Kinetic Mechanism of Human dUTPase, an Essential Nucleotide Pyrophosphatase Enzyme*

Human dUTPase is essential in controlling relative cellular levels of dTTP/dUTP, both of which can be incorporated into DNA. The nuclear isoform of the enzyme has been proposed as a promising novel target for anticancer chemotherapeutic strategies. The recently determined three-dimensional structure of this protein in complex with an isosteric substrate analogue allowed in-depth structural characterization of the active site. However, fundamental steps of the dUTPase enzymatic cycle have not yet been revealed. This knowledge is indispensable for a functional understanding of the molecular mechanism and can also contribute to the design of potential antagonists. Here we present detailed pre-steady-state and steady-state kinetic investigations using a single tryptophan fluorophore engineered into the active site of human dUTPase. This sensor allowed distinction of the apoenzyme, enzyme-substrate, and enzyme-product complexes. We show that the dUTP hydrolysis cycle consists of at least four distinct enzymatic steps: (i) fast substrate binding, (ii) isomerization of the enzyme-substrate complex into the catalytically competent conformation, (iii) a hydrolysis (chemical) step, and (iv) rapid, nonordered release of the products. Independent quenched-flow experiments indicate that the chemical step is the rate-limiting step of the enzymatic cycle. To follow the reaction in the quenched-flow, we devised a novel method to synthesize γ-32P-labeled dUTP. We also determined by indicator-based rapid kinetic assays that proton release is concomitant with the rate-limiting hydrolysis step. Our results led to a quantitative kinetic model of the human dUTPase catalytic cycle and to direct assessment of relative flexibilities of the C-terminal arm, critical for enzyme activity, in the enzyme-ligand complexes along the reaction pathway.

dUTPase is the unique enzyme that specifically hydrolyzes the ␣-␤ pyrophosphate bond of dUTP to yield dUMP and PP i (1). The enzyme is essential in maintaining DNA integrity in dividing cells (2,3). Its activity is responsible for setting the physiological dUTP/dTTP concentration ratios (1:24) (4), thus preventing high rates of uracil incorporation into newly synthesized DNA. Although uracil in DNA is tolerated to a certain level by the base excision DNA repair mechanisms, higher levels of uracil in DNA trigger double-strand breaks and lead to cell death (5). Several lines of evidence show that up-regulated dUTPase is responsible for desensitizing tumors to drugs inhibiting the thymidylate synthase pathway, thus acting as an important survival factor for tumor cells (6,7). Increased levels of the nuclear isoform of the enzyme correlate to worsened prognosis of several tumors, as revealed by detailed analysis of tissue samples (8,9). dUTPase has therefore emerged as a high potential anticancer drug target, which possesses several additional, possibly advantageous features for drug design. Unlike most nucleotide-metabolizing enzymes, dUTPase is extremely specific to its substrate nucleotide, potentially allowing construction of substrate analogue antagonists with similarly high specificity. The nuclear isoform of the enzyme is under strict cell cycle control; its expression is mostly limited to rapidly dividing (including cancer) cells (10,11). In addition to the fact that the enzyme is an important focus in biomedical research, dUTPase also serves as a model system for detailed analysis of enzyme-catalyzed nucleotide pyrophosphorolysis.
Current knowledge of the dUTPase mechanism is mainly based on three-dimensional structural approaches. Most dUTPases are homotrimers with a unique active site architecture, where all three monomers contribute to each of the three catalytic sites. High resolution crystal structures of the human (hDUT) 3 (12,13) and other (14 -18) dUTPases provided important mechanistic clues. The catalytic site is formed by five conserved motifs, four of which are contributed by two adjacent monomers. The fifth motif, positioned on the C-terminal arm, is usually provided by the third monomer. The C terminus, associated with an increased conformational freedom, was suggested to close upon the active site during the chemical step (12,14,19). Cleavage of the ␣-␤ pyrophosphate linkage is initiated by a nucleophilic attack from the catalytic water molecule coordinated by a conserved aspartate (Asp 102 in the human enzyme) within the third motif accommodating the uracil and deoxyribose moieties of dUTP (16).
