Nucleoside-diphosphate kinase-mediated signal transduction via histidyl-aspartyl phosphorelay systems in Escherichia coli.

Nucleoside-diphosphate kinase (NDP kinase), a key enzyme in nucleotide metabolism, is also known to be involved in growth and developmental control and tumor metastasis suppression. Interestingly, we find that coexpression of NDP kinase with Taz1, a Tar/EnvZ chimera, in the absence of its native signal, can activate a porin gene ompC-lacZ expression in Escherichia coli. Further studies show that NDP kinase can act as a protein kinase to phosphorylate histidine protein kinases such as EnvZ and CheA which are members of the His-Asp phosphorelay signal transduction systems in E. coli. Instead of ATP, the exclusive phosphodonor for histidine kinases, GTP can be utilized in vitro in the presence of NDP kinase to phosphorylate EnvZ and CheA, which then transfer the phosphoryl group to OmpR and CheY, the respective response regulators. The direct involvement of GTP for the phosphorylation of EnvZ through NDP kinase was further demonstrated by the use of a mutant EnvZ, which lost ability to be autophosphorylated with ATP. Phospho-OmpR thus formed can bind specifically to an ompF promoter sequence. These results suggest that NDP kinase may play a physiological role in signal transduction.

Nucleoside-diphosphate kinase (NDP kinase) 1 is considered to play a key role in nucleotide metabolism to generate all nucleoside triphosphates from their corresponding nucleoside diphosphates using the ␥-phosphate from ATP or other (d)NTPs via a phosphoenzyme intermediate (1). The genes and structures of NDP kinases are highly conserved from Escherichia coli to human (43% identity) (2,3), and NDP kinase is believed to be a housekeeping enzyme essential for DNA and RNA synthesis (4).
Besides its role in nucleotide metabolism, NDP kinase is also involved in a number of cellular regulatory functions such as growth and developmental control and tumor metastasis suppression. Genetic analysis of NDP kinase from Myxococcus xanthus, a Gram-negative bacterium, has indicated that it may be essential for cell growth (5). The ndk gene from Schizosaccharomyces pombe was shown to regulate gene expression in sexual development in response to mating pheromone signaling (6). A null mutation of the awd gene, a Drosophila homologue of NDP kinase, results in abnormalities in larvae development (7). A cAMP receptor-stimulated NDP kinase activity was found in the cytoplasmic membrane of Dictyostelium dis-coideum which mediates the hormone action for the activation of G proteins (8). In addition, the genes of human and mouse NDP kinases/Nm23 have been shown to be involved in tumor metastasis suppression (9 -11). High metastatic potential of low nm23-expressing murine melanoma and human breast carcinoma cell lines was inhibited by transfection with nm23 cDNA. A human DNA-binding protein, PuF, identified as NDP kinase Nm23-H2, was shown to bind to the promoter region of c-myc in vitro and to activate c-myc transcription (12). Surprisingly, the ndk genes from E. coli, yeast, and S. pombe could be knocked out without affecting cell viability (6,13,14). However, the disruption of ndk, the only structural gene for this enzyme in E. coli, results in a mutator phenotype (13). It is highly interesting to investigate and better understand the precise molecular mechanisms of these diverse regulatory functions of NDP kinases in normal growth and development and tumor metastasis.
Recently, NDP kinase was found to exhibit a phosphotransferase activity by phosphorylating other cellular proteins. NDP kinase from rat liver was able to phosphorylate ATP-citrate lyase from PC12 cell cytosol on a histidine residue in vitro (15). Nm23/NDP kinase preparations from human, Drosophila, yeast, and Dictyostelium were also shown to exhibit a serine/ threonine-specific protein phosphotransferase activity when incubated with colon carcinoma cell lysate in the presence of urea (16). Since NDP kinase can be autophosphorylated with (d)NTP at a specific histidine residue to form a high-energy phosphorylated enzyme intermediate, it is intriguing whether the high-energy phosphate on NDP kinase can be transferred to other regulatory proteins inside the cell and whether such a phosphorelay through NDP kinase may be a cause for a number of unexplained roles of the enzyme discussed above.
