The structure of the S. aureus
TAG–3-MeA complex was determined to 1.8 Å resolution and that of the Y16F TAG–3-MeA complex to 2.22 Å resolution. The structure of the native 3-MeA complex is very similar to the crystal structure of the S. typhi
TAG–3-MeA–abasic DNA complex (Metz et al.
) and the NMR structure of the E. coli
TAG–3-MeA complex (Cao et al.
). Relative to apo TAG (Oke et al.
), Glu38 has rotated to make 2.7 Å contacts with the exocyclic N atom and N7 of 3-MeA. Tyr16 moves to make a 2.8 Å contact with the exocyclic N atom of 3-MeA (Fig. 3
). Trp46 stacks with the bound purine ring of 3-MeA, while Phe6, Tyr13 and Tyr21 make edge-on contacts. His41 rotates 80° to create space for 3-MeA to bind. The Y16F-mutant complex revealed that 3-MeA adopts a different orientation, although it preserves a bidentate hydrogen bond to Glu38 and a stacking interaction with Trp46 (Fig. 3
). This conformation is unlikely to be physiologically relevant, as it would require a very different orientation of the DNA to that observed in the S. typhi
complex (Metz et al.
). Using a fluorescence assay, we measured 3-MeA binding (Fig. 2
), obtaining a similar result at pH 7.8 (K
= 78 µM
) to that for the E. coli
enzyme at pH 7.5 (K
= 42 µM
; Cao et al.
). However, the assay is flawed for the S. aureus
enzyme as the E38Q mutant gave the same result as for the native protein (Fig. 2
), which is physically unreasonable. ITC (Figs. 2
) showed clear differences between the native and mutant S. aureus
enzymes (Y16F, K
= 1.2 mM
; E38Q, no binding) and gave K
values of 220 µM
at pH 7.8 and 471 µM
at pH 5.8 for the native enzyme. We did not detect adenine binding.
Figure 3 (a) Structure of the 3-MeA–TAG complex (C atoms, yellow; N atoms, blue; O atoms, red) showing the key interactions. The apo structure is shown with C atoms in white. (b) Structure of the 3-MeA–Y16F TAG complex (C atoms shown in pink); (more ...)
3-Methyldeoxyadenosine is positively charged in DNA, whilst deoxyadenosine is neutral; simple charge–charge recognition was therefore the original explanation for the specificity of TAG (Labahn et al.
; Lau et al.
; Hollis et al.
). However, it has been shown that E. coli
TAG binds 3-MeA but not adenine and binds protonated 3-MeA (pH 5.7) more weakly than neutral 3-MeA (pH 7.5) (Cao et al.
; Drohat et al.
), establishing that charge–charge recognition is not the sole explanation (Cao et al.
). We suggest that a particular hydrogen-bond pattern contributes to the selection of a specific but favoured (Sharma & Lee, 2002
) neutral tautomer of 3-MeA (Fig. 3
) that is not available to adenosine (Fig. 3
) and that is disfavoured for protonated 3-MeA (Fig. 3
). Our hypothesis implies that there is an energetic penalty in reorganizing the hydrogen-bond network around Tyr16 to avoid a van der Waals clash (Fig. 3
). In DNA, 3-methyldeoxyadenosine can adopt a tautomer that has the same hydrogen arrangement as neutral 3-MeA and has positive charge (Fig. 3
), which is favoured at the active site (Metz et al.
). A clash of H atoms was observed between the amide of His136 and the amino group of adenine in human AAG and is used to preferentially select the damaged purine base (O’Brien & Ellenberger, 2004
). Higher resolution data or neutron diffraction are required to further test the hypothesis for the TAG enzyme.