Identification of four amino acid residues essential for catalysis in ...

2 downloads 145 Views 122KB Size Report
tion (Müller and Zahn, 1979; Bouffard et al., 1993). Hence the cytostatic effect of the analogues on various tissues is highly dependent on the level of CDA in the ...
Protein Engineering vol.11 no.1 pp.59–63, 1998

Identification of four amino acid residues essential for catalysis in human cytidine deaminase by site-directed mutagenesis and chemical modifications

A.Cambi, S.Vincenzetti, J.Neuhard1, S.Costanzi, P.Natalini and A.Vita2 Dipartimento di Biologia MCA, Universita` di Camerino, 62032 Camerino, Italy and 1Center for Enzyme Research, Department of Biological Chemistry, University of Copenhagen, 1307 Copenhagen K, Denmark 2To

whom correspondence should be addressed

By site-directed mutagenesis on human cytidine deaminase (CDA), five mutant proteins were obtained: C65A, C99A, C102A, E67D and E67Q. The three cysteine mutants were completely inactive, whereas E67D and E67Q showed a specific activity about 200- and 200 000-fold lower, respectively, than the wild-type CDA. Zinc analysis revealed that only E67D, E67Q and C65A contained 1 mol Zn2F/mol subunit as in the wild-type CDA. Kinetic measurements with the specific carboxylic group reagent N-ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline performed on wild-type CDA suggest that Glu67 is essential for the catalytic process. Furthermore, when both native and denatured CDA was titrated with 5,59-dithiobis(2nitrobenzoic acid) six sulfhydryl groups were detected, whereas in the denatured and reduced enzyme nine such groups were found, according to the sequence data. When phydroxymercuriphenyl sulfonate was used, nine sulfhydryl groups were detectable and the release of 1 mol of zinc per mole of CDA subunit was revealed by the metal indicator dye 4-(2-pyridylazo)resorcinol. It seems plausible that the limiting step for the maintenance of zinc in the active site is the formation of coordination between Cys99 and Cys102, whereas Cys65 could lead the zinc to the correct position and orientation within the active site. Keywords: carboxylic reagent/deaminase/mutagenesis/ sulfhydryl titration/zinc protein

Introduction The antitumor or antiviral activity of cytosine nucleoside analogues depends on their intracellular phosphorylation to the corresponding 59 triphosphates, which exert their cytostatic effect by inhibiting DNA polymerase and reverse transcriptase in competition with their natural substrate. The same analogues are susceptible to metabolic degradation by cytidine deaminase (CDA, EC 3.5.4.5), leading to their pharmacological inactivation (Mu¨ller and Zahn, 1979; Bouffard et al., 1993). Hence the cytostatic effect of the analogues on various tissues is highly dependent on the level of CDA in the respective cells (Ho, 1973; Rivard et al., 1981). To overcome the negative effect of CDA with respect to cytosine-based drugs two strategies are possible: (a) to synthesize nucleoside analogues not susceptible to the deaminating activity of the enzyme; or (b) to design potent non-toxic CDA inhibitors to co-administer with cytidine analogue pro-drugs. This requires a detailed knowledge of the CDA catalytic mechanism and hence of the active site molecular structure. Recently it has been shown © Oxford University Press

