| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 281, Issue 8, 4983-4992, February 24, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1
2
From the
Division of Molecular Biosciences and the Centre for Structural Biology, Imperial College London, South Kensington, London SW7 2AZ, United Kingdom, the ¶Department of Biochemistry, University of Bristol, University Walk, Bristol BS8 1TD, United Kingdom, and the ||School of Crystallography, Birkbeck College, Malet Street, London WC1E 7HX, United Kingdom
Received for publication, August 18, 2005 , and in revised form, November 18, 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-fluoro-2'-deoxyuridine). Equilibrium DNA binding revealed that both human and HSV-1 UNG enzymes bind to abasic DNA and both substrate analogs more weakly than to uracil-containing DNA. Structure determination of HSV-1 D88N/H210N UNG in complex with uracil revealed detailed information on substrate binding. Together, these results suggest that a significant proportion of the binding energy is provided by specific interactions with the target uracil. The kinetic parameters for human UNG indicate that it is likely to have activity against both U·A and U·G mismatches in vivo. Weak binding to abasic DNA also suggests that UNG activity is unlikely to be coupled to the subsequent common steps of base excision repair. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
UNG is one of the initiating DNA glycosylases of the base excision repair pathway and is responsible for the specific recognition and removal of uracil. It removes uracil by cleavage of the N-glycosidic bond between the uracil and the deoxyribose backbone, leaving an apyrimidinic (AP) site (2). AP sites are themselves highly mutagenic and are further repaired by the common downstream enzymes of the base excision repair pathway (3). The first common repair step is an AP endonuclease, which cleaves the DNA backbone at the abasic site (4).
UNG enzymes have been identified in a wide variety of organisms, including human, Escherichia coli, and herpes simplex virus type 1 (HSV-1). They are well conserved; the HSV-1 and human enzymes have 39% sequence identity, whereas the HSV-1 and E. coli enzymes are 49% identical. They are also highly conserved at the structural level (5-7). The structure of human (h) UNG in complex with DNA has revealed that it binds uracil by flipping the target nucleotide out of the double helix and into the active-site pocket (6, 8). The DNA-binding pocket of UNG is highly specific for uracil in DNA. Other DNA bases and uracils in RNA are excluded from this DNA-binding pocket due to unfavorable steric and hydrogen bond interactions (5, 6).
A number of previous studies on both the human and HSV-1 UNG enzymes have indicated possible differences in their catalytic behavior. Notably, hUNG has been reported to bind tightly to its abasic product DNA, and it has been proposed that the action of hUNG is coupled to the common base excision repair steps via AP endonuclease (9). In comparison, HSV-1 UNG has been observed to bind product DNA only weakly (10, 11), making coupling of these two events unlikely in the viral enzyme. Furthermore, hUNG has been cited as having a higher Km value (8, 12, 13) than we observed for HSV-1 UNG (10), and this has informed the description of hUNG as having a single function in post-replicative repair of U·A mismatches (14).
In light of these reported differences between the viral and human UNG enzymes, we decided to investigate both enzymes in kinetic and DNA binding studies. The objective was to precisely identify any differences between these two highly related DNA repair enzymes, with a view to further defining their functional roles as well as exploring the potential of viral UNG as a therapeutic target. A previous report has indicated that nucleotide analogs may provide specific inhibitors for the HSV-1 enzyme (15). We therefore wished to investigate the DNA binding of both human and HSV-1 UNG enzymes to non-cleavable substrate analogs both as a probe to investigate protein-DNA interactions and to further explore the possibility of finding specific inhibitors of viral UNG.
