Characterization of a 5'-polynucleotide kinase/3'-phosphatase from bacteriophage RM378.

A polynucleotide kinase from the thermophilic bacteriophage RM378 that infects the thermophilic eubacterium Rhodothermus marinus was identified, expressed, and purified. This polynucleotide kinase was demonstrated to have a 5'-kinase domain as well as a 3'-phosphohydrolase domain. The RM378 polynucleotide kinase had limited sequence similarity to the 5'-kinase domain of the T4 bacteriophage polynucleotide kinase, but apparent homology was not evident within the 3'-phosphohydrolase domain. The domain order of RM378 polynucleotide kinase was reversed relative to that of the T4 polynucleotide kinase. The RM378 phosphohydrolase domain displayed some sequence similarity with the bacterial poly(A) polymerase family, including an HD motif characteristic of the diverse superfamily of metal-dependent HD phosphohydrolases. The RM378 polynucleotide kinase was biochemically characterized and shown to possess 5'-kinase activity on RNA and single- and double-stranded DNA at elevated temperatures. It also showed phosphohydrolase activity on 2':3'-cyclic adenosine monophosphate. This description of the RM378 polynucleotide kinase, along with the recently described RM378 RNA ligase, suggests that the RM378 bacteriophage has to counter a similar anti-phage mechanism in R. marinus as the one that the T4 phage has to counter in Escherichia coli.

Recently, we described the identification and characterization of a thermostable RNA ligase from the thermophilic bacteriophage RM378 that is homologous to the T4 RNA ligase 1 and infects the thermophilic eubacterium Rhodothermus marinus (1)(2)(3)(4). The biological role of T4 bacteriophage RNA ligase 1 is to counteract certain Escherichia coli host defenses that are based on the degradation of tRNA molecules (5)(6)(7)(8). The best known of these systems is the suicidal anticodon nuclease system, which is activated upon inhibition of the EcoprrI DNA restriction-modification system by the T4 viral Stp polypeptide. The anticodon nuclease cleaves tRNA Lys 5Ј at its wobble position, yielding 2Ј:3Ј-cyclic phosphate and a 5Ј-hydroxyl group (5,7,8). Whereas the RNA ligase 1 is essential for ligation of the cleaved tRNA molecules, the polynucleotide kinase (PNK) 1 has the important role of making the tRNA fragments appropriate substrates for the ligation step (6, 9 -12). T4 PNK is a nucleic acid processing enzyme that has two major functions as follows: (i) to remove the 2Ј:3Ј-cyclic phosphate from the 5Ј tRNA fragment; and (ii) to add a phosphate group to the 5Ј-hydroxyl group of the 3Ј-tRNA fragment using ATP as the phosphate donor (6, 10 -12). The RNA ligase 1 and the PNK are thus part of the same system, acting in concert to repair dysfunctional translational machinery.
The T4 PNK is the founding member of a large 5Ј-kinase/3Јphosphohydrolase family that has been studied extensively for the last 35 years. The PNK family includes polynucleotide kinases from wide variety of organisms. The biological role of the eukaryotic 5Ј-kinase/3Ј-phosphohydrolase is to mend broken nucleic acids strands, making them appropriate substrates for repair by nucleic acid ligases and playing a notably important role in the repair of DNA nicks and gaps (13)(14)(15).
Biochemical and mutational analysis have shown that T4 PNK is a homotetramer with no kinetic cooperativity (11, 16 -18). The 5Ј-kinase and 3Ј-phosphohydrolase activities have been shown to reside in distinct and separate domains, with an N-terminal 5Ј-kinase domain and a C-terminal 3Ј-phosphohydrolase domain. The 5Ј-kinase domains of PNK enzymes contain the nucleotide binding motif GXXXXGK(S/T) (a Walker A box or P-loop), which is a common motif in phosphotransferases as well as in other proteins containing nucleotide binding domains (16,17,19). Mutational data from T4 PNK has shown that in addition to residues in the P-loop motif (Lys-15 and Ser-16), residues Asp-35, Arg-38, Asp-85, and Arg-126 are all essential for the 5Ј-kinase activity (20). In addition, studies have suggested a role of residues Asp-85 and Asn-87 in the quaternary structure integrity, resulting in mixture of dimers and tetramers when these positions are changed by sitedirected mutagenesis (16,17,20).
