Polyubiquitinated PCNA triggers SLX4-mediated break-induced replication in alternative lengthening of telomeres (ALT) cancer cells

Abstract

Replication stresses are the major source of break-induced replication (BIR). Here, we show that in alternative lengthening of telomeres (ALT) cells, replication stress-induced polyubiquitinated proliferating cell nuclear antigen (PCNA) (polyUb-PCNA) triggers BIR at telomeres and the common fragile site (CFS). Consistently, depleting RAD18, a PCNA ubiquitinating enzyme, reduces the occurrence of ALT-associated promyelocytic leukemia (PML) bodies (APBs) and mitotic DNA synthesis at telomeres and CFS, both of which are mediated by BIR. In contrast, inhibiting ubiquitin-specific protease 1 (USP1), an Ub-PCNA deubiquitinating enzyme, results in an increase in the above phenotypes in a RAD18- and UBE2N (the PCNA polyubiquitinating enzyme)-dependent manner. Furthermore, deficiency of ATAD5, which facilitates USP1 activity and unloads PCNAs, augments recombination-associated phenotypes. Mechanistically, telomeric polyUb-PCNA accumulates SLX4, a nuclease scaffold, at telomeres through its ubiquitin-binding domain and increases telomere damage. Consistently, APB increase induced by Ub-PCNA depends on SLX4 and structure-specific endonucleases. Taken together, our results identified the polyUb-PCNA-SLX4 axis as a trigger for directing BIR.

Graphical Abstract

Figure 1.

Ub-PCNA increases APBs with DNA synthesis in the G2 phase of ALT + cancer cells. (A) A diagram of PCNA ubiquitination and de-ubiquitination. The triangle on the DNA represents a DNA lesion. (B–E) U2OS-ATAD5^AID cells were treated with auxin for 48 h and fixed at the G2 phase for immunostaining with telomere FISH. (B,D) Representative images of localization of PCNA (B) or Ub-PCNA (D) at APBs. (C,E) The percent localization of PCNA (C) or Ub-PCNA (E) at APBs per G2 phase cell was quantified. (F–H) U2OS-ATAD5^AID cells were treated with auxin for 48 h and fixed at the G2 phase for an ATSA assay. (I,J) After siRNA transfection, G2-synchronized U2OS cells were subjected to an ATSA assay. (G–J) Quantification ATSA results were displayed. (B,D,F) The dotted line represents the edge of the nucleus. Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: two-tailed unpaired Student’s t-test (C, E, G, H); one-way ANOVA (I,J). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 2.

Ub-PCNA increases BIR-like MiDAS events in ALT + cancer cells. (A) Representative images of MiDAS after APH treatment. (A–Q) After transfection, cells were treated with DMSO or 0.4 mM aphidicolin (APH) for 16 h and subjected to a MiDAS assay. (B–Q) The number of MiDAS at telomeres or CFS regions per metaphase cell was quantified. (H) RAD52 expression was induced by doxycycline (DOX) for 48 h. Cell lines: U2OS (B,C,F–H,K–O), GM0637 (D), SK-LU-1 (E), RAD52^−/− U2OS (H) and HeLa_LT (I,J,P,Q). Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: One-way ANOVA (B–Q). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 3.

Ub-PCNA increases break-induced telomere synthesis. (A) A diagram of the TRF1-FokI system. (B–K) After siRNA transfection, U2OS-mCherry-TRF1-FokI cells were enriched in the G2 phase and treated with tamoxifen and Shield1 for 2 h (B–F,G,I) or indicated times (H,J,K) before fixation for immunostaining. (B) Representative immunostaining images of Ub-PCNA, EdU and TRF1-FokI co-localization after telomeric break induction. (C,G,I) Representative images showing localization of Ub-PCNA (C,I) or PCNA (G) at TRF1-FokI foci. Scale bar 10 μm. (D–F,H,J,K) The percent localization of Ub-PCNA (D,J), EdU (E,F,K) or PCNA (H) at TRF1-FokI foci was quantified. (B,C,G,I) The dotted line represents the edge of the nucleus. Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: One-way ANOVA (D–F); two-tailed unpaired Student’s t-test (H,J,K). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 4.

