DNA-thumb interactions and processivity of T7 DNA polymerase in comparison to yeast polymerase eta.

The replicative polymerase of bacteriophage T7 is structurally and mechanistically well characterized. The crystal structure of T7 DNA polymerase or gene 5 protein complexed to its processivity factor, Escherichia coli thioredoxin, a primer-template, and a dideoxynucleotide reveals how this enzyme interacts with the 3'-end of the primer-template, but does not show how thioredoxin confers processivity to the polymerase. In the crystal structure highly conserved amino acids Asn(335) and Ser(338) of the thumb subdomain of T7 DNA polymerase are seen to interact with phosphates 7 and 8 of the DNA template strand. Results with a mutant T7 DNA polymerase in which aliphatic residues are substituted for these amino acids and experiments with different length and methylphosphonate-modified primer-templates demonstrate that these interactions are essential for processive synthesis and d(A.T)(n) tract bypass. Our data with methylphosphonate-modified DNA suggests that thioredoxin confers processivity to T7 DNA polymerase in part by causing an interaction with the phosphate backbone or minor groove of DNA. Residues Asn(335) and Ser(338) may also function with a nearby helix-loop-helix motif located at residues 339-372 to enclose the DNA during processive synthesis. Our results suggest that this structure must be held close to the DNA by ionic interactions to function. These interactions also allow for DNA sliding but physically block the passage of a 3T bulge in the template. In contrast, yeast polymerase eta, a polymerase that non-mutagenically repairs cis-syn thymidine dimers, allows the same bulge to slide past its thumb subdomain during synthesis. A relaxed thumb interaction with the DNA could account for the notably low processivity of polymerase eta.

An important biological process is the movement of proteins across DNA and RNA. For this to occur, the specific protein needs to maintain association with the polymer during translocation. Translocation by RNase II, an Escherichia coli 3Ј-5Ј exonuclease, is an example of a simple processive reaction. A protein "anchor" holds the downstream RNA while the catalytic site sequentially releases 5Ј-monophosphates from the 3Ј-end of the RNA (1,2). The more complex yeast RNA polymerase II stabilizes the elongation complex during catalytic site translocation at two positions. Jaws and clamp close on the down-stream DNA while the nascent RNA binds to the underneath side of the protein (3)(4)(5). To achieve highly processive synthesis, replicative polymerases employ accessory proteins to maintain association with the DNA. The beta subunit of E. coli DNA polymerase III and proliferating cell nuclear antigen of human and yeast polymerase ␦ form sliding clamps around the DNA (6,7). In contrast, the gap-filling DNA polymerase ␤ binds to the 5Ј-phosphate located at the far end of a gap through an 8-kDa domain to anchor the polymerase in processive synthesis (8 -10). How DNA polymerases bind to template primers is fundamental to eventually understanding how polymerases behave when they encounter DNA damage, such as DNA photoproducts.
The DNA polymerase of T7 bacteriophage (gene 5 protein) is a structurally and mechanistically well characterized enzyme that becomes highly processive upon binding to E. coli thioredoxin (11,12). The crystal structure of the T7 DNA polymerase complexed to E. coli thioredoxin, a 22-mer primer/26-mer template, and a dideoxynucleotide triphosphate has been solved (13). Only the first 11 bp of the primer-template, along with the templating nucleotide and its 5Ј-neighbor, could be assigned coordinates. The rest of the primer-template could not be seen due to disorder. The thioredoxin is seen to bind to an extended loop between ␣-helices H and H1 (262-338), but, because of the disorder in the crystal, residues 294 -318 cannot be located and it is thus not possible to determine how the thioredoxin confers processivity to T7 DNA polymerase. It has been suggested that the thioredoxin either enhances binding through electrostatic interactions or possibly swings across the primer template so as to encircle it (13) in a similar fashion to how proliferating cell nuclear antigen, the processivity factor for polymerase ␦, forms a ring around the DNA (6,14).