The sequence analysis and mutational data on the T4 PNK/ 3Ј-phosphohydrolase domain have shown that the metal-dependent phosphatase family motif DXDXT is found in the PNK family and is essential for the phosphohydrolase activity of the domain (16,17,20). Sequence analysis of PNK shows that the 3Ј-phosphatase domain of T4 PNK is distantly related to other phosphatase families like the histidinol phosphatase family and the acid phosphatase (HAD) superfamily (16,20). The crystal structure of the T4 PNK was solved by Galburt et al. and confirmed that there were two functionally distinct structural domains (18). The N-terminal 5Ј-kinase domain was structurally similar to those of the adenylate kinase (Adk) * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) NC_004735.
Here we describe a polynucleotide kinase from the thermophilic bacteriophage RM378 that infects R. marinus. As compared with T4 PNK, the RM378 PNK has analogous activity while utilizing a phosphohydrolase domain of different origin and having the kinase and phosphohydrolase domains in a reversed relative order as compared with those of T4 PNK.

EXPERIMENTAL PROCEDURES
Cloning of the pnk Gene from RM378 Bacteriophage Genome-All standard molecular biology protocols were done as described by Sambrook et al. (21). Chemicals and media were purchased from Sigma or Merck unless otherwise noted. Oligonucleotides were purchased from MWG Biotech Inc. and Eurogentech Inc. The full-length polynucleotide kinase gene was amplified by standard PCR from RM378 bacteriophage DNA using the KinR-Ase forward primer (5Ј-dCCAATTGATTAATATGC-CGAACTTCATTACAAACATC-3Ј) and the KinR-Bam reverse primer (5Ј-dCGCGGATCCAAGCTACTCTCAACACAT-3Ј) with Dynazyme TM DNA polymerase (Finnzymes Oy) as recommended by the manufacturer. The PNK PCR product was cloned into pJOE 3075 vector with a connected His tail to the C-terminal. Five clones were sequenced for verification of the DNA sequence, and a vector-PNK clone named pJOE-PNK was selected for expression experiments. pJOE-PNK was transformed into CodonPlus® BL21 RIL E. coli cells (Stratagene Inc.), the strain was cultivated at 37°C in a 10-liter BioFlow 3000 fermentor, and the expression was induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside. The cells were harvested after 4 h of incubation and disrupted by sonication. The crude cell extract was centrifuged in a SA-600 rotor (Sorvall Inc.) at 10,400 ϫ g for 1 h. The supernatant was collected and applied to an XK 26/10 50-ml column (Amersham Biosciences) packed with chelating Sepharose and charged with nickel ions. The column was washed with buffer (10 mM Tris-HCl buffer, pH 7.5, 0.5 M NaCl, and 25 mM imidazole). Elution was performed in the same buffer with 500 mM imidazole in a stepwise manner (20 and 40%). The eluted protein was then put through HiPrep Sephacryl 26/60 S200 high resolution column (Amersham Biosciences), eluted with 2ϫ kinase storage buffer (20 mM Tris, pH 9, 100 mM KCl, 2 mM dithiothreitol, 0.2 mM EDTA and 0.4 M ATP), and subsequently mixed at a 1:1 volume ratio with 100% glycerol and stored at Ϫ20°C.
Effects of different pH, divalent ions, salt ions, PEG 6000, spermidine, and different substrates were analyzed as described under "Results." Thermostability assays were done by taking samples of an ongoing reaction at different temperatures over a time scale and looking for a linear accumulation of product.
The effect of ADP was tested on three kinds of reactions, namely an exchange reaction, a 5Ј-kinase reaction, and dephosphorylation. This was done by titrating different amounts of ADP to reaction mixtures containing the following: (i) 10 M phosphorylated single-stranded DNA oligomer and 10 M [␥-32 P]ATP; (ii) 10 M 5Ј-hydroxylated singlestranded DNA oligomer with 10 M [␥-32 P]ATP; or (iii) with 5Ј-32 Plabeled single-stranded DNA oligomer without ATP.
