Kinetic and Structural Impact of Metal Ions and Genetic Variations on Human DNA Polymerase ι*

DNA polymerase (pol) ι is a Y-family polymerase involved in translesion synthesis, exhibiting higher catalytic activity with Mn2+ than Mg2+. The human germline R96G variant impairs both Mn2+-dependent and Mg2+-dependent activities of pol ι, whereas the Δ1–25 variant selectively enhances its Mg2+-dependent activity. We analyzed pre-steady-state kinetic and structural effects of these two metal ions and genetic variations on pol ι using pol ι core (residues 1–445) proteins. The presence of Mn2+ (0.15 mm) instead of Mg2+ (2 mm) caused a 770-fold increase in efficiency (kpol/Kd,dCTP) of pol ι for dCTP insertion opposite G, mainly due to a 450-fold decrease in Kd,dCTP. The R96G and Δ1–25 variants displayed a 53-fold decrease and a 3-fold increase, respectively, in kpol/Kd,dCTP for dCTP insertion opposite G with Mg2+ when compared with wild type, substantially attenuated by substitution with Mn2+. Crystal structures of pol ι ternary complexes, including the primer terminus 3′-OH and a non-hydrolyzable dCTP analogue opposite G with the active-site Mg2+ or Mn2+, revealed that Mn2+ achieves more optimal octahedral coordination geometry than Mg2+, with lower values in average coordination distance geometry in the catalytic metal A-site. Crystal structures of R96G revealed the loss of three H-bonds of residues Gly-96 and Tyr-93 with an incoming dNTP, due to the lack of an arginine, as well as a destabilized Tyr-93 side chain secondary to the loss of a cation-π interaction between both side chains. These results provide a mechanistic basis for alteration in pol ι catalytic function with coordinating metals and genetic variation.

Genomic DNA is continuously attacked by a variety of endogenous and exogenous agents in cells, and the persistent unrepaired lesions can lead to genomic mutations and related diseases such as cancer. DNA polymerases (pols), 2 as well as DNA repair enzymes, are key enzymes for maintaining or altering genomic integrity against DNA lesions during various DNA transactions in organisms. The DNA replicative mechanisms linked to DNA damage and repair are believed to contribute to producing various mutational signatures in human cancer genomes (1). At least 17 different human DNA polymerases have been identified to date, which differ in their functions in DNA replication, repair, recombination, and damage tolerance (2,3).
Y-family DNA polymerases, including pols , , , and REV1, are specialized in replicating through DNA lesions, so-called translesion DNA synthesis (TLS). These polymerases have low fidelity with undamaged DNA templates but have spacious and solvent-accessible active sites to allow the accommodation and replicative bypass of bulky and distorted DNA lesions (4). Individual Y-family polymerases play error-free or error-prone roles in TLS, depending on DNA lesion types in cells (5). At bulky carcinogen-derived N 2 -G DNA lesions (e.g. benzo[a]pyrene-diol epoxide-G), both pol and REV1 catalyze error-free TLS, but both pols and catalyze error-prone TLS (6 -14). By contrast, at UV-induced cyclobutane thymine dimers (T-T), only pol (but not the other Y-family pols) can catalyze errorfree TLS (15,16). Therefore, the overall balance toward errorfree TLS with all working polymerases at various DNA lesions might be crucial in preventing mutations from numerous genotoxic agents. Recently, we reported that catalytic (either hypoactive or hyperactive) alterations are found in a considerable number of human germline non-synonymous variants of Y-family pols , , and REV1 (17)(18)(19), which might potentially influence on the overall TLS capacity in the affected individuals.