We investigated the catalytic rates and DNA binding properties of both wild-type and mutant forms of human and viral UNG using both biochemical and biophysical assays. These included quench reactions to monitor the rate of N-glycosyl hydrolysis and a fluorescence assay to follow steady-state reactions. Furthermore, the binding of the human and HSV-1 UNG enzymes to substrate DNA and abasic product DNA was examined by fluorescence anisotropy. Catalytically inactive mutant enzymes were used in studies of the binding of UNG to uracil-containing substrates. In addition, the binding activity of wild-type UNG was measured using the non-hydrolyzable substrate analogs 2'-deoxypseudouridine (d
rd) and 2'--fluoro-2'-deoxyuridine (-FdUrd). We also performed crystallographic analysis of HSV-1 mutant UNG to compare the active-site structure and mode of uracil binding to the wild-type enzyme.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The recombinant hUNG plasmid pUNG15 was purchased from American Type Culture Collection. The truncated hUNG gene encoding amino acids 84-304 of hUNG was subcloned into expression vector pRSET-B (Invitrogen) using NcoI/HindIII restriction sites. The DNA fragment encoding the enterokinase recognition site located between the N-terminal polyhistidine tag and the truncated hUNG gene was deleted by NcoI/NheI digestion, fill-in reaction (16), and plasmid religation, resulting in pHUNG4. The D145N and H268N mutations were introduced into pHUNG4 by site-directed mutagenesis. Human D145N/H268N UNG was subsequently generated by site-directed mutagenesis using the plasmid containing the D145N hUNG gene as a template. All mutants were verified by sequencing.
Protein Overexpression and Purification—HSV-1 UNG was overexpressed and purified as described previously (10). The D88N, H210N, and D88N/H210N mutant enzymes were expressed and purified from a construct in exactly the same manner. hUNG and its mutant enzymes were expressed in E. coli BH156. The cultures were grown in LB broth at 37 °C to A600 = 0.8 before induction with 1 mM isopropyl 1-thio-
-D-galactopyranoside. The cultures were grown for an additional 3 h at 37 °C before the cells were harvested. The harvested cells were lysed by sonication in buffer containing 20 mM Tris (pH 7.6), 50 mM imidazole, 300 mM NaCl, and a tablet of EDTA-free protease inhibitors (Roche Applied Science). The lysate was clarified by centrifugation at 40,000 rpm for 45 min. The clarified lysate was loaded onto a nickel-Sepharose column equilibrated in 20 mM Tris (pH 7.6), 50 mM imidazole, and 300 mM NaCl. After washing, the UNG enzyme was eluted from the column with a gradient of 50-500 mM imidazole. The fractions were assayed by SDS-PAGE, and those containing UNG were pooled and concentrated by ultrafiltration. The concentrated fractions were diluted with 20 mM Tris (pH 7.6), 1 mM EDTA, and 10% glycerol so that the NaCl concentration was <0.02 M. The partially purified protein was then applied to a poly(U)-Sepharose affinity column equilibrated in 20 mM Tris (pH 7.6), 1 mM EDTA, and 10% glycerol; and after washing, the UNG enzyme was eluted with a gradient of 0-1 M NaCl. The pure enzyme was then concentrated by ultrafiltration. The concentration of the protein was calculated from the absorbance at 280 nm based on an extinction coefficient of 48.9 x 103 M-1 cm-1.
DNA Synthesis and Purification—Oligodeoxynucleotides were synthesized by MWG Biotech. All modifications were incorporated into oligonucleotides at the point of synthesis using conventional phosphoramidite chemistry. The phosphoramidites of 2'-dUrd, drd, -FdUrd, 2-aminopurine (2-AP), and hexachlorofluorescein were also supplied by MWG Biotech. Oligonucleotides were purified by high pressure liquid chromatography as described previously (17). Double-stranded substrates and abasic oligonucleotides were made as described previously (10).
Kinetic Assays—Steady-state fluorescence spectra were measured on a HORIBA Jobin Yvon FluoroMax 3 spectrofluorometer. The 2-AP fluorescence was recorded with an excitation wavelength of 310 nm and an emission wavelength of 370 nm using a 5-mm cuvette. Steady-state fluorescence assays were performed at 25 °C in 50 mM Tris-HCl (pH 7.6), 1 mM EDTA, 50 mM NaCl, 0.1 mg ml-1 bovine serum albumin, and 2 mM MgCl2 using the hUNG enzyme to give a final concentration of 0.5, 1.0, or 2.0 nM. The endpoint of the reaction was determined by addition of the second aliquot of high concentration UNG. The initial velocity (µM s-1) was then calculated from the initial linear progression in fluorescence and the total change in fluorescence. The rate (s-1) was calculated by dividing the initial velocity by the total enzyme concentration.
