The Gene for Nucleoside Diphosphate Kinase Functions as a Mutator Gene inEscherichia coli

J. Mol. Biol. (1995) 254, 337–341

COMMUNICATION

The Gene for Nucleoside Diphosphate Kinase Functions as a Mutator Gene in Escherichia coli Qing Lu1, Xiaolin Zhang2, Niva Almaula1, Christopher K. Mathews2 and Masayori Inouye1* 1

Department of Biochemistry Robert Wood Johnson Medical School, Piscataway, NJ 08854, USA 2

Department of Biochemistry and Biophysics, Oregon State University, Corvallis, OR 97331, USA

Nucleoside diphosphate (NDP) kinase is a key enzyme in the control of cellular concentrations of nucleoside triphosphates, and has been shown to play important roles in various cellular activities such as developmental control, signal transduction and metastasis in eukaryotic systems. In this study, the gene for NDP kinase of Escherichia coli (ndk) was disrupted and surprisingly found to be dispensable without any discernible effects on cell growth or morphology. However, a mutator phenotype was found in ndk-disruption strains; frequencies of spontaneous mutations to rifampicin resistance and nalidixic acid resistant significantly increased. A higher frequency in reversion mutations was observed with use of an amber mutation in the kanamycin-resistance gene in an ndk-disruption strain. Imbalance in dNTP pools, in particular a significant increase of the dCTP content was observed, which is likely to result in the higher spontaneous mutation rates. These results suggest that NDP kinase, although not essential, plays an important role in the appropriate balance of intracellular dNTP pools to maintain a high DNA replication fidelity. Strains with ndk− pykA− pykF− as well as ndk− scs− were constructed without any discernible effect on cell growth, indicating that there is yet another enzyme(s) catalyzing nucleoside triphosphate synthesis, in addition to NDP kinase, pyruvate kinases and succinyl CoA synthetase. 7 1995 Academic Press Limited

*Corresponding author

Keywords: nucleoside diphosphate kinase; mutator gene; dNTP pool; pyruvate kinases; succinyl CoA synthetase

Nucleoside diphosphate (NDP) kinase (EC 2.7.4.6) is a ubiquitous enzyme highly conserved from prokaryotes to higher eukaryotes. It catalyzes the synthesis of (d)NTPs from (d)NDPs (Parks & Agarwal, 1973) and is known as a housekeeping enzyme for maintenance of intracellular levels of all (d)NTPs and NTPs for DNA and RNA synthesis in the cell. The genes coding for NDP kinases cloned from bacteria to humans (Hama et al., 1991; MunozDorado et al., 1990; Lacombe et al., 1990; Fukuchi et al., 1993; Biggs et al., 1990; Rosengard et al., 1989; Kimura et al., 1990; Ishikawa et al., 1992; Teng et al., 1991; Steeg et al., 1988; Stahl et al., 1992) and three-dimensional structures of NDP kinases (Dumas et al., 1992; Williams et al., 1993; Chiadmi et al., 1993) all show remarkable similarity, indicating that NDP kinase evolution has been mostly conservative. In addition to maintenance of (d)NTP levels, NDP kinase has been suggested to have a number of other important cellular and 0022–2836/95/480337–05 $12.00/0

developmental functions. Studies of NDP kinase from Myxococcus xanthus, a Gram-negative bacterium, have suggested that the enzyme is essential for cell growth (Munoz-Dorado et al., 1990). The transcript level of NDP kinase is downregulated upon the starvation-induced developmental cycle of Dictyostelium discoideum (Wallet et al., 1990). NDP kinase in Drosophila melanoganster, Awd, is essential for differentiation of normal wing disks at the larval stage, and a null mutation of the awd gene is lethal for larvae at the third instar stage (Biggs et al., 1990). Human and mouse NDP kinases, Nm23, have been identified as a putative tumor metastasis suppressor (Rosengard et al., 1989; Leone et al., 1991). Transfection of nm23 cDNA was shown to suppress malignant progression in Drosophila and mammalian cells in vivo. A DNA-binding protein PuF, identified as an NDP kinase Nm23-H2 homologue, was shown to bind to the promoter region of c-myc in vitro, suggesting its regulatory role in oncogene expression (Postel et al., 1993). Association of NDP 7 1995 Academic Press Limited

