DNA polymerase III mis-insertion may, where not corrected by its 3'→ 5' exonuclease or the mismatch repair (MMR) function, result in all possible non-cognate base pairs in DNA generating base substitutions. The most thermodynamically unstable base pair, the cytosine (C)⋅C mismatch, destabilizes adjacent base pairs, is resistant to correction by MMR in Escherichia coli, and its repair mechanism remains elusive. We present here in vitro evidence that C⋅C mismatch can be processed by base excision repair initiated by the E. coli formamidopyrimidine-DNA glycosylase (Fpg) protein. The k cat for C⋅C is, however, 2.5 to 10 times lower than for its primary substrate 8-oxoguanine (oxo8G)⋅C, but approaches those for 5,6-dihydrothymine (dHT)⋅C and thymine glycol (Tg)⋅C. The K M values are all in the same range, which indicates efficient recognition of C⋅C mismatches in DNA. Fpg activity was also exhibited for the thymine (T)⋅T mismatch and for N 4- and/or 5-methylated C opposite C or T, Fpg activity being enabled on a broad spectrum of DNA lesions and mismatches by the flexibility of the active site loop. We hypothesize that Fpg plays a role in resolving C⋅C in particular, but also other pyrimidine⋅pyrimidine mismatches, which increases survival at the cost of some mutagenesis.

Fine-tuned regulation of the cellular nucleotide pools is indispensable for faithful replication of Deoxyribonucleic Acid (DNA). The genetic information is also safeguarded by DNA damage recognition and repair processes. Uracil is one of the most frequently occurring erroneous bases in DNA; it can arise from cytosine deamination or thymine-replacing incorporation. Two enzyme activities are primarily involved in keeping DNA uracil-free: dUTPase (dUTP pyrophosphatase) activity that prevent thymine-replacing incorporation and uracil-DNA glycosylase activity that excise uracil from DNA and initiate uracil-excision repair. Both dUTPase and the most efficient uracil-DNA glycosylase (UNG) is thought to be ubiquitous in free-living organisms. In the present work, we have systematically investigated the genotype of deposited fully sequenced bacterial and Archaeal genomes. We have performed bioinformatic searches in these genomes using the already well described dUTPase and UNG gene sequences. For dUTPases, we have included the trimeric all-beta and the dimeric all-alpha families and also, the bifunctional dCTP (deoxycytidine triphosphate) deaminase-dUTPase sequences. Surprisingly, we have found that in contrast to the generally held opinion, a wide number of bacterial and Archaeal species lack all of the previously described dUTPase gene(s). The dut- genotype is present in diverse bacterial phyla indicating that loss of this (or these) gene(s) has occurred multiple times during evolution. We discuss potential survival strategies in lack of dUTPases, such as simultaneous lack or inhibition of UNG and possession of exogenous or alternate metabolic enzymes involved in uracil-DNA metabolism. The potential that genes previously not associated with dUTPase activity may still encode enzymes capable of hydrolyzing dUTP is also discussed. Our data indicate that several unicellular microorganisms may efficiently cope with a dut- genotype lacking all of the previously described dUTPase genes, and potentially leading to an unusual uracil-enrichment in their genomic DNA.

The cellular methyl donor S-adenosylmethionine (SAM) and other endo/exogenous agents methylate DNA bases non-enzymatically into products interfering with replication and transcription. An important product is 3-methyladenine (m3A), which in Escherichia coli is removed by m3A-DNA glycosylase I (Tag) and II (AlkA). The tag gene is constitutively expressed, while alkA is induced by sub-lethal concentrations of methylating agents. We previously found that AlkA exhibits activity for the reactive oxygen-induced thymine (T) lesion 5-formyluracil (fU) in vitro. Here, we provide evidence for AlkA involvement in the repair of oxidized bases by showing that the adenine (A) ⋅ T → guanine (G) ⋅ cytosine (C) mutation rate increased 10-fold in E. coli wild-type and alkA - cells exposed to 0.1 mM 5-formyl-2'-deoxyuridine (fdU) compared to a wild-type specific reduction of the mutation rate at 0.2 mM fdU, which correlated with alkA gene induction. G⋅C → A⋅T alleviation occurred without alkA induction (at 0.1 mM fdU), correlating with a much higher AlkA efficiency for fU opposite to G than for that to A. The common keto form of fU is the AlkA substrate. Mispairing with G by ionized fU is favored by its exclusion from the AlkA active site.