Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Nucleoside analogs AZT and D4T are powerful anti-HIV drugs that are now generally used in combination with 3TC (-L-2',3'-dideoxy-3'-thiocytidine) and either an antiprotease or a non-nucleosidic reverse transcriptase inhibitor in tri-therapy protocols (Foudraine et al., 1998). The pharmacological target of AZT and D4T is the HIV reverse transcriptase because both analogs act as DNA chain terminators due to the lack of 3'-OH in their ribose moiety. However, to exert their antiviral activity, the nucleoside analogs must be activated into triphosphate derivatives by cellular kinases. Intracellular accumulation of mono or diphospho compounds due to kinetically limiting steps in this phosphorylation pathway could be responsible for some of the toxic effects of the drugs, in particular at the mitochondrial level (Zhu et al., 1998). The last step in this pathway is catalyzed by NDP kinase. Although this enzyme is generally not specific for the base and accepts deoxyribonucleotides, we showed that the phosphorylation of AZT-DP and ddNDP by NDP kinase is very slow and could constitute a limiting step in the overall phosphorylation pathway (Bourdais et al., 1996; Schneider et al., 1998b).

NDP kinase catalyzes the phosphotransfer from NTP to NDP with the transient phosphorylation of a histidine at the active site (Garces and Cleland, 1969). The X-ray structures of NDP kinases from several species have been determined at high resolution (Williams et al., 1991; Dumas et al., 1992; Chiadmi et al., 1993; Webb et al., 1995), showing that both the subunit fold and structure of the active site are highly conserved. The mode of nucleotide binding to the protein is different from that of most kinases because the base makes no polar interactions with the protein and is at the protein surface. The ribose and phosphate moieties are located deeper inside the NDP kinase active site, forming numerous bonds with a Mg²⁺ ion and with protein side chains. The enzyme-nucleotide complex is also stabilized by a hydrogen bond network between the 3'-OH of the sugar, the -phosphate, and two conserved residues of the active site, Lys16 and Asn119 (Tepper et al., 1994).

All eukaryotic NDP kinases are hexamers made of identical 17-kDa monomers. In humans, in whom five isoforms have been reported, the major forms are NDPK-A and NDPK-B, which are encoded, respectively, by the genes nm23-H1 and nm23-H2 and display 88% amino acid identity. The other members of the family include DR-nm23, involved in apoptosis (Venturelli et al., 1995), the mitochondrial nm23-H4 (Milon et al., 1997), and nm23-H5, found in testis germ cells but devoid of enzymatic activity (Munier et al., 1998). Human A and B enzymes present similar kinetic parameters for natural substrates (Schaertl et al., 1998) as well as for a series of thymidine diphosphate derivatives (Gonin et al., 1999). Both the structure and the kinetic parameters of the NDP kinase from the lower eukaryote Dictyostelium discoideum (Dd-NDPK) are also very similar to those of human NDPK-A or NDPK-B, and this has allowed us to use Dd-NDPK as a reliable model for eukaryotic NDP kinases. The structure of Dd-NDPK was also solved in the presence of ADP and AlF₃ which led to modeling of the transition state and the catalytic mechanism (Xu et al., 1997a).

NDP kinase has a very high turnover with k_cat around 1000 s¹ for "natural" ribo- or deoxyribonucleotides. The fluorescence properties of NDP kinases allowed us to monitor the degree of phosphorylation of the catalytic histidine (Deville- Bonne et al., 1996), and we have shown by steady-state enzymatic assays that AZT-TP and ddTTP are poor substrates for NDP kinase, with a loss in catalytic efficiency in the 10⁴ to 10⁵ range as compared with natural nucleotides (Bourdais et al., 1996). Using stopped-flow experiments, we showed that this loss is due to a dramatic decrease in the phosphate transfer rate between the analog and the enzyme (Schneider et al., 1998b).