pol is exceptionally error-prone in DNA synthesis among polymerases, particularly opposite template bases G and T, due to its uniquely restricted active-site and related non-Watson-Crick base pairing (20 -22). pol is able to catalyze nucleotide insertion opposite a variety of DNA lesions, including N 2 -and O 6 -alkyl and aralkyl G adducts, 8-oxo-7,8-dihydroG (8-oxoG), pyrimidine dimers, and abasic sites, but mediates largely miscoding TLS (albeit occasionally accurate) with varied nucleotide selectivity depending on lesion type (13,23,24). Both C and T are inserted opposite template N 2 -and O 6 -alkyl and aralkyl G adducts by pol (13,25). C is only slightly favored over G opposite template 8-oxoG by pol , A is favored opposite the 3Ј T of (6-4) T-T photoproducts, and both G and T are favored opposite 5Ј T of (6-4) T-T photoproducts and abasic sites (23,24). A possible implication of pol in mutation and cancer has been suggested by substantial evidence from several knock-out mouse studies (26 -28), as well as from multiple studies verifying frequent pol dysregulation in various types of human cancers (29 -32). In this respect, the appropriate catalytic function of pol in cells might be required for preventing cancer. One of the distinctive catalytic properties of pol is the metal ion preference. Unlike other Y-family polymerases, pol is known to prefer Mn 2ϩ over Mg 2ϩ as the metal ion cofactor for catalysis (33,34), but the structural mechanistic details remain speculative. In addition, substantial alterations were reported in metaldependent DNA polymerase activity from two rare human germline pol variants, i.e. severe impairment of both Mg 2ϩdependent and Mn 2ϩ -dependent activities in the R96G variant and moderate enhancement of only the Mg 2ϩ -dependent activity in the ⌬1-25 variant for matched and mismatched nucleotide incorporations opposite normal and lesion templates (17). Detailed structural and kinetic mechanisms of catalytic alterations of pol by these genetic variants still remain unclear. Any disease associations have not been reported yet, but the catalytically altered pol genetic variants might be of potential importance in that they would change the TLS capacity of pol and consequently modify mutation phenotypes to genotoxic agents in genetically predisposed individuals.
To elucidate both the kinetic and the structural basis for alterations in the catalytic function of pol by different two metal ions, Mg 2ϩ and Mn 2ϩ , as well as by two human germline non-synonymous variants, R96G and ⌬1-25, we performed pre-steady-state kinetic analysis for nucleotide insertion by pol and also determined x-ray crystal structures of ternary pol complexes in the presence of either Mg 2ϩ or Mn 2ϩ ions, using the recombinant human pol core (residues 1-445) proteins of wild type and two variants with a simple model of a correct dCTP incorporation opposite normal G. The combined presteady-state kinetic and structural results indicate that Mn 2ϩ enables pol to adopt more ideal octahedral coordination in the active site and achieve much higher catalytic efficiency than Mg 2ϩ , whereas the R96G variant results in the loss of hydrogen bond interactions of residues Gly-96 and Tyr-93 with an incoming nucleotide as well as conferring a much greater reduction in its catalytic efficiency. Our detailed kinetic and structural results are discussed in the context of understanding the possible mechanistic and functional aspects of metal ions and genetic variations on pol .

Pre-steady-state Kinetic Analysis of dCTP Incorporation
Opposite G by pol , pol (26 -445), and R96G pol  Enzymes in the Presence of Mg 2ϩ or Mn 2ϩ -Pre-steadystate kinetic methods were used to quantify the alterations in catalytic efficiency and the apparent nucleotide binding affinity of pol by metal ions (Mg 2ϩ or Mn 2ϩ ) and known human genetic variants (⌬1-25 or R96G). Pre-steady-state kinetic parameters were determined for dCTP incorporation opposite template G into 18-mer/36-mer duplexes by pol (1-445), pol (26 -445), and R96G pol (1-445) enzymes under single turnover conditions (where pol was present in 10-fold excess over DNA substrate), in the presence of either 0.15 mM MnCl 2 or 2 mM MgCl 2 , which is in the optimal range for pol activity (33,34), using a rapid quench flow instrument. Analysis of the change of the observed rate (k obs ) as a function of increasing dCTP concentration yielded k pol , the maximal rate of nucleotide incorporation, and K d,dCTP , a measure of the binding affinity of dCTP to the pol⅐DNA binary complex to form a ternary complex poised for catalysis (Table 1 and supplemental Fig. S1). pol (1-445) displayed