In prokaryotes, histidine protein kinases play the major role in signal transduction required for adaptive responses to numerous environmental stresses (17). Recently, histidine protein kinases were also found in yeast (18,19) and in plants (20), suggesting that histidine protein kinases may prevail from prokaryotes to eukaryotes. During adaptation processes, the major players in E. coli are the sensor kinase (usually a transmembrane receptor) and the response regulator which mediates changes in gene expression and/or locomotion. The sensor kinase utilizes ATP to phosphorylate a histidine residue. Subsequently, the phosphoryl group is transferred to an aspartyl residue on the response regulator which causes a functional switch in the response regulator. Signal transduction is uniquely carried out by the reversible phosphorelay of a high energy phosphate between histidine and aspartate residues, which defines the histidyl-aspartyl signal transduction system (also known as the two-component signal transduction system, Ref. 21).
In the present study, we investigate whether NDP kinase can activate histidine protein kinases as an upstream phosphodonor in the signal transduction pathway. In E. coli, EnvZ is a trans-inner membrane histidine protein kinase in the EnvZ-OmpR phosphorelay signal transducing system that serves as an osmosensor in response to osmolarity, and CheA is a cytoplasmic histidine kinase required for controlling bacterial chemotaxis. Conserved features that are shared by members of the histidine protein kinase family include: the conserved amino acid residues, His(H), the autophosphorylation site, Asn(N), and two glycine-rich segments (G1 and G2) which are involved in nucleotide binding (22)(23)(24). EnvZ consists of a periplasmic sensor domain, transmembrane domain, and a cytoplasmic signaling domain. Once activated, EnvZ becomes autophosphorylated at a histidine residue (His-243) in the cytoplasmic signaling domain (25). EnvZ has the ability to act as both kinase and phosphatase to regulate the level of phosphorylated OmpR. OmpR receives the phosphate from phosphorylated EnvZ onto a conserved aspartate residue (Asp-55) (26,27). Following its activation, phospho-OmpR acts as a cytoplasmic transcription factor to bind upstream sites on porin promoters to differentially regulate the expressions of the outer membrane porin genes ompF and ompC (28,29). Although both NDP kinase and histidine protein kinases are phosphorylated at a specific histidine residue to form a high-energy phosphoenzyme, histidine protein kinases can only use ATP as the phosphodonor in contrast to NDP kinase which can use all (d)NTPs. This ATP limitation of histidine protein kinases may be overcome if phosphorylated NDP kinase could serve as the phosphodonor for histidine kinase.
Here, we demonstrate that NDP kinase can indeed mediate bacterial signal transduction by activation of a histidine protein kinase in the E. coli EnvZ-OmpR system in vivo and in vitro. Phosphorylation of histidine protein kinases such as E. coli EnvZ and CheA by NDP kinase using GTP as a phosphodonor can be observed, and the phosphorylated histidine protein kinases result in the transfer of the high-energy phosphate to their cognate response regulators. The present finding raises an intriguing possibility that NDP kinase may play an important physiological role under certain stress conditions acting as a phosphodonor for the His-Asp phosphorelay signal transducing systems.
In Vivo Activation of ompC-lacZ Expression-RU1012 cells were transformed with pKT8P3 and/or pYY0401 followed by plating onto lactose MacConkey agar plates (35) with or without the addition of 5 mM aspartate. Transformants were selected by using 50 g/ml ampicillin for pKT8P3, 25 g/ml chloramphenicol for pYY0401, or the addition of both ampicillin and chloramphenicol when plating cotransformants. Lac ϩ colonies were red, and Lac Ϫ colonies were white. Plates were incubated for 15 h at 37°C, and only portions of the plates are shown.
Protein Expression and Purification-EnvZ(C), containing a C-terminal fragment of EnvZ from residues Ile-179 to Gly-450, was expressed using a T7 expression system. BL21(DE3) strain was transformed with pET11a-EnvZ(C), and the production of EnvZ(C) was induced in the presence of 1 mM isopropyl-␤-D-thiogalactoside. EnvZ(C) was purified to homogeneity by a modified procedure described previously (24); the DE52 column was substituted with a Q-Sepharose ion exchange resin (Bio-Rad), and the Green A affinity column was substituted with a Blue Sepharose CL-6B chromatography column (Pharmacia Biotech Inc.). Proteins were further purified by hydroxylapatite chromatography (Bio-Rad), and S-100 Sephadex gel filtration (Sigma). Similarly, His6-EnvZ(C) and His6-EnvZ(C)⅐G1 proteins were purified through Ni 2ϩ -affinity chromatography performed on a Bio-Rad Econo system. The purity of the purified proteins was Ͼ95% as judged by Coomassie Brilliant Blue staining.