that the L-enantiomers of cytidine analogues have pronounced antiviral or antitumor activities (Coates et al., 1992; Furman et al., 1995) and that compounds such as L-ddC and L-FddC are stronger inhibitors of HIV and HBV than the corresponding D-enantiomers (Gosselin et al., 1994). This may, at least partially, be explained by the finding that whereas 29-deoxycytidine kinase is a non-enantioselective enzyme that phosphorylates D- and L-enantiomers with the same efficiency (Verri et al., 1997), CDA cannot deaminate the L-enantiomeric forms of cytidine analogues (M.Shafiee, J.-F.Griffon, G.Gosselin, A.Cambi, S.Vincenzetti, A.Vita, S.Eriksson, J.-L.Imbach and G.Maury, in preparation). Human CDA has been purified and partially characterized: it is a tetramer of identical subunits (16.2 kDa/subunit) and contains a tightly bound zinc atom per subunit (Vincenzetti et al., 1996). By comparing the amino acid sequences of CDA from different sources (Yang et al., 1992), two highly conserved regions are apparent. The first consists of the sequence TVH/CAE and was shown for the Escherichia coli enzyme (Betts et al., 1994) to include the zinc-coordinating residue H102 and the residue E104 which is essential for catalysis; the second conserved region consists of the sequence PCXXC where two cysteines (C129 and C132) are coordinated to the zinc atom. In human CDA the glutamate residue is conserved (E67) and H102 is replaced by C65. Thus, the sequence data suggested that in human CDA zinc may be coordinated by three cysteines: C65, C99 and C102. In the present work we employed chemical modifications and site-directed mutagenesis to evaluate the importance of residues E67, C65, C99 and C102 of human CDA in catalysis and in zinc binding. Materials and methods Chemicals N-Ethoxycarbonyl-2-ethoxy-1,2-dihydroquinoline (EEDQ), isopropyl-1-thio-β-D-galactopyranoside (IPTG), 1,10-phenanthroline (OP), 5,59-dithiobis(2-nitrobenzoic acid) (DTNB), p-hydroxymercuriphenyl sulfonate (PMPS), p-chloromercuribenzoic acid (PCMB), 4-(2-pyridylazo)resorcinol (PAR) and iminodiacetic acid Sepharose 6B were obtained from Sigma Chemical (St Louis, MO). Other chemicals were of reagent grade from J.T. Baker Chemicals (Deventer, The Netherlands). PM10 membranes were purchased from Amicon (Beverly, MA) and protein markers from Bio-Rad (Richmond, CA, USA). pTrc99-A vector was supplied by Pharmacia (Uppsala, Sweden). Oligodeoxyribonucleotide primers were synthesized by DNA Technology, ApS (Aarhus, Denmark). The 42 bp Histag linker, cloned into the pET-3d NcoI site, and the pET-3d plasmid were a kind gift from Dr Giuseppa Levantino (University of Pisa, Italy). Restriction nucleases were obtained from either Promega (Madison, WI) or New England Biolabs (Boston, MA). Bacterial strains and growth media Escherichia coli DH5α was used as host for all clonings (Sambrook et al., 1989). For complementation tests the pyrimi59

A.Cambi et al.

Table I. Specific primers used in the two-step site directed mutagenesis and the mutant protein obtained Mutation

Specific primer

Second primer

E67D E67Q H35Q C99A C102A C65A C65H

c ggt ccg atc agc aca gat gcc ggt ccg ttg agc aca gat g c cac agg aaa ctg act gta g c tct cca gct ggg gcc tgc ggg gcc gcc agg caa gtc g ttc agc agc gat gcc cag g ttc agc atg gat gcc ca

PNco PNco PNco PBam PBam PNco PNco

The mutated codons are in bold.

dine-requiring cytidine deaminase negative derivative of MC1061, SØ5201 (MC1061cdd::Tn10 pyr::Kan), was employed. E.coli B strain BL21(DE3) (F– ompT rB–mB–) (Studier and Moffatt, 1986; Grodberg and Dunn, 1988) was employed as host for cloning and expression of pET-3d vectors. Phosphate-buffered AB medium (Clark and Maaløe, 1967) supplemented with 0.2% glucose and 0.2% vitamin-free casamino acids was used as minimal medium; L broth (Miller, 1972) was used as rich medium. When required, supplements were added at the following final concentrations: thiamine 1, uracil 20, CdR 40, ampicillin 100–200, kanamycin 30 and tetracycline 10 µg/ml. DNA techniques Sub-cloning and sequencing of CDA cDNA constructs were performed as described by Vincenzetti et al. (1996). To facilitate production of the wild-type protein, the CDA cDNA was cloned as a 446 bp NcoI/BamHI fragment into pTrc-99A, yielding plasmid pTrcHUMCDA (Vincenzetti et al., 1996). Specific amino acid substitutions were obtained by site-directed mutagenesis of the wild-type CDA cDNA, using a two-step PCR procedure. In the first step, specific primers containing the appropriate mutation (Table I) and either the 59-primer (PNco) cagaCCATGGcccagaagcgtc or the 39-primer (PBam) ccGGATCCaggtggctgttac (Vincenzetti et al., 1996) were used to produce specific ‘megaprimers’. In the second PCR step, the megaprimer was used together with either (PNco) or (PBam) primers, producing the whole CDA cDNA inserts with the correct mutations. Subsequently, the PCR fragment was digested with NcoI and BamHI and cloned into BamHI/NcoI digested pTrc-99A such that the CDA coding region was transcribed from the inducible trc promoter of the vector and translation was initiated from the SD sequence located immediately upstream of the NcoI site. The primary structure of the inserts was confirmed by specific restriction analysis and DNA sequencing. Plasmid DNA was isolated using the Qiagen DNA Kit and PCR products were purified with the Qiagen PCR Purification Kit. Endonuclease digestion and ligation of DNA were done according to the suppliers and the procedure used for transformation was that described by Sambrook et al. (1989). DNA sequence analysis was performed by the chain-terminating method of Sanger et al. (1977). Construction of oligo-histidine-domain-containing vector pHUMCDA-H6 A 42 bp synthetic linker encoding a histidine hexapeptide followed by a tetrapeptide containing the cleavage site for protease FXa was cloned into the NcoI site of the pET-3d vector (Figure 1). In pET-3d(His6) the unique NcoI cloning site is located such that the linker sequence encoding the 60

