DNA pol (polymerase) ϵ is thought to be the leading strand replicase in eukaryotes. In the present paper, we show that human DNA pol ϵ can efficiently bypass an 8-oxo-G (7,8-dihydro-8-oxoguanine) lesion on the template strand by inserting either dCMP or dAMP opposite to it, but it cannot bypass an abasic site. During replication, DNA pols associate with accessory proteins that may alter their bypass ability. We investigated the role of the human DNA sliding clamp PCNA (proliferating-cell nuclear antigen) and of the human single-stranded DNA-binding protein RPA (replication protein A) in the modulation of the DNA synthesis and translesion capacity of DNA pol ϵ. RPA inhibited the elongation by human DNA pol ϵ on templates annealed to short primers. PCNA did not influence the elongation by DNA pol ϵ and had no effect on inhibition of elongation caused by RPA. RPA inhibition was considerably reduced when the length of the primers was increased. On templates bearing the 8-oxo-G lesion, this inhibitory effect was more pronounced on DNA replication beyond the lesion, suggesting that RPA may prevent extension by DNA pol ϵ after incorporation opposite an 8-oxo-G. Neither PCNA nor RPA had any effect on the inability of DNA pol ϵ to replicate past the AP site, independent of the primer length.
8-Oxo-G (7,8-dihydro-8-oxoguanine) and abasic sites (AP site) are two of the most frequent endogenous DNA lesions generated during normal cell growth. On the basis of structures of the 8-oxodG:dC pair in double-stranded DNA, 8-oxo-G does not appear to cause a serious distortion in the overall helical structure of the DNA [1,2], and this also seems to hold true when the structure of the miscoding base pair 8-oxodG:dA is investigated [3,4]. The 8-oxodG:dA mispair can give rise to G-C→T-A transversion during DNA replication with adverse consequences for the cell. To obviate this potent source of mutagenesis, cells have evolved two different BER (base excision repair) systems that ensure repair of the 8-oxo-G lesion. The first is an OGG1 (8-oxoguanine DNA glycosylase 1)-dependent pathway, which targets the 8-oxodG:dC pair, removes the lesion and leaves an intact DNA strand to act as template for the resynthesis step ; the second is a MUTYH (mutY homologue)-dependent pathway which targets the 8-oxodG:dA mispair and removes the adenine . However, these repair systems are not infallible, so that 8-oxo-G in the template strand is sometimes encountered by DNA pols (polymerases) during DNA replication or by resynthesis associated with DNA repair. Indeed, a number of eukaryotic DNA pols can bypass in vitro the 8-oxo-G lesion in a mutagenic way by incorporating adenine and cytosine during the bypass. These enzymes do not exclusively belong to the Y family pols, specialized for TLS (translesion synthesis), but include also replicative DNA pols such as DNA pols α and δ [7,8].
Abasic sites arise frequently by spontaneous hydrolysis of purines in the DNA and represent a common intermediate of BER systems. Compared with the 8-oxo-G lesion, an abasic site poses a more serious problem to the advancement of a DNA pol since the modified base has lost its coding capacity. Accordingly, its replication requires the intervention of one or more TLS DNA pols and often results in incorporation of dAMP opposite the lesion ( and references therein). However, it has been reported that an AP site can also be bypassed in vitro, albeit inefficiently, by replicative DNA pol α  and by DNA pol δ in the presence of the processivity factor PCNA (proliferating-cell nuclear antigen) .
Efficient and accurate replication of the eukaryotic genome requires not only DNA pols α and δ, but also DNA pol ϵ. Recent work in yeast supports a model wherein, during normal DNA replication, DNA pol ϵ is primarily responsible for copying the leading strand and DNA pol δ is primarily responsible for copying the lagging strand . Although much information is already available in the literature concerning the capacity of DNA pol δ to deal with 8-oxo-G and AP lesions, such information is still scarce concerning DNA pol ϵ, particularly for the human enzyme.
