Structure-Function Analysis of the Phosphoesterase Component of the Nucleic Acid End-Healing Enzyme Runella slithyformis HD-Pnk

5′-end healing and 3′-end healing are key steps in nucleic acid break repair in which 5′-OH ends are phosphorylated by a polynucleotide kinase, and 3′-PO₄ or 2′,3′-cyclic-PO₄ ends are hydrolyzed by a phosphoesterase to generate 5′-PO₄ and 3′-OH termini needed for joining by DNA and RNA ligases. This study interrogates, biochemically and via mutagenesis, the phosphoesterase activity of Runella slithyformis HD-Pnk, a bifunctional bacterial 5′- and 3′-end-healing enzyme composed of HD phosphoesterase and P-loop kinase modules. HD-Pnk homologs are found in 129 bacterial genera from 11 phyla. In 123/129 instances, HD-Pnk is encoded in an operon-like gene cluster with a putative ATP-dependent polynucleotide ligase (LIG), suggesting that HD-Pnk and LIG are agents of a conserved bacterial nucleic acid repair pathway.

ABSTRACT

Runella slithyformis HD-Pnk is the prototype of a family of dual 5′ and 3′ nucleic acid end-healing enzymes that phosphorylate 5′-OH termini and dephosphorylate 2′,3′-cyclic-PO₄, 3′-PO₄, and 2′-PO₄ ends. HD-Pnk is composed of an N-terminal HD phosphohydrolase module and a C-terminal P-loop polynucleotide kinase module. Here, we probed the phosphoesterase activity of HD-Pnk by querying its ability to hydrolyze non-nucleic acid phosphoester substrates and by conducting a mutational analysis of conserved amino acid constituents of the HD domain. We report that HD-Pnk catalyzes vigorous hydrolysis of p-nitrophenylphosphate (K_m = 3.13 mM; k_cat = 27.8 s⁻¹) using copper as its metal cofactor. Mutagenesis identified Gln28, His33, His73, Asp74, Lys77, His94, His127, Asp162, and Arg166 as essential for p-nitrophenylphosphatase and DNA 3′ phosphatase activities. Structural modeling places these residues at the active site, wherein His33, His73, Asp74, His94, and His127 are predicted to coordinate a binuclear metal complex and Lys77 and Arg166 engage the scissile phosphate. HD-Pnk homologs are distributed broadly (and exclusively) in bacteria, usually in a two-gene cluster with a putative ATP-dependent polynucleotide ligase (LIG). We speculate that HD-Pnk and LIG comprise the end-healing and end-sealing components of a bacterial nucleic acid repair pathway.

IMPORTANCE 5′-end healing and 3′-end healing are key steps in nucleic acid break repair in which 5′-OH ends are phosphorylated by a polynucleotide kinase, and 3′-PO₄ or 2′,3′-cyclic-PO₄ ends are hydrolyzed by a phosphoesterase to generate 5′-PO₄ and 3′-OH termini needed for joining by DNA and RNA ligases. This study interrogates, biochemically and via mutagenesis, the phosphoesterase activity of Runella slithyformis HD-Pnk, a bifunctional bacterial 5′- and 3′-end-healing enzyme composed of HD phosphoesterase and P-loop kinase modules. HD-Pnk homologs are found in 129 bacterial genera from 11 phyla. In 123/129 instances, HD-Pnk is encoded in an operon-like gene cluster with a putative ATP-dependent polynucleotide ligase (LIG), suggesting that HD-Pnk and LIG are agents of a conserved bacterial nucleic acid repair pathway.

INTRODUCTION

End-healing enzymes convert nucleic acid breaks with 3′-PO₄ (or 2′,3′-cyclic-PO₄) ends and 5′-OH ends to 3′-OH and 5′-PO₄ termini that serve as substrates for subsequent repair reactions. Polynucleotide kinases (Pnks) are 5′-end-healing enzymes that transfer the γ phosphate of a nucleoside triphosphate (NTP) donor to a 5′-OH polynucleotide acceptor. In many nucleic acid repair systems, a Pnk module is fused within a single polypeptide to a 3′-end-healing enzyme. The 3′-end-healing components belong to a variety of structurally and mechanistically distinct phosphoesterase enzyme families. For example, a DXDXT superfamily acyl-phosphatase in bacteriophage T4 Pnkp (1–3), a binuclear metallophosphoesterase superfamily enzyme in Clostridium thermocellum Pnkp (4, 5), and a 2H phosphoesterase superfamily enzyme in fungal and plant tRNA ligases (6, 7).