Fig. 1. Nucleotide sequence of the synthetic linker region of pET-3d(His6). Capital letters in italics, the 42 bp linker; capital letters in roman, nucleotides of pET-3d. The underlined ATG codon will be the start codon of the CDA coding region of the cloned CDA cDNA. The bold ATG codon is the start codon for translation of the His-tagged fusion proteins.

decapeptide is in reading frame with the ATG codon of the NcoI site. The pET-3d(His6) vector was checked by PCR to verify the presence of only one copy of the insert. Subsequently it was sequenced on both strands using primers complementary to the sequences upstream and downstream of the inserted oligonucleotide. The 446 bp NcoI–BamHI fragment containing the human CDA cDNA, from both wild-type and mutants, was isolated from the pTrcHUMCDA plasmids (Vincenzetti et al., 1996), inserted into pET-3d(His6), yielding the pHUMCDA-H6 plasmids, and transformed into E.coli strain BL21(DE3) for expression of the His6–CDA fusion proteins. Expression and purification of the His–CDA fusion proteins by immobilized metal affinity chromatography Transformed E.coli BL21(DE3) strains were grown at 37°C in L broth supplemented with 200 µg/ml ampicillin and His6– CDA expression was induced during late exponential growth (A600 5 0.6) by the addition of 1 mM IPTG. After 3 h of vigorous shaking at 37°C, cells were harvested by centrifugation at 5000 g and washed with 0.9% NaCl. The cell pellet was treated as described by Van Dyke et al. (1992) and the mutated CDAs were purified by metal chelate affinity chromatography on iminodiacetic acid Sepharose 6B charged with nickel (Ni21–IDA). Since the mutant proteins in most instances were catalytically inactive, the eluted fractions were analyzed by the Bradford protein assay (Bradford, 1976). The final preparation was dialyzed against 50 mM Tris–HCl, pH 7.5, 5 mM β-mercaptoethanol, 1 mM EDTA, concentrated by ultrafiltration on PM-10 membranes (Amicon) and analyzed by SDS–PAGE. CDA assay Spectrophotometric assay. The decrease in absorbance of the reaction mixture was followed at 37°C at an appropriate wavelength as described previously (Cacciamani et al., 1991). One enzyme unit is defined as the amount of enzyme which catalyzes the deamination of 1 µmol of CR per minute at 37°C. HPLC assay. To estimate the very low CDA activity of some of the mutant enzymes, the following procedure was used. Mixtures of 1 ml, containing 100 mM Tris–HCl, pH 7.5, 100 mM KCl, mutant enzyme (3 µg) and 0.167 mM CR, were incubated at 37°C. At various time intervals, aliquots of 100 µl were deproteinized by precipitation with 0.4 M HClO4, centrifugation at 14 000 g for 5 min, neutralization with 0.167 M K2CO3 and further centrifugation at 14 000 g for 5 min. The supernatant (100 µl) was injected on to a reversedphase LC-18 5 µm HPLC column (25034.6 mm i.d.; Supelco) equilibrated with 100 mM potassium hydrogenphosphate, pH 6, 8 mM tetrabutylammonium hydrogensulfate and eluted with the same buffer (1.3 ml/min). CR and uridine (UR) showed retention times of 3.4 and 4.8 min, respectively. Elution of the substrate and product was monitored at 254 nm and the integrated peak areas of the substrate and product were

