Presence of 18-Å Long Hydrogen Bond Track in the Active Site of Escherichia coli DNA Polymerase I (Klenow Fragment)

The analysis of the active site region in the crystal structures of template-primer-bound KlenTaq (Klenow fragment equivalent of Thermus aquaticus polymerase I) shows the presence of an ∼18-Å long H-bonding track contributed by the Klenow fragment equivalent of Asn845, Gln849, Arg668, His881, and Gln677. Its location is nearly diagonal to the helical axis of the template-primer. Four base pairs in the double stranded region proximal to 3′ OH end of the primer terminus appear to interact with individual amino acid components of the track through either the bases or sugar moieties. To understand the functional significance of this H-bonding network in the catalytic function of Klenow fragment (KF), we generated N845A, N845Q, Q849A, Q849N, R668A, H881A, H881V, Q677A, and Q677N mutant species by site-directed mutagenesis. All of the mutant enzymes showed low catalytic activity. The kinetic analysis of mutant enzymes indicated that K m .dNTP was not significantly altered, but K D .DNA was significantly increased. Thus the mutant enzymes of the H-bonding track residues had decreased affinity for template-primer, although the extent of decrease was variable. Most interestingly, even the reduced binding of TP by the mutant enzymes occurs in the nonproductive mode. These results demonstrate that an H-bonding track is necessary for the binding of template-primer in the catalytically competent orientation in the pol I family of enzymes. The examination of the interactive environment of individual residues of this track further clarifies the mode of cooperation in various functional domains of pol I.

erase I has been extensively used as a model system for understanding the structural basis for the polymerase reaction mechanism. The crystal structures have shown the 68-kDa KF folds into two distinct domains, a ϳ200-amino acid long 3Ј-5Ј exonuclease domain at the N-terminal, and a ϳ400-residue polymerase domain at the C-terminal (8 -10). The polymerase domain, which anatomically resembles a half-open right-hand, consists of a cleft formed by three subdomains designated as fingers, palm, and thumb (8,10). These subdomains have been implicated in different functions of polymerases (7). The active site constituted by motifs A and C resides at the palm subdomain, the dNTP binding site at the fingers subdomain, and the template-primer binding site at the thumb subdomain (7).
The crystal structures of several DNA polymerases complexed with template-primer and dNTP have shown the participation of individual amino acid residues in the respective polymerase function (4,6,11,12). In particular, the crystal structures of binary and ternary complexes of the Klenow fragment of Thermus aquaticus DNA polymerase I (KlenTaq) can be considered to represent the snapshots of initial steps of the nucleotidyltransferase reaction pathway of DNA polymerases (11). The analyses of these crystal structures together with the crystal structure of apo-KlenTaq revealed that the binding of template-primer to the enzyme induces a conformational rearrangement of the side chains of several residues present at the active site. We noted that resulting from this conformational change in the side chains is the formation of a predominately hydrogen bonded network of ϳ18 Å long at the bottom of the cleft (Fig. 1). This network of hydrogen bonding (called "Hbonding track" hereafter) is present in both binary and in the "open" and "closed" conformations of the ternary complexes. The residues that participate in the formation of the H-bond track are highly conserved in the pol I family of DNA polymerases. They are equivalent to the E. coli pol I residues Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 .
To understand the significance of the H-bonding track residues at the active site of pol I and their functional contribution, we carried out site-directed mutagenesis of Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 . These residues were mutated to both conserved and nonconserved (alanine) residues. Properties of the mutant proteins were then investigated to assess their individual participation in the catalytic function of KF. All of the mutants showed decreased catalytic activity, albeit to different degrees, suggesting the involvement of these residues in the catalytic function of pol I. Most importantly, all of the mutant enzymes showed significantly reduced (15-75-fold) template-primer binding affinity compared with wild-type enzyme. Furthermore, neither the conservative nor the nonconservative mutant enzymes showed ability to form prepolymerase binary or ternary complexes. Similarly, unlike the wildtype enzyme, the enzyme-TP covalent complexes generated for individual mutant species of all five residues failed to catalyze in situ addition of substrate dNTP. These results together with the structural analysis of the binary and ternary complexes of pol I family DNA polymerases strongly implicate the function of H-bonding track residues in the binding of template-primer at the proper position and orientation, which is prerequisite for the nucleotidyltransferase function of polymerases. An examination of the location of some of the highly conserved residues present in the different motifs of pol I, in the context of the H-bonding track, further reveals the coordinated participation of all of these motifs in the catalytic function.
plate-primers (Chart 1). The reaction was carried out in a final volume of 100 l containing 50 mM Tris-HCl, pH 7.8, 1 mM DTT, 100 g/ml bovine serum albumin, 250 nM of the template-primer, 5 mM MgCl 2 , and 25 M [ 32 P]dNTP (0.5 Ci/assay) corresponding to the homo-and heteropolymeric template-primers. Reactions with the heteropolymeric template-primer contained all four dNTPs at a concentration of 25 M, with two of them radiolabeled. Unless otherwise indicated, the final enzyme concentration was 7.5 nM. The reaction was initiated by the addition of MgCl 2 and terminated by the addition of 5% ice-cold trichloroacetic acid containing 10 mM PP i . The acid-precipitated material was collected on Whatman GF/B filters and counted for radioactivity in a liquid scintillation counter, as described elsewhere (18).
K D.DNA Determination-To determine the binding affinity of template-primer to the desired enzymes, gel mobility shift assays were performed as described previously (19). The binding of 33/20 ϩ ddC-mer and 21/12 ϩ ddC, the dideoxy-terminated template-primers (Chart 1), to various concentrations of enzyme was carried out in a reaction mixture containing 50 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 10% (v/v) glycerol, and 0.1 mg/ml bovine serum albumin. The concentration of 32 P-labeled template/primer was 50 -100 pM. Different protein concentrations were used to bracket the K D.DNA value. Samples were electrophoresed at 100 V for 1.5 h at 4°C on a 6% nondenaturing polyacrylamide gel, using 89 mM Tris borate, pH 8.2, buffer. Gels were dried, then scanned in a PhosphorImager and quantitated by ImageQuant software (Amersham Biosciences). The percent enzyme-TP binding was calculated by quantifying the amount of uncomplexed TP in each lane. Percent binding values were then used for the determination of K D.DNA by interpolation, using nonlinear regression for one-site binding (hyperbola) with GraphPad Prism software.