338

Communication

Figure 1. A representation of construction of ndk-disruption mutants in E. coli. A 1.3 kb EcoRI fragment encoding the kanamycinresistance gene was inserted into the unique EcoRI site within the ndk gene on plasmid pKT8P3 (Hama et al., 1991) to construct plasmid pNdkDkm. A partial deletion of the ndk coding region was achieved by replacing the ndk fragment between BssHII sites from nucleotide position 73 bp on pKT8P3 with a 2.6 kb PstI fragment containing the chloramphenicol-resistance gene, which is derived from pINIII-ompA3-cmr (Ghrayeb et al., 1984). The plasmid thus constructed was designated pNdkDcm. pNdkDkm and pNdkDcm were linearized by BamHI-HindIII digestion or SmaI digestion, respectively. The resulting linearized fragments carrying the disrupted ndk genes were used for transforming E. coli JC7623 harboring helper pIE/PuvII (ndk+, ts) at 30°C to disrupt the chromosomal ndk gene by homologous recombination as previously described (March et al., 1988). Candidates for the chromosomal ndk disruption mutants were screened on LB agar plates containing either kanamycin (50 mg/ml) or chloramphenicol (25 mg/ml) at 42°C. The colonies thus obtained were then screened for ampicilin sensitivity (Aps ) to generate strains NA7623 (JC7623 ndk::km) and QL7623 (JC7623 ndk::cm) for ndk-disrupted mutants with the km r gene and cm r gene, respectively. B, BssHII; E, EcoRI.

kinase with signal transducing pathways mediated by GTP-binding proteins has been proposed (Ishikawa et al., 1992; Teng et al., 1991). NDP kinase may participate in multi-protein complexes of the DNA replication machinery. E. coli NDP kinase has been reported to be associated with several T4 phage-coded enzymes of dNTP synthesis (Allen et al., 1983) and this complex may be associated with the replication machinery. Some evidence supports the existence of such complexes in uninfected bacteria (Mathews, 1993).

ments to strain JC7623 without the helper plasmid were also performed to make ndk-disruption mutants. Southern blot analysis demonstrated that the chromosomal ndk genes of these strains were indeed disrupted and that they had no extra intact ndk gene (Figure 2). PCR analysis confirmed the disruption of the ndk gene for all the constructions described above (not shown). Hybridization using the same filter under low stringency conditions did not detect any other hybridizable DNA fragments, indicating that E. coli does not contain another gene homologous to ndk.

Characterization of ndk -disruption mutants In the present study, we examined the role of the gene for the E. coli NDP kinase (ndk) by chromosomal disruption and complementation. Surprisingly, ndk-disrupted mutants grew normally and showed no apparent morphological changes. However, in ndk-disruption strains (ndk− ) frequencies of spontaneous mutations were found to be significantly higher, indicating that ndk gene functions as a mutator gene. To make an ndk-null mutation in E. coli, a helper plasmid, pIE/PuvII, carrying the wild-type ndk gene and temperaturesensitive origin and ampicillin-resistance gene was first constructed. Using E. coli strain JC7623 (recBrecCsbcB) carrying this helper plasmid , two independent chromosomal ndk disruption strains were isolated through homologous recombination as shown in Figure 1; in NA7623 the kanamycin-resistance gene (km r ) was inserted into the unique EcoRI site within the ndk coding region; and in QL7623 the 236 bp BssHII fragment within the ndk coding region was replaced with the chloramphenicol-resistance gene (cm r ). Both strains grew well at 42°C without the helper plasmid as judged by their ampicillin sensitivity (not shown). Direct transformations with linearized drug-disrupted ndk frag-

Figure 2. Southern blot analysis of disruptions of the chromosomal ndk gene. The NruI digests of the genomic DNA from strains JC7623, NA7623, QL7623, QL1387 and DSN1 were hybridized with ndk gene fragment (NdeINruI fragment from plasmid pET11ndk: Almaula et al., 1995) labeled by random priming (Amersham). The 1.9 kb fragment agrees with the size expected from the ndk::km r construct as shown in Figure 1 (lane 2). Similarly, the mutants of ndk-disruption with the cm r gene, QL7623, QL1387 and DSN1, were also confirmed as shown in lanes 3, 4 and 5, respectively, yielding a 2.9 kb fragment. In both cases there was no band at 0.5 kb expected for the undisrupted chromosomal ndk gene (lane 1).