Site-directed mutagenesis on human cytidine deaminase

Fig. 2. Release of zinc from wild-type CDA during titration with the mercurial reagent PMPS. After each addition of PMPS, the absorbance at 500 nm (j) due to the reaction of zinc with PAR was monitored. The enzymatic activity (d) was calculated after each addition of PMPS. The absorbance of PMPS at 250 nm is not shown in this plot.

compared with those of standard samples of known concentration. Other analytical procedures The molecular mass and the purity of wild-type (wt) and mutant His6–CDA were estimated by SDS–PAGE (Laemmli, 1970) using 15% acrylamide; the markers used were lowrange from Bio-Rad. Proteins were stained with Coomassie Brilliant Blue or silver. Sulfhydryl group titrations on CDAwt were performed using specific thiol reagents such as DTNB (Habeeb, 1973), PMPS (Giedroc and Coleman, 1986) and PCMB (Boyer, 1954). The amount of zinc released during PMPS titration on CDAwt was determined using PAR (Giedroc and Coleman, 1986) The metal-free CDA was obtained following extensive dialysis of 0.037 mg/ml of enzyme against 16 mM OP in the ratio 1 mol enzyme subunit/7000 mol of OP using dialysis tubes treated as described by Auld (1988). The zinc content of the purified proteins (20 µg) was determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) using a Jobin Yvon Model 24R instrument. Results Involvement of sulfhydryl groups in zinc binding Deduced from the cDNA, human CDA contains nine cysteine residues per subunit (Ku¨hn et al., 1993). Titration of the native enzyme with both PCMB and DTNB showed the presence of six free SH groups per subunit. After titration the enzyme was dialyzed extensively against metal free buffer and analyzed for zinc content. It showed the presence of 1 mole of zinc per mole of subunit. Six sulfhydryl groups per subunit were also available for titration by DTNB after denaturation in 8 M urea. Three additional cysteine residues were made DTNB reactive provided that denaturation with 8 M urea was performed in the presence of sodium borohydride as reducing agent, yielding a total of nine SH groups per subunit. Titration of the enzyme with the strong dissociating sulfhydryl reagent PMPS revealed nine SH groups per subunit both by following the increase in absorbance at 250 nm and by following the loss of enzyme activity during the titration. The displacement of thiol-coordinated zinc by PMPS probably promotes the release of zinc atom which was monitored spectrophotometrically at 500 nm (Figure 2) by performing the PMPS titration in the presence of the high-affinity metal indicator dye PAR. As shown in

