Regulatory Mechanisms Differ in UMP Kinases from Gram-negative and Gram-positive Bacteria*

In this work, we examined the regulation by GTP and UTP of the UMP kinases from eight bacterial species. The enzyme from Gram-positive organisms exhibited cooperative kinetics with ATP as substrate. GTP decreased this cooperativity and increased the affinity for ATP. UTP had the opposite effect, as it decreased the enzyme affinity for ATP. The nucleotide analogs 5-bromo-UTP and 5-iodo-UTP were 5–10 times stronger inhibitors than the parent compound. On the other hand, UMP kinases from the Gram-negative organisms did not show cooperativity in substrate binding and catalysis. Activation by GTP resulted mainly from the reversal of inhibition caused by excess UMP, and inhibition by UTP was accompanied by a strong increase in the apparent Km for UMP. Altogether, these results indicate that, depending on the bacteria considered, GTP and UTP interact with different enzyme recognition sites. In Gram-positive bacteria, GTP and UTP bind to a single site or largely overlapping sites, shifting the T ⇄ R equilibrium to either the R or T form, a scenario corresponding to almost all regulatory proteins, commonly called K systems. In Gram-negative organisms, the GTP-binding site corresponds to the unique allosteric site of the Gram-positive bacteria. In contrast, UTP interacts cooperatively with a site that overlaps the catalytic center, i.e. the UMP-binding site and part of the ATP-binding site. These characteristics make UTP an original regulator of UMP kinases from Gram-negative organisms, beyond the common scheme of allosteric control.

Bacterial UMP kinases represent a particular subfamily of NMP 2 kinases (1, 2). They do not share any significant sequence homology with other known NMP kinases and exist in solution as stable hexamers. A first structural model of Escherichia coli UMP kinase (3) based on the conservation of the carbamate kinase and N-acetylglutamate kinase folds (4,5) helped to better rationalize previous site-directed mutagenesis experiments (6). The crystal structure of E. coli UMP kinase (7) indicated a similar fold between its monomers and N-acetylglutamate kinase, a dimeric enzyme (4,5). However, the quaternary structure assembly of these two proteins is completely different (7). Deposited crystal structures of UMP kinases from other bacteria such as Pyrococcus furiosus (8), Neisseria meningitidis (Protein Data Bank code 1YBD), Hemophilus influenzae (code 2AIF), and Streptococcus pyogenes (code 1Z9D) show threedimensional structures very similar to that of the E. coli enzyme. The residues essential for binding nucleotide substrates and catalysis are conserved among all bacterial UMP kinases ( Fig. 1) (3,9). Consequently, the active sites of these enzymes and the phosphoryl transfer mechanisms are most probably similar.
Comparison of the biochemical properties of recombinant UMP kinases from Gram-negative E. coli (1,2) and Gram-positive Streptococcus pneumoniae (10) indicated significant differences in their kinetic properties particularly in their regulation by nucleotides. Unlike the E. coli enzyme, UMP kinase from S. pneumoniae exhibited cooperative kinetics with respect to ATP, and its activation by GTP resulted in a decrease in cooperativity and an increase in affinity for ATP.
To substantiate and eventually extend these observations to other UMP kinases from Gram-negative or Gram-positive bacteria, the corresponding pyrH genes were cloned, and the recombinant proteins were studied for their kinetic properties in both forward and reverse reactions. Thus, GTP and UTP are so far effectors for all the investigated UMP kinases. They act on the kinetic parameters mostly via conformational changes induced in the protein. Consequently, the regulating effects of GTP and UTP on UMP kinases from both Gram-negative and Gram-positive organisms are strongly related to the quaternary structures of these proteins.

EXPERIMENTAL PROCEDURES
Chemicals-Nucleotides, restriction enzymes, T4 DNA ligase, Vent and Tfu DNA polymerases, and coupling enzymes were purchased from Roche Applied Science, New England Biolabs, Qbiogene Inc., or Sigma. UTP and UMP analogs halogenated at position 5 in the heterocycle were purchased from Jena Bioscience GmbH. NDP kinase from Dictyostelium discoideum (2000 units/mg of protein) was kindly provided by I. Lascu.
