|Home | About | Journals | Submit | Contact Us | Français|
In Escherichia coli the highly conserved DNA damage regulated dinB gene encodes DNA Polymerase IV (DinB), an error prone specialized DNA polymerase with a central role in stress-induced mutagenesis. Since DinB is the DNA polymerase with the highest intracellular concentrations upon induction of the SOS response, further regulation must exist to maintain genomic stability. Remarkably, we find that DinB DNA synthesis is inherently poor when using an RNA primer compared to a DNA primer, while high fidelity DNA polymerases are known to have no primer preference. Moreover, we show that the poor DNA synthesis from an RNA primer is conserved in DNA polymerase Kappa, the human DinB homolog. The activity of DinB is modulated by interactions with several other proteins, one of which is the equally evolutionarily conserved recombinase RecA. This interaction is known to positively affect DinB’s fidelity on damaged templates. We find that upon interaction with RecA, DinB shows a significant reduction in DNA synthesis when using an RNA primer. Furthermore, with DinB or DinB:RecA a robust pause, sequence and lesion independent, occurs only when RNA is used as a primer. The robust pause is likely to result in abortive DNA synthesis when RNA is the primer. These data suggest a novel mechanism to prevent DinB synthesis when it is not needed despite its high concentrations, thus protecting genome stability.
The SOS gene network in Escherichia coli is a highly conserved global stress response induced by DNA damage caused by either exogenous sources or byproducts of cellular metabolism (Horii et al., 1981; Goodman and Woodgate, 2013). The SOS response is upregulated when single-stranded DNA (ssDNA), the signal of DNA damage, accumulates and is bound by the protein RecA. The interaction between RecA and ssDNA results in a new activity for the RecA nucleoprotein filament or RecA. The newly acquired co-protease activity of RecA promotes the autocleavage of LexA, the SOS global transcriptional repressor (Little et al., 1980; Horii et al., 1981). LexA repressor cleavage de-represses expression of >40 genes comprising the network, including the dinB, recA, and lexA genes themselves (Fernandez De Henestrosa et al., 2000; Courcelle et al., 2001; Khil and Camerini-Otero, 2002; Friedberg et al., 2006). RecA, a multifunctional protein, plays essential roles in maintaining genomic integrity; its activities include mediating strand exchange in homologous recombination (Chintapalli et al., 2013), coordinating the access of DNA polymerases at the replication fork (Indiani et al., 2013), and slowing down DNA replication upon DNA damage (Tan et al., 2015).
DNA polymerase IV, an evolutionarily conserved error-prone DNA polymerase, is among the very first proteins upregulated during the SOS response as a consequence of the weak affinity between the LexA repressor and its binding site on the dinB promoter (Fernandez De Henestrosa et al., 2000; Courcelle et al., 2001). DinB is well-known for its ability to catalyze TLS, during which it bypasses specific DNA lesions on the template strand that would otherwise result in lethal replication fork stalling (Goodman, 2002; Friedberg et al., 2006; Jarosz et al., 2009). TLS is enabled by DinB’s open active site, which allows the enzyme to accommodate DNA lesions, though at the forfeiture of reliable geometric basepair checking (Yang, 2003; Jarosz et al., 2009). DinB is thus error-prone [at least a 10-fold higher mutation frequency on undamaged DNA compared to high fidelity DNA polymerase III’s (Tang et al., 2000)], possessing a mutational signature of -1 frameshifts and specific base pair substitutions (Kim et al., 2001; Nohmi, 2006). Accordingly, DinB activity has been associated with acquisition of bacterial antibiotic resistance (Bull et al., 2001; Cirz et al., 2005).
Despite LexA regulation, the basal level of DinB in E. coli is still relatively high [~250 nM (Kim et al., 2001)] when compared to other DNA polymerases in the cell [e.g., 40 nM for the catalytic alpha subunit of DNA polymerase III (Pol IIIα) (Maki et al., 1985)]. As a consequence, other means of regulating DinB activity are physiologically important. One of these is the formation of a higher-order protein complex with RecA and a dimer of full-length UmuD, an accessory subunit. While in this complex, the enzyme generates fewer -1 frameshift mutations and has increased catalytic activity on properly aligned templates (Godoy et al., 2007). Structural docking based on peptide interaction data suggests that the binding of RecA and UmuD encloses the DinB active site. The interactions likely reduce DinB-mediated mutagenesis by restricting template looping that is required for frameshift events (Godoy et al., 2007). Recently, we showed that DinB also forms complexes in vivo exclusively with RecA (Cafarelli et al., 2013), and that this positively affects DinB fidelity (Cafarelli et al., 2014).