Characterization of the PNK/3Ј-Phosphohydrolase Activity-After the determination of pH, the cation, and the apparent temperature optimum, the standard phosphohydrolase assay conditions used were potassium acetate buffer, pH 6, 5 mM MnCl 2 , 1 mM dithiothreitol, 10 mM KCl, 0.1-5 mM cyclic 2Ј:3Ј-cyclic adenosine monophosphate (cAMP), and 0.05 mg/ml PNK. Reaction time was 30 -60 min at 65-70°C. The reaction was quenched by adding 90 l of Biomol Green reagent (Biomol Research Laboratories, Plymouth Meeting, PA) to the 10-l reaction volume. The release of phosphate was measured at A 620 nm in a Sunrise Absorbance Reader (Tecan Group Ltd, Maennedorf, Switzerland) and compared with a phosphate standard curve.
Characterization of the phosphohydrolase activity was done using two substrates, thymidine 3Ј-monophosphate (3Ј-TMP) and cAMP. Determination of pH optimum was done with MOPS and potassium acetate buffers using 0.1 mM cAMP and 0.05 mg/ml PNK. Apparent tem-perature optimum was also done under the standard condition by assaying at different temperatures. Substrate concentration curves were done for cAMP and 3Ј-TMP under the standard conditions using different amount of substrates. All assays were done in triplicate, and the mean value was calculated.

RESULTS
Sequence Analysis, Overexpression, and Purification of the RM378 PNK-Although the initial screening of the RM378 bacteriophage genome (GenBank TM accession number NC_004735) using standard BLAST analysis (22) identified the RnlA gene, which encodes a homolog to the RNA ligase 1 in the T4 phage (2), we could not find an open reading frame with similarity to the T4 PNK (pseT) gene by using standard bioinformatic protocols. Our working model was that R. marinus contained a tRNA degradation system similar to that of E. coli, and we therefore hypothesized that a polynucleotide kinase would be present in the RM278 genome as part of a system to override a tRNA degradation anti-phage mechanism. However, we could not rule out the possibility that the tRNA degradation system in R. marinus was very different from the one in E. coli, given the distant evolutionary relationship between these two bacterial species (3). We therefore redefined our search by looking specifically for T4-like 5Ј-kinase domains and found a hit with open reading frame tag Rm378p110 for a putative PNK within the RM378 genome. This putative polynucleotide kinase gene designated pnkp (GenBank TM accession number NP_835697), which has previously been described as having similarity to poly(A) polymerases from eubacteria, was 1086 bp in length and coded for a 361-amino acid polypeptide with a calculated mass of 42.1 kDa. The putative PNK had reversed domain orientation relative to T4 PNK, with the 5Ј-kinase domain on the C terminus of the putative PNK protein. Amino acid sequence similarity searches with the N terminus of the pnkp showed sequence similarity to the superfamily of HD metal-dependent phosphohydrolase (23), which meant that RM378 PNK could also contain a 3Ј-phosphohydrolase activity analogous to the activity observed in T4 PNK (24).
The amino acid sequence of the putative RM378 PNK/5Јkinase domain was compared with those of the T4 and mycobacteriophage CJW1 5Ј-kinase domains ( Fig. 1A) (19,25). Alignments were done using the ClustalX program (26). Although the overall similarity was low, only 13 and 20% to T4 and mycobacteriophage CJW1, respectively, a characteristic nucleotide binding P-loop motif was conserved. The putative RM378 PNK/3Ј-phosphohydrolase domain was compared with those of Clostridium acetobutylicum poly(A) polymerase and Desulfitobacterium hafniense tRNA nucleotidyltransferase/ poly(A) polymerase (Fig. 1B). Again, similarity was low, but an HD box, which is the main characteristic for the superfamily, was present (23). It was our hypothesis that this HD phosphohydrolase domain might have similar activity as that of the T4 PNK family-like phosphohydrolase even if they do not share any sequence similarity.