The DNA-protein interactions most remote from the active site of T7 DNA polymerase that can be located in the crystal structure are between highly conserved amino acids Asn 335 and Ser 338 (15), located between beta strand 6b and ␣-helix H1 of the polymerase thumb and template phosphates 7 and 8. These two amino acids are also in close proximity to a nearby helixloop-helix (HLH) 1 motif (amino acids 339 -372, domains H1-6c-6d-H2). Similarly, the Klenow editing complex contains 50 amino acid residues between helices H and I of the thumb subdomain, which includes Asn 579 and Ser 582 , which are close to template phosphates 7 and 8 and also a nearby HLH motif (16). A deletion of the entire loop of the HLH (amino acids 590 -613) limits processive synthesis and leads to frameshift mutations (17). The authors concluded that the deleted peptide somehow tracks the minor groove during processive synthesis. Consistent with this finding, the p66 helix H of the HIV-1 reverse transcriptase thumb subdomain has been linked to processivity, and the amino acid residues interacting with the DNA have been designated a "minor groove binding track" (MGBT). Mutations within the MGBT effect processivity and increase the possibility of frameshift mutations (18 -23). DNA mapping shows that HIV-1 reverse transcriptase protects the same number of nucleotides after limited synthesis (24). Based on this observation, it was suggested that the events in forming a single phosphodiesterase bond strip the MGBT from the DNA, allowing it to translocate one position upstream (25).
In contrast to T7 DNA polymerase, which belongs to the family of replicative polymerases, which have high processivity and fidelity, polymerase belongs to a group of DNA damage repair enzymes that have low processivity and fidelity and have been the subject of intensive study since its discovery (26). Despite its low processivity, polymerase can efficiently and non-mutagenically synthesize past cis-syn thymidine dimers, and a condition called xeroderma pigmentosum occurs when this activity is missing in humans (27). T7 DNA polymerase, on the other hand, bypasses the thymine dimer very inefficiently (28 -30) and only in the presence of the thioredoxin processivity factor and in the absence of its exonuclease activity. The crystal structure of the catalytic core of yeast pol has been solved (31) and contains a small thumb subdomain that consists of only 90 residues and alpha helices L, M, N, O, P, and Q, which bear no structural similarity to any polymerase in the pol A family of which T7 DNA polymerase is a member.
Herein we show that the T7 DNA polymerase thumb residues Asn 335 and Ser 338 interact with the primer-template stem and are critical for processive synthesis. These interactions may function with a nearby HLH motif to enclose the DNA. Our data suggest this structure must be drawn close to the DNA by ionic interactions to function. These interactions allow free sliding of the DNA during processive synthesis but physically block the passage of a 3T bulge. Our results also support the idea that thioredoxin confers processivity by enhancing the interactions between the phosphate backbone of the DNA, or possibly by causing the polymerase to track the minor groove of DNA, rather than by encircling the DNA. In contrast, the catalytic core of yeast pol allows easy passage of a 3T bulge past its thumb subdomain, which could be explained by a relaxed thumb-DNA interaction, which could also account for it's low processivity.

EXPERIMENTAL PROCEDURES
Materials-Thioredoxin crystals were from U. S. Biochemical Corp. and taken up in 50% glycerin with 20 mM Tris-HCl at pH 7.5 and stored at Ϫ20°C at 5 mg/ml. DNase I, in crystal form, was obtained from Sigma, and was resuspended and stored in 50% glycerin with 10 mM Tris-HCl, pH 7.5, and 10 mM MgCl 2 at Ϫ20°C at a concentration of 5 mg/ml. T4 polynucleotide kinase was from New England Biolabs.
Oligodeoxynucleotides-DNA oligonucleotides were purchased from IDT (Coralville, IA) or synthesized in our laboratory with phosphoramidites or methylphosphonamidites from Glen Research on an ABI 8909 synthesizer. The methylphosphonate containing primer-templates were synthesized and desalted according to the suppliers protocol. When necessary, the ODNs were purified by acrylamide gel electrophoresis, and full-length ODNs were detected by UV shadowing, eluted, purified from salts and acrylamide by two ethanol precipitations, and quantified by UV absorption.
DNA Polymerases-The exo Ϫ T7 DNA polymerase used in this study was the double mutant (D5A, E7A) described earlier (32). The gene 5 protein was overproduced in E. coli A179 cells containing two plasmids. One plasmid (pGp5-3) contained the T7 DNA polymerase (gene 5) preceded by the T7 RNA polymerase Ø10 promoter and a second plasmid (pGp1-3) contained the T7 RNA polymerase (gene 1) preceded by the pl promoter (12). The plasmid pGp5-3 contains a gene for ampicillin resistance, and pGp1-3 contains a gene for tetracycline resistance. The E. coli A179 strain contains a thioredoxin gene (trxA) inactivated by a kanamycin insert (33). One liter of cells was grown and heat-induced as previously described (32).
