Physical and functional interplay between PCNA DNA clamp and Mre11–Rad50 complex from the archaeon Pyrococcus furiosus

Abstract

Several archaeal species prevalent in extreme environments are particularly exposed to factors likely to cause DNA damages. These include hyperthermophilic archaea (HA), living at temperatures >70°C, which arguably have efficient strategies and robust genome guardians to repair DNA damage threatening their genome integrity. In contrast to Eukarya and other archaea, homologous recombination appears to be a vital pathway in HA, and the Mre11–Rad50 complex exerts a broad influence on the initiation of this DNA damage response process. In a previous study, we identified a physical association between the Proliferating Cell Nuclear Antigen (PCNA) and the Mre11–Rad50 (MR) complex. Here, by performing co-immunoprecipitation and SPR analyses, we identified a short motif in the C- terminal portion of Pyrococcus furiosus Mre11 involved in the interaction with PCNA. Through this work, we revealed a PCNA-interaction motif corresponding to a variation on the PIP motif theme which is conserved among Mre11 sequences of Thermococcale species. Additionally, we demonstrated functional interplay in vitro between P. furiosus PCNA and MR enzymatic functions in the DNA end resection process. At physiological ionic strength, PCNA stimulates MR nuclease activities for DNA end resection and promotes an endonucleolytic incision proximal to the 5′ strand of double strand DNA break.

INTRODUCTION

Extremophile organisms provide remarkable study systems for understanding cellular processes that allow them to live in conditions likely to cause a high rate of DNA damage. Several archaeal species prevalent in extreme environments are particularly exposed to such stressors, including hyperthermophilic archaea (HA) living at temperatures >70°C. However, several studies have shown that HA such as Pyrococcus furiosus and Sulfolobus solfataricus can fully restore their genomes if they are fragmented by γ-radiation (1–3). They arguably have efficient strategies and robust genome guardians to repair DNA damage threatening genome integrity. These guardians are proteins working and interacting together in a carefully orchestrated ballet. Archaea employ ‘eukaryotic’ DNA replication and repair complex proteins (4), but several DNA repair protein families that are broadly conserved among Bacteria and Eukarya have not been found in Archaea (5). Using in silico, genetic or biochemical approaches, several studies revealed new actors or new complexes involved in genomic integrity in Archaea (6–15), leading to the characterization of new helicases and nucleases, like Hel308/Hjm, GAN, NucS/EndoMS and more recently NerA, thus, improving our understanding of genomic maintenance processes in Archaea. To contribute to this effort to discover new actors or new complexes involved in archaeal genomic integrity, we characterized a protein-protein interaction network sustaining genome maintenance in Pyrococcus abyssi (15). In this previous study, we identified a physical association between the Proliferating Cell Nuclear Antigen (PCNA) and the Mre11–Rad50 (MR) complex: the characterization of this association is addressed in the present study.

Structurally conserved between Archaea and Eukarya, PCNA is a multimeric, ring-shaped factor that encircles DNA duplex. Archaeal Replication Factor C complex (RFC), which functions as a clamp loader, stimulates PCNA assembly around DNA even though archaeal PCNA can spontaneously load in vitro onto DNA (16,17). First reported as a processivity factor for DNA polymerases, PCNA is essential for cell viability. As a DNA-clamp, PCNA is a moving platform for numerous partners involved in DNA replication and repair pathways (18). Extensive lists of PCNA partners have been given in reviews (19–21).

The MR complex has a broad influence on the DNA damage response network, especially on repair of DNA Double-Strand Breaks (DSB) (22). As DSBs are a particularly threatening type of DNA damage, induced by external agents as well as by internal molecular events, cells have evolved a highly sophisticated DNA damage response system. For the recognition and repair of DNA breaks, the two major mechanisms are Homologous Recombination (HR) and Non-Homologous End-Joining (NHEJ). In HR, the broken ends are resected into 3′ single-strand tails and then used as templates for a homology search; whereas in NHEJ, the broken ends are directly rejoined (for reviews see (23–25)). In Eukarya, DSBs are repaired by either HR or NHEJ depending on the cell cycle (26). NHEJ was assumed to be absent from Archaea until Bartlett et al. reconstituted an archaeal NHEJ apparatus in vitro similar to that of bacterial machinery (27). In contrast to eukaryotes and other archaea, HR appears to be a vital pathway in HA, since genetic analyses have shown that the mre11, rad50 and radA genes are essential for Thermococcus kodakaraensis and Sulfolobus islandicus (28,29).

The eukaryotic macromolecular machine MR(N/X) is composed of two core proteins, Mre11 and Rad50, with an additional component: Nbs1 for higher eukaryote or Xrs2 for yeast, which are found in neither Bacteria nor Archaea. Both Mre11 and Rad50 are highly conserved in all three domains and even exist as gp46/47 in some virus such as T4 phage (30). The MR complex engages the HR pathway by tethering and resecting DNA ends through a combination of nuclease and ATPase activities tightly related to conformational changes (31). However, the MR complex nuclease functions, 3′→5′ double strand (ds) DNA exonuclease activity and single strand (ss) DNA endonuclease activity, are not sufficient to generate a long 3′ ssDNA tail and require additional partners to catalyse efficient DSB resection (32). Moreover, the MR complex appears to be essential in replication fork restart in eukaryote cells, but to date the biochemical and regulation mechanisms remain partially understood.

Given the role of PCNA to orchestrate DNA replication and other DNA processes, we wondered whether this newly discovered interaction with the MR complex would regulate MR enzymatic functions in the DNA end resection process. By performing co-immunoprecipitation and SPR analyses, we demonstrated physical association between P. furiosus (Pfu) PCNA and the MR complex and identified a short motif in the C terminal portion of PfuMre11 that interacts with PfuPCNA and corresponds to a variation on the PCNA-Interaction Peptide (PIP) motif theme. Enzymatic assays, at physiological ionic strength, showed that PfuPCNA stimulates nuclease activity of the PfuMR complex on dsDNA substrates and promotes an endonucleolytic incision proximal to the 5′ strand of a DNA double strand break in a manner still consistent with HR process requirements.