EnvZ(M) was expressed using pDR200 containing envZ under the control of the lpp promoter and purified as associated with the inner membrane (34). OmpR was purified to homogeneity according to the previously published method (36). Purified CheA and CheY were obtained from Dr. A. Stock (Robert Wood Johnson Medical School).
Phosphorylation of E. coli NDP Kinase-E. coli NDP kinase was purified as described in Ref. 32. The phosphorylated form of NDP kinase was generated by incubating 2 g of NDP kinase with 40 Ci of [␥-32 P]GTP for 15 min at 30°C in 20-l mixture containing 50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 50 mM KCl, 5% glycerol. Phospho-NDP kinase was separated from free [␥-32 P]GTP by gel filtration on a Sephadex G-50 column (Pharmacia) equilibrated in 50 mM KCl and 20 mM Tris-HCl, pH 8.0. Phosphoenzyme was collected at 50 l of each fraction from the column. Phosphoenzyme concentration was determined by the method of Bradford using a reagent purchased from Bio-Rad.
Gel Mobility Shift Assay-Oligonucleotides used for the assay contain the ompF Ϫ100 to Ϫ64 regulatory sequences (29). Oligonucleotides (5Ј-GATCCTTTTACTTTTGGTTACATATTTTTTCTTTTTGAAAC-3Ј and 5Ј-GATCGTTTCAAAAAGAAAAAATATGTAACCAAAAGTAAAA-G-3Ј) were annealed and labeled by incubation with the Klenow fragment in the presence of [␣-32 P]dGTP. The probe was purified using a Nuctrap (Stratagene) column to remove unincorporated nucleotides followed by ammonium acetate and ethanol precipitation. Binding reactions were carried out in binding buffer (50 mM Tris-HCl (pH 8.0), 50 mM KCl, 5 mM CaCl 2 , 5% glycerol, 1 mM DTT, and 0.2 g/reaction poly(dI:dC)). Purified proteins were combined (0.48 g of NDP kinase, 1.8 g of EnvZ(C), and 0.25 g of OmpR) using a 2:1 molar ratio of EnvZ(C):NDP kinase and incubated for 20 min at 37°C. ATP or GTP was added to a final concentration of 0.3 mM, and reactions were carried out for 50 min at 37°C. Labeled DNA was then added to each reaction using 5000 cpm/reaction, and samples were incubated for another 20 min at 25°C. The final reaction volume was 15 l. Samples were immediately loaded onto a 5% acrylamide/bisacrylamide (40:1.2) gel which was run in 1 ϫ TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA) buffer with recirculation at 25°C using 120 V. The dried gel was exposed to autoradiography overnight at Ϫ80°C.

RESULTS
In Vivo Activation of Signal Transduction through NDP kinase-To demonstrate whether NDP kinase can function as a protein kinase in E. coli, we examined whether coexpression of the ndk gene together with the taz1 gene can activate an E. coli EnvZ-OmpR signal transduction pathway in vivo. Taz1, a hybrid chimeric kinase consisting of the N-terminal Tar chemoreceptor domain and the C-terminal EnvZ signaling domain had been previously constructed to examine activation of the EnvZ signaling pathway because the natural ligand for EnvZ has not been identified (31). Taz1 has been shown to activate ompC-lacZ expression in response to aspartate which is the natural ligand for Tar (31). Interestingly, transformation of strain RU1012 (⌽ompC-lacZ, ⌬envZ::Km r ) with both plasmid pYY0401, harboring the taz1 gene, and plasmid pKT8P3, harboring the wild type ndk gene as shown in Fig. 1, can significantly activate ompC-lacZ expression in the absence of 5 mM aspartate, as evident from the red colony formation on lactose MacConkey agar plates (35). In contrast, cells with pYY0401(taz1) alone formed white colonies in the absence of aspartate and formed red colonies only in the presence of aspartate, and cells with pKT8P3(ndk) alone were unable to form red colonies either in the presence or in the absence of aspartate (Fig. 1). These results demonstrated that Taz1 can be activated by another mechanism rather than by transmembrane signaling with its natural ligand. Clearly, NDP kinase is responsible for this activation and the ompC expression observed above was through the EnvZ(C) signaling domain of Taz1.
