Holding it together: DNA end synapsis during non-homologous end joining

Abstract

DNA double strand breaks (DSBs) are common lesions whose misrepair are drivers of oncogenic transformations. The non-homologous end joining (NHEJ) pathway repairs the majority of these breaks in vertebrates by directly ligating DNA ends back together. Upon formation of a DSB, a multiprotein complex is assembled on DNA ends which tethers them together within a synaptic complex. Synapsis is a critical step of the NHEJ pathway as loss of synapsis can result in mispairing of DNA ends and chromosome translocations. As DNA ends are commonly incompatible for ligation, the NHEJ machinery must also process ends to enable rejoining. This review describes how recent progress in single-molecule approaches and cryo-EM have advanced our molecular understanding of DNA end synapsis during NHEJ and how synapsis is coordinated with end processing to determine the fidelity of repair.

1. Introduction

To survive, cells must be able to tolerate a wide range of endogenous and exogeneous insults that chemically damage their DNA¹. Accordingly, cells employ a broad set of DNA repair and damage tolerance pathways. DNA double strand breaks (DSBs) are a particularly toxic form of DNA damage, as failures to repair even a single DSB can result in cell death. Eukaryotes possess multiple DSB repair pathways with nonhomologous end joining (NHEJ) and homologous recombination (HR) being the two predominate pathways². In NHEJ broken DNA strands are directly rejoined by a ligase³. By contrast, HR uses a sister chromatid to template repair⁴. Given the absence of a sister chromatid for most of the cell cycle, NHEJ is the primary DSB repair pathway in human cells and is estimated to repair up to 80% of spontaneous DSBs⁵. In addition, NHEJ also plays critical roles in the repair of programmed DSBs generated during antigen receptor development in T and B cells⁶.

Properly functioning NHEJ is essential to human health⁷. Loss of NHEJ due to mutations in core NHEJ factors results in both deficiencies in DSB repair and defective V(D)J recombination which causes severe combined immunodeficiency (SCID)⁸. These patients also frequently exhibit radiosensitivity and developmental abnormalities. In contrast, overactive NHEJ due to overexpression of core NHEJ factors in tumors is associated with resistance to DSB-inducing therapies^9,10.

Upon formation of a DSB, DNA ends are rapidly tethered by a multi-protein synaptic complex which ultimately ligates them together (Fig. 1). Assembly of the NHEJ machinery begins within the presynaptic complex before DNA ends are paired. Subsequent formation of intermolecular interactions between factors on each end leads to formation of the synaptic complex. Often DNA ends are not initially compatible for ligation as common agents that generate DSBs, such as reactive oxygen species or ionizing radiation, can alter DNA end chemistry¹¹. To address these challenges, NHEJ employs numerous end processing enzymes that act on DNA ends to enable ligation.

Figure 1:

An overview of the NHEJ pathway.

Rapid and stable DNA end synapsis is critical for maintaining the correct pairing of DNA ends as the NHEJ pathway lacks a mechanism to determine if two DNA segments belong together. Mispairing of DNA ends leads to potentially oncogenic chromosome translocations of which NHEJ is the primary causative pathway in human cells^12,13. In this review, I will describe our current understanding of DNA end synapsis during NHEJ and how it promotes efficient and faithful repair.

2. Core and Accessory NHEJ Factors

NHEJ is carried out by a multi-protein complex consisting of core and accessory NHEJ factors³. DSBs are recognized by the ring-shaped Ku70/Ku80 heterodimer (Ku)¹⁴. Ku is extremely abundant in vertebrates and rapidly binds to DNA ends¹⁵. As it interacts with many NHEJ factors, Ku acts as a recruitment platform that assembles the NHEJ machinery^16–18. One of these downstream factors is the large protein kinase DNA-PKcs, which binds to DNA-bound Ku to form the DNA-PK holoenzyme¹⁹. A member of the phosphoinositide 3-kinase (PI3K) – related kinase family, DNA-PKcs kinase activity is activated upon DNA binding^20,21. While numerous factors are phosphorylated by DNA-PKcs, it remains unclear whether many of these modifications have any significant functional consequences. Likely the most important target of DNA-PKcs phosphorylation is itself. Extensive autophosphorylation of DNA-PKcs occurs within two clusters known as ABCDE and PQR, and ablating phosphorylation sites within these clusters leads to substantial repair defects in cells^22,23.