MATERIALS AND METHODS

Proteins and peptides

A gene coding for P. furiosus PCNA was inserted into pET19b to add an N-terminal 10xHis-tag (plasmid provided by B. Connolly (33)) and expressed in Escherichia coli Rosetta pLysS. The P. furiosus MR complex was coexpressed from a bicistronic pET27b vector (gift from J. Tainer and T. Paull (34)) in BL21 DE3 codonplus E. coli adding a 6xHis-tag in the N-terminal region of Mre11. This P. furiosus Mre11–Rad50 plasmid was used to build a P. furiosus Mre11–Rad50 ΔPIP mutant (1–411) using Q5^® Site-Directed Mutagenesis Kit (BioLabs). P. furiosus Mre11–Rad50^link1, Rad50^link2 and Mre11^core proteins were provided by G. Williams and J. Tainer.

Cells were grown at 37°C to an OD₆₀₀ 0.7–0.8, and expression was induced by addition of 1 mM IPTG (final concentration). Four hours after induction, cells were harvested by centrifugation and re-suspended in a buffer containing (i) for PfuPCNA proteins: 10 mM Tris–HCl pH 7.4, 500 mM NaCl, 20 mM imidazole, 1 mM DTT, (ii) for PfuMR wt and PfuMR ΔPIP: 20 mM Tris–HCl pH 8.0, 500 mM NaCl, 10 mM imidazole, 1 mM DTT, supplemented by EDTA-free protease inhibitor (Roche). Cells were lysed by applying 1.9 kbar pressure (One shot, Constant Systems). PfuPCNA, PfuMR wt and ΔPIP supernatants were incubated overnight with DNase I at 37°C and then heated at 75–80°C for 20 min. After centrifugation, soluble fractions were loaded onto a HisPrepFF 16/10 (GE Healthcare) nickel resin column. After a wash step, elution was performed with a linear gradient from 10 to 500 mM imidazole. Peak fractions were run on 15% SDS-PAGE gels (Bio-Rad), then pooled and concentrated using Vivaspin columns (10 or 30 MWCO) before running on a Superdex 200 10/300 GL column (GE Healthcare). PfuPCNA was eluted in 10 mM Tris–HCl pH 7.4, 200 mM NaCl, 1 mM DTT and PfuMR wt and ΔPIP in 20 mM Tris–HCl pH 8.0, 150 mM NaCl, 1 mM DTT complemented with 20% glycerol and stored at −20°C. Proteins were quantitated using DC protein assay (BioRad) for PfuPCNA, and absorbance measurement at 280 nm for PfuMR wt and ΔPIP complexes. All molar concentrations indicated in this study corresponded to the homotrimeric PCNA and to the heterotetrameric form of the M₂R₂ complex.

Peptides to be used as competitors for PCNA-binding in pull-down assays were synthesized and purified (>90% purity) by Genepep (St-Jean-de-Védas, France). The PIP-like Mre11 peptide was derived from the PfuMre11 sequence (412–424): Ac-KKKRGTLDSWLGG-NH₂. The peptide used as the negative control (PIP–) was: Ac-KEVKEEYKRFLEE-NH₂ (12).

Co-immunoprecipitation experiments

To study the physical interaction between PfuPCNA and PfuMR wt or ΔPIP, 8.33 μg of anti-P. abyssi (Pab) PCNA polyclonal antibody was immobilized onto 1.5 mg of magnetic Dynabeads Protein A (ThermoFischer). Subsequently, antibodies were covalently anchored using 57 μg of BS3 crosslinker (Thermo Scientific); such an amount of beads was determined in order to specifically bind 1 μg of PfuPCNA input (data not shown). In a 20 μl reaction volume, 1 μg PfuPCNA was incubated for 30 min at 4°C with 10 μg PfuMR wt or ΔPIP complexes in binding buffer (25 mM HEPES pH 7.0, 150 mM NaCl, 1 mM DTT, 0.05% Tween 20). The resulting protein complexes were trapped by anti-PCNA Dynabeads over 10 min at 4°C. Beads were washed three times with binding buffer (100 μl) before final elution for 10 min at 95°C in denaturating XT loading buffer (Bio-Rad). Proteins were then separated on SDS-PAGE (4–20% Pierce) and visualized using Coomassie Blue dye. Pull-down assays were performed following the same procedure for the three PfuMR constructs, with the exception that protein complexes were formed during 1 h at 4°C and incubated with the bead-antibody complex for 5 min at 4°C and then washed three times at 25°C. A 1:3 molar ratio PfuPCNA:PfuMR proteins was used. As input controls, 1 μg of proteins was loaded onto SDS-PAGE.

For co-immunoprecipitation experiments in presence of PIP-like peptide, in 20 μl reaction, 1 μg PfuPCNA was incubated with a 100 molar excess of competitor peptide or negative control peptide at least 1 h at 4°C in binding buffer. Then, 10 μg PfuMR complex was added to the PCNA/peptide solution for 5 min at 4°C. The resulting protein complexes were trapped by anti-PCNA Dynabeads, washed three times with 50 μl binding buffer at 4°C and eluted as described above. After SDS-PAGE separation, proteins were transferred onto a PVDF membrane (Thermo Scientific). PfuPCNA and PfuMre11 were simultaneously probed using anti-His monoclonal antibody (Invitrogen). Proteins were revealed by immunofluorescence using an ECL 2 blot kit (Thermo Scientific). Image acquisition was done with a ChemiDoc XRS+ (BioRad) and quantifications carried out using QuantityOne software (Bio-Rad).

DNA substrates

Oligonucleotides were purchased from Eurogentec and purified by RP-HPLC for S50/50, S50/50s, S50s/50s, S87/87s and S87s/87s or by PAGE for substrates containing reporter-quencher pairs, RQ-S87s/87s and RQ23-S87s/87s (Sequences in supplementary data). DNA substrates were annealed, at a 1:1 primer:template ratio, in presence of 10 mM HEPES pH 7.5 and 100 mM NaCl by heating at 95°C for 5 min and cooling to room temperature.

