DNA methylation is a major epigenetic mechanism of controlling gene expression in mammalian cells (14
). Even though DNA methylation changes in human cancers are characterized by generalized genomic hypomethylation and hypermethylation of focal CpG islands, and both events are able to alter gene expression, emphasis has been largely focused on functional significance of hypermethylation in CpG islands on promoters (16
). Because of this intense focus on hypermethylation events, the role of genome-wide DNA hypomethylation in human cancer has been under-studied. Recent studies have focused on repetitive DNA elements such as LINE-1
that are located throughout the human genome. These elements remain heavily methylated in normal cells, and their methylation levels, as measured by PCR-based methods has been shown to correlate well with overall genomic 5-methylcytosine content (17
) indicating that analysis of Alu and LINE-1 methylation serves as a surrogate measure of genomic methylation levels. LINE repeat regions located on specific chromosomes (e.g., 22q11-q12) are shown to be a consistent indicator of global methylation status (18
). Yang et al. demonstrated that changes in the methylation levels of abundant LINE-1 repetitive elements (GeneBank accession no. x58075) could also be used as a surrogate marker of genome-wide methylation changes (20
). Studies have shown that methylation levels measured in these abundant LINE-1 regions, which are easy to characterize by pyrosequencing technology, do not vary significantly with time within an individual, and therefore changes in their methylation levels could potentially be attributed to dietary or lifestyle factors or interventions with such factors (9
). In addition to LINE-1 genome-wide methylation assays, gene-specific assays are also used to assess the influence of folate status on gene methylation. In one such study, there was no effect of high serum folate levels on methylation patterns of the ec-SOD gene promoter in healthy subjects (21
Hypomethylation-mediated reactivation of active forms of LINEs (LINE-1) in relation to human cancer risk has been under investigation for several years. LINE-1 retrotransposons are shown to be hypomethylated in many cancers (22
) suggesting potential activation of LINE-1 in these cancers. Based on these observations, it is biologically plausible that reactivation of LINE-1 elements through hypomethylation could potentially induce genomic instability, thus transforming these cells into a precancerous or cancerous state. In support of this, studies have shown that LINE-1 retrotransposition is able to induce DNA damage and apoptosis in a breast cancer cell lines (25
). In addition, hypomethylation of LINE-1 elements are shown to be associated with increased genomic instability in non-small cell lung cancer (26
) and colorectal carcinomas (12
In addition to the genome-wide hypomethylation in cancer tissues, results support the notion that degree of LINE-1 methylation in “normal” cells such as peripheral blood derived DNA is a biomarker for susceptibility for head and neck squamous cell carcinoma (27
). This study assessed whole blood global DNA methylation levels in one long interspersed nuclear element repeat region, LRE1
, located on 22q11-q12 using a modified version of the combined bisulfite restriction analysis. An inverse association between leukocyte global DNA methylation (assessed by using a combination of high performance capillary electrophoresis (HPCE) and Hpa II digestion of DNA) and bladder cancer has been reported previously (28
). A recent study suggested that leukocyte DNA hypomethylation determined by liquid chromatography-electrospray ionization tandem mass spectrometry is independently associated with development of breast cancer (29
). Genomic methylation of leukocyte DNA determined by liquid chromatography-mass spectrometry is also shown as a potential etiologic factor for early stage of colorectal adenomas (30
). Reverse causation is theoretically possible in cross-sectional studies of cancer, which is the presence of cancer may lower leukocyte DNA methylation, but it is unlikely that this occurs with precancerous lesions such as adenomas or, as is the case of this study, CIN2+. No previous studies have been able to rule out this possibility because of lack of methylation data in peripheral cells and in cancer or pre-cancer cells of the same individuals. We addressed this gap in knowledge by measuring LINE-1 methylation in paired samples of PBMCS and CCs in our study population.