DNA end ligation is ultimately carried out by DNA ligase IV (LIG4) and its cofactors. LIG4 forms a constitutive complex with XRCC4²⁴. While XRCC4 possesses no known catalytic activity, it plays critical roles in NHEJ by stabilizing LIG4²⁵ and facilitating DNA end synapsis through its interaction with its paralog XRCC4-like factor (XLF)^26,27. Like XRCC4, the major role of XLF seems to be in facilitating DNA end synapsis. Paralog of XRCC4 and XLF (PAXX) also participates in synapsis although its contributions are genetically redundant with XLF^28–30.

In addition to the core NHEJ factors, a number of accessory NHEJ factors are recruited to DNA ends^16,17. Accessory factors include end processing enzymes and other factors whose roles in NHEJ are more illdefined. In contrast to core factors, which display significant defects in end joining upon ablation in cells, accessory factors have more modest roles in repair and are likely only needed under certain conditions. Processing factors can be divided into two classes: damage correction enzymes, such as polynucleotide kinase 3´-phosphatase (PNKP) and tyrosyl-DNA phosphodiesterase 1 (TDP1), which act on DNA ends in an error-free manner and error-prone factors which include NHEJ associated polymerases (e.g. pol λ and pol μ) and nucleases (e.g. Artemis) which can lead to insertions and deletions. A more thorough treatment of the various end processing enzymes and their activities can be found in other reviews^3,31.

3. Synapsis is mediated by structurally distinct intermediates

During NHEJ DNA ends pass through structurally and functionally distinct intermediates. Single-molecule and cryo-EM studies of the NHEJ machinery have enabled the investigation of these intermediates during repair reactions. Single-molecule Förster resonance energy transfer (smFRET)³² between donor and acceptor labeled DNA ends reports on the distance between ends in real time^33–35. Combining these smFRET assays with multi-color imaging of fluorophore-labeled protein factors allows one to observe the stoichiometry and structural dynamics of the NHEJ machinery^36–39. Nanomanipulation-based single-molecule methods have also contributed to our understanding of synapsis by measuring how the stability of the synaptic complex is influenced by the application of biologically relevant forces^40,41. Complementary high-resolution cryo-EM imaging of the NHEJ machinery provides atomic resolution of the intermolecular interactions that maintain synapsis^42,43.

Working in the physiologically complex model system of Xenopus laevis egg extracts, we showed that the NHEJ machinery passes through at least two structurally distinct synaptic complexes³³. Ends are initially weakly tethered in a long-range (LR) synaptic complex in which they are held at a distance greater than a hundred Angstroms. Initial formation of the LR complex requires Ku and DNA-PKcs but does not require other core factors. Structural remodeling of the LR complex results in the formation of the short-range (SR) complex in which the ends are poised to be ligated. In contrast to the LR complex, the SR complex is quite stable and persists for tens to hundreds of seconds when ligation is blocked. Transition from LR to SR synapsis requires DNA-PKcs kinase activity and the presence of XLF and XRCC4-LIG4. Besides their clear structural differences, the LR and SR synaptic complexes are also functionally distinct. DNA ends are largely protected within the LR complex but become accessible for both ligation and DNA end processing within the SR complex⁴⁴.

