Reaction of human UMP-CMP kinase with natural and analog substrates Claudia Pasti 1, * , †, Sarah Gallois-Montbrun 1, †, He ´ le ` ne Munier-Lehmann 2 , Michel Veron 1 , Anne-Marie Gilles 2 and Dominique Deville-Bonne 1 1 Unite ´ de Re ´ gulation Enzymatique des Activite ´ s Cellulaires and 2 Laboratoire de Chimie Structurale des Macromole ´ cules, CNRS URA 2185, Institut Pasteur, Paris, France UMP-CMP kinase catalyses an important step in the phosphorylation of UTP, CTP and dCTP. It is also involved in the necessary phosphorylation by cellular kinases of nucleoside analogs used in antiviral therapies. The reactivity of human UMP-CMP kinase towards natural substrates and nucleotide analogs was reexamined. The expression of the recombinant enzyme and conditions for stability of the enzyme were improved. Substrate inhibition was observed for UMP and CMP at concentrations higher than 0.2 m M , but not for dCMP. The antiviral analog L -3TCMP was found to be an efficient substrate phosphorylated into L -3TCDP by human UMP-CMP kinase. However, in the reverse reaction, the enzyme did not catalyse the addition of the third phosphate to L -3TCDP, which was rather an inhibitor. By molecular modelling, L -3TCMP was built in theactivesiteoftheenzymefromDictyostelium.Human UMP-CMP kinase has a relaxed enantiospecificity for the nucleoside monophosphate acceptor site, but it is restricted to D -nucleotides at the donor site. Keywords: UMP-CMP kinase; antiviral analog; 3TC; AraC; phosphorylation. Nucleoside analogs constitute a familly of important antiviral and anticancer drugs. Analogs of thymidine like AZT (2¢3¢-deoxy-3¢azido thymidine), d4T (2¢3¢-dideoxy-2¢3¢- didehydro-thymidine) have been used to treat HIV infection for a number of years as well as analogs with an L -conformation in the sugar like L -3TC. All of these analogs are delivered as nucleosides to the patients and need to be phosphorylated within cells in order to be active on their viral or cellular target. UMP-CMP kinase catalyses the reversible transfer of phosphate between UMP (or CMP) and ATP in the presence of magnesium ions according to the following reaction: UMP þ ATP-Mg 2þ  ! UDP þ ADP-Mg 2þ The enzyme is involved in both the de novo and the salvage pathway of nucleosides. It catalyses a compulsory step for the synthesis of UTP, CTP and dCTP, that are incorporated into nucleic acids. UMP-CMP kinase is therefore also involved in the activation of nucleoside analogs and its kinetic properties are important for the efficacy and the selectivity of these drugs. The UMP-CMP kinase of several eukaryotic organisms has been studied including yeast [1], the slime mold Dictyostelium discoideum [2,3] and, recently, in man [4]. The UMP-CMP kinase mRNA was ubiqui- tously present in human tissues with highest levels in pancreas, muscle and liver [4]. The enzyme belongs to the large family of eukaryotic nucleoside and NMP kinases for which AMP kinase is a prototype [5,6]. In contrast, in prokaryotes and, in particular in E. coli, UMP kinase and CMP kinase are different proteins sharing no sequence similarity [7,8]. The 3D structure of UMP-CMP kinase from the Saccharomyces cerevisiae [9] and from D. discoid- eum [3] have been solved by X-ray crystallography and compared to the structure of yeast S. cerevisiae adenylate kinase [10]. All NMP kinases are globular proteins with a Ôcore domainÕ made of a five-stranded parallel b-sheet surrounded by eight helices. The protein is made of two subdomains, each containing a nucleotide binding site. The site for the nucleotide donor of phosphate (in general ATP) presents a typical ÔP-loopÕ and a Ôlid domainÕ that closes down over the active site during the catalytic cycle. The protein, which is flexible as several structural changes occur upon ligand binding, moves from an ÔopenÕ to a ÔclosedÕ conformation during the catalytic cycle [10]. UMP-CMP kinase is involved in the phosphorylation of several anticancer and antiviral drugs that are given to patients suffering from AIDS and B hepatitis. Correspondence to D. Deville-Bonne, Institut Pasteur, Unite ´ de Re ´ gulation Enzymatique des Activite ´ s Cellulaires, 25, rue du Dr Roux, 75724 Paris Cedex 15, France. Fax: + 33 1 45 68 83 99, Tel.: + 33 1 40 61 35 35, E-mail: ddeville@pasteur.fr Abbreviations: AraC, cytosine D -arabinofuranoside; AZT, 2¢,3¢-deoxy- 3¢-azido thymidine; GST, glutathione S-transferase; dCK, deoxycyti- dine kinase; L -dC, b- L -2¢-deoxcytidine; d4T, 2¢3¢-dideoxy-2¢3¢- didehydro-thymidine; L -dT, b- L -2¢-deoxythymidine; L -3TC, b- L -2¢,3¢-dideoxy-3¢-thiacytidine; L -3TCMP, b- L -2¢,3¢-dideoxy-3¢-thi- acytidine monophosphate; L -3TCDP, b- L -2¢,3¢-dideoxy-3¢-thiacyti- dine diphosphate; L -3TCTP, b- L -2¢,3¢-dideoxy-3¢-thiacytidine triphosphate; NMP, nucleoside monophosphate; NDP, nucleoside diphosphate; NTP, nucleoside triphosphate. Enzyme: Human UMP-CMP kinase (EC 2.7.4.14). *Present address:UniversitadegliStudidiFerrara,Dip.Biochimicae Biologia Molecolare,Via Borsari, 46, 44100 Ferrara, Italy. Note: Both authors contributed equally to this work. (Received 18 December 2002, revised 19 February 2003, accepted 25 February 2003) Eur. J. Biochem. 