The MRE11–RAD50–NBS1 Complex Conducts the Orchestration of Damage Signaling and Outcomes to Stress in DNA Replication and Repair

Abstract

Genomic instability in disease and its fidelity in health depend on the DNA damage response (DDR), regulated in part from the complex of meiotic recombination 11 homolog 1 (MRE11), ATP-binding cassette–ATPase (RAD50), and phosphopeptide-binding Nijmegen breakage syndrome protein 1 (NBS1). The MRE11–RAD50–NBS1 (MRN) complex forms a multifunctional DDR machine. Within its network assemblies, MRN is the core conductor for the initial and sustained responses to DNA double-strand breaks, stalled replication forks, dysfunctional telomeres, and viral DNA infection. MRN can interfere with cancer therapy and is an attractive target for precision medicine. Its conformations change the paradigm whereby kinases initiate damage sensing. Delineated results reveal kinase activation, posttranslational targeting, functional scaffolding, conformations storing binding energy and en-abling access, interactions with hub proteins such as replication protein A (RPA), and distinct networks at DNA breaks and forks. MRN biochemistry provides prototypic insights into how it initiates, implements, and regulates multifunctional responses to genomic stress.

INTRODUCTION

At the cellular level, life is constantly challenged by genotoxic stress originating from endogenous and exogenous sources, with stalled replication forks (RFs) and double-strand breaks (DSBs) being among the most deleterious and extreme forms of DNA damage. Failure to properly respond to genomic distress can be mutagenic and toxic to cells as well as fatal to the organism. Cells have therefore evolved to counteract genotoxic stress by integrating DNA repair mechanisms and cell cycle regulation through checkpoint signaling (1). In this network of DNA damage responses (DDR), the meiotic recombination 11 homolog 1 (MRE11)–DNA repair protein RAD50–Nijmegen breakage syndrome protein 1 (NBS1) (MRN) complex (Figure 1a,b) holds a key position. It is among the first responders and regulates both signaling and damage responses to at least four types of extreme cellular stress: DNA damage at DSBs, stalled RFs, dysfunctional telomeres, and viral invasion. MRN also has a prime role in kinase [ataxia-telangiectasia mutated (ATM) and ataxia-telangiectasia and RAD3-related (ATR)] signaling that helps orchestrate cell cycle progression and damage responses (2).

Figure 1.

MRN domains, complex assembly, and mutations. (a) Key domains of MRN subunits MRE11 (orange), RAD50 ( green), and NBS1 (cyan). Mutated sites (red arrows) for human inherited diseases (color coded for ATLD, NBS, and NBSLD; mutations corresponding to each disease are color coded with respective abbreviations) are shown on each subunit with respective substitutions, where X represents the stop codon. (b) Modeled assembly of MRN complex and dsDNA based on experimentally derived partial crystal structures (PDB IDs: 5F3W, 5GOX, and 3HUE; color coding as in panel a) showing a dimer of a heterotrimer. The allosterically linked Zn-hook is highlighted in the inset along with CxxC motifs (red ) on two RAD50 monomers. (c) The extent of mutations (reported as mutation factor) in subunits of the MRN complex in cancers documented in the International Cancer Genome Consortium (ICGC) projects. A mutation factor is defined as a number of observed mutations multiplied by the fraction of mutation-positive cases. ICGC projects with high mutation factors (>15 in total) include the following: #13, Breast Cancer, ER+ and HER2−, European Union/United Kingdom; #20, Esophageal Adenocarcinoma, United Kingdom; #28 Liver Cancer, China; #31, Liver Cancer, RIKEN, Japan; #36, Malignant Lymphoma, Germany; #52, Skin Adenocarcinoma, Brazil; #53, Skin Cancer, Australia; and #55, Soft Tissue Cancer, Leiomyosarcoma, France. A complete list of the 58 projects is provided in Supplemental Table 1. Abbreviations: ABM, ATM binding motif; ATLD, ataxia-telangiectasia-like disorder; BRCT, BRCA1 C-terminal domain; dsDNA, double-stranded DNA; FHA, forkhead-associated domain; MBM, MRE11 binding motif; MRE11, meiotic recombination 11 homolog 1; NBS, Nijmegen breakage syndrome; NBSLD, NBS-like disorder; PDB, Protein Data Bank; RAD50, DNA repair protein RAD50; RBD, RAD50 binding domain.

MRN functional significance in both signaling and repair is underscored by the observation that the core Mre11–Rad50 (MR) catalytic subcomplex is conserved across all domains of life, encompassing archaea (e.g., Pyrococcus furiosus), eubacteria (Escherichia coli SbcC/D), some phage (T4 phage gp46/47), and eukaryotes including homologs in Saccharomyces cerevisiae and Schizosac-charomyces pombe (3). The NBS1 (Xrs2 in S. cerevisiae) component is conserved only in eukaryotes. However, mutations in any of the genes encoding MRN components cause severe consequences in humans. Germline mutations (Figure 1a) in MRE11, NBS1, or RAD50 cause ataxia-telangiectasia (A-T)–like disorder (ATLD), Nijmegen breakage syndrome (NBS), or NBS-like disorder (4–10). A-T is caused by inherited mutations in ATM kinase. Aside from germline mutations, all 3 MRN complex components are mutated in >50 cancers, as assessed from the International Cancer Genome Consortium projects (Figure 1c and Supplemental Table 1). Furthermore, the MRN complex is frequently found absent in patients with epithelial ovarian cancer (11). MRN functional disruption may be a synthetic lethal combination with poly(ADP-ribose) glycohydrolase (PARG) inhibition (12) providing a possible strategy to target MRN-defective cancers by inhibition of PARG, whose structure is defined (13). However, upregulation of RAD50 increases cellular radioresistance, and knockdown sensitizes some cancer cells to radiation, suggesting MRN may be a target for precision medicine (14). Overall, its conservation across all domains of life coupled to the catastrophic consequences of its mutations affirms the role of the MRN complex in maintaining essential aspects of genomic stability for all cells.

In this review, we consider exemplary recent developments and paradigm-shifting results linking MRN structural biochemistry to its biology to complement and build upon prior excellent reviews (2, 4, 15, 16). We integrate recent advances to delineate how MRN’s multidomain three protein composition enables its central enzymatic, sensing, signaling, architectural, and scaffolding functions in damage responses. MRN conformations are key to damage signaling through the ATM and ATR kinases, regulating the MRE11 nuclease and opening DNA for nuclease incision. By activating ATM, the MRN complex regulates phosphorylation of >900 sites on >700 proteins, including MRE11, RAD50, and NBS1 (17, 18). NBS1 interacts with ATM kinase plus its N-terminal phosphopeptide-interacting domains that link MRN to other partners including the C-terminal binding protein (CtBP)-interacting protein (CtIP) (19, 20). As it directly binds NBS1, DNA, and RAD50, the MRE11 nuclease dimer is the core of the MRN complex (21–23). We therefore update the two-step MRE11 nuclease incision model to explain its endo- and exonuclease properties in light of recent structures, including the eukaryotic RAD50 Zn-hook and CtIP N-terminal domain. We describe factors affecting MRN functions at DSBs, stalled RFs, dysfunctional telomeres, and complexes combating exogenous viral DNA infection. We discuss how MRN acts more as a dynamic molecular machine for the expeditious formation of protein–nucleic acid functional scaffolds at DSBs, RFs, dysfunctional telomeres, and viral DNA than as part of linear pathways, as also proposed for nonhomologous end joining (NHEJ) complexes (24). As a multifunctional macromolecular machine, the structural biochemistry of the MRN complex initiates sensing of stalled forks and DSBs, cell cycle checkpoint signaling cascades, incision for and commitment to break repair pathways, reestablishment of posttranslational phosphorylations via ATM kinase, and functional regulation of chromatin remodeling. The collective results outlined here suggest how MRN conformations and interactions enable both its specificity and multiple functions in cells with implications for science and biomedicine.