The putative pnkp gene was amplified from the RM378 viral genome cloned into the pJOE-His expression vector with the pnkp open reading frame in-frame with a His tag. The gene product was overexpressed in E. coli and purified to near homogeneity using nickel affinity chromatography. The yield of purified recombinant RM378 PNK was 2 mg/g cells (wet weight). Aliquots from the purification procedure were collected and run on 12% SDS-PAGE and stained with Coomassie blue R-250 (27). The purification was estimated as Ͼ90% by SDS-PAGE analysis (Fig. 2). The recombinant PNK protein was studied by analytical gel filtration to evaluate the oligomeric state of the protein. Conclusive results were not obtained, because peaks from the gel filtration column ranging in size from 34 to 200 kDa all displayed 5Ј-kinase activity (data not shown).
Characterization of the RM378 PNK/5Ј-Kinase Activity-We utilized the PNK assay of Richardson (12) to characterize the RM378 PNK/5Ј-kinase activity. Using MOPS buffers of different pH values, the enzyme was assayed under standard conditions as described under "Experimental Procedures." The apparent pH optimum was determined to be 8.5, but Ͼ80% of the activity was observed in a range between pH 7.5 and 10 ( Fig.  3A). Temperature optimum was determined by doing the standard assay at different temperatures for 1 h. The apparent temperature optimum of the enzyme was determined to be between 60 and 70°C under the given conditions (Fig. 3B). To determine whether the enzyme was stable at this temperature, the enzyme was assayed at 50, 60, 65, and 70°C, and samples were taken for activity analysis at the time points 0, 30, 60, and 120 min. The results showed relatively good stability up to 60 -65°C, but at 70°C enzymatic activity was only observed for 1 h (Fig. 3C). The effects of divalent cations were tested by doing the standard assay in presence of Mg 2ϩ or Mn 2ϩ at different concentrations. The results showed that a divalent cation was essential for the 5Ј-kinase reaction and that the maximum activity was reached in 10 and 100 M for Mn 2ϩ and Mg 2ϩ , respectively (data not shown). Activity was similar for Mn 2ϩ and Mg 2ϩ , but Mn 2ϩ concentrations of Ͼ100 M severely inhibited the reaction (data not shown). The effects of varying NaCl and KCl concentrations on the PNK activity were also examined, and the results showed that both salts caused a steady decrease in activity at concentrations of Ͼ10 mM (data not shown). The effect of spermidine was limited, with a Ͻ10% increase in activity at a 1 mM concentration (data not shown). On the other hand, PEG 6000 significantly increased the 5Јkinase activity (3-4-fold) in a 5-15% concentration but inhibited the 5Ј-kinase reaction at higher concentrations (Fig. 3D).
The effect of ADP concentration on 5Ј-kinase activity, dephosphorylation, and phosphate exchange was studied in an assay with 10 M ATP and an oligomer and by titrating the ADP concentration. As seen in Fig. 4A, the 5Ј-kinase activity decreased as the ADP concentration was increased to complete inhibition at 1 mM ADP. If only ADP and 32 P-labeled oligomer were assayed with the PNK, the level of dephosphorylation increased as the ADP concentration was increased as seen in Fig. 4A. Additionally, when labeling an already phosphorylated oligomer, the exchange reaction was ϳ2-4% overall when compared with kinase activity on a 5Ј-hydroxylated oligonucleotide independent of the ADP concentration from 0 to 0.1 mM ADP. At a 1 mM ADP concentration the reaction was inhibited.