Bacterial Strains, Plasmids, Growth Conditions, and DNA Manipulations-General DNA manipulations were performed as described by Sambrook et al. (11). Open reading frames from the pyrH gene from different organisms (E. coli, Salmonella typhimurium, H. influenzae, N. meningitidis, Bacillus subtilis, S. pneumoniae, Staphylococcus aureus, and Enterococcus faecalis) were amplified from chromosomal DNA as template using the corresponding primers ( Table 1). The PCR products were inserted into the vector pET24a (between the NdeI and EcoRI restriction sites) or the vector pET28a (between the NdeI and XhoI or HindIII restriction sites) (Novagen). The resulting plasmids were introduced into strain BL21(DE3)/pDIA17 (12) to overproduce the UMP kinase. The recombinant strains were grown in 2ϫ yeast extract tryptone medium supplemented with kanamycin (70 g/ml) and chloramphenicol (30 g/ml) to an absorbance of 1.5 at 600 nm, and then overproduction was triggered by isopropyl ␤-D-thiogalactopyranoside induction (1 mM final concentration) for 3 h at 37°C. The cells were harvested by centrifugation and served as a source for protein purification.
The single mutants T135A and N137A and the double mutant T135A/N137A of B. subtilis UMP kinase were constructed by the one-tube PCR-based mutagenesis method (13) using the plasmid harboring the corresponding UMP kinase gene as template, Tfu DNA polymerase, the dNTPs, and the following mutagenic oligonucleotides: 3Ј B. subtilis UMP kinase T135A, 5Ј-GAAATATGGGTTTCCAGCGCCCGCA-GCGAAAAT-3Ј; 3Ј B. subtilis UMP kinase N137A, 5Ј-AGTT-GAGAAATATGGAGCTCCTGTGCCCGCAGC-3Ј; and 3Ј B. subtilis UMP kinase T135A/N137A, 5Ј-AGTTGAGAAAT- ATGGAGCTCCAGCGCCCGCAGCGAAAAT-3Ј. The PCR product was cloned at the NdeI and XhoI restriction sites of the pET28a vector. All plasmids were sequenced to verify either their integrity or the incorporation of the desired modifications.
Purification of UMP Kinases and Activity Assay-The different N-terminally His-tagged UMP kinases (E. coli D159N soluble variant, H. influenzae, N. meningitidis, B. subtilis, S. pneumoniae, S. aureus, and E. faecalis) overproduced in E. coli were purified by nickel-nitrilotriacetic acid affinity chromatography using the QIAexpress expression system (14). The recombinant proteins (purity Ͼ95% as indicated by SDS-PAGE) were stored at ϩ4°C in buffer (pH 8.0) containing 50 mM Na 2 HPO 4 , 150 mM imidazole, and 300 mM NaCl. Recombinant S. typhimurium UMP kinase was purified as described previously for wild-type UMP kinase from E. coli (1). Protein concentration was measured according to Bradford (15). Ion spray mass spectra of purified proteins were recorded on an API-365 quadrupole mass spectrometer (PerkinElmer Life Sciences) equipped with an ion spray (nebulizer-assisted electrospray) source. SDS-PAGE was performed as described by Laemmli (16).
UMP kinase activity was determined at 30°C using coupled spectrophotometric assays (0.5-ml final volume) on an Eppendorf ECOM photometer (17). The reaction medium in the forward direction contained 50 mM Tris-HCl (pH 7.4); 50 mM KCl; 1 mM phosphoenolpyruvate; 0.2 mM NADH; 2 units each of lactate dehydrogenase, pyruvate kinase, and NDP kinase; and various concentrations of MgCl 2 , ATP, and UMP. The UMP kinase appropriately diluted in 50 mM Tris-HCl (pH 7.4) was then added, and the decrease in absorbance was recorded at 340 nm. The reaction medium in the reverse direction contained 50 mM Tris-HCl (pH 7.4); 50 mM KCl; 1 mM glucose; 0.4 mM NADP ϩ ; 2 units each of hexokinase and glucose-6-phosphate dehydrogenase; and various concentrations of MgCl 2 , ADP, and UDP. The appropriately diluted UMP kinase was then added, and the increase in absorbance was recorded at 340 nm. One unit of UMP kinase corresponds to 1 mol of product formed per min.
The thermal stability of UMP kinases was tested by incubating the purified enzymes (1 mg/ml) in 50 mM Tris-HCl (pH 7.4 or 8.5) containing 0.1 M NaCl at a temperature between 30°C and 80°C for 10 min in the presence or absence of various nucleotides. The results (expressed as the percentage of residual activity compared with non-incubated controls) were used to calculate the temperature of half-inactivation (T m ).
