Deoxynucleoside kinases encoded by the yaaG and yaaF genes of Bacillus subtilis. Substrate specificity and kinetic analysis of deoxyguanosine kinase with UTP as the preferred phosphate donor.

The overlapping yaaG and yaaF genes from Bacillus subtilis were cloned and overexpressed in Escherichia coli. Purification of the gene products showed that yaaG encoded a homodimeric deoxyguanosine kinase (dGK) and that yaaF encoded a homodimeric deoxynucleoside kinase capable of phosphorylating both deoxyadenosine and deoxycytidine. The latter was identical to a previously characterized dAdo/dCyd kinase (Møllgaard, H. (1980) J. Biol. Chem. 255, 8216-8220). The purified recombinant dGK was highly specific toward 6-oxopurine 2'-deoxyribonucleosides as phosphate acceptors showing only marginal activities with Guo, dAdo, and 2',3'-dideoxyguanosine. UTP was the preferred phosphate donor with a Km value of 6 microm compared with 36 microm for ATP. In addition, the Km for dGuo was 0.6 microm with UTP but 6.5 microm with ATP as phosphate donor. The combination of these two effects makes UTP over 50 times more efficient than ATP. Initial velocity and product inhibition studies indicated that the reaction with dGuo and UTP as substrates followed an Ordered Bi Bi reaction mechanism with UTP as the leading substrate and UDP the last product to leave. dGTP was a potent competitive inhibitor with respect to UTP. Above 30 microm of dGuo, substrate inhibition was observed, but only with UTP as phosphate donor.

Biosynthesis of 2Ј-deoxyribosyl groups occurs solely through reduction of the 2Ј-hydroxyl group of ribonucleoside di-or triphosphates, catalyzed by ribonucleotide reductases (1,2). In addition, a number of organisms possess deoxynucleoside kinases that provide a salvage pathway for the utilization of preformed deoxynucleosides as precursors of DNA. Because the cytotoxicity of a large variety of deoxynucleoside analogs depends on the conversion of these compounds to the corresponding deoxynucleotide analogs, characterization of the substrate specificity and regulation of deoxynucleoside kinases from various sources have received considerable attention (3).
Among eubacteria only two genera, Lactobacilli (14) and Bacilli (15), have been shown to phosphorylate all four deoxynucleosides, whereas it has been established that a number of bacteria, including Escherichia coli and Salmonella enterica serovar Typhi, are lacking dGK, dAK, and dCK activities (16). In Lactobacillus acidophilus strain R26, a strain that appears to lack a functional ribonucleotide reductase (14), the deoxynucleoside kinase activities are organized as three enzymes. In addition to a separate dTK, the remaining three activities are located on two heterodimeric proteins, dGK/dAK and dCK/ dAK, with each subunit being highly specific for the individual substrates, and specifically feedback inhibited by its respective dNTP end-product (17). The three subunits of the two heterodimeric enzymes are encoded by two tandem genes, dak and dgk, where dak encodes the dAK subunit of both enzymes, and dgk encodes both dGK and dCK. The only difference between the amino acid sequences of the two latter subunits is that dCK lacks the N-terminal amino acid residues 2 and 3. The mechanism responsible for the co-or post-translational deletion event has not yet been identified (18). In contrast, the organization of the activities in Bacillus subtilis is quite different. dTK and dGK are genetically distinct and different from the enzyme that phosphorylates dAdo and dCyd (15).
The present report concerns the molecular cloning of two overlapping B. subtilis genes encoding two homodimeric enzymes, dGK and dAK/dCK, and describes the purification and characterization of the recombinant dGK overexpressed in E. coli. The dAK/dCK enzyme appeared to be identical to the deoxyadenosine/deoxycytidine kinase previously purified and characterized from B. subtilis (15).

Methods
Bacterial Strains and Growth Conditions-E. coli SØ5110 (MC1061 cdd::Tn10 (19)) was used as host strain and pUC18 as cloning vector throughout. Cells were grown at 37°C in LB medium (20). When required, ampicillin was added to the medium at 100 g/ml.
DNA Techniques-The methods used for preparing chromosomal DNA and for transforming E. coli have been described previously by Sambrook et al. (21). Plasmids were isolated from E. coli by the alkaline/SDS lysis procedure (22). Fragments from digested plasmid DNA were isolated and purified from agarose gels by the QIAquick gel extraction kit (Qiagen, Germany). Endonuclease digestion and ligation of DNA was done according to the recommendations of the suppliers. DNA was sequenced by the chain termination method (23) on polymerase chain reaction (PCR) products, using the BigDye Terminator cycle sequencing kit (PE Applied Biosystems, Warrington, United Kingdom) and an ABI Prism 310 genetic analyzer (PE Applied Biosystems).