Titration To investigate the completeness of the 5Ј-kinase reaction, 20 M r(A 20 ) and d(A 20 ) oligos were labeled with a 10 M 32 Plabeled ATP mixture under optimal conditions. The results, shown in Fig. 4D, demonstrate that the ATP was depleted when labeling was done with a limited amount of ATP. Also, hydroxylated oligomers (1-25 M) were fully phosphorylated (Ͼ95%) in the reaction using 100 M ATP (data not shown). The RM378 PNK therefore appears to be an excellent enzyme for nucleic acid labeling at an elevated temperature. Characterization of the RM378 PNK/3Ј-Phosphohydrolase Activity-The characterization of the 3Ј-phosphohydrolase activity was done using both cAMP and 3Ј-TMP. We hypothesized that cAMP might work better than 3Ј-TMP as a substrate for HD phosphohydrolase because cAMP was the preferred substrate for the HD domain of E. coli tRNA nucleotidyl transfer-ase as described by Yakunin et al. (24). Consistent with our hypothesis, the RM378 PNK had more phosphohydrolase activity on cAMP relative to 3Ј-TMP (Fig. 5, C and D). We studied the pH profile and found using MOPS that activity was only seen from pH 6 -7, which is out of the MOPS stable pH range. We subsequently used potassium acetate buffer from pH 4 -6 in the comparison and observed a pH optimum of 6, with Ͼ35% activity in the range pH 5.5-7.0 as seen in Fig. 5A. We only observed activity on cAMP during the initial experiments and, therefore, only used this substrate in the general characterization of the 3Ј-phosphohydrolase activity. Activity was relatively low on cAMP at pH 6 using Mg 2ϩ . When Mg 2ϩ was replaced with Mn 2ϩ in the 3Ј-phosphohydrolase assay on the cAMP substrate, the activity increased, resulting in an ϳ10-fold increase in activity at 1 mM concentration as seen in Fig. 5D.
Temperature optimum of the 3Ј-phosphohydrolase activity was 75°C, although the enzyme showed good activity (Ͼ50%) from 65-80°C as seen in Fig. 5B. As before, the enzyme was stable up to 60 -65°C for 2 h (data not shown). We also titrated the cAMP and 3Ј-TMP substrates in the 3Ј-phosphohydrolase reaction and found V max to be 13.5 mol⅐mg Ϫ1 ⅐h Ϫ1 and 1.5 mol⅐mg Ϫ1 ⅐h Ϫ1 for cAMP and 3Ј-TMP, respectively. The K m constants for cAMP and 3Ј-TMP were 0.7 and 0.06 mM, respectively. The substrate titrations curves for the PNK phosphohydrolase activity are shown in Fig. 5C. Turnover numbers were estimated at 0.16/sec and 0.017/sec for cAMP and 3Ј-TMP, respectively. The RM378 was able to dephosphorylate 80% of the 2Ј:3Ј-cAMP substrate under optimal conditions using 1 mM cAMP (data not shown).
A comparison was done between RM378 and T4 PNK using the cAMP, 3Ј-TMP, and d(A 15 )-PO 4 Ϫ oligomer as substrates, all at a 0.2 mM concentration using 20 units (by the PNK unit definition by Richardson) (12) in 20-l reaction volumes with potassium acetate buffer, pH 6, with both Mg 2ϩ and Mn 2ϩ as the divalent cation. The reactions were carried out for 2 h at 37 and 70°C for T4 and RM378 PNK, respectively. As seen in Fig.  5D, the two enzymes exhibited different substrate preferences. Although T4 PNK showed good phosphohydrolase activity on 3Ј-TMP and the oligomer 3Ј-phosphate, the RM378 PNK revealed much better activity on the cAMP versus 3Ј-TMP and no detectable activity on the d(A 15 )-3Ј-PO 4 Ϫ oligomer under the experimental conditions. DISCUSSION The RM378 polynucleotide kinase is a bifunctional enzyme that carries out the same or very similar processes as those of the T4 polynucleotide kinase. The T4 PNK and RNA ligase will mend and ligate broken tRNA molecules after cleavage with the anticodon nuclease in prr ϩ E. coli strains. The anticodon nuclease suicidal mechanism has the purpose of limiting the T4 phage infection and is an interesting example of altruistic behavior among bacteria (7,8). Our studies suggest that the phage RM378 needs to counter similar RNA degradation mechanisms in R. marinus. This observation is based on the fact that the bacteriophage seems to be armed with both RNA ligase (2) and polynucleotide kinase.
