Bacillus subtilis encodes a discrete flap endonuclease that cleaves RNA-DNA hybrids


Frances Caroline Lowder, et al.


The current model for Okazaki fragment maturation in bacteria invokes RNA cleavage by RNase H, followed by strand displacement synthesis and 5′ RNA flap removal by DNA polymerase I (Pol I). RNA removal by Pol I is thought to occur through the 5′-3′ flap endo/exonuclease (FEN) domain, located in the N-terminus of the protein. In addition to Pol I, many bacteria encode a second, Pol I-independent FEN. The contribution of Pol I and Pol I-independent FENs to DNA replication and genome stability remains unclear. In this work we purified Bacillus subtilis Pol I and FEN, then assayed these proteins on a variety of RNA-DNA hybrid and DNA-only substrates. We found that FEN is far more active than Pol I on nicked double-flap, 5′ single flap, and nicked RNA-DNA hybrid substrates. We show that the 5′ nuclease activity of B. subtilis Pol I is feeble, even during DNA synthesis when a 5′ flapped substrate is formed modeling an Okazaki fragment intermediate. Examination of Pol I and FEN on DNA-only substrates shows that FEN is more active than Pol I on most substrates tested. Further experiments show that ΔpolA phenotypes are completely rescued by expressing the C-terminal polymerase domain while expression of the N-terminal 5′ nuclease domain fails to complement ΔpolA. Cells lacking FEN (ΔfenA) show a phenotype in conjunction with an RNase HIII defect, providing genetic evidence for the involvement of FEN in Okazaki fragment processing. With these results, we propose a model where cells remove RNA primers using FEN while upstream Okazaki fragments are extended through synthesis by Pol I. Our model resembles Okazaki fragment processing in eukaryotes, where Pol δ catalyzes strand displacement synthesis followed by 5′ flap cleavage using FEN-1. Together our work highlights the conservation of ordered steps for Okazaki fragment processing in cells ranging from bacteria to human.


RNA-DNA hybrids form during transcription and DNA replication in all cells and many viruses. While their formation is essential, persistence of RNA-DNA hybrids can lead to genome instability by causing nicks and double-stranded breaks when the 2′ OH on the ribose sugar reacts with the phosphodiester bond, forming a 2′, 3′ cyclic phosphate [1]. During normal growth, RNA-DNA hybrids occur in several different forms within cells. R-loops represent one type of RNA-DNA hybrid, occurring when newly transcribed mRNA remains base-paired with the complementary DNA strand following transcription [2,3]. RNA-DNA hybrids also occur when ribonucleotides are covalently joined to DNA. Sugar errors can occur when single ribonucleotide monophosphates (rNMPs) or stretches of rNMPs are covalently nested in DNA by replicative polymerases [4,5] or under stress conditions when error-prone synthesis occurs [6]. Covalent RNA-DNA hybrids also form when replicative polymerases extend the primase-synthesized RNA primers at the initiation of leading strand synthesis and for synthesis of each Okazaki fragment during lagging strand replication [79]. Due to the frequency of Okazaki fragment synthesis, RNA primers occur predominantly on the lagging strand, resulting in the incorporation of approximately 20,000 rNMPs per genome replication event for bacteria [10]. Because RNA-DNA hybrids can lead to genomic instability and disease, all cells enlist a wide array of proteins to recognize and resolve the different types of hybrids that form in vivo [11]. The ribonuclease H (RNase H) proteins are a well characterized group of enzymes involved in the repair of all types of RNA-DNA hybrids occurring in cells. RNase HI, HII, and HIII are all capable of cleaving substrates with stretches of four or more ribonucleotides, which includes the RNA in Okazaki fragments. RNase HI and HIII are also able to resolve R-loops, while RNase HII is responsible for cleaving at single ribonucleotide errors in genomic DNA [12,13].

Several studies have provided evidence that RNase Hs participate in the removal of RNA primers during Okazaki fragment maturation [1214], however, studies in Escherichia coli identified DNA polymerase I (Pol I) as the major bacterial protein involved in this process [14,15]. Pol I is composed of three domains: a 5′-3′ polymerase, a 3′-5′ exonuclease, and a 5′-3′ nuclease [16,17]. The well-studied Klenow fragment consists of the 5′-3′ polymerase and 3′-5′ exonuclease located in the C-terminus of the protein and is responsible for DNA synthesis and error “proofreading”, while the 5′-3′ nuclease domain is located at the N-terminus of Pol I [1520]. Coordination of Pol I activities during lagging strand synthesis should allow Pol I to synthesize DNA from an upstream Okazaki fragment through the downstream RNA primer followed by primer removal. The remaining nick between adjacent DNA fragments would then be sealed by DNA ligase to complete repair [15,21]. Due to the importance of Pol I activities during DNA replication, Pol I is very well conserved across bacterial species [22], although many bacteria, including Bacillus subtilis, lack the catalytic residues associated with an active 3′-5′ exonuclease [23], suggesting that the polymerase activity and 5′ nuclease activity of Pol I are the most critical in vivo [24]. The current model, that Okazaki fragment maturation in bacteria is primarily accomplished by Pol I [15], stems from studies in E. coli and B. subtilis showing the accumulation of ribonucleotides in short DNA fragments from cells expressing various mutants of polA [2528]. Based on these early studies, the overarching model for Okazaki fragment maturation in bacteria invokes RNA incision by RNase HI (or HIII) with DNA synthesis and the bulk of RNA removal catalyzed by Pol I.

More recent studies have used sequence and structural homology to identify the N-terminal nuclease domain of Pol I as part of a larger superfamily of proteins with 5′-3′ flap endonuclease and 5′-3′ exonuclease activity, known as FENs [29]. FEN activity is essential for replication; as such, proteins in this family are found across all domains of life [3037]. Notable features of the FEN family include the binding of two or three divalent cations in the active site, a helical or unstructured arch near the active site, and a C-terminal DNA binding motif [30,3841]. FENs are structure-specific nucleases best known for their endonuclease activity on substrates generated during strand-displacement synthesis such as 5′ flap overhang substrates or double-flap substrates, which form when a 5′ flap is generated with a 3′ single nucleotide flap occurring on the upstream strand [4246]. Proteins in the FEN family have also been shown to have activity on a variety of other substrates, including nicked duplex DNA [29,32,43]. In all three kingdoms of life, organisms exist that encode multiple FENs [29,37,47]. In many bacteria, one of the FENs is a part of Pol I while the other exists as an independently encoded protein [22,29,35]. The FENs found as a domain of Pol I represent the most well-studied bacterial FENs to date, while the discrete FENs have received much less attention, making their contribution to DNA replication and genome instability unclear.

B. subtilis is one such organism known to encode a discrete FEN in addition to Pol I. This protein, YpcP, was initially identified as a protein homologous to the N-terminal domain of Pol I [22,48]. YpcP has been described as an exonuclease, and due to its similarity to the N-terminal domain of Pol I renamed ExoR [49]. Sequence homology and structural predictions of the Pol I N-terminal domain and YpcP show that both proteins belong to the wider FEN family [29]. For this reason, we will adopt the original nomenclature proposed for Pol I-independent, bacterial FENs [29] and herein refer to B. subtilis YpcP as FEN and its gene as fenA. Initial work characterizing the relationship between Pol I and FEN demonstrated that B. subtilis polA and fenA mutants are synthetically lethal [35,48] and that overexpression of polA or fenA was able to partially suppress the filamentous phenotype of an RNase H-deficient strain [35]. However, overexpression of fenA was unable to rescue a temperature sensitive strain lacking the N-terminal domain of Pol I [35]. Through these experiments, it was concluded that the two proteins have overlapping functions but are not redundant. Biochemical assays using purified FEN suggest that it has nucleolytic activity on a variety of substrates, with some preference shown for RNA-DNA hybrids [13]. Other evidence from protein-DNA binding assays suggests that FEN has the highest binding affinity for double-stranded DNA with a 5′ overhang, suggesting that FEN might prefer to act on DNA [49]. Together these data suggest that FEN may actively contribute to Okazaki fragment processing and DNA repair, although the relationship between Pol I and FEN remains unknown.

In this work, we examined the substrate preferences of Pol I and FEN using a variety of RNA-DNA hybrids and DNA-only substrates. We also examined the phenotype of polA and fenA during genotoxic stress and in RNase H deficient backgrounds to determine the contribution of Pol I and FEN 5′ nuclease activity to genome integrity. We found that FEN shows the most robust activity on RNA-DNA hybrid substrates modeling Okazaki fragment intermediates and that the strong polA phenotypes are rescued by simply expressing the C-terminal domain lacking 5′ nuclease activity. With our results, we conclude that discrete B. subtilis FEN functions as the major Okazaki fragment nuclease, with Pol I DNA polymerase activity contributing to both DNA repair and Okazaki fragment resynthesis. Because discrete FENs are present in a wide-range of bacterial genomes, we suggest that Pol I-independent FENs provide the 5′ nuclease activity important for RNA removal during lagging strand replication for many bacteria. Further, our work suggests that lagging strand processing in many bacteria is more congruent with eukaryotes than previously thought.


