BRCA1 Haploinsufficiency Is Masked by RNF168-Mediated Chromatin Ubiquitylation

Abstract

BRCA1 functions at two distinct steps during homologous recombination (HR). Initially, it promotes DNA end resection, and subsequently it recruits the PALB2 and BRCA2 mediator complex, which stabilizes RAD51-DNA nucleoprotein filaments. Loss of 53BP1 rescues the HR defect in BRCA1-deficient cells by increasing resection, suggesting that BRCA1's downstream role in RAD51 loading is dispensable when 53BP1 is absent. Here we show that the E3 ubiquitin ligase RNF168, in addition to its canonical role in inhibiting end resection, acts in a redundant manner with BRCA1 to load PALB2 onto damaged DNA. Loss of RNF168 negates the synthetic rescue of BRCA1 deficiency by 53BP1 deletion, and it predisposes BRCA1 heterozygous mice to cancer. BRCA1^+/-RNF168^-/- cells lack RAD51 foci and are hypersensitive to PARP inhibitor, whereas forced targeting of PALB2 to DNA breaks in mutant cells circumvents BRCA1 haploinsufficiency. Inhibiting the chromatin ubiquitin pathway may, therefore, be a synthetic lethality strategy for BRCA1-deficient cancers.

Keywords: BRCA1; PALB2; RAD51; RNF168; cancer; chromatin ubiquitylation; genome stability; haploinsufficiency; homologous recombination; replication fork protection; resection.

Figure 1.. RNF168 sustains organismal viability and genome maintenance when BRCA1 is inactivated.

(A) Model of the γ-H2AX-RNF8-RNF168 chromatin ubiquitylation pathway and downstream effectors. (B) Breeding strategy to generate mice with combined deficiencies in BRCA1 and the DNA damage response (DDR) factors H2AX, RNF8, RNF168 or 53BP1. (C-E) Summary of breeding outcomes from the BRCA1^+/Δ1153BP1^+/− X BRCA1^+/Δ1153BP1^−/− intercross (C), the BRCA1^+/Δ11RNF168^+/− X BRCA1^+/Δ11RNF168^+/− intercross (D), and three intercrosses: BRCA1^+/Δ11RNF168^+/−53BP1^+/− X BRCA1^+/Δ11RNF168^+/−53BP1^+/−; BRCA1^+/Δ11RNF168^+/−53BP1^+/− X BRCA1^+/Δ11RNF168^+/−53BP1^−/−; BRCA1^+/Δ11RNF168^+/−53BP1^−/− X BRCA1^+/Δ11RNF168^+/−53BP1^−/− (E). (F) The average number of chromosomal radials per metaphase spread in WT, BRCA1^{FΔ11/FΔ11; CD19Cre} RNF168^−/− and BRCA1^{FΔ11/FΔ11; CD19Cre} RNF168^−/− B cells exposed to PARPi. (G) The percentage of EdU-positive (S-phase) WT, BRCA1^{FΔ11/FΔ11; CD19Cre}, RNF168^−/− and BRCA1^{FΔ11/FΔ11; CD19Cre} RNF168^−/− B cells that stained positive for RAD51 foci 4 hours post γ-irradiation (5 Gy). Data in F and G are presented as mean ± SD. In C-E and F-G, statistical significance was calculated using χ² test for goodness of fit and unpaired two-tailed Student’s t-test, respectively. See also Figures S1 and S2.

Figure 2.. RNF168 supports BRCA1-independent survival in human cells.