Phosphorylation of OmpR by NDP Kinase and EnvZ(C) in the Presence of GTP-We next examined in vitro whether NDP kinase can directly phosphorylate the EnvZ(C) signaling domain of Taz1, which results in the concomitant phosphorylation of OmpR and activation of the ompC transcription in vivo. To avoid autophosphorylation of EnvZ(C) by [␥-32 P]ATP, [␥-32 P]GTP was used as the phosphate donor for the phosphorylation of NDP kinase in reaction mixtures. To assess the kinase activity of NDP kinase in the presence of [␥-32 P]GTP toward histidine protein kinases, the purified C-terminal domain of EnvZ, EnvZ(C), and a kinase Ϫ /phosphatase ϩ EnvZ(C) mutant, EnvZ⅐N347D(C) (24), were examined. As shown in Fig Taz1), or pKT8P3 (expressing Ndk), or both plasmids. Transformants were plated on MacConkey agar plates in the absence (Ϫ) and the presence (ϩ) of 5 mM aspartate. Expression of Taz1 or Ndk in the absence of aspartate resulted in Lac Ϫ phenotype (white colonies). Coexpression of Ndk and Taz1, however, resulted in a Lac ϩ phenotype (red colonies) even in the absence of aspartate. In the presence of aspartate, cells with pKT8P3 (ndk) were still white (Lac Ϫ ), while cells with pYY0401 (taz1) become red (Lac ϩ ).  8), which was also used as a phosphodonor for OmpR (lane 9). At the end of the incubation, 5 l of gel sample buffer (80 mM Tris-HCl (pH 6.8), 4% SDS, 4 mM ␤-mercaptoethanol, and 20% (v/v glycerol)) was added to each reaction mixture. Samples were heated to 37°C for 5 min and applied to 16% SDS-PAGE. Autoradiogram of phosphorylated proteins in dried SDS-PAGE gel is shown. Protein amounts were EnvZ(C), 0.9 g (lanes 3, 6, and 7); EnvZ⅐N347D(C), 0.9 g (lanes 10 -12); NDP kinase (Ndk), 0.37 g (lanes 1, 2, 5-7, 11, and 12); OmpR, 0.78 g (lanes 7 and 9); bovine serum albumin, 0.7 g (lane 2). Radioisotopes used were [␥- 32  In addition, the phosphorylated form of NDP kinase was isolated from the reaction mixture of NDP kinase and [␥-32 P]GTP through gel filtration to remove free [␥-32 P]GTP. When the isolated phospho-NDP kinase was incubated with EnvZ(C) in reaction buffer A at 30°C for 30 min, EnvZ(C) was found to be phosphorylated (Fig. 2B, lane 2), indicating that phosphotransfer occurred directly from phospho-NDP kinase to EnvZ(C).
To further test whether NDP kinase is able to phosphorylate EnvZ(C) mutant protein incapable of autophosphorylation, we used a nucleotide binding defective EnvZ(C) mutant protein, H6-EnvZ(C)⅐G1 with triple point mutations (G375A, G377A, and A379S) in the G1 domain of EnvZ(C) (23). As shown in  5 and 6). This result further excludes the possibility that EnvZ(C) was phosphorylated by [␥-32 P]ATP which might have been generated through NDP kinase and thus clearly supports the phosphotransfer from NDP kinase to EnvZ(C) using [␥-32 P]GTP. It is should be noted that when taz1 with the G1 mutations was coexpressed with ndk in pKT8P3, colonies on a MacConkey plate became reddish (not shown), indicating that a weak His-Asp phosphorelay can be established with EnvZ(C)⅐G1 in the presence of NDP kinase. However, taz1 with the G1 mutations when expressed alone gave white colonies on the plate.
