RNA Template-dependent 5′ Nuclease Activity ofThermus aquaticus and Thermus thermophilus DNA Polymerases*

DNA replication and repair require a specific mechanism to join the 3′- and 5′-ends of two strands to maintain DNA continuity. In order to understand the details of this process, we studied the activity of the 5′ nucleases with substrates containing an RNA template strand. By comparing the eubacterial and archaeal 5′ nucleases, we show that the polymerase domain of the eubacterial enzymes is critical for the activity of the 5′ nuclease domain on RNA containing substrates. Analysis of the activity of chimeric enzymes between the DNA polymerases from Thermus aquaticus(TaqPol) and Thermus thermophilus (TthPol) reveals two regions, in the “thumb” and in the “palm” subdomains, critical for RNA-dependent 5′ nuclease activity. There are two critical amino acids in those regions that are responsible for the high activity of TthPol on RNA containing substrates. Mutating glycine 418 and glutamic acid 507 of TaqPol to lysine and glutamine, respectively, increases its RNA-dependent 5′ nuclease activity 4–10-fold. Furthermore, the RNA-dependent DNA polymerase activity is controlled by a completely different region of TaqPol and TthPol, and mutations in this region do not affect the 5′ nuclease activity. The results presented here suggest a novel substrate binding mode of the eubacterial DNA polymerase enzymes, called a 5′ nuclease mode, that is distinct from the polymerizing and editing modes described previously. The application of the enzymes with improved RNA-dependent 5′ nuclease activity for RNA detection using the invasive signal amplification assay is discussed.

The structure-specific 5Ј nucleases are involved in DNA replication and nucleotide excision repair, where the primary function of these enzymes is to remove the RNA primers of Okazaki fragments or damaged DNA fragments, respectively (1)(2)(3)(4)(5)(6). The activity of the 5Ј nucleases is controlled by DNA polymerases that create for them an optimal substrate by displacing the 5Ј-end of the downstream strand during DNA synthesis (7)(8)(9). The precise removal of the displaced arm by the 5Ј nuclease creates a nicked structure repaired by DNA ligase (9). Previously, we studied the substrate specificity of seven eubacterial and archaeal structure-specific 5Ј exonucleases with DNA substrates (9). The enzymes showed very similar specificities, de-spite their limited level of sequence similarity and different structural organization; eubacterial 5Ј nucleases are discrete domains of the DNA polymerases, whereas in archaea the 5Ј nucleases are separate polypeptides. Comparison of the 5Ј nuclease activities of Taq DNA polymerase (TaqPol) 1 and its isolated nuclease domain (TaqExo) showed that the 5Ј nuclease domain can function independently, although the polymerase domain influences the 5Ј nuclease activity by imposing additional stringency on substrate recognition and by increasing substrate binding (9,10).
A specific substrate for the 5Ј nucleases that resembles DNA undergoing displacement synthesis can be created with synthetic oligonucleotides. This substrate, termed the "overlapping substrate," is formed by annealing adjacent upstream and downstream oligonucleotides on a template strand, also called the target strand. The duplexes formed by the two oligonucleotides must overlap by at least one base pair for efficient cleavage by the 5Ј nucleases (9 -11). The upstream oligonucleotide in this substrate is completely annealed to the target except for the 3Ј terminal nucleotide that may interact with the 5Ј nuclease. The downstream oligonucleotide consists of the target specific region and the 5Ј arm region that is not annealed to the target and cleaved by the 5Ј nuclease to create the nicked substrate (9,11). The ability of the 5Ј nucleases to specifically recognize and cleave the overlapping substrate has been utilized in assays for quantitative DNA detection and single nucleotide polymorphism analysis (11)(12)(13)(14). In this method, called the invasive signal amplification reaction, cleavage of the downstream oligonucleotide is dependent upon the presence of a particular target sequence in a sample. Each target molecule can generate more than 10 3 reporter molecules by rapid turnover of the downstream oligonucleotide at elevated temperatures (11,12). 2 By adding a second invasive reaction, total signal amplification can reach a factor of 10 7 (12). The invasive signal amplification reaction can also be used for quantitative RNA detection, although this would require an investigation of the ability of the structure-specific 5Ј nucleases to cleave the overlapping substrate with an RNA target.
