Lesion search and recognition by thymine DNA glycosylase revealed by single molecule imaging

Abstract

The ability of DNA glycosylases to rapidly and efficiently detect lesions among a vast excess of nondamaged DNA bases is vitally important in base excision repair (BER). Here, we use single molecule imaging by atomic force microscopy (AFM) supported by a 2-aminopurine fluorescence base flipping assay to study damage search by human thymine DNA glycosylase (hTDG), which initiates BER of mutagenic and cytotoxic G:T and G:U mispairs in DNA. Our data reveal an equilibrium between two conformational states of hTDG-DNA complexes, assigned as search complex (SC) and interrogation complex (IC), both at target lesions and undamaged DNA sites. Notably, for both hTDG and a second glycosylase, hOGG1, which recognizes structurally different 8-oxoguanine lesions, the conformation of the DNA in the SC mirrors innate structural properties of their respective target sites. In the IC, the DNA is sharply bent, as seen in crystal structures of hTDG lesion recognition complexes, which likely supports the base flipping required for lesion identification. Our results support a potentially general concept of sculpting of glycosylases to their targets, allowing them to exploit the energetic cost of DNA bending for initial lesion sensing, coupled with continuous (extrahelical) base interrogation during lesion search by DNA glycosylases.

Figure 1.

Specific complexes of hTDG at G:U lesion sites. (A) AFM image of the catalytically inactive hTDG variant N140A on DNA substrate containing a G:U mismatch positioned at 46% of the DNA fragment length. Arrows point to specific TDG complexes bound at the target site (yellow) and nonspecific complexes bound elsewhere on homoduplex DNA (white). The insert shows a higher magnification of a representative (specific) hTDG–DNA complex. (B) TDG binding position distribution on DNA demonstrating a moderate binding preference for the G:U mismatch over nonspecific DNA (enhanced occupancy at ∼46% DNA length). Fractional occupancies are plotted for ∼22 bp long sections of the 549 bp DNA substrate from DNA fragment ends (0%) to DNA center (50%). (C) Distribution of DNA bend angles induced by hTDG-N140A at G:U sites. A double Gaussian fit (R² = 0.88) centered at (29 ± 10) and (68 ± 10)°. Bend angles are binned at intervals of 8 bp. (D) Distribution of intrinsic DNA bend angles at the G:U mismatch in the absence of protein revealed a similar DNA bend angle of (30 ± 8)° as in (C) in addition to a predominant straight DNA conformation (0 ± 8)°. Results were pooled from three individual experiments (n = total number of data points) and are summarized in Table 3.

Figure 2.

Specific complexes of hTDG-R275A at G:U^F lesion sites. (A) AFM image of hTDG-R275A on DNA substrate containing a G:U^F mismatch at 46% of DNA fragment length. Arrows point to specific hTDG-R275A–DNA complexes bound to the target site (yellow) and nonspecific complexes (white). (B) A Gaussian fit to the hTDG-R275A binding position distributions on G:U^F DNA revealed a lesion specificity of S = (358 ± 101). (C) Distribution of DNA bend angles induced by hTDG-R275A at G:U^F sites. A triple Gaussian fit (R² = 0.89) centered at (0 ± 12), (37 ± 12) and (66 ± 12)°. (D) Distribution of intrinsic DNA bend angles at the G:U^F mismatch in the absence of protein revealed a similar slightly bent state (33 ± 12)° as in (C) and a predominant linear state (2 ± 12)°. Results were pooled from three individual experiments (n = total data points) and are summarized in Table 3.

Figure 3.

Base flipping activities of TDG. Increasing concentrations of (A) hTDG wt and (B) hTDG-R275A were titrated to 170 nM 2-AP DNA substrates. Steady-state fluorescence emission spectra (340–400 nm) were recorded at λ_ex = 320 nm. The inset in (A) shows a schematic of the 2-AP (2, cyan) DNA substrate containing G:U^F (yellow). (C) Quantification of 2-AP fluorescence intensity increases in (A) and (B) in the presence of 5 μM protein. Results were derived from two individual titrations. The error bars indicate the SD. Significance is classed as *P < 0.05.

Figure 4.

Characterization of TDG complexes with nonspecific DNA. (A) Representative AFM image of hTDG-N140A on nonspecific DNA (containing no lesion sites). (B) Distribution of DNA bend angles for hTDG-N140A complexes bound to nonspecific DNA. A double Gaussian fit (R² = 0.94) centered at (31 ± 12), (65 ± 12)°. (C) Distribution of DNA bend angles of hTDG-R275A bound to nonspecific DNA. A triple Gaussian fit centered at (-2 ± 13), (34 ± 13) and (65 ± 13)°. (D) Intrinsic DNA bend angles of nonspecific DNA substrates. A semi-Gaussian fit centered at (0 ± 21)°. Bend angle results were pooled from at least three individual experiments (n = total number of data points). (E and F) Increasing concentrations of (E) hTDG wt and (F) hTDG-R275A were titrated to 170 nM 2-AP DNA substrates. Steady-state fluorescence emission spectra (340–400 nm) were recorded at λ_ex = 320 nm. The inset in (E) shows a schematic of the DNA substrate used containing a 2-AP (2, cyan) neighboring a Watson–Crick G:C base pair (yellow). (G) Quantification of the 2-AP fluorescence intensity increase in (E) and (F) for protein concentrations of 5 μM. Results were derived from two individual titrations. The error bars indicate the SD. Significance is classed as ***P < 0.005.

Figure 5.

hTDG damage search model. Our proposed model for DNA lesion search and recognition by glycosylases involves three different types of TDG–DNA complexes: search complex (SC), interrogation complex (IC) and excision complex (EC). In the model, the glycosylase scans the DNA for intrinsically flexible sites, switching between an SC and an IC state. While homoduplex DNA is actively bent in the conformation of the SC (arrows indicate applied force), TDG target sites G:T and G:U form wobble base pairs (blue star) that display enhanced flexibility or intrinsic pre-bending that matches the conformation in the SC complex (passive bending). The energetic cost for DNA bending may hence serve as an initial damage-sensing mechanism that may result in a longer residence time of the glycosylase at a potential target site. In the IC, the enzyme probes base pairs via exertion of additional force on the DNA (e.g. phosphate pinching, arrows), resulting in a more strongly bent DNA structure. Protein–DNA interactions in the IC conformation may expedite spontaneous base flipping (question marks in figure) and stabilize the extrahelical base (for TDG e.g. Arg²⁷⁵). Correct target bases (or mismatched pairs) are then identified in the catalytic pocket of the glycosylase in the EC conformation and catalysis of base excision can occur.