Both single-molecule and structural studies using reconstitutions of purified human NHEJ factors provide further support for the LR and SR synaptic complexes. Magnetic tweezers experiments demonstrated that minimal reconstitutions of Ku and DNA-PKcs could weakly tether DNA ends, similar to initial observations of the LR complex⁴⁰. Stable synapsis, consistent with the SR complex, required the addition of XLF and XRCC4-LIG4. Cryo-EM studies have identified structurally distinct LR and SR complexes and have identified the intermolecular interactions that assemble these states^42,45,46. The next sections will describe the role these intermolecular interactions play in structurally remodeling the NHEJ machinery during repair.

4. The presynaptic NHEJ complex

Assembly of the presynaptic NHEJ complex is initiated with the recruitment of Ku to DNA ends. While multiple copies of Ku will readily load onto DNA in minimal in vitro reconstitutions⁴⁷, the stoichiometry of Ku is limited to approximately a single Ku molecule per DNA end in physiological systems¹⁵. Hyperloading of Ku is minimized by the binding of DNA-PKcs to the presynaptic complex³⁸ which associates with DNA ends at a similar rate as Ku⁴⁸. DNA-PKcs interacts with Ku via a number of interfaces including the C-terminal domain of Ku80 and also makes extensive contacts with the DNA end^49–53. Through these interactions, DNA-PKcs acts to retain Ku at the DNA end and likely sterically blocks further Ku loading. Ku stoichiometry may also be influenced by unloading of Ku through an active mechanism of polyubiquitination and subsequent p97-mediated removal^54,55. However, experiments in Xenopus egg extracts and human cells suggest that active Ku removal is not likely the major determinant of Ku stoichiometry³⁸.

The recruitment of NHEJ factors to DNA ends is largely mediated through interactions with Ku and does not require end synapsis³⁹. Many of these factors interact with Ku through peptide motifs known as Ku binding motifs or KBMs. These KBMs have been classified into two groups based on their similarities to those present in Aprataxin and PNKP-like factor (APLF) and XLF⁵⁶. The XLF-like motif is comprised of a basic patch of residues and a conserved phenylalanine at the C-terminus of XLF and binds relatively weakly to Ku. Binding of the XLF KBM to Ku80 requires a rotation of the vWA domain to expose its binding site⁵⁷. PAXX also contains an XLF-like motif although it binds to the vWA domain of Ku70, which similarly must open to enable PAXX binding^58–60. The APLF KBM contains a basic RKR patch followed by a hydrophobic patch of PXW and binds ~150-fold more tightly to Ku80 than the XLF motif. As it occupies a unique site near the periphery of the vWA domain, APLF and XLF should not compete for Ku binding. APLF KBMs present in the accessory factors Werner protein (WRN) and modulator of retroviral infection (MRI) bind the same site on Ku80 as APLF and thus Ku binding is likely competitive amongst these factors⁶¹.

XRCC4-LIG4 is recruited to DSBs through a unique interaction between the Ku70/80 core domains and the N-terminal BRCT domain of LIG4^42,62. The NHEJ associated polymerases, pol λ, pol μ and TdT also have BRCT domains which interact with Ku^63–66. However, stable recruitment of NHEJ polymerases require additional interactions with XRCC4^67,68. The end processing enzymes PNKP⁶⁹ and Aprataxin⁷⁰ also interact with XRCC4 via their forkhead associated (FHA) domains. In addition to its KBM, APLF uses its FHA-XRCC4 interaction to stabilize LIG4 at DNA ends^71–73.

While core NHEJ factors and many accessory proteins are recruited to the presynaptic complex, it does not have a well-defined composition. Instead, Ku and likely DNA-PKcs form a recruitment platform on which other factors are rapidly binding and unbinding. Single-molecule imaging experiments in Xenopus egg extracts showed that while Ku remains stably associated within the presynaptic complex, XLF and XRCC4-LIG4 are highly dynamic and typically dissociate within a few seconds³⁹. The short lifetimes of these factors arise from the relatively weak interactions that the XLF KBM and LIG4 BRCT domain make with Ku. While modest in strength, these interactions play an important role in concentrating XLF and XRCC4-LIG4 near DNA ends and are necessary for efficient end joining^37,74. Upon recruitment to DNA ends, intermolecular interactions between the XLF and XRCC4 head domains stabilizes both factors within the presynaptic complex; this interaction has been directly observed in “single-DNA end” cryo-EM structures (Fig. 2A)⁴³. Furthermore, LIG4 is stabilized through direct binding of the DNA end which can occur within the presynaptic complex³⁷. It is the formation of essential intermolecular interactions between factors bound on each DNA end that drives formation of the synaptic complex and enables its subsequent structural remodeling.

