Base-Flipping Propensities of Unmethylated, Hemimethylated, and Fully Methylated CpG Sites

Methylation of C5 of cytosines at CpG dinucleotide sites of the DNA is one of the most important factors regulating the expression of genes. The interactions of these CpG sites with proteins are essential for recognition and catalysis and in many cases are characterized by the flipping of either of the cytosine bases out of the DNA helix. In this paper, we present results from molecular dynamics simulations indicating that methylation of CpG sites suppresses spontaneous extra-helical conformations of either of the two cytosines. Thus, cytosines in unmethylated sites flip out easier than in hemimethylated sites and the latter flip out easier than in fully methylated sites. The different propensities for base flipping is observed not only between the cytosines that differ in their methylation states but also between the cytosines on the complementary strand. From alchemical mutation calculations, we find that methylation of one of the cytosines increases the free energy of the extra-helical conformation by 10.3–16.5 kJ/mol and this increase is additive with respect to the second methylation. Potential of mean force calculations confirm these results and reveal that cytosines in unmethylated sites favor flipping via the major-groove pathway. We perform several analyses to correlate this behavior with structural changes induced by the different methylation states of the CpG site. However, we demonstrate that the driving force for these propensities is the change in the electronic distribution around the pyrimidine ring upon methylation. In particular, unmethylated cytosine interacts more favorably (primarily via electrostatic forces) with solvent water molecules than methylated cytosine. This is observed for, both, extra-helical cytosines and intra-helical cytosines in which the cytosine on the complementary strand flips out and water molecules enter the DNA double-helix and substitute the hydrogen bonds with the orphan guanine. On the basis of these results of spontaneous base flipping, we conjecture that the mechanism for base flipping observed in complexes between hemimethylated DNAs and proteins is not likely to be passive.