Pre-steady-state Kinetic Analysis of a Family D DNA Polymerase from Thermococcus sp. 9°N Reveals Mechanisms for Archaeal Genomic Replication and Maintenance*

Background: Family D DNA polymerase (polD) is important for replication in most archaea, excluding Crenarchaeota. Results: We report a detailed kinetic characterization of polD nucleotide incorporation, mismatch discrimination, and 3′-5′ exonuclease hydrolysis. Conclusion: Despite evolutionary divergence, polD kinetic pathways share similarities to other DNA polymerase families. Significance: This work contributes to unifying our understanding of DNA polymerase function.

DNA polymerases play central roles in genome replication, maintenance, and repair and are therefore critical for genome integrity. Consequently, DNA polymerases have been the subject of extensive and widespread research for over 60 years (1). Multiple sequence alignment studies have classified DNA polymerases into seven different families: A, B, C, D, X, Y, and reverse transcriptases (RTs) (2)(3)(4). Kinetic studies of Family A, B, C, X, Y, and RT DNA polymerases have proven to be a powerful tool in understanding polymerase function (5)(6)(7). Importantly, such studies reveal the kinetic basis of nucleotide selection and mismatch discrimination mechanisms, offering critical insight into how DNA polymerases accurately and efficiently synthesize and maintain genomes.
Although the majority of DNA polymerase families have been well characterized, studies of Family D DNA polymerases (polDs), 2 found in all known archaea (excluding the Crenarchaea branch), have been limited (8). Previous polD characterization studies have revealed that this polymerase is heterodimeric, composed of a large polymerase subunit and small 3Ј-5Ј exonuclease proofreading subunit (9 -11). The activities of the two subunits are co-dependent, and the presence of both is required for activity of either unit, a feature unique to the polD family (12,13). Gene deletion studies in Thermococcus kodakarensis and Methanococcus maripaludis suggest that polD is essential for cell viability and indicate that polD may be the major replicative polymerase responsible for leading and lagging strand synthesis in these organisms (14,15). Furthermore, polD forms complexes with many replisome components, including the minichromosome maintenance helicase, proliferating cell nuclear antigen processivity factor, and DNA ligase, providing further support for the belief that polD is a replicative polymerase in Archaea (15)(16)(17)(18). It is still unclear what specific role(s) polD plays in archaeal genome replication and which roles are played by other DNA polymerases. For example, efficient Okazaki fragment maturation is dependent on a Family B DNA polymerase (19).
The biochemical properties of a well expressed polD from Thermococcus sp. 9°N have been characterized in our laboratory (20). In this work, a qualitative assessment of 9°N polD * This work was supported by New England Biolabs, Inc. The authors declare 3Ј-5Ј exonuclease activity showed a dependence on Mg 2ϩ or Mn 2ϩ for catalytic activity, with a preference for Mn 2ϩ (20). Despite its 3Ј-5Ј exonuclease, recombinant polD has relatively low fidelity compared with most other characterized DNA polymerases (20).
To gain further insight into this unique multisubunit polymerase, we have performed a detailed kinetic characterization of the polymerase activity of an exonuclease-deficient mutant of polD, including kinetics of correct nucleotide incorporation and pyrophosphorolysis as well as incorrect nucleotide and ribonucleotide discrimination kinetics. Furthermore, we kinetically characterized the 3Ј-5Ј exonuclease activity of wild type (WT) polD at 60°C in the presence of Mn 2ϩ and Mg 2ϩ . This detailed kinetic analysis performed on a Family D DNA polymerase provides insight into polymerase and exonuclease reaction pathways and suggests mechanisms for archaeal genome maintenance.

Experimental Procedures
Enzymes, Oligonucleotides, and Reagents-WT Thermococcus sp. 9°N polD and an exonuclease-deficient polD mutant (H554A) (abbreviated as polD Ϫ for clarity) were expressed and purified as described previously (20). polD small and large subunits form a heterodimeric complex in a ϳ1:1 molar ratio, as determined by SDS-PAGE gel analysis and quantitation by densitometry (data not shown). Oligonucleotides used in this study were purchased from Integrated DNA Technologies (Coralville, IA) ( Table 1). For oligonucleotide detection, the primer strand was 5Ј-labeled with a 5-carboxyfluorescein (FAM) fluorophore (Integrated DNA Technologies). Buffers used in this study were from New England Biolabs (Ipswich, MA).