Loss of fenA confers sensitivity to DNA damage in the absence of ribonuclease HIII

Since B. subtilis encodes two proteins with hypothesized FEN activity, we began by investigating the importance of fenA for growth during conditions that result in DNA damage or dNTP depletion to test the model that FEN contributes to genome maintenance under stress conditions [49,50]. As shown in Fig 1A, a single deletion of fenA does not result in any detectable phenotype for cells grown on hydroxyurea (HU) or cells exposed to UV radiation (S1 Fig). It was previously shown that fenA disruption does not confer cold sensitivity nor sensitivity to mitomycin C (MMC) or methyl methanesulfonate (MMS) [13]. Given these results, it seems clear that B. subtilis FEN does not have an appreciable role during DNA repair in vivo. Another model is that FEN is involved in Okazaki fragment maturation. If true, we would expect a strain lacking both rnhC and fenA to confer a phenotype. Ribonuclease HIII (RNHIII), encoded by rnhC, is a protein known to participate in Okazaki fragment maturation and the resolution of RNA-DNA hybrids [12,13]. The absence of both proteins resulted in cells that were more sensitive to HU or UV than either single deletion (Figs 1A and S1), suggesting that FEN could be involved in the resolution of RNA-DNA hybrids on the lagging strand, and any defects due to the absence of FEN are exacerbated with the combined absence of RNHIII and presence of genotoxic stress. Notably, loss of fenA does not increase the sensitivity of a ΔrnhB strain (Figs 1A and S1), indicating that the sensitivity observed in the ΔrnhC, ΔfenA strain is due to a failure to repair RNA-DNA hybrids that occur from patches of ribonucleotides consistent with Okazaki fragments.


Fig 1. FEN contributes to RNA-DNA hybrid repair in vivo.

(A) Spot titer assay to determine a phenotype of cells exposed to hydroxyurea (HU) in the absence of fenA. (B) Alphafold model of FEN, with (C) Site 1 conserved carboxylate residues and (D) Site 2 carboxylate residues shown in pink. Coordinated divalent metals are shown in blue and were added via alignment to Mycobacterium smegmatis FEN (PDB: 6C33). (E) Spot titer assay testing ectopically expressed mutants of fenA lacking conserved carboxylate residues for rescue of ΔrnhC, ΔfenA HU sensitivity.

FEN is reliant on residues conserved across bacterial FENs

FEN contains a conserved group of eight carboxylate resides that are predicted to be a part of the active site [29], which is modeled in Fig 1B–1D, with a multiple sequence alignment provided (S2 Fig). To test the importance of these amino acids, we designed a series of mutations to disrupt conserved residues and expressed each variant from the ectopic chromosomal amyE locus under the Pspank promoter, which allows induction of expression in the presence of isopropyl ß-D-1-thiogalactopyranoside (IPTG). As discussed above, the single deletion of fenA did not result in an obvious phenotype, so mutants were expressed in the ΔrnhC, ΔfenA background.

Overexpression of WT fenA rescued the double deletion phenotype (Figs 1E and S3). With the ΔrnhC, ΔfenA complementation assay established, four fenA mutants were created, two for each metal-binding site. The first mutant, fenAE114Q,D116N, was designed to abrogate binding of the Site 1 divalent cation without changing the overall shape of the active site. The binding at Site 1 was also altered in the second mutant, where the EADD motif (residues 114 to 117) was changed to AAAA, a construct designated as fenASite1. In addition to loss of cation binding, we predict that this mutant would have alterations to the shape of the active site due to the hydrophobic residues, as well as changes to the flexible arch, since sequence similarity suggests that the D117 residue interacts with R82 to form a salt-bridge [51]. As shown in Figs 1E and S3, neither mutant was able to rescue ΔrnhC, ΔfenA as well as WT fenA, suggesting that these residues are important for function of FEN in vivo. The third mutant, fenAD192N, has one residue from metal-binding Site 2 changed. Previous work suggested that this mutant was catalytically inactive [13], however, overexpression in the double deletion strain rescues to nearly the same degree as the WT gene (Figs 1E and S3). Adding a second mutation, creating fenAD189N,D192N, once again led to a failure to rescue the ΔrnhC, ΔfenA strain. Together this suggests that not every conserved residue is critical for activity, however, perturbations that significantly affect metal-binding render fenA inactive in vivo.

An observation from our experiments is that expression of fenAE114Q,D116N or fenAD189N,D192N in the ΔrnhC, ΔfenA strain results in cells that are more sensitive to HU than the parent strain or those expressing WT fenA (Fig 1E). To test if these mutations were dominant negative, we used the same system to express each mutant or the WT allele from an ectopic locus in WT or a ΔrnhC strain. There was no detectable phenotype in WT cells (S4 Fig), however, induced expression of either fenAE114Q,D116N or fenAD189N,D192N in the ΔrnhC background resulted in cells that were demonstrably more sensitive to HU than the parent strain (S5 Fig). Even when fenAE114Q,D116N or fenAD189N,D192N were induced on LB plates with no stressor, the colonies were smaller than those of the parent strain (S5 Fig). We speculate that the metal-binding variants are still able to bind substrates in vivo but are unable to bind one of the metal ions and therefore lack the catalytic activity required for turnover. We hypothesize that in WT cells, the mutants are outcompeted by the overlapping functions of Pol I, FEN, and RNHIII, and cause no phenotype. Loss of RNHIII disrupts this balance, which is then exacerbated by exposure to HU [52], ultimately leading to a phenotype during mutant overexpression.

FEN is more active on canonical double-flap structures than Pol I

The fenA phenotype supports a contribution to Okazaki fragment processing as opposed to DNA repair. Given this, we chose to assay FEN and Pol I in vitro to compare their nuclease activity on a series of substrates to identify the substrates preferred by each enzyme. To do so, we expressed and purified each protein with a His6-SUMO tag, which we cleaved prior to use (detailed in Methods). This allowed purification of full-length proteins without any additional residues (S6 Fig). Using fluorescently labeled oligonucleotides (S1 Table), we generated a series of biologically relevant substrates. Each substrate had a DNA-only version, as well as a chimera / RNA-DNA hybrid version, where the first 12 bases at the 5′ end of the labeled strand are equivalent ribonucleoside monophosphates (rNMPs). In Figs 27, the assayed substrates are depicted at the top of the gels, with ribonucleotides indicated by zig-zag lines. As the activity of FENs is dependent on the divalent cations at the protein’s active site, we used a buffer containing physiologically relevant concentrations of Mg2+ and Mn2+ as described previously [12], which were based on studies of free magnesium [53] and estimated values necessary for magnesium and manganese transporter activation in B. subtilis [54,55].


Fig 2. FEN is more active on canonical double-flap substrates than Pol I.

(A) Products of nuclease activity of FEN, (B) FEND192N, or (C) Pol I on nicked, 5′ flaps with an abutting single nucleotide 3′ flap resolved using denaturing urea-PAGE. The mean for percentage of intact substrate across three replicates is graphically shown beneath the respective gel with error representing the standard deviation. Asterisks indicate significance as follows: * p<0.05, ** p<0.01, or *** p<0.001. (D) Representative gels of FENE114Q,D116N, FENSite1, or FEND189N,D192N activity on the described substrate from at least three replicates. For all assays, substrates consist of oligonucleotides oFCL11 and oJR368 with oJR339 (RNA-DNA hybrid; RNA indicated by zigzags) or oJR348 (DNA-only). Time points for all assays are 0 s, 10 s, 1 min, and 15 min with ladders generated by alkaline hydrolysis.


Fig 3. FEN and Pol I are active on gapped double-flap substrates.

(A) Nuclease assay results of FEN, (B) FEND192N, (C) or Pol I on gapped double-flap substrates separated by urea-PAGE. The mean percentage of uncut substrate is quantified under the respective gel, based on three replicates with error bars representing the standard deviation. Significance is indicated as follows: * p<0.05, ** p<0.01, or *** p<0.001. (D) Gels of FENE114Q,D116N, FENSite1, or FEND189N,D192N activity on the gapped double-flap substrate. Gels are representative of at least three replicates Substrates were made by annealing oFCL10 and oJR368 with oJR339 (RNA-DNA hybrid; RNA indicated by zigzags) or oJR348 (DNA-only). Time points for all assays are 0 s, 10 s, 1 min, and 15 min with ladders obtained by incubating the hybrid structure with sodium hydroxide.


Fig 4. FEN is active on flap overhang structures.

(A) Endonuclease activity of FEN, (B) FEND192N, or (C) Pol I on 5′ flap overhang structures visualized using denaturing urea-PAGE. The mean percentage of substrate remaining uncut by each protein across three replicates is quantified with standard deviation shown beneath a representative gel; asterisks indicate significance: * p<0.05, ** p<0.01, or *** p<0.001. (D) Assays of activity on 5′ flap structures by the catalytic mutants FENE114Q,D116N, FENSite1, or FEND189N,D192N. For all assays, flap substrates consist of oligonucleotides oJR366 and oJR368 with oJR339 (RNA-DNA hybrid; RNA indicated by zigzags) or oJR348 (DNA-only). Time points for all assays are 0 s, 10 s, 1 min, and 15 min with ladders generated by alkaline hydrolysis.


Fig 5. FEN can remove RNA from Okazaki fragment-like nicks.