Nuclease assays

Nuclease activities of PfuMR wt and ΔPIP complexes on linear dsDNA substrates was followed in 10 μl reactions containing 25 nM DNA in 25 mM HEPES pH 7.0, 1 mM DTT, 0.5 mg/ml BSA complemented with 1 mM ATP, 5 mM MgCl₂, 5 mM MnCl₂ and 150 or 300 mM NaCl, as indicated in the figure captions. Pre-incubation was performed with 25 nM dsDNA substrates and the indicated concentrations of PfuPCNA at ambient temperature for 5 min followed by an incubation with the indicated concentrations of PfuMR wt or ΔPIP complexes at 70°C for different times as indicated in the figures. Reactions were stopped by addition of 85% deionized formamide, 0.01 N NaOH, 10 mM EDTA, 2 μM Trap (RC50 or RC87) and by heating samples at 95°C for 5 min. DNA products were separated by electrophoresis on a gel composed of 15% or 18% polyacrylamide 19:1, 7 M urea, 16% deionized formamide and 1× Tris Borate EDTA (TBE). Labeled fragments were analysed with a fluorimager Typhoon 9500 (GE Healthcare) and quantified with Image Quant software.

Real time fluorescence DNA unwinding assays

Unwinding assays were carried out at 55°C using dsDNA RQ-S87s/87s and RQ23-S87s/87s, which contain a fluorophore-quencher pair (6-FAM, 6-carboxyfluorescein/DDQI, Deep Dark Quencher I or BHQ-1, Black Hole Quencher 1) positioned at the DNA end or 23 nt from the extremity. Emission of fluorescence was triggered by unwinding quenched DNA duplex substrates. Unwinding assays were performed in 50 μl of 25 mM HEPES pH 7.0 buffer containing 25 nM DNA, 300 mM NaCl, 1 mM DTT, 0.5 mg/ml BSA, 5 mM MgCl₂ and 500 nM Trap 3′ 87RC, complemented with 1 mM ATP and 5 mM MnCl₂ when indicated. 50 nM PfuPCNA were pre-incubated with the DNA mix for 5 min at ambient temperature before adding 25 nM PfuMR wt or ΔPIP. Fluorescence emission was monitored using Q-PCR equipment (StepOnePlus™ Real Time PCR System Thermo Fisher Scientific). After 30 min, the temperature was increased to 95°C to induce complete unwinding to determine the maximum fluorescence intensity (100% unwinding signal, Q_max).

Unwinding percentage was calculated as follows: , where Q was the real-time detected fluorescence and Q₀ corresponds to the fluorescence measured at the beginning of the reaction. Unwinding assays were repeated at least three times.

SPR experiments

Data were obtained using a Reichert SR7000DC spectrometer instrument (Reichert Inc., Buffalo, NY, USA). The running buffer was 25 mM HEPES pH 7.0, 300 mM NaCl, 1 mM DTT and 0.05% Tween 20, and flow rate was 25 μl/min. PfuPCNA was immobilized on a mixed self-assembled monolayer (10% C11-(OEG)6-COOH: 90% C11-(OEG)3-OH), Reichert Inc.) via classic amine coupling chemistry and 25 nM of PfuMR wt or ΔPIP complexes were injected over the PfuPCNA surface at 25°C. The chip was regenerated after serial injections of 100 mM H₃PO₄ (3 × 30 s). Each curve displayed was double referenced with a set of blank buffer injections.

RESULTS

P. furiosus PCNA physically interacts with the MR complex

We first considered using the homologous system with P. abyssi recombinant proteins but failed to achieve a proper level of production of the MR complex. As an alternative we used the recombinant proteins from the close species Pyrococcus furiosus. Complexes of PfuPCNA and PfuMR from P. furiosus were produced and purified (Supplementary Figure S1A). Using co-immunoprecipitation, we demonstrated that these two components also formed a complex in P. furiosus, as shown by the co-precipitation of the PfuMR complex with PfuPCNA (Figure 1A). We then explored some conditions where this association could take place. As illustrated in Supplementary Figure S1C, the complex could form in presence or absence of a metallic co-factor and ATP. As protein samples were DNA free after DNase I treatment, we assumed that the interaction between PfuMR complex and PfuPCNA is not dependent on DNA substrates.

Figure 1.

Physical association of P. furiosus PCNA and Mre11–Rad50. Protein-protein interactions were determined, in vitro, using a bead-based co-immunoprecipitation assay. (A) Full length PfuMR co-immunoprecipitated with PfuPCNA, (B) co-immunoprecipitation assays with PfuPCNA and different PfuMre11 and PfuRad50 protein constructs (details given in Supplementary Figure S1B). 1 μg of protein was loaded on SDS-PAGE as Input. IP corresponds to the immunoprecipitation assays in presence of beads coated with PCNA antibodies. Fraction bound to the beads were analysed by Coomassie blue staining. Assays were performed in buffer with 150 mM NaCl.

To investigate the surface of interaction, we were able to use three PfuMre11 and PfuRad50 deletion constructs (35). PfuMre11^core (residues 1–342) lacks 84 C-terminal residues including its Rad50 binding domain (RBD); PfuRad50-link2, an untagged version of PfuRad50 with shortened coiled coils (unable to bind Mre11) connected by an intramolecular Gly-Gly-Ser-Gly-Gly sequence; and PfuMre11–Rad50-link1 a complex of another shortened version of PfuRad50 (able to bind Mre11 RBD) purified with full-length PfuMre11 (Supplementary Figure S1B). Only the PfuMre11–Rad50-link1 construct could form a complex with PfuPCNA in solution (Figure 1B, compare lanes 4, 8 and 12). These data indicated that PfuPCNA does not interact directly with PfuRad50 and that the coiled-coil domain of PfuRad50 is not required for recruitment of the MR complex onto PfuPCNA. On the other hand, these observations raised two non-exclusive hypotheses: PfuPCNA/MR interaction requires prior PfuMR complex formation and/or is mediated by the C-terminal region of PfuMre11 absent in the Mre11^core construct. With the exception of a domain interacting with the base of PfuRad50’s coiled-coils (348–381), the C-terminal domain of PfuMre11 is predicted to be disordered or flexible and is thought to be responsible for protein-protein and protein–DNA interactions (31).

