The essential elements for the noncovalent association of two DNA ends during NHEJ synapsis

Abstract

One of the most central questions about the repair of a double-strand DNA break (DSB) concerns how the two free DNA ends are brought together - a step called synapsis. Using single-molecule FRET (smFRET), we show here that both Ku plus XRCC4:DNA ligase IV are necessary and sufficient to achieve a flexible synapsis of blunt DNA ends, whereas either alone is not. Addition of XLF causes a transition to a close synaptic state, and maximum efficiency of close synapsis is achieved within 20 min. The promotion of close synapsis by XLF indicates a role that is independent of a filament structure, with action focused at the very ends of each duplex. DNA-PKcs is not required for the formation of either the flexible or close synaptic states. This model explains in biochemical terms the evolutionarily central synaptic role of Ku, X4L4, and XLF in NHEJ for all eukaryotes.

Fig. 1

Ku and X4L4 can mediate efficient noncovalent synapsis of blunt dsDNA ends. a Schematic of the smFRET assay for blunt end synapsis. An 85 bp duplex with a terminal 5′-OH and a Cy5 dye located 4 bp away from the end was immobilized on the surface of the imaging coverslip. The 74 bp incoming dsDNA with a terminal 5′-OH and a Cy3 dye located 4 bp away from the end was added to the reaction via an injected solution that also contained 25 nM Ku and 50 nM X4L4. The figure illustrates a lateral configuration that may permit movement of the duplexes along one another [rather than an end-to-end configuration (see below)]. b Representative single-molecule time traces of donor (green) intensity, acceptor (magenta) intensity, and the corresponding E_FRET values (blue) for synapsis mediated by Ku and X4L4. Histograms of donor intensity (green), acceptor intensity (magenta), and E_FRET values (blue) within synapsis period are indicated at the right part of the figure. c Left panel: representative trace for dwell time (t_synapsis) calculation. Right panel: histogram and corresponding exponential fit of total synapsis time mediated by Ku and X4L4. Cy3 signal lifetime (including zero-FRET and detectable FRET portions) of the synaptic complex was used to calculate the total dwell time for each synapsis event and only the synapsis events (n = 139) with both start and end time points within the detection time window were included. Synapsis time shown on graph is represented as mean ± SD of two replicates. d Histogram of E_FRET values of all synapsis events mediated by Ku and X4L4. The E value shown on the graph was obtained from a Gaussian fit of the highest peak. n = 515 molecules. e One of the dynamic intensity traces (donor: green; acceptor: magenta) and corresponding E_FRET values (blue) of synapsis mediated by Ku and X4L4. The black line represents the smoothed trace of corresponding donor, acceptor, or E_FRET trace. Source data are provided as a Source Data1 file

Fig. 2

DNA-PKcs has little effect on synapsis mediated by Ku plus X4L4. a Normalized synapsis efficiency mediated by different NHEJ factors −25 nM Ku, 50 nM X4L4, and 12.5 nM DNA-PKcs using the glucose plus gloxy oxygen scavenger system. Data are represented as mean ± SD of at least four independent replicates. T-test (unpaired, two-tailed) was applied for p value calculation (p = 0.08). b Histogram of E_FRET values of all synapsis events mediated by 25 nM Ku, 50 nM X4L4, and 12.5 nM DNA-PKcs using the glucose plus gloxy oxygen scavenger system in the solution. The E value shown on the distribution was obtained by a Gaussian fit of the highest peak. n = 792 molecules. c Histogram and corresponding exponential fit of total synapsis time mediated by Ku, X4L4, and DNA-PKcs using the glucose plus gloxy oxygen scavenger system in the solution. Cy3 signal lifetime (including zero-FRET and detectable FRET portions) of the synaptic complex was used to calculate the total dwell time for each synapsis event, and only the synapsis events (n = 231) with both start and end time points within the detection time window were included. Synapsis time shown on graph is represented as mean ± SD of three replicates. d Summarized dwell times of synaptic complexes formed by Ku and X4L4, and by Ku, X4L4, and DNA-PKcs using glucose plus gloxy oxygen scavenger system in the solution. Error bars represent SD of three replicates. T-test (unpaired, two-tailed) was applied for p value calculation. The corresponding dwell distributions and exponential fits are shown in c and Supplementary Fig. 3c. Source data are provided as a Source Data1 file

