Escherichia coli Rep DNA helicase and error-prone DNA polymerase IV interact physically and functionally

Authors


E-mail plfoster@indiana.edu; Tel. (+1) 812 855 4084; Fax (+1) 812 855 6705.

Summary

Escherichia coli DNA polymerase IV, encoded by the dinB gene, is a member of the Y family of specialized DNA polymerases. Pol IV is capable of synthesizing past DNA lesions and may help to restart stalled replication forks. However, Pol IV is error-prone, contributing to both DNA damage-induced and stress-induced (adaptive) mutations. Here we demonstrate that Pol IV interacts in vitro with Rep DNA helicase and that this interaction enhances Rep's helicase activity. In addition, Pol IV polymerase activity is stimulated by interacting with Rep, and Pol IV β clamp-binding motif appears to be required for this stimulation. However, neither Rep's helicase activity nor its ability to bind DNA is required for it to stimulate Pol IV's polymerase activity. The interaction between Rep and Pol IV is biologically significant in vivo as Rep enhances Pol IV's mutagenic activity in stationary-phase cells. These data indicate a new role for Rep in contributing to Pol IV-dependent adaptive mutation. This functional interaction also provides new insight into how the cell might control or target Pol IV's mutagenic activity.

Introduction

DNA damage in Escherichia coli results in the transcriptional induction of over 40 genes, a phenomenon known as the SOS response. Many of these genes encode enzymes that promote DNA repair, recombination, and DNA synthesis past lesions that block replication [reviewed in Janion (2008)]. Three DNA polymerases that assist in the replication of damaged DNA are induced as part of the SOS response: DNA polymerase II (Pol II, encoded by the polB gene), DNA polymerase IV (Pol IV, encoded by the dinB gene) and DNA polymerase V (Pol V, encoded by the umuDC operon). Pol II is postulated to play a major role in the error-free restart of stalled replication forks (Rangarajan et al., 1999). Pol IV and Pol V are members of the Y family of specialized DNA polymerases (Ohmori et al., 2001), which are characterized by their ability to catalyse synthesis past DNA lesions. However, these polymerases are poorly processive and error-prone when replicating both damaged and undamaged DNA (Yang, 2005).

After Pol V is induced its activity is tightly controlled and targeted. Pol V consists of two subunits, UmuC and UmuD; the polymerase subunit, UmuC, is active only when complexed with a dimer of the truncated form of UmuD, UmuD′. In addition, lesion bypass catalysed by Pol V requires RecA and the DNA polymerase III (Pol III) processivity unit, the β clamp (Patel et al., 2010). In the absence of DNA damage, Pol V levels are low [< 15 molecules cell−1 (Woodgate and Ennis, 1991)], and they contribute little to spontaneous mutation rates [e.g. see Wagner and Nohmi (2000); Kuban et al. (2006)].

In contrast, Pol IV does not require cofactors to synthesize DNA, although its processivity is dramatically enhanced by association with the β clamp (Wagner et al., 2000). Pol IV is also strikingly abundant in normally growing cells [∼250 molecules cell−1 (Kim et al., 2001)], which suggests that it has a role during normal replication. One such role may be replication restart; like Pol II (Becherel and Fuchs, 2001), Pol IV is able to elongate from misaligned primer/template termini (Wagner et al., 1999). Thus, Pol IV and Pol II may compete for access to stalled replication forks.

The relatively high cellular levels of Pol IV also raise an important question regarding the control of its mutagenic activity. Pol IV contributes little to spontaneous mutations on the chromosome of normally growing cells (Strauss et al., 2000; McKenzie et al., 2003; Kuban et al., 2004; Wolff et al., 2004). However, Pol IV is responsible for 50–80% of both adaptive and non-adaptive mutations that occur in stationary-phase cells (Foster, 2000; McKenzie et al., 2001; Tompkins et al., 2003). In addition, overproducing Pol IV in growing cells increases spontaneous mutation rates as much as 100-fold (Kim et al., 1997; Wagner and Nohmi, 2000; Kuban et al., 2004; Stumpf and Foster, 2005). Thus, it seems likely that in normally growing cells the mutagenic activity of Pol IV is tightly regulated, but that in stationary-phase cells, or when Pol IV is overproduced, this regulation is circumvented.

Both the expression and the activity of Pol IV are controlled by a number of regulatory systems within the cell. These include the SOS response (Kenyon and Walker, 1980; Courcelle et al., 2001), the RpoS-dependent general stress response (Layton and Foster, 2003; Storvik and Foster, 2010), the heat-shock response (Layton and Foster, 2005) and the Ppk-mediated nutrient limitation response (Stumpf and Foster, 2005). These regulatory pathways ensure that Pol IV's mutagenic activity is highest when cells are under stress. To date, however, little is known about the actual molecular mechanisms by which Pol IV's activity is controlled. It has recently been proposed that the activity of Pol IV is regulated by physical interactions with UmuD and RecA (Godoy et al., 2007) as well as the transcription factor NusA (Cohen et al., 2009). Here, we report a new Pol IV interacting partner, the DNA helicase Rep.

Rep is a member of the ubiquitous and highly conserved superfamily 1 (SF1) class of DNA helicases and has 3′–5′ polarity (Lohman and Bjornson, 1996). Originally identified as required for ΦX174 replication (Arai and Kornberg, 1981), Rep also has at least two important functions for chromosomal replication in E. coli. The first of these is to stabilize the replication fork: cells lacking Rep have a 50–60% reduction in the rate of replication fork movement (Lane and Denhardt, 1975). Evidence suggests that in the absence of Rep, the replication fork stalls more frequently, resulting in its collapse and the formation of double-stranded breaks (DSBs) (Michel et al., 1997). Rep's primary role in preventing fork stalling appears to be removal of blocking nucleoprotein complexes, a task that the replicative helicase, DnaB, is apparently unable to accomplish alone (Guy et al., 2009; Boubakri et al., 2010). Critical to this unblocking role is the interaction of Rep with DnaB, which serves to increase the local concentration of Rep at the replication fork (Atkinson et al., 2011).

The second function of Rep is to participate in the restart of replication forks after they collapse (Sandler, 2000). Replication restart at sites other than oriC, the chromosomal replication origin, requires the primosome proteins PriA, PriB, PriC and DnaT (Marians, 2004). Rep participates in a PriA-independent restart pathway in which PriC is thought to co-ordinate the loading of Rep and DnaB onto stalled fork structures. Rep may serve to remove the Okazaki fragments from the lagging-strand template to create a platform for DnaB loading (Heller and Marians, 2007).

In this report, we demonstrate that Rep associates physically with Pol IV and that this association enhances Rep's DNA helicase activity. In addition, the interaction with Rep stimulates Pol IV's DNA polymerization activity, and this stimulation appears to require the Pol IV β clamp-binding motif. We also demonstrate that the effects we see in vitro are biologically relevant as mutations in Rep affect Pol IV's mutagenic activity in vivo.

Results

Rep is required for the maximal levels of adaptive mutation in E. coli strain FC40

To find proteins that interact with Pol IV, we used the yeast two-hybrid system to screen a library of random fragments of the E. coli chromosome. A DNA fragment expressing a truncated region of the Rep DNA helicase, amino acids 110 to 254 (of 673), tested positive. The truncated protein fragment consists of half of domain 1A and all of domain 1B of the protein; it retains the conserved SF1 helicase motifs II (ATPase Walker B) and III (ssDNA interaction) (Korolev et al., 1997; Lee and Yang, 2006). To determine the functional significance of the interaction between Pol IV and Rep, we assayed rep mutant strains for their ability to revert the Lac- phenotype of E. coli strain FC40. The F′128 episome carried by this strain has a lacI-lacZ fusion that is inactive because of a +1 frameshift mutation; Lac+ revertants accumulate during incubation on lactose minimal medium, a process known as adaptive mutation (Cairns and Foster, 1991). As mentioned above, Pol IV is responsible for 50–80% of the Lac+ adaptive mutations (Foster, 2000; McKenzie et al., 2001). As shown in Fig. 1A, adaptive mutation was reduced about threefold in a Δrep mutant strain.

Figure 1.

Adaptive mutation to Lac+ and levels of Pol IV in wild-type and rep mutant strains. The graphs show the accumulation of Lac+ mutants among Lac- cells incubated on lactose minimal medium. Data are the cumulative numbers of Lac+ colonies appearing each day from days 2–5 divided by the number of viable Lac- cells on the plate 2 days earlier; no loss of viability of any strain was detected over the course of the experiment. Each point is the mean of three cultures ± SEM (error bars are smaller than the symbols). A. The accumulation of Lac+ mutants in dinB+ strains. Wild-type = FC40, Δrep = PFG528, repK28I = PFB842. B. The accumulation of Lac+ mutants in ΔdinB strains. ΔdinB = PFB243, ΔdinBΔrep = PFB895, ΔdinB repK28I = PFB846. C. Western blot visualized with antibodies to Pol IV and GroEL (as a loading control) showing the relative levels of Pol IV in cells grown to stationary phase in glycerol minimal medium. Relative intensities were calculated as the ratio of the Pol IV band to the loading control for each strain divided by the same ratio for the wild-type strain.

