This strongly suggests the existence of variant influencing the variability of DNA Epoxomicin methylation levels at the SLC19A2 gene. How the rs970740 T/C genetic variation affects SLC19A2 DNA methylation remains an open question. This could be through the creation of a transcription factor binding site, the modification of the local CpG sites distribution, or more complex phenomena. The SLC19A2 gene codes for a thiamine transporter protein that has been associated with human anemia syndrome. Our results suggest that genetically determined DNA methylation sensitive mechanisms are involved in this disease susceptibility. Several conclusions could be drawn from this work. First, three identified CpG sites were found to be strongly associated with the plasma variability of two quantitative biomarkers of the coagulation cascade, supporting the potential of genome-wide DNA methylation data to identify epigenetic marks associated with biological phenotypes involved in thrombotic disorders. Nonetheless, this works highlights the need for careful analyses of associations between genetic variants, biological phenotypes, and methylation at CpG sites to avoid false inference on functional variant, in particular due to LD extending over large genomic distances. Integrating MWAS, GWAS and biological data from the same individuals, as illustrated here, is key to elucidating these relationships. Second, if such cautions are taken, DNA methylation data can help to dissect the functional mechanisms associated with known disease-causing SNPs. Several limitations must be acknowledged. First, the design of our study may not be optimal. As we did not have access to a casecontrol study for VT with genome-wide DNA methylation data, we adopted a ‘case-only’ approach for our discovery stage. Such approach has been shown to be a valid alternative to detect gene6 environment or gene 6 gene interactions. We here used this strategy with the aim of identifying epigenetic factors that interact with the FV Leiden mutation to modulate the risk of VT. Since, in our replication study, 45 carriers were VT patients and the remaining 8 carriers were healthy individuals, we also looked into this dataset for specific methylation patterns associated with VT but the low sample size precludes from identifying any significant association. Second, because homozygosity for FV Leiden mutation was an exclusion criteria for the MARTHA study and no homozygote was observed in the F5L-families, our analysis only included heterozygous carriers which may have reduced our power to identify CpG sites under the strong influence of the mutation. Third, while extremely dense, the used Illumina array does not cover all sites of the genome that could be subject to DNA methylation, we cannot exclude that some relevant methylation association has been missed. Fourth, the sample size of our discovery study was large enough to detect, at the genome-wide level of 1.29 1027, a 0.05 increase in the methylation b-value. Whether an increase of smaller magnitude in DNA methylation marks detected in whole blood could be biologically relevant remains an open question. Whole blood DNA methylation levels reflect the average levels resulting from the epigenetic state.
[title][/title]