The PNK and the RNA ligase are two components with a common purpose and presumably act in conjunction with each other. It is therefore not surprising to find the corresponding genes located close to each other in the genome of bacteriophage T4, separated by ϳ1400 bp and running in the same direction. It may seem logical to expect similar organization in the genome of bacteriophage RM378, although the overall general organization of the two phage genomes is not similar as judged from identified homologous genes (4). Intensified efforts to locate the PNK gene after our discovery of the RNA ligase gene did not identify a likely candidate open reading frame in the immediate vicinity of the RNA ligase gene, but the PNK gene was finally located still relatively nearby, ϳ6400 bp downstream but running in the opposite direction.
The RM378 PNK only shared its 5Ј-kinase domain with the PNK family and revealed no apparent homology to the 3Јphosphatase domain in that family. Similarity with the T4 5Ј-kinase is still low, but they share the P-loop GXXXXGK(S/T) motif, which is characteristic of many phospho-transferase families. The 5Ј-kinase domain is located on the C-terminal end of the RM378 PNK in contrast to the T4 PNK, suggesting that some kind of domain rearrangement had taken place, because known polynucleotide kinases from viral origin have the same domain arrangement as that of the T4 PNK. T4 PNK and homologs found previously in other bacteriophages, including coliphage RB69, phage Aeh1, and, very recently in the mycobacteriophages Omega and Cjw1 and the vibriophage KVP40, all have the kinase-phosphatase order of domains in contrast to the phosphatase-kinase order in mammalian PNKs (25). Although distinctly different from both of these groups because of its unique phosphatase domain, RM378 PNK resembles the mammalian PNKs rather than other phage PNKs in terms of domain arrangement (25). This is a very interesting observation, because eukaryotic PNKs function as repair enzymes on double-stranded DNA. The eukaryotic PNKs show relatively good homology to the viral 5Ј-kinase domains but less to the 3Ј-phosphohydrolase domains and contain relatively analogous activity, although they only phosphorylate DNA.
BLAST results revealed that the 3Ј-phosphohydrolase domain of RM378 PNK was related to the HD superfamily of phosphohydrolases, defined by the characteristic HD motif (23). The HD phosphohydrolase superfamily is distantly related to the phosphodiesterase family. The natural substrates of members of the phosphodiesterase family are 3Ј:5Ј-cyclic AMP and GMP, but they are inhibited by other forms of NMP FIG. 4. Characterization of the PNK/5-kinase. A, effect of ADP on the 5Ј-kinase activity on 5Ј-hydroxylated single-stranded DNA (white bars), dephosphorylation of 5Ј-phosphorylated single-stranded DNA (black bars), and phosphate exchange reaction (gray bars). The 5Ј-kinase reaction was inhibited as the concentration of ADP increased. The dephosphorylation increased as the ADP concentration increased, and the exchange reaction was similar from 0 to 100 M ADP but showed Ͻ10% relative activity at 1 mM ADP. B, Lineweaver-Burk plot of the effect of ATP concentration on the 5Ј-kinase reaction. The such as AMP and 2Ј:3Ј-cyclic AMP. We did not discover any relationship between the RM378 PNK and the phosphodiesterase family. Sequence analysis and BLAST searches revealed that the RM378 PNK N-terminal domain shares similarity with the HD domains of poly(A) polymerases of bacterial origin. Poly(A) polymerases are responsible for mRNA adenylation, which is a control mechanism for RNA degradation in bacteria. Poly(A) polymerases belong to the nucleotidyl transferase superfamily, which includes CCA and the nucleotidyl transferases poly(A) polymerase and DNA polymerase ␤ (28 -30). Recently, Yakunin et al. characterized a template-independent nucleotidyltransferase, (ATP(CTP):tRNA nucleotidyltransferase that comprises two domains, an N-terminal nucleotidyltransferase domain and a C-terminal HD domain containing 2Ј:3Ј-cyclic phosphodiesterase and phosphatase activities (24). The RM378 PNK phosphohydrolase domain shows similar activity at elevated temperatures on 2Ј:3Ј-cyclic AMP, confirming the phosphohydrolase activity of the domain. The HD phosphohydrolase domains are found together with a variety of other types of domains displaying various domain architectures. Many of the proteins thus formed seem to have a function in nucleotide metabolism through a fusion of a HD domain to a nucleotidyl transferase, a helicase, or an RNA binding domain (23). The RM378 PNK is thus another variation on this theme.