Calculation of Magnesium⅐Nucleotide Complexes and Kinetic Data Analysis-The concentration of MgCl 2 in the assay medium in which coexisted nucleotides differing in the number of phosphate units was found to be critical for several reasons. The dissociation constant (K d ) of metal⅐nucleotide complexes varies within 2 orders of magnitude from 0.1 mM for MgNTP, 1 mM for MgNDP, and 20 mM for MgNMP (18,19). On the other hand, as some nucleotides played multiple roles, this resulted in mixed kinetic effects. Numerical simulations with different concentrations of MgCl 2 and nucleotides showed that an acceptable compromise in the forward reaction was to use a 2 mM excess of MgCl 2 above the concentration of NTPs. Thus, for the range of ATP (0.2-25 mM) and UMP (0.1-2 mM) concentrations used in most experiments, MgATP represented 95.7 Ϯ 0.9% of the total ATP, and magnesium-free UMP represented 89.6 Ϯ 2% of the total UMP (Table 2). Furthermore, the free metal ion (between 1.8 and 2.8 mM) was held at a sufficiently high but non-inhibitory concentration. For the sake of simplicity, the calculation of the kinetic constants in the forward reaction employed the actual concentration of various nucleotides. In this case, a K m or K 0.5 of 2 mM for ATP corresponds approximately to a K m for MgATP of 1.9 mM. Similarly, a K m for UMP of 0.1 mM corresponds approximately to a K m of 0.09 mM for magnesium-free UMP. In the reverse reaction, the concentration of MgCl 2 (millimolar) was related to the concentrations of UDP and ADP (or GDP when present) by the following relationship: [MgCl 2 ] t ϭ 4 ϩ 0.8 [NDP] t . Under these conditions, the concentration of MgNDPs represented 80% of the total nucleotide concentration, and the concentration of free magnesium cation was always 4 mM. When GMP or GMP-PNP was used, this relation changed as follows: [  The kinetic results were fitted to one of the following three equations by nonlinear least-squares fitting analysis using KaleidaGraph software (Equations 1-3), where v is the steady-state velocity, V m is the maximal rate, [S] is the substrate concentration (i.e. ATP or UMP in the forward reaction and ADP or UDP in the reverse reaction), K m is the Michaelis-Menten constant, K 0.5 is the substrate concentration at half-saturation, K I is the inhibition constant, and n (or n H ) is the Hill number (indicating the cooperativity index). The accuracy of the constants calculated by these fittings (on average, they varied within Ϯ10%) depended on the experimental errors (protein concentration and stability, purity of the commercially available nucleotides, and efficiency of the coupling enzymes in the assay system) and the computed concentration of the "active" metal-free or metal-complexed nucleotides from the corresponding dissociation constants.

RESULTS
Purification and Specific Activity of Recombinant UMP Kinases-Because we did not observe significant differences in the specific activities of the wild-type or His-tagged forms of E. coli (7), H. influenzae (this work), and B. subtilis (9) UMP kinases, the recombinant enzyme from the other bacterial species was overproduced with an N-terminal His tag and purified by affinity chromatography on nickel-nitrilotriacetic columns. We assumed that the His tag does not affect the activity of other bacterial UMP kinases. Gel permeation chromatography on Sephacryl S-300 and ultracentrifugation by sedimentation equilibrium confirmed that all variants exist as hexamers. Table 3 indicates the specific activity of UMP kinases from eight bacterial species at a single concentration of UMP (1 mM) and GTP (0.5 mM) and at two concentrations of ATP (2 and 8 mM). The highest concentration of nucleotides (8 mM ATP, 1 mM UMP, and 0.5 mM GTP) was selected arbitrarily to reach the maximal activity for all bacterial species. In Gram-negative organisms (E. coli, S. typhimurium, H. influenzae, and N. meningitidis), the ratio of UMP kinase activity in the presence and absence of GTP was practically independent of the concentration of ATP, whereas in Gram-positive organisms (B. subtilis, S. pneumoniae, S. aureus, and E. faecalis), the activating effect of GTP was much higher at a low concentration of ATP. Further kinetic analysis of UMP kinases from various species clarified the origin of this difference.
Dependence of UMP Kinase Activity on ATP Concentration-E. coli and S. typhimurium UMP kinases were shown to exhibit hyperbolic dependence of activity as a function of ATP concentration in both the absence and presence of GTP (1,20). The same was true for N. meningitidis UMP kinase. H. influenzae UMP kinase slightly deviates from this rule, as the kinetics with ATP as variable substrate were best fitted by the Hill equation. However, the n H values did not exceed 1.30 (Table 4). In the case of UMP kinases from the Gram-positive bacteria, the plot of activity versus the concentration of ATP was clearly sigmoidal in the absence of GTP, with n H varying from 1.7 for S. aureus, 2.0 for B. subtilis, and 2.5 for S. pneumoniae. In the presence of GTP, the cooperativity index decreased to almost 1.00, and the K 0.5 for ATP decreased by a factor of 3 for S. aureus and 8 for B. subtilis and S. pneumoniae. At saturating concen-

Concentrations (millimolar) of free (ATP f and UMP f ) and metal-complexed (MgATP and MgUMP) nucleotides as a function of their total concentrations (ATP o and UMP o ) and the total concentration of MgCl 2 (Mg o )
The dissociation constants (K d ) of MgATP and MgUMP complexes (0.1 and 20 mM, respectively) were taken from Alberty (18,19) and assuming that only the phosphate chain contributes to the strength of the metal⅐nucleotide complex.  trations of ATP, the V m values measured were similar with or without GTP (Table 4), in agreement with the previously published results on S. pneumoniae UMP kinase (10).
