Coordinating the Initial Steps of Base Excision Repair

DNA glycosylases initiate base excision repair by removing damaged or mismatched bases, producing apurinic/apyrimidinic (AP) DNA. For many glycosylases, the AP-DNA remains tightly bound, impeding enzymatic turnover. A prominent example is thymine DNA glycosylase (TDG), which removes T from G·T mispairs and recognizes other lesions, with specificity for damage at CpG dinucleotides. TDG turnover is very slow; its activity appears to reach a plateau as the [product]/[enzyme] ratio approaches unity. The follow-on base excision repair enzyme, AP endonuclease 1 (APE1), stimulates the turnover of TDG and other glycosylases, involving a mechanism that remains largely unknown. We examined the catalytic activity of human TDG (hTDG), alone and with human APE1 (hAPE1), using pre-steady-state kinetics and a coupled-enzyme (hTDG-hAPE1) fluorescence assay. hTDG turnover is exceedingly slow for G·T (kcat = 0.00034 min-1) and G·U (kcat = 0.005 min-1) substrates, much slower than kmax from single turnover experiments, confirming that AP-DNA release is rate-limiting. We find unexpectedly large differences in kcat for G·T, G·U, and G·FU substrates, indicating the excised base remains trapped in the product complex by AP-DNA. hAPE1 increases hTDG turnover by 42- and 26-fold for G·T and G·U substrates, the first quantitative measure of the effect of hAPE1. hAPE1 stimulates hTDG by disrupting the product complex rather than merely depleting (endonucleolytically) the AP-DNA. The enhancement is greater for hTDG catalytic core (residues 111–308 of 410), indicating the N- and C-terminal domains are dispensable for stimulatory interactions with hAPE1. Potential mechanisms for hAPE1 disruption of the of hTDG product complex are discussed.

The nucleobases in DNA are continuously modified by processes involving deamination, methylation, and oxidation, generating mutagenic and cytotoxic lesions that are implicated in aging and diseases including cancer and neurodegeneration (1,2). Such damage is handled by the highly conserved base excision repair (BER) 3 pathway, initiated by a damage-specific DNA glycosylase (3,4). These remarkable enzymes use a nucleotideflipping mechanism to find lesions and cleave the N-glycosidic bond, releasing the damaged base and producing an abasic or apurinic/apyrimidinic (AP) site in the DNA (5). Many DNA glycosylases remain tightly bound to their AP-DNA product, impeding enzymatic turnover. This may reflect a need to protect against the mutagenic and cytotoxic properties of AP sites, which impede some DNA polymerases, lack base coding information if replicated, and lead to single-strand breaks. Previous studies of BER in organisms ranging from Escherichia coli to humans have shown that AP endonucleases stimulate the activity (turnover) of many DNA glycosylases (6 -10). These important findings suggest some degree of coordination in the initial steps of BER, yet the mechanism remains largely unknown.
We address this question here for human thymine DNA glycosylase (hTDG), which removes T from G⅐T mispairs and excises many additional lesions, with specificity for damaged bases that are paired with guanine and located in a CpG sequence context (11)(12)(13)(14)(15). A crystal structure of the hTDG catalytic domain (hTDG cat , residues 111-308) bound to abasic DNA indicates the CpG specificity arises from interactions that select for guanine as the pairing partner of the target base and for guanine 3Ј to the target base (i.e. 5Ј-CpG-3Ј/5Ј-XpG-3Ј, where X is the target base) (16). The specificity of hTDG for damage in a CpG context suggests the predominant biological substrate is G⅐T mispairs arising from deamination of 5-methylcytosine (m 5 C) to thymine (17), because DNA methylation occurs at cytosine (C5) of CpG dinucleotides. CpG methylation is an epigenetic modification that plays a central role in regulating gene expression and maintaining genomic stability. Another human DNA glycosylase exhibits specificity for G⅐T mispairs at CpG sites, methyl binding domain IV (18 -21), which may reflect a biological imperative to preserve the integrity of CpG sites. Nevertheless, CpG sites exhibit a disproportionately high frequency of mutations (C 3 T) in human cancers and genetic disease (22)(23)(24), and it was suggested this may be attributable in part to the slow turnover of hTDG (12). The importance of understanding hTDG-initiated BER is underscored by findings that it may participate in the active demethylation of m 5 CpG sites, hence transcriptional regulation, by processing G⅐T mispairs created by active deamination of m 5 C to T (25,26). Such a role is consistent with a preliminary report that homozygous knock-out of the TDG gene is embryoniclethal in mice (13). We note that if demethylation of m 5 CpG sites involves active deamination, it would dramatically increase the burden of G⅐T mispairs, providing an alternative explanation for the high mutational frequency observed at CpG sites.
Previous studies have shown that hTDG exhibits exceedingly slow turnover after converting a stoichiometric amount of G⅐T or G⅐U substrate to G⅐AP product (7,27,28). This is attributable to very slow product release; a dissociation constant of k off ϭ ϳ0.0016 min Ϫ1 (half-life of 7 h) was estimated for G⅐AP-DNA on the basis of electrophoretic mobility shift studies (7). Accordingly, hTDG binds abasic DNA with high affinity (29), and abasic DNA is a potent inhibitor of hTDG (27). In contrast, hTDG is not inhibited by the nucleobases that it removes from DNA, including thymine or 3,N 4 -ethenocytosine (⑀C), even at a concentrations of up to 5 mM (17), indicating they do not bind hTDG with significant affinity.
It has been shown that human AP endonuclease 1 (hAPE1, also known as Ref-1) stimulates the turnover of hTDG for G⅐T, G⅐U, and G⅐⑀C substrates (7,17,30). However, the magnitude of the hAPE1 effect is unknown because the steady-state turnover (k cat ) of hTDG in the presence and absence of hAPE1 has not been reported for any substrate. Moreover, the mechanism of hAPE1 stimulation has remained elusive, although two basic ideas have been proposed. It has been suggested that hAPE1 stimulates hTDG by depleting (endonucleolytically) the concentration of AP-DNA, involving no interaction between hAPE1 and the hTDG product complex (13,31), a mechanism that we will refer to as "passive enhancement." On the other hand, the "active displacement" mechanism requires that hAPE1 interacts with hTDG and/or AP-DNA to disrupt the product complex (6,7,17,30,32) and may also involve hAPE1mediated depletion of AP-DNA.
