Eur. J. Biochem. 268, 1972±1981 (2001) q FEBS 2001

Structural and catalytic properties and homology modelling of the human nucleoside diphosphate kinase C, product of the DRnm23 gene Muriel Erent1,*, Philippe Gonin1, Jacqueline Cherfils2, Pierre Tissier1,², Giuseppe RaschellaÁ3, Anna Giartosio4, Fabrice Agou5, Claude Sarger1, Marie-Lise Lacombe6, Manfred Konrad7 and Ioan Lascu1 1

Institut de Biochimie et GeÂneÂtique Cellulaires (UMR 5095), Centre National de la Recherche Scientifique et Universite de Bordeaux-2, France; 2Laboratoire d'Enzymologie et Biochimie Structurales, CNRS, Gif-sur-Yvette, France; 3Ente Nuove Tecnologie Energia Ambiente (ENEA) Casaccia, Section of Toxicology and Biomedical Sciences, Rome, Italy; 4Istituto Pasteur±Fondazione Cenci Bolognetti and Dipartimento di Scienze Biochimiche A. Rossi Fanelli, UniversitaÁ La Sapienza, Roma, Italy; 5Institut Pasteur, Unite de ReÂgulation Enzymatique des ActiviteÂs Cellulaires, Paris, France; 6INSERM U 402, CHU Saint-Antoine, Paris, France; 7Max-Planck-Institut fuÈr Biophysikalische Chemie, GoÈttingen, Germany

The human DRnm23 gene was identified by differential screening of a cDNA library obtained from chronic myeloid leukaemia-blast crisis primary cells. The over-expression of this gene inhibits differentiation and induces the apoptosis of myeloid precursor cell lines. We overproduced in bacteria a truncated form of the encoded protein lacking the first 17 N-terminal amino acids. This truncated protein was called nucleoside diphosphate (NDP) kinase CD. NDP kinase CD had similar kinetic properties to the major human NDP kinases A and B, but was significantly more stable to denaturation by urea and heat. Analysis of denaturation by urea, using size exclusion chromatography, indicated unfolding without the dissociation of subunits, whereas

renaturation occurred via a folded monomer. The stability of the protein depended primarily on subunit interactions. Homology modelling of the structure of NDP kinase CD, based on the crystal structure of NDP kinase B, indicated that NDP kinase CD had several additional stabilizing interactions. The overall structure of the two enzymes appears to be identical because NDP kinase CD readily formed mixed hexamers with NDP kinase A. It is possible that mixed hexamers can be observed in vivo.

Correspondence to I. Lascu, Institut de Biochimie et GeÂneÂtique Cellulaires (UMR 5095), Centre National de la Recherche Scientifique and Universite de Bordeaux-2, 1 rue Camille Saint-SaeÈns, 33077 Bordeaux, France. Fax/ Tel.: 133 5 56 99 90 11, E-mail: [email protected] or M. Konrad, Max-Planck-Institute for Biophysical Chemistry, Department of Molecular Genetics, Am Fassberg 11, 37077 GoÈttingen, Germany. Fax: 149 5512011718, Tel.: 149 5512011706, E-mail: [email protected]

Nucleoside diphosphate (NDP) kinases catalyse the reversible phosphorylation of nucleoside diphosphates to nucleoside triphosphates via a phosphohistidine intermediate [1,2]. The eukaryotic NDP kinases are also involved in the regulation of various functions including differentiation and cell proliferation [3±5]. The expression level of human NDP kinase A (product of the nm23-H1 gene) has been shown to correlate with the metastatic potential of some tumours and cell lines [5,6]. Human NDP kinase B (product of the nm23-H2 gene) binds to DNA and regulates transcription of the c-myc proto-oncogene [7±9]. NDP kinases A and B are the most abundant isoforms in humans [10]. Other human genes encoding NDP kinases have been identified. Nm23-H4 encodes a mitochondrial NDP kinase, with an amino-acid sequence 58±59% identical to those of the NDP kinases A and B [11,12] (Fig. 1). Recently the nm23-H5 gene, which is specifically expressed in testis, has been found to encode a protein devoid of NDP kinase activity [13]. The Nm23-H6 [14], Nm23-H7 (GenBank, accession number AF153191) and Nm23-H8proteins have not yet been characterized. Two NDP kinase-like domains are present in tandem in the putative product of the nm23H7 gene and three are present in that of nm23-H8. This situation is reminiscent of the remarkable tandem repeat of three such domains in dynein intermediate chain 1 [15]. These proteins display weak sequence similarity to the canonical NDP kinases (about 30% identical residues). For a recent discussion on human NDP kinases see Lacombe et al. [16].

Abbreviations: NDP, nucleoside diphosphate; NDP kinase CD, NDP kinase C with the first 20 N-terminal amino acids replaced by Met-Ala-Asn. Enzymes: NDP kinase (ATP: nucleoside diphosphate phosphotransferase; E.C. 2.7.4.6). *Present address: Department of Biophysics, Max-Planck-Institute for Medical Research, Heidelberg, Germany. ²Present address: Laboratoire de ReÂsonance MagneÂtique des SysteÁmes Biologiques, UMR5536 CNRS/Universite Bordeaux-2, France. Note: In this paper, the name NDP kinase C is used, consistent with the names of the two major isoforms, NDP kinases A and B. The gene name DRnm23 is not appropriate because `nm' means nonmetastatic, and this is not a property of this protein. The human NDP kinase from mitochondria (product of the nm23-H4 gene) is called NDP kinase D. In addition, for simplicity, residue numbering is as for NDP kinases A and B (see Fig. 1). Sequencing made it possible to correct the published sequence (lysine 114 was missing in [17], in GenBank accession no. Q13232and in Sequence 2 from the US patent 5817783). (Received 4 December 2000, revised 1 February 2001, accepted 2 February 2001)

Keywords: Drnm23; hybrid; nucleoside diphosphate kinase; oligomeric proteins; protein stability.

q FEBS 2001

Human NDP kinase C, product of the DRnm23 gene (Eur. J. Biochem. 268) 1973

Fig. 1. Sequence alignment for NDP kinases A, B, C and D. The secondary structures, based on the crystal structure of NDP kinase B (protein database file 1nue) are indicated. X, Active site residues.