270, 1784–1790 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03537.x Arabinofuranosylcytocine (AraC) has a hydroxyl group in the trans configuration at the 2¢-position of the sugar. b- L -2¢- deoxynucleosides were shown to inhibit specifically hepatitis B replication, along with b- L -thymidine ( L -dT) and b- L - deoxycytidine ( L -dC) which are the most potent inhibitors [11]. b- L -NTP and b- D -NTP are enantiomers resulting from the asymmetry of the C4¢ and C1¢ of the ribose. L -cytidine analogs used in therapies are substrates of deoxycytosine kinase. This enyzme, which adds the first phosphate to cytidine derivatives with variable sugar structures, has a low enantioselectivity although it generally favors the D -enantio- mer [12]. Human UMP/CMP kinase was recently shown to phosphorylate monophosphate derivatives of deoxycyti- dine analogs in either the L -or D - conformers [13]. The third phosphate addition to L -cytidine analogs is catalysed by phosphoglycerate kinase [14] but not by NDP kinase, which reacts only with D -nucleotide analogs [15]. Another NMP kinase, AMP kinase also called myokinase, was reported to complement NDP kinase deficiency in E. coli upon gene disruption [16]. AMP kinase catalyses the reversible ATP- dependent synthesis of ADP from AMP. Therefore the reverse reaction forms ATP and AMP from two ADP molecules. We ask the question whether UMP-CMP kinase also allows the addition of the third phosphate to analog diphosphate substrates in the reverse reaction. We analyze here the specificity of human UMP-CMP kinase for different substrates including L -3TCMP in the forward and backward reactions. Materials and methods Chemicals All nucleotides were purchased from Sigma chemicals (St Louis, MO, USA). L -3TC was extracted from EpivirÒ tablets (GlaxoWellcome) with methanol. After purification by silica gel column chromatography (dichloromethane/ methanol, 95 : 5, v/v), the nucleoside was phosphorylated into the 5¢-monophosphate by overnight incubation at 20 °C with GTP in the presence of deoxyadenosine- deoxycytidine kinase from Bacillus subtilis (a kind gift of J. Neuhard, University of Copenhagen, Denmark). L -3TCMP was purified by successive chromatography on a G-10 column (240 mL total volume) in water and by reverse phase on RP-18 silica gel (50 mL). The purity of L -3TCMP was characterized by nuclear magnetic resonance ( 1 H, 31 P) spectroscopy and mass spectrometry. Cloning of the gene coding for UMP-CMP kinase, expression and purification of His-tagged UMP-CMP kinase The 588-bp fragment corresponding to the gene coding for the human UMP-CMP kinase was amplified by PCR [17] using cDNA from SK melanoma cells as the matrix. The two synthetic oligonucleotides used for amplification were: 5¢-GGAATTCCATATGAAGCCGCTGGTCGTGTTC GTCCTCGGCGGCCCCGGCGCCGGCAAGGGGA CC-3¢ and 5¢-CCGCTCGAGTTAGCCTTCCTTGTCA AAAATCTG-3¢. During amplification, NdeIandXhoI restriction sites (in bold letters in the oligonucleotide sequences) were created at both ends of the amplified fragment. After digestion by NdeIandXhoI, the amplified gene was inserted into the pET28a plasmid (Novagen, Inc.) digested with the same enzymes. Two clones containing the gene coding for the human UMP- CMP kinase and overexpressing the human UMP-CMP kinase with a His-tag at the N-terminal part were characterized. One of them was kept for further studies and the corresponding plasmid named pHL60-5. The DNA insert was sequenced [18] in order to verify the absence of any mutational events in the course of amplification. BL21 (DE3)/pDIA17 E. coli strain [19] transformed with pHL60-5 plasmid was grown in 2YT medium [17] contain- ing 30 lgÆmL )1 chloramphenicol and 70 lgÆmL )1 kana- mycine until D 600 ¼ 1. After induction with 1 m M isopropyl thio-b- D -galactoside and growth for 3 h at 30 °C, cells were harvested, resuspended in 40 mL lysis phosphate buffer 50 m M pH 8 containing 300 m M NaCl, 10 m M imidazole, protease inhibitors (Complete EDTA-free, Roche) and 1m M dithiothreitol and kept at )80 °C. Cells were broken in a French Press at 100 Mpa and centrifuged for 30 min at 12 000 g at 4 °C. The supernatant was added onto a 20-mL Ni-nitrilotriacetic acid column (Qiagen, Germany) preequili- brated with lysis buffer. The column was washed with lysis buffer (100 mL). The protein was then eluted by a linear imidazole gradient (10–250 m M ) pH 8.0. Fractions contain- ing the enzymatic activity were pooled and dialyzed against Tris/HCl 50 m M pH 7.5 containing 20 m M NaCl, 1 m M dithiothreitol and 50% glycerol and then kept at )20 °C. The protein was > 99% pure as judged by SDS/PAGE on 15% acrylamide gels (Biorad). The homogeneity of the enzyme preparation was meas- ured by dynamic light scattering experiments using a