Dependence of UMP Kinase Activity on UMP Concentration-At pH 7.4 and higher, the activity of E. coli UMP kinase with UMP as variable substrate exhibits a biphasic behavior (1,9), which is best fitted by the equation v ϭ V m UMP/(UMP ϩ K m UMP ϩ UMP 2 /K i ). In the presence of GTP, the plot of activity versus the concentration of nucleoside monophosphate became hyperbolic. The calculated V m values using the two different plots are closely similar, suggesting that the major effect of GTP on E. coli UMP kinase with UMP as variable substrate is the reversal of inhibition caused by excess nucleoside monophosphate. The H. influenzae and N. meningitidis UMP kinases exhibited similar properties; nevertheless, for the latter enzyme, GTP increased also significantly the V m ( Table 5). The "V m " effect of GTP on N. meningitidis UMP kinase was also demonstrated by measuring its activity at several con- At saturating concentrations of ATP or in the presence of GTP, the Gram-positive B. subtilis and S. pneumoniae UMP kinases exhibited hyperbolic dependence of activity with UMP as variable substrate. In the absence of activator and at concentrations of ATP below the K 0.5 , inhibition by excess UMP was also observed (Table 5). Moreover, the K m for UMP increased significantly from low to saturating concentrations of ATP or in the presence of GTP, suggesting a complex relationship between the substrates and regulatory nucleotide. To minimize the effect of one substrate or effector on the kinetic parameters of the second substrate, the activity of B. subtilis UMP kinase  Specificity of Bacterial UMP Kinases for GTP as Activator-GTP appeared to be the common positive effector for all investigated bacterial UMP kinases (Tables 3-5). The concentration of nucleotide required for half-maximal activation (K a ) was independent of the concentration of Mg 2ϩ ions. At single concentrations of substrates (2 mM ATP and 1 mM UMP), the K a of UMP kinases from E. coli, H. influenzae, B. subtilis, and S. pneumoniae varied between 70 and 120 M. N. meningitidis UMP kinase exhibited a higher K a for GTP (ϳ300 M). When other guanine nucleotides or related compounds such as dGTP, 7-deaza-dGTP, 3Ј-anthraniloyl-dGTP, GMP-PNP, ITP, and XTP were tested as activators, a variety of effects were observed (data not shown). Thus, GMP-PNP and dGTP activated all forms of UMP kinases but to a variable extent and affinity compared with GTP. N. meningitidis UMP kinase was less sensitive to this activation by GMP-PNP than the other enzymes. GMP was ineffective on B. subtilis, S. pneumoniae, or N. meningitidis UMP kinase, but did activate H. influenzae or E. coli UMP kinase. 3Ј-Anthraniloyl-dGTP, a fluorescent analog of dGTP (9), was the strongest activator of B. subtilis and S. pneumoniae UMP kinases, with a 4-fold lower K a than for GTP, but was less effective on UMP kinases from the Gram-negative organisms (data not shown).
Inhibition of UMP Kinase Activity by UTP-One of the earliest observations regarding E. coli UMP kinase that made this enzyme unique among the other NMP kinases was the inhibition by UTP and its reversal by GTP or high concentrations of MgCl 2 (1). These results suggested that the true inhibitor of the bacterial enzyme was the magnesium-free UTP and that GTP acted as an antagonist of the former nucleotide. On the other hand, high concentrations of UMP partly protected the enzyme against inhibition by UTP (2). These observations were confirmed with UMP kinase from N. meningitidis (Fig. 3A) or H. influenzae (data not shown). The I 50 value for inhibition by magnesium-free UTP of the N. meningitidis enzyme at 0.05 mM UMP was 10 M. A 40-fold increase in UMP concentration shifted the I 50 to 130 M magnesiumfree UTP. Under the same experimental conditions, inhibition of B. subtilis (Fig. 3B) or S. pneumoniae (data not shown) UMP kinase by UTP was very little affected by high concentrations of UMP or Mg 2ϩ ions.