Single turnover quench reactions were performed at 25 °C with 32P-labeled Urd substrate (0.1 µM; see Table 1) and mutant UNG enzymes (1 µM) in standard reaction buffers (HSV-1 UNG: 50 mM Tris-HCl (pH 8.0), 1 mM EDTA, 150 mM NaCl, and 0.1 mg ml-1 bovine serum albumin; and hUNG: 50 mM Tris-HCl (pH 7.6), 1 mM EDTA, 50 mM NaCl, and 0.1 mg ml-1 bovine serum albumin). The assay was otherwise performed as described previously (18).
|
Equilibrium DNA Binding Assays—Equilibrium DNA binding assays were performed by fluorescence anisotropy using the same spectroscopic settings described above to examine the binding of the wild-type and mutant UNG enzymes to the 5'-hexachlorofluorescein-labeled oligonucleotides (Table 1). The binding assays were conducted at 25 °C in reaction buffer by titrating fixed concentrations of the fluorescent DNA with increasing amounts of the enzyme. The observed anisotropy was plotted against the concentration of enzyme, and the data were fitted to the quadratic binding equation (Equation 1) using GraFit Version 5.0,
 |
Crystallization, Data Collection, and Structure Determination and Refinement—Crystals of HSV-1 D88N/H210N UNG were obtained in microbatch experiments at 25 °C at 30 mg ml-1 protein in 6% polyethylene glycol 4000 and 40 mM imidazole (pH 6.8). The D88N/H210N UNG-uracil complexes were obtained by soaking native crystals in 11% polyethylene glycol 4000, 42 mM imidazole (pH 6.8), and 20 mM 2'-dUrd (Sigma) for 16 h.
Diffraction data were collected on a Quantum 4 CCD detector (Area Detection Systems Corp.) with 0.978-Å radiation on beamline 9.6 of the Daresbury Synchrotron Radiation Source (Cheshire, UK). Diffraction images were processed using Denzo and Scalepack (HKL Research, Inc). Parameters for the data collection and processing are given in Table 2. The structure of HSV-1 D88N/H210N UNG was determined by molecular replacement using molrep (20) and the structure of HSV-1 wild-type UNG (Protein Data Bank code 1UDG) as a search model. The best molecular replacement solution gave a correlation coefficient of 0.470 and an R-factor of 0.438, with corresponding values for the next highest peak of 0.252 and 0.528. The protein model transformed by the molecular replacement solution was refined against the observed diffraction data using CNS (21). The Fo - Fc maps reveal electron density features consistent with the Asn210 and Asn88 mutations.
|
Data for HSV-1 D88N/H210N UNG in complex with uracil were collected on the Quantum 4 CCD detector with 0.87-Å radiation on beamline 9.6 of the Daresbury Synchrotron Radiation Source. The complex crystal had the same space group as the native crystal. Data collection statistics are shown in Table 2. The structure of D88N/H210N UNG in complex with uracil was determined by molecular replacement using molrep (20) and the structure of native D88N/H210N UNG as a search model. The correct solution gave the highest correlation coefficient of 0.507 and an R-factor of 0.400, with corresponding values for the next solution of 0.336 and 0.501, respectively. The model of the complex was refined and rebuilt as described above. The Fo - Fc maps indicate the presence of a uracil base in the active site of UNG. The uracil was constructed and refined with the protein model using CNS (21). The final model gave refinement statistics as shown in Table 2.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
hUNG reactions were performed with a single-stranded oligonucleotide substrate (1U) and double-stranded forms of the substrate containing either a U·G mismatch or a U·A base pair (1U·G and 1U·A, respectively). The rate constants showed a hyperbolic dependence upon the substrate concentration, and the data were fitted to the Michaelis-Menten equation (Fig. 1), giving the steady-state parameters kcat and Km.We also determined that removal of the N-terminal His tag from hUNG had no impact on the catalytic constants (data not shown). The kinetic parameters for hUNG with different substrates are shown compared with those for HSV-1 UNG (Table 3) (10). The kinetic data show an 8-50-fold higher kcat/Km value for hUNG compared with that for HSV-1 UNG. This represents a significantly increased catalytic efficiency for hUNG.