Plasmid Constructions-The overlapping yaaG and yaaF genes were amplified from genomic B. subtilis DNA by PCR using Vent polymerase (New England BioLabs Inc.). The 5Ј-primer, BsdNT4 (5Ј-GAAGGTCG-GATCCTGACGCACG-3Ј), was designed to contain a BamHI site (underlined) 35 bp upstream of the translational start codon of yaaG. The 3Ј-primer, BsdNT5 (5Ј-GTTTTGAAGCTTGCATTCGGTGCGGCG-3Ј), was designed to contain a HinDIII (underlined) 40 bp downstream of the yaaF stop codon. The resulting amplicon was digested with BamHI and HindIII and ligated into similarly digested pUC18, yielding pRA1. Because the reading frames of yaaG and the vector-borne lacZЈ happened to be the same on pRA1, the BamHI site of the plasmid (located between the lac promoter and the cloned fragment) was opened, filled in with the Klenow fragment of E. coli DNA polymerase, and religated. The lacZЈ reading frame of the resulting plasmid, pRA8, was changed, and the BamHI site was replaced by a unique ClaI site (see Fig. 1). Plasmid pRA5 was obtained by subcloning the 825-bp PstI-HindIII fragment of pRA8 into pUC18. Plasmid pRA9 was obtained by cloning the 713-bp EcoRI fragment into the EcoRI site of pUC18. Plasmids with the correct orientation relative to the lac promoter were identified by PstI digestion (see Fig. 1).
Separation of dGK and dAK/dCK-A 1-liter culture of E. coli SØ5110 containing plasmid pRA8 was grown overnight at 37°C in LB medium containing ampicillin, harvested by centrifugation, and subjected to sonication in 15 ml of 50 mM Tris-HCl, pH 7.8 (buffer A). Following centrifugation at 7000 ϫ g for 10 min at 4°C, the supernatant was centrifuged at 100,000 ϫ g for 1 h in a Beckman L8 -80 ultracentrifuge. The supernatant was subjected to ammonium sulfate fractionation, and the precipitate formed between 45 and 70% saturation was redissolved and dialyzed overnight against buffer A. The dialyzed fraction was applied to a Dyematrex RedA column (1.5 ϫ 16 cm), washed with 10 volumes of buffer A, and the column was eluted with 5 volumes of a linear KCl gradient (0 -2 M) in buffer A. Fractions with dNK activity were pooled, concentrated by ammonium sulfate precipitation, redissolved, dialyzed against buffer A, and applied to a Dyematrex Blue A column (1.5 ϫ 16 cm). The column was washed with 10 volumes of buffer A and eluted as above. Fractions 34 -37, containing the highest dGK activity, and fractions 46 -51, containing the highest dAK/dCK activity, were pooled separately, concentrated by ammonium sulfate precipitation, redissolved, and dialyzed against buffer A.
Purification of dGK-One liter of an early stationary phase culture of E. coli SØ5110/pRA9 in LB medium containing ampicillin was harvested by centrifugation and washed once with 0.9% NaCl, and the resultant cell pellet was suspended in 15 ml of buffer A containing 5 mM dithiothreitol and 1 mM EDTA. The suspension was subjected to sonication followed by centrifugation to remove cell debris. The supernatant was further centrifuged for 1 h at 100,000 ϫ g in an ultracentrifuge, and the 100,000 ϫ g supernatant was applied to a Dyematrex Blue A column (1.5 ϫ 16 cm). The column was washed with 5 volumes of buffer A followed by 30 ml of a linear gradient of KCl (0 -0.2 M) in buffer A. The activity was recovered by subsequent isocratic elution with 0.2 M KCl in buffer A. Fractions containing dGK activity were pooled and applied to a phenyl-Sepharose Hi Performance column (1.5 ϫ 15 cm). The column was washed with 50 ml of 25 mM Tris-HCl, pH 7.8, followed by 50 ml of 5 mM Tris-HCl, pH 7.8, 20% 2-propanol, 5% ethanediol, and subsequently eluted with 1% Triton X-100. Fractions with the highest specific activity were pooled, dialyzed against buffer A, and concentrated by ultrafiltration using a Centriprep centrifugal concentrator (Amicon, Inc.). The concentrated enzyme was stored in 50% (v/v) glycerol at Ϫ20°C.