To better understand these differences between Gram-positive and Gram-negative species, the effect of UTP on individual kinetic constants was further investigated with E. coli, H. influenzae, and B. subtilis UMP kinases. A first series of experiments was conducted at constant concentrations of ATP (around the K m or K 0.5 values of individual enzymes) and UTP and at variable concentrations of UMP (Fig. 4, A-C). In the case of E. coli and H. influenzae UMP kinases, the curves converged at high concentrations of UMP (Fig. 4, A and B), in accordance with the  observed protective effect against inhibition by UTP of high UMP concentrations (2). Until a 0.1 mM concentration of UTP, i.e. ϳ4.5 M magnesium-free nucleotide, the apparent K m for UMP was almost unchanged. Above this concentration, a dramatic increase in the apparent K m for UMP was observed, which suggested a strongly cooperative effect. Thus, in the presence of 0.25 or 0.50 mM UTP, the apparent K m for UMP of E. coli UMP kinase increased by a factor of 6 or 34, respectively.
On the other hand, inhibition by excess UMP declined at high concentrations of UTP as indicated by the increase in the K I value. The cooperativity of inhibition by UTP of UMP kinases from Gram-negative organisms was expressed quantitatively by an equation similar to that describing the competitive inhibition: K m Ј ϭ K m (1 ϩ [UTP] n /K UTP n ). K m and K m Ј are the apparent K m UMP in the absence and presence of a given concentration of UTP, respectively; n is the cooperativity index; and K UTP is a constant that corresponds to the concentration of UTP doubling the apparent K m UMP . Transformed in a linear form, log 10 (K m Ј /K m Ϫ 1) ϭ n⅐log 10 [UTP] Ϫ log 10 K UTP n ; this equation allows calculation of the two constants. From the example described in Fig. 4A for E. coli UMP kinase, n ϭ 2.7 and K UTP ϭ 140 M (i.e. 6.3 M in terms of magnesium-free UTP). It is obvious that for n equal or close to unity, we return to the common figure of "noncooperative" competitive inhibition.
In the case of B. subtilis UMP kinase, the different curves obtained with UMP as variable substrate evolved in parallel with only a slight inhibition by excess UMP. The apparent V m and K m UMP decreased simultaneously, whereas their ratio remained almost constant (Fig. 4C). No increase in the apparent K m UMP was noticed even at the strongest inhibitory concentrations of UTP. With ATP as variable substrate, inhibition by UTP resulted in an increase in the apparent K m or K 0.5 for ATP. Thus, in the presence of 0.5 mM UTP, the apparent K m or K 0.5 for ATP increased by a factor of 4 for E. coli UMP kinase, a factor of 3.2 for H. influenzae UMP kinase, and a factor of 3.1 for B. subtilis UMP kinase (Fig. 4, D-F). GTP in excess of UTP restored the kinetic parameters of bacterial UMP kinases to the values observed in the absence of UTP (Table 6).
Among the UTP analogs tested as inhibitors, dUTP was five times weaker than the corresponding ribonucleotide, whereas TTP was completely ineffective even in the millimolar range.   E. coli (A and D), H. influenzae (B and E), and B. subtilis (C and F)  affinity than the reference nucleotide (Fig. 5). 5-Bromo-UTP and 5-iodo-UTP exhibited the most interesting effects. Thus, B. subtilis and S. pneumoniae UMP kinases were 5-10 times more sensitive to these nucleotides than to the parent nucleotide, whereas E. coli and H. influenzae UMP kinases were much less sensitive to inhibition by 5-bromo-UTP and 5-iodo-UTP (Fig.  5). It should also be mentioned that the corresponding monophosphates (5-bromo-UMP and 5-iodo-UMP), unlike 5-fluoro-UMP (9), were not substrates of bacterial UMP kinases.
UMP Kinase Activity in the Reverse Reaction-An essential condition in achieving meaningful quantitative data in the reverse reaction was to maintain "controlled" concentrations of different nucleotide species while varying one single nucleotide. We assumed that ADP, UDP, and GDP form complexes with MgCl 2 with similar K d values, i.e. 1 mM (18). When GDP substituted efficiently for GTP or GMP-PNP as activator of UMP kinase in the reverse reaction, we used mixtures of these three nucleotides and adjusted the concentration of MgCl 2 according to the relationship indicated under "Experimental Procedures." Both N. meningitidis (Fig. 6A) and B. subtilis (Fig. 6B) UMP kinases exhibited biphasic kinetics with magnesium-free UDP as variable substrate. The apparent K m values for the nucleotide in the absence of activators were 4 M (B. subtilis) and 3.4 M (N. meningitidis). GDP or GMP-PNP increased considerably the reaction rates, reversing almost completely the inhibition caused by excess magnesium-free UDP. As in the forward reaction, GDP or GMP-PNP also increased the apparent K m values for magnesium-free UDP to 8.5 M (N. meningitidis) and 24.1 M (B. subtilis). In the absence of GMP-PNP, the activity of B. subtilis UMP kinase with MgADP as variable substrate was very low even at the highest concentrations of nucleoside diphosphate (Fig. 6C). The major effect of GMP-PNP on the reverse reaction rate was apparently to reverse the inhibition exhibited by both the magnesiumfree and magnesium-complexed forms of UDP and consequently to increase the affinity for MgADP.
