The Essential C Family DnaE Polymerase Is Error-prone and Efficient at Lesion Bypass*

DnaE-type DNA polymerases belong to the C family of DNA polymerases and are responsible for chromosomal replication in prokaryotes. Like most closely related Gram-positive cells, Streptococcus pyogenes has two DnaE homologs Pol C and DnaE; both are essential to cell viability. Pol C is an established replicative polymerase, and DnaE has been proposed to serve a replicative role. In this report, we characterize S. pyogenes DnaE polymerase and find that it is highly error-prone. DnaE can bypass coding and noncoding lesions with high efficiency. Error-prone extension is accomplished by either of two pathways, template-primer misalignment or direct primer extension. The bypass of abasic sites is accomplished mainly through “dNTP-stabilized” misalignment of template, thereby generating (–1) deletions in the newly synthesized strand. This mechanism may be similar to the dNTP-stabilized misalignment mechanism used by the Y family of DNA polymerases and is the first example of lesion bypass and error-prone synthesis catalyzed by a C family polymerase. Thus, DnaE may function in an error-prone capacity that may be essential in Gram-positive cells but not Gram-negative cells, suggesting a fundamental difference in DNA metabolism between these two classes of bacteria.

The replicative DNA polymerase of Escherichia coli, the ␣ subunit, is encoded by dnaE and functions with a sliding clamp processivity factor and a clamp loader for efficient chromosome duplication (1)(2)(3). Most Gram-positive organisms have two essential polymerase subunits with high sequence similarity to the E. coli ␣ subunit, called Pol C 1 and DnaE (4 -7). Pol C is essential and has long been known to be required for replication of the chromosome (6, 8 -11). Both Pol C and DnaE function with the ␤ sliding clamp and a clamp loader (12).
The dnaE gene is essential in Gram-positive bacteria that are closely related to Streptococcus pyogenes, such as Enterococcus faecalis, Staphylococcus aureus, and Bacillus subtilis (4,5,7). Furthermore, at nonpermissive temperature, a ts B. sub-tilis dnaE mutant is disrupted in lagging strand synthesis but not leading strand synthesis of plasmid DNA (4). These data suggest that Pol C and DnaE are both replicative polymerases, with Pol C polymerase synthesizing the leading strand and DnaE polymerase synthesizing the lagging strand. Pol C contains an intrinsic 3Ј-5Ј proofreading exonuclease, whereas DnaE, like E. coli ␣, lacks proofreading activity. However, E. coli ␣ binds tightly to ⑀, a small 3Ј-5Ј exonuclease that supplies proofreading function to ␣ (13). S. pyogenes also contains a homolog to ⑀, but whether it associates with DnaE has not been tested.
E. coli DNA polymerase III is a high fidelity polymerase that is severely blocked by DNA lesions (1,14). Specialized repair polymerases are required to synthesize across DNA lesions in vivo (15). Bypass of DNA lesions, or "translesion synthesis," is performed by the Y family of DNA polymerases, which are characterized by their ability to bypass lesions in synthetic DNA templates and lack an associated exonuclease (16).
In this report, we examine the biochemical properties of the S. pyogenes DnaE polymerase and find that it has low fidelity, efficiently misincorporates nucleotides into DNA, and extends mismatched termini of primers. DnaE also proficiently extends a primer across an abasic lesion. Moreover, S. pyogenes DnaE does not associate with the S. pyogenes ⑀ homolog. Thus, S. pyogenes DnaE polymerase is an error-prone C family polymerase and may constitute the first example of an error-prone polymerase that is essential for cell viability.
Purification of S. pyogenes Exonuclease Encoded by DnaQ-A search of the S. pyogenes genome data base using the sequence of several known DnaQ proteins yielded a weak homolog of 26% identity. PCR primers were designed as follows. The upstream 31-mer, 5Ј-GCG AAA TAA TCA CAA CAT ATG ACG AAG GCC C-3Ј, contains an NdeI site, and the downstream 29-mer, 5Ј-GAC CTA GCG GGA TCC GCA AGG GCT AAC CC-3Ј contains a BamHI site. A PCR product obtained with these primers was digested with NdeI and BamHI, purified (624 bp), and ligated into pET 11a vector to produce pET11a.SpdnaQ. The S. pyogenes DnaQ gene encodes a 208-residue protein of 23,176 Da.