ATAD5 deficiency extends telomere length in ALT + cancer cells and promotes recombination-associated ALT phenotypes. (A,B) U2OS-ATAD5^AID or HeLa-ATAD5^AID cells were treated with auxin for the indicated days and subjected to a Flow-FISH assay. (C) ATAD5^+/+ and ATAD5^−/− clones of each cell line were subjected to a Flow-FISH assay. (A–C) Normalized telomere length was quantified. (D) ATAD5^+/+ and ATAD5^−/− U2OS and HeLa-LT cells were subjected to a terminal restriction fragment assay to measure actual telomere length. (E–G) U2OS-ATAD5^AID cells were treated with auxin and subjected to an SMAT assay. (E) Representative images of an SMAT assay. After the continuous long ssDNA fibers that were aligned with the telomere and CldU signal, and the terminal CldU signal was defined as telomere extension. (F,G) The total length of telomeres with the extension and extended telomere length (F), and the percentage of telomere extension events (G) were quantified from at least 350 fibers per each replicate. Data represent the mean ± s.d. of three independent experiments. (H–J,L–O) After transfection, cells were subjected to a CO-FISH to measure T-SCE (H–J, N) and C-circle assay (L,M,O). (H) Representative images of CO-FISH. A circle indicates a T-SCE event. (I,J,N) The percentage of T-SCE per metaphase was quantified. (K) U2OS-ATAD5^AID or HeLa-ATAD5^AID cells were treated with auxin for 48 h and subjected to a C-circle assay. (K,L,M,O) The intensity of the C-circle level was quantified. (N,O) U2OS cells were transfected with a combination of ATAD5 siRNA and wild type (WT), UAF1 interaction-defective (ΔUAF1) or PCNA unloading-defective (E1173K) ATAD5 cDNA. Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: two-tailed unpaired Student’s t-test (A–C, F, G, J–M); One-way ANOVA (O). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 5.

Ub-PCNA triggers telomere damage leading to APB formation and subsequent BIR in the G2 phase of ALT + cancer cells (A–C) U2OS-ATAD5^AID cells were synchronized in the G2 phase, treated with auxin and/or USP1i for 6 h, and chromatin-bound protein extracts were prepared for immunoblotting (A) or cells were subjected to an ATSA assay (B,C). (D,E) After transfection, U2OS cells were synchronized in the G2 phase, treated with USP1i for 6 h, and subjected to an ATSA assay. (F–J) RAD52 doxycycline-inducible RAD52^−/− U2OS cells were synchronized in the G2 phase, treated with USP1i for 6 h, and chromatin-bound protein extracts were prepared for immunoblotting (F) or cells were fixed for gH2AX immunostaining to measure TIFs (G,H) or an ATSA assay (I,J). RAD52 expression was induced by doxycycline (DOX) treatment for 48 h before cell harvest or fixation. (G,I) Representative images. The dotted line represents the edge of the nucleus. Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: One-way ANOVA (B-E, H, J). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 6.

PolyUb-PCNA increases SLX4 recruitment to telomeres in ALT + cancer cells. (A,B) After transfection, U2OS cells were synchronized in G2 phase and fixed for SLX4 immunostaining with telomere FISH. (A) Representative images showing SLX4 localization at telomeres. The dotted line represents the edge of the nucleus. (B–F,H) The number of SLX4 localization at telomeres in the G2 phase was quantified. (C) U2OS-ATAD5^AID synchronized in the G2 phase were treated with auxin and USP1i for 6 h and fixed. (D) HeLa_LT or U2OS cells synchronized in the G2 phase were treated with USP1i for 6 h and fixed. (E) G2-synchronized U2OS-ATAD5^AID cells were treated with auxin, USP1i, and UBE2Ni for 6 h, and cells were fixed. (F,G) After siRNA transfection, U2OS cells were treated with PDS for 24 h and UBE2Ni for 6 h, respectively, before fixation. (F) Representative images. (H,I) U2OS cells expressing GFP-tagged full-length SLX4 (FL), N-terminal fragment of SLX4 (N), SLX4^N with a mutation in UBZ4-1 (UBZ4-1m), UBZ4-2 (UBZ4-2m) or both UBZ4 domains (UBZ4-1/2m), respectively, were synchronized in the G2 phase, treated with USP1i for 6 h and fixed. (H) Representative images showing GFP-SLX4 localization at telomeres. The dotted line represents the edge of the nucleus. Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: One-way ANOVA (B,C,E,G,I); two-tailed unpaired Student’s t-test (D). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 7.