In some experiments, we used the same exo Ϫ T7 polymerase fused at the N-terminal end to the His 6 tag containing peptide MGSSHHHH-HHSSGLVPRGSH 2 and will be referred to as His-tagged T7 polymerase. The gene encoding the exo Ϫ T7 polymerase was obtained by PCR amplification from pGp5-3 with Pfu DNA polymerase (Stratagene) and cloned into a pET-28a vector (Novagen) through the NdeI and XhoI restriction sites. This plasmid was overproduced in cell line BL21(DE3)pLysS (Stratagene), which contains T7 RNA polymerase as a lysogen under control of the lac operator and a plasmid with genes for chloramphenicol resistance and lysozyme to inhibit low levels of background T7 RNA polymerase. Because the pET-28a vector carries the gene for kanamycin resistance, an overnight culture was grown in LB media with both kanamycin and chloramphenicol (each 40 g/ml). The next day cells were grown only with kanamycin before induction with 0.5 mM IPTG.
The catalytic core of yeast polymerase is an N-terminal His 6 -tagged fusion protein PCR-cloned from the yeast expression plasmid pEGUh6-RAD30 (34) into the pET-28a vector through the same NdeI and XhoI restriction sites used for the construction of the His 6 -tagged T7 polymerase. 3 The core polymerase contains residues 1-513 of the native enzyme, for which a crystal structure has been reported (31), fused to MGSSHHHHHHSSGLVPRGSH on the N-terminal end. The plasmid was overproduced in BL21-CodonPlus(DE3)-RIL cells (Stratagene), which contain extra copies of three tRNAs (argU, ileY, and leuW) necessary for optimal production of the polymerase. Cells were grown in LB media with kanamycin and induced with IPTG.
DNA Polymerase Isolation-Thioredoxin-free T7 polymerase was isolated from E. coli A179 cells was purified by standard column chromatography at 4°C with a 40-ϫ 2.5-cm Sepharose-200 column, a 20-ϫ 2.5-cm Q-Sepharose column, and a 10-ϫ 2.5-cm S-Sepharose column in that order (Sigma). A pH of 7.5 was maintained with 20 mM Tris-HCl throughout the purification except for the S-Sepharose column, for which a 10 mM Mes buffer, pH 5.8, was used. In our experience the polymerase appears to be susceptible to sulfhydryl rearrangements, which leads to inactive enzyme. To avoid this, high salt and reducing conditions were maintained throughout the purification. Cells were lysed in 0.5 M salt by sonication on ice (5 times 20 s). The sizing column was equilibrated and run with 0.5 M NaCl, and both ion exchange columns were equilibrated and loaded in 0.25 M NaCl. The enzyme was eluted from the Q-Sepharose column in a step gradient at 0.7 M NaCl and batch-eluted from the S-Sepharose column in 1 M NaCl. The enzyme (1 mg/ml) was stored at Ϫ20°C with 0.3 M NaCl and 5 mM DTT with 10 mM Tris-HCl, pH 7.5, and 50% glycerin. Polymerase was stored under the same conditions. Alternatively, the T7 DNA polymerase could be stored without NaCl at pH 5.8 in 10 mM Mes buffer containing 5 mM DTT and 50% glycerin.
The His 6 -tagged proteins were purified from BL21(DE3)pLysS cells by affinity chromatography using a nickel-nitrilotriacetic acid-agarose column (Qiagen). Cells were lysed in high salt at pH 7.5 as described above. After centrifugation, however, the supernatant was diluted to 0.3 M NaCl before loading onto a 2-ml column. We followed the general protocol suggested by the manufacturer for loading, washing, and eluting enzyme from the column. Briefly, the column was washed with 10 column volumes of 25 mM imidazole in pH 7.5 Tris-HCl following loading of the extract, and then the protein was eluted with 250 mM imidazole, pH 7.5.
Mutant T7 Polymerase-The N-terminal His 6 -tagged exo Ϫ T7 polymerase described above was mutated from 334 FNPSS 338 to GLPIA to give T7M. The altered sequence was made to encode a unique BamH1 site, which was used to screen plasmids after site-directed mutagenesis. We followed the general protocol described in the Stratagene catalog for QuikChange site-directed mutagenesis. Proline was not changed in the altered amino acid sequence to maintain peptide structure.