To further determine if the Rep activity is important for adaptive mutation, we made a mutation in Rep (repK28I) that changes a conserved lysine in the putative helicase active site to isoleucine. In UvrD, a close homologue of Rep, this lysine (K35) is in the Walker A motif, and changing it to isoleucine abolishes the ATPase and helicase activities, but not the DNA binding properties, of UvrD (Maluf et al., 2003). However, we found that in the RepK28I mutant protein all three activities are abolished (Fig. S1). As shown in Fig. 1A, the repK28I mutant allele conferred a more severe phenotype for adaptive mutation than did the Δrep allele, reducing adaptive mutation about sevenfold compared with the wild-type levels.

To determine whether Rep and Pol IV are in the same pathway for adaptive mutation, we compared a ΔdinBΔrep double mutant strain with a ΔdinB mutant strain. As shown in Fig. 1B, the ΔdinBΔrep double mutant strain showed no further reduction in adaptive mutation compared with the ΔdinB mutant strain, indicating that Rep acts in the same pathway with Pol IV (although comparing Fig. 1A and B reveals that a component of Pol IV-dependent adaptive mutation is Rep-independent). Interestingly, the repK28I mutant allele resulted in a further decrease in adaptive mutation, suggesting that RepK28I inhibits a Pol IV-independent pathway for adaptive mutation.

Decreased adaptive mutation could result from a reduction either in Pol IV's mutagenic activity or in the amount of Pol IV in the cell. To distinguish between these two, we performed a Western blot with wild-type and rep mutant strains grown to saturation in minimal medium. As shown in Fig. 1C, Pol IV levels in the rep mutant strains were similar to that of the wild-type strain. Thus, reduction in Pol IV's activity, not in its amount, accounts for the decrease in adaptive mutation seen in rep mutant strains.

Mutations in rep increase growth-dependent spontaneous mutation

We measure growth-dependent spontaneous mutation in a strain, FC722, which contains a mutational target similar to the target used for adaptive mutation. The F′128 episome carried by FC722 contains a tetA gene that is inactive because of a +1 frameshift mutation (Foster, 1997). Spontaneous growth-dependent mutation rates are determined by measuring the rate of reversion from tetracycline sensitivity (TcS) to tetracycline resistance (TcR) using classical fluctuation tests (Stumpf and Foster, 2005; Storvik and Foster, 2010). About 70% of the spontaneous growth-dependent TcR mutations are due to Pol IV (Williams and Foster, 2007; Storvik and Foster, 2011).

FC722 derivatives carrying the Δrep mutant allele or the repK28I mutant allele had elevated spontaneous mutation rates compared with the wild-type strain (Fig. 2A). Previous studies found that the SOS response was partially induced in rep mutant strains (Ossanna and Mount, 1989). SOS induction in the absence of DNA damage can increase mutation rates (George et al., 1975; Witkin, 1976), and both of E. coli's inducible error-prone DNA polymerases, Pol IV and Pol V, can contribute to this increase (Kuban et al., 2006). In a separate experiment we found that the mutation rate of a ΔrepΔdinB double mutant strain was only 20% that of the Δrep mutant strain (data not shown). Thus, a low level of SOS induction in the rep mutant strains could explain the elevated spontaneous mutation rates of these strains.

Figure 2.

Growth-dependent spontaneous mutation rate and SOS expression in wild-type and rep mutant strains. A. Growth-dependent mutation rate to TcR of the wild-type strain (FC722), the isogenic Δrep mutant strain (PFB902) and the isogenic repK28I mutant strain (PFB894). Error bars are 95% confidence limits. B. Expression of sulAp-gfp among cells of the wild-type strain and rep mutant strains. Wild-type = PFB903, Δrep = PFB905, repK28I = PFB904. The bottom panels show sulAp-gfp as green, and the top panels are sulAp-gfp images merged with phase contrast images that are pseudocolored red.

The sulA gene is tightly regulated by the LexA repressor and is induced 100-fold during the SOS response (Friedberg et al., 2006); the sulA promoter (sulAp) fused to gfp is a useful reporter for SOS induction (McCool et al., 2004). As shown in Fig. 2B, sulAp-gfp was upregulated in exponentially growing cells of the Δrep and repK28I mutant strains. Most of the cells of the rep mutant strains showed some level of increased sulAp-gfp expression relative to the wild-type strain; 14% (74/537) of the cells of the Δrep mutant strain and 19% (71/380) of the cells of the repK28I mutant strain, versus < 0.3% (0/294) cells of the wild-type strain, expressed sulAp-gfp at the maximum detected intensity (i.e. the brightest cells in Fig. 2B). This pattern of continuous distribution in the levels of expression of sulAp-gfp, with about 20% of the cells displaying maximum expression, is typical of other strains mutant in replication, recombination and DNA repair functions (McCool et al., 2004). Taken together, these results suggest that the elevated growth-dependent spontaneous mutation rates of the rep mutant strains are due to induction of the SOS response and not to any functional role of Rep in the mutagenic process. However, because adaptive mutation was decreased in the rep mutant strains (Fig. 1A), induction of SOS does not explain the effect of loss of Rep on adaptive mutation.

Translesion synthesis activity of Pol IV is unaffected in Rep mutants

In vitro, Pol IV efficiently bypasses certain bulky lesions at the N2 position of guanine residues (Neeley et al., 2007), and strains lacking Pol IV are sensitive to chemicals, such as 4-nitroquinoline-1-oxide (NQO), which produce these lesions (Jarosz et al., 2006). Thus, Pol IV's translesion synthesis activity can be measured by testing cell survival on medium containing NQO. As shown in Fig. 3, whether or not Pol IV was also present in the strain, neither the Δrep nor the repK28I mutant allele affected NQO resistance. Thus, loss of Rep does not appear to affect the ability of Pol IV to bypass NQO-induced lesions in vivo.

Figure 3.

NQO sensitivity of strains mutant for dinB and rep compared with the wild-type strain. Spot tests were performed as described in the Experimental procedures; only the informative dilutions are shown. Wild type = FC40; ΔdinB = PFB243; Δrep = PFG528; ΔdinBΔrep = PFB895; repK28I = PFB842; ΔdinB repK28I = PFB846.

Rep interacts with Pol IV in vitro, and this interaction involves Pol IV's β clamp-binding motif

Our yeast two-hybrid results predicted an interaction between Rep and Pol IV. To confirm this interaction, a Rep-GST fusion protein was purified and used in pull-down assays on glutathione-sepharose resin, to which the GST moiety binds. Although this fusion protein does not complement a rep mutant allele for adaptive mutation (see Fig. S2), it nonetheless was able to pull down potential interacting partners. We first tested full-length untagged Pol IV and Pol IV with a Strep-FLAG-tag on its N-terminus (S-FLAG Pol IV) (these purified proteins are shown in Fig. 4A, lanes 2, 3 and 4; that S-FLAG Pol IV compliments a dinB mutant allele for adaptive mutation, and thus has in vivo activity, is shown in Fig. S2); GST protein, which should not interact with the polymerases, was used as a control (Fig. 4A, lanes 8–10). Both tagged and untagged Pol IV proteins were pulled down with Rep-GST (Fig. 4A, lanes 11 and 12), indicating that there is a stable protein–protein interaction between Rep and Pol IV. However, about half as much S-FLAG Pol IV as Pol IV was pulled down (Fig. 4B), suggesting that, although the tagged protein has in vivo activity (Fig. S2), the N-terminal tag interferes with binding to Rep to some extent.

Figure 4.

Pull-down assays with Rep-GST and Pol IV, Pol IVΔC6, Pol II and Pol IIIα. A. A Coomassie blue-stained SDS-polyacrylamide gel is shown. Lanes 2–7: the purified proteins used in the pull-down assays; lanes 8–10: negative control assays with GST protein purified on glutathione-sepharose resin; lanes 11 to 15: pull-down assays with Rep-GST purified on glutathione-sepharose medium. Assays included the following polymerases: Pol IV (lanes 8, 11), S-FLAG-Pol IV (lane 12), S-FLAG-Pol IVΔC6 (lane 13), FLAG-Pol II (lanes 9, 14) and FLAG-Pol IIIα (lanes 10, 15). Rep-GST or GST and polymerases were added at equal mass ratios in these assays (see Experimental procedures). B. A density scan of the polymerase bands in lanes 11–14 of the Coomassie blue-stained SDS-polyacrylamide gel shown in A: Pol IV, lane 11; S-FLAG Pol IV, lane 12; S-FLAG-Pol IVΔC6, lane 13; FLAG-Pol II, lane 14 and FLAG-Pol IIIα, lane 15. The density of each polymerase band was divided by the density of the corresponding Rep band, and then each of these ratios was divided by the same ratio for Pol IV and multiplied by 100 to give the % of the Pol IV value. C. A Western blot of Rep-GST pull-down assays visualized with anti-FLAG antibody. Lanes 1, 3, 5 and 7 show the negative control assays with GST protein. Pull-down assays included the following polymerases: S-FLAG-Pol IV, lanes 1, 2; S-FLAG-Pol IVΔC6, lanes 3, 4; FLAG-Pol II, lanes 5, 6; and FLAG-Pol IIIα, lanes 7, 8. D. Density scans of lanes 2, 4, 6 and 8 of the Western blot shown in C: S-FLAG Pol IV, lane 2; S-FLAG-Pol IVΔC, lane 4; FLAG-Pol II, lane 6 and FLAG-Pol IIIα, lane 8. The density of each band was divided by the density of the S-FLAG Pol IV band, and each of these values was multiplied by 100 to give the % of the S-FLAG Pol IV value (A shorter exposure than shown was used for the density scan).