In order to gain insight into the mechanism of substrate recognition by the structure-specific 5Ј nucleases and to develop an enzyme for RNA detection using the invasive signal amplification assay, we compared the activity of eubacterial and archaeal 5Ј nucleases using an overlapping substrate containing an RNA target. We find that only the eubacterial DNA polymerases TaqPol and TthPol possess RNA templatedependent 5Ј nuclease activity, although it is significantly reduced compared with the 5Ј nuclease activity with a DNA target. Of the two enzymes, TthPol has a higher RNA-dependent 5Ј nuclease activity than TaqPol. We have used this observation to investigate the activity of chimeric enzymes constructed from TthPol and TaqPol, and we have identified two regions in the polymerase domain involved in substrate recognition of 5Ј nuclease substrates. Site-directed mutagenesis studies reveal that lysine 420 and glutamine 509 of TthPol are the amino acids in these regions that are critical for the 5Ј nuclease activity on RNA-containing substrates. Mutating the analogous amino acids of TaqPol (G418K and E507Q) to match TthPol at those positions increases the RNA-dependent 5Ј nuclease activity of TaqPol 4 -10-fold but has no effect on its DNAor RNA-dependent polymerase activities. Based on these results, we propose a novel 5Ј nuclease-specific mode of substrate binding by eubacterial DNA polymerases.

EXPERIMENTAL PROCEDURES
Materials-Polymerase chain reaction amplification was done with the Advantage cDNA polymerase chain reaction kit (CLONTECH). Restriction enzymes were purchased from New England Biolabs. Chemicals and buffers were from Fisher unless otherwise noted.
Site-directed Mutagenesis-Site-directed mutagenesis of the TaqPol and TthPol genes was performed with the Transformer Site-Directed Mutagenesis kit (CLONTECH) according to the manufacturer's protocol. Using site-directed mutagenesis, three new unique restriction sites, NotI, BstBI, and NdeI, were created in the TaqPol gene at positions approximately corresponding to amino acids 328, 382, and 443, respectively. The same restriction sites were also introduced at the homologous positions 330, 384, and 443 of the TthPol gene. The NotI sites were created by using mutagenic primers 5Ј-GCCGCCAGGGGCGGC-CGCGTCCACCGGGCC and 5Ј-GCCTGCAGGGGCGGCCGCGTGCAC-CGGGCA, which correspond to the sense strands of the TaqPol and TthPol genes, respectively. The BstI and NdeI sites were introduced into both genes using sense strand mutagenic primers 5Ј-CTCCTG-GACCCTTCGAACACCACCCC and 5Ј-GTCCTGGCCCATATGGAG-GCCAC. The mutated nucleotides are shown in boldface type, and the corresponding restriction sites are underlined.
Construction and Purification of Chimeric Enzymes-Chimeric constructs were created by substituting homologous fragments between the cloned TaqPol and TthPol genes using the common unique restriction sites EcoRI, NotI, BstBI, NdeI, BamHI, and SalI and standard cloning techniques. Expression and purification of the chimeric enzymes was done as described (9 -11) except that the His Bind resin chromatography purification step was replaced by affinity chromatography using an Econo-Pac heparin cartridge (Bio-Rad) and a Dionex DX 500 HPLC instrument. Briefly, the cartridge was equilibrated with 50 mM Tris-HCl, pH 8, 1 mM EDTA, and enzyme extract dialyzed against the same buffer was loaded on the column and then eluted with a linear gradient of NaCl (0 -2 M) in the same buffer. The HPLC-purified protein was dialyzed and stored as described above. Expression of the TthPol-TaqPol chimeric genes yielded proteins of the expected size. The enzymes were purified to homogeneity according to SDS-polyacrylamide gel electrophoresis (9) and were shown to cleave the invasive substrates with both DNA and RNA targets.