The C-terminal region of P. furiosus Mre11 contains a putative PCNA interacting motif

PCNA-binding partners generally possess a PCNA-Interaction Peptide (PIP) motif, usually located at the extreme N- or C-terminus (36). The results obtained prompted us to look for a potential PIP motif at the C-terminal portion of PfuMre11. The core element of the archaeal PIP-box is a peptide with a sequence Qxxϕ (ϕ being hydrophobic residues L, M or I), which in most cases is C-terminally flanked by the sequence xxΩΩ (Ω being aromatic residues F or W) (37). Based on the alignment of PIP-Box like sequences (QXXϕXXΩΩ) from a subset of Pyrococcus sp. proteins whose affinity for PCNA has already been described (19), we identified a candidate PCNA-interacting peptide in the C-terminal region of PfuMre11 (Figure 2A). Located in the extreme C-terminal portion of PfuMre11 (positions 412–422) the motif exhibits the conserved hydrophobic residues but lacks the otherwise conserved glutamine residue. In addition, N-terminal extension from the motif is composed of a stretch of basic residues known to interact with the positively charged outer surface of PCNA (38). As Meslet-Cladiere et al. described, high affinity peptides for PCNA tend to be positively charged (12). Here the identified peptide has a predicted Isoelectric Point (pI) of 10.29 consistent with this property. 3D structure of the complete C-terminal Mre11 region has not been resolved to date; however, Hydrophobic Cluster Analysis identified this motif as a globular region and showed that the hydrophobic residues clustered with a similar shape to that observed for canonical PIP motif sequences (data not shown). Although this motif lacks the glutamine conserved residues, these features strongly suggest that it could act as a hydrophobic anchor on PCNA.

Figure 2.

Identification of a putative PCNA binding motif in the C-terminal region of PfuMre11. (A) PfuMre11 displays a C-terminal PIP-like motif when compared with the consensus sequence (x, any residues; Φ, hydrophobic; Ω, aromatic or hydrophobic) and various sequences of proteins from Pyrococcus sp. known to interact with PCNA through the PIP motif. Positions within the sequences are indicated and PfuMre11 domains are illustrated below the table. (B) The interaction PfuPCNA/MR is inhibited by an excess of Mre11 C-terminal peptide in competitive co-immunoprecipitation assays. Western blot signals of PfuMre11 were normalized with corresponding signals obtained with a peptide control. Experiments were performed in binding buffer containing 150 mM NaCl. Error bars represent standard deviation of three independent reactions. (C) Surface plasmon sensograms obtained after injection of 25 nM PfuMR wt or ΔPIP over an immobilised PfuPCNA surface. The running buffer contained 300 mM NaCl.

We then looked at occurrence of this motif in Mre11 sequences among species of the order Thermococcales. Remarkably, all available sequences displayed this putative PCNA-interacting motif in the C-terminal region (Supplementary Figure S2). From this alignment, we could derive a pattern [PK]-x-[KRNA]-x-[GSPNK]-x(1,3)-[IL]-x(2)-[WFY]-[ILV]) for a motif search using Scanprosite on archaeal protein sequences from Swiss-prot and TrEMBL databases. This pattern was detected in Mre11 sequences from the order Thermococcales and from an Archeoglobale, Archeoglobus fulgidus. We also identified that Hel308/Hjm archaeal members of helicase superfamily 2 possessed a similar motif in the extreme C-terminal region (Supplementary Figure S2). Presence of this PIP-like motif in Hel308/Hjm sequences showed similar repartition as in sequences of Mre11, as only Thermococcale Hel308/Hjm helicases harboured the motif, with the exception of Thermococcus litoralis. In addition, alignment of the C-terminal region of Hel308/Hjm and Mre11 proteins showed a strong conservation of hydrophobic residues in the C-terminal extension of the motif (Supplementary Figure S2), suggesting that this structural element could also serve as a hydrophobic plug on PCNA surface. Most noticeably, interaction between PCNA and Hel308/Hjm from P. furiosus has already been described. Using a deletion mutant lacking the 20 residues at the extreme C-terminal region (39), the authors proposed that this portion could mediate interaction with PCNA which is consistent with our analysis.

Mre11 C-terminus region is essential for PfuPCNA/MR interaction

Taken together, these observations prompted us to verify the assumption that the C-terminus of PfuMre11 contributes to PfuPCNA/MR complex formation. To this aim, we performed pull-down competitive assays with the Mre11 peptide corresponding to the revealed motif (412-KKKRGTLDSWLGG-424). Figure 2B shows that an excess of the Mre11 peptide (Mre11 PIP-like) significantly blocked PfuPCNA/MR interaction compared with the control peptide. As shown in the graph, the amount of immunoprecipitated Mre11 decreased drastically in the assay with the competitive peptide suggesting that this peptide inhibits assembly of the PfuPCNA/MR complex. To test whether the C-terminal sequence of PfuMre11 could mediate interaction of the PfuMR complex with PfuPCNA, we used Surface Plasmon Resonance (SPR) with PfuPCNA immobilized on a chip. SPR measurements indicated that the Mre11 peptide physically interacts with PfuPCNA at a micromolar range of concentrations (Supplementary Figure S3A). It is interesting to note that the binding value (4.05 μM) correlates with K_D value obtained with a canonical PIP motif peptide derived from the sequence of another nuclease, P. abyssi NucS (40). Together, these results suggest that the primary docking site of PfuMre11 on PfuPCNA could be similar to that described for PIP-motif containing proteins.