The pET 11a.SpdnaQ plasmid was transformed into E. coli BL21(DE3)recA. A single colony was used to inoculate 12 liters of LB broth supplemented with 200 g/ml ampicillin. Cells were grown at 37°C to A 600 ϭ 0.5, at which point the temperature was lowered to 15°C and 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside was added. Induction proceeded for 16 h at 15°C, and then cells were collected by centrifugation and resuspended in 50 mM Tris-HCl, pH 7.5, 10% su-crose, 1 M NaCl, 30 mM spermidine, 5 mM dithiothreitol (DTT). Cells were lysed by two passages through a French press (15,000 p.s.i.) followed by centrifugation at 13,000 rpm for 30 min. Ammonium sulfate was added to the lysate to a final concentration of 0.226 g/ml. The precipitate was collected by centrifugation, and the protein pellet was resuspended in buffer A (20 mM Tris-HCl (pH 7.5), 4% glycerol, 0.1 mM EDTA, 5 mM DTT) and then dialyzed against buffer A. The dialyzed protein (250 mg) was applied to a 30-ml Fast Flow Q column and then eluted with a linear gradient of 50 -500 mM NaCl in Buffer A. The peak fractions (fractions 27-42; 121 mg) were pooled, dialyzed, and applied to a 15-ml heparin-Affi-Gel column and eluted with a linear gradient of 50 -500 mM NaCl in Buffer A. The peak fractions were pooled (30 mg) and dialyzed against buffer A containing 50 mM NaCl, and then onehalf was applied onto a 1-ml MonoQ column. The MonoQ column was developed with a linear gradient of 50 -500 mM NaCl in Buffer A. The peak fractions were pooled and stored frozen at Ϫ80°C.
Steady-state Kinetics-Reaction conditions were optimized for reaction times and enzyme concentration, such that less than 20% of the primed template was extended. Several time courses were performed to determine the range over which nucleotide insertion was linear. Reactions contained 6.4 nM 32 P-labeled primer-template in 5 l of Reaction Buffer (40 mM Tris-HCl (pH 7.5), 8% glycerol, 0.2 mM EDTA, 10 mM DTT, 80 g/ml bovine serum albumin, 16 mM MgCl 2 ). Primer-template mixtures were incubated with various concentrations of DnaE (0.033-16.9 nM) for 1 min at 37°C, and then synthesis was initiated by mixing with an equal volume of dNTP buffer (reaction buffer containing dNTPs). To measure nucleotide insertion, the polymerase/primer-template (5 l) was added to 5 l of dNTP buffer containing various concentrations of correct nucleotide (0.03 M to 1.2 mM) or incorrect nucleotide (0.01-20 mM). To measure mismatch extension kinetics, the concentration of TTP was 0.03 M to 1.2 mM for correct primer-template termini and 0.01-20 mM for incorrect primer-template termini. Reactions were incubated at 37°C for 1-10 min and were quenched with a 20-l solution of 96% formamide, 40 mM EDTA, 0.001% bromphenol blue, 0.001% cyanol blue. Reaction products were resolved on a 12% polyacrylamide gel (8 M urea), exposed to a PhosphorImager screen, and analyzed using ImageQuant software. Apparent V max /K m values were determined as described (18 -20). The intensity of the bands corresponding to the starting primer (I 0 ) and the primer extended by one nucleotide (I 1 ) were used to calculate velocities. Relative velocity was calculated using v ϭ 100(I 1 /(I 0 ϩ 0.5I 1 )t). Apparent V max and K m values were determined from a Hanes-Woolf plot of the data. A Hanes-Woolf plot of [dNTP]/v versus [dNTP] was fitted by linear least squares to obtain the intercept (K m /V max ) and slope (1/V max ). The nucleotide misincorporation efficiency (f inc ) and intrinsic mismatch extension frequency (f ext 0 ) are expressed as ratios: (V max /K m )w/(V max /K m )r, where w and r refer to either correct (r) or incorrect (w) deoxyribonucleoside pair formation.
