Kinetic Analysis of the Effect of N-terminal Acetylation in Thymine DNA Glycosylase

Abstract

Thymine DNA glycosylase (TDG) is tasked with initiating DNA base excision repair by recognizing and removing T, U, the chemotherapeutic 5-fluorouracil (5-FU) and many other oxidized and halogenated pyrimidine bases. TDG contains a long, unstructured N-terminus that contains four known sites of acetylation: lysine (K) residues 59, 83, 84, and 87. Here, K to glutamine (Q) mutants are used as acetyl-lysine (AcK) analogues to probe the effect of N-terminal acetylation on the kinetics of TDG. We find that mimicking acetylation affects neither the maximal single-turnover rate k_max, nor the turnover rate k_TO, indicating that the steps after initial binding, through chemistry and product release, are not affected. Under sub-saturating conditions, however, acetylation changes the processing of U substrates. Subtle differences among AcK analogues are revealed with 5-FU in single-stranded DNA. We propose that the subtleties observed among the AcK analogues may be amplified on the genomic scale, leading to regulation of TDG activity. N-terminal acetylation, though, may also play a structural, rather than kinetic role in vivo.

Graphical Abstract

INTRODUCTION

Genomic DNA is under constant onslaught from a vast array of damaging agents from both exogenous and endogenous sources, resulting in lesions that may prove mutagenic, cytotoxic, or transformative for the cell. Hydrolytic deamination of nucleobases, namely deamination of cytosine (C) to uracil (U) or the epigenetic marker 5-methylcytosine (5-mC) to thymine (T), results in a U:guanine (G) or T:G mismatch that, if left unrepaired, results in a C to T transition mutation.^1-3 The base excision repair (BER) pathway is responsible for repairing these and other nucleobase lesions.^4,5 BER is initiated by a glycosylase specific to the nucleobase lesion, which excises the damaged base leaving an abasic (apurinic/apyrimidinic; AP) site. AP endonuclease 1 (APE1) nicks the DNA backbone, allowing for insertion of a C by DNA polymerase β (polβ), and restoration of the strand by DNA ligase.^4,5 There are four human glycosylases that recognize and remove uracil from the genome: uracil DNA glycosylase (UDG/UNG), single strand-selective monofunctional uracil DNA glycosylase (SMUG1), thymine DNA glycosylase (TDG), and methyl binding domain 4 (MBD4). ^5,6 Within this group, both TDG and MBD4 also remove T from T:G mispairs, specifically within the 5’-CpG-3’ dinucleotide context.^5-16

TDG is a 46 kDa enzyme with three structural regions: the unstructured N-terminus (residues 1-108), the catalytic domain (residues 109-308), and the unstructured C-terminus (residues 309-410).^12,13,17-19 NMR studies have demonstrated that the N-terminus has a different conformation when TDG is bound to a T substrate versus a U substrate: when bound to T, this region folds back upon the catalytic domain, while when bound to U, this region is extended and in contact with the DNA.²⁰ In addition to U and T, TDG excises several nucleobases from the genome, including ethenocytosine, oxidized and halogenated uracil derivatives such as 5-fluorouracil (5-FU), and the oxidized forms of 5-mC 5-formylcytosine (5-fC) and 5-carboxycytosine (5-caC), with specificity for 5’-CpG-3’ dinucleotide contexts.^{7-11,15,16,21-24} TDG, furthermore, displays differential excisional rates depending on the substrate: those substrates that more readily flip into the active site and have decreased N-glycosidic bond stability are excised faster.^10,25 Importantly, TDG exhibits high levels of end-product inhibition regardless of substrate, that is, high affinity for the AP site.^26,27 Its removal of 5-fC and 5-caC ties TDG to active demethylation, which is crucial during epigenetic programming and reprogramming, especially during gametogenesis and embryogenesis.^{14,15,21,28,29} Because of this essential epigenetic role, TDG is the only glycosylase known to exhibit embryonic lethality.²⁸ TDG, additionally, may have roles that are unrelated to its glycosylase activity, as it has been shown to interact with a wide array of other proteins, including transcription factors, chromatin remodelers, proliferating cell nuclear antigen (PCNA), kinases, the acetyltransferase CBP/p300, and the deacetylase SIRT1.^30-40

TDG not only interacts with kinases and acetyltransferases, but also undergoes post-translational modification (PTM), on both the N- and C- termini. Lysine (K) 330 in the C-terminus serves as a site of SUMOylation.^41-44 This PTM significantly hinders the ability of TDG to perform chemistry,⁴⁴ while also decreasing the affinity of TDG for the product AP site while maintaining affinity for the substrate.^41,43 It has been proposed that SUMOylation may enable TDG to serve a chromatin “reader” function, rather than as a glycosylase,⁴⁴ or conversely, to allow for turnover of the enzyme.^41,43,45 The N-terminus, on the other hand, contains several PTM sites: K59, K83, K84, and K87 are sites of acetylation by CPB/p300, while serine (S) residues 85 and 88 are phosphorylated by protein kinase C (Figure 1a).^31,36,46 Phosphorylation does not change the Michaelis-Menten constants of TDG, though it is mutually exclusive with acetylation, leading to the hypothesis that the role of phosphorylation is to prevent acetylation of the nearby K.⁴⁶ The role of acetylation, however, is less clear. Acetylation has been suggested to decrease the activity of TDG on T and U substrates, while simultaneously to increase the activity on 5-FU substrates.³¹ These studies, however, were performed on enzymatically acetylated TDG, and therefore it remains to be determined how acetylation of each K contributes to these observations. Furthermore, CBP/p300 is known to acetylate K in clusters,^{33, 47} thereby indicating the possibility that K83, K84, and K87 may be acetylated together.

Figure 1.

Graphical representation of TDG and structural comparison of amino acids used as analogs in this study. (A) The unstructured N-terminus of TDG contains several post-translational modification (PTM) sites: acetylation (Ac) at lysine (K) residues 59, 83, 84, and 87; and phosphorylation (℗) at serine (S) residues 85 and 88. (B) Structural comparison of K, acetyl-lysine (AcK), and glutamine (Q), the AcK analog used in this study.

TDG is deacetylated by the NAD⁺-dependent deacetylase SIRT1, which has been implicated in the cellular and organismal response to caloric restriction, healthspan and lifespan extension, and the response of certain neoplasms to chemotherapeutics.^31,48-53 SIRT1 targets have been identified throughout the BER pathway, including other glycosylases, APE1, and polβ.^54-56 The redox and metabolic states of the cell can determine NAD⁺ levels and availability. Metabolism, therefore, directly influences the enzymes, pathways, and processes regulated by SIRT1, including BER.^53,57 Furthermore, knockdown of SIRT1 in several cell lines sensitized the cells to the chemotherapeutic 5-FU. ^49,58 The authors suggest that the increased sensitivity is due to the acetylation and therefore increased excisional activity of TDG on 5-FU.⁵⁸ In this case, increased excision of 5-FU without subsequent completion of BER leads to an overwhelming DNA damage response (DDR) resulting in cell death.