Here, we explore the mechanism by which hAPE1 stimulates hTDG turnover using an experimental approach that has not previously been employed to address this question in BER. Using pre-steady-state multiple-turnover kinetics, we determined the maximal turnover rate (k cat ) for hTDG acting upon G⅐T, G⅐U, and G⅐FU substrates. We developed a coupled-enzyme (hTDG-hAPE1) fluorescence assay to measure the steady-state kinetic parameters of hTDG in the presence of hAPE1. Together, these methods provide a quantitative measure of the hAPE1 effect. We also examined the stimulatory effect of hAPE1 on the turnover of hTDG cat , which lacks the Nand C-terminal domains of hTDG (110 and 102 residues, respectively). Our results rule out the passive enhancement mechanism and require that hAPE1 actively displaces AP-DNA from the hTDG product complex. The coupled-enzyme assay described here provides a new approach for studying the mechanism of hTDG using steady-state kinetics.

EXPERIMENTAL PROCEDURES
Materials-DNA oligonucleotides were synthesized at the Keck Foundation Biotechnology Resource Laboratory of Yale University, purified by HPLC or Glen-Pak purification cartridges (Glen Research), and quantified by absorbance (260 nm) as described (14). Purity was verified by analytical anion-exchange HPLC under denaturing (pH 12) conditions (14). Phosphoramidites for special nucleotides were obtained from Glen Research. The duplex DNA substrates ( Fig. 1) were hybridized by rapid heating to 80°C followed by slow cooling to room temperature.
Pre-steady-state Kinetics-Transient kinetics experiments for hTDG (and hTDG cat ) were conducted at room temperature (ϳ22°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. Reactions were performed manually or by using a three-syringe rapid chemical quenched-flow instrument (RQF-3, Kintek Corp.). The extent of product formation was analyzed using an HPLC assay as described (14,15). . DNA substrates used in this work. U represents 2Ј-deoxyuridine (dU), FU represents 5-fluoro-dU, DabT and FAMT represent dabcyl-dT and 6-carboxyfluorescein-dT, respectively, where dabcyl and FAM are conjugated to the thymine base (C5 carbon) of dT (see Fig. 3). For each substrate the target base (bold) is located in a CpG dinucleotide context (underlined).
Single turnover experiments were collected for hTDG (and hTDG cat ) under saturating enzyme conditions ([E] Ͼ Ͼ [S] Ͼ Ͼ K D ) to obtain rate constants (for the maximal rate of AP product formation) that are not impacted by product release or the association of enzyme and substrate (14,15). Data were fitted by non-linear regression to Equation 1 using Grafit 5 (33), where A is the amplitude, k obs is the rate constant, and t is the reaction time (min). We use a concentration of enzyme, 5 M, that is Ͼ100-fold higher than previously reported K D values for hTDG (17,34) and 0.5 M substrate. These concentrations are sufficiently high to ensure saturating enzyme conditions, thereby providing the maximal rate constant for (enzymebound) product formation (i.e. k obs Ϸ k max ), as shown in our previous studies (14,15). To confirm saturating enzyme conditions, single turnover experiments were in some cases repeated with higher (10 M) and lower (2.5 M) enzyme concentrations, yielding the same result (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 and excess substrate ([S] Ͼ [E] Ͼ Ͼ K D ) such that k cat was not limited by the association of enzyme and substrate (35). Progress curves exhibited "burst" kinetics, with a rapid exponential phase followed by a slow linear phase, indicating that the rate of product formation (enzyme-bound) greatly exceeds that of product release (35). Data were fitted to Equation 2, ͓Product͔ ϭ A͕1 Ϫ exp͑Ϫk obs t͖͒ ϩ vt (Eq. 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 obs /k cat Ն 150, indicating the rate-limiting step occurs after chemistry. We used a high enzyme concentration (500 nM) and excess substrate (Ն1 M) to obtain maximal k cat values (35), as confirmed by repeating experiments at lower enzyme concentrations.
Coupled-enzyme Fluorescence Assay-We developed a coupled-enzyme (hTDG-hAPE1) fluorescence assay to accurately determine the steady-state kinetic parameters of hTDG (k cat and K m ) in the presence of hAPE1 ( Fig. 2A). As is typical for a coupled-enzyme assay (36), the product of the primary enzyme (hTDG) is a substrate for the coupling enzyme (hAPE1), and the steady-state velocity of hTDG is monitored by the formation of hAPE1 product. As shown below and as expected for a coupled-enzyme assay (36), the initial velocity (v 0 ) directly reflects the steady-state turnover of hTDG because we used a sufficiently high concentration of hAPE1 to ensure that hTDG turnover was rate-limiting. It is important to note that the steady-state turnover of hAPE1 (alone) is at least 10,000-fold faster than that observed here for hTDG (alone) (37). Using the coupled-enzyme assay, the steady-state kinetic parameters for hTDG (k cat and K m ) were obtained by determining the initial velocity (v 0 ) as a function of substrate concentration and fitting the data to the Michaelis-Menten equation (Equation 3) using non-linear regression with Grafit 5 (33), where k obs ϭ v 0 /[E] tot , and [E] tot is the total concentration of hTDG (or hTDG cat ). The kinetic parameters k cat and K m directly reflect the steady-state activity of hTDG because the experiments (v 0 determinations) were collected in the presence of sufficient hAPE1 such that the hTDG reaction was rate-limiting and v 0 was independent of [hAPE1], as shown below.