Analysis of DnaE and DnaQ by Gel Filtration-Gel filtration analysis of protein mixtures was performed using an HR 10/30 Superose 6 column (PerkinElmer Life Sciences) equilibrated with buffer A containing 100 mM NaCl. Protein mixtures were incubated at 25°C for 10 min, and then the sample was injected onto the column. After the first 6.6 ml, fractions of 170 l were collected. Fractions were analyzed in 10% SDS-polyacrylamide gels. Analysis of DnaE and DnaQ subunit was performed using either 800 g of DnaE alone, 400 g of DnaQ alone, or a mixture of both proteins (800 g of DnaE and 400 g of DnaQ) in 200 l of buffer A containing 100 mM NaCl.

S. pyogenes DnaE Is a Low Fidelity
Polymerase-To examine the fidelity of the S. pyogenes DnaE polymerase, DnaE was incubated with a synthetic primer-template, but in the presence of only one dNTP. Surprisingly, S. pyogenes DnaE polymerase incorporated dNTP in each case and in fact extended the primer up to 7 bases (Fig. 1A). We examined the S. pyogenes DnaE preparation for terminal transferase activity but found no activity in the absence of a primed site (data not shown). As controls, we examined E. coli core (exo-) and S. pyogenes Pol C at the same concentrations as S. pyogenes DnaE. E. coli core (exo-) contains ␣, , and a mutant ⑀ subunit, in which two residues are replaced by alanines to eliminate exonuclease activity. The results indicate that these polymerases appear to be less efficient in incorporation of a mismatched dNTP (Fig. 1, B and C, all lanes except lanes 1, 6, 11, and 16). In fact, unlike S. pyogenes DnaE, neither E. coli core (exo-) nor S. pyogenes Pol C can extend the primer to the end of the template in presence of all four dNTPs. S. pyogenes Pol C has an intrinsic exonuclease activity, which accounts for the degradation products observed in Fig. 1C. Overall, these observations suggest that S. pyogenes DnaE has quite low fidelity, especially compared with established replicative polymerases.
The above experiments were performed under conditions using excess polymerase and a long incubation time. Next, we performed a similar experiment except under single hit conditions, accomplished using a low enzyme concentration such that less than 20% of the primers were extended, thereby ensuring that primer-template had a productive encounter with the polymerase only once (18). The first unpaired template residue and the incoming dNTP were systematically varied. The concentration of incoming dNTP was titrated in separate reactions, providing the steady state kinetic data necessary to determine the apparent V max /K m and the efficiency of misincorporation (Table I, template residues A, T, C, and G) (18 -20). Consistent with the results of experiments in Fig. 1, S. pyogenes DnaE misincorporates dNTP at a high rate (f inc ϭ 1.4 ϫ 10 Ϫ3 to 5.8 ϫ 10 Ϫ2 ). These misincorporation rates are 100 -1000-fold higher than those reported for replicative polymerases, such as ␣ subunit from E. coli (10 Ϫ4 to 10 Ϫ5 ) (21). These rates are also higher than those reported for errorfree repair polymerases, such as Pol II (misincorporation efficiency ϳ10 Ϫ4 ) (22)(23)(24). Fig. 1 and Table I demonstrate that S. pyogenes DnaE has an impressive capacity to misincorporate dNTPs, thus forming a DNA-mismatched 3Ј primer terminus. Next, we studied DnaE action on DNA substrates that contain a preformed mismatch at the 3Ј primer terminus ( Fig. 2A). In the presence of all four dNTPs, S. pyogenes DnaE efficiently extended almost all of the mismatched primer to near the end of the template (Fig. 2A, lane  1). In contrast, the replicative E. coli core (exo-) polymerase could not extend the mismatched primer at all under identical conditions ( Fig. 2A, lane 6). These results are consistent with those of previous reports that demonstrate that neither E. coli DNA polymerase III (exo-) nor isolated ␣ subunit extend mismatches (25)(26)(27).