5-FU is a chemotherapeutic agent often used with other agents in the treatment of many cancers, including colorectal carcinoma and acute myeloid leukemia.^59-63 It was initially designed as an inhibitor of thymidylate synthase (TS) that would disrupt the balance of nucleotides in the dNTP pool and result in a “thymine-less death”, however further mechanisms of action, including those involving BER, contribute to its efficacy.^59,62 Serving as a TS inhibitor, the relative dUTP:dTTP ratio in the dNTP pool increases, resulting in increased incorporation of U into the genome during replication. 5-FU also finds its way into the dNTP pool as 5-FdUTP, and can also be incorporated into the genome during replication.^59,62 A recent study demonstrated that much of U and 5-FU incorporated during replication is removed at the replication fork by UNG.⁶⁴ If, however, the lesions are not removed during replication, such as with UNG inhibition, they persist.⁶⁴ The persistent U and 5-FU are subject to repair, including removal by TDG, resulting in AP sites that may be close to one another and on opposite strands. These AP sites may be processed by APE1 or may undergo hydrolysis prior to APE1 processing. The number of strand breaks in this case exceeds the capabilities of polβ and ligase, resulting in single- and double-strand breaks that initiate the DDR. An overwhelming DDR triggers cell death.^59,62

Given the possible implication of TDG, and specifically acetylated TDG, in the cellular response to 5-FU, we set out to understand how acetylation modulates the kinetics of this enzyme through mutants that mimic acetyl-lysine (AcK; AcK analogues). We have found that neither the maximal single turnover rate (k_max) nor the rate of turnover (k_TO) of TDG is changed in these analogues, though there is likely a decrease in the affinity for U substrates. Additionally, we have determined that 5-FU in single-stranded DNA is a substrate for TDG, and that this substrate reveals subtle distinctions among the different sites of acetylation in the AcK analogues. The potential biological implications of these findings are discussed.

MATERIALS AND METHODS

Oligonucleotide synthesis and purification

Oligonucleotide sequences are based on those of M. T. Bennett et al.¹⁰ and M. T. Morgan et al.¹¹ The sequence of the lesion-containing strand is as follows: 5’-CAC TGC TCA XGT ACA GAG C, in which “X” denotes T, U, or 5-FU. The sequence of its complement is 5’- GCT CTG TAC GTG AGC AGT G, such that the lesion “X” is paired opposite G in the single CpG dinucleotide of the resulting duplex. Oligonucleotides were synthesized on a MerMade 4 (BioAutomation) using standard phosphoramidite chemistry. All reagents were purchased from Glen Research. The final dimethoxytrityl group was retained for HPLC purification at 90 °C (Agilent PLRP-S column, 250 mm × 4.6 mm; A= 100 mM triethylammonium acetate [TEAA] in 5% [v/v] aqueous acetonitrile [MeCN], B= 100 mM TEAA in MeCN; 5:95 to 30:70 A:B over 30 min at 1 mL/min). Oligonucleotides were subject to detritylation by incubation for 60 min at room temperature in 20% (v/v) aqueous glacial acetic acid. The reaction was quenched by precipitation of the oligonucleotides in room-temperature ethanol. A second HPLC-purification was then performed at 90 °C (Agilent PLRP-S column, 250 mm × 4.6 mm; A= 100 mM triethylammonium acetate [TEAA] in 5% [v/v] aqueous acetonitrile [MeCN], B= 100 mM TEAA in MeCN; 0:100 to 25:75 A:B over 40 min at 1 mL/min). The purified oligonucleotides were flash-frozen with liquid nitrogen, lyophilized, and resuspended in storage buffer (10 mM Tris-HCl pH 8.0, 30 mM NaCl, 1 mM EDTA). The concentration of purified oligonucleotide was determined by absorbance at 260 nm, using extinction coefficients calculated by the OligoAnalyzer 3.1 (Integrated DNA Technologies; https://www.idtdna.com/calc/analyzer).

Site-directed mutagenesis of TDG

The pET-28a plasmid containing wild-type human TDG (pET-28a_TDG-WT) was a generous gift from the laboratory of Alexander C. Drohat (University of Maryland School of Medicine). Site-directed mutagenesis (SDM) was accomplished using the Q5 Site-Directed Mutagenesis Kit (New England BioLabs) with primers (Integrated DNA Technologies) designed by the NEBaseChanger tool (New England BioLabs; http://nebasechanger.neb.com/) (Table S1) following the manufacturer’s instructions. Briefly, Q5 2X Master Mix, forward primer (final concentration 0.5 μM), reverse primer (final concentration 0.5 μM), pET-28a_TDG-WT (29.4 ng), and ddH₂O were mixed in a PCR tube and placed on the thermocycler. The PCR reaction was performed: initial denaturation at 98 °C for 30 s; 25 cycles of denaturation for 10 s at 98 °C, annealing at 58 °C (K83Q; K84Q; 3Q), 59 °C (K87Q), or 62 °C for 30 s, and extension at 72 °C for 195 s; final extension at 72 °C for 2 min. Following PCR, the kinase, ligase, and DpnI (to remove WT plasmid) treatments were performed by mixing the PCR product, 2X KLD Reaction Buffer, 10X KLD enzyme mix, and ddH₂O and incubating at room temperature for 5 min. The mutagenized plasmids were transformed into chemically-competent DH5α E. coli (New England BioLabs) by first incubating the preceding reaction with thawed E. coli on ice for 30 min. The bacteria were then heat-shocked at 42 °C for 30 s and immediately placed on ice for 5 min. To this mixture room-temperature SOC media (New England BioLabs) was added, and the tube was shaken at 37 °C and 250 rpm for 60 min. The cells were carefully mixed, and were spread on Luria Broth (LB) agar bacterial selection plates containing 50 μg/mL kanamycin sulfate. Plates were incubated overnight at 37 °C.

The following day, eight colonies per mutagenesis were selected and each used to inoculate 5 mL autoclaved LB media containing 50 μg/mL kanamycin sulfate in vented bacterial culture tubes. Tubes were shaken at 225 rpm overnight at 37 °C. Plasmids were extracted from the bacteria using the QIAprep Spin Miniprep Kit (Qiagen) following the manufacturer’s protocol. Success of the mutagenesis was confirmed by Sanger sequencing of the mutagenized plasmids by GeneWiz using the universal T7 primer.

Glycosylase expression and purification

Recombinant TDG (WT and mutants) was expressed and purified as previously reported with slight modifications.¹³ BL21(DE3) E. coli cells (New England BioLabs) were transformed with TDG-containing plasmid (WT or mutants) according to the manufacturer’s instructions and spread onto LB agar bacterial selection plates containing 50 μg/mL kanamycin sulfate. A single colony was chosen to inoculate 5 mL LB media with 50 μg/mL kanamycin sulfate in a vented bacterial culture tube. This tube was shaken at 225 rpm overnight at 37 °C. The following morning, the full 5 mL were used to inoculate 100 mL LB with 50 μg/mL kanamycin sulfate which was then shaken at 37 °C and 180-200 rpm until OD₆₀₀=0.6-0.8. At this time, the 100 mL of growing E. coli were used to inoculate two 1 L LB with 50 μg/mL kanamycin each (50 mL of E. coli into each flask). The mixtures were shaken at 180 rpm, 37 °C until OD₆₀₀=0.6-0.8. At this time, the temperature was lowered to 15 °C and the cultures shook for 30 min at 180 rpm. Expression was then induced with isopropyl β-D-1-thiogalactopyranoside (IPTG) to a final concentration of 0.4 mM and allowed to continue for 16 h. Bacteria were pelleted by centrifugation at 5,000 × g for 15 min at 4 °C, flash-frozen in liquid nitrogen, and stored at −80 °C until lysis.