We recently identified and characterized Runella slithyformis and Deinococcus radiodurans HD-Pnk enzymes as the founders of a new dual 3′- and 5′-end-healing enzyme family (8, 9). HD-Pnk consists of an N-terminal module of the HD phosphoesterase domain clade fused to a C-terminal Pnk domain (Fig. 1). HD domain enzymes are binuclear metal-dependent phosphodiesterases and monoesterases that act on diverse substrates (10–24). HD enzymes participate in a broad range of biological functions.

FIG 1.

Genomic context and primary structure of Runella HD-Pnk. (Top) The ORF encoding Runella slithyformis HD-Pnk is preceded by three cooriented upstream ORFs encoding enzymes with potential roles in nucleic acid repair. (Bottom) The amino acid sequence of R. slithyformis (Rsl) HD-Pnk (NCBI accession AEI48611) is aligned to that of its homolog from E. coli UTI89 (Eco) (NCBI accession ABE08504). Positions of side chain identity/similarity are denoted by dots. Gaps in the alignments are denoted by dashes. The phosphoesterase-essential His73 and kinase-essential Asp254, and Arg257 residues identified previously by alanine scanning are denoted by “|” symbols. Asp254 and Arg257 are shaded blue. Amino acids in the HD domain that were targeted for further mutagenesis in the present study are highlighted in green shading. The kinase P-loop motif (GXXGXGK) that coordinates the NTP phosphates is outlined by the blue box.

The invocation of a DNA or RNA repair function for HD-Pnk was underscored by its gene context in the R. slithyformis and D. radiodurans chromosomes (8, 9). The HD-Pnk open reading frame (ORF) is located immediately downstream of a cooriented ORF encoding a polynucleotide ligase (LIG), in an operon-like fashion in which the 3′ end of the LIG ORF overlaps the 5′ end of the HD-Pnk ORF (Fig. 1). Whereas HD-Pnk is inessential for D. radiodurans growth, deletion of the gene encoding HD-Pnk sensitizes D. radiodurans to ionizing radiation and mitomycin C (9). Because HD-Pnk homologs are present in many bacterial taxa (8), often in an operon with LIG, we envision that the HD-Pnk and LIG activities are functionally related.

In the R. slithyformis chromosome, the LIG⋅HD-Pnk operon is separated by a 152-nucleotide spacer from another two-gene cassette that encodes Tpt1 and a putative methyltransferase–histidine-triad enzyme (Fig. 1). Runella Tpt1 is a biochemically well-characterized bona fide ortholog of yeast Tpt1 (25). Tpt1 is an RNA repair enzyme that removes an internal RNA 2′-phosphate (e.g., at the splice junction formed by fungal-type tRNA ligases) via a phosphoryl transfer to NAD⁺ (26, 27). Thus, it is conceivable that R. slithyformis HD-Pnk, Tpt1, and LIG might act together in a nucleic acid break repair pathway.

The primary structure of the 368-amino-acid R. slithyformis HD-Pnk polypeptide is shown in Fig. 1, aligned to the homologous protein from E. coli UTI89. The C-terminal half of HD-Pnk contains two hallmark motifs found in polynucleotide kinases: the GXXGXGK “P-loop” and a DXXR motif (28). The N-terminal portion of HD-Pnk contains the signature HD (His-Asp) motif that coordinates a metal cofactor.

Previously, we showed that recombinant Runella slithyformis HD-Pnk has 3′- and 5′-end-healing activities (8). The kinase activity of HD-Pnk phosphorylates 5′-OH polynucleotides (≥9 nucleotides in length) in the presence of magnesium using any standard NTP as the phosphate donor. HD-Pnk dephosphorylates RNA 2′,3′-cyclic-PO₄, RNA 3′-PO₄, RNA 2′-PO₄, and DNA 3′-PO₄ ends in the presence of a transition metal, which can be either nickel, copper, or cobalt. Copper was the most effective of the three metals in supporting DNA 3′ phosphatase activity (8). The 3′- and 5′-end-healing activities are functionally separable, i.e., mutation of Asp254 or Arg257 of the DXXR motif effaced kinase activity without affecting the phosphatase and mutation of His73 of the HD motif abolished 3′ phosphatase activity without affecting the kinase (8).