The DNA substrates (Fig. 1) contain a fluorescence quencher (dabcyl) at the 5Ј end of the substrate (i.e. dU-containing) strand and a fluorophore (FAM) at the 3Ј end of the complementary strand, such that FAM fluorescence is initially quenched (Fig. 2A). The substrates are similar to those used for previous "molecular beacon" assays for hAPE1 (38,39) but differ in one important respect as described below. The sequential activity of hTDG (creates an AP site) and hAPE1 (nicks the DNA backbone 5Ј to the AP site) releases a short dabcyl-con-FIGURE 2. Coupled-enzyme fluorescence assay for hTDG activity. A, schematic for the coupled-enzyme fluorescence assay. hTDG removes the target base (x), creating an AP site, and hAPE1 cleaves the DNA backbone 5Ј to the AP site, releasing the short dabcyl-containing oligonucleotide and generating a fluorescence increase. B, fluorescence emission spectra for the G⅐U20 substrate (1000 nM, lower curve) and the corresponding 5Ј-incised AP product generated by the coupled hTDG-hAPE1 reaction (upper curve). An 18-fold fluorescence increase was observed (519 nM). C, a typical progress curve (black) for the steady-state activity of hTDG cat (50 nM) acting upon the G⅐U20 substrate (500 nM) in the presence of hAPE1 (50 nM). The inset shows that the initial velocity (v 0 ) is highly linear over the first 50 s (r ϭ 0.998). The fluorescence of the G⅐U20 substrate is not altered by hAPE1 alone (red curve). The blue dotted line indicates the fluorescence after full conversion of G⅐U20 substrate to 5Ј-incised AP-DNA product. D, initial velocities (v 0 ) determined for fixed concentrations of hTDG cat (50 nM) and G⅐U20 substrate (1500 nM) and hAPE1 concentrations varying from 0 to 50 nM (F) show that v 0 does not increase for [hAPE1] Ͼ 50 nM. Initial velocities determined for a fixed concentration of hAPE1 (50 nM) and G⅐U20 (1500 nM) and varying hTDG cat concentrations (E) show that v 0 increases with [hTDG cat ].
taining oligonucleotide, giving a large fluorescence increase (Fig. 2B). Experiments were collected in 3-mm fluorescence cells (Starna Cells) maintained at 22°C using a QM-4 spectrofluorometer (Photon Technology International), with excitation and emission wavelengths of 491 and 516 nm.
The individual reactions (i.e. v 0 determinations) were initiated by adding hTDG to HEMN.1 buffer containing the DNA substrate and hAPE1 and were monitored by the change in fluorescence intensity. A typical progress curve obtained from the coupled-enzyme assay is shown in Fig. 2C. As expected, hAPE1 alone does not generate a fluorescence change for the hTDG substrates, which do not contain an AP site (Fig. 2C, red data). The initial velocities (v 0 ) were obtained by linear regression of fluorescence intensity (cps) versus time (s) for ϳ50 s at the beginning of the progress curve, which is highly linear (Fig.  2C, inset). The v 0 values were converted from units of fluorescence (cps⅐s Ϫ1 ) to product concentration (nM⅐s Ϫ1 ) using a conversion factor obtained from a plot of total fluorescence change (⌬F tot ) versus substrate concentration. These standard curves (⌬F tot versus [S]) were determined for [S] Յ 500 nM and were quite linear (r Ͼ 0.99). Values for ⌬F tot were determined by bringing the coupled reaction to full completion. For the G⅐U20 and G⅐FU20 substrates, reactions were rapidly brought to completion by adding 25 nM human uracil DNA glycosylase (hUNG) after measuring v 0 for hTDG. The fluorescence cells were carefully cleaned after each experiment to thoroughly remove hUNG. This was verified by ensuring the fluorescence of substrate with hAPE1 did not change over time (ϳ5 min) before adding hTDG. The presence of residual hUNG would be clearly indicated by a fluorescence increase due to the sequential activity of hUNG and hAPE1.
Control experiments were routinely preformed to confirm that the hAPE1 concentration was sufficiently high such that v 0 is independent of [hAPE1] and reflects the steady-state turnover of hTDG, i.e. that hTDG turnover is rate-limiting. For example, a series of v 0 determinations collected with 50 nM hTDG cat , 1500 nM G⅐U20 substrate, and varying hAPE1 concentrations (Fig. 2D) shows that v 0 is independent of hAPE1 concentration for [hAPE1] Ն 50 nM. As expected, v 0 increases linearly with hTDG cat concentration (for 12-150 nM hTDG cat ) in the presence of a fixed (50 nM) concentration of hAPE1 (and 1500 nM substrate). A similar approach was used for G⅐T and G⅐FU substrates to ensure the v 0 values are independent of [hAPE1] and reflect the steady-state turnover of hTDG.
hAPE1 Removes FAM and Dabcyl Linked to the 5Ј-or 3Ј Terminal Oxygen of DNA-Our coupled enzyme assay differs from previous molecular beacon assays involving hAPE1 (38,39), for which the fluorophore (FAM) and quencher (dabcyl) were linked by phosphodiester bond to the 5Ј-or 3Ј-terminal oxygen of the DNA (Fig. 3). We prepared a G⅐U substrate using this approach and found that hAPE1 alone generates a rapid fluorescence increase even though the DNA contained no AP site. In contrast, hAPE1 did not change the fluorescence of a G⅐U duplex with FAM and dabcyl linked to the thymine base of a terminal dT nucleotide, which involves no phosphodiester bond (Fig. 3). Control experiments with four (non-AP) duplexes in which one group (either FAM or dabcyl) was linked (phosphodiester bond) to the 3Ј-or 5Ј-terminal oxygen and the other group was dT-linked revealed a fluorescence increase for hAPE1 alone in all cases (i.e. for 3Ј-or 5Ј-linked FAM and for 3Ј-or 5Ј-linked dabcyl). We conclude that hAPE1 removes FAM or dabcyl linked to either the 3Ј-or 5Ј-terminal oxygen of DNA, presumably by hydrolyzing the phosphodiester bond. This activity is significant, up to 13% of the AP endonuclease activity (not shown). Accordingly, FAM-dT and dabcyl-dT were used for our coupled assay (Fig. 1).

