Role of Two Strictly Conserved Residues in Nucleotide Flipping and N-Glycosylic Bond Cleavage by Human Thymine DNA Glycosylase*

Thymine DNA glycosylase (TDG) promotes genomic integrity by excising thymine from mutagenic G·T mismatches arising by deamination of 5-methylcytosine, and follow-on base excision repair enzymes restore a G·C pair. TDG cleaves the N-glycosylic bond of dT and some other nucleotides, including 5-substituted 2′-deoxyuridine analogs, once they have been flipped from the helix into its active site. We examined the role of two strictly conserved residues; Asn140, implicated in the chemical step, and Arg275, implicated in nucleotide flipping. The N140A variant binds substrate DNA with the same tight affinity as wild-type TDG, but it has no detectable base excision activity for a G·T substrate, and its excision rate is vastly diminished (by ∼104.4-fold) for G·U, G·FU, and G·BrU substrates. Thus, Asn140 does not contribute substantially to substrate binding but is essential for the chemical step, where it stabilizes the transition state by ∼6 kcal/mol (compared with 11.6 kcal/mol stabilization provided by TDG overall). Our recent crystal structure revealed that Arg275 penetrates the DNA minor groove, filling the void created by nucleotide flipping. We found that the R275A and R275L substitutions weaken substrate binding and substantially decrease the base excision rate for G·T and G·BrU substrates. Our results indicate that Arg275 promotes and/or stabilizes nucleotide flipping, a role that is most important for target nucleotides that are relatively large (dT and bromodeoxyuridine) and/or have a stable N-glycosylic bond (dT). Arg275 does not contribute substantially to the binding of TDG to abasic DNA product, and it cannot account for the slow product release exhibited by TDG.

The chemically reactive bases in DNA are continuously modified by oxidation, alkylation, and deamination, creating mutagenic and cytotoxic lesions that are implicated in aging and diseases including cancer (1,2). Such damage is handled predominantly by the base excision repair pathway, initiated by one of many damage-specific DNA glycosylases (3). These enzymes use a nucleotide flipping mechanism to find damaged bases within the vast excess of normal DNA and cleave the base-sugar (N-glycosylic) bond to release the base, and follow-on base excision repair proteins complete the repair process.
Thymine DNA glycosylase (TDG) 2 removes thymine from mutagenic G⅐T mispairs, one of the few glycosylases that removes a normal base from DNA. Consistent with the need to avoid acting upon undamaged DNA, TDG activity is 18,000fold greater for G⅐T mispairs relative to A⅐T pairs (4). TDG is also specific for a particular DNA sequence, exhibiting the highest activity for G⅐T mispairs (and other lesions) with a 5Ј-flanking C⅐G base pair, i.e. damage located in CpG site (4 -7). Because cytosine methylation in vertebrates occurs selectively at CpG sites, the specificity of TDG indicates its predominant biological substrate is a G⅐T mispair arising from deamination of 5-methylcytosine (8).
Cytosine methylation at CpG sites is central for regulating gene expression and maintaining genomic stability (9). Another human DNA glycosylase has specificity for G⅐T mispairs arising at CpG sites, methyl-binding domain IV (10 -12), suggesting a biological imperative to maintain the integrity of CpG sites. Nevertheless, CpG sites exhibit a disproportionately high frequency of mutations (C 3 T) in human cancers and genetic disease (13)(14)(15), and it was suggested that the slow enzymatic turnover of TDG may contribute (16). An alternative explanation is raised by reports that TDG may participate in the demethylation of CpG sites by processing G⅐T mispairs created by active deamination of 5-methylcytosine (17,18), which would dramatically increase the burden of G⅐T mispairs. A preliminary report that homozygous knock-out of the TDG gene in mice results in embryonic lethality is consistent with such a role (19). Although these findings remain to be substantiated, they underscore the importance of understanding the catalytic mechanism of TDG.