Substrate Preparation and Purification-The downstream and upstream oligonucleotides and the IL-6 DNA target were synthesized on a PerSeptive Biosystems instrument using standard phosphoramidite chemistry (Glen Research). The synthetic RNA-DNA chimeric IrT target labeled with biotin at the 5Ј-end was synthesized utilizing 2Ј-ACE RNA chemistry (Dharmacon Research). The 2Ј-protecting groups were removed by acid-catalyzed hydrolysis according to the manufacturer's procedure. The downstream probes labeled with 5Ј-fluorescein or 5Јtetrachlorofluorescein at their 5Ј ends were purified by reverse phase HPLC using a Resource Q column (Amersham Pharmacia Biotech). A fragment of human IL-6 cDNA (nucleotides 64 -691 of the sequence published in Ref. 15) was cloned using a TOPO-TA Cloning Kit (Invitrogen), and the 640-nucleotide IL-6 RNA target was synthesized by T7 RNA polymerase run-off transcription of the cloned fragment using a Megascript Kit (Ambion). All oligonucleotides were finally purified by separation on a 20% denaturing polyacrylamide gel followed by excision and elution of the major band. Oligonucleotide concentration was determined by measuring absorption at 260 nm. The biotin-labeled IrT target was incubated with a 5-fold excess of streptavidin (Promega) in a buffer containing 10 mM MOPS, pH 7.5, 0.05% Tween 20, 0.05% Nonidet P-40, and 10 g/ml tRNA at room temperature for 10 min.
RNA Template-dependent 5Ј Nuclease Activity Assay-Unless otherwise indicated, RNA template-dependent probe cycling reactions were carried out in 10 l of a reaction buffer containing 10 mM MOPS, pH 7.5, 0.05% Tween 20, 0.05% Nonidet P-40, 10 g/ml tRNA, 100 mM KCl, and 5 mM MgSO 4 . The substrates and 5Ј nuclease enzymes were mixed with the reaction buffer lacking MgSO 4 and overlaid with Chill-out liquid wax (MJ Research). The enzyme and substrate concentrations are specified in the figure legends of Figs. 1, 3, 4, 5, and 6. Reactions were brought up to reaction temperature, started by the addition of MgSO 4 , and incubated for the specified length of time. Reactions were stopped by the addition of 10 l of 95% formamide containing 10 mM EDTA and 0.02% methyl violet (Sigma). Samples were heated to 90°C for 1 min immediately before electrophoresis through a 20% denaturing acrylamide gel (19:1 cross-linked) containing 7 M urea in a buffer of 45 mM Tris borate, pH 8.3, 1.4 mM EDTA. Gels were then scanned on an FMBIO-100 fluorescence gel scanner (Hitachi) using a 505-nm emission filter. The fraction of cleaved product was determined from intensities of bands corresponding to uncut and cut substrate with FMBIO Analysis software (version 6.0; Hitachi). The fraction of cleavage product did not exceed 20% to ensure that measurements approximated initial cleavage rates. The cycling cleavage rate was defined as the concentration of the cut downstream probe divided by the target concentration and the time of the reaction (in minutes).
Michaelis-Menten Kinetics Assays-Assays consisted of 10-l reactions containing 10 mM MOPS, pH 7.5, 0.05% Tween 20, 0.05% Nonidet P-40, 10 g/ml tRNA, 4 mM MgCl 2 , 1 nM enzyme (TaqPol, TthPol, or TaqG418K/E507Q), and different concentrations (0.125, 0.25, 0.5, or 1 M) of an equimolar mixture of the IrT target and the probe. The cleavage kinetics for each enzyme and each substrate concentration were measured at 46°C. The fraction of cleaved product was determined as described above and plotted as a function of reaction time. The initial cleavage rates were determined from the slopes in the linear range of the cleavage kinetics and were defined as the concentration of cut product divided by the enzyme concentration and the time of the reaction (in minutes). The Michaelis constant K m and the maximal catalytic rate k cat of each enzyme with the IrT substrate were determined from the plots of the initial cleavage rate versus the substrate concentration.

