XLF and APLF bind Ku80 at two remote sites to ensure DNA repair by non-homologous end joining

Abstract

The Ku70-Ku80 (Ku) heterodimer binds rapidly and tightly to the ends of DNA double-strand breaks and recruits factors of the non-homologous end-joining (NHEJ) repair pathway through molecular interactions that remain unclear. We have determined crystal structures of the Ku-binding motifs (KBM) of the NHEJ proteins APLF (A-KBM) and XLF (X-KBM) bound to a Ku-DNA complex. The two KBM motifs bind remote sites of the Ku80 α/β domain. The X-KBM occupies an internal pocket formed by an unprecedented large outward rotation of the Ku80 α/β domain. We observe independent recruitment of the APLF-interacting protein XRCC4 and of XLF to laser-irradiated sites via binding of A- and X-KBMs, respectively, to Ku80. Finally, we show that mutation of the X-KBM and A-KBM binding sites in Ku80 compromises both the efficiency and accuracy of end joining and cellular radiosensitivity. A- and X-KBMs may represent two initial anchor points to build the intricate interaction network required for NHEJ.

Figure 1.. Crystal structure of the APLF KBM (A-KBM) bound to the Ku80 vWA domain.

(a) Positions of the A-KBM (magenta) and X-KBM (blue) motifs in APLF, XLF, WRN and CYREN. The C-terminal domain of PAXX contains a P-KBM that interacts with Ku70 subunit. NTD: N-terminal domain. (b) Overall view of the quaternary complex Ku70/Ku80/hDNA/(APLF peptide). The A-KBM (magenta) binds at the periphery of the Ku80 (light green) vWA domain. The Ku70 subunit and hDNA are represented respectively in orange and red. The hairpin part of the DNA has been removed for clarity. (c) The N-terminal part of the A-KBM motif has an extended conformation whereas the C-terminal residues form a turn. (d-e) Zoom of the interactions made by (d) the hydrophobic patch and (e) the basic patch of the A-KBM. (f) The A-KBM binding site is delineated by conserved residues of Ku80 vWA domain. The binding site is represented in surface mode with amino acids colored according to their conservation rate: red (highly conserved) to white (not conserved)). The conservation rate was measured using sequences of metazoan Ku80. The orientation is the same as in (c).

Figure 2.. Crystal structure of the XPLF KBM (X-KBM) bound to the Ku80.

(a) Crystal structure of the quaternary complex Ku70-Ku80-DNA-(X-KBM peptide). The X-KBM (blue) binds in an internal site of the Ku80 subunit created upon an outward rotation of the vWA domain. The Ku80 vWA opening creates a large groove between the Ku80 vWA and the rest of the heterodimer. (b) The crystal structure of Ku70/Ku80/DNA in presence of the A-KBM is shown with the same orientation. (c-d) Comparison of the X-KBM binding site in presence of X-KBM (c) or A-KBM (d) peptides. The X-KBM interacts with Ku80 residues involved in Ku intramolecular contacts in the closed state of Ku observed with the A-KBM or with no peptide. The last GLFS residues of the X-KBM interact with the bottom of the groove formed in the open state. The glutamic acid presents an atypical hydrophobic environment and could be at the origin of the vWA instability. The X-KBM residues occupy the position of the helix 236–241 of Ku80 in the closed conformation and some X-KBM side chains (R295^X, L297^X and F298^X) mimic the intramolecular interactions made by Ku80 residues with the vWA domain. (e) Gel shift assay with XLF and Ku in presence of a 50bp DNA with a FAM in 5’and competition with pXLF containing the X-KBM motif. The arrow indicates the XLF Ku-DNA complex. Uncropped gel image is shown in Supplementary Data Set 1. (f) The pair distributions P(r) obtained in solution by SAXS analysis indicates an opening of the Ku70/Ku80/DNA complex with higher Dmax and Rg in presence of the X-KBM (blue line) compared to the Ku/DNA complex without peptide (grey line) and to the A-KBM complex (magenta line). Values deduced from SAXS analysis are reported beside the curves.

Figure 3.. Life cell imaging of A-KBM and X-KBM recruitment after nuclear micro-irradiation.

(a) Wild-type (WT) and mutant CFP-(A-KBM) behaviour at 0 s and 50 s after laser nuclear micro-irradiation. The white rectangle and arrows mark irradiated areas. Magnification: X40. (b) Dynamics of wild-type and mutant CFP-(A-KBM) at laser-induced damage sites in U2OS cells. Mean values of relative fluorescence with s.e.m. were calculated from data obtained in several individual cells: n=23 and 19 cells for WT and mutant A-KBM, respectively). p values at last time point were calculated using unpaired two-tailed t-test: WT vs W189G p<0.0001. (c) Dynamics of wild-type and mutant CFP-(X-KBM) at laser-damaged sites as in b). n=27, 21, and 24 cells for WT, L297E and L297W X-KBM, respectively. p values at last time point : WT vs L297W p=0.8574; WT vs L297E p=0.0021. (d-e) Dynamics of CFP-(A KBM) (d) and (X-KBM) (e) at laser damaged sites in cells expressing wild-type or I122R mutant Ku80 as in b). n=20, 13 cells for A-KBM in WT or I122R Ku80, and n=48 and 39 cells for X-KBM in WT or I122R Ku80, respectively. p values at last time point : (d) WT vs I112R p=0.0002; (e) WT vs I112R p=0.5692 (f-g) Dynamics of wild-type and mutant CFP-(X-KBM) at laser-damaged sites in cells expressing I112R mutant Ku80 (f) or treated with a shAPLF (g) as in b). n=26, 28, and 21 cells for WT, L297E and L297W X-KBM in (f), and n=15 cells for each of WT, L297E and L297W X-KBM in (g). p values at last time point : (f) WT vs L297W p=0.023; WT vs L297E p=<0.0001; (g) WT vs L297W p=0.0144; WT vs L297E p=0.2654.