MRN COMPLEX: STRUCTURAL BIOCHEMISTRY AND BIOLOGY

The MRN complex is an intriguing chemo-mechanical molecular machine that converts energy from binding to DNA, ATP, and protein partners into conformational changes that control nuclease activities and network assemblies (2). MRN binds DNA and acts as a sensor of DNA damage (25). It is furthermore the activator for ATM and ATR kinase cascades (26), a tethering scaffold for recognition and stabilization of protein interactions, a processor of RFs, a licensing factor for homologous recombination (HR), and a facilitator of fork restart and DNA repair processes (2, 27, 28). At the basic level, MRN consists of two RAD50 ATP-binding cassettes (ABC)–ATPase sub-units, two MRE11 subunits forming a DNA structure-specific endo/exo/hairpin nuclease, and the NBS1 regulatory docking protein with phosphopeptide-interacting forkhead associated (FHA) and BRCA1 C-terminal (BRCT) domains flexibly linked to an MRE11 interface and adjacent C-terminal ATM kinase interaction motif (16, 29) (Figure 1a,b).

The core MR complex exists as a hetero-tetrameric assembly (M2R2), whose dynamic morphology emerges from shifts in its modular head, coil, and hook domains (Figure 1b). The globular DNA binding head is formed by the two Rad50–ATPase domains and the Mre11 nuclease that bind to the base of the Rad50 helical coiled-coil domains (Figure 2a). The heart of the MR (in prokaryotes) and MRN (in eukaryotes) assembly originates from the Mre11 dimerization interface (29). Yet, RAD50 is the largest subunit: It belongs to the structural maintenance of chromosomes (SMC) protein family (30). Each Rad50 polypeptide assembles into a globular ATPase domain despite a large intramolecular antiparallel coiled-coil sequence insertion. This coiled-coil domain extends far from the head domain (∼500 Å for eukaryotic Rad50 homologs) with a central CxxC motif forming a Zn-hook and reverse turn that caps its distal end and mediates Zn-dependent Rad50 subunit interactions (31–33). The resulting Rad50 bipartite ATPase dimerizes in a Mg²⁺and ATP-dependent manner (30). Notably, Rad50 conformation at the dimer interface is controlled by binding two ATP molecules: Each is sandwiched between signature motifs conserved in ABC–ATPases on one subunit and the Walker A, Walker B, Q-loop, and D-loop motifs from the other subunit (34, 35).

Figure 2.

The Mre11–Rad50 (MR) heterotetramer: DNA-binding interface, ATP-mediated conformational changes, and a unified model for MRN’s functions in DSB resection (color coding as in Figure 1) are shown. (a) Crystal structures of ATPγS-bound Methanococcus jannaschii (Mj) Mre11—Rad50–DNA complex (PDB ID: 5F3W) and the symmetric (PDB ID: 3DSC) and asymmetric (PDB ID: 3DSD) Pyrococcus furiosus Mre11 DNA-bound states, highlighting the DNA-binding interface at the Rad50 and Mre11 dimerization interface. (b) Crystal structures of the MR complex in the nucleotide-free state (PDB ID: 3QG5, Thermotoga maritima MR) and in the ATPγS-bound state (PDB ID: 3AV0, MjMR), highlighting the conformational switch in the MRN core complex. (c) Human CtIP model from the recent crystal structure of CtIP N-terminal tetramerization domain (PDB ID: 4D2H). Plausible NBS1 binding sites on four CtIP and C-terminal DBMs, along with CxxC motifs and Zn coordination, are shown. (d ) A testable pathway for DSB resection by the MRN complex. For clarity, only MRE11 (M) and RAD50 (R) subunits are shown. First, two molecules of MRN bind to DSB ends in an ATP-bound state, with MRE11 nuclease activity inhibited by the closed RAD50 conformation. Then, MRN forms an interlinked complex through a RAD50 coil swap. At this point, CtIP–NBS1 interaction allows MRN dimers to slide away from the break, and MRE11 makes the initial licensing cut through its endonuclease activity. Upon complete ATP hydrolysis, the RAD50 dimerization interface opens and dsDNA becomes accessible to the MRE11 exonuclease. DNA resection begins in the 3΄–5΄ direction and proceeds toward the break. Upon reaching the break, the MRN complex may be dislodged from DSB with the help of CtIP and NBS1 (NBS1 not shown for clarity). Because only M and R are shown in the model, NBS1 (N) is colored gray in MRN notation. Abbreviations: CTD, C-terminal domain; CtIP, C-terminal binding protein–interacting protein; DBMs, DNA binding motifs; DSBs, double-strand breaks; dsDNA, double-stranded DNA; MRE11, meiotic recombination 11 homolog 1; MRN, MRE11–RAD50–NBS1 complex; NBS1, Nijmegen breakage syndrome protein 1; NTD, N-terminal domain; PDB, Protein Data Bank; RAD50, DNA repair protein RAD50.

MRE11 key features are an N-terminal core nuclease domain followed by a cap domain that restricts access to the nuclease active site and a flexibly linked RAD50 binding motif (21, 22, 36–39). X-ray structures of P. furiosus Mre11 revealed the two-domain architecture for the catalytic core (40). The N-terminal calcineurin-like phosphoesterase domain bears the Mn²⁺-coordination and nuclease catalytic motifs. The phosphoesterase is further decorated by an alpha-beta fold C-terminal domain positioned to enforce DNA binding specificities (Figure 1a) (40). The ternary complex of P. furiosus Mre11 bound to Mn²⁺ and a 5΄҆-dAMP nucleotide hydrolysis product supports the biochemically observed 3΄–5΄ directionality of the Mre11 exonuclease activity. Yet, as noted below, Mre11 nuclease activity also enables 5΄–3΄ resection of DNA ends to generate 3΄ single-stranded DNA (ssDNA) overhangs for template-mediated HR and homologous repair (27). The partly disordered MRE11 C terminus is regulated by posttranslational modifications (Figure 1a).

The NBS1 subunit is specific to eukaryotes. It has one FHA and two BRCT domains in its N terminus (Figure 1a). These FHA and BRCT domains bind multiple phosphorylated proteins to regulate MRN interactions (29). The rest of NBS1 is unstructured except for the functionally critical C-terminal MRE11 and ATM-binding subdomains and motifs (23, 41). The NBS1 ATM-binding motif resembles the C-terminal binding motifs of Ku80 and ATR-interacting protein (ATRIP), which interact with and recruit the DNA-dependent protein kinase catalytic subunit (DNA-PKcs) and ATR, respectively (42), suggesting an interaction conserved across the three kinases (43). A recent DNA-PKcs crystal structure with bound Ku80 C terminus showed how Ku80 binds near DNA-PKcs autophosphorylation sites (the PQR cluster) (44). Interestingly, the NBS1 C terminus may also bind near ATM autophosphorylation sites (S367, S1893, and S1891) (42), suggesting that analogous modes of phosphorylation regulate DNA-PKcs and ATM activities. Yet, this NBS1 C-terminal region may not be critical for ATM-mediated checkpoint activation but helps in proper phosphorylation of some of the ATM substrates such as SMC1 and proapoptotic factor BID (45, 46).

The combination of at least two dimer states for structure-specific nuclease (MRE11) and ATPase enzyme (RAD50) activities, long-range allostery from NBS1 and the Rad50 Zn-hook binding domains, and regulatory phosphorylation plus flexible interface linkages enables multiple MRN conformational states. Together, these functional characteristics make MRN a dynamic machine that not only can sense and process DNA damage but also can participate in different pathways and networks, such as NHEJ, alternative nonhomologous end joining (A-NHEJ), or HR, for DNA damage repair. Owing to its dynamic and extended nature, structural analyses of MRN have employed multiple combined methods and biophysical techniques, including X-ray crystallography, small-angle X-ray scattering, electron microscopy (EM), and atomic or scanning force microscopy. Furthermore, structural information on MR has emerged from prokaryotes that can provide more accessible structures than have been analyzed for their eukaryotic counterparts. In fact, prokaryotic Mre11 is used as a molecular avatar (an embodiment of human MRE11–essential features) for design of inhibitors against human MRE11 (47). Excitingly, high-resolution structures of eukaryotic MRN protein domains are also being solved (33, 38, 39). For eukaryotes, CtIP is also emerging as a key player in MRN functions in HR on the basis of structural analyses of the CtIP N-terminal domain (48–50). From these collective MRN structural and biochemical results, we propose a testable integrated model for MRN structure–function relationships (Figure 2).

