br Materials and methods br Results
Materials and methods
Discussion Since 1990s, solving structures for most DNA glycosylases by X-ray crystallography significantly advanced mechanistic understanding of the catalytic activity and substrate specificity of this important class of DNA repair enzymes. However, static structures do not present the whole picture of substrate recognition and catalysis, and had been supplemented with various kinds of molecular dynamics and evolutionary analysis to reveal functionally important interactions not immediately evident from the structures. These in silico methods, combined with biochemical analysis of their predictions, provide a powerful way to detect important residues and interactions in enzymes and enzyme–substrate complexes. Pairs of coevolved Trehalose formed a network spanning both Fpg domains; only two pairs (His181–Asp183 and Glu140–Leu189) were not connected to this network. As can be seen from Fig. 1, both the β-sandwich domain and the H2TH domain showed intra- and interdomain correlation. Notably absent were coevolved residues in three important parts of the protein: the zinc finger, the oxoG-binding loop, and the interdomain linker; almost no correlations were observed either within these regions or between them and other parts of the protein (Fig. 1). Since these regions are not absolutely conserved, the lack of coevolution with other parts may indicate that they function as independent modules in the Fpg molecule. In particular, although we had specifically limited the search to Fpg homologs with an intact Cys4 zinc finger, the lack of coevolution in this motif is consistent with the existence of eukaryotic Fpg homologs carrying a variety of poorly conserved β-hairpins termed “zincless fingers” [78,79]. The three coevolved pairs selected for our analysis form seemingly important connections in all four molecules residing in the unit cell in the crystal structure of E. coli Fpg . Nevertheless, one of them, Y170–S208, turned out to be functionally dispensable and unstable in the MD runs, underscoring the importance of dynamics in the functional interpretation of features observed in static X-ray structures. The Tyr170–Thr214 interaction that formed instead in the MD runs seems to be much more important, since the Y170 F mutant had its activity reduced by more than three orders of magnitude, DNA binding decreased 34-fold, and the melting temperature moderately lowered. Of note, while Ser208 resides close to the oxoG-binding loop, Thr214 resides in the loop itself, so its interactions with Tyr170 can still be considered necessary for the correct conformation of the Fpg active site. Interestingly, while E. coli Fpg and Nei both have Ser in the position corresponding to Ser208, and both form a bond Ser–Tyr, Fpg from L. lactis, G. stearothermophilus, and Th. thermophilus have Ala instead of Ser, and in all these enzymes Tyr accepts a bond from Ser homologous to Thr214 in E. coli Fpg. Disruption of the remaining coevolved pairs had different consequences for Fpg stability and activity. Elimination of the R54–E131 bridge had no effect on the overall stability of the protein but totally abrogated substrate DNA binding and cleavage. The importance of interactions centered in the structure around this coevolved pair suggests that the correct maintenance or dynamics of interdomain connections is vital for Fpg activity. Notably, the R54–E151 bridge is not part of the interdomain DNA-binding groove, rather being exposed at the opposite side of the protein. It has been discussed for a long time whether Fpg can exist in the open (free) and closed (DNA-bound) form. These two conformations with different orientation of the domains have been found in the structures of Nei [70,80] but the only known structure of free Fpg, that of Th. thermophilus, is in the closed form . E. coli Nei in the closed form possesses a similar interdomain bridge involving Arg50, Asp128, and several adjacent residues, and the network of connecting bonds is reorganized in the open form. An increase in cooperativity of R54E mutants unfolding is suggestive of a loss of interactions formed by Arg54 and the linker residues with the resulting destabilization of partially unfolded states [67,68]. Overall, the mutagenesis, thermodynamic, and computational data presented here are consistent with the idea that DNA binding by Fpg/Nei family enzymes requires highly precise remodeling of interaction between the domains.