A number of hypothetical genes, previously annotated as being similar to poly(A) polymerases such as those in E. coli (AAN81695), C. acetobutylicum (GenBank TM NP_347389), and, more interestingly, in Deinococcus radiodurans DRB0098 (GenBank TM NP_051631), showed limited similarity to the whole RM378 PNK protein and shared similar size and same domain orientation. These putative genes share both the 5Јkinase P-loop and the 3Ј-phosphohydrolase HD motif, which suggests that they are in fact polynucleotide kinases like the RM378 PNK. We therefore propose that these gene products constitute a new subfamily of polynucleotide kinases. Apart from the data presented here, evidence supporting existence of this HD-Adk PNK subfamily is already accumulating (31,32).
First, D. radiodurans contains a RNA ligase of family 2 involved in tRNA repair, revealing a potential need for a PNK involved in the mending of RNA ends (33). Second, Liu et al. showed that the D. radiodurans DRB0098 hypothetical gene, which is in an operon with two genes, including an ATP-dependent DNA ligase-like gene, showed a recA-like expression pattern. That is, the gene was induced in the recovering phase after the irradiation of D. radiodurans cells. This may suggest that the DRB0098 gene product plays a part in nucleic acid repair in D. radiodurans (31,32).
Although the fold of the phosphatase domains in T4 and RM378 PNKs may be quite different and the order of the kinase and phosphatase domains reversed, it is not excluded that the relative spatial location of these domains in the two enzymes may be quite similar within the monomer or within a tetramer. Interestingly, the structural organization of the T4 PNK tetramer suggests that different ends at the site of the cut in cleaved tRNA may access both kinase and phosphatase active sites simultaneously (16). However, it is still possible that the domains could be acting independently in the sense that a certain spatial relationship between the domains is not strictly required and may thus be different in the two phage proteins.
Characterization of the RM378 PNK protein showed that it was a moderately thermostable protein with an apparent temperature optimum of 70°C and was stable up to 60 -65°C. This temperature optimum is consistent with the R. marinus natural environment as well as with R. marinus optimum growth conditions at 65°C (1). The PNK exhibits high activity on both RNA and DNA and has similar activity on double-stranded blunt end DNA as compared with single-stranded DNA oligomers. The 5Ј-kinase activity was dependent upon a divalent cation, and both Mg 2ϩ and Mn 2ϩ worked equally well. The main component that increased the activity of the protein was PEG 6000, where a 5-15% addition resulted in increasing the activity 3-4-fold. Similar effects of addition have been reported from nucleotidyl transferase enzymes like T4 PNK and from T4 FIG. 5. Characterization of the 3-phosphohydrolase activity of RM378 PNK. A, pH profile, measured in potassium acetate buffer from pH 4 to 6 (squares) and MOPS from pH 6 to 9 (diamonds). pH optimum was determined to be 6. B, apparent temperature optimum was 75°C measured for 1 h, but the enzyme is not stable for an extended time at temperatures of Ͼ65°C. C, titration of cAMP (squares) and 3Ј-TMP (diamonds) under standard assay conditions with Mn 2ϩ as the cation clearly shows that the 3Ј-phosphohydrolase activity is severalfold higher on cAMP than on 3Ј-TMP. D, comparison of T4 and RM378 PNK/3Ј-phosphohydrolase activity using 0.2 mM cAMP (white bar), 3Ј-TMP (black bar), or d(A 15 )-3ЈPO 4 Ϫ oligomer (gray bar) using Mn 2ϩ or Mg 2ϩ as the cation for the reaction. The substrate specificity of the two enzymes is clearly different. RM378 PNK showed high activity on cAMP and some activity on 3Ј-TMP but no activity on the oligomer and much less (20%) activity on cAMP in the presence of Mg 2ϩ . On the other hand, T4 PNK had similar activity on 3Ј-TMP and the oligomer but much less activity on the cAMP. DNA and RNA ligase (34 -37). This work did not reveal whether this increase in activity was due to stabilization effects on the protein or to the steric exclusion caused by PEG and many other bio-polymers (38).