Dynapro-800 instrument (Proteinsolutions). The protein was > 99.6% monomer with a hydrodynamic radius of 2.2 nm and a molecular mass estimated as 22 000. The Stokes radius was measured by gel filtration on a Superdex 75H 10/30 (Amersham Biosciences) in Tris/HCl 50 m M , pH 7.5 containing 200 m M KCl and 1 m M dithiothreitol, using bovine albumin, ovalbumin and chymotrypsinogen as markers. The elution volume of UMP-CMP kinase indica- ted a Stokes radius of 2.8 nm, in agreement with a monomer species. Expression and purification of human UMP-CMP kinase as a fusion protein with GST BL21 gold (DE3) E. coli cells (Stratagene) were transformed with the plasmid umpk-pGEX5X, a kind gift from A. Kar- lsson (Karolinska Institute, Stockholm, Sweden), coding for a fusion protein with the GST at the N-terminus of UMP- CMP kinase [4]. Cells were grown in 2YT medium [17] containing ampicillin 100 lgÆmL )1 until D 600 ¼ 0.4. After induction with 1 m M isopropyl thio-b- D -galactoside and growth for 5 h at 22 °C, cells were harvested, washed once in 100 m M Tris/HCl pH 8 containing 10 m M MgCl 2 and 1m M dithiothreitol, resuspended in 50 m M Tris/HCl pH 8 containing 1 m M EDTA, 5 m M MgCl 2 ,5m M KCl, 1 m M dithiothreitol and 0.5 m M phenylmethanesulfonyl fluoride and kept at )80 °C. The cells were broken as indicated above, centrifuged for 30 min at 12 000 g and the super- natant was incubated with glutathione-agarose for 1 h at Ó FEBS 2003 Substrate specificity of human and Dictyostelium UMP-CMP kinase (Eur. J. Biochem. 270) 1785 4 °C according to the Amersham Biosciences procedure. The gel was poured into a column and washed with phosphate buffer 100 m M pH 7.4 containing 100 m M NaCl, 1m M EDTA and 1 m M phenylmethanesulfonyl fluoride. The fusion protein was eluted by phosphate buffer 100 m M pH 7.4 containing 100 m M NaCl, 5 m M MgCl 2 and 10 m M glutathione. The fractions containing UMP-CMP kinase activity were pooled; glycerol was added to a 20% final concentration and kept at )20 °C. The protein was found 95% pure by SDS/PAGE on 15% acrylamide gels (Biorad). No loss in activity was observed after 2 months. The typical yield was 10 mg pure fusion protein per litre of culture. Expression and purification of Dictyostelium UMP-CMP kinase Plasmid pIMS5, a kind gift from L. Wiesmu ¨ ller (University of Hamburg, Germany), was expressed in XL 1 Blue E. coli strain in Luria Broth. After growth at 37 °CtoD 600 ¼ 0.5, isopropyl b- D -thiogalactoside (0.5 m M ) induction was car- ried out for 4 h at 22 °C. After centrifugation, cells were resuspended in 50 m M Tris/HCl buffer pH 7.4 containing 1m M EDTA and 2 m M dithiothreitol (buffer A). The cells were sonicated (3 · 2 min) and centrifuged (17 000 g, 30 min). The supernatant was applied onto a Blue-Seph- arose column (Amersham Biosciences) (100 mg protein for 6 mL of gel), washed with 50 mL of buffer A and 10 mL of buffer A containing 0.3 M NaCl, before elution of the protein with buffer A containing 1 M NaCl. Further purification by gel filtration chromatography (Ultrogel AcA 54, Sigma) in buffer A without dithiothreitol resulted in pure enzyme. It was stored at )20 °C in 20% glycerol. Enzymatic activity of UMP-CMP kinase The forward reaction of UMP-CMP kinase was followed in a spectrophotometer by measuring ADP formation as described [20] in the presence of dithiothreitol [4,13,21]. The assay contained 50 m M Tris/HCl pH 7.4, 50 m M KCl, 5m M MgCl 2 ,1m M ATP, 0.2 m M NADH, 10 m M dithio- threitol, 1 m M phosphoenolpyruvate, 1 m M ATP and the auxiliary enzymes: pyruvate kinase (4 U), lactate dehydro- genase (4 U) and NDP kinase from Dictyostelium (4 U). In the case of L -3TCMP as substrate, no NDP kinase was added, as L -3TCDP is not a substrate for this enzyme [15]. The reaction was started at 37 °C by addition of 1 m M dCMP or another phosphate acceptor (1 m M )andenzyme (final concentration 8.4 lgÆmL )1 ¼ 0.17 l M ). In order to avoid limitation by the coupled system, the rates were below 0.2 units at A 340 per min. The rate was calculated assuming that two ADP are generated during the reaction (due to the presence of NDP kinase), except when L -3TCMP was the phosphoacceptor where only one ADP was produced. Curve-fit was performed using KALEIDAGRAPH (Abelbeck Software) for a hyperbolic progress equation unless indica- ted. Alternatively the enzyme-coupled assay was automated in 96-wells microplates in 250 lL final volume with four times less enzyme and the reactions were monitored during 10 min on a microplate reader (BIO-TEK Elx808). The reaction rates were calculated using the KC4 software (BIO- TEK Instruments, USA). The reverse reaction catalysed by the human UMP- CMP kinase was assayed with [ 3 H]CDP (Amersham Biosciences) and CDP, dCDP or L -3TCDP as substrates. The formation of dCTP (or L -3TCTP) catalysed by UMP- CMP kinase was studied under steady state conditions. The amount of [ 3 H]CMP and [ 3 H]CTP formed in the presence of 1 m M [5- 3 H]CDP (0.2 CiÆmmol )1 ) and 0.25– 2m M of dCDP or L -3TCDP was monitored for 2, 4 and 6 min. The assays were started by adding the enzyme (final concentration 8.4 lgÆmL )1 ¼ 0.17 l M )toareaction mixture (15 lL) containing 50 m M Tris/HCl pH 7.5, 5m M MgCl 2 and the substrates at 37 °C. The reaction was stopped at 2, 4 and 6 min by adding 3 lL aliquots of the reaction mixture to 2 lL stop solution consisting of 1 m M formic acid and 10 m M each of CDP and CTP. Radioactive CMP, CDP and CTP were separated on TLC aluminum sheets (Silica gel 60 F 254 ,Merck),which were developed with isopropanol/NH 3 30%/H 2 O, 3 : 1 : 1, v/v/v. After drying, the plates were counted on a BETA-Imager 2000 counter (Biospace, Paris, France) [21] and analyzed. As [ 3 H]CMP was made in parallel in the reaction from two [ 3 H]CDP molecules, the corres- ponding background was subtracted from the results obtained in the presence of dCDP or L -3TCDP. Results Analysis of the forward reaction catalysed by UMP-CMP kinase with natural substrates and nucleoside analogs The purified recombinant human UMP-CMP kinase expressed as a His-tag fusion was analyzed by dynamic light scattering and gel filtration on Superdex 75, yielding an hydrodynamic radius of 2.2 nm and 2.8 nm, respectively (results not shown). This corresponded to a monomeric state of the protein. Figure 1A and Table 1 show the activity of human UMP-CMP kinase as a function of natural substrates, and the corresponding kinetic param- eters. The reaction rates were found to increase as a function of UMP and CMP as expected for Michaelis curves, but did not reach a plateau (the maximum rate V) due to substrate inhibition at concentrations above 0.2 m M . The correpond- ing inhibition constants K i were 0.5 m M for CMP and 1.5 m M for UMP, with K m values of 22 l M and 44 l M , respectively. In contrast, no excess substrate inhibition was foundwithdCMP(K m ¼ 0.9 m M ), indicating that the presence of the 2¢OH of the ribose is presumably involved in the inhibition mechanism. Both UMP and CMP were found to be better substrates than dCMP with smaller K m and higher V, resulting in considerably improved catalytic efficiencies (k cat /K m ). These data were found for concentra- tions of 5 m M Mg 2+ ions and above. The substrate specificity of the His-tagged enzyme was investigated by assaying a series of nucleoside monophos- phates: AMP was found to be a slow substrate with K m higher than 5 m M and a catalytic efficiency about 1000 M )1 Æs )1 while dTMP and GMP were not substrates (k cat /K m < 100 M )1 Æs )1 ) (not shown). The activity of human UMP-CMP kinase was then analyzed using nucleo- side analogs as substrates. As shown in Fig. 1B, ( L )-3TCMP and AraCMP were substrates with K m in the 0.2 m M range. 1786 C. Pasti et al. (Eur. J. Biochem. 270) Ó FEBS 2003 No inhibition by excess of substrate was detected. The turnover number for AraCMP was similar to that of UMP (around about 150 s )1 ) and it was significantly lower for L -3TCMP (36 s )1 ) (Table 1). According to catalytic effi- ciencies, NMPs fall in two groups with catalytic efficiencies intherangeof10 7 M )1 Æs )1 for UMP and CMP, and of 10 5 M )1 Æs )1 for AraCMP, L -3TCMP and dCMP (Table 1). Careful characterization of the reaction catalysed by human UMP-CMP kinase indicated that the type of fusion protein made for purification purposes could interfere with the kinetic properties of the enzyme. Thus, when expressed as a GST fusion, the kinetic parameters of human UMP- CMP kinase were different from those of the His-tagged fusion protein. As shown in Table 2, the catalytic efficiencies calculated from K m and k cat for both natural substrates or analogs resulted in an average fivefold decrease in catalytic efficiencies as compared to the His-tagged enzyme shown in Table 1. As NMP kinases are very flexible proteins [10], this decrease in activity with the GST-fusion enzyme is likely to correspond to a decrease in the protein dynamics due to the presence of the bulky additional GST domain. We also compared the kinetic properties of the human UMP-CMP kinase with those of the enzyme from Dicty- ostelium whose tertiary structure was determined by X-ray crystallography [3] (Table 3). Although both proteins show 50% sequence identity, several differences in the kinetic properties were observed (Table 3). For natural substrates (UMP and CMP), no inhibition by substrate excess is found. While the turnover is increased for the Dictyostelium enzyme by two fold, a decrease in catalytic efficiencies from 10 7 M )1 Æs )1 for human UMP-CMP kinase to 10 6 M )1 Æs )1 is observed for the Dictyostelium enzyme. Table 3 shows that Dictyostelium UMP-CMP kinase also accepts L -3TCMP as substrate, although it reacts with the analog about 500 times less than the human enzyme. The Dictyostelium enzyme seemed a more appropriate structural model for the human enzyme [3,22] than the protein from yeast, which differs significantly in specificity since it also