TDG is relatively permissive and can remove 5-halogenated uracils (5-fluorouracil, 5-chlorouracil, 5-bromouracil, and 5iodouracil), many other 5-substituted uracils, 3,N 4 -ethenocytosine, hypoxanthine, and other damaged bases (see Ref. 4 and references therein). The CpG sequence specificity is retained for these substrates, consistent with the conclusion that CpG⅐T is the predominant biological substrate (8). Nevertheless, comparing the kinetic and binding parameters for a select group of substrates can provide much insight into the catalytic mechanism of TDG (4,20,21), and such an approach was used here. We previously observed robust activity for the excision of 5-halogenated uracils, including FU, ClU, and BrU, even from DNA sequence contexts other than CpG, with implications for the catalytic mechanism of TDG and for the cytotoxicity associated with the incorporation of these bases in DNA (4). Consistent with the potent in vitro activity for excising FU (4, 20 -22), a recent study finds that FU excision by TDG contributes to the DNA-mediated toxicity of FU in human and mouse cancer cells (23). Detailed studies of the TDG catalytic mechanism should further illuminate its ability to promote the toxicity of FU, a widely used anticancer agent. Toward this end, we have investigated the role of two strictly conserved active site residues of TDG, Asn 140 and Arg 275 .
Together, human TDG and the Escherichia coli mismatch-specific uracil DNA glycosylase (MUG, 32% identical) are the most thoroughly characterized members of the TDG-MUG family (5,19,24), part of the uracil DNA glycosylase (UDG) superfamily (1,25). We investigated the catalytic role of Asn 140 , which is found in a motif ( 138 GINPG 142 ) that is strictly conserved in TDG-MUG enzymes (25). Previous structural and biochemical studies indicate that Asn 140 is important, probably for the chemical step (22, 26 -29), yet its catalytic role had not been rigorously investigated for any member of the TDG-MUG family. To address this deficiency, we examined the role of Asn 140 by determining the effect of the N140A substitution on substrate binding and the base excision rate for a number of different substrates.
All of the DNA glycosylases employ nucleotide flipping to extrude the target nucleotide from the helix and gain access to the damaged base and the scissile N-glycosylic bond (30). Most glycosylases use a bulky side chain (Asn, Arg, Gln, Leu, or Tyr) to plug the helical space created by nucleotide flipping, although exceptions include 3-methyladenine DNA glycosylase, which has a Gly at this position (31). These "nucleotide flipping" residues can promote forward progress of the target nucleotide toward the active site and/or hinder its reverse flipping back into the helix, increasing its lifetime in the active site (30,32,33). A structure of TDG bound to abasic DNA shows that the Arg 275 side chain penetrates the minor groove, plugging the space vacated by the flipped abasic nucleotide and contacting two DNA backbone phosphates that flank the target site ( Fig. 1) (29). Although Arg 275 is strictly conserved in TDG from vertebrates and lower eukaryotes (19), the closely related MUG enzymes have a Leu in the corresponding position, as do the more distantly related UDG enzymes. We have examined the catalytic role of Arg 275 by determining the effect of R275A and R275L substitutions on substrate binding, the base excision rate (k max ), and the binding and release of abasic DNA product.

EXPERIMENTAL PROCEDURES
Materials-The DNA used for this work is shown in Fig. 2. Duplex DNA was hybridized by rapid heating to 80°C followed by slow cooling to room temperature. DNA oligonucleotides were synthesized at the Keck Foundation Biotechnology Resource Laboratory of Yale University (trityl-on), purified using Glen-Pak purification cartridges (Glen Research) following the manufacturer's instructions, and quantified by absorbance (260 nm) as previously described (20). Purity was verified by analytical anion exchange HPLC under denaturing (pH 12) conditions (20). Phosphoramidites for the noncanonical DNA nucleotides were obtained from Glen Research. The phosphoramidite used for incorporation of 2Ј-deoxy-2Ј-fluoroarabinouridine (U F ; Fig. 2) was obtained from ChemGenes Corp. (Wilmington, MA), and the phosphoramidite for 2Ј-deoxy-2Ј-fluoroarabinothymidine (T F ) (34) was obtained from Link Technologies (Lanarkshire, UK). These phosphoramidites were incorporated using standard phosphoramidite chemistry with coupling times extended to 5 min. Control experiments demonstrate that duplex DNA containing U F or T F (paired with guanine) are not cleaved by TDG when incubated for an  The target nucleotide (x) is located in a CpG dinucleotide sequence context. T represents dT, U represents dU, and tetrahydrofuran abasic site analog (THF) represents the tetrahydrofuran abasic site analog. The structure of the U F analog is shown. T F is identical to U F except that the uracil base is replaced by thymine. extended time (Ͼ48 h) under single turnover conditions, with a DNA concentration of 0.5 M and an enzyme concentration of 5 M (not shown), consistent with previous findings for TDG (35) and MUG (27). Previous studies show that a single T F substitution slightly elevates the DNA duplex melting temperature (T m increased ϳ0.6°C) and does not significantly alter the overall structure of B-DNA and that the 2Ј-fluoroarabino substitution stabilizes an O4Ј-endo (east) sugar pucker that is compatible with B-DNA geometry (36).