polD Ϫ Steady-state Single Nucleotide Incorporation Assay-A steady-state kinetic assay was performed to determine whether polDϪ followed burst kinetics and to calculate an active enzyme concentration. The primer⅐template used to monitor steady-state kinetics was prepared by annealing the 50-mer 5Ј-FAM primer (10 M) (Table 1) to the 62-mer "C Template-1" (15 M) ( Table 1) in 1ϫ ThermoPol buffer (20 mM Tris-HCl, 10 mM (NH 4 ) 2 SO 4 , 10 mM KCl, 2 mM MgSO 4 , 0.1% Triton X-100, pH 8.8, at 25°C) by heating to 95°C for 3 min, followed by cooling to room temperature. A 250-l polD Ϫ / DNA aliquot was prepared by mixing ThermoPol buffer (1ϫ final concentration), primer⅐template DNA (80 nM final concentration), and polD Ϫ (20 nM final concentration) to ensure a 4-fold excess of DNA to polD Ϫ . A second 250-l aliquot was prepared by mixing ThermoPol buffer (1ϫ final concentration) and dGTP (200 M final concentration). Using a rapid chemical quench instrument (RQF) (KinTek Corp., Snow Shoe, PA) set to 62.5°C with a circulating water bath to achieve a final 60°C reaction temperature, the polD Ϫ /DNA and dGTP solutions were rapidly mixed from 0.07 to 10 s and quenched with 50 mM EDTA. After mixing of equal volumes of polD Ϫ /DNA and dGTP solutions by the RQF, the final reaction concentrations were 40 nM DNA, 10 nM polD Ϫ , and 100 M dGTP in 1ϫ Ther-moPol buffer. A negative control reaction was performed in which the dGTP aliquot was replaced with 1ϫ ThermoPol buffer and reacted in the RQF with polD Ϫ /DNA for 10 s. To ensure that the DNA substrate (40 nM) was in molar excess to satisfy steady-state requirements, control experiments were performed with a fixed polD Ϫ concentration (10 nM final concentration) and varying DNA concentrations (10 -80 nM final concentration) as described above (data not shown). Furthermore, the dependence of burst amplitude on enzyme amount was confirmed (data not shown). Reaction products were separated by capillary electrophoresis using a 3730xl Genetic Analyzer (Applied Biosystems), and fluorescent peaks were analyzed using Peak Scanner software version 1.0 (Applied Biosystems). The concentration of product (51 nt of DNA with a FAM label) was graphed as a function of time, and the data were fit to burst Equation 1 using the nonlinear regression program Kaleidagraph (Synergy Software).
From the fit, one can extract the active enzyme concentration (A), the initial rate of product formation (k obs ), and steady-state turnover rate (k ss ), which is obtained by dividing k 2 by A. All kinetic assays described in this work were performed at least twice to ensure experiment reproducibility.
polD Ϫ Pre-steady-state Single Nucleotide Incorporation-To obtain the rates of correct nucleotide incorporation, incorrect nucleotide incorporation, and ribonucleotide incorporation by polD Ϫ , pre-steady-state single nucleotide assays were performed. The primer⅐template substrates used in these assays were prepared as described above. A 150-l polD Ϫ /DNA aliquot was prepared by mixing ThermoPol buffer (1ϫ final concentration), primer⅐template DNA (30 nM final concentration), and a 3-fold excess of polD Ϫ (90 nM active enzyme final concentration); a control experiment demonstrated that polymerase saturation was reached at 3-fold excess polD Ϫ , satisfying pre-steady-state requirements (data not shown). A second 150-l aliquot was prepared by mixing ThermoPol buffer (1ϫ final concentration) and dNTP (5-200 M). Higher concentrations (100 -4000 M) were required for incorrect nucleotide and ribonucleotide incorporation assays. Using the RQF, the  polD Ϫ /DNA construct was rapidly mixed with dNTP and quenched with 50 mM EDTA. After mixing an equal volume of polD Ϫ /DNA and dNTP solutions by the RQF, the final reaction concentrations were 15 nM DNA, 45 nM active polD Ϫ , and 2.5-100 M dNTP for correct nucleotides (or 50 -2000 M dNTP for incorrect nucleotides or ribonucleotides) in 1ϫ ThermoPol buffer. A control reaction was performed in which dNTP was replaced with 1ϫ ThermoPol buffer and reacted in the RQF with polD Ϫ /DNA for 10 s. Reaction products were separated by capillary electrophoresis as described above. Because reaction products for dCTP incorporation did not resolve into two fully distinguishable substrate and product peaks by capillary electrophoresis, we instead separated the products of dCTP insertion by denaturing gel electrophoresis (20% polyacrylamide, 8 M urea, 1ϫ TBE buffer), visualized by Typhoon TRIO (GE Healthcare), and quantitated with ImageQuant software (Molecular Dynamics). In order to confirm that both capillary and gel electrophoresis give comparable results, the 100 M dTTP incorporation time course was analyzed by both capillary electrophoresis and gel electrophoresis (data not shown).
The product concentration was graphed as a function of time, and the data were fit to the single-exponential Equation 2 to obtain the observed rate constant of nucleotide incorporation (k obs ) using the nonlinear regression program Kaleidagraph (Synergy Software).
To obtain the maximum rate of polymerization constant (k pol ) and apparent equilibrium dissociation constant (K d(dNTP) ), the k obs values were graphed as a function of dNTP (or rNTP) concentration, and the data were fit to the hyperbolic Equation 3 using Kaleidagraph.