Determination of Steady-state Kinetics Constants-Determination of steady-state kinetic parameters was carried out as described earlier (15,18) on homopolymeric dC 60 /dG 18 and dA 36 /dT 18 template-primers, in the presence of varying amounts of complementary dNTPs (dGTP and dTTP, respectively). Saturating concentrations of a homopolymeric dA 36 /dT 18 and dC 60 /dG 18 were used as the template-primers, to mimic the first-order reaction conditions. The reaction mixture contained 50 mM Tris-HCl, pH 7.8, 50 mM KCl, 1 mM DTT, 0.01% bovine serum albumin, 1 M respective template-primers, 1 Ci/reaction [␣-32 P]dNTP, 1-1000 M respective dNTPs, 5 nM WT KF, and activity normalized mutant enzymes. The reaction was initiated by the addition of 5 mM MgCl 2 . The acid-precipitable material was filtered on a Whatman glass fiber filter and the radioactivity was quantitated by scintillation counting. Steady-state parameters K m and V max were determined from Eadie-Hoftsee plots, using the Enzyme Kinetic program. Steady-state values of k cat were calculated from the equation: Stable Ternary Complex Formation Assay-Our method for assessing stable ternary complex formation is based on the procedure described by Scott and colleagues (20,21), where nondenaturing gel electrophoresis is used to identify the enzyme-TP-dNTP ternary complex. Briefly, in a final volume of 10 l, the desired enzyme quantity (1.2 nM of the wild-type and different amount of mutant enzymes) is mixed with 21-mer 5Ј-end labeled dideoxynucleotide-terminated primer annealed to a 33-mer template (final concentration of primer is ϳ100 pM) in a standard buffer containing 50 mM Tris-HCl, pH 7.8, 10% (v/v) glycerol, 5 mM MgCl 2 , and 0.1 mg/ml bovine serum albumin. The enzyme-DNA complexes were allowed to form on ice. To one of the duplicate samples, cognate nucleotide was added at 200 M concentration to CHART 1   Template-primer used for polymerase activity  49/17-mer  5Ј-GGAAAATCAGTCACACC-3Ј  3Ј-CCTTTTAGTCAGTGTGGAAAATCTCTAGCAGTGGCGCCCGAA-CAGGGAC-5Ј   Template-primers used for binary and ternary complex formation assay  33/ Template-primer used for in situ nucleotide incorporation 5Ј-CGTTAGCCACTCCGAAGTG-3Ј 3Ј-GCAATCGGTGAGGCTTCACGCATATTGCGCG-5Ј induce ternary complex formation. The stability of the enzyme-TP binary complex and enzyme-TP-dNTP ternary complex was then assessed by the degree of persistence of DNA in the enzyme-TP complex, upon addition of 500-fold excess of the same unlabeled template-primer. We find that under these conditions, DNA from enzyme-TP binary complexes completely dissociates, whereas ternary complexes exhibit significant retention of radiolabeled template-primer by enzyme protein.
All samples were resolved on a native 6% polyacrylamide gel, and the positions of the radioactive bands representing enzyme-DNA and free DNA were visualized by exposure in a PhosphorImager. Effect of Complementary or Noncomplementary Incoming Nucleotide on Template-primer Binding Affinity-The effect of the complementary and noncomplementary nucleotide on the binding affinity of wild-type and mutant enzymes was assessed in a manner similar to that used for the stable ternary complex formation assay. The enzyme-TP binary complexes were allowed to form in 50 mM Tris-HCl, pH 7.8, 10% (v/v) glycerol, 5 mM MgCl 2 , and 0.1 mg/ml bovine serum albumin for 10 min at 4°C in duplicate. To one of the duplicate samples, the correct incoming dNTP (200 M dGTP) was added. The incorrect dNTPs (dATP, dCTP, and dTTP, 100 M each) were added to the second sample. The reaction mixture was further incubated on ice for 10 min, prior to loading on a 6% nondenaturing polyacrylamide gel, under the same conditions as those used for K D.DNA determination.
Enzyme-Template-primer Cross-linking-The UV-mediated photochemical enzyme-TP cross-linking was performed essentially as described before (18,22,23). The reaction mixtures containing 50 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 1 mM DTT, 32 P-5Ј-end labeled 33/20 ϩ ddC or 21/12 ϩ ddC template-primer and the desired amount of enzyme were incubated on ice for 10 min. The samples were exposed to UV light (wavelength ϭ 254 nm) at a dose rate of 3700 mJ/cm 2 for 3 min. The wild-type KF was 30 pmol, and the mutant proteins were varied to obtain nearly equal extent of the enzyme-template-primer cross-linked species. Measurement of covalent attachment of labeled TP or dNTP to enzyme protein was assessed by 8% SDS-polyacrylamide gel electrophoresis, followed by the exposure of gel to PhosphorImager and analysis of exposed gel by ImageQuant software (Amersham Biosciences).
In Situ Addition of Nucleotide on Immobilized TP with Enzyme-Nucleotidyltransferase activity of the wild-type and mutant enzymes, containing covalently cross-linked template-primer, was carried out as described previously (22). The wild-type KF and various mutant enzymes were cross-linked to 32/19-mer template-primer in a reaction mixture, in a final volume of 50 l, containing 35 pmol of the unlabeled TP, 50 mM Tris-HCl, pH 7.8, 1 mM DTT, and 5 mM MgCl 2 . The reaction contained 30 pmol of the wild-type KF, whereas the amount of individual mutant enzymes was varied to yield near equal quantities of enzyme-TP cross-linked species. The nucleotidyltransferase reactions were then initiated by the addition of 1 M NaCl together with 5 Ci of complementary [␣-32 P]dTTP at a final concentration of 0.5 M. The reaction mixture was incubated for 30 min at room temperature and terminated by the addition of protein solubilizing solution. An aliquot of the reaction mixture was subjected to SDS-polyacrylamide gel electrophoresis, followed by phosphorimaging and analysis by ImageQuant software (Amersham Biosciences).

RESULTS
The structural analysis of the template-primer and template-primer-dNTP bound crystal structures of KlenTaq and Bacillus stearothermophilus (11,24) showed the presence of a network of interactions (mainly H-bonds) near the active site of these polymerases. The constituent residues in these two enzymes (KlenTaq and Bst) are equivalent to Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 of E. coli pol I (Fig. 1). In the template-primer-bound binary complex of KlenTaq (Protein Data Bank file 4ktq), the H-bonds exist between Asn 845 -Gln 849 , Gln 849 -Arg 668 , and His 881 -Gln 677 . The interaction between Arg 668 and His 881 (which would complete the H-bonding track) is van der Waals in nature. As mentioned by Franklin et al. (12) for T7 DNA polymerase I, the equivalent residue to Arg 668 in KlenTaq (Arg 573 ) is buttressed by the His 881 equivalent amino acid (His 784 ). The fact that these crystal structures have been solved at 2.3 Å or lower resolution, it is possible that slightly altered modeling of the side chain conformation of His 784 (equivalent to His 881 of KF) would result in an H-bond between Arg 573 and His 784 . Therefore, for simplicity, we have the interactions among these five residues designated as the H-bonding track.
Site-directed Mutagenesis and Purification of the Mutant Enzymes-A total of 9 mutants, viz. N845A, N845Q, Q849A, been generated from the crystal structure of the enzyme-TP binary complex of KlenTaq (11) (Protein Data Bank file 4ktq). In panel A, the phosphate backbones of the template (blue) and primer (gold) are shown as ribbons. The constituent amino acid residues of the Hbonding track are colored in cyan and are shown as sticks. The two metals (in magenta balls marked A and B) at the active site and two essential aspartic acids (shown as green sticks) as seen in the ternary complex of KlenTaq have also been included as reference points. Because Asp 610 of KlenTaq (equivalent to Asp 705 of KF) assumes two different conformations in the binary and ternary complexes, its orientation in this figure is as seen in the "closed ternary complex" crystal structure (Protein Data Bank file 3ktq). The dotted lines among H-bonding track residues represent the hydrogen bond. An extra pink dotted line between Arg 668 and His 881 equivalent residues of KlenTaq represents the van der Waals interactions between these two side chains. The potential donor atoms of H-bonding track residues that can form the hydrogen bond with the minor groove moieties of template-primer (purine N3 and pyrimidine C ϭ O groups) are shown as blue balls. The amino acid residues and positions in Klen-Taq are labeled with KF numbers based on their structural and positional equivalence. In panel B, a stereo-diagram depicting another view of the H-bonding track residues is shown. This view has been generated by rotating the figure in panel A by about 90°in the clockwise direction. In this figure, the template and primer are colored blue and gold, respectively. The bonds of the template and primer are rendered in sticks, whereas the bases and sugar moieties have been filled by plates. The amino acid residues, rendered in sticks, are also shown with their van der Waals surface in different colors. These figures were generated by MolMol (44). Q849N, R668A, H881A, H881V, Q677A, and Q677N of Klenow fragment were generated by site-directed mutagenesis as described under "Experimental Procedures" (18,22). All side chains were replaced by alanine, a nonhomologous side chain, or by homologous side chain, except in the H881V mutant. The H881V mutation results in a KF sequence (VDE), which mimics the sequence present at topologically conserved positions in reverse transcriptases (VDD/MDD). The mutant enzymes were greater than 95% pure as judged by SDS-polyacrylamide gel electrophoresis. The levels of expression, solubility, and yield as well as the chromatographic characteristics of all mutant proteins were identical to that of the wild-type enzyme, suggesting no significant change in the folding pattern of the mutant enzymes.