Figure 2:

The early stages of DNA end synapsis. (A) Core NHEJ factors are initially recruited to the presynaptic complex (PDB: 7NFE). Ends are initially held at a distance within the (B) “domain swap” (PDB: 6ZHE) or the (C) “XLF-mediated” (PDB: 7LT3) complexes. (D) A cartoon depiction of potential assembly pathways of the NHEJ machinery.

5. DNA-PK dimerization within the long-range complex

In the LR complex, synapsis is primarily mediated by dimerization of DNA-PK which holds the DNA ends at a distance. Initial smFRET experiments characterized the LR complex by the absence of energy transfer between the colocalized DNA ends, which indicates that they are held at least ~100 Angstroms apart³³. Cryo-EM structures have refined this distance to 115 Angstroms and surprisingly identified two structurally distinct LR complexes that maintain this distance^42,43,46. In both LR complexes the major drivers of synapsis are interaction interfaces between DNA-PK on each DNA end. In the first of these structures, termed the “domain swap” LR complex, the C-terminal domain of Ku80 reaches across the DNA break and interacts with the cradle of the opposing DNA-PKcs⁴⁶. The second structure, named the “XLF-mediated” LR complex, is comprised of an extensive interface between DNA-PKcs protomers along with a bridge comprised of XRCC4-LIG4 and XLF that spans DNA ends (Fig. 2C). Within the “XLF-mediated” LR complex the kinase domain of DNA-PKcs is positioned to phosphorylate the ABCDE and PQR clusters on the opposing DNA-PKcs protomer⁴². Introduction of mutations into both the “domain swap” and “XLF-mediated” structures sensitized cells to various DSB inducing agents which suggests that both states are functionally important in cells^43,45. How these distinct LR complexes assemble remains unclear (Fig. 2D). As the LR complex forms efficiently in the absence of XLF and XRCC4-LIG4³³, it is possible that formation of the “domain swap” complex may proceed the “XLF-mediated” complex. Alternatively, these LR complexes may have distinct roles in the repair of complex DNA ends and thus may function in parallel pathways⁴⁵. Further work is necessary to elucidate the roles and assembly of these structurally distinct LR complexes.

While numerous studies have proposed a role for DNA-PKcs in DNA end synapsis, its exact catalytic and non-catalytic roles in NHEJ remain contested. To complicate matters, DNA-PKcs contributes to numerous cellular processes including ribosomal RNA processing⁷⁵, telomere maintenance⁷⁶, and chromatin remodeling near DSBs^77,78. Independent structural and biochemical studies showed that in the absence of other factors, DNA-PK will readily dimerize to bring DNA ends together^{40,46,79–81}. However, in vitro studies have reached opposing conclusions on the necessity of DNA-PKcs for end joining. Depletion of DNA-PKcs or the inhibition of its kinase activity largely prevents end joining in both human cell and Xenopus egg extracts^33,82,83. DNA-PKcs mediated chromatin remodeling has been proposed as a necessary mechanism in extracts that support chromatinization⁸⁴. However, experiments in Xenopus egg extracts were performed on 100 base pair DNAs that don’t permit nucleosome assembly and thus argue for a structural role for DNA-PKcs in end synapsis³³. Within human reconstitutions of NHEJ, DNA-PKcs has been found to be important for end joining in certain studies^40,85–87 but not in others^34,88. Genetic studies indicate an important role for DNA-PKcs in NHEJ, albeit not absolutely essential in every context. Deletion of DNA-PKcs or ablating its autophosphorylation sites sensitizes cells to DNA double strand breaks although not as severely as loss of LIG4^22,23,89. However, loss of DNA-PKcs catalytic activity leads to an embryonically lethal phenotype in mice in line with LIG4 knockouts⁸⁹. Within error-free end joining reporter assays, loss of DNA-PKcs led to variable defects in repair ranging from modest to more moderate (10x decrease) effects depending on the cell line and the expression level of DNA-PKcs⁹⁰. These defects were substantially enhanced when combined with mutations that weakened the interaction between XLF monomers which suggests that loss of LR synapsis is amplified when SR complex formation is also attenuated.

