Protein-DNA interactions are mainly electrostatic in nature, which makes them sensitive to salt. If the inhibitory effect of salt is due to competition of ions with charged groups of Polε and DNA involved in interaction, it should be proportional to the salt concentration. Despite this prediction, we have shown that salt severely affects the interaction of hPolεCD with DNA. In particular, a 1.5-fold increase in salt concentration from 0.1 M to 0.15 M results in a 25-fold reduction in stability of the hPolεCD/DNA complex (Table 1). Such a dramatic effect of salt on affinity to DNA can be attributed to changes in protein solvation and elevated conformational dynamics of the flexible parts/domains. The catalytic domains of B-family DNA polymerases have the universal “right-hand” DNA polymerase fold (32) consisting of five subdomains: N-terminal, exonuclease, palm, fingers, and thumb. Among them, the fingers are the most flexible and change the conformation from “open” to “closed” upon dNTP binding. The thumb domain grips the distant part of the DNA duplex and demonstrates significant flexibility according to structural studies of hPolα (29). An ordered movement of the palm domain during polymerase translocation along the duplex was proposed from structural studies of yeast Polα (33).
This study has shown that hPolεCD exhibits low affinity to the template:primer at physiological salt concentration, with a 19-fold higher KD value than the 79 nM obtained in the absence of salt (16). This finding supports the idea that PCNA is required for processive synthesis of the leading strand, according to the current model of the eukaryotic replication fork (34). The recently published structure of yeast Polε revealed that accessory subunits do not interact with the DNA duplex (5), indicating that the holoenzyme footprint on a DNA duplex is only ten base-pairs, the same as for the catalytic domain (22, 24). Polε is flexibly attached to the CMG helicase through interaction with the C-terminal domain of the catalytic subunit and the N-terminal domain of the second accessory subunit (4, 7). In contrast to PCNA, interaction with CMG does not prevent the Polε catalytic domain from occasional dissociation from the growing primer end.
The main role of Polε is a synthesis of the leading DNA strand, but it may handle RNA and chimeric RNA-DNA strands in certain circumstances. One example is a ribonucleotide at the growing primer end, which Polε can insert and extend (26, 27). Another example is the ability of Polε to extend the 3'-ends of R-loops, serving as a possible way to restart DNA replication (28). In addition, it is possible that Polα preliminarily terminates DNA synthesis due to generation of a mismatch, like purine-purine, which is difficult to extend. In this case, Polε may bind the chimeric RNA-DNA primer, proofread the non-cognate nucleotide, and extend the corrected primer with dNMPs until the length of a primer is enough for RFC to load PCNA. A similar mechanism was suggested for the lagging strand where Polδ proofreads the Polα-generated mismatches (35). Intriguingly, some portion of hPolε is associated with RNA polymerase II independently of the cell cycle (36). Moreover, according to results of UV cross-linking studies, hPolε is located in close proximity to the newly synthesized RNA strand (36). These data suggest that Polε could play a currently unknown role in RNA transcript processing.
As we found, hPolεCD shows approximately 3,000-fold lower activity in extension of RNA versus DNA primers, making it unlikely that Polε plays a role in restarting the replication fork from the R-loops. The kobs value of 0.019 s− 1 indicates that half of RNA primers will be extended with one dNMP in ~ 40s. The best candidate for R-loop extension would be Polα, which displays a similar rate of DNA and RNA primer extension (29). On the other hand, upon extension of an RNA primer with dNMPs, hPolεCD activity gradually increases, and the chimeric primer with seven dNMPs is extended fairly well (Fig. 4). Our studies have shown that, in comparison to template:primer binding, the rate of dNMP incorporation is more sensitive to the presence of ribonucleotides in the primer. It is interesting that despite their similarly organized DNA binding sites (29), Polα and Polε demonstrate such a significant difference in selectivity to a DNA primer upon catalysis of phosphodiester bond formation.
Incorporation of ribonucleotides to the nascent DNA strand by replicative DNA polymerases could be a challenge for genome integrity (37, 38). It was reported previously that hPolεCD readily inserts ribonucleotides, and almost half of them escape proofreading by intrinsic exonuclease (27). Interestingly, activity of hPolε was reduced only two- to three-fold when one–three consecutive ribonucleotides were added to the primer 3'-end (27). Such a minimal effect of 3'-rNMPs on primer extension may be due to the absence of salt in reaction and to the type of assay where the rate of enzyme/DNA complex formation limits the reaction rate due to low enzyme concentration (1 nM) and the presence of 10% glycerol. Our study, performed in the presence of 0.1 M salt, demonstrates much stronger sugar selectivity, with a 38-fold reduction in activity by 3'-rNMP (Fig. 5). Due to the balance between DNA polymerase and exonuclease activities (39), the hampered extension of 3'-ribonucleotides will result in increased probability of their excision.
A notable finding of the current study is the role of incoming dNTP in the discrimination of hPolε against primers containing ribonucleotides. When dNTP is not present, hPolε binds DNA, RNA, and chimeric primers with similar affinity. In the presence of dNTP, selectivity for DNA primers increases two- to four-fold. This is an interesting example of how one substrate modulates enzyme selectivity for another. This discriminative feature reduces stability of the hPolεCD complex with an RNA-containing primer. On the other hand, CMG should facilitate Polε reloading on those primers. Of note, there is no discrimination against RNA primers extended with seven or more dNMPs.
Available structural data provide clues as to why, upon template:primer binding, hPolε discriminates against the chimeric primers with one or three dNMPs with similar efficiency and does not discriminate against primers containing seven dNMPs at the 3' end. Polε interacts with a ten base-pair duplex, where the first four base pairs from the growing primer end interact with the rigid DNA-binding pocket established mainly by the palm subdomain with significant contributions by the main-chain atoms (22–24). The remainder of the template:primer interacts with the flexible thumb, which can adjust its conformation to accommodate a wider DNA:RNA duplex. The absence of discrimination against RNA when dNTP is not added may be due to the alternative mode of duplex binding, which reduces steric hindrance between the wide DNA:RNA duplex and the rigid DNA-binding cleft. dNTP binding and the fingers closing likely forces hPolε interaction with the template:primer in a specific way, which results in steric hindrance and reduced affinity to RNA and chimeric primers.