Pre-steady-state Kinetics-Transient kinetics experiments were conducted at 22°C in HEMN.1 buffer (20 mM HEPES, pH 7.5, 0.1 M NaCl, 0.2 mM EDTA, 2.5 mM MgCl 2 ) with 0.1 mg/ml bovine serum albumin, quenched with 50% (v:v) 0.3 M NaOH, 0.03 M EDTA and heated for 15 min at 85°C to induce cleavage of the DNA backbone at abasic sites. The extent of product formation was analyzed using an HPLC assay as described (4,20). The maximal rate constant for product formation (k max ) was determined using single turnover experiments collected under saturating enzyme conditions ([E] Ͼ Ͼ [S] Ͼ Ͼ K d ), providing a rate constant that is not impacted by product release or the association of enzyme and substrate (4,20). The data were fitted by nonlinear regression to Equation 1 using Grafit 5 (37), where A is the amplitude, k obs is the rate constant, and t is the reaction time (min). We used a DNA substrate concentration of 0.5 M and a high enzyme concentration (typically 5 M) that is Ͼ100-fold higher than the K d values observed here and reported previously (8,35). Previous studies in our laboratory (4,20) indicate that these conditions are saturating and therefore provide the maximal rate constant for product formation (i.e. k obs Ϸ k max ). This was confirmed by collecting the data at an enzyme concentration of 10 M, which provided the same rate constant (within experimental error).
The maximal rate constant for enzymatic turnover (k cat ) was determined using pre-steady-state multiple-turnover kinetics, conducted with a high enzyme concentration (500 nM) and is not limited by the association of enzyme and substrate (21,38). Progress curves exhibited "burst" kinetics, with a rapid exponential phase followed by a slow linear phase, indicating that the rate of product formation greatly exceeds that of product release. The data were fitted to Equation 2, where A and k obs are the amplitude and rate constant of the exponential phase, v is the steady-state velocity, and t is the reaction time (min). The steady-state rate constant (k cat ) was obtained by dividing the steady-state velocity (v) by the amplitude (A). In all cases, k max /k cat Ͼ 60, indicating that the ratelimiting step occurs after the chemical step.
Isothermal Titration Calorimetry-ITC experiments were performed using a MicroCal VP-ITC instrument essentially as described previously (29). Protein and DNA samples were dialyzed versus ITC buffer (10 mM Tris-HCl, pH 7.5, 0.1 M NaCl, 0.5 mM tris(2-carboxyethyl)phosphine hydrochloride) and degassed prior to data collection. The experiments were collected (at 5°C) by titrating small volumes of DNA (concentration of 15-20 M) into a 1.4-ml sample cell containing enzyme (concentration of 2-3 M). The data were analyzed using pyITC, a python-based program with improved base-line estimates in raw heat data and automatic evaluation of dilution heat as a fitting parameter (29,39).

RESULTS
Our aim in this work was to probe the catalytic role of two strictly conserved residues of TDG, Asn 140 and Arg 275 , using site-directed mutagenesis, pre-steady-state kinetics, and equilibrium binding experiments. As we and others have shown previously, the steady-state turnover of TDG is severely limited by a step following chemistry, i.e. release of the AP-DNA product (7,21,40). Accordingly, we used single turnover kinetics experiments under saturating enzyme conditions to determine the effect of Asn 140 and Arg 275 substitutions on k max , which reflects the maximal rate of (enzyme-bound) product formation and is not impacted by product release (Fig. 3) (20).