Our results showed that only the degree of PBMC LINE-1 methylation is independently associated with CIN 2+. Lack of significant correlation between PBMC and CC LINE-1 methylation indicates that PBMC methylation is not a surrogate marker for CC methylation. Higher LINE-1 methylation of PBMCs and its association with lower likelihood of CIN 2+ diagnosis may be related to the ability of those individuals to mount a protective immune response to HR-HPVs. The study of Hsiung et al supports this idea by demonstrating that PBMC methylation has a positive influence on immune response against HPV related head and neck cancer (31
). Therefore, our results suggest that susceptibility to HR-HPV related cervical carcinogenesis may also be an epigenetically modified process operating via a central mechanism rather than via a mechanism operating at target tissues. Immune response has been shown to play a crucial role in eliminating HPV infections (32
). Cell-mediated immune responses, especially Th1 responses leading to increased IL-2 and IFN- production, are likely to be the major component that contributes to the development of cell-mediated immunity against HPV infections and HPV-associated neoplasms (33
). The wider range of immune responses demonstrated in these studies, however, suggests the possibility that host factors such as nutritional status may modify the immune responsiveness at an individual level. In this regard, β-hydroxy-β-methylbutyrate, which is an intermediate of leucine metabolism, impaired lymphocyte proliferation and progression through the cell cycle and modified the Th1/Th2 cytokine production towards a Th2 profile when HMB was present at 10 mM (34
). Several lines of evidence from animal studies suggested that folate deficiency affects cell-mediated immune functions (35
) and supplementation with folic acid is able to improve the immune response (37
We observe that the combination of supra-physiologic concentrations of folate and sufficient vitamin B12 has a positive influence on PBMC LINE-1 methylation and higher PBMC methylation is associated with lower risk of CIN 2+. Therefore, our current observations strengthen our previous finding that supra-physiologic plasma folate concentrations seen in the post US folic acid fortification era do not increase the risk of CIN in premenopausal women of childbearing age, especially in the presence of sufficient vitamin B12 status, because the same micronutrient combination is able to exert a positive influence on PBMC methylation which in turn reduces the risk of CIN 2+. However, these results do not exclude the possibility of adverse effects of exposure to higher folate on other health conditions in this population.
In addition to folate and vitamin B12, we observe that other lifestyle factors and micronutrients may have an influence on PBMC methylation. Higher physical activity seems to be associated with a lower degree of PBMC DNA methylation, but interpretation of this finding is unclear. More extensive data on physical activity rather than self-reported data may be needed to explain these contradictory findings. Higher vitamin C is positively associated with PBMC LINE-1 methylation, while gamma-tocopherol is inversely associated. In our previous studies, these two micronutrients were not independently associated with CIN risk (4
). Therefore, it is unlikely that these micronutrients have significant influence on CIN risk via PBMC methylation.
In summary, even though several biological mechanisms by which folate could modify the risk associated with HR-HPV are highly plausible and include folate's effects on cancer biomarkers such as methylation of DNA and histones and DNA damage in target cells, or folate's effects on immune responses, all of these are largely unproven mechanisms. We propose PBMC LINE-1 methylation as a possible intermediate biomarker for the effect of folate and vitamin B12 on the occurrence of CIN 2+ in our study population. Our previous studies documented a reduced prevalence of CIN 2+ among women with supra-physiologic folate and sufficient vitamin B12 levels (4
). In this report, we show that increased PBMC LINE-1 methylation is strongly associated with supra-physiologic concentrations of folate and sufficient vitamin B12 concentrations among non-cases, suggesting that higher folate and vitamin B12 affect PBMC LINE-1 methylation before the development of CIN. In addition, in the comparison of cases and non-cases, PBMC LINE-1 methylation is a strong predictor of decreased CIN risk. These findings are consistent with the hypothesis that PBMC LINE-1 methylation mediates the effects of folate and vitamin B12 in the pathway leading to CIN in women exposed to HR-HPVs. Documenting causal pathways in epidemiologic analyses is difficult, and our analysis of this cross-sectional study cannot conclusively address the questions of whether PBMC LINE-1 methylation is a real intermediary of the protective effect of folate and vitamin B12, and whether other pathways exist. The presence of a residual, non-significant protective effect of folate and vitamin B12 among women with <70% PBMC LINE-1 methylation is consistent with other mechanisms for the effect of the two micronutrients, but cannot exclude that PBMC LINE-1 methylation is the only pathway. The results from our ongoing follow-up study may provide both a more appropriate temporal relation, and more adequate numbers to address these issues.