In the present study, we advance our characterization of the phosphoesterase activity of HD-Pnk in two directions, by querying its ability to hydrolyze non-nucleic acid phosphoester substrates and by conducting a mutational analysis of conserved amino acids of the HD domain.

RESULTS AND DISCUSSION

HD-Pnk is a copper-dependent p-nitrophenylphosphatase.

To probe the repertoire of HD-Pnk, we tested its activity with the generic phosphomonoesterase substrate p-nitrophenylphosphate. Hydrolysis of p-nitrophenylphosphate liberates p-nitrophenol, which is easily quantified by its absorbance at 410 nm. We found that HD-Pnk readily converted 10 mM p-nitrophenylphosphate into p-nitrophenol in the presence of 1 mM CuCl₂ (Fig. 2A). No activity was detected in the absence of a divalent cation or in the presence of 1 mM magnesium, calcium, or zinc. Activity in the presence of 1 mM nickel and cobalt was 8 and 2% of the level attained with 1 mM copper (Fig. 2A). No activity was detected in the presence of 1 mM FeCl₂ or FeCl₃ (not shown). Divalent cation titration showed that phosphomonoesterase activity was optimal at 0.25 to 1 mM copper and declined as the copper concentration was increased (Fig. 2B). Product formation at 5 mM copper was 54% of the value at the copper optimum. Nickel supported feeble activity at 0.25 to 5 mM concentration, i.e., 9 to 10% of the activity seen at the copper optimum (Fig. 2B). In parallel, we tested HD-Pnk for phosphodiesterase activity with the generic substrate bis-p-nitrophenylphosphate. No release of p-nitrophenol was detected in reaction mixtures containing 1 mM copper, nickel, magnesium, calcium, or zinc. Hydrolysis of bis-p-nitrophenylphosphate was detectable in the presence of 1 mM manganese or cobalt; however, the extents of p-nitrophenol formation in manganese and cobalt were 8 and 0.8% of the extent of p-nitrophenol release during reaction of HD-Pnk with p-nitrophenylphosphate in the presence of copper (not shown).

FIG 2.

HD-Pnk is a copper-dependent p-nitrophenylphosphatase. (A) Metal specificity. Reaction mixtures (25 μl) containing 100 mM Tris-acetate (pH 6.0), 10 mM (250 nmol) p-nitrophenylphosphate, 10 pmol (0.4 μM) of HD-Pnk, and either no added metal cofactor (lane –) or 1 mM the indicated divalent cation (added as chloride salt) were incubated at 37°C for 10 min. The extents of p-nitrophenol product formation are plotted in bar graph format. Each datum is the average from three separate experiments ± the standard errors of the mean (SEM). (B) Metal concentration dependence. Reaction mixtures (25 μl) containing 100 mM Tris-acetate (pH 6.0), 10 mM (250 nmol) p-nitrophenylphosphate, 20 pmol (0.8 μM) of HD-Pnk, and 0, 0.1, 0.25, 0.5, 1.0, 2.0, 3.0, 4.0, or 5 mM CuCl₂ or NiCl₂ as indicated were incubated at 37°C for 10 min. p-Nitrophenol formation is plotted as a function of divalent cation concentration. (C) His73 is essential for p-nitrophenylphosphatase activity. Reaction mixtures (25 μl) containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 10 mM (250 nmol) p-nitrophenylphosphate, and 10 pmol (0.4 μM) wild-type (WT) HD-Pnk or mutants H73A or D254A as specified were incubated at 37°C for 10 min. (D) Enzyme titration. Reaction mixtures (25 μl) containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 10 mM (250 nmol) p-nitrophenylphosphate, and 0, 1. 2.5, 5, 7.5, 10, or 20 pmol of HD-Pnk were incubated at 37°C for 10 min. p-Nitrophenol formation is plotted as a function of input enzyme. Each datum is the average from three separate experiments ± the SEM. (E) Time course. A reaction mixture (250 μl) containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 10 mM p-nitrophenylphosphate, and 0.2 μM HD-Pnk was incubated at 37°C. Aliquots (25 μl, containing 250 nmol of input p-nitrophenylphosphate) were withdrawn at the times specified and quenched immediately with EDTA. p-Nitrophenol formation is plotted as a function of time. (F) p-Nitrophenol concentration dependence. Reaction mixtures containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 0.2 μM HD-Pnk, and 0.625, 1.25, 2.5, 5 or 10 mM p-nitrophenylphosphate were incubated at 37°C. Aliquots (25 μl) were withdrawn at 1, 2, 5, and 10 min and quenched immediately with EDTA. p-Nitrophenol formation was plotted as a function of time, and the initial rate (nmol min⁻¹) at each substrate concentration was calculated by linear regression in Prism. The averages ± the SEM of initial rates (V) from three separate experiments are plotted as a function of p-nitrophenol concentration. The data were fit to the Michaelis-Menten equation by nonlinear regression in Prism.