Single Round Synthesis Experiments-Each reaction mix of 10 l contained 1 pmol of 5Ј-end-labeled 32 P-labeled primer and 3 pmol of template at pH 7.5 in 10 mM Tris-HCl, 10 mM MgCl 2 , 5 mM DTT, and where indicated 0.2 nmol of thioredoxin. The reactions were conducted at 4°C unless stated otherwise. 5 pmol of T7 polymerase or polymerase were added and allowed to equilibrate with the primer-template. After a 10-min preincubation, 15 l of 500 M dNTPs containing 100 g of calf thymus DNA as a trap was added. The reaction was allowed to proceed for 10 s before quenching with 80 l of formamide containing xylene-cyanol dye, 250 mM EDTA, and 100 pmol of cold primer. The trap efficiency was tested with each primer template by adding the DNA trap prior to the polymerase and dNTPs. The samples were heated to 100°C for 5 min before loading 5 l of each sample on a 40-cm sequencing acrylamide gel. The gel was then exposed to a PhosphorImager screen and scanned with a Bio-Rad FX PhosphorImager, and the primer extension was quantified by the Bio-Rad Quantity One program. The processivity index (M/(NϩM)) was calculated for each nucleotide insertion in single round synthesis, where N is the integrated volume of band N, whereas M is the summation of the volume integral of all subsequent bands. The processivity index approaches 1 (most processive) as N approaches 0.
DNase I Footprinting-DNase I footprinting was used to map polymerase interactions with the DNA. 1 pmol of 32 P-labeled primer and 3-fold excess template was preincubated with either 5 pmol of T7 DNA polymerase and 40-fold excess thioredoxin or 10 pmol of polymerase at 4°C. DNase I was added in the amounts indicated in the figure with 10 mM Mg 2ϩ in a 5-s footprinting experiment. The reaction was quenched with 200 mM EDTA and of 0.1% SDS.

Determinants of Processivity of T7 DNA Polymerase-We
were interested in determining what structural features and interactions of T7 DNA polymerase were important in conferring processivity in DNA synthesis. T7 DNA polymerase becomes highly processive when complexed to E. coli thioredoxin. Only the first 11 of the 21 bp of the primer-template stem can been seen in the crystal structure of T7 polymerase complexed with a DNA template primer and thioredoxin (13). The crystal structure reveals a number of minor groove interactions with the first 4 bp and contacts to the first six phosphates in the primer and to the first eight phosphates in the template. To determine which of these interactions might be important in conferring processivity, we decided to make use of changes in primer-template length and sequence ( Fig. 1), salt concentration, methyphosphonate substitution, bulge-containing templates, and site-directed mutagenesis, along with binding constant determinations and DNase I footprinting. To separate the effects of the processivity factor, E. coli thioredoxin, we prepared the exo Ϫ T7 polymerase by a previously described method from thioredoxin-deficient E. coli strain to which we could then add thioredoxin. An exo Ϫ mutant was used to eliminate degradation of the primer terminus, which would complicate analysis of the processivity of the enzyme. The D5A, E7A mutant was chosen because it causes the minimum alteration of the exonuclease domain, and its kinetic properties have been thoroughly characterized (32). We also decided to compare the properties of T7 DNA polymerase with that of yeast pol , which as a member of the DNA damage bypass family, and is much less processive and more error-prone.
Effect of Primer-template Length on Processivity of T7 DNA Polymerase-We first wanted to see if we could localize the sites of important interactions on T7 DNA polymerase in the absence of thioredoxin by determining the effect of stem length on the processivity. We compared single round synthesis at 4°C with different length primer templates stems (Fig. 2, a-d) by preincubating the T7 DNA polymerase with the primer templates and initiating reaction with a mixture of dNTPs and many-fold excess calf thymus DNA to act as a trap for dissoci- ated enzyme. A processive burst on a poly(dA) template with a 10-bp primer-template stem results in extension up to 7 nucleotides (Fig. 2d). A shorter primer-template stem of 9 bp gave the same extension as 10 bp, but 8 was less processive and 7and 6-nucleotide stems were nearly distributive with 80% of the synthesis ending after the insertion of one nucleotide (Fig.  2, a-c). To better quantify the processivity, we have made use of the processivity index given by k pol /(k pol ϩ k off ), which represents the fraction of template-primer bound polymerase that goes on to add another nucleotide rather than to dissociate from the primer-template. A purely processive enzyme would have a processivity index of one, and only full-length product would be seen in the presence of a DNA trap for free enzyme. On the other hand, a purely distributive polymerase would have a processivity index of zero, and only the product of a single nucleotide insertion would be observed in the presence of a trap. Fig. 3a compares the processivity index for each step of nucleotide insertion for the reactions shown in Fig. 2 (a-d) and clearly shows an increase in processivity with increasing length of the primer-template stems, up to a maximum of about 0.8 for a 10-bp stem.