Site-directed Mutagenesis Experiments-Structure analysis of E. coli UMP kinase indicated that the vicinal amino acid residues Thr 138 and Asn 140 are involved in the cross-talk between two adjacent dimers in the hexameric structure (7). The main chain oxygen of Thr 138 from one subunit is hydrogen-bonded to the side chain nitrogen of Asn 140 from the neighboring subunit. The two residues also interact with the base moiety of UMP. As expected, the T138A and N140A variants of E. coli UMP kinase exhibited a much lower thermodynamic stability than the reference protein (7). Substituting Thr 138 , the side chain of which is hydrogenbonded to uracil, results in a 4-fold higher K m for UMP. In contrast, the K m is not altered by the N140A substitution, as this residue binds uracil only through its main chain carbonyl. The two single residue mutations induce a moderate loss of sensitivity to inhibition by UTP (7). However, the cooperativity of this inhibition appears to be significantly altered. Thus, the cooperativity index of the N140A variant of E. coli UMP kinase declined to 1.5, and the K UTP increased to 300 M. As in the case of the reference enzyme, GTP restored the kinetic constants of the UTP-inhibited N140A variant to the values observed in the absence of UTP (Table 6).
Because Thr 138 and Asn 140 of E. coli UMP kinase are conserved as Thr 135 and Asn 137 in the B. subtilis enzyme, we investigated the kinetic properties of the similar variants obtained by site-directed mutagenesis experiments. All three modified variants (T135A, N137A, and T135A/N137A) of B. subtilis UMP kinase exhibited T m values 10°C lower than that of the wild-type protein. The double mutant T135A/N137A was also the most affected in its stability because, upon dilution in 50 mM Tris-HCl (pH 7.4), it was irreversibly inactivated within several hours. The major kinetic changes (Table 7) are the following: (a) loss of cooperativity with ATP as variable substrate (all modified variants of B. subtilis UMP kinase exhibited hyperbolic dependence of activity in either the absence or presence of GTP); (b) significant increase in the K m for UMP of the T135A variant compared with that of the wildtype enzyme or the N137A variant; and (c) continued sensitivity of both T135A and N137A variants of B. subtilis UMP kinase to activation by GTP, with a 3-fold increase in the K a for activator compared with that of the wild-type enzyme.

DISCUSSION
Bacterial UMP Kinases, an Original Family of Catalysts-Bacterial UMP kinases are unique members of the NMP kinase family of en-  H. influenzae (A) and  B. subtilis (B) UMP kinases at constant concentrations of ATP (2 (A) and 15 (B) mM) and UMP (0.1 (A) and  0.3 (B) mM). f, UTP; F, 5-fluoro-UTP; OE, 5-bromo-UTP; ࡗ, 5-iodo-UTP.  (21) and B. subtilis (22) and most probably from all other bacterial species suggests a specific role of these enzymes in the synthesis of membrane or cell wall constituents. The cooperative kinetics with respect to ATP of UMP kinase from S. pneumoniae (10), a Gram-positive organism, shed new light on this family of catalysts and prompted us to explore or re-examine other UMP kinases from either Gram-positive and Gram-negative bacteria. Our results show that bacterial UMP kinases can indeed be classified in two subfamilies with significantly different regulatory mechanisms. This is not an unprecedented case as, for instance, E. coli aspartate transcarbamoylase, the paradigm of allosteric enzymes, exhibits both homotropic and heterotropic interactions (23), whereas B. subtilis aspartate transcarbamoylase, a homotrimer, lacks both homotropic and heterotropic interactions (24). Finding the structural basis of these differences in UMP kinases and deciphering the mechanism of regulation are challenging issues. For this purpose, a selection of several representative UMP kinases, some belonging also to pathogenic strains for humans, was a necessary step.
Common Properties of UMP Kinases from Gram-positive and Gram-negative Bacteria-Despite the diversity of responses to nucleotides acting as substrates or effectors, the UMP kinases from Gram-negative and Gram-positive bacteria share several common traits. (i) GTP is the common positive effector for all explored enzymes. It reverses the inhibition of excess UMP (forward) or UDP (reverse) and increases the affinity for ATP (forward) or ADP (reverse). (ii) UTP has an opposite effect by decreasing the affinity for ATP/ADP. Whereas in Gram-positive organisms, inhibition by UTP is independent of Mg 2ϩ ions, in Gram-negative organisms, inhibition by UTP occurs only via magnesium-free nucleotide.
