MRN-CtIP, EXO1, and DNA2-WRN/BLM act bidirectionally to process DNA gaps in PARPi-treated cells without strand cleavage

Abstract

Single-stranded DNA (ssDNA) gaps impact genome stability and PARP inhibitor (PARPi) sensitivity, especially in BRCA1/2-deficient tumors. Using single-molecule DNA fiber analysis, electron microscopy, and biochemical methods, we found that MRN, CtIP, EXO1, and DNA2-WRN/BLM resect ssDNA gaps through a mechanism different from their actions at DNA ends. MRN resects ssDNA gaps in the 3'-to-5' direction using its pCtIP-stimulated exonuclease activity. Unlike at DNA ends, MRN does not use its endonucleolytic activity to cleave the 5'-terminated strand flanking the gap or the ssDNA. EXO1 and DNA2-WRN/BLM specifically resect the 5' end of the gap independent of MRN-CtIP. This resection process alters ssDNA gap repair kinetics in BRCA1-proficient and -deficient cells. In BRCA1-deficient cells treated with PARPis, excessive resection results in larger ssDNA gaps, hindering their repair and leading to DNA breaks in subsequent cell cycle stages due to ssDNA gaps colliding with DNA replication forks. These findings broaden our understanding of the role of human nucleases in DNA metabolism and have significant implications for defining the mechanisms driving PARPi sensitivity in BRCA-deficient tumors.

Keywords: BRCA; DNA replication; DNA replication stress; MRE11; PARP inhibitor; genome stability; nucleases; single-stranded DNA gaps.

Figure 1.

ssDNA gaps cannot be repaired in BRCA1-deficient cells treated with PARPis and are larger than in BRCA1-proficient cells (see also Supplemental Fig. S1). (A, left) Schematic of the DNA fiber spreading assay with the S1 nuclease. (Right) Representative images of DNA fibers in SUM149PT cells treated with 10 μM olaparib ± S1. Scale bar, 10 μm. (B) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells ±10 μM olaparib (1 h) and ±S1. The S1 nuclease was added immediately after (0 min [T0]) and 30 min (T30) and 60 min (T60) after olaparib removal (n = 3). At least 180 tracts were scored for each sample. Statistics: Kruskal–Wallis followed by Dunn's multiple comparisons test. (ns) Nonsignificant, (****) P < 0.0001. (C, left) Representative electron micrograph of a replication fork containing an internal daughter strand ssDNA gap. (Middle) Magnified image of the fork junction. Red arrows indicate the daughter strand ssDNA gap. (P) Parental strand, (D) daughter strand. (Right) Schematic of the electron microscopy experiments of D and E and of a replication fork containing an internal gap. (D) Percentage of replication forks with daughter strand gaps in SUM149PT + BRCA1 and SUM149PT cells ±10 μM olaparib for 1 h. Cells were collected immediately after (T0) and 30 min (T30) after PARPi removal. “# RI” indicates the number of analyzed replication intermediates. (n = 3). Mean values are shown above each data set. Columns indicate mean ± SD. Statistics: unpaired t-test. (ns) Nonsignificant, (*) P < 0.0332, (**) P < 0.0021. (E) Length of daughter strand ssDNA gaps in nucleotides in SUM149PT and SUM149PT + BRCA1 cells treated as in D. “# RI” indicates the number of analyzed replication intermediates. (n = 3). Statistics: unpaired t-test with Welch correction. (ns) Nonsignificant, (****) P < 0.0001. Horizontal bars indicate median. Median values are shown above each data set.

Figure 2.