The specific activities for correct nucleotide, incorrect nucleotide, and ribonucleotide incorporation were calculated using Equation 4, and the nucleotide selectivities for incorrect nucleotide incorporation and ribonucleotide incorporation were calculated using Equation 5.
polD Ϫ Pyrophosphorolysis-Under certain circumstances, DNA polymerases can perform the reverse reaction (pyrophosphorolysis) to remove nucleotides from the 3Ј-end of the DNA molecule in the absence of exonuclease activity. To obtain the rate of polD Ϫ pyrophosphorolysis, a 3-fold excess of polD Ϫ was preincubated with 5Ј-FAM primer⅐C Template-1 and was rapidly mixed with 50 -500 M (final concentration after RQF mixing) inorganic pyrophosphate (PP i ) using the RQF at 60°C. Reaction products were separated by capillary electrophoresis and analyzed as described above. The concentration of pyrophosphorolysis product (Ͻ50 nt of DNA) was graphed as a function of time and fit to Equation 2 to obtain the observed rate constants (k obs ), followed by fitting the k obs as function of PP i concentration in Equation 3 to obtain a maximal rate for pyrophosphorolysis (k pyro ) and apparent equilibrium dissociation constant of PP i (K d(PPi) ). In reactions with multiple pyrophosphorolysis events, all product peaks were summed as total product.
polD 3Ј-5Ј Exonuclease Activity-The kinetics of polD 3Ј-5Ј exonuclease activity was measured by monitoring shortening of a FAM-labeled DNA primer. To obtain the rate of 3Ј-5Ј exonuclease activity on single-stranded DNA, a pre-steady-state single-turnover assay was performed. A 40 nM aliquot of 5Ј-FAM primer was prepared in 1ϫ ThermoPol II buffer (20 mM Tris-HCl, 10 mM (NH 4 ) 2 SO 4 , 10 mM KCl, 0.1% Triton X-100, pH 8.8, at 25°C) supplemented with either 2 mM MgSO 4 or MnSO 4 . A second aliquot containing polD (120 nM) was prepared in 1ϫ ThermoPol II buffer supplemented with either 2 mM MgSO 4 or MnSO 4 . Using the RQF, polD and 5Ј-FAM labeled primer were rapidly mixed at 60°C from 0.002 to 5 s and quenched with 0.1 N H 2 SO 4 . Final reaction conditions after mixing by the RQF were 20 nM primer, 60 nM polD in 1ϫ ThermoPol II buffer with 2 mM MgSO 4 or MnSO 4 . After quenching, samples were neutralized with 1 N NH 4 OH, and reaction products were separated by capillary electrophoresis and analyzed by peak scanner as described above. All product peaks (Ͻ50 nt) resulting from 3Ј-5Ј exonuclease hydrolysis were summed and graphed as a function of time, followed by fitting of data to Equation 2 as described previously. It is important to note that 0.05, 0.1, and 1 M EDTA did not sufficiently quench 3Ј-5Ј exonuclease reactions at rapid time scales, but 0.1 N H 2 SO 4 was sufficient to quickly and reliably quench activity (data not shown).

Analysis of polD Ϫ Multiple-turnover Steady-state Nucleotide Incorporation Kinetics Reveals a Postchemistry Rate-limiting
Step-Steady-state kinetic assays are used to infer the rate-determining step of a catalytic pathway and are performed under multiple-turnover conditions (i.e. [substrate] [enzyme]). Previous steady-state kinetic studies performed on a host of different DNA polymerases have revealed that nucleotide incorporation is fast and DNA release is rate-limiting (21)(22)(23)(24)(25)(26)(27). In order to determine if the same is true for polD Ϫ , a steadystate multiple-turnover assay was performed in which polD Ϫ was preincubated with a 4-fold excess of DNA and rapidly mixed with dGTP at 60°C using an RQF. Under these conditions, only one dGTP is added to the primer because the next position requires incorporation of a nucleotide that is absent in the reaction mixture. polD Ϫ must then dissociate from the DNA and bind to a new primer⅐template to repeat the addition of the single dGTP. DNA substrate and product were analyzed by capillary electrophoresis. Concentration of the 51-nt DNA product was graphed as a function of time, and the steady-state kinetic parameter, k ss , was determined. We observe a rapid initial accumulation of product, designated the burst phase, followed by a slower linear phase of product formation (Fig. 1). The burst phase represents the fast chemistry step of nucleotide incorporation and is proportional to the concentration of the active prebound polymerase. The linear phase represents a slow step occurring after chemistry. This slow phase, probably dom-inated by the rate of DNA release between nucleotide addition reactions, defines the polD Ϫ steady-state turnover rate, 0.3 s Ϫ1 (Table 2). Furthermore, by extrapolating the linear phase through the y axis, the concentration of active enzyme can be determined. For this preparation of polD Ϫ , 4.8 nM of the 10 nM was active. This active enzyme concentration (4.8 nM) was used to calculate the active polD Ϫ concentration in subsequent presteady-state studies using this preparation of enzyme.
