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

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.

109 In this work, we examined the substrate preference for Pol I and FEN using a variety of RNA-DNA 110 hybrids and DNA only substrates. We also examined the phenotype of polA and fenA during 111 genotoxic stress and in RNase H deficient backgrounds to determine the contribution of Pol I and 112 FEN 5′ nuclease activity to genome integrity. We found that FEN shows the most robust activity on 113 RNA-DNA hybrid substrates modeling Okazaki fragment intermediates and that the strong polA 114 phenotypes are rescued by simply expressing the C-terminal domain lacking 5′ nuclease activity. With 115 our results, we conclude that discrete B. subtilis FEN functions as the major Okazaki fragment 116 nuclease, with Pol I DNA polymerase activity contributing to both DNA repair and Okazaki fragment 117 resynthesis. Because discrete FENs are present in a wide-range of bacterial genomes, we suggest 118 that Pol I-independent FENs provide the 5′ nuclease activity important for RNA removal during 119 lagging strand replication in many bacteria. Our work suggests that lagging strand processing in these 120 bacteria occurs through a process more similar to eukaryotes than previously appreciated. 129 hydroxyurea (HU) or cells exposed to UV radiation (S1 Fig). It was previously shown that fenA 130 disruption does not confer cold sensitivity nor sensitivity to mitomycin C (MMC) or methyl 131 methanesulfonate (MMS) [13]. Given these results, it seems clear that B. subtilis FEN does not have 132 an appreciable role during DNA repair in vivo. Another model is that FEN is involved in Okazaki 133 fragment maturation. If true, we would expect a strain lacking both rnhC and fenA to confer a 134 phenotype. Ribonuclease HIII (RNHIII), encoded by rnhC, is a protein known to participate in Okazaki 135 fragment maturation and the resolution of RNA-DNA hybrids [12,13]. The absence of both proteins 136 resulted in cells that were more sensitive to HU or UV than either single deletion (Fig 1A and S1 Fig), 137 suggesting that FEN could be involved in the resolution of RNA-DNA hybrids on the lagging strand, 138 and any defects due to the absence of FEN are exacerbated with the combined absence of RNHIII 139 and presence of genotoxic stress. To test the importance of these amino acids, we designed a series of mutations to disrupt 145 conserved residues and expressed each variant from the ectopic chromosomal amyE locus under the 146 Pspank promoter, which allows induction of expression in the presence of isopropyl ß-D-1-147 thiogalactopyranoside (IPTG). As discussed above, the single deletion of fenA did not result in an 148 obvious phenotype, so mutants were expressed in the ΔrnhC, ΔfenA background.

149
150 Overexpression of WT fenA rescued the double deletion phenotype (Fig 1D and S3 Fig). With the 151 ΔrnhC, ΔfenA complementation assay established, four fenA mutants were created, two for each 152 metal-binding site. The first mutant, fenA E114Q,D116N , was designed to abrogate binding of the Site 1 153 bivalent cation without changing the overall shape of the active site. The binding at Site 1 was also 154 altered in the second mutant, where the EADD motif (residues 114 to 117) was changed to AAAA, a 155 construct designated as fenA Site1 . In addition to loss of cation binding, we predict that this mutant 156 would have alterations to the shape of the active site due to the hydrophobic residues, as well as 157 changes to the flexible arch, since sequence similarity suggests that the D117 residue interacts with 158 R82 to form a salt-bridge [50]. As shown in Fig 1D and S3 Fig, neither  161 Previous work suggested that this mutant was catalytically inactive [13], however, overexpression in 162 the double deletion strain rescues to nearly the same degree as the WT gene (Fig 1D and S3 Fig). 163 Adding a second mutation, creating fenA D189N,D192N , once again led to a failure to rescue the ΔrnhC, 164 ΔfenA strain. Together this suggests that not every conserved residue is critical for activity, however, 165 perturbations that significantly affect metal-binding render fenA inactive in vivo.

166
167 An interesting observation from our experiments is that expression of fenA E114Q,D116N or fenA D189N,D192N 168 in the ΔrnhC, ΔfenA strain results in cells that are more sensitive to HU than the parent strain or those 169 expressing WT fenA (Fig 1D). To test if these mutations were dominant negative, we used the same 170 system to express each mutant or the WT allele from an ectopic locus in WT or a ΔrnhC strain. There We speculate that the metal-binding variants are still able to bind substrates in vivo but are 176 unable to bind one of the metal ions and therefore lack the catalytic activity required for turnover. We

