Human BRCA1-BARD1 ubiquitin ligase activity counteracts chromatin barriers to DNA resection

Abstract

The opposing activities of 53BP1 and BRCA1 influence pathway choice in DNA double-strand-break repair. How BRCA1 counteracts the inhibitory effect of 53BP1 on DNA resection and homologous recombination is unknown. Here we identify the site of BRCA1-BARD1 required for priming ubiquitin transfer from E2∼ubiquitin and demonstrate that BRCA1-BARD1's ubiquitin ligase activity is required for repositioning 53BP1 on damaged chromatin. We confirm H2A ubiquitination by BRCA1-BARD1 and show that an H2A-ubiquitin fusion protein promotes DNA resection and repair in BARD1-deficient cells. BRCA1-BARD1's function in homologous recombination requires the chromatin remodeler SMARCAD1. SMARCAD1 binding to H2A-ubiquitin and optimal localization to sites of damage and activity in DNA repair requires its ubiquitin-binding CUE domains. SMARCAD1 is required for 53BP1 repositioning, and the need for SMARCAD1 in olaparib or camptothecin resistance is alleviated by 53BP1 loss. Thus, BRCA1-BARD1 ligase activity and subsequent SMARCAD1-dependent chromatin remodeling are critical regulators of DNA repair.

Figure 1. A basic residue of BARD1 promotes Ub-transfer from BRCA1-E2˜Ub

A. Western blots comparing the ability of BRCA1-BARD1 heterodimer containing Wild type (WT) WT-BARD1 or R99E-BARD1 to catalyse the formation of Ub chains. Ub mix refers to E1, E2, Ub, ATP and ligase reaction buffer. Probed for BRCA1, 6xHis (BARD1) and Ub. B. Yeast two hybrid assays showing the BRCA1-BARD1 heterodimer is not disrupted by BARD1-R99 variants. Yeast strains expressed VP16-BRCA1 1-300 with WT and substituted LexA-BARD1 27-146 (100 mM 3AT). L44R is included as a heterodimer-disruptive control [19]. C. Top, western blots showing improved activity of R99E-BARD1 heterodimer with D32R mutant Ub. Graph below shows quantification of high molecular weight Ub from 4 independent experiments, bars are s.e.m (Note R99E-heterodimer reaction exposed longer than control). * indicates p<0.05, Student’s t-test throughout. D. In vitro assays showing ability of WT and R99E-BARD1 heterodimers to discharge Ub from a loaded E2˜Ub dimer. Results show mean and s.e.m from 4 independent experiments.

Figure 2. Other Type-I RING E3 ligases require R or K residues on the partner protomer.

A. Structural models showing views in the same orientation of Type 1 RING:RING structures: BRCA1-BARD1 (PDB: 1JM7 ref 18) with BMI1-RING1B (PDB: 2CKL ref 40), RAD18 (PDB:2Y43 ref 41), TRIM37 (PDB:3LRQ42), and RBBP6 (PDB:3ZTG ref 43), residues equivalent to R99-BARD1 are shown in pink. B. Western blot indicating the ability to induce mono-ubiquitination of PCNA is lost by R76A-RAD18. Cells transfected with either WT-RAD18 or R76A-RAD18 mutant in otherwise untreated 293 cells or those treated with 40 J UV. Lysates were probed for PCNA and controls as shown. C. Western blots indicating mono-ubiquitination of H2A in nucleosomes by incubation with WT BMI1-RING1B and K73E-BMI1 bearing RING1B heterodimer.

Figure 3. BRCA1-BARD1 ligase activity promotes survival to a subset of DNA damaging agents.

A. Structural model BARD1 (orange) and BRCA1 (green) (PDB: 1JM7) illustrating the location of L44 (blue) and R99 (pink) residues. Zinc ions are filled spheres (black). Lower image is 90° rotation about the horizontal. B. Immunoblot of cells treated with Non-targeting control siRNA (NTC) or BARD1 siRNA and complemented with siRNA-resistant BARD1 variants shown. The graphs show cell survival relative to NTC control of cells depleted and complemented in this way were exposed to the agents shown, plated and clones counted 10-14 days later. Colony numbers are expressed as % of untreated cells. 4 replicates per experiment and means from minimum of 3 experiments shown, bars are s.e.m. C. As in B, depleted with siRNA to BRCA1 and complemented with siRNA resistant WT-BRCA1 and I26A-BRCA1. 4 replicates per experiment, 4 experiments, bars are s.e.m.

Figure 4. BRCA1-BARD1 ligase activity promotes DNA resection in the presence of 53BP1.

A. RPA and RAD51 foci in EdU positive cells treated with BARD1 siRNA and complemented with siRNA resistant WT, R99E- or L44R-BARD1 variants. Images show representative cells. Scale bars are 10 μm throughout. Graphs (right) show the mean number of foci per cell RPA = 60 cells, Rad51 = 100 cells, bars are s.e.m. ***p<0.005. B. The graph shows colony cell survival in Olaparib (10 μM) relative to NTC control of cells treated with BARD1 siRNA and complemented with WT or R99E-BARD1 and treated with additional siRNAs shown. 3, replicates per experiment, 4 experiments, bars are s.e.m. *p<0.05. Western blots (right) show detection of BARD1, CtIP and 53BP1 in treated cells. C. Graph shows colony survival of BRCA1 depleted cells and cells co-depleted with 53BP1 and treated with Hydroxyurea (3 mM) relative to NTC treated controls. 3 replicates per experiment, 3 experiments, bars are s.e.m. * indicates p<0.05. Western blots (right) show detection of 53BP1, BRCA1 in treated cells. D. High-resolution images of BRCA1 and 53BP1 in cells treated with BARD1 siRNA and complemented with WT- or R99E-BARD1, exposed to 2 Gy IR and fixed 8 hours later. An average across 30 profiles is shown, over 3 experimental repeats. Bars =1 standard deviation.