|
The activity of the UNG mutants was monitored under single turnover conditions using the quench assay with the 32P-labeled ssU oligonucleotide (18). Under these conditions, the substrate is in excess over the enzyme; and hence, only a single turnover of the reaction occurs: under saturating conditions, the accumulation of product corresponds directly to the rate of N-glycosidic bond cleavage. The observed rate of cleavage (kobs) was determined by fitting the accumulation of product to either a single exponential or a linear equation.
Reactions were performed under saturating conditions with a 10-fold excess of the single mutant UNG enzymes (human D145N and H268N and HSV-1 D88N and H210N) over the 32P-labeled substrate. The reactions demonstrated that the single mutants of UNG contain some residual activity (Fig. 2). However, the human D145N/H268N and HSV-1 D88N/H210N UNG double mutants showed no UNG activity even after 24 h with a 10-fold excess of the enzyme (data not shown).
|
We used two non-cleavable substrate analogs (drd and -FdUrd). The pucker of the ribose sugar is largely determined by the electronegativity of the 2'-substituent; hence, RNA has a C-3'-endo sugar pucker, whereas DNA has a C-2'-endo sugar pucker (29). Similarly, both the - and -2'-fluorouridine derivatives favor the C-3'-endo pucker (30-32). The -isomer has been suggested to be the better uridine analog (33); in 2-AP fluorescence binding assays with E. coli UNG, it gave a stronger signal; although the -isomer was also observed to be bound by this enzyme, a Kd was not reported for this isomer. The -isomer has also been reported to be a specific inhibitor of HSV-1 UNG (15). In this study, we used the commercially available -isomer.
The binding of DNA by UNG has been observed using 5'-hexachlorofluorescein-labeled DNA substrates. When proteins bind the fluorescently labeled DNA, it leads to a change in the rotational diffusion of the fluorophore due to the change in the molecular mass of the protein-DNA complex relative to free DNA. This in turn results in an increase in fluorescence anisotropy and has been shown to be an ideal method for monitoring the binding of DNA by proteins in solution (34). The advantage of this technique is that the DNA is free to move in solution. Moreover, the fluorescence reporter group is far removed from the target uracil and thus provides a measure of DNA binding that is independent of other conformational transitions associated with damage recognition.
It is of course important to determine that the hexachlorofluorescein reporter group does not interfere with DNA binding by UNG. We have therefore examined this by performing a competition binding experiment. Sufficient HSV-1 D88N/H210N UNG was mixed with the labeled HU oligonucleotide to form an enzyme-DNA complex, and the labeled DNA was then competed away with the unlabeled ssU oligonucleotide. This resulted in a decrease in the observed anisotropy, which could be fitted to a competitive binding model following the method described by Reid et al. (19). This gave a Kd of 4.9 nM for the unlabeled ssU oligonucleotide (Fig. 3), compared with a Kd of 10 nM for the hexachlorofluorescein-labeled HU oligonucleotide (Table 4; see below). This demonstrates that the hexachlorofluorescein reporter group does not interfere with DNA binding by UNG for the substrates used in this study.
|
|
The inactive human D145N/H268N UNG enzyme produced a tight binding curve with both the single- and double-stranded uracil-containing substrates (HU and HU·G) (Fig. 4A). When the wild-type enzyme was examined with the abasic DNAs, the binding was found to be 2 orders of magnitude weaker (Fig. 4B). Similar results were also found with the human H268N UNG single mutant (Table 4).
|
rd and -FdUrd. The wild-type enzyme bound both rd and -FdUrd DNAs 
2 orders of magnitude more weakly compared with the mutant enzymes with uracil-containing DNA (Fig. 5A). We therefore considered that this discrepancy could be due to two causes: the mutant UNG enzymes might bind uracil-containing DNA anomalously tightly; alternatively, the mutants may bind DNA in a very similar manner to the wild-type enzyme, but the non-cleavable analogs that we tested are not good substrate mimics. Intriguingly, a strain hypothesis has been previously proposed for UNG (35), and it is possible that the wild-type enzyme might bind its substrate more weakly compared with a mutant that is unable to exert any strain on the substrate.