211
212 FEN is more active on nicked substrates than Pol I 213 We next investigated nuclease activity on a nicked substrate (Fig 3) 217 showed high activity on the RNA-DNA hybrid nick substrate (Fig 3A), fully removing the rNMPs.
218 While FEN was also fully active on the DNA-only substrate, it appears to use primarily 5′-3′ 219 exonuclease activity, resulting in a sequential removal of nucleotides from the 5′ end rather than the 220 single band associated with flap endonuclease activity. Regardless of the specific type of incision 221 used, FEN had an overall preference for the hybrid structure. The FEN D192N mutant (Fig 3B) shows 222 both flap endonuclease and 5′ exonuclease activities on the hybrid structure, but not at the same rate 223 observed for the WT enzyme. Despite the differences in rate, FEN and the FEN D192N mutant generate 224 similar products for each substrate by the 15-minute time point. Conversely, Pol I had markedly less 225 activity on either variant of the nicked substrate than FEN (Fig 3C). Pol I engaged minimally in 5′ 226 exonuclease activity on each nick, removing single nucleotides from each substrate. Interestingly, Pol 227 I had significantly more activity on the DNA-only substrate than the hybrid, demonstrating a 228 preference opposite that of FEN. As before, the remaining FEN mutants appear catalytically inactive 229 and do not have detectable activity on the nicked substrate (Fig 3D).
230 231 FEN is more active on 3′ overhang structures than Pol I 232 The third canonical substrate we tested was a 3′ overhang (Fig 4). This substrate is primarily 233 associated with Okazaki fragment maturation [15] but can be formed during other cellular processes 234 [58]. FEN was highly active on the hybrid version of this substrate, which most closely mimics an 235 Okazaki fragment, cleaving all rNMPs from more than 80% of the substrate within 10 seconds (Fig   236 4A). FEN was also active on the DNA-only substrate, although the cleavage pattern suggests that this 237 is primarily 5′ exonuclease activity. As with the other substrates, the FEN D192N mutant had reduced 238 activity on both the hybrid and DNA-only substrates, producing products consistent with those of WT 239 FEN (Fig 4B). Pol I had minimal exonuclease activity on either substrate, engaging primarily in 5′ 240 exonuclease activity (Fig 4C). Despite the high activity of WT FEN on the 3′ overhang, the remaining 241 FEN mutants were not demonstrably active, further highlighting the essentiality of these residues to 242 the protein's function (Fig 4D). These substrates: 5′ flaps, nicked duplex DNA, and 3′ overhangs, 243 represent the three most common structures associated with Okazaki fragment maturation. We  255 5B), although the enzymatic efficiency was reduced. Pol I had no remarkable activity on either 256 substrate (Fig 5C). The FEN mutants were not active on the blunt substrates (Fig 5D).

257
258 Pol I has more activity on 5′ overhangs than FEN 259 We also tested activity on a 5′ overhang structures (Fig 6), which make up a portion of double 260 stranded breaks in the cell [59]. While not as striking as with the previous substrates, FEN showed 261 exonuclease activity on both the hybrid and the DNA-only substrate (Fig 6A). This activity was not 262 detected with the FEN D192N mutant (Fig 6B), suggesting that the 5′ overhang is not a preferred 263 substrate of FEN. The remaining FEN mutants also had no discernable activity on the substrates (Fig   264 6D). Unlike FEN, Pol I was able to perform endonucleolytic cleavage on both the hybrid structure and 265 the DNA-only 5′ overhang (Fig 6C). Pol I had overall lower activity on the DNA-only substrate than 266 the hybrid, however, a distinct band indicative of the removal of multiple rNMPs was present for both. 272 273 FEN has more activity on single-stranded substrates than Pol I 274 The last construct that we investigated is a single-stranded substrate with no double-stranded regions 275 (Fig 7). It has been established that EcPol I 5′ to 3′ exonuclease activity requires duplex DNA [60], 276 however this is not the case for all FEN proteins [29]. FEN had exonuclease activity on both versions 277 of this substrate, and there was no detectable band indicative of full removal of the rNMPs (Fig 7A).
278 The FEN D192N mutant behaved similarly to WT FEN, but again with reduced overall activity (Fig 7B) 281 While the three catalytically inactive FEN mutants initially appear to have some activity on this 282 substrate (Fig 7D and S7A Fig) 313 the strong phenotypes associated with ΔpolA strains exposed to DNA damage is due to loss of 314 polymerase activity rather than loss of the protein's nuclease activity. This result reaffirms that the 315 FEN domain of Pol I is not critical for normal cell growth, and that FEN is sufficient even when cells 316 are exposed to DNA damage.

317
318 Pol I cannot directly substitute for FEN 319 We have established that, in general, FEN has more nuclease activity than Pol I on biologically 320 relevant substrates assayed in vitro. We considered the implications that this might have in the cell. 325 As before, overexpression of either fenA or fenA D192N resulted in rescue (Fig 10) D192N , or (C) Pol I on single-stranded RNA-DNA hybrid or DNA. Percent of intact substrate was quantified from three replicates and shown below the appropriate gel. Significance is indicated by asterisks (* p<0.05, ** p<0.01, or *** p<0.001) and standard error bars are provided. (D) FEN mutants were assayed on the same substrate, with representative gels shown. Single-stranded RNA-DNA hybrid (RNA indicated by zigzags) was oligonucleotide oJR339 while single-stranded DNA was oJR348. 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 10 s, 1 m, and 15 m. Ladder was generated by adding oJR363 and oJR364 to sodium hydroxide treated substrate.  D192N and polA Klenow 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 .   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.  D116N and fenA D189N,D192N is not detectable in a WT background. WT B. subtilis ectopically expressing fenA or fenA mutants with the indicated changes. Cells were imaged after growth at 30˚C for 16 hours. Fig. fenA E114Q,D116N and fenA D189N,D192N are dominant negative in the absence of rnhC. Overexpression of ectopic fenA or fenA mutants in cells lacking rnhC.