Figure 2, all nine sulfhydryl groups present in each subunit have to be titrated before all the zinc is released. EEDQ inhibition studies on wild-type CDA In preliminary studies it was found that incubation of CDA with the specific carboxyl reagent EEDQ resulted in progressive inhibition of CDA. To investigate the inhibition further, the Km and Vmax values were determined in the presence of different concentrations of the inhibitor (data not shown). EEDQ decreased the Vmax value for CR deamination but did not affect Km significantly, suggesting that binding of EEDQ to wild-type CDA did not modify the affinity of CDA for its substrate but rather inactivated the enzyme. In order to calculate the number of inhibitor molecules which can react per active site of CDA, the enzyme was incubated with EEDQ at different fixed concentrations for different periods of time and the halftime of inactivation (t1/2) was determined for each inhibitor concentration (Figure 3A). By plotting log(1/t1/2)against log(EEDQ concentration) a straight line was obtained with a slope corresponding to the number of EEDQ molecules reacting with each active site to form an inactive complex (Levy et al., 1963). As shown in Figure 3B, the slope was 0.89, suggesting that the binding of one molecule of EEDQ per active site is sufficient to inactivate the enzyme. Moreover, it was found that addition of the nucleophile dithiothreitol (DTT) after EEDQ stopped the course of inactivation but did not restore lost enzyme activity (data not shown). Construction, complementation test and expression of mutant CDAs Alignment of the deduced amino acid sequences of the E.coli and the human CDA (Betts et al., 1994) suggested Cys65, Cys99 and Cys102 of the human enzyme as putative zinc coordinating residues and Glu67 as the essential Glu residue corresponding to Glu104 of the E.coli enzyme. To establish the importance of these residues for the activity of the human enzyme, they were individually changed by site-directed mutagenesis. Cys65, Cys99 and Cys102 were all changed to Ala and, in addition, Cys65 was mutated to a His residue, creating a constellation of zinc coordination residues similar to that of the E.coli enzyme. The conserved Glu67 residue was mutated to both Asp and Gln. Moreover, the unique His residue found in human CDA, His35, was mutated to Gln, to test the importance of this residue in catalysis. The mutated cDNAs were cloned into pTrc99-A as described in Materials and methods and the resulting plasmids were transformed into SØ5201. The pyrimidine requirement of SØ5201 cannot be satisfied by 29-deoxycytidine (CdR) owing to lack of CDA activity caused by the cdd::Tn10 mutation. This forms the basis of the complementation test for plasmids encoding CDA. By testing SØ5201, containing each of the mutant recombinant plasmids, for growth on plates containing minimal medium with CdR and IPTG, it was observed that only strains carrying pTrcHUMCDA(E67D) and pTrcHUMCDA(H35Q) showed detectable growth. Strains harboring plasmids with the C65A, C65H, E67Q, C99A and C102A mutations showed no visible growth after 48 h at 37°C. The CDA activity present in crude extracts of the various strains confirmed the complementation data. SDS–PAGE of the same extracts revealed a certain difference between the various mutants in their level of expression of the CDA polypeptide, suggesting that some of the mutant proteins may be unstable in the cells (data not shown). 61

A.Cambi et al.

Fig. 3. Wild-type CDA activity at different concentrations of EEDQ for different periods of time. (r) 0, (j) 0.4, (m) 0.6, (u) 0.8 and (d) 1 mM EEDQ; n is the number of EEDQ molecules that react with each enzyme subunit. EEDQ was used as a methanolic solution and appropriate controls were run with methanol alone. EEDQ was preincubated for 10 min with wild-type CDA in 50 mM Tris–HCl, pH 7.5, 1 mM EDTA, 0.1 mM DTT, prior to adding the substrate.

Table II. Characteristics of purified wild-type and mutant CDAs. Enzyme

mol Zn21/mol CDA subunit

CDA wild-type H35Q E67D E67Q C65A C99A C102A

1.16 0.91 0.80 1.20 1.01 0.16 0.23

aOne

6 6 6 6 6 6 6

0.015 0.015 0.020 0.050 0.050 0.020 0.002

Activity (%)

Km (310–5 M)

Vmax (U/mg)a

Vmax/Km (3105)

100 100 0.45 0.00045 0 0 0

1.0 2.5 1.2 4.3 – – –

257 133 1.27 0.0015 – – –

257 53.2 1.06 0.00034 – – –

unit of enzyme activity is the amount of enzyme which catalyzes the deamination of 1 µmol of substrate in 1 min at 37°C and pH 7.5.

Purification and characterization of the mutant CDAs H35Q was the only mutant enzyme which could be purified by the procedure developed for the wild-type enzyme (Vincenzetti et al., 1996), involving chromatography on an affinity column. To facilitate the purification of the other mutant enzymes, the respective cDNAs were cut out of the pTrc99-A-based plasmids and cloned into pET-3d(His6) producing pHUMCDA-H6 plasmids. They were transformed into E.coli BL21(DE3) and expression was induced by IPTG. The fusion proteins were purified in one step from crude extracts by using an Ni21– IDA Sepharose 6B column and they were used without prior removal of the His tag, since there was no difference between the enzymatic characteristics of His-tagged and non-His-tagged wild-type CDA. The purity of the mutant proteins was checked by SDS–PAGE. After prolonged dialysis against metal-free buffer, the zinc content of each purified mutant CDA was determined by ICPOES and the results are shown in Table II. The H35Q, E67D, E67Q and C65A mutants contained about 1 mole of zinc per mole of subunit, like the wild-type enzyme, whereas C99A and C102A mutants contained only about 0.2 mole of zinc per mole of subunit. Table II also gives the CDA activities of purified wild-type and mutant enzymes. As shown, only the H35Q, E67D and E67Q mutant enzymes were found to be active, H35Q showing the same activity as wild-type, whereas E67D and E67Q showed 0.45 and 0.00045% of the wild-type CDA activity, respectively. The kinetic parameters for the 62