Copper-dependent p-nitrophenylphosphatase activity was abolished by alanine mutation of His73 in the HD domain that is critical for the end-healing phosphoesterase activities of HD-Pnk (Fig. 2C). In contrast, p-nitrophenylphosphatase activity was unaffected by alanine mutation of Asp254 that is essential for the kinase function of HD-Pnk (Fig. 2C). We conclude that p-nitrophenylphosphate hydrolysis is inherent to the HD component of HD-Pnk and this reaction provides a simple and inexpensive gauge of HD phosphoesterase activity.

The extent of p-nitrophenol production was proportional to input HD-Pnk (Fig. 2D). From the slope of the titration curve, we calculated that 15.3 ± 0.3 nmol of p-nitrophenylphosphate was hydrolyzed per pmol of HD-Pnk during a 10-min reaction, which translates to a turnover number of ∼25.5 s⁻¹. The kinetic profile of the reaction of 5 pmol of HD-Pnk with 10 mM p-nitrophenylphosphate is shown in Fig. 2E; the observed initial rate of 7.27 ± 0.11 nmol of substrate hydrolyzed per min corresponds to a turnover number of 24.2 s⁻¹. The rate of p-nitrophenol production by 5 pmol of HD-Pnk displayed a hyperbolic dependence on p-nitrophenylphosphate concentration (Fig. 2F). From a nonlinear regression curve fit of the data to the Michaelis-Menten equation, we derived an apparent K_m of 3.13 ± 0.51 mM p-nitrophenylphosphate and a k_cat of 27.8 ± 1.8 s⁻¹.

Mutagenesis of the HD domain of HD-Pnk.

We selected 12 amino acids in the HD domain of HD-Pnk (highlighted in Fig. 1) for a mutational analysis entailing replacement of each position with alanine as well as at least one conservative substitution. The amino acids were chosen based on their potential to serve as transition metal ligands (histidine, aspartate) or as ligands for the scissile phosphate (arginine, lysine), as well as their conservation in HD-Pnk enzymes from a broad range of bacterial taxa. Eight new alanine mutants—Q28A, H33A, T42A, K77A, H94A, H127A, R166A, and D170A—were produced in E. coli and purified from soluble extracts in parallel with wild-type (WT) HD-Pnk (Fig. 3A). Three other alanine mutants (D74A, R103A, and D162A) were insoluble when produced in E. coli and were not amenable to purification.

FIG 3.

Alanine scanning mutagenesis of the HD domain of HD-Pnk. (A) Aliquots (5 μg) of the recombinant wild-type HD-Pnk (WT) and the indicated HD-Pnk-Ala mutants were analyzed by SDS-PAGE. A Coomassie blue-stained gel is shown. The positions and sizes (kDa) of marker polypeptides are indicated on the left. (B) p-Nitrophenylphosphatase reaction mixtures (25 μl) containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 10 mM p-nitrophenylphosphate, and 20 pmol (0.8 μM) of WT or mutant HD-Pnk as specified were incubated at 37°C for 10 min. The extents of p-nitrophenol product formation are plotted in bar graph format. Each datum is the average from four separate experiments ± the SEM. (C) Polynucleotide kinase reaction mixtures (10 μl) containing 100 mM Tris-acetate (pH 6.0), 5 mM MgCl₂, 2.5 mM dithiothreitol, 100 pmol (10 μM) of 21-mer 5′-OH DNA oligonucleotide, 100 μM [γ-³²P]ATP, and 5 pmol (0.5 μM) of wild-type or mutant HD-Pnk were incubated at 37°C for 40 min. The extents of label transfer to the DNA oligonucleotide (5′ ³²P-DNA) are plotted in bar graph format. Each datum is the average from three separate experiments ± the SEM.