MRE11 inhibition rescues ssDNA gap repair without affecting ssDNA gap formation (see also Supplemental Fig. S2). (A, top) Schematic of the DNA fiber spreading assay with the S1 nuclease in the presence and absence of Mirin. (Bottom) Representative images of DNA fibers in SUM149PT treated with S1 ± 10 μM olaparib for 1 h and ±50 μM Mirin. Scale bar, 10 μm. (B) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib and ±50 μM Mirin for 1 h. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). At least 180 tracts were scored for each sample. Statistics: Kruskal–Wallis followed by Dunn's multiple comparisons test. (ns) Nonsignificant, (*) P < 0.0332, (**) P < 0.0021, (***) P < 0.0002, (****) P < 0.0001. (C, left) Schematic of electron microscopy experiment in the presence and absence of Mirin and REV1i (JH-RE-06). (Right) Percentage of replication forks with daughter strand gaps in SUM149PT + BRCA1 and SUM149PT cells treated with 10 μM olaparib ± 50 μM Mirin and ±2 μM REV1i (JH-RE-06) for 1 h. Cells were collected immediately after PARPi removal (T0). The first and fifth columns are repeated data from Figure 1D used for easier comparison between samples in different figures. “# RI” indicates the number of analyzed replication intermediates. (n = 3). Columns indicate mean ± SD. Mean values are shown above each data set. Statistics: unpaired t-test. (ns) Nonsignificant, (*) P < 0.0332, (**) P < 0.0021, (***) P < 0.0002. (D) Length of daughter strand ssDNA gaps in nucleotides in SUM149PT and SUM149PT + BRCA1 cells treated as in C. The first and fifth columns are repeated data from Figure 1D used for easier comparison between samples in different figures. “# RI” indicates the number of analyzed replication intermediates. (n = 3). Statistics: unpaired t-test with Welch correction. (ns) Nonsignificant, (****) P < 0.0001. Horizontal bars indicate median. Median values are shown above each data set. (E, top) Schematic of the DNA fiber assay performed by using the spreading technique with the S1 nuclease in the presence and absence of PRIMPOL. (Bottom) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib and ±siPRIMPOL. The S1 nuclease was added immediately after olaparib removal (time 0) (n = 3). (F, top) Schematic of the DNA fiber assay performed by using the combing technique with the S1 nuclease in the presence and absence of PRIMPOL. (Bottom) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib and ±siPRIMPOL. The S1 nuclease was added immediately after olaparib removal (time 0) (n = 3). At least 130 tracts were scored for each sample in E and F. Statistics in E and F: Kruskal–Wallis followed by Dunn's multiple comparisons test. (ns) Nonsignificant, (**) P < 0.0021, (****) P < 0.0001.

Figure 3.