Essential Role for Asn 140 -We used single turnover kinetics and substrates including G⅐T, G⅐U, G⅐FU, and G⅐BrU to exam- FIGURE 3. Minimal kinetic mechanism for TDG. The association of enzyme and DNA gives a collision complex (E⅐S), and nucleotide flipping (involving conformational changes in E and S) gives the reactive enzyme-substrate complex (EЈ⅐SЈ). The chemical step (k chem ) involves cleavage of the N-glycosylic bond and addition of the nucleophile (water), giving the product complex (EЈ⅐B⅐PЈ). Product release involves a conformational change in E and P (not shown) and is slow. Also indicated are the reaction steps that contribute to the kinetic parameters k max , the maximal rate of product formation, and k cat , the maximal rate of steady-state turnover and to the binding constant (K d ) for substrate (analog).
ine the role of Asn 140 in the catalytic mechanism of TDG. We find no evidence of thymine excision from a G⅐T substrate (0.5 M) incubated with a saturating concentration of TDG-N140A (10 M), even after 48 h. The absence of measurable activity is not due to improper folding or instability of the N140A variant, as indicated by results from equilibrium binding studies presented below. Given the high sensitivity of our HPLC assay (20), we can place an upper limit on the activity of TDG-N140A for a G⅐T substrate, k max Ͻ 7 ϫ 10 Ϫ6 min Ϫ1 (i.e. Ͻ1% product in 24 h). Thus, the N140A substitution diminishes the base excision rate by at least 10 4.4 -fold (Table 1). We also examined the single turnover activity of TDG-N140A for other substrates. In contrast to a previous report (22), TDG-N140A retains some activity for a G⅐U substrate, albeit exceedingly low, k max ϭ 0.00011 Ϯ 0.00001 min Ϫ1 (Fig. 4A), revealing a 10 4.4 -fold loss in k max for the N140A substitution ( Table 1). The N140A substitution causes decreases in k max of 10 4.4 -and 10 4.1 -fold for the G⅐FU and G⅐BrU substrates, respectively ( Fig. 4 and Table 1). These huge effects on k max , together with the results of binding studies below, indicate an essential role for Asn 140 in the chemical step of the TDG reaction.
Asn 140 Is Dispensable for Substrate Binding-Given the essential role of Asn 140 in base excision, we next sought to determine its role in substrate binding, i.e. in forming the reactive Michaelis complex (EЈ⅐SЈ in Fig. 3). To determine the binding affinity (K d ) of wild-type TDG to substrate in the absence of base excision, we used DNA containing a 2Ј-deoxyuridine (dU) analog, U F , which flips into the active site but is resistant to N-glycosylic bond cleavage, as demonstrated previously for human TDG and for MUG and UDG from E. coli (27,35,41). As shown in Fig. 5, ITC experiments show that TDG binds the G⅐U F substrate analog with high affinity, K d ϭ 5 Ϯ 1 nM ( Table  2). The binding is endothermic, driven by favorable entropy (T⌬S Ͼ 0) that compensates for large unfavorable enthalpy (⌬H Ͼ 0), consistent with our previous results for TDG binding to abasic DNA product (29) and with the thermodynamic signature of other proteins that bind the minor groove of DNA (42). Remarkably, TDG-N140A binds the G⅐U F substrate analog with nearly the same affinity and thermodynamic parame-ters as wild-type TDG (Fig. 5 and Table 2). The exceedingly low base excision activity of TDG-N140A allows a determination of its binding affinity for natural substrate in the absence of base excision (i.e. Ͻ1% product formation in 2 h at 5°C). The affinity of TDG-N140A for a natural G⅐U substrate, K d ϭ 9 Ϯ 2 nM, is very similar to that observed for TDG-N140A and wild-type TDG binding to the G⅐U F substrate analog (Fig. 5 and Table 2). Our results are consistent with previous findings that U F is a good mimic of the natural dU substrate (27,35,41).
We also examined the binding of TDG and the N140A variant to DNA containing a dT substrate analog, T F . TDG binds the G⅐T F substrate analog with high affinity, K d ϭ 36 Ϯ 7 nM   Table 1 and represent the average values obtained from at least three independent experiments. (  (8). Given the absence of thymine excision activity for TDG-N140A, we determined its binding affinity for a natural G⅐T substrate, finding K d ϭ 22 Ϯ 1 nM (Table 2). In addition, TDG-N140A binds the G⅐BrU substrate with similar affinity (Table 2). Our observations are in contrast with a previous study reporting that TDG-N140A has weaker binding affinity than native TDG for G⅐U and G⅐T substrate DNA (22). Our finding that the N140A substitution halts the chemical step while not altering substrate binding indicates the N140A variant is a useful platform for investigating the role of other residues in substrate binding, as illustrated for Arg 275 below. Although the U F analog has been employed in the study of many DNA glycosylases (27,35,41), use of the T F analog has not previously been reported for these enzymes. Our finding that the N140A variant binds with similar affinity and thermodynamic parameters to G⅐T and G⅐T F (Table 2) indicates that T F is a good mimic of dT and should be useful in future structural and mechanistic studies of TDG and other glycosylases that excise thymine from DNA.