The inhibition caused by excess UMP is variable from one enzyme to another and might depend on pH, the concentration of cosubstrate, or the presence of GTP. At pH 6, inhibition by excess UMP was less apparent or abolished for most examined UMP kinases. At saturating concentrations of ATP, B. subtilis UMP kinase was insensitive to inhibition by excess UMP, in contrast to E. coli or H. influenzae UMP kinase. In all cases, GTP reversed inhibition by excess UMP. Because inhibition by excess nucleoside monophosphate has also been observed with other NMP kinases such as E. coli adenylate kinase (25,26) and CMP kinase (27) and yeast GMP kinase (28), several common causes might be invoked to explain this phenomenon. Binding of UMP to the MgATP site is excluded, as inhibition is not competitive with MgATP. Binding of UMP to the allosteric site also seems less probable, as isothermal calorimetry showed that UMP binds to a single site of E. coli and H. influenzae UMP kinases. 3 The most probable explanation would be the occurrence of an abortive UMP kinase⅐MgADP⅐UMP complex, which slows down the release of MgADP. Whatever the true explanation, inhibition by excess UMP of the bacterial UMP kinases also depends on their quaternary structure as demonstrated by site-directed mutagenesis experiments with E. coli (7) and B. subtilis (this study) UMP kinases.
Another property common to various UMP kinases (E. coli appears to be an exception) is that activation by GTP results also in a decrease in the K m or K 0.5 for ATP. In other words, the positive effector acts simultaneously on the kinetic constants of both nucleotide substrates, irrespective of the cooperativity or noncooperativity existing toward the phosphate donor. As a corollary, the complex kinetic effects exhibited by the negative effector (UTP), i.e. a significant increase in the apparent K m or K 0.5 for ATP and a change in the apparent K m for UMP, were not surprising. The fact that the apparent K m for UMP of E. coli and H. influenzae UMP kinases increased dramatically at concentrations of UTP above 0.1 mM is related to the cooperative binding of UTP to its site, which is consistent with the fluorescence properties of the E. coli UMP kinase⅐UTP complex (1,6).
Differences between UMP Kinases from Gram-positive and Gram-negative Organisms-The major difference between UMP kinases from Gram-negative and Gram-positive organisms is the lack of cooperativity with ATP in the former organisms. Although, with H. influenzae UMP kinase, the best fittings of reaction rates with ATP as variable substrate were obtained using the Hill equation, the n H values never exceeded 1.3. On the other hand, the activation of Gram-negative N. meningitidis UMP kinase by GTP is a combination of several effects: enhancement of V m , increase in affinity for ATP, and reversal of inhibition by excess UMP. In this respect, it is worth mentioning that cooperativity in allosteric enzymes is mediated via changes in affinity for substrates (K systems) or via changes in the maximal velocity (V systems) (29). UMP kinases from Gram-positive organisms belong clearly to the K systems, i.e. both T and R states have the same V m values, but different affinities for ATP. In the absence of effectors, the binding of ATP is cooperative, and the positive homotropic interaction is lowered in the presence of GTP or its analogs (10). A factor that might contribute to the cooperativity with ATP of UMP kinase from Gram-positive bacteria might be the dissociation of active hexamers into lower molecular mass oligomers. Such reversible dissociation of hexamers was never observed with E. coli or H. influenzae UMP kinase.
Another major difference between UMP kinases from Grampositive and Gram-negative organisms is related to their sensitivity to inhibition by UTP and its halogenated analogs. In Gram-positive bacteria, inhibition by UTP is not sensitive to high concentrations of Mg 2ϩ or UMP, whereas in Gram-negative organisms, inhibition by UTP is reversed by high concentrations of divalent ion or UMP. On the other hand, the halogen-substituted UTP analogs demonstrate strikingly different effects on UMP kinases from Gram-positive and Gram-negative organisms, suggesting that they interact with different sites in the UMP kinases from these two families of bacteria.