(A) Visualization of nuclease activity of FEN, (B) FEND192N, or (C) Pol I on nicked duplex substrates. Three replicates of each assay are quantified beneath a representative urea-PAGE gel with standard deviation shown. Asterisks are used to indicate significance: * p<0.05, ** p<0.01, or *** p<0.001. (D) Results of incubation of nicked duplex substrates with the FEN mutants FENE114Q,D116N, FENSite1, or FEND189N,D192N. Nicked duplex substrates consist of oJR338 and oJR340 with oJR339 (RNA-DNA hybrid; RNA indicated by zigzags) or oJR348 (DNA). Ladders were generated by treating hybrid structure with sodium hydroxide. Assay time points for all reactions are 0 s, 10 s, 1 min, and 15 min.


Fig 6. FEN is more active on 3′ overhangs than Pol I.

(A) Nuclease activity of FEN, (B) FEND192N, or (C) Pol I on 3′ overhang structures shown using 20% urea-PAGE. The mean percent of substrate left uncut by the protein, based on three replicates, is shown beneath the respective gel. Significance is indicated by asterisks: * p<0.05, ** p<0.01, or *** p<0.001. Standard deviation across replicates is indicated by the black bars. (D) Failure of FEN mutants FENE1414Q,D116N, FENSite1, and FEND189N,D192N to cleave 3′ overhang substrates. Representative gels of at least three replicates are shown. Substrates assayed were generated by annealing oJR340 with oJR339 (RNA-DNA hybrid; RNA indicated by zigzags) or oJR348 (DNA). The triangle indicates time point progression of 0 s, 10 s, 1 min, and 15 min; the ladder was produced via alkaline hydrolysis of the hybrid substrate.


Fig 7. FEN has strong activity on blunt duplex substrates.

(A) Denaturing urea-PAGE gels showing assays of FEN, (B) FEND192N, or (C) Pol I activity on blunt duplex structures. The mean percent uncut substrate remaining from three replicates is quantified under the appropriate gel. Black bars represent standard deviation and asterisks symbolize significance: * p<0.05, ** p<0.01, or *** p<0.001. (D) Assays of FEN mutants: FENE1414Q,D116N, FENSite1, and FEND189N,D192N on the blunt substrates. Representative gels of at least three replicates are shown. Substrates were composed of oligonucleotides oJR365 and either oJR339 (RNA-DNA hybrid; RNA indicated by zigzags) or oJR348 (DNA only), with a ladder generated by treating the hybrid version with sodium hydroxide. Time points for all reactions are 0 s, 10 s, 1 min, and 15 min.

The first substrates we tested were variations of a 5′ flap, as this structure can be generated in vivo during strand-displacement synthesis by Pol I [56]. This structure is often associated with Okazaki-fragment maturation [27,28], nucleotide excision repair (NER) or long patch base excision repair (BER) [57]. The first substrate tested was a nicked double-flap, composed of a single nucleotide 3′ flap directly abutting a downstream 5′ flap (Fig 2). This is the canonical structure associated with FEN activity, as some FENs have been shown to contain a 3′ binding pocket, capable of binding a single displaced nucleotide from the upstream DNA strand [5861]. Utilization of this pocket stimulates protein activity, increasing substrate turnover [58,60]. As demonstrated by Fig 2A, FEN is capable of cleaving both the RNA-DNA hybrid and DNA-only versions of the substrate, with a preference for the hybrid as quantified below the gel. The FEND192N mutant was previously suggested to be catalytically inactive because it showed no activity when assayed at a lower concentration [13]. However, at a higher concentration, FEND192N generated similar products to those of the WT FEN, although it had reduced activity on both substrates (Fig 2B). This is consistent with the in vivo results shown in Figs 1E and S3. Surprisingly, Pol I had minimal activity on either version of the canonical substrate (Fig 2C). The other three FEN mutants surveyed in Fig 1 were also assayed on the nicked double-flap substrate and had no significant activity on either variant (Fig 2D). We conclude that FEN is much more active on the canonical substrate compared with Pol I.

FEN and Pol I have similar activity on gapped double-flap structures

The next substrate we investigated was like the previously described canonical structure, although the upstream and downstream pieces were separated by a four-nucleotide gap. WT FEN had strong activity on both substrate types, although it preferred the RNA-DNA hybrid (Fig 3A). The FEND192N mutant behaved similarly (Fig 3B), and the cleavage pattern of the hybrid suggests that FEN is less specific with the incision point of this substrate and may engage more in exonuclease activity. In a stark contrast to the nicked double-flap, Pol I had high activity on both the RNA-DNA hybrid and DNA-only substrates when the flaps were separated by a gap (Fig 3C). Even with this substrate there is a clear difference in the distribution of products produced by FEN and Pol I. FEN shows complete cleavage of the flap at earlier timepoints as compared with Pol I which engages in exonucleolytic degradation of the flap. The FENE114Q,D116N and FENSite1 mutants did not have significant activity on the substrates, however, the FEND189N,D192N mutant did have some weak activity on the hybrid substrate (Fig 3D). FEND189N,D192N activity was minimal, even compared to the FEND192N mutant, but suggests that our abrogation of the second metal binding site did not result in a complete loss of catalytic activity on the gapped double flap substrate.

FEN is more active on 5′ flap substrates than Pol I

We next tested a 5′ flap overhang with a four-nucleotide gap between the upstream piece and the flap junction (Fig 4). Unlike the prior substrates, this substrate lacks the 3′ flap. As shown in Fig 4A, FEN was able to cleave both versions of this substrate well, rapidly generating a low molecular weight product for both, although it had significantly more activity on the hybrid flap structure. The FEND192N mutant was also active on both substrates (Fig 4B), although they produced less product than the wild-type protein. Like FEN, Pol I preferred the hybrid flap substrate and generated a product consistent with flap endonuclease activity, however overall it had much lower catalytic activity than FEN (Fig 4C). The three remaining mutants, FENE114Q,D116N, FENSite1, and FEND189N,D192N show no noticeable activity on the substrate, consistent with their failure to rescue ΔrnhC, ΔfenA strains (Fig 4D). The activity of Pol I on the DNA substrate is more indicative of a role in DNA repair while the activity of FEN, cleaving the RNA-DNA hybrid to completion, is more indicative of a role for FEN in Okazaki fragment maturation.

Previous work has shown that some FENs have increased activity when potassium ions are bound to the protein [40,41]. We also assayed for nuclease activity on this simple flap structure in the presence of potassium. As shown in S7 Fig, neither FEN nor Pol I was appreciably stimulated by the addition of potassium to the reaction mixture.

FEN is more active on nicked substrates than Pol I

After testing the series of flapped intermediates, we investigated nuclease activity on a nicked substrate which would occur prior to flap formation (Fig 5). Nicks are a common type of DNA damage, where a missing phosphodiester bond occurs in otherwise continuous DNA. Nicks can result from many DNA repair pathways, including NER, BER, and mismatch repair [62]. Nicks can also be created following RNase H activity [63] or during Okazaki fragment maturation [64]. FEN showed high activity on the RNA-DNA hybrid nick substrate (Fig 5A), fully removing the rNMPs. While FEN was also fully active on the DNA-only substrate, it appears to use primarily 5′-3′ exonuclease activity, resulting in a sequential removal of nucleotides from the 5′ end rather than the single band associated with flap endonuclease activity. Regardless of the specific type of incision used, FEN has an overall preference for the hybrid structure. The FEND192N mutant (Fig 5B) shows both flap endonuclease and 5′ exonuclease activities on the hybrid structure, but not at the same rate observed for the WT enzyme. Despite the differences in activity, FEN and the FEND192N mutant generate similar products for each substrate by the 15-minute time point. Conversely, Pol I had much less activity on either variant of the nicked substrate than FEN (Fig 5C). Pol I engaged minimally in 5′ exonuclease activity on each nick, removing single nucleotides from each substrate. Interestingly, Pol I had significantly more activity on the DNA-only substrate than the hybrid, demonstrating a preference opposite to that shown by FEN. As before, the remaining FEN mutants appear catalytically inactive and do not have detectable activity on the nicked substrate (Fig 5D).

FEN is more active on 3′ overhang structures than Pol I

The final Okazaki fragment-like substrate we tested was a 3′ overhang (Fig 6). This substrate is primarily associated with Okazaki fragment maturation [15] but can be formed during other DNA processes [65]. FEN was highly active on the hybrid version of this substrate, which most closely mimics an Okazaki fragment, cleaving all rNMPs from more than 80% of the substrate within 10 seconds (Fig 6A). FEN was also active on the DNA-only substrate, although the cleavage pattern suggests that this is primarily 5′ exonuclease activity. As with the other substrates, the FEND192N mutant had reduced activity on both the hybrid and DNA-only substrates, producing products consistent with those of WT FEN (Fig 6B). Pol I had minimal exonuclease activity on either substrate, engaging primarily in 5′ exonuclease activity (Fig 6C). Despite the high activity of WT FEN on the 3′ overhang, the remaining FEN mutants were not demonstrably active, further highlighting the essentiality of these residues to FEN function (Fig 6D). These substrates: 5′ flaps (with or without a 3′ flap), nicked duplex DNA, and 3′ overhangs, represent the most common structures associated with Okazaki fragment maturation. In all cases, FEN is more active than Pol I on the RNA-DNA hybrid versions of the substrates tested. Based on these data, we suggest that FEN contributes significantly to the resolution of Okazaki fragments in vivo.