Modern pharmacology demands knowledge of the precise mechanism of action of target enzymes. However, structural data have not yet been complemented by detailed solution kinetic studies for any eukaryotic dUTPase, possibly due to the lack of suitable optical signals reporting enzymatic events. Relying on proton escape during nucleotide pyrophosphorolysis, pH indicator-based assays were used to continuously follow dUTP hydrolysis (20,21), but these methods are transparent to conformational changes of the enzyme. In this study, we took advantage of an intrinsic tryptophan sensor that we had recently engineered in the C-terminal arm of hDUT (Trp 158 ) (13) to resolve the fundamental steps of the enzymatic cycle using fast kinetic methods. Trp 158 replaces a conserved phenylalanine residue that interacts with the uracil ring of dUTP ( Fig.  1C) (12,13). The mutational replacement of the benzene ring with an indole moiety did not perturb the enzyme activity (13). In the present study, the active site Trp 158 sensor also allowed assessment of the proposed structural ordering of the C-terminal arm in the distinct enzyme-ligand complexes, relevant for the reaction cycle. Furthermore, we have developed a protocol for quenchedflow analysis, which is the first to allow the direct monitoring of the hydrolysis step of a dUTPase. We unambiguously show that the chemical step is rate-limiting and that the C-terminal arm is predominantly ordered in all enzymatic states.
Enzyme Activity-Enzyme activity was measured in steadystate pH indicator-based assays as described in Ref. 20 and was typically found to be 6 Ϯ 2 s Ϫ1 . Active site titration was used to determine K M and also to evaluate the active fraction of hDUT and hDUT W158 preparations. In the absorbance stopped-flow setup, an assay buffer containing 100 M phenol red indicator and 1 mM HEPES, pH 7.5, provided optimal monitoring of dUTP hydrolysis. To avoid mixing artifacts, the enzyme was dialyzed in this assay buffer prior to active site titration. Measured time courses (cf. Fig. 3C) were subjected to global fit analysis using GEPASI (24). The floated parameters were k 1 , k Ϫ1 , k 2 , and [E] of the Michaelis-Menten scheme, where K M ϭ (k Ϫ1 ϩ k 2 )/k 1 and k 2 ϭ k cat .
The inactive protein fraction in the measured hDUT or hDUT W158 preparation was only in the range of the uncertainty of protein concentration determination (5-10%).
Fluorescence Spectra and Intensity Titrations-Fluorescence spectra and intensity titrations were recorded on a Jobin Yvon Spex Fluoromax-3 spectrofluorometer with excitation at 297 nm (slit 1 nm), emission between 320 and 400 nm (slit 5 nm), or at 347 nm. Because large concentrations of nucleotides were used, care was taken to correct for any additional fluorescence or inner filter effect imposed on the measured intensities by the nucleotide stock solutions.
Acrylamide Quenching-Acrylamide quenching was carried out by the addition of minute volumes of a 5 M acrylamide solution to the enzyme, enzyme-ligand, or N-acetyl-L-tryptophanamide (NATA) solutions. Raw data were corrected for the fluorescence arising from the acrylamide solution itself. F 0 /F versus [Q] curves were analyzed using a modified Stern-Volmer equation (Equation 1), where F 0 is the unquenched and F is the quenched fluorescence; Q is the quencher; K SV is the dynamic (bimolecular) quenching constant; and V is the static (sphere of action) component of quenching (cf. Ref. 25).
Fluorescence Anisotropy-Fluorescence anisotropy was measured by the single-channel method in an Edinburgh Instruments FLS920P spectrofluorometer equipped with Glan-Thompson prism polarizers. Tryptophan emission spectra ( ex ϭ 295 nm, em ϭ 320 -400 nm) were recorded at four different polarizer configurations (VV, VH, HV, and HH, where V and H denote vertical and horizontal polarizer configurations, respectively, the first letter being designated to the excitation, the second to the emission polarizer). After base-line correction, anisotropy was calculated for the entire spectrum using Equation 2, where r is anisotropy, I is fluorescence intensity, and G ϭ I HV / I HH is a wavelength-dependent parameter of the instrument setup.
Stopped-flow Experiments-Measurements were done using either an SF-2004 (KinTek Corp., Austin, TX) or a SFM-300 (Bio-Logic SAS) stopped-flow apparatus. Tryptophan fluorescence was excited at 297 nm, and emission was selected with a band-pass filter having a peak in transmittance at 340 nm. Time courses were analyzed using the curve fitting software provided with the stopped-flow apparatus or by Origin 7.5 (OriginLab Corp., Northampton, MA).