We therefore investigated the effect of 8-oxo-G or AP site lesions on in vitro DNA elongation catalysed by human DNA pol ϵ. To this end, we used primer-extension assays of DNA templates containing either an 8-oxo-G lesion or a tetrahydrofuran moiety mimicking an abasic site. We performed experiments under both running and standing start conditions and our results indicate that DNA pol ϵ can efficiently bypass the 8-oxo-G lesion by inserting either dCMP or dAMP opposite to it, whereas its primer elongation activity is arrested by the synthetic abasic site. Furthermore, since replicative DNA pols associate with accessory factors during DNA replication that affect their properties and may alter their bypass ability, we investigated the role of the processivity factor PCNA and of the single-stranded DNA-binding protein RPA (replication protein A) in modulating the DNA synthesis and translesion capacity of DNA pol ϵ.
Recombinant His6-tagged human heterotrimeric RPA, PCNA and heterotetrameric DNA pol δ were expressed and purified as described in [13–15]. Human DNA pol ϵ was purified from HeLa cells through six purification steps as described in . Briefly, 90 litres of HeLa cells in suspension at (5–8)×l05 cells/ml were harvested and homogenized and then DNA pol ϵ was purified by conventional methods. Steps included (NH4)2SO4 fractionation, ion-exchange chromatography on DEAE-Sephacel and phosphocellulose, adsorption chromatography on HA (hydroxyapatite), and GG (glycerol gradient) centrifugation. A DNA pol assay using poly(dA)/oligo(dT) as template–primer  as well as Western blotting against DNA pols α, δ and ϵ were used to monitor the purification. The antibodies used were: mouse monoclonal antibodies 1ct102 and 1ct25 against the catalytic subunit of DNA pol α ; rat monoclonal antibody 1E8 or rabbit polyclonal antibodies K30, K31, K32 and K33 against the catalytic subunit of DNA pol δ ; mouse monoclonal antibodies 93G1A, 93H3B and 93E24A against DNA pol ϵ p262 ; mouse monoclonal antibodies 102E2 and 102F1 against DNA pol ϵ p59 (raised against recombinant p59 expressed in Escherichia coli); rabbit polyclonal antibodies p12CR1 and p12CR2 as well as p19DH1 and p19DH2 raised against recombinant p12 and p19 respectively expressed in E. coli. Proteins for antibody production were purified by combination of conventional purification and preparative SDS/PAGE. The HA fraction had a specific activity of 3900 units/mg, and the GG fraction had a specific activity of 24000 units/mg. DNA pol ϵ consisted of all four subunits as can be seen by Western blot analysis (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/429/bj4290573add.htm). The protein content of the HA fraction was estimated to be 50 μg/ml by standard protein measurement, of which an estimated 5 μg/ml (13.5 fmol/μl) constituted DNA pol ϵ (~10% purity). The HA fraction was free from immunoreactive DNA pol α that eluted at higher salt from HA, and essentially free from immunoreactive DNA pol δ that was mainly separated in the previous columns, and was also eluted before DNA pol ϵ from HA. The concentration of the GG fraction was estimated to be 5.5 μg/ml (15 fmol/μl) by quantifying different fractions from fluorescently stained gels using the average intensity of the molecular mass marker bands as a calibrator. Purity was estimated to be >50% and the fraction was devoid of other replicative DNA pols.
All oligonucleotides were from Eurogentec and purified from a polyacrylamide denaturing gel. The 100-mer templates, either undamaged or containing an 8-oxo-G or a synthetic AP site (tetraydroxyfurane moiety), and the primers were purified further on 12% (w/v) polyacrylamide, 7 M urea and 30% formamide gels. After elution and ethanol precipitation, their concentrations were determined spectrophotometrically. The sequence of DNA templates and primers is indicated in Figure 1. Primers were 5′-labelled with T4 polynucleotide kinase (New England Biolabs) in the presence of [γ-32P]ATP according to the manufacturer's protocol. Each primer was mixed with the complementary template oligonucleotide at a 1:1 (M/M) ratio in the presence of 20 mM Tris/HCl (pH 8) and 50 mM KCl, heated at 90 °C for 5 min and then slowly cooled.