Differences in the dependency of DNA-PKcs in biochemical studies may result from the fact that synapsis mediated by DNA-PK is relatively weak. Single-molecule magnetic tweezers experiments showed that upon the application of low forces (~1 pN) synaptic complexes composed of Ku and DNA-PKcs rupture within a few hundred milliseconds⁴⁰. Addition of PAXX led to a ~10-fold increase in lifetime suggesting that other NHEJ factors not typically included in reconstitutions but present in cells and cell extracts may play important roles in stabilizing the LR complex. Within Xenopus egg extracts the lifetime of the LR complex is on the order of a second to a few seconds which is much shorter than the lifetime of the SR complex³³. Structural studies have visualized additional interactions with core NHEJ factors that may stabilize DNA-PKcs at DNA ends and thus contribute to LR complex stability. Of particular note, is the observation that the C-terminal tail of XRCC4 interacts with the FAT domain of DNA-PKcs⁴².

DNA-PKcs is rapidly recruited to breaks⁹¹ where it regulates Ku stoichiometry³⁸. Given its presence at DNA ends and that numerous studies have shown that DNA-PK can dimerize, it is likely that DNA-PKcs is involved in the early stages of end synapsis. However, the presence of the LR complex is not strictly required for end joining to occur in cells. Such robustness is commonly found in multiprotein complexes which typically have multiple assembly pathways. In cells, as opposed to in vitro experiments, the massive size of chromosome domains prevent DNA ends from rapidly diffusing apart. Furthermore, higher order chromatin interactions and possibly loop extrusion⁹² keep ends tethered, reducing the need for LR synapsis. Single-molecule experiments within human reconstitutions of NHEJ lacking DNA-PKcs demonstrate that synapsis can occur although it is unclear with what efficiency compared to more physiologically complex reconstitutions^34,88. Indeed, end joining efficiency in Xenopus egg extract is remarkably efficient with nearly 100% of substrates joined within 30 minutes⁹³. Reconstitutions of human NHEJ factors have commonly reported much lower joining efficiencies^{65,66,85,86,88,94}. Collectively, the bulk of evidence points to the LR complex as being a physiological intermediate during NHEJ in cells that promotes assembly of the SR complex and thus stimulates end joining.

6. DNA-PKcs kinase activity facilitates remodeling of the LR complex

Substantial structural remodeling occurs during the transition from the LR to SR states. In contrast to the LR complex, DNA ends are stably held and closely aligned for ligation within the SR complex^33,42. Initial observations in Xenopus egg extracts identified the necessity of DNA-PKcs kinase activity and the core factors XLF and XRCC4-LIG4 for the LR to SR transition³³. How DNA-PKcs kinase activity remodels the LR complex remains unknown. However, a substantial body of literature suggests that DNA-PKcs autophosphorylation alters its interaction with DNA ends⁹⁵. Initially, DNA-PKcs acts as a block to both NHEJ associated end processing factors^44,85,87 and to the long-range resection required for HR⁹⁶. Phosphorylation of DNA-PKcs modulates the affinity of DNA-PKcs for the DNA end which results in increased end accessibility and ultimately DNA-PKcs dissociation^91,97. On this basis, biochemical reconstitutions of the SR complex for cryo-EM studies lacked DNA-PKcs, although it has not been directly observed when in the repair reaction DNA-PKcs is lost from the NHEJ machinery. Therefore, the emerging model is that DNA-PKcs autophosphorylation is required to relieve end protection within the LR complex which facilitates SR complex formation.