The human DRnm23 gene was identified by differential screening of a cDNA library prepared from chronic myeloid leukaemia-blast crisis primary cells [17]. DRnm23 over-expression in 32Dc13 cells prevented differentiation induced by granulocyte colony-stimulating factor and induced apoptosis in these cells. The gene product has been shown to be present in the cytosol [18] or in the heavy membranes containing mitochondria [19]. In neuroblastoma cells, DRnm23 gene expression appears to be associated with differentiation. Synthesis of the cellular adhesion protein, integrin, seems to be regulated by DRnm23 [20]. Homologues of DRnm23 have been found in mouse (accession no. AAD38976) and zebrafish (accession no Q9PTF3), but not in yeast or in the invertebrates Drosophila and Caenorhabditis elegans. To gain insight into the molecular mechanisms of action of this protein, we studied the biochemical properties of NDP kinase C2, the product of the DRnm23 gene. The protein is < 65% identical to the NDP kinases A and B, the major isoforms of human NDP kinase, over the length of the sequence common to all these isoforms, with no gaps or insertions. The N-terminus of NDP kinase C is 17 amino acids longer than those of the other isoforms (Fig. 1). We describe here the enzymatic and stability properties of the recombinant, N-terminally truncated NDP kinase C, equivalent in size to NDP kinases A and B, and referred to as NDP kinase CD. We demonstrate that this protein is structurally similar to NDP kinase A, by the formation of mixed hexamers. Molecular modelling shows why mixed hexamer formation was possible and revealed probable structural determinants of the heat-stability of NDP kinase C.

started with the sequence Met-Ala-Asn, based on the N-terminal consensus sequences of human and mouse NDP kinases A and B (Fig. 1). Escherichia coli, BL21(DE3) strain, was cultured at 37 8C in 2YT medium in the presence of 200 mg´L21 ampicillin. At 1.2±1.5 D600, gene expression was induced by adding 1 mm isopropyl thio-b-d-galactoside and incubating for 4 h. The bacterial cell pellets were kept at 220 8C until use. Full-length protein was produced in very small amounts and was found to be insoluble. It was not studied further here. The N-terminally modified NDP kinase CD was expressed to . 50 mg´L21 culture and was soluble. It was purified by affinity chromatography on Cibacron blue Sepharose. Preliminary experiments showed that NDP kinase CD protein had a very high affinity for commercially available Cibacron blue Sepharose; elution was difficult and yields were low. Therefore we used a home-made affinity gel produced as described previously [22], with a low degree of substitution (1 mmol dye per g wet gel, measured as described [23]). The working pH was . 8.0, conditions in which the affinity of the protein for Cibacron blue Sepharose is lower [24] and elution was performed by reversing the flow to prevent multiple binding-dissociation equilibria. The crude bacterial extract (obtained by sonication) was clarified by centrifugation and applied to the Cibacron blue Sepharose column equilibrated with starting buffer (SB: 50 mm Tris/HCl pH 8.2, containing 1 mm dithiothreitol). The column was washed with SB, then with SB containing 0.2 m NaCl and again with SB. The enzyme was eluted with 3 mm ATP in SB. The enzyme preparation was dialysed against Tris/HCl buffer containing 50% glycerol and kept at 220 8C. Specific activity declined to < 50% in 4 months under these conditions.

M AT E R I A L S A N D M E T H O D S Expression and purification

Biochemical techniques

The full-length cDNA (provided by B. Calabretta, Kimmel Cancer Centre, Philadelphia, USA), was used as a template for the cloning by PCR of the NDP kinase C coding region, which was then inserted into the pJC20 expression vector, a pET derivative [21]. Both a full-length construct, encoding a 168 amino acid protein, and a truncated version, designated NDP kinase CD, were generated. The truncated protein lacked the first 20 amino acids of the N-terminus [17] and

Enzymatic activity was assayed using the coupled spectrophotometric assay [25] at 25 8C. Residual activity in heat or urea denaturation experiments was measured under the same conditions, quickly diluting the enzyme 100-fold into the assay mixture. The reaction was linear for several minutes, indicating that no reactivation occurred. Differential scanning calorimetry was performed with a MicroCal MC2 (MicroCal, Inc., Northampton, MA, USA) differential

1974 M. Erent et al. (Eur. J. Biochem. 268)

scanning calorimeter and analysed with Origin software. Proteins were dialysed overnight against the working buffer (50 mm Tris/HCl, 100 mm NaCl, 1 mm EDTA, 1 mm dithiothreitol, pH 8.0) and degassed. The reference cell was filled with dialysis buffer. The cells were kept under nitrogen pressure to prevent bubbling and were heated at a rate of 60 8C´h21. Thermograms were corrected by subtracting the instrumental baseline and normalized for protein concentration. Protein aggregation following heat denaturation produced an exothermic signal that prevented determination of the post-transitional baseline. Tm, the temperature of maximum heat capacity, was calculated by subtracting a linear baseline obtained by extrapolating the pre-transitional baseline to the end of the thermal transition. Protein concentration was estimated using an absorbance coefficient of 1.55 at 280 nm for a 1 mg´mL21 solution, calculated from the amino acid composition as described previously [26]. Molar enzyme concentration is given as subunit concentration, assuming a molecular mass of 17 166. The N-terminal methionine was absent, as shown by electrospray MS. This is a common feature of recombinant NDP kinases expressed in E. coli. In calorimetric experiments, molar enzyme concentration was expressed as hexamer concentration. Analytical ultracentrifugation Sedimentation equilibrium experiments were performed at 12 000 r.p.m., 4 8C in a Beckman Optima XL-A analytical ultracentrifuge, using an An-60 Ti rotor and a double-sector cell of path-length 12 mm. Enzyme samples (400 mL) were dialysed exhaustively against three changes of 500 mL 20 mm potassium phosphate pH 7.5 buffer containing 50 mm potassium chloride and 1 mm dithiothreitol. Equilibrium was checked by superimposing duplicate scans recorded at 4 h intervals. The experimental data were first fitted with a model for a single homogeneous species, according to the following equation [27]: c…r† ˆ c…ref † exp{‰m…1 2 v r†v2 =2RTŠ…r 2 2 r 2ref †} 1 d where c(r) is the concentration of the protein at radial position r, c(ref ) is the concentration of the protein at an arbitrary reference distance rref, m is the molecular mass, v is the partial specific volume of the solute, r is the density of the solvent, v is the angular velocity, R and T are the molar gas constant and the absolute temperature, respectively, and d is the base-line offset. The partial specific volume (0.732 mL´g21) was calculated from the amino acid composition of the protein, and the density of the buffer (1.005 g´mL21 at 4 8C) was determined from published tables [27]. After collecting data at sedimentation equilibrium, samples were centrifuged at 40 000 r.p.m. for 12 h to sediment the protein and radial scans were performed to obtain a baseline correction for each cell. In all cases, the inclusion of a second virial coefficient did not improve the fit, and therefore nonideality in the system was not detected. Antibodies Polyclonal antibodies against NDP kinase CD were raised in rabbits by Davids Biotechnologie (Regensburg, Germany). They were purified by affinity chromatography using a