Several factors may account for the discrepancy between total and active enzyme concentrations. The concentration of polD Ϫ was first determined spectrophotometrically, which may account for inaccurate apparent protein concentration. Alternatively, incorrect stoichiometry between small and large subunits, misfolding of the subunits, or multiple DNA polymerases binding to a single DNA substrate may also account for a low apparent active enzyme concentration.
Analysis [substrate]) and are used to determine the kinetic parameters associated with steps masked by the ratelimiting step, such as single nucleotide incorporation (7). To obtain rates for nucleotide incorporation, including maximal rate of polymerization, k pol , and apparent equilibrium dissoci-ation constant, K d(dNTP) , polD Ϫ pre-steady-state kinetic assays were performed. Such parameters provide insight into how polymerases discriminate against incorrect nucleotides and incorporate correct nucleotides. The k pol reflects how fast the polymerase will incorporate a nucleotide, whereas the K d(dNTP) reflects how tightly the polymerase binds a nucleotide, where a lower K d reflects tighter nucleotide binding. Although k pol and K d(dNTP) can be obtained from the burst phase of steady-state kinetics, such methods may result in parameters with large sources of error (5). Therefore, to measure pre-steady-state kinetics, a 3-fold excess of active polD Ϫ was preincubated with DNA, rapidly mixed with single dNTP solutions at 60°C, and analyzed as described above. A schematic of the incorporation assay and expected capillary electrophoresis results are depicted in Fig. 2. For each dNTP concentration, the concentration of product was graphed as a function of time to obtain the observed rate of nucleotide incorporation, k obs . The k pol as well as the K d(dNTP) for each dNTP was obtained by graphing k obs versus dNTP concentration.
As shown in Fig. 3A for dTTP incorporation paired with template dA (dTTP:A), we observe an increase in k obs with increasing dTTP concentration, where we reach a maximal k obs at high (100 M) dTTP concentration. When k obs is graphed as a function of dTTP concentration, the maximal rate of polymeriza-

Reaction observed
Parameter Value . polD ؊ pre-steady-state dTTP incorporation. A, reaction scheme, a 50-nt 5Ј-FAM primer was annealed to template DNA and then preincubated with a 3-fold excess of active polD Ϫ , followed by rapid mixing with dTTPs at 60°C and incubation from 0 to 10 s. Because only dTTP is present, the primer can only be extended 1 nt unless an incorrect nt is added. B, substrates (50 nt) and incorporation products (51 nt) were resolved by capillary electrophoresis after denaturation of the double-stranded DNA.  tion is obtained from the plateau, whereas the K d(dNTP) is the dTTP concentration at half k pol (Fig. 3B). Table 3 shows the obtained k pol and K d(dNTP) values for correct nucleotide incorporation using separate matched templates for each possible dNTP. All values for k pol are within 2-fold, ranging from 1.8 to 3.1 s Ϫ1 , suggesting that polD incorporates all correctly base-paired nucleotides at a similar rate. Similarly, the obtained K d(dNTP) values are within 3-fold, ranging from 0.9 to 2.5 M, indicating that polD binds each correct nucleotide with similar affinity (Table 3). Furthermore, we calculated the specificity constant for each correct nucleotide, obtained by dividing k pol by the corresponding K d(dNTP) ( Table  3). The specificity constant (k pol /K d(dNTP) ), a reflection of both the incorporation rate and dNTP binding, is used to compare the efficiency of incorporation for each nucleotide by a polymerase as well as to compare the efficiency of incorporation with other polymerases. A larger specificity constant reflects more efficient binding and incorporation (5). For polD Ϫ , specificity constants for correct nucleotide incorporation range from 1.0 to 2.3 M Ϫ1 s Ϫ1 (Table 3), again demonstrating that the enzyme lacks preference among Watson-Crick base-paired dNTP substrates.
Pre-steady-state Kinetic Analysis of polD Ϫ Incorrect Nucleotide Incorporation Reveals the Presence of a Nucleotide Discrimination Mechanism-DNA polymerases have evolved specific mechanisms to discriminate against the incorporation of incorrect nucleotides during synthesis (reviewed in Ref. 4). Such mechanisms are critical for faithful replication of the genome and to ensure the transfer of accurate information to subsequent generations. Importantly, nucleotide selectivity mechanisms have not been characterized in polD. Therefore, we performed a thorough kinetic characterization of incorrect nucleotide incorporation to understand polD fidelity. Presteady-state single-turnover kinetic experiments were performed on the 12 possible incorrect base pairing combinations ( Table 3).