FEN is more active on blunt substrates than Pol I

In addition to Okazaki fragment-like substrates, we assayed the proteins for activity on less canonical substrates. The first of these (Fig 7) mimicked blunt DNA, which can occur in vivo after a nick is converted to a double-stranded break when encountered by a replication fork [66]. As Fig 7A shows, FEN had high activity on both the hybrid and DNA-only substrates, with a significant preference for the hybrid. Furthermore, FEN demonstrates different activities on the two variants of the substrate: on the hybrid it showed endonuclease activity and cleaved all the rNMPs while on the DNA-only version it cleaved exonucleolytically. The catalytically reduced FEND192N mutant produced similar results (Fig 7B), although the enzymatic efficiency was reduced. Pol I had no remarkable activity on either substrate (Fig 7C). The FEN mutants were not active on the blunt substrates (Fig 7D).

Pol I has more activity on 5′ overhangs than FEN

We also tested activity on a 5′ overhang structures (S8 Fig), which make up a portion of double-stranded breaks in the cell [66]. While not as striking as with the previous substrates, FEN showed exonuclease activity on both the hybrid and the DNA-only substrate (S8A Fig). This activity was not detected with the FEND192N mutant (S8B Fig), suggesting that the 5′ overhang is not a preferred substrate of FEN. The remaining FEN mutants also had no discernable activity on the substrates (S8D Fig). Unlike FEN, Pol I was able to perform endonucleolytic cleavage on both the hybrid structure and the DNA-only 5′ overhang (S8C Fig). Pol I had overall lower activity on the DNA-only substrate than the hybrid, however, a distinct band indicative of the removal of multiple nucleotides was present for both. The differences between the activities of FEN and Pol I on the 5′ overhang substrate are striking. Interestingly, FEN from bacteriophage T5 was shown to have activity on 5′ overhangs as seen here with Pol I, while FEN from Staphylococcus aureus behaved similarly to B. subtilis FEN [29]. This leads us to conclude that, while FEN and Pol I belong to the same family, these proteins have distinct roles in maintaining genome integrity.

FEN has more activity on single-stranded substrates than Pol I

The last construct that we investigated is a single-stranded substrate with no double-stranded regions (S9 Fig). It has been established that EcPol I 5′ to 3′ exonuclease activity requires duplex DNA [67], however this is not the case for all FEN proteins [29]. FEN had exonuclease activity on both versions of this substrate, but there was no detectable band indicative of full removal of the rNMPs (S9A Fig). The FEND192N mutant behaved similarly to WT FEN, but with reduced overall activity (S9B Fig). Pol I had negligible activity on either variant and did not create a distinct product, in agreement with prior work showing that neither Pol I from E. coli nor Thermus aquaticus is active on ssDNA [67,68]. While the three catalytically inactive FEN mutants initially appear to have some activity on this substrate (S9D–S9E Fig), assays lacking protein suggest that minor loss of signal is attributable to background degradation of the RNA substrate (S9G Fig) as it is more prone to spontaneous hydrolysis when single-stranded. We conclude that FEN is active on ssDNA and ssRNA/DNA hybrids.

Pol I nuclease activity is not stimulated with extension

Given the striking differences in the nuclease activities of FEN and Pol I, we asked whether these differences could be attributed to the presence of the Klenow fragment on Pol I. While FEN is composed of a single domain, Pol I also has the Klenow fragment, which is approximately twice as large as the Pol I FEN-domain. In the established model for Okazaki fragment maturation, there must be coordination between the FEN-domain and the Klenow fragment [21,64,69]. To account for this, we tested Pol I activity on a dual-labeled substrate mimicking an Okazaki fragment. As Fig 8 (left) shows, in the absence of dNTPs, Pol I was unable to extend the free 3′ end of the 5′-labeled substrate and did not noticeably cleave the 3′-labeled substrate. In the presence of dNTPs, Pol I was able to extend the 5′-labeled substrate and cleave minimal downstream nucleotides, while most of the substrate remained intact. We assayed FEN under the same conditions Fig 8 (middle) and found that it cleaved the primer-like portion of the 3′-labeled substrate regardless of whether dNTPs were included in the reaction buffer. Combining both proteins at equimolar concentrations Fig 8 (right) resulted in extension of the 5′-labeled substrate and cleavage of the 3′-labled substrate. Thus, even under conditions that promote engagement of both activities of Pol I, FEN had greater catalytic activity on the substrate, which is consistent with our biochemical characterization of FEN. Together, our biochemical work suggests that FEN is the major contributor to the repair of Okazaki fragments in B. subtilis.


Fig 8. Pol I has lower nuclease activity than FEN even with concurrent DNA synthesis.

Denaturing urea-PAGE of products generated by Pol I or FEN activity on a substrate resembling an Okazaki fragment, where upstream and downstream fragments are separated by ten bases. Extension by synthesis from the upstream, 3′ hydroxyl group is tracked by a 5′ fluorescent labeled substrate (green, oFCL8) while nuclease activity on the downstream Okazaki-fragment is tracked with the red colored, 3′ labeled oligonucleotide (oJR367). Labeled oligonucleotides were annealed to oFCL6. Each protein was tested with or without 50 μM dNTPs and time points of 0 s, 10 s, 1 min, and 15 min. Ladder was generated by adding oFCL11 annealed to oFCL6 to sodium hydroxide treated substrate. Images are representative of at least three replicates.

ΔpolA strain sensitivity is due to loss of the Klenow fragment

Unlike the ΔfenA strain described earlier, a single deletion of polA results in sensitivity to DNA damage from MMC and MMS [13,23,35]. It has been shown in Streptococcus pneumoniae and Haemophilus influenzae that 5′ to 3′ FEN of Pol I is essential, although these two bacteria lack fenA or any other protein with a FEN domain apart from Pol I [29,70,71]. Given our biochemical results, we hypothesized that the ΔpolA phenotype is due to the loss of the Klenow fragment rather than loss of the FEN-domain. Since prior work has shown that Pol I activity can be reconstituted even with the fragments physically separated [17], we constructed IPTG-inducible strains to express Pol I (polA), the Pol I FEN-domain (polAFEN), or the Pol I Klenow fragment (polAKlenow) in the ΔpolA background. As shown in Fig 9, overexpression of either full-length Pol I or the Klenow fragment was sufficient to rescue the polA deletion strain. Overexpressing the Pol I FEN domain did not reduce MMC sensitivity, indicating that the strong phenotypes associated with ΔpolA strains exposed to DNA damage is due to loss of polymerase activity rather than loss of nuclease activity. This result reaffirms that the FEN domain of Pol I is not critical for normal cell growth, and that FEN is sufficient even when cells are exposed to DNA damage.


Fig 9. Sensitivity of polA deletion strains is due to loss of the Klenow fragment.

Rescue of polA deletion strain sensitivity to mitomycin C (MMC) by ectopic, full-length Pol I (polA) or Pol I polymerase domains (polAKlenow and polAFEN). Plates were incubated at 30°C.

Pol I cannot directly substitute for FEN

We have established that FEN has more nuclease activity than Pol I on biologically relevant substrates assayed in vitro. We also considered the implications that this might have in the cell. As Fig 1E demonstrated, overexpression of FEN or the catalytically compromised FEND192N mutant was able to rescue the ΔrnhC, ΔfenA strain. This led us to question if the biochemical difference we show between FEN and Pol I was realized in cells. We created ΔrnhC, ΔfenA strains expressing IPTG-inducible fenA, fenAD192N, polA, polAFEN, or E. coli xni and assayed for rescue of HU sensitivity. As before, overexpression of either fenA or fenAD192N resulted in rescue (Fig 10). Neither the FEN domain of Pol I nor full-length Pol I were able to rescue the strain as well as fenA or fenAD192N, indicating that FEN’s function in cells cannot be fully compensated for by Pol I alone. Expression of xni, which encodes the E. coli FEN paralog ExoIX, also failed to rescue the strain and instead showed a small dominant negative phenotype, likely caused by binding SSB in B. subtilis or causing an inactive complex on substrates in vivo. This protein has been shown to have some catalytic activity [41] although it is missing three of the carboxylate residues that typify bacterial FENs (S2 Fig); as such, it may have reduced activity compared to FENs that contain all eight residues [29]. We conclude that the high catalytic activity of FEN cannot be directly substituted for by Pol I, especially in the absence of RNase HIII.


Fig 10. Expression of polA does not rescue deletions of fenA.

Ectopic expression of different proteins with FEN activity in a ΔrnhC, ΔfenA strain sensitive to hydroxyurea (HU). fenA and polA encode the full-length proteins from B. subtilis, fenAD192N and polAKlenow express derivatives of the native protein, and xni encodes the E. coli homolog of FEN, ExoIX, which lacks three of the carboxylate residues associated with bacterial FENs.