DNA templates used in the present study
Primer extension assays
Reactions of 10 μl were incubated at 37 °C for 30 min and contained 0.1 pmol of DNA templates, 50 mM Hepes (pH 7.5), 5 mM MgCl2, 1 mM DTT (dithiothreitol), 200 μg/ml BSA, 100 μM of each dATP, dCTP, dGTP and dTTP, and 23 fmol of HA fraction of DNA pol ϵ unless indicated otherwise in the Figure legends. The amount of RPA and PCNA used is indicated in the Figure legends. The reactions were stopped by adding a 0.5 reaction volume of stop solution containing 0.1% xylene cyanol and 0.1% Bromophenol Blue in 90% formamide. Before loading on to the gel, the reactions were denatured by heating at 100 °C for 3 min. The reaction products were resolved by denaturing PAGE (7 M urea, 10–15% acrylamide and 30% formamide) and then visualized and quantified using Molecular Dynamics PhosphorImager and ImageQuant software. The percentage of extension was calculated as the ratio of the intensity of the bands greater than the primer to the intensity of all bands, including the primer.
Bypass of 8-oxo-G by DNA pol ϵ under running start conditions; influence of PCNA and RPA
To carry out this study we used either the HA or the highly purified GG fractions of the DNA pol ϵ purified as described in the Experimental section. Initial titration experiments with the HA fraction indicated that at least 15 fmol of DNA pol ϵ were required to obtain significant extension (40–60%) of the DNA template–primers employed (results not shown).
To investigate the capacity of DNA pol ϵ to replicate past the 8-oxo-G lesion when acting alone or in the presence of PCNA, we first used as DNA template–primer the 100-mer A shown in Figure 1 annealed to its corresponding 17-mer primer. The 3′-hydroxy group of the primer is 22 nucleotides upstream of the 8-oxo-G lesion and its elongation was monitored under conditions defined as running start. As can be seen by comparing lanes 1 and 5 of Figure 2(A), the enzyme paused at the 8-oxo-G and at the nucleotide preceding it, as indicated by the appearance of a distinct doublet in the gel; however, longer products up to full-length template were also easily detected. By comparing lanes 1 and 5, the bypass efficiency of the 8-oxo-G lesion was estimated to be 90%, indicating that, after an initial pause, the lesion is efficiently bypassed by DNA pol ϵ. Next, we asked whether addition of the sliding clamp PCNA had any effect on the capacity of DNA pol ϵ to replicate past the 8-oxo-G lesion. Increasing concentrations of human PCNA did not significantly affect the bypass efficiency of the 8-oxo-G lesion by DNA pol ϵ, as can be seen in lanes 2–4 and 6–8 of Figure 2(A) and in Figure 2(C). In a parallel control experiment, we found that 2.8 pmol of PCNA caused roughly 4-fold stimulation of DNA synthesis performed by human DNA pol δ, whereas, under identical experimental conditions with the same undamaged DNA template–primer, no stimulation was observed for DNA pol ϵ (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/429/bj4290573add.htm). This shows that the PCNA utilized in this study is functionally active. Moreover, since PCNA dynamically slides on and off of linear DNA templates, we used the highest PCNA concentration tested (2.8 pmol) for all subsequent experiments.
Effect of PCNA and RPA on the primer-extension activity of DNA pol ϵ with DNA templates undamaged or containing a single 8-oxo-G residue
Next, the effect of the single-stranded DNA-binding protein RPA on the extension capacity of DNA pol ϵ was investigated. Biochemical studies have shown that human RPA binds in vitro to single-stranded DNA in at least two modes that interact with eight to ten nucleotides or with 30 nucleotides respectively [20,21]. As seen in Figures 2(B) and 2(D), increasing concentrations of human RPA resulted in increased inhibition of primer extension on both undamaged and damaged 100-mer template A annealed to a 17-mer primer. With 0.1 pmol of this template–primer, 0.28 pmol of RPA corresponds to one molecule of RPA for 30 nucleotides of single-stranded DNA, whereas 0.96 pmol corresponds to one molecule of RPA per eight nucleotides. It is worth noting that, at the highest concentration of RPA, the inhibition of the synthesis seemed to be more pronounced for the reaction products beyond the lesion when compared with the products before the lesion (compare the lower and upper part of lanes 5–8 in Figure 2B and see also Figure 2E and Supplementary Figure S3 at http://www.BiochemJ.org/bj/429/bj4290573add.htm). For all subsequent experiments, RPA concentration was kept to 0.96 pmol corresponding to 1 RPA for eight to ten nucleotides of single-stranded DNA.