[␥-32 P]dUTP Synthesis-All synthesis reactions were carried out in a buffer containing 25 mM Tris, pH 7.4, and 100 mM NaCl. Autophosphorylation of 20 M nucleoside diphosphate kinase (NDPK; from yeast; catalog number N0379; Sigma) was carried out in 5 mM EDTA at 30°C for 10 min in a final volume of 100 l using 20 M [␥-32 P]ATP, according to Ref. 26. To remove ADP and [␥-32 P]ATP from the reaction in a quick manner, we applied batch adsorption on anion exchanger resin in an expectation that the resin will only remove the negatively charged nucleotides and will not bind NDPK. The NDPK isoenzyme used in the experiment has a calculated pI of 8.65 and therefore carries a net positive charge at pH 7.4. According to the expectation, 25 l of washed Q-Sepharose (Amersham Biosciences) added to the autophosphorylation reaction mixture immobilized all nucleotides without binding NDPK. The Q-Sepharose beads were then removed from the reaction by a 30-s centrifugation step. Subsequently, 25 M dUDP and 10 mM Mg 2ϩ were added to [ 32 P]NDPK to yield [␥-32 P]dUTP (incubation for 10 min at 30°C). The enzyme was then completely removed from the [␥-32 P]dUTP-containing solution by phenol extraction that was carried out according to Ref. 27. For the analysis of the synthesis products (see "Results"), the radioactive nucleotide and phosphate contents were separated from each other using charcoal adsorption (as in Ref. 28). The advantage of using charcoal is that it binds all nucleotides but not P i and PP i . Radioactivity was counted in water in a Wallac 1409 liquid scintillation counter. We used a [␥-32 P]ATP stock solution of high specific activity (0.4 MBq/l, 111 GBq/ mol) to obtain a similarly high specific activity [␥-32 P]dUTP sample suitable for tracing. The synthesized [␥-32 P]dUTP was added to a large molar excess of nonlabeled bulk dUTP in a 1:100 volume ratio. For further details of the analysis of the synthesis products, see "Results." To capture the dUTP-bound cycling steady state, a high excess of dUTP was used, and the spectrum was recorded within 30 s after dUTP was added. For analysis of fluorescence spectral parameters, see Table 1. B, fluorescence equilibrium titration of hDUT W158 with its ligands. ex ϭ 295 nm, em ϭ 347 nm; solid lines represent quadratic fits to the data except for the dUMP⅐PP i curve, where a Hill equation with n ϭ 1.7 provided a better fit. K d values from the presented fits are as follows: 31 Ϯ 4 M for dUMP (triangles), 12 Ϯ 2 M for dUDP (circles), 1.9 Ϯ 0.2 M for dUPNPP (diamonds) (in the inset, 146 Ϯ 15 M for PP i (stars) and 494 Ϯ 25 M for dUMP⅐PP i (crosses)). The dUMP⅐PP i titration was carried out by titrating dUMP-saturated enzyme (in 500 M dUMP) with PP i . C, three-dimensional structure of hDUT in complex with dUPNPP (figure produced using Protein Data Bank code 2HQU (13) and PyMOL). The three monomers (A-C) are represented by color-coded schematic diagrams. One of the three active sites is shown with the bound dUPNPP (stick model; yellow carbons and otherwise atomic coloring). The Phe 158 residue (monomer C) stacks over the uracil ring. Other coordinating residues from monomers A and B and the catalytic water molecule (red sphere, labeled W cat ) are shown for orientation purposes.

dUTPase Catalysis Reported by an Intrinsic Tryptophan Sensor
Quenched-flow Experiments-Quenched-flow experiments were carried out using the RQF-3 (KinTek Corp., Austin, TX) quenched-flow apparatus. 2 M HCl ( 2 ⁄ 3 M in the reaction) was used as the chemical quencher of the dUTPase reaction. Hydrolysis products were separated according to Ref. 28. The amount of the resulting 32 PP i product was counted in water using a Wallac 1409 liquid scintillation counter (PerkinElmer Life Sciences).
Data Analysis and Numerical Simulations-Data analysis and numerical simulations were done using Origin 7.5 (Origin-Lab Corp., Northampton, MA) or the freely available GEPASI 3 biochemical kinetics simulation software (24), respectively.