RESULTS
Kinetics of hTDG Alone-To advance our understanding of how hAPE1 stimulates the turnover of hTDG, it was necessary to first determine the kinetic parameters for hTDG in the absence of hAPE1. Fig. 4 shows a minimal kinetic mechanism for the hTDG reaction. After DNA binding, nucleotide flipping brings the target base into the hTDG active site, and cleavage of the N-glycosidic bond (and addition of the water nucleophile) gives the ternary enzyme-product complex. Two possible pathways for product release are shown, where AP-DNA can be released before or after the excised base. The kinetic parameter k max reflects the maximal rate of (enzyme-bound) product formation and is governed by the reaction steps after DNA binding and before product release (i.e. nucleotide flipping and base excision). The kinetic parameter k cat reflects the maximal steady-state turnover of hTDG and reflects the same steps as k max plus product release (Fig. 4). Although k max is readily  NOVEMBER 21, 2008 • VOLUME 283 • NUMBER 47 determined for hTDG by using single turnover kinetics experiments (14,15,17,27), k cat is difficult to obtain from conventional steady-state kinetics due to the exceedingly slow product release. Accordingly, we determined k cat using pre-steady-state multiple-turnover (or burst) kinetics (35), an approach that has not been used previously for hTDG but has been employed for other glycosylases (40 -42). For enzymes exhibiting rate-limiting product dissociation (i.e. k max Ͼ Ͼ k cat ), burst kinetics experiments exhibit an exponential (burst) phase, reflecting rapid formation of enzyme-bound product followed by a slow steadystate phase that provides the maximal turnover rate (k cat ) and reflects the rate of product release (35). Fig. 5A shows a typical result from the burst kinetics experiments collected for hTDG (500 nM) acting upon the G⅐U20 substrate (1000 nM), which give k obs ϭ 1.4 Ϯ 0.1 min Ϫ1 for the exponential burst phase and k cat ϭ 0.005 Ϯ 0.001 min Ϫ1 for the steady-state phase (Table 1). Identical k cat values (within exper-imental error) were obtained for a lower hTDG concentration of 100 nM (not shown), indicating that our conditions provide the maximal turnover rate for hTDG (alone). We used single turnover kinetics to more accurately determine the maximal rate of (enzyme-bound) AP-DNA product formation, obtaining k max ϭ 2.2 Ϯ 0.3 min Ϫ1 (Fig. 5B). This is consistent with the exponential phase of the burst kinetics experiment (above) and all of our previous results from single turnover experiments for similar G⅐U substrates (14 -16, 43). As a control, the kinetics experiments were repeated with the G⅐U19 substrate, which is identical to G⅐U20 but does not contain the terminal FAM-dT or dabcyl-dT groups (which are needed for the coupled-enzyme fluorescence assay, discussed below). The results are essentially the same, k max ϭ 2.6 Ϯ 0.3 min Ϫ1 and k cat ϭ 0.007 Ϯ 0.001 min Ϫ1 (not shown), indicating FAM-dT and dabcyl-dT do not perturb hTDG activity.

APE1 Actively Increases Enzymatic Turnover of TDG
We also collected the burst kinetics and single turnover experiments for a G⅐T substrate, finding the steady-state turnover is exceedingly slow, k cat ϭ 0.00034 Ϯ 0.00007 min Ϫ1 (Fig.  5C), much slower than the maximal rate of product formation, k max ϭ 0.09 Ϯ 0.01 min Ϫ1 (Fig. 5D). The observation that k max exceeds k cat by 440-and 270-fold for G⅐U and G⅐T substrates, respectively, demonstrates that the turnover of hTDG (k cat ) is limited by a step after chemistry, likely AP-DNA product release (i.e. k cat Ϸ k off,AP ).
Kinetics for hTDG Catalytic Domain-We were also interested in examining the potential for hAPE1 to stimulate the turnover of hTDG cat , which contains the region of high similarity (32% identical) to E. coli mismatch-specific uracil glycosylase (16,28,44,45). Previous studies (28,43) and our findings here (Table 1) show that hTDG cat has nearly the same catalytic activity as hTDG for most substrates even though it lacks the Nand C-terminal domains (110 and 102 residues, respectively). We find the steady-state turnover of hTDG cat (500 nM) for the G⅐U substrate (1000 nM) is slow, k cat ϭ 0.006 Ϯ 0.002 min Ϫ1 (Table 1), and is identical to that observed for full-length hTDG. Identical k cat values (within error) were obtained for a lower hTDG cat concentration (100 nM, not shown). Using single turnover experiments, we find the maximal rate of AP product formation for hTDG cat is much (150-fold) faster, k max ϭ 0.9 Ϯ 0.1 min Ϫ1 for the same G⅐U20 substrate (not shown). Control experiments using the G⅐U substrate which lacks the FAM and dabcyl groups (G⅐U19) gave essentially the same result, k max ϭ 1.2 Ϯ 0.1 min Ϫ1 and k cat ϭ 0.005 Ϯ 0.001 min Ϫ1 , indicating that hTDG cat activity is not altered by FAM-dT or dabcyl-dT. These experiments were not collected for G⅐T20 because the G⅐T activity (k max ) was significantly lower for hTDG cat relative to hTDG.
hAPE1 Does Not Affect hTDG Reaction Steps before Product Release-Having established the baseline kinetic parameters k max and k cat for hTDG alone, we sought to establish which step(s) of the hTDG reaction is enhanced by hAPE1 and to quantify the magnitude of the effect. Previous studies suggested that hAPE1 stimulates hTDG product release (7,17), but it was not established whether hAPE1 may also influence earlier steps of the hTDG reaction (i.e. nucleotide flipping or chemistry). To directly address this question, we repeated the single turnover experiments for hTDG and the G⅐U20 substrate in the presence . Kinetic mechanism for hTDG. The minimal kinetic mechanism is shown, including the steps that contribute to k max , which reflects the maximal rate of product formation, and k cat , which reflects the maximal steady-state turnover rate. The association of enzyme and DNA gives the collision complex (E⅐D), and nucleotide flipping gives the reactive enzyme-substrate complex (E⅐ B D). The chemical step (k chem ) involves cleavage of the base-sugar (N-glycosidic) bond and the addition of water (nucleophile) giving the product complex (E⅐B⅐apD; where B is the excised base, apD is AP-DNA). Two potential pathways for product release are shown. Our finding that the excised base influences the dissociation rate of AP-DNA suggests AP-DNA is released before the base (upper pathway). FIGURE 5. Pre-steady-state single and multiple turnover kinetics for hTDG alone. A, representative data from pre-steady-state multiple turnover (burst) kinetics experiments for hTDG (500 nM) and G⅐U20 (1000 nM). B, representative data from single turnover kinetics experiments collected for hTDG (5 M) and G⅐U20 substrate (500 nM). C, burst kinetics data collected for hTDG (500 nM) and G⅐T20 (1500 nM). D, single turnover kinetics data collected for hTDG (5 M) and G⅐T20 (500 nM).
of hAPE1 (500 nM). We found that the maximal rate of AP product formation is essentially unchanged in the presence of hAPE1, k max ϭ 2.0 Ϯ 0.2 min Ϫ1 (data not shown). Similarly, hAPE1 (500 nM) did not increase k max for hTDG cat acting upon the G⅐U20 substrate (not shown). Thus, hAPE1 did not influence steps of the hTDG reaction before product release, indicating the stimulation of hTDG turnover by hAPE1 involves product release (and/or a conformational change in hTDG and/or AP-DNA required for product release). We note that the 500 nM hAPE1 concentration used for these single turnover experiments is high, ϳ10-fold higher than that needed to realize the maximal hAPE1 affect on hTDG turnover, as shown below.