As shown in Fig. 3C for dATP:A incorporation for dATP incorporation paired with template dA, k obs increases with higher dATP concentrations. It is important to note that maximal k obs for dATP:A incorporation is reached at exceedingly high dATP concentration, 2000 M, compared with 100 M for correct dTTP:A incorporation. Furthermore, maximal product incorporation is achieved after 60 s for incorrect dATP:A incorporation, 12 times slower than for correct dTTP:A incorporation. By plotting the observed k obs versus dATP concentration, the k pol and K d(dNTP) for incorrect nucleotide incorporation were obtained (Fig. 3D and Table 3). These results suggest that for misincorporation, high substrate concentrations and longer reaction times are required to drive the incorporation of the incorrect nucleotide and imply that polD contains specific mechanisms to exclude incorrect nucleotides, similar to other DNA polymerases. Table 3 shows k pol and K d(dNTP) values for all 12 incorrect nucleotide incorporation combinations. Obtained k pol rates range from 0.07 to 1.3 s Ϫ1 , resulting in a 1.5-45-fold decrease in k pol between correct and incorrect dNTP incorporation ( Table  3). The K d(dNTP) ranges from 300 to 1400 M, a 120 -1500-fold increase in K d(dNTP) between correct and incorrect binding ( Table 3). These results suggest that polD Ϫ prevents the incorporation of incorrect nucleotides primarily through weaker binding of the incorrect nucleotide, as reflected in the large increase in K d .
As done for correct nucleotide incorporation assays, the specificity constant was calculated for incorrect nucleotide incorporation. Nucleotide selectivity, obtained by dividing the specificity constant for the correct nucleotide by the specificity constant for the incorrect nucleotide, reflects overall DNA polymerase fidelity (5). Of the 12 potential mismatches, polD discriminates most strongly against a dGTP:G mismatch (8100fold) and weakly discriminates against a dATP:C mismatch (1000-fold) ( Table 3).
Analysis of polD Ϫ Single-turnover Ribonucleotide Incorporation Kinetics Reveals the Presence of a Ribonucleotide Discrimination Mechanism-In addition to incorrect nucleotide discrimination, polymerases discriminate against ribonucleotide incorporation (28,29). Excluding rNTPs is especially important due to the intracellular excess of rNTPs over dNTPs (30). Furthermore, rNTPs incorporated during replication may lead to genome instability by inducing strand breaks. Therefore, to understand whether and how polD Ϫ discriminates against ribonucleotides, a pre-steady-state single-turnover assay was performed examining rATP incorporation paired to dT (rATP:T) as described above. For rATP incorporation, k pol was 0.16 s Ϫ1 and K d(rNTP) was 360 M, which correspond to a 16-fold decrease in k pol and a 144-fold increase in K d(rNTP) compared with incorporation of dATP (Table 4). Such results indicate that ribonucleotide discrimination occurs primarily through reduced binding affinity of the rNTP. The specificity constant (k pol /K d(rNTP) ), 4.4 ϫ 10 Ϫ4 , and nucleotide selectivity constant ((k pol /K d(dNTP) )/(k pol /K d(rNTP) )), 2.3 ϫ 10 3 , suggest that dATP is incorporated 2,300-fold more efficiently than rATP ( Table 4).
Analysis of polD Ϫ Pyrophosphorolysis-During the incorporation of dNTPs by a DNA polymerase, PP i is produced and released. If PP i stays bound within the enzyme active site, reversal of chemistry can occur, resulting in pyrophosphorolysis, which shortens the primer and releases a dNTP. In the presence of a high concentration of PP i , the polymerase can remove multiple dNTPs during a single DNA polymerase- DNA binding event. Single-turnover kinetic assays were performed to test whether polD Ϫ undergoes pyrophosphorolysis and to obtain the maximal rate of pyrophosphorolysis, k pyro , and the apparent dissociation rate constant for PP i , K d (PPi) . A schematic of the pyrophosphorolysis assay and the expected capillary electrophoresis results are shown in Fig.  4. We observed pyrophosphorolysis of multiple nucleotides at longer time points (Fig. 4B, VI-VIII). Concentration of product (Ͻ50 nt of DNA) was graphed as a function of time to obtain the observed rate constants, k obs (data not shown).
To obtain k pyro and K d (PPi) , k obs was graphed as a function of PP i concentration (Fig. 4C). For polD Ϫ , k pyro is 0.4 s Ϫ1 and K d(PPi) is 190 M ( Table 2).