This hypothesis was therefore tested by examining the binding of the mutant hUNG enzymes with the non-cleavable substrate analogs rd and -FdUrd. The human D145N/H268N UNG double mutant bound both substrate analogs with a weak affinity similar to that observed with the wild-type enzyme (Fig. 5B). Experiments were also performed with the human D145N and H268N UNG single mutants. In both cases, similar weak binding affinities were also observed (Table 4). This therefore indicates that the inability of UNG to bind rd and -FdUrd tightly is not due to substrate strain because the mutants, which bound the uracil-containing DNA substrates tightly, bound the DNA analogs weakly. The mutants thus appear to be unable to bind the bases of rd and -FdUrd in the same manner in which they bind the uracil base in the HU and HU·G oligonucleotides. Indeed, it is striking that the binding affinities of the analog DNAs closely resemble those observed with the abasic product DNA, and it is possible that the DNA binding with rd and -FdUrd is due to nonspecific interactions with the DNA backbone.
We also extended our previous work on DNA binding by HSV-1 UNG to include the substrate analogs and thus further explore this issue as well as the relatedness of the human and HSV-1 UNG enzymes. The equivalent active-site mutations were investigated with the substrate analog DNAs rd and -FdUrd. In all cases, HSV-1 UNG exhibited exactly the same behavior as its human counterpart (Table 4): tight binding of mutant enzymes to uracil-containing DNA and weak binding of wild-type and mutant enzymes to product DNA and both substrate analogs rd and -FdUrd.
Energetic Analysis of DNA Binding—The equilibrium dissociation constants determined above can be considered in energetic terms because the dissociation constant provides a measure of the Gibbs free energy for the binding reaction according to Equation 2,
 |
The binding of human D145N/H268N UNG to the single-stranded HU oligonucleotide has a free energy of -10.6 kcal mol-1. The equivalent abasic oligonucleotide has a 
G of -7.4 kcal mol-1. The absence of the uracil base thus leads to a G of -3.2 kcal mol-1, which is a significant proportion of the binding energy. A more modest loss in binding energy of -1.2 kcal mol-1 was observed upon going from single- to double-stranded DNA (HU versus HU·G), which is most likely due to the distortions that the enzyme must exert on the DNA. However, with the abasic DNAs, there was no loss of binding energy between single- and double-stranded forms (HAP versus HAP·G), suggesting that the DNA is less distorted in the enzyme-product complex. These values are remarkably consistent across both the human and HSV-1 UNG enzymes.
Further analysis of binding energies between the wild-type and single and double mutant forms of the human and HSV-1 UNG enzymes with rd and -FdUrd analogs did not reveal any significant coupling of binding energy between these two mutations: for hUNG with rd, the coupling energy (G) between the D145N and H268N mutations is +0.62 kcal mol-1. The absence of any negative coupling energy implies that these catalytic residues do not cooperate in straining the substrate prior to catalysis. Alternatively, UNG may not be able to fully flip these base analogs into the active site, so they may not be able to interact with the catalytic residues.
Structural Comparison of the HSV-1 Wild-type and D88N/H210N UNG Enzymes—The DNA binding data presented above indicate that the mutant enzymes bind uracil-containing DNA tightly, whereas abasic product DNA is bound weakly. To further explore the mechanism of base recognition by mutant UNG compared with wild-type UNG, we used x-ray crystallography to investigate these interactions at the atomic level.
The crystal structure of free HSV-1 D88N/H210N UNG was refined to 2.1-Å resolution. The structure lacks electron density for the first 16 amino acid residues, and these residues are presumed to be disordered in the crystal. This phenomenon is also observed in the HSV-1 wild-type UNG crystal (5). Superimposition of the mutant structure with that of HSV-1 wild-type UNG indicates that the double mutation of Asp88 and His210 to Asn caused no major changes in either the overall structure or the positions of critical amino acid residues involved in the formation of the uracil-binding pocket (Fig. 6A).