a k pol of 0.74 Ϯ 0.06 s Ϫ1 and a K d,dCTP of 1.5 Ϯ 0.4 M in the presence of Mn 2ϩ . Thus, the catalytic efficiency (k pol /K d,dCTP ) of pol (1-445) with Mn 2ϩ was estimated to be 4.9 ϫ 10 5 M Ϫ1 s Ϫ1 , which was 770-fold higher than that with Mg 2ϩ , mainly due to a 450-fold lower K d,dCTP . Similar trends of kinetic results were also observed with pol (26 -445) and the R96G variant, indicating that pol binds nucleotide much more tightly and catalyzes nucleotide insertion much more efficiently in the presence of Mn 2ϩ than in the presence of Mg 2ϩ . pol (26 -445) had a k pol /K d,dCTP value similar to that of pol  in the presence of Mn 2ϩ but displayed a 3-fold increase in k pol /K d,dCTP in the presence of Mg 2ϩ when compared with pol (1-445), indicating a slight enhancement only in the Mg 2ϩ -dependent catalytic efficiency of pol due to the deletion of N-terminal 25 residues. The R96G variant showed a 53-fold decrease in k pol /K d,dCTP for dCTP insertion opposite G in the presence of Mg 2ϩ when compared with that with wild type, while showing a 9-fold decrease of that value in the presence of Mn 2ϩ . This mitigation effect of Mn 2ϩ on the Mg 2ϩ -de-  (26 -445), and R96G pol (1-445) with DNA containing template G and dCMPNPP in the presence of either Mg 2ϩ or Mn 2ϩ . The strategy, employing a nonhydrolyzable dCTP analogue, dCMPNPP, as well as DNA substrate having an intact 3Ј-OH at the primer end, was utilized to capture pre-catalytic ternary pol complexes that preserve metal-coordinating ligands in the active site while preventing catalysis in the presence of active-site metal ions, as successfully applied with pol ␤ (35). Crystals of ternary complexes of pol (26 -445), pol (1-445), and R96G pol (1-445) diffracted to about 2.5-2.6, 3.2-3.6, and 2.8 Å resolution, respectively ( Table  2). All ternary complex structures of pol contained two metal ions with relatively lower occupancy in the A-site metal ions in the active site, except for the R96G pol (1-445)⅐Mg 2ϩ ternary complex that missed an Mg 2ϩ ion at the metal A-site near the primer terminus 3Ј-OH. To our knowledge, our structures represent the first ternary pol structures containing the primer end 3Ј-OH entity, as well as defining the position of two activesite metal ions of either Mg 2ϩ or Mn 2ϩ , which provides geo-metric information in the pre-catalytic state. However, electron density was not observed for the N-terminal 25 residues in all pol (1-445) ternary complexes, suggesting the disordered nature of this negatively charged N-terminal region. Thus, all the refined structures of pol ternary complexes contained pol residues 51-439 as observed previously with the ternary pol (26 -445) complex structure (36). The overall structures of six ternary pol complexes were almost identical, except for subtle variations near metal ions in the active site and at the amino acid substitution site (supplemental a Values for highest resolution shell are given in parentheses. b R merge : R linear ϭ SUM(ABS(I Ϫ ͗I͘))/SUM(I), where I is the integrated intensity of a given reflection. c One non-catalytic metal ion is present in the structure. SEPTEMBER   39) (Table 3 and Fig. 2, D and E). In strong contrast, the Mg 2ϩ at the A-site in the pol active site showed a considerable deviation from the ideal octahedral coordination geometry, yielding average coordination distance (2.45 Å) and distance RMSD (0.474 Å) values that were substantially higher than those with Mn 2ϩ (Table 3 and Fig. 2, C and D). These more optimal fea-  tures with Mn 2ϩ than Mg 2ϩ for octahedral coordination geometry at the A-site were similarly observed with the refined pol (1-445) ternary complex structures (Table 3), albeit having lower resolution (3.6 and 3.2 Å). The poor coordination with Mg 2ϩ at the A-site appeared to be related to subtle displacement of the side-chain carboxyl group of Asp-59 away from the A-site Mg 2ϩ as well as a slight shift of the A-site Mg 2ϩ to the B-site Mg 2ϩ , yielding a 0.2 Å shortening of inter-metal distance when compared with that with Mn 2ϩ in the pol active site (Fig.  2, C, D, and F). Interestingly, the C3Ј-endo conformation was equally observed at the sugar moieties at the 3Ј primer end (Fig.  2), as well as the nucleotide 5Ј to the primer end, and three nucleotide pairs at positions n-2 to n-4 of the primer/template duplex in both pol ternary complex structures, unlike the pre-viously reported structures of the pol binary and ternary complexes lacking the primer end 3Ј-OH (Protein Data Bank (PDB) IDs 2FLP and 2ALZ) (36,40), indicating a distinctive pattern of sugar pucker changes induced in the pre-catalytic pol ternary complex.