For studies involving characterization of the DNA polymerization reaction, template-primer binding, cross-linking of dNTP, and template-primer and in situ addition, the WT and H-bonding track mutant proteins also contained the D424A mutation at the 3Ј-5Ј exonuclease active site, rendering them exonuclease-deficient. Thus, the mutant protein D424A is referred to as WT and R668A (also containing D424A mutant) as R668A mutant for the polymerase activity studies.
Specific Activities of the Mutant Enzymes with Different Template-primers-The polymerase activity of mutant enzymes was determined with two homopolymeric (dA 36 /dT 18 and dC 60 / dG 18 ) and one heteropolymeric (49/17-mer) template-primer in the presence of Mg 2ϩ as metal cofactor. The activities of different mutant enzymes expressed as a percentage of WT are summarized in Table I. All mutants showed a significantly decreased activity with dA 36 /dT 18 ranging between 0.2 and 32% of the level in the wild-type enzyme. Generally, the two His 881 mutants (H881A and H881V) appeared to retain relatively higher activity with all three template-primers, namely dA 36 / dT 18 , dC 60 /dG 18 (homopolymers), and 49/17-mer (heteropolymer). Among the three template-primers, the least activity of individual mutant enzymes was seen with dA 36 /dT 18 . The most severe effect on the catalytic activity was noted with the N845Q, Q849N, R668A, and Q677N mutant species of KF. Surprisingly, the conserved substitutions of residues Asn 845 , Gln 849 , and Gln 677 consistently exhibited greater loss of activity with all of the template-primers, compared with nonconserved substituents at these positions. These activity data suggested that the side chains of Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 have some important role in the polymerase function of KF.
Determination of Steady-state Kinetics Constants-The steady-state kinetic parameters, k cat for the polymerase reaction and K m for dNTP utilization, were determined for the WT and each mutant protein, with two homopolymeric DNA substrates (dA 36 /dT 18 and dC 60 /dG 18 ). The rates of incorporation of ␣-32 P-labeled dNTP into products at increasing concentrations of dNTP were measured, and k cat for the polymerase reaction and K m for dNTP were determined by Eadie-Hofstee plots from the rate data. The results are summarized in Table II. The wild-type enzyme exhibited a K m.dNTP of ϳ5 M with both homopolymeric template-primers. The mutant N845A showed almost no change in K m.dNTP with either template-primer, whereas the catalytic efficiency was decreased by ϳ6-fold with dA 36 /dT 18 template-primer. The catalytic efficiency with dC 60 / dG 18 template-primer was not changed significantly for the N845A mutant. However, the conserved mutant, N845Q, showed a drastic change of ϳ85-fold in catalytic efficiency. The mutants of Gln 849 exhibited a moderate change in K m.dNTP only with dC 60 /dG 18 template-primer (ϳ5-10-fold). However, the steady-state catalytic rate (k cat ) was significantly reduced with both template-primers (20-fold with dA 36 /dT 18 and 60-fold with dC 60 /dG 18 ). The change in both K m.dNTP and k cat was reflected in significantly decreased catalytic efficiency (14 -75-fold) of Gln 849 mutants. For both, Asn 845 and Gln 849 , the greater effect was noted with conserved substitutions suggesting that not only the chemical nature of the side chain but also the size of the side chain at a specific position is critical for KF to function optimally. The two His 881 mutants did not show significant change in K m.dNTP ; however, the k cat of both enzymes (H881A and H881V) was significantly (ϳ65-fold) reduced with dA 36 / dT 18 template-primer. This effect was not as pronounced with dC 60 /dG 18 template-primer, as only a moderate (ϳ6-fold) change was noted in the catalytic rate of both His 881 mutants.
The mutants of Gln 677 (Q677A and Q677N) also showed a defect in their catalytic efficiency. The efficiency of the Q677A mutant was reduced by ϳ10and 120-fold on dA 36 /dT 18 and dC 60 /dG 18 template-primers, respectively. As noted for Asn 845 and Gln 849 , the efficiency of the homologous mutant of Gln 677 (Q677N) was significantly more reduced than that of its nonhomologous counterpart (Q677A). The catalytic efficiency was reduced by nearly 3000-fold on dA 36 /dT 18 and by ϳ300-fold with dC 60 /dG 18 template-primer. Similarly, a reduction of about 300-fold in the catalytic efficiency of the R668A mutant enzyme was noted with dA 36 /dT 18 template-primer. Similar (type of) template-specific inactivation was also noted with the mutants of O-helix residues (18). These steady-state kinetic data revealed that most of the mutant enzymes have a compromised DNA polymerization efficiency, suggesting some defect in the mutant of enzymes at one or more steps of the catalytic mechanism.
Ability of Mutant Enzymes to Form E⅐TP Binary Complexes-Because the mutants of Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 amino acid residues of the KF displayed reduced polymerase activity, we examined if this decrease was because of a defect in template-primer binding ability of mutant proteins. To determine the binding affinity of various mutant enzymes with the template-primer, we carried out gel mobility shift assay of the wild-type and mutant enzymes with 33/20 ϩ ddC and 21/12 ϩ ddC, two template-primers that differ in duplex length (Chart 1). The primer moiety of 33/20 ϩ ddC template-primer was 5Ј-end 32 P-radiolabeled, whereas in the 21/12 ϩ ddC template-primer, the template strand was radiolabeled. The individual enzyme-TP complexes were resolved on 6% nondenaturing polyacrylamide gels. The representative patterns of the migration of enzyme-TP complexes for each of the two template-primers (33/20 ϩ ddC and 21/12 ϩ ddC) are  Fig. 2) could bind ϳ55% of the 33/20 ϩ ddC template-primer. These data suggest that the affinity of wild-type KF for a template-primer with a short duplex region (21/12 ϩ ddC with only 13 base pairs) is greater than that for a template-primer containing a longer duplex region (33/20 ϩ ddC with 21 base pairs). The K D.DNA values for 33/20 ϩ ddC and 21/12 ϩ ddC template-primers were 0.60 and 0.14 nM, respectively, calculated by plotting the concentration of enzyme against that of complexed template-primer and fitting the data to the hyperbolic curve shown in Fig. 2, panels C and D. Another interesting observation pertaining to the binding of KF with two template-primers is the gel mobility pattern of the enzyme-TP complex, particularly at high enzyme concentrations. It appears that with increasing concentrations of KF, a regular 1:1 enzyme-TP complex forms first, which is then shifted to a slower migrating species (supershift position) (Fig.  2, panel A). This kind of supershift is not seen with the use of 21/12 ϩ ddC template-primer ( Fig. 2, panel B). Note that the last two lanes in each panel show the pattern obtained with 75 and 150 nM enzyme concentration. Thus it appears that the template-primer containing 21 base pairs may support sequential binding of two enzyme molecules, albeit with different affinity.