FEN, previously referred to as YpcP or ExoR, is a member of the large FEN family of proteins that are found across all domains of life [45,72]. These proteins play an essential role in genome maintenance, particularly during Okazaki fragment maturation as FEN activity is required for viability [73]. For bacteria including Streptococcus pneumonia and Haemophilus influenzae, the Pol I FEN domain is essential although these organisms lack a second FEN [29,70,71]. In the case of B. subtilis, an organism that encodes two proteins with FEN activity, at least one must be present for viability [35]. Our biochemical analysis reveals surprising differences between the nuclease activities of FEN and Pol I. We show that FEN has more nuclease activity than Pol I on substrates that resemble Okazaki fragments. This is the case even for canonical structures associated with Pol I activity, nicked double-flaps, 5′ flaps, nicks, and 3′ overhangs, with FEN showing more activity on RNA. In fact, across all tested structures except the 5′ overhang, FEN showed a strong preference for RNA-DNA hybrids over dsDNA further suggesting that FEN is the dominant nuclease during Okazaki fragment processing in B. subtilis. We conclude that FEN’s major biological function is the removal of RNA primers during lagging strand synthesis. There is evidence suggesting that FEN contributes to DNA repair, particularly during UV damage [49,50]. Given the poor activity of FEN on DNA substrates and the lack of a DNA damage phenotype for ΔfenA, we conclude that FEN does not provide a significant contribution to the repair of UV-induced DNA damage.

An important consideration regarding the activities of FEN and Pol I is that the latter is physically constrained by the Klenow fragment. The two major activities of Pol I are present in separate domains, however they are joined by a flexible linker. Biophysical modeling suggests that the Klenow fragment has lower affinity for the flap structure, which allows it to be replaced by the FEN domain [21,69,74,75]. To account for potential interference by the Klenow fragment, we also assayed for nuclease activity under conditions where extension occurs from an upstream 3′ hydroxyl. While extension by Pol I occurred in the presence of dNTPs, the nuclease activity of Pol I remained much lower than FEN under the same conditions. The relatively low nuclease activity of Pol I that we observe is consistent with other studies, as purifications of Pol I from B. subtilis were noted to have lower levels of nuclease activity than purifications from E. coli [26,76]. It is important to note that these early results were possibly influenced by proteolysis during protein purification [17,27,77]; to avoid this, the proteins used in this study were confirmed to be full length using SDS-PAGE (S6 Fig). Additionally, our group has previously shown that cleavage of an Okazaki fragment-like substrate by Pol I is minimal, and that repair of that substrate is increased when RNase HIII is included in the reaction [13]. Together, our biochemical data show that FEN is the major nuclease during Okazaki fragment maturation in B. subtilis, a role previously attributed to DNA polymerase I.

Despite FENs apparent role in Okazaki fragment maturation, ΔfenA cells do not demonstrate a detectable phenotype, though a ΔrnhC ΔfenA strain is more sensitive to genotoxic stress than either of the single deletions. With these results, we suggest that FEN is actively involved in the resolution of RNA-DNA hybrids, although some of this activity can be compensated for by other repair pathways [13]. Unlike the ΔfenA strain, ΔpolA strains are sensitized to DNA damage, including damage caused by MMC and UV [13,23,35]. Given our biochemical results, we suspect that this is due to loss of the polymerization activities of Pol I rather than loss of the nuclease. Indeed, overexpression of the Klenow fragment rescued the ΔpolA strain as well as full-length polA. Similarly, MMS sensitivity in E. coli was attributed to mutations affecting the polymerase activity of Pol I [25,78]. Given that Pol I is known to be broadly involved in the repair of DNA damage [62], it is likely that loss of the Klenow fragment and its strand-displacement synthesis activity leads to an accumulation of unrepaired DNA, resulting in genotoxic stress.

It has been accepted that Pol I is the major polymerase involved with Okazaki fragment maturation in bacteria [15,18], due to its dual function as polymerase and nuclease [15]. We show that ΔrnhC, ΔfenA strains are best rescued by WT FEN or a mutant with reduced catalytic activity (FEND192N), and that overexpression of Pol I or Pol IFEN fails to rescue the strain. Our work complements the results of Fukushima et al. [35], and further advances the idea that, while FEN and Pol I can degrade similar substrates, the proteins have different contributions and substrate preferences in the cell. We suggest that the removal of RNA primers during Okazaki fragment maturation in B. subtilis is carried out primarily by FEN while the upstream Okazaki fragment is extended by Pol I, as modeled in Fig 11. In the absence of FEN, the combined activities of Pol I and RNHIII could operate as a secondary pathway for maturation of the fragments yielding WT growth. We show that Pol I has levels of biochemical activity very similar to FEND192N, yet fenD192N complements the ΔrnhC, ΔfenA phenotype better than polA. We interpret this result to mean that while a complete loss of 5′ nuclease activity from FEN and Pol I is lethal [35], maximal FEN activity is not necessary for normal growth. We conclude that B. subtilis cells grow well when some FEN activity is provided, and the existence of two pathways for Okazaki fragment maturation explains the lack of an obvious phenotype for the ΔfenA strain.


Fig 11. A model of the proposed role of FEN in maturation of Okazaki fragments.

During lagging strand synthesis, maturation of Okazaki fragments can occur via two pathways. In the primary pathway (left, solid arrows), FEN removes RNA primers (pink) from Okazaki fragments while Pol I synthesizes from the upstream 3′ hydroxyl. DNA synthesis continues until Pol I loses affinity for the substrate and is outcompeted by DNA ligase, which seals the remaining nick. In the alternate pathway (right, dashed arrows), RNase HIII internally cleaves the RNA primer, shortening the RNA-DNA hybrid to facilitate Pol I nuclease activity. As Pol I approaches the remaining ribonucleotides, it can remove them with its FEN domain following strand-displacement synthesis or nick translation. Synthesis and nucleotide removal continue until Pol I is outcompeted by DNA ligase, which repairs the final nick.

When we compare the activities of B. subtilis Pol I to Pol I from other organisms, we find some interesting differences. B. subtilis Pol I lacks 3′ to 5′ exonuclease [23] activity and we show here that Pol I FEN is weak. For organisms like S. pneumoniae, H. influenzae or E. coli that either lack a discrete FEN or have a discrete FEN with little to no measurable activity (in the case of E. coli Exo IX), the Pol I FEN domains show robust nuclease activity [29,70,71]. We suggest that when bacteria contain an active, discrete FEN, the 5′ nuclease activity of Pol I is no longer necessary, and Pol I could diverge to a more specialized role in Okazaki fragment resynthesis and DNA repair. In the case of organisms that lack an active, discrete FEN, these bacteria remain reliant on the FEN activity of Pol I. Therefore, we suggest that the differences in Pol I activity observed between organisms with a robust, discrete FEN may reflect the adaptation and specialization of each organism’s genetic toolbox to manage the steps of Okazaki fragments and DNA repair more efficiently. For E. coli, S. pneumoniae and H. influenzae both tasks are completed by Pol I; for B. subtilis and the wide-spread group of bacteria that encode Pol I and FEN, we suggest that Okazaki fragments are matured using FEN and Pol I.

Our work shows that the major contribution of Pol I to the cell is through its DNA polymerase domain, for both Okazaki fragment maturation and DNA repair. Based on all available evidence, the contribution of FEN is specific to completion of Okazaki fragments. We propose that the combination of FEN and Pol I provide the most efficient process for primer removal and resynthesis during completion of the lagging strand. The two-protein system may provide B. subtilis with alternate pathways for lagging strand synthesis, which could be advantageous under conditions of genotoxic stress. Further, this mechanism, with separate proteins contributing FEN and DNA polymerase activity, is reminiscent of the mechanism of Okazaki fragment maturation employed by eukaryotic cells where FEN-1 (Rad27) and DNA polymerase δ coordinate during replication [73]. The maintenance of multiple members of the FEN family in the same organism can be seen beyond prokaryotes, as demonstrated by the FEN-1 and EXO1 proteins in eukaryotes [79]. As such, the differences between FEN and Pol I that we show here may support a broad pattern of specialized FENs in bacterial DNA replication and repair. Given the similarities we find between B. subtilis and eukaryotes, further study of lagging strand replication in B. subtilis may provide new mechanistic insights that will be relevant to Okazaki fragment processing in more complex systems.


Plasmid cloning

All primers used are listed in S2 Table. Plasmids were constructed using either pDR110 or pE-SUMO as the vector, which were amplified with oJR262 and 263 or oJR46 and 47, respectively. WT B. subtilis genes were amplified from PY79 genomic DNA, while FEN mutants were created using site-directed mutagenesis of plasmid carrying the WT gene. polAFEN or polAKlenow in the pDR110 vector were generated by selectively amplifying the appropriate regions from the polA-expressing plasmid. xni was amplified from E. coli BL21(DE3) cells. Genes were inserted into the appropriate vector using Gibson assembly, and subsequently used to transform E. coli MC1061 cells. Transformants with spectinomycin (pDR110 vector) or kanamycin (pE-SUMO vector) resistance were confirmed using colony PCR. Plasmid sequences were confirmed via Sanger sequencing.

