Characterization of Polynucleotide Kinase/Phosphatase Enzymes from Mycobacteriophages Omega and Cjw1 and Vibriophage KVP40*

Coliphage T4 Pnkp is a bifunctional polynucleotide 5′-kinase/3′-phosphatase that catalyzes the end-healing steps of a RNA repair pathway. Here we show that mycobacteriophages Omega and Cjw1 and vibriophage KVP40 also encode bifunctional Pnkp enzymes consisting of a proximal 5′-kinase module with an essential P-loop motif, GXGK(S/T), and a distal 3′-phosphatase module with an essential acyl-phosphatase motif, DX- DGT. Biochemical characterization of the viral Pnkp proteins reveals several shared features, including an alkaline pH optimum for the kinase component, an intrinsic RNA kinase activity, and a homotetrameric or homodimeric quaternary structure, that distinguish them from the monomeric DNA-specific phosphatase/kinase enzymes found in mammals and fission yeast. Whereas the phage 5′-kinases differ from each other in their preferences for phosphorylation of 5′ overhangs, blunt ends, or recessed ends, none of them displays the preference for recessed ends reported for mammalian DNA kinase. We hypothesize that Pnkp provides phages that have it with a means to evade an RNA-damaging antiviral host response. Genetic complementation of the essential end-healing steps of yeast tRNA splicing by the Omega and Cjw1 Pnkp enzymes establishes their capacity to perform RNA repair reactions in vivo. A supportive correlation is that Omega and Cjw1, which are distinguished from other mycobacteriophages by their possession of a Pnkp enzyme, are also unique among the mycobacteriophages in their specification of putative RNA ligases.

When breakage of a 3Ј-5Ј phosphodiester results in the formation of 5Ј-PO 4 and 3Ј-OH termini at the break site, the ends can be rejoined to each other (or to novel partner strands) by the action of DNA-specific or RNA-specific polynucleotide ligases. However, when breakage occurs with the opposite polarity, resulting in 5Ј-OH and 3Ј-PO 4 (or 2Ј,3Ј-cyclic PO 4 ) termini, the broken ends must be "healed" before they can be sealed. Healing entails two steps: (i) hydrolysis of the 3Ј-PO 4 (or 2Ј,3Ј-cyclic phosphate) to form a 3Ј-OH and (ii) phosphorylation of the 5Ј-OH to form a 5Ј-PO 4 end.
The bacteriophage T4 proteins polynucleotide kinase/phosphatase (Pnkp) and RNA ligase 1 (Rnl1) are the prototypal RNA healing and sealing enzymes (1-3, 8 -12). T4 Pnkp and Rnl1 function in vivo to repair a break in the anticodon loop of Escherichia coli tRNA Lys triggered by phage activation of a host-encoded anticodon nuclease PrrC (18). Depletion of tRNA Lys by PrrC blocks phage protein synthesis and arrests the infection before it can spread. However, Pnkp and Rnl1 repair the broken tRNAs and thereby evade the host defense mechanism.
T4 Pnkp is a homotetramer of a 301-aa 1 polypeptide (15, 16, 19 -21) (Fig. 1B). Essential constituents of the separate active sites for the 5Ј-kinase and 3Ј-phosphatase activities have been identified by mutagenesis and crystallography (15,16,21,22). Amino acids required for catalysis of the 5Ј-kinase reaction map to the N-terminal half ( Fig. 1) (21,22). The kinase active site is composed of (i) a classical P-loop motif ( 9 GXXGXGKS 16 ) that coordinates the ␤-phosphate of the NTP donor; (ii) essential side chain Arg 126 , which also coordinates the NTP ␤-phosphate; (iii) essential side chain Arg 38 , which coordinates the 3Ј-phosphate of the 5Ј-OH acceptor; and (iv) essential side chain Asp 35 , which has been suggested to function as a general acid to activate the 5Ј-OH for direct nucleophilic attack on the NTP ␥-phosphate. Residues essential for the 3Ј-phosphatase function cluster in the C-terminal half ( Fig. 1) (21,22). Two of the essential phosphatase residues, Asp 165 and Asp 167 , are located within a 165 DXDXT 169 motif that defines a superfamily of phosphotransferases that act through a covalent aspartyl phosphate intermediate (23). Mutational results and comparison of acyl-phosphatase structures (24,25) suggest a mechanism whereby Asp 165 acts as the nucleophile to attack the polynucleotide 3Ј-PO 4 and form the acyl-phosphoenzyme and Asp 167 serves as a general acid catalyst to expel the polynucleotide 3Ј-O leaving group. The native T4 Pnkp tetramer is assembled via pairwise kinase-kinase and phosphatase-phosphatase dimer interfaces (15,16).