Pol IV contacts the β clamp processivity factor via a 6-amino-acid β clamp-binding motif located at Pol IV's extreme C-terminus, and this motif is required for full Pol IV activity (Lenne-Samuel et al., 2002; Bunting et al., 2003) (Fig. S2). To determine if this motif is important for Pol IV's interaction with Rep, we tested whether Rep-GST could pull down an N-terminal S-FLAG-tagged Pol IV deleted for its C-terminal 6 amino acids (S-FLAG-Pol IVΔC6) (the purified protein is shown in Fig. 4A, lane 5; that this mutant protein does not compliment a dinB mutant allele for adaptive mutation is shown in Fig. S2). As shown in Fig. 4A, lane 13, and Fig. 4B, compared with full-length Pol IV, less of the truncated Pol IV was pulled down with Rep-GST. The greater sensitivity of a Western blot revealed that Rep-GST pulled down about 75% as much S-FLAG-Pol IVΔC6 as S-FLAG-Pol IV (Fig. 4C, lanes 2 and 4; Fig. 4D). These results suggest that the β clamp-binding motif of Pol IV contributes to its interaction with Rep, but one or more other sites are also involved. Alternatively, loss of the C-terminal 6 amino acids may simply alter the structure of Pol IV so that its interaction with Rep is weakened.

Rep also interacts with DNA Pol II and weakly with DNA Pol IIIα

To determine if Rep can interact with other DNA polymerases, we tested purified FLAG-tagged Pol II (FLAG-Pol II) and purified FLAG-tagged α subunit (the polymerase subunit) of Pol III (FLAG-Pol IIIα) in pull-down assays with Rep-GST (these purified proteins are shown in Fig. 4A, lanes 6 and 7; neither FLAG-Pol II nor FLAG-Pol IIIα was tested for in vivo activity). As detected by Coomassie blue staining, very little FLAG-Pol II and FLAG-Pol IIIα were pulled down with Rep-GST (Fig. 4A, lanes 14 and 15; Fig. 4B). The more sensitive Western blot revealed that Rep-GST pulled down about 30% and 4% of the amounts of FLAG-Pol II and FLAG-Pol IIIα, respectively, as S-FLAG-Pol IV (Fig. 4C, lanes 6 and 8; Fig. 4D). This assay was repeated using equimolar, rather than equal mass, amounts of protein, with the same results (Fig. S3). Thus, Pol II and Pol IIIα interact with Rep-GST, but apparently with reduced affinity compared with Pol IV.

Pol IV stimulates Rep helicase activity independently of its β clamp-binding motif

To determine if the interaction between Pol IV and Rep is functional, we tested Pol IV for its ability to stimulate Rep helicase activity. Helicase activity was monitored by using purified untagged Rep in a molecular beacon-based helicase assay (MBHA) (Belon and Frick, 2008). Briefly, an oligonucleotide capable of forming an intramolecular hairpin modified with a fluorescent moiety (Cy5) at the 5′ terminus and a fluorescence quencher (BHQ) at the 3′ terminus is annealed to a complementary single-stranded oligonucleotide with a 3′ overhang. Upon unwinding, the modified oligonucleotide forms the hairpin, bringing Cy5 and BHQ into proximity and quenching fluorescence. Helicase activity is measured by the decay of the fluorescence signal. The great advantages of this assay are: (i) because one product of unwinding forms a hairpin, the reaction is essentially irreversible, precluding reannealing of the unwound strands, and (ii) the strand on which the helicase travels is not modified, thus minimizing the potential of interference with the reaction (Belon and Frick, 2008).

As shown in Fig. 5A, when purified S-FLAG-Pol IV was included with untagged Rep in the MBHA, Rep helicase activity was enhanced (Fig. 5A; compare diamonds with circles); based on the slopes of the linear parts of the curves, this enhancement was approximately threefold. On its own, Pol IV did not exhibit any helicase activity (Fig. 5A, squares). The stimulation of Rep helicase activity by Pol IV was confirmed using a conventional gel-based unwinding assay; consistent with the fluorescence-based helicase assay, Pol IV stimulated Rep helicase activity approximately twofold (Fig. 5B and C). Including SSB in this assay had no effect (Fig. S4).

Figure 5.

The effect of Pol IV and Pol IVΔC6 on Rep helicase activity. Rep helicase activity was measured using a MBHA (A) or a gel-based radioactive unwinding assay (B) to detect the ssDNA products of double-stranded DNA unwinding (see Experimental procedures). Data points and error bars in the fluorescence-based assays are the means and standard deviations of triplicate reactions. A. A comparison of Rep helicase activity stimulated by S-FLAG-Pol IV (●) and S-FLAG-Pol IVΔC6 (▴) using MBHA. Rep in the absence of polymerase (◆) and S-FLAG-Pol IV without Rep (inline image) are also shown. The reaction was initiated by adding ATP to 200 nM of untagged Rep at the time point indicated in the figure. The polymerases were present at 1000 nM. Fluorescence was measured in arbitrary units (A.U.); loss of fluorescence indicates helicase activity. B. An autoradiogram of an SDS-polyacrylamide gel showing the radiolabelled products of DNA unwinding. S-FLAG-Pol IV stimulation of Rep DNA helicase activity (left lanes) is compared with Rep in the absence of polymerase (right lanes). Untagged Rep was at 20 nM, and S-FLAG-Pol IV was at 100 nM in these assays. The reaction products are indicated to the left of the gel; decreases in the upper band and increases in the lower band indicate helicase activity. C. A density scan of the bands in the gel shown in B: Rep plus S-FLAG-Pol IV (●) and Rep (▴). The density of each lower band was divided by the sum of the densities of the corresponding upper and lower bands.

Because a Pol IV mutant lacking the β clamp-binding motif appears to have reduced affinity for Rep (Fig. 4), we tested whether the β clamp-binding motif is necessary for Pol IV to stimulate Rep helicase activity. As shown in Fig. 5A, S-FLAG-Pol IVΔC6 stimulated Rep helicase activity, but, based on the slopes of the linear parts of the curves, was about 30% less active in stimulating Rep than was S-FLAG-Pol IV (compare triangles with circles). This difference, although small, was reproducible (data not shown). Thus, while the Pol IV β clamp-binding motif is not required to stimulate Rep helicase activity, it is required for maximal stimulation.

Exonuclease-defective Pol II, but not Pol IIIα, stimulates Rep helicase activity

Because both Pol II and Pol III are involved in adaptive mutation (Foster et al., 1995; Foster, 2000; Hastings et al., 2010), we tested whether Pol II and/or the α subunit of Pol III can stimulate Rep helicase activity. To prevent the DNA substrate from being degraded during the experiment, we used an exonuclease-defective mutant of Pol II that has previously been characterized (Foster et al., 1995), although we did not test the fusion protein (FLAG-Pol II Exo-) for in vivo activity. As shown in Fig. 6, Pol II Exo- stimulated Rep to approximately the same degree as Pol IV (compare open squares and open circles with closed diamonds), but Pol IIIα had no stimulatory effect (compare closed squares with closed diamonds). This negative result also establishes that the FLAG tag itself has no stimulatory effect on Rep helicase activity.

Figure 6.

The effect of Pol IV, Pol II Exo- and Pol IIIα on Rep helicase activity. Rep helicase activity was measured using a MBHA; data points and error bars are the means and standard deviations of triplicate reactions. Fluorescence was measured in arbitrary units (A.U.); loss of fluorescence indicates helicase activity. The reaction was initiated by adding ATP to 200 nM of untagged Rep at the time point indicated in the figure. The following polymerases were present at 1000 nM: S-FLAG-Pol IV (○), FLAG-Pol II Exo- (□), and FLAG-Pol IIIα (inline image). FLAG-Pol II Exo- (●) and FLAG-Pol IIIα (▴) in the absence of Rep, and Rep (◆) in the absence of polymerase are also shown.