q FEBS 2001

HiTrap NHS activated column (Amersham-Pharmacia) as described previously [28]. Cross-reacting antibodies were eliminated on columns containing immobilized E. coli NDP kinase and human NDP kinases A and B. The crude serum was passed through the three columns before the column containing immobilized NDPkinase CD. Molecular modelling A model of human hexameric NDP kinase C was built from the crystal structure of NDP kinase (protein database file 1 nue) by side chain replacement. All side chains could be fitted into the monomer and at subunit interfaces with minor graphical adjustment using turbo [29]. The NDP kinase C model was refined by 150 cycles of energy minimization using xplor [30]. A mixed NDP kinase hexamer containing one subunit of NDP kinase C and five NDP kinase B subunits was modelled by replacing one subunit in the NDP kinase B hexamer by one subunit of refined NDP kinase C. The hexamer was then refined by 150 cycles of energy minimization using x-plor. Cell culture The human neuroblastoma cell line LAN-5 [31] was maintained in RPMI 1640 supplemented with 10% fetal bovine serum at 37 8C in an atmosphere containing 5% CO2. Western blot analysis Cells were lysed in 50 mm Tris/HCl pH 7.4, 250 mm NaCl, 5 mm EDTA, 50 mm NaF, 0.1 mm Na3VO4, 0.1% Triton X-100, 1 mm phenylmethanesulfonyl fluoride and 10 mg´mL21 leupeptin. Cell debris was collected by centrifugation, supernatants were transferred to a fresh tube and electrophoresis buffer added. Polyacrylamide (10%)/SDS gel electrophoresis and protein transfer to a poly(vinylidene difluoride) membrane were carried out as described previously [32]. The polyclonal antibody against NDP kinase CD was used at a dilution of 1 : 500. The secondary horse radish peroxidase-conjugated anti-(rabbit IgG) (SantaCruzBiotechnology Inc., SantaCruz, CA, USA) was used at a dilution of 1 : 2000. Immunodetection was carried out using the ECL-Plus reagent kit (Amersham-Pharmacia) according to the manufacturer's instructions.

R E S U LT S A N D D I S C U S S I O N Catalytic properties and molecular size of NDPkinase CD The truncated protein was produced in large amounts in bacteria and was purified in one step only, to . 90% purity, as shown by SDS/PAGE (data not shown). The protein was fully active. Active site titration [25] showed that 0.86 mol phosphate was bound per mole protein. The kinetic parameters of NDP kinase CD were remarkably similar to those of NDP kinases A and B. The apparent Km for thymine-5 0 -diphosphate (TDP) (at a fixed concentration of 1.0 mm ATP) were 220, 125 and 100 mm for NDP kinase CD, NDP kinase A and NDP kinase B, respectively. The corresponding apparent Km for ATP (at a fixed concentration of 100 mm TDP) was 25, 45, and 40 mm, respectively, and kcat was 192, 166, and 107´s21, respectively. The kcat

q FEBS 2001

Human NDP kinase C, product of the DRnm23 gene (Eur. J. Biochem. 268) 1975

Fig. 2. Sedimentation equilibrium analysis for NDP kinase CD. The relative concentration of NDP kinase CD measured by absorbance at 280 nm (W) is plotted as a function of radial distance after sedimentation at 12 000 r.p.m., 4 8C. Initial concentration was 0.15 mg´mL21. The line shows the best-fit curve for an ideal monospecies. The distribution of residuals is shown.

values were obtained by varying both ATP and TDP at a constant ratio of 5.0. We have shown previously that the substrate specificity for a series of TDP analogues, modified in the 3 0 position, is also similar between the three NDP kinases [33]. Sedimentation equilibrium ultracentrifugation experiments indicated a monodisperse species corresponding to a molecular mass of 102 200 ^ 4200 Da (Fig. 2). The random dispersion of residuals indicates that this model was correct. NDP kinase CD is therefore hexameric in solution, a common feature of eukaryotic NDP kinases. This shows that the absence of the N-terminal 17 amino acids has no effect on the quaternary structure of the protein. In analysis by size-exclusion chromatography on a Superose 12 column, the NDP kinase CD elution volume suggested a lower-order oligomer (not shown). Other mammalian NDP kinases behave abnormally in size-exclusion chromatography. For example, NDP kinase B was claimed to be a tetramer [34] but other studies in solution [9] and crystallographic data indicate a hexameric structure [35,36]. We assume that the aberrant elution from the size exclusion column is due to interaction of the protein with the gel matrix, and not to dissociation of the subunits. Subunit interactions generate a very stable hexamer. Indeed, NDP kinase CD unfolded in urea without the accumulation of dissociated species (see below). NDP kinase A, which is very similar, behaves normally in size-exclusion chromatography. For a recent discussion on the quaternary structure of NDP kinases, see Lascu et al. [37]. Thermal stability NDP kinase CD appears to be remarkably thermostable. Samples of NDP kinase CD were incubated at various

Fig. 3. Thermal stability of NDP kinase CD. (A) Inactivation of NDP kinase CD (X), NDP kinase A (K), NDP kinase B (W) and NDP kinase CD renatured in the presence of an excess of catalytically inactive H118N mutant of NDP kinase A (O). Aliquots of enzymes (10 mg´mL21) were incubated at the temperatures indicated for 15 min in 50 mm phosphate buffer (pH 7.0) containing 1 mm EDTA, 1 mm dithiothreitol and 1 mg´mL21 BSA. Residual activity was measured under standard conditions at 25 8C. (B) Differential scanning calorimetry profiles with NDP kinase CD (right) and NDP kinase A (left).