It is noteworthy that the host bacterium E. coli appears not to encode homologs of T4 polynucleotide kinase and RNA ligase; nor do most other bacterial species for which genomic sequence is available. The closest relative of the T4 RNA repair enzyme system is the multifunctional Rnl1 protein of the Autographa californica nucleopolyhedrovirus (AcNPV). AcNPV is a prototype of the baculovirus family of eukaryotic DNA viruses, which replicate in arthropod hosts. The 694-aa baculovirus Rnl1 protein consists of an autonomous N-terminal RNAspecific ligase domain and an autonomous C-terminal endhealing domain (26). The end-healing domain is composed of a proximal polynucleotide kinase module with an essential Ploop motif and a distal polynucleotide 3Ј-phosphatase module with an essential acyl-phosphatase motif. The Rnl1 kinasephosphatase domain has a homodimeric quaternary structure. The similarities between the phage T4 and AcNPV enzymes prompted the speculation that AcNPV may have acquired an RNA repair system to contend with an RNA-damaging antiviral pathway that exists in its natural ecological niche (26). Although T4 Pnkp is an RNA repair enzyme in vivo, it modifies either RNA or DNA ends in vitro (1)(2)(3). The baculovirus homolog also can act on DNA ends in vitro (26). Mammalian cells have separate DNA-specific and RNA-specific polynucleotide kinase activities (27)(28)(29)(30). The mammalian RNA kinase has an alkaline pH optimum (30), whereas the DNA-specific kinase has a characteristic acidic pH optimum (27)(28)(29). Whereas the human RNA kinase does not have an associated 3Ј-phosphatase (30), the eukaryal DNA kinases do have an intrinsic DNA-specific 3Ј-phosphatase function (31)(32)(33). In metazoans and fission yeast, such bifunctional DNA-specific Pnkp enzymes assist in the repair of DNA damage induced by oxidation, radiation, and topoisomerase I poisons (34 -37). The eukaryal DNA-specific Pnkp enzymes are distinguished structurally from the phage T4 Pnkp in two major respects. First, the domain organization is inverted in the eukaryal enzymes (i.e. the acyl-phosphatase module precedes a C-terminal kinase module) (Fig. 1A). Second, the eukaryal Pnkp has a monomeric quaternary structure (38). Based on available data, we can classify two groups of Pnkp proteins that arose via fusions of ancestral P-loop and DXDXT phosphoryl transfer domains: (i) the oligomeric T4-like kinase-phosphatase flavor that appears to mediate RNA repair reactions and (ii) the monomeric eukaryal-type phosphatase-kinase group that is dedicated to DNA repair reactions.
A major impediment to evaluating the merits of this classification and, indeed, to understanding the prevalence and relevance of RNA repair pathways is that T4-like Pnkp proteins have only been characterized from T4 and AcNPV viruses. Broadly speaking, RNA-based "immune" responses to viral infection are common in virology, as exemplified by the key role of double-stranded RNA, RNase L, and the RNA-activated protein kinase PKR in the interferon response pathway in animals (39). By analogy to the T4 tRNA restriction and eukaryal RNase L systems, it is conceivable that many viruses trigger the incision of an essential host or viral RNA molecule, thereby impeding virus replication or pathogenesis, in which case such viruses might encode T4-like RNA repair functions. Here we identify and characterize homologs of T4 Pnkp from two mycobacteriophages, Omega and Cjw1 (40), and the vibriophage KVP40 (41). The recurrence of Pnkp enzymes in disparate viral pathogens suggests that RNA damage-based responses to virus infection may be widespread.

Recombinant Omega Pnkp and Cjw1 Pnkp
The Omega gp136 and Cjw1 gp89 genes encoding Pnkp homologs were amplified by PCR with Pfu DNA polymerase using primers designed to introduce an NdeI restriction site at the start codon and a BamHI site 3Ј of the stop codon. The template DNAs used for amplification (a pBluescript plasmid containing the Omega gp136 gene and Cjw1 genomic DNA, respectively) were a generous gift of Dr. Graham Hatfull (University of Pittsburgh). The PCR products were digested with NdeI and BamHI and inserted into pET16b (Novagen) to generate expression plasmids encoding the phage Pnkp polypeptides fused to an N-terminal His 10 tag. Alanine substitution mutations were introduced by PCR using the two-stage overlap extension method. The inserts of the wild-type and mutant pET-Pnkp plasmids were sequenced completely to exclude the acquisition of unwanted changes during amplification and cloning.
Wild-type and mutant pET-Pnkp plasmids were transformed into E. coli BL21(DE3). Cultures (100 ml) of E. coli BL21(DE3)/pET-Pnkp were grown at 37°C in Luria-Bertani medium containing 0.1 mg/ml ampicillin until the A 600 reached 0.6. The cultures were adjusted to 0.5 mM isopropyl-D-thiogalactopyranoside and then incubated at 17°C for 15 h. Cells were harvested by centrifugation, and the pellets were stored at Ϫ80°C. All subsequent steps were performed at 4°C. Thawed bacteria were resuspended in 10 ml of lysis buffer (50 mM Tris-HCl, pH 7.5, 0.5 M NaCl, 10% sucrose, 15 mM imidazole). Lysozyme and Triton X-100 were added to final concentrations of 50 g/ml and 0.1%, respectively. The lysates were sonicated to reduce viscosity, and insoluble material was removed by centrifugation. The supernatants were applied to 1-ml columns of Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen, Chatsworth, CA) that had been equilibrated with lysis buffer. The columns were washed with 10 ml of lysis buffer and then eluted stepwise with 2-ml aliquots of buffers containing 50, 100, 200, 300, and 500 mM imidazole in 50 mM Tris-HCl (pH 7.5), 0.2 M NaCl, 10% glycerol. The polypeptide compositions of the fractions were monitored by SDS-PAGE. Omega Pnkp was recovered predominantly in the 200 mM imidazole eluate fraction (containing 3 mg of protein). Cjw1 Pnkp was recovered predominantly in the 300 mM imidazole eluate fraction (containing 5 mg of protein). The protein concentrations were determined by using the Bio-Rad dye reagent with bovine serum albumin as the standard. The recombinant Pnkp proteins were stored at Ϫ80°C.

Recombinant KVP40 Pnkp
KVP40 gene 90 encoding a Pnkp homolog was amplified by PCR from KVP40 genomic DNA (a generous gift of Dr. Eric Miller, North Carolina State University) using primers designed to introduce an NdeI restriction site at the start codon and a BamHI site 3Ј of the stop codon. The PCR product was digested with NdeI and BamHI and inserted into pET16b to generate an expression plasmid encoding KVP40 Pnkp fused to an N-terminal His 10 tag. The pET-Pnkp plasmid was transformed into E. coli BL21(DE3). A 1-liter culture of E. coli BL21(DE3)/pET-Pnkp was grown and induced to express the KVP40 protein as described above for the mycobacteriophage proteins. The thawed bacterial pellet was resuspended in 20 ml of buffer A (50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 10% sucrose). Lysozyme and Triton X-100 were added to final concentrations of 50 g/ml and 0.1%, respectively. The lysate was sonicated to reduce viscosity and insoluble material was removed by centrifugation. The soluble extract was applied to a 2-ml column of Ni 2ϩ -nitrilotriacetic acid-agarose that had been equilibrated with buffer A containing 0.1% Triton X-100. The column was washed with 20 ml of the same buffer and then eluted stepwise with 3-ml aliquots of 50, 100, 200, 500, 1000, and 1200 mM imidazole in buffer B (50 mM Tris-HCl, pH 8.0, 0.2 M NaCl, 10% glycerol, 0.05% Triton X-100). The polypeptide compositions of the column fractions were monitored by SDS-PAGE. KVP40 Pnkp was recovered predominantly in 1000 mM imidazole eluate fraction (containing 12 mg of protein).