S. pyogenes DnaE Extends Mismatches at a Similar Efficiency as Misincorporation-The experiments in
The S. pyogenes DnaE often stops one nucleotide short of the end of the template (lane 1). This observation indicates that mismatch extension on this substrate appears to proceed primarily by template-primer misalignment. This observation is further supported by the efficient incorporation of dTTP in lane 3, which may be explained by "flip-out" of the G in the template strand, thereby pairing the 3Ј-A to the T in the template. This can also explain the lack of primer extension in lanes 4 and 5, since there are no flip-out solutions for dC or dG. Lane 2 shows that dATP can also be incorporated efficiently by direct extension of the mismatched terminus.
To examine the mismatch extension frequency of S. pyogenes DnaE polymerase in greater detail, the primer terminus and "paired" template nucleotides were each systematically varied to create four correctly paired and 12 mismatched primer templates (Fig. 2B). Only the correct incoming dTTP was included, and conditions were optimized such that less than 20% of the primer was extended. Mismatch extension frequency was calculated by titrating dTTP into different reactions and determining V max /K m from the resulting time courses, as described (18,20). Of the 12 mismatches, the frequencies of mismatch extension varied considerably (1.3 ϫ 10 Ϫ1 to 8.4 ϫ 10 Ϫ5 ) but were generally high compared with estimates of mismatch extension frequencies by E. coli core (exo-) (Ͼ10 Ϫ5 to 10 Ϫ6 ) (21,25,27). In contrast, mismatch extension by error-prone polymerases occurs over a range of 10 Ϫ1 to 10 Ϫ2 for Pol , Pol , and Rev1 and of 10 Ϫ4 to 10 Ϫ5 for Pol IV (28 -30). Fig. 2B shows that four mismatched primed templates have high frequencies of mismatch extension (between 1.3 ϫ 10 Ϫ1 and 1 ϫ 10 Ϫ2 ), and the other eight have more moderate efficiencies (between 1 ϫ 10 Ϫ2 and 8.4 ϫ 10 Ϫ5 ). The four mismatched primed templates that have high extension efficiencies are pyrimidine/purine mismatches, and the eight with moderate extension efficiencies are purine/purine or pyrimidine/pyrimidine mismatches. The purine/pyrmidine mismatches may form a better steric fit within the enzyme active site, enabling DnaE to extend the mismatch more readily. The same pattern of mismatch extension has been shown for Pol IV polymerase (28). These may be compared with previous studies of E. coli core and ␣ subunit in which mismatch extension efficiency is exceedingly poor (10 Ϫ6 ) (21,25,27).
In general, repair polymerases typically extend mismatches at a similar rate at which they incorporate mismatched nucleotides (14,15). One exception to this rule is Pol , which extends mismatches more frequently than it misincorporates them, and thus it has been proposed that the in vivo role of Pol is to extend from mismatches (30). Human immunodeficiency virus reverse transcriptase has also been shown to extend mismatched primer termini with high efficiency (31, 32). In  Incorporation Specificity at an Abasic Site-The formation of abasic lesions in the chromosome is a fairly common event. An abasic site is considered "noncoding," since the original base information is lost (33)(34)(35). Early studies suggested that the preferred dNTP incorporated opposite an abasic site followed the A-rule, where dATP is preferentially inserted opposite an abasic site (36). Recent studies have revealed that some polymerases do not follow the A-rule (15,30,37,38). Instead, some of these polymerases skip over abasic lesions, thereby generating (Ϫ1) deletions. This mechanism is referred to as dNTPstabilized template misalignment (28,37).