reacts with adenosine [23,24]. The structure of Dictyostelium UMP-CMP kinase has been determined as a complex with ATP, CMP as well as with AlF 3 (PDB accession no. 3UKD), providing a model for the active state, as aluminum fluoride resembles a transition state of a phosphoryl transfer reaction [22]. Figure 2 shows a model in which CMP was replaced by L -3TCMP in the acceptor site of Dictyostelium UMP-CMP kinase. Starting from the superimposition of the a-phosphates, the oxathiolan ring of L -3TCMP could fit over the CMP ribose. This results in a translation of the cytosine ring of about 1.2 A ˚ . This shift seems to be easily tolerated by the binding site, due to the absence of direct interactions between the base moiety and the side chains of the enzyme [3]. In conclusion, L -3TCMP can be modeled into the CMP binding site of UPM-CMP Fig. 1. Measure of the steady-state kinetics parameters of human UMP- CMP kinase. (A) Kinetic for natural substrates UMP, CMP, dCMP. (B) Kinetic for substrate analogs D-AraCMP and L -3TCMP. V is expressedinunit(lmol productÆmin )1 )Æmg )1 of the fusion protein UMP-CMP kinase with GST (PM ¼ 51 000). The reaction rate v with UMP and CMP as a substrate [S] were best fitted with Eqn (1). v ¼ V Á S½ K m þ S ½ þ s ½ 2 K i ð1Þ Table 1. Kinetic parameters of human His-tagged UMP-CMP kinase in the forward reaction. Activity measurements were carried out using a spectrophotometric coupled assay under initial rates conditions in the presence of 1 m M ATP and 5 m M Mg 2+ as described in Material and methods section (conditions of Fig. 1). Values are means of two to four independent measurements. Even when the Michaelis curve reached a plateau, catalytic efficiencies have been calculated from the linear fit of the first values (initial rate as a function of substrate concentration). Substrate K m (m M ) V (UÆmg )1 ) k cat (s )1 ) k cat /K m ( M )1 Æs )1 ) K i (m M ) UMP 0.045 ± 0.010 350 ± 30 130 ± 10 (2.8 ± 0.2) · 10 7 1.5 ± 0.4 CMP 0.020 ± 0.005 350 ± 30 130 ± 10 (6.5 ± 0.5) · 10 7 0.50 ± 0.1 dCMP 0.90 ± 0.10 200 ± 20 73 ± 8 (8.0 ± 0.5) · 10 4 No inhibition L -3TCMP 0.15 ± 0.02 100 ± 10 36 ± 4 (2.8 ± 0.3) · 10 5 No inhibition araCMP 0.26 ± 0.05 400 ± 40 150 ± 20 (5.8 ± 0.5) · 10 5 No inhibition Ó FEBS 2003 Substrate specificity of human and Dictyostelium UMP-CMP kinase (Eur. J. Biochem. 270) 1787 kinase, confirming the absence of stereospecificity observed at the acceptor site level. The UMP-CMP kinase reverse reaction We have examined the possibility that UMP-CMP kinase could also catalyse the addition of the third phosphate in the reverse reaction (see scheme above). Clearly this would depend on the stereospecificity of the donor site. The terms acceptor and donor are used here according to the forward reaction, with the donor site binding a nucleotide triphos- phate (ATP but also other NTPs) and the acceptor site binding a nucleotide monophosphate (preferentially CMP and UMP). In the reverse reaction, both donor and acceptor sites bind a nucleotide diphosphate. We have studied this reaction using [ 3 H]CDP as a substrate. Product formation was measured with a BETA Imager after separation on TLC plates as described under Materials and methods. Under our experimental conditions, when [ 3 H]CDP was used alone as substrate, both [ 3 H]CMP and [ 3 H]CTP were produced in similar amounts (Fig. 3A, lane 2–4). Because no radioactive L -3TCDP was available, we Table 2. Kinetic parameters of human GST fusion UMP-CMP kinase in the forward reaction. Activity measurements were carried out under the conditions described in Table 1. Substrate K m (m M ) V (UÆmg )1 ) k cat (s )1 ) K cat /K m ( M )1 Æs )1 ) K i (m M ) UMP 0.14 ± 0.10 110 ± 10 92 ± 9 (6.5 ± 0.5) · 10 5 0.5 ± 0.2 CMP 0.07 ± 0.01 86 ± 10 72 ± 6 (1.0 ± 0.2) · 10 6 0.12 ± 0.02 dCMP 1.10 ± 0.10 36 ± 4 30 ± 4 (3.0 ± 0.3) · 10 4 No inhibition L -3TCMP 0.30 ± 0.05 17 ± 3 14 ± 3 (4.6 ± 0.4) · 10 4 No inhibition ara-CMP 0.34 ± 0.03 55 ± 5 45 ± 4 (1.3 ± 0.3) · 10 5 No inhibition Table 3. Kinetic parameters of Dictyostelium UMP-CMP kinase in the forward reaction. Activity measurements and analysis were carried out under the conditions described in Table 1 except for the determination of catalytic efficiencies which obtained from the ratio of the kinetic parameters k cat and K m . Substrate K m (m M ) V (UÆmg )1 ) k cat (s )1 ) k cat /K m ( M )1 Æs )1 ) ðk cat =K m Þ Dd ðk cat =K m Þ human UMP 0.2 ± 0.1 520 ± 30 217 ± 10 (1.1 ± 0.3) · 10 6 0.40 CMP 0.2 ± 0.1 1000 ± 30 410 ± 20 (2.0 ± 0.4) · 10 6 0.30 dCMP 2.0 ± 0.5 55 ± 10 22 ± 5 (1.1 ± 0.2) · 10 4 0.13 L -3TCMP 2.0 ± 0.5 25 ± 5 10 ± 3 (5.0 ± 0.5) · 10 3 0.02 Fig. 2. Illustrations of L -3TCMP and CMP superimposed into the acceptor site in Dictyostelium UMP-CMP kinase. The docking was performed on a Silicon Graphics workstation using O software [33]. CMP is represented according to the conformation in the complex with thekinaseasfoundin3UKD.PDB.b- L -3TCMPwaspositionedinthe active site