7. The role of XLF and XRCC4-LIG4 in short-range synapsis

XLF and XRCC4-LIG4 play essential structural roles in the transition from the LR to SR complexes. XRCC4 and its paralogs exist as homodimers with globular N-terminal head domains, a dimerization domain mediated by a coiled-coil stalk and unstructured C-terminal tails^98,99. At the extreme C-terminus of each XLF monomer is a KBM that acts to concentrate XLF near DSBs^56,100,101. The C-terminal tail of XLF also acts as a flexible tether that facilitates formation of a head-to-head interaction between XLF-XRCC4⁷⁴. This interaction is important for NHEJ, as ablating it via point mutants on XLF or XRCC4 strongly attenuates end joining in a variety of biochemical and cell-based models and blocks formation of the SR complex in Xenopus egg extracts^36,102,103.

Different models propose how the interaction between XLF and XRCC4 facilitates end synapsis. Alternating filaments of XLF and XRCC4 form in vitro and have been suggested to act as end bridging structures that allow the DNA ends to slide relative to each other^41,104–106. Within fixed cells XLF and XRCC4 form elongated foci which have been interpreted as extended filaments³⁴. Different organizations of these filaments have been suggested including large cylindrical bundles¹⁰⁵. However, it is not clear if these filaments could allow factors including DNA-PKcs, LIG4 or end processing factors to access DNA ends. In contrast, direct imaging of XLF in Xenopus egg extracts revealed that a single XLF dimer mediates formation of the SR complex³⁶. Importantly, end joining efficiency and synapsis were strongly attenuated by asymmetric XLF heterodimers containing only one monomer capable of interacting with XRCC4. These results support a model in which a single XLF interacting with XRCC4-LIG4 complexes via each of its head domains bridges DNA ends prior to SR synapsis. Structural studies have now directly visualized this ternary complex of XLF and XRCC4-LIG4 in both the “XLF-mediated” LR and SR complexes (Figs 2C and 3A)^42,43.

Figure 3:

The short-range synaptic complex. (A) Cryo-EM structure of the SR complex (PDB: 7LSY). (B) Synapsis within the SR complex is maintained by multiple independent protein networks.

XLF plays a central role in SR synapsis. In addition to its role in the XRCC4 – XLF “bridge”, the SR complex structure shows that the KBMs of XLF interact with Ku on each DNA end⁴². While this interaction is not required for SR complex formation⁷⁴ it likely acts as one of a series of redundant connections (Fig. 3B) that maintains synapsis. Furthermore, the XLF stalk interacts with each Ku70 subunit⁴². All together XLF contributes to three independent protein networks that tether DNA ends within the SR complex and likely positions LIG4 to capture DNA ends when they become accessible.

Like DNA-PKcs, loss of XLF leads to variable defects in end joining in cells. Depending on the particular cell line and assay, XLF loss phenocopies XRCC4²⁶ or results in a more modest defect⁹⁰. The mechanistic basis for this variability remains unknown. However, it likely reflects that similar to LR synapsis, SR synapsis is not strictly required for NHEJ and instead stimulates ligation to varying degrees based on cellular conditions, such as the expression level of DNA repair factors.