Figure 5. BRCA1-BARD1 ligase activity is required after HR-commitment.

A. BrdU track lengths in cells were treated with BrdU, and Olaparib (10 µM). 50 fibres, horizontal line represents mean, bars= s.e.m. ***p<0.005. B. Graph shows quantification of GFP recovery in NHEJ-substrate cells depleted for BARD1 and transfected with siRNA resistant variants, 3 replicates per experiment, 4 experiments, bars are s.e.m ***= p<0.005. C. Graph shows quantification of GFP recovery in NHEJ-substrate cells depleted for BARD1, CtIP or both 3 replicates per experiment, 6 experiments, bars are s.e.m ***= p<0.005. D. Graph shows quantification of GFP recovery in HR-substrate cells depleted for BARD1, CtIP or both. 3 replicates per experiment, 6 experiments, bars are s.e.m ***= p<0.005. E. Graph shows colony survival of cells treated with IR (0.5 Gy) and depleted for BARD1 and complemented with WT or R99E-BARD1 or co-depleted for BARD1 and CtIP relative to control treated cells. 3 replicates per experiment, 3 experiments, bars are s.e.m ***= p<0.005.

Figure 6. BRCA1-BARD1 dependent modification of nucleosomes.

A. Western blot of selected proteins in whole cell extracts (WCE) and Nickel-column precipitations (Ni²⁺ IP) of lysates from cells transfected with His-Myc-Ub and BARD1 constructs and irradiated (30 Gy). B. Western blot of HA-H2A of cells transfected with HA-H2A or HA-H2A-Ub fusion. In the fusion Histone 2A carried lysine to arginine mutations at lysines K13 15 118 119 125 127 and 129 and in the Ub the seven lysines of Ub were mutated to R (illustrated top). Cells were lysed sequentially in buffer containing increasing amounts of salt. Sol = soluble fraction Nuc= nuclear fraction and Pel= Pellet. C. Quantification of HA-H2A-Ub intensity (expression) and number of RAD51 foci in BARD1 depleted S-phase cells after exposure to IR (5 Gy) and 2 hours recovery (for representative immunofluorescence images see Supplemental Fig 6E) n=30 cells. Western blots (right) shows BARD1 and HA-H2A-Ub expression in treated cells. D. Graph below shows quantification of RAD51 foci in BARD1 depleted cells expressing HA-H2A or various fusions of H2A and pulsed with EdU fixed 2 hours after 5 Gy IR. n=75 cells per condition from 3 experiments, bars= s.e.m. Images top show cells stained with RAD51, HA and incubated with click-it detection reagents to detect EdU labelled S-phase cells.

Figure 7. The nucleosome remodeler SMARCAD1 is in the same pathway as BRCA1-BARD1 ligase activity.

A. Illustration of SMARCAD1 protein. B. Western blot showing the detection of HA-H2A (WT) or H2A-Ub expressed in cells and bound, or not bound, to in vitro purified WT or mutant SMARCAD1 CUE domains (amino acids 98-318). C. Western blot showing expression of full length WT and CUE-domain mutant SMARCAD1 in SMARCAD1 depleted cells. D. Representative images (left) of cells expressing siRNA resistant SMARCAD1 variants treated with SMARCAD1 siRNA or co-depleted with siRNA to BARD1 and bearing laser-line induced DNA damage. Cells labelled ATMi = 10 μM KU55933 ATM inhibitor 4 hours prior to damage. White ‘X’ marks the laser-line path. The graph shows intensity of myc-SMARCAD1 measured in the region of the γH2AX laser line compared to nucleoplasm intensity. 25 lines per experiment, 2 experiments, bars S.E. **= p<0.05, ***= p<0.005 Student t-test. Scale bars 10 µm E. Length of BrdU DNA fibres after treatment with Olaparib (10 µM) in cells treated with the siRNAs shown. Western blots (right) shows detection of BARD1 and SMARCAD1 proteins in treated cells. F. Representative images (left) of cells transfected with siRNA targeting SMARCAD1 with siRNA resistant WT and CUEm myc-SMARCAD1 exposed to 2 Gy IR and fixed 8 hours later. Graphs (right) show averages of 30 foci profiles over 3 experimental repeats. Bars =1 standard deviation. G. Quantification RAD51 foci in cells transfected with SMARCAD1 and 53BP1 siRNA together with siRNA resistant forms of SMARCAD1, n=70 cells, bars= s.e.m. ***p<0.005). H. The graph shows colony cell survival in Olaparib (10 μM) and Camptothecin (2.5 μM) relative to NTC control of SMARCAD1 complemented cells, 3 replicates per experiment, 6 experiments, bars are s.e.m. p<0.005. Western blots (right) show detection of SMARCAD1 and 53BP1 in treated cells.

Figure 8. Proposed model for the BRCA1-BARD1 Ub ligase in promoting resection at DSB-damaged chromatin.

1. Limited resection occurs in the absence of BRCA1-BARD1 activity dependent on CtIP-Mre11. 2. BRCA1-BARD1 dependent Ub modification of H2A promotes SMARCAD1 interaction with damage-proximal nucleosomes. 3. SMARCAD1 activity to reposition or evicts nucleosome, and move 53BP1 and its effector proteins to release 53BP1-mediated inhibition of DNA resection. 4. Long range resection can proceed.