The other critical regulatory interaction made by DinB is with the beta processivity clamp, a key player of the cell’s replication machinery (Wagner et al., 2000). The beta clamp increases the processivity of various DNA polymerases (Wagner et al., 2000; Maor-Shoshani and Livneh, 2002; Vidal et al., 2004) and is partially responsible for coordinating polymerase switching at the replication fork (Kath et al., 2014). DinB’s affinity for the beta clamp is lower than that of Pol IIIα [KD = 460 nM (Heltzel et al., 2012) and 108 nM (Heltzel et al., 2009) respectively]. It is possible that under non-DNA damaging conditions, this ~4-fold difference contributes to preventing DinB from accessing the replication fork instead of Pol IIIα, though there is no direct evidence that speaks to this. However, upon DNA damage DinB is clearly the most abundant DNA polymerase in the E. coli cell at a concentration of approximately 2500 nM (Kim et al., 2001). Remarkably, in the DNA damage induced cell, the ratio of DinB:Pol IIIα is at approximately 60:1 (Maki et al., 1985; Kim et al., 2001). Given that the relative affinity of the beta clamp for DinB is only approximately four times lower than that for Pol IIIα and that DinB is in such excess in the cell, other mechanisms are likely necessary to prevent DinB synthesis on undamaged DNA.
Here, we report that DinB performs poor DNA synthesis with RNA primers and that this synthesis is further impeded upon interaction with RecA. The mechanism of this inhibition of synthesis is through a seemingly robust pause that is independent of template and lesion and is likely to result in abortive DNA synthesis when RNA is the primer. Poor synthesis of DinB using RNA primers might represent a way to prevent DNA synthesis by DinB when it is not needed. Our data provide novel insight into the mechanisms of regulation of error-prone DinB, which will, in turn, permit a deeper understanding of the relationship between DNA damage, mutagenesis, and genomic stability.
The strain TMCΔT: BL21-AI (Life Technologies, Carlsbad, CA, USA) ΔdinB, ΔumuDC, ΔrecA, was used for protein purification, and was constructed by P1 transduction (Thomason et al., 2007) using as a host the BL21-AI ΔdinB, ΔumuDC strain (Cafarelli et al., 2013). The construction of the RecA overproducing plasmid (pILRecA) is described below while the DinB overproducing plasmid (pDFJ1) has been previously published (Jarosz et al., 2006). The TMCΔT strain with the overproducing plasmids was grown in Luria broth medium with ampicillin [TMCΔT/pDFJ1 (Jarosz et al., 2006); 100 μg/mL] or kanamycin (TMCΔT/pILRecA; 35 μg/mL). Protein induction conditions are described below. All oligonucleotides used in this work are listed in Supplementary Table 1.
The native DinB overproducer plasmid, pDFJ1 (Jarosz et al., 2006), was introduced by transformation into the TMCΔT strain. Native DinB was overexpressed by autoinduction (Studier, 2005; Cafarelli et al., 2013) and cells were lysed using a cell homogenizer (Cafarelli et al., 2013). DinB was purified by ion exchange and hydrophobic interaction chromatography as previously published (Beuning et al., 2006). This will be referred to as naïve DinB in this report, since it has never been in contact with RecA.