AZT and D4T are both thymidine analogs and are presumably phosphorylated by the same cellular kinases. However, their patterns of intracellular phosphorylation are not the same because, in particular, D4T is phosphorylated to its 5'-monophosphate at a level 500-fold lower than AZT is (Balzarini et al., 1989; Gao et al., 1994). In this work we have used pre-steady-state and steady-state experiments to investigate kinetic parameters of human NDPK-A with D4T-DP or D4T-TP as a substrate. We show that D4T-DP is more easily phosphorylated by NDPK-A than is AZT-DP. The use of an inactive mutant of Dd-NDPK allowed us to investigate for the first time the binding of NTPs to NDP kinase in the absence of catalysis. We show that D4T-TP binds to the NDP kinase active site with an affinity similar to that of natural substrates, whereas AZT-TP is a weaker ligand.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. Natural nucleotides were from Roche Molecular Biochemicals. The chemical synthesis of phosphoderivatives of nucleoside analogs was as described previously (Bourdais et al., 1996). The double mutation F64W-H122G in Dd-NDPK was made by site-directed mutagenesis according to the method of Kunkel (1985), with the oligonucleotides 5'-GAAAGACCATGGTTCGGTGGTT-3' and 5'-GAAACATCATCGGCGGTTCTGATTC-3' for the F64W and H122G mutations, respectively. Altered bases as compared with the wild-type sequence are bold underlined. The mutations were verified by DNA sequencing.

Enzyme Purification. Wild-type human NDPK-A was overexpressed in Escherichia coli (BL21-DE3) by using plasmid pJC20 (gift from M. Konrad) after induction for 5 h by 1 mM isopropylthiogalactoside when the culture reached A_{600 nm}  0.5. Cells were collected, washed twice, and resuspended in extraction buffer (50 mM Tris-HCl, pH 8.0, 20 mM KCl, 1 mM EDTA, 5 mM MgCl₂, and 1 mM DTE) containing 1 mM phenylmethylsulfonyl fluoride. The cell extract was diluted 2.5-fold with extraction buffer and centrifuged at 10,000g for 20 min at 4°C. The supernatant was treated by 0.15% Polymin P (BASF) to precipitate most of the nucleic acids. After centrifugation at 10,000g for 15 min, the supernatant was loaded at pH 8.0 onto a POROS HQ column (PerSeptive Biosystems, Cambridge, MA; 2.6 × 12 cm) equilibrated in 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, and 1 mM DTE. The recombinant enzyme was separated from the E. coli NDP kinase by a linear KCl gradient (0.02-1 M). Active fractions measured as in Lascu et al. (1992) were pooled, diluted 2-fold with 100 mM Bis-Tris, pH 6.0 and DTE 1 mM, and loaded onto a ceramic hydroxyapatite column (American International Chemical, 1.6 × 10 cm) equilibrated in 10 mM potassium phosphate, pH 6.5, and 1 mM DTE. NDPK-A was eluted by a linear potassium phosphate gradient (0.01-1 M). After concentration and desalting by ultrafiltration in 50 mM Tris-HCl, pH 7.5, 1 mM DTE, and 20 mM KCl, the enzyme was extensively dialyzed against the same buffer containing 50% glycerol and stored at 20°C. NDPK-A was > 97% pure as judged by SDS-PAGE. Wild-type and mutant Dd-NDPK were overexpressed in E. coli (XL1-Blue) and purified as previously described (Schneider et al., 1998a). Enzyme concentration expressed as 17-kDa subunits was determined either by the method of Bradford (1976) or by using an absorbance coefficient of A₂₈₀ = 1.238, 0.55, and 0.85 for a 1 mg/mL solution of NDPK-A, Dd-NDPK, and the double mutant F64W-H122G, respectively (Gill and Von Hippel, 1989). F64W-H122G mutant NDP kinase had no measurable NDP kinase activity but had a very slight ATPase activity (k_cat/K_M = 450 M¹ s¹).

Fluorometric Binding Studies. The affinity of NDP and NTP derivatives for NDP kinase was determined by following the variation of the intrinsic fluorescence upon nucleotide binding as described in Schneider et al. (1998b). The fluorescence of the F64W-H122G enzyme in T₁ buffer (50 mM Tris-HCl, 75 mM KCl, 5 mM MgCl₂, pH 7.5) was measured at 330 nm with excitation at 310 nm (2-nm excitation slit and 4-nm emission slit). Successive aliquots of nucleotide were added to a 1 µM enzyme solution. The inner filter effect was negligible. Experimental binding curves were fitted to a quadratic equation for ligand-protein curve after correction for dilution. By using the double mutant with NTP, the true NTP concentration was measured in order to take into account the small residual ATPase activity. The correction was less than 2%.