RESULTS
DNA and RNA Template-dependent 5Ј Nuclease Activities of Structure-specific 5Ј Nucleases-The 5Ј nuclease activity assay was designed similarly to the invasive signal amplification reaction described previously (11). The IL-6 substrate used in these experiments consists of the downstream and upstream oligonucleotides annealed with either RNA or DNA target (the terms "target" and "template" will be used interchangeably) as shown in Fig. 1A. The fluorescently labeled downstream oligonucleotide, also called the probe, is specifically cleaved by the 5Ј nuclease at the site where the upstream oligonucleotide overlaps or "invades" the probe (9, 10). The invasive signal amplification reaction is carried out under conditions of limiting target and excess probe. At the optimal reaction temperature, which occurs at the melting temperature of the probe, a single target molecule gives rise to multiple cleaved probes due to a rapid probe exchange. 2 The cleavage rate of this reaction is determined as the number of the probe molecules cleaved per target molecule per minute.
In agreement with the previous results, 2 all enzymes showed high cleavage activity with the IL-6 DNA target (Fig. 1B). When the DNA target was replaced with the IL-6 RNA target, only TthPol and TaqPol were able to cleave the probe, although at a reduced rate compared with the DNA target (Fig. 1C). The cleavage rates of TaqPol and TthPol with the RNA target are only 0.2 and 0.8 min Ϫ1 , respectively, which is almost 2 orders of magnitude lower than the cleavage rates on the DNA substrate with the same sequence (the rates were determined from the data shown in Fig. 1, B and C, as described under "Experimental Procedures"). All other enzymes, including the TaqExo nuclease domain and the archaeal 5Ј nucleases, showed no RNAdependent 5Ј nuclease activity. The common feature of these 5Ј nucleases is the absence of a polymerase domain. Thus, it would be logical to assume that the polymerase domain is involved in the ability of the eubacterial enzymes to cleave the substrate with an RNA target.
Chimeric Enzymes between TaqPol and TthPol-DNA polymerase enzymes from Thermus species, like all DNA polymerase I homologues, consist of two distinctive domains, the 5Ј nuclease and the polymerase domains, shown schematically in Fig. 2. The polymerase domain comprises the C-terminal two-thirds of the proteins, and it is responsible for both DNA-dependent and RNA-dependent DNA polymerase activities, whereas the N-terminal one-third contains the 5Ј nuclease domain. According to a commonly accepted nomenclature (17), the polymerase domain has been described as having a physical form resembling a right hand. The palm region consists of, roughly, amino acids 300 -500 of a genus Thermus DNA polymerase I, the thumb region includes amino acids 500 -650, and the fingers region is formed by the remaining amino acids from 650 to 830 (Fig. 2).
The amino acid sequences of TaqPol and TthPol share about 87% identity and greater than 92% similarity. We took advantage of this high degree of sequence similarity between the enzymes to construct a series of chimeric enzymes between TthPol and TaqPol. The chimeric constructs shown in Fig. 2 were created by swapping DNA fragments defined by the restriction endonuclease sites, EcoRI and BamHI, common for both genes, the SalI site in the cloning vector, and the new sites, NotI, BstBI, and NdeI, created at the homologous positions of both genes by site-directed mutagenesis (see "Experimental Procedures" and Fig. 2). Since TthPol has a 4-fold higher cleavage rate with the IL-6 RNA template than TaqPol (Fig. 1C), the activities of the chimeric enzymes were rated relative to TthPol and used as a parameter to identify the region(s) affecting RNA template-dependent 5Ј nuclease activity.
RNA-dependent 5Ј Nuclease Activity of the Chimeric Enzymes-The activity of each chimeric enzyme was evaluated using the invasive signal amplification assay with the IL-6 RNA target (Fig. 1A); the cleavage rates shown in Fig. 3 were determined as described under "Experimental Procedures." Comparison of the cleavage rates of the first two chimeras, TaqTth(N) and TthTaq(N), created by swapping the polymerase and 5Ј nuclease domains at the NotI site (Fig. 2), shows that TaqTth(N) has the same activity as TthPol, whereas its counterpart TthTaq(N) retains the activity of TaqPol (Fig. 3). This result indicates that the higher cleavage rate of TthPol is associated with its polymerase domain, which is consistent with an important role for the polymerase domain in the 5Ј nuclease activity (9, 10).