B. subtilis strain construction

All strains used in this work are described in S3 Table, and all B. subtilis strains are derivatives of PY79. Competent B. subtilis was generated by inoculating LB (supplemented with 3 mM Mg2SO4) with one colony of desired strain and grown in a rolling rack at 37°C until A600nm ~ 0.7. Pre-warmed MD minimal media (1x PC buffer [10x PC buffer: 107 g/L K2HPO4, 60 g/L KH2PO4, 10 g/L trisodium citrate • (H2O)5], 2% glucose, 50 μg/mL tryptophan, 50 μg/mL phenylalanine, 11 μg/mL ferric ammonium citrate, 2.5 μg/mL potassium aspartate, 3 mM MgSO4) was inoculated with culture and grown in a rolling rack at 37°C for 6 hours. IPTG-inducible strains were constructed by transforming the appropriate parent strain with the NcoI linearized, pDR110-based plasmid containing the desired gene (plasmids are listed in S4 Table). Insertion at the amyE locus was confirmed by testing for a loss of starch-hydrolysis and through PCR amplification using oPEB866 and 867.

Protein expression

Competent BL21(DE3) were transformed with the appropriate plasmid (S4 Table), spread on LB agar supplemented with 25 μg/mL kanamycin, and incubated overnight at 37°C. A starter culture was made by inoculating LB containing 25 μg/mL kanamycin with a single colony, which was grown shaking at 200 rpm in a 37°C incubator overnight. A portion of the starter culture was used to inoculate each liter of LB supplemented with 25 μg/mL kanamycin. Cultures were grown at 37°C with shaking until A600nm ~ 0.6. Each liter of culture expressing FEN or FEN mutants was induced with 0.5 mM IPTG, cooled on ice for 20 min, and grown for 18 hours at 25°C. Cultures expressing Pol I were induced with 0.5 mM IPTG and grown at 37°C for 3 hours. Induced cells were harvested by centrifugation at 4°C and 4000 xg for 25 min. Supernatant was discarded and pellets were stored at -80°C.

Protein purification

Two 1 L pellets were thawed on ice, then resuspended in Lysis Buffer (20 mM Tris, pH 8.0, 400 mM NaCl, 1 mM DTT). One protease inhibitor tablet (Pierce Protease Inhibitor Tablets, EDTA-free, A32965) was added, then cells were sonicated on ice for 2.5 minutes total on time (cycles of 10 seconds on and 20 seconds off) at 4°C. Lysed cells were centrifuged at 4°C and 14,000 xg for 45 minutes, then supernatant was decanted into a clean conical. Ni-NTA agarose (Qiagen, 30210) resin was equilibrated with Lysis Buffer in a gravity column, before clarified lysate was passed over it. The column was washed with 5 volumes of Lysis Buffer, followed by a wash with 10 column volumes of Wash Buffer (20 mM Tris, pH 8.0, 2 M NaCl, 1 mM DTT, 15 mM imidazole). Fractions containing protein were eluted with 3 column volumes of Elution Buffer (20 mM Tris, pH 8.0, 400 mM NaCl, 1 mM DTT, 300 mM imidazole), and the most concentrated fractions were determined using a NanodropLite spectrophotometer. Selected fractions were pooled and treated with SUMO-protease for 2 hours at room temperature to remove the 6x-His-SUMO tag. Digested protein was transferred to 10K molecular cut off weight (MCOW) dialysis tubing and dialyzed overnight against Dialysis Buffer (20 mM Tris, pH 8.0, 300 mM NaCl, 1 mM DTT) at 4°C overnight. The protein was passed over a Ni-NTA gravity column equilibrated with Dialysis Buffer, and the flowthrough was collected. The column was further washed with 3 column volumes each of Dialysis Buffer, Wash Buffer, and Elution Buffer. Each flowthrough was collected, and samples were analyzed using a 10% SDS-PAGE gel stained with Coomassie brilliant blue. For all proteins, the initial flowthrough and dialysis-wash flowthrough was pooled and concentrated using a 10K MCOW centrifugal filter unit (Amicon, UFC9010). Protein was stored in 25% glycerol at -80°C. Further purification was completed using a HiTrap Q FF anion exchange column (Cytivia, 17515601) on an AKTA FPLC using Q Buffer A (20 mM Tris, 1 mM DTT, 5% glycerol) and Q Buffer B (20 mM Tris, 500 mM NaCl, 1 mM DTT, 5% glycerol). Briefly, the column was equilibrated with 10% Q Buffer B and protein was diluted to 50 mM NaCl in Q Buffer A before loading. Protein was fractionated using an increasing gradient of Q Buffer B. Peak fractions were determined using SDS-PAGE, then peak fractions were pooled and concentrated using a centrifugal filter. Glycerol was added to 25%, followed by flash freezing aliquots in liquid nitrogen before storing at -80°C. Gel demonstrating protein purity was generated using by loading 1 μg of each protein onto a Mini-PROTEAN TGX 4–20% gradient gel (BIO-RAD, 4561096) and then staining with Coomassie blue.

Nuclease activity assays

Substrates were generated by combining 1 μM labeled oligonucleotide with 2 μM of each appropriate unlabeled oligonucleotide in 1x Dilution Buffer (20 mM Tris, pH 8.0, 50 mM NaCl) and boiling for 1 minute at 98°C before being allowed to cool to room temperature. The following oligonucleotides (synthesized by IDT) were used: nicked double-flap substrates contained oJR368, oFCL12 and either oJR339 or oJR340, gapped double-flap substrates used oJR368, oFCL10, and either oJR339 or oJR348, 5′ flap substrates included oJR366, oJR368, and either oJR339 or oJR348, nicked substrates contained oJR338, oJR340 and either oJR339 or oJR348, 3′ overhang structures consisted of oJR340 and either oJR339 or oJR348, blunt substrates consisted of oJR365 and either oJR339 or oJR348, 5′ overhang structures were oJR365 with oFCL4 or oFCL5, and single-stranded oligos were either oJR339 or oJR348. Sequences for all oligonucleotides are listed in S1 Table. Purified proteins were diluted to 250 nM in 1x Metals Buffer (20 mM Tris, pH 8.0, 50 nM NaCl, 1 mM MgCl2, and 10 μM MnCl2). Each reaction contained 100 nM substrate, 50 nM protein and 1x Metals Buffer. Assays were performed at 25°C, and reactions were terminated with an equivalent volume of Stop Buffer (95% formamide, 20nM EDTA, bromophenol blue) at 0 s, 10 s, 1 min, and 15 min (unless otherwise noted in the figure caption). Stopped samples were incubated at 98°C for 5 minutes, immediately followed by snap-cooling on ice. Each ladder was generated by incubating 500 nM of the chimeric form of a substrate in 200 nM NaOH at 37°C for 8 minutes (5 minutes for the 5′ overhang), then stopped as described. Products were resolved using 20% denaturing urea-PAGE and visualized using the 800 nM channel of a LiCor Odyssey imager. Each reaction was repeated at least three times, using different preparations of substrate and different aliquots of protein. FEN and Pol I activity on the 5′ flap substrate in the presence of potassium (S7 Fig) was with the addition of 10 mM KCl to the 1x Metals Buffer. Our use of KCl was based on studies with E. coli ExoIX [41], although the referenced study used a higher KCl concentration.

Okazaki repair assays

Repair assays were performed using the method previously described [12,13]. Briefly, substrates were generated by combining 1 μM oJR367, 1 μM oFCL8, and 2 μM oFCL6 in 1x Dilution Buffer then boiling for 1 minute at 98°C and cooling to room temperature [9,13]. As noted in S1 Table, oFCL8 contains phosphorothioate bonds at the 5′ end to prevent off-target 5′ exonuclease activity. Reactions were carried out in 1x Extension Buffer (40 mM Tris-acetate, pH 7.8, 12 mM magnesium acetate, 300 mM potassium glutamate, 3 μM ZnSO4, 2% (w/v) polyethylene glycol, 0.02% pluronic F68), using 50 nM protein and 100 nM substrate, with or without 50 μM deoxynucleotide triphosphates (dNTPs). Assays were completed as described above for the Nuclease Activity Assays; however, an appropriate ladder was generated by combining the substrate (treated with sodium hydroxide) with an equal volume of 0.5 μM oFCL11 annealed to oFCL6. The dual-colored substrate was imaged using the 700 nM and 800 nM channels of a LiCor Odyssey imager.