temperatures and enzymatic activity was found to be abolished at 85 8C, whereas NDP kinases A and B were inactivated at 60±65 8C (Fig. 3A). Differential scanning calorimetry also demonstrated protein unfolding at 87.7 8C (Fig. 3B). Under the same conditions, the human NDP kinases A and B unfold at 58.1 and 61.0 8C, respectively. Stability to denaturation by urea NDP kinase CD denaturation and renaturation by urea, as followed by intrinsic tryptophan fluorescence and by enzymatic activity are shown in Fig. 4. Denaturation occurred at much higher urea concentration than renaturation. This has been noted previously with other hexameric NDP kinases [38]. The origin of this phenomenon was investigated by studying denaturation and renaturation by size-exclusion chromatography (Fig. 5A and B). Denaturation occurred without the accumulation of dissociated species. No transition was apparent between the native oligomer and the unfolded monomer. The native hexamer

1976 M. Erent et al. (Eur. J. Biochem. 268)

Fig. 4. Denaturation and inactivation by urea. (A) Unfolding of NDP kinase CD (X), NDP kinase A (W), NDP kinase B (A) and refolding of NDP kinase CD (O), followed by intrinsic protein fluorescence (lex 295 nm, lem 340 nm). (B) Inactivation/reactivation of NDP kinases. Symbols as in (A). The protein (10 mg´mL21) was incubated with the indicated concentration of urea overnight at 25 8C.

and the unfolded monomer had the same Stokes' radii. The protein was inactive in 8 m urea and had a fluorescence spectrum typical of an unfolded protein. When renaturation was studied by injecting the unfolded protein into the column equilibrated with a different urea concentrations a different picture emerged. The species appearing in the presence of a low urea concentration was identified as the folded monomer. The native monomeric species has the smallest Stokes' radius of all possible species and therefore its identification is unambiguous. The existence of one peak only, the position of which changed during the transition, is indicative of a rapid equilibrium between the unfolded and folded monomer (with respect to the time scale of the experiment). In the absence of urea or the presence of low concentrations of urea, the hexamer appeared as a separate peak. The existence of separate peaks indicates a slow equilibrium between the two species. The reaction is probably under kinetic control. This is reasonable because three second-order association steps are required to generate the hexamer from the native monomers. No association states, such as dimers or trimers, accumulated under these conditions. The urea concentrations at which the transition was observed are different from those shown in Fig. 4A, because the incubation time was much shorter (< 30 min rather than 20 h as in the equilibrium experiment). NDP kinase A behaved very differently from NDP kinase CD (Fig. 5C and D). Denaturation of the hexamer yielded a dissociated species. During renaturation, a species

q FEBS 2001

Fig. 5. Dissociation/unfolding and refolding/association followed by size-exclusion chromatography. Native (A) or urea-unfolded (B) NDP kinase CD (10 mg in 100 mL) was injected into a Sepharose 12 column equilibrated with 50 mm Tris/HCl pH 7.4 containing 1 mm dithiothreitol, 100 mm NaCl and urea, from 0 (front) to 8 m (back), in steps of 0.4 m. The flow rate was 0.5 mL´min21. A similar experiment was performed with NDP kinase A (C, denaturation; D, renaturation). The estimated elution volumes of the hexamer and of the native monomer are 12.5 mL and 15.3 mL, respectively, using external standards (NDP kinase from Dictyostelium and horse myoglobin, respectively).

with the same elution volume appeared, but persisted at lower urea concentrations. This species was not the native monomer because its elution volume was smaller than that of the native monomer (i.e. its size was larger). Detailed studies (Ph. Gonin, unpublished results) indicated that this non-native intermediate had the properties of a molten globule. Interestingly, in both the naturally occurring S120G mutant of NDP kinase A [39] and the S122P mutant of NDP kinase B [40], this intermediate form persisted even in the absence of urea. NDP kinase C appears to behave in a manner typical of hexameric NDP kinases, such as the Dictyostelium [38,41], Drosophila [42] yeast (P. Gonin, I. Lascu, P. Tissier, A. Giartosia, L. Giangiacomo, C. Sarger and M. Konrad, unpublished data) and probably the human NDP kinase D [12]: it unfolds without the accumulation of lower-order oligomers, but the native monomer accumulates as an intermediate in the refolding/association pathway. The increase in tryptophan fluorescence at intermediate concentrations of urea (Fig. 4A) indicates conformational changes. The same phenomenon has been observed with human mitochondrial NDP kinase D [12]. The presence of four tryptophan residues in NDP kinase CD makes structural interpretation difficult. Dictyostelium cytosolic NDP kinase, which contains a single tryptophan residue, shows no such fluorescence change [38,43]. In addition, the enzymatic activity of NDP kinase CD increased in the presence of low concentrations of urea. Similar results have been obtained with enzymes from thermophilic organisms and may indicate a favourable effect on enzymatic activity or greater mobility of part of the enzyme.

q FEBS 2001

Human NDP kinase C, product of the DRnm23 gene (Eur. J. Biochem. 268) 1977

Formation of mixed hexamers with NDP kinase A We investigated the structural similarity of NDP kinase CD to the well-studied NDP kinase A by analysing the formation of mixed hexamers of the two enzymes from the reactivation kinetics of urea-unfolded enzymes. The reactivation curve of NDP kinase CD was concentration-dependent (Fig. 6), indicating the importance of the association of inactive monomers to yield the active species. Preliminary experiments indicate that enzymatic activity is associated with hexameric structure. The specific activity of the native monomer was only 2±3% that of the hexamer (P. Gonin, unpublished data). To distinguish between the two enzymes, the specific activities of which are very similar, we used the H118N mutant of NDP kinase A. This protein is completely inactive because the histidine phosphorylated during the catalytic cycle is absent. On mixing unfolded NDP kinase CD with an excess of the unfolded H118N mutant of NDP kinase A and diluting in renaturation buffer, reactivation occurred much more rapidly than with NDP kinase CD alone, at the same protein concentration. Because of the large excess of inactive NDP kinase A, and assuming a similar rate of oligomer formation for the two proteins, the product is probably a mixture of inactive NDP kinase A homohexamer plus mixed hexamers containing one subunit of NDP kinase CD and five inactive NDP kinase A subunits. The reactivation experiments showed that the absence of the active site histidine has no major effect on hexamer assembly. This has been suggested based on in vivo evidence [44]. The incorporation of active NDP kinase CD into heterohexamers with NDP kinase A was also demonstrated by analysis of the renaturation product generated with the H118N mutant of NDP kinase A, which carries a His-tag at the N-terminus. A small amount of NDP kinase CD reconstituted with a large excess of this inactive enzyme was retained by a Co21 column (TALON, Clontec) whereas NDP kinase CD renatured alone was not. This confirms the physical association of the two species of subunit (data not shown). We also determined the thermal stability of the heterohexamers and found it to be more similar to that of NDP