The bacterial pellets were thawed on ice, followed by resuspension in 25 mL lysis buffer (50 mM NaP_i pH 8.0, 10 mM imidazole, 300 mM NaCl, 10 mM β-mercaptoethanol [βME]) containing complete protease inhibitor tablet (1 tablet per 25 mL; Millipore Sigma). The cell suspension was stirred on ice at 4 °C for 20 min to homogenize the suspension, flash-frozen in liquid-nitrogen, and stored at −80 °C until purification.

The cell suspension was thawed in a room-temperature bath for approximately 30 min and was transferred to ice to finish thawing. To the thawed cell suspension, powdered lysozyme (Fisher Scientific) was added to a final concentration of 1 mg/mL and gently stirred on ice in the cold room for 20 min. At this time, benzonase nuclease (Millipore Sigma) was added and the mixture stirred for 30 min (until the viscosity was similar to water). The lysate was cleared by centrifugation at 23,000 ×g for 30 min at 4 °C and syringe-filtered through a 0.22 μM poly-ethylene sulfonate (PES) membrane in preparation for purification.

All following purification steps were performed by fast protein liquid chromatography on an ÄKTA FPLC (Amersham Biosciences). The filtered lysate was first loaded by syringe onto a HiTrap His 5 mL nickel affinity column (Cytiva) equilibrated with lysis buffer (50 mM NaP_i pH 8.0, 10 mM imidazole, 300 mM NaCl, 10 mM β-mercaptoethanol [βME]). The column was washed with 50 mL High Salt Ni²⁺ Buffer (50 mM NaPi pH 8, 10 mM imidazole, 1M NaCl, 10 mM β-ME) followed by 50 mL Low Salt Ni²⁺ Buffer (50 mM NaP_i pH 8, 20 mM imidazole, 300 mM NaCl, 10 mM β-ME). TDG was eluted from the nickel column with the following gradient: A=Low Salt Ni²⁺ Buffer, B=Ni²⁺ Elution Buffer (50 mM NaP_i pH 8, 500 mM imidazole, 300 mM NaCl, 10 mM β-ME); 100:0 to 0:100 A:B over 100 mL at 5 mL/min. A final 25 mL wash of 100% Ni²⁺ Elution Buffer at 5 mL/min ensured elution of all TDG. Fractions containing TDG, as determined by absorbance at 280 nm and SDS-PAGE, were combined for further ion-exchange purification.

In preparation for anion exchange, TDG-containing fractions were dialyzed against Exchange Buffer A (EB-A; 20 mM HEPES pH 7.0, 75 mM NaCl, 1 mM dithiothreitol [DTT], 0.2 mM ethylene diamine tetraacetic acid [EDTA]) using 6K MWCO cellulose dialysis tubing (Fisher Scientific) for 90 min, with a buffer change at 60 min. Dialyzed TDG was syringe-filtered through 0.22 μm PES membrane and loaded by syringe onto a 5 mL HiTrap Q Anion Exchange Column (Cytiva) equilibrated with EB-A. The column was then washed with 50 mL EB-A. TDG was eluted from the anion exchange column with the following gradient: 100:0 to 0:100 EB-A: Elution Buffer B (EB-B; EB-A + 1 M NaCl) over 100 mL at 5 mL/min. Fractions containing TDG, as determined by absorbance at 280 nm and SDS-PAGE, were combined and the salt concentration in the buffer dropped to <100 mM NaCl by the addition of Salt Drop Buffer (EB-A+10 mM NaCl). The TDG-containing solution was then syringe-filtered through a 0.22 μm PES membrane and loaded by syringe onto a HiTrap SP Cation Exchange Column (Cytiva). The column was washed with 50 mL EB-A, and TDG was eluted from the column with the following gradient: 100:0 to 0:100 EB-A: Elution Buffer B (EB-B; EB-A + 1 M NaCl) over 90 mL at 3 mL/min. Fractions containing TDG as determined by absorbance at 280 nm were analyzed by SDS-PAGE. Only those fractions with >95% purity determined by staining were combined. Combined TDG fractions were dialyzed against 2X Storage Buffer (40 mM HEPES pH 7.5, 200 mM NaCl, 2 mM DTT, 1 mM EDTA) overnight at 4 °C using 5K MWCO dialysis tubing. The following morning, fresh 2X Storage Buffer was added and TDG dialyzed for 3 h.

Dialyzed TDG was concentrated using a Vivaspin™ 5K MWCO centrifugal concentrator (Sartorius) at 4,000 rpm at 4 °C until the final volume was <1 mL. A small volume of the TDG was reserved for analysis by ESI mass spectrometry, and to the remainder an equal volume of glycerol was added to generate 1X Storage Buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 1 mM DTT, 0.5 mM EDTA, 50% [v/v] glycerol). TDG in 1X Storage Buffer was aliquoted into 10 μL single-use fractions, flash-frozen with liquid nitrogen, and stored at −80 °C until use. Concentration was determined by the Bradford assay using bovine γ-globulin standards.

Glycosylase kinetics experiments

Samples of DNA substrate and glycosylase were prepared. The lesion-containing strand carrying a T, U, or 5-FU was 5’−³²P end-labeled using T4 polynucleotide kinase (New England BioLabs) following the manufacturer’s protocol. For duplex substrates, radiolabeled lesion strand was annealed to its complement (1:1.5 molar ratio) in ANL Buffer (10 mM Tris-HCl, pH 8.0, 30 mM NaCl) by heating to 95 °C for 4 min and cooling at a rate of 1 °C/min to 25 °C. For ss5-FU substrates, radiolabeled lesion strand was subjected to the same annealing conditions as duplex substrates in ANL Buffer. For single-turnover kinetic experiments, 50 nM-2 μM substrate solutions and 500 nM-4 μM TDG solutions were prepared separately in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). For multiple-turnover kinetic experiments, 0.5-4 μM substrate solutions and 20-600 nM TDG solutions were prepared separately in Buffer TNK. All solutions were pre-incubated at 37 °C for 2 min prior to initiating time courses. For experiments with T and U duplex substrates and ss5-FU, enzyme and substrate solutions were mixed and quenched by hand. At time t=0, equal volumes of DNA substrate and TDG were mixed for final concentrations of 25 nM-1 μM DNA and 250 nM-2 μM TDG (single-turnover) or 0.25-2 μM DNA and 10-300 nM TDG (multiple-turnover) and incubated at 37 °C. At the designated time points, a small volume of the reaction solution was removed and quenched with equal volume 1 M NaOH (final concentration 500 mM NaOH). This method was used for two replicates each of T and U duplex substrates under STO and MTO conditions with TDG^WT, TDG^K83Q, TDG^K84Q, and TDG^K87Q. Alternatively, equal volumes of DNA substrate and TDG were mixed for individual time points, incubated at 37 °C, and quenched with 1 M NaOH (final concentration 500 mM NaOH). This method was used for all other reactions under STO and MTO conditions with T and U duplex substrates, and ss5-FU. The time points for T duplex substrates for both STO and MTO conditions were the following (min): 0.167, 0.5, 1, 2, 3, 5, 10, 15, 30, 45, 60. The time points for U duplex substrates for both STO and MTO were the following (min): 0.167, 0.333, 0.5, 0.75, 1, 1.5, 2, 3, 5, 7.5, 10, 15, 20, 30 (and 45, 60 for 25 nM U under STO conditions only). The time points for ss5-FU substrates under STO and MTO conditions were the following (min): 0.5, 1, 2, 3, 5, 7.5, 10, 15, 20, 30, 45, 60.