Fig. 3

XLF increases end proximity of dsDNA within synaptic complexes. a, b Representative single-molecule time traces of donor (green) intensity, acceptor (magenta) intensity, and corresponding E_FRET values (blue) for synapsis mediated by 25 nM Ku, 50 nM X4L4, and 50 nM XLF. The right parts are histograms of donor intensity (green), acceptor intensity (magenta), and E_FRET values (blue) within synapsis period. c Histogram and corresponding exponential fit of lag time between synapsis starting and transition to high E_FRET. Only the events with a detectable transition from low E_FRET (E < 0.6) to high E_FRET (E ≥ 0.6) were included. n = 23 traces. Error represents the SEM of the fit. d Histogram of E_FRET values of all synapsis events mediated by Ku, X4L4, and XLF. The E value shown was obtained from a Gaussian fit of the highest peak of each kind of synaptic complex (FS or CS). n = 423 molecules. The diagram illustrates an end-to-end configuration of the CS state which is ready for ligation. e Histogram and corresponding exponential fit of synapsis time of high E_FRET (E ≥ 0.6) events mediated by Ku, X4L4, and XLF. Only the high E_FRET events (n = 63) with both start and end time points within the detection time window were included. Error represents the SD of two independent replicates. f XLF concentration-dependent synaptic complex formation. The reaction contains 25 nM Ku, 50 nM X4L4, and varied XLF. FS complex: E_FRET < 0.6, CS complex: E_FRET ≥ 0.6. Data are represented as mean ± SD of three independent replicates. Source data are provided as a Source Data1 file

Fig. 4

PAXX increases end proximity of two dsDNA but with modest efficiency. a Representative single-molecule time traces of donor (green) intensity, acceptor (magenta) intensity, and corresponding E_FRET values (blue) for synapsis mediated by 25 nM Ku, 50 nM X4L4, and 50 nM PAXX. The right parts are histograms of donor intensity (green), acceptor intensity (magenta), and E values (blue) within the synapsis period. b Histogram of E_FRET values of all synapsis events mediated by Ku, X4L4, and PAXX. The E value shown was obtained from a Gaussian fit of the highest peak of each kind of synaptic complex (FS and CS). n = 223 molecules. c Histogram of E_FRET values of all synapsis events with E_FRET ≥ 0.6. The E value was obtained from a Gaussian fit of the corresponding peak. n = 33 events. The data used here is the same batch as that used in b. d Dwell time of CS complex stimulated by PAXX alone or XLF plus PAXX. Only the high E_FRET (E ≥ 0.6) events with both start and end time points within the detection time window were included. Error bars represent SD of two replicates. The corresponding dwell distributions and exponential fits are shown in Supplementary Figs. 6e, f. e Histogram of E_FRET values of all synapsis events mediated by Ku, X4L4, XLF, and PAXX. The E value shown was obtained from a Gaussian fit of the highest peak of each type of synaptic complex (FS and CS). n = 429 molecules. f Fraction of low E_FRET (FS) complex and high E_FRET (CS) complex mediated by different combinations of NHEJ factors. Data are represented as the mean ± SD of at least three replicates. Source data are provided as a Source Data1 file

Fig. 5

Time-dependent accumulation of CS complex. a, b Kinetics of CS complex formation with 25 nM Ku, 50 nM X4L4, and 50 nM XLF (a) or with 25 nM Ku, 50 nM X4L4, 50 nM XLF, and 50 nM PAXX (b). Data are represented as the mean ± SD of two independent replicates. Source data are provided as a Source Data1 file

Fig. 6

Ensemble confirmation and the proposed model of NHEJ synapsis. a (Top) Gel results of blunt end ligation mediated by different combinations of NHEJ components in the presence and absence of 10% (v/v) PEG-8000. (Bottom) Quantified ligation efficiency. Reactions 1–12 correspond to the reactions represented by lanes 1–12 in the top panel, respectively. Data are represented as mean ± SD of two independent replicates. PEG-8000 is a volume excluder, which increases the collision frequency of the two dsDNA together ready for covalent ligation. With this volume excluder in the solution, the ligation efficiency was similar with or without stimulation from XLF, PAXX, or both (lanes 3–6). Without the volume excluder in the solution, where only NHEJ proteins would be available to bridge the two dsDNA ends for ligation, we find that Ku and X4L4 cannot mediate the ligation of blunt end dsDNA (lane 9), and XLF or PAXX can stimulate the covalent ligation (lanes 10–12). BZ15 was synthesized with a 5′ PO₄. b Ku and X4L4 mediate a flexible synapsis (FS), in which two dsDNA are brought into a lateral configuration. Aligning the two dsDNA within the FS complex to an end-to-end state is stimulated by the XLF or PAXX protein, and XLF is the more efficient one in this respect (as reflected by the length of the reaction arrow). One XLF dimer either interacting with X4 in X4L4 or with Ku may be sufficient for close synapsis (CS), but we speculate that more than one XLF dimer binding to both Ku and X4 may more fully stabilize the CS complex. We have not shown filament formation in this diagram, but for chromatinized DNA templates, it is possible that filaments may be important. Source data are provided as Source Data1 and Source Data2 files