Supporting information


  1. 1.
    Oivanen M, Kuusela S, Lönnberg H. Kinetics and mechanisms for the cleavage and isomerization of the phosphodiester bonds of RNA by Brønsted acids and bases. Chem Rev. 1998;98: 961–990. pmid:11848921
  2. 2.
    Thomas M, White RL, Davis RW. Hybridization of RNA to double-stranded DNA: Formation of R-loops. Proc National Acad Sci. 1976;73: 2294–2298. pmid:781674
  3. 3.
    Aguilera A, Gómez-González B. DNA–RNA hybrids: The risks of DNA breakage during transcription. Nat Struct Mol Biol. 2017;24: 439–443. pmid:28471430
  4. 4.
    Joyce CM. Choosing the right sugar: How polymerases select a nucleotide substrate. Proc National Acad Sci. 1997;94: 1619–1622. pmid:9050827
  5. 5.
    Yao NY, Schroeder JW, Yurieva O, Simmons LA, O’Donnell ME. Cost of rNTP/dNTP pool imbalance at the replication fork. Proc National Acad Sci. 2013;110: 12942–12947. pmid:23882084
  6. 6.
    Henrikus SS, Oijen AM van, Robinson A. Specialised DNA polymerases in Escherichia coli: Roles within multiple pathways. Curr Genet. 2018;64: 1189–1196. pmid:29700578
  7. 7.
    Okazaki R, Okazaki T, Sakabe K, Sugimoto K, Sugino A. Mechanism of DNA chain growth. I. Possible discontinuity and unusual secondary structure of newly synthesized chains. Proc National Acad Sci. 1968;59: 598–605. pmid:4967086
  8. 8.
    Rowen L, Kornberg A. A ribo-deoxyribonucleotide primer synthesized by primase. J Biological Chem. 1978;253: 770–4. pmid:340459
  9. 9.
    Sanders GM, Dallmann HG, McHenry CS. Reconstitution of the B. subtilis replisome with 13 proteins including two distinct replicases. Mol Cell. 2010;37: 273–281. pmid:20122408
  10. 10.
    Schroeder JW, Randall JR, Matthews LA, Simmons LA. Ribonucleotides in bacterial DNA. Crit Rev Biochem Mol. 2014;50: 181–193. pmid:25387798
  11. 11.
    McLean EK, Nye TM, Lowder FC, Simmons LA. The impact of RNA-DNA hybrids on genome integrity in bacteria. Annu Rev Microbiol. 2022;76. pmid:35655343
  12. 12.
    Randall JR, Hirst WG, Simmons LA. Substrate specificity for bacterial RNases HII and HIII is influenced by metal availability. J Bacteriol. 2018;200: e00401–e00415. pmid:29084857
  13. 13.
    Randall JR, Nye TM, Wozniak KJ, Simmons LA. RNase HIII is important for Okazaki fragment processing in Bacillus subtilis. J Bacteriol. 2019;201: e00686–e00699. pmid:30670546
  14. 14.
    Ogawa T, Okazaki T. Function of RNase H in DNA replication revealed by RNase H defective mutants of Escherichia coli. Mol Gen Genetics Mgg. 1984;193: 231–237. pmid:6319961
  15. 15.
    Kornberg A, Baker TA. DNA replication. New York: W.H. Freeman; 1992.
  16. 16.
    Setlow P, Brutlag D, Kornberg A. Deoxyribonucleic acid polymerase: Two distinct enzymes in one polypeptide I. A proteolytic fragment containing the polymerase and 3′ → 5′ exonuclease functions. J Biol Chem. 1972;247: 224–231.
  17. 17.
    Setlow P, Kornberg A. Deoxyribonucleic acid polymerase: Two distinct enzymes in one polypeptide II. A proteolytic fragment containing the 5′ → 3′ exonuclease function. Restoration of intact enzyme functions from the two proteolytic fragments. J Biol Chem. 1972;247: 232–240.
  18. 18.
    Okazaki R, Arisawa M, Sugino A. Slow joining of newly replicated DNA chains in DNA polymerase I-deficient Escherichia coli mutants. Proc National Acad Sci. 1971;68: 2954–2957. pmid:4943548
  19. 19.
    Lehman IR, Uyemura DG. DNA polymerase I: Essential replication enzyme. Science. 1976;193: 963–969. pmid:781842
  20. 20.
    Nagata Y, Mashimo K, Kawata M, Yamamoto K. The roles of Klenow processing and flap processing activities of DNA polymerase I in chromosome instability in Escherichia coli K12 strains. Genetics. 2002;160: 13–23. pmid:11805041
  21. 21.
    Xie P, Sayers JR. A model for transition of 5′-nuclease domain of DNA polymerase I from inert to active modes. Plos One. 2011;6: e16213. pmid:21264264
  22. 22.
    Sayers JR. Computer aided identification of a potential 5′-3′ exonuclease gene encoded by Escherichia coli. J Theor Biol. 1994;170: 415–421. pmid:7996866
  23. 23.
    Duigou S, Ehrlich SD, Noirot P, Noirot-Gros M. DNA polymerase I acts in translesion synthesis mediated by the Y-polymerases in Bacillus subtilis. Mol Microbiol. 2005;57: 678–690. pmid:16045613
  24. 24.
    Kiefer JR, Mao C, Hansen CJ, Basehore SL, Hogrefe HH, Braman JC, et al. Crystal structure of a thermostable Bacillus DNA polymerase l large fragment at 2.1 Å resolution. Structure. 1997;5: 95–108. pmid:9016716
  25. 25.
    Gass KB, Hill TC, Goulian M, Strauss BS, Cozzarelli NR. Altered deoxyribonucleic acid polymerase activity in a methyl methanesulfonate-sensitive mutant of Bacillus subtilis. J Bacteriol. 1971;108: 364–374.
  26. 26.
    Gass KB, Cozzarelli NR. Further genetic and enzymological characterization of the three Bacillus subtilis deoxyribonucleic acid polymerases. J Biol Chem. 1973;248: 7688–7700.
  27. 27.
    Tamanoi F, Okazaki T, Okazaki R. Persistence of RNA attached to nascent short DNA pieces in Bacillus subtilis cells defective in DNA polymerase I. Biochem Bioph Res Co. 1977;77: 290–297. pmid:407908
  28. 28.
    Kurosawa Y, Ogawa T, Hirose S, Okazaki T, Okazaki R. Mechanism of DNA chain growth XV. RNA-linked nascent DNA pieces in Escherichia coli strains assayed with spleen exonuclease. J Mol Biol. 1975;96: 653–664. pmid:1104864
  29. 29.
    Allen LM, Hodskinson MRG, Sayers JR. Active site substitutions delineate distinct classes of eubacterial flap endonuclease. Biochem J. 2009;418: 285–292. pmid:19000038
  30. 30.
    Tomlinson CG, Atack JM, Chapados B, Tainer JA, Grasby JA. Substrate recognition and catalysis by flap endonucleases and related enzymes. Biochem Soc T. 2010;38: 433–437. pmid:20298197
  31. 31.
    Harrington JJ, Lieber MR. Functional domains within FEN-1 and RAD2 define a family of structure-specific endonucleases: implications for nucleotide excision repair. Gene Dev. 1994;8: 1344–1355. pmid:7926735
  32. 32.
    Harrington JJ, Lieber MR. The characterization of a mammalian DNA structure-specific endonuclease. Embo J. 1994;13: 1235–1246. pmid:8131753
  33. 33.
    Kimura S, Takenouchi M, Hatanaka M, Seto H, Kouroku Y, Sakaguchi K. An ATP-inhibited endonuclease from cauliflower (Brassica oleracea var. botrytis) inflorescence: Purification and characterization. Planta. 1998;206: 641–648.
  34. 34.
    Sommers CH, Miller EJ, Dujon B, Prakash S, Prakash L. Conditional lethality of null mutations in RTH1 that encodes the yeast counterpart of a mammalian 5′- to 3′-exonuclease required for lagging strand DNA synthesis in reconstituted systems. J Biol Chem. 1995;270: 4193–4196. pmid:7876174
  35. 35.
    Fukushima S, Itaya M, Kato H, Ogasawara N, Yoshikawa H. Reassessment of the in vivo functions of DNA polymerase I and RNase H in bacterial cell growth. J Bacteriol. 2007;189: 8575–8583. pmid:17905985
  36. 36.
    Larsen E, Gran C, Sæther BE, Seeberg E, Klungland A. Proliferation failure and gamma radiation sensitivity of Fen1 null mutant mice at the blastocyst stage. Mol Cell Biol. 2003;23: 5346–5353. pmid:12861020
  37. 37.
    Burkhart BW, Cubonova L, Heider MR, Kelman Z, Reeve JN, Santangelo TJ. The GAN exonuclease or the flap endonuclease Fen1 and RNase HII are necessary for viability of Thermococcus kodakarensis. J Bacteriol. 2017;199: e00141–17. pmid:28416706
  38. 38.
    Finger LD, Atack JM, Tsutakawa S, Classen S, Tainer J, Grasby J, et al. The wonders of flap endonucleases: Structure, function, mechanism and regulation. The eukaryotic replisome: A guide to protein structure and function. 2012. pp. 301–326. pmid:22918592
  39. 39.
    Uson ML, Carl A, Goldgur Y, Shuman S. Crystal structure and mutational analysis of Mycobacterium smegmatis FenA highlight active site amino acids and three metal ions essential for flap endonuclease and 5′ exonuclease activities. Nucleic Acids Res. 2018;46: gky238–. pmid:29635474
  40. 40.
    AlMalki FA, Flemming CS, Zhang J, Feng M, Sedelnikova SE, Ceska T, et al. Direct observation of DNA threading in flap endonuclease complexes. Nat Struct Mol Biol. 2016;23: 640–646. pmid:27273516