The difference in processivity is not due to dissociation of the primer-template into single strands, because we found that close to 100% of the primers were in a duplex with the template DNA under the conditions of the experiment. Fig. 2g is a 10% native acrylamide gel comparing the 10-nucleotide primer alone and along side the same primer with a 3-fold excess of template. All the primer was incorporated in a duplex with the template. These same results were observed with the 7-mer primer and its primer-template complex.
Effect of Primer-template Length on T7 DNA Polymerase Binding-Dissociation constants for the T7 DNA polymerase were calculated from synthetic bursts with constant primertemplate over a range of polymerase concentrations at 4°C (Fig. 3, b and c). Polymerase shows about the same K d for both the 7-and 10-mer primer-templates (307 Ϯ 52 and 240 Ϯ 26 nM, respectively) indicating that its recognition is within 7 nucleotides. In contrast, the K d for the His-tagged T7 DNA polymerase with no added thioredoxin for the 7-mer was 4.5 times higher than for the 10-mer primer-template (299 Ϯ 60 versus 63 Ϯ 10 nM, respectively) showing that the primertemplate stem beyond 7 nucleotides contributes significantly to binding. The K d for the 10-mer primer-template is similar to that of 65 nM reported for oligo(dT) 20 ⅐poly(dA) 380 (12).
Effect of Methylphosphonate Substitutions on Processivity of T7 DNA Polymerase-Methylphosphonates neutralize the negative charge in DNA and can be used to identify electrostatic interactions between DNA and protein (35), although as a secondary effect methylphosphonates narrow the minor groove of DNA (36) and can bend DNA (37). Because we had found that the processivity of synthesis dropped when the primer template stem was less than 8 bp, we decided to determine what the effect of changing phosphates 7 and 8 in both primer and template to methyl phosphonates would be on the processivity of T7 DNA polymerase in the absence of thioredoxin. We found that synthesis with P22m2/T46m2 is nearly distributive (Fig.  2e) and similar to the synthesis with the 7-mer primer-template stem (Fig. 2b).
Effect of Salt on T7 DNA Polymerase Processivity-Given that many of the interactions between the T7 polymerase and the primer-template involve phosphates, we wanted to see to what extent the processivity of DNA synthesis on different templates are modulated by salt. Increasing salt concentration has been previously shown to decrease the binding constant of the T7 DNA polymerase-thioredoxin complex for poly(dA)⅐ oligo(dT) (12). Fig. 4a shows single round synthesis with T7 DNA polymerase in the presence of thioredoxin on the 10-mer primer-template stem (P10/T20) with increasing salt concentration. As the salt approaches 500 mM, the reaction becomes almost completely distributive, similar to the reaction with the 7-mer primer-template stem (Fig. 2b). We then examined the processivity of T7 polymerase complexed to thioredoxin on a series of 22-mer primer-template stems with different template sequences for comparison. Processivity also decreased with ionic strength with a mixed sequence DNA template (Fig. 4b) but even more drastically for a poly(dA) or poly(T) template (Fig. 4, c and d). In each case, synthesis terminated between 7 and 9 nucleotides. Surprisingly, synthesis on a poly(dC) template was highly processive and virtually insensitive to salt (Fig. 4e).