Polynucleotide Kinase Assays
Oligonucleotide Substrate-Reaction mixtures contained a 5Ј-OH oligodeoxyribonucleotide or oligoribonucleotide substrate, [␥-32 P]ATP, and other components as specified in the figure legends. Oligodeoxyribonucleotides were purchased from BIOSOURCE International (Camarillo, CA). Oligoribonucleotides were purchased from Dharmacon (Lafayette, CO) and deprotected as instructed by the vendor. Concentrations of oligonucleotide stock solutions were determined by UV absorbance at 260 nM, using extinction coefficients calculated according to their base compositions. The kinase reactions were initiated by adding Pnkp and terminated with formamide/EDTA. The products were analyzed by electrophoresis through a 15-cm polyacrylamide gel (either 12 or 15% acrylamide) containing 7 M urea in 90 mM Tris borate, 2.5 mM EDTA. The radiolabeled oligonucleotide products of expected size were visualized by autoradiography of the gel and quantified by scanning the gel with a Fujix BAS2500 imaging apparatus. Positions of amino acid side chain identity or similarity in all seven proteins are indicated by filled circles. Conserved residues identified by mutational analysis as essential for the 5Ј-kinase and 3Ј-phosphatase activities of T4 Pnkp activity are shaded in gray. The secondary structure of T4 Pnkp is shown above the sequence, with ␤-strands depicted as arrows and ␣-helices as cylinders.
were visualized and quantified by scanning the TLC plate with a Fujix BAS2500 imaging apparatus.
Linear Plasmid DNA Substrate-Circular pUC19 plasmid DNA was digested with BamHI, KpnI, or SmaI, and the linear products were recovered by extraction with phenol/chloroform and precipitation with ethanol. The 5Ј-PO 4 termini were removed by treatment of the DNA with calf intestine alkaline phosphatase (New England Biolabs), and the DNA was again recovered by extraction with phenol/chloroform and precipitation with ethanol. The DNA was resuspended in water, and its concentration was determined by UV absorbance at 260 nM. Kinase reaction mixtures (20 l) containing 50 mM Tris-HCl (pH 7.5 for Cjw1 Pnkp, pH 9.0 for Omega Pnkp, or pH 7.6 for T4 Pnkp), 5 mM MgCl 2 , 2 mM DTT, 1 g of linear 5Ј-OH pUC19 DNA, either 1 M or 50 M [␥-32 P]ATP, and 1 g of Cjw1, Omega, or T4 Pnkp were incubated for 20 min at 37°C. The reactions were quenched with EDTA, and the DNA was purified using a QIAquick PCR purification kit (Qiagen) according to the vendor's instructions. The isolated DNA was analyzed by electrophoresis through a horizontal 1% agarose gel in 40 mM Tris acetate, 10 mM EDTA buffer with 0.02% ethidium bromide. The ethidiumstrained DNA was visualized by UV transillumination. The gel was then dried under vacuum on DEAE paper (Whatman), and the radiolabeled DNA was visualized by autoradiography.

Assay of Reversal of the Polynucleotide Kinase Reaction
Reaction mixtures containing 50 mM Tris (pH 6.5), 5 mM MgCl 2 , 5 mM DTT, 1 mM ADP, 5Ј-32 P-labeled oligodeoxyribonucleotide d(ATTC-CGATAGTGACTACA) as specified and Omega or Cjw1 Pnkp as specified were incubated at 37°C. Reactions were initiated by adding Pnkp. Aliquots (6 l) were withdrawn at the times specified and quenched immediately by adding 2 l of 100 mM EDTA. The quenched samples were applied to a polyethyleneimine-cellulose TLC plate, which was developed with 1 M formic acid, 0.5 M LiCl. Unlabeled ATP, ADP, and AMP markers were chromatographed in parallel. The [ 32 P]ATP product and the 32 P-labeled oligodeoxyribonucleotide substrate were visualized by autoradiography and quantified by scanning the plate with a Fujix BAS2500 imaging device.

3Ј-Phosphatase Assay
Reaction mixtures (25 l) contained 3Ј-TMP (Sigma) and other components as specified in the figure legends. The reactions were quenched by adding 75 l of cold water and 1 ml of malachite green reagent (purchased from BIOMOL Research Laboratories, Plymouth Meeting, PA). Release of phosphate was determined by measuring A 620 and interpolating the value to a phosphate standard curve.