Electron density maps reveal that the His residue in the active site (His210) was successfully mutated to asparagine and adopted the same conformation. Amino acid 88 in D88N/H210N UNG adopted a different orientation in comparison with Asp88 in the wild-type enzyme (Fig. 6A). The Asn88 side chain in D88N/H210N UNG is turned away from the active site, whereas the Asp88 side chain in wild-type UNG is pointed toward the active site. The position of the Asn88 side chain in the D88N/H210N UNG crystal is similar to the Asp88 side chain in the trinucleotide complex crystal (5), where Asp88 is rotated away from deoxythymidine, and this orientation is presumed to be involved in the rejection of thymidine as a substrate (5). Consistent with this, the position of the side chain of Asn88 is most likely a factor causing a decrease in the activity of D88N UNG.
The crystal structure of HSV-1 D88N/H210N UNG soaked with uridine has been solved to 1.8-Å resolution. Electron density maps show that there is electron density corresponding to the N-glycosidic bond, consistent with the lack of catalytic activity observed with this mutant in biochemical assays (Fig. 3). Some disordered density is observed in the region of the deoxyribose. The disordering is most likely due to rotation around the N-glycosidic bond and possibly also to fragmentation of the labile deoxyribose by high intensity ionizing radiation during data collection.
The D88N/H210N UNG-uracil complex structure reveals that the uracil is bound in exactly the same position as in the binding pocket of the wild-type enzyme (Fig. 6B). The uracil contributes a number of hydrogen bonds with amino acid residues in the binding pocket (Fig. 6C). These interactions in HSV-1 D88N/H210N UNG are similar to those in the wild-type enzyme (Fig. 6C). The rotation of Asn88 away from the position of Asp88 in the wild-type structure does not affect the binding of the uracil, nor does the mutation of His210 to Asn. These results clearly demonstrate that HSV-1 D88N/H210N UNG binds uracil in the same manner as the wild-type enzyme.
We also attempted to solve the structure of the rd nucleotide in complex with UNG with the objective of investigating the atomic interactions of this analog with the active site of the enzyme. The structure of hUNG has been previously solved in complex with an oligonucleotide containing rd, and a distorted conformation of the N-glycosidic bond was observed (35). Here, we were concerned primarily with interactions of the base with the active-site pocket of the mutant enzyme and whether the free nucleotide would adopt the same distorted conformation in the D88N/H210N mutant enzyme. However, despite efforts to both soak and co-crystallize the mutant enzyme with rd under a wide range of conditions, we were unable to obtain a complex. We consider that the weak binding of the rd oligonucleotide reflects a poor interaction of the enzyme with the pseudouracil base and that, for this reason, we were unable to obtain a complex of HSV-1 D88N/H210N UNG with rd.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
20 s-1) may reflect the processivity of the enzyme releasing many uracil bases from the same DNA molecule. The Km values obtained from this assay range from 0.1 to 4.5 µM, but vary according to the conditions used and cannot be related to specific sequences due to the nature of the assay. These high Km values have led to the proposition that hUNG acts solely in the post-replicative repair of U·A mismatches (14).
|
In principle, kcat/Km provides a measure of the relative efficiency with which an enzyme discriminates between two substrates present in the same reaction mixture (36). The kcat/Km value for hUNG with U·G mismatches (74 µM-1 s-1) is significantly higher than the value for Xenopus SMUG1 with U·G mismatches (0.04 µM-1 s-1) (37). However, Xenopus SMUG1 has been shown to be rate-limited by its product dissociation from double-stranded DNA, but Xenopus SMUG1 is stimulated by AP endonuclease, the subsequent enzyme in the base excision repair pathway (13, 14). The specificity constant may thus not provide an accurate picture of substrate discrimination in vivo.
A comparison of our Km value for hUNG with U·G mismatches (0.1 µM) with the Km value for Xenopus SMUG1 with U·G mismatches (0.035 µM) shows that there is not a significant difference. Although Xenopus SMUG1 is evidently capable of removing uracil from U·G mismatches from substrates at low concentration (14), our data demonstrate that there will not be stringent discrimination between U·A and U·G substrates by hUNG and Xenopus SMUG1 based on their kinetic parameters.