Stopped-Flow Kinetic Experiments. Stopped-flow kinetic experiments were performed with a Hi-Tech DX2 microvolume stopped-flow reaction analyzer equipped with a high intensity xenon lamp as described (Schneider et al., 1998b). The excitation wavelength was 304 nm, with a 2-mm excitation slit and a 320-nm cutoff filter at the emission. Mixing was achieved in less than 2 ms. After mixing NDPK (1 µM) and NTP (10-500 µM) or phosphorylated NDPK and NDP, the intrinsic protein fluorescence was recorded for 10 to 200 s. In each experiment 400 pairs of data were recorded, and the data from three to four identical experiments were averaged and fitted to a number of nonlinear analytical equations using the software provided by Hi-Tech. All curves fitted to single exponentials.

Analysis of Kinetic Results. As previously described (Schneider et al., 1998b), the data were analyzed with the reaction scheme:

(1)

The observed single step could be attributed to the phosphotransfer between the nucleotide and the enzyme in both directions. In both the forward and the reverse reactions, the product concentration remains very low; thus, the product binding can be neglected. The binding steps are expected to be fast in both directions and are not detectable in the time range of the stopped-flow experiments. Saturation could not be obtained with the concentrations of AZT-TP and D4T-TP used here. Therefore, we measured the catalytic efficiency of the enzyme phosphorylation (CE_phos) and the catalytic efficiency of phosphoenzyme dephosphorylation (CE_dephos), which are equivalent to second order constants and allow a reliable comparison of a variety of NDP kinase substrates.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Kinetics of D4T Phosphorylation and Dephosphorylation by NDP Kinase. The intrinsic fluorescence of NDP kinase decreases upon phosphorylation of the catalytic histidine, providing a convenient signal for probing the state of phosphorylation of the enzyme (Schneider et al., 1998b). Figure 1A shows the time course of this fluorescence decrease when human NDPK-A is mixed rapidly with different phosphodonors. Addition of D4T-TP results in an exponential decrease in fluorescence leading to a plateau which corresponds to the steady-state of the phosphorylation reaction. Each recorded fluorescence time course follows a single decay with no evidence of lag, burst, or biphasicity. Whereas the amplitude of the fluorescence change is almost constant, the constant of the exponential representing the rate constant of the phosphorylation reaction is linear with the concentration of the phosphodonor in the range 10 to 500 µM (Fig. 1B).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Pre-steady-state kinetics of phosphotransfer between NTP analogs and NDPK (A and B) and between the phosphorylated NDPK and NDP analogs (C and D). A, kinetics of phosphorylation of NDPK-A by D4T-TP. The enzyme (1 µM, final concentration) in T2 buffer (50 mM Tris-HCl, pH 7.5, 5 mM MgCl2, 75 mM KCl, 1 mM DTE, and 5% glycerol) at 20°C was rapidly mixed with D4T-TP (50, 100, and 200 µM, respectively, for curves a, b, and c). The decrease in fluorescence was monitored with time on the stopped-flow device (exc = 304 nm, emission filter < 320 nm). Each reaction was monitored during 200 s. The data were plotted on the scale 0 to 80 s for clarity. The solid lines represent the best fit of each curve to a monoexponential. B, concentration dependence of the phosphorylation rate constant upon D4T-TP or AZT-TP concentration. The pseudo-first order rate constant for the reaction (kobs) was plotted against analog triphosphate: D4T-TP () and AZT-TP (). The linear fit indicates that data can be analyzed as a second order reaction (see Experimental Procedures). The apparent second order rate constant is 700 M1 s1 for D4T-TP and 75 M1 s1 for AZT-TP. C, kinetics of dephosphorylation of phosphorylated NDPK-A by D4T-DP. Phosphorylated enzyme was prepared and the stoichiometry of phosphorylation was determined as described (Deville-Bonne et al., 1996) with small modifications for human NDPK. NDPK-A was preincubated with a saturating amount of ATP in T2 buffer. The phosphorylated enzyme (1 µM, final concentration) in T2 buffer at 20°C was rapidly mixed with D4T-DP (40, 100, and 150 µM, respectively for curves a, b, and c). The increase in fluorescence was monitored with time on the stopped-flow device. The solid lines represent the best fit of each curve to a monoexponential. D, concentration dependence of the phosphotransfer rate constant on D4T-DP or AZT-DP concentration. The pseudo-first order rate constant for the reaction (kobs) was plotted against analog diphosphate: D4T-DP () and AZT-DP (). The linear fit indicates that data can be analyzed as a second order reaction with an apparent constant of 2600 M1 s1 for D4T-DP and 200 M1 s1 for AZT-DP.