New Bacteriophage Homologs of T4 Pnkp-
The recently completed genome sequence of the T4-like vibriophage KVP40 (41) revealed a predicted 305-aa gene product with 33% identity to the 301-aa T4 Pnkp polypeptide (Fig. 1). The conservation of Pnkp in KVP40 is notable given that only 26% of the predicted KVP40 proteins have homologs in phage T4 and only 36% of T4 proteins have homologs in KVP40 (41). The genome sequences of 10 mycobacteriophages isolated from diverse niches by selection for lytic growth in Mycobacterium smegmatis (40) revealed that two of these DNA viruses, Omega and Cjw1, also encoded putative homologs of T4 Pnkp. The Omega and Cjw1 genomes consist of 239 and 143 genes, respectively, of which 22 appear to be shared when stringent cut-off values are applied (40). The predicted Omega and Cjw1 Pnkp polypeptides are 317 and 315 aa long, respectively, and they have 64% sequence identity to each other but only 30% identity to T4 Pnkp. Additional Pnkp homologs are encoded by the coliphage RB69 and the Aeromonas hydrophila phage Aeh1. The similarities between T4 Pnkp and its phage and baculovirus homologs extend throughout the entire length of the Pnkp polypeptide. (The 57 positions of side chain identity/similarity in all of the aligned proteins are indicated by filled circles in Fig. 1B). The conserved positions include five amino acids in the N-terminal domain that were shown to be essential for the kinase activity of T4 Pnkp and nine amino acids in the C-terminal domain shown to be essential for the 3Ј-phosphatase activity of T4 Pnk (21,22). 2 The order of the catalytic modules in these phage proteins is consistently kinase-phosphatase, not the phos-2 H. Zhu and S. Shuman, unpublished data. phatase-kinase arrangement seen in the eukaryal DNA endhealing enzymes. To address whether the newly identified phage Pnkp homologs possess the functions imputed to them on the basis of sequence comparisons and to evaluate their functional and physical properties, we produced, purified, and characterized recombinant versions of the Omega, Cjw1, and KVP40 Pnkp proteins.
Mycobacteriophage Omega Pnkp-The wild-type 317-aa mycobacteriophage Omega Pnkp protein was produced in E. coli as a His 10 -tagged fusion and purified from the crude soluble bacterial extract by adsorption to nickel-agarose and elution with buffer containing imidazole. SDS-PAGE analysis showed that the protein preparation consisted principally of a 40-kDa polypeptide corresponding to His 10 -Pnkp ( Fig. 2A). Mutated versions K16A (in the putative kinase motif 13 GSGKS 17 ) and D175A (in the phosphatase motif 175 DIDGT 179 ) were produced and purified in parallel ( Fig. 2A).
The 5Ј-polynucleotide kinase activity of wild-type Omega Pnkp was demonstrated by the transfer of 32 P i from 12.5 M [␥-32 P]ATP to the 5Ј-OH terminus of an 37-mer oligodeoxyribonucleotide (at a concentration of ϳ4 M 5Ј-OH ends) to form a 5Ј-32 P-labeled oligonucleotide product that was resolved from free ATP by polyacrylamide gel electrophoresis. The extent of phosphoryl transfer was proportional to input Omega protein (Fig. 2B). From the slope of the titration curve, we calculated a specific activity of 0.1 pmol of phosphorylated 5Ј-ends/ng of protein, which corresponds to a turnover number of ϳ0.2 min Ϫ1 . The K16A mutation in the P-loop abolished 5Ј-kinase activity, whereas the D175A mutant retained 60% of wild-type specific activity (Fig. 2B). These findings confirmed that the polynucleotide kinase activity was intrinsic to the recombinant Omega protein and implicated the P-loop lysine in catalysis. The extent of end-labeling increased with ATP concentration up to 20 M and reached a plateau at 20 -60 M, at which point ϳ90% of the input 5Ј-OH substrate was phosphorylated (Fig.  3A). The reaction was half-saturated at ϳ7 M ATP. The 5Ј-kinase activity of Omega Pnkp was also evinced by the transfer of 32 P from 25 M [␥-32 P]ATP to 1 mM 3Ј-CMP to form [5Ј-32 P]pCp, which was resolved from the ATP substrate by PEI-cellulose TLC. Formation of pCp was proportional to the amount of Omega protein added (Fig. 2C). From the slope of the titration curve, we estimated a turnover number of ϳ1.6 min Ϫ1 . Whereas the K16A mutant protein was defective in phosphorylating 3Ј-CMP, the D175A mutant retained 68% of wild-type specific activity (Fig. 2C). The 3Ј-CMP kinase activity of Omega Pnkp was optimal at pH 8.5-9.5 in 50 mM Tris-HCl buffer and declined gradually to half of the peak value at pH 6.0 (not shown). The divalent cation requirement was satisfied best by magnesium in the range of 1-2 mM concentration; higher magnesium concentrations resulted in reduced activity (Fig. 3B). Manganese was about half as effective as magnesium in supporting kinase activity.
3Ј-Phosphatase activity was measured by the release of inorganic phosphate from deoxythymidine 3Ј-monophosphate (3Ј-TMP), which serves as a convenient substrate for bacteriophage T4 Pnkp (11,12). Phosphate release was measured colorimetrically using the malachite green method (42). Omega Pnkp hydrolyzed 3Ј-TMP, and the yield of product depended on input enzyme (Fig. 2D). We calculated a specific activity of 19 pmol of P i formed per ng of enzyme, corresponding to a turnover number of ϳ38 min Ϫ1 . The D175A mutation in the DX-DXT motif abolished the phosphatase activity of Omega Pnkp, whereas the K16A mutation in the P-loop had no effect (Fig.  2D). These findings confirmed that the observed 3Ј-phosphatase activity was intrinsic to the Omega protein and implicated Asp 175 as the active site nucleophile in phosphoryl transfer. The 3Ј-phosphatase activity was optimal at pH 5.5 in Tris acetate buffer and declined sharply as the pH was either lowered to Յ5 or raised to Ն7.0 (Fig. 3C). The phosphatase reaction required a divalent cation cofactor, and this requirement was satisfied by either magnesium, manganese, or cobalt at their optimal concentrations of Ն10 mM divalent cation (Fig. 3D).