We also repeated the burst kinetics experiments for hTDG (200 nM) and the G⅐U20 substrate (2000 nM), this time in the presence of hAPE1, with concentrations ranging from 0 to 50 nM (Fig. 6). We found that hAPE1 does not significantly alter the rate or amplitude of the exponential burst phase, consistent with the conclusion from single turnover experiments that hAPE1 did not affect steps of the hTDG reaction before product release. However, a significant increase was observed for the steady-state phase, indicating that hAPE1 increases the turnover of hTDG by increasing the rate of product release.
hAPE1 Dramatically Increases hTDG Turnover-To rigorously quantify the effect of hAPE1 on the steady-state turnover of hTDG (i.e. its effect on k cat ), we developed a coupled enzyme (hTDG-hAPE1) fluorescence assay (Fig. 2). The DNA substrates were designed with the molecular beacon approach (38,39), where the sequential activity of hTDG and hAPE1 generates a large fluorescence increase, and the coupled reaction is monitored in real-time. As expected for a properly executed coupled-enzyme assay (36), the rate constant obtained from the hTDG-hAPE1 assay reflects the steady-state turnover of the primary enzyme, hTDG, and is independent of the coupling enzyme, hAPE1, because we use a sufficiently high concentration of hAPE1 to ensure that hTDG turnover is rate-limiting. Importantly, the steady-state activity of hAPE1 alone, k cat Ͼ 60 min Ϫ1 (37), is at least 10,000-fold greater than that reported here for hTDG alone (Table 1). Nevertheless, because the rate-limiting step of the coupled-enzyme reaction is likely to be hAPE1-stimulated dissociation of the hTDG product complex, it was necessary to determine the amount of hAPE1 required to obtain the maximal stimulatory effect such that the observed rate constants are independent of hAPE1 concentration (Fig. 2D).
The utility of our coupled-enzyme assay for determining the steady-state kinetic parameters of the hTDG reaction, as stimulated by hAPE1, is illustrated by our results for hTDG cat acting upon the G⅐U20 substrate (Fig. 7). Using the coupled-enzyme assay, we determined the steady-state rate constant as a function of G⅐U20 substrate concentration (Fig. 7) and fitted these data to the Michaelis-Menten equation (Equation 3), yielding   k cat ϩAPE1 ϭ 0.46 Ϯ 0.02 min Ϫ1 and K m ϭ 162 Ϯ 17 nM (where k cat ϩAPE1 denotes k cat for hTDG in the presence of hAPE1). A comparison of the rate constants for the steady-state turnover of hTDG cat alone (k cat ) and in the presence of hAPE1 (k cat ϩAPE1 ) reveals that hAPE1 enhances the turnover of hTDG cat for G⅐U20 by 77-fold (Table 1). In fact, the hAPE1 effect on product release is so large that k cat ϩAPE1 approximates the maximal rate of AP-DNA product formation, k max ϭ 0.9 Ϯ 0.1 min Ϫ1 . Thus, in the presence of hAPE1, product release is no longer fully ratelimiting for hTDG cat acting upon the G⅐U substrate.
We also used the coupled-enzyme assay to examine the effect of hAPE1 on the steady-state turnover of full-length hTDG for the G⅐U20 substrate. The hAPE1-stimulated turnover of hTDG is lower than that of hTDG cat for G⅐U20, which was somewhat unexpected given that k max is 2-fold greater for hTDG versus hTDG cat (Table 1). This precluded accurate measurements at low concentrations of G⅐U20 substrate (i.e. below 50 nM). Nevertheless, the data provide an accurate value for the maximal hAPE1-stimulated turnover of hTDG, k cat ϩAPE1 ϭ 0.13 Ϯ 0.01 min-1 , and an estimated Michaelis constant of K m ϭ ϳ10 nM (not shown). A comparison of k cat and k cat ϩAPE1 reveals that hAPE1 enhances hTDG turnover by a remarkable 26-fold for the G⅐U substrate. Although this hAPE1 effect is large, the observation that k max exceeds k cat ϩAPE1 by 14-fold indicates product release is still rate-limiting for hTDG processing of G⅐U substrates, even in the presence of hAPE1.
We also examined the hAPE1-stimulated turnover of hTDG for the G⅐T substrate using the coupled-enzyme assay. The coupled reaction was slow, precluding rate measurements at low substrate concentrations. Nevertheless, rate constants were determined for substrate concentrations of 750, 1000, and 1500 nM and were found to be essentially the same (not shown), indicating saturating substrate conditions. Accordingly, the average of these data is taken as the maximal steady-state rate constant, k cat ϩAPE1 ϭ 0.014 Ϯ 0.001 min Ϫ1 (Table 1). Thus, hAPE1 increases hTDG turnover by 42-fold for the G⅐T substrate. The observation that k cat ϩAPE1 is merely 6-fold lower than k max indicates product release is much less rate-limiting in the presence of hAPE1 for hTDG processing of the G⅐T substrate. Notably, we find that a 100 nM concentration of hAPE1 provides the maximal stimulatory effect on hTDG turnover for the G⅐T substrate (not shown).
We note that the stimulatory effects of hAPE1 observed here are much larger than those reported previously for other DNA glycosylases. hAPE1 increases the turnover of hUNG by 2-4fold (6,46) and has a 4-fold effect on the turnover of human 8-oxoguanine DNA glycosylase (hOGG1) (9).
Kinetics for a G⅐FU Substrate-Given the much higher k max observed previously for G⅐FU relative to G⅐U substrates (14), it was of interest to examine the enzymatic turnover (k cat ) of hTDG for a G⅐FU substrate and the potential stimulatory effect of hAPE1. For hTDG alone and G⅐FU20, the maximal rate of product formation, k max ϭ 278 Ϯ 35 min Ϫ1 , was 440-fold greater than steady-state turnover, k cat ϭ 0.63 Ϯ 0.17 min Ϫ1 (Fig. 8A, Table 1), indicating k cat is limited by product release. Using our coupled-enzyme assay, we obtained k cat ϩAPE1 ϭ 0.54 Ϯ 0.03 min Ϫ1 and K m ϭ 39 Ϯ 8 nM (Fig. 8B). Thus, hAPE1 does not enhance the turnover of hTDG for the G⅐FU substrate.