Effects of Metal Ion and Genetic Variation on pol
Active-site Structures of Human R96G pol (1-445) Ternary Complexes in the Presence of Mg 2ϩ or Mn 2ϩ -To reveal the structural mechanism of severe catalytic impairments in the R96G pol variant, we compared the active-site structures of R96G pol (1-445) ternary complexes with those of pol (26 -445) ternary complexes of relatively high resolution. Interestingly, the R96G variant structures showed substantial alterations not only at the amino acid substitution site (Gly-96) but also at the nearby residue site (Tyr-93), when compared with pol (26 -445) (Fig. 3). The F o Ϫ F c simulated annealing omit maps for two key residues (Gly-96 and Tyr-93) in R96G pol ⅐Mg 2ϩ and R96G pol ⅐Mn 2ϩ ternary complexes showed no clear electron density for two long side chains of Arg-96 and Tyr-93 (Fig. 3, B and D), which were obviously observed in those with the pol (26 -445) complexes (Fig. 3, A and C). This collateral destabilization of the Tyr-93 side chain seems to be related to the loss of a cation-interaction between Arg and Tyr side chains due to the loss of an Arg side chain in the R96G variant. Accordingly, two hydrogen bonds of the Arg-96 side chain with a ␤,␥-bridging oxygen and a ␥-phosphate oxygen of dCMPNPP, as well as one hydrogen bond of the Tyr-93 side chain with a ␥-phosphate oxygen, were lost in R96G variant structures when compared with those in pol (26 -445) ternary complex structures (Fig. 3, E-H). Interestingly, only Mn 2ϩ but not Mg 2ϩ was observed at the A-site in the R96G pol active site (Fig. 3, B and D), possibly reflecting an inferior A-site binding ability of Mg 2ϩ when compared with Mn 2ϩ , which might in part be attributed to a 6-fold greater decrease in catalytic efficiency of the R96G variant with Mg 2ϩ than that with Mn 2ϩ (Table 1).

Discussion
In this study, we provide structural and pre-steady-state kinetic evidence that Mn 2ϩ is more optimal for the two-metal ion binding site configuration and catalysis of pol than Mg 2ϩ . Our pol ternary complex structures containing intact coordinating ligands such as the primer end 3Ј-OH reveal that Mn 2ϩ ions achieve more ideal octahedral coordination geometry than Mg 2ϩ ions in the pol active site, specifically in the catalytic metal A-site. Moreover, our pre-steady-state kinetic data demonstrate that Mn 2ϩ ions confer much higher (2-3 orders of magnitude) efficiency of pol catalysis than Mg 2ϩ ions, mainly through augmenting nucleotide binding affinity. We also confirmed that the R96G variant, displaying a severe reduction in catalytic efficiency, loses three hydrogen bond interactions of two residues (Gly-96 and Tyr-93) with an incoming nucleotide in the pol active site. Interestingly, Mn 2ϩ substantially rescued the catalytic impairment in the Gly-96 variant by restoring the apparent nucleotide binding affinity. The ⌬1-25 variant displayed a small alteration (3-fold increase) only in Mg 2ϩ -dependent catalytic efficiency, although the structural effects of N-terminal 25 residues are not discernible due to the absence of their electron density.