Effect of Arg 275 Substitutions on Base Excision-We used single turnover kinetics experiments to determine the effect of Arg 275 substitutions on k max . For R275A the steric bulk and positive charge (or at least the polarity) of the Arg side chain are both removed (Fig. 1). Some bulk is retained for R275L, but the shapes of the Leu and Arg side chains differ substantially, and Leu is nonpolar. The R275L variant was also chosen because the related MUG and UDG enzymes have strictly conserved Leu residues in the positions corresponding to Arg 275 of TDG (25,26,29). We used several different target nucleotides to probe the contribution of Arg 275 to nucleotide flipping and the chemical step of the TDG reaction, which can both impact k max (Fig.  3) (4, 20). For the G⅐FU substrate, the R275A and R275L substitutions give very small effects of 1.6-and 1.4-fold, respectively (Table 1). This indicates essentially no effect on the chemical step for the cleavage of FdU, which is relatively small and is rapidly cleaved by TDG, consistent with our previous findings (4,20). Likewise, for a G⅐U substrate, we find small effects on k max of 2-and 3-fold for the R275A and R275L substitutions. In contrast, for the G⅐BrU substrate, k max is decreased 4-and 16-fold by the R275A and R275L substitutions, respectively. The largest effects are observed for the G⅐T substrate, where k max is decreased 8-fold for R275A and 30-fold for R275L ( Fig. 6 and Table 1). As discussed in detail below, these observations FIGURE 6. Effect of R275A and R275L substitutions on k max for a G⅐T substrate. Shown are representative data from single turnover kinetics experiments for processing the G⅐T substrate by TDG (A), the R275A variant (B), and the R275L variant (C). Rate constants (k max ) determined from these experiments are provided in Table 1 and represent the averages obtained from at least three independent experiments. The enzyme concentration was 5 M, and the DNA substrate concentrations were 0.5 M.

TABLE 2 Binding of TDG and enzyme variants to DNA containing substrate and product sites
The values reported represent the averages and standard deviations from at least two and typically three independent ITC experiments. indicate that Arg 275 serves to promote and/or stabilize nucleotide flipping (i.e. it increases k 2 and/or decreases k Ϫ2 ; Fig. 3).

DNA
Arg 275 Substitutions Decrease Substrate Binding-We used ITC to examine the effect of R275A and R275L substitutions on binding of TDG to G⅐U F and G⅐T F substrate analogs (Fig. 7). The R275A and R275L substitutions weaken the binding for G⅐U F by 2.9-and 6.3-fold (Table 2). Similarly, the R275A and R275L variants bind 3.3-and 3.6-fold more weakly to the G⅐T F substrate analog (Table 2). To examine the effect of the R275L substitution on the binding affinity for natural substrates, we produced the N140A/R275L enzyme, which seems reasonable because the N140A substitution alone halts the chemical step while not significantly altering substrate binding (Tables 1  and 2). The binding affinity for a G⅐U substrate is 5.6-fold weaker for N140A/R275L versus N140A, the same effect on K d as observed for binding of R275L to the G⅐U F substrate analog (Table 2). Similarly, for the G⅐T and G⅐BrU substrates, the binding affinity is about 4-fold weaker for N140A/R275L relative to N140A ( Table 2). Because Arg 275 is unlikely to contribute to the initial enzyme-substrate (collision) complex (E⅐S; Fig. 3), the effects of Arg 275 substitutions on substrate binding likely reflect a change in nucleotide flipping, i.e. a decrease in k 2 or an increase in k Ϫ2 (Fig. 3), consistent with the effects of Arg 275 substitutions on k max .