PolyUb-PCNA induces APB formation and MiDAS at telomeres and CFS. (A,B) G2-synchronized (G2) U2OS-ATAD5^AID cells were treated with auxin, USP1i and UBE2Ni for 6 h, and cells were fixed for an ATSA assay. (C–H) After transfection, U2OS cells were subjected to a MiDAS assay (C,D), immunoblotting with chromatin-bound protein extracts (E), or immunostaining (F–H). (C,D,F–H) U2OS cells were treated with 0.4 mM aphidicolin (APH) for 16 h, and 10 mM UBE2Ni for 17 h before fixation as indicated. (C,D) The number of MiDAS at telomeres (C) or CFSs (D) was quantified. (F) Representative images showing SLX4 and FANCD2 localization in metaphase cells. (G,H) The number of SLX4 and FANCD2 co-localization (G) and FANCD2 foci (H) per metaphase cell was quantified. (I,J) After transfection, U2OS cells synchronized in the G2 phase were fixed for an ATSA assay. Data in all graphs represent the mean ± s.d. of at least three independent experiments. Statistical analysis: One-way ANOVA (A–D, G–J). ****P<0.0001, ***P<0.001, **P<0.01, *P<0.05 and ns.: not significant.

Figure 8.

Graphical model of polyUb-PCNA-induced generation of telomere breakage and Ub-PCNA-induced promotion of the BIR processes in ALT + cancer cells. [A, (i,ii)] Telomeres in ALT + cells suffer more frequent replication stress than those in TEL + cells. MonoUb- and polyUb-PCNA, formed under telomeric replication stress, remain on chromatin at a significant level until the G2 phase due to an unknown mechanism that restricts unloading and deubiquitination of Ub-PCNA. Although it is not clear whether monoUb- and polyUb-PCNA in the G2 phase have completed their roles in TLS- and template switch-mediated damage bypass, respectively, the K63-linked-polyubiquitin chain of PCNA additionally acts as a binding site for SLX4 via the UBZ4 domain and consequently acts as a working platform for the SLX4-associated SMX tri-nuclease complex. All three endonucleases comprising the SMX tri-nuclease complex (SLX1, MUS81, and XPF) cooperatively generate telomere breakage. [A, (iii)] Broken telomeric DNA ends induce APB formation with a help of BLM helicase and PML, and then RAD52- and POLD3-dependent BIR is initiated in APBs. [A, (ii)] Telomeric breakage by polyUb-PCNA-SLX4 is inhibited by balanced deubiquitination and/or unloading of polyUb-PCNA by the USP1-UAF1-ATAD5-RLC. [B, (i,ii)] When this counterbalance is disrupted by acute ATAD5 depletion or USP1 inhibition, even in the G2 phase, the excess polyUb-PCNA increases the abundance of the SLX4 and SMX tri-nuclease complex at telomeres, which leads to sequential increases in telomeric breakage, APB formation, ALT phenotypes, and telomere length. [A, (iv)] During the BIR process, PCNA-polymerase δ encounters replication stress, and, consequently, PCNA is ubiquitinated. Although whether monoUb-PCNA or polyUb-PCNA is present and the role of Ub-PCNA during BIR is unclear, our results suggest that Ub-PCNA facilitates the BIR process. [B, (iii)] Consistently, Ub-PCNA left unprocessed due to depletion of ATAD5 or USP1 increased TRF1-FokI-induced telomeric DNA synthesis and telomeric MiDAS.