The template-primer binding affinity of mutant enzymes of Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 was determined in a similar fashion as described above for the wild-type KF. A representative profile of DNA binding by R668A and Q849N with two template-primers is shown in Fig. 3. The estimated K D.DNA values for various mutant enzymes with 33/20 ϩ ddC template-primer are listed in Table II. It is clear from this table that the template-primer binding affinity of all mutant proteins was significantly decreased. The K D.DNA difference varied in the range of 15-75-fold with this template-primer. The difference in K D.DNA was further increased when measured with 21/12 ϩ ddC template-primer. For example, the K D.DNA for the R668A mutant with this template-primer was ϳ100-fold greater than that for wild-type enzyme. Curiously, the migra-  Fig. 3. It is clear from the results that both R668A and Q849N mutant enzymes fail to form a distinct species resembling the enzyme-TP complex, although the amount of free TP is continually decreasing as the concentration of the protein is increased. A shade of gray above the free template-primer for R668A mutant enzyme (possibly a dissociated complex) is nonetheless seen at higher concentrations of the enzyme. This shade of gray, which is not so obvious for the Q849N mutant enzyme, suggests that there are further differences among different mutants with regards to the binding of the shorter duplex containing template-primer. Panel C of Fig. 4 shows curves representing the binding affinity of R668A for 33/20 ϩ ddC and 21/12 ϩ ddC template-primers and Q849N for 33/20 ϩ ddC. The K D.DNA values of the R668A mutant for two templateprimers (33/20 ϩ ddC and 21/12 ϩ ddC), estimated from these plots, are 14 and 33 nM, respectively. The K D.DNA of the Q849N mutant enzyme with 33/20 ϩ ddC template-primer is 31 nM. The affinity of Q849N for 21/12 ϩ ddC template-primer appears very low and could not be calculated from these data. These results strongly suggest that there are two binding modes of KF to template-primer and that the mutants of Hbonding track residues bind only in the supershifted mode.
Effect of Correct and Incorrect dNTPs on E-dNTP Binary Complex-The other means that we employed to investigate   (20). This assay has been successfully used to demonstrate the ability or inability of mutant enzymes to form a stable ternary complex, as judged by In contrast, no supershifted species is formed when 21/12 ϩ ddC is used at any enzyme concentration (panel B). In panels C and D, the amount of enzyme-TP complex has been plotted as a function of increasing concentrations of the wild-type KF. The percent enzyme-TP complexes were calculated by subtracting the free template-primer in the individual reaction from the total input value. The data were fit to a hyperbolic function by nonlinear regression for the determination of K D.DNA . the effect of the trap on the bound template-primer (15,19). The results of this investigation for the wild-type and three mutant enzymes are shown in Fig. 5. The lane marked input represents the mobility of the template-primer alone. The next four consecutive lanes marked "enzyme ϩ TP," "enzyme ϩ TP ϩ trap," "enzyme ϩ TP ϩ dNTP," and "enzyme ϩ TP ϩ dNTP ϩ trap" show: (i) the binding of template-primer alone, (ii) the effect of the addition of ϳ500-fold nonradioactive template-primer (trap), (iii) the putative ternary complex in the presence of complementary incoming dNTP, and (iv) the susceptibility of the ternary complex to trap, respectively. It is clear from the figure that enzyme-TP complex between the wild-type (or mutant species) is readily competed out by the addition of the trap template-primer (see lanes marked Enzyme ϩ TP ϩ Trap and compare with Enzyme ϩ TP lanes). However, the complex between wild-type KF and template-primer ϩ dNTP could not be competed out by the same (trap) templateprimer (see lane marked Enzyme ϩ TP ϩ dNTP ϩ Trap for WT enzyme). For the mutant enzymes, the enzyme ϩ TP complex in the presence of dNTP was competed out by the nonradioactive template-primer trap suggesting that the addition of dNTP to the enzyme-TP complex failed to form a stable ternary com-plex. A small amount (ϳ5% of wild-type) of stable ternary complex was noted for H881A. This is consistent with some catalytic activity seen with His 881 mutants of KF.
UV-mediated Cross-linking of Enzyme-TP and the Addition of Complementary dNTP in Situ by E-TP Complex-The data presented above suggest that the mutant enzymes of Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 are generally defective in template-primer binding and that the little binding that occurs is probably in the nonpolymerase mode. To further assess if the small extent of binding of template-primer to various mutant enzymes is in the polymerase mode, we used UV-mediated photocross-linking of enzyme to template-primer and examined the ability of individual enzyme-TP complexes for their ability to incorporate the first complementary nucleotide. It has been shown previously that the Klenow fragment of E. coli DNA polymerase I, covalently cross-linked to template-primer, is capable of adding one nucleotide onto the photocross-linked template-primer (18,22). When covalently cross-linked to the template-primer, translocation ability of the enzyme along the template-primer is compromised. Under these conditions, the nucleotidyl transfer reaction is restricted to the incorporation of the first incoming nucleotide, provided that the 3Ј-OH of R668A and Q849N) for template-primer was determined in a manner similar to that used for wild-type enzyme (Fig. 2). Whereas the template-primer concentration was maintained at ϳ100 pM, much higher quantities of mutant enzymes were required, because of their decreased and variable affinity for template-primer. The panels A and B show the binding of 33/20 ϩ ddC and 21/12 ϩ ddC template-primers, respectively, to the indicated enzymes. The concentrations of both R668A and Q849N mutant enzymes varied between 4.6 and 153.6 nM. The quantitation of enzyme-TP complex formation with the mutant enzymes was carried out based on the quantity of the uncomplexed template-primer, because no clear-cut band migrating at the enzyme-TP position can be detected. For example, in panel A, the binding of R668A to 33/20 ϩ ddC shows a progressive decrease in the free template-primer with increasing concentration of R668A. However, in panel B, with the same mutant enzyme and 21/12 ϩ ddC template-primer, a progressive decrease in free template-primer is accompanied by a "smudge" above the free template-primer position. We have considered the smudge as a part of enzyme-TP complex, although the strength and stability of this binary complex is certainly compromised. In all cases, the quantitation of the area corresponding to free TP position has been used to draw the curves shown in panel C. Because of the extremely low affinity of Q849N for 21/12 ϩ ddC template-primer, the binding curve could not be obtained. The K D.DNA was estimated by fitting the data to a hyperbolic function by nonlinear regression. To demonstrate that the mutants bind the long duplex containing template-primer only in the supershifted mode, two wild-type enzyme concentrations (lanes 8 and 9) were included along with R668A. Three concentrations of wild-type (lanes 8 -10) enzyme were included to show the same phenomenon with Q849N. The concentrations of wild-type corresponding to lanes 8 and 9 of the left two figures was 4.8 and 75 nM, respectively. The concentrations of wild-type enzyme for lanes 8 -10 was 3.2, 7.2, and 25 nM, respectively. The comparison of the migration of enzyme-TP complexes with mutant enzymes and wild-type shows that the mutant enzymes bind 33/20 ϩ ddC only in supershifted form. This observation was noted for all mutants of the H-bonding track.