The native size of Omega Pnkp was gauged by sedimentation through a 15-30% glycerol gradient. Marker proteins catalase (248 kDa), BSA (66 kDa), and cytochrome c (13 kDa) were included as internal standards. His 10 -Pnkp (calculated to be a 40-kDa polypeptide) sedimented as a discrete peak between BSA and catalase (Fig. 4A). A plot of the S values of the three standards versus fraction number yielded a straight line (not shown), and an S value of 6.1 was determined for Omega Pnkp by interpolation to the internal standard curve. The 5Ј-kinase and 3Ј-phosphatase activity profiles were coincident with the Pnkp polypeptide (Fig. 4B). Because the apparent S value of Omega Pnkp was larger than expected for an 80-kDa globular homodimer and less than expected for a globular 160-kDa homotetramer (and less than the S value of 7.5 observed for the homotetrameric recombinant T4 Pnkp; data not shown), we suspect that the Omega protein is either a stable but asymmetrically shaped homotetramer or a metastable tetramer that is prone to dissociate to a homodimer during the velocity sedimentation procedure.
Mycobacteriophage Cjw1 Pnkp-The wild-type 315-aa mycobacteriophage Cjw1 Pnkp protein was produced in E. coli as a His 10 -tagged fusion and purified from the crude soluble bacterial extract by Ni 2ϩ -agarose chromatography; mutants K15A and D169A were purified in parallel (Fig. 5A). Wild-type Cjw1 Pnkp catalyzed the transfer of 32 P i from 12.5 M [␥-32 P]ATP to the 5Ј-OH terminus of a 37-mer oligodeoxyribonucleotide. We calculated a specific activity of 0.13 pmol of phosphorylated 5Ј-ends/ng of protein (Fig. 5B), which corresponds to a turnover number of ϳ0.26 min Ϫ1 . The K15A mutant was seemingly inert for 5Ј-kinase activity, but the D169A mutant retained half of the wild-type specific activity (Fig. 5B). The extent of oligonucleotide end-labeling increased with ATP concentration up to 100 M and saturated at 200 -400 M ATP, at which point all of the input 5Ј-OH ends were apparently phosphorylated (Fig.  6A). Half-saturation was attained at ϳ40 M ATP.
Cjw1 Pnkp also transferred 32 P from 25 M [␥-32 P]ATP to 1 mM 3Ј-CMP to form [5Ј-32 P]pCp with a turnover number of ϳ0.5 min Ϫ1 . The K15A mutation suppressed the 3Ј-CMP kinase activity, but the D169A change had little effect (Fig. 5C). The 3Ј-CMP kinase activity of Cjw1 Pnkp was optimal at pH 5.5-7.5 in 50 mM Tris acetate or Tris-HCl buffer and remained active up to pH 9.5. Activity declined sharply at pH Յ4.5 (Fig.  6B). The kinase reaction was divalent cation-dependent. Optimal activity was observed at 0.6 -2.5 mM cobalt; magnesium and manganese were about 60% as effective as cobalt at equivalent peak concentrations (Fig. 6C).
Cjw1 Pnkp catalyzed the release of P i from 3Ј-TMP with a turnover number of ϳ35 min Ϫ1 (Fig. 5D). The D169A mutation suppressed the 3Ј-phosphatase activity, but the K15A protein retained 58% of wild-type specific activity (Fig. 5D). The 3Јphosphatase activity was optimal at pH 5.5 in Tris acetate buffer and declined sharply as the pH was either lowered to Յ4.5 or raised to Ն7.0 (Fig. 6D). The Cjw1 phosphatase required a divalent cation cofactor, with the order of preference being cobalt Ͼ magnesium Ͼ manganese (Fig. 6E).
Cjw1 Pnkp (calculated to be a 39-kDa polypeptide) was sedimented in a 15-30% glycerol gradient with internal standards. The Cjw1 protein migrated as a discrete peak heavier than BSA (Fig. 4C). An S value of 5.7 was determined by interpolation to the internal standard curve. The 5Ј-kinase and 3Јphosphatase activities tracked with the Cjw1 Pnkp polypeptide (not shown). As discussed above for the Omega protein, the sedimentation behavior of Cjw1 Pnkp is consistent with either an asymmetrical homotetramer or a metastable homotetramer that dissociates to a homodimer during sedimentation.
Mycobacteriophage Pnkp Proteins Phosphorylate 5Ј-OH RNAs-The 5Ј-RNA kinase activity of Cjw1 Pnkp was demonstrated by the transfer of 32 P i from [␥-32 P]ATP to the 5Ј-OH terminus of a synthetic RNA oligonucleotide to form a 5Ј-32 Plabeled RNA product that was resolved from free ATP by poly-acrylamide gel electrophoresis (Fig. 7). RNAs with chain lengths of 18, 15, 12, or 9 nucleotides were phosphorylated to similar extents by Cjw1 Pnkp and T4 Pnkp (Fig. 7) and by Omega Pnkp (data not shown). The yields of 32 P-labeled RNA were 50 -90% of the input 5Ј-OH ends.

Omega and Cjw1 Pnkp Have Distinct Preferences for Protruding Versus Recessed 5Ј-OH Ends-The Omega and Cjw1
Pnkp enzymes resemble T4 Pnkp in their ability to phosphorylate the 5Ј-OH ends of DNA and RNA polynucleotides and nucleoside 3Ј-monophosphates. In contrast, the mammalian Pnkp enzyme phosphorylates 5Ј-OH DNA polynucleotides but fails to phosphorylate 3Ј-dNMPs (33). T4 Pnkp prefers to phosphorylate RNAs or DNAs with 5Ј single strand extensions and is poorly active on blunt duplex 5Ј-OH DNA termini or 5Ј-OH DNA termini that are recessed within duplex regions or at the junction of a 3Ј single strand tail. In contrast, the mammalian kinase preferentially phosphorylates 5Ј-ends recessed within duplex DNA structures (32).