We then asked whether the inhibitory effect of RPA is affected by PCNA. To this aim, we monitored the effect of RPA and PCNA together on the activity of the DNA pol ϵ HA fraction using the same template–primer either intact or damaged (Figure 2E, quantified in Figure 2G). PCNA did not influence the inhibitory effect of RPA in a detectable manner (compare lanes 2–4 and 6–8 of Figure 2E). As in the absence of PCNA (Figure 2B), RPA inhibition is stronger for the synthesis beyond the 8-oxo-G lesion also in the presence of PCNA. We varied PCNA concentration from 0.4 to 2.8 pmol and found no influence on the RPA inhibition of synthesis on the damaged template (results not shown). Comparable results were obtained when the experiment depicted in Figure 2E was repeated with the highly purified GG fraction of DNA pol ϵ (see Supplementary Figure S3). These results indicate that, under our experimental conditions, the presence of PCNA did not relieve the inhibitory effect of RPA and confirmed that inhibition of TLS past 8-oxo-G by RPA is more pronounced in the presence of both proteins.
In summary, our results show that human DNA pol ϵ efficiently replicates past the 8-oxo-G lesion. PCNA did not influence primer extension of either intact or damaged templates. High concentrations of RPA inhibited synthesis by DNA pol ϵ on both templates. Moreover, inhibition by RPA appeared to be more pronounced for reaction products representing elongation past the 8-oxo-G lesion. PCNA did not relieve the inhibitory effect of RPA under any experimental conditions.
Bypass of 8-oxo-G by DNA pol ϵ under standing start conditions: influence of PCNA and RPA
In order to investigate further the effect of RPA and PCNA on the 8-oxo-G bypass by DNA pol ϵ, we examined the effect of these accessory proteins in a standing start reaction, where the 3′-hydroxy group of the primer is situated in the immediate vicinity before the lesion. For this purpose, we used the 100-mer template A annealed to the 39-mer primer (Figure 1).
Overall, the results of such experiments (Figure 2F, quantified in Figure 2H) appeared to be comparable with those performed under running start conditions. The lesion did not affect the activity of DNA pol ϵ, and PCNA had no influence on this. RPA had only a modest inhibitory effect, which could not be reverted by PCNA. Clearly, the inhibitory effect of RPA on the overall primer extension is not that strong as it is under running start conditions (Figures 2E and 2G). This difference may be due to the shorter primer used in running start reactions (see the Discussion). We have repeated the experiment in Figure 2(F) with the GG fraction of DNA pol ϵ and obtained comparable results (see Supplementary Figure S4 at http://www.BiochemJ.org/bj/429/bj4290573add.htm).
Taken together, these results indicate that under standing start conditions the 8-oxo-G lesion did not affect the activity of DNA pol ϵ. As in running start reactions, primer extension by DNA pol ϵ on a template, either undamaged or bearing the 8-oxo-G lesion, appears to be inhibited by RPA, although to a noticeably lesser extent than under running start conditions. This inhibition is not affected by PCNA.