The recombinant WT HD-Pnk and HD-Pnk-Ala mutants were assayed for copper-dependent hydrolysis of 10 mM p-nitrophenylphosphate using 20 pmol of input enzyme, a level that sufficed to consume 85% of the input substrate (Fig. 3B). Whereas the D170A mutant was as active as wild-type HD-Pnk, the extents of product formation by the other Ala mutants was sharply curtailed, as follows: Q28A (0.8% of wild type), H33A (1.2%), T42A (6.1%), K77A (0.5%), H94A (3.1%), H127A (1.0%), and R166A (2.9%) (Fig. 3B). As a means of checking whether the alanine mutations might have resulted in a global effect on HD-Pnk folding, we also assayed the wild-type and Ala mutant proteins for polynucleotide kinase activity by reacting the protein with magnesium, [γ-³²P]ATP, and a 21-mer 5′-OH DNA oligonucleotide. This resulted in label transfer from ATP to the 21-mer DNA, as gauged by denaturing PAGE and quantified by scanning the gel (8). The level of wild-type enzyme used in the assays sufficed to phosphorylate ∼60% of the 5′-OH DNA substrate. The K77A, H127A, and D170A mutations had no effect on kinase activity (Fig. 3C). The Q28A, H94A, T42A, R166A, and H33A kinase activities were 86, 79, 71, 64, and 47% of the wild type, respectively (Fig. 3C). Thus, the alanine mutations exerted highly selective effects on the phosphomonoesterase activity of HD-Pnk.

Structure-activity relations at targeted amino acids of the HD domain.

In order to illuminate structure-activity relations, we introduced “conservative” amino acid changes at the 12 positions indicated in Fig. 1. An SDS-PAGE analysis of the 18 recombinant proteins is shown in Fig. 4A. None of the conservative HD domain changes affected HD-Pnk polynucleotide kinase activity (Fig. 4C). The mutational effects on p-nitrophenylphosphatase activity are shown in Fig. 4B.

FIG 4.

Conservative mutations of the HD domain of HD-Pnk. (A) Aliquots (5 μg) of the recombinant WT HD-Pnk and the indicated mutants were analyzed by SDS-PAGE. Coomassie blue-stained gels are shown. The positions of marker polypeptides are indicated on the left of each panel. The sizes of the marker polypeptides (in kilodaltons) are indicated in the leftmost panel. (B and C) p-Nitrophenylphosphatase and kinase reactions were performed as described for Fig. 2B and C, respectively. The extents of product formation are plotted in bar graph format. Each datum is the average from three separate experiments ± the SEM.

Replacing His33 or His73 with Gln effaced phosphomonoesterase activity, as did substituting His33, His94, or His127 by Asn. The H127Q mutation reduced activity to 8.7% of the wild-type HD-Pnk (Fig. 4B). Because the Asn and Gln side chain amide nitrogens are sterically akin to the Nδ and Nε atoms of histidine, we surmise that the essential roles of His33, His73, His94, and His127 entail more than service as hydrogen bond donors. The data are consistent with some or all of these four histidines serving as ligands for the copper cofactor (or conceivably as an acid-base catalyst). Replacement of Asp74 or Asp162 with Asn eliminated phosphomonoesterase activity, signifying that the carboxylate is essential at these positions, consistent with a role in metal coordination. The inactivity of the D74E mutant suggests that the active site cannot accommodate the increased main chain to carboxylate distance in Glu versus Asp. In contrast, changing Asp170 to Glu or Asn had no effect on p-nitrophenylphosphatase activity (Fig. 4B).