To examine the end preferences of the mycobacteriophage kinases, we compared the recombinant T4, Omega, and Cjw1 Pnkp proteins for their ability to catalyze label transfer from [␥-32 P]ATP to linear pUC19 DNA with different end structures (Fig. 8). The plasmid substrates were prepared by digestion of pUC19 with one of three restriction endonucleases that cleave uniquely in the polylinker to leave either a 4-nucleotide 5Ј overhang (BamHI), a 4-nucleotide 3Ј overhang (KpnI), or a blunt duplex end (SmaI), followed by treatment with calf intestine phosphatase to remove the 5Ј-PO 4 groups. The DNAs were phenol-extracted and ethanol-precipitated to remove the phosphatase and then reacted with the various phage Pnkp enzymes in the presence of either 1 M ATP or 50 M ATP. The DNAs were freed of ATP and then analyzed by agarose gel electrophoresis. The DNA content of the gel was visualized by staining with ethidium bromide (shown for the 1 M ATP reaction in Fig. 8, bottom), and the extent of DNA end labeling was gauged by autoradiography of the dried gel (Fig. 8, top and  middle).
In agreement with Lillehaug et al. (43), we found that T4 Pnkp preferentially labeled the 5Ј single-stranded BamHI overhang when the reactions were performed at 1 M ATP (Fig. 8,  middle), and that the end specificity was relaxed when the ATP concentration was increased to 50 M. This contrasts with Omega Pnkp, which labeled 5Ј-OH termini at 5Ј overhangs, 3Ј overhangs, and blunt ends with similar efficiencies at either 1 or 50 M ATP. Cjw1 Pnkp phosphorylated either 5Ј overhangs or blunt ends but was poorly active on a recessed 5Ј-OH end at the junction of a 3Ј KpnI tail. This bias of Cjw1 Pnkp against recessed ends was evident at both 1 and 50 M ATP. The extent of labeling of the BamHI ends by Cjw1 Pnkp was about 10-fold higher at 50 M ATP versus 1 M ATP; this is consistent with the higher half-saturation point for ATP seen for the Cjw1 kinase activity on the single-stranded oligonucleotide substrate versus the Omega kinase activity. These experiments show that different bacteriophage polynucleotide kinases display distinctive phosphate acceptor specificities.
The Mycobacteriophage Polynucleotide Kinase Reaction Is Reversible-The crystal structure of the kinase domain of T4 Pnkp highlights a tunnel-like active site through the heart of the enzyme. The NTP and the 5Ј-OH polynucleotide substrates enter the active site from opposite ends of the tunnel and are coordinated in their respective donor and acceptor sites to achieve phosphoryl transfer via an in-line single-displacement mechanism entailing direct attack of the 5Ј-O on the ␥-phosphorus of ATP with inversion of stereochemical configuration at the transferred phosphate (44). The T4 kinase reaction is reversible (i.e. the enzyme can catalyze attack by ADP on a 32 P-end-labeled 5Ј-32 PO 4 -terminated polynucleotide to form [␥-32 P]ATP) (45).
We assessed the reversibility of the Omega and Cjw1 kinase reactions by incubating the recombinant enzymes with a 5Ј-32 PO 4 -terminated 18-mer DNA in the presence of 1 mM ADP. The reaction products were analyzed by PEI-cellulose TLC under conditions in which the oligonucleotide remains near the chromatographic origin and ATP, ADP, and AMP are resolved. Both mycobacteriophage enzymes transferred the label to ADP to generate a product comigrating with an ATP standard (Fig.  9, A and B). [ 32 P]ATP formation was time-dependent, and the reactions attained an end point at which up to 95% of the input 5Ј-32 P-DNA label was converted to [ 32 P]ATP. Omission of ADP from the reaction prevented formation of this product (Fig. 9A). The transfer of label from DNA to ADP by Omega Pnkp to form ATP was suppressed by the K16A mutation of the P-loop but unaffected by the D175A mutation of the DXDXT phosphatase motif (Fig. 9C). Similar effects were seen for the K15A and D169A mutations of Cjw1 Pnkp (not shown). We surmise that the forward and reverse kinase reactions are catalyzed by the same active site. The reverse kinase reactions of Omega and Cjw1 Pnkp were optimal at pH 5.5 and required a divalent cation cofactor (not shown). The amount of [ 32 P]ATP generated at pH 9.0 was about one-tenth of the yield of ATP at pH 5.5 (not shown). The acidic shift in the pH optimum of the reverse kinase reaction relative to the alkaline pH optimum of the forward kinase reaction of mycobacteriophage Pnkp recapitulates the findings of van de Sande et al. (45) for the reverse kinase of T4 Pnkp. The extent of [ 32 P]ATP formation by Omega Pnkp in a reaction containing 0.2 M 5Ј-32 P-DNA increased with the concentration of the ADP phosphate acceptor (Fig.  9D). Half-saturation was attained between 10 and 100 M ADP.
Mycobacteriophage Pnkp Enzymes Can Perform RNA Repair Functions in Vivo-The enzymatic steps in bacteriophage T4 tRNA restriction/repair are broadly similar to those of yeast tRNA splicing, the process whereby introns are removed seam- lessly from the tRNA anticodon loop (57). The incision steps in both cases result in the formation of 2Ј,3Ј cyclic phosphate and 5Ј-OH termini. tRNA splicing requires two breaks in the backbone of the pre-tRNA to excise the intron, whereas tRNA restriction involves a single break in the mature tRNA. The end-healing and strand-sealing steps of the phage T4-encoded tRNA repair pathway are performed by Pnkp and Rnl1, whereas a single enzyme Trl1 performs these steps in yeast tRNA splicing (57). The healing and sealing activities of Trl1 are essential for yeast viability (58,59). We showed recently that RNA repair systems are portable in vivo (i.e. that a lethal trl1⌬ mutation of S. cerevisiae can be rescued to normal growth by coexpression of bacteriophage T4 Rnl1 and Pnkp) (48). This approach provides a means to identify new RNA repair enzymes by functional complementation in yeast.