The key differences between the pol (26 -445)⅐Mg 2ϩ and pol (26 -445)⅐Mn 2ϩ ternary complex structures were located at and near the catalytic metal A-site in the pol active site, particularly the configuration of the Asp-59 side chain as well as the positioning of the A-site metal (Fig. 2F), which substantially modified the A-site metal coordination. First, in the presence of Mg 2ϩ ions as opposed to Mn 2ϩ ions, the side-chain carboxyl group of Asp-59 is rotationally displaced to locate a ligand atom (one carboxyl oxygen of Asp-59) more distant (0.5 Å) from the A-site metal (despite no alteration in coordination distance between the other carboxyl oxygen and the B-site metal) in the pol active site. Second, Mg 2ϩ (when compared with Mn 2ϩ ) at the A-site was positionally shifted closer (ϳ0.2 Å) to the B-site metal in the pol active site, yielding an inter-metal distance shorter than that with Mn 2ϩ ions (Table 3). Thus, in the pol active site, only Mn 2ϩ ions but not Mg 2ϩ ions seem to achieve an inter-metal distance similar to that with Mg 2ϩ ions usually observed in active sites of ternary complex structures of other human pols and ␤ with normal G⅐dCMPNPP (38,39) (Table  3). Superposition of the active-site metal binding sites of pol ⅐Mg 2ϩ , pol ⅐Mn 2ϩ , and pol ⅐Mg 2ϩ (PDB ID 4DL3) ternary complexes appears to reflect some gradational changes in the extent of positional shift of the A-site metal (toward the B-site metal) as well as in the angle of rotational displacement of the side-chain carboxyl group of Asp-59 (or Asp-13 for pol ), in the order of pol ⅐Mg 2ϩ Ͻ pol ⅐Mn 2ϩ Ͻ pol ⅐Mg 2ϩ . (Fig. 2F). Consequently, Mg 2ϩ ions led to a considerable deviation from the ideal octahedral coordination geometry at the A-site, yielding the values of average coordination distance and distance RMSD (2.45 and 0.474 Å, respectively), which were quite higher than those with Mn 2ϩ (Table 3). These coordination parameters with Mg 2ϩ ions seem to be slightly improved in the aspect of average coordination distance but worsened in the aspect of distance RMSD by the presence of the primer terminus 3Ј-OH, when compared with those (2.66 and 0.456 Å, respectively) of the A-site Mg 2ϩ (albeit involving only four coordinating atoms) in the previously reported structure (PDB ID 2ALZ) of pol ternary complex lacking the primer end 3Ј-OH (36). This Mg 2ϩ -induced geometric alteration at the A-site in the pol active site contrasts with the near perfect octahedral coordination of the A-site Mg 2ϩ observed in ternary complex structures of other pols and ␤ with the correct incoming non-hydrolyzable nucleotide (Table 3) (38,39).
The Mn 2ϩ requirement for achieving the optimal octahedral coordination geometry at the A-site in ternary complex structures with the correct incoming nucleotide seems to be unique in pol . Other DNA polymerases such as bacteriophage pol RB69 and human pol ␤ achieve good octahedral coordination geometry not only with Mg 2ϩ but also with Mn 2ϩ at the A-site, as observed in their ternary complex structures with the correct incoming non-hydrolyzable nucleotide (PDB IDs 3SJJ, 3SPY, 2FMS, and 3C2K) (35,41,42). It is notable that the superior coordinating ability of Mn 2ϩ when compared with Mg 2ϩ at the A-site is observed in the previously reported structures of ternary pol ␤ complexes (PDB IDs 4PGQ, 4PGX, 4PHA, and 4PHD) with an incoming incorrect non-hydrolyzable nucleotide (43). Our data and that of others suggest that Mn 2ϩ is more tolerant of atypical pol active sites (e.g. the inherently restricted pol active site and the distorted pol active site due to base pair mismatch) than Mg 2ϩ ) and thus able to form good octahedral coordination geometry not only at the B-site but also at the A-site in the pol active site. This effect might be attributed to a relaxed coordination requirement of Mn 2ϩ when compared with Mg 2ϩ (37). It is of interest to perform further studies to verify whether our results of pol with normal base pairs of template G and incoming dCTP are valid for other base pairs involving DNA lesions or mismatches such as G:T and T:G pairs.
The Mg 2ϩ -induced distortion of the A-site coordination geometry in the pol active-site structure seems to be closely related to a severe (220 -770-fold) diminution in catalytic efficiency (k pol /K d,dCTP ) of pol in the presence of Mg 2ϩ when compared with Mn 2ϩ (Table 1). This severe catalytic impairment with Mg 2ϩ was mainly due to a severe reduction in the affinity of productive nucleotide binding of pol in the presence of Mg 2ϩ when compared with Mn 2ϩ , as reflected by large (220 -450-fold) increases in the apparent equilibrium dissociation constant (K d,dCTP ) for incoming dCTP (Table 1). These results suggest that substitution of Mn 2ϩ for Mg 2ϩ might boost the catalytic efficiency of pol for correct nucleotide insertion, mainly through improving the binding affinity of nucleotide, by achieving optimal octahedral coordination at the A-site in the active site. Our finding is in good agreement with previous studies with other polymerases, e.g. RB69 and pol ␤ (42,44). It is also notable that the extent of increases (220 -450-fold) in the nucleotide binding affinity by Mn 2ϩ substitution appears to be much higher with pol (Table 3) than those (3-4-and 8 -19fold, respectively) observed with RB69 and pol ␤ (42, 44), implying a more marked effect of Mn 2ϩ on pol than other polymerases. Our combined structural and kinetic data suggest that the optimal octahedral coordination of two active-site metal ions is essential for proper catalytic function of pol , and Mn 2ϩ is superior in this respect when compared with Mg 2ϩ , particularly at the A-site in the pol active site. The superiority of Mn 2ϩ for pol function is also supported by the biochemical property of pol (1-445) to more tightly bind DNA substrates in the presence of a low level of Mn 2ϩ than Mg 2ϩ (17). Although DNA polymerases most likely utilize physiologically abundant Mg 2ϩ ions for catalysis in vivo, from our and previous studies (17,33,34), it may be relevant to postulate that pol would inherently employ physiologically low levels of Mn 2ϩ for catalysis in kinetic and structural preference to Mg 2ϩ in vivo. Similarly, pol has been also suggested to use Mn 2ϩ as the preferred activating metal ion in vivo (45). Mn 2ϩ has also been reported to increase the activity of non-canonical DNA polymerases such as Sulfolobus solfataricus Dpo4 and human Prim-Pol in vitro (46,47). We also note a very recent study suggesting the requirement of a third metal ion for pol catalysis, with slight preference for Mg 2ϩ (48). It would be of interest to investigate whether this is the case in pol catalysis.