To confirm the essentiality of the PIP-like motif for the interaction, we produced a deleted version of the PfuMR complex lacking the last 15 amino acids of the PfuMre11 subunit (Supplementary Figure S3B, left panel). The co-IP experiments clearly showed that the mutant PfuMR ΔPIP complex did not bind to PfuPCNA (Supplementary Figure S3B, right panel, compare lanes 4 and 10), providing conclusive evidence that this variation in the PIP motif mainly contributes to the interaction of the PfuMR complex with PfuPCNA. We then looked at the stability of the interaction at a higher salt condition to test specificity and to get closer to the reported physiological ionic strength of Pyrococcus furiosus (41). Supplementary Figure S3B indicates that the interaction between PfuMR and PfuPCNA is stable at 300 mM NaCl (lanes 5–6) and that the mutant PfuMR ΔPIP complex did not interact with PfuPCNA under the same conditions (lanes 11–12). As a final point on the physical interaction, the direct association between PfuPCNA and the full length PfuMR complex was confirmed by SPR analysis at 300 mM NaCl. PfuMR specifically bound to PfuPCNA anchored on a sensor chip, while the PfuMR ΔPIP complex did not bind under the same conditions (Figure 2C). In addition, we conducted a kinetic experiment over the full range of PfuMR complex concentrations and determined an apparent dissociation constant value (K_Dapp) of ∼4.07 ± 1.46 nM (Supplementary Figure S3C).

PfuPCNA stimulates PfuMR activity for dsDNA cleavage

In archaea, as in some other organisms, the MR complex is implicated in early steps of the HR pathway. Thanks to a combination of Mre11 nuclease and Rad50 ATPase activities, the MR complex initiates DNA end resection to provide suitable DNA template used by subsequent HR components. In several in vitro studies, PfuMre11 displayed distinct activities: 3′→5′ dsDNA exonuclease, ssDNA endonuclease and endonucleolytic cleavage on the 5′ strand at a break (30,34,42–44). Given PCNA preference for dsDNA substrate, we tested influence of the PfuPCNA on PfuMR complex nuclease activities to resect synthetic dsDNA oligonucleotides.

To this end, we performed nuclease assays on a linear blunt-end dsDNA substrate (S50/50s). The unlabeled complementary strand has phosphorothioate bonds at its 3′ end to block 3′→5′ exonuclease activity, in order to characterize exo- and endonuclease activities on the top strand irrespective of exonuclease degradation of the complementary strand (Figure 3A). We first confirmed that the nuclease activities of the PfuMR complex were in accordance with previously reported activities at low salt conditions (Supplementary Figure S4A) and observed DNA products as already described on the same dsDNA substrates (42). Intracellular salt concentration was determined to be ∼350 mM in P. furiosus (41), we then decided to test the activity of the PfuMR complex and impact of PfuPCNA at 300 mM NaCl, close to the reported ionic strength. Increased concentrations of PfuMR complex were incubated for 30 min at 70°C with dsDNA substrate. DNA fragments from the 5′-labeled strand was revealed by fluorescence (Figure 3B). As shown in lanes 6–7, the 5′-labeled strand was degraded by the PfuMR complex, although to a much lesser extent than at low salt condition (compare with Supplementary Figure S4A), and defined products ranging from 15 to 37 nt were generated.

Figure 3.

PfuPCNA stimulates DNA degradation activity of PfuMR. In vitro nuclease assays (A) 5′-labeled 50bp dsDNA substrate used. SSSSS represents phosphorothioate bonds. Black arrows indicate potential endo- and exo-nuclease cleavage activities. (B) 25 nM of DNA substrate were incubated at 70°C for 30 min with increasing concentrations of PfuMR wt. (C) 25 nM of DNA substrate were pre-incubated with indicated concentrations of PfuPCNA at room temperature for 5 min before adding 25 nM PfuMR wt. Reactions were performed for 30 min at 70°C. (D) 25 nM of DNA substrate were pre-incubated with 50 nM PfuPCNA at room temperature for 5 min before adding 25 nM PfuMR wt or ΔPIP. Reactions were performed for 30 min at 70°C in 300 mM NaCl. DNA products were resolved in 18% PAGE and fluorescence revealed using Typhoon 9500 (GE Healthcare).

From a PfuMR:DNA ratio of 1:1, conditions in which PfuMR was inactive, we tested increased concentrations of PfuPCNA and observed strong activation of DNA degradation activity (Figure 3C, lanes 7–9) with about 94% of the substrate used. Moreover, addition of PfuPCNA, changed the degradation pattern and increased the specificity of the enzyme towards generation of major products ranging from 27 to 29 nucleotides (compare lanes 7 in Figure 3B and Figure 3C). A time course experiment confirmed that the products accumulated over time and were the major end products of the reaction (Supplementary Figure S4C). As expected, this activation was not observed in presence of the mutant PfuMR ΔPIP (Figure 3D, compare lanes 3 and 5). To confirm that the interaction of PfuPCNA with the PfuMR complex is responsible of nuclease activation, we demonstrated that the ΔPIP mutant is not affected in its nuclease activity (compare Figure 3B and C with Supplementary Figure S4 D and E), indicating that direct interaction was necessary to stimulate PfuMR nuclease activity. However, faint bands at 27–29 nt could still be observed on Figure 3D when PfuMR ΔPIP was in presence of PfuPCNA, indicating that the secondary domain of interaction might account for a weak resilient interaction with PfuPCNA.

At 150 mM NaCl, we also a noticed change in the degradation pattern of DNA caused by PfuPCNA, but this shift came with a strong inhibition of substrate utilization by PfuMR, contrary to what was observed at 300 mM NaCl, and most importantly, we found no significant differences between PfuMR wt and ΔPIP activities in presence of PfuPCNA (Supplementary Figure S4B). These led us to test, for both salt concentrations, the ability of PfuMR to bind dsDNA. By performing EMSA experiments, we confirmed its binding onto DNA at 150 and 300 mM NaCl and noticed no significant change caused by PfuPCNA (Supplementary Figure S5), indicating that nuclease inhibition or activation effect was not due to an improvement or a blockage of PfuMR fixation onto DNA substrates. This demonstrated the relevance of characterizing the functional interplay between PfuMR and PfuPCNA in higher salt concentration conditions than for previous reported characterizations of the MR complex from P. furiosus, especially as we were approaching physiological ionic strength with this treatment (41).