To test whether the mycobacteriophage Pnkp proteins could substitute in vivo for the end-healing components of yeast tRNA ligase, we cloned the Omega and Cjw1 Pnkp genes into a yeast centromeric plasmid under the control of the yeast SLU7 promoter. A plasmid shuffle complementation assay (59) revealed that coexpression of T4 Rnl1 and either Omega or Cjw1 Pnkp rescued the trl1⌬ mutation (not shown). trl1⌬ strains expressing T4 Rnl1 and Omega Pnkp grew on rich medium (yeast extract/peptone/dextrose agar) at 30 and 37°C, whereas trl1⌬ strains expressing T4 Rnl1 and Cjw1 Pnkp grew at 30°C but not at 37°C (not shown). Thus, Omega and Cjw1 Pnkp are both capable of healing the ends of broken tRNA molecules in vivo in yeast, although Cjw1 Pnkp function in yeast is temperature-sensitive.
Vibriophage KVP40 Pnkp-KVP40 Pnkp was produced in E. coli as a His 10 -tagged fusion and purified from a soluble bacterial extract by adsorption to nickel-agarose and step elution with imidazole. The recombinant protein was recovered in the 1 M imidazole eluate (Fig. 10A). The native size of KVP40 Pnkp was gauged by sedimentation through a 15-30% glycerol gradient with catalase, BSA, and cytochrome c included as internal standards. The His 10 -Pnkp polypeptide (with a calculated mass of 38 kDa) sedimented as a discrete peak coincident with BSA (Fig. 10B). The 5Ј-polynucleotide kinase activity profile (assayed by the transfer of 32 P i from 1 mM [␥-32 P]ATP to the 5Ј-OH terminus of a 34-mer oligodeoxyribonucleotide) and the 3Ј-phosphatase activity profile (measured as P i release from 3Ј-TMP) paralleled the distribution of the Pnkp polypeptide (Fig. 10C). We surmise that KVP40 Pnkp is a homodimer in solution.
KVP40 3Ј-phosphatase activity was proportional to input Pnkp (Fig. 11A); from the slope of the titration curve, we calculated a specific activity of 1.1 nmol of P i formed per ng of enzyme, corresponding to a turnover number of ϳ35 s Ϫ1 . KVP40 3Ј-phosphatase activity was optimal at pH 5.5 and declined as the pH was either lowered to Յ4.5 or raised to Ն8.0 (Tris-HCl) (Fig. 11B). KVP40 polynucleotide kinase activity was optimal at pH 7.5-9.5, declined steadily as the pH was lowered, and then was abolished at pH Յ5.0 (Fig. 11C). The extent of oligonucleotide end-labeling by the kinase increased with ATP concentration up to 50 M and reached a plateau at 100 -500 M ATP (Fig. 11D).

DISCUSSION
The present study advances our understanding of the endhealing phase of nucleic acid break repair on two fronts. First, by identifying and characterizing three new bacteriophage Pnkp enzymes, we provide evidence for family of T4-like proteins with shared domain organization, oligomeric quaternary structure, core enzymatic activities, and capacity to participate in RNA repair reactions in vivo. Second, the presence of T4-like Pnkp enzymes in diverse bacterial and eukaryotic viral niches implicates RNA repair as a widespread facet of virus-host dynamics.
The biochemical characterization of new bacteriophage Pnkp enzymes reveals both similarities and differences with respect to the well studied phosphatase and kinase reactions of T4 Pnkp and the recently described baculovirus Pnkp. The 3Јphosphatase components consistently display an acidic pH optimum: pH 6.0 for T4 and baculovirus (12,26); pH 5.5 for Omega, Cjw1, and KVP40. It is sensible that these 3Ј-phosphatases, as members of the DXDX(T/V) acyl-phosphatase superfamily, would display optimal activity at mildly acidic pH, because the phosphoryl transfer mechanism calls for an unprotonated aspartate nucleophile and a protonated aspartate gen-eral acid catalyst (24,25). The T4 and KVP40 phosphatases displayed greater catalytic power in hydrolyzing 1.6 mM 3Ј-TMP, with turnover numbers of 24 s Ϫ1 (22) and 35 s Ϫ1 , respectively, than did the Omega and Cjw1 enzymes, which had turnover numbers of 38 and 35 min Ϫ1 , respectively.
The kinase components of the bacteriophage Pnkp enzymes have in common the ability to phosphorylate the 5Ј-OH of DNA, RNA, and 3Ј-CMP. The crystal structure of the T4 polynucleotide kinase revealed the existence of a bidentate electrostatic interaction of the essential Arg 38 side chain to a sulfate ion occupying the acceptor site that was proposed to correspond to the 3Ј-phosphate of the terminal nucleotide of the 5Ј-OH acceptor (15). It was proposed that this dedicated contact to the 3Ј-phosphate accounts for the ability of T4 Pnkp to phosphorylate 3Ј-NMPs, whereas the lack of an equivalent arginine in the mammalian Pnkp enzyme explains its inability to utilize 3Ј-dNMP as a substrate. In this light, it is noteworthy that the side chain corresponding to Arg 38 of T4 Pnkp is conserved in all of the phage Pnkp proteins (Fig. 1). All of the phage Pnkp enzymes have an alkaline pH optimum for their kinase components, which distinguishes them from the mammalian DNA kinases, which display optimal activity at acidic pH (27)(28)(29). The most notable difference among the phage kinases is their preference for DNA substrates with 5Ј single strand overhangs, blunt duplex termini, or recessed 5Ј-ends with 3Ј single strand tails. We find that T4 Pnkp prefers 5Ј overhangs (especially at lower [ATP]), whereas Cjw1 Pnkp acts on 5Ј overhangs and blunt ends and is biased against a 3Ј overhang. Omega Pnkp is catholic in its ability to label all three types of ends with similar efficacy. Note that none of the phage Pnkp enzymes evinced the bias against a 5Ј single strand overhang and the accompanying preference for a recessed 5Ј-OH end that are characteristic of the mammalian DNA-specific kinase (32). The crystal structure of the T4 kinase reveals a tunnel-like aperture leading to the 3Ј-phosphate binding site of the phosphate acceptor that appears to be too narrow to allow facile ingress of duplex nucleic acid but can readily accommodate a single-stranded polynucleotide (15,16,56). The fact that high concentrations of the ATP phosphate donor can influence the accommodation of blunt or 3Ј recessed ends in the 5Ј-OH acceptor site of T4 kinase remains mysterious. The distinctive end preferences of the two mycobacteriophage kinases argues that their phosphate acceptor sites have a less constrained architecture compared with the T4 enzyme.