The MRE11 complex extends DNA gaps through CtIP-stimulated exonuclease activity (see also Supplemental Fig. S3). (A) Schematic of the plasmid-based DNA substrate with a 10 nt long DNA gap, indicating various degradation scenarios. See the text for details. (B) Nuclease assays with a 10 nt long gapped DNA substrate with MRX and increasing concentrations of phosphorylated Sae2 (pSae2). (Lane 1) The linearized substrate is included for reference. Shown is a representative gel from three independent experiments. (C) Nuclease assays with a 68 nt long gapped DNA substrate with MRN, phosphorylated CtIP (pCtIP), and RPA. (Lane 8) The linearized substrate is included for reference. Shown is a representative gel from two independent experiments. (D) Annealing resection assays with a 10 nt long gapped DNA substrate and MRX and pSae2, as indicated. (Top) Zoomed-in view of the ssDNA gap, indicating the positions of the probes used to detect DNA resection. (Middle) Quantitation of resection efficiency measured with the 3′-specific probe (red; left) or the 5′-specific probe (blue; right). Averages are shown; n ≧ 3; error bars indicate SEM. (Bottom) Representative gels from at least three independent experiments. (E) Annealing resection assays with a linearized DNA substrate (otherwise identical to D) and MRX and pSae2, as indicated. (Top) Zoomed-in view of the DSB, indicating the positions of the probes used to detect DNA resection. (Middle) Quantitation of resection efficiency measured with the 3′-specific probe (red; left) or the 5′-specific probe (blue; right). Averages are shown; n ≧ 3; error bars indicate SEM. (Bottom) Representative gels from at least three independent experiments. (F) Annealing resection assays of the gapped plasmid-based DNA substrate with the T7 exonuclease. Shown is a representative gel from three independent experiments showing that both probes exhibit a similar annealing efficacy. (G) Exonuclease assays with the indicated oligonucleotide-based DNA substrates and increasing concentrations of MRN, as indicated. (Top) Cartoons of the various DNA substrates. The red asterisk represents the position of the ³²P label. (Bottom) Representative gels from three independent experiments. (H) Quantitation of experiments shown in G. Averages are shown; n = 3; error bars indicate SEM. (I) Exonuclease assays with 1 nt long gapped DNA substrate and MRN, as indicated. (Top) A cartoon of the substrate. The red asterisk represents the position of the ³²P label. (Bottom) Representative gel from three independent experiments. (J) Exonuclease assays as in I but with a substrate containing eight phosphorothioate bonds at the 3′ side of the gap, represented by the orange line in the cartoon. (Top) A cartoon of the substrate. The red asterisk represents the position of the ³²P label. (Bottom) Representative gel from three independent experiments. (K) Exonuclease assays with a 1 nt long gapped DNA substrate, MRN, and increasing concentrations of pCtIP, as indicated. (Top) A cartoon of the substrate. The red asterisk represents the position of the ³²P label. (Middle) Quantitation of DNA degradation. Averages are shown; n = 3; error bars indicate SEM. (Bottom) Representative gel from three independent experiments. (L) Exonuclease assays with a 1 nt long gapped DNA substrate, MRE11, and increasing concentrations of pCtIP, as indicated. (Top) A cartoon of the substrate. The red asterisk represents the position of the ³²P label. (Middle) Quantitation of DNA degradation. Averages are shown; n = 3; error bars indicate SEM. (Bottom) Representative gel from three independent experiments. (M) Quantitation of exonuclease assays with substrates having gaps of different lengths, with MRN, in the absence or presence of human RPA, as shown in Supplemental Figure S3F. Averages are shown; n = 3; error bars indicate SEM.

Figure 4.

EXO1 and DNA2 extend DNA gaps in the 5′–3′ direction (see also Supplemental Fig. S4). (A) Annealing DNA end resection assays with the 10 nt long gapped plasmid-based DNA substrate and EXO1, BLM–DNA2, or WRN–DNA2, as indicated. All samples contained 267.7 nM RPA. The signal resulting from the degradation in the 3′ (red; left) and 5′ (blue; right) directions is presented. Representative gels from three independent experiments are shown. (B) Annealing DNA end resection assays with the linear plasmid-based DNA substrate and EXO1, BLM–DNA2, or WRN–DNA2 without or with MRN and pCtIP. All samples contained 267.7 nM RPA. (Top) Quantitation of resection efficiency measured with the 5′-specific probe. Averages are shown; n = 3; error bars indicate SEM. Statistics: unpaired t-test. (ns) Nonsignificant, (*) P < 0.0332, (**) P < 0.0021, (***) P < 0.0002, (****) P < 0.0001. (Bottom) Representative gels from three independent experiments. (C) Annealing DNA end resection assays with the 10 nt long gapped plasmid-based DNA substrate and EXO1, BLM–DNA2, or WRN–DNA2 without or with MRN and pCtIP. All samples contained 267.7 nM RPA. (Top) Quantitation of resection efficiency measured with the 5′-specific probe. Averages are shown; n = 3; error bars indicate SEM. Statistics: unpaired t-test. (ns) Nonsignificant, (*) P < 0.0332, (**) P < 0.0021, (***) P < 0.0002. (Bottom) Representative gels from three independent experiments.

Figure 5.