For each time course, three negative controls were included. The no treatment (NT) sample was prepared by incubating DNA at room temperature for the duration of the reaction; this sample was not quenched with NaOH. The no enzyme (−E) sample was prepared by mixing equal volume of DNA substrate and Buffer TNK, incubating at 37 °C for the longest time point, and adding 1 M NaOH quench (final concentration 500 mM NaOH). The quench control (QC) was prepared by mixing DNA substrate and 1 M NaOH (double the volume of DNA), followed by immediate addition of TDG (equal volume to DNA) and incubation at 37 °C for the longest time point. For experiments with 5-FU:G, DNA substrate and TDG were mixed at 37 °C and quenched with 1 M NaOH in a Rapid Quench Flow apparatus (RQF-3; KinTek Corp.). The −E sample was prepared by mixing DNA substrate and Buffer TNK in the RQF for the longest time point followed by quench with 1 M NaOH. The time points under STO conditions were the following (s): 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 5. The time points under MTO conditions were the following (s): 0.025, 0.05, 0.1, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 5, 7.5, 10, 12, 15.

Following completion of the time course, all samples (except NT) were heated to 90 °C for 2 min to induce strand breaks at abasic sites. DNA substrate and product were resolved by 12% denaturing PAGE, imaged by phosphorimagery, and quantitated by densitometry.

The fraction product, F_p, at each time point t, was determined as follows:

$$F_p(t)=\frac{{\delta }_p(t)}{{\delta }_s(t)+{\delta }_p(t)}$$

in which δ_s(t) and δ_p(t) represent the densities of substrate and product bands, respectively, at time t. The product yield at each time point, P(t), was corrected for baseline levels of damage as follows:

$$P(t)=\frac{F_p(t)-F_p(0)}{1-F_p(0)}$$

in which F_p(0) is the fraction product in the QC sample for manually-quenched experiments, or the fraction product in the −E sample for the RQF-quenched experiments. For STO experiments, the product yield for each time course was plotted as a function of time and fit using nonlinear least-squares regression to the modified first-order integrated rate law:

$$P(t)=P(\infty )(1-e^{-k_{\text{obs}}t})$$

where P(t) is the product yield at time t, P(∞) is the maximum product yield, and k_obs is the observed rate of product formation (Kaleidagraph). For MTO experiments, the product yield for each time course was plotted as a function of time and fit using nonlinear least-squares regression to burst plus linear function:

$$P(t)=A(1-e^{-k_{\text{obs}}t})+vt$$

where P(t) is the product yield at time t, A is the amplitude of the burst, k_obs is the observed rate of the burst, and v is the steady-state rate of the reaction (Kaleidagraph). The overall rate of the reaction k_TO is calculated by dividing v by the concentration of enzyme, [E], in the reaction. The time courses depicted for k_TO are single representative time courses at the same concentration of enzyme over a wide range of enzyme concentrations measured. The reported rates represent mean ± standard error. Individual A and v values for the MTO time courses are given in Table S2.

Extrapolation of k_max and K_D for ss5-FU substrates

To extrapolate k_max and K_D values for TDG^WT and AcK analogues, manual time courses under STO conditions were performed as described above with the following final concentrations: 25 nM or 250 nM DNA, 1 μM TDG; and 500 nM or 1 μM DNA, and 2 μM TDG. The individual k_STO rates were determined by fitting the fraction product by nonlinear least-squares regression to a modified first-order integrated rate law as described above. The rates were then plotted as a function of potential [ES] and fit to a hyperbolic curve by nonlinear least-squares regression (Kaleidagraph):

$$k_{\text{STO}}=\frac{k_{\max }[ES]}{K_{\mathrm{D}}+[ES]}$$

where k_STO is the observed STO rate, k_max the extrapolated maximal STO rate, [ES] the potential [ES] concentration, and K_D the extrapolated binding constant.¹²

Statistical Analysis

Values for k_max and k_TO, and were compared individually within each substrate by analysis of variance (ANOVA) followed by a post hoc Tukey test for pairwise comparison (α=0.01) using R.⁶⁵ Here we consider p<0.01 to be significant.

RESULTS AND DISCUSSION

Rationale

To examine the effect of acetylation on K residues in the N-terminus of wild-type TDG (TDG^WT), we chose to mutate K to Q (Figure 1b). Q is similar both in size and charge to AcK, and has been experimentally validated as a suitable analogue.^66,67 Each AcK site was mutated individually to Q. Given the propensity of CBP/p300 to acetylate K near AcK,^33,47 we also mutated the cluster of three lysine residues (K83, K84, K87) to Q, which here will be called 3Q (Table 1). Collectively, we refer to these K to Q mutants as TDG-AcK analogues.

Table 1.

List of TDG AcK analogs and corresponding amino acid sequence used in this study.

TDG	Sequence
WT	N- …59K… 82SKKSGKS…-C
K59Q	N- …59Q… 82SKKSGKS…-C
K83Q	N- …59K… 82SQKSGKS…-C
K84Q	N- …59K…82SKQSGKS…-C
K87Q	N- …59K… 82SKKSGQS…-C
3Q (K83Q-K84Q-K87Q)	N- …59K… 82SQQSGQS…-C

We chose to use a previously published 19mer oligonucleotide sequence^10,11 that contains a centrally located CpG site and allows for binding of only a single TDG molecule (Figure 2a). It is important to note that the substrates examined in this study—T, U, and 5-FU—are in the 5′-XpG/CpG-3′ sequence context (in which X is a lesion). TDG contains residues responsible for recognizing the 5′-CpG-3′ on the complement strand, and prefers substrates in this sequence context.^12,13

Figure 2.

(A) DNA duplex substrate^10,11 used in this study. X represents T, U, or 5-FU, positioned in a 5’-XpG/CpG-3’ sequence context. (B) Representative denaturing PAGE depicting the excision of T, U, or 5-FU from duplex substrates by TDG^WT under STO conditions. NT, no treatment; −E, no enzyme; QC, quench control. (C) Corresponding graphs depicting the excision of T (circles), U (squares), or 5-FU (diamonds) from duplex substrate by TDG^WT. Data were fit to a single exponential equation and error bars represent standard error.

We probed the effects of N-terminal acetylation on TDG using two broad sets of conditions: single turnover (STO) conditions, in which [TDG]>>[DNA], and multiple turnover (MTO) conditions, in which [DNA]>>[TDG]. In the minimal kinetic scheme of TDG (Scheme 1), TDG first binds to the DNA substrate. The lesion X—either T, U, or 5-FU—is flipped into the active site, the equilibrium of which is described as K_flip. The N-glycosidic bond is cleaved (k_chemistry; k_chem), generating an AP site. The product is then released. It is important to note that TDG has significant affinity for abasic sites,^26,68 and the AP-containing DNA may be re-bound by TDG.

Scheme 1.

Minimal kinetic scheme of TDG. TDG first binds to the DNA substrate containing the lesion (X). TDG flips the lesion into the active site, the equilibrium of which is given by K_flip. TDG then cleaves the N-glycosidic bond (k_chem). The free nucleobase and AP-containing DNA products are released. Notably, TDG can re-bind the AP-containing DNA (end-product inhibition). The observed STO rate (k_STO) is representative of steps up to and including chemistry. The maximal STO rate (k_max) is representative of the steps after initial binding up to and including chemistry. The MTO rate of turnover (k_TO) is representative of all steps after initial binding, including any end-product inhibition.