The bacteriophage Pnkp enzymes differ in their quaternary structures, as gauged by zonal velocity sedimentation. Whereas T4 Pnkp is a stable homotetramer, the KVP40 protein sediments as a homodimer under the same conditions. The T4 Pnkp homotetramer is formed by two pairs of kinase-kinase and phosphatase-phosphate homodimer interfaces (15,16). A dimeric KVP40 Pnkp could form either via a single phosphatasephosphatase or kinase-kinase interface or through a pair of intermolecular kinase-phosphatase interfaces. (We attempted to address this issue by producing various recombinant ver-sions of the N-terminal kinase and C-terminal phosphatase domain segments of KVP40 Pnkp but found that the truncated proteins were either insoluble or poorly soluble and without detectable enzymatic activity when purified.) The Omega and Cjw1 Pnkp proteins behave either as asymmetrically shaped homotetramers or metastable tetramers that dissociate to homodimers during sedimentation. Initial studies had also suggested that T4 Pnkp vacillates between tetrameric and dimeric states (18,46,47). These subtleties notwithstanding, we conclude that viral Pnkp enzymes of the T4-like flavor consistently adopt an oligomeric quaternary structure that distinguishes them from the monomeric DNA-specific Pnkp enzymes found in mammals.
T4 Pnkp is inessential in standard laboratory strains of E. coli used to study T4 replication, which lack the prr locus encoding the cellular anticodon nuclease, but essential for phage replication in E. coli host strains that carry the prr operon (49,50). Given the similarities between the T4 Pnkp and the newly described Pnkp enzymes of Omega, Cjw1 and KVP40, it is sensible to speculate that many bacterial viruses might possess an RNA repair enzyme system to contend with an RNA-damaging antiviral pathway that exists in their natural ecological niche. The biological target of the T4 RNA repair pathway is a broken tRNA (18). It is conceivable that Omega and Cjw1 (and KVP40) infection also trigger tRNA restriction or perhaps incision of a different essential host or viral RNA molecule, which impedes virus replication. The agent of the posited RNA damage response may be surmised to a certain extent from comparative genomics. The sequenced strain of Vibrio cholerae does not encode an obvious homolog of the E. coli anticodon nuclease PrrC, but other species of Vibrio do, specifically Vibrio vulnificus and Vibrio parahaemolyticus (GenBank TM accession numbers NP_933608 and NP_800765, respectively). Indeed, V. parahaemolyticus was the source strain for the isolation of KVP40 (51). Putative PrrC-like anticodon nuclease homologs are also found in the proteomes of Neisseria meningitidis, Helicobacter pylori, Campylobacter jejuni, Corynebacterium diphtheriae, Streptococcus mutans, Xanthomonas campestris, Staphylococcus aureus, Pseudomonas putida, Chlorobium tepidum, Photorhabdus luminescens, Ralstonia solanacearum, and Magnetococcus. The sequenced Mycobacterium tuberculosis and Mycobacterium leprae strains do not encode obvious homologs of E. coli PrrC. However, as noted for E. coli, it is possible that "wild" strains of mycobacteria (including M. smegmatis) do possess PrrC-type RNA restriction enzymes. Alternatively, mycobacteria may have inducible RNA restriction systems unrelated to PrrC.
The limited distribution of Pnkp among the 10 DNA mycobacteriophages selected for growth on M. smegmatis suggests that RNA repair is an acquired capacity that may be relevant to virus host range. If M. smegmatis (or some other natural host for these 10 mycobacteriophages) does have an RNA restriction system, we would predict that only the mycobacteriophages with an RNA repair pathway are capable of inducing an RNA damage response. This idea is consistent with the findings that activation of the PrrC anticodon nuclease by T4 infection depends on a specific T4-encoded effector protein Stp (52).
The general hypothesis that Pnkp provides phages that have it with a means to evade an RNA damage host response is plausible only if there is an RNA ligase available to seal the ends that are healed by phage Pnkp. T4 encodes two different RNA ligase enzymes, Rnl1 and Rnl2 (53). Rnl1 is clearly implicated in the tRNA restriction/repair pathway (18). The role of T4 Rnl2 in vivo is unclear. Vibriophage KVP40, which we show specifies a Pnkp enzyme, encodes its own homologs of both Rnl1 and Rnl2 (41,54). The provocative correlation is that Omega and Cjw1, which are distinguished from other mycobacteriophages by their possession of a Pnkp enzyme, are also unique among the mycobacteriophages in their specification of putative Rnl2-like ligases (Fig. 12). The primary structures of the Omega gp162 (446-aa) and Cjw1 gp93 (441-aa) Rnl2-like proteins have 67% pairwise side chain identity, and they include counterparts of nucleotidyl transferase motifs I, III, IIIa, IV, and V that comprise the active site of T4 Rnl2 (17,55). The p209 protein of Aeromonas salmonicida bacteriophage 44RR2.8t is another putative viral Rnl2-like ligase with extensive similarity to the mycobacteriophage Omega and Cjw1 proteins (Fig. 12). Individual side chains identified as essential for the ligase activity of T4 Rnl2 are conserved in the Omega and Cjw1 proteins (these are indicated by vertical lines in Fig. 12). Omega and Cjw1 appear not to have homologs of T4 Rnl1. Moreover, they do not encode any DNA ligase homolog. Thus, we speculate that the mycobacteriophage Pnkp enzymes collaborate with mycobacteriophage Rnl2-like enzymes to comprise an RNA repair system.