Mutating Lys77 in the HDXXK motif to Gln eliminated phosphomonoesterase activity. The K77R change reduced activity to 1.9% of wild-type HD-Pnk, implying that a basic side chain is essential and the longer Arg side chain cannot fulfill this role. Changing Arg103 or Arg166 to Gln abolished phosphomonoesterase activity. At position Thr42, where alanine substitution was deleterious (6.1% of the wild type), installation of serine elicited a partial restoration of function to 39% of wild-type activity, and valine substitution resulted in even more activity (87% of the wild type). These results indicate space occupancy by the β branched side chain is the pertinent contribution of Thr42, not hydrogen bonding by the Oγ atom.

Mutational effects on the DNA 3′ phosphatase activity of HD-Pnk.

DNA 3′-end-healing activity was gauged by reaction of HD-Pnk with a ³²P-labeled 10-mer pDNAp substrate. We showed previously that wild-type HD-Pnk converts pDNAp to a pDNA_OH product in the presence of copper (8). Here, the WT and mutant HD-Pnk proteins were titrated for DNA 3′ phosphatase activity. The results for the HD-Pnk-Ala mutants are shown in Fig. 5A. The specific activities were calculated in Prism from the slopes of the titration curves in the linear range of enzyme dependence and then normalized to the specific activity of wild-type HD-Pnk (defined as 100%). The values for the mutants were as follows: D170A, 82%; H73A, 4%; T42A, 2%; H127A, 0.6%; R166A, 0.6%; H33A, 0.5%; Q28A, 0.5%; H94A, 0.4%; and K77A, 0.4%. The effects of alanine mutations on DNA 3′ end healing accord with their impact on hydrolysis of p-nitrophenylphosphate.

FIG 5.

Mutational effects on DNA 3′ phosphatase activity of HD-Pnk. Reaction mixtures (10 μl) containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 0.1 μM (1 pmol) 5′ ³²P-labeled 10-mer pDNAp (shown at the bottom) and wild-type or mutant HD-Pnk as specified were incubated at 37°C for 10 min. The products were analyzed by urea-PAGE. The extents of conversion of pDNAp to pDNA_OH (the reaction depicted at bottom) are plotted as a function of input enzyme. Each datum in the graphs is the average from three or more independent experiments ± the SEM. (A) Alanine mutants. (B) Conservative mutants.

The DNA 3′ phosphatase activities of 16 conservative mutants are shown in Fig. 5B. The specific activities of the mutants vis-à-vis wild type were as follows: T42V, 32%; T42S, 12%; H127Q, 5.8%; H127N, 0.3%; D74N, 3.4%; D74E, 1.1%; Q28N, 2.1%; D162N, 1.6%; H73Q, 1.6%; H33Q, 0.7%; H33N, 0.6%; H94N, 0.4%; K77R, 0.4%; K77Q, 0.3%; R103Q, 1.2%; and R166Q, 0.4%. Here too the hierarchy of mutational effects on DNA 3′ phosphatase activity mimics that seen on p-nitrophenylphosphatase activity. (We eschewed assaying D170E and D170N, given that these mutations had no effect on p-nitrophenylphosphatase activity and that D170A was nearly as active as the wild type in DNA 3′ phosphate hydrolysis.)

Structure modeling of HD-Pnk aids interpretation of the mutational effects.

After accruing the results of the mutational analysis, we submitted the amino acid sequence of the HD domain to the Phyre2 structure prediction server (29). The top model (100% confidence), which was templated on an HD domain structure from Aquifex aeolicus tRNA nucleotidyltransferase (30) (PDB 3WFP and 3WFR), spans HD-Pnk amino acids 9 to 193, and is shown in stereo view in Fig. 6A. The HD-Pnk domain (the query sequence) and the 3WFP template sequence were aligned by Phyre2 as shown in Fig. 6B, with 39 positions of amino acid identity plus 19 positions of side chain similarity. The Phyre2 model places essential HD-Pnk residues His73, Asp74, Lys77, His94, His127, Asp162, Arg166, His33, and Gln28 in or near the active site (Fig. 6A). In contrast, inessential residue Asp170 is predicted to be located in a surface loop far away from the active site. The essential Arg103 side chain is predicted to be located in an α helix and remote from the active site. The model suggests that Arg103 plays a structural role in the HD domain, insofar as Arg103 is in position to form a salt-bridge network to Glu99 (in the same α helix) and Glu118 (in the neighboring α helix). These two glutamates are conserved in the E. coli HD-Pnk homolog (Fig. 1). Thr42 is predicted to contribute to the hydrophobic core of the HD fold.