The plasmid overproducing RecA was constructed as follows: the recA gene was amplified by PCR using the plasmid pCA24N (Konola et al., 1998) as template. The recA gene-containing amplicon was cloned into the pET His6 TEV LIC cloning vector (plasmid 29653; Addgene, Cambridge, MA, USA) by ligation independent cloning (LIC). PCR cycling conditions were based on the melting temperature of primer pairs. Primer sequences are listed in Supplementary Table 1. The resulting plasmid, pILRecA, was introduced by transformation into the TMCΔT strain. Cells were grown to saturation at 37°C in Luria broth medium with 35 μg/mL kanamycin and 0.05% (v/v) glucose. A 1 L culture of the same medium was inoculated with a 1:1,000 dilution from the saturated culture and grown at 37°C with agitation (250 rpm) until it reached an OD600 of 0.7. Protein overexpression was then induced by adding 0.05% (v/v) of L-arabinose. The culture was incubated at 20°C with agitation as before for approximately 12 h. Native naïve RecA (never in contact with DinB) was purified following a protocol previously described in Chen et al. (2008), but an approximate 1:15 (w/w, TEV protease/substrate) dilution of tobacco etch virus (TEV) protease (QB3 Macrolab, University of California, Berkeley, Berkeley, CA, USA) was used to cleave the His-tag from the RecA preparations. Purity of all proteins was determined by SDS-PAGE (Supplementary Figure 1).
Primer extension assays were carried as previously described (Cafarelli et al., 2013) using Cy3 5′ labeled DNA or RNA primers annealed to various DNA templates (Supplementary Table 1 and Figure Figure1A1A). We used undamaged templates containing adenine (A) at the primer-template junction as well as a lesion-containing template with a 3-deaza-3-methyladenine (3d-meA) lesion at the same position (Supplementary Table 1). Reactions contained a mixture of 500 μM dNTPs (Takara, Otsu, Shiga, Japan), buffer and similar concentrations of the different DNA polymerases. For DinB:RecA reactions, naïve DinB and RecA were incubated in buffer SA [50 mM Hepes, 10% glycerol (v/v), 2 mM 2-mercaptoethanol, pH 7.5] at a 1:1 molar ratio at room temperature for 1 h before 0.6 μM were added to the primer extension reaction mix. Polymerase Kappa was a gracious gift of Janice Pata (Wadsworth Center, NYS Department of Health) and DNA polymerase I was obtained from New England Biolabs (Ipswich, MA, USA). The undamaged C3 DNA template (Supplementary Table 1) was used unless otherwise indicated. All reactions were performed in triplicate and direct comparisons were made only of samples separated in the same gel. Percent full extension was calculated as the fraction of the FE product divided by the total obtained from all products (Figure Figure11). Percent total extension was calculated as the fraction of extended products minus the primer divided by the total obtained from all products (Figure Figure33).
To our knowledge there is no information published regarding DinB’s ability to perform DNA synthesis using an RNA primer. It is known that in vitro DinB considerably slows DNA replication by Pol IIIα specifically on the lagging strand of synthesis (Indiani et al., 2009) and can switch with Pol IIIα at the replication fork during DNA synthesis (Heltzel et al., 2009, 2012). Due to DinB’s high intracellular concentration [lowest levels: 250 nM (Kim et al., 2001), compared to 40 nM of Pol IIIα (Maki et al., 1985)], and the high availability of RNA primers to copy the lagging strand, we were interested in assessing DinB’s inherent ability to carry out DNA synthesis using RNA primers.
We first used naïve DinB, which has never been in contact with RecA, to measure extension of a fluorescently labeled DNA or RNA primer annealed to the same unlabeled DNA template (Figure Figure1A1A). The reactions conditions and enzyme concentrations were consistent for primer extension reactions involving DNA or RNA primers. The products of these reactions were then examined by denaturing gel electrophoresis. We did not observe a significant difference between DNA and RNA primer extension in experiments with the high fidelity DNA polymerase I (Supplementary Figure 2). In contrast, we find that naïve DinB extended significantly less from an RNA than from a DNA primer (Figure Figure1B1B). Remarkably, we find that this property is conserved in DNA Pol Kappa (Figure Figure1C1C), the human homolog of DinB. We also noted a prominent band at the third nucleotide insertion (+3 site, see +3 in Figure Figure1B1B) only in DinB’s RNA primer extension products. The synthesis beyond this third addition was significantly diminished as compared to synthesis from a DNA primer (compare bands above +3 in lanes 2 and 4, Figure Figure1B1B).