  41. 41.
    Anstey-Gilbert CS, Hemsworth GR, Flemming CS, Hodskinson MRG, Zhang J, Sedelnikova SE, et al. The structure of Escherichia coli ExoIX—implications for DNA binding and catalysis in flap endonucleases. Nucleic Acids Res. 2013;41: 8357–8367. pmid:23821668
  42. 42.
    Murante RS, Rust L, Bambara RA. Calf 5′ to 3′ exo/endonuclease must slide from a 5′ end of the substrate to perform structure-specific cleavage. J Biol Chem. 1995;270: 30377–30383. pmid:8530463
  43. 43.
    Garforth SJ, Sayers JR. Structure-specific DNA binding by bacteriophage T5 5′→3′ exonuclease. Nucleic Acids Res. 1997;25: 3801–3807. pmid:9380501
  44. 44.
    Hosfield DJ, Mol CD, Shen B, Tainer JA. Structure of the DNA repair and replication endonuclease and exonuclease FEN-1 coupling DNA and PCNA binding to FEN-1 activity. Cell. 1998;95: 135–146. pmid:9778254
  45. 45.
    Kaiser MW, Lyamicheva N, Ma W, Miller C, Neri B, Fors L, et al. A comparison of eubacterial and archaeal structure-specific 5′-exonucleases. J Biol Chem. 1999;274: 21387–21394. pmid:10409700
  46. 46.
    Kao H-I, Henricksen LA, Liu Y, Bambara RA. Cleavage specificity of Saccharomyces cerevisiae flap endonuclease 1 suggests a double-flap structure as the cellular substrate. J Biol Chem. 2002;277: 14379–14389. pmid:11825897
  47. 47.
    Williams RS, Kunkel TA. FEN nucleases: Bind, bend, fray, cut. Cell. 2011;145: 171–172. pmid:21496637
  48. 48.
    Thomaides HB, Davison EJ, Burston L, Johnson H, Brown DR, Hunt AC, et al. Essential bacterial functions encoded by gene pairs. J Bacteriol. 2006;189: 591–602. pmid:17114254
  49. 49.
    Hernández-Tamayo R, Oviedo-Bocanegra LM, Fritz G, Graumann PL. Symmetric activity of DNA polymerases at and recruitment of exonuclease ExoR and of PolA to the Bacillus subtilis replication forks. Nucleic Acids Res. 2019;47: 8521–8536. pmid:31251806
  50. 50.
    Patlán AG, Corona SU, Ayala-García VM, Pedraza-Reyes M. Non-canonical processing of DNA photodimers with Bacillus subtilis UV-endonuclease YwjD, 5′→3′ exonuclease YpcP and low-fidelity DNA polymerases YqjH and YqjW. Dna Repair. 2018;70: 1–9. pmid:30096406
  51. 51.
    Ceska TA, Sayers JR. Structure-specific DNA cleavage by 5′ nucleases. Trends Biochem Sci. 1998;23: 331–336. pmid:9787638
  52. 52.
    Wozniak KJ, Simmons LA. Hydroxyurea induces a stress response that alters DNA replication and nucleotide metabolism in Bacillus subtilis. J Bacteriol. 2021;203: e00171–21. pmid:34031038
  53. 53.
    Froschauer EM, Kolisek M, Dieterich F, Schweigel M, Schweyen RJ. Fluorescence measurements of free 6Mg2+9 by use of mag-fura 2 in Salmonella enterica. Fems Microbiol Lett. 2004;237: 49–55. pmid:15268937
  54. 54.
    Wakeman CA, Goodson JR, Zacharia VM, Winkler WC. Assessment of the requirements for magnesium transporters in Bacillus subtilis. J Bacteriol. 2014;196: 1206–1214. pmid:24415722
  55. 55.
    Kliegman JI, Griner SL, Helmann JD, Brennan RG, Glasfeld A. Structural basis for the metal-selective activation of the manganese transport regulator of Bacillus subtilis. Biochemistry-us. 2006;45: 3493–3505. pmid:16533030
  56. 56.
    Kelly RB, Atkinson MR, Huberman JA, Kornberg A. Excision of thymine dimers and other mismatched sequences by DNA polymerase of Escherichia coli. Nature. 1969;224: 495–501.
  57. 57.
    Baute J, Depicker A. Base excision repair and its role in maintaining genome stability. Crit Rev Biochem Mol. 2008;43: 239–276. pmid:18756381
  58. 58.
    Xu Y, Potapova O, Leschziner AE, Grindley NDF, Joyce CM. Contacts between the 5′ nuclease of DNA polymerase I and its DNA substrate. J Biol Chem. 2001;276: 30167–30177. pmid:11349126
  59. 59.
    Chapados BR, Hosfield DJ, Han S, Qiu J, Yelent B, Shen B, et al. Structural basis for FEN-1 substrate specificity and PCNA-mediated activation in DNA replication and repair. Cell. 2004;116: 39–50. pmid:14718165
  60. 60.
    Finger LD, Blanchard MS, Theimer CA, Sengerová B, Singh P, Chavez V, et al. The 3′-flap pocket of human flap endonuclease 1 is critical for substrate binding and catalysis. J Biol Chem. 2009;284: 22184–22194. pmid:19525235
  61. 61.
    Pauszek RF, Lamichhane R, Singh AR, Millar DP. Single-molecule view of coordination in a multi-functional DNA polymerase. Elife. 2021;10: e62046. pmid:33704066
  62. 62.
    Wozniak KJ, Simmons LA. Bacterial DNA excision repair pathways. Nat Rev Microbiol. 2022;20: 465–477. pmid:35210609
  63. 63.
    Vaisman A, McDonald JP, Noll S, Huston D, Loeb G, Goodman MF, et al. Investigating the mechanisms of ribonucleotide excision repair in Escherichia coli. Mutat Res Fundam Mol Mech Mutagen. 2014;761: 21–33. pmid:24495324
  64. 64.
    Lundquist RC, Olivera BM. Transient generation of displaced single-stranded DNA during nick translation. Cell. 1982;31: 53–60. pmid:6218882
  65. 65.
    Lenhart JS, Schroeder JW, Walsh BW, Simmons LA. DNA repair and genome maintenance in Bacillus subtilis. Microbiol Mol Biol R. 2012;76: 530–564. pmid:22933559
  66. 66.
    Ayora S, Carrasco B, Cárdenas PP, César CE, Cañas C, Yadav T, et al. Double-strand break repair in bacteria: A view from Bacillus subtilis. Fems Microbiol Rev. 2011;35: 1055–1081. pmid:21517913
  67. 67.
    Cozzarelli NR, Kelly RB, Kornberg A. Enzymic synthesis of DNA XXXIII. Hydrolysis of a 5′-triphosphate-terminated polynucleotide in the active center of DNA polymerase. J Mol Biol. 1969;45: 513–531. pmid:4917367
  68. 68.
    Lyamichev V, Brow MAD, Varvel VE, Dahlberg JE. Comparison of the 5′ nuclease activities of Taq DNA polymerase and its isolated nuclease domain. Proc National Acad Sci. 1999;96: 6143–6148. pmid:10339555
  69. 69.
    Xu Y, Grindley NDF, Joyce CM. Coordination between the polymerase and 5′-nuclease components of DNA polymerase I of Escherichia coli. J Biol Chem. 2000;275: 20949–20955. pmid:10806216
  70. 70.
    Díaz A, Lacks SA, López P. The 5′ to 3′ exonuclease activity of DNA polymerase I is essential for Streptococcus pneumoniae. Mol Microbiol. 1992;6: 3009–3019. pmid:1336089
  71. 71.
    Bayliss CD, Sweetman WA, Moxon ER. Destabilization of tetranucleotide repeats in Haemophilus influenzae mutants lacking RnaseHI or the Klenow domain of PolI. Nucleic Acids Res. 2005;33: 400–408. pmid:15653640
  72. 72.
    Mueser TC, Nossal NG, Hyde CC. Structure of bacteriophage T4 RNase H, a 5′ to 3′ RNA–DNA and DNA–DNA exonuclease with sequence similarity to the RAD2 family of eukaryotic proteins. Cell. 1996;85: 1101–1112. pmid:8674116
  73. 73.
    Balakrishnan L, Bambara RA. Flap endonuclease 1. Annu Rev Biochem. 2013;82: 119–138. pmid:23451868
  74. 74.
    Turner RM, Grindley NDF, Joyce CM. Interaction of DNA polymerase I (Klenow fragment) with the single-stranded template beyond the site of synthesis. Biochemistry-us. 2003;42: 2373–2385. pmid:12600204
  75. 75.
    Delagoutte E, Hippel PH von. Function and assembly of the bacteriophage T4 DNA replication complex: Interactions of the T4 polymerase with various model DNA constructs. J Biol Chem. 2003;278: 25435–25447. pmid:12700227
  76. 76.
    Okazaki T, Kornberg A. Enzymatic synthesis of deoxyribonucleic acid: XV. Purification and properties of a polymerase from Bacillus subtilis. J Biological Chem. 1964;239: 259–68.
  77. 77.
    Falaschi A, Kornberg A. Biochemical studies of bacterial sporulation II. Deoxyribonucleic acid polymerase in spores of Bacillus subtilis. J Biol Chem. 1966;241: 1478–1482.
  78. 78.
    Lucia PD, Cairns J. Isolation of an E. coli strain with a mutation affecting DNA polymerase. Nature. 1969;224: 1164–1166. pmid:4902142
  79. 79.
    Vallur AC, Maizels N. Distinct activities of Exonuclease 1 and Flap endonuclease 1 at telomeric G4 DNA. Plos One. 2010;5: e8908. pmid:20126648
  80. 80.
    Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag New York; 2016. Available from:
  81. 81.
    Team RC. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2022. Available from:

Source link