Nucleotide incorporation opposite the 8-oxo-G lesion by DNA pol ϵ
Next we asked which nucleotide was incorporated opposite the 8-oxo-G by DNA pol ϵ and whether RPA and PCNA would influence such a reaction. To this end, a standing start assay with the 100-mer template A annealed to the 39-mer primer was employed (Figure 1) and the insertion of a single nucleotide opposite the lesion was monitored. Figure 3(A) shows that on the undamaged template (lanes 2–5) the enzyme incorporated only the correct dCMP opposite the undamaged G, as expected (lane 3). Since the G implicated in the lesion is followed by a second one, the position of the main radioactive band corresponded to the position of the second G (lane 3). Some incorporation was also detected at a subsequent G that may be the result of a frameshift of the C nucleotide located between two Gs (see sequence on the left of Figure 3A). With 8-oxo-G, incorporation of A, but not of G and T, was detected in addition to C (lanes 7–10). We then monitored the effect of RPA and PCNA (Figure 3B). Inclusion of RPA or PCNA separately, or both together, did not change the pattern of incorporation by DNA pol ϵ.
Nucleotide incorporation opposite the 8-oxo-G lesion by the DNA pol ϵ alone or in the presence of PCNA and RPA
Taken together, these data show that DNA pol ϵ inserts both C and A opposite the 8-oxo-G lesion and that RPA and PCNA do not change the pattern of single nucleotide insertion catalysed by the pol.
Bypass of an AP site by DNA pol ϵ, and influence of PCNA and RPA
Next we investigated the primer extension capacity of DNA pol ϵ on a template containing a specific AP site. For our initial experiments, we used the 100-mer template B annealed to its corresponding 17-mer primer with the 3′-hydroxy group at a distance of 22 nucleotides from the AP site (see Figure 1). Time course experiments of 30, 60 and 90 min indicated that DNA pol ϵ was severely blocked at the nucleotide preceding the AP site, although some incorporation opposite the lesion was also detected (results not shown). Moreover, in contrast with what was observed with the 8-oxo-G lesion, no synthesis was detectable past the AP site. Experiments performed in the presence of PCNA gave similar results (results not shown).
We then analysed the effect of the AP site on the elongation capacity of DNA pol ϵ using longer primers and with RPA and PCNA present. As templates for these experiments, we used the 100-mer template B annealed either to a 44-mer primer 22 nucleotides from the AP site (running start) or to a 45-mer primer where the 3′-hydroxy group of the primer is situated at the nucleotide preceding the lesion (standing start) (see Figure 1). RPA and PCNA, separately or together (Figure 4), had no detectable effect on the synthesis by DNA pol ϵ with the undamaged 100-mer–44-mer and 100-mer–45-mer template–primers (Figures 4A and 4B, lanes 1–4), showing that the inhibitory effect of RPA is influenced by the length of the primers used (see the Discussion). Inclusion of the two accessory factors did not change the replication of the damaged template by DNA pol ϵ under running start conditions, which appeared to be mainly arrested at the nucleotide preceding the AP site, with some incorporation in front of the lesion (Figure 4A, lanes 5–8). The same results were obtained with the standing start reaction (Figure 4B). We also investigated the effect of the AP site using the GG fraction of DNA pol ϵ under the experimental conditions of Figure 3 and obtained comparable results (see Supplementary Figure S5 at http://www.BiochemJ.org/bj/429/bj4290573add.htm).
Effect of RPA and PCNA on the primer extension activity of DNA pol ϵ with DNA templates undamaged or containing a single AP site
While this work was in progress, it was reported that yeast DNA pol ϵ was capable of TLS past an abasic site . Notably, with a 76-mer–58-mer template–primer in which the 3′-hydroxy group was placed three bases from the lesion, the bypass efficiency was estimated to be up to 17%. We therefore decided to compare, under the same experimental conditions, the capacity of the human DNA pol ϵ preparations used in our study to replicate the AP site placed on our template–primers and on the same DNA template–primer used by Sabouri and Johansson  (template–primer C–C in Figure 1). We found that synthesis by both the HA fraction (Figure 5) and the GG fraction (results not shown) of the DNA pol ϵ was arrested at the AP site on all the templates tested, and no synthesis was detectable beyond the AP site.