We investigated the response of S. pyogenes DnaE polymerase upon encountering an abasic site. In Fig. 3A, S. pyogenes DnaE was incubated with a primed template containing an abasic site and all four dNTPs. The results show that S. pyogenes DnaE is capable of extending most of the primers past the abasic site to the end of the template, although the product analysis shows that it pauses at the abasic site. E. coli core (exo-) was incapable of inserting a dNTP opposite the abasic site. S. pyogenes Pol C, which contains an intrinsic 3Ј-5Ј exonuclease activity, degraded rather than extended the primed template. The intrinsic ability of S. pyogenes DnaE to extend a primer past the abasic site suggests that it may function in some bypass capacity.  1-5) and E. coli core (exo-) (lanes 6 -10). Reactions contained 32 P-labeled primertemplate and equimolar amounts of either S. pyogenes DnaE or E. coli core (exo-). Proteins were incubated for 5 min. B, intrinsic mismatch extension efficiencies (f ext 0 ) were determined for all 16 primer-template combinations. f ext 0 is the ratio of (V max /K m ) mispair /(V max /K m ) correct pair . C, DnaE is more efficient at misincorporation (f inc ) than at mismatch extension (f ext 0 ). The incorporation data are from Table I, and the mismatch extension data from B are plotted against one another. Data points that fall above the line indicate more efficient mismatch extension. Data points that fall below the line indicate more efficient misincorporation. We next asked whether S. pyogenes DnaE skips over the abasic lesion (like Pol IV), or whether it incorporates a specific dNTP opposite the abasic site (like Pol II). If DnaE skips over the abasic site, the final length of the extended primer should be at least one nucleotide shorter than the template strand. Results in Fig. 3B show that most of the fully extended primer strand is in fact one to two nucleotides shorter than the template strand. Next, we asked whether DnaE can use any dNTP to extend the primer terminus positioned at an abasic site. The results in lanes 2-5 of Fig. 3B demonstrate that S. pyogenes DnaE can utilize dA, dT, and dG. Incorporation of dT is probably due to dNTP-stabilized template misalignment, by pairing with the A just beyond the abasic site. Incorporation of dA may be the result of direct misincorporation following the A-rule. However, it would appear that DnaE does not only follow the A-rule, since dGTP is also incorporated, and even dCTP to some extent. In fact, multiple incorporation events are observed in the presence of either dCTP or dGTP. We then included only one dNTP in a reaction under single hit conditions for single nucleotide extension as in Table I. Efficiency of nucleotide incorporation was calculated for each dNTP with the result that all dNTPs are incorporated with high frequency (Table I). dTTP is incorporated with a much higher efficiency compared with the other dNTPs, consistent with the results of Fig. 3B discussed above.

S. pyogenes DnaE Does Not Skip over a Template Base
Mismatch-To further examine the properties of DnaE, we constructed a set of primer-templates with well defined homopolymeric extensions (Fig. 4). S. pyogenes DnaE efficiently extends the primer to the end of a homopolymeric dT template in the presence of all four dNTPs (Fig. 4A, lane 1). The primer is not extended very far if only an incorrect dNTP is added (lanes 3-5). The one or two nucleotides incorporated in lanes 3-5 are most likely due to misincorporation. Interestingly, S. pyogenes DnaE did not reach the end of the template when only dA was added to the reaction (lane 2), suggesting that slippage on the template occurs. We presume that in the presence of all four dNTPs, the template has time to realign into the proper register when the polymerase binds an incorrect dNTP and that this action prevents slippage products. Specifically, in the presence of all four dNTPs, one of the three noncognate dNTPs will often occupy the active site and will need to diffuse back out, which may buy sufficient time for the template to realign.
Next, we examined DnaE on a homopolymeric dC template (Fig. 4B). In the presence of all four dNTPs or dGTP (lanes 6 and 10, respectively), DnaE extends the primer to either full length or 1 nucleotide shy of full length, indicating that some slippage occurs whether or not all four dNTPs are present. dTTP and dCTP are misincorporated on this template with similar efficiency as dATP, supporting the conclusion that DnaE directly misincorporates nucleotides with high efficiency (lanes 7-9).