The RM378 PNK/5Ј-kinase activity was inhibited in the presence of ADP. When only ADP and 5Ј-phosphorylated oligomer were present, dephosphorylation of the oligomer was observed. These results demonstrate that the catalyzed reaction is reversible, as is the case with T4 PNK (39 -41). When ATP, ADP, and a 5Ј-hydroxylated oligomer were incubated together in a mixture, an increasing concentration of ADP inhibited the 5Ј-kinase reaction. These observations suggest that the two components (ADP and ATP) are competing for the nucleotide-binding site. Interestingly, the exchange reaction was not significantly affected by the ADP concentration apart from inhibition of the kinase reaction at a high concentration. The overall exchange phosphorylation ranged from 2 to 4% as compared with the kinase activity on a 5Ј-hydroxylated oligomer in presence of 0 to 0.1 mM ADP (Fig. 4D). This exchange activity is relatively low, and the ADP independence may be caused by the following: (i) substrate-independent conversion of ATP to ADP and P i upon binding of the ATP and protein as seen with T4 PNK (18); or (ii) the possibility that the high temperature causes some minor degradation of the 5Ј-phosphate group of the oligomer, causing it to be re-phosphorylated and, in the process, generating free ADP and starting the exchange reaction.
The structure of T4 PNK reveals a narrow entrance to the kinase active site (18). Modeling of substrate binding suggest that only a single-stranded nucleic acid substrate can be accommodated at the binding site and, thus, provides a possible explanation for the relatively lower efficiency of the enzymecatalyzed phosphorylation using blunt end, double-stranded DNA substrates. An enzyme working at higher temperatures, where strand separation in double-stranded nucleic acids is facilitated, might thus be more efficient for phosphorylation of double-stranded DNA with blunt ends or 3Ј-overhangs. We observed similar activity on blunt end, double-stranded DNA and single-stranded DNA in our 5Ј-kinase activity assays.
The 3Ј-phosphohydrolase activity had somewhat different physical properties as compared with the kinase activity characteristics. Although the two domains shared thermostability properties and apparent temperature optima, they had little else in common. The first striking observation was that the pH optimum of the phosphohydrolase was 6 compared with 8 -9 of the kinase domain. Both domains were, however, active around pH 7. This is a similar scenario as the one seen with the T4 PNK domains and may suggest that the protein domains merged relatively late in evolutionary history. This, along with sequence data and data from the structural organization of the T4 PNK, suggested that the kinase and phosphohydrolase activity are on different protein domains (16 -18,20). Another difference was the preference for Mn 2ϩ compared with Mg 2ϩ , resulting in much more phosphohydrolase activity. Even though the activity was less in presence of Mg 2ϩ on cAMP, the enzyme was able to remove 2Ј:3Ј-cyclic phosphate from cAMP in the presence of Mg 2ϩ .
When comparing the virus-derived tRNA repair systems known today, it becomes clear how a variety of solutions can be applied to solve the tRNA degradation problem. Both T4 and RM378 are armed with RNA ligase and PNK proteins, but RM378 has replaced one of the two PNK domains with an analogous domain and changed the orientation of the domains, resulting in similar activity (2,42). The existence of similar genes in bacteria like the DRB0098 gene in D. radiodurans suggests that the RM378 phage has retrieved the gene from a bacterial genome along its evolutionary history. The solution in the ACNV RNA ligase/PNK system has been to merge the two polypeptides into one multidomain protein with the same basic functions as the two above. In fact, the variety of phage solutions to the suicidal RNA degradation mechanism suggests that uptake and manipulation of host and environmental genetic material may be a major factor in host-virus relationship development.