339

Communication

Table 1. Spontaneous mutation frequencies for ndk cultures grown in L-broth and M9 minimal medium No. of mutants/108 cells Mediuma

Strain +

JC7623 (ndk ) −

NA7623 (ndk ) QL7623 (ndk− )

LB M9 LB M9 LB M9

Rifampicinr 0.8520.32 1.4720.53 30.227.9 (35)b 32.126.1 (22) 24.526.4 (29) 40.0211.3 (27)

Nalidixic acidr 0.2620.13 0.2420.08 13.122.1 (50) 10.622.7 (44) 7.322.0 (28) 7.421.5 (31)

For routine screening of ndk phenotype, ten independent colonies of the ndk disruptants and their parent strains were inoculated at 37°C with agitation in 5 ml of L-broth medium and M9 minimal medium to near stationary phase. Then appropriate amounts of these cultures were plated on selective plates containing rifampicin (100 mg/ml) or nalidixic acid (30 mg/ml). They were also plated on non-selective LB plates to determine total viable cell counts after appropriate dilution. Mutation frequencies (mean values2S.D.) were calculated by dividing the drug-resistant cell count by the total cell count. a LB, L broth; M9 minimal medium. b Values in parentheses are the ratios of the mutation frequency of ndk mutants to those of the respective wild-type strains.

Both ndk-disrupted strains grew normally in both LB rich medium and M9 minimal medium and showed no apparent morphological changes in cell shape under the light microscope. It is also interesting to note that T4 phage were able to normally form plaques on the disrupted strains in contrast to the previous proposal that NDP kinase may be required for T4 DNA replication (Ray & Mathews, 1992). Cell extracts prepared from the wild-type and mutant cells were assayed for NDP kinase activity using (g-32P) transfer reaction from ATP to CDP (Hama et al., 1991). The activity was reduced to approximately 15% in both strains NA7623 and QL7623 when compared with the wild-type cells (not shown). Mutator phenotype of ndk -disruption mutants Since NDP kinase plays important roles in the maintenance of the intracellular dNTP pool, the present ndk mutation may perturb intracellular nucleotide pool size and result in mutator phenotype. The spontaneous mutation frequencies of ndk - strains to rifampicin resistance or nalidixic acid resistance were indeed significantly higher (20 to 50-fold) than that of ndk+ strains (Table 1). The addition of thymidine or adenosine (5 mM) to the cultures did not increase the spontaneous mutation frequencies of the wild-type strains (not shown). The mutator activity is not medium-dependent, since frequencies of spontaneous mutation in ndk mutants are high for both LB and M9 minimal medium cultures (Table 1). Note that two independent ndk− strains (NA7623 and QL7623) gave similar results. The higher spontaneous mutation frequencies of strain QL1387 (6.0 × 10−8 ) were reduced to the wild-type levels (0.35 × 10−8 ), when the ndk− strains were transformed with a plasmid carrying the ndk+ gene, clearly indicating that the mutator phenotype is due to the ndk gene. We also tested the reversion frequency of an

amber mutation in the kanamycin-sensitive gene. The plasmid pAKE63 contains an ampicillin-resistance gene as well as a mutated kanamycin-resistance gene in which the codon for Glu63 (GAG) was changed to an amber codon (TAG). This plasmid, pAKE63 (AprKms ) was transformed into the wild-type strain (JC7623) and the ndk-disruption strain (QL7623). The reversion frequencies of these strains from Kms to Kmr were then obtained by plating them on the L-broth plates containing 50 mg/ml kanamycin. The reversion frequencies in strain QL7623 (3.0 × 10−7 ) were approximately 20 times higher than that in strain JC7623 (1.5 × 10−8 ). Thirteen independent revertants were analyzed for their DNA sequences and in all of them the TAG stop codon had reverted back to GAG (Glu) caused by T:A to G:C transversion, suggesting that Glu at position 63 may be essential for kanamycin resistance. Pool size of (d)NTP in ndk+ and ndk− strains Since NDP kinase is considered a major enzyme for generating dNTPs and NTPs, mutation of NDP kinase may cause an intracellular pool size imbalance, which may in turn decrease replication fidelity (Hibner & Alberts, 1980; Kunz & Kohalmi, 1991). We examined the pool size of (d)NTP in the wild-type and ndk mutant strains. Indeed as shown in Table 2, compared with the wild-type, ndk mutant QL7623 showed imbalance of intracellular dNTP level, in particular, dCTP and dGTP levels are elevated about 20-fold and sevenfold, respectively, while the levels of dCTP and dTTP in a triple mutant, QL1387 (ndk− pykA− pykF− ) increase about threefold, although both strains still have mutator phenotype, while rNTP levels remain normal. These results suggest that NDP kinase may play an important role in maintaining the balance of dNTPs pool to keep a high level of DNA replication fidelity. Spontaneous mutation frequency in an ndk mutant