active mutant enzymes were determined by HPLC assay. The Km values are almost the same for the mutants and the wildtype CDA, whereas the Vmax and Vmax/Km values are very different for the four enzymes. In fact, the catalytic efficiency (Vmax/Km) was reduced approximately 260-fold for E67D and 800 000-fold for E67Q. Since both mutants contain a normal amount of zinc, the large difference between the CDA activities is most likely due to the lack of the appropriately localized carboxyl group of Glu67 required for protonating both the leaving amino group and N-3 of the pyrimidine ring during catalysis, as proposed by Betts et al. (1994) for the E.coli enzyme. Regarding the three cysteine mutant enzymes, they are all inactive although C65A maintains 1 mole of Zn21 per mole of subunit, suggesting a different role for these cysteine residues in zinc coordination. Discussion We have previously identified the human cytidine deaminase as a metalloprotein containing one zinc atom per enzyme subunit (Vincenzetti et al., 1996) and suggested that the metal was involved in catalysis (Vincenzetti et al., 1997a). From pH studies on human placenta CDA (Vincenzetti et al., 1997b), two ionizable groups were found, one of them having a pKa value of 3.8, suggesting the presence of a carboxylic group in the catalytic pocket. In this work we have demonstrated the importance of a COOH group by the inhibition exerted by the specific carboxylic reagent EEDQ. This inhibitor decreased the

Site-directed mutagenesis on human cytidine deaminase

Vmax value but did not affect the Km for cytidine significantly, suggesting that binding of EEDQ to wild-type CDA did not modify the affinity of the enzyme for its substrate. The large difference between the catalytic efficiency (Vmax/Km) of the E67D and E67Q mutants and that of the wild-type CDA suggested that the essential carboxylic group is provided by the Glu67 residue. Since the Km value is not greatly affected by these mutations, Glu67 seems not to be essential for substrate recognition. As shown for the E.coli enzyme (Betts et al., 1994), the deamination mechanism consists of an enzyme-assisted direct water attack at the C-4 position of the nucleoside substrate, forming a hydroxylic group that interacts with the zinc atom and the carboxylate group. Therefore, the carboxylic group and the zinc atom are both necessary for the catalytic event and have to be appropriately placed in the active site. The properties of the C65A mutant enzyme suggested that Cys65 is required for the enzymatic activity but not for the maintenance of zinc in the active site. Substitution of this cysteine with a histidine, as in the bacterial enzyme, produced a mutant protein, C65H, which was inactive like C65A (data not shown). Moreover, the H35Q mutant enzyme showed a content of zinc and kinetic properties similar to those of the wild-type, ruling out the His residue as a possible zinc ligand in human CDA. DTNB titration of the native enzyme showed that six of the nine SH groups present per subunit could be titrated and that they were not involved in zinc binding. The remaining three SH groups, probably buried in the catalytic pocket, were only made DTNB accessible after denaturation in presence of a strong reducing agent. Titration of all the nine SH groups with the strong dissociating sulfhydryl reagent PMPS in the presence of the high-affinity metal indicator dye PAR resulted in the release of 1 mole of zinc per mole of CDA subunit, suggesting involvement of cysteine residues in zinc binding. This was further confirmed by replacing the two cysteines of the conserved PCXXC region (Cys99 and Cys102) by alanine residues. The resulting mutant proteins, C99A and C102A, were completely inactive and exhibited a zinc content less than 25% of that found for the wild-type enzyme. It is plausible that the main forces holding zinc in the catalytic site are the coordinations to Cys99 and Cys102, whereas Cys65 is important in guiding the zinc ion to the correct position and orientation within the active site. When zinc was totally removed from the enzyme after extensive dialysis against OP, the zinc ion and the enzymatic activity were completely lost. The addition of exogenous zinc or cobalt did not restore the activity (data not shown).