Rep stimulates Pol IV polymerase activity, and this stimulation requires the β clamp-binding motif

To determine whether Rep can modulate Pol IV's polymerase activity, we tested the ability of S-FLAG-Pol IV and S-FLAG-Pol IVΔC6 to extend a radiolabelled primer in the presence of Rep and ATP. Consistent with previous reports (Tang et al., 2000; Lenne-Samuel et al., 2002), Pol IV exhibited only weak polymerase activity on its own (Fig. 7, lane 2), but its activity was stimulated by SSB, resulting in about 6% of the primers being extended to full-length product (Fig. 7, lane 3). The addition of Rep further stimulated Pol IV's polymerase activity, resulting in 30% of the primers being fully extended (Fig. 7, lane 4). The same result was obtained with untagged Pol IV (Fig. S5). Even in the absence of SSB, Rep was able to stimulate S-FLAG-Pol IV, resulting in about 1% of the primers being extended to full-length product (Fig. 7, lane 5). This last result suggests that although SSB is required for full Pol IV activity, Rep can stimulate Pol IV activity in its presence or in its absence.

Figure 7.

The effect of Rep on the polymerization activity of S-FLAG-Pol IV and S-FLAG-Pol IVΔC6. An autoradiogram of a polyacrylamide-urea gel with the radiolabelled products of DNA primer-extension reactions is shown (see Experimental procedures). Lane 1: Rep and SSB; lanes 2 and 6: S-FLAG-Pol IV and S-FLAG-Pol IVΔC6 activity, respectively, without SSB; lanes 3 and 7: S-FLAG-Pol IV and S-FLAG-Pol IVΔC6 activity, respectively, in the presence of SSB; lanes 4 and 8: S-FLAG-Pol IV and S-FLAG-Pol IVΔC6 activity, respectively, in the presence of Rep and SSB; lanes 5 and 9: S-FLAG-Pol IV and S-FLAG-Pol IVΔC6 activity, respectively, in the presence of Rep without SSB. The concentrations of the polymerases were 20 nM in lanes 2–9; lanes 10–13 correspond to lanes 6–9 but with the concentration of S-FLAG-Pol IVΔC6 increased to 30 nM, at which concentration S-FLAG-Pol IVΔC6 displays similar basal activity to S-FLAG-Pol IV at a concentration of 20 nM (e.g. compare lanes 3 and 11). Where indicated the reactions included 100 nM of untagged Rep and/or 1000 nM SSB protein.

While the Pol IV β clamp-binding motif only slightly enhanced the stimulation of Rep's helicase activity by Pol IV (Fig. 5A), it was possible that this motif was required for Rep to stimulate Pol IV's polymerase activity. Like full-length S-FLAG-Pol IV, the polymerase activity of S-FLAG-Pol IVΔC6 was stimulated upon addition of SSB (Fig. 7, lanes 6 and 7); however, unlike full-length S-FLAG-Pol IV, Rep failed to stimulate S-FLAG-Pol IVΔC6 regardless of whether or not SSB was present (Fig. 7, lanes 8 and 9). Thus, the β clamp-binding motif is required for Rep to stimulate Pol IV's polymerase activity. Note that we observed reduced polymerase activity with S-FLAG-Pol IVΔC6 compared with the full-length S-FLAG-Pol IV (compare Fig. 7, lanes 2 and 6 and lanes 3 and 7). However, when the concentration of S-FLAG-Pol IVΔC6 was increased so that its polymerase activity was more comparable with that of full-length S-FLAG-Pol IV (Fig. 7, lanes 10 and 11), Rep was still unable to stimulate its activity (Fig. 7, lanes 12 and 13). Nonetheless, the low activity of S-FLAG-Pol IVΔC6 complicates the interpretation of its insensitivity to stimulation by Rep.

Stimulation of Pol IV polymerase activity does not require Rep helicase activity

As mentioned above, the RepK28I mutant protein is defective for ATP hydrolysis, DNA binding and helicase activity (Fig. S1). Nonetheless, as shown in Fig. 8, Pol IV interacted with RepK28I-GST in a pull-down assay (Fig. 8A, lane 3). Based on a density scan of the Western blot, about 60% of the amount of Pol IV was pulled down with RepK28I as was pulled down with wild-type Rep (Fig. 8B). Pol II also interacted with RepK28I (Fig. 8A, lane 5), but, as seen with wild-type Rep (Fig. 4), only about 40% as much Pol II was pulled down as was Pol IV (Fig. 8B). We detected no interaction between Pol IIIα and RepK28I (Fig. 8A, lane 7; Fig. 8B). RepK28I was also able to stimulate Pol IV in a primer-extension assay. As shown in Fig. 9, RepK28I was as active as wild-type Rep in enhancing Pol IV's polymerase activity; the addition of either protein resulted in about a 15-fold increase in full-length product (compare lanes 2 and 3). These results suggest that neither helicase activity nor DNA-binding activity is required for Rep to be able to stimulate DNA polymerization by Pol IV.

Figure 8.

Pull-down assays of Pol IV, Pol II and Pol IIIα with RepK28I-GST. A. A Western blot of RepK28I-GST pull-down assays visualized with anti-FLAG antibody. Lanes 1, 4 and 6, the negative control assays with GST protein purified on glutathione-sepharose resin; lanes 3, 5 and 7, pull-down assays with RepK28I-GST purified on glutathione-sepharose resin. Assays included the following polymerases: S-FLAG-Pol IV, lanes 1, 2 and 3; FLAG-Pol II, lanes 4 and 5; and FLAG-Pol IIIα, lanes 6 and 7. FLAG-Pol IV pulled down with Rep-GST is shown for comparison (lane 2). RepK28I-GST, Rep-GST, or GST and polymerases were added at equal mass ratios in these assays (see Experimental procedures). B. A density scan of polymerase bands in lanes 2, 3, 5 and 7 of the Western blot shown in A. S-FLAG Pol IV, lanes 2, 3; FLAG-Pol II, lane 5; and FLAG-Pol IIIα, lane 7. The density of each band was divided by the density of the band of S-FLAG Pol IV pulled down with Rep-GST (lane 2), and then each of these values was multiplied by 100 to give the % of the S-FLAG Pol IV value. (A shorter exposure than shown was used for the density scan.)

Figure 9.

The effect of RepK28I on the polymerization activity of S-FLAG-Pol IV. An autoradiogram of a polyacrylamide-urea gel with the radiolabelled products of DNA primer-extension reactions is shown (see Experimental procedures). Lane 1: S-FLAG-Pol IV; lane 2; Rep and S-FLAG-Pol IV; lane 3, RepK28I and S-FLAG-Pol IV. All reactions included 20 mM DNA template, 1250 nM SSB protein, 20 nM S-FLAG-Pol IV, and 100 nM untagged Rep or RepK28I.

Discussion

In this report we identify a new binding partner for Pol IV, Rep DNA helicase, which stimulates Pol IV's polymerization activity in vitro. Addition of Rep to Pol IV primer-extension assays increased both the amount of primer extended and the yield of full-length product. Further experimentation is required to determine if this stimulation is due to an increase in Pol IV's affinity for nucleotides, in its rate of nucleotide incorporation, in its processivity, or all three.

Pol IV has a number of biological roles. In addition to its TLS activity, Pol IV is postulated to play a major role in restarting stalled and collapsed replication forks (Goodman, 2002; Lovett, 2006), and it is responsible for generating adaptive and non-adaptive mutations in stationary-phase cells (Foster, 2000; McKenzie et al., 2001; Tompkins et al., 2003). We sought to determine in which of these processes Pol IV and Rep might function together.

As mentioned above, Pol IV can bypass DNA lesions produced by NQO at the N2 positions of guanines; thus, sensitivity to NQO is an assay for Pol IV's TLS activity in vivo (Jarosz et al., 2006). However, strains defective for Rep were not any less resistant to NQO than was the wild-type strain (Fig. 3); thus, Rep does not appear to play an essential role in helping Pol IV to bypass NQO-produced DNA adducts. In contrast, Rep enhanced Pol IV's ability to produce adaptive mutations about threefold (Fig. 1). These results support and extend previous findings (Godoy et al., 2007; Wagner et al., 2009) that Pol IV's TLS activity is genetically and mechanistically distinct from its role in generating spontaneous (untargeted) mutations.

By a current model of adaptive mutation (Foster, 2007), the collapse of replication forks produces DSBs that are repaired by E. coli's homologous recombination pathway for DSB repair. DNA synthesis is then re-initiated by an oriC-independent primosome that utilizes either Pol IV or Pol II. The majority of errors that give rise to adaptive mutations occur when Pol IV out-competes Pol II for access to the primer-template DNA. Thus, by this model, Pol IV's role in adaptive mutation is essentially the same as its role in replication restart. We found that strains with mutant alleles of rep were deficient for adaptive mutation (Fig. 1A) and that these phenotypes were not due to a decrease in the amount of Pol IV (Fig. 1C). A null allele of dinB was epistatic to a null allele of rep, indicating that Rep is in a DinB-dependent pathway for adaptive mutation (Fig. 1B). There is, in addition, a Pol IV-dependent but Rep-independent component of adaptive mutation (compare Fig. 1A and B). Surprisingly, the repK28I mutant allele was more inhibitory for adaptive mutation than was the rep null allele, reducing even the residual level of Pol IV-independent adaptive mutation (Fig. 1B). Thus, it appears that the RepK28I mutant protein can inhibit adaptive mutations that are due to one or more other polymerases (for example, Pol II, which interacts with RepK28I; Fig. 8), although these mutations do not normally require Rep for their production. Nonetheless, our results indicate that during the process that produces adaptive mutations, Rep helicase normally helps to enhance or target Pol IV's mutagenic activity. This finding presents a new biological role for Rep.