Role of Arg 275 in Binding AP-DNA Product-Previous studies by us and others have shown that TDG binds AP-DNA with high affinity and that AP-DNA product release is slow (7,21,29,40,43). The crystallographic observation that Arg 275 likely promotes and/or stabilizes nucleotide flipping (Fig. 1) raised the possibility that Arg 275 might contribute to the tight binding and slow release of AP-DNA. Remarkably, the binding affinity for DNA product is the same (ϳ10 nM) for TDG and the R275A and R275L variants (Table 2), indicating no significant contribution for Arg 275 to product binding. We also collected pre-steady-state multiple-turnover (burst) kinetics experiments to determine the effect of the Arg 275 substitutions on steady-state turnover (k cat ). For enzymes exhibiting rate-limiting product release (like TDG), progress curves for these experiments exhibit an exponential (burst) phase, reflecting the rapid formation of enzymebound product, followed by a slow steady-state phase that provides the maximal turnover rate (k cat ) (38). As shown in Fig. 8, the R275A and R275L substitutions have essentially no effect on k cat for TDG acting upon a G⅐U substrate (k cat values given in Table 1). We note that the k cat values observed here are essentially the same as those observed previously for a similar G⅐U20 substrate (21). These findings indicate that the DNA interactions involving the Arg 275 side chain do not contribute substantially to the tight binding or slow release of AP-DNA product by TDG.

Asn 140 Is Essential for the Chemical
Step of the TDG Reaction-Asn 140 is strictly conserved in TDG and MUG enzymes, and previous structural and biochemical studies suggested that this  Burst kinetics experiments were collected using a 1000 nM concentration of the G⅐U substrate and 500 nM concentration of either TDG (E), TDG-R275A (F), or TDG-R275L (Ⅺ). The rate of the steady-state phase was not substantially changed by the Arg 275 substitutions (k cat values given in Table 1). residue is important for the chemical step of the reaction (26 -29), yet its catalytic role had not been rigorously examined for any member of the TDG-MUG enzyme family. Our results demonstrate that Asn 140 is essential for the chemical step but does not contribute substantially to prior steps, including formation of the reactive enzyme-substrate complex in which the target nucleotide is flipped into the active site (EЈ⅐SЈ in Fig. 3).
Indeed, the N140A substitution causes a huge decrease of at least 10 4.4 -fold in the base excision rate (k max ) for a G⅐T substrate (Table 1), indicating that Asn 140 stabilizes the transition state by Ͼ6 kcal/mol. For comparison, the total rate enhancement provided by TDG for G⅐T substrate is k max /k non ϭ 10 8.5 , corresponding to 11.6 kcal/mol transition state stabilization (20). A similarly large role for Asn 140 in the chemical step is observed for the G⅐U, G⅐FU, and G⅐BrU substrates (Table 1). Remarkably, the N140A substitution does not substantially alter the binding affinity of TDG for the G⅐T F or G⅐U F substrate analogs ( Table 2), indicating that Asn 140 does not contribute to substrate binding or nucleotide flipping. Similar results were obtained for the corresponding residue in UDG, which is an Asp rather than Asn. The catalytic Asp of UDG stabilizes the transition state by ϳ5 kcal/mol, compared with 17 kcal/mol transition state stabilization provided by UDG in total (44), and the Asp does not contribute to substrate binding (44). Thus, Asn 140 of TDG accounts for a larger fraction of transition state stabilization than the catalytic Asp of UDG.
Our findings, together with previous studies, suggest a potential role for this essential Asn of the TDG-MUG enzyme family. Many previous studies show that hydrolysis of the N-glycosylic bond follows a highly dissociative or stepwise mechanism for both enzymatic and noncatalyzed reactions (45). The reactions catalyzed by UDG and MutY (from E. coli) are stepwise, where rupture of the N-glycosylic bond precedes the addition of the nucleophile (water) (46 -48). For UDG, N-glycosylic bond cleavage produces a highly unstable glycosyl cation-uracil anion intermediate, and the catalytic Asp appears to play three key roles: stabilizing the glycosyl cation, positioning the nucleophilic water molecule, and activating the nucleophile for attack (via general base catalysis) (44,49,50).