Next, we designed a template that is homopolymeric in T (similar in sequence to that in Fig. 4A) but is interrupted at two positions with dC (Fig. 4C). S. pyogenes DnaE still extends the primer to the end of this template in the presence of all four dNTPs (Fig. 4C, lane 11). However, if only dATP is added, two dark bands are visible, one representing a primer extension product of 6 nucleotides and the other representing a primer extension product of 12 nucleotides (lane 12). The C residues are positioned 6 and 12 bases from the primer terminus. Thus, the S. pyogenes DnaE incorporates a dATP opposite the C residue at each position and then pauses before proceeding (Fig. 2B). It remains possible that instead of directly incorporating dATP opposite C, DnaE may flip-out the C residue and incorporate dATP at the next T, followed by template realignment to form a mismatched 3Ј terminus, the extension of which causes the polymerase to pause. Furthermore, there is a light band representing full primer extension to the end of the template, confirming that no bases are skipped (lane 12).

S. pyogenes DnaE Efficiently Bypasses 8-oxo-dG and 2-Aminopurine (2-AP) Bases-We next determined how S. pyogenes
DnaE responds to the lesion, 8-oxo-7,8-dihydro-2Ј-deoxyguanosine (8-oxo-dG) (Fig. 5). 8-oxo-dG is a common DNA lesion that occurs as a product of oxidative damage and exists in both syn and anti conformations (39). In the presence of all four dNTPs, S. pyogenes DnaE extends essentially all of the primer past the 8-oxo-dG lesion (Fig. 5, lane 1). It also efficiently misincorporates bases opposite 8-oxo-dG in each of the single nucleotide addition reactions (lanes [2][3][4][5]. A comparison with the E. coli core (exo-) is shown in Fig. 5B. Results using individual dNTPs in Fig. 5 show that DnaE can utilize any of the four dNTPs to extend the primer past 8-oxo-dG and does so more efficiently than E. coli core (exo-).
In the favored syn conformation, 8-oxo-dG forms base mispairs with dA. This mispair contributes significantly to mutation rates in oxidatively damaged cells by generating G:C to T:A transversion mutations. In the anti conformation, 8-oxo-dG correctly pairs with dC. Results in Table I   high efficiency. These results suggest that S. pyogenes DnaE can repair genomic DNA that has suffered oxidative damage, but in a potentially mutagenic fashion.
We next examined the response of S. pyogenes DnaE to the nucleotide analog 2-AP. 2-AP is a strong genotoxic mutator, which efficiently induces the SOS response in prokaryotic cells and is also highly toxic to eukaryotic cells (40,41). S. pyogenes efficiently bypasses the 2-AP nucleotide analog, extending all of the primers in the presence of the four dNTPs (Fig. 5, lane 6).
2-AP is predominantly incorporated opposite T. However, in subsequent steps of replication, 2-AP can mispair with C, thereby generating A:T to G:C transitions. S. pyogenes DnaE incorporates either dT or dC opposite 2-AP with similar high efficiency (Table I), consistent with the well established mutagenic potential of 2-AP (42)(43)(44). Results in Fig. 5 (lanes 7-10) demonstrate that dT is efficiently incorporated opposite 2-AP by S. pyogenes DnaE, and the other dNTPs are incorporated to some extent (Table I). Thus, S. pyogenes DnaE can bypass several different DNA lesions and nucleotide analogs in a potentially error-prone manner. E. coli core (exo-) incorporates nucleotides much less efficiently opposite the same lesion (Fig.  5, lanes 11-20).
Lack of Interaction of DnaE with the DnaQ Homolog-The E. coli ⑀ exonuclease subunit (encoded by dnaQ) interacts tightly with the ␣ polymerase (product of the dnaE) of DNA polymerase III (45). This 3Ј-5Ј exonuclease subunit increases the fidelity of the E. coli DNA polymerase III and further diminishes its ability to bypass lesions (21,26). Since S. pyogenes DnaE polymerase is a homolog of the E. coli ␣ subunit, we reasoned that perhaps DnaE polymerase and the DnaQ homolog of S. pyogenes might also interact, like E. coli ␣ and ⑀, thus providing proofreading function to DnaE.