To eliminate the possibility of any contaminating ADP which could be converted to [␥-32 P]ATP that in turn phosphorylates EnvZ(C) by NDP kinase in the presence of [␥-32 P]GTP, the purified NDP kinase was incubated in the presence of nonradioactive GTP to 25 M containing 5 Ci of [␥-32 P]GTP at 37°C for 30 min. By thin layer chromatography (TLC), no [␥-32 P]ATP formation was detected (Fig. 3C, lane 2), indicating that all the purified NDP kinase, carrier GTP, and radiolabel [␥-32 P]GTP used in the present study were free of ADP contamination. Even in the reaction mixture including purified EnvZ(C), NDP kinase, and 25 M [␥-32 P]GTP, no [␥-32 P]ATP was found to be generated (lane 3). In contrast, when exogenous 0.1 M ADP was added to the above reaction mixture, the [␥-32 P]ATP was clearly synthesized (lane 4). These results further demonstrated that the purified EnvZ(C), NDP kinase, and GTP used in the reaction mixture did not contain ADP.
We further examined the phosphorylation of a full-length EnvZ membrane preparation, EnvZ transfer between NDP kinase and EnvZ, phospho-EnvZ(M) was generated by [␥-32 P]ATP, and free [␥-32 P]ATP was completely removed by extensive washing followed by centrifugation. As shown in Fig. 4B, when NDP kinase was added to the phosphorylated EnvZ(M) preparation, NDP kinase was subsequently phosphorylated (lane 2). NDP kinase could not be phosphorylated when incubated with the supernatant of the EnvZ(M) preparation (lane 1), confirming that the phosphorylated EnvZ membrane preparation did not contain free [␥-32 P]ATP. These results indicate that the reversible transfer of a phosphoryl group between EnvZ and NDP kinase may occur through a direct protein-protein interaction. The complex formation between the purified H6-EnvZ(C) and NDP kinase was found by means of Ni 2ϩ -His6 tag affinity chromatography with the similar binding affinity of dimer formation of EnvZ(C) (data not shown).
Characterization of the Phosphorylation of EnvZ(C) by NDP Kinase-Divalent cations are generally required for bacterial histidine kinases involved in the His-Asp phosphorelay system. Therefore, we examined the effects of various divalent cations on the phosphorylation of EnvZ(C) by NDP kinase. As shown in Fig. 5, in contrast to an inhibitory effect which EDTA exerted on the phosphoryl transfer from NDP kinase to EnvZ(C) (lane 1), enhanced phosphorylation of EnvZ(C) was observed in the presence of Mg 2ϩ , Ca 2ϩ , and Mn 2ϩ . The addition of manganese resulted in the most enhanced phosphorylation of EnvZ(C) (lane 4). The strong preference for manganese to stimulate phosphorylation has been demonstrated for other bacterial histidine kinases such as the FrzE chemotaxis sensor (37), FixL nitrogen fixation sensor (38), and EnvZ osmosensor (39). An active site mutation of NDP kinase with replacement of histidine 117 to glutamine resulted in defective autophosphorylation of NDP kinase and led to a loss of phosphotransfer activity from NDP kinase to EnvZ(C) (data not shown). This result suggests that protein phosphotransfer activity is dependent on the phosphorylation of the catalytic site (His-117) of E. coli NDP kinase.
A time course of phosphorylation of EnvZ(C) is shown in Fig.  6. EnvZ(C) was incubated with NDP kinase prephosphorylated with 25 M [␥-32 P]GTP. Aliquots of the reactions were removed at the indicated time points and subjected to 16% SDS-PAGE. EnvZ(C) labeled with 32 P was quantitated by phosphor Image analysis. EnvZ phosphorylation with [␥-32 P]GTP by NDP kinase was linear for the first 30 min and reached a steady state. Phosphorylation of EnvZ(C) with 25 M [␥-32 P]ATP was about 3 times faster than that with NDP kinase and GTP under the condition used. Furthermore from the levels of the steady state phosphorylation, phosphorylation of EnvZ(C) with ATP was approximately 2.5-fold more effective than that with NDP kinase and GTP.