We have previously shown that DinB and RecA co-purify (Cafarelli et al., 2013) and that their interaction enhances DinB’s fidelity (Cafarelli et al., 2014). We investigated whether RecA interaction also affects DinB’s extension of RNA primers. We find that, remarkably, DinB:RecA’s primer preference (i.e., difference between full extension using DNA or RNA primers) is significantly higher than that of naïve DinB. While RecA may be capable of binding ssDNA that is present in the template, we have previously shown that RecA in a 1:1 molar ratio with DinB does not inhibit DinB activity with a DNA primer [(Cafarelli et al., 2014), also compare lane 2 in Figure Figure1B1B to lane 1 in Figure Figure1D1D)]. Thus, it is unlikely that RecA’s ssDNA binding is responsible for reduced DinB activity in the RNA primer experiments (Figure Figure1D1D). Notably, the prominent band at the third nucleotide insertion that is observed in the naïve DinB extension assays was also seen here (compare +3: lane 4 in Figure Figure1B1B and lane 2 Figure Figure1D1D).
We noticed that mixing DinB and RecA in a 1:1 molar ratio has a significant effect on DinB’s full extension (compare FE band in Figure Figure1B1B, lane 4 and Figure Figure1D1D, lane 2). However, DinB interaction with RecA had little effect on the addition of the first three nucleotides (compare +3 band and below in Figure Figure1B1B, lane 4 and Figure Figure1D1D, lane 2), which accounted for most of the extension products observed in these reactions. As the uninduced RecA concentration in E. coli is approximately fourfold higher than that of DinB (Salles and Paoletti, 1983; Kim et al., 2001), we sought to determine how higher RecA concentration would alter DinB’s extension. To account for the first three-nucleotide insertions, we compared in these experiments the percent total extension (i.e., the quantification of all extension products divided by the quantification of the extension products plus the primer). Interestingly, increasing concentrations of RecA significantly inhibits DinB’s total extension using RNA primers, while having no significant effect on its extension using DNA primers (Figure Figure1E1E). These data indicate that interaction with RecA further inhibits DinB’s already poor ability to synthesize DNA using RNA primers.
To ensure that DinB poor RNA primer extension was not due to a difference in concentration or degradation of the RNA primers, the un-annealed DNA or RNA primers were separated using denaturing gel electrophoresis (Figure Figure2A2A). The full-length band for either primer was quantified as the percent of full-length primer in relation to lower bands and found to be comparable (graph in Figure Figure2A2A).
We next sought to examine another possible technical explanation for the data in Figure Figure11: that the RNA primer is not efficiently annealing to the DNA template. Therefore, RNase H was used to digest the RNA primer annealed to the DNA template. Since RNase H only digests RNA-DNA hybrids, inefficient annealing would result in undigested RNA primer (similar to the undigested control, lane 1 in Figure Figure2B2B). However, we find only smaller fragments (lane 2 in Figure Figure2B2B) and negligible evidence of the full-length RNA primer after RNase H digestion, indicating that most of the RNA primer had successfully annealed to the DNA primer.
Finally, as an RNase contamination of the purified protein preparation could possibly account for poor RNA primer extension, we incubated purified naïve DinB or RecA with the RNA primer annealed to DNA template in the absence of dNTPs and found no visible RNA primer degradation, suggesting this possibility is unlikely (Figure Figure2C2C).
These data led us to conclude that naïve DinB’s poor RNA primer extension (Figures 1B,D) arises from bona fide inefficient synthesis from an RNA primer.
We measured the effect of adding increasing concentrations of unlabeled DNA: DNA on DinB extension from a labeled RNA primer annealed to the same DNA template (RNA: DNA). We predicted that the unlabeled DNA: DNA should decrease the extension from the RNA: DNA due to DinB’s preference for DNA primers, while having less of an effect on DinB extension from a labeled DNA primer annealed to a DNA template (DNA: DNA). Quantification of the data gathered from separating the extension products in a denaturing gel (lanes 4–6 in Figure Figure3A3A) indicate that the addition of 50-fold higher concentration of DNA: DNA reduced naïve DinB’s total extension from the RNA: DNA by ~34% (graph in Figure Figure3A3A). As predicted, the addition of DNA: DNA had little effect on the total extension of DinB on DNA: DNA (lanes 1–3 in Figure Figure3A3A).