The rates obtained from STO experiments (k_STO) give information about the events up to and including the first irreversible step, which for TDG is glycosidic bond cleavage. Under conditions in which [TDG]>[DNA] and [TDG]>>K_D, k_STO collapses to the maximal STO rate (k_max), which is reflective of the rates after binding and up to and including chemistry, which can be expressed in as in Eq. 1:^25,69

$$k_{\max }=k_{\text{chem}}\frac{K_{\text{flip}}}{1+K_{\text{flip}}}.$$ (1)

MTO experiments give information about the entire catalytic cycle. As is true for many glycosylases and other BER enzymes, the rate-determining step of TDG occurs after chemistry.^12,13,26 This results in biphasic product accumulation (also called burst kinetics) prior to attainment of the steady-state phase of the reaction (pre-steady state).^27,70 The burst represents the first turnover of the enzyme, with the subsequent linear phase indicative of subsequent turnovers and is equivalent to the initial velocity (v₀). Under conditions in which [DNA]>>K_m, v₀ collapses to k_TO[E], where k_TO represents the rate of turnover of the enzyme. Given the high affinity of TDG for the AP-containing DNA product,^26,68 we cannot assume that k_TO is equivalent to k_cat, which assumes no end-product inhibition of the enzyme.⁷¹ In performing experiments under both STO and MTO conditions, we gain insight into the full catalytic cycle of TDG in both the acetylated and non-acetylated states.

k_max is not affected in the N-terminal AcK analogues

To evaluate the kinetic steps up to and including chemistry, we performed time courses under STO conditions. To ensure [TDG]>>K_D for determination of k_max, experiments were performed with at least two different enzyme and substrate concentrations. As expected, TDG^WT exhibits monophasic kinetics and converts all substrates fully to product over the course of the reaction, though at much different time scales (Figure 2b, c). This significant difference in the rates of excision is reflected in the k_max values: TDG-WT excises T with k_max=0.30 ± 0.3 min⁻¹ and U and 5-FU ~14 and ~3,300 times faster, respectively (Table 2), similar to previously reported rates.^{10,11,13,25,26,44,72} The increased rates of excision have been attributed both to increased base flipping²⁵ and increased rate of chemistry due to decreased N-glycosidic bond stability¹⁰ for U and 5-FU relative to T.

Table 2.

Rates of processing duplex substrates by TDG^WT and AcK analogs. ^aRates are reported as mean ± standard error (n). ^b25 nM U:G substrate, 1000 nM TDG. Asterisk (*) denotes p<0.01 versus slow phase of 3Q by ANOVA and post-hoc Tukey test for pairwise comparisons for U:G 25 nM substrate.

Enzyme	Substrate	kmax (min−1)a	kmax relative to T:G	kSTO (min−1)a,b , yield (%)	kTO (min−1)a	kTO relative to T:G	kmax/kTO
TDGWT	T:G	0.30 ± 0.03 (4)	1		0.007 ± 0.002 (7)	1	43
	U:G	4.2 ± 0.2 (3)	14	4.1 ± 0.6*, 91 (4)	0.14 ± 0.05 (5)	20	30
	5-FU:G	1000 ± 290 (3)	3300		2.5 ± 1.0 (3)	360	400
TDGK59Q	T:G	0.42 ± 0.01 (7)	1		0.003 ± 0.001 (7)	1	140
	U:G	4.4 ± 0.4 (5)	10	3.6 ± 0.10*, 96 (3)	0.04 ± 0.01 (7)	10	150
	5-FU:G	940 ± 130 (3)	2200		3.4 ± 0.8 (6)	100	280
TDGK83Q	T:G	0.36 ± 0.02 (4)	1		0.002 ± 0.001 (9)	1	180
	U:G	4.4 ± 0.3 (3)	12	1.9 ± 0.2*, 92 (3)	0.09 ± 0.03 (9)	45	50
	5-FU:G	1260 ± 320 (3)	3500		2.2 ± 0.5 (3)	1100	570
TDGK84Q	T:G	0.46 ± 0.05 (4)	1		0.009 ± 0.002 (6)	1	51
	U:G	4.2 ± 0.3 (3)	9.1	3.6 ± 0.5*, 93 (4)	0.03 ± 0.01 (4)	3.3	140
	5-FU:G	980 ± 280 (3)	2100		3.4 ± 0.7 (3)	110	290
TDGK87Q	T:G	0.42 ± 0.05 (4)	1		0.006 ± 0.003 (5)	1	70
	U:G	3.9 ± 0.2 (3)	9.3	2.1 ± 0.3*, 90 (4)	0.11 ± 0.04 (4)	18	35
	5-FU:G	790 ± 14 (3)	1900		2.3 ± 0.4 (3)	380	340
TDG3Q	T:G	0.43 ± 0.05 (4)	1		0.004 ± 0.002 (3)	1	110
	U:G	3.1 ± 0.3 (5)	7.2	2.9 ± 0.2*, 75 (3)	0.05 ± 0.01 (7)	13	62
				0.04 ± 0.01, 25 (3)
	5-FU:G	750 ± 130 (3)	1700		4.5 ± 0.9 (4)	1100	170

All TDG-AcK analogs also exhibit monophasic kinetics for T, U, and 5-FU substrates (Figure 3, Table 2). For the excision of T and U, k_max is not statistically different for any of the TDG-AcK analogs compared to TDG^WT (Figure 3a, b). Though these values are not statistically different from WT or each other in the context of this short oligonucleotide, it is interesting to consider the slight variation among the TDG-AcK analogs. The k_max for TDG^WT is 14 times faster for U, while TDG^K84Q, TDG^K87Q, and TDG^3Q are less than 10 times faster (Table 2). The difference in relative k_max values suggests a possible shift in the kinetic discrimination of the enzyme between T and U substrates with acetylation. NMR studies have demonstrated that the N-terminal tail of TDG—specifically the region containing the acetylation and phosphorylation sites—exhibits different conformations when bound to a T or U substrate.²⁰ When bound to T, the N-terminal tail folds back upon the catalytic domain and likely contacts acidic residues on the surface of the enzyme, while when bound to U, the tail is extended, and may be in contact with the DNA.²⁰ Acetylation of the N-terminus may change the interaction of the tail with the catalytic domain and/or the adjacent DNA, resulting in the observed discrepancies in relative k_max.

Figure 3.

k_max is not affected in the TDG-AcK analogues. STO time courses were used to determine the effect of N-terminal acetylation of TDG on the excision of T (A), U (B), and 5-FU (C) from duplex substrates. Reactions consisted of 25 nM (T, 5-FU) or 250 nM (U) duplex substrate and 1000 nM TDG in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). Data were fit to a single exponential equation and error bars represent standard error. Black, TDG^WT; purple, TDG^K59Q; green, TDG^K83Q; red, TDG^K84Q ; blue, TDG^K87Q; orange, TDG^3Q.

For every TDG-AcK analogue, the excision of 5-FU is much faster than T and U, with k_max values of ~1000 min⁻¹ (Figure 3c). Similar to the excision of T and U, the k_max values are not significantly different from each other on this substrate.

Taken together, these data indicate that acetylation of the N-terminus of TDG does not affect k_max on short oligonucleotides.

STO kinetics are changed in acetylation analogues at lower U substrate concentrations

In determining k_max, we performed experiments at varying enzyme and substrate concentrations to ensure measurement of the maximal single turnover rate. Though we did not observe differences among these titrations for T and 5-FU, there was a striking difference between TDG^3Q and all other TDG-AcK analogs in the excision of U (Figure 4). In contrast to TDG^WT and all other analogs at the substrate concentrations examined, TDG^3Q exhibits biphasic kinetics. The fast phase, comprising 75% of the product accumulation, proceeds at a rate k_STO-fast=2.9 ± 0.2 min⁻¹, which is not statistically different from k_STO for the other TDG-AcK analogs under these conditions nor from k_max for U excision by any TDG-AcK analog. The slow phase, however, which accounts for 25% of the product, occurs at a rate of k_STO-slow=0.04 ± 0.01 min⁻¹. This rate is significantly different from both the k_STO-fast for TDG^3Q, k_STO under these conditions for all other TDG-AcK analogs, and every k_max for excision of U, indicating the measurement of a substrate-dependent rate-determining step prior to chemistry.