(A) The auxin-induced BRCA1 degradation system in human TK6 cells. (B) The growth profile of BRCA1^AID/AID, BRCA1^AID/AID53BP1^−/− and BRCA1^AID/AIDRNF168^−/− TK6 cells in the absence and presence of 0.5 mM auxin. BRCA1 degradation induced by addition of auxin resulted in severe growth inhibition in both BRCA1^AID/AID and two independent clones of BRCA1^AID/AIDRNF168^−/− TK6 cells (p<0.0001 compared to no auxin). Loss of 53BP1 rescued the growth defect in BRCA1-depleted cells. (C) RAD51 foci formation in BRCA1^AID/AID, BRCA1^AID/AID53BP1^−/− and BRCA1^AID/AIDRNF168^−/− TK6 cells 2 hours post γ-irradiation (2 Gy); cells were pre-treated or not with 0.5 mM auxin. (D) The average number of RAD51 foci per cell among irradiated Cyclin A-positive (S/G2) TK6 cells. (E) Outline of the Multicolor Competition Assay (MCA). (F) MCA in Cas9⁺BRCA1^-/−53BP1^−/− human hTERT-RPE1 cells transduced with a specific guide RNA targeting RNF168 or an empty vector (sgCTL). RPE1 cells transduced non-targeting guides (sgLacZ) were used as the competitor. Deletion of RNF168 significantly attenuated the growth of BRCA1^−/− 53BP1^−/− cells following PARPi treatment (p<0.0001). (G) Efficient knockdown of RNF168 in BRCA1^-/−53BP1^−/− hTERT-RPE1 cells by CRISPR-Cas9. A representative blot is shown. (H) RAD51 foci formation in BRCA1^-/−53BP1^−/− hTERT-RPE1 cells transduced with either sgCTL or sgRNF168 4 hours post γ-irradiation (5 Gy). (I) The percentage of BRCA1^-/−53BP1^−/− hTERT-RPE1 cells stained positive for RAD51 foci 4 hours post γ-irradiation (5 Gy). Data in B, D, F and I are presented as mean ± SD. In B/F, D and I, statistical significance was calculated using two-way ANOVA, Mann-Whitney test and two-tailed Student’s t-test, respectively. See also Figure S3.

Figure 3.. Loss of RNF168 unmasks BRCA1 haploinsufficiency.

(A) BRCA1 protein expression level (full-length and delta-11 isoforms) in BRCA1 heterozygous cells (BRCA1+/Δ11 and BRCA1+/FΔ11; CD19Cre); WT and BRCA1FΔ11/FΔ11; CD19Cre cells were used as controls. (B) Summary of breeding outcomes from the BRCA1^+/Δ11RNF168^+/− X BRCA1^+/Δ11RNF168^+/− intercross. (C) Representative morphology of E16.5 WT and BRCA1^+/Δ11RNF168^−/− embryos; the latter exhibited growth retardation as well as exencephaly. (D) Staining of E16.5 embryos for senescence-associated β-galactosidase activity. (E) Kaplan-Meier survival analysis of WT (n=8), BRCA1^+/Δ11 (n=16), RNF168^−/− (n=13) and BRCA1^+/Δ11RNF168^−/− (n=19) mice. Significantly shorter lifespan was observed in BRCA1^+/Δ11RNF168^−/− mice compared to the RNF168^−/− counterparts (p<0.0001). (F) Kaplan-Meier survival analysis of p53^−/− (n=11), BRCA1^+/Δ11p53^−/− (n=22), RNF168^−/−p53^−/− (n=4) and BRCA1^+/Δ11RNF168^−/−p53^−/− (n=3) mice. Significantly shorter tumor-free survival was observed in BRCA1^+/Δ11RNF168^−/−p53^−/− mice compared to BRCA1^+/Δ11p53^−/− and RNF168^−/−p53^−/− counterparts (p<0.0001 and p=0.01, respectively). (G) Kaplan-Meier survival analysis of p53^+/− (n=10), BRCA1^+/Δ11p53^+/− (n=11), RNF168^−/−p53^+/− (n=10) and BRCA1^+/Δ11RNF168^−/−p53^+/− (n=8) mice. Significantly shorter tumor-free survival was observed in BRCA1^+/Δ11RNF168^−/−p53^+/− mice compared to BRCA1^+/Δ11p53^−/− and RNF168^−/−p53^−/− counterparts (p<0.0001 and p=0.003, respectively). (H) Growth of WT, BRCA1^+/Δ11, RNF168^−/− and BRCA1^+/Δ11RNF168^−/− primary mouse embryonic fibroblasts (MEFs) in culture. BRCA1^+/Δ11RNF168^−/− cells grew significantly slower than BRCA1^+/Δ11 and RNF168^−/− counterparts (p=0.02, Kruskal-Wallis test). (I) The average number of chromosomal radials per metaphase spread in WT, BRCA1^+/Δ11, RNF168^−/− and BRCA1^+/Δ11RNF168^−/− MEFs exposed to PARPi. (J) The average number of chromosomal radials per metaphase spread in PARPi-treated BRCA1^+/Δ11RNF168^−/− MEFs stably expressing WT or catalytic mutant (R57D) forms of RNF168. BRCA1^+/Δ11RNF168^−/− MEFs transduced with empty vector (EV) were used as control. Data in G-I are presented as mean ± SD. In B and H, statistical significance was calculated using χ² test for goodness of fit and one-way ANOVA, respectively. In E-G and I-J, statistical significance was calculated using Mantel-Cox test and unpaired two-tailed Student’s t-test, respectively. See also Figure S4 and S5.