Enzyme Assays-Deoxyguanosine kinase activity was measured with one of two procedures. Assay 1 measured the conversion of [ 3 H]dGuo to [ 3 H]dGMP. The reaction mixture contained 50 mM Tris-HCl, pH 8.5, variable concentrations of [ 3 H]dGuo and nucleoside triphosphate, 5 mM MgCl 2 , and 1 mg/ml BSA in a final volume of 50 l. The reaction was started by addition of enzyme and performed at 37°C. Four 10-l aliquots of the reaction mixture were taken at different times after addition of enzyme. These were mixed with 5 l of 1.5 M formic acid to stop the reaction, and the entire mixture was pipetted onto DE-81 filter discs and dried. The discs were washed in deionized water for 30 min and dried, and the bound radioactivity was eluted from the filters with 1 ml of 0.1 M HCl, 0.2 M KCl and counted in a Tri-Carb liquid scintillation analyzer (Packard Instruments, Inc., Downers Grove, IL) using 5 ml of Ecoscint A scintillation fluid (National Diagnostics). The amount of enzyme used in the assays was such that less than 10% of the substrate was converted to product during the reaction. All samples were taken within the linear part of the reaction, and initial velocities were determined from the slope of the lines. Assay 2 measured spectrophotometrically the oxidation of NADH coupled enzymatically to the phosphorylation of nucleoside to nucleotide. The reaction mixture (500 l) contained 50 mM Tris-HCl, pH 7.8, 100 mM KCl, 5 mM MgCl 2 , 1.5 mM NTP, 0.1-0.4 mM nucleoside, 0.5 mM phosphoenolpyruvate, 0.18 mM NADH, 2 units of pyruvate kinase (10 units when UTP or CTP was used as phosphate donor), 2.5 units of lactate dehydrogenase, and enzyme. The reaction was started by addition of nucleoside, and the decrease in absorbance at 340 nm was followed at 37°C. One enzyme unit is defined as the amount of enzyme that catalyzes the formation of 1 mol of dGMP per min. Protein was determined by the method of Lowry et al. (24) using BSA as a standard.
Data Treatment-K m and V max values were obtained by fitting steady-state kinetic data to the rate equation for a sequential Bi Bi mechanism with the Biosoft program UltraFit 3.0 for Macintosh. K i values were calculated from secondary plots of inhibitor concentration versus intercept or slope of the double-reciprocal plots using linear least-squares regression analysis to determine the best-fit line describing the data.
Mass Spectrometry-The mass spectrometry was performed by Dr. Jette Wagtberg Sen at the Statens Seruminstitut, Copenhagen, using a Mariner electrospray time-of-flight spectrometer (PerSeptive Biosystems).

RESULTS
Cloning of the B. subtilis yaaG and yaaF Genes-The amino acid sequences of dAK and dGK from L. acidophilus, deduced from the nucleotide sequence of the dak/dgk tandem genes (27), were used to search the B. subtilis genome sequence (28) for open reading frames encoding putative deoxynucleoside kinases. Two genes, yaaG and yaaF, were identified at coordinates 22497-23767 bp, which encoded putative polypeptides with 25-32% amino acid sequence identity to the L. acidophilus deoxynucleoside kinases. The nucleotide sequence indicated that the TGA stop codon of yaaG overlapped 2 bp with the ATG start codon of yaaF, and that both open reading frames were preceded by typical B. subtilis Shine-Dalgarno sequences (29). The deduced amino acid sequences of the two gene products were 30% identical.
The putative yaaG/yaaF operon was amplified from B. subtilis genomic DNA by PCR and cloned into pUC18, resulting in pRA8 (Fig. 1). Subsequently, yaaG and yaaF were subcloned separately in pUC18 using the internal PstI and EcoRI sites of pRA8, yielding plasmids pRA9 and pRA5, respectively (Fig. 1). In all constructs the cloned genes were transcribed from the lac promoter (lacP) of the vector and translated from the native ribosomal binding regions. DNA sequencing of the cloned fragments of pRA9 and pRA5 confirmed the published genomic sequence and showed that they encoded potential polypeptides of 207 and 217 amino acid residues, respectively. SDS-PAGE of total proteins from cultures of E. coli harboring pRA9 or pRA5 revealed a major band with a mobility corresponding to a 24-kDa polypeptide, in accordance with the molecular masses of the yaaG and yaaF gene products as deduced from the nucleotide sequence (data not shown).