Figure 4.

TDG^3Q exhibits biphasic kinetics under STO conditions at lower [U]. Reactions consisted of 25 nM U duplex substrate and 1000 nM TDG in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). Data were fit to a single or double exponential equation and error bars represent standard error. Black, TDG^WT; purple, TDG^K59Q; green, TDG^K83Q; red, TDG^K84Q; blue, TDG^K87Q; orange, TDG^3Q.

Furthermore, it is worthwhile to note that, except for TDG^WT, the k_STO values at 25 nM are generally slower than at 250 nM (Table 2). These results indicate that the pre-chemistry complex (TDG•DNA-X_flipped, Scheme 1) is not entirely saturated at 25 nM U:G substrate. The single TDG-AcK analogs still display monophasic product accumulation kinetics, however, suggesting the impact of a substrate-dependent step within a single progress curve only when the tail is acetylated. Taken with the slight disparities in relative k_max values for T and U excision observed for TDG^K84Q, TDG^K87Q, and TDG^3Q, these data point to a shift in kinetic discrimination of T and U with acetylation.

k_TO is not affected in N-terminal AcK analogues

To evaluate all steps of the reaction after initial binding, we conducted kinetic time courses under MTO conditions. For all substrates, TDG^WT and every TDG-AcK analog exhibit burst kinetics, in which the first turnover of the enzyme is represented in the exponential rise of the curve, followed by a linear phase reflective of subsequent turnovers (Figure 5). As described above, the biphasic nature of pre-steady state product accumulation indicates the presence of a rate-determining step following chemistry. For the excision of T, k_TO is incredibly slow, with k_TO<0.01 min⁻¹ for TDG^WT and the TDG-AcK analogs (Figure 5a, Table 2). This rate translates to k_TO<15 day⁻¹. The k_TO rates on the short oligonucleotides are not statistically different from each other.

Figure 5.

k_TO is not affected in the TDG-AcK analogues. Representative MTO time courses depict the effect of N-terminal acetylation of TDG on the excision of T (A), U (B), and 5-FU (C) from duplex substrates. Reactions consisted of 1000 nM duplex substrate and 100 nM TDG in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). Data were fit to a biphasic exponential plus linear equation. Black, TDG^WT; purple, TDG^K59Q; green, TDG^K83Q; red, TDG^K84Q; blue, TDG^K87Q; orange, TDG^3Q.

Similar to what was observed under STO conditions, TDG processes and subsequently turns over faster on U and 5-FU compared to T substrates (Figure 5b,c). Indeed, TDG^WT has a k_TO ~20 and ~350-fold greater for U and 5-FU, respectively, relative to T. Notably, k_TO for 5-FU is within the same order of magnitude as k_max for U. As observed for T substrates, for U and 5-FU there is not a statistical difference for the k_TO values of TDG^WT and the TDG-AcK analogs.

TDG^K84Q, however, stands out as having a k_TO for U that is less than 5-fold greater than k_TO for T, and TDG^3Q has relative k_TO ~10, again suggesting a possible change from TDG^WT in T and U kinetic discrimination. On the other hand, TDG^K83Q displays the greatest difference of ~50 between U and T k_TO, indicating a potential change from TDG^WT, but clearly differently from TDG^K84Q.

Furthermore, slight differences among TDG^WT and the TDG-AcK analogs are revealed by the k_max/k_TO values. For TDG^WT, k_max/k_TO are 43, 30, and 400 for T, U, and 5-FU, respectively. TDG^K87Q is the only TDG-AcK analogue for which k_max/k_TO are similar to TDG^WT for every substrate. TDG^K59Q, on the other hand, is the only analogue with k_max/k_TO over 100 for both T and U. These ratios would indicate that TDG^K87Q is most similar to TDG^WT in processing these three substrates, while TDG^K59Q is the least similar.

Importantly, there are striking differences in k_TO values for T, U, and 5-FU excision for all analogs. As discussed above, k_TO is reflective of the slowest step(s) of the reaction, which for TDG occurs after chemistry. In considering the minimal kinetic scheme, these steps encompass nucleobase release, AP site release, and any rebinding of TDG to the AP site (end-product inhibition). Molecular dynamics simulations have suggested that nucleobase release occurs on the nano- to microsecond time scale immediately following chemistry while the AP site is still bound by the enzyme.⁷³,⁷⁴ In this case, it is unlikely that nucleobase release is rate-limiting, and thus step(s) beginning with AP-bound TDG contribute to k_TO. One may assume, therefore, that the steps governing AP site release and/or rebinding to the AP site would be the same regardless of starting substrate since the enzyme has reached the same step. Under these assumptions, k_TO would be comparable regardless of the substrate. As demonstrated previously²⁶,²⁷,⁶⁸ and above, these assumptions do not hold. In this case, there may be more steps to release of an AP site from a T substrate than U, which in turn has more steps than release from a 5-FU substrate. Such step(s) could include, for example, unfurling of the N-terminal tail from the catalytic domain in the release of an AP site from T, a step that would not be required in the release of the AP site from U. Under this interpretation, the lower limit for AP site release by TDG is revealed by 5-FU k_TO ~2-5 min⁻¹.

5-FU in single-stranded DNA is a substrate for TDG

Experimental evidence indicates that duplex substrates are required for the excision of T and U by TDG.⁹,¹⁷,²¹,⁶⁸ This observation is supported by co-crystal structures of TDG bound to DNA where residues specifically recognize the complementary strand. ¹²,¹³ Given how readily not only TDG^WT, but also all the TDG-AcK analogs, both excise and turn over 5-FU in duplex DNA, we asked whether 5-FU in single-stranded DNA (ss5-FU) is a substrate for TDG. To test this potentially novel capability of TDG, we incubated ss5-FU with TDG^WT and the TDG-AcK analogs. The 5-FU duplex substrate used in the experiments described above served as the positive control. Much to our surprise, we observed excision of 5-FU from the single-stranded substrate by TDG^WT and all the TDG-AcK analogs (Figure 6).

Figure 6.

5-FU in ssDNA is a substrate for TDG. Denaturing PAGE depicting reactions consisting of 25 nM 5-FU duplex (ds) or 250 nM ssDNA (ss) substrate incubated with 1000 nM TDG for 1 min (1) or 60 min (60) in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). NT, no treatment; −E, no enzyme; QC, quench control.

We next endeavored to measure the k_max values for excision of ss5-FU, starting with STO conditions that were suitable for the measurement of k_max on duplex substrates. To confirm that the rates determined correspond to k_max, we performed additional STO experiments at increasing enzyme and substrate concentrations (Figure 7, Figure S3).

Figure 7.

Excision of 5-FU from ssDNA by TDG^WT under STO conditions. Reactions were performed with 25 nM DNA, 1000 nM TDG^WT (circles), 250 nM DNA, 1000 nM TDG^WT (squares), 500 nM DNA, 2000 nM TDG^WT (diamonds), and 1000 nM DNA, 2000 nM TDG^WT (triangles) in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). Data were fit to a single exponential curve.