Fig. 6. Formation of hybrids between NDP kinases CD and A followed by reactivation kinetics. The kinetics of reactivation of ureaunfolded NDP kinase CD were followed at a protein concentration of 12 nm (W), 176 nm (X) and at 12 nm in the presence of 470 nm of the inactive mutant H118N of NDP kinase A (A).

kinase A than to that of NDP kinase C (Fig. 3A). This indicates the importance of intersubunit interactions for hexamer stability (see below). Structure of the NDP kinase CD hexamer, model based on the structure of NDP kinase B NDP kinase structure has been conserved during evolution. Eukaryotic and bacterial NDP kinases display almost identical monomer folding despite having differences in assembly [45]. Eukaryotic NDP kinases are hexamers and Myxococcus NDP kinase is a tetramer, built by a different assembly of dimers assembled in an identical way. Moreover, the formation of stable hexamers incorporating different types of subunit indicates that the interfaces between the subunits are also well conserved. We used this information to model the structure of NDP kinase CD based on the structure of human NDP kinase B. Previous attempts to model human NDP kinase B from the structure of Dictyostelium NDP kinase have been successful, except for the side-chain conformation of a few amino acids in the C-terminal part, from which one amino acid is missing in the Dictyostelium NDP kinase [46]. Hexameric NDP kinase C and the mixed hexamer containing one NDP kinase C subunit and five NDP kinase B subunits were modelled from the crystal structure of human NDP kinase B. The atomic coordinates of the human NDP kinase A are not available yet. As human NDP kinases A and B differ exclusively in surface residues [35], we used the X-ray structure of NDP kinase B, rather than that of NDP kinase A, for comparisons throughout the analysis. No rearrangement of the main chain was necessary to accommodate the side chains of NDP kinase C Ê for Ca). In addition, side (root mean square was 0.45 A chain changes did not interfere with oligomer formation. We first analysed the number of internal hydrogen bonds in the monomer (Table 1). We found that the NDP kinase A /B monomer can potentially form three additional internal hydrogen bonds and one more salt bridge than the NDP kinase C monomer. Conversely, two additional hydrogen bonds and one more salt bridge could be formed at subunit interfaces in NDP kinase C. All are located at a cluster of sequence changes including Gly19Arg and Ser144Asp, which interact with each other and replace five to six water molecules in a buried interface cavity in NDP kinase B (Fig. 7). Human NDP kinase D also has an Arg in position 19 and an Asp in position 144. Its structure is known (protein database file 1ehw). However, the C-terminal part is missing from the electron density map, starting with residue 143 [12]. This is probably due to the absence in NDP kinase D of the consensus Tyr151±Glu152, present in all hexameric NDP kinases with the exception of mitochondrial NDP kinases. The interaction of Arg19 with Asp144 is not sufficient for the rigidity of the entire C-terminal segment. Hydrogen bonds and packing are major factors correlated with the thermal stability of proteins [47,48]. Our structural analysis suggests that the monomer of NDP kinase A /B is more stable than that of NDP kinase C because it forms more electrostatic interactions. However, subunit±subunit interactions stabilize the NDP kinase C hexamer to a larger extent than NDP kinase A /B because more hydrogen bonds and salt bridges are formed. In addition, a favourable

1978 M. Erent et al. (Eur. J. Biochem. 268)

q FEBS 2001

Table 1. Comparison of internal and interfacial interactions in NDP kinase A /B, NDP kinase C and in the mixed hexamer. The prime indicate residues from neighbouring subunits.

Interfaciala

NDP kinases A and B

NDP kinase C

Gly19

Arg19

Ser144 Cys145 Phe61 Internalb Asn69 Glu79 Gly87 Ser99 Glu127 Mixed hexamerc Gly19

None

None None

Asp144 Ser145 Tyr61 Hydrogen bond with main chain Ala69 One hydrogen bond plus one salt link with Arg34 Gln79 None Ser87 Hydrogen bond with main chain Ala99 Two hydrogen bonds with Arg6 Arg127 Arg19

Ser144 Cys145 a

Subunit±subunit interactions.

Asp144 Ser145 b

Salt bridge with Asp144 0 Hydrogen bond with main chain Replaces five to six water molecules in interfacial cavity Salt bridge with Arg19 0 Hydrogen bond with main chain Hydrogen bond with Arg49 None One hydrogen bond with Lys33 Replaces a buried water molecule None None Salt bridge with Asp144 0 Hydrogen bond with main chain Replaces five to six water molecules in interfacial cavity None Hydrogen bond with main chain (A /B)

Internal interactions. c Subunit±subunit interactions in the mixed hexamer.

Fig. 7. NDP kinase C has additional interface interactions. (A) Overall view of hexameric NDP kinases with the threefold axis vertical. Interface interactions present in NDP kinase C only are located at the dimer interface and are indicated by an asterisk. The N-terminus (amino acid 2) of one subunit is marked by N. (B) Comparison of the subunit±subunit interaction at residues 19 (chain A) and 144 (chain D) between NDP kinase A /B and NDP kinase C. The main trace for NDP kinase C is shown for clarity. The water molecules that hydrate the buried cavity in NDP kinase A /B are shown as dots.