Phosphorylation of CheA by NDP Kinase in the Presence of GTP-In E. coli, there are at least 17 documented members of His-Asp phosphorelay signal transduction systems (40). In addition to the EnvZ-OmpR system, we also evaluated another system for its interaction with NDP kinase. CheA-CheY phosphorelay system is similar to the EnvZ-OmpR phosphorelay system. CheA is a cytoplasmic histidine kinase required for the chemotactic response (41). While NDP kinase cannot directly phosphorylate CheY (Fig. 7, lane 5 Phosphorylated Response Regulator OmpR Is Biologically Active-To demonstrate that phospho-OmpR generated via NDP kinase in the presence of GTP was biologically active and able to bind the ompF promoter region, a gel mobility shift assay was performed (Fig. 8) (lanes 2-4), or incubated in combinations of OmpR plus EnvZ or OmpR plus NDP kinase, specific protein-DNA complexes were not formed (lanes 5 and 6).

DISCUSSION
Prokaryotic signal transducing systems contain a large family of protein histidine kinases and their response regulators. These histidyl-aspartyl phosphorelay systems and signaling circuits enable bacteria to adapt to rapidly changing environ-ments (42). While extensive studies have been directed toward the phosphotransferase activity of histidine protein kinases, understanding of the regulation of histidine kinases is still incomplete. In this report, we demonstrate for the first time that NDP kinase can act as a protein kinase to activate bacterial histidine kinases in the His-Asp phosphorelay signal transduction system not only in vivo but also in vitro. We have also shown that, besides ATP, GTP can be used as a phosphate donor for the His-Asp phosphorelay signal transduction in the presence of NDP kinase. It is important to note that as listed below, one can exclude the possibility that NDP kinase converted ADP contaminated in the reaction mixtures used in the present experiments into [␥-32 P]ATP, which in turn phosphorylated EnvZ(C). (a) Although an EnvZ(C)⅐G1 mutant protein was unable to be phosphorylated with either ATP or GTP, it can be phosphorylated with GTP in the presence of NDP kinase. This result can only be explained by the protein kinaselike function of NDP kinase. (b) When phosphorylated NDP kinase free of nucleotides was mixed with EnvZ(C), EnvZ(C) was effectively phosphorylated, indicating that the phosphate group was directly transferred from phosphorylated NDP kinase to EnvZ(C). (c) When NDP kinase was incubated with [␥-32 P]GTP or NDP kinase was incubated with EnvZ(C) and [␥-32 P]GTP, these reaction mixtures were unable to generate any detectable [␥-32 P]ATP. In addition, after NDP kinase was incubated with [␥-32 P]GTP, the flow-through fraction obtained by centrifuging the reaction mixture with use of Microcon-3 (3-kDa molecular mass cutoff) was unable to phosphorylate EnvZ(C) (data not shown).
While EnvZ and CheA represent typical bacterial histidine protein kinases in the His-Asp phosphorelay system, there are two other classes of bacterial histidine kinases: phosphoenolpyruvate phosphotransferase systems and metabolite-activated histidine kinases such as NDP kinase. Phosphorelay between two histidine residues has been shown in the bacterial phosphoenolpyruvate:sugar phosphotransferase system (43) and within a single molecule in the case of the E. coli ArcB protein (44). The present results indicate that the histidinehistidine phosphorelay between NDP kinase and the C-terminal signaling domain of EnvZ is reversible and quite efficient.  5, 6, and 8). After incubation for another 30 s, 5 l of gel sample buffer was immediately added to each reaction mixture. Samples were then heated to 37°C for 1 min and applied directly to 18% SDS-PAGE. Protein amounts used were: CheA, 1.25 g; NDP kinase (Ndk), 0.7 g; CheY, 1.1 g. Radioisotopes used were [␥-32 P]GTP (6 Ci, 5000 Ci/ mmol) (lanes 1-6) and [␥-32 P]ATP (6 Ci, 6000 Ci/mmol) (lanes 7 and 8).
Autoradiogram of a dried SDS-PAGE gel is shown. This interaction appears to be specific, since a mutation in EnvZ(C) (Asn-347 3 Asp) blocks the phosphorelay reaction even if the histidine residue for the phosphorylation site remains intact. It should be noted that direct phosphorylation does not occur between NDP kinase and cognate response regulators for the His-Asp phosphorelay system such as OmpR and CheY.