Analysis of polD 3Ј-5Ј Exonuclease Kinetics Reveals a 40-Fold Preference for
Mn 2ϩ over Mg 2ϩ -Along with nucleotide incorporation activity, many polymerases, including polD, contain a metal-dependent 3Ј-5Ј exonuclease activity, which facilitates removal of incorrectly incorporated nucleotides, thereby increasing the overall fidelity of the polymerase (31). In vivo, a DNA polymerase melts duplex DNA to shuttle the singlestranded DNA primer to the exonuclease active site prior to hydrolysis (31). Importantly, melting of duplex DNA during DNA polymerase exonuclease hydrolysis was previously determined to be rate-limiting (32). In order to focus on 3Ј-5Ј exonuclease hydrolysis rates rather than DNA duplex melting dynamics, experiments were designed using a single-stranded DNA substrate (32). To assess polD exonuclease activity in the presence of Mn 2ϩ and Mg 2ϩ , we performed pre-steady-state assays in which we rapidly mixed a 3-fold excess of polD with single-stranded 5Ј-FAM DNA in the presence of either metal ion. A schematic of this assay is shown in Fig. 5A, and represen- . polD ؊ pyrophosphorolysis. A, reaction scheme, a 50-nt 5Ј-FAM primer was annealed to template DNA and preincubated with a 3-fold excess of active polD Ϫ followed by rapid mixing with PP i at 60°C using an RQF. Pyrophosphorolysis results in production of a smaller FAM-labeled oligonucleotide. B, pyrophosphorolysis products (Ͻ50 nt) were resolved by capillary electrophoresis after DNA denaturation. C, the dependence of the reaction rate k obs on PP i concentration was fit to Equation 3 to obtain the maximum rate of k pyro and the equilibrium binding constant of K d(PPi) , ϳ0.4 s Ϫ1 and ϳ190 M, respectively. tative CE traces for exonuclease kinetics done in the presence of Mn 2ϩ or Mg 2ϩ are shown in Fig. 5B. The concentration of exonuclease product (Ͻ50 nt of DNA) was graphed as a function of time and fit to a single-exponential burst equation to obtain pre-steady-state kinetic rates of exonuclease hydrolysis (Fig. 5). Due to rapid exonuclease hydrolysis by polD, multiple dNMPs were excised from the 5Ј-FAM substrate at each reaction time. We therefore report a lower limit for polD k exo , Ն110 and Ն2.5 s Ϫ1 , in the presence of Mn 2ϩ and Mg 2ϩ , respectively. The results obtained here suggest that there is a Ն44-fold increase in the exonuclease hydrolysis rate in the presence of Mn 2ϩ compared with Mg 2ϩ (Table 2). Several alternative RQF experiments were unsuccessful in limiting polD exonuclease hydrolysis to removal of a single dNMP, including prebinding polD to 5Ј-FAM ssDNA and reaction initiation by the addition of Mn 2ϩ or Mg 2ϩ in the presence of a large excess of trap DNA to prevent rebinding to the FAM-labeled DNA (data not shown).

Discussion
Family D DNA Polymerase Kinetic Scheme-The overall polymerization kinetic pathway is highly conserved among different DNA polymerase families (Fig. 6, top path). In this pathway, DNA polymerase first binds to DNA (E⅐DNA n ), followed by dNTP binding (E⅐DNA n ⅐dNTP), which induces a conformational change from an "open" polymerase conformation to a "closed" conformation. Once in a closed conformation, the ␣-phosphate of the bound dNTP is within close proximity to the 3Ј-OH of primer DNA. Here, nucleotide incorporation chemistry occurs when the 3Ј-OH of primer DNA attacks the ␣-phosphate of bound dNTP, incorporating dNMP and generating PP i (E⅐DNA n ϩ 1 ⅐PP i ). Following nucleotide incorporation, the polymerase undergoes two conformation changes to release PP i and then DNA (E ϩ DNA n ϩ 1 ). All steps within the polymerase kinetic scheme are reversible. Furthermore, it is generally agreed upon that the conformational changes are faster than chemistry, and DNA release is the slow rate-limiting step of the kinetic pathway (5).
polD, however, is structurally and genetically unrelated to any other DNA polymerase family. Notably, there are few structural or functional studies for this family of polymerases in the literature to date. Thus, caution must be applied in making assumptions about the polD reaction mechanism through anal-ogy to other studied polymerases. To address this issue, we have performed a variety of kinetic assays to elucidate the polD reaction pathway and allow comparisons with other polymerase families. Our findings demonstrate that polD does, in fact, follow a similar overall polymerization kinetic pathway as other families of DNA polymerases.
Under multiple-turnover conditions, we observed a burst kinetic profile, indicating that polD shares the feature of fast nucleotide incorporation chemistry followed by a rate-limiting post-phosphoryl transfer reaction. Pre-steady-state kinetic experiments revealed that the k pol is 1.8 -3.1 s Ϫ1 and the K d(dNTP) is 0.9 -2.5 M (Table 2) for all correct nucleotide basepairing combinations. Furthermore, pre-steady-state pyrophosphorolysis kinetics revealed that nucleotide incorporation is indeed reversible, providing further evidence that Family D DNA polymerases follow the generally observed cross-family polymerase kinetic scheme (Table 2 and Fig. 6). Importantly, maximal nucleotide incorporation rates are faster than maximal pyrophosphorolysis rates, and the K d(dNTP) for correct nucleotide binding is 95-fold tighter than K d (PPi) , suggesting that the equilibrium of correct nucleotide incorporation chemistry strongly favors dNTP incorporation over dNTP removal (Table 2).