When k_STO does not increase with increasing [DNA] and [TDG], one knows k_max has been reached. At even the highest concentrations tested, which were limited by the concentration of TDG, we could not confirm attainment of k_max. Such titrations under STO conditions, however, allow for modeling of both k_max and K_D values by fitting the k_STO as a function of the potential [ES] concentration:⁷⁵

$$k_{\text{STO}}=\frac{k_{\text{max}}[ES]}{K_{\mathrm{D}}+[ES]}.$$ (2)

In modeling of TDG^WT (Figure 8), the projected k_max=2.3 min⁻¹, which is within an order of magnitude of k_max for U excision from duplex and is comparable to k_TO for 5-FU on duplex.

Figure 8.

Extrapolation of k_max and K_D values from STO experiments with ss5-FU. The observed single turnover rate (k_STO) is plotted as a function of potential [ES] and data were fit to a hyperbolic curve. Black, TDG^WT; purple, TDG^K59Q; green, TDG^K83Q; red, TDG^K84Q; blue, TDG^K87Q; orange, TDG^3Q.

Furthermore, the predicted K_D for TDG^WT is 4 μM, which is over two orders of magnitude greater than previously measured values for T in duplex substrates (18 nM)^13,76 and over four orders of magnitude greater than K_D values for U in duplex substrates (0.6 nM)^12,76. Though the k_max and K_D values for the TDG-AcK analogs are not as well defined as for TDG^WT, comparison of the curves to TDG^WT allows for generalizations of relative k_max and K_D values (Figure 8). Namely, the TDG^K59Q and TDG^3Q curves track below TDG^WT, suggesting that for these two TDG-AcK analogs, k_max may be lower, K_D may be higher, or a combination of the two. On the other hand, the curves of the single AcK analogs in the cluster—TDG^K83Q, TDG^K84Q, and TDG^K87Q—track with or above WT, indicating that k_max may be comparable to or higher than TDG^WT, and K_D may be comparable to or lower than TDG^WT. Though ss5-FU reveals differences among TDG^WT and the TDG-AcK analogs that were less apparent on the duplex substrates, it is important to note that 5-FU is excised from single-stranded DNA at rates orders of magnitude slower than from duplex.

The rate-determining step of ss5-FU processing occurs prior to or at chemistry

Since the modeled k_max of ~2 min⁻¹ for TDG^WT on ss5-FU is comparable to k_TO on 5-FU duplex substrates, we asked whether product release is rate-limiting for ss5-FU. If product release is rate limiting, a burst would be observed under pre-steady state MTO conditions, as we observed for T, U, and 5-FU in duplex (Figure 5). On the other hand, if product release is not rate limiting, the burst will be severely depressed or lost, and pre-steady state product will accumulate in a linear manner. We captured the full reaction curve, which demonstrates that the single-site TDG-AcK analogs process ss5-FU faster than TDG^WT under these conditions, while TDG^3Q processes ss5-FU similarly to TDG^WT (Figure 9). By zooming in to the pre-steady state portion of the curve (Figure 9, inset), no burst is observed for any of the TDG enzymes. Therefore, in processing ss5-FU, the rate-determining step of the reaction has shifted from product release to a step before or at chemistry.

Figure 9.

The RDS in the processing of ss5-FU occurs before or at chemistry. Reactions were carried out with 1000 nM ss5-FU and 250 nM TDG in Buffer TNK (20 mM Tris-HCl (pH 7.6), 25 mM NaCl, 75 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg/mL BSA). The first 3 min of the reaction were fit to a linear equation. Black, TDG^WT; purple, TDG^K59Q; green, TDG^K83Q; red, TDG^K84Q; blue, TDG^K87Q; orange, TDG^3Q.

Biological Implications

Potential implications of TDG acetylation at the molecular level

Here we have demonstrated that AcK analogues in the N-terminus of TDG affect neither k_max nor k_TO on short oligonucleotide substrates for processing T, U, and 5-FU. However, ss5-FU and decreased [U] substrates have revealed subtle differences among TDG^WT and the TDG-AcK analogues. Furthermore, mimicking acetylation seems to cause a shift in kinetic discrimination between T and U, namely decreased processing of T relative to the unacetylated state, in accordance with a previous study by Madabushi et al.³¹ Our results indicate that this shift is the consequence of decreased turnover (k_TO) relative to base-flipping and chemistry (k_max), which may be from decreased rates of AP site release and/or increased affinity for the free AP site. In a cellular context, this characteristic would translate to increased protection of the highly labile AP site^4,26,27 until processing by APE1.

In addition to decreased processing of T by AcTDG compared to the acetylated state, Madabushi and colleagues also found that AcTDG increased activity on 5-FU substrates.³¹ Though our kinetic data with 5-FU in duplex DNA do not reveal differences among TDG^WT and the AcK analogs, experimental procedures may account for the seeming disparate results. The TDG used in the investigation by Madabushi et al. was enzymatically acetylated by p300 and AcCoA, which could lead to a potentially heterogeneous population of acetylation sites in the N-terminus. Furthermore, the concentrations of both duplex DNA and TDG were much lower than those in our reactions. It is worthy of note, however, that we did observe subtle differences among TDG^WT and the AcK analogues on the ss5-FU substrate. The much lower concentrations of the experiments by Madabushi et al. may be revealing the same subtleties we observe with ss5-FU. That is, only conditions that “challenge” TDG with sub-saturating concentrations of substrate or sub-optimal substrate uncover the effect of N-terminal acetylation.

In considering the TDG-AcK analogues, TDG^K87Q stands out as processing substrates most similarly to TDG^WT, while TDG^K59Q is the single AcK analogue that is least similar to TDG^WT. Interestingly, comparable conclusions regarding the role of acetylation sites were drawn regarding the glycosylase NEIL2.⁷⁷ NEIL2, which is also a target of CBP/p300, has a similar acetylation pattern in its N-terminus to TDG: a more N-terminal site (K49), and a cluster of three residues closer to the catalytic domain (K149, K150, K153).⁷⁷ It was reported that while both K49 and K153 were the main targets for acetylation, decreased activity was observed only when K49 was acetylated, and thus the acetylation of K153 may play a different biological role.⁷⁷ NEIL2 K49 is analogous to TDG K59. Mimicking acetylation of TDG K59, as discussed above, leads to activity that is least similar to TDG^WT, suggesting that single acetylation sites outside of clusters may function as a “switch” for changes in activity. TDG K87, the mutation of which leads to the least change in activity from TDG^WT, is analogous to NEIL2 K153. TDG K87, and NEIL2 K153, for that matter, may therefore function as “sentinel” acetylation sites, i.e. this K is the first to undergo acetylation in the cluster of three. The acetylation of K87 in TDG would serve as the anchor for the bromodomain of CPB/p300, which would then acetylate K83 and/or K84. In this case, the cluster of three K residues serve as tunable activity for TDG, culminating in the acetylation of all three K residues. As demonstrated above, TDG^3Q clearly processes U differently than the other TDG-AcK analogues and TDG^WT. The acetylation of TDG at K59 versus at the cluster of three may therefore be differentially regulated and serve distinct purposes in vivo.