Given that k cat is already quite fast for G⅐FU20 (ϳ100-fold faster than G⅐U20), this may reflect an upper limit for the effect of hAPE1 on the dissociation of AP-DNA from hTDG (i.e. k off,AP ϩAPE1 Ͻ 0.54 min Ϫ1 ).
For hTDG cat and the G⅐FU20 substrate, we find k max ϭ 111 Ϯ 14 min Ϫ1 (not shown) and k cat ϭ 0.58 Ϯ 0.12 min Ϫ1 (Fig. 8C). The observation that k max exceeds k cat by 191-fold again indicates rate-limiting product release. Using the coupled enzyme assay we find k cat ϩAPE1 ϭ 5.9 Ϯ 0.2 min Ϫ1 and K m ϭ 59 Ϯ 7 nM (Fig. 8D). Thus, hAPE1 increases the turnover of hTDG cat by 10-fold for the G⅐FU substrate, a smaller effect than observed for the G⅐U substrate (77-fold). This may be explained in part by the fact that turnover of hTDG cat alone is ϳ100-fold faster for G⅐FU relative to G⅐U ( Table 1). The faster turnover observed for G⅐FU relative to G⅐U substrates has implications for the mechanism of hTDG, as discussed below.

Implications for the Kinetic Mechanism of hTDG-
The presteady-state kinetics experiments collected here for hTDG (and hTDG cat ) provide important new insight into its catalytic mechanism. hTDG is widely regarded as a "single turnover" enzyme, because in many previous studies the reaction reaches a plateau as the [product]/[enzyme] ratio approaches unity for G⅐U and G⅐T substrates (13,27). Nevertheless, steady-state turnover (k cat ) can be measured using pre-steady-state multiple turnover (burst kinetics) experiments, collected with a high enzyme concentration and excess substrate (Fig. 5, A and C). We find hTDG turnover is very slow for the G⅐U substrate, k cat ϭ 0.005 Ϯ 0.001 min Ϫ1 , and strikingly slow for the G⅐T substrate, k cat ϭ 0.00034 Ϯ 0.00007 min Ϫ1 , corresponding to a half-life of 34 h for dissociation of the product complex. . Effect of hAPE1 on hTDG and hTDG cat activity for a G⅐FU substrate. A, burst kinetics data collected for hTDG (500 nM) and the G⅐FU20 substrate (1000 nM). B, saturation curve for steady-state activity of hTDG (5 or 10 nM) and the G⅐FU20 substrate (25-500 nM), collected in the presence of hAPE1 (25 nM) using the coupled-enzyme assay. C, burst kinetics data collected for hTDG cat (500 nM) and G⅐FU20 (1000 nM). D, saturation curve for steady-state activity of hTDG cat (1-10 nM) and G⅐FU20 substrate (10 -500 nM) obtained in the presence of hAPE1 (25 nM) using the coupled-enzyme assay.
Our results provide the first comparison of single turnover (k max ) and steady-state (k cat ) activity of hTDG reported for any substrate. In all cases (G⅐T, G⅐U, and G⅐FU substrates), we find k max Ͼ Ͼ k cat , with k max /k cat ranging from 150 to 440 (Table 1). The very large difference in these rate constants confirms the rate-limiting step(s) of the hTDG reaction occurs after the chemical step (35), i.e. product release, which includes dissociation of AP-DNA and the excised base and may require a conformational change for hTDG and/or AP-DNA (Fig. 4).
Previous studies and our unpublished observations indicate that release of AP-DNA, rather than release of the excised base, is the rate-limiting step of product release. AP-DNA binds tightly to hTDG (29), dissociates very slowly from the binary hTDG⅐AP-DNA complex (7), and is a potent inhibitor of hTDG (7,27,29). In contrast, we and others find hTDG is not inhibited by the nucleobases that it removes from DNA (including uracil, thymine, FU, and ⑀C), even at concentrations of up to 5 mM (17), indicating they do not bind hTDG with significant affinity.
Although these results indicate AP-DNA dissociation is ratelimiting, they do not indicate the order of product release, i.e. whether the base dissociates before or after AP-DNA (Fig. 4). For MutY, which excises adenine mispaired with 8-oxoguanine, adenine dissociates rapidly, k off Ͼ 5 min Ϫ1 , much faster than AP-DNA, k off,AP Ϸ 0.005 min Ϫ1 (40,41,47). For mismatchspecific uracil glycosylase, the E. coli ortholog of hTDG that also exhibits rate-limiting product release (48), an "escape route" for release of uracil before AP-DNA was suggested by a crystal structure (49), although this has not been confirmed. Our findings suggest a different mechanism for hTDG.
The burst kinetics experiments collected here indicate the excised base influences the dissociation rate of AP-DNA from the hTDG product complex. Our finding that k cat for G⅐FU20 is 126-and 1850-fold faster than k cat for G⅐U20 and G⅐T20 (Table  1) indicates a similarly large difference in the rate-limiting dissociation of AP-DNA (because k off,AP Ϸ k cat ). These striking differences were not expected, given that the AP-DNA product is identical, and they suggest the excised base remains trapped in the product complex by AP-DNA. Consistent with this idea, our recent crystal structure of hTDG cat bound to abasic DNA reveals no obvious pathway for departure of the excised base prior to AP-DNA absent a significant enzyme conformational change (16).
One explanation for the large differences in k off,AP for the G⅐T, G⅐U, and G⅐FU reactions is that the ionization state of the excised base may differ when bound in the product complex. Our previous studies indicate the hTDG reaction is highly dissociative and that the departing base is negatively charged, indicating the active site stabilizes the anionic base to some extent (14). The pK a (N1 nitrogen) is much lower for FU (pK a N1 ϭ 8.43) relative to U (pK a N1 ϭ 9.76) and T (pK a N1 ϭ 10.19); thus, FU is more likely than U (and U more likely than T) to be anionic in the product complex. The anionic base may promote the dissociation of AP-DNA by repulsive interactions with the DNA phosphates. This idea could be examined by determining the ionization state of FU, U, and T in the ternary product complex using NMR spectroscopy (50). Such experiments unexpectedly revealed that uracil is anionic when bound in the product complex of uracil DNA glycosylase at neutral pH (50).