RAD50 forms a unique SMC-type structure in which subunits interact through a Zn-hook at the tip of antiparallel coiled-coil arms to form an M₂R₂N₂ intralinked complex (Figure 1b) (30). In the ATP-free or hydrolyzed state seen in the Thermotoga maritima MR complex,the Rad50 ATPase subunits are flexible and relatively open with the arms relaxed (51, 52) (Figure 2b). In contrast, upon ATP binding, Rad50 closes into a single, more rigid conformation in which both head domains (N- and C-terminal) are interacting with each other in trans and form a central groove that can accommodate double-stranded DNA (dsDNA) (Figure 2a) (21, 38, 51). Notably, an added DNA binding site at the coiled-coil domain on Rad50 subunits is seen for the complex of Rad50 (nucleotide binding domain) and Mre11 (helix-loop-helix domain) with AMPPNP and dsDNA (53). Single-molecule imaging of the human MRN complex provided additional evidence for this DNA-binding mode. Fluorescence imaging of MRN on nucleosome coated DNA suggested that this binding mode facilitates an efficient search for DNA lesions via facilitated diffusion along chromatin (34). Also, as seen in EM images for human and P. furiosus MRN complexes, the coiled-coil domains can form interlinked complexes (two intralinked MRN complexes bridged through RAD50 Zn-hooks), where they may tether two distant DNA sites, and in this configuration, the arms can extend to 1,200 Å (32). This observation was supported by the crystal structure of the P. furiosus Rad50 Zn-hook domain with a small coiled-coil region, where the Zn²⁺-coordinating cysteines are arranged such that either intra- or interlinked complex coiled-coil domains can form (32). Recently, however, crystal structure and X-ray scattering analyses of human RAD50 Zn-hook with a portion of coiled-coil domain suggest that RAD50 predominantly forms an intralinked complex (Figure 1b) (33) having a novel interaction interface within the coiled-coil domain that supports a rigid rod-shaped RAD50 dimer.

The Zn-hook and coiled-coil domain are critical for functional RAD50 DNA tethering and ATM activation (54–56). Yet, two Rad50 subunits are also held flexibly by the Mre11 dimer, whose subunits form a flexibly linked helix-bundle interface at the base of two Rad50 subunits that allow the Rad50 dimer to open upon ATP hydrolysis and to expose the Mre11 DNA-binding channel for nuclease excision. Although limited coiled-coil domain–length deletions are tolerated, mutations in the Zn-hook domain share the phenotype of a null mutation in yeast (32, 57), suggesting that long-range allostery connecting the MR core enzymatic activity with the Zn-hook is essential for MRN function in vivo.

Importantly, the closed-state (ATP-bound) Rad50 renders dsDNA inaccessible to the Mre11 nuclease active site (Figure 2a) (21). However, ATP hydrolysis opens dsDNA so that flexible ssDNA may access the Mre11 active site (21). In another ATPase machine, FlaI (58), a structurally defined intermediate with bound phosphate ion (PO₄^3–) released from ADP partially opens subunit contacts: This structure suggests that RAD50 conformational intermediates may allow MRE11 ssDNA endonuclease activity while restricting dsDNA and MRE11 exonuclease excision. By analogy, this as yet unseen RAD50 state would involve binding of arginine on the ABC–ATPase signature helix to the PO4^3– released by ATP hydrolysis (35). After ATP hydrolysis, Rad50 forms a structurally defined and more open state that makes DNA accessible to Mre11 for 3΄–5΄ exonuclease activity. The switch between closed and open Rad50 states orchestrates conformational changes to the MRN complex and its bound DNA (Figure 2b) that result in end protection or processing, as supported by structure-guided mutations (52). Furthermore, functional analyses of the Rad50 mutants show that the correct Rad50 interface with ATP and Mre11 are required for appropriate DSB processing (35, 59). Moreover, RAD50 mutation L1237F resulted in an exceptional patient response to an otherwise failed clinical trial of a checkpoint kinase-1 (CHK1) inhibitor combined with the DNA-damaging agent irinotecan (60). Our analyses suggest that this mutation promotes an open RAD50 that could activate MRE11 nuclease. We argue that matching the biochemistry of this mutation with a structure-based small-molecule inhibitor suggests a strategy to generalize the exceptional clinical response to other patients for precision medicine.

Crystal structures of Mre11 core domain are solved for both the Mre11 alone and in complex with Rad50 from different prokaryotes and eukaryotes (21, 29, 38, 39). All crystal structures, irrespective of source, show dimer formation, yet angles for the protomer arrangement vary. The dimerization surface is in the N terminus of T4 phage Mre11, followed by a capping domain and an acidic linker to the Rad50 binding domain, where the linker can act as an auto-inhibitor when Mre11 is not bound to Rad50 (61). From the X-ray crystal structures of P. furiosus Mre11 bound to a two-ended dsDNA and to a one-ended DNA fork (36), MR rearrangements can reshape DNA to access the nuclease active site (62). Existing DNA complex structures suggest that the Mre11 interface helps distinguish two-ended dsDNA breaks (symmetric dimer) from one-ended DNA forks (asymmetric dimer) (36) (Figure 2a). In fact, structure-guided small-molecule inhibitors (27), which bind adjacent to the dimer interface, show that blocking MRE11 endonuclease activity promotes DSB repair through NHEJ. An endonuclease nick distal to the DSB end licenses HR by initiating 3΄–5΄ MRE11 exonuclease activity back toward the DSB. Furthermore, the MRE11 endonuclease provides an entry point for long-range EXO1- and DNA2-dependent 5΄–3΄ resection activity away from the DSB. Consequently, inhibition of MRE11 exonuclease activity led to a DNA repair defect owing to a DSB repair defect that could be partially rescued by inhibiting EXO1/BLM resection needed to form extended ssDNA as the committed step for HR (27). Thus, MRN enzyme activities can help determine pathway choice between NHEJ and HR.

Nbs1 structural analyses define the phosphoprotein binding sites and uncover their flexible linkage to the Nbs1 C terminus (19). Furthermore, the crystal structure of Mre11 fused with the Nbs1 C terminus indicates that the Nbs1 C terminus interacts across the Mre11 dimer interface and on the face opposite to the DNA binding groove (23). As the Nbs1 N terminus interacts with several phosphorylated proteins, its flexible connection to its C-terminal interface across the Mre11 dimer extends the reach and network of the MRN complex for orchestrating DDRs (16, 29). For example, comparison of Nbs1-bound versus free structures unveils a long-range conformational change transmitted from the FHA domain bound to the phospho-CtIP-peptide through the BRCT domains, which may impact their binding to their phospho-serine protein targets, to drive an ∼30◦ domain rotation. These changes, which pull the linker to the Mre11 interface by ∼30 Å (19), are postulated to alter the Mre11 dimer symmetry thought to distinguish nuclease binding and activity for two-ended dsDNA breaks (symmetric dimer) and one-ended DNA forks (asymmetric dimer) (19). Overall, NBS1 (Xrs2 in S. cerevisiae) acts a chaperone and adaptor coordinating interactions of the MR complex with other proteins (63, 64) and as a regulated activator of Mre11 nuclease (65).

Consistent with the implications from structures and biochemistry, CtIP binds to the NBS1 FHA domains and evidently promotes resection in HR (50, 66–70) as one mode of conformational control for MRN. Recently, structure analyses of the N terminus for human CtIP and S. pombe Ctp1 reveal its assembly (48, 49). The full length and N terminus of the CtIP predominantly form a tetramer in solution. Furthermore, an extended Zn-binding motif in the coiled-coil domain adjacent to the N terminus extends the dimeric interface in both directions (Figure 2c). This CtIP interface resembles the MRN architecture and may tune the MRN complex in the interlinked complex that joins two sets of MRN–DNA complexes.