340

Communication

Table 2. Nucleoside triphosphate pools (pmol/108 cells) of wild-type and ndk mutants E. coli strain +

JC7623 (ndk ) QL7623 (ndk− ) HW1387 (ndk+pykA−pykF− ) QL1387 (ndk−pykA−pykF− )

dATP a

24 11 12 9

dCTP 53 1241 37 95

dGTP

dTTP

ATP

CTP

GTP

UTP

9 63 14 14

33 56 19 46

1600 1540 NDb ND

1830 2050 ND ND

2000 2000 ND ND

1940 1860 ND ND

a Each value reported represents the average of triplicate aliquots counted from the same DNA polymerase reaction mixture. Each pool size was determined in at least three separate experiments involving extracts prepared and analyzed from different cultures as described (Sargent & Mathews, 1987). dNTP measurements used the DNA polymerase based assay as described by North et al. (1980), and rNTP levels were determined by HPLC separation on a Whatman Partisil-1 SAX column followed by spectrophotometric quantification (Chen et al., 1977). Individual readings agreed within210%. b Not determined.

(strain QL7623) is greatly decreased in anaerobic cells compared with that in an aerobic environment (unpublished results). A similar phenotype, with lower frequencies of mutagenesis in an anaerobic environment, was observed in an E. coli MutT mutator strain (Fowler et al., 1994). At present, it is not known whether similar mechanisms are responsible for these effects of anaerobiosis on mutator phenotype. Dispensability of pyk A, pyk F and scs in an ndk -disruption strain Since pyruvate kinases are known to have a sufficiently broad specificity to generate all kinds of (d)NTPs (Saeki et al., 1974), we examined whether pyruvate kinases might account for the residual NDP kinase activity. For this purpose, the ndk::cm r gene fragment from strain QL7623 was introduced into a double pyk-deletion strain HW1387 (pykA− pykF−; Garrido-Pertierra & Cooper, 1983) carrying the temperature-sensitive helper plasmid (ndk+ ) by P1 transduction. A triple mutant strain designated QL1387 thus isolated was still able to grow normally at 42°C, indicating that pyruvate kinases are not complementing the ndk function. The NDP kinase activity of QL1387 was approximately 15% of that of HW1387 (not shown). Southern analysis of QL1387 is shown in Figure 2, lane 4, demonstrating that its chromosomal ndk was disrupted. In Pseudomonas aeruginosa nucleoside diphosphate kinase was shown to form a complex with succinyl CoA synthetase (Scs; Kavanaugh-Black et al., 1994), an enzyme of the tricarboxylic acid cycle, which catalyzes substrate-level phosphorylation to give ATP or GTP in the TCA cycle. The products of ndk and scs may cooperatively play an important role in intracellular energy metabolism. However, when we transduced the ndk::cm fragment by P1 phage to the strain TK3D18, a scs detection strain (Dscs; Rhoads et al., 1978), to make a double null mutation of ndk and scs genes as shown in Figure 2, lane 5, the resulting strain DSN1 (TK3D18 ndk::cm r ) grew normally without any discernible morphological change. These results indicate that E. coli possesses another enzyme having NDP kinase activity in addition to NDP kinase, pyruvate kinases and succinyl CoA synthetase. The present result, that ndk is dispensable in E. coli, suggests that there is a compensating

enzyme(s) catalyzing synthesis of (d)NTPs. We exclude pyruvate kinases and succinyl CoA synthetase as the compensating enzymes. The compensating enzyme yet to be identified seems to be sufficiently active to support cell growth, but is deficient in adequate maintenance of the (d)NTP pool. It may have a substrate specificity quite distinct from that of the ndk gene product. The observed mutator phenotype of ndk− strains supports the notion that the ndk product plays a role in maintaining the fidelity of DNA replication, although that role may be indirect, e.g. mediated through control of normal dNTP pool sizes (Hibner & Alberts, 1980; Kunz & Kohalmi, 1991). Imbalance of (d)NTP pools caused by the ndk disruptions is likely to be responsible for the higher spontaneous mutation frequencies in bacterial populations. A human ndk homologue, the nm23 gene, was identified as a metastasis suppressor gene in certain cell lines, while the biochemical function of the Nm23 protein, responsible for its suppressive effect on the metastatic process, remains unknown (De La Rosa et al., 1995). It would be interesting to examine whether disruption of the nm23 gene results in a mutator phenotype in human cell lines.