Identification of the yaaG and yaaF Gene Products-Crude cellular extracts of E. coli SØ5110 harboring pRA8, pRA9, and pRA5 were assayed for deoxynucleoside kinase activity with dGuo, dAdo, and dCyd as substrates (Table I). The results suggested that yaaG encoded a kinase specific for deoxyguanosine, whereas the yaaF gene product was capable of phosphorylating both deoxyadenosine and deoxycytidine.
To establish whether all three kinase activities were part of a single heterooligomeric enzyme in conditions where both yaaG and yaaF were present in the cells, crude extracts of E. coli SØ5110/pRA8 were subjected to both ammonium sulfate fractionation and dye-affinity chromatography. Table II shows that the kinase activities for dGuo, dAdo, and dCyd copurified upon ammonium sulfate precipitation and on RedA dye chromatography. However, on a BlueA dye column deoxyguanosine kinase activity eluted clearly before the deoxyadenosine and deoxycytidine kinase activities. Together with the results presented in Table I, this indicated that the gene products of yaaG and yaaF formed two physically distinct enzymes, dGK and dAK/dCK. Preliminary characterization of the recombinant dAK/dCK from the BlueA dye column indicated that it was identical to the deoxyadenosine/deoxycytidine kinase previously purified and characterized from B. subtilis (15).
Purification of Recombinant dGK-Having established that deoxyguanosine kinase was encoded by the yaaG gene, the recombinant enzyme was purified from cells of E. coli SØ5110 harboring pRA9. Because preliminary experiments showed that ammonium sulfate precipitation resulted in heavy losses, this step was avoided. The final procedure adopted involved fractionation by BlueA dye affinity chromatography and separation according to hydrophobicity on a phenyl-Sepharose column producing an enzyme that was Ͼ99% pure as judged by SDS-PAGE, with an overall yield of 25% (Table III).
Molecular Weight-The molecular mass of the subunit was determined to be 24,147 Da by electrospray mass spectrometry (data not shown), in accordance with the molecular mass of the yaaG gene product deduced from the DNA sequence (24,145 Da). The molecular mass of purified recombinant dGK was about 49 kDa as estimated by gel filtration on a Sephacryl S-300 column equilibrated with 50 mM Tris-HCl, pH 7.8, suggesting that the enzyme was a homodimer under the conditions employed.
Stability-In the presence of 50% v/v glycerol, concentrated solutions of the enzyme (Ͼ1 mg/ml) could be stored for at least 1 year at Ϫ20°C without loss of activity. Upon dilution for assays the enzyme was highly unstable at 0°C unless BSA (1 mg/ml) was added to the dilution buffer. In the presence of BSA the diluted enzyme (Ͼ1 g/ml) was stable for months at 0°C.
pH Optimum-With saturating concentrations of dGuo and UTP, dGK showed a broad pH optimum centered around pH 9. At pH 7.5 and 11.5 more than 80% of maximal activity was still observed. At pH 6.0, 60% activity remained, whereas the enzyme was completely inactive below pH 5.6 (data not shown).
Metal Ion Requirement-As observed for most kinases, dGK required Mg 2ϩ for activity, suggesting that the true phosphate donor is MgNTP 2Ϫ . Excess Mg 2ϩ did not inhibit the enzyme. Thus, all assays were performed with Mg 2ϩ at 5 mM to ensure that all NTP present in the reaction was complexed with magnesium.
Phosphate Acceptor Specificity-The specificity of the enzyme toward the nucleoside substrate was determined using the coupled assay and GTP as the phosphate donor. Naturally occurring ribo-and deoxyribonucleosides as well as certain nucleoside analogs were tested. As shown in Table IV, the enzyme is highly specific toward 6-oxopurine 2Ј-deoxynucleosides. Only marginal activity was observed with guanosine, and neither 2Ј,3Ј-dideoxynucleosides nor hypoxanthine arabinoside served as substrates.
Phosphate Donor Specificity-Previous studies with dAK/ dCK from B. subtilis indicated that GTP and dGTP were the preferred phosphate donors with both dAdo and dCyd as substrates; ATP and UTP were only half as effective (15). Preliminary experiments with dGK indicated that all naturally occurring ribo-and deoxyribonucleoside triphosphates, with the exception of dGTP, were active as phosphate donors. Under the conditions used (20 M dGuo and 25 M NTP) UTP was slightly  more efficient than the other NTPs. With dGTP no measurable activity was observed (data not shown).