The abrupt termination observed on the poly(dA) or poly(T) templates is probably due to the formation of a d(A⅐dT) n tract. That the termination is due to the specific structure that the d(A⅐T) n tract adopts rather than the stability of its duplex form is indicated by the inability of the randomized template sequence d(TAATTATAT) (Fig. 4g) to cause termination. In this later case, termination occurs about 3 to 5 nucleotides later within the poly(dA) template. Likewise, when a primer ending FIG. 3. Processivity index and binding constants for the primer-templates. a, the processivity index for extension of the primer corresponding to each band shown in Fig. 2 (a-d). Extension from primers P10, P8, P7, and P6 are indicated by circles, diamonds, x's, and squares, respectively. Synthetic burst amplitudes at 4°C were quantified for different concentrations of His-tagged T7 DNA polymerase (b) or His-tagged polymerase (c) with 100 nM primer-templates P7/T17 and P10/720 (circles and squares, respectively) used to calculate dissociation constants (K d values). Each data point was normalized to the maximum amplitude determined at the highest enzyme concentration and plotted as a fraction of that amount. in 4 Ts is used opposite the poly(dA) template, synthesis by T7 DNA polymerase terminates after the insertion of only 3-5 nucleotides (Fig. 4f), and with a template containing 5 dCs before the dA template, synthesis terminates 7-9 nucleotides after entering the poly(dA) template (Fig. 4h).
Effect of Mutating Asn 335 and Ser 338 of T7 DNA Polymerase on Processivity-The crystal structure of T7 DNA polymerasethioredoxin complex with DNA indicates that Asn 335 and Ser 338 of T7 gene 5 protein interact with template phosphates 7 and 8 (13). To investigate the role of these amino acids in processivity we constructed a His-tagged T7 polymerase mutant in which these residues were mutated. We decided to change 5 residues between 334 -338 from FNPSS to GLPIA to incorporate a BamH1 site into the DNA for screening purposes. The alteration also eliminates Ser 337 , which might otherwise be able to compensate for the loss of Ser 338 , although this residue appears distant from the DNA in the crystal structure. The proline in the sequence was retained to maintain the peptide conformation, and modeling of the mutant suggested that the mutations would not cause any significant structural distortions. Because the expression vector for the mutant T7 M DNA polymerase did not contain the appropriate antibiotic resistance gene for expression in the thioredoxin-deficient host used to prepare the unmodified T7 DNA polymerase, we had to use a host that was not deficient in thioredoxin. As a result, we could not directly test the effect of thioredoxin on the processivity of the mutant T7 M DNA polymerase, but we could compare the processivity of T7 DNA polymerase and the mutant on longer primer templates in the presence of thioredoxin.
The mutant His-tagged T7 M DNA polymerase was nearly distributive with the 10-mer primer-template stem in the absence of added thioredoxin (Fig. 2f) giving results similar to the 7-mer primer-template stem and the methylphosphonate substituted primer-template (Fig. 1e). The processivity of the mutated His-tagged T7 M DNA polymerase on a 22-mer primer template in the presence of excess thioredoxin and low salt were similar to those observed for the His-tagged T7 DNA polymerase in the presence of excess thioredoxin under high salt conditions (Fig. 5). Processivity was very high with the dC template, but much lower with the mixed template. With the dA or dT templates, synthesis terminated sharply after 4 or 5 nucleotides. The dissociation constants for the P22/T46 primer template in the presence of excess thioredoxin at 23°C were found to be 0.9 Ϯ 0.3 nM for the His-tagged T7 DNA polymerase and 3 Ϯ 1.5 nM for the mutant polymerase. A similar 4-fold decrease in binding affinity for the P10/T20 template at 23°C was found for the His-tagged T7 DNA polymerase (3.3 Ϯ 1.5 nM) the mutant His-tagged T7 M DNA polymerase (12 Ϯ 4.6 nM) in the absence of added thioredoxin.
Processivity of T7 DNA Polymerase in the Absence of Thioredoxin on Long Template Primers-Early termination in dA or T templates was also observed for T7 DNA polymerase in the absence of thioredoxin under low salt conditions with the 22-mer primer template stem (Fig. 6). The processivity FIG. 4. Effect of salt concentration on processivity. Single round synthesis with His-tagged T7 DNA polymerase with excess thioredoxin (0.2 nM) at different salt concentrations with the dA template at 4°C. In this experiment the His-tagged T7 DNA polymerase was 20-fold excess over primer-template to ensure complete binding of the gene 5 protein-thioredoxin complex even in the presence of salt, which interferes with thioredoxin-DNA binding. The NaCl concentration (in millimolar) is indicated above the lanes and is in addition to ϳ30 mM that comes from the polymerase stock solution. Lane C is a control lane in which the polymerase was omitted from the reaction. index for T7 DNA polymerase under these conditions remained constant at about 0.8 with the mixed template for about 19 nucleotides until it neared the end of the template. In contrast, synthesis on the poly(dA) and poly(T) templates (T46 and T46A) terminated sharply at 7 and 6 nucleotides, respectively, and behaved in a very similar manner to the P10/T20 template (Fig. 3a). As expected, synthesis on the template beginning in d(TAATTATAT), T46AT, stopped after 13 nucleotides, which is 4 nucleotides within the poly(dA) template. Unexpectedly, synthesis on the C template, which was highly processive with both T7 DNA polymerase and the mutant T7 M DNA polymerase in the presence of thioredoxin, was almost completely distributive with the T7 DNA polymerase in the absence of thioredoxin.