The next step was to identify the minimal region of TthPol polymerase domain that would give rise to the TthPol-like 5Ј nuclease activity when substituted for the corresponding region of the TaqPol sequence. To this end, we selected the TaqTth(N) chimera to generate a series of new constructs by replacing its TthPol sequence with homologous regions of TaqPol. First, we substituted the N-terminal and C-terminal segments of the TaqPol polymerase domain for the corresponding regions of TaqTth(N) using the common BamHI site as a breaking point to create TaqTth(N-B) and TaqTth(B-S) chimeras, respectively (Fig. 2). TaqTth (N-B), which has the TthPol sequence between amino acids 328 and 593, is approximately 3 times more active than the TaqTth(B-S) and 40% more active than TthPol (Fig. 3). This result demonstrates that the NotI-BamHI region of the TthPol polymerase domain is responsible for the high 5Ј nuclease activity of TthPol with RNA targets.
The NotI-BamHI region of TthPol was further subdivided into two approximately equal parts using NdeI (Fig. 2), and the effect of the substitution of each of these sequences on the cleavage rate of TaqPol was investigated by measuring the activities of the TaqTth(N-Nd) and TaqTth(Nd-B) chimeras. Each of these chimeric enzymes showed TaqPol-like activity (Fig. 3), suggesting that both NotI-NdeI and NdeI-BamHI regions of TthPol contain amino acids indispensable for the high cleavage rate of TthPol. The substitution of the BstBI-BamHI region of TthPol for the homologous sequence of TaqPol produced the TaqTth(Bs-B) chimera, which exhibited the same activity as TthPol (Fig. 3). Thus, this study of the chimeric enzymes limited the portion of the TthPol sequence determining its high RNA template-dependent 5Ј nuclease activity to the BstB-BamHI region located approximately between amino acids 382 and 593 (Fig. 2).
Site-directed Mutagenesis of the Chimeric Enzymes-Comparison of the TthPol and TaqPol amino acid sequences between the BstBI and BamHI sites reveals only 25 differences (Fig. 4A). Among these, there are 12 conservative substitutions and 13 substitutions resulting in an alteration of charge. Since the analysis of the chimeric enzymes suggested that the critical mutations are located in both BstBI-NdeI and NdeI-BamHI regions of TthPol, we used chimeric enzymes that have TthPol sequence in one of these regions and introduced TthPol-specific amino acids by site-directed mutagenesis in the other region that has TaqPol sequence. For example, six TthPol-specific substitutions changing amino acid charge were created in the BstBI-NdeI region of the TaqTth(Nd-B) chimera by single or double amino acid mutagenesis (Fig. 4A). Only one of these substitutions, double mutation W417L/G418K, was able to restore the TthPol activity with the IL-6 RNA target, whereas the other four mutations were neutral ( Fig. 4B and data not  shown). Similarly, all 12 TthPol amino acid substitutions were introduced at the homologous positions of the NdeI-BamHI region of the TaqTth(N-Nd) chimera, and only one, E507Q, increased the cleavage rate to the TthPol level, whereas the other 10 mutations were neutral, and one, G499R, showed a smaller increase (Fig. 4B and data not shown).
To confirm that the W417L, G418K, and E507Q substitutions are sufficient to increase the TaqPol activity to the TthPol level, TaqPol variants carrying these mutations were created, and their cleavage rates with the IL-6 RNA target were compared with that of TthPol. Fig. 4C shows that the TaqPol W417L/G418K/E507Q and TaqPol G418K/E507Q mutants have a 1.4 times higher activity than TthPol and more than a 4-fold higher activity than TaqPol, whereas the TaqPol W417L/ E507Q mutant has the same activity as TthPol. Thus, these results provide strong evidence that the RNA-dependent 5Ј nuclease activity of TaqPol with the RNA IL-6 target can be significantly increased by just two critical amino acid substitutions, G418K and E507Q.