We also explored the effect of PfuPCNA on metal dependence of nucleolytic degradation. Mre11 has two metal binding sites for which manganese has a higher affinity (43) whereas Rad50 needs magnesium to hydrolyse ATP (45). Supplementary Figure S6A showed that at 70°C, dsDNA nucleolytic cleavage strongly required both ATP and manganese and that change in cleavage specificity observed in presence of PfuPCNA was not dependent on the presence or absence of magnesium. Hereinafter we focus on our data obtained at 300 mM NaCl, with 10 mM ATP, 5 mM MgCl₂ and 5 mM MnCl₂ (similar results were obtained with KCl instead of NaCl, data not shown).

From the degradation pattern observed, it is tempting to speculate that this product occurred through endonucleolytic cleavage, which might be promoted upon association with PfuPCNA. However, at this point it is not possible to conclude whether PfuPCNA had an effect on the exonuclease, the endonuclease or regulated both activities of the PfuMR complex.

Cleavage of the 5′-terminated DNA strand is promoted by PfuPCNA/MR association

To address this question, we used dsDNA substrates protected against exonuclease activity at both 3′ ends (S50s/50s), which consequently could be only degraded by endonuclease activity. As observed previously, major DNA products of 27–29 nt were generated from blocked 3′ end substrate (Figure 4A), confirming that the pattern observed stems from the initial endonuclease cut. To question whether phosphorothioates bonds in synthetic oligonucleotides may have an impact on PfuPCNA effect, we used unprotected blunt dsDNA substrate. Supplementary Figure S7A clearly shows that addition of PfuPCNA generated a similar change of cleavage specificity in that it led to the accumulation of main products around 27–29 nt. We then asked whether a non-specific protein, such as streptavidin, bound to the 5′ end might bring about a change in cleavage pattern or efficiency. To this end, we tested a biotinylated synthetic oligonucleotide with streptavidin to block the 5′ end (Supplementary Figure S7B) and noticed no cleavage pattern difference on this particular substrate. Finally, we looked at cleavage products from 3′ end labeled substrates. The generated fragment sizes (21–23 nt) were complementary to 5′ end labeled products (27–29 nt) confirming the specificity of the endonucleolytic cut (Supplementary Figure S7C and D).

Figure 4.

PfuPCNA/MR interaction generates endonucleolytic cleavage at dsDNA ends. In vitro nuclease assays. 25 nM of DNA substrate were pre-incubated with 50 nM PfuPCNA at room temperature for 5 min before adding 25 nM PfuMR wt or ΔPIP. Reactions were performed for 30 min at 70°C. DNA products were resolved in 18% PAGE and fluorescence revealed using Typhoon 9500 (GE Healthcare). DNA substrates used are indicated at the top of the panels. (A) 50 bp DNA substrate with a phosphorothioate stretch at both 3′ ends and (B) 87 bp DNA substrate with protected (right panel) or unprotected (left panel) 3′ end.

Next, we set out to examine whether there was any sequence and/or length dependence concerning the DNA substrate. To address this, we used blunt end dsDNA substrate of 87 bp that differed in sequence from the shorter substrate. Comparison of nuclease activity of the PfuMR complex in presence of PfuPCNA clearly showed a similar activation on this longer substrate, either blocked or accessible from the 3′ end (Figure 4B). Regardless of the size and sequence of the DNA substrates, PfuPCNA stimulated the nuclease activity of the PfuMR complex at the 5′ end of a DNA break to generate major 27–29 nt products through its direct interaction via the PIP-like motif. This suggests that alteration of nuclease activity by PfuPCNA seems to be promiscuous and independent of length or sequences of substrates. These results are in agreement with the mechanism by which the endonuclease activity of PfuMre11 acts to incise the 5′ strand DNA close to the DSB ends (46). Taken together, these data suggest that upon association with PfuPCNA, PfuMR could promote endonucleolytic activity to process DNA ends.

After cleavage, the endonucleolytic product is released via PfuMR/PCNA strand opening activity

Next, we investigated whether the endonucleolytic cleavage was coupled with DNA melting by the PfuMR complex. As ATP-dependent DNA unwinding activity has already been reported for both prokaryotic and eukaryotic MR complexes (47–50), we made a fluorescence-based helicase assay to monitor the activity of PfuMR and assess the influence of PfuPCNA. We used modified S87s/87s substrate with reporter-quencher pair at the 5′ end to monitor emission of fluorescence during double strand opening in real time. To prevent partial DNA melting at high temperature, we performed unwinding assays at 55°C on the 3′ end protected substrate. In the previous cited studies (47,48,50,51), authors showed that addition of ATP/MgCl₂ was sufficient to detect a limited DNA unwinding activity of the MR complex. Yet, in our case, PfuMR did not exhibit bona-fide DNA unwinding activity dependent on ATP/MgCl₂ cofactors (Figure 5A, compare lanes 1 and 4). However, in presence of PfuPCNA and MnCl₂, conditions for which PfuMR also displayed nuclease activity, strand opening activity was detected (Figure 5A, compare lanes 4 and 8). We obtained similar results with the same substrate but containing the reporter-quencher pair located at internal position 23 nt from 5′ end of the top strand (Figure 5B), indicating that local DNA melting extended from the cleavage site down to the 5′ extremity and suggesting that this DNA fragment was displaced by the PfuPCNA/MR complex. To confirm this, we performed nuclease assays using the same substrates in similar conditions and resolved DNA products in native PAGE (Supplementary Figure S8). Data showed generation of a single strand DNA fragment of about 30 nt, corresponding to the displaced FAM-labeled fragment from the 5′ end. From these results, we can propose that coordination of endonuclease and local strand opening activities of PfuPCNA/MR complex leads to the release of a small 5′ end ssDNA fragment to generate a DNA product with a 3′ ssDNA overhang of about 30 nt exposed.

