Abstract
In the active cytosine demethylation pathway, 5-methylcytosine (5mC) is oxidized sequentially to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC). Thymine DNA glycosylase (TDG) selectively excises 5fC and 5caC but not cytosine (C), 5mC, and 5hmC. We propose that the electron-withdrawing properties of -CHO and -COOH in 5fC and 5caC increase N3 acidity, leading to weakened hydrogen bonding and reduced base pair stability relative to C, 5mC, and 5hmC, thereby facilitating the selective recognition of 5fC and 5caC by TDG. Through (13)C NMR, we measured the pKa at N3 of 5fC as 2.4 and the two pKa's of 5caC as 2.1 and 4.2. We used isotope-edited IR spectroscopy coupled with density functional theory (DFT) calculations to site-specifically assign the more acidic pKa of 5caC to protonation at N3, indicating that N3 acidity is increased in 5fC and 5caC relative to C. IR and UV melting studies of self-complementary DNA oligomers confirm reduced stability for 5fC-G and 5caC-G base pairs. Furthermore, while the 5fC-G base pair stability is insensitive to pH, the 5caC-G stability is reduced as pH decreases and the carboxyl group is increasingly protonated. Despite suggestions that 5fC and 5caC may exist in rare tautomeric structures which form wobble GC base pairs, our two-dimensional infrared (2D IR) spectroscopy of 5fC and 5caC free nucleosides confirms that both bases are predominantly in the canonical amino-keto form. Taken together, these findings support our model that weakened base pairing ability for 5fC and 5caC in dsDNA contributes to their selective recognition by TDG.