Extension activity of DNA pol ϵ on different DNA templates either undamaged or containing a single AP site
Taken together, these results indicate that, under our experimental conditions, human DNA pol ϵ cannot synthesize past an AP site, but is arrested mainly at the base preceding the lesion with minor incorporation opposite the lesion. This inhibitory effect of the AP site was observed with both running and standing start reactions, and with different DNA template–primers. RPA and PCNA had no effect on the inability of the DNA pol ϵ to bypass the AP site.
We have investigated the capacity of the human replicative DNA pol ϵ to bypass the 8-oxo-G lesion and a model abasic site in oligonucleotide DNA substrates. The influence of the accessory replicative proteins PCNA and RPA on the bypass capacity of DNA pol ϵ was also studied. DNA pol ϵ was purified from HeLa cells through six purification steps (see the Experimental section and Supplementary Figure S1 for a protein gel image of the last two steps). Given the relative paucity of the material obtained at the last step, most of the experiments were conducted with the fraction preceding it (HA) that had a specific activity of 3900units/mg and was essentially devoid of immunoreactive DNA pols α and δ. Key experiments were repeated with the most purified GG fraction and led to the same conclusions as those obtained with the HA fraction (Supplementary Figures S3–S5).
8-Oxo-G is a non-distorting DNA lesion that has high mutagenic miscoding potential and can be bypassed in vitro by the replicative DNA pols α and δ by incorporating both dCMP and dAMP opposite the lesion [7,8]. The efficiency of DNA pol δ to bypass 8-oxo-G, compared with bypass of the undamaged G, can vary from 15 to 90%, depending on whether the assays are performed under single- or multiple-hit conditions [23–25].
By monitoring the replication of 8-oxo-G by the human DNA pol ϵ under running start conditions, we found that, after an initial pause both at the base preceding and across from the lesion, DNA pol ϵ efficiently replicated the damaged substrate (Figures 2A, 2B and 2E). Results in Figure 2(A) also indicate that increasing concentrations of human PCNA did not significantly affect either the extent of synthesis or the bypass capacity of DNA pol ϵ. It should be noted that, although DNA pol ϵ was originally isolated as a large form of DNA pol δ insensitive to stimulation by PCNA [16,26], PCNA was later found to stimulate [27,28] or not [17,29] the in vitro processivity of DNA pol ϵ, possibly depending on the type of DNA substrates and experimental conditions used. However, to ensure that the PCNA we used was functionally active, we have performed a control experiment showing that PCNA stimulated synthesis by DNA pol δ under our experimental conditions (Supplementary Figure S2)
We then monitored the effect of human RPA on the extension and the 8-oxo-G-bypass capacity of the human DNA pol ϵ. Increasing the concentration of RPA resulted in increased inhibition of primer extension on both undamaged and damaged templates (Figures 2B and 2D). However, in experiments with DNA primers longer than the 17-mer used in Figures 2(A), 2(B) and 2(E), the inhibitory effect of this RPA concentration was nearly undetectable (Figures 2F, 2H and 4 and Supplementary Figure S4). We confirmed this further by performing the experiment depicted in Figure 6, showing that, with the same template, increasing the length of the primer from 17- to 44-mer nearly abolishes the RPA inhibition.
Differential inhibitory effect of RPA on the extension activity by DNA pol ϵ on an undamaged DNA template with primers of different lengths
Since RPA has been shown to possess a strand-displacement activity , a trivial explanation for this observation could be that a short primer is more easily displaced than a longer one, resulting in reduction of DNA synthesis. However, this displacement activity is abolished at a concentration of MgCl2 as low as 1 mM and is almost undetectable under DNA polymerization conditions . Therefore a perhaps more appealing explanation could come from the elegant work by the Johansson laboratory , who showed that a minimal length of double-stranded template–primer (≈40 bp) is required to maximize processivity of the yeast DNA pol ϵ, possibly by reducing pol dissociation. Therefore it may be that, with the 17-mer primer, binding of RPA in the vicinity of the primer reduces reassociation by the DNA pol ϵ, in this way decreasing the net amount of synthesis.
