RAD50 is required for efficient initiation of resection and recombinational repair at random, gamma-induced double-strand break ends

Abstract

Resection of DNA double-strand break (DSB) ends is generally considered a critical determinant in pathways of DSB repair and genome stability. Unlike for enzymatically induced site-specific DSBs, little is known about processing of random "dirty-ended" DSBs created by DNA damaging agents such as ionizing radiation. Here we present a novel system for monitoring early events in the repair of random DSBs, based on our finding that single-strand tails generated by resection at the ends of large molecules in budding yeast decreases mobility during pulsed field gel electrophoresis (PFGE). We utilized this "PFGE-shift" to follow the fate of both ends of linear molecules generated by a single random DSB in circular chromosomes. Within 10 min after gamma-irradiation of G2/M arrested WT cells, there is a near-synchronous PFGE-shift of the linearized circular molecules, corresponding to resection of a few hundred bases. Resection at the radiation-induced DSBs continues so that by the time of significant repair of DSBs at 1 hr there is about 1-2 kb resection per DSB end. The PFGE-shift is comparable in WT and recombination-defective rad52 and rad51 strains but somewhat delayed in exo1 mutants. However, in rad50 and mre11 null mutants the initiation and generation of resected ends at radiation-induced DSB ends is greatly reduced in G2/M. Thus, the Rad50/Mre11/Xrs2 complex is responsible for rapid processing of most damaged ends into substrates that subsequently undergo recombinational repair. A similar requirement was found for RAD50 in asynchronously growing cells. Among the few molecules exhibiting shift in the rad50 mutant, the residual resection is consistent with resection at only one of the DSB ends. Surprisingly, within 1 hr after irradiation, double-length linear molecules are detected in the WT and rad50, but not in rad52, strains that are likely due to crossovers that are largely resection- and RAD50-independent.

Figure 1. Ionizing radiation–