Steady-state Kinetics-Initial velocity experiments were carried out at varying concentrations of the four individual ribonucleoside triphosphates and fixed concentrations of dGuo. For each of the NTPs, double reciprocal plots of the initial rate as a function of the NTP concentration for various fixed dGuo concentrations yielded a series of intersecting lines, diagnostic of a sequential reaction mechanism (data not shown). The data obtained were fitted to the rate equation for a sequential Bi Bi mechanism. The kinetic constants obtained (Table V) showed that the true V max values for the various NTPs were quite similar, whereas large variations in the true K m values for the four NTPs were observed. The k cat /K m measure of substrate efficiency indicated a preference for UTP as phosphate donor by a factor of 4 -6. Furthermore, the data indicated that the K m for dGuo was highly dependent on the nature of the phosphate donor (Table V). In the presence of UTP the true K m for dGuo was 0.6 M compared with 6.5 with ATP. This synergism between the high affinity for UTP and the low K m for dGuo with UTP as phosphate donor makes UTP 50 -60 times more efficient than ATP as a phosphate donor. In all further kinetic studies UTP was employed as phosphate donor.
Product inhibition studies with dGuo and UTP as substrates were performed to determine the binding order of substrates and products (Fig. 2, A-D; Table VI). The secondary plots of the slopes or intercepts versus the concentration of inhibitor were in all cases linear (see insets in Fig. 2, A-D). With UTP as the variable substrate, inhibition by UDP was competitive ( Fig.  2A), whereas inhibition by dGMP was noncompetitive with K is being of the same order of magnitude as K m for UTP, and K ii about 25-fold higher (Fig. 2B). With dGuo as the variable substrate, both products produced noncompetitive inhibition (Fig. 2, C and D). This inhibition pattern suggested that dGK followed a sequential ordered reaction mechanism in which UTP had to bind before dGuo, and dGMP was leaving the enzyme complex before UDP. Such an ordered reaction mechanism predicted uncompetitive inhibition by dGMP with UTP as variable substrate, under conditions of full saturation with respect to the second substrate, dGuo. Because of substrate inhibition exerted by dGuo (see below) this prediction could not be tested. Whether the reaction with other NTPs as phosphate donors followed the same ordered reaction mechanism was not investigated.
Substrate inhibition was observed by dGuo at concentrations above 30 M. As shown in Fig. 3, the inhibition was exclusively observed with UTP as phosphate donor, and it was found to be linear (Fig. 3, inset). This indicated that the inhibition resulted from the formation of a dead-end complex of dGuo with a form of the enzyme that it is not supposed to react with (30). The inability of high UTP concentrations to reverse the inhibition (data not shown) suggested that dGuo, the second substrate in the ordered reaction, combined with the E⅐UDP complex to form the dead-end complex.
Inhibition of dGK by dGTP-As mentioned above, dGTP was the only naturally occurring NTP that did not function as phosphate donor for dGK. In contrast, dGTP was a potent competitive inhibitor of the enzyme with respect to UTP, with a K i value of 0.4 M (data not shown). The pattern of dGTP inhibition with dGuo being the variable substrate was noncompetitive at lower dGuo concentrations with K ii and K is values of 1.6 and 0.7 M, respectively.

DISCUSSION
The genetic and enzymatic basis for the ability of B. subtilis to phosphorylate dCyd, dAdo, and dGuo was established by cloning and overexpressing the overlapping yaaG and yaaF genes in E. coli, an organism unable to phosphorylate deoxyribonucleosides other than dThd (16). Simultaneous expression of both genes yielded two separable enzymes, one specific for dGuo and the other capable of phosphorylating both dCyd and dAdo. Expression of each gene separately showed that yaaG encoded dGK and yaaF encoded dCK/dAK, and that both enzymes were homodimeric proteins. Preliminary studies indicated that dCK/dAK was identical to the previously characterized deoxyadenosine/deoxycytidine kinase of B. subtilis (15). Based on these findings we propose that the yaaG and yaaF genes be renamed as dgk and dak, respectively.