The loss of hydrogen-bonding interactions of two structurally altered residues, the substituted Gly-96 and the nearby destabilized Tyr-93, with the incoming nucleotide in the crystal structures of the R96G pol ternary complexes (Fig. 3) may provide a molecular explanation for severe diminution of the catalytic efficiency in the R96G variant ( Table 1). The disordered electron density of the Tyr-93 side chain in R96G pol crystal structures seems to be related to the destabilization of the Tyr-93 side chain due to the absent cationinteraction with Gly-96. Both Arg-96 and Tyr-93 residues, which are conserved and important residues for nucleotide binding in all Y-family pols (20), seem to co-stabilize their side chain conformations by forming the cationinteraction between their parallel side chains as observed in our pol ternary complexes (Fig.  3). In good accordance with our data, individual missense mutations of two homologous Arg-67 and Tyr-64 residues in yeast pol severely diminish its catalytic activity (49). It can be postulated that the weakened interaction of the R96G variant with incoming nucleotide likely diminishes its nucleotide binding affinity and catalytic efficiency. Interestingly, the attenuat-ing effect of the R96G variation on the apparent nucleotide binding affinity of pol was observed much more strongly in the presence of Mg 2ϩ (Table 1). Mn 2ϩ substitution fully restored the apparent nucleotide binding affinity of the R96G variant to a level similar to that of wild-type pol (Table 1), indicating a rescuing effect of Mn 2ϩ to offset the destabilized nucleotide interaction in the R96G variant. Accordingly, the extent of reduction of catalytic efficiency in the R96G variant was considerably lessened in the presence of Mn 2ϩ when compared with Mg 2ϩ ( Table 1).
The variant of pol (26 -445) lacking the N-terminal 25 residues retained an Mn 2ϩ -dependent catalytic efficiency almost similar to that of the pol (1-445) but displayed a 3-fold increase in selectivity in Mg 2ϩ -dependent catalytic efficiency when compared with that of the pol (1-445) ( Table 1). These results are in good agreement with our previous steady-state kinetic data (17). However, the structural effects of the N-terminal 25 residues are not clear due to their disordered nature in the x-ray crystal structure. Although it may not be obvious due to its low resolution (3.6 Å), interestingly, our refined crystal structure of pol (1-445) ternary complex in the presence of Mg 2ϩ appeared to have slightly altered conformations of Arg-96 and Tyr-93 side chains when compared with the pol (26 -445) ternary complex (supplemental Fig. S1C). This subtle structural alteration, observed only with Mg 2ϩ ions in the pol (1-445) active site, may in part explain a slightly lower efficiency for Mg 2ϩ -dependent catalysis by pol (1-445) when compared with pol (26 -445).
In summary, we have investigated the effects of different metal ions (Mg 2ϩ and Mn 2ϩ ) and genetic variations (R96G and ⌬1-25) on both the structure and the catalytic function of pol .