FIG 6.

Phyre2 model of the HD domain of HD-Pnk. (A) Stereo view of the modeled HD domain structure (amino acids 1 to 193); the side chains targeted for mutagenesis in this study are shown as stick models. (B) The HD-Pnk domain (the query sequence) and the 3WFP template sequence were aligned by Phyre2 as shown. Positions of amino acid identity are denoted by a “:” above the alignment. The positions of side chain similarity are denoted by a “.” above the alignment.

Alignment of the model of Runella HD-Pnk to the actual crystal structures of exemplary HD phosphoesterases with two transition metals and a phospho-substrate or inorganic phosphate in the active site allowed us to infer specific atomic interactions of the equivalent residues in HD-Pnk. Figure 7 shows a stereo view of the 1.5-Å resolution structure of the active site of an HD hydrolase enzyme from Clostridium acetobutylicum in complex with two Fe(III) metal ions (labeled M1 and M2) and inorganic phosphate (PDB 3CCG). The amino acids in the figure are numbered according to their position in HD-Pnk. We regard this active-site structure as a plausible mimetic of the Michaelis complex of HD-Pnk’s phosphomonoesterase reaction, except that copper serves as the catalytic metal for HD-Pnk. The octahedral coordination complex of the M1 ion includes the equivalents of HD-Pnk His33-Nε, His73-Nε, Asp162-Oδ1, and Asp74-Oδ2, a phosphate oxygen, and a water that is simultaneously coordinated by the M2 ion. The octahedral M2 coordination complex includes Asp74-Oδ1, His94-Nε, His127-Nε, and phosphate oxygen, the bridging water to M1, and a second water (Fig. 7). Thus, the drastic effects of mutating His33, His73, Asp74, His94, His127, and Asp162 on HD-Pnk’s phosphatase activities are readily explained by loss or perturbation of essential contacts to the metal cofactors. The contacts of M1 and M2 to the phosphate O1 and O4 atoms are construed to mimic metal interactions with the two nonbridging oxygens of the phosphoester substrate; such contacts would stabilize the extra negative charge developed on the presumptive pentacoordinate phosphorane transition state of the hydrolysis reaction. The equivalents of essential HD-Pnk amino acids Lys77 and Arg166 make additional electrostatic interactions with the phosphate that would stabilize the transition state. The water that bridges M1 and M2 likely corresponds to the nucleophile in the phosphoesterase reaction, based on the following considerations: (i) simultaneous coordination by two metals should reduce the pKa of the water, promoting its attack on the scissile phosphate (Asp162-Oδ2 is in position to accept a proton from the water nucleophile) and (ii) the bridging water is situated 3.2 Å from the phosphorus atom in an apical orientation to the O3 atom of the phosphate (wat–P–O3 angle of 169°), which corresponds to the oxygen of the leaving group during phosphoester hydrolysis.

FIG 7.

Model of the HD active site. Stereo view of the active site of an HD hydrolase enzyme from Clostridium acetobutylicum in complex with two Fe(III) metal ions (magenta spheres, labeled M1 and M2) and inorganic phosphate (stick model), from the 1.5-Å resolution crystal structure reported in PDB 3CCG. Waters in the metal coordination complexes as depicted as red spheres. The amino acids in the figure are numbered according to their position in HD-Pnk. The atomic contacts in the metal coordination complexes and to the phosphate anion are shown as black dashed lines. The magenta dashed line indicates the 3.2-Å distance between the metal-bridged water nucleophile and the phosphorus atom (colored yellow).

Prevalence of a bacterial HD-Pnk–LIG operon.

Homologs of Runella slithyformis HD-Pnk are distributed widely in bacteria (8). They are also encoded by bacterial viruses, including Rhodothermus bacteriophage RM378 (31) and a variety of bacteriophages that infect Campylobacter (32). HD-Pnk homologs are not (to our inspection) found in archaea or eukarya. The HD-Pnk ORF is frequently in an operon-like arrangement in bacterial chromosomes with the ORF encoding a polynucleotide ligase/adenylyltransferase domain.