Figure 4.. RNF168-mediated PALB2 recruitment is essential for viability and genome maintenance when the BRCA1-PALB2 pathway is compromised.

(A) The average fluorescence intensity of PALB2 stripes in WT, BRCA1^+/Δ11, RNF168^−/− and BRCA1^+/Δ11RNF168^−/− MEFs stably expressing GFP-PALB2. Signals were normalized to the background noise. (B) The percentage of EdU-positive (S-phase) WT, BRCA1^{+/FΔ11; CD19Cre}, RNF168^−/− and BRCA1^{+/FΔ11; CD19Cre} RNF168^−/− B cells that stained positive for RAD51 foci 4 hours post γ-irradiation (5 Gy). (C) Breeding strategy for the generation of mice lacking RNF168 in the context of an abrogated BRCA1-PALB2 interaction (PALB2^CC6). (D) Summary of breeding outcomes from the PALB2^+/CC6RNF168^+/− X PALB2^+/CC6RNF168^+/− intercross. (E) Strategy for forced targeting of PALB2 to DSB sites. (F) Formation of PALB2 and RAD51 foci in BRCA1^+/Δ11RNF168^−/− MEFs stably expressing GFP-PALB2^FHA. (G) The percentage of EdU-positive (S-phase) BRCA1^+/Δ11RNF168^−/− MEFs stably expressing GFP-PALB2^FHA that stained positive for PALB2 and RAD51 foci 4 hours post γ-irradiation (10 Gy). WT MEFs and BRCA1^+/Δ11RNF168^−/− MEFs transduced with empty vector (EV) were used as controls. (H) The average number of chromosomal radials per metaphase spread in PARPi-treated BRCA1^+/Δ11RNF168^−/− MEFs and BRCA1^{+/FΔ11; CD19Cre} RNF168^−/− B cells stably expressing WT PALB2 or PALB2^FHA. (I) Colony formation capacity of BRCA1^+/Δ11RNF168^−/− MEFs stably expressing WT PALB2 or PALB2^FHA after treatment with PARPi. PALB2^FHA expression significantly rescued PARPi hypersensitivity in BRCA1^+/Δ11RNF168^−/− MEFs (p<0.0001). In H and I, MEFs and B cells transduced with empty vector (EV) were used as controls. Data in A, B and G-I are presented as mean ± SD. In A and D, statistical significance was calculated using Mann-Whitney test and χ² test for goodness of fit, respectively. In G-H and I, statistical significance was calculated using unpaired two-tailed Student’s t-test and two-way ANOVA, respectively. See also Figure S6.

Figure 5.. RNF168 function is dispensable in a BRCA1 mutant that retain interaction with PALB2.