FIG. 3. The template-primer binding affinity and its mode of binding with H-bonding track mutant enzymes. The affinity of H-bonding track mutant enzymes (represented by
the primer is correctly positioned in the complex for an in-line attack on the ␣-phosphate of incoming dNTP. In general, a larger quantity of mutant protein was required to obtain approximately equal extent of cross-linking as that of wild-type KF. To determine the efficiency of enzyme-template-primer cross-linking by mutants compared with wild-type KF, the radioactive template-primers were used to produce enzyme-TP cross-link adducts. The quantitation of these adducts with varying concentration of individual mutant enzymes (data not shown) provided the guidance for the generation of approximately equal quantity of enzyme-TP complexes for all mutant enzymes. For the dNTP addition reaction, all mutant enzymes as well as the wild-type KF were covalently cross-linked to nonradioactive 32/19-mer template-primer and addition of [␣-32 P]dCTP (the first incoming substrate dNTP for this template-primer) was assessed. The in situ catalytic addition of dCTP was performed in 1 M NaCl to ensure that the addition reaction would be restricted only to the cross-linked enzymetemplate-primer species. The results are shown in Fig. 6. A comparison of the intensity of radiolabeled band (representing the addition of radiolabeled nucleotide) shows that none of the mutant enzymes were able to incorporate dNTP onto the crosslinked template-primer to any significant extent (Fig. 6). Compared with the wild-type enzyme, less than 2% incorporation was seen with N845A, R668A, and H881A mutant enzymes. No visible incorporation was noted for Q849A and Q677A enzymes. These results strongly suggest that the 3Ј-OH of the templateprimer cross-linked or bound to mutant proteins was not properly placed at the active site for catalysis to occur. DISCUSSION The reaction mechanism of DNA polymerases has been well studied (27)(28)(29). In addition, the crystal structures of several template-primer and/or dNTP-bound DNA polymerases have provided significant insight into the understanding of the DNA replication mechanisms offered by the nucleic acid polymerases (4 -7, 11, 24, 30, 31). The first step in the DNA replication kinetic pathway is the binding of enzyme with TP. The com- FIG. 4. The effect of correct dNTP and three incorrect dNTPs on the binding affinity of wild-type and mutant enzymes. This figure shows the effects of incoming complementary dNTP (dGTP for both template-primers) and three incorrect dNTPs (dATP, dCTP, and dTTP) on the affinity of wild-type (panels A and B), R668A (panels C and D), and Q849N (panels E and F) for two template-primers. Panels A, C, and E represent the migration of the complexes of 33/20 ϩ ddC with wild-type, R668A, and Q849N mutant enzymes, respectively. Similarly, panels B, D, and F represent the binding characteristics of 21/12 ϩ ddC with wild-type, R668A, and Q849N mutant enzymes, respectively. The effect of dNTPs on template-primer binding by the wild-type and mutant enzymes was assessed using preformed enzyme-TP binary complexes. To one group of the enzyme-TP complexes, 200 M dGTP was added, whereas the duplicate received three incorrect dNTPs (100 M each). The concentration of the wild-type (0.3-10 nM in panel A and 0.04 -1.24 nM in panel B) and mutant enzymes (4.8 -153.6 nM) was the same for lanes marked TP only, TP ϩ dGTP, and TP ϩ 3dNTP-dGTP. The comparison of lanes marked TP only with TP ϩ dGTP and/or TP ϩ 3dNTP Ϫ dGTP in panels A and B shows that the presence of correct incoming substrate enhances the affinity of template-primer to wild-type enzyme for both template-primers. In contrast, the presence of 3 noncomplementary dNTPs inhibits the template-primer binding to the wild-type enzyme. It is interesting to note that such effects are not seen for the mutant enzymes, suggesting a difference in the template-primer binding characteristics of mutant enzymes, compared with wild-type enzyme. In panels C-F, the lanes marked WT represent the wild-type enzyme, which was included as control.
parison of the crystal structures of KlenTaq and B. stearothermophilus (Bst) DNA polymerases in apo (Protein Data Bank files 1ktq and 1xwl) and DNA bound forms (Protein Data Bank files 4ktq and 2bdp), suggests that the thumb domain of these polymerases undergoes a conformational change to accommodate the double stranded region of the template-primer. As a result of this conformational change, the thumb undergoes two changes: (i) a 12°opening followed by 7°movement toward the cleft, and (ii) the transition of J-helix to a coiled structure. In addition to these large conformational changes, the binding of template-primer also induces new intramolecular network(s) involving the side chains of several amino acid residues in its vicinity. Most prominent among these interactions is a cluster of five invariant amino acid residues interacting via H-bonding in almost a straight line fashion. The involved residues are the KF equivalents of Asn 845 , Gln 849 , Arg 668 , His 881 , and Gln 677 (Fig. 1). The distance between the C␣ atoms of residues at either end of the straight line (Gln 677 at one end and Asn 845 at the other) is ϳ18 Å and hence we have described it as an ϳ18-Å long H-bonding track. The binding of the substrate dNTP to enzyme-TP complex causes a major conformational change in the fingers domain of pol I (5,11). The process of enzyme-TP-dNTP complex formation is popularly referred to as "fingers closing" or ''prepolymerase" ternary complex formation (32). Interestingly, the straight line H-bonding network of interactions remains unaltered in the ternary complex. To understand the function of this interaction network, we employed sitedirected mutagenesis to generate mutant derivatives of all five residues and biochemically characterized individual mutant enzymes for its polymerase activity, steady-state kinetic properties, template-primer binding affinity, and the ability to incorporate dNTP substrate onto the primer terminus in situ.
Architecture of the H-binding Track-In the crystal structure of KF and analogous DNA polymerases, the Arg 668 equivalent residue is at the center of the H-bonding track and is located on ␤8. This residue is a part of the highly conserved TGR motif found in the polymerase I family of enzymes (33). In the template-primer bound crystal structures of KlenTaq, the Arg 668equivalent residue (Arg 573 ) is flanked by Gln 754 (Gln 849 in KF), Gln 849 and His 786 (His 881 in KF). The two distal residues are the equivalents of Asn 845 and Gln 677 (Fig. 1). In the binary complex of KlenTaq, the amino acid residue equivalent to Arg 668 of KF forms a hydrogen bond with Gln 849 (3.44 Å), which in turn forms a hydrogen bond with Asn 845 (3.69 Å). On the other side, Arg 668 has van der Waals interactions with His 881 . The distance between two of the nearest atoms of Arg 668 (N) and His 881 (C␦2) is 3.38 Å. There also exists a possibility of a H-bond between atom N2 of Arg 668 and N␦1 of His 881 . However, the distance between the two H-bond forming atoms (4.16 Å) is slightly unfavorable. Furthermore, His 881 forms a hydrogen bond with Gln 677 (3.33 Å). There are also two intermolecular hydrogen bonds mediated by Arg 668 in the enzyme-TP binary complex (11). One is with the primer base (2.39 Å) and the other is with template nucleotide (3.8 Å) base. Similarly,

FIG. 5. Ability of wild-type KF and H-bonding track mutants to form a stable ternary complex.