Aubertin,A.M., Kirne,A. and Imbach,J.L. (1994) Antimicrob. Agents Chemother., 38,1292–1297. Grodberg,J. and Dunn,J.J. (1988) J. Bacteriol., 170, 1245–1253. Habeeb,A.F.S.A. (1973) Anal. Biochem., 56, 60–65. Ho,D.H.W. (1973) Cancer Res., 33, 2816–2820. Ku¨hn,K., Bertling,W.M. and Emmrich,F. (1993) Biochem. Biophys. Res. Commun., 190, 1–7. Laemmli,U.K. (1970) Nature, 277, 680–683. Levy,H.M., Leber,P.D. and Ryan,E.M. (1963) J. Biol. Chem., 238, 3654–3659. Miller,J. (1972) Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Mu¨ller,W.E.G. and Zahn,R.K. (1979) Cancer Res., 39, 1102–1107. Rivard,G.E., Momparler,R.L., Demers,J., Benoit,P., Raymond,R., Lin,K.T. and Momparler,L.F. (1981) Leuk. Res., 5, 453–462. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) In Nolan,C. (ed.), Molecular Cloning. A Laboratory Manual. 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sanger,F., Nicklen,S. and Coulson,A.R. (1977) Proc. Natl Acad. Sci. USA, 74, 5463–5467. Studier,F.W. and Moffatt,B.A. (1986) J. Mol. Biol., 189, 113–130. Van Dyke,M.W., Sirito,M. and Sawadogo,M. (1992) Gene, 111, 99–104. Verri,A., Focher,F., Priori,G., Gosselin,G., Imbach,J.L., Capobianco,M., Garbesi,A. and Spadari,S. (1997) Mol. Pharmacol., 51, 132–138. Vincenzetti,S., Cambi,A., Neuhard,J., Garattini,E. and Vita,A. (1996) Protein Express. Purif., 8, 247–253. Vincenzetti,S., Cambi,A., Balducci,E., Natalini,P., Volpini,R. and Vita,A. (1997a) Biochem. Mol. Biol. Int., 42, 469–476. Vincenzetti,S., Angeletti,M., Lupidi,G., Cambi,A., Natalini,P. and Vita,A. (1997b) Biochem. Mol. Biol. Int., 42, 477–486. Yang,C., Carlow,D., Wolfenden,R. and Short,A. (1992) Biochemistry, 31, 4168–4174. Received August 12, 1997; revised October 31, 1997; accepted November 13, 1997

References Auld,D.S. (1988) Methods Enzymol., 158, 110. Betts,L., Xiang,S., Short,S.A., Wolfenden,R. and Carter,C.W.Jr (1994) J. Mol. Biol., 235, 635–656. Bouffard,D.Y., Laliberte´,J. and Momparler,R.L. (1993) Biochem. Pharmacol., 45, 1857–1861. Boyer,P.D. (1954) J. Am. Chem. Soc., 76,4331–4337. Bradford,M.M. (1976) Anal. Biochem., 72, 249–254. Cacciamani,T., Vita,A., Cristalli,G., Vincenzetti,S., Natalini,P., Ruggieri,S., Amici,A. and Magni,G. (1991) Arch. Biochem. Biophys., 290, 285–292. Clark,D.J. and Maaløe,O. (1967) J. Mol. Biol., 23, 99–112. Coates,J.A.V., Cammack,N., Jenkinson,H.J., Mutton,I.M., Pearson,D.A., Storer,R., Cameron,J.M. and Penn,C.R. (1992) Antimicrob. Agents Chemother., 36, 202–205. Furman,P.A., Wilson,J.E., Reardon,J.E. and Painter,G.R. (1995) Antimicrob. Agents. Chemother., 6, 345–355. Giedroc,D.P. and Coleman,J.E. (1986) Biochemistry, 25, 4969–4978. Gosselin,G., Schinazi,R.F., Sommadossi,J.P., Mathe,C. Bergogne,M.C.,

63