Figure 5.

The endonucleolytic product from PfuPCNA/MR activity is displaced from dsDNA. Results from real-time fluorescence DNA unwinding assays are presented on histogram chart with the percentage of unwound quenched substrates at 55°C after 30 min of time reaction. The dsDNA substrates used are illustrated above each histogram: RQ-S87s/87s substrate (A) contained a reporter-quencher pair at one extremity (+5 of the 3′ strand) consisting of dsDNA duplex; (B) RQ23-S87s/87s substrate had the reporter-quencher pair located at internal position 23 of dsDNA. Kinetics of DNA unwinding assays are displayed in the panels below, curves numbering corresponds to the histogram lane number (from 1 to 11). 25 nM DNA substrate were pre-incubated with 50 nM PfuPCNA at room temperature for 5 min before adding 25 nM PfuMR wt or ΔPIP. Reactions were performed for 30 min at 55°C in buffer with 300 mM NaCl, 5 mM MgCl₂ complemented with 1 mM ATP and 5 mM MnCl₂ as indicated in the table below histograms (see ‘Materials and methods’ for complete protocol). Experiments were performed in triplicate and error bars correspond to standard deviation.

PfuPCNA does not interfere with ATP hydrolysing cycle of PfuRad50

At this point, we demonstrated that PfuPCNA stimulates PfuMR nuclease activity depending on ATP presence. A member of the ATP Binding Cassette (ABC) protein superfamily, Rad50 contains a conserved Nucleotide Binding Domain (NBD) that dimerizes upon two ATP molecules binding at dimer interface (45). By binding and hydrolysing ATP, Rad50 drives dynamic structural transitions of the MR complex controlling DNA unwinding and nuclease activities (49,52–56). To examine effect of the PfuPCNA/MR complex interaction on ATPase activity, we measured radioactive phosphate released by ATP hydrolysis in conditions where PfuMR complex was deficient in nuclease activity (without MnCl₂). As observed in several other studies, PfuMR alone exhibited weak ATPase activity that was not stimulated by the addition of dsDNA substrate (Supplementary Figure S9). In addition, we showed that PfuPCNA did not significantly regulate ATPase activity of PfuMR and hence may have no effect on PfuMR conformational change; similar data (not shown) were obtained in presence of MnCl₂.

DISCUSSION

MR complex is involved in various aspects of DSB repair, including sensing DSB triggering signal pathways and facilitating DSB repair through different pathways. Among hyperthermophilic archaea, gene deletions of mre11 and rad50 are lethal for cells arguing for their apparent essentiality, which distinguishes HA from all other cellular organisms, including mesophilic archaea (for a review, see (5)). Here we revealed in vitro physical association and functional interplay between the DNA clamp PCNA and the recombination MR complex of P. furiosus. To support this conclusion, we found that (i) PfuMR directly interacts with PfuPCNA via a PIP-like motif and (ii) PfuPCNA regulates PfuMR ATP-dependent nuclease activity to promote endonucleolytic cleavage about 30 nt from 5′-terminated dsDNA.

We demonstrated that PCNA binds directly to the Mre11–Rad50 complex of P. furiosus and that the interaction is mediated by a non-canonical PIP motif located in the C-terminal region of PfuMre11. The conspicuous difference between the PfuMre11 motif and the canonical PIP motif is the absence of the well-conserved glutamine residue. For several PCNA-interacting partners, the PIP motif contains a glutamine residue involved in multiple interactions with PCNA surface residues (38). Recently, a number of additional proteins that bind to PCNA have been reported (reviewed in (19–21,57)). Among them, some PCNA partners showed a PIP motif lacking this glutamine residue, and its absence has also been observed for homologue proteins, such as RFC in Archaea (58). Interestingly, the N-terminal part of the PfuMre11 peptide contains a basic region composed mainly of lysine (K) amino acids. As described for the PIP motif of RFC from P. furiosus, these residues could establish electrostatic interactions with the negatively charged surface of the PCNA C-terminal tail and compensate for the absence of glutamine residue (59,60). These results suggest that the peptide detected on PfuMre11 sequences is a variation on a theme of the canonical PIP motif, lacking the otherwise conserved glutamine residue, and mediates interaction between PfuPCNA and the PfuMR complex.

Remarkably, the putative PCNA interacting motif is conserved among Mre11 sequences of all Thermococcales species and of Archaeoglobus fulgidus. Similar PCNA interaction motifs were not found in other archaea or in eukaryotic Mre11 sequences. However, a search for a canonical PIP motif (Q-x(2)-[LIVM]-x(2)-[FYW]-[FYWLIVM]) in generic databases Swiss-Prot and TrEMBL, using Mre11 as description filter, identified Mre11 sequences from the orders Methanosarcinales and Halobacteriales as potentially harbouring a canonical PIP motif at the extreme C-terminus region (data not shown). This suggests that the interaction between PCNA and the MR complex could be a more general feature, not only restricted to these phylogenetic branches. In addition, proteins may use additional regions to interact with the DNA clamp and novel PCNA interacting motifs have been identified since the classic PIP-box discovery (21,61). In this context, this hypothesis deserves to be tested, particularly for eukaryotic Mre11 for which a co-localisation with PCNA was observed in human during the S phase of the cell cycle (62).