Although polD has been implicated as a replicative polymerase in Euryarchaea, the maximal rate of polymerization, k pol , for polD is one of the slowest for kinetically characterized replicative polymerases (Family A, B, and C) and repair and translesion polymerases (Family X and Y) ( Table 5). On the contrary, polD has one of the smallest, and therefore tightest, K d(dNTP) values of kinetically characterized polymerases (Table 5). Importantly, the resulting specific activity for polD, ϳ1.8 M Ϫ1 s Ϫ1 , is average compared with other replicative, repair, and translesion polymerases (Table 5).
Metal Ion Dependence of polD 3Ј-5Ј Exonuclease ssDNA Hydrolysis-The proposed 3Ј-5Ј exonuclease ssDNA hydrolysis kinetic pathway for polD is presented in Fig. 6 (bottom path). Importantly, polD exonuclease hydrolysis requires a divalent metal ion for catalysis, typically Mg 2ϩ , and the reaction is nonreversible at the 3Ј-5Ј exonuclease active site.
Previously reported qualitative data on 9°N polD showed that 3Ј-5Ј exonuclease activity was dependent on Mg 2ϩ or Mn 2ϩ , with an increase in activity seen for Mn 2ϩ (20). The quantitative FIGURE 6. Schematic of polD kinetic pathway. During polymerization (top pathway), DNA polymerase binds to DNA (E⅐DNA n ), followed by dNTP binding (E⅐DNA n ⅐dNTP). A proposed conformational change from an "open" polymerase conformation to a "closed" conformation brings the ␣-phosphate of the bound dNTP within close proximity to the 3Ј-OH of the primer DNA. Here, nucleotide incorporation chemistry occurs when the 3Ј-OH of primer DNA attacks the ␣-phosphate of the bound dNTP, incorporating dNMP, extending the primer DNA by 1 nucleotide and generating PP i (E⅐DNA n ϩ1 ⅐PP i ). Following nucleotide incorporation, the polymerase undergoes two proposed conformation changes to release PP i and then DNA (E ϩ DNA n ϩ1 ). Alternatively, during 3Ј-5Ј exonuclease hydrolysis (bottom pathway), the 3Ј-5Ј exonuclease active site binds ssDNA (ssDNA n ) and hydrolysis occurs, shortening the DNA to release a dNMP from the enzyme⅐DNA complex (E⅐ssDNA n Ϫ 1 ). Kinetic constants derived from this study are averages from Table 3.
pre-steady-state exonuclease kinetics performed here reveal a Ն44-fold increase in k exo in the presence of Mn 2ϩ versus Mg 2ϩ ( Table 2). Similar to 9°N polD, the 3Ј-5Ј exonuclease activity of polD from Pyrococcus horikoshii is more robust in the presence of Mn 2ϩ over Mg 2ϩ , whereas polD from Methanococcus jannaschii requires Mn 2ϩ and is inactive with Mg 2ϩ (13,33,34). Differences in metal requirements between species suggest that the 3Ј-5Ј exonuclease active site of the polD small subunit may differ between different archaeal species.
The polD 3Ј-5Ј exonuclease small subunit bears homology to the calcineurin-like phosphoesterase family (35)(36)(37). This calcineurin-like phosphoesterase superfamily requires the presence of two divalent metal ions per active site for catalysis, typically Mn 2ϩ , Ni 2ϩ , Ca 2ϩ , Fe 2ϩ , Fe 3ϩ , or Zn 2ϩ , and contains a wide array of members, including phosphoserine/threonine phosphodiesterases, nucleotidases, and nucleases (38). The polD homology with calcineurin-like phosphoesterases and its preference for Mn 2ϩ over Mg 2ϩ suggest that the polD small subunit may have been an independently active exonuclease enzyme. Other archaeal replisome proteins, including the GINS-associated nuclease (GAN), a ssDNA 5Ј-3Ј exonuclease, rely upon Mn 2ϩ for exonuclease activity, with limited activity observed in the presence of Mg 2ϩ (17). Together, these data suggest a putative role of Mn 2ϩ as a cofactor during archaeal replication. However, it is still unclear which metal ion, Mn 2ϩ or Mg 2ϩ , is bound within the polD 3Ј-5Ј exonuclease site in vivo. The in vivo concentrations of Mn 2ϩ and Mg 2ϩ as well as the relative binding affinities of the two metals in the 3Ј-5Ј exonuclease active site (both unknown parameters) determine which metal is bound in vivo.
Comparison of Nucleotide Selectivity among Different Polymerase Families-Polymerases have evolved highly specific mechanisms to ensure selection and incorporation of the correct nucleotide. The pre-steady-state nucleotide incorporation kinetics performed in this work confirm the presence of nucleotide discrimination mechanisms within polD and suggest that discrimination is achieved through weak binding and slow catalysis of incorrect nucleotides. Due to the lack of polD structural characterization studies, it is not yet possible to easily identify active site determinants for nucleotide discrimination.