Characterization of the TaqPol G418K/E507Q Mutant-Next, we compared the TaqPol G418K/E507Q, TaqPol, and TthPol enzymes in the RNA template-dependent 5Ј nuclease assay while varying temperature and the concentrations of salt and divalent ions. The upstream DNA oligonucleotide and the RNA target of the substrate used in this study were linked into a single molecule, called the IrT target, as shown in Fig. 5A, and the labeled probe was present in large excess. The 5Ј-end of the IrT target was blocked with a biotin-streptavidin complex to prevent nonspecific degradation by the enzyme during the reaction (18,19). 2 The cleavage rates for TaqPol G418K/ E507Q, TaqPol, and TthPol are plotted versus temperature in Fig. 5B. The activity difference between TthPol and TaqPol with the IrT target is even greater than that found with the IL-6 RNA target. The G418K/E507Q mutations increase the activity of TaqPol more than 10-fold and by 25% as compared with TthPol. All three enzymes show a typical temperature profile of the invasive signal amplification reaction 2 and have the same optimal temperature. We found no significant effect of the G418K/E507Q mutations on DNA template-dependent 5Ј nuclease activity of TaqPol with an all-DNA target analogous to IrT under the same conditions (data not shown).
The effects of KCl and MgSO 4 concentrations on the 5Ј nuclease activity of TaqPol G418K/E507Q, TaqPol, and TthPol with the IrT target are shown in Fig. 5, C and D. The activities of all enzymes have similar salt dependences with an optimal KCl concentration of 100 mM for TaqPol G418K/E507Q and TthPol and 50 mM for TaqPol. The optimal MgSO 4 concentration for all enzymes is approximately 8 mM. The analysis of the data presented in Fig. 5 suggests that the properties of TaqPol G418K/E507Q are much closer to those of TthPol than TaqPol, confirming the key role of the G418K/E507Q mutations in the recognition of the substrate with an RNA target.
To understand the mechanism of the reduction of the 5Ј nuclease activity in the presence of an RNA versus a DNA target, we determined the Michaelis constant, K m , and the maximal catalytic rate, k cat , of TaqPol, TthPol, and TaqPol G418K/E507Q using an excess of the IrT target and the probe and a limiting enzyme concentration as described under "Experimental Procedures." It was found that all three enzymes have similar K m values in the range of 200 -300 nM and k cat values of approximately 4 min Ϫ1 for TaqPol and TthPol and of 9 min Ϫ1 for TaqPol G418K/E507Q (data not shown).
Polymerase Activities of TthPol, TaqPol, and TaqTth Chimeric Enzymes-To determine whether the RNA template-dependent 5Ј nuclease activity of the Thermus DNA polymerase enzymes is related to their RNA-dependent polymerase activity, we have reversed the D785N and D787N mutations used to create the polymerase-deficient versions of TaqPol and TthPol, respectively (9). The polymerase activities were then evaluated by extension of a dT 25-35 -oligonucleotide primer with fluorescein-labeled dUTP in the presence of either poly(dA) or poly(A) template as described under "Experimental Procedures." As shown in Fig. 6, the DNA-dependent polymerase activities are very similar for all constructs used in this experiment, whereas the RNA-dependent polymerase activities of TthPol, TaqTth(N), and TaqTth(B-S) are at least 6-fold higher than the activities of TaqPol, TaqTth (N-B), and the TaqPol W417L/ G418K/E507Q mutant. From the analysis of these results, it can be concluded that the high RNA-dependent DNA polymerase activity of TthPol is determined by the C-terminal half of the polymerase domain (amino acids 593-830) and that the RNA-dependent 5Ј nuclease and polymerase activities are controlled by different regions and thus are not related to each other. An approach similar to that described in this work can be used to identify the critical amino acids affecting the reverse transcriptase activity of the Thermus DNA polymerase enzymes.