We also observed that RPA seemed to preferentially inhibit synthesis past the 8-oxo-G lesion in the running start reaction (Figure 2E and Supplementary Figure S3). Here too, the pausing of the enzyme at the lesion could transitorily increase its dissociation from the primer so that the inhibitory effect of RPA becomes more evident. These results may indicate that extension past the 8-oxo-G lesion by the human DNA pol ϵ is the limiting step in the presence of RPA. Experiments performed to monitor the 8-oxo-G bypass capacity of DNA pol ϵ under standing start conditions (Figure 2F) gave results overall similar to those obtained with running start conditions, except for a diminished inhibitory effect of RPA on the overall DNA extension, as discussed above. Finally, results of experiments performed with the GG fraction of the human DNA pol ϵ paralleled those observed with the HA fraction (Supplementary Figures S3 and S4).
Next we addressed the question of which nucleotide was incorporated opposite the 8-oxo-G. The experiment of single nucleotide insertion shown in Figure 3(A) indicates that, as expected, only C was incorporated on the undamaged template, whereas both C and A, but not T and G, were incorporated opposite the lesion. Addition of RPA or PCNA separately or together did not change the pattern of nucleotide insertion obtained with the pol alone (Figure 3B). Crystal structures of the T7 DNA pol with 8-oxo-G-containing templates revealed that incorporation of either dC or dA in front of this lesion induces local perturbation of the DNA backbone or formation of a Hoogsteen-like base pair respectively . Such alterations of the DNA structure could interfere with the binding of DNA pol ϵ on to DNA and account for its reduced bypass efficiency of 8-oxo-G in the presence of RPA.
In the present study, we also found that human DNA pol ϵ is unable to replicate past a synthetic abasic site and stopped mainly at the base preceding the lesion, with some residual incorporation opposite it. Some incorporation opposite an abasic site has been also observed with the yeast DNA pol δ alone , whereas the replicative T4 DNA pol was shown to completely stop at the base preceding the lesion . This different behaviour may be related to the higher efficiency of the 3′→5′ exonuclease activity associated with the T4 pol. The inability of DNA pol ϵ to bypass an AP site was observed with all primers tested, spanning from 17- to 58-mers, upon addition of RPA or PCNA, separately or in combination, and under running and standing start conditions (Figures 4 and 5 and Supplementary Figure S5).
Unlike what was reported with the yeast enzyme in the experiments performed by Sabouri and Johansson , we did not detect synthesis past the AP site with the human DNA pol ϵ, even with the template primer they used, with both the HA (Figure 5) and the GG fraction (results not shown). Therefore, although we cannot rule out that different experimental conditions may play a role, we show that this discrepancy is not dependent on the use of different substrates, and our results may reflect a different bypass capability of an AP site for the yeast and human form of DNA pol ϵ.
If proven to be physiologically relevant, the documented arrest of the synthesis by the human DNA pol ϵ in the presence of accessory proteins at the endogenously frequent AP site lesions could greatly contribute to the activation of DNA-damage checkpoints.
Giada Locatelli, Nicolas Tanguy Le Gac and Giuseppe Villani provided the damaged templates and performed the experiments. Sinikka Parkkinen, Helmut Pospiech and Juhani Syväoja purified and characterized the DNA pol ϵ. Barbara van Loon and Ulrich Hubscher purified and characterized the DNA pol δ, PCNA and RPA. Juhani Syväoja, Helmut Pospiech, Ulrich Hubscher, Nicolas Tanguy Le Gac and Giuseppe Villani designed and interpreted the experiments and contributed to writing the paper and preparing Figures.
This work was supported by the Academy of Finland [grant numbers 106986 and 123082 (to H. P. and J. E. S. respectively)], and by the Association pour la Recherche sur le Cancer [grant number 4969] and EDF (Électricité de France) (to G. V.). G. A. L. was supported by a fellowship from the Fondation pour la Recherche Médicale. B. v. L. and U. H. were supported by the Swiss National Science Foundation [grant number 3100-109312/2] and by the University of Zurich.