A stepwise mechanism seems reasonable for the TDG reaction (20,45), but the catalytic Asn likely plays a very different role from the Asp of UDG. Our findings and previous studies suggest that the primary role of the essential Asn in TDG-MUG enzymes is to bind and properly position the nucleophilic water molecule. As illustrated in Fig. 9, the crystal structures of TDG and MUG (free enzymes) reveal a water molecule bound to the catalytic Asn via hydrogen bonds to its side chain and backbone carbonyl oxygens, and structures of enzyme-DNA complexes show the Asn would position the water for nucleophilic attack at C1Ј of the flipped nucleotide (26,27,29). However, unlike the catalytic Asp of UDG, the Asn of TDG cannot activate the nucleophile via general base catalysis, and its side chain carbonyl oxygen (O␦1; Fig. 9) does not appear close enough (to C1Ј) to stabilize the glycosyl cation (26 -29). Thus, for TDG-MUG enzymes the glycosyl cation is likely stabilized by the anionic base (leaving group) and neighboring DNA phosphates, as observed for UDG (48,51), and potentially by the oxygen of the nucleophilic water molecule.
In the absence of nucleophile activation by a general base, it seems reasonable that a proton may be transferred from the nucleophile to the anionic base, which would also serve to stabilize the leaving group (preventing its reassociation with the glycosyl cation). A recent computational study finds that the noncatalyzed hydrolysis of 2Ј-deoxyuridine involves transfer of a proton (via intervening water molecules) from the nucleophile to O 2 of the uracil anion (52). As noted previously, the structure of the MUG enzyme-substrate complex shows uracil O 2 is poised to abstract a proton from the nucleophile (Fig. 9), which could be favorable for the anionic base that results from N-glycosylic bond cleavage (27,45). Of course, alternative pathways for proton transfer are possible and could potentially involve the backbone carbonyl oxygen of Asn 140 (Fig. 9). Such a mechanism for nucleophile activation and leaving group stabilization may be important for TDG-MUG enzymes, which do not provide many interactions to stabilize the leaving group or a side chain for general base catalysis (26,27,29). The proper positioning of the nucleophile by Asn 140 may be particularly important for such a mechanism.
Role for Arg 275 in Nucleotide Flipping-Our results indicate that the Arg 275 side chain promotes and/or stabilizes nucleotide flipping, and its contribution to base excision is greater for target nucleotides that have a relatively large base (i.e. dT, BrdU) and/or a stable N-glycosylic bond that resists cleavage (such as dT). This is illustrated by considering the differing effects of Arg 275 substitutions on the base excision rate (k max ) for the various substrates used in this work.
We find essentially no effect on k max for the G⅐FU substrate, indicating that Arg 275 does not contribute substantially to transition state stabilization in the chemical step. It also indicates that any effect of the Arg 275 substitutions on nucleotide flipping (k 2 or k Ϫ2 ; Fig. 3) does not diminish the active site lifetime of FdU to an extent that k max is altered. We note that FdU is the most rapidly cleaved substrate for TDG (Table 1).
In contrast to the absence of an effect for FdU, the R275A and R275L substitutions substantially decrease k max (by 4-and FIGURE 9. Catalytic Asn in TDG-MUG enzymes. Crystal structures of TDG and MUG show that the catalytic Asn binds a water molecule, positioning it for nucleophilic attack C1Ј of the target nucleotide (dU). Shown is the catalytic Asn and the flipped dU nucleotide from crystal structures of free MUG (Protein Data Bank code 1MUG) and MUG bound to DNA containing the U F analog shown in Fig. 2 (Protein Data Bank code 1MWJ). The Asn-bound water is observed in the structure of free MUG, which was aligned with MUG from the U F -DNA complex (for which the water is not observed) to generate the figure. The catalytic Asn from the MUG-U F -DNA complex is shown but obscured by the Asn from free MUG (cyan). The figure was produced with PyMOL.
16-fold) for the G⅐BrU substrate (Table 1). To account for this difference, it is important to note that wild-type TDG cleaves FdU 17-fold faster than BrdU (Table 1). As we suggested previously, the difference in wild-type TDG activity is likely due to steric effects (20). BrdU and FdU exhibit the same N-glycosylic bond stability and the same noncatalyzed rate of bond cleavage (k non ), but BrdU is much larger and is likely more difficult to flip and/or maintain in the TDG active site than FdU (20). Thus, given the equivalent susceptibility of FdU and BrdU to N-glycosylic bond cleavage (same k non ), the much larger effect of Arg 275 substitutions on k max for G⅐BrU versus G⅐FU substrates indicates an important role for Arg 275 in promoting and/or stabilizing nucleotide flipping for bulky substrates such as BrdU.