Mechanism for hAPE1 Stimulation of hTDG-Previous studies of BER for organisms ranging from E. coli to humans have shown that AP endonucleases enhance the turnover of DNA glycosylases (6 -10, 51, 52). A consistent observation is that the AP endonuclease affects product release rather than the chemical step of the glycosylase reaction. Although the detailed molecular mechanism has remained unknown, two basic ideas have emerged. For the passive enhancement mechanism, the AP endonuclease stimulates glycosylase turnover by simply depleting the concentration of AP-DNA, converting it to 5Ј-nicked AP-DNA, which should be less inhibitory to the glycosylase (8). The active displacement mechanism requires that the AP endonuclease interacts with the glycosylase and/or AP-DNA to disrupt the product complex (6) and may also involve endonucleolytic depletion of AP-DNA. A passive mechanism was proposed for hAPE1 stimulation of hOGG1 (human 8-oxoguanine DNA glycosylase) (9,51), although a recent study suggests active displacement may contribute (42). An active mechanism was proposed for hAPE1 stimulation of hUNG (6), and hAPE1 stimulation of human MutY homolog appears to involve protein-protein interactions (53,54). Active displacement is indicated for AP endonuclease (Exo III and Endo IV) stimulation of MutY in E. coli (10). For the hAPE1 stimulation of hTDG, both passive (13,31) and active (6,7,17,30,32) mechanisms have been proposed.
Our findings demonstrate that hAPE1 stimulates hTDG turnover by active displacement of AP-DNA, as illustrated by our results for the G⅐U substrate. The burst kinetics experiments provide a good estimate for the spontaneous AP-DNA dissociation rate for the G⅐U reaction, where k off,AP Ϸ k cat ϭ 0.005 min Ϫ1 (true because k max Ͼ Ͼ k cat ). Meanwhile, the coupled-enzyme assay provides a lower limit for the hAPE1-stimulated dissociation rate of AP-DNA from the hTDG product complex (k off,AP ϩAPE1 Ն k cat ϩAPE1 ϭ 0.13 min Ϫ1 ). This is true because, by definition, no individual step of the coupled-enzyme reaction can be slower than the observed steady-state rate constant (k cat ϩAPE1 ). Our results exclude the passive displacement mechanism, because the spontaneous AP-DNA dissociation rate (k off,AP Ϸ 0.005 min Ϫ1 ) is not kinetically competent with the hAPE1-stimulated turnover rate (k cat ϩAPE1 ϭ 0.13 min Ϫ1 ). The same argument applies to hTDG acting upon the G⅐T substrate (k off,AP Ϸ 0.00034 min Ϫ1 and k cat ϩAPE1 ϭ 0.014 min Ϫ1 ) and hTDG cat processing the G⅐U substrate (k off,AP Ϸ 0.006 min Ϫ1 and k cat ϩAPE1 ϭ 0.46 min Ϫ1 ). Active displacement is required, because hAPE1 increases the dissociation rate of AP-DNA, presumably by contacting hTDG and/or AP-DNA to disrupt the product complex.
Other evidence supports an active displacement mechanism. The observation that k cat (hence k off,AP ) is identical for hTDG cat and hTDG (Table 1), whereas the hAPE1-effect is greater for hTDG cat (Table 1) indicates active displacement, because one would expect the same hAPE1 effect for a passive displacement mechanism (i.e. if hAPE1 does not alter k off,AP ). Active displacement is also consistent with the previous observation that Endo IV, an E. coli AP endonuclease that is structurally unrelated to hAPE1, does not stimulate hTDG for G⅐T or G⅐⑀C substrates (7,30). If a passive mechanism prevailed, i.e. depletion of sponta-APE1 Actively Increases Enzymatic Turnover of TDG NOVEMBER 21, 2008 • VOLUME 283 • NUMBER 47 neously released AP-DNA, one would expect Endo IV (or any AP endonuclease) to stimulate hTDG turnover.
hAPE1 Effect Depends on AP-DNA Release Rate-Previous studies and our findings suggest that when the spontaneous dissociation rate of AP-DNA (k off,AP ) is increased (i.e. due to enzyme or substrate modifications), a lower concentration of hAPE1 provides the maximal stimulatory effect. For example, AP-DNA dissociates more rapidly from SUMO-modified hTDG than from hTDG, and a 5 nM concentration of hAPE1 increased the turnover of SUMO-hTDG but had no effect on turnover of unmodified hTDG (for a G⅐U substrate) (32). Similarly, k off,AP is faster for C⅐AP-DNA versus G⅐AP-DNA, and the stimulatory effect of 10 nM hAPE1 on hTDG turnover is much greater for C⅐U versus G⅐U substrates (7). We find that k cat is much faster for G⅐FU relative to G⅐U substrates (indicating faster k off , AP for the G⅐FU reaction), and the concentration of hAPE1 required for maximal enhancement is lower for G⅐FU (25 nM) versus G⅐U (50 nM). Similarly, a 100 nM concentration of hAPE1 provides maximal enhancement of hTDG turnover for the G⅐T substrate, and k cat is lower for G⅐T relative to G⅐U. These observations also indicate that hAPE1 interacts with the hTDG product complex to disrupt it (i.e. active displacement) and that the interaction is more likely to be productive if the inherent AP-DNA release rate (k off,AP ) is faster. They are not consistent with passive enhancement because more rather than less hAPE1 should be required when k off,AP is faster. Further studies are needed to uncover the mechanistic underpinnings of these intriguing observations.
Nature of the Stimulatory Interaction-Our finding that hAPE1 enhances hTDG cat turnover by 77-fold (for G⅐U20) reveals that any stimulatory interactions with hAPE1 do not require the N-or C-terminal domains of hTDG (residues 1-110 or 309 -410). Indeed, the smaller hAPE1-effect for hTDG versus hTDG cat (Table 1) indicates the N-and/or C-terminal regions tend to diminish the stimulatory effect of hAPE1.
Previous studies using yeast two-hybrid and electrophoretic mobility shift experiments found no evidence for a stable bimolecular interaction for hTDG and hAPE1 (7,30). Attempts to visualize a stable complex of hTDG, AP-DNA, and hAPE1 using electrophoretic mobility shift and surface plasmon resonance were also unsuccessful (7,30). This is perhaps not surprising because hAPE1 displaces hTDG and rapidly converts AP-DNA to 5Ј-nicked AP-DNA. The stimulatory interactions of hAPE1 with hTDG and/or AP-DNA are probably transient and weak and may involve selective recognition of the hTDG product complex rather than free hTDG (7). Indeed, a stable complex seems incompatible with the robust stimulation of hTDG (and hTDG cat ) observed here.