Our results also indicate a functional interaction between Rep and Pol IV and, to a lesser degree, between Rep and Pol II. Rep interacts physically with both polymerases (Fig. 4), and its helicase activity is stimulated by these interactions (Figs 5 and 6). Rep also interacts weakly with Pol IIIα, but this interaction does not stimulate Rep's helicase activity and thus may not be functionally significant (Fig. 6). Surprisingly, the interaction between Rep and Pol IV is reciprocal – Pol IV's polymerase activity is enhanced by its interaction with Rep (Fig. 7). Because the mutant RepK28I protein likewise stimulates Pol IV (Fig. 9), this stimulation does not require Rep's helicase, ATPase, or DNA-binding activities. Yet, a repK28I mutant strain is defective for Pol IV-dependent adaptive mutation (Fig. 1). Thus, the ability of Rep to stimulate Pol IV's polymerase activity is not sufficient for adaptive mutation, and Rep must play another role in promoting Pol IV-dependent adaptive mutation. Whether Rep's stimulation of Pol IV polymerase activity, while not sufficient, might be necessary for adaptive mutation is not addressed by our data. For example, Rep's helicase activity may be needed to produce a substrate for Pol IV extension, which is then stimulated by Rep independently of its helicase activity. Our results also leave open the possibility that the stimulatory activity of Rep may be important for Pol IV functions that we have not yet investigated.

To our knowledge, this is the first report of an interaction between Rep and a DNA polymerase. Although Rep is required for the replication of bacteriophage ΦX174, it interacts with the initiator nuclease, gpA, not with the replicase (Duguet et al., 1979; Arai et al., 1981; Kornberg and Baker, 2005). Other functional helicase–polymerase interactions have been well characterized in a number of model systems. These interactions are essential for a variety of cellular processes including optimization of replication fork progression and toleration of DNA damage. For example, highly processive replication by the bacteriophage T7 replisome depends on dynamic interactions between DNA polymerase gp5 and DNA helicase gp4, a member of the hexameric DnaB-like family of helicases (Hamdan et al., 2007). [In contrast, DnaB itself contacts not the polymerase, but the tau subunit of the replication complex (Kim et al., 1996).] Additionally, the activities of gp4 and gp5 are interdependent as DNA polymerization by gp5 increases the rate of unwinding by gp4 (Stano et al., 2005). The eukaryotic WRN helicase, a member of the SFII-RecQ DNA helicase family, interacts with the TLS polymerases Polη, Polκ and Polι, increasing their polymerase and mutagenic activities (Kamath-Loeb et al., 2007). WRN also interacts with the highly processive and accurate Polδ, enhancing both its polymerase activity (Kamath-Loeb et al., 2000) and its processivity over DNA sequences that are difficult to replicate (Shah et al., 2010). These sequences in eukaryotes, known as common fragile sites, are a major source of genomic instability that can lead to cancer (Smith et al., 2007). Thus, WRN may serve as a model for Rep as it plays an important role in maintaining genomic integrity by interacting with both error-prone and accurate polymerases. Interestingly, WRN's helicase activity is not required for it to be able to stimulate the polymerase activities of Polδ or Polηin vitro (Kamath-Loeb et al., 2000; Shah et al., 2010), just as we have found for Rep's stimulation of Pol IV's polymerase activity (Fig. 9).

As mentioned above, Rep has two known roles in DNA replication: (i) preventing the collapse of replication forks by removing impediments to replication (Guy et al., 2009; Boubakri et al., 2010; Atkinson et al., 2011) and (ii) helping collapsed replication forks to restart by facilitating the loading of the replicative helicase, DnaB (Sandler, 2000; Heller and Marians, 2005; 2007). Either or both of these functions of Rep could promote the production of adaptive mutations by Pol IV. As discussed above, adaptive mutations are postulated to be produced when Pol IV initiates DNA synthesis from recombinational intermediates during DSB repair of collapsed replication forks. Rep participates in an accessory pathway for restarting replication forks that is PriA-independent but PriC-dependent. It is hypothesized that in this pathway Rep, together with PriC, recognizes a stalled fork; Rep removes the nascent lagging strand, allowing PriC to load DnaB (Sandler, 2000; Heller and Marians, 2005; 2007). Similarly, Rep might assist Pol IV and Pol II in extending from recombination intermediates by recognizing the substrate and helping to load the polymerases; as Rep promotes adaptive mutations (Fig. 1), in this process it would presumably favour error-prone Pol IV. Alternatively, Pol IV may load Rep and stimulate it to remove nascent lagging strands ahead of DNA extension. In either case, Rep would be the intermediate that would eventually allow the replicative helicase, DnaB, to load and nucleate the assembly of the full replication complex.

A third possibility is that a Pol IV-Rep complex might be the active form of Pol IV required in processes such as adaptive mutation. Like all of E. coli's DNA polymerases, Pol IV contacts the processivity factor, the β clamp, by a β clamp-binding motif located at its extreme C-terminus (Lenne-Samuel et al., 2002; Bunting et al., 2003). Pol IV has an additional interaction with β via its ‘little-finger’ domain that appears to hold it in an inactive position (Bunting et al., 2003; Wagner et al., 2009). The β clamp stimulates Pol IV's activity and processivity in vitro (Wagner et al., 2000), and loss of the β clamp-binding motif severely diminishes Pol IV's ability to produce spontaneous mutations and to perform both mutagenic and error-free TLS in vivo (Becherel et al., 2002; Lenne-Samuel et al., 2002; Fig. S2). When acting as a dimer, Rep has helicase activity, but its processivity is limited (Cheng et al., 2001). However, as a monomer Rep can translocate 3′ to 5′ along ssDNA at a rate of 298 nucleotides per second and with a processivity of 700 nucleotides (Brendza et al., 2005). Given this relatively high processivity and the fact that Rep contacts Pol IV, in part, via the β clamp-binding motif (Fig. 4), an attractive hypothesis is that Rep is an alternative processivity factor that enhances Pol IV's processivity in situations when the β clamp may be absent, such as during re-establishment of stalled replication forks. However, the fact that Rep's DNA binding activity is not required for stimulation of Pol IV's polymerase activity (Fig. 9) argues against this hypothesis. Thus, we favour the alternative hypothesis that Rep's role is to clear the DNA of obstructions, allowing Pol IV extension. Further studies are needed to define Rep's exact mechanism of action.

Evidence suggests Pol IV's mutagenic activity is at its maximum when cells are under stress (Foster, 2007), but the molecular mechanisms of this control are poorly understood. The regulation of Pol IV activity by directly interacting cofactors, such as the β clamp, has been well established (Wagner et al., 2000). It has recently been proposed that Pol IV's mutagenic activity is modulated by interactions with UmuD and RecA (Godoy et al., 2007) and enhanced by the transcription factor NusA (Cohen et al., 2009). Our results suggest that Pol IV activity is also controlled by its association with Rep helicase. It would be interesting if the Rep-Pol IV complex is part of one, or more, larger protein assemblies that direct and regulate Pol IV's activities for its diverse biological roles.

Experimental procedures

Bacterial strains

The bacterial strains and their relevant genotypes used in this work are given in Table 1. Genetic manipulations were performed as described (Miller, 1992). The Δrep::Kn allele was from strain JJC213 (Michel et al., 1997). The ilvA::Kn allele was from the Keio collection (Baba et al., 2006). Strains PFG528, PFB895, PFB902 and PFB839 were constructed from donor strains containing the relevant alleles using P1 bacteriophage transduction, selecting for resistance to kanamycin. The repK28I mutant allele was constructed in strain SW102 (Warming et al., 2005) using the cat-sac recombineering method as described (Sawitzke et al., 2007). The sequences of the oligonucleotides used are given in Table 2. A PCR product that included sequence upstream and downstream of repK28 and a flanking cat-sac cassette was generated by amplification from the template plasmid pELO4 using the oligonucleotides 5′ REP CAT and 3′ REP SAC (Table 2). This PCR product was transformed into SW102, selecting for chloramphenicol resistance, which generated a rep allele disrupted with cat-sac. A 200 bp PCR product containing the repK28I point mutation (a change of the 28th codon from AAA to ATC) was generated by amplification from genomic DNA using overlap extension PCR mutagenesis (Ho et al., 1989). The oligonucleotides used for this method were: 5′ 100 bp US REP K28, 3′ 100 bp DS REP K28, 5′ REP K28I and 3′ REP K28I (Table 2). The PCR product was then transformed into the SW102 derivative carrying the disrupted rep allele by selecting for resistance to sucrose. This procedure placed the repK28I mutant allele on the chromosome. This strain was transformed with a plasmid carrying RecA+ (pPF2041), and a P1 bacteriophage lysate of this strain was used to transduce PFB839 to isoleucine prototrophy, thus creating strain PFB842. The presence of the repK28I mutant allele in strain PFB842 was confirmed by sequencing.