Consistent with this idea, the effects of the Arg 275 substitutions are greater for G⅐BrU versus G⅐U, even though the N-glycosylic bond is much more readily cleaved for BrdU than for dU. Evidently the Arg 275 substitutions have a substantially greater impact on the active site lifetime of BrdU than dU.
The Arg 275 substitutions exert the greatest effect on base excision rate (k max ) for the G⅐T substrate (Table 1). Although dT is slightly smaller than BrdU, the N-glycosylic bond of dT is much more resistant to cleavage (20). The significant effect on k max for both substrates indicates that substitution of Arg 275 impairs flipping for dT and BrdU nucleotides. The larger k max effect for dT likely reflects the greater impact of decreased active site lifetime for a nucleotide with a more stable N-glycosylic bond.
The observation that stabilization of nucleotide flipping by Arg 275 is most important for bulky nucleotides with a stable N-glycosylic bond (dT) is consistent with previous findings that the effects on k max of altering the CpG context or changing the partner of the target base are largest for cleavage of dT and smallest for FdU (4). Although our findings confirm an important role for Arg 275 in nucleotide flipping, as predicted by our recent crystal structure, future studies will be required to further elucidate the detailed role of Arg 275 , i.e. whether it serves to promote forward flipping (increase k 2 ) and/or slow the rate of reverse flipping of the target nucleotide back into the DNA helix (decrease k Ϫ2 ).
Leu Is a Poor Substitution for Arg 275 -As discussed above, the base excision activity (k max ) of TDG for G⅐T and G⅐BrU substrates is substantially decreased by the Arg 275 substitutions. Surprisingly, the activity is ϳ4-fold lower for the Leu variant (R275L) as compared with the Ala variant (R275A) ( Table 1). This result was unexpected because most DNA glycosylases employ a bulky residue to stabilize nucleotide flipping (30) and because the related MUG and UDG enzymes use Leu rather than Arg for nucleotide flipping (25,26,29). One potential explanation for the lower activity of R275L versus R275A involves steric hindrance between methyl groups of the Leu side chain and the nucleotides that flank the target nucleotide (Fig. 1B). To explore this idea, we replaced Arg 275 with Leu in the TDG-AP-DNA crystal structure and produced a model structure for each of the four common Leu rotamers using the program Coot (53). These model structures reveal steric hindrance between the 3Ј and 5Ј-flanking nucleotides and the Leu side chain; the location and extent of the steric clashes vary with the Leu rotamer (not shown). In all models the Leu side chain clashes to some extent with the 3Ј-dG nucleotide. Importantly, TDG interacts with the 3Ј-dG to obtain specificity for G⅐T mispairs and other lesions in a CpG sequence context (5Ј-CpG/5Ј-TpG) (29), and perturbing these interactions significantly diminishes base excision activity (4). Thus, replacement of Arg 275 with Leu could potentially perturb TDG interactions with the 3Ј-dG. In contrast, the Arg side chain threads nicely through the gap between the 3Ј-and 5Ј-flanking nucleotides, and, unlike Leu, the Arg side chain interacts favorably with the anionic phosphates (Fig. 1). Although the electrostatic interactions and steric bulk of Arg are absent for the Ala substitution (R275A), Ala does not clash with the 3Ј and 5Ј nucleotides, which may explain the smaller effect on base excision for R275A versus R275L.
Although replacement of Arg 275 with Leu gives a modest 3-fold reduction in G⅐U activity, it causes a much larger 30-fold reduction in G⅐T activity (Table 1). It is also observed that Arg 275 is strictly conserved in TDG enzymes from vertebrate species, for which the biological substrate is likely a G⅐T mispair in a CpG site, whereas Leu is strictly conserved in the nucleotide flipping position of MUG enzymes, which act upon G⅐U but have essentially no G⅐T activity. Together, these observations indicate that an Arg in the nucleotide flipping position is necessary for the efficient cleavage of dT. Perhaps the extended length and/or DNA backbone contacts provided by the Arg side chain are needed to stabilize nucleotide flipping and increase the active site lifetime of dT, which is bulkier and more resistant to cleavage than dU.