DISCUSSION
Whereas all structure-specific 5Ј nucleases tested in this work efficiently cleave the overlapping substrate with a DNA target, only TthPol and TaqPol enzymes retain partial activity with the substrate in which the DNA target is replaced with an RNA target (Fig. 1). These were the only enzymes that had both 5Ј nuclease and DNA polymerase domains, thereby implying an important role for the polymerase domain in substrate recognition. This conclusion agrees with our previous finding that the polymerase domain of TaqPol affects the activity and specificity of its 5Ј nuclease domain and that both TthPol and TaqPol bind the overlapping DNA substrate more strongly than the archaeal 5Ј nucleases and the isolated 5Ј nuclease domain of TaqPol (9, 10). 2 Using the chimeric constructs between TthPol and TaqPol and site-directed mutagenesis, we identified two groups of amino acids that are important for the RNA template-dependent 5Ј nuclease activity and that are possibly involved in recognition of the target strand of the overlapping substrate. In the TthPol sequence, the first group includes leucine 419 and lysine 420, and the second one con- sists of glutamine 509. Mutation of the homologous amino acids in TaqPol (W417L, G418K, and E507Q) increases its RNA template-dependent 5Ј nuclease activity 4 -10-fold and makes it even higher than the activity of TthPol. To understand the role of these mutations in the interactions of TaqPol with the overlapping substrate, we used the known three-dimensional structures of binary/ternary complexes of DNA polymerases and primer-template DNA duplexes (20,21,22). The primertemplate duplex of those structures is analogous to the upstream duplex of the overlapping substrate that formed by the upstream and target strands.
The positions of the W417L, G418K, and E507Q mutations in the crystal structure of a ternary complex of the polymerase domain of TaqPol (Klentaq1), dideoxynucleotide, and a primertemplate DNA (20) are shown in Fig. 7. The E507Q mutation is located at the tip of the thumb subdomain at a nearest distance of 3.8 Å and 18 Å from the backbone phosphates of the primer and template strands, respectively. The interaction between the thumb and the minor groove of the DNA primer-template was previously suggested by the binary complex structures of Klenow fragment of DNA polymerase I (21) and TaqPol (22) with DNA. Deletion of a 24-amino acid portion of the tip of the thumb in Klenow fragment, corresponding to amino acids 494 -518 of TaqPol, reduces the DNA binding affinity by more than 100-fold (23). These observations are consistent with the hypothesis that the thumb region, which includes the E507Q mutation, is involved in interactions with the upstream duplex of the overlapping substrate.
The W417L and G418K mutations present in the palm region of TaqPol (Fig. 7) are located approximately 25 Å from the nearest phosphates of the primer-template duplex according to the binary/ternary complex structures of TaqPol with DNA bound in polymerizing mode (20,22). The same distance was observed between the analogous W513 and P514 amino acids of Klenow fragment and the template strand of DNA bound in the editing mode (21). Thus, no interactions between amino acids 417-418 of TaqPol and the overlapping substrate can be suggested from the available binary complex studies.
To explain the data described here, we propose that the amino acids at positions 417 and 418 in the palm region of TaqPol interact with the upstream duplex of the overlapping substrate only when the enzyme functions as a 5Ј nuclease but no interaction of the substrate with these amino acids occurs when TaqPol switches into polymerizing mode. This hypothesis suggests a novel mode of substrate binding by DNA polymerases, which we call the 5Ј nuclease mode. Several lines of evidence support this hypothesis. The chimeric enzyme study clearly separates regions of the polymerase domain involved in the 5Ј nuclease and polymerase activities. Accordingly, the W417L and G418K mutations, together with the E507Q mutation, affect the 5Ј nuclease activity of TaqPol on substrates having an RNA target strand (Fig. 4C) but have no effect on either RNA-or DNA-dependent DNA polymerase activities (Fig. 6). Conversely, mutations in the active site of TaqPol, such as R573A, R587A, E615A, R746A, N750A, and D785N, which correspond to substitutions in Klenow fragment that affect both polymerase activity and substrate binding affinity in the polymerizing mode (24 -26), have little or no effect on the 5Ј nuclease activity (data not shown). Superposition of the polymerase domains of TaqPol (22), Escherichia coli DNA polymerase I (21), and Bacillus stearothermophilus DNA polymerase I (27) using DALI (28,29) and Insight II (Molecular Simulation Inc.) programs shows that a portion of the palm subdomain of TaqPol between amino acids 402 and 451, including Trp 417 and Gly 418 , is highly conserved between the three polymerases, although there is no structural similarity between the rest of the palm subdomains. This observation sug- gests an important role for this region in eubacterial DNA polymerases.