Table 1.  Bacterial strains and plasmids.
Strain/PlasmidRelevant genotype or descriptionReference
Strain  
 P90CF-araΔ(gpt-lac)5 thiCoulondre and Miller (1977)
 FC29P90C/F′Δ(lacIZ) Pro+Cairns and Foster (1991)
 FC36P90C RifRCairns and Foster (1991)
 FC40FC36/F′Φ(lacI33-lacZ) Pro+Cairns and Foster (1991)
 FC722FC40 carrying a TcSdTn10 on the episomeFoster (1997)
 PFB243FC40 ΔdinB::Zeo on the chromosome and the episomeStumpf and Foster (2005)
 PFB839FC40 ΔilvA::KnThis work
 PFB842FC40 repK28IThis work
 PFB846PFB243 repK28IThis work
 PFB894FC722 repK28IThis work
 PFB895PFB243 Δrep::KnThis work
 PFB902FC722 Δrep::KnThis work
 PFB903FC40 gal-76::Cm ΔattB::psulA-gfpThis work
 PFB904PFB842 gal-76::Cm ΔattB::psulA-gfpThis work
 PFB905PFG528 gal-76::Cm ΔattB::psulA-gfpThis work
 PFC1530P90C ΔaraBADThis work
 PFG528FC40 Δrep::KnThis work
 HC101RM1094 polBex1Slater and Maurer (1993); Foster et al. (1995)
 SW102galK[λ cI857 Δ(cro-bioA)<>tet]Warming et al. (2005)
 BL21(DE3)F-, ompT, hsdSB(rB-, mB-), dcm, gal, λ(DE3)Studier and Moffatt (1986)
Plasmid  
 pBAD24Plasmid for arabinose-inducible gene expressionGuzman et al. (1995)
 pBR322Plasmid for cloningBolivar et al. (1977)
 pELO4Plasmid carrying the cat-sac cassetteSawitzke et al. (2007)
 pET22bPlasmid containing a T7 inducible promoterNovagen
 pTB146Plasmid carrying HIS-tagged SUMO protein with a T7 inducible promoterBendezu et al. (2009)
 pPF2041pBR322 carrying recA+ expressed from its own promoterStumpf and Foster (2005)
 pDEST27Plasmid carrying GSTInvitrogen
 pPFV391pBAD24 expressing STREP-FLAG Pol IVThis work
 pPFV392pBAD24 expressing STREP-FLAG Pol IVΔC6This work
 pPFV393pET22b expressing FLAG Pol IIThis work
 pPFV394pET22b expressing FLAG Pol II Exo-This work
 pPFV395pET22b expressing FLAG Pol IIIαThis work
 pPFV396pSUMO expressed Pol IV from the T7 promoterThis work
 pPFV397pBAD24 expressing GST-PreSc Prot-RepThis work
 pPFV399pBAD24 expressing GST-PreSc Prot-RepK28IThis work
Table 2.  Oligonucleotides.
OligonucleotideSequence (Written 5′ to 3′)
5′ REP CATGTCGAATTCGTTACCGGCCCCTGCCTGGTGCTGGCGGGCGCGGGTTCCGGTCCTGTGACGGAAGATCACTTCG
3′ REP SACCCGCAACCGCGGATCAGATGGGCGATTTTATTGGTGATAACACGAGTTTTCTGAGGTTCTTATGGCTCTTG
5′ 100 bp US REP K28CGAAGATTGAGCAATACACC
3′ 100 bp DS REP K28CTCGCGCGCTGCTTTATTAG
5′ REP K28ICGGGTTCCGGTATCACTCGTGTTATC
3′ REP K28IGATAACACGAGTGATACCGGAACCCG
5′ EcoRI STREP/FLAGCGGAATTCACCATGTGGAGCCATCCCCAGTTTGAAAAAGATTACAA AGATGATGATGATAAA
5′ FLAG DINBGATTACAAAGATGATGATGATAAAATGCGTAAAATCATTCATGTGG
3′ DINB NcoICATGCCATGGTCATAATCCCAGCACCAGTTGTCTTTC
3′ DINB DELTA6 NcoICATGCCATGGTCATCTTTCCATTTGCGGGTCAAG
5′ NdeI FLAG NheI POLBGGAATTCCATATGGATTACAAAGATGATGATGATAAAGCTAGCGTG GCGCAGGCAGGTTTTATCTTA
3′ EcoRI POLBCGGAATTCTCAAAATAGCCCAAGTTGCCCGGT
5′ NheI DNAECTAGCTAGCATGTCTGAACCACGTTTCGTACACC
3′ EcoRI DNAECGGAATTCTTAGTCAAACTCCAGTTCCACCTGC
5′ SUMO DINBGAGAACAGATTGGTGGTATGCGTAAAATCATTCATG
3′ SUMO DINBCATGAATGATTTTACGCATACCACCAATCTGTTCTC
5′ KpnI PreSc GSTGGGGTACCCTGGAAGTTCTGTTCCAGGGGCCCATGTCCCCTATACTA GGTTATTGG
3′ HindIII GSTCCCAAGCTTTTAATCCGATTTTGGAGGATGGTCGCC
5′ SD NheI REPCTAGCTAGCAGGAGGAATTCACCATGCGTCTAAACCCCGGCC
3′ KpnI REPCGGGGTACCTTTCCCTCGTTTTGCCGCC
MOLECULAR BEACON-1Cy5/CCTACGCCACCAGCTCCGTAGG/BHQ
BEACON SUBSTRATE-1GGAGCTGGTGGCGTAGGTTTTTTTTTTTTTTTTTTTT
BEACON COMPLEMENTCCTACGGAGCTGGTGGCGTAGG
89 bp TEMPGCTCTAGAAATAATTTTGTTTAACTTTAAGAAGGAGATATACCATGGATTACAAAGATGATGATGATAAACATATGGC TAGCGGATCGG
89 bp COMPCCGATCCGCTAGCCA

Plasmid construction

All DNA manipulations were performed according to standard procedures (Sambrook et al., 2000). Restriction enzymes were purchased from New England Biolabs. Platinum PCR SuperMix was purchased from Invitrogen. Oligonucleotides were purchased from Integrated DNA Technologies. All constructs were confirmed by DNA sequencing. To construct pPFV391 and pPFV392 carrying STREP-FLAG-Pol IV (S-FLAG-Pol IV) and STREP-FLAG-Pol IVΔC6 (S-FLAG-Pol IVΔC6), respectively, the dinB gene was amplified from the DNA of E. coli strain FC36 using the oligonucleotide pairs 5′ EcoRI STREP/FLAG, 3′ DINB NcoI and 5′ FLAG DINB, 3′ DINB DELTA6 NcoI (Table 2). The PCR products were cloned into pBAD24 using the restriction enzymes EcoRI and NcoI. To construct pPFV393, carrying FLAG-Pol II, and pPFV394, carrying FLAG-Pol II Exo-, the polB gene was amplified from the DNA of E. coli strains PFB842 and HC101 [which carries the Pol II Exo- allele (Foster et al., 1995)] using the oligonucleotide pair 5′ NdeI FLAG NheI POLB, 3′ EcoRI POLB (Table 2). The PCR products were cloned into pET22b using the restriction enzymes NdeI and EcoRI. This manipulation replaced the pelB leader sequence in pET22b with a sequence encoding the 8 amino acid FLAG peptide followed by a NheI restriction enzyme recognition sequence. To construct pPFV395, carrying Pol IIIα, the dnaE gene was amplified from the DNA of E. coli strain FC36 using the oligonucleotide pair 5′ NheI DNAE, 3′ EcoRI DNAE (Table 2). The PCR product was cloned into pPFV393 behind the sequence encoding FLAG using the restriction enzymes NheI and EcoRI. To construct pPFV396, carrying Pol IV, the dinB gene was amplified from the DNA of E. coli strain FC36 using the oligonucleotide pair 5′ SUMO DinB, 3′ SUMO DinB (Table 2). The PCR product was cloned into pSUMO using the restriction enzymes NheI and SacI. To construct pPFV397 and pPFV399, the DNA encoding GST fused to an N-terminal Prescission Protease site was amplified from pDEST27 (Invitrogen) using the oligonucleotide pair 5′ KpnI PreSc, 3′ HindIII GST (Table 2). The PCR product was cloned into pBAD24 using the restriction enzymes KpnI and HindIII. In a subsequent step, the wild-type rep gene or the repK28I mutant gene was amplified from the genomic DNA of E. coli strain FC36 or PFB842 using the oligonucleotide pair 5′ SD NheI REP, 3′ KpnI REP. This PCR product was cloned into the intermediate plasmid using the restriction enzymes NheI and KpnI.