Here we report the first x-ray crystal structures acquired in the presence of either Mg 2ϩ or Mn 2ϩ of pol ternary complexes having intact coordinating metals and ligands such as the primer terminus 3Ј-OH. Comparisons of active-site conformations between pol ⅐Mg 2ϩ and pol ⅐Mn 2ϩ ternary complexes revealed that pol adopts near perfect octahedral coordination geometries for two metal ions in the active site only in the presence of Mn 2ϩ . This structural feature with Mn 2ϩ is clearly consistent with the pre-steady-state kinetic observation that Mn 2ϩ greatly bolsters the apparent nucleotide binding affinity and the catalytic efficiency of pol when compared with observations with Mg 2ϩ . Moreover, our combined structural and presteady-state kinetic analysis also revealed that the catalytic impairment in the R96G variant is related to the lack of hydrogen-bonding interactions of Gly-96 and Tyr-93 with incoming nucleotides in the active site. Our comparison between pol (1-445) and pol (26 -445) also suggests a potential role of the disordered N-terminal 25 amino acids in selectively improving (albeit slightly) the Mn 2ϩ -dependent catalytic efficiency in the wild-type pol . Overall, our study provides insights into the delicate structural and kinetic features of different metal coordination and genetic variants that contribute to understanding of the molecular basis of the catalytic function of pol .
Pre-steady-state Reactions-The 18-mer primer (5Ј-AGC CAG CCG CAG ACG CAG-3Ј) was 5Ј end-labeled using T4 polynucleotide kinase with [␥-32 P]ATP and annealed with 36-mer template (3Ј-CGG AGC TCG GTC GGC GTC TGC GTC GCT CCT GCG GCT-5Ј). Rapid quench experiments were performed using a model RQF-3 KinTek Quench Flow instrument (KinTek Corp., Snow Shoe, PA). All DNA polymerase reactions were performed in 50 mM Tris-HCl (pH 7.5) buffer containing 5 mM dithiothreitol, 100 g ml Ϫ1 BSA (w/v), 10% glycerol (v/v), and 2 mM MgCl 2 (or 0.15 mM MnCl 2 ). Reactions were initiated by rapid mixing of 32 P-primer/template/polymerase mixtures (18-mer/36-mer, 100 nM; pol , 1 M, in 10-fold excess to DNA substrate to ensure single turnover conditions) with the metal-dCTP mixtures (2 mM MgCl 2 or 0.15 mM MnCl 2 ; dCTP, in varying concentrations), and then quenched with 0.15 M EDTA at time intervals from 0.15 to 30 -120 s (or from 2 to 240 -480 s for the R96G variant). MgCl 2 was supplemented by as much as the increase of dCTP to counterbalance the Mg 2ϩ -chelating effect of dCTP for the reactions at high levels of dCTP. Reaction products were mixed with formamide-dye solution (20 mM EDTA, 95% formamide (v/v), 0.5% bromphenol blue (w/v), and 0.05% xylene cyanol (w/v)) and separated using an 8 M urea-containing denaturing gel with 16% polyacrylamide (w/v), and then quantified by a Bio-Rad Personal Molecular Imager instrument and the Quantity One software. Pre-steady-state data obtained under the single turnover condition were fit to the single-exponential equation y ϭ A(1 Ϫ exp(Ϫk obs t)), where y ϭ concentration of product, A ϭ reaction amplitude, k obs ϭ observed rate of nucleotide incorporation, and t ϭ time (50,51), using nonlinear regression analysis in GraphPad Prism software.
Determination of k pol and K d,dCTP -The pre-steady-state kinetic parameters k pol and K d,dCTP were estimated by analyzing the dCTP dependence on the observed pre-steady-state rates of dCTP insertion under single turnover conditions. A graph of the observed rate (k obs ) versus dCTP concentration was fit to the hyperbolic equation k obs ϭ [k pol [dNTP]/([dNTP] ϩ K d )], where k pol is the maximal rate of nucleotide incorporation and K d,dCTP is the equilibrium dissociation constant for dCTP (50,51).
Structure Determination and Refinement-X-ray diffraction data were collected on the 21-ID-F (Life Sciences Collaborative Access Team (LS-CAT)) beam line at the Advanced Photon Source (Argonne National Laboratory, Argonne, IL). Collected data were indexed, integrated, and scaled using HKL2000 (52). All crystal types belonged to space group P6 5 22. Structures were determined by molecular replacement phasing using the program Phaser MR (53) and the pol structure (PDB ID 2ALZ) as a search model. Structure refinements and model building were performed using PHENIX (54) and COOT (55). Metal coordination geometry was analyzed with UCSF Chimera (56). Structural illustrations were prepared with PyMOL (Schrödinger, LLC).