While we have demonstrated that EnvZ and CheA histidine protein kinases can be phosphorylated by NDP kinase in the presence of GTP, it is also possible that NDP kinase could phosphorylate other histidine protein kinases as well. Crosstalk among members of the His-Asp phosphorelay has been documented in other E. coli systems such as CheA which can serve as a phosphodonor for NtrC (45), OmpR (46), and SpoOA (47), and EnvZ can serve as a phosphodonor for NtrC (48). This report now demonstrates that cross-talk can also exist between families of histidine kinases as well. Recently, NDP kinase from rat liver was found to phosphorylate ATP-citrate lyase from PC12 cell cytosol on a histidine residue (15). Nm23/NDP kinase preparations from various higher eukaryotic species were shown to phosphorylate proteins on serine and threonine residues when incubated with colon carcinoma cell lysate in the presence of urea (16), suggesting that NDP kinases can utilize the same type of high-energy phosphohistidine intermediate to not only phosphorylate protein histidine kinases in prokaryotic systems, but also to phosphorylate a number of proteins on histidine or on serine and threonine residues in higher eukaryotic systems. Together, these results provide biochemical evidence for the diverse role of NDP kinase in cellular regulation.
Besides its key function in synthesizing cellular (d)NTP for biosynthesis of DNA and RNA, NDP kinase also engages in many other important cellular and developmental functions in eukaryotes. There are at least two different functional properties for NDP kinase, one is NDP kinase activity-dependent such as the biosynthesis of (d)NTPs (1) and the regulation of signal transduction (8), the second property is NDP kinase activity-independent function such as DNA binding (12) and the inhibition of differentiation (49). Human NDP kinase homologue (Nm23) has been shown to play an important role in tumor metastasis suppression. However, the link between tumor metastasis suppression and NDP kinase activity still remains to be established (11). The present results suggest that NDP kinase activity is required for the activation of phosphorelay signal transduction system in E. coli since mutation of a catalytic histidine in the active site of NDP kinase resulted in a loss of phosphotransfer activity from NDP kinase to EnvZ(C).
NDP kinase has a broad substrate specificity enabling the utilization of all (d)NTPs (1). NDP kinase has a preference for GTP (50), and NDP kinase from M. xanthus was originally identified as a GTP-binding protein (5). In addition, NDP kinase has a 100-fold higher affinity for ATP than histidine kinases involved in the bacterial His-Asp phosphorelay signal transduction (51). ATP concentrations in E. coli range from 2-5 mM, and GTP concentrations are approximately 1 mM (52). Therefore, during certain stress conditions when ATP levels are too low to be used by a histidine kinase such as EnvZ, NDP kinase could still activate a His-Asp phosphorelay system by using either GTP or low levels of ATP. While the majority of NDP kinase has been localized to the cytosol, NDP kinase has recently been found to be membrane-associated (8,53,54), which supports its interaction with G proteins (8,55). Similarly, the proximity of NDP kinase may enhance its ability to interact with inner membrane-associated histidine kinases such as EnvZ. It is also possible that the localization of NDP kinase may change during exposure to various growth environments. Under starvation conditions, membranes from Dictyos-telium cells contain a cAMP receptor-stimulated NDP kinase which produces GTP from exogenous GDP, and in turn activates a G-protein signaling pathway (8). From the early growth phase to late stationary phase, there is a progressive increase in the level of the truncated form of NDP kinase which is found to be membrane-associated and to synthesize GTP preferentially in Pseudomonas aeruginosa (54). In the present work, we have shown that by coexpression of ndk and taz1, that His-Asp phosphorelay signal transduction involved in the osmoregulation of the porin protein synthesis can be activated to stimulate ompC-lacZ expression, and, furthermore, our results in vitro suggest that NDP kinase functions as a protein kinase to phosphorylate the cytoplasmic signaling domain of EnvZ. However, it still remains to be elucidated under which physiological conditions NDP kinase plays a role in the stress response and adaptation by activating a His-Asp phosphorelay signal transduction pathway.