Mutation and viability assays

Media and protocols were used as described previously (Cairns and Foster, 1991; Foster, 1994). For adaptive mutation experiments (Fig. 1A and B), three independent cultures of each strain were grown to saturation in M9-glycerol minimal medium at 37°C with aeration. From these cultures, about 2 × 108 cells of strains FC40, PFG528 and PFB842 and about 5 × 108 cells of strains PFB243, PFB846 and PFB895 were plated on lactose minimal plates together with about 2 × 109 cells of the non-revertible scavenger strain, FC29. Plates were incubated at 37°C, and Lac+ colonies that appeared from day 2 to day 5 were counted. The viabilities of the Lac- cells during the experiment were determined from day 0 to day 3 by removing plugs from between Lac+ colonies, suspending the cells in 0.85% saline and plating dilutions of these suspensions on Luria–Bertani (LB) plates supplemented with 100 µg ml−1 rifampicin, which selects against FC29. To calculate the number of Lac+ revertants per cell, the number of Lac+ colonies was divided by the mean number of Lac- cells present on the lactose plate 2 days earlier (it takes 2 days for a Lac+ colony to become visible). Statistical calculations were as described (Rodriguez et al., 2002). To determine growth-dependent mutation rates (Fig. 2A), 25 independent cultures of strains FC722, PFB894 and PFB902 were grown to saturation in LB broth at 37°C with aeration. Aliquots from nine cultures were combined into three samples, and appropriate dilutions of these were plated on LB agar plates to determine the total number of cells. 0.1 ml aliquots of each culture (approximately 108 cells) were spread on LB agar plates supplemented with 40 µg ml−1 tetracycline (Tc); the plates were incubated at 37°C, and colonies appearing after 24 h were counted. The mean numbers of mutations per culture and their confidence limits were obtained with the Ma-Sandri-Sarkar maximum likelihood method (Sarkar et al., 1992) implemented by the FALCOR web tool found at http://www.mitochondria.org/protocols/FALCOR.html (Hall et al., 2009), corrected for plating only 1/10th of the culture. These values were divided by twice the total number of cells per culture to obtain the mutation rates per cell per generation and their confidence limits (Foster, 2006).

Microscopy

Strains carrying the plasmid expressing the sulA-gfp reporter (McCool et al., 2004) were grown to an OD600 of 0.4 in LB broth with aeration at 37°C, concentrated by centrifugation, and resuspended in 0.85% saline. 3 µl aliquots of each culture were spotted on a glass microscope slide. Phase and fluorescence images were taken on a Nikon eclipse 80i microscope with a CoolSNAP HQ2 camera and Metamorph image capture software (Universal Imaging). Green fluorescent protein fluorescent signals were visualized using a C-FL HYQ fluorescein isothiocyanate filter cube (excitation filter wavelength, 460–500 nm; barrier filter wavelength, 515–550 nm) with a Nikon Plan Apo 100X objective. Phase contrast and fluorescence images were false coloured and superimposed using Metamorph software. All images were taken with equal exposure time and were not adjusted for digital contrast.

NQO sensitivity assays

4-nitroquinoline-1-oxide (Sigma-Aldrich) stocks were prepared at 10 mM in N,N-dimethylformamide (Sigma-Aldrich). LB agar plates containing 12 µM NQO were prepared the day of the experiments. Strains were grown to saturation in LB broth at 37°C with aeration. These cultures were sequentially diluted in 0.85% saline from 10−1 to 10−6, and 10 µl aliquots of these dilutions were spotted on LB agar plates with and without NQO. Plates were incubated at 37°C in the dark for 18 h and then imaged.

Protein purification

S-FLAG Pol IV and S-FLAG Pol IVΔC6 were purified from strain PFC1530 carrying the pBAD24-derived plasmids pPFV391 and pPFV392. Overnight cultures of the strains were diluted 1:50 into 1 l LB broth containing 100 µg ml−1 ampicillin and grown to mid-log phase (OD600 ∼ 0.4). Expression was induced by adding 0.05% arabinose, after which the cultures were incubated an additional 3 h at 37°C. The cells were harvested by centrifugation and resuspended in 100 ml lysis buffer [50 mM Tris-HCl, pH 8.0, 300 mM NaCl, 1 mM EDTA, 1 mM sodium azide (Sigma-Aldrich), 0.1% Triton X-100, 1 mM DTT, 5 mM benzamidine HCl, 1 mM phenylmethanesulfonyl fluoride], treated with 1 mg ml−1 lysozyme for 30 min at 4°C, and then sonicated. The lysate was clarified by centrifugation, incubated by rocking with ∼0.5 ml Anti-FLAG M2 Affinity Gel (Sigma-Aldrich) for 2 h at 4°C, and then passed through a 20 ml Poly-Prep Chromatography Column (Bio-Rad Laboratories). The column was washed four times with 10 ml lysis buffer and eluted with lysis buffer containing 100 µg ml−1 FLAG Peptide (Sigma-Aldrich). Elution fractions containing the fusion protein (determined by electrophoresis of small aliquots) were pooled and successively dialysed two times in 1 l of dialysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM sodium azide, 1 mM DTT), then dialysed into storage buffer (150 mM NaCl, 1 mM sodium azide, 1 mM DTT, 50% glycerol) and stored at −20°C. One round of FLAG purification produced protein with purity > 95% as judged by Coomassie G250 staining. Pol IV preparations were checked for β clamp contamination by assaying for polymerase processivity on primed M13mp18 single-stranded DNA (New England Biolabs); no β clamp contamination was observed.

FLAG-Pol II, FLAG-Pol II Exo- and FLAG-Pol IIIα were purified from E. coli strain BL21(DE3) carrying the pET22b-derived plasmids pPFV393, pPFV394 and pPFV395. Overnight cultures were diluted 1:50 into 0.5 l LB broth containing 100 µg ml−1 ampicillin and grown to mid-log phase (OD600 ∼ 0.4). Expression was induced by adding 1 mM IPTG, and the cultures were incubated an additional 18 h at 16°C. Cells were harvested and lysed, and protein was purified as described above. One round of FLAG purification produced protein with purity > 95% as judged by Coomassie G250 staining.

Pol IV-SUMO was purified from E. coli strain BL21(DE3) carrying the pSUMO-derived plasmid pPFV396. Expression was induced as described above for the pET22b-derived plasmids. Cells were harvested and lysed as described above in HIS lysis buffer (50 mM Tris-HCl, pH 8.0, 300 mM NaCl, 1 mM sodium azide, 0.1% Triton X-100, 1 mM DTT, 5 mM benzamidine HCl, 1 mM phenylmethanesulfonyl fluoride, 10 mM imidazole). The lysate was clarified by centrifugation, incubated by rocking with ∼2.0 ml Ni-NTA His-Bind Resin (Novagen) for 2 h at 4°C, and then passed through a 20 ml Poly-Prep Chromatography Column (Bio-Rad Laboratories). The column was washed four times with 10 ml HIS lysis buffer and eluted using a stepwise elution of HIS lysis buffer containing 50–500 mM Imidazole. Elution fractions containing the fusion protein (determined by electrophoresis of small aliquots) were pooled and dialysed two times successively in 1 l of SUMO protease dialysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.2% NP-40, 1 mM DTT). SUMO was removed from the Pol IV protein by incubating with SUMO Protease (a gift from L. Chen) for 48 h at 4°C. The free SUMO (which retains the HIS tag) was then removed by passing the sample over ∼2.0 ml Ni-NTA His-Bind Resin. The protein was dialysed in storage buffer (150 mM NaCl, 1 mM sodium azide, 1 mM DTT, 50% glycerol) and stored at −20°C. The purity of native Pol IV was > 95% as judged by Coomassie G250 staining. The preparations were tested for β clamp contamination as described above; no β clamp contamination was detected.

The Rep-PreSc-GST and the RepK28I-PreSc-GST fusion proteins were purified from strain PFC1530 carrying the pBAD24-derived plasmids pPFV397 or pPFV399, respectively. Expression was induced as described above for pBAD24-derived plasmids. Cells were harvested and lysed as described above. The lysate was clarified by centrifugation, incubated by rocking with ∼0.5 ml glutathione sepharose high performance resin (GE Healthcare) for 2 h at 4°C, and passed through a 20 ml Poly-Prep Chromatography Column (Bio-Rad Laboratories). The column was washed four times with 10 ml lysis buffer and eluted with lysis buffer containing 10 mM L-glutathione (Sigma-Aldrich). Elution fractions containing the fusion protein (determined by electrophoresis of small aliquots) were pooled and successively dialysed two times in 1 l of Rep dialysis buffer (50 mM Tris-HCl, 300 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 4 mM MgCl2, 1.5 mM ATP, 2.5% glycerol, 1 mM sodium azide). GST was removed from Rep or RepK28I by incubation with PreScission Protease (GE Healthcare) overnight at 4°C. Free GST was removed by passing the samples over ∼0.5 ml glutathione sepharose high performance resin. The proteins were dialysed in storage buffer (150 mM NaCl, 1 mM sodium azide, 1 mM DTT, 50% glycerol) and stored at −20°C.