Because there is a larger difference in DinB:RecA’s ability to synthesize from DNA versus RNA primers compared to naïve DinB, we predicted that smaller concentrations of competing DNA: DNA would be required to reduce DinB:RecA’s extension on RNA: DNA. Notably, addition of only 10-fold higher concentration of DNA: DNA reduced total extension of the RNA: DNA by ~23% (graph in Figure Figure3B3B), though total extension from DNA: DNA is similar to that observed and quantified in naïve DinB (compare lanes 1–3 in Figure Figure3A3A to lanes 1–3 in Figure Figure3B3B). These results support the assertion that DinB:RecA synthesizes using RNA primers even more poorly than naïve DinB and suggests that there might be a change in the DinB active site when interacting with RecA. This is consistent with our previous findings that interaction with RecA affects DinB function (Godoy et al., 2007; Cafarelli et al., 2013, 2014).
From the experiments described above, we know that when both naïve DinB and DinB:RecA (1:1) extend from an RNA primer there is a clear accumulation of a +3 insertion product (lane 4 Figure Figure1B1B and lane 2 in Figure Figure1D1D). This is not observed on the DNA primer (lane 2 Figure Figure1B1B and lane 1 in Figure Figure1D1D). To rule out the possibility that this accumulation is due to a template sequence effect, we changed the sequence of the template at the nucleotides opposite from the +3 insertion site from C to G (C3G DNA template, Supplementary Table 1). This sequence change had no effect on the accumulation of the +3 insertion product for either naïve DinB (C3G template, Figure Figure4A4A) or DinB:RecA (C3G template, Figure Figure4B4B). We also changed every purine in the template sequence to a pyrimidine (All pyr DNA template, Supplementary Table 1) and we observe no effect on the +3 insertion product regardless of whether we used naïve DinB (All pyr template, Figure Figure4A4A) or DinB:RecA (All pyr template, Figure Figure4B4B). These data indicate that the +3 insertion product accumulation is independent of template sequence.
Due to DinB’s activity as a TLS polymerase, we determined whether both naïve DinB and DinB:RecA have the ability to bypass a lesion from a template that contained 3d-meA at the primer:template junction (Supplementary Table 1). We observed that there is lesion bypass, but the accumulation of the +3 insertion product for naïve DinB or for DinB:RecA (lesion template, Figures 4A,B respectively) occurs regardless of the template lesion indicating that the +3 pause is lesion independent.
The +3 insertion pause is reminiscent of one previously described for the derivative DinB(Y79L) (Jarosz et al., 2009) in experiments of lesion bypass. This pause was interpreted as being part of an important regulatory mechanism for TLS with cell survival consequences; strains with the plasmid-borne DinB(Y79L) are susceptible to DNA damage generated by the antibiotic nitrofurazone (Jarosz et al., 2009).
It is possible that during RNA primer extension DinB’s active site resembles that of DinB(Y79L), in which tyrosine (Y) 79 has been changed to leucine (L), resulting in a similar pause. Pausing and aborted DNA synthesis, like that predicted in the DinB(Y79L) variant (Jarosz et al., 2009), might represent a mechanism by which DinB is prevented from synthesizing on the lagging strand of DNA synthesis. This preferential activity suits the role of DinB as a specialized DNA polymerase with DNA damage tolerance activities.
DinB, the most evolutionarily conserved TLS DNA polymerase (Ohmori et al., 2001), is prone to causing mutations (Tang et al., 2000). Therefore, dinB gene expression (Fernandez De Henestrosa et al., 2000; Courcelle et al., 2001; Khil and Camerini-Otero, 2002; Friedberg et al., 2006) and DinB activity (Godoy et al., 2007; Cafarelli et al., 2013, 2014) are tightly regulated. In particular, DinB activity is governed by protein–protein interaction with RecA and a dimer of UmuD (Godoy et al., 2007). The binding of these interacting partners visibly alters the mutagenic potential of DinB in vivo and in vitro (Godoy et al., 2007; Cafarelli et al., 2013, 2014). We have also established that RecA binds to DinB prior to UmuD and proposed that DinB interacting with RecA may occur more frequently than the DinB•RecA•UmuD2 complex in vivo (Cafarelli et al., 2013). We have also shown that RecA interaction positively alters DinB fidelity (Cafarelli et al., 2014), which we presumed is due to a change in the DinB active site conformation upon binding to RecA.