by choosing the same orientation for the cytidine base moiety as in the complex with CMP. Carbon atoms of CMP and L -3TCMPareinblackandgrey,respectively,withsulfurinyellow, phosphorus in magenta and oxygens in blue (for the base) and red (for the sugar and the phosphate). Left, overview on the base moiety; right, overview on the sugar moiety. Fig. 3. Kinetic of backward reaction of His-tagged human UMP-CMP kinase. (A) Time-course of products formation after separation on a TLC silica plate visualized on the BETA Imager. The substrates are [ 3 H]CDP plus CDP (lanes 2–4), or plus dCDP (lanes 5–7) or plus L -3TCDP (lanes 8–10). (B) Initial rates of [ 3 H]CMP formation upon addition of dCDP or L -3TCDP. 1788 C. Pasti et al. (Eur. J. Biochem. 270) Ó FEBS 2003 developed a competitive assay in order to analyze the ability of the enzyme to use nucleotide analogs in the reverse reaction. Indeed when an unlabelled NDP was added to [ 3 H]CDP, the reaction was expected to produce [ 3 H]CMP and NTP and/or [ 3 H]CTP and NMP in equal amounts. The relative affinities of the added NDP for the donor and the acceptor sites are the determinant for the ratio [ 3 H]CMP/ [ 3 H]CTP. Upon cold CDP addition, this ratio was still found to be close to 1. In contrast, upon dCDP addition, the [ 3 H]CMP formation was found to increase with dCDP, while [ 3 H]CTP remained very low (Fig. 3A, lane 5–7 and 3B). This indicates that dCDP binds preferentially to the donor site where it is transformed into dCTP. We then examined the effect of L -3TCDP addition. As shown in Fig. 3, the amount of [ 3 H]CMP decreased in a dose-dependent way and [ 3 H]CTP remained very low. We conclude that, in contrast to dCDP, L -3TCDP was not phosphorylated at the donor site, but rather that it constitutes a competitor in the reaction (Fig. 3A, lane 8–10 and Fig. 3B). Each nucleotide site has a different specificity, the donor site binding D -nucleotides only and the acceptor site recognizing UMP, CMP and L -analogs like L -3TCMP and L -3TCDP. Discussion Human UMP-CMP kinase has been recently a subject of interest for several research groups. In 1999, Van Rompay et al. reported the first cloning of the human gene and characterization of the corresponding protein, UMP-CMP kinase [4]. The enzyme was expressed as a GST fusion protein to facilitate its purification. During the course of the present study, Cheng’s laboratory also cloned the human enzyme and provided a measure of kinetic parameters. They also showed that it was predominantly localized in the cytoplasm [13]. We observed that the enzyme with an N-terminal His-tag presents better kinetic performances than the GST-fusion protein. The enzyme was found to be active in Tris buffer during a few hours only but stabiliza- tion by glycerol significantly improves the reproducibility of data. Addition of dithiothreitol in the assay also improved the measures as previously shown [4,13,23]. Under these conditions, the turnover of human UMP-CMP kinase (k cat ¼ 130 s )1 ) is the highest of the NMP kinases family. This high activity is particularly striking when compared to human TMP kinase (k cat ¼ 1s )1 ) [24,25]. Inhibition by excess of substrate was observed for UMP and CMP but not for dCMP, AraCMP and L -3TCMP. Previous studies did not report such an inhibition, probably because the range of concentrations investigated was too narrow. Such a substrate inhibition can be attributed to a non-productive binding of CMP and UMP, or to the products CDP and UDP, forming a dead-end complex. It is correlated to the presence of the 2¢-OH of the sugar in the cis position. The latter is absent in dCMP and L -3TCMP, and is in trans in the arabinose epimer. The nonproductive binding of NMP due to the 2¢-OH in cis could occur at the donor site. A similar situation has been observed by X-ray analysis in the Drosophila deoxynucleoside kinase with an additional ATP bound to the acceptor site [26]. Such an inhibition is an important parameter to take into account when the kinase activity is characterized in cellular extracts. Indeed, the kinase specificity profile is used to characterize the contribution of the cellular kinases in tissues and is often determined using only one saturating concentration of each substrate. Obviously, choosing a high substrate concentra- tion, as one usually does to perform enzyme assay, may provide misleading results. The reactivity of both sites of UMP-CMP kinase for L -3TC mono or diphosphate was studied. The acceptor site is able to bind and to phosphorylate L -3TCMP in the forward reaction (Figs 1B and 2). At the donor site, L -3TCDP was found to be an inhibitor, while dCDP was a substrate (Fig. 3). Therefore L -3TCTP cannot be synthes- ized by human UMP-CMP kinase. Solving the structure of human UMP-CMP kinase by X-ray crystallography is currently underway and will help to understand the molecular basis of this substrate specificity. Several enzymes of the nucleoside kinases family have