Using MRN’s structural biology, we propose a testable model for MRN endo- and exonuclease activity during HR (Figure 2d). Other nucleases, proteins, and spatial constraints likely impact this mechanism. Following initial recruitment to the damaged region by NBS1 N-terminal interactions (in eukaryotes) (71), ATP-bound MRN complexes load onto both sides of the DSB with RAD50 coiled-coil domains as intralinked complexes. Owing to their proximity, the Zn-hook and coiled-coil domains may switch into an interlinked complex, as seen in EM images (32). The ring-like conformation is suitable to permit MRN to move back from the break to allow access to other proteins, as the coiled-coil domain spans ∼1,200 Å , or ∼300 base pairs (bp) (32). Alternatively, a recent single-molecule imaging study (34) shows that the MRN complex can diffuse along the DNA, even in the presence of nucleosomes, to find the DNA end and carry out resection in association with EXO1. This study shows that Rad50 is critical for promoting facilitated diffusion along the nucleosomes-coated DNA, whereas Mre11 recognizes the DNA ends. This separation of function suggests one possible reason for the coevolution of Mre11 and Rad50 in all domains of life. Nbs1 is dispensable for both activities, allowing recruitment of the MR complex to the break in lower organisms for which Nbs1 is absent.

Through its interactions with both MRN and DNA (72, 73), the tetrameric CtIP configuration suggests a mechanical role in geometrically activating an MRE11 endonuclease cut in the 5΄ strand of the duplex with possible associated or sequential ATP hydrolysis. Upon ATP hydrolysis, RAD50 opens to allow MRE11 to carry out 3΄–5΄ resection toward the break, and the arms may be freed from the interlinked complex. At the nick, other 5΄–3΄ enzymes, such as EXO1/DNA2, resect away from the break to form extended ssDNA as the committed step and 3΄ single-strand substrate for HR. Upon reaching the DSB end, the CtIP dissociation upon a posttranslational modification of MRN may regulate MRN off rate and avoid release of unprotected DNA ends. To gain sequence-level information to test this model and other possible MRN mechanisms, it will likely be necessary to employ cryo–electron microscopy (cryo-EM), as enabled by advances in sample preparation for large complexes, plus EM detection technology and data analysis (74, 75). Integrated models of NHEJ from component structures have proved useful to guide both biochemical and cellular experiments (76), so we hope this may be the case for MRN.

MRN AND DOUBLE-STRAND BREAK REPAIR

MRN’s role in DDR is extensively studied for DSB repair. Unrepaired DSBs are toxic and can lead to chromosomal rearrangements (cancer hallmarks) and cell death through apoptosis (77). DSBs are caused by endogenous events (such as RF stalling; see Section 4) and severe damage from oxidation or ionizing radiation (IR) (78). MRN functions in both DSB sensing and repair: It is recruited to DSBs through interaction between NBS1 and cell cycle checkpoint protein RAD17 (71). RAD17 is phosphorylated by ATM and recruited to DSBs independent of the MRN complex (ATM-activation mechanisms are thoroughly reviewed in 79). Once at DSBs, the MRN complex recruits and activates ATM through its NBS1 interaction (Figure 3), with the likely participation of other unidentified interfaces. Activated ATM phosphorylates mediator of DNA damage checkpoint 1 (MDC1), which in turn recruits more MRN to DSBs and thereby amplifies MRN/ATM signaling (80).

Figure 3.

The MRN complex activates possible (alternative) pathways at DSBs. The MRN complex can initially be recruited to a DSB by RAD17 and activates the ATM kinase. Once activated, ATM phosphorylates downstream substrates including MDC1, which in turn recruits MRN, and the ATM signal is further amplified at the DSB. ATM checkpoint signaling leads to cell cycle regulation through CHK2 activation ( far left). On the basis of the cell cycle phase and nature of the DSB ends, DSBs are channeled through either HR or NHEJ for repair (central paths). In HR, the MRN complex initiates the 5΄ resection followed by long-range resection by EXO1 or DNA2. The resulting 3΄ ssDNA overhangs are loaded with ssDNA-binding protein RPA. The RPA-coated ssDNA can activate a second checkpoint pathway: ATR signaling. ATR is recruited to the RPA-coated ssDNA through ATRIP and RPA interaction. Then the RAD17–RFC clamp loading complex loads the 9–1–1 (RAD9–HUS1–RAD1) checkpoint clamp at the ssDNA and dsDNA junction on the RPA-coated ssDNA arms. The ATR activator TopBP1 is loaded by RAD9, leading to activation of CHK1 kinase and cell cycle regulation. For high-fidelity DSB repair, the BRCA2–RAD51 complex nucleates the formation of RAD51 nucleofilaments and displaces RPA from ssDNA. RAD51 nucleofilaments promote homology search and strand invasion in association with BRCA1–BARD1 for error-free HR repair through the resolution of repair intermediates. The two NHEJ pathways (right) depend upon DNA-PKcs and MRN–CtIP. In C-NHEJ, DSB ends are bluntly fused through Ku70/80–DNA-PKcs complex tethering, Artemis-mediated end processing, ligase IV, and XLF-XRCC4 scaffolding factor–mediated ligation. The C-NHEJ is independent of the MRN complex; however, MMEJ operates through MRN–CtIP-mediated resection followed by priming based on microhomology, flap removal, and gap filling. Abbreviations: ATM, ataxia-telangiectasia mutated; ATR, ataxia-telangiectasia and RAD3-related; ATRIP, ATR-interacting protein; C-NHEJ, canonical nonhomologous end joining; DNA-PKcs, DNA-dependent protein kinase catalytic subunit; DSB, double-strand break; HR, homologous recombination; MMEJ, microhomology-mediated end joining; MRE11, meiotic recombination 11 homolog 1; MRN, MRE11–RAD50–NBS1 complex; NBS1, Nijmegen breakage syndrome protein 1; NHEJ, nonhomologous end joining; RAD50, DNA repair protein RAD50; RFC, replication factor C; RPA, replication protein A; ssDNA, single-stranded DNA.

MRN/ATM signaling is regulated in part by factors that influence the stability and loading of the complex to DSBs. Interestingly, ATM and NBS1 are protected by the molecular chaperone heat shock protein 90 (Hsp90) alpha from polyubiquitination-mediated proteasomal degradation (81). Similarly, PIH1D1, a subunit of Hsp90 cochaperone R2TP complex, stabilizes MRE11 through its C-terminal interactions (82). Ubiquitination of NBS1 by the ring finger protein RNF8 promotes optimal binding to DSBs (83), whereas MRE11 phosphorylation by Polo-like kinase 1 (Plk1) inhibits recruitment of the MRN complex to DSBs (84). In addition, ATM-interacting protein (ATMIN) competes with NBS1 for ATM binding, therefore decreasing DSB-related ATM signaling (85, 86). In contrast, the recently discovered MRN-interacting protein (MRNIP) promotes MRN loading onto DSBs, thus contributing positively to MRN/ATM signaling (87). Consequently, activated ATM phosphorylates many downstream substrates to regulate checkpoint signaling and allow cells time to repair damage (88).

There are two major pathways through which cells repair DSBs (89) (Figure 3): the error-prone canonical nonhomologous end joining (C-NHEJ) and the error-free HR. NHEJ mends DSBs by end joining with minimal end processing. This process is initiated by Ku heterodimer (Ku70/Ku80) binding at DSB ends and recruitment of DNA-PKcs to phosphorylate various substrates and aid end processing and ligation by the DNA ligase IV complex (90). Two added pathways that repair DSBs with minimum end processing are resection-dependent NHEJ (rd-NHEJ) and microhomology-mediated end joining (MMEJ) (91, 92). In rd-NHEJ, an initial resection step involves MRE11 exonuclease, EXD2, Plk3-phosphorylated CtIP/BRCA1, EXO1, and Artemis endonuclease before repair through C-NHEJ occurs. MRE11 endonuclease activity may be dispensable in rd-NHEJ (92), whereas in MMEJ, initial resection involves both endo- and exonuclease activity of the MRN complex plus other enzymes, such as CtIP, PARP1, FEN1, and DNA ligases I and III (93, 94). The minimum homology for ligation is ∼5 bp (94). However, Werner syndrome helicase (WRN) and Bloom syndrome helicase (BLM) regulate these alternative pathways by blocking MRE11 and CtIP recruitment to DSB ends to promote repair through C-NHEJ (95, 96).