The ability of the wild-type and three mutant enzymes, H881A, N845A, and Q677A, to form stable ternary complexes in the presence of cognate incoming dNTP was assessed following the protocol described under "Experimental Procedures." The sensitivity of radiolabeled template-primer (33/20 ϩ ddC) in the desired enzyme-TP complexes to the addition of 500-fold excess unlabeled DNA, (which completely dissociates the radiolabeled DNA) provided an indication of ternary complex formation. The radiolabeled template-primer in the complex formed by the wild-type enzyme in the presence of cognate nucleotide (lane labeled Enzyme ϩ TP ϩ dNTP) is not competed out by addition of excess unlabeled DNA (shown in lane Enzyme ϩ TP ϩ dNTP ϩ Trap). In contrast, except for the H881A enzyme, which forms ϳ5% stable ternary complex compared with the wild-type KF, all other mutant enzymes failed to form stable ternary complexes. Note that the addition of excess nonradiolabeled template-primer readily competes out the bound template-primer in the binary as well as ternary complexes for all mutant enzymes (lanes marked enzyme ϩ TP ϩ Trap). The concentration of wild-type, H881A, N845A, and Q677A enzymes was 1. Gln 849 and Asn 845 form hydrogen bonds with template strand. Gln 849 hydrogen bonds with the base moiety of the template nucleotide (3.15 Å) paired with the base at the primer terminus and Asn 845 forms a hydrogen bond with the sugar oxygen of the same template base. Amino acid residue Gln 677 is positioned between the 3rd and 4th base pairs from the primer terminus and has potential to form a hydrogen bond with the base moiety of 3rd primer base in the double strand region of the templateprimer. In the binary complex crystal structures of KlenTaq (11) and Bst polymerase (24), Gln 677 equivalent residues do not form a hydrogen bond through their side chains. However, the possibility of a water-mediated hydrogen bond with the base moieties of the 3rd and 4th base pairs in the double stranded region from the primer end cannot be ruled out as the side chain of Gln 677 equivalent residues is positioned within the minor groove of the template-primer. A possibility of hydrogen bond between Gln 677 and base moiety of the primer strand can also be extrapolated from the ternary complex crystal structure of T7 DNA polymerase. Gln 439 in the J-helix region of T7 DNA polymerase, which can be considered equivalent to Gln 677 of KF, forms a hydrogen bond with base moiety of the 3rd nucleotide from the primer end (5).
From the detailed analysis of the properties of mutant species of various residues, it is clear that the substitution of the H-bonding side chain in any of the five residues disrupts the track; however, the degree of disturbance is probably different, because the catalytic activity of the individual mutant protein shows variation (Table I). We shall first discuss the participation of individual members and its mutant derivatives deduced from the structural and the experimental results.
Properties and Environment of Asn 845 and Gln 849 -In KF, the sequence 845 NAPMQG 850 contains two highly conserved uncharged polar residues common to all members of the polymerase I family of enzymes. These are Asn 845 and Gln 849 . Previous studies have suggested that both Asn 845 and Gln 849 play some role in catalysis of the polymerase reaction (13,14). Our kinetic characterization shows that the K m.dNTP for N845A and N845Q did not change significantly. However, a significant change was noted in k cat values for the two Asn 845 mutant enzyme species. It is rather curious that the N845Q mutant enzyme was significantly more affected compared with N845A. The loss of catalytic activity of the Gln 849 mutant derivatives (Q849A and Q849N) was more severe suggesting that the side chain of Gln 849 participates directly or indirectly in the catalytic process. Previously, Polesky et al. (13) have suggested that Gln 849 interacts with the DNA primer terminus and has some role in dNTP turnover, because Q849A and Q849E exhibited deficiency in the turnover of ␣-thio-dNTPs. The latter observation also indicates function of Gln 849 at the chemical step of the reaction. However, the crystal structures of KlenTaq (Klenow fragment of T. aquaticus DNA polymerase I) in three different forms (binary, open ternary, and closed ternary complexes) (11) as well as the crystal structure of T7 DNA polymerase (5) show that the Gln 849 equivalent glutamine interacts with template nucleotides and it does not have any interaction with dNTP. Therefore, the suggested participation of Gln 849 in the chemical step (13) is most likely a consequence of a loss in the templateprimer binding affinity and/or altered position of 3Ј OH relative to the active site.
Properties and Environment of Arg 668 -Extensive biochemical characterization of R668A has been reported previously by Polesky et al. (13). The results showed a moderate change in K m.dNTP (ϳ2-fold). Significant changes in K D.DNA (ϳ20-fold) and k cat (ϳ400-fold) were noted for the R668A mutant enzyme. Our estimations of K m.dNTP , K D.DNA , and k cat for the same mutant enzyme show similar changes. In fact, the K D.DNA was increased by ϳ25and ϳ160-fold with 33/20 ϩ ddC and 21/12 ϩ ddC, respectively. The k cat was reduced by ϳ350-fold with dA 36 /dT 18 template-primer. These data suggest that the defect in the catalytic ability of the R668A mutant was predominately because of the defect in the binding of the template-primer. The interaction of Arg 668 seen in the crystal structure complexes with template-primer supports the decreased affinity of the R668A mutant enzyme. The side chain position of Arg 668 in ternary complex crystal structures is favorable for a hydrogen bond with the sugar moiety of dNTP. However, the major participants in the binding of dNTP to enzyme-TP complex appear to be the members of the O-helix and metal ligated Asp 705 and Asp 882 . Therefore, the weak hydrogen bond between Arg 668 and the sugar oxygen of dNTP may not be contributing significantly to binding and stabilization of dNTP at the active site.
The role of an amino acid residue at the chemical step of nucleotide incorporation by DNA polymerases has been judged by the difference in utilization of normal versus ␣-thio-substituted dNTPs (32). This is called an elemental effect. R668A mutant protein has been shown to utilize dNTP␣S about 10fold less efficiently than the wild-type enzyme, suggesting the involvement of Arg 668 at the chemical step. Arg 668 has been seen to form the ion pairs with Glu 710 , which has been proposed to ligate a metal ion coordinating with the 3Ј-OH of dNTP (34). Thus the negative effect of R668A on the chemical step may be because of the loss of its reactivity with Glu 710 , and the elemental effect seen for the R668A mutant may be the reflection of some indirect effect mediated through Glu 710 .
Properties and Environment of His 881 -The biochemical characterization of His 881 shows that the H881A mutant is ϳ30% as active as WT. There is no significant change in K m.dNTP , whereas a 6 -66-fold decrease in k cat with different template-primers was noted, suggesting the requirement of His 881 for optimal catalysis by KF. The location of His 881 is on a ␤-turn between ␤12 and ␤13 in the palm subdomain. It is highly conserved among pol I type DNA polymerases. In the crystal structure of T7 and TaqDNA polymerases, the equivalent residues interact with the sugar moiety of the primer terminus nucleotide (5,11). Therefore, this residue has been proposed to position the primer terminus for catalytic activity (5). Comparison of the apo and binary complex of TaqDNA polymerase crystal structures shows that the turn between ␤12 and ␤13 undergoes a "springboard-like" conformational change upon DNA binding, which does not change further upon ternary complex formation. The C␣ distance between two His 881 conformations (in apo-KlenTaq and template-bound KlenTaq) is ϳ3 Å. Superposition of apo onto the binary complex shows that if such a sharp springboard-like conformational change of the turn between ␤12 and ␤13 were not to occur, His 881 would interfere with the sugar moiety of the primer terminus nucleotide. Analyses of these crystal structures suggest that two additional residues probably contribute to induce a springboard-like conformational change. These residues are Gln 677 and Gln 879 . In the binary and/or ternary complex, His 881 side chain forms a hydrogen bond with Gln 677 , whereas the main chain carbonyl forms a hydrogen bond with Gln 879 . In addition, Gln 879 forms a hydrogen bond with Gln 677 . It appears that the triangular H-bond arrangement at this site may be responsible for the conformational change in the ␤-turn, which can then accommodate the template-primer terminus at a proper position.