Fluorescence Spectral Properties of hDUT W158 and Its
Ligand-bound Complexes-Recently, fluorescence emission from the Trp 158 fluorophore was shown to be significantly and characteristically quenched in dUTPase-dUPNPP and dUTPase-dUMP complexes as compared with the apoenzyme (13), in agreement with the expectation that the stacking between conserved residue Phe 158 and the substrate uracil ring (cf. Fig. 1C) is also present in the Trp 158 mutant enzyme. Following these observations, we quantified maximal fluorescence changes and spectral shifts of Trp 158 upon binding to physiological ligands and to the nonhydrolyzable substrate analogue dUPNPP ( Fig. 1A and Table 1). These data yield information on the interaction of Trp 158 with the uracil moiety of any bound nucleotide and will allow interpreting the fluorescence-based kinetic experiments. The Trp 158 fluorescence emission maximum of the apoenzyme was at 353 nm, a typical value for a nonburied protein tryptophan ( max, NATA ϭ 355 nm) (25) (Fig.  1A). Fig. 1A also shows that the binding of different uracil nucleotides but not that of PP i to hDUT W158 quenches Trp 158 fluorescence, probably due to aromatic stacking between the indole and uracil rings (shortest distances between atoms of the uracil moiety and those of the Phe 158 benzene ring in hDUT are 3.4 -3.7 Å, as determined in the crystal structure of the enzyme-dUPNPP complex, Protein Data Bank code 2HQU) (13) (Fig. 1C). The magnitude of the nucleotide-induced quench and blue shift increased in the order dUMP 3 dUDP 3 dUPNPP (Table 1). This implies that the presence of the ␤and ␥-phosphates causes the C-terminal arm to form more interactions with the phosphate chain of the substrate (in agreement with the structural description (13)), whereby the arm may become less flexible and may stabilize the stacking interaction between Trp 158 and the uracil ring. Inter-estingly, Trp 158 fluorescence was even more quenched during steady-state dUTPase cycling than in any of the other ligandbound states (Fig. 1A). This suggests that there is at least one major steady-state intermediate that cannot be produced by the addition of the above ligands (e.g. the prehydrolysis mimic dUPNPP or the posthydrolysis mimic dUMP⅐PP i states). A possible explanation for this finding is that a particular protein conformational change occurs in the presence of dUTP (but not in the presence of dUPNPP or other nucleotides), leading to the hydrolysis-competent state (see below).
The large fluorescence increase in the presence of PP i indicates that the binding of this ligand also causes a conformational change in the active site. Control experiments conducted with bovine serum albumin and NATA (data not shown) ascertained that the effect of PP i on our tryptophan sensor was specific. Interestingly, a rather similar phenomenon was observed in an earlier study in which a tryptophan engineered into the entrance of the nucleotide binding site of myosin (Trp 129 , in close proximity to the adenine moiety of ATP) exhibited a large quench on nucleotide binding and a large fluorescence increase on PP i binding (29). We probed the potential interaction of hDUT W158 with phosphate (P i ) (used at high excess), but no signal change was detected.
Fluorescence Intensity Titrations to Determine Enzyme-Ligand Dissociation Constants- Fig. 1B shows fluorescence intensity titrations of 4 M hDUT W158 with various ligands (enzyme active site concentrations are used throughout this paper). Dissociation constants are given in Table 1. K d values illustrate that the affinity of hDUT W158 increases in the order dUMP 3 dUDP 3 dUPNPP, with dUDP and dUPNPP binding being 3 and 10 times stronger than that of dUMP, respectively. dUTP binding cannot be measured using this equilibrium method, but we anticipate that its K d value may be equal to or lower than that of dUPNPP (K d ϳ 1 M). Nucleotide-free hDUT W158 exhibited a K d for PP i of 146 M. The dissociation constant of PP i for the ternary enzyme products complex (E⅐dUMP⅐PP i ) was about 3 times larger than that for E⅐PP i . This moderate antagonistic effect between the binding of dUMP and PP i to the enzyme is probably due to the repulsion between the negative charges of dUMP and PP i .
Acrylamide Quenching-We have performed acrylamide quenching experiments with hDUT W158 to monitor the solvent accessibility of the Trp 158 reporter ( Fig. 2A, Table 1). For reference and control, we also measured properties of NATA, a model compound for a rotationally free and maximally solvent- ), Trp 158 exhibits markedly reduced solvent accessibility (K SV ϭ 6.6 M Ϫ1 ) even in the apoenzyme. Such a low K SV was unexpected, considering that Trp 158 is situated in the C-terminal arm of hDUT W158 , six residues away from the terminal amino acid. On the other hand, differences between K SV values of various ligand-bound states of hDUT W158 are relatively small but significant (K SV , ranging from 5.0 to 6.8 M Ϫ1 ) ( Fig. 2A and Table 1). This finding suggests that large conformational changes of the C-terminal arm upon ligand binding are unlikely to occur. The solvent accessibility of Trp 158 decreases in the order dUMP 3 dUDP 3 dUPNPP, suggesting a gradual movement of the C-terminal arm toward the nucleotide. This observation is in line with our experiments shown in Fig. 1. Importantly, the solvent accessibilities of the E⅐dUMP⅐PP i and E⅐dUMP states were very similar, indicating that dUMP but not PP i induces shielding of the active site. PP i binding alone does not perturb the solvent accessibility of Trp 158 , probably due to a relatively open active site conformation (Fig. 1).