What could be the reason for a 5Ј nuclease binding mode? As we discussed above, the 5Ј nuclease and polymerase activities should be precisely synchronized in order to create a nicked structure rather than a gap or an overhang that could result in a deletion or an insertion during Okazaki fragment processing or DNA repair. According to the previously proposed model (9), the 3Ј terminal nucleotide of the upstream strand in the overlapping substrate is sequestered by the 5Ј nuclease domain to prevent its extension, thus halting synthesis. This interaction with the 3Ј nucleotide activates the 5Ј nuclease, which then endonucleolytically removes the displaced 5Ј arm of the downstream strand by precise incision at the site defined by the 3Ј nucleotide to create the nick. This model requires a substantial rearrangement of the substrate-enzyme complex that may include a translocation of the complex to the 5Ј nuclease mode to separate the primer-template duplex from the polymerase active site.
The hypothesized translocation into the 5Ј nuclease mode could be accomplished through an interaction of the downstream duplex formed between the template and downstream strands with the crevice formed by the finger and thumb subdomains. Such an interaction could force conformational transitions in the thumb that would bring the upstream duplex into close contact with the Trp 417 and Gly 418 amino acids. Significant flexibility of the thumb has been previously reported that might explain such changes (17, 20 -22, 30, 31). Additional conformational changes in the fingers domain that might help to open the crevice, such as the transition from the "closed" to the "open" structure described by Li et al. (20), are consistent with this model. To answer the question as to why the 5Ј nuclease binding mode was not observed in any of the published co-crystal structures of a DNA polymerase I, we would argue that the majority of the structures were solved for the polymerase domain only and with a primer-template substrate rather than with an overlapping substrate.
The K m values of 200 -300 nM determined in this work for TaqPol, TthPol, and TaqPol G418K/E507Q for the RNA-containing substrate are much higher than the K m value of Ͻ1 nM estimated previously for TthPol with an all-DNA overlapping substrate, 2 suggesting that the RNA target adversely affects substrate binding. The reduced affinity for RNA-containing substrates can be explained by the unfavorable interaction between the enzyme and the A-form duplex adopted by the substrate with an RNA target or by inhibition of binding by the ribose 2Ј hydroxyls of the RNA target. Among these two factors, the latter looks more attractive, since the 5Ј nucleases of eubacterial DNA polymerases can efficiently cleave substrates with an RNA probe (18), which would presumably have an A-form, and since the binary complex structural studies suggest that the primer-template duplex partially adopts a conformation close to A-form in its complex with DNA polymerase (20,22,27). The G418K/E507Q mutations increase the k cat of TaqPol more than 2-fold but have little effect on K m . Such an effect would be expected if the mutations positioned the sub-  (20). Amino acids Trp 417 , Gly 418 , and Glu 507 are displayed in a space-filling view, and the rest of the polypeptide is shown as a white ribbon. Heavy atoms of the DNA template and primer strands are shown in green. A portion of the tip of the thumb, corresponding to amino acids 494 -518, and a part of the structurally conserved region of the palm subdomain (see "Discussion") corresponding to amino acids 402-435 are shown as red ribbons.
strate in an orientation more appropriate for cleavage rather than simply increasing the binding constant.
In conclusion, we have identified specific mutations in two regions of the polymerase domain of TaqPol important for RNA template-dependent 5Ј nuclease activity. We propose a novel 5Ј nuclease mode of substrate binding that is distinct from the previously described editing and polymerizing modes. According to the model, the transition of the enzyme-substrate complex from the polymerizing to the 5Ј nuclease mode requires the presence of the 5Ј nuclease domain and a specific overlapping substrate. The transition is probably accompanied by a dislocation of the thumb subdomain that brings the substrate in direct contact with the structurally conserved portion of the palm subdomain. This work opens up the possibility of development of 5Ј nucleases that can specifically cleave signal probes in the presence of an RNA target and, therefore, can be used for RNA analysis with the invasive signal amplification reaction.