Considering regulation of and roles for TDG acetylation at the cellular level

It is interesting to consider when TDG may undergo acetylation within the cellular context. Studies on other BER enzymes have demonstrated increased acetylation when CBP/p300 is activated.^{33,47,54,55,89} These circumstances include cellular stress (e.g. oxidative stress and hyperosmolarity), in response to hormones and certain growth factors, and at times of acetyl-coenzyme A (AcCoA) abundance (Figure 10a).^33,47,78,79 In the case of cellular stress, transcription of genes responsible for responding to the stress increases.^66,79,90 Acetylation of TDG and decreased processing of T substrates may therefore be a way to maintain heterochromatin and preserve cellular resources while the cell is trying to maintain homeostasis. It would follow that regulation of TDG in this way may be through acetylation of K59, which would allow for the “switch” in activity necessary for the cell.

Figure 10.

Potential roles and regulation of TDG acetylation within the cell. (A) TDG undergoes acetylation by CBP/p300, which requires the cofactor AcCoA. Cellular stress, including oxidative stress by hydroxyl radicals (·OH), and cellular growth and signaling factors can induce the activity of CBP/p300,^33,47,78,79 leading potentially to greater acetylation of TDG. The NAD⁺-dependent SIRT1 deacetylates TDG, which is under the control of small molecules, nutrient availability, and NAD⁺/NADH levels.^31,51,80 CBP/p300 and/or SIRT1 can be dysregulated in cancer.^48,49,81-83 (B) Acetylation of the N-terminus may promote interactions between TDG and its many protein binding partners.^30-40 (C) Acetylation of the N-terminus may promote LLPS of TDG into the heterochromatic compartment.^84-88 This allows for TDG to access substrates in heterochromatin. (D) The acetylation of the N-terminus leads to differential processing of U duplex and ss5-FU substrates under sub-saturating conditions.

Furthermore, SIRT1 and/or CBP/p300 are found to be under/overexpressed and/or dysregulated in a variety of cancers (Figure 10a).^48,49,81-83 These states would lead to loss of regulation of acetylation of both chromatin and associated enzymes, such as TDG. This could result in an increase of “missed repair” opportunities by TDG. Indeed, one of the most common mutations found in cancers is the C to T transition mutation, especially in 5′-CpG-3′ dinucleotide contexts.^1-3,49,91 On the other hand, the primary role of acetylation may not be kinetic, but rather structural. As mentioned above, TDG interacts with a multitude of proteins,^30-40 and PTM may promote and/or prevent the association of TDG with certain binding partners (Figure 10b).

The N-terminus of TDG shares many features with intrinsically disordered proteins/regions (IDP/Rs):^84,85 its amino acid composition is rich in glycine (G), S, Q, and tyrosine (Y); it contains several PTM sites, including clustered phosphoserine and AcK sites;^31,36,46 and it is unstructured via NMR.²⁰ One major characteristic of IDP/Rs is the ability to undergo liquid-liquid phase separation (LLPS); that is, to create a distinct compartment within a liquid mixture, such as oil droplets in water or the nucleolus inside the nucleus.^84,85 For many IDP/Rs, K acetylation promotes phase separation, while serine phosphorylation prevents phase separation.^84,85 Interestingly, previous work has demonstrated that phosphorylation and acetylation of the N-terminus of TDG are mutually exclusive, and the authors suggest the primary purpose of phosphorylation is the inhibition of acetylation.⁴⁶

Heterochromatin is known to exist in a phase-separated state inside the nucleus.^84,86,87 Acetylation of TDG, therefore, may promote the separation of the enzyme into the heterochromatic phase, where it can access its substrates (Figure 10c). The Sczepanski laboratory demonstrated the ability of TDG to reversibly condense chromatin, suggesting the enzyme, even in the absence of PTM, is capable of phase separation.⁸⁸ Acetylation and phosphorylation of TDG may regulate the phase separation properties of TDG, thereby determining the genomic compartment in which it is active.

This property may be especially important in the removal of the chemotherapeutic 5-FU from the genome (Figure 10d). As demonstrated above, 5-FU in duplex is excised and turned over by TDG at rates orders of magnitude greater than T and U. This brings into question whether the rate of AP site release on which the cell typically relies for regulation of TDG-initiated BER is lost. In the event of incomplete or slower repair by Polβ and ligase, multiple strand breaks proximal to each other could initiate the double-strand break response. If this response overwhelms the cell, it will die by apoptosis (if the machinery is intact), or necrosis. As such, it has been suggested that TDG is the primary glycosylase responsible for cellular sensitivity or resistance to 5-FU.^88,92 A recent study, however, identified UNG in particular, and not TDG, as a modulator of sensitivity to 5-FU incorporated into DNA in a wide variety of colon cancer cell lines.⁶⁴ These compelling results suggest that, if TDG does have a role in the cellular response to 5-FU, its role is much more complex than previously appreciated. Even in the case that TDG plays a minor role in the sensitivity of a tumor to 5-FU, our results indicate that, if 5-FU nucleobases persist until discovery by TDG, the enzyme will remove and turnover 5-FU duplex substrates significantly faster than T and U. The typical regulation of TDG as the cell uses for T and U are lost, as TDG is “distracted” by 5-FU over its native substrates.

It is therefore important to consider whether ss5-FU is a relevant substrate for TDG in cellulo. We have demonstrated that TDG is capable of excising ss5-FU, though at rates orders of magnitude slower than 5-FU in duplex, indicating that the cellular substrate of 5-FU for TDG is duplex DNA. Ss5-FU does demonstrate, however, that there are subtle differences among TDG^WT and the TDG-AcK analogues. Though these differences are likely insignificant on short oligonucleotide substrates, they would be amplified on the genomic scale. That is, discrepancies in rates that are insignificant on a short oligonucleotide may prove significant inside the cell. As such, amplification of minute differences may translate to fine-tuned regulation of TDG enzymatic activity so that key epigenetic information is maintained or (purposefully) lost by the cell.

Conclusions

In this study, we have demonstrated that acetylation analogues in the N-terminus of TDG exhibit changes neither in k_max nor in k_TO on short oligonucleotides, as well as the novel capability of TDG to excise 5-FU from ssDNA. Furthermore, the processing of ss5-FU by TDG exhibits different kinetics than the duplex substrates, with the rate-determining step occurring prior to or at chemistry. Subtle differences among the AcK analogues are revealed by ss5-FU, differences that may prove significant when placed on the genomic scale. Furthermore, acetylation of the N-terminus of TDG may play regulatory and/or structural roles, with kinetic variation playing a secondary role. These observations augment the argument for the consequential role of TDG in cellular sensitivity and/or resistance to 5-FU in the treatment of cancer.

ABBREVIATIONS

5-caC

5-carboxycytosine

5-fC

5-formylcytosine

5-FU

5-fluorouracil

5-mC

5-methylcytosine

AcCoA

acetyl coenzyme A

AcK

acetyl-lysine

ANOVA

analysis of variance

AP

apurinic/apyrimidinic

APE1

AP endonuclease 1

BER

base excision repair

C

cytosine

DDR

DNA damage response

G

guanine, glycine

IDP/R

intrinsically disordered proteins/regions

K

lysine

K _D

dissociation constant

k _max

maximal single turnover rate

k _STO

observed single turnover rate

k _TO

turnover rate

LLPS

liquid-liquid phase separation

MBD4

methyl binding domain 4

MTO

multiple turnover

NAD⁺/NADH

nicotinamide adenine dinucleotide

PAGE

polyacrylamide gel electrophoresis

PCNA

proliferating cell nuclear antigen

polβ

DNA polymerase β

PTM

post-translational modification

Q

glutamine

S

serine

STO

single turnover

T

thymine

TS

thymidylate synthase

TDG

thymine DNA glycosylase

U

uracil

Y

tyrosine