A recent study demonstrated that Ung-/-/SMUG
mouse embryonic fibroblast cells are more sensitive to 
-radiation compared with cell lines that carry just the individual mutations (38). This clearly indicates degeneracy in the activity of these two enzymes against a lethal lesion. The same study also showed that C-to-T mutation frequency is largely, although not completely, additive. The C-to-T mutation frequency in Ung-/- cells relative to SMUG cells indicates that UNG is involved in the repair of U·G mismatches in vivo, which might be due to an increase in deamination during DNA replication. The largely additive response suggests some specialization for the two enzymes, possibly through compartmentalization and control (38).
There is evidence for the control of hUNG from its localization at replication foci (39) and controlled degradation during S phase (40). However, the high catalytic turnover of hUNG means that, even at barely detectable levels, it may still be a significant anti-mutator, and there is evidently no survival disadvantage from degeneracy in DNA repair systems. The observed control may thus reflect the requirement to significantly up-regulate repair during DNA replication. Although it is clear that hUNG has a role in the post-replicative repair of uracil (39-41), our data support the hypothesis that hUNG will operate on both U·A and U·G lesions in vivo and will thus have some overlapping activity with SMUG1 against U·G mismatches and other lesions.
Base Excision by UNG Is Uncoupled from Abasic Site Recognition— DNA binding assays demonstrated that the catalytically inactive human D145N/H268N and HSV-1 D88N/H210N UNG double mutants were able to bind uracil-containing DNA tightly (Table 4). The crystal structure of HSV-1 D88N/H210N UNG in complex with uracil shows that the base is bound in the active site of the enzyme in exactly the same way as it is bound in the wild-type enzyme (Fig. 6, B and C). The specific interactions between the base and the active-site amino acids remain unaltered. We therefore conclude that the active-site mutations do not interfere with substrate binding, but solely with the catalytic mechanism.
In contrast to the tight binding to the uracil-containing DNA, both the human and HSV-1 wild-type and mutant UNG enzymes exhibited 100-fold weaker binding to the abasic product DNA (Table 4). These results contradict those previously obtained with hUNG predicting abasic site protection and coupling to the subsequent base excision repair machinery (9). This previous study used surface plasmon resonance to measure DNA binding, whereas we employed a solution-based technique that is free of many of the deviations that can arise with surface plasmon resonance. Furthermore, our results (Table 4) are completely consistent with our previous data obtained with UNG from HSV-1 (10) as well as with the observed lack of AP site inhibition with hUNG (13, 42).
The crystal structures of UNG in complex with uracil show that the amino acids in the active site of UNG can form a number of interactions with the bound uracil (Fig. 6C). These interactions are likely to be responsible for the tight binding of UNG to the uracil-containing DNA. Consistent with this, the lack of these specific interactions in abasic product DNA gives rise to the low affinity of UNG for AP sites described above.
UNG activity has been shown to increase upon addition of human AP endonuclease (HAP-1), suggesting that HAP-1 promotes AP site release by UNG and couples the steps of base excision repair (9). However, the enhancement of UNG activity increases with increasing concentrations of AP endonuclease up to 50-molar excess, whereas a stoichiometric amount of HAP-1 does not increase activity. Furthermore, the difference in activity is more apparent at later reaction time points. Because UNG is weakly inhibited by abasic DNA in vitro (13, 42), the previously observed enhancement of UNG activity by HAP-1 is consistent with the removal of accumulating AP sites. Our data clearly show that UNG has a low affinity for abasic sites and that the enzyme has a high turnover rate, which is inconsistent with AP site protection by UNG. Although there is a clear appeal to having a coupling of DNA glycosylase and AP endonuclease activities, we can find no data that support such a theory for either human or HSV-1 UNG.