Besides the physical interaction, we described a functional interplay between the recombination complex and PfuPCNA. First, we demonstrated that PfuMR is weakly active at moderate ionic strength, the condition prevailing in P. furiosus cells. We also demonstrated that, in these particular conditions the DNA sliding clamp stimulates and modulates nuclease activity of PfuMR complex. Our data thus indicate that DNA resection initiated by PfuMR could be regulated by the DNA sliding clamp. We showed that upon association with PfuPCNA, the PfuMR complex generates a major internal incision in the 5′ strand proximal to DSB ends. Our findings are consistent with recent study in budding yeast, wherein a distinct MR complex partner, Sae2, influenced the activity of MR in a comparable manner, by activating only the endonuclease activity of the Mre11–Rad50–Xrs2 complex (46). In addition, it was shown that the 5′ strand cleavage by eukaryotic MR was strongly enhanced by a protein block mimicking a covalently bound topoisomerase-like protein (46). While MR nuclease activities are dispensable for the resection of ‘clean’ DSBs (63), endonuclease activity is essential to clear obstructed DNA ends (64). Consistent with this property, we also demonstrated that P. furiosus MR complex was able to cleave ds break DNA intermediates with a streptavidin block at the 5′ end or with both ends blocked with phosphorothioate residues. The reported behaviour of T4 phage system gp46/47 (MR homologues) in presence of the protein factors gp32 and UvsY was also comparable to what we observed for the archaeal PfuPCNA/MR complex, since the degradation profile of gp46/47 was shifted towards the generation of endonuclease products between 15 and 25 nt (30). More recently, Deshpande et al., demonstrated for human proteins the role of Nbs1, the third component of the MR/N complex, in regulating MR activities by inhibiting exonuclease activity on clean ends, whereas phosphorylated CtIP, the orthologue of Sae2, stimulates endonucleolytic cleavage (65).

These different reports emphasize the importance of regulating endo- and exonuclease activities of Mre11 within the MR complex for DSB repair and that this regulation is tightly coupled with ATP-dependent conformational changes of the MR complex. A major contribution of the present study is that we revealed PfuPCNA as a new interacting partner stimulating nuclease activity of the PfuMR complex. The molecular mechanisms that would explain how PfuPCNA triggers this internal incision by PfuMR remain to be determined.

As mentioned earlier, the cycle of ATP binding and hydrolysis is associated with conformational changes of the MR complex, with transition between ATP-bound and ATP-hydrolysed states, where the ATP-bound form mediates DNA end binding, whereas ATP-hydrolyzed state renders the Mre11 nuclease active site accessible. Here, we showed that the endonucleolytic cleavage induced by the PfuPCNA/MR complex is ATP-dependent, but that PfuPCNA had no effect on ATP hydrolysis or DNA binding activities of PfuMR. Interestingly, the PIP-like motif in PfuMre11 is located close to the RBD domain involved in binding of the Rad50 subunit. PCNA and the Rad50 subunit slide or diffuse along DNA whereas Mre11 is required for DNA end recognition and nuclease activities (66). Further structural and biochemical analysis would be required to determine if one of the two proteins has an effect on its partner's diffusion along the DNA or if PfuPCNA interacts preferentially with a conformation of the PfuMR complex. In support of the latter hypothesis, we showed by SPR that a more stable PfuPCNA/MR complex was formed in presence of ATP, and in the case of the PfuMR construct with truncated PfuRad50, the complex was faster to dissociate from PfuPCNA (data not shown), indicating that the PfuRad50 coiled-coil domain also has a role in the stabilisation of the interaction. Structural determination of the PfuPCNA/MR complex with DNA would be important for understanding how PfuPCNA controls or binds a specific conformational state of PfuMR complex.

According to bidirectional resection model of DNA DSB, upon the initial endonuclease cleavage, the Mre11 exonuclease proceeds back towards the DNA end via its 3′→5′ exonuclease activity (model discussed in (32,67). Astonishingly, in our results no 5′-products shorter than 15 nt were observed in the nuclease assays, suggesting an absence of extensive 3′→5′ resection for the second step. The most likely explanation for this observation is that archaeal PfuMR displaced the 5′ end through a combination of melting and endonuclease activities. This explanation is consistent with the reported ability of the eukaryotic MR(N/X) complex to open the DNA helix on ∼20 base pairs at the end of the duplex in an ATP-dependent manner (47,50).

In our conditions, the PfuPCNA/MR complex did not display genuine DNA helicase activity dependent on ATP and Mg²⁺. To date, ATP-dependent DNA unwinding activity has only been reported for bacterial MR and eukaryotic MRN complexes. This finding suggests that this unwinding activity is not conserved in the archaeal MR complex. However, in conditions suitable for DNA cleavage, the 5′ labeled product (27–29 nt) was displaced from the initial dsDNA substrate. We thus propose that PfuPCNA stimulates PfuMR DNA end processing leading to an internal cleavage coupled with 5′ end removal. As described by Liu and collaborators, ATP-dependent DNA melting facilitates the access of Mre11 for DNA cleavage (49). Here, we cannot elucidate which, from the cleavage or the DNA opening event, occurs first. Altogether, we assume that the short-range processing by PfuPCNA/MR would generate 3′-tailed substrate that could be suitable for additional partners responsible for extended resection. The helicase/nuclease HerA/NurA complex was found in all thermophilic archaea, clustered in the operon encoding Mre11 and Rad50 (68), and in vitro experiments demonstrated that Rad50, Mre11, HerA and NurA co-operate for resection of the 5′ strand at a DNA double strand break, generating a 3′ ssDNA suitable for the recombinase RadA (42). In Pyrococcus abyssi, at least 20 nt are required to bind one RPA trimer onto DNA efficiently (unpublished data). Here, the PfuPCNA/MR interplay generates a 3′ overhang of 27–29 nt suitable for RPA loading and thus for 3′ tail protection from degradation by NurA, while the complex HerA/NurA can still extend 5′ strand resection. Taken together, our findings indicate that PfuMR interacts with PfuPCNA physically and functionally in a manner consistent with an end resection process for the HR pathway.

This study leads to questions about the role of PCNA/MR interplay in hyperthermophilic archaea. The precise employment of the different process components in cases of either DSB damage or replication fork stalling remains to be clarified. Deciphering the role of PCNA/MR interplay in HA will require further genetic and structural studies, with the hope that it might provide clues to improve understanding of recombinational repair processes in archaea.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Ifremer; CNRS; University of Western Britany. This work was supported by the French “Agence Nationale pour la Recherche” [ANR-16-CE12-0016 to B.C.O. and D.F.]. Funding for open access charge: Ifremer.

Conflict of interest statement. None declared.