GST pull-down assays

Pull-down assays (Figs 4 and 8 and S3) were performed as described (Dionne et al., 2003). 8 µg Rep-GST, RepK28I-GST or GST (purchased from Santa Cruz Biotechnology) was preincubated with 10 µl bed volume Glutathione Sepharose High Performance resin (GE Healthcare) in TBSTM [10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl2, 0.1% Tween-20 (Sigma-Aldrich)] containing 100 µg ml−1 BSA at room temperature for 15 min. 8 µg of the indicated protein was added and incubated at room temperature with rocking for 30 min. Mixtures were washed six times with 1 ml TBSTM. The resins were boiled in 30 µl 1× SDS-PAGE sample loading buffer. The tubes were centrifuged, and half of each supernatant was electrophoresed on 8%–10% denaturing polyacrylamide gels. The samples were visualized by Coomassie blue G250 staining or western analysis (see below). Band intensities were quantified using ImageJ version 1.43µ software (Rasband, 2007).

Immunoblots

Standard molecular biology techniques were used as described (Ausubel et al., 1987). For Pol IV and GroEL Western blots (Fig. 1C), the bacterial strains were diluted 10−3 in M9 glycerol minimal medium and grown to saturation at 37°C with aeration for 24 h. Cells were harvested by centrifugation, and the total protein content was determined using Bradford assays (Bio-Rad Laboratories). After adjusting for equal protein content, the pellets were resuspended in 1× SDS-PAGE sample loading buffer, boiled for 5 min, and 40 µg of total protein was electrophoresed on an 8% SDS-polyacrylamide gel. Proteins were then transferred to an Immobilon-P membrane (pore size, 0.45 µm; Millipore) and probed with a 1:2000 dilution of rabbit anti-Pol IV polyclonal antiserum [obtained from H. Ohmori and clarified using acetone powder made from a ΔdinB strain according to the procedure in (Harlow and Lane, 1988)] in PBS-TM (1× PBS, 0.1% Tween-20, 5% non-fat milk powder, 0.005% sodium azide). Bands were visualized using a 1:5000 dilution of alkaline phosphatase-conjugated goat anti-rabbit secondary antibody (Promega) in PBS-TM and developed using Western-Star Immunodetection System (Applied Biosystems). The blot was then stripped using BlotFresh Western Blot Stripping Reagent Ver. II (SignaGen Laboratories), probed with a 1:5000 dilution of mouse anti-GroEL monoclonal antibody in PBS-TM (Stressgen), and visualized as described above except using alkaline phosphatase-conjugated goat anti-mouse secondary antibody (Promega). Band intensities were quantified using ImageJ version 1.43µ software (Rasband, 2007).

For anti-FLAG Western blots of pull-down assays shown in Fig. 4C and S3C, protein was prepared as described in the ‘GST pull-down assays’ section; 5 µl aliquots of each sample were electrophoresed on an 8% SDS-polyacrylamide gel and transferred as described above. The membranes were probed with a 1:5000 dilution of anti-FLAG monoclonal antibody (Sigma-Aldrich) in PBS-TM. Blots were visualized as described above using alkaline phosphatase-conjugated goat anti-mouse secondary antibody (Promega) and Western-Star Immunodetection System (Applied Biosystems). For the anti-FLAG Western blot of pull-down assays shown in Fig. 8A, the procedure was modified as follows: 15 µl aliquots of each sample were electrophoresed on an 10% SDS-polyacrylamide gel; after transfer, the membrane was probed with a 1:500 dilution of anti-FLAG monoclonal antibody in TBST; visualization was with ECL Western Blotting Substrate (Thermo Scientific). For all the Western blots, band intensities were quantified using ImageJ version 1.43µ software (Rasband, 2007).

MBHA

The helicase substrate used for the MBHA is described in Belon and Frick (2008). The substrate was prepared by combining MOLECULAR BEACON-1 and BEACON SUBSTRATE-1 at a 1:1 molar ratio and a final concentration of 10 µM in 10 mM Tris-HCl, pH 8.0. The strands were annealed by transferring the tube to a beaker of 3 l of water at 95°C and allowing the water to cool to room temperature. Reactions (100 µl) contained 25 mM Tris(OAc), pH 7.5, 5% glycerol, 10 mM Mg(OAc)2, 2 mM DTT, 100 nM DNA substrate, 200 nM untagged Rep helicase and, where indicated, 1000 nM polymerase. Reactions were initiated by adding ATP to a final concentration of 2 mM and incubated at room temperature for 16 min. The data were collected every 30 s using a SpectraMax M5 microplate reader (Molecular Devices). Cy5 fluorescence was measured for excitation/emission at 643/667 nm. The data points shown in Figs 5A, 6 and S1 are the means of triplicate reactions; error bars represent standard deviations of the triplicate reactions.

Radioactive unwinding assays

The helicase substrate used for radioactive unwinding assays (Figs 5B and S4A) was prepared by combining 5′-32P-labelled BEACON COMPLEMENT and unlabelled BEACON SUBSTRATE-1 (Belon and Frick, 2008) at a 1:1 molar ratio and a final concentration of 100 nM in 10 mM Tris-HCl, pH 8.0. The strands were annealed as described above. Reactions (10 µl) contained 25 mM Tris(OAc), pH 7.5, 5% glycerol, 10 mM Mg(OAc)2, 2 mM DTT, 10 nM DNA substrate, 20 nM untagged Rep helicase, where indicated, 100 nM polymerase, and, where indicated, 1000 nM SSB (Sigma-Aldrich). Reactions were initiated by adding ATP to a final concentration of 2 mM, incubated at room temperature for 18 min, and quenched by adding a 1× volume of helicase stop buffer (10 mM Tris-HCl, pH 8.0, 20 mM EDTA, 20% glycerol, 0.5% SDS, 0.1% bromophenol blue). The samples were then electrophoresed on an 18% SDS-polyacrylamide gel, which was then dried and visualized on film. Band intensities were quantified using ImageJ version 1.43µ software (Rasband, 2007).

Primer-extension assays

The substrate used for primer-extension assays was unmodified, undamaged DNA and was prepared by combining oligonucleotide 89 bp TEMP and 5′-32P-labelled 15 base-long oligonucleotide 89 bp COMP (Table 2) at a 1:1 molar ratio and final concentration of 200 nM in 10 mM Tris-HCl, pH 8.0. The strands were annealed as described above. Excess radiolabelled primer was removed by purifying the annealed substrate with a QIAquick PCR Purification Kit (Qaigen), eluting in 10 mM Tris-HCl, pH 8.0, and reannealing the strands. Reactions (10 µl) contained 25 mM Tris(OAc), pH 7.5, 5% glycerol, 10 mM Mg(OAc)2, 2 mM DTT, and 20 nM DNA substrate. Where indicated, the reactions also contained 100 nM untagged Rep or RepK28I helicase and 1000 nM SSB (except for the assays in Fig. 9, which contained 1250 nM SSB). Concentrations of polymerases were as follows: Fig. 7, left, 20 nM S-FLAG Pol IV and S-FLAG Pol IVΔC6; Fig. 7, right, 30 nM S-FLAG Pol IVΔC6; Fig. 9, 20 nM S-FLAG Pol IV; Fig. S5, 50 nM untagged Pol IV. Reactions were initiated by adding ATP to a final concentration of 2 mM and dNTPs to a final concentration of 0.5 mM, incubated at room temperature for 2 min, and quenched by adding a 1× volume of stop buffer (95% formamide, 20 mM EDTA, 0.1% bromophenol blue). Samples were then electrophoresed on a 13% urea gel [13% polyacrylamide, 7 M urea, 0.1 M Tris base, pH 8.0, 0.1 M boric acid, 2 mM EDTA (1× TBE); running buffer was 0.5× TBE], which was then dried and visualized on film. Band intensities were quantified using ImageJ version 1.43µ software (Rasband, 2007).

Acknowledgements

We thank T.G. Bernhardt, L. Chen, D.L. Court, D.B. Kearns, B. Michel, J.H. Miller, H. Ohmori, S.J. Sandler, and the National BioResource Project (Japan) for bacterial strains, plasmids, and reagents. We are grateful to D.B. Kearns for the use of his microscope, and to S.R. Chinnaswany, K. Hu, A. Heaslip, C. Kao and D.B. Kearns for technical advice and assistance. Finally, we thank the past and present members of our laboratory for their encouragement, advice and technical help; and, in particular, we thank Shera Lesly and Ashley Williams for doing the original yeast two-hybrid screen. This work was supported by USPHS NIH Grant GM065175 to P.L.F.

Ancillary