Nevertheless, a stable interaction was reported for murine TDG and APE1 (using glutathione S-transferase pulldown assays) involving residues 92-121 of mTDG (55). This region of mTDG is acetylated (Lys residues) by the transcriptional coactivator CBP/p300, and acetylation of mTDG disrupted its bimolecular interaction with mAPE1 (55). The corresponding (nearly identical) acetylation domain of hTDG, residues 81-110, is not present in hTDG cat . Yet, we find that hTDG cat is strongly stimulated by hAPE1, 77-fold for the G⅐U substrate. Clearly, the acetylation domain of hTDG is not required for stimulatory interactions with hAPE1, a finding that conflicts with the previous suggestion that acetylation of TDG regulates the recruitment of APE1, hence the second step of BER (55). Our results indicate that any hTDG-hAPE1 interaction mediated by the acetylation domain, if adopted in vivo, may be important for some function other than promoting hTDG turnover. Additional experiments are warranted to determine the effect of acetylation on the turnover of hTDG in the presence and absence of hAPE1.
How Might hAPE1 Actively Stimulate hTDG?-Our finding that hAPE1 actively disrupts the hTDG product complex raises the question of exactly how this occurs. It was suggested that hAPE1, which forms extensive minor groove interactions, stimulates the turnover of hUNG by binding the minor groove and processing toward hUNG to "pry" it from the AP site (6,56). Such a mechanism seems consistent with our findings, given some limitations. The 19-bp DNA substrates used here (Fig. 1) are not much longer than the overall footprint (ϳ12 bp) for one hTDG cat subunit, which binds the minor groove and occupies less than half of the total circumference of the helix (16). Thus, hAPE1, with a footprint of about 8 bp (56,57), could bind the minor groove immediately adjacent to hTDG and on 3Ј side of the AP site (Fig. 9). This would position hAPE1 to initially contact the "insertion loop" of hTDG, which plays key roles in FIGURE 9. Structure of hTDG cat bound to AP-DNA. Shown is the recent crystal structure of hTDG cat (semitransparent surface mode) bound to AP-DNA (16) whose length corresponds to the substrates used in this study (Fig.  1). A protein dimer is seen in the crystal structure, one subunit bound at the AP site (product complex) and one at an undamaged site (nonspecific complex). For the product complex, the insertion loop residues of hTDG (270 -280) are colored cyan, and Lys-246 and Lys-248 are blue. The AP strand of the DNA is orange (3Ј and 5Ј ends labeled), and the complementary strand is yellow. One possibility for active disruption of the hTDG product complex is that hAPE1 binds the minor groove adjacent to hTDG and 3Ј to the AP site, as indicated, such that it might contact the insertion loop and/or Lys-246/Lys-248 to disrupt the product complex.
substrate recognition and nucleotide flipping and accounts for most of the DNA contacts formed by hTDG (16), and may also disrupt the DNA phosphate contacts involving Lys-246 and Lys-248 (Fig. 9). Disruption of the insertion loop and/or the Lys-246/Lys-248 interactions could conceivably promote dissociation of the product complex. hAPE1 binding to the 5Ј side of the AP site may be precluded by the other (nonspecific) hTDG subunit if the product complex involves 2:1 (protein: DNA) binding (Fig. 9) as indicated by our recent structural and biochemical studies (16). Moreover, hAPE1 binding 5Ј to the AP site is not consistent with observation that the hAPE1 effect is the same for G⅐U20 and a shorter G⅐U17 substrate (not shown). G⅐U17 is identical to G⅐U20 (Fig. 1) but has just three base pairs located 5Ј to the AP site, offering very little foothold for binding of hAPE1 5Ј to the AP site.
Of course, other active displacement mechanisms are possible. A direct handoff of AP-DNA from hTDG to the active site of hAPE1 seems unlikely because both enzymes flip the abasic deoxyribose deep into their active site and contact the same five phosphates flanking the AP site (16,56). hAPE1 could potentially contact hTDG alone to promote dissociation of the product complex, and then compete with hTDG to bind freely released AP-DNA. Additional biochemical and structural studies are needed to fully elucidate the mechanism by which hAPE1 disrupts the hTDG product complex.
Implications for SUMO Modification of hTDG-hTDG binds to and is covalently modified by SUMO-1 and SUMO-2/3, which decreases its DNA binding affinity, apparently by stabilizing an ␣-helix (residues 317-329) that clashes with DNA (32,45). For G⅐U substrates, k cat is higher for SUMO-hTDG versus hTDG, due likely to faster AP-DNA product release, and the stimulatory effect of hAPE1 is greater for SUMO-hTDG (32). Consistent with weaker substrate binding, a 5 nM concentration of SUMO-hTDG exhibited no G⅐T activity, but it was not reported whether G⅐T activity could be recovered at higher enzyme concentrations. It was suggested that product-bound hTDG is SUMOylated to increase the dissociation rate of AP-DNA and that the SUMO group is subsequently removed (32). Our studies provide a measure of the hAPE1-stimulated turnover of hTDG for a G⅐T substrate, k cat ϩAPE1 ϭ 0.014 min Ϫ1 , which corresponds to a half-life of ϳ50 min for dissociation of the product complex. Assuming a similar rate in vivo, our finding suggests ample time and a potential need for SUMOylation of hTDG to further enhance the effect of hAPE1 on product release for processing G⅐T mispairs. SUMOylation of hTDG seems less important for G⅐U processing, because the hAPE1stimulated turnover is relatively fast (k cat ϩAPE1 ϭ 0.13 min Ϫ1 ). Additional studies are needed to quantitatively establish the effect of TDG SUMOylation on product release in the presence and absence of hAPE1.
Studying hTDG Using Steady-state Kinetics-In addition to illuminating the mechanism for hAPE1 stimulation of hTDG, our coupled-enzyme fluorescence assay provides a method for monitoring steady-state kinetics of hTDG in real time. This assay could potentially be useful for screening and evaluating hTDG inhibitors, determining the damaging effect of removing specific side chains of hTDG by site-directed mutagenesis and structure-activity correlations using modified substrates. Such studies may be most productive by using hTDG cat and the G⅐FU20 substrate, which provide the highest turnover (k cat ϩAPE1 ϭ 6 min Ϫ1 ), hence the largest dynamic range for determining the effect of enzyme or substrate modifications.