entropic contribution is provided by the removal of the hydrated interface cavity [49]. In addition, residues Ile53, Asn69, Val89 and Ser131 (in NDP kinase A /B), which are located in a helices, are replaced by alanine in NDP kinase C. Alanine has a stronger tendency to form an a helix. This is consistent with experimental data that shows that no monomeric intermediate forms during urea denaturation of NDP kinase C whereas such an intermediate, in the form of a molten globule, is observed during urea denaturation of NDP kinase A [39]. Thus, the high thermal stability of NDP kinase C, like that of other NDP kinases, probably results from the stability of the hexamer rather than from the stability of the monomer [41]. Modelling also showed that interfacial sequence changes in the NDP kinase C hexamer, because they fill a buried cavity, are accommodated without structural rearrangements. This implies that hybrid hexamers can be formed readily, as is the case with the A and B subunits [10]. Three stabilizing interactions are retained between the C subunit and the A /B subunits in the mixed hexamer (Table 1) vs. 18 in the NDP kinase C hexamer. This may explain why no gain in stability is achieved by allowing NDP kinase C to renature in the presence of a large excess of inactive NDP kinase A. NDP kinase C is remarkably similar to the other eukaryotic NDP kinases. All residues previously found to be involved in substrate binding and catalysis [50,51] are present. Also, the residues known to be necessary for hexamer assembly are conserved: Pro96 [42], Pro101 [38], Lys31, Tyr151 and Glu152 [52,53]. Although an attractive hypothesis, the putative RGD motif (residues 105±107) cannot have a functional role in the interaction with integrins because the three amino acids are buried within the protein: they are not exposed to the solvent even after hexamer dissociation. Arg105 is essential for catalysis [50]. In addition, major changes in side chain conformation are also probably not possible. The conservative mutation to lysine has a dramatic effect on hexamer stability [50]. Another putative motif present in the NDP kinase sequence,

q FEBS 2001

Human NDP kinase C, product of the DRnm23 gene (Eur. J. Biochem. 268) 1979

a leucine zipper-like motif, is not consistent with the structure: leucines are not in helices. Although NDP kinase B binds DNA and stimulates c-myc gene transcription [3], the structural basis of this interaction differs from that of leucine zipper transcription factors.

Biological significance The catalytic properties of NDP kinase C are very similar to those of the human NDP kinases A and B, but NDP kinase C is considerably more stable. It is clear that thermostability is not a requirement for biological function, but merely a consequence of strong subunit interaction. Some differences between NDP kinases B and C appear to be similar to those between proteins from mesophilic and thermophilc organisms, such as a higher arginine/lysine ratio (8 : 10 in NDP kinase A, 9 : 13 in NDP kinase B, 16 : 6 in NDP kinase C), filling in of cavities and more amino acids in helices, with a marked preference for this secondary structure. It is probably not just one single difference that accounts for the considerable difference in thermostability between NDP kinases B and C, but rather the accumulation of several small contributions. Much less NDP kinase C than NDP kinases A and B, the major human isoforms, is produced in cells (M.-L. Lacombe, unpublished data). The involvement of this enzyme in the cellular metabolism of nonadenine nucleotides is probably insignificant. However, the situation may be different if the protein has a special location. In this case, it may be involved in the local synthesis of nucleoside triphosphates. NDP kinase C has a longer N-terminus than NDP kinases A and B. Almost all of the 17 extra amino acids are hydrophobic and might be organized as a transmembrane a helix, anchoring NDP kinase C in membranes. In all hexameric NDP kinase structures, the N-terminus is located at the hexamer surface. An anchoring function is therefore compatible with the protein's structure. NDP kinase C forms mixed hexamers with NDP kinase A in vitro. This is probably also the case in vivo, in which case, one or two subunits of NDP kinase C (anchored to membranes) may form heterohexamers with NDP kinases A and B, generating membrane-bound NDP kinase activity. Such activity was observed long ago, but no explanation for the interaction with membranes was provided. One important question was whether the N-terminal part of the full-length NDP kinase C is exposed towards the exterior of the hexamer. This is so in all crystal structures of hexameric NDP kinases, and in the model on NDP kinase CD as well (Fig. 7). In some such structures a few amino acids are not seen in the electron density map because of their mobility. A second key question concerned whether the N-terminal part was present in the mature protein, or cleaved, as in mitochondrial NDP kinases [12,54]. We studied NDP kinase C in neuroblastoma cells by Western blotting with specific polyclonal antibodies. Most of the NDP kinase C in neuroblastoma was longer than the recombinant NDP kinase CD (Fig. 8). At least part of the N-terminal extension was present in the mature protein. A faint band migrated to a position identical to that of NDP kinase CD. It is unknown whether proteolysis occurs in vivo, or is an artefact of sample preparation. Detailed analysis of the

Fig. 8. Western blot analysis of NDP kinase C levels in a neuroblastoma cell line LAN-5. Lanes 1, 2 and 3, 500 pg, 1 ng and 10 ng of recombinant NDP kinase CD; lanes 4 and 5, cell lysates from 75 000 and 150 000 LAN-5 cells in basal growth conditions. Numbers on the right indicate the molecular masses (kDa) of the marker proteins (full range Rainbow molecular weight marker, Amersham-Pharmacia). Immunodetection was carried out as described.

chemical structure and subcellular location of NDP kinase C in cells is underway.

ACKNOWLEDGEMENTS This work was supported in part by a fellowship from the Association pour la Recherche contre le Cancer (grant 9633 to I. L. and fellowships to M. E. and to P. G.) and by a PROCOPE programme (1995±1997). We are grateful to B. Calabretta (University of Philadelphia, PA, USA) for providing us with the DRnm23 cDNA and to E. SchininaÁ (University of Rome, Italy) for MS analysis.