To test this idea, we identified, cloned, overproduced, and purified the S. pyogenes DnaQ homolog. First we examined S. pyogenes DnaQ for exonuclease activity by incubating DnaQ with a 5Ј-labeled 30-mer DNA oligonucleotide. The results, in Fig. 6A, show the disappearance of the full-length 30-mer DNA and the appearance of the degradation ladder, indicating the presence of the 3Ј-5Ј exonuclease activity.
To test the physical interaction between S. pyogenes DnaE and DnaQ, we incubated the two proteins and then analyzed them for complex formation using a gel filtration sizing column. The results in Fig. 6B show that S. pyogenes DnaQ does not coelute with DnaE, indicating that they do not interact. Although E. coli ␣⅐⑀ complex is stable and can be isolated by gel filtration, it is possible that S. pyogenes DnaE and DnaQ interact but that the complex is too weak to detect by this tech-nique. Therefore, we studied S. pyogenes DnaQ for an effect on the DNA polymerase activity of DnaE.
The ability of E. coli ␣ to bypass a lesion is inhibited by ⑀ (46,47). To examine whether S. pyogenes DnaQ affects lesion bypass activity of DnaE, we incubated DnaE and DnaQ on a template containing an abasic site. Fig. 6C shows that DnaQ neither stimulates nor inhibits DnaE-catalyzed DNA synthesis. The result is consistent with inability of DnaQ to interact with DnaE.

S. pyogenes DnaE Is a Promiscuous Polymerase-S. pyogenes
DnaE and E. coli ␣ subunit share 34% sequence identity and 47% sequence similarity. Despite this high degree of sequence similarity, the two proteins exhibit markedly different biochemical properties, suggesting that the proteins serve different functional roles in vivo. E. coli ␣ is known to be the primary replicative polymerase, and its biochemical properties are typical of a protein designed for replication: high fidelity synthesis of undamaged DNA templates, poor ability to bypass DNA lesions, and poor ability to extend from DNA mismatches (1,14). Moreover, the replicative polymerase binds to a proofreading subunit, ⑀, which further enhances fidelity and retards its ability to bypass lesions. Finally, the replicative polymerase binds tightly to a ring-shaped clamp, which enhances the processivity of the polymerase by preventing it from falling off of the DNA (2, 3).
The current study demonstrates that S. pyogenes DnaE does not share the biochemical properties of a replicative polymerase. In contrast, S. pyogenes DnaE has biochemical properties that are typical of a repair polymerase: low fidelity synthesis of undamaged DNA templates, efficient bypass of DNA lesions, and efficient extension from DNA mismatches. Furthermore, DnaE polymerase does not bind to the S. pyogenes DnaQ homolog of the E. coli ⑀ proofreading subunit encoded by dnaQ. However, we cannot exclude the possibility that S. pyogenes DnaE might bind DnaQ in vivo, perhaps through a third bridging subunit that has yet to be identified. E. coli core contains a third subunit, . However, ␣ binds ⑀ directly, and does not serve this role (48,49). Further, is dispensable in E. coli (50), and homologs of do not appear to be present in many organisms.
There are two different paths a polymerase may take to bypass an abasic lesion: misincorporation opposite the abasic site, catalyzed by Pol II and Pol V, or skipping the abasic site by dNTP-stabilized template misalignment, catalyzed by the Y family polymerases, Pol IV (DinB), and Pol . DnaE prefers to skip over an abasic site (Fig. 7). On undamaged templates, DnaE most likely utilizes a combination of direct misincorporation and template-primer misalignment to extend either a mismatched or a correctly paired primer (Fig. 7). The mechanism of misalignment could be either dNTP-stabilized or misincorporation misalignment (51).