The recombinant Bs-dGK was purified to homogeneity and characterized. Compared with other enzymes with deoxyguanosine kinase activity the selectivity of Bs-dGK for the nucleoside substrate is unique. Only 6-oxopurine 2Ј-deoxyribonucleosides (dGuo and dIno) are phosphorylated by the enzyme at significant rates, with a K m for dGuo of 0.6 M and a k cat of 1.4 s Ϫ1 . The homodimeric mammalian mitochondrial dGK, which in addition to dGuo and dIno can use dAdo as a substrate, shows K m and k cat values of 7.6 M and 0.002 s Ϫ1 , respectively, with dGuo as a substrate (9), and the homodimeric human dCK, which uses dCyd, dAdo, and dGuo, phosphorylates dGuo with a K m of 150 -430 M and a k cat of   0.4 -6.0 s Ϫ1 (7,31). Recently, a monomeric multisubstrate dNK from Drosophila melanogaster (Dm-dNK) was characterized (12). Its K m for dGuo was found to be very high (654 M), but at the same time the enzyme displayed the highest k cat reported for any deoxynucleoside kinase (18 s Ϫ1 ). The specificity constant (k cat /K m ) for dGuo phosphorylation by Bs-dGK is thus 10 2 to 10 4 higher than the corresponding values of the various eukaryotic deoxyguanosine kinases. Expression in E. coli of the L. acidophilus dgk gene, encoding the dGK subunit of the heterodimeric dAK/dGK, results in the production of a loosely associated homodimeric dGK, with significant similarity to Bs-dGK regarding its specificity toward the nucleoside sub- strate. However, the specificity constant of this "unnatural" dGK is only 1/20th of that of Bs-dGK (18).
In contrast to its strict phosphate acceptor requirement, Bs-dGK can use most NTPs as phosphate donors. However, the results presented in Table V indicate that UTP is the preferred phosphate donor. Compared with the other NTPs the efficiency of UTP results from a 6-fold lower K m for UTP than for the other NTPs and a much lower K m for dGuo when UTP is employed as donor. The steady-state kinetic analysis of the reaction with dGuo and UTP indicated that the reaction follows an Ordered Bi Bi mechanism with UTP as the leading substrate and release of UDP last (Scheme 1). The competitive inhibition by dGTP with respect to UTP is consistent with an ordered addition of substrates with UTP binding first (32). The substrate inhibition data (Fig. 3) are also consistent with the proposed kinetic model and suggest formation of the dead-end complex E⅐UDP⅐dGuo. The observation that substrate inhibition with dGuo only occurred with UTP as phosphate donor shows that the reaction with other NTPs as phosphate donors may follow a different kinetic mechanism.
It was previously observed that UTP, CTP, and dTTP were more efficient phosphate donors than ATP for the bovine liver mitochondrial dGK at physiological pH (33), and that UTP was the preferred phosphate donor for human dCK with dCyd and dAdo as acceptors (34,35). More recently, a detailed kinetic analysis of human dCK with dCyd and UTP was reported (36). It showed that the preference for UTP as phosphate donor was primarily due to large differences in the true K m values for the phosphate donor with values of 1.2 and 54 M for UTP and ATP, respectively. Only a 2-fold difference in the true K m values for dCyd was observed (0.5 and 1.0 M with UTP and ATP, respectively). The kinetic mechanism of dCyd phosphorylation also appeared to depend on the phosphate donor showing a random Bi Bi reaction sequence with ATP but an ordered addition-random release sequence with UTP (31,36). Unfortunately, the kinetic properties of the enzyme with dAdo or dGuo as substrates were not investigated.
Most deoxynucleoside kinases are end-product-inhibited by the corresponding dNTP, and it has been shown in a number of cases that the dNTP inhibitor behaves like a bisubstrate analog    a Derived from the inserted secondary plots in Fig. 2, A-D. (17,37,38). The kinetics of dGTP inhibition and the observation that all NTPs and dNTPs except dGTP function as phosphate donors, suggest that dGTP inhibition of Bs-dGK is due to binding of the inhibitor to both the deoxynucleoside acceptor site and the triphosphate binding site of UTP. The intracellular amounts of nucleoside triphosphates in B. subtilis have been measured previously (39). For UTP and dGTP they were estimated to be 0.15 and 1.6 nmol/mg of dry weight, which corresponds to 188 M and 2 mM respectively. 2 Insertion of these values, the K i value for dGTP of 0.4 M, and the K m for UTP of 6 M, into the rate equation for competitive inhibition shows that the enzyme would be inhibited by ϳ60% under these conditions, indicating that this feedback control may be of physiological importance. Recently, the nucleotide sequence of the entire Deinococcus radiodurans genome was determined (40). It revealed that this Gram-positive bacterium contains two overlapping genes encoding putative polypeptides with 32 and 35% identity to the yaaG and yaaF genes of B. subtilis, respectively. Thus, D. radiodurans may represent a third genus of eubacteria possessing dGuo, dAdo, and dCyd kinase activities. It is noteworthy that homologous genes have so far not been identified in Gram-negative bacteria.