Fang et al. ⁽ Reference Fang, Chen and Sun ^{43 ,} Reference Fang, Chen and Yang ^{44 )} reported that genistein dose-dependently inhibited DNMT activity and partially reactivated genes repressed by hypermethylation such as retinoic acid receptor β, p16 and O-6-methylguanine-DNA methyltransferase in oesophageal squamous cell carcinoma cell lines. Similar effects were observed from the cancer cell lines of colon⁽ Reference Wang and Chen ^{45 )}, prostate⁽ Reference Vardi, Bosviel and Rabiau ^{46 )} and cervix⁽ Reference Jha, Nikbakht and Parashar ^{47 )}. These observations indicate that genistein can modify DNA methylation tissue specifically, gene specifically and life cycle specifically.

Epigallocatechin-3-gallate (EGCG), the primary polyphenol in green tea, is known to have anti-cancer effects through many different mechanisms⁽ Reference Suganuma, Saha and Fujiki ^{48 )}. One candidate mechanism is the inhibition of DNMT1 leading to hypomethylation and de-repression of epigenetically silenced genes⁽ Reference Lee, Shim and Zhu ^{49 )}. Nandakumar et al. ⁽ Reference Nandakumar, Vaid and Katiyar ^{50 )} reported that EGCG reactivates silenced tumour suppressor genes p21 and p16 by reducing DNA methylation in human skin cancer cells, resulting in re-expression of mRNA and proteins of silenced tumour suppressors. Wong et al. ⁽ Reference Wong, Nguyen and Noh ^{51 )} also reported that physiologically relevant concentrations of EGCG can induce the expression of forkhead box P3, a master switch that controls the development and function of regulatory T-cells and Jurkat T-cells in vitro. These cells play a critical role in the maintenance of tolerance and the control of autoimmunity. The expression of forkhead box P3 was associated with reduced DNMT expression and DNA demethylation in EGCG-treated cells, suggesting that EGCG may epigenetically modify forkhead box P3 methylation and promote regulatory T-cell induction and expansion to potentially support the immune response to cancer. The DNMT inhibitory effects of EGCG were not as potent as pharmacologic agents such as 5-aza-2′-deoxycytidine, though this is not unexpected as nutrients are not specifically designed as therapeutics. It appears that EGCG provides potentially sustained and longer-term exposure effects with lower toxicity compared with pharmacologic agents, demonstrating the potential benefits of natural substances as chemotherapeutics.

In addition to the direct inhibitory effects of DNMT1, it is also reported that consumption of polyphenols could increase the formation of SAH, which supports an additional mechanism inhibiting DNA methylation by EGCG⁽ Reference Lee, Shim and Zhu ^{52 )}. Animal studies also demonstrated that EGCG consumption through drinking water can moderately decrease the level of SAM in the intestine⁽ Reference Fang, Chen and Yang ^{44 )}. Both observations indicate that the inhibitory effects of EGCG on DNMT1 could be conveyed indirectly by modifying one-carbon metabolism.

Apigenin from parsley⁽ Reference Fang, Chen and Yang ^{44 )}, curcumin from turmeric⁽ Reference Fu and Kurzrock ^{53 )}, lycopene from tomato⁽ Reference King-Batoon, Leszczynska and Klein ^{54 )} and sulforaphane from cruciferous vegetables⁽ Reference Meeran, Patel and Tollefsbol ^{55 )} are also known to have an inhibitory effect on DNMT, though their effects are weaker than that of tea polyphenols and genistein. Further studies are needed to determine the optimal nutrients intakes to effectively regulate the epigenome in health.

Bioactive food components and histone modifications

Histone proteins are essential for the packaging of DNA into chromosomes within the nucleus of a cell⁽ Reference Esteller ^{56 )}. Post-translational modifications of histones have been highlighted due to their function to regulate gene expression, especially synergistic interactions with DNA methylation⁽ Reference Cheng and Blumenthal ^{5 )}. Further, dysregulated histone acetylation patterns have been associated with many diseases including cancer, cardiac hypertrophy and asthma⁽ Reference Esteller ^{56 –} Reference North and Ellis ^{58 )}. Among many different types of histone tail modifications, histone acetylation has been the most frequent target to evaluate the epigenetic effects of nutrients, bioactive components and aging⁽ Reference Davis and Ross ^{38 ,} Reference Bandyopadhyay and Medrano ^{59 )}. Histone acetylation is an epigenetic phenomenon that acetylates lysine residues at the histone tail to alter local DNA compaction, leading to site-specific changes in gene expression⁽ Reference An ^{60 )}. The histone acetylation status is regulated by a family histone acetyltransferases (HAT) and histone deacetylases (HDAC).