In HR, DSB ends are extensively resected to generate 3 ssDNA overhangs that inhibit NHEJ repair (Figure 3). As noted above, the evolving general model suggests that the MRN complex can initiate and license resection through an MRE11 endonuclease cut (nicking several hundred base pairs back from the DSB). Subsequent MRE11 exonuclease (3΄–5΄ direction toward the DSB) acts in association with phosphorylated CtIP (Sae2 in yeast). At the nick, 5΄–3΄ nucleases, such as EXO1 or DNA2 in conjunction with BLM or WRN, promote extended resection away from the DSB to generate 3΄ ssDNA overhangs more than 1,000 bp in length that constitute the committed step for HR (29). EXO1 catalyzes long-range resection of MRN-processed DNA ends in human cells (97, 98). However, EXO1 is strongly inhibited by replication protein A (RPA) binding to the resulting ssDNA (99, 100). Single-molecule imaging identified an additional, noncatalytic role for MRN in long-range resection. Here, MRN physically interacts with EXO1, traveling with the resection machinery as a processivity factor that prevents RPA inhibition (34).

Via interaction with the MRN complex or CtIP, extended DNA resection is stimulated by additional enzymes, including RECQL4 helicase and possibly by EXD2 exonuclease (101, 102), which has alternatively been localized to the cytoplasm and regulates translation in mitochondria (103). Regardless, the biochemistry for these regulating partners suggests changes in protein–DNA interaction stability and off rates. As described in the proposed model (Figure 2d), Sae2 (CtIP in humans) may dislodge the yeast MRX complex (Mre11–Rad50–Xrs2, MRN in humans) from ssDNA ends after resection (104, 105). In the absence of Sae2, DSB repair is impaired due to prolonged MRX complex binding. This defect is rescued by depleting the yeast 53BP1-ortholog RAD9, as 53BP1 promotes NHEJ by inhibiting dsDNA resection for HR (106, 107). In humans, the tumor suppressor BRCA1 inhibits 53BP1 at breaks in association with MRN and CtIP to promote HR over NHEJ (107–110). Extended RPA-coated ssDNA can activate ATR signaling (discussed in Section 4). For HR, RPA is displaced from ssDNA by RAD51 that forms nucleofilaments after loading by BRCA2 (111). RAD51 filaments enable a homology search, subsequent strand invasion, and template-dependent HR repair, as reviewed previously (111–113). Besides contributing to HR pathway choice, the BRCA1–BARD1 complex also plays an indispensable role in stimulating RAD51 recombinase activity (114). On the basis of the in vitro EM imaging data, the MRN complex may support HR by clamping the invading strand and HR template independent of its nuclease properties (32). Also, Mre11 nuclease-deficient or null TK46 cells show delays in Rad51 foci resolution after IR-induced breaks compared with wild-type cells. This delay suggests that the MRN complex may function downstream (as well as upstream) of RAD51-filament formation for proper resolution of HR intermediates (115).

Key factors regulating pathway choice for DSB repair are the cell cycle (116) and the nature of DNA ends (69). Whereas NHEJ-based repair pathways are active throughout the cell cycle, HR requires a DNA template for the repair and is active in only the S and G2 phases, during which the homologous sister chromosome is present. The nature of the DSB ends also regulates the repair pathway choice (69). If the DSB ends are free of protein adducts and damaged nucleotides, they are likely channeled through NHEJ for repair in human cells. Otherwise, they require processing through resection and the phosphatase activity of polynucleotide phosphatase/kinase (PNKP) (117). MRN biochemical endo- and exonuclease activities act on both 5΄ and 3΄ strands of DNA with protein adducts, and its endonuclease activity is stimulated by CtIP (72). In contrast, NBS1 inhibits the MR 3΄–5΄ exonuclease activity on clean DSB ends (72). The MRN complex can thus remove covalently linked topoisomerase complexes from genomic DNA in a mechanism requiring CtIP and BRCA1 (115, 118, 119). The MRN complex in association with CtIP also protects genomic loci at common fragile sites and palindromic repeats (120). Indeed, DSBs occurring at these sites often lead to the formation of secondary structures that can attract structure-specific nucleases and result in chromosomal rearrangements (120). At such sites, however, MRE11 nuclease activity promotes HR for error-free repair.

MRN AND REPLICATION FORK DYNAMICS

Accurate replication of genetic information is a prime genetic selection, and stress during replication can slow, stall, or collapse the RF (121). Sources for replication-induced stress include damaged DNA, chromatin compaction, non-B-DNA (i.e., DNA structures that do not conform to the canonical Watson-Crick double helix) forming sequences (G-quadruplex, small inverted repeats, trinucleotides repeats), covalent protein-DNA adducts, and DNA/RNA hybrids (121). Cells can overcome these barriers through RF stabilization, protection, remodeling, and restart (121–124) along with dormant origin firing in higher organisms (125–128). HR is implicated in fork transactions during fork recovery; however, HR protein responses at forks can be independent of dsDNA break repair (124). Although the HR response restores the replication robustness under stress, it can also contribute to genomic instability as a result of its intrinsic reliance on a homologous sequence (reviewed in 124). Given its HR-initiating roles, the MRN complex helps orchestrate RF stress responses (Figure 4). Its role at forks can be both beneficial and detrimental and thus requires regulation.

Figure 4.

The MRN complex with RPA and ATR orchestrates replication fork protection, remodeling, and restart as well as HR rescue of collapsed forks. At stalled forks, the MRN complex participates in fork protection through non-nuclease functions (e.g., DNA tethering), and its ssDNA endonuclease activity is inhibited by RPA. RPA-coated ssDNA initiates ATR checkpoint signaling that provides cells time for repair and restart. Replication barriers are removed by remodeling through fork reversal that must then be processed for fork restart. The nascent ssDNA at reversed forks is protected against MRN exonuclease activity by RAD51 loading in association with BRCA2. For restart, MRE11 excision of one strand may facilitate fork reversal and restart. Unrepaired collapsed forks can become DSBs where the MRN complex initiates homologous recombination by initiating resection. Abbreviations: ATR, ataxia-telangiectasia and RAD3-related; ATRIP, ATR-interacting protein; MRE11, meiotic recombination 11 homolog 1; MRN, MRE11–RAD50–NBS1 complex; NBS1, Nijmegen breakage syndrome protein 1; RAD50, DNA repair protein RAD50; RPA, replication protein A; ssDNA, single-stranded DNA.

At forks, protein-coated ssDNA is a platform for binding by MRN and other DNA repair proteins that initiate damage signaling, fork protection, and repair (129). An encounter between the replication machine (or replisome) and a replication barrier can uncouple the replicative helicase and polymerase. This uncoupling results in long stretches of ssDNA, leading to an ATR checkpoint response that helps to stabilize the fork and allows time to resolve the replication barrier by arresting cell cycle progression (Figure 4) (121). Stretches of ssDNA are protected by ssDNA-binding proteins such as RPA or human single-stranded DNA binding protein 1 (hSSB1) (130, 131). RPA binding to ssDNA inhibits annealing of short homologous sequences (as in short repeats) through microhomology-mediated repair and reduces secondary structures recognized by nucleases (132). The resulting protein-coated ssDNA not only is protected but also promotes binding of MRN and other DNA repair proteins (133). ATR, along with ATRIP, is recruited to ssDNA-RPA regions through ATRIP and RPA interactions (134). At the damaged forks, RAD9–RAD1–HUS1 (also known as the 9–1–1 clamp), is loaded onto the DNA end adjacent to RPA-bound ssDNA by the clamp loader RAD17–RFC complex (134). The 9–1–1 clamp recruits the ATR-activator protein topoisomerase-binding protein 1 (TopBP1) that assists ATR in phosphorylating its downstream substrates, including CHK1 and RPA (Figure 3) (134). Notably, TopBP1 was recruited independently of the 9–1–1 complex to the ATR-activating sites at ssDNA-dsDNA junctions at the RFs through the MRN complex in Xenopus egg extracts (135). This result is consistent with the observation that the MRN-depleted Xenopus extracts show a decrease but not complete loss of CHK1 phosphorylation (136).