Acknowledgements The present work was supported by a grant from the National Institutes of Health (GM19043) to M.I. and NSF grants DMB 91-19854 and DMB 92-18618 to C.K.M. We are indebted to Drs N. Grinter, W. Epstein, J. Guest and J. Nishimura for sending us bacterial strains and plasmids. We thank Mr Yongho Kim for the construction of plasmid pAKE63.

References Allen, J., Lasser, G., Booth, J., Goldman, D. & Mathews, C. K. (1983). T4 phage deoxyribonucleotide synthesizing enzyme complex. J. Biol. Chem. 258, 5746–5755. Almaula, N., Lu, Q., Delgado, J., Belkin, S. & Inouye, M. (1995). Nucleoside diphosphate kinase from Escherichia coli. J. Bacteriol. 177, 2524–2529. Biggs, J., Hersperger, E., Steeg, P., Liotta L. & Shearn, A. (1990). A Drosophila gene that is homologous to a mammalian gene associated with tumor metastatis coded for a nucleoside diphosphate kinase. Cell, 63, 933–940.

Communication

Chen, S. C., Brown, P. R. & Rosie, D. M. (1977). Extraction procedures for use prior to HPLC nucleotide analysis using microparticle chemically bonded packings. J. Chromatog. Sci. 15, 218–221. Chiadmi, M., Morera, S., Lascu, I., Dumas, C., Le Bras, G., Veron, M. & Janin, J. (1993). Crystal structure of the Awd nucleotide diphosphate kinase from Drosophila. Structure, 1, 283–293. De La Rosa, A., Williams, R. L. & Steeg, P. S. (1995). Nm23/nucleoside diphosphate kinase: toward a structural and biochemical understanding of its biological function. BioEssays, 17, 53–62. Dumas, C., Lascu, I., Morera, S., Glaser, P., Fourme, R., Wallet, V., Lacombe, M., Veron, M. & Janin, J. (1992). X-ray structure of nucleoside diphosphate kinase. EMBO J. 11, 3203–3208. Fowler, R. G., Erickson, J. A. & Isbell, R. J. (1994). Activity of the Escherichia coli mutT mutator allele in an anaerobic environment. J. Bacteriol. 176, 7727-7729. Fukuchi, T., Nikawa, J., Kimura, N. & Watanabe, K. (1993). Isolation, overexpression and disruption of a Saccharomyces cerevesiae YNK gene encoding nucleoside diphosphate kinase. Gene, 129, 141–146. Garrido-Pertierra, A. & Cooper, R. (1983). Evidence for two distinct pyruvate kinase genes in E. coli K-12. FEBS Letters, 162, 420–424. Ghrayeb, J., Kimura, H., Takahara, M., Hsiung, H., Masui, Y. & Inouye, M. (1984). Secretion cloning vectors in Escherichia coli. EMBO J. 3, 2437–2442. Hama, H., Almaula, N., Lerner, C., Inouye, M. & Inouye, S. (1991). Nucleoside diphosphate kinase from Escherichia coli. Its overexpression and sequence comparison with eukaryotic enzymes. Gene, 105, 31–36. Hibner, U. & Alberts, B. M. (1980). Fidelity of DNA replication catalysed in vitro on a natural DNA template by the T4 bacteriophage multi enzyme complex. Nature, 285, 300–305. Ishikawa, N., Shimada, N., Munakata, Y., Watanaba, K. & Kimura, N. (1992). Isolation and characterization of a gene encoding rat nucleoside diphosphate kinase. J. Biol. Chem. 267, 14366–14372. Kavanaugh-Black, A., Connolly, D. M., Chugani, S. A. & Chakrabarty, A. M. (1994). Characterization of nucleoside-diphosphate kinase from Pseudomonas aeruginosa: complex formation with succinyl-CoA synthetase. Proc. Natl Acad. Sci. USA, 91, 5883–5887. Kimura, N., Shimada, N., Nomura, K. & Watanabe, K. (1990). Isolation and characterization of a cDNA clone encoding rat nucleoside diphosphate kinase. J. Biol. Chem. 265, 15744–15749. Kunz, B. & Kohalmi, S. (1991). Modulation of mutagenesis by deoxyribonucleotide levels. Annu. Rev. Genet. 25, 339–359. Lacombe, M. L., Wallet, V., Troll, H. & Veron, M. (1990). Functional cloning of a nucleoside diphosphate kinase from Dictyostelium discoideum. J. Biol. Chem. 265, 10012–10018. Leone, A., Flatow, U., Richter-King, C. M., Sandeen, M., Margulies, I., Liotta, L. & Steeg, P. (1991). Reduced tumor incidence, metastatic potential and cytokine responsiveness of nm23-transfected melanoma cells. Cell, 65, 26–34.