been shown to recognize L -derivatives as substrates, in particular at the acceptor site, including deoxycytidine kinase, human mitochondria thymidine kinase 2 and herpes thymidine kinase 1 and 2 [27–29]. Only human deoxycyti- dine kinase was previously shown to have a relaxed enantioselectivity at the donor site for L -ATP [30]. This is not the case for human UMP-CMP kinase, as L -3TCDP is not a substrate at the donor site. Enzymes from virus and phages including HIV reverse transcriptase, HBV DNA polymerase, human DNA primase and T4 DNA ligase have been also reported to use L -NTPs as substrates [31]. The three steps of the cellular activation of L -3TC involve deoxycytidine kinase [30], UMP-CMP kinase [13] (data not shown) and phosphoglycerate kinase [14]. Most human degradation enzymes like cytidine deaminase present a strict enantioselectivity for D -nucleotides [11,29] that probably contribute to the high stability of the phosphorylated forms of L -3TC in cells and to a high cellular level of L -3TCTP. Several laboratories have tried to improve the catalytic properties of herpes virus thymidine kinase and human thymidylate kinase for antiviral drugs. An improved herpes virus enzyme in combination with AZT could be used as a suicide enzyme for cancer cells [32]. Taking into account its high efficiency in phosphorylating several nucleotide ana- logs, human UMP-CMP kinase could provide a model for new approaches in suicide enzymes therapies. Acknowledgements We wish to thank A. Karlsson (Karolinska Institute, Stockholm, Sweden) for kind gift of the plasmid coding for GST-fusion UMP- CMP kinase, and our colleagues from the Institut Pasteur for their kind help: M. Delarue for modelling L -3TCMP complexed to UMP-CMP kinase, A. Haouz for DLS measurements, A. Cardona for BETA- Imager counting, C. Guerreiro for kind advises about L -3TCMP purification, E. Seclaman for helping cloning the gene of human UMP- CMP kinase, N. Duchange for the kind gift of human cDNA bank and O. Barzu for stimulating discussions. This work was supported in part by a grant from Agence Nationale pour la Recherche contre le SIDA (France). HML was supported by Institut National de la Sante ´ et de la Recherche Me ´ dicale (France) and CP by University of Ferrara (Italy). References 1. Liljelund, P. & Lacroute, F. (1986) Genetic characterization and isolation of the Saccharomyces cerevisiae gene coding for uridine monophosphokinase. Mol. Gen. Genet. 205, 74–81. Ó FEBS 2003 Substrate specificity of human and Dictyostelium UMP-CMP kinase (Eur. J. Biochem. 270) 1789 2. Wiesmu ¨ ller,L.,Noegel,A.A.,Barzu,O.,Gerisch,G.&Schleicher, M. (1990) cDNA-derived sequence of UMP-CMP kinase from Dictyostelium discoideum and expression of the enzyme in Escherichia coli. J. Biol. Chem. 266, 6339–6345. 3. Scheffzek, K., Kliche, W., Wiesmu ¨ ller, L. & Reinstein, J. (1996) Crystal structure of the complex of UMP/CMP kinase from Dictyostelium discoideum and the bisubstrate inhibitor P 1 -(5¢- adenosyl) P 5 -(5¢-uridyl) pentaphosphate (UP 5 A) and Mg 2+ at 2.2 A ˚ : implications for water-mediated specificity. Biochemistry 35, 9716–9727. 4. Van Rompay, A.R., Johansson, M. & Karlsson, A. (1999) Phos- phorylation of Deoxycytidine analog monophosphates by UMP- CMP kinase: molecular characterization of the human enzyme. Mol. Pharmacol. 56, 562–569. 5. Eriksson, S., Munch-Petersen, B., Johansson, K., & Eklund, H. (2002) Structure and function of cellular deoxyribonucleoside kinases. Cell. Mol. Life Sci. 59, 1327–1346. 6. Van Rompay, A.R., Johansson, M. & Karlsson, A. (2000) Phos- phorylation of nucleosides and nucleoside analogs by mammalian nucleoside monophosphate kinases. Pharmacol. Therapeut. 87, 189–198. 7. Serina, L., Bucurenci, N., Gilles, A.M., Surewicz, W.K., Fabian, H., Mantsch, H.H., Takahashi, M., Petrescu, I., Batelier, G. & Baˆ rzu, O. (1996) Structural properties of UMP-kinase from Escherichia coli: modulation of protein solubility by pH and UTP. Biochemistry 35, 7003–7011. 8. Bucurenci, N., Sakamoto, H., Briozzo, P., Palibroda, N., Serina, L.,Sarfati,R.Z.,Labesse,G.,Briand,G.,Danchin,A.,Baˆ rzu, O. & Gilles, A.M. (1996) CMP kinase from Escherichia coli is structurally related to other nucleoside monophosphate kinases. J. Biol. Chem. 271, 2856–2862. 9. Muller-Dieckmann, H.J. & Schultz, G.E. (1994) The structure of uridylate kinase with its substrates, showing the transition state geometry. J. Mol. Biol. 236, 361–367. 10. Vonrhein, C., Schlauderer, G.J. & Schulz, G.E. (1995) Movie of the structural changes during a catalytic cycle of nucleoside monophosphate kinases. Structure 3, 483–490. 11. Bryant, M.L., Bridges, E.B., Placidi, L., Faraj, A., Loi, A G., Pierra, C., Dukhan, D., Gosselin, G., Imbach, J.L., Hernandez, B., Juodawlkis, A., Tennant, B., Korba, B., Cote, P., Marion, P., Cretton-Scott, E., Shinazi, R.F. & Sommadossi, J P. (2001) Antiviral 1-nucleosides specific for Hepatitis B virus infection. Antimicrob. Agents Chemotherapy 45, 229–235. 