Properties and Environment of Gln 677 -Gln 677 belongs to the J-helix in pol I-type DNA polymerases. This helix is formed by five amino acid residues flanked by proline on either ends. The sequence of this region is highly conserved. The J-helix under-goes a conformational change (from helix to coil) upon TP binding (5,11). However, the helical structure is maintained when the binding of the template-primer occurs in the 3Ј-5Ј exonuclease mode (9). We have previously shown that helix to coil transition (or vice versa) of the J-helix plays a crucial role in the switching of the template-primer from the polymerase to the exonuclease site, as judged from the properties of P680G (35). Upon the binding of template-primer to the enzyme in the polymerase mode, repositioning of the side chain of Gln 677 has also been noted. This in turn allows Gln 677 to make contacts with His 881 and Gln 879 . The mutant Q677A enzyme is an inactive enzyme (35,36). Our activity and steady-state kinetic data also show severe loss in the polymerase activity of both nonhomologous Q677A and homologous Q677N mutant enzymes (Table II). In fact, the polymerase activity of the P680G mutant of KF was also reduced to the same extent as that seen for Gln 677 mutant enzymes (35). Therefore, it is possible that the reduced activity of P680G was a reflection of the activity of the enzyme because of the altered position of Gln 677 . A change in the position of Gln 677 would result in the destabilization of the H-bonding track arrangement, which in turn would not permit the binding of enzyme to template-primer. Whereas the polymerase activity of the P680G mutant enzyme was severely reduced, the mutant enzyme was extremely efficient in the 3Ј-5Ј proofreading activity (35). We have noted a similar increase in the efficiency of 3Ј-5Ј exonuclease activity with the Q677A mutant species. 2 Thus, the comparison between the activity and steady-state kinetic constants of the wild-type and mutant enzymes (Table  II) suggests that all of the H-bonding track residues participate in the catalytic function of Klenow fragment of E. coli DNA polymerase I. The major participation of H-bonding track residues is in the process of the template-primer binding, as judged by the significant decrease in K D.DNA (Table II). These observations are consistent with the crystal structure data of the binary and ternary complexes of KlenTaq, which show that these residues interact with the template-primer (11). The K D.DNA for some of the mutant enzymes studied here was also determined by others (13). They noted that the N845A mutant showed no change in template-primer binding affinity compared with the wild-type KF. However, the other mutant of Asn 845 (N845D) had a 15-fold decrease in template-primer binding affinity. The K D.DNA for H881A was altered by only 2.5-fold, whereas this value for Q849A and R668A mutants was reported to be ϳ10and 17-fold greater, respectively (13). The change in K D.DNA for N845A, Q849A, R668A, H881A, and Q677A with 33/20 ϩ ddC, in our study, is 16-, 43-, 22-, 25-, and 70-fold, respectively. Thus, the K D.DNA values determined here are significantly different from those reported by Polesky et al. (13). The differences in K D.DNA values may be attributed to two different methods and the template-primers used in two studies. In fact, we have noted that K D.DNA values vary significantly, based upon the method of determination as well as the concentration of template-primer (19).
The structural data on enzyme-DNA binary complexes of KlenTaq and B. stearothermophilus Klenow fragment and the ternary complexes of KlenTaq and T7 DNA polymerase I show that 8 -10 base pairs of the double stranded DNA are occluded by these enzyme proteins. The template-primers in these crystal structure complexes have between 30 and 45 interactions. A majority of these interactions are in the double stranded region of the template-primer. Despite the large number of interactions, the interactions provided by the H-bonding track residues appear to be the most significant for retaining the bound template-primer. The fact that mutation at any one of the five residues produces significant loss in binding affinity for TP further suggests that these residues operate in concert and that the formation of H-bonding track concurrent with TP binding in the active center may provide an appropriate geometrical conformation for loading of the correct dNTP substrate.

Template-primer Binding by the Mutant Species of H-bonding Track
Residues-Another interesting feature of all the mutant enzymes that we noted is that they are not only defective in template-primer binding, but quality of their binding is also different from that seen with the wild-type enzyme. For example, they bind the longer duplex containing template-primer in a supershifted mode (Figs. 2 and 3). It has been noted previously that increasing the concentration of KF results in the spontaneous shift in the retardation pattern of the enzyme-TP complex species from a regular to supershift position (15,25,(37)(38)(39). It has been suggested that the supershifted species of enzyme-TP complex corresponds to the binding of two or more enzymes to the same template-primer. The DNA footprinting and crystallographic data on KlenTaq, T7 DNA polymerase, and B. stearothermophilus have implicated that the Klenow fragment occludes 8 -10 base pairs (5,11,24) of the templateprimer. Furthermore, these mutant enzymes appear unable to stably bind the template-primer containing a shorter (13 base pair) double stranded region (Figs. 3 and 4).
These observations suggest that the duplex region in the 33/20 ϩ ddC template-primer, which is not occluded by the enzyme, can support the binding of another KF molecule to produce a supershifted position of migration. A plausible way that the second molecule of KF can bind the unoccupied region of duplex is its binding at the blunt end of the duplex with significantly lower affinity, thus requiring a higher quantity of enzyme protein. Obviously, this situation is not possible for the shorter duplex containing template-primer, for its length of the double stranded region may not be enough to permit the binding of a second molecule. The relatively lower affinity of the wild-type KF for 33/20 ϩ ddC compared with 21/12 ϩ ddC template-primer is therefore likely to be because of the presence of two sites in the former template-primer compared with only one site in the latter. The binding of two (mutant) enzymes to the longer duplex containing template-primer could conceivably occur in an "exit mode" because of their inability to bind at the 3Ј terminus flanked by a single stranded template overhang at the appropriate orientation and position. This might permit the binding of two mutant enzymes in a nonproductive manner (back-to-back binding), using both termini of the template-primer. The wild-type enzyme, in contrast, would bind the bona fide 3Ј terminus with high affinity and the nonproductive 3Ј terminus (of the same template-primer) with low affinity. The capability of Taq polymerase to bind a duplex DNA in the latter mode is well documented by Eom et al. (31). In the crystal structure, the authors have shown that the 3Ј-OH of the duplex is bound close to the active site and that this type of binding has been called the exit mode binding.
The observations on the effect of dNTP addition on the stability of the enzyme-TP binary complex also supports the notion that the binding of the template-primer to mutant enzymes is of a different nature than that seen with wild-type. For example, the addition of the next incoming dNTP complementary to the template nucleotide increases the affinity of the wild-type enzyme for template-primer, as judged by a ϳ3-4fold decrease in the K D.DNA . In contrast, the addition of noncomplementary dNTPs significantly reduces the binding affinity of the wild-type enzyme for the template-primer (Fig. 4,  panels A and B). All of the mutant enzyme-TP complexes re-mained unaltered by the addition of the next incoming dNTP or by the addition of 3 noncomplementary dNTPs (Fig. 4, panels C-F). Similarly, none of the H-bonding track mutants could form a pre-polymerase ternary complex when an appropriate dNTP was added to the enzyme-TP complex (Fig. 5). The pattern of trap effect with or without dNTP remained unchanged for all of the mutant enzymes, except for H881A. The mutant H881A did exhibit a small but consistent ability to produce a trap-resistant ternary complex as judged by the "shade of gray" seen in Fig. 5 (lane marked enzyme ϩ TP ϩ dNTP ϩ Trap for H881A mutant).