In order to update our appreciation of the phylogenetic distribution of the HD-Pnk and the HD-Pnk plus LIG gene cluster, we have exploited the Integrated Microbial Genomes Database (33) to survey their presence in available bacterial genomes/proteomes (see the supplemental material). HD-Pnk is found in exemplary species from 129 bacterial genera representing 11 bacterial phyla. In 119/129 (92%) of these species, HD-Pnk is encoded in a two-gene cluster with LIG. In 4/129 instances (3%), HD-Pnk and LIG are clustered but separated by a single cooriented ORF. It is noteworthy that the gene order within the cluster is variable, being LIG→HD-Pnk in 107/129 taxa and HD-Pnk→LIG in 16/129 taxa. This degree of conservation of the gene cluster suggests that HD-Pnk and LIG comprise the end-healing and end-sealing components of a bacterial nucleic acid repair pathway.

MATERIALS AND METHODS

HD-Pnk purification and mutagenesis.

Missense mutations were introduced into the pET28b-His₁₀Smt3-HD-Pnk expression vector by quick-change PCR. The inserts of all plasmids were sequenced to confirm that no unwanted coding changes were acquired. Recombinant wild-type and mutant Runella HD-Pnk proteins were produced in Escherichia coli BL21(DE3) as His₁₀-Smt3 fusions and isolated from soluble bacterial extracts by adsorption to Ni-agarose resin and elution with imidazole as described previously (8). The His₁₀-Smt3 tag was removed with the Smt3-specific protease Ulp1 and the tagless HD-Pnk proteins were separated from the tag by a second round of Ni-agarose chromatography, followed by a final purification step of Superdex-200 gel filtration (8). Peak HD-Pnk fractions were concentrated by centrifugal ultrafiltration and stored at –80°C. Protein concentrations were determined by using the Bio-Rad dye reagent with bovine serum albumin as the standard.

Hydrolysis of p-nitrophenylphosphate.

Reactions mixtures (25 μl) containing 100 mM Tris-acetate (pH 6.0), 10 mM (250 nmol) p-nitrophenylphosphate, 1 mM divalent cation, and HD-Pnk as specified were incubated at 37°C. The reactions were quenched by adding 50 μl of 50 mM EDTA and then 0.9 ml of 1 M Na₂CO₃. The release of p-nitrophenol was determined by measuring the A₄₁₀ and interpolating the value to a p-nitrophenol standard curve.

Polynucleotide kinase assay.

Reaction mixtures (10 μl) containing 100 mM Tris-acetate (pH 6.0), 5 mM MgCl₂, 2.5 mM dithiothreitol, 100 μM [γ-³²P]ATP, 100 pmol (10 μM) of 21-mer 5′-OH DNA oligonucleotide d(CTAGAGCTACAATTGCGACCG), and HD-Pnk as specified were incubated at 37°C for 40 min. The reactions were initiated by adding HD-Pnk and quenched by adding an equal volume of 90% formamide, 50 mM EDTA, and 0.01% bromophenol blue-xylene cyanol. The mixtures were analyzed by electrophoresis (at 15-W constant power) through a 15-cm 20% polyacrylamide gel containing 7 M urea in 45 mM Tris-borate–1 mM EDTA. The radiolabeled 21-mer oligonucleotide products visualized by scanning the gel with a Fujifilm FLA-7000 imaging device and quantified using ImageJ software.

Assay of DNA 3′ phosphatase activity.

Reactions mixtures (10 μl) containing 100 mM Tris-acetate (pH 6.0), 1 mM CuCl₂, 100 nM (1 pmol) 10-mer 5′ ³²P-labeled pDNAp substrate (5′-pATCACGCTTCp; prepared by enzymatic phosphorylation of a synthetic 10-mer _HODNAp oligonucleotide by phosphatase-dead T4 Pnkp-D167N and then gel purified), and HD-Pnk as specified were incubated at 37°C for 10 min. The reactions were quenched by adding an equal volume of formamide and EDTA. The products were analyzed by electrophoresis at 55 W constant power through a 40-cm 20% polyacrylamide gel containing 7 M urea in 45 mM Tris-borate–1 mM EDTA, visualized by scanning the gel with a FujiFilm FLA-7000 imager, and quantified using ImageJ software.