REFERENCES 1. Parks, R.E. Jr & Agarwal, R.P. (1973) The nucleoside diphosphokinases. In The Enzymes 8, 307±334. 2. Lascu, I. & Gonin, P. (2000) The catalytic mechanism of nucleoside diphosphate kinases. J. Bioenerg. Biomembr. 32, 237±246. 3. Postel, E.H. (1996) NM23/Nucleoside diphosphate kinase as a transcriptional activator of c-myc. Curr. Topics Microbiol. Immunol. 213, 233±252. 4. Timmons, L. & Shearn, A. (2000) Role of AWD/nucleoside diphosphate kinase in Drosophila development. J. Bioenerg. Biomembr. 32, 293±300. 5. de la Rosa, A., Williams, R.L. & Steeg, P.S. (1995) Nm23/ nucleoside diphosphate kinase: toward a structural and biochemical understanding of its biological functions. Bioessays 17, 53±62. 6. Hartsough, M. & Steeg, P.S. (2000) Nm23/nucleoside diphosphate kinase in human cancers. J. Bioenerg. Biomembr. 32, 301±308. 7. Postel, E.H., Berberich, S.J., Flint, S.J. & Ferrone, C.A. (1993) Human c-myc transcription factor PuF identified as nm23-H2 nucleoside diphosphate kinase, a candidate suppressor of tumor metastasis. Science 261, 478±480. 8. Postel, E.H. (1998) NM23-NDP kinase. Int. J. Biochem. Cell Biol. 30, 1291±1295. 9. Agou, F., Raveh, S., Mesnildrey, S. & VeÂron, M. (1999) Single strand DNA specificity analysis of human nucleoside diphosphate kinase B. J. Biol. Chem. 274, 19630±19638. 10. Gilles, A.M., Presecan, E., Vonica, A. & Lascu, I. (1991) Nucleoside diphosphate kinase from human erythrocytes. Structural characterization of the two polypeptide chains responsible for heterogeneity of the hexameric enzyme. J. Biol. Chem. 266, 8784±8789. 11. Milon, L., Rousseau-Merck, M.F., Munier, A., Erent, M., Lascu, I., Capeau, J. & Lacombe, M.L. (1997) nm23-H4, a new member of the family of human nm23/nucleoside diphosphate kinase genes localised on chromosome 16p13. Hum. Genet. 99, 550±557. 12. Milon, L., Meyer, P., Chiadmi, M., Munier, A., Johansson, M., Karlsson, A., Lascu, I., Capeau, J., Janin, J. & Lacombe, M.L.

1980 M. Erent et al. (Eur. J. Biochem. 268)

13.

14.

15.

16.

17.

18.

19.

20.

21.

22. 23.

24.

25.

26.

27.

28.

(2000) The human nm23-H4 gene product is a mitochondrial nucleoside diphosphate kinase. J. Biol. Chem. 275, 14264±14272. Munier, A., Feral, C., Milon, L., Pinon, V.P., Gyapay, G., Capeau, J., Guellaen, G. & Lacombe, M.L. (1998) A new human nm23 homologue (nm23-H5) specifically expressed in testis germinal cells. FEBS Lett. 434, 289±294. Mehus, J.G., Deloukas, P. & Lambeth, D.O. (1999) NME6: a new member of the nm23/nucleoside diphosphate kinase gene family located on human chromosome 3p21.3. Hum. Genet. 104, 454±459. Ogawa, K., Takai, H., Ogiwara, A., Yokota, E., Shimizu, T., Inaba, K. & Mohri, H. (1996) Is outer arm dynein intermediate chain 1 multifunctional? Mol. Biol. Cell 7, 1895±1907. Lacombe, M.-L., Milon, L., Munier, A., Mehus, J.G. & Lambeth, D.O. (2000) The human Nm23/nucleoside diphosphate kinases. J. Bioenerg. Biomembr. 32, 247±258. Venturelli, D., Martinez, R., Melotti, P., Casella, I., Peschle, C., Cucco, C., Spampinato, G., Darzynkiewicz, Z. & Calabretta, B. (1995) Overexpression of DR-nm23, a protein encoded by a member of the nm23 gene family, inhibits granulocyte differentiation and induces apoptosis in 32Dc13 myeloid cells Proc. Natl Acad. Sci. USA 92, 7435±7439. Martinez, R., Venturelli, D., Perrotti, D., Veronese, M.L., Kastury, K., Druck, T., Huebner, K. & Calabretta, B. (1997) Gene structure, promoter activity, and chromosomal location of the DR-nm23 gene, a related member of the nm23 gene family. Cancer Res. 57, 1180±1187. Negroni, A., Venturelli, D., Tanno, B., Amendola, R., Ransac, S., Cesi, V., Calabretta, B. & RaschellaÁ, G. (2000) Neuroblastoma specific effects of DR-nm23 and its mutant forms on differentiation and apoptosis. Cell Death Differ. 7, 843±850. Amendola, R., Martinez, R., Negroni, A., Venturelli, D., Tanno, B., Calabretta, B. & RaschellaÁ, G. (1997) DR-nm23 gene expression in neuroblastoma cells: relationship to integrin expression, adhesion characteristics, and differentiation. J. Natl Cancer Inst. 89, 1300±1310. Clos, J. & Brandau, S. (1994) pJC20 and pJC40 ± two high-copynumber vectors for T7 RNA polymerase-dependent expression of recombinant genes in Escherichia coli. Prot. Expr. Purif. 5, 133±137. Scopes, R.K. (1994) Protein Purification, 3rd edn. Springer Verlag, New York. Chambers, G.K. (1977) Determination of Cibacron Blue F3gA substitution in blue Sephadex and blue dextran-Sepharose. Analyt. Biochem. 83, 551±556. Lascu, I., Porumb, H., Porumb, T., Abrudan, I., Tarmure, C., Petrescu, I., Presecan, E., Proinov, I. & Telia, M. (1984) Ionexchange properties of Cibacron Blue 3G-A Sepharose (Blue Sepharose) and the interaction of proteins with Cibacron Blue 3G-A. J. Chromatogr. 283, 199±210. Lascu, I., Pop, R.D., Porumb, H., Presecan, E. & Proinov, I. (1983) Pig heart nucleosidediphosphate kinase. Phosphorylation and interaction with Cibacron blue 3GA. Eur. J. Biochem. 135, 497±503. Gill, S.C. & von Hippel, P.H. (1989) Calculation of protein extinction coefficients from amino-acid sequence data Analyt. Biochem. 182, 319±326. Laue, T.M., Bhaivari, D.S., Ridgeway, T.M. & Pelletier, S.L. (1992) Computer-aided interpretation of analytical sedimentation data for proteins. In Analytical Ultracentrifugation in Biochemistry and Polymer Science (Harding, S.E., Rowe, A.J. & Horton, J.C., eds), pp. 90±125. Royal Society of Chemistry Press, London. Phang-Ba, V., Munier, A., de Lascu, I., Lajarte-Thirouard, A.-S., PreÂvot, S., Capeau, J. & Lacombe, M.-L. (1998) Preparation of specific polyclonal antibodies against protein isoforms: Application to nucleoside diphosphate kinase A. Science Tools from Amersham Pharmacia Biotech 3, 10±11.