LIG4 plays a structural role in end synapsis that is independent of its catalytic activity in ligation^{33,82,88,107–109}. LIG4 contains an N-terminal DNA binding domain, catalytic and OB domains and a tandem BRCT domain at the C-terminus which interacts with XRCC4 and Ku¹¹⁰. Experiments in Xenopus egg extracts and in human reconstitutions of NHEJ showed that LIG4 is important for stable SR synapsis^33,40,88. DNA ends are closely aligned during SR synapsis suggesting that the structural role of LIG4 may be to span DNA ends in the SR complex. Work by our laboratory has shown that, consistent with this hypothesis, DNA binding by LIG4 is necessary for SR complex formation³⁷. Furthermore, three-color smFRET experiments allowed us to simultaneously observe SR synapsis and end binding by LIG4. These experiments showed that shortly before SR synapsis two LIG4 molecules are present with each bound to a DNA end. This stoichiometry is consistent with a bridge of XRCC4-XLF-XRCC4 being a critical structural intermediate to enable SR complex formation^36,42. During the transition from the LR to SR complexes, one LIG4 molecule unbinds, liberating a DNA end. This end is rapidly captured by the remaining ligase, such that a single LIG4 spans the DSB at the moment of SR synapsis³⁷. Consistent with this model, the cryo-EM structure of the SR complex (Fig 3A)⁴² shows a single LIG4 catalytic core binding both DNA ends. The SR complex structure also shows a second LIG4 molecule although only the tandem BRCT domain is resolved; this structure may reflect an intermediate that precedes collapse of the XRCC4-XLF-XRCC4 bridge and loss of the unbound ligase or reassociation of a second ligase following establishment of SR synapsis. Coordinating SR synapsis with the engagement of LIG4 with DNA ends facilitates the rapid ligation of compatible DNA ends and minimizes unnecessary end processing.

8. End synapsis and the regulation of end processing

As some end processing enzymes generate insertions or deletions of genetic information, NHEJ is often considered to be error-prone although this is likely too simplistic¹¹¹. Sequencing of repair junctions indicates that the physiological context of DSBs is important for determining their repair fidelity. V(D)J coding joints resulting from the repair of programmed breaks display highly variable sequences for identical DNA ends^112,113. Based on these results it was proposed that each DNA end is processed independently of the other in an iterative fashion, with rounds of processing followed by synapsis and attempted ligation (Fig. 4, left)⁶⁵. In contrast to programmed breaks, a large body of work in cells and cell extracts demonstrated that DNA ends that are initially compatible for ligation are largely joined without processing and ends that must be processed are acted on minimally^{44,83,114–117}. Collectively these results imply that ligation is prioritized over end processing and that processing is tightly coordinated with synapsis to minimize mutagenesis during the repair of spontaneous breaks. The discrepancy between programmed and spontaneous breaks likely arises due to the different apparent goals involved in the repair of these breaks. During V(D)J recombination, error-prone NHEJ contributes to the sequence diversity at the coding joint which leads to a broader repertoire of antigen receptors. In contrast, during the repair of spontaneous breaks, high fidelity NHEJ minimizes mutagenesis and genome instability.

Figure 4:

The coordination of end processing with end synapsis. DNA end sensing by DNA-PKcs biases end processing to occur outside the context of end synapsis for hairpin ends generated during programmed breaks (left) or tightly coordinated with synapsis for spontaneous breaks (right).