DNA Binding and Base Recognition by UNG—Previous crystallographic studies with hUNG clearly showed the enzyme flipping its target nucleotide out of the DNA double helix and into the enzyme active site (8). Our binding data with both human and HSV-1 UNG mutants demonstrate that it gains a large part of its binding energy from interactions with the uracil base. Furthermore, our crystallographic studies demonstrated that the double mutant enzyme binds the uracil base in exactly the same way as the wild-type enzyme. We therefore conclude that the active-site His and Asp residues do not contribute to uracil binding, but have a solely catalytic role. The double mutant enzymes thus provide a sound basis for investigating protein-DNA interactions of UNG enzymes.
An unexpected finding of this study was the weak binding of both non-cleavable substrate analogs that we tested: drd and -FdUrd. The drd analog has been previously used to obtain the structure of an enzyme-substrate complex (35). In this structure, hUNG is bound to DNA, with the drd flipped out of the DNA double helix into the active site. The ring of drd is rotated 90° on the glycosidic bond, and C-1 adopts a tetrahedral conformation. This is a highly unusual nucleotide conformation and led to the proposition that UNG exerts a physical strain on the substrate, which leads to enhanced electron orbital overlap, thus facilitating catalysis (35).
An alternative explanation for the conformation of drd observed in the complex with UNG is that a tautomerization has taken place, which leads to sp3-hybridized C-1. It is noteworthy that N-5 of drd becomes significantly more acidic than N-1 in uracil (43), and this may provide a mechanism for protonation of C-1. This is an energetically unfavorable process because it would lead to the loss of aromaticity. However, the crystals with the drd oligonucleotides took 2 months to grow compared with 1 week for crystals with native DNA (6, 8). It is thus possible either that this unfavorable tautomer is slowly produced and then captured by UNG or that the tautomerization is relatively facile and is the reason for the poor binding that we observed. Our results demonstrate that caution needs to be used when interpreting experiments with both drd and -FdUrd because UNG clearly does not behave in the same way with these base analogs as it does with native uridine.
Conclusion—We have performed a kinetic analysis of both human and HSV-1 UNG enzymes using oligonucleotide substrates to monitor both DNA binding and catalytic activity. The striking feature of this analysis is the similarity of the two enzymes. hUNG has a notably higher kcat/Km, suggesting that it is evolutionarily better adapted to removing minor bases from a large genome. In all other respects, these two enzymes possess a remarkable similarity in function given their 39% sequence identity. Considering also the great similarity of their structures, it seems that the viral enzyme would be a difficult target for specific therapeutics.
The kinetic parameters that we determined for hUNG also have significant implications for the proposed function of this enzyme. We have demonstrated that hUNG has a low Km and a high kcat/Km for both U·A and U·G substrates. We therefore conclude that it is unlikely to be a specific U·A glycosylase in vivo, but that it will most likely have overlapping U·G activity with the SMUG1 DNA glycosylase. Furthermore, the high Kd that we observed for hUNG with abasic DNA suggests that its activity will be uncoupled from the subsequent common steps of base excision repair initiated by the AP endonuclease.
|
FOOTNOTES |
|---|
* This work was supported in part by the Biotechnology and Biological Sciences Research Council. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 
1 Supported by a studentship from the Royal Thai Government.
2 To whom correspondence should be addressed: Div. of Molecular Biosciences, Imperial College London, Sir Alexander Fleming Bldg., South Kensington, London SW7 2AZ, UK. Tel.: 44-20-7594-5228; Fax: 44-20-7584-2056; E-mail: g.baldwin{at}imperial.ac.uk.
3 The abbreviations used are: UNG, uracil-DNA glycosylase; AP, apyrimidinic; HSV-1, herpes simplex virus type 1; h, human; drd, 2'-deoxypseudouridine; -FdUrd, 2'--fluoro-2'-deoxyuridine; 2-AP, 2-aminopurine.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
), and U·A (
). The data are shown with the best fit to the Michaelis-Menten equation with the following values: 1U(—), kcat = 13.9 ± 0.7 s-1 and Km = 0.22 ± 0.04 µM; 1U·G (- - -), kcat = 8.0 ± 0.3 s-1 and Km = 0.12 ± 0.02 µM; and 1U·A(····), kcat = 7.4 ± 0.4 s-1 and Km = 0.1 ± 0.02 µM.