(A) Co-immunoprecipitation of BRCA1-interacting proteins in BRCA1-null human MDA-MB436 cells stably expressing full-length (FL), ΔRING and Δ11q isoforms of BRCA1. Cells expressing empty vector (mCherry) were used as control. (B) Breeding strategy for the generation of mice lacking RNF168 in the context of homozygous BRCA1^Δ2/Δ2 mutation. (C) Summary of breeding outcomes from two intercrosses: BRCA1^+/Δ2RNF168^+/− X BRCA1^+/Δ2RNF168^+/− and BRCA1^+/Δ2RNF168^+/− X BRCA1^+/Δ2RNF168^-/−. (D) Kaplan-Meier survival analysis of WT (n=6), RNF168^−/− (n=9) and BRCA1 ^Δ2/Δ2RNF168^−/− (n=6) mice. Overall survival was comparable between BRCA1 ^Δ2/Δ2RNF168^−/− and RNF168^−/− mice (p=0.31). (E) The average number of chromosomal radials per metaphase spread in WT, BRCA1^{FΔ2/FΔ2; CD19Cre}, RNF168^−/− and BRCA1^Δ2/Δ2 RNF168^−/− B cells exposed to PARPi. (F) RAD51 and BARD1 foci formation in WT (BRCA1^FΔ2/FΔ2 no Cre), BRCA1^Δ2/Δ2 (BRCA1^FΔ2/FΔ2+Ad-Cre), RNF168^−/− and BRCA1^Δ2/Δ2 RNF168^−/− MEFs 4 hours post γ-irradiation (5 Gy). Note that a small fraction (<10%) of BRCA1^FΔ2/FΔ2+AdCre MEFs retain robust BARD1 foci formation under these conditions. The majority of such cells also stain positive for RAD51 foci. (G). The percentage of EdU-positive (S-phase) WT (BRCA1^FΔ2/FΔ2 no Cre), BRCA1^Δ2/Δ2 (BRCA1^FΔ2/FΔ2+Ad-Cre), RNF168^−/− and BRCA1^Δ2/Δ2 RNF168^−/− MEFs that stained positive for RAD51 (left panel) or BARD1 (right panel) foci 4 hours post γ-irradiation (5 Gy). For BRCA1^FΔ2/FΔ2+AdCre MEFs, RAD51 foci formation was assessed in BARD1-negative cells. Data in E and G are presented as mean ± SD. In C, D and E/G, statistical significance was calculated using χ² test for goodness of fit, Mantel-Cox test and unpaired two-tailed Student’s t-test, respectively.

Figure 6.. RNF168 does not cooperate with BRCA1 in the protection of stalled replication forks.

(A) Ratio of IdU vs. CldU incorporation in WT, BRCA1^{+/FΔ11; CD19Cre}, RNF168^−/− and BRCA1^{+/FΔ11; CD19Cre} RNF168^−/− B cells following hydroxyurea (HU) treatment. BRCA2^{F/F; CD19Cre} B cells were used as a positive control for HU-induced nucleolytic degradation of nascently replicated DNA. Schematic for labeling B cells with CldU and IdU is shown at the top. (B) Ratio of IdU vs. CldU incorporation in WT (BRCA1^FΔ2/FΔ2 no Cre), BRCA1^Δ2/Δ2 (BRCA1^FΔ2/FΔ2+Ad-Cre), RNF168^−/− and BRCA1^Δ2/Δ2 RNF168^−/− MEFs following HU treatment. Schematic for labeling MEFs with CldU and IdU is shown at the top. Data shown in A and B are compiled from two independent experiments. Statistical significance was calculated using the Mann-Whitney test. (C) A working model depicting how RNF168 cooperates with BRCA1 during HR. RNF168 regulates HR at two distinct steps. First, RNF168 recruits 53BP1 to limit end resection. Once nucleolytic processing of the break is underway, RNF168 additionally recruits PALB2 to the ssDNA compartment or chromatin flanking the break site. In BRCA1-proficient cells, loading of RAD51 is likely to be primarily carried out by the BRCA1/PALB2/BRCA2 complex that accumulates on the processed ssDNA, while RNF168/PALB2 may also assist in RAD51 assembly. As a result, loss of RNF168 in BRCA1-proficient cells produces only relatively subtle HR defects. However, if the canonical BRCA1/PALB2/BRCA2 pathway is absent or limiting in its functionality, RNF168-mediated PALB2 recruitment to ssDNA or chromatin provides an essential alternative route for RAD51 loading. Abrogation of RNF168 activity in BRCA1 compromised cells results in dramatically elevated genome instability, which may promote tumorigenesis.