Effect of a 3T Bulge on Processivity- Fig. 7 compares synthesis with two primer-templates, one containing a 3T bulge (T47T3) in the template strand located 3 nucleotides from the site of Asn 335 -template interaction in the T7 DNA polymerase, and one without (T47). His-tagged T7 DNA polymerase synthesis in the presence of excess thioredoxin (0.2 nM) with this primer-template progressed only two nucleotides (Fig. 7a), which is what one would expect if the Asn 335 -DNA interaction, or physical interaction with the thumb structure, blocks the passage of the extrahelical bulge. In contrast, synthesis with polymerase did not appear to be affected by the template bulge (Fig. 7b).
Effect of Methylphosphonates on Thioredoxin-mediated Processivity-We tested the effectiveness of thioredoxin as a processivity factor for T7 DNA polymerase with a primer-template containing methylphosphonates. Fig 8a shows that thioredoxin does not promote processive synthesis with a 10-bp primertemplate stem, whereas Fig. 8b shows that thioredoxin can act as a processivity agent with a longer primer-template stem of 22 nucleotides. On the other hand, a 22-bp primer-template stem containing 22 methylphosphonates distributed throughout the putative thioredoxin binding site (13) did not promote processive synthesis and behaved like an unmodified 10-bp primer-template stem, which is thioredoxin-insensitive. A time-course experiment with the modified primer-template shows only short synthetic products until ϳ80% of the initial substrate is extended, after which only full-length product is seen (Fig. 8c). The modified primer migrated as a fuzzy band in acrylamide electrophoresis, which might be due to the presence of so many diastereomeric methylphosphonates. We therefore performed single round synthesis with an unmodified primer and modified template, P22/T46m11 (Fig. 8d), and obtained a synthetic burst, which was the same as seen for the unmodified 10-bp primer-template stem (Fig. 8a). Although this reduction in processivity might be due to interference with protein-phos- phate interactions, it is also possible that it is due to narrowing of the minor groove (36) or bending of DNA, which is known to result from asymmetrical incorporation of methylphosphonates into only one strand of a DNA duplex (37).
DNase I Footprinting of the Polymerase Binding Sites-DNase I footprinting was used to map the pol and T7 DNA polymerase-thioredoxin-DNA complexes using two different primers opposite the same 5Ј-end labeled template, T51. The P46 primer corresponds to the P42 primer extended by 4 nucleotides and was used to see if the footprint tracked with the position of the primer along the template. Fig. 9a, lane C, shows a limit digest of 13 nucleotides or less with the P42/T primer-template in the absence of polymerase. This same primer-template with T7 DNA polymerase-thioredoxin shows an intense band located 28 nucleotides from the 5Ј-end of the template. This corresponds to cleavage 19 nucleotides from the 3Ј-end of the primer and probably defines the outer border of the T7 DNA polymerase-thioredoxin interaction with the DNA and is consistent with the crystal structure (13). This footprint is smaller than what was determined by burst experiments on synthetic primer-templates, which concluded that the optimal primer-template stem was 25-26 bp (38). Footprinting with the P46 primer shows a different cleavage pattern consisting of a faint band at 28 nucleotides and more intense bands at 27 and 25 nucleotides reflecting the positional change of 4 nucleotides in the primer. In contrast, polymerase only protects 6 or 7 nucleotides of the P42/T51 primer-template stem at the highest DNase I concentrations. The strong 13-mer cleavage band found in all lanes corresponds to cleavage in the template that corresponds to a position that is 4 nucleotides from the 3Ј-end of the primer and results from the presence of unbound primer-template. DISCUSSION According to the crystal structure, Asn 335 and Ser 338 of the thumb domain of T7 DNA polymerase interact with template phosphates 7 and 8 of the primer-template stem (13). Results with different sizes or methylphosphonate-modified primertemplates demonstrate that these interactions are essential for processive synthesis (Figs. 2-4). These interactions may also function with a nearby helix-loop-helix motif located at residues 339 -372 to enclose the DNA. It has been proposed that the analogous structure found in the Klenow fragment limits DNA dissociation during processive synthesis (17). The thumb structure must be held in close proximity to the DNA by weak ionic interactions. Salt interferes with these interactions, and amino acid substitutions of non-polar aliphatic residues for Asn 335 and Ser 338 render the structure non-functional except on a poly(dC) template (Figs. 4 and 5). Despite a close fit, duplex DNA must pass freely to keep pace with synthesis, although details of how this might occur is not available. Conversely, the same protein subdomain blocks passage of a 3T template bulge (Fig. 7a), which may be how it suppresses frameshift mutations as does the Klenow fragment (17).
