This is the first population-based randomized intervention study to describe the global DNA methylation response to the initiation of and withdrawal from folic acid supplementation. In this study, no global methylation changes were associated with up to 4,000 µg/day folic acid supplementation for 6 months; when the analysis was completed using DNA extracted from uncoagulated blood. However, significant changes in DNA methylation were found in response to supplementation with folic acid and the withdrawal of supplementation in DNA extracted from coagulated blood that were time-, dose-, and genotype-dependent, suggesting that folic acid intake might impact a coagulation-mediated process.
Previous small-scale feeding studies have provided evidence that global DNA methylation decreases in response to moderate folate intake/depletion 
and increases upon subsequent repletion 
, thus an initial increase in DNA methylation was anticipated in the current study. However, no changes in DNA methylation levels were found in uncoagulated samples, which might reflect more closely circulating blood and the samples used in previous studies. One possible explanation for the lack of change observed in this study was that DNA methylation was already optimal among our participants, and would not have been expected to increase with folic acid supplementation. Alternatively, the differential results between studies possibly could be explained by the fact that previous analytical methodology used to estimate DNA methylation lacked the accuracy and precision of the LC-MS/MS method to measure the true level of genome-wide methylation 
Surprisingly, the data from this study also suggested that folic acid supplementation led to changes in DNA methylation levels in blood cells that were allowed to coagulate and that these changes were time-, dose-, and MTHFR genotype-dependent. The differential results by blood sample type were unexpected and have not been reported previously. To rule out methodological error, we replicated the quantitation of DNA methylation from blood clots by LC MS/MS up to four times, including reblinding and side-by-side analysis from extraction through analysis, with high reproducibility of results between analytical runs. We also estimated DNA methylation in the coagulated samples by using a completely different methodology (3H methyl acceptance), which confirmed the pattern and magnitude of change in the coagulated samples. Moreover, the analogous change in DNA methylation observed between the two assays, the decrease in %MdCty (LC-MS/MS), and the increase in unmethylated dCyt ([3H] methyl acceptance assay), could not be explained by changes in the quality of the DNA used in the assay. The [3H] methyl acceptance assay requires that the DNA be high quality and double stranded; thus, low-quality DNA would result in a decrease in [3H] methyl acceptance capacity (not an increase, as was observed in this study). In contrast, the LC-MS/MS assay is independent of DNA quality, as the DNA is digested to nucleosides as part of the assay. Therefore, these data supported the conclusion that the discrepancy in DNA methylation levels between blood sample types could not be attributed to methodological error.
The differential results from coagulated and uncoagulated blood were present only in association with folic acid supplement use and the large decreases in DNA methylation observed in coagulated blood after supplement withdrawal were associated strongly with the MTHFR genotype and consistently replicated and confirmed across time using different analytical methodology. The fact that the level of DNA methylation was the same in both sample types at baseline, with differences observed only after the initiation of folic acid supplementation, confirms the influence of folic acid supplementation and MTHFR genotype in DNA exposed to the clotting process. Therefore, the time-, dose-, and MTHFR genotype-dependent changes in DNA methylation levels in the coagulated DNA samples in response to folic acid supplementation and withdrawal cannot be considered an artifact. These data raise a series of questions related to the possible influence of coagulation and the MTHFR 677C→T genotype on DNA methylation when folic acid exposure is altered and sets a research agenda for these intriguing new questions to be investigated.
Why do global methylation levels differ between coagulated and uncoagulated blood samples?
Differential Cell Type Distribution
In this study we utilized uncoagulated (EDTA) blood (spun down with the plasma removed) as one source of DNA. The buffy coat (the cell fraction most often used in genetic studies) should be contained completely in this fraction. Ideally, the uncoagulated and the coagulated samples (obtained through simply allowing a clot to form, then spinning down and removing the sera) should contain similar amounts and types of the nucleated blood cells (all of them), which would contribute to the DNA assayed for methylation status; however; in practice, this might not be the case. In our study, it is possible that some of the observed changes in DNA methylation in response to supplementation might have been due to differences in leukocyte cell types or specific cell subtype responses from which DNA was extracted, given that whole blood contains a mixture of cells with different turnover rates (e.g., neutrophils, lymphocytes, and monocytes) 
whose DNA methylation patterns might differ 
. The fractionation procedures could have left different populations of cells in the aqueous fraction (plasma or sera, respectively). To address this issue in the available samples, we compared the DNA yields from coagulated and uncoagulated blood and correlated DNA yield in the clots with methylation level. Samples yielded the expected amount of DNA based on the blood volume of the initial blood draw. There was variation in DNA yield, which is to be expected when using clots 
; however; there was no correlation between DNA yield and global methylation levels in the samples. It seems unlikely that different cell types in the clots would have been a driving mechanism for the observed reductions in DNA methylation for a number of reasons. First, to achieve even a 14% decrease in methylation (as observed with supplementation), an extensive shift in cell type distribution within the leukocyte cell population would have been required (this is possible, but not likely). Second, achievement of a large genotype-dependent reduction (up to 90%) without a concomitant drop in DNA yield seems improbable. Additionally, many different human tissue types (e.g., heart, liver, lung, spleen, lymphocyte, brain, and thymus) among humans appear to exhibit little variation in global DNA methylation (mean
in contrast to ~14% promoter of CpG sites have been reported to vary between tissues 
. Moreover, the level of DNA methylation in monocytes from a U.S. study (4.45+/−0.12%; unpublished data) was similar to that observed for whole blood DNA at enrollment in this study in the clotted DNA samples (4.42+/−0.12%). The fact that no differences were detected in DNA methylation at baseline between the coagulated and uncoagulated samples argues against there being differences in the cell population between the two sample types.