Inhibition of MRE11 3′–5′ exonuclease activity and loss of CtIP rescue ssDNA gap repair in PARPi-treated cells (see also Supplemental Fig. S5). (A, top) Schematic of the DNA fiber spreading assay with the S1 nuclease in the presence and absence of PFM39. (Bottom) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib for 1 h and ±100 μM PFM39. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). (B, top) Schematic of the DNA fiber spreading assay with the S1 nuclease in the presence and absence of CtIP. (Bottom) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with the S1 nuclease ± 10 μM olaparib for 1 h and ±siCtIP. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). At least 170 tracts were scored for each sample in A and B. Statistics in A and B: Kruskal–Wallis followed by Dunn's multiple comparisons test. (ns) Nonsignificant, (*) P < 0.0332, (***) P < 0.0002, (****) P < 0.0001.

Figure 6.

EXO1, DNA2, BLM, and WRN regulate ssDNA gap repair in PARPi-treated cells (see also Supplemental Fig. S6). (A) Schematic of the DNA fiber spreading assay with the S1 nuclease in the presence and absence of EXO1, DNA2, WRN, and BLM. (B) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib for 1 h and ±siEXO1 #1. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). (C) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib for 1 h and ±siDNA2. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). (D) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib for 1 h and ±siBLM. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). (E) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib for 1 h and ±siWRN. The S1 nuclease was added immediately after (time 0) and 30 min (time 30) after olaparib removal (n = 3). (F) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 control sgAAVS1 or pooled sgEXO1 KO cells treated with S1 ± 10 μM olaparib for 1 h and ±C5 inhibitor. (G) Dot plot and median of IdU tract lengths in SUM149PT and SUM149PT + BRCA1 cells treated with S1 ± 10 μM olaparib for 1 h and ±siWRN and siBLM. At least 110 tracts were scored for each sample in B–G. Statistics in B–G: Kruskal–Wallis followed by Dunn's multiple comparisons test. (ns) Nonsignificant, (*) P < 0.0332, (**) P < 0.0021, (***) P < 0.0002, (****) P < 0.0001.

Figure 7.

DNA breaks only form after prolonged PARPi treatment (see also Supplemental Fig. S7). (A, left) Representative immunofluorescence images of γ-H2AX foci in SUM149PT and SUM149PT + BRCA1 cells treated with 10 μM olaparib for the indicated durations. Scale bar, 11 μm. (Right) Dot plot of normalized γ-H2AX focus intensity (A.U). At least 300 cells were analyzed per condition per biological replicate (n = 3). Statistics: Kruskal–Wallis followed by Dunn's multiple comparisons test. (**) P < 0.0021, (****) P < 0.0001. (B, left) Representative immunofluorescence images of γ-H2AX foci in SUM149PT and SUM149PT + BRCA1 cells treated with 10 μM olaparib plus 10 μM palbociclib for the indicated durations. Scale bar, 11 μm. (Right) Dot plot of normalized γ-H2AX focus intensity (A.U). At least 300 cells were analyzed per condition per biological replicate (n = 3). Statistics: Kruskal–Wallis followed by Dunn's multiple comparisons test. (ns) Nonsignificant, (****) P < 0.0001. (C, left) Schematic (top) and representative image (bottom) of metaphase spread experiment. Red boxes indicate representative images of triradial and chromatid breaks (ctb). (Right) The percentage of triradials (top) and chromatid breaks (bottom) in SUM149PT + BRCA1 and SUM149PT cells ±10 μM olaparib for 4 and 48 h. (D) Working model for the mechanism of ssDNA gap resection. MRN–CtIP resect ssDNA gaps in the 3′–5′ direction, whereas EXO1 and DNA2/WRN/BLM operate in the 5′–3′ direction. ssDNA gaps are overresected in BRCA1-deficient cells, preventing their repair. The unrepaired ssDNA gaps collide with ongoing replication forks, leading to fork breakage and promoting cell death in the absence of BRCA1.