341 March, P. E., Lerner, G. E., Ahnn, J., Ciu, X. & Inouye, M. (1988). The Escherichia coli Ras-like protein (Era) has GTPase activity and is essential for cell growth. Oncogene, 2, 539–544. Mathews, C. K. (1993). The cell-bag of enzymes or network of channels? J. Bacteriol 175, 6377–6381. Munoz-Dorado, J., Inouye, M. & Inouye, S. (1990). Nucleoside diphosphate kinase from Myxococcus xanthus. I. Cloning and sequencing of the gene. J. Biol. Chem. 265, 2702–2706. North, T. W., Bestwick, R. K. & Mathews, C. K. (1980). Detection of activities that interfere with the enzymatic assay of deoxyribonucleoside 5'-triphosphates. J. Biol. Chem. 255, 6640–6645. Parks, R. E., Jr & Agarwal, R. (1973). Nucleoside diphosphokinase. In The Enzymes (Boyer, P. D., ed.), vol. 8, pp. 307–333, Academic Press, New York. Postel, E., Berberich, S., Flint, S. & Ferrone, C. (1993). Human c-myc transcription factor PuF identified as nm 23-H2 nucleoside diphosphate kinase, a candidate suppressor of tumor metastasis. Science, 261, 478–480. Ray, N. & Mathews, C. K. (1992). Nucleoside diphosphokinase: a functional link between intermediary metabolism and nucleic acid synthesis. Curr. Topics Cell. Reg. 33, 343–357. Rhoads, D. B., Laimins, L. & Epstein, W. (1978). Functional organization of the kdp genes of Escherichia coli K-12. J. Bacteriol. 135, 445–452. Rosengard, A., Krutzsch, M., Shearn, A., Biggs, J. M., Burker, E., Margulies, I., King, C., Liotta, L. & Steeg, P. (1989). Reduced Nm23/Awd protein in tumor metasis and aberrant Drosophila development. Nature, 342, 177–180. Saeki, T., Hori, M. & Umerezawa, H. (1974). Pyruvate kinase of E. coli. J. Biochem. 76, 631–637. Sargent, R. G. & Mathews, C. K. (1987). Imbalanced deoxyribonucleoside triphosphate pools and spontaneous mutation rates determined during dCMP deaminase-defective bacteriophase T4 infections. J. Biol. Chem. 262, 5546–5553. Stahl, J. A., Leone, A., Rosegard, A. M., Porter, L., Ritcher-King, C. & Steeg, P. S. (1992). Identification of a second human nm23 gene, nm23-H2. Cancer Res. 51, 445–449. Steeg, P. S., Bevilacqua, G., Kopper, L., Thorgeirsson, J., Talmudge, Liotta, E. L. & Sobel, M. (1988). Evidence for a novel gene associated with low tumor metastatic potential. J. Natl Cancer Inst. 80, 200–204. Teng, D., Engele, C. & Venkatesh, T. (1991). A product of the prune locus of Drosophila is similar to mammalian GTPase-activating protein. Nature, 353, 437–440. Wallet, V., Mutzel, R., Troll, H., Barzu, O., Wurster, B., Veron, N. & Lacombe, M. (1990). Dictyostelium nucleosdie diphosphate kinase highly homologous to Nm23 and Awd proteins involved in mammalian tumor metastasis and Drosophila development. J. Natl Cancer Inst. 82, 1199–1202. Williams, R. J., Oren, D. A., Munoz-Dorado, J., Inouye, S., Inouye, M. & Arnold, E. (1993). Crystal structure of Myxococcu xanthus nucleoside diphosphate kinase ˚ and its interaction with a nucleotide substrate at 2.0 A resolution. J. Mol. Biol. 234, 1230–1247.

Edited by N. Sternberg (Received 12 June 1995; accepted 20 September 1995)