q FEBS 2001 29. Roussel, A., Inisan, A.G. & Cambillaud, C. (1989) TURBO. AFMB and BioGraphics, Marseille, France. 30. BruÈnger, A.T., Karplus, M. & Petsko, G.A. (1989) Crystallographic refinement by simulated annealing: application to crambin. Acta Crystallogr. A45, 50±61. 31. Seeger, R.C., Danon, Y.L., Rayner, S.A. & Hoover, F. (1982) Definition of a Thy-1 determinant on human neuroblastoma, glioma, sarcoma, and teratoma cells with a monoclonal antibody. J. Immunol. 128, 983±989. 32. Baldi, A., Luca, A., Claudio, P.P., Baldi, F., Giordano, G.G., Tommasino, M., Paggi, M.G. & Giordano, A. (1995) The RB2/ p130 gene product is a nuclear protein whose phosphorylation is cell cycle regulated. J. Cell. Biochem. 59, 402±408. 33. Gonin, P., Xu, Y., Milon, L., Dabernat, S., Morr, M., Kumar, R., Lacombe, M.L., Janin, J. & Lascu, I. (1999) Catalytic mechanism of nucleoside diphosphate kinase investigated using nucleotide analogues, viscosity effects, and X-ray crystallography. Biochemistry 38, 7265±7272. 34. Schaertl, S. (1996) Quaternary structure of human nucleoside diphosphate kinase isoforms HA and HB in solution. FEBS Lett. 394, 316±320. 35. MoreÂra, S., Lacombe, M.L., Xu, Y., LeBras, G. & Janin, J. (1995) X-ray structure of human nucleoside diphosphate kinase B Ê resolution. Structure 3, 1307±1314. complexed with GDP at 2 A 36. Webb, P.A., Perisic, O., Mendola, C.E., Backer, J.M. & Williams, R.L. (1995) The crystal structure of a human nucleoside diphosphate kinase, NM23-H2. J. Mol. Biol. 251, 574±587. 37. Lascu, I., Giartosio, A., Ransac, S. & Erent, M. (2000) Quaternary structure of nucleoside diphosphate kinases. J. Bioenerg. Biomembr. 32, 227±236. 38. Lascu, I., Deville-Bonne, D., Glaser, P. & VeÂron, M. (1993) Equilibrium dissociation and unfolding of nucleoside diphosphate kinase from Dictyostelium discoideum. J. Biol. Chem. 268, 20268±20275. 39. Lascu, I., Schaertl, S., Wang, C., Sarger, C., Giartosio, A., Briand, G., Lacombe, M.L. & Konrad, M. (1997) A point mutation of human nucleoside diphosphate kinase A found in aggressive neuroblastoma affects protein folding. J. Biol. Chem. 272, 15599±15602. 40. Schaertl, S., Geeves, M.A. & Konrad, M. (1999) Human nucleoside diphosphate kinase B (Nm23-H2) from melanoma cells shows altered phosphoryl transfer activity due to the S122P mutation. J. Biol. Chem. 274, 20159±20164. 41. Giartosio, A., Erent, M., Cervoni, L., Morera, S., Janin, J., Konrad, M. & Lascu, I. (1996) Thermal stability of hexameric and tetrameric nucleoside diphosphate kinases. Effect of subunit interaction. J. Biol. Chem. 271, 17845±17851. 42. Lascu, I., Chaffotte, A., Limbourg-Bouchon, B. & VeÂron, M. (1992) A Pro/Ser substitution in nucleoside diphosphate kinase of Drosophila melanogaster (mutation killer of prune) affects stability but not catalytic efficiency of the enzyme. J. Biol. Chem. 267, 12775±12781. 43. Deville-Bonne, D., Sellam, O., Merola, F., Lascu, I., Desmadril, M. & VeÂron, M. (1996) Phosphorylation of nucleoside diphosphate kinase at the active site studied by steady-state and time-resolved fluorescence. Biochemistry 35, 14643±14650. 44. Sellam, O., VeÂron, M. & Hildebrandt, M. (1995) Overexpression of wild-type and mutant NDP kinase in Dictyostelium discoideum. Mol. Microbiol. 16, 79±85. 45. Janin, J., Dumas, C., MoreÂra, S., Xu, Y., Meyer, P., Chiadmi, M. & Cherfils, J. (2000) Three-dimensional structure of nucleoside diphosphate kinase. J. Bioenerg. Biomembr. 32, 215±225. 46. Harrison, R.W., Chatterjee, D. & Weber, I.T. (1995) Analysis of six protein structures predicted by comparative modeling techniques. Proteins 23, 463±471. 47. Hubbard, J.S. & Argos, P. (1995) Evidence on close packing and cavities in proteins. Curr. Opin. Biotechnol. 6, 375±381. 48. Vaughan, C.K., Buckle, A.M. & Fersht, A.R. (1999) Structural

q FEBS 2001

Human NDP kinase C, product of the DRnm23 gene (Eur. J. Biochem. 268) 1981

response to mutation at a protein±protein interface. J. Mol. Biol. 286, 1487±1506. 49. Dunitz, J.D. (1994) The entropic cost of bound water in crystals and biomolecules. Science 264, 670. 50. Tepper, A.D., Dammann, H., Bominaar, A.A. & Veron, M. (1994) Investigation of the active site and the conformational stability of nucleoside diphosphate kinase by site-directed mutagenesis. J. Biol. Chem. 269, 32175±32180. 51. Xu, Y.W., Morera, S., Janin, J. & Cherfils, J. (1997) AlF3 mimics the transition state of protein phosphorylation in the crystal structure of nucleoside diphosphate kinase and MgADP. Proc. Natl Acad. Sci. USA 94, 3579±3583.

52. Karlsson, A., Mesnildrey, S., Xu, Y., Morera, S., Janin, J. & Veron, M. (1996) Nucleoside diphosphate kinase. Investigation of the intersubunit contacts by site-directed mutagenesis and crystallography. J. Biol. Chem. 271, 19928±20034. 53. Mesnildrey, S., Agou, F., Karlsson, A., Bonne, D.D. & Veron, M. (1998) Coupling between catalysis and oligomeric structure in nucleoside diphosphate kinase. J. Biol. Chem. 273, 4436±4442. 54. Troll, H., Winckler, T., Lascu, I., Muller, N., Saurin, W., Veron, M. & Mutzel, R. (1993) Separate nuclear genes encode cytosolic and mitochondrial nucleoside diphosphate kinase in. Dictyostelium discoideum. J. Biol. Chem. 268, 25469±25475.