Butyrate is generated during the fermentation of dietary fibre in the large intestine⁽ Reference Alvaro, Solà and Rosales ^{61 )}. Early in 1977, Riggs et al. ⁽ Reference Riggs, Whittaker and Neumann ^{62 )} reported that addition of sodium n-butyrate to tissue culture media increases global histone acetylation in cancer cell lines. Thereafter, sulforaphane, an isothiocyanate from broccoli, broccoli sprouts and cabbage as well as allyl compounds from garlic such as diallyl disulfide and S-allyl mercaptocysteine have been demonstrated to have HDAC inhibitory effects⁽ Reference Meeran, Ahmed and Tollefsbol ^{39 )}. In light of the development of chemotherapeutic HDAC inhibitors, this suggests potential functional significance of this family of bioactive food components as chemopreventive agents to regulate histone acetylation status⁽ Reference Fang ^{63 )}.

Curcumin, a yellow pigment present in the spice turmeric (Curcuma longa), has been linked with multiple beneficial activities with anti-inflammatory, antioxidant and anti-cancer properties⁽ Reference Goel and Aggarwal ^{64 )}. Curcumin is known to have inhibitory activity against HDAC and HAT, with a specifically strong inhibition of HAT in cancer models⁽ Reference Goel and Aggarwal ^{64 ,} Reference Marcu, Jung and Lee ^{65 )}. Kang et al. ⁽ Reference Kang, Cha and Jeon ^{66 )} demonstrated that curcumin-mediated HAT inhibitory activity is associated with decreased histone 3 and 4 acetylation in both glioblastoma cancer cells and adult neural-derived stem cells. Functionally, these epigenetic changes were associated with increased apoptosis in cancer cells and promoted neuronal differentiation in stem cells, suggesting possible therapeutic potential in cancer and neurodegenerative diseases.

Nutritional influences on age-related diseases through epigenetics

Epigenetic patterns are heavily influenced by the environment; due to dietary requirements for sustenance, it follows that nutrition represents a substantial daily environmental input⁽ Reference Kennedy ^{84 )}. Nutrition influences our physiology over the course of the entire lifecycle, with particular phases representing times that are more sensitive to nutritional inputs⁽ Reference Stipp ^{85 )}. Most recent studies indicate that the effects of nutrition in early life alters programmes leading to differential disease susceptibilities later in life which may be conveyed through epigenetic mechanisms⁽ Reference Zeisel ^{86 )}. Furthermore, the majority of age-related chronic diseases in the developed world are multi-factorial with substantial lifestyle components, indicating a significant role for nutritional epigenetics in their development⁽ Reference Feinberg ^{87 )}.

Sie et al. ⁽ Reference Sie, Medline and van Weel ^{88 )} investigated the effect of maternal and post-weaning folic acid supplementation on colorectal cancer risk in the offspring. The data suggested for the first time that maternal folic acid supplementation at North American post-fortification levels recommended to women at reproductive age protects against the development of colorectal cancer in the offspring. This protective effect may be mediated in part by increased global DNA methylation, decreased epithelial proliferation and reduced DNA damage in the colorectum. However, the same group reported that high intrauterine and post-weaning dietary exposure to folic acid may increase the risk of mammary tumours in the offspring, mediated in part by altered DNA methylation and DNMT activity⁽ Reference Ly, Lee and Chen ^{89 )}. These results indicate that different tissues have variable responses to folic acid supplementation, again emphasising the tissue specificity of epigenetic regulation.

The disruption of the HAT and HDAC balance can also be a major mechanism underlying changing epigenetic patterns with functional disease output, including cancer and neurodegeneration⁽ Reference Saha and Pahan ^{90 )}. In a rodent study, aged mice display specific deregulation of histone H4 at lysine 12 (H4K12) acetylation during learning and fail to initiate a hippocampal gene expression programme associated with memory consolidation. Restoration of deregulated histone acetylation reinstates the expression of learning-induced genes and recovers cognitive functions, purporting the importance of epigenetic histone regulation in neurological function⁽ Reference Peleg, Sananbenesi and Zovoilis ^{91 )}. Interestingly, Govindarajan et al. ⁽ Reference Govindarajan, Agis-Balboa and Walter ^{92 )} reported that butyrate, which is known to have an HDAC inhibitory effect, improves memory function in an Alzheimer's disease mouse model when administered at an advanced stage of disease.