Structural work investigating the DNA-PK-Artemis complex points to an intriguing mechanism by which end sensing by DNA-PKcs may be used to switch the fidelity of NHEJ¹¹⁸. Within the cradle of DNA-PKcs an end blocking helix interacts with the DNA end altering the conformation and activity of the kinase domain. On hairpin DNA ends mimicking those formed during V(D)J recombination by the RAG recombinase, the kinase adopts a conformation that enables rapid cis autophosphorylation of the ABCDE cluster. As phosphorylation of this cluster is associated with increased end accessibility^23,85,87,119, processing can occur on ends in a manner largely uncoupled from synapsis. Consistent with this model, hairpin ends only promote ABCDE phosphorylations and not other DNA-PK sites or NHEJ factors¹²⁰. In contrast, on ends generated by spontaneous breaks, melting of the terminal basepair by the end blocking helix positions the kinase domain such that DNA-PKcs autophosphorylation of the ABCDE cluster is favored to occur in trans, thereby coupling end accessibility with synapsis. Earlier work had also suggested trans autophosphorylation¹²¹. We showed in Xenopus egg extracts that for a diverse set of DNA ends containing either single-stranded DNA overhangs or adducts, end processing is largely restricted to the SR complex even though processing factors can be recruited to the LR complex (Fig. 4, right)⁴⁴. Consistent with this observation, perturbations that block SR synapsis also block processing in human cells and human cell extracts^122–126. LIG4 must bind both ends to initially form the SR complex³⁷. Therefore, restricting processing to the SR complex prioritizes ligation over end processing and results in the rapid ligation of compatible ends. Furthermore, restricting end processing to the ligation-competent SR complex minimizes processing beyond what is required for ligation. This is particularly advantageous for polymerase activity, as DNA ends can be paired within the SR complex and a polymerase can utilize any existing microhomology. Collectively these studies point to DNA-PKcs playing a central role in regulating end processing during the repair of both programmed and spontaneous breaks. Other factors impact the diversity of repair junctions including the nature of the DNA ends. For example, 3´ noncomplementary overhangs which mimic those generated by Artemis opening of RAG-induced hairpins show much more diversity than similar 5´ overhangs⁴⁴. Repair fidelity of programmed breaks may also be impacted by the RAG recombinase, which has been implicated in end synapsis and handoff to the NHEJ machinery during V(D)J recombination^127–130.

9. Concluding Remarks and Future Directions

NHEJ is carried out by a dynamic multiprotein machine that coordinates DNA end synapsis with end processing to tune the fidelity of repair. Upon formation of a DSB, Ku and DNA-PKcs rapidly bind DNA ends to form a stable platform on which other factors are assembled. Given the relatively weak affinity these downstream factors have for Ku, they are readily turned over within the presynaptic complex. Establishing intermolecular interactions between these factors stabilizes them at DNA ends and drives formation of the synaptic complex. Synapsis is initiated by the formation of the LR complex which tethers DNA ends at a distance through distinct DNA-PK interfaces. Subsequent assembly of a bridge composed of two copies of XRCC4-LIG4 and one copy of XLF facilitates formation of the SR complex through positioning of LIG4. SR synapsis further requires DNA-PKcs kinase activity which likely modulates the accessibility of factors to bind DNA ends and the engagement of both ends by a single copy of LIG4. As end processing of spontaneous breaks is largely restricted to the SR complex, LIG4 is given preferential access to ligate compatible DNA ends. Furthermore, end processing is minimized by carrying out processing within a ligation competent synaptic complex.

Our mechanistic understanding of NHEJ has increased dramatically with the emergence of single-molecule and cryo-EM studies and yet many questions remain. While autophosphorylation of DNA-PKcs is a critical step to allow for progression of the NHEJ reaction, it remains unclear when DNA-PKcs dissociates from DNA ends. DNA-PKcs is also known to phosphorylate numerous other core and accessory factors. As these phosphorylations affect the stability of Ku¹³¹ and XLF-XRCC4¹³² on DNA ends, DNA-PKcs catalytic activity may regulate disassembly of the NHEJ machinery more broadly. Other outstanding issues include further defining the network of intermolecular interactions that hold DNA ends together within the various synaptic complexes and how accessory factors, such as APLF, and noncoding RNAs contribute to these networks^133–135. In addition, it remains unknown how SR synapsis is maintained in the switch from attempted ligation to end processing³⁷ and how processing enzymes are selected to minimize mutagenesis^44,117. Answering these questions are necessary to fully understand the molecular organization and dynamics of this critical DNA repair machinery.

Highlights.

• NHEJ repairs DNA double strand breaks by a direct ligation mechanism.

• The NHEJ machinery tethers DNA ends in a dynamic synaptic complex.

• End synapsis progresses through structurally and functionally distinct states.

• End processing is coordinated with end synapsis to maximize repair fidelity.