12. Wang, J., Choudhury, D., Chattopadhyaya, J. & Eriksson, S. (1999) Stereoisomeric selectivity of human deoxyribonucleoside kinases. Biochemistry 38, 16993–16999. 13. Liou,J Y.,Dutschman,G.E.,Lam,W.,Jiang,Z.&Cheng,Y C. (2002) Characterization of human UMP/CMP kinase and its phosphorylation of D -and L -form deoxycytidine analogue monophosphates. Cancer Res. 62, 1624–1631. 14. Krishnan, P., Fu, Q., Lam, W., Liou, J Y., Dutschman, G. & Cheng, Y C. (2002) Phosphorylation of pyrimidine deoxy- nucleoside analog diphosphates. J. Biol. Chem. 277, 5453– 5459. 15. Kreimeyer, A., Schneider, B., Sarfati, R., Faraj, A., Sommadossi, J., Veron, M. & Deville-Bonne, D. (2001) NDP kinase reactivity towards L-3TC nucleotides. Antiviral Res. 50, 147–156. 16. Lu, Q. & Inouye, M. (1996) Adenylate kinase complements nucleoside diphosphate kinase deficiency in nucleotide meta- bolism. Proc. Natl Acad. Sci. USA 93, 5720–5725. 17. Sambrook, J., Fritsch, E.F. & Manniatis, T. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, New York, USA. 18. Sanger, F., Nicklen, S. & Coulson, A.R. (1977) DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463–5467. 19. Munier, H., Gilles, A.M., Glaser, P., Krin, E., Danchin, A., Sarfati, R. & Barzu, O. (1991) Isolation and characterization of catalytic and calmodulin-binding domains of Bordetella pertussis adenylate cyclase. Eur. J. Biochem. 196, 469–474. 20. Blondin, C., Serina, L., Wiesmuller, L., Gilles, A M. & Baˆ rzu, O. (1994) Improved spectrophotometric assay of nucleoside mono- phopshate kinase activity using the pyruvate kinase/lactate dehy- drogenase coupling system. Anal. Biochem. 220, 219–221. 21. Charpak, G., Dominik, W. & Zaganidis, N. (1989) Optical ima- ging of the spatial distribution of betaparticles emreging from surfaces. Proc Natl Acad Sci. USA. 86, 1741–1745. 22. Schlichting, I. & Reinstein, J. (1997) Structures of active con- formations of UMP kinase from Dictyostelium discoideum suggest phosphoryl transfer is associative. Biochemistry 36, 9290–9296. 23. Muller-Dieckmann, H.J. & Schulz, G.E. (1995) Substrate specifi- city and assembly of the catalytic center derived from two struc- tures of ligated uridylate kinase. Journal of Molecular Biology 246, 522–530. 24. Yan, H. & Tsai, M D. (1999) Nucleotide monophosphate kinases: structure, mechanism, and substrate specificity. In Advances in Enzymology and Related Areas of Molecular Biology (Purich, D.L., ed.), pp. 103–135. John Wiley & Sons, New York, USA. 25. Brundiers, R., Lavie, A., Veit, T., Reinstein, J., Schlichting, I., Ostermann, N., Goody, R.S. & Konrad, M. (1999) Modifying human thymidylate kinase to potentiate azidothymidine activa- tion. J. Biol. Chem. 274, 35289–35292. 26. Johansson,K.,Ramaswamy,S.,Ljungcrantz,C.,Knecht,W., Piskur, J., Munch-Petersen, B., Eriksson, S. & Eklund, H. (2001) Structural basis for substrate specificities of cellular deoxy- ribonucleotide kinases. Nat. Struct. Biol. 8, 616–620. 27. Spadari, S., Maga, G., Verri, A., Bendiscioli, A., Tondelli, L., Capobianco, M., Colonna, F., Garbesi, A. & Focher, F. (1995) Lack of stereospecificity of some cellular and viral enzymes involved in the synthesis of deoxyribonucleotides and DNA: molecular basis for the antiviral activity of unnatural 1-beta- nucleosides. Biochimie 77, 861–867. 28. Furman, P.A., Wilson, J.E., Reardon, J.E. & Painter, G.R. (1995) The effect of absolute configuration on the anti-HIV and anti- HBV activity of nucleoside analogues. Antiviral Chem. Chemo- therapy 6, 345–355. 29. Maury, G. (2000) The enantioselectivity of enzymes involved in current antiviral therapy using nucleoside analogues: a new strategy? Antiviral Chem. Chemotherapy 11, 165–190. 30. Verri, A., Montecucco, A., Gosselin, G., Boudou, V., Imbach, J.L.,Spadari,S.&Focher,F.(1999) L -ATP is recognized by some cellular and viral enzymes: does chance drive enzymic enantio- selectivity? Biochem. J. 337, 585–590. 31. Focher, F., Maga, G., Bendiscioli, A., Capobianco, M., Colonna, F., Garbesi, A. & Spadari, S. (1995) Stereospecificity of human DNA polymerases a, b, c, d and e, HIV-reverse transcriptase, HSV-1 DNA polymerase, calf thymus terminal transferase and E. coli DNA polymerase I in recognizing D -and L -thymidine 5¢-triphosphate as substrate. Nucleic Acids Res. 23, 2840–2847. 32. Encell, L.P., Landis, D.M. & Loeb, L.A. (1999) Improving enzymes for cancer gene therapy. Nat. Biotechnol. 17, 143–147. 33. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991) Improved methods for binding protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119. 1790 C. Pasti et al. (Eur. J. Biochem. 270) Ó FEBS 2003 . or L -3TCDP. Results Analysis of the forward reaction catalysed by UMP-CMP kinase with natural substrates and nucleoside analogs The purified recombinant human UMP-CMP kinase expressed. cellular kinases of nucleoside analogs used in antiviral therapies. The reactivity of human UMP-CMP kinase towards natural substrates and nucleotide analogs