Asn 335 , Ser 338 , and thioredoxin-thumb interactions with the DNA are also required for bypassing d(A⅐T) n tracts (Figs. 3-6), which are known to cause bending of the DNA and constrict the minor groove (39). Constriction of the minor groove is considered the primary reason for HIV-1 reverse transcriptase termination in d(A⅐T) n tracts. The HIV-1 reverse transcriptase terminates at nucleotide positions 6 -8 beginning from the start of dA tracts in the template strand and 4 -6 nucleotides beginning with T tracts (40). Similarly, T7 DNA polymerase alone or with thioredoxin but missing Asn 335 and Ser 338 interactions terminates synthesis 4 -9 nucleotides within the poly(dA) or poly(T) templates (Figs. 4 and 5). Termination must be caused by the specific structure formed by d(A⅐T) n tracts, because termination does not occur within a randomized poly-(dA,T) template (Fig. 4g).
Thioredoxin does not promote processive synthesis with a primer-template containing methylphosphonates distributed throughout the putative thioredoxin binding site (Fig. 8, c and  d). Methylphosphonates neutralize the phosphates in DNA but as a secondary effect may narrow its minor groove (36). One or both of these effects could interfere with thioredoxin-mediated interaction with the DNA. It seems clear that thioredoxin works together with Asn 335 and Ser 338 in the T7 DNA polymerase processive reaction. The polymerase-thioredoxin complex missing any interaction of Asn 335 and Ser 338 with the DNA is only marginally processive with a mixed DNA template as well as poly(dA) and poly(T) (Fig. 5). What was completely unexpected, however, is that synthesis opposite the poly(dC) template appears to be almost unaffected by the mutant polymerase or by high salt. What makes this more perplexing is that the unmodified T7 DNA polymerase in the absence of thioredoxin terminates synthesis with the poly(dC) template after inserting only 3 nucleotides (Fig. 6). The explanation for this unusual behavior may have to do with the specific structure of poly(dC) (41) and the spatial positioning of its phosphate backbone and/or base stacking and its interactions with the polymerase.
In contrast to the T7 DNA polymerase, which appears to bind 10 bp of a primer template, yeast polymerase appears to only bind 7 bp or less. This conclusion is based on the affinity of pol for different size DNA stems (Fig. 3c) and DNase I footprinting (Fig. 9b) and is consistent with a model of the polymerase bound to a template primer that is based on the crystal structure (31). The openness of the active site of polymerase would also explain how it is able to easily slide by a 3T extrahelical bulge (Fig. 7b), indicating that it might be prone to frameshift mutations.
In conclusion, we have shown that Asn 335 and Ser 338 of the thumb domain of T7 DNA polymerase are involved in binding to the primer-template and may function with a nearby HLH motif (residues 339 -372) to enhance processive synthesis. The T7 DNA polymerase thumb domain also interacts with thioredoxin, and we have shown this complex interacts directly with the DNA. It seems clear that thioredoxin, Asn 335 , and Ser 338 interactions with DNA are required for efficient bypass of d(A⅐T) n tracts and even for optimal processive synthesis with mixed sequence DNA. In contrast, polymerase only binds to a 7-bp primer-template stem, and its weak hold on DNA is consistent with a loosely interacting thumb domain.