A quantitative assessment of incorrect nucleotide incorporation is reflected by nucleotide selectivity constants. Polymerases with high nucleotide selectivity are less likely to incorporate incorrect nucleotides and are therefore less likely to introduce mutations. Therefore, high nucleotide selectivity (incorporation of correct versus incorrect dNTP) is an important polymerase feature to ensure accurate genome replication. Table 6 compares the average nucleotide selectivity of several DNA polymerase families. Selectivity values range from as high as 2.7 ϫ 10 5 in the replicative Family B RB69 DNA polymerase (39) to as low as 3.0 ϫ 10 3 in the translesion Family Y Dpo4 DNA polymerase (40). polD nucleotide selectivity is relatively low compared with other families, especially for a polymerase implicated in playing a replicative role, with an average nucleotide selectivity of 4.3 ϫ 10 3 . Likewise, the data obtained here correspond well to previously published fidelity data, which suggests that polD has a higher error rate than typical replicative polymerases (41). We suspect the low fidelity observed for 9°N polD is due to the slow rate of polymerization observed for correct nucleotide incorporation, k pol between 1.8 and 3.1 s Ϫ1 , in conjunction with the small decrease observed in k pol from correct to incorrect nucleotide incorporation (from 1.5-to 45-fold). Although the large increase in K d(dNTP) from correct to incorrect nucleotide binding in 9°N polD (from 120-to 1500fold) somewhat offsets the small change in k pol , replicative polymerases typically have a significant -fold change in both k pol and K d(dNTP) during nucleotide discrimination. It is possible, and quite likely, that the k pol and, consequently, the nucleotide selectivity and fidelity of polD are highly dependent upon the presence of replisome components, such as proliferating cell nuclear antigen, which are absent in our in vitro studies.
Ribonucleotide Discrimination-In the cell, rNTPs are present in a 1000-fold excess over dNTPs (30). To cope with this imbalance and to maintain genome integrity during replication and repair, DNA polymerases have evolved specific mecha-  nisms to exclude rNTPs (29). Family A, B, and Y DNA polymerases and RTs exclude rNTPs by a clash between a conserved bulky side chain "steric gate" amino acid and the rNTP C2Ј-OH.
In Family X DNA polymerases (pol and pol␤), an active site backbone carbonyl clashes with the ribose C2Ј-OH, preventing rNTP incorporation (42,43). These clashes prevent the binding and incorporation of the rNTP (25, 44 -46). The kinetic basis of rNTP discrimination in Family A, B, and RT is due to weak binding (high K d ) and slower incorporation (low k pol ) compared with dNTPs. Family Y DNA polymerases bind rNTPs and dNTPs with similar affinity; therefore, discrimination occurs during catalysis (Ͼ6000-fold reduction in k pol for rNTPs) (47,48). The kinetic analysis of rATP incorporation by polD performed here shows that discrimination is primarily due to weak rATP binding (144-fold higher K d ) and slower incorporation (16-fold slower k pol ) compared with dATP. Kinetic data are similar to Family A, B, Y, and RT discrimination kinetics and suggest that a polD active site amino acid may block rNTP. However, the amino acid(s) important for rNTP discrimination in polD are currently unknown. polD lacks conserved steric gate motifs (28,29), and in the absence of polD structural information, the location and identity of steric gate amino acids for rNTP discrimination within the polD family remain elusive.
Conclusions and Future Directions-The kinetics performed here on polD (the most recent DNA polymerase family to be identified) allows a comprehensive and direct comparison of nucleotide incorporation and 3Ј-5Ј exonuclease activities of all currently known DNA polymerase families. polD employs nucleotide discrimination mechanisms to prevent misincorporations and contains an active site steric gate amino acid to prevent ribonucleotide incorporation, similar to other DNA polymerase families. Due to a lack of polD structural information, the identity of nucleotide discrimination and steric gate active site determinants remains unclear and will be the focus of future structural studies. Furthermore, structural data will help to identify the polD active site, needed to explore the proposed conformational changes associated with nucleotide incorporation, including open and closed polymerase conformations. Although polD has been implicated as the major replicative polymerase in much of Archaea, the kinetic work performed here suggests low polymerase fidelity. We propose that the presence of other replisome components, such as proliferating cell nuclear antigen, will increase polD fidelity. The effect of these replisome components on polD fidelity will be explored in future work. Finally, although polD contains a 3Ј-5Ј exonuclease hydrolysis activity similar to many other polymerase families, the reliance on Mn 2ϩ for optimal activity is unique to the polD family and importantly suggests that Mn 2ϩ may play an important role in polD fidelity. Our data demonstrate that polD follows the same overall kinetic pathway as the other DNA polymerase families despite being a two-subunit enzyme with little sequence similarity to other families of DNA polymerases. Importantly, despite active site divergence among families, DNA polymerases, including polD, have each evolved specific mechanisms to accurately and faithfully replicate genomes.