Identity of the Effector-binding Site(s) and Mechanism of Regulation of Bacterial UMP Kinases-The existence of either a unique or two distinct binding sites for GTP and UTP was raised at the very beginning of our study of bacterial UMP kinases (1). From the kinetic experiments described in this work, corroborated by previous spectroscopic and site-directed mutagenesis experiments (2,6) and the x-ray analysis of E. coli UMP kinase in complex with GTP, 4 we can confidently assume that each subunit of bacterial UMP kinase, irrespective of its origin, has three distinct nucleotide-binding sites. The fundamental difference between Gram-positive and Gram-negative organisms is related to the occupancy of these sites by nucleotides and their corresponding analogs. Two of these sites conserved throughout different bacterial species belong to the catalytic center. They interact with ATP or ADP, either as magnesium complexes or magnesium-free nucleotides, and with UMP or UDP, only as magnesium-free nucleotides. The third site, less conserved than the previous ones, interacts primarily with GTP, either as magnesium-free or magnesiumcomplexed species, and in, some particular cases, with GDP, GMP, and even cGMP or guanosine (1). The GTP-binding site, located at the interface of two vicinal monomers, is most probably common to all bacterial species sensitive to activation by guanine nucleotides and/or related analogs. Comparison of a E. coli UMP kinase⅐GTP complex with other structurally known bacterial UMP kinases (H. influenzae and S. pyogenes) indicates an identical fold and distribution of amino acid residues critical for binding of this effector. 4 In Gram-positive organisms, the GTP-binding site corresponds to the single allosteric site, commonly described for the vast majority of regulatory proteins. It can therefore be designed as the "GTP/ UTP site" or "effector site." Binding of effectors to this site shifts the T % R equilibrium to either the R form (GTP) or the T form (UTP), i.e. to either the "high" or "low affinity" form for ATP. The bulky substituents (bromo and iodo) in the UTP heterocycle can be accommodated in the relatively large GTP-binding pocket with an even better affinity than that of the natural nucleotide. Unlike its closest derivative (dUTP), TTP does not inhibit UMP kinase activity, indicating that the hydrophobic methyl group at position 5 in the pyrimidine ring precludes binding to the allosteric site.
In Gram-negative organisms, GTP and UTP, although mutually exclusive, bind to different sites. The allosteric regulation implies a conformational adjustment caused by an effector that affects indirectly the binding ability for the other effector. The binding site of UTP in Gram-negative bacteria overlaps the UMP/UDP site and part of the ATP/ADP site, as indicated by the x-ray data of E. coli UMP kinase in complex with UTP (7). This explains also why UTP increases simultaneously and cooperatively the apparent K m for both nucleotide substrates. Among the UTP analogs, only 5-fluoro-UTP satisfies the structural requirements for fitting the catalytic site and for substituting with similar efficiency for the natural nucleotide. The fact that 5-bromo-UTP and 5-iodo-UTP are still inhibitors of E. coli and H. influenzae UMP kinases might be explained by their "promiscuous" interaction with the GTP-binding site as suggested by fluorescence experiments conducted with H. influenzae UMP kinase (37).
Physiological Relevance of Regulatory Effects of GTP and UTP on Bacterial UMP Kinases-UMP kinase is the first of the three enzymes involved in the conversion of UMP to UTP and CTP, the last two being NDP kinase and CTP synthetase. Both UMP kinase and CTP synthetase are oligomeric proteins positively regulated by GTP and use nucleotides as substrates (1, 30 -32). The structural and functional complexity of these two bacterial enzymes and the inhibition of their activity by the end products UTP and CTP, respectively, indicate that they might be submitted also in vivo to a closely similar control of activity by nucleotides and Mg 2ϩ ions. Assuming that, in bacteria such as E. coli and B. subtilis, the concentrations of ATP, GTP, UTP, and UMP oscillate around 2-3 mM (ATP), 0.8 -1.2 mM (GTP and UTP), and 0.050 -0.1 mM (UMP) (33,34) and that the concentration of soluble Mg 2ϩ is ϳ15 mM (35,36), we might speculate about a role of these nucleotides in modulating the activity of individual UMP kinases. In the case of Gram-positive bacteria, the UMP kinases of which exhibit low affinity for ATP in the absence of GTP, it is obvious that the latter nucleotide is a major participant besides the two substrates. Once the UTP pool is saturated, it competes with GTP for the allosteric site, lowering UMP kinase activity. In Gram-negative organisms, the situation appears to be different. Thus, in bacteria such as H. influenzae and N. meningitidis, the K m for ATP of the corresponding UMP kinases is of the same order of magnitude as the cellular concentration of this nucleotide. Consequently, the cooperative inhibition by magnesium-free UTP at Ͼ10 M might be physiologically relevant. The role of GTP would be rather compensating as an "antagonist of the inhibitor." E. coli has apparently the most "buffered" UMP kinase system, with the enzyme operating always under saturating concentrations of ATP (10 times the corresponding K m ). One of the future tasks in addition to the precise identification of the allosteric site of UMP kinases from Gram-positive organisms will be to determine the in vivo coupling of the UMP kinase and CTP synthetase activities, as much as both enzymes represent valuable targets for antibacterial agents.