We report here that DinB poorly extends from RNA primers (Figure Figure11). During non-DNA damaging conditions the affinity of the beta clamp for DinB, though lower than the affinity for DNA Pol IIIα (Heltzel et al., 2009, 2012), may be sufficient to prevent DinB from accessing the forming replication fork at RNA primers, even though DinB is present at a significantly higher concentration (Maki et al., 1985; Kim et al., 2001). We hypothesize that the drastically increased concentration of DinB upon DNA damage would overwhelm the affinity for the beta clamp allowing DinB access to the lagging strand. In fact, upon SOS induction, the dramatically higher concentration of DinB (Kim et al., 2001) slows the rate of replication by Pol IIIα (Tan et al., 2015), likely occurring at the lagging strand, suggesting that the elevated DinB intracellular concentration overwhelms the selectivity of the beta clamp and allows DinB to access the replication fork. Indeed, previous studies have shown that DinB slows replication by Pol IIIα specifically on the lagging strand (Indiani et al., 2009). We find that DinB:RecA synthesizes poorly using RNA primers compared to DNA primers beyond the third nucleotide insertion and that this poor synthesis is independent of template sequence or of a lesion (Figure Figure44). DinB:RecA (or DinB) would access the replication fork on the lagging strand, but would not efficiently synthesize from RNA primers providing a mechanism to slow down replication with a lowered mutagenic cost.
The pausing observed in DinB (or DinB:RecA) using an RNA primer (Figures Figures11, 33, 44) resembles that caused by a DinB variant, DinB(Y79L), which also pauses exactly three nucleotides after a template lesion (Jarosz et al., 2009). In the native DinB protein, the Y79 residue interacts with another aromatic residue, phenylalanine (F) 13, which in turn interacts with the incoming nucleotide (Nevin et al., 2015). During TLS, the loss of the large Y aromatic residue in DinB(Y79L) is hypothesized to cause a conformational change of the active site, which leads to the DinB pausing three nucleotides after encounter of a lesion (Jarosz et al., 2009; Nevin et al., 2015). It is remarkable that the use of an RNA primer results in a similar DinB pausing (Figures Figures11, 33, 44), suggesting the RNA primer may induce a similar active site conformation as that of DinB(Y79L). Moreover, the DinB interaction with RecA would stabilize this change. The pausing observed in the DinB(Y79L) variant is hypothesized to be part of an abortive TLS mechanism, which would lead to a futile cycle of DinB pausing and DNA polymerase III excision repair (Jarosz et al., 2009). It is possible that native DinB undergoes a similar cycle of abortive synthesis with an RNA primer. This mechanism could account for the sickness and lethality that has been observed upon DinB overproduction (Benson et al., 2014).
The data reported here identify a pausing in DinB’s DNA synthesis while using an RNA primer. This novel finding represents an important potential mechanism for preventing the high intracellular concentration of DinB from inducing unnecessary mutagenesis.
TT: designed and performed experiments, analyzed data, and wrote paper; IL: performed and designed RecA protein purification experiments and analyzed data; VB: performed initial RNA primer extension experiments; TC: designed initial experiments and analyzed data; VG: designed experiments, analyzed data and wrote paper.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We are grateful to D. Jarosz (Stanford University) and K. Knight (University of Massachusetts Medical School) for critical reading of the manuscript. We would like to thank S. Gradia for plasmid 29653: pET His6 TEV LIC cloning vector (1B) and Janice Pata (Wadsworth Center, NYS Department of Health) for the generous contribution of purified DNA polymerase Kappa. We would also like to thank members of the Godoy Lab for critically reading the manuscript.
Funding. National Institute of General Medical Sciences (RO1GM088230 to VG). Funding for open access charge: NIGMS (RO1GM088230 to VG).
The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb.2017.00288/full#supplementary-material