As aberrant histone deacetylation has been demonstrated to silence critical genes in carcinogenesis, HDAC inhibitors have great potential as new anti-cancer drugs due to their ability to modulate transcription⁽ Reference Tanji, Ozawa and Kikugawa ^{93 )}. HDAC inhibitors such as trichostatin-A induce apoptosis and suppress cancer cell growth by affecting the acetylation status of tumour suppressor genes in cancer cell lines, though their specificity to gene targets is not well understood⁽ Reference Fortson, Kayarthodi and Fujimura ^{94 ,} Reference Xiong, Du and Zhang ^{95 )}. Sulforaphane and curcumin have similar effects on cancer cells by modifying histone acetylation⁽ Reference Goel and Aggarwal ^{64 ,} Reference Clarke, Hsu and Yu ^{96 )}. Sulforaphane inhibits HDAC activity in vivo and suppresses tumorigenesis in APC^min mice⁽ Reference Myzak, Dashwood and Orner ^{97 )}. In vitro, sulforaphane exerts differential effects on cell proliferation, HDAC activity and downstream targets in both normal and cancer cells⁽ Reference Clarke, Hsu and Yu ^{96 )}. Accompanied by reduced HDAC4 expression and activity, curcumin induces apoptosis and cell cycle arrest in medulloblastoma cells in vitro and reduced tumour growth in in vivo medulloblastoma xenografts⁽ Reference Lee, Krauthauser and Maduskuie ^{98 )}. These observations indicate that both sulforaphane and curcumin or any other bioactive components that have histone modification effects have the potential to be developed as cancer chemotherapeutic agents. Similar to trichostatin-A, it is important to determine the gene specificity of this nutrient-induced HDAC repression. Furthermore, studies need to determine whether the amounts of these nutrients can be consumed in the whole diet at physiological quantities or if supraphysiological supplements are required. The timing of nutrient exposure needs to be determined as well; are these nutrients effective at reducing carcinogenesis prior to cancer development, during early carcinogenesis or in late-stage disease?

Nutrition, systemic metabolism and epigenetics

In contrast to single-nutrient–single-gene interactions, systemic metabolism also plays a role in determining diet–gene interactions⁽ Reference Cyr and Domann ^{8 )}. In light of the increasing prevalence of obesity and subsequent development of type II diabetes mellitus, a large proportion of the population is being exposed to chronic hyperglycaemia and impaired lipid homoeostasis leading to a substantially different cellular milieu⁽ Reference Ren, Pulakat and Whaley-Connell ^{99 )}. Epigenetic mechanisms are currently being investigated within the scope of the metabolic profiles characteristic of obesity and its associated sequellae⁽ Reference Cooper and El-Osta ^{100 )}. Exposure of macrophages to high glucose to model hyperglycaemia caused recruitment of lysine-specific demethylase 1H3 and a reduction in H3K9 methylation at the NF-κB-p65 gene promoter leading to an increased expression of this transcription factor⁽ Reference Cooper and El-Osta ^{100 )}. Similar results were observed in both endothelial cells and vascular smooth muscle cells⁽ Reference Fernandez, Siebel and El-Osta ^{101 )}. These epigenetic marks were sustained following return to normoglycaemic conditions, lending support to the notion of metabolic memory that has been observed in clinical studies describing persistent vascular injury following previous poor metabolic control^{( 102 )}. Furthermore, human vascular smooth muscle cells treated with high-glucose conditions demonstrated activating histone 3 at lysine 4 (H3K4) dimethylation marks and increased gene expression at the NF-κB targets, monocyte chemotactic protein-1 and IL-6, indicating epigenetic mechanisms underlying the hyperglycaemia-induced vascular inflammation⁽ Reference Teperino, Schoonjans and Auwerx ^{103 )}.

Hypercholesterolaemia and circulating oxidised LDL are also systemic metabolic characteristics that potentiate the development of age- and obesity-related chronic diseases⁽ Reference Tonstad and Després ^{104 )}. Consistent with the notion that chronic inflammation plays an aetiological role in atherogenesis, incubation of human umbilical-cord vein endothelial cells with oxidized LDL also led to altered epigenetic marks at inflammation-related genes⁽ Reference N'Guessan, Riediger and Vardarova ^{105 )}. Cells exposed to oxidized LDL demonstrated recruitment of cAMP-response-element-binding protein-binding protein/p300 and reduced HDAC1 and HDAC2 binding leading to increased activating histone marks at the IL-8 and monocyte chemoattractant protein-1 promoters. In utero exposure to hypercholesterolemia from ApoE ^−/− mothers led to differential global histone methylation patterns in offspring vascular smooth muscle cells and endothelial cells⁽ Reference Alkemade, van Vliet and Henneman ^{106 )}. These effects were further potentiated when offspring were fed a high-fat diet inducing hypercholesterolemia. Studies in primate models have demonstrated that maternal high-fat diet leads to altered hepatic histone modifications and epigenetic enzyme expression, with gene-specific changes localised to genes involved in circadian rhythms, lipid metabolism and heat shock responses⁽ Reference Aagaard-Tillery, Grove and Bishop ^{107 )}. Taken together, these studies demonstrate that both trans-generational and post-natal exposure to the metabolic characteristics of high-fat diet and hypercholesterolaemia lead to differential epigenetic patterns in various tissues which may potentiate chronic disease development later in life.