Does DnaE Function at the Replication Fork?-
The current study demonstrates that S. pyogenes DnaE is an error-prone polymerase. It seems unlikely that an error-prone polymerase would participate in bulk chromosomal synthesis. Perhaps DnaE is more accurate in vivo, but it is also possible that it may function in another capacity. For example, DnaE may serve to move a stalled replication fork past a blocking lesion. When a replication fork encounters a DNA lesion, the replication apparatus is expected to stall, and repair proteins must be recruited to correct or bypass the lesion before replication can be restarted. An example of this type of action in E. coli is promoted by Pol IV and Pol V. E. coli Pol IV and Pol V are induced as part of the SOS response, and they function to bypass lesions in an error-prone fashion (28,38). These polymerases are not essential to E. coli, and in fact most lesions are either repaired before the replication fork encounters them or are avoided by replication restart or recombinative repair rather than bypassed by an error-prone DNA polymerase (52)(53)(54)(55). Gram-positive cells also contain Y family DNA polymerases. However, the fact that DnaE is essential, at least in most Gram-positive cells examined to date, may suggest that some blocks to replication occur frequently in Gram-positive cells and that DnaE is needed to circumvent them.
Like Pol IV and Pol V in E. coli, S. pyogenes DnaE may also be a participant in the SOS response in Gram-positive bacteria. The S. pyogenes dnaE gene contains a putative SOS response regulon in its sequence, a dinR binding box (data not shown) (56). Furthermore, in the distantly related species M. tuberculosis, transcription of dnaE is stimulated 14-fold during the SOS response (57,58). Thus, S. pyogenes DnaE may catalyze translesion synthesis in Gram-positive cells during the SOS response.
Translesion synthesis by S. pyogenes DnaE is probably quite costly to the organism, since the current study demonstrates that it is highly error-prone. Thus, DnaE action on undamaged templates may also result in frequent base substitutions or (Ϫ1) frameshifts. If this is the case for DnaE, there must be some mechanism to harness its action, directing it only as needed to bypass blocks to replication that are not solved by error-free processes. How Can a Polymerase in the C Family Replicative Polymerase Class Function as a Sloppy Polymerase?-Studies of polymerase fidelity suggest that the size and the geometry of the active site of polymerases correlate with the fidelity of the polymerase (14,59). Generally, polymerases with tight active sites are able to accommodate only correctly paired bases, conferring high fidelity (60,61). DNA polymerases with large flexible active sites have sufficient extra space to accommodate DNA lesions and thereby may bypass the lesion, but at a cost to fidelity because the loose active site does not provide exquisite base discrimination. Thus, one may expect that S. pyogenes DnaE may have a larger active site compared with Pol C or E. coli ␣ subunit. However, it has also been shown that polymerase fidelity can be drastically lowered by simply mutating a few residues around the active site. A good example of this point is a mutation study of the active site of the Thermus aquaticus DNA polymerase I (62). Therefore, it is also possible that the differences between Pol C and DnaE active site architecture will be subtle.
Why Is an Error-prone Polymerase Essential for Cell Viability?-Several studies in closely related Gram-positive bacteria have demonstrated that the dnaE gene is essential for viability (5,7). Further, studies by Dervyn et al. (4) suggest that DnaE may be responsible for lagging strand synthesis, although alternative explanations may exist. The low fidelity of DnaE and slow speed (60 nucleotides/s), even with the sliding clamp present, are not consistent with the fidelity and speed (1 kb/s) required for chromosomal DNA replication. It is still possible that DnaE functions directly in replication, but if so, one may expect that other factors increase its speed and fidelity. We have examined the effect of the clamp loader and ␤ clamp on DnaE, but they had no effect on its ability to bypass an abasic site or extend a mismatch (data not shown).
Why would an error-prone polymerase be essential to the cell? Perhaps, at some phase of the cell cycle, an error-prone process is absolutely required. However, the catalytic activity of a DNA polymerase is not always the defining essential activity, as demonstrated in the case of E. coli DNA polymerase I and yeast DNA polymerase ⑀. The polymerase activity of E. coli DNA polymerase I is not essential, but the unique 5Ј-3Ј exonuclease activity is essential, being required to remove the RNA from the ends of Okazaki fragments (63). Eukaryotic Pol ⑀ is essential, but one report on this subject indicates that the region encoding the polymerase activity is dispensable (64). Despite these arguments, DnaE could still be the lagging strand polymerase and thus reflect major differences and requirements between Gram-positive and Gram-negative cells. In that case, it will be interesting to see what protein-protein interactions target DnaE to the lagging strand.