A final support for the interpretation that mutant enzymes bind TP in a nonproductive mode is obtained by examination of the catalytic competence of the enzyme-TP covalent complexes of individual mutant enzymes. All of the H-bonding track mutant proteins exhibited nearly complete inability to catalyze in situ addition of a single dNTP onto the template-primer in the enzyme-TP complex (Fig. 6). Even the addition of dNTP by the H881A mutant was only ϳ2% of the wild-type enzyme level. These results reinforce the suggestion that the participation of the H-bonding track members is required for the productive binding of TP within the active site of KF.
H-bonding Track Residues Are Spanned Across the Known Functional Motifs of pol I-The compilation and alignment of the amino acid sequence of many DNA polymerases have shown that certain motifs (motifs A-E) are conserved in different members of the DNA polymerase family (1-3, 40, 41). Two of these motifs (A and C) are conserved across all known DNA polymerases. In the pol I family of DNA polymerases, five sets of conserved amino acid sequences were identified (1). These conserved sequences were numbered as regions 1-5. The conserved regions 3-5 were designated as motifs A-C, respectively (1). In the pol I class of enzymes, the catalytically essential aspartates (Asp 705 and Asp 882 ) are present in motifs A and C, whereas motif B contains 4 catalytically important residues of the O-helix. Region 1 is part of the thumb subdomain and appears to be somewhat isolated from other motifs. A part of region 2 contains the conserved sequence TGR; the Arg residue in this sequence is Arg 668 of KF. Interestingly, this motif was later specified as the (T/D)XXGR motif that occurs in T7 RNA polymerase (33). In addition to these identified conserved regions, it appears that three additional conserved motifs could be added to this list based on their functional importance in the pol I family. These motifs are represented by (i) the J-helix region, (ii) the hinge region between M and N helices, and (iii) a part of the Q-helix containing the NXXXQG sequence. We would like to suggest that the sequence PXXQXXG (corresponding J-helix) be defined as J-motif, where P is proline, Q is glutamine, and X is a variable residue. The second motif represented by GXDXH, which was previously recognized as the GXD hinge region in KF and Mycobacterium tuberculosis (15) pol I, may now be called the H-motif. The importance of the third motif represented by the NXXXQG sequence was also recognized by Loeb and co-workers (42) and they labeled it as region 6. We suggest that because of the functional importance of both Asn and Gln in this sequence, it be called the NQ motif. These three motifs and their conservation in the polymerase I family of enzymes are shown in Fig. 7A. Thus, there are now at least 8 functionally important conserved regions in the pol I family of enzymes. The location of these conserved motifs also provides some insight into their functional importance, as most of these residues are located surrounding the active site in the polymerase domain (Fig. 7C). Using the biochemical data presented here and those reported previously, coupled with structural data on KF and related enzymes, we find that during the catalytic process, all of the conserved motifs interact with each other to effect the optimal binding of substrates leading to efficient chemical reaction. The scheme of their interactions is shown in Fig. 7B. The center of the interaction is at Arg 668 . During the template-primer binding, the conformational change is accompanied by interactions among motifs TGR, A, C, J, and NQ. This is followed by the binding of dNTP, which induces the conformational change mediated by Asp 732 of Hmotif (15) such that a new interaction between motifs A and B is established to form a closed pre-polymerase ternary complex (6,11).
Does H-bonding Track Provide Flexibility to the 3Ј-5Ј Exonuclease Domain-Another unique aspect of pol I-type enzymes is the presence of a 3Ј-5Ј exonuclease domain, which operates at some 30 -35 Å from the polymerase active site. This domain is conserved even among the members (e.g. T. aquaticus pol I and M. tuberculosis polymerase I) that do not exhibit catalytic activity through this center. Nevertheless, in the pol I family, the importance of the 3Ј-5Ј exonuclease domain is recognized by its proofreading activity (i.e. exonucleolytic removal of mismatched nucleotide). Of the two models for the proofreading function of pol I, one involves the shuttling of a primer strand containing terminal mismatched nucleotide from the polymerase site to the exonuclease site (43). The second mode suggests a dissociation of enzyme from the template-primer upon misincorporation, followed by rebinding of an enzyme to the mismatched template-primer, in the 3Ј-5Ј exonuclease mode (43). Regardless of which mode is operative, the binding of mismatched template-primer to enzyme probably requires part of the same binding track with some alterations, whereby part of the primer strand may be oriented toward the exonuclease site. Based on the crystal structure of the ternary complex of T7 DNA polymerase, Doublie et al. (5) have proposed that such an alteration in template-primer binding contact points may be required in directing the primer terminus to the exonuclease site and that residues equivalent to Arg 668 and Gln 849 of KF may be involved in this process. Similar shuttling between polymerase and exonuclease sites with similar changes in the contact points have been proposed for RB69 DNA polymerase by Franklin et al. (12). Earlier, we have reported that alteration in the J-helix region (Gln 677 is a member of this region) also triggers preferential binding of the primer strand to the exonuclease site (35). The fact that 3 of 5 members of the Hbonding track are also involved in the binding or shifting of template-primer to the exonuclease mode strongly suggests that the 18-Å long H-bonding track may be uniquely found in enzymes that contain both polymerase and exonuclease domains. Furthermore, the binding of primer moiety (containing mismatched terminal nucleotide) could conceivably occur via alteration in the H-bonding track. The RB69 DNA polymerase, a member of the pol ␣ family (40), has both polymerase and 3Ј-5Ј exonuclease activity domains and it also contains an Hbonding track, similar to pol I, which can be readily discerned from its ternary complex crystal structure (Fig. 7D). Conversely, such a track is absent in enzymes lacking the 3Ј-5Ј exonuclease domain, such as HIV-1 reverse transcriptase and mammalian DNA polymerase ␤. Therefore, it is tempting to speculate that an H-bonding track may have been involved to permit shuttling of primer from the polymerase to the exonuclease site.
In summary, we have identified an important structural feature in the form of an 18-Å long H-bonding track in the pol I family of enzymes. The track is formed by the concerted action of 5 residues upon the binding of template-primer in the active center of pol I. The mutation of any one of the five residues involved in H-bonding abolishes the binding of templateprimer, because the formation of the H-bonding track involves the interaction among of all its members, as well as with the sugar and base moieties of four terminal base pairs of templateprimer. We conclude that the formation of this track not only provides the stability to bound template-primer, but may be responsible for the reported changes in the conformation of DNA from the B to A-like form at the active center. Inherent flexibility in the H-bonding network may readily permit the displacement of the water molecules near the active site to effect the structural transition of bound template-primer. Such a change may be pre-requisite for the positioning of the primer, orientation of templating base or single stranded template strand to facilitate the binding of the appropriate dNTP. The presence of the H-bonding track seen in both enzyme-TP and enzyme-TP-dNTP ternary complexes strongly supports this contention. The stabilization of the template-primer via the H-bonding track interactions is also likely to permit easy sliding during synthetic reaction as opposed to conventional interactions involving backbone phosphate groups of template-primer. A detailed analysis of the role played by individual members of the H-bonding track is currently under investigation.