Figure 4.. Biophysical and cellular analyses of XLF mutants in X-KBM.

(a) SwitchSENSE kinetic analysis of the WT XLF interaction with Ku-DNA complexes. Solid grey lines represent raw data (from 1 to 8 μM; light grey to dark grey; averages of triplicates). Global fitting was performed, following a single-exponential function (solid orange lines) yielding kinetic rate constants; k_ON=4.7 ± 1.7 10⁵ M⁻¹s⁻¹ and k_OFF = 9.1 ± 0.4 10⁻²s⁻¹ for XLF(wt). (b) Dynamics of wild-type and mutant CFP-XLF at laser-damaged sites in BuS cells as in Figure 3b. n=20 cells for WT, L297E and L297W XLF. p values at last time point : WT vs L297W p=0.0093; WT vs L297E p<0.0001. (c) Representative super-resolution images of WT, L297E mutant, and L297W mutant BuS nucleus, with XLF and Ku displayed in green and magenta, respectively (scale = 2500 nm). Right: zoomed-in areas (scale = 250 nm). (d) Representative pair correlation function calculated from the 8×8 μm² center square of one XLF nucleus image of WT (green), L297E (red), and L297W (blue) mutants. WT XLF shows bigger correlation radius (arrow). (e) Statistics of XLF foci size. Each plot represents the average XLF foci size (indicated as radius translated from the correlation radius) in one nucleus. Box’s height displays the s.d. with the mean value labelled in the middle. n=116, 95, 104 nuclei for WT, L297E, and L297W. The two-sample unpaired t-test between WT and L297E is p=10⁻¹³ while that between WT and L297W is p=0.03. (f) Cell survival of BuS cells complemented with vector (EV) or WT or mutated XLF. y axis is log scale. Error bars represent s.d., n=5 to 6 independent experiments. p values were calculated using unpaired two-tailed t-test: WT vs EV p=1.788e-06; WT vs LW p=0.068; WT vs LE p=0.021. (*p< 0.05, **p < 0.01, ***p < 0.001).

Figure 5.. Effects of Ku80 mutations in APLF and XLF binding sites.

(a-b) Dynamics of CFP-(A-KBM) (a) and (X-KBM) (b) at laser damaged sites, in U2OS cells expressing wild-type or I112R/E133M mutant Ku80. n=20 and 9 cells in (a) and n=48 and 11 cells in (b) for WT and mutant Ku80. p values at last time point: (a) WT vs I112R/E133M p=0.001; (b) WT vs I112R/E133M p=0.0111. (c-d) Dynamics of CFP-XRCC4 (c) and XLF (d) in cells expressing wild-type, I112R, E133M or I112R/E133M mutant Ku80. n=38, 27, 28, and 24 cells for WT, E133M, I112R and I112R/E133M Ku80 conditions in (c) and n=24, 26, 20 and 23 cells for I112R, WT, I112R/E133M and E133M Ku80 conditions in (d). p values at last time point: (c) WT vs E133M p=0.532; WT vs I112R p=0.0133; WT vs I112R/E133M p=0.0048; (d) WT vs I112R p=0.246; WT vs E133M p=0.0048; WT vs I112R/E133M p=0.0248. (e) End-joining activity in U2OS cells expressing mutated or WT Ku80. Error bars represent s.d., n=4 independent experiments. p values (unpaired two-tailed t- test): WT vs E133M p=0.0004; WT vs I112R p=0.0052; WT vs I112R/E133M p=0.0002. (f) Distal end-joining in U2OS cells containing mutated or WT Ku80. Error bars represent s.d., n=7 independent experiments. p values (unpaired two-tailed t-test): WT vs E133M p=7.49 e-05; WT vs I112R p=2.21 e-06; WT vs I112R/E133M p=4.05 e-06. (g) Survival of U2OS cells expressing WT or mutated Ku80. y axis is log scale. Error bars represent s.d., n=7 to 10 independent experiments. p (unpaired two-tailed t-test): WT vs I112R p=1.47 e-06; WT vs E133M p=6.32 e-05; WT vs I112R/E133M p=2.52 e-13; I112R vs E133M p=0.011. Significant p-values are indicated as follows: *p< 0.05, **p < 0.01, ***p < 0.001. (h) Model for APLF and XLF KBMs function during NHEJ.