Other processes associated with de-/hypo-methylation
Apoptosis is the primary consideration in any discussion of de-/hypomethylation because of the clear connection between hypomethylation and apoptosis 
. This study was not designed to assess if the samples allowed to coagulate were undergoing apoptosis. A limited number of samples run on agarose gels did not show a definitive apoptotic banding pattern or degradation that was limited to samples with decreases in assayed DNA methylation level (data not shown). However, since this study was not designed to directly assay for evidence of apoptosis, a potential influence of this process cannot be ruled out and warrants further study.
Generally, decreases in global methylation are thought to be a result of a reduced availability of substrate during DNA replication. However, it is unlikely that the hypomethylation seen in our study is replication-dependent because of the limited window of time (1–2 hours) and the large decreases reported. Active demethylation has been observed embryonically; there is a wave of demethylation as DNA methylation patterns are reset in the early embryo 
; however, the identity the demethylase remains unknown 
. It is possible that the hypomethylation observed in our study may be the result of active demethylation, as part or consequence of the 2 hours allowed for coagulation.
The study of coagulation has centered largely on factors in the plasma that induce coagulation. Elevated homocysteine concentration consistently has been shown to be associated with venous thrombosis; however, the nature of the association remains an active area of investigation 
. Among populations with higher folate/folic acid intake, such as those in North America and Europe, there has been no consistent association between thrombosis and MTHFR
. However, among populations in Asia and Latin America, the MTHFR
T allele has been shown to be associated more consistently with an increased risk of venous thrombosis 
. It is interesting to consider the fact that the largest decreases in DNA methylation in our present study were following withdrawal of supplementation in those with the MTHFR
C allele, which might be more resistant to coagulation. In our previous study, elevated homocysteine was associated strongly with the MTHFR
T genotype among this population 
. Work in human cell lines suggests that hypomethylation is a mechanism of hyperhomocysteinemia-induced cell damage 
. Intriguingly, a small plasma proteomics study of folic acid supplementation (1.2 mg for12 weeks) found that folic acid supplementation was associated with changes in proteins involved in increased coagulation 
. Additionally, taking a folic acid supplement during pregnancy has been shown to be associated with a reduction in hypercoagulability 
. There is a balance between coagulation and hemorrhage, and MTHFR
genotype and folic acid use and DNA methylation could affect this equilibrium. Future research will be needed to determine any effects on coagulation of folic acid supplementation and the termination of supplementation.
Additional basic science research is needed to replicate our findings and provide insight into additional biological mechanisms that might explain our results. The differential results between sample types were not present at baseline, becoming evident only after initiation of the folic acid supplementation trial. Currently, studies are underway to determine if there are differences in the coagulated and uncoagulated samples of individuals who were B12-deficient and anemic who were enrolled in the study, but not eligible to participate in the intervention trial. The results from this study will help to determine if this phenomenon is limited to folic acid exposure or might be a broader effect of micronutrient status.
Although we have shown that in the uncoagulated blood there are no global methylation changes, this does not rule out changes in specific genes or genetic loci. Tests currently are under way in uncoagulated samples to determine if changes in specific genes in response to folic acid supplementation or the termination of supplementation, or both, can be detected. Additionally, controlled studies are needed to determine if these results can be replicated among populations both exposed and unexposed to folic acid, as well as other racial and ethnic and nutritional status groups.
Although the present study does not provide definitive explanations regarding the observed influence of blood clotting, folic acid supplementation, and the MTHFR genotype on DNA methylation, the findings do inform future studies of DNA methylation. Specifically, these results provide new opportunities for basic researchers to identify pathways of DNA demethylation that occur during coagulation that are sensitive to MTHFR genotype and folic acid exposure. These data have implications for the design of future studies since clotted samples often are available as a source of genomic DNA for DNA methylation analyses in human metabolic and population surveys. DNA methylation analysis currently requires genomic DNA in relatively high quantities (several hundred ng) which can be problematic for large studies involving human participants using archival samples. The clots from serum collection are an often available and unutilized source of DNA; however, our study strongly suggested that this sample type does not show the same results as those obtained from uncoagulated blood (which might represent more closely what is present in circulating blood). The study highlights the importance of sample selection for those planning epigenetic studies. Additionally, these data provide the impetus for further investigations to clarify the observed differences in DNA methylation between coagulated and uncoagulated blood that are suggestive of an effect of folic acid on a coagulation-mediated process.