Fluorescence Anisotropy-Fluorescence anisotropy is routinely used to describe the dynamic properties of a protein environment. Freely rotating small fluorophores are depolarized at room temperature and therefore exhibit anisotropies close to zero (cf. NATA in Fig.  2B and Table 1). We measured the steady-state anisotropies of apo-hDUT W158 and its ligand-bound complexes to gain further insights into the dynamic behavior of the C-terminal arm in various enzymatic states. The steady-state anisotropy of apo-hDUT W158 (r ϭ 0.077) increased upon ligand binding, which reflects a steric hindrance of the fluorophore. Similarly to the previously described experiments (Figs. 1 and 2A), a correlation of the measured effect to the length of the phosphate chain of the nucleotide was observed (i.e. the value of r increased in the order apo 3 dUMP 3 dUDP 3 dUPNPP) (Fig.  2B). The largest increase in anisotropy was detected in the PP i -  bound species (E⅐PP i and E⅐dUMP⅐PP i ), although we previously showed that these are the most "open" and solvent-accessible enzyme states ( Fig. 2A). Taken together, the anisotropy data indicate that (i) ligand binding to the polyphosphate binding site causes structural ordering of the C-terminal arm, proportionally to the length of the polyphosphate chain (without shielding Trp 158 from the solvent); (ii) the lower anisotropy of uracil nucleotide-bound states compared with that of the E⅐PP i state show that aromatic stacking to uracil slightly depolarizes Trp 158 (Fig. 2B and Table 1).
Rapid Kinetics of hDUT W158 Followed by Intrinsic (Trp 158 ) and Extrinsic (Proton Release) Signals-In the knowledge of the fluorescence characteristics of individual enzyme-substrate (substrate analogue) and enzyme-product complexes (Fig. 1), progress curves obtained by monitoring Trp 158 fluorescence during the interaction of hDUT W158 with dUTP in the stopped-flow yielded significantly more information than pH detectionbased (proton release) methods. Fig. 3 shows single and multiple dUTP turnovers obtained using Trp 158 fluorescence (A and B) or proton release (C) signals. Trp 158 fluorescence traces of single dUTP turnovers ([E] Ͼ [S]) consisted of three exponential phases (Fig. 3A). A fast initial quench in fluorescence (k obs ϳ 900 s Ϫ1 ) was followed by an additional slower decrease (k obs ϳ 20 s Ϫ1 ), and then the fluorescence intensity returned to a closeto-initial value with a k obs of 6.8 Ϯ 2.0 s Ϫ1 . In light of the steadystate fluorescence data of Fig. 1A, we interpret the first fast phase as the initial binding of the nucleotide in which Trp 158 quenching occurs by stacking over the uracil ring. Considering the difference between the fluorescence intensity of the enzyme-dUPNPP complex and that during steady-state dUT-Pase cycling, the second slower phase can be interpreted as a dUTP-induced structural change that precedes or is concomi- ). Measurements carried out using near-equimolar concentrations of enzyme and substrate indicated that the time course of this phase does depend on concentration (k obs values of force-fitted exponentials were 400 -1200 s Ϫ1 in the applied 2.5-15 M concentration range). Numerical simulations in which this phase was assigned to a second-order binding step showed good agreement with the experimental traces, and the fundamental rate constants could be extracted (cf. Fig. 6A and Table 2). We did not observe systematic concentration dependence of the k obs (termed k ISO,obs in Table 2) of the second exponential phase (20 Ϯ 18 s Ϫ1 ), which confirms the first order nature of this proposed isomerization step (Fig. 3B). The k obs value of the third phase did not exhibit concentration dependence in the single turnover concentration regime. The k obs of this phase was in good agreement with the previously determined steady-state k cat of hDUT (8 Ϯ 3 s Ϫ1 ) (13), indicating that it represents the rate-limiting step of the dUTPase cycle. Furthermore, the duration of the steady state (t ss ) in multiple turnover Trp 158 fluorescence traces (i.e. the time elapsed between the start of the reaction and the inflection point of the fluorescence restoration phase) (Fig. 3B) was consistent with the above third phase k obs and steady-state k cat values (t ss Ϸ [S] initial /([E] total k cat ), if [S] initial Ͼ Ͼ K M ). Fig. 3C shows single and multiple turnovers detected by a proton release assay in an absorbance stopped-flow setup. The amplitude of the curves was directly proportional to the initial substrate (and thus the released proton) concentration. In single turnover conditions ([E] Ͼ [S], lower two curves in Fig. 3C), the time courses corresponded to single exponentials, and k obs values (6.5 Ϯ 0.1 s Ϫ1 ) were identical to the steadystate k cat of the enzyme ( Table 2). In multiple turnovers (upper two curves in Fig. 3C) a linear steadystate phase was observed without any burst of proton release. These profiles altogether imply that the enzymatic cycle is limited by a single rate-limiting step that occurs before   Table 1. B, kinetic model of the hDUT enzymatic cycle. Daggers and stars indicate fluorescence decrease or increase compared with the apoenzyme, respectively. The rate constants shown in the model were used as parameters of the kinetic simulation (A) and are compiled in Table 2. For the k ϪPM /k PM and k ϪM /k M rate constant pairs, only the ratios (defined by K d values of Tables 1 and 2) and the lower bounds for the rate constant pairs are known. These lower bounds were used in the numerical simulations as shown. Increases in the values of these rate constants (while keeping their respective ratios constant) did not cause any detectable change in the enzyme mechanism.
or is concomitant with proton release. We could model these proton release events with Michaelis-Menten kinetics in which a rapid equilibrium (k 1 , k Ϫ1 ) precedes the rate-limiting step (k 2 ϭ k cat ). Global fits to the single and multiple turnover time courses using the k 1 , k Ϫ1 , and k 2 floating parameters of Scheme 1 yielded K M ϭ 3.6 Ϯ 1.9 M, k cat ϭ 6.7 Ϯ 0.2 s Ϫ1 , k cat /K M ϳ 1.9 ϫ 10 6 M Ϫ1 s Ϫ1 , for both hDUT and hDUT W158 proteins.
[␥-32 P]dUTP Synthesis-[␥-32 P]dUTP is not commercially available. We therefore developed a straightforward synthesis method (Fig. 4A) using NDPK that converts [␥-32 P]ATP and dUDP into [␥-32 P]dUTP and ADP by a ping-pong mechanism (26,30). We took advantage of the fact that the phosphorylated enzyme intermediate of the NDPK reaction is long lived in the absence of Mg 2ϩ and thus can be separated from the phosphate donor nucleotides (26). The resulting synthesis product (after step 4 in Fig. 4A) contains [␥-32 P]dUTP, dUDP, and inorganic phosphate. To test for the presence of any non-dUTP-derived radiolabeled species that would compromise radiochemical purity, aliquots of the synthesis product were fully hydrolyzed by (i) dUTPase (extremely specific for dUTP), (ii) dUTPase ϩ myosin (hydrolyzes NTPs (31)), or (iii) apyrase (hydrolyzes (d)NTPs and (d)NDPs (32)). All three enzyme conditions resulted in liberation of the same 32 P i content of the total radioactive material, demonstrating that practically all hydrolyzable radioactive nucleotide species in the synthesis product was [␥-32 P]dUTP. The synthesis product contained 15 Ϯ 3% nonnucleotide 32 P i (measured in samples from which all nucleotides had been removed). Analysis showed that this fraction originated from (i) carryover from the original [␥-32 P]ATP solution (5%), (ii) spontaneous hydrolysis of ␥-32 P-labeled nucleotides during the four-step procedure, and possibly (iii) slow 32 P i release from the phosphorylated NDPK in the absence of phosphate acceptor (during step 2). The ϳ15% 32 P i in the synthesis product does not reflect the P i content of the bulk solution to be used in quenched-flow experiments, because a subsequent large dilution of the synthesis product in nonla-beled dUTP decreased the P i /dUTP concentration ratio to less than 1:10,000 in the reagent solution used in the quenched-flow assay. The only noticeable effect of the condition that ϳ15% of the total radioactivity was radioactive 32 P i was the reduction of the maximal expected signal change from 100 to 85%, which did not impede the evaluation of quenched-flow results. Similarly, the dUDP concentration of the synthesis product was drastically reduced by the dilution of [␥-32 P]dUTP in a large molar excess of nonlabeled dUTP (the dUDP/dUTP molar ratio was less than 1:1600 in the assay reagent). In the above described experimental conditions, the most important factor in providing chemical purity was the use of high quality nonlabeled nucleotide to be traced with a high specific activity radioactively labeled one. The total [␥-32 P]dUTP yield was calculated following the analysis of the radioactive constitution of the synthesis product and was found to be 25% (i.e. one-quarter of the [␥-32 P]ATP was converted specifically into [␥-32 P]dUTP). Considering that both reactions 1 and 3 (Fig.  4A) are fully reversible, this yield indicates that the procedure was highly efficient. Fig. 4B shows a single turnover experiment with a single exponential fit to the data points. For both wild-type hDUT and hDUT W158 constructs and depending on the protein preparation, the k H of single turnovers was determined to be 5.5 Ϯ 2.5 s Ϫ1 , in agreement with the k obs values observed in the fluorescent and proton release turnovers ( Table  2). There was no systematic difference between the k H values of hDUT and hDUT W158 . When excess dUTP was mixed with hDUT ( Fig. 4C), we observed a linear steady-state phase without any burst, clearly arguing that the rate-limiting step of the dUTPase enzymatic cycle is identical to (or precedes) the chemical step.

Direct Observation of the Chemical Step by Quenched-flow Using [␥-32 P]dUTP-
Product Release-The large fluorescence intensity change of Trp 158 induced by PP i binding allowed us to follow the dissociation of PP i from the enzyme. We carried out dUTP chase experiments to avoid rebinding of the dissociated PP i . Upon mixing the E⅐PP i complex with excess dUTP in the stoppedflow, double exponential curves were recorded (Fig. 5, upper black trace) having a fast phase of 740 Ϯ 66 s Ϫ1 and a slow phase of 24 Ϯ 6 s Ϫ1 . The amplitude of the fast but not the second slow phase depended upon the concentration of PP i (Fig. 5, inset).
The first phase can therefore be attributed to PP i dissociation, whereas the second phase arises from the isomerization of the ES complex occurring after the initial dUTP binding step (cf. Fig. 3A). The K d value resulting from a one-binding site quadratic fit to the amplitude data (327 Ϯ 117 M) was similar to that obtained from equilibrium titrations (cf. Fig. 1B and Table 1). Product dissociation was measured also from the E⅐dUMP⅐PP i complex (Fig. 5, gray trace). Curves exhibited k obs values (684 Ϯ 84 s Ϫ1 ) similar to those of the E⅐PP i curves, showing that the rate constants of PP i dissociation from E⅐PP i and E⅐dUMP⅐PP i are similar (Table 2). In line with the lower initial fluorescence level of E⅐dUMP⅐PP i compared with that of E⅐PP i (cf. Fig. 1A), the amplitudes of the E⅐dUMP⅐PP i chasing traces were lower than those of the E⅐PP i chasing traces (Fig. 5). Binding and dissociation of dUMP was too fast to observe by stopped-flow. Kinetic Modeling of the hDUT Enzymatic Cycle-The measured accessible parameters of the hDUT enzymatic cycle (Tables 1 and 2) allowed us to propose a model that provided good fits to our experimental data (Fig. 6B). Using this model, kinetic simulations of the hDUT W158 fluorescence profile during dUTP hydrolysis yielded time courses that were very similar to the measured ones (Fig. 6A). Parameters for the binding (k B , k ϪB ), isomerization (k ISO , k ϪISO ) and hydrolysis (k H ) steps were floating parameters, given that these are the events that primarily determine the fluorescence profiles during dUTP turnovers. Kinetic parameters of the product release steps were fixed so that the ratios of the dissociation and association rate constants of the individual steps yield the K d values shown in Table 1 and thus determine the final fluorescence levels. The rate constants of product release are so fast compared with the rate-limiting step that they do not influence the turnover curves.

DISCUSSION
A central aspect of the present study is that the fluorescent signal of a single tryptophan engineered into the C-terminal arm of hDUT (Trp 158 ), which forms part of the active site, allowed precise resolution and characterization of practically all key enzymatic steps. These steps include (i) a rapid, probably diffusion-limited substrate binding, (ii) a subsequent substrate-induced structural change (isomerization) required for the formation of the catalytically competent conformation, (iii) the rate-limiting hydrolysis step, and (iv) rapid, nonordered release of the hydrolysis products ( Fig. 6B and Table 2). The second isomerization step was not foreseen or suggested earlier due to the lack of conformationally sensitive assays to follow the reaction. Importantly, in the present work, two independent lines of evidence argue in favor of the existence of this isomerization step. First, the different extent of quenching and blue shift associated with the enzyme-dUPNPP and enzyme-dUTP (steady-state) complexes (cf. Fig. 1A) indicate the existence of at least two distinct prehydrolysis conformations of the active site. Second, the kinetic analysis of time courses in Figs. 3, A and B, and 6 clearly shows the presence of a second slower exponential component following the initial fast binding of dUTP. An intriguing feature of the mechanism is that two different dUTP-bound intermediates will be significantly populated during steady-state dUTP hydrolysis (E⅐dUTP † † † will be predominant, but about 30% of the enzyme molecules will populate E⅐dUTP † † ). This steadystate distribution results from the k ISO rate constant being in the same order of magnitude as the rate-limiting hydrolysis rate constant (k H ) (Fig. 6B).
We confirmed the rate-limiting nature of the chemical (hydrolysis) step by [␥-32 P]dUTP-based quenched-flow transient kinetic analysis (Fig. 4). To obtain the commercially unavailable [␥-32 P]dUTP, we developed a simple synthesis method based on the ping-pong phosphate transfer mechanism of NDPK (26). The novelty in our synthesis is that isolation of the [ 32 P]NDPK intermediate and the final ␥-32 P-labeled nucleotide product takes place in an Eppendorf tube, requires no instrumentation, and results in a radiochemical purity that is suitable for many applications. Laborious purification of the synthesis products is not necessary, because the donor and acceptor nucleotides are spatially and temporally separated. This straightforward method may be of great help in studying enzymes that use pyrimidine-triphosphates as substrate (e.g. dCTP deaminase, dTTPase, tRNA cytidyltransferase, etc.), since none of these relevant ␥-labeled pyrimidine nucleotides are commercially available (or they may be purchased only as expensive custom synthesis orders). Due to the substrate promiscuity of NDPK, even base-modified nucleotide analogs may be radioactively labeled using this method for additional specific applications.
Interestingly, the estimated intracellular dUTP concentration (ϳ0.7 M (4)) is in the same range as the K M of hDUT for dUTP (Tables 1 and 2). This indicates that dUTPase function is highly sensitive to cellular dUTP fluctuations around the physiological level. Our data show that product inhibition by dUMP at its estimated physiological concentration (ϳ2.7 M (4)) is probably not significant due to its relatively low affinity for hDUT and its rapid release from the enzyme-products complex (Tables 1 and 2).
We probed in silico F158W mutations in available structures and found that the replacement of Phe 158 with a Trp residue does not cause a steric hindrance within the active site. Accordingly, we found that hDUT W158 retains the enzymatic activity of the wild-type enzyme. The presence of an aromatic residue at this location has been suggested to be important for enzyme activity (12), and thus the fact that a Trp residue can functionally replace the native Phe 158 implies that the fluorescence signal reports events that are highly relevant to the physiological activity of the enzyme.
Trp 158 is sensitive to the precise nucleotide (or other ligand) content of the active site (Fig. 1A). The solvent shielding of Trp 158 increases, whereas the structural flexibility of this residue decreases with increasing length of the polyphosphate chain of the nucleotide ligand (Fig. 2). PP i binding into the binding site, however, causes a structural ordering of the arm without a solvent shielding effect. Based on our observation that E⅐PP i exhibits elevated fluorescence compared with the apoenzyme (Fig. 1A), we speculate that PP i binding may cause disruption of a quenching interaction of Trp 158 , supposed to be present in the apo state. A possibility for such a quenching interaction is a cationtype interaction (33) between Trp 158 and the positively charged guanidino moiety of an arginine residing in its close proximity. Candidate arginines are residues 85, 128, and 135, all contributing to the binding and stabilization of the polyphosphate chain of the substrate nucleotide ( Fig. 7) (13). These groups are separated by about 9 -10 Å from Trp 158 via the intercalation of the uracil group in the hDUT-dUPNPP structure. In the apo state, however, one of the candidate arginines might move closer to the phenylalanine (tryptophan) to neutralize the positive charge via cationstacking; hence the intermediate fluorescence level observed in the apo-hDUT W158 .
It is noteworthy that, whereas the binding of dUMP, dUDP, and dUPNPP (and even more that of dUTP) to the enzyme causes marked quenching of Trp 158 as compared with the apo state, the posthydrolysis E⅐dUMP⅐PP i complex has an enhanced Trp 158 fluorescence. We surmise that this fluorescence increase reflects a structural state in which the stacking interaction between Trp 158 and the uracil moiety is at least partially disrupted, aiding the rapid release of products from this posthydrolytic complex.
In addition to its utility in the determination of the kinetic and thermodynamic parameters, the Trp 158 signal also provided much information about the structural dynamics of the C-terminal arm during catalysis. Evidence is presented that this protein segment is at least partially closed upon the active site in all enzymatic states (even including the apoenzyme) (Fig. 2), and therefore its conformational freedom may be well restricted. Fig. 7 shows that the C-terminal arm of monomer A is anchored in a ␤-sheet with the N-terminal residues of monomer C. At its very C terminus, the arm also interacts with monomer B via strong hydrogen bonds (second anchor). Interactions of the arm with the ligand dUPNPP are mainly formed by residues situated between the monomer-monomer interacting regions. This arrangement rationalizes (i) the proposed proximity of the C terminus to the protein core even in the absence of nucleotide (in the apoenzyme) and (ii) that the C-terminal arm still conveys a significant flexibility in the apoenzyme (between the two anchor regions), as suggested in previous studies (12,13).
We described a complex methodology to assess the fundamental steps of dUTP hydrolysis as catalyzed by human dUTPase, an important chemotherapeutic target protein.
The resulting novel insights underline the importance of the dynamic behavior of the C-terminal arm during catalysis and advocate the targeting of this enzyme segment for perturbation of dUTPase function.