1

1

1

1

1

1

Measure of the Affinity of NDP Kinase for NTP. Up to now, all previous affinity measurements of nucleotides for NDPK have been made with NDP, as well as AZT-DP and ddNDP (Schneider et al., 1998b), because for NTP the -phosphate hydrolysis is too fast. However diphospho-compounds lead to the formation of "dead-end" complexes that are not normally part of the reaction, and it would be important to determine the true affinities of NDPK for its NTP substrates. To adress this question, we combined two previously characterized single mutations in the Dd-NDPK. The H122G mutation results in an inactive protein because it is unable to autophosphorylate. Crystal structure of the complex between the H122G enzyme and ADP... Pi-Mg²⁺ shows that the mutation does not affect the overall structure of the catalytic site (Admiraal et al., 1999). In contrast, the single mutant F64W, where the Phe 64 at the entrance of the active site is substituted to a Trp residue, is fully active. Moreover this Trp constitutes a fluorescent probe that is sensitive to NDP nucleobase stacking (Schneider et al., 1998b). Figure 2 (inset) shows the intrinsic fluorescence of F64W-H122G mutant NDP kinase. Upon excitation at 310 nm, the fluorescence intensity of the mutant protein increases by 50% upon any NTP and NDP addition.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Binding curve of F64W-H122G NDP kinase with D4T-TP. Fluorescence change of F64W-H122G NDPK (1 µM) upon binding of D4T-TP () at 20°C in buffer T1, pH 7.5 (exc = 310 nm, em = 330 nm). The solid line is the best fit of the data to a quadratic saturation curve. It corresponds to a KD of 1.2 ± 0.1 µM for D4T-TP (see also Table 3). Inset: fluorescence emission spectra of F64W-H122G NDPK (1 µM) in the absence and in the presence (dotted line) of 0.5 mM D4T-TP.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

In this paper, we show that the DP and the TP forms of D4T are considerably better substrates for NDP kinase than are AZT derivatives. D4T-TP also has a high affinity toward HIV reverse transcriptase and human DNA polymerase  (Ono et al., 1989), possibly due to the planarity of the glycone cycle resulting in a unique conformation of the nucleotide, which allows the formation of a reaction complex with DNA polymerase and participates in the formation of the transition state (Krayevsky and Watanabe, 1998). In contrast, the presence of an azido group on the sugar moiety may restrict its mobility and prevent a planar conformation from occurring within the catalytic cycle. In agreement with this hypothesis, the deoxyribose is found C2'-endo in AZT-DP (Xu et al., 1997b), whereas it is C3'-endo for natural NDP (Cherfils et al., 1994; Morera et al., 1994, 1995). A precise structural explanation for the different efficiencies of AZT and D4T derivatives as substrates for NDP kinase is presently not available. However, we hypothesize a role of the conserved Lys (K16 in Dd-NDPK, K12 in NDPK-A) that participates in catalysis (Tepper et al., 1994). Indeed, the side chain of this lysine moves by more than 2.6Å upon binding of the bulky azido group (Xu et al., 1997b), and this probably results in the poor phosphorylation of AZT-DP by NDP kinases. The absence of steric hindrance in the D4T analog may leave the lysine in place, resulting in a better phosphorylation efficiency.

The three enzymes involved in the cellular phosphorylation pathway of thymidine derivatives, i.e., TK, thymidylate kinase, and NDP kinase, have very different efficiencies toward D4T and AZT (Balzarini et al., 1989). Whereas the critical steps in the AZT phosphorylation pathway are the reactions catalyzed by thymidylate kinase (Lavie et al., 1997) and NDP kinase (Schneider et al., 1998b), the pharmacological activation of D4T is mostly dependent upon its phosphorylation in D4T-MP (Balzarini et al., 1989). In vitro studies also show that D4T is a poor substrate for purified human TK1 and TK2 (Munch-Petersen et al., 1991). By using recombinant enzymes, we showed that, although D4T is a poor substrate for TK from E. coli, D4T-MP is a good substrate for E. coli thymidylate kinase (not shown). We show here that the phosphorylation of D4T-DP by NDP kinase is 10 times more efficient than that by AZT or ddNDPs. Our results highlight the potential interest for designing a prodrug to deliver D4T-MP inside cells to bypass the TK-dependent reaction step (Balzarini et al., 1996; Egron et al., 1998). Because D4T is already known to present less toxicity (Kaul et al., 1999) and to elicit less resistance than AZT at the reverse transcriptase level, a prodrug of D4T should be efficiently phosphorylated after intracellular maturation into D4T-MP.