In fact, ATR activation is a critical function for MRN in damage signaling. MRE11 (in addition to ATM) is required for ATR activation upon IR-induced damage (137, 138). Furthermore, NBS1 activates ATR in vitro independently of MRE11 (139). Notably, for replication-associated DSBs, a substantial amount of RPA phosphorylation by ATR is dependent on the NBS1–RPA interaction and is independent of RAD17 (140). These functions are likely conserved in yeast, in which RPA recruits the MRX complex to stalled forks and DSBs (141). Upon recruitment, MRX-mediated sister-chromatid tethering is essential for survival of genomic stress at both forks and breaks.

ATR activation is also regulated by RAD50 (142). Once activated, ATR phosphorylates RAD50 at S635, which is required for cohesin complex (that holds together newly replicated chromatids) loading at the forks. Aside from phosphorylation, the neighboring Zn-hook sequence influences cohesin loading at replication sites to promote fork restart (143). These collective data suggest that multiple MRN components can orchestrate ATR signaling events that may themselves regulate the temporal sequence for MRN biochemical activities at RFs.

At forks, MRN-mediated resection plays a major functional role. MRN-mediated resection of nascent DNA likely frees the replisome from the stalled forks by promoting strand removal, fork inversion, and restart (123). MRE11 can generate 3΄ overhangs to initiate HR in association with DNA2/EXO1. For example, in a recent study in a Werner syndrome–derived fibroblast cell line, the phosphorylation status of the S1133 of the WRN helicase influences the resection at breaks generated at the RF (144). Inhibition of S1133 phosphorylation in WRN results in impaired interaction with MRE11, defective MRN foci formation, and resection at the forks. Consequently, breaks at the fork can be repaired through DNA fusion by NHEJ with associated genome instability.

Although the MRN nuclease function can help resolve stalled RFs, unregulated resection can be catastrophic and lead to fork degradation through MRE11 exonuclease activity (145, 146). Thus, other proteins help regulate MRN-mediated nascent DNA degradation (147–150). The Fanconi anemia and breast cancer protein BRCA2 blocks MRE11-mediated resection of nascent ssDNA by promoting formation of stable nucleofilaments of RAD51 on nascent DNA (147, 151). However, cyclin-dependent kinase 2 (CDK2)–dependent phosphorylation of the BRCA2 C terminus results in disassembly of RAD51 filaments (152). This CDK2 activity is controlled by a complex formation with large tumor suppressor homolog 1 (LATS1) as part of the regulation of BRCA2 phosphorylation. Similarly, WRN helicase interacting protein 1 (WRNIP1) can stabilize RAD51 and thereby inhibit MRN-mediated degradation, a function independent of WRNIP1 ATPase activity (153). Like MRE11, WRN has an exonuclease domain with an associated coiled-coil region allowing assembly and partner interactions (154–156). Interestingly, the WRN exonuclease domain appears essential to inhibit MRE11/EXO1-dependent fork degradation under replication stress induced with low doses of the TOP1 inhibitor camptothecin (157). MRN regulation at forks is thus enforced by protein interactions that block productive MRE11 nuclease binding.

By inhibiting Mre11 using endonuclease- and exonuclease-specific inhibitors, BRCA2 deficiency is rescued in tumor cells lacking functional BRCA2. Although BRCA-defective cancer cells can be killed by PARP inhibitors (PARPi), resistance to PARPi arises in cancer cells from the inactivation of proteins (such as PTIP, CHD4, and PARP1) that recruit MRE11 protein for nuclease processing at stalled forks (158). This unraveled resistance mechanism underscores the critical role for MRE11 in the regulated and error-free fork restart and how loss of MRE11 or its regulation at forks can rebalance damage responses in cancer cells. In fact, this may be the basis for genome instability from p53 mutations (159).

MRN AND DYSFUNCTIONAL TELOMERES

MRN responds to the disfunction of telomeres, which protect linear eukaryotic chromosome ends from DSB-related DDR and suppress replication instability (160). Telomeres consist of long (thousands of base pairs in length) duplex-DNA repeats (-TTAGGG-in humans) with shorter (up to several hundred base pairs) single-stranded 3΄ overhangs and associated telomere protection–related proteins collectively termed the shelterin complex (161). Owing to the strict polarity restrictions of DNA polymerase, the 3΄-DNA end sequence from the lagging strand is not duplicated at telomere ends, leading to shortening of the net telomere length with each replication cycle (162). Conversely, duplication of the leading strand always results in a blunt-ended dsDNA owing to correct polarity (Figure 5). Both leading and lagging strand telomeres are processed by nucleases Apollo and EXO1 to generate functional 3΄ overhangs (163).

Figure 5.

Telomere structure and dynamics regulate MRN interactions and distinguish chromosome ends from breaks. Human telomeres consist of a long stretch of repetitive TTAGGG sequences and interacting proteins. Telomeres forming from the replication of a leading strand result in tails with blunt ends, whereas the replication of a lagging strand generates telomeres with 3΄ overhangs. The 5΄ resection of a blunt-ended telomere inhibits telomere fusion through NHEJ. The telomeres with 3΄ overhangs form transient T-loop structures. The shelterin complex, which helps protect telomere ends, consists of telomeric repeat-binding factors 1 and 2 (TRF1 and TRF2), repressor and activator protein (RAP1), TRF1-interacting nuclear protein 2 (TIN2), protection of telomeres 1 (POT1), and TIN2- and POT1-interacting protein (TPP1). Abbreviations: ATM, ataxia-telangiectasia mutated; ATR, ataxia-telangiectasia and RAD3-related; CtIP, C-terminal binding protein–interacting protein; MRE11, meiotic recombination 11 homolog 1; MRN, MRE11–RAD50–NBS1 complex; NBS1, Nijmegen breakage syndrome protein 1; NHEJ, nonhomologous end joining; RAD50, DNA repair protein RAD50; T-loop, telomere loop.

In addition to the shelterin complex, the reverse transcriptase telomerase maintains telomere length in budding yeast, stem cells, and many cancer cells (164). Yet interestingly, in most somatic cells, telomerase is inactive, leading to a limited number of replicative cycles and thereby regulating mutational accumulation and tumorigenesis (165–167). In budding yeast, the MRX complex may act in recruiting telomerase to telomeres (168). Also, telomere elongation may depend on X-ray repair cross-complementing protein 3 (XRCC3) and NBS1 in stem cells (169). The MRX complex may also assist Rad6 and Bre1 ubiquitin-conjugating enzymes in successful telomere replication (170).

In humans, telomeres function in tumor suppression. Cells either enter a replicative senescence state or undergo apoptosis once telomere length becomes critical, to avoid losing essential genes from linear chromosome ends (166). This regulatory mechanism relies on MRN and other proteins that assist the shelterin complex to avoid DSB repair at functional telomeres (reviewed in 171, 172).

At functional telomeres, DNA ends are protected. Since telomeres are G-rich sequences, they can form non-B-DNA structures such as G-quadruplexes that protect against strand degradation (173). As a result, replication in telomeric regions may have more frequent fork stalling (174) that may require resolution by MRN, but this role is uncertain.

In addition, the 3΄ overhang can fold back on to the dsDNA to form a T-loop similar to strand invasion in HR (161, 175) (Figure 5). The six-protein shelterin complex contains telomeric repeat-binding factors 1 and 2 (TRF1 and TRF2), repressor and activator protein (RAP1), TRF1-interacting nuclear protein 2 (TIN2), protection of telomeres 1 (POT1), and TIN2- and POT1-interacting protein (TPP1) (176). TRF2 binds to dsDNA, protects the T-loop from helicases and resolvases (176), and inhibits ATM signaling, whereas POT1 preferentially binds ssDNA and inhibits ATR signaling (177) (Figure 5). Shelterin defects and telomerase abnormalities lead to dysfunctional telomeres, promoting MRN damage signaling and repair within a few cycles of replication (178).

The MRN complex may thus play dual roles at telomeres, where it may cause or inhibit genomic instability at dysfunctional telomeres by either promoting or inhibiting the NHEJ pathway, respectively (171). NBS1 co-crystallizes with TRF2 (179). NBS1 phosphorylation at S432 residue inhibits this interaction and promotes C-NHEJ at telomeres lacking TRF2 (179). Yet, NBS1 phosphorylation at the same site promotes A-NHEJ at telomeres lacking POT1–TPP1 (179). In addition to defective shelterin components, critically short or deleted ends lead to dysfunctional telomeres. Although most cancer cells use telomerase to replenish dysfunctional telomere lengths, some utilize MRE11-dependent and recombination-based alternative lengthening of telomere (ALT) for telomere maintenance (180, 181). Loss of both telomerase and the ALT pathway causes telomere dysfunction. In yeast cells with these characteristics, Mre11 opposes proliferation of cells lacking telomeres in the absence of Exo1 and checkpoint signaling (182).

In budding yeast lacking telomerase, the MRX complex promotes replicative senescence in association with Tel1 (ATM in humans) and RAP1-interacting factor 2 (Rif2) (183). Tel1 promotes MRX recruitment, whereas Rif2 acts as a negative regulator by enhancing ATP hydrolysis that opens Rad50, facilitating MRX release from DNA (184). In line with insights from yeast experiments, changes in expression and posttranslational modification of MRN and telomeric proteins are associated with various diseases in humans (185–187). In mouse embryonic fibro-blasts, the MRN complex is required for the response to telomere dysfunction, consistent with the idea that initiating this response shares common elements with the response to interstitial DNA damage (188). Interestingly, in patients with rheumatoid arthritis for which T cell aging occurs prematurely, low MRE11 levels lead to telomeric damage, juxtacentromeric heterochromatin unravelling, and upregulation of senescence marker. MRE11 inhibition in healthy T cells induced the aging phenotype, whereas MRE11 overexpression in rheumatoid arthritis T cells reverses it. The observation that the premature T cell aging is linked to MRE11 and telomere deprotection suggests MRE11 as a therapeutic target for immune aging and suppressing dysregulated inflammation (189). We expect that further knowledge of MRN biochemical mechanisms will be key to an improved understanding and intervention in multiple diseases including immune system aging.

MRN AND THE INNATE IMMUNE RESPONSE

MRN functions that sense, signal, and defend against viral infection (190) underscore the importance of deciphering the biochemical basis for its multifunctionality. MRN and other damage recognition receptors detect viral DNA and initiate a cascade of innate immune and DNA damage signaling (191) (Figure 6a). In the genome, activation of ATM by MRN is amplified over a mega-base region leading to formation of DNA damage foci, cell cycle arrest, and DNA repair (192). Upon viral infection, local but not global ATM activation and signaling are initiated by the MRN complex owing to the small size of viral DNA (193). This local ATM signaling avoids cell cycle arrest and global damage responses when a virus infects the cell (Figure 6a) (193). Thus, MRN biochemical responses allow cells to differentiate genome damage signaling (global ATM signaling) for DNA repair compared with foreign-genome signaling (local ATM signaling) to block viral replication in the nucleolus (193). As part of its repair-related activities, the MRN complex in association with NHEJ repair proteins can further limit viral DNA replication in the nucleus by forming long viral DNA concatemers (190, 194, 195), which are too large to be packaged. Concatemers also interfere with replication: Replication origins generally present at the viral DNA termini are buried inside heterogeneously fused viral DNA and mutated through error-prone NHEJ (190, 196, 197) (Figure 6a).

Figure 6.

MRN in host and virus responses. (a) The MRN complex inhibits viral replication via pathways linking DNA repair proteins and the innate immune response. The interaction between the MRN complex and viral DNA leads to local ATM activation that avoids cell cycle arrest and apoptosis, which is characteristic of global ATM signaling in DNA damage. The MRN complex promotes the formation of viral DNA concatemers (randomly fused long chains of viral DNA) in association with NHEJ proteins to help inhibit viral replication. MRN components furthermore promote the innate immune response by inducing interferon and cytokine production through pathways involving STING, IRF3, and NF-κβ. (b) Viruses neutralize the MRN complex by relocalization and/or degradation. The viral protein complex of E4-ORF6 and E1B-55K directs the MRN complex for degradation through ubiquitination (Ub). The viral protein E4-ORF3 relocalizes the MRN complex to the nucleus through sumoylation (Su) to promote viral replication by limiting MRN-promoted viral DNA concatemer formation. Abbreviations: ATM, ataxia-telangiectasia mutated; CARD9, caspase recruitment domain-containing protein 9; DNA-PKcs, DNA-dependent protein kinase catalytic subunit; E1B-55K, a 55-kDa protein of adenovirus E1B gene; E4-ORF3, early region 4 of adenovirus open reading frame 3 protein; E4-ORF6, early region 4 of adenovirus open reading frame 6 protein; IFNs, type I interferons; IL-1β, interleukin 1 beta; IL-33, interleukin 33; IRF3, interferon regulatory factor 3; MRE11, meiotic recombination 11 homolog 1; MRN, MRE11–RAD50–NBS1 complex; NBS1, Nijmegen breakage syndrome protein 1; NF-κβ, nuclear factor κβ; NHEJ, nonhomologous end joining; RAD50, DNA repair protein RAD50; STING, stimulator of interferon genes.

DNA in the cytoplasm is recognized as a potential viral infection and induces an immune response. MRN binding to viral DNA in the cytoplasm leads to production of type I interferons (IFNs) through activation of nuclear factor κβ (NF-κβ), a stimulator of interferon genes (STING), and interferon regulatory factor 3 (IRF3) (Figure 6a) (198). Both MRE11 and RAD50, but not NBS1, are essential for producing IFNs. A cell line derived from an ATLD patient containing an MRE11 mutation that abrogates its DNA binding properties has defects in IFN production (198). Interestingly, RAD51 absence can also induce an innate response (199). RAD51 protects DNA inside the nucleus from MRE11-mediated nucleolytic degradation. Thus, in the absence of RAD51, MRE11 nuclease can degrade DNA, and the resulting DNA fragments are released into the cytoplasm where they activate the innate immune response. For interleukin-33 (IL-33), a member of the IL-1 superfamily of cytokines whose structure has been defined (200), stimulation upon adenoviral transduction depends on MRE11 status (201). Similarly, upon encountering cytosolic DNA, RAD50 forms a complex with innate immune system adaptor protein CARD9 in dendritic cells (202), which are the antigen-presenting cells that link the innate and adaptive immune systems. The cytosolic dsDNA-induced signaling complex (dsDNA–RAD50–CARD9) stimulates production of proinflammatory cytokine IL-1β through NF-κβ pathway activation (Figure 6a) (202). However, viruses encode specific proteins that bind MRN and other host DNA repair proteins rendering them inactive through protein relocalization and degradation (Figure 6b) by sumoylation and ubiquitination (190, 203). Yet, viruses can also rely on MRN and other host DNA repair proteins for successful lytic replication.

As a result, manipulation of MRN biochemistry by interactions with virus proteins is under active investigation, and several of these have been uncovered in human adenovirus (Ad) infection. Ad early region 4 (E4) encodes two proteins from open reading frame 3 (E4-ORF3) and open reading frame 6 (E4-ORF6). These proteins (with viral protein E1b-55K) inhibit concatemer formation by the MRN complex via relocalization and degradation (190, 204–207), avoiding MRN-mediated defects in late viral replication (205, 207). Upon recognition, E4-ORF3 protein localizes the MRN complex to specific nuclear locations in the host cells known as promyelocytic leukemia protein domains (or PML bodies), rendering the complex inactive (Figure 6b) (208). X-ray crystallography and mutational analyses show that E4-ORF3 assemble into both linear and branched polymeric structures by C-terminal tail swapping (209). This polymeric network spans the entire nuclear volume, forming multi-site interactions with PML bodies and MRN-binding interface in the C-terminal tail. Upon relocalization to PML, E4-ORF3 induces sumoylation of MRN complex components (210, 211). The E4-ORF6 in association with E1b-55K recruits other cellular proteins, such as cullin 5, Rbx1, and elongin B and C to form an active ubiquitin ligase complex (212, 213). This activated ligase complex ubiquitinates the MRN complex for proteasome-dependent degradation (190).

Building upon the Ad results, MRN roles are increasingly being elucidated in virus infection associated with human diseases—for example, high-risk human papillomavirus (HPV) that is implicated in cervical cancer and as well as in genital, head, and neck cancers (214). HPV-31 increases MRN complex half-life for productive viral replication (215). HPV-16 encodes transcription and replication factor E2 that interacts with RAD50-interacting protein 1 (Rint1) (216). Overexpression of exogenous Rint1 enhances HPV-16 replication, and a truncated mutant that lacks a RAD50-interacting domain reduces HPV-16 replication. Thus, HPV-16 may target MRN complex inactivation through Rint1 for productive replication. Similarly, the MRN complex inhibits replication of Kaposi sarcoma herpesvirus (KHSV) (217), which is implicated in Kaposi sarcoma, the most common cancer in HIV-infected individuals (218, 219). KHSV encodes a cytoplasmic isoform of latency-associated nuclear antigen (LANA). This isoform inhibits the host innate immune response by neutralizing MRN and cyclic GMP-AMP synthase DNA sensors (217).

The relationship linking viral infection, DNA repair, and innate immunity has biomedical importance, as oncolytic virus-based therapies are implemented in several cancers. For example, a modified oncolytic adenovirus mutant that lacks an antiapoptotic gene (E1B19K) sensitizes pancreatic cells in association with DNA-damage drugs (220), and the proposed mechanism is MRE11 inactivation. For oncolytic adenovirus type 5 being used in ovarian cancer cells (221), virus cytotoxicity is proposed to enable its ability to redistribute MRN (and other HR components). Overall, our evolving knowledge of interactions between viral infection, DNA repair, and the immune response requires defining multiple roles for MRN biochemistry.

SUMMARY AND PERSPECTIVES

A key aspect of the DNA double-helical structure, initially absent in the Watson and Crick model, is that it contains information not only for replication but also for repair of genomic damage (222) that is recognized by biologically conserved complexes such as MRN. Thus, MRN-DNA biochemistry is uncovering primary origins of normal and pathological cellular activities for DNA stress responses. Yet, this requires a path for a comprehensive characterization of proteins and nucleic acids, including their biologically relevant complexes and conformations (223). These observations have motivated our efforts to develop high-throughput and objective X-ray methods for comprehensive measurements of complexes and conformations and flexibility under near-physiological conditions (e.g., see 224–229). Thus, X-ray scattering and diffraction methods coupled to cryo-EM and single-molecule data are helping to define the MRN structural biochemistry described here.

At the cellular level, MRN defects can cause chromosome instability. At the systems level, we see the importance of maintaining the B-DNA double helix structure, as potential non-B-DNA structures (PONDS) occur at high frequency in cancer as translocations and deletion break points (230). Furthermore, at RFs where the double helix is opened for processing, precise control of DNA ends is needed to avoid not only mutations but also template switching and large repeat expansions (231, 232). These and other data provide compelling support for the notion that DNA structures at forks, ends, and breaks are intrinsic risk factors that complexes such as MRN are under strong evolutionary pressure to tightly control.

Combined genetic and biochemical data suggest that MRN has evolved to coordinate cellular network responses to DNA damage linked to mutation, cancer, and cell death. Its modular components are integrated into complexes and pathway networks in different ways, but MRN core structures and activities are maintained and conserved across all domains of life. Collective data show that multiple DNA stress responses are dynamically interrelated by MRN in ways that depend on its interactions and conformations. These results have broad implications for damage responses, as we are learning that even direct damage reversal enzymes, such as ALKBH3 (233), surprisingly function within large complexes for DNA and RNA integrity (234).

The evolutionarily conserved MRN complex is thus decoding a biological language of dynamic shapes controlling chemistry, connections, and complex functionalities from self-assembled, sequence-encoded building blocks whose fidelity is maintained by complexes such as MRN. For example, mutants modulating RAD50 dynamic conformations can switch biological outcomes from DNA excision to end joining (52). The data reviewed here demonstrate how MRN interactions and resulting activities feed into cellular decision making in DNA stress responses including viral defenses (typically considered to be only broadly related to the DDR). MRN subunit biochemical activities depend upon their three-dimensional folds, but MRN biological functions are governed by its complexes and conformationally regulated interaction networks. Thus, HR and MMEJ are activated in irradiated human cells owing to phosphorylation-dependent formation of either NBS1 or XRCC1 repair complexes with MRE11–RAD50 (93). In fact, the use of the same MRN subunits in multiple response networks provides a physical means to coordinate sensing, signaling, chromatin remodeling, and repair processing. This helps ensure appropriate responses to DSBs and stalled forks that avoid destructive interference and thus chromosome breakage and aberrations. As its cellular selectivity and function are determined partly by its assembly with partners and damaged DNA, we have focused on what we know and would like to know about dynamic MRN complexes and their control of biological outcomes. From the results to date, we can expect that defining the biochemical activities of MRN components and their assemblies will provide a critical foundation for understanding multiple DNA stress responses.

As a consequence of its dynamic conformations and assemblies, MRN responses are DNA structure specific rather than sequence specific. For example, MRN complexes confer specificity to distinct nuclease processing at DNA breaks and forks. These processes include an endonuclease nick and the initiation of exonuclease processing that is not adjacent to the DSB. Moreover, the resulting combined MRE11 endo- and exonuclease activities can act as go/no-go controls in the selection of HR versus NHEJ pathways (27). MRN can thus orchestrate DDRs as a sensor, regulator, and effector within distinct DSB repair, RF, and telomere rescue complexes as well as in viral response networks.

We propose that MRN conducts, coordinates, and connects responses to DNA stress largely by orchestrating three processes: assembly (localization and protection at damage), conformation (creation of a productive geometry for interactions and active site chemistry), and disassembly (controlling access and handoffs of DNA intermediates). In chemistry, transition state theory instructs us on the control of the reaction barrier as the expected rate-limiting step. In the cellular biochemistry of DNA replication and repair, however, the rate-limiting step is often not chemistry, which has been evolutionarily optimized for high efficiency. Instead, we see that protein and DNA binding energy is converted into conformational change to regulate network responses on the basis of the cell cycle state and the nature of the DNA damage. At the nanoscale of MRN, charge, chemical bonding, molecular mechanics, and thermal energy share similar scales: This allows their efficient interconservation to regulate specific activities within metastable MRN–DNA and MRN–protein intermediates. We propose that this ability to transfer binding energy into functionally important conformational states is an evolutionarily optimized and emergent MRN property. In practical terms, targeting conformational switching, as proposed here for RAD50 on the basis of the exceptional patient response in an otherwise failed clinical trial (60), suggests a novel biochemical strategy for precision medicine.

More broadly, current and emerging MRN results are aiding in addressing grand challenges of the postgenomic era for biochemistry and cancer biology: (a) an accurate prediction of pathology for damage response sequence variants, (b) an actionable mechanistic understanding of oncogenic replication and repair stress for biology and medicine, and (c) an advanced assessment of functional biochemical complexes and networks controlling outcomes from molecules to cells (235). Existing insights are already changing our concepts of damage detection, signaling, and repair with implications for strategies of controlling MRN activities for cancer, T cell aging, immune responses, and viral infections. Going forward, we will increasingly need to link structures of intact dynamic assemblies from cryo-EM and X-ray scattering studies to outcomes in cells and furthermore test this multiscale understanding with separation-of-function mutations and activity-specific chemical inhibitors, as was done for MRE11 (27). Emerging robust methods for the coexpression of large, multiprotein complexes provide enabling technologies for examining functional assemblies(75). The resulting foundational knowledge defining how these molecular machineries act may allow control of DNA stress responses for biology and medicine. We suggest that targeting MRN biochemical activities in DNA damage and the innate immune response together may overcome the mutational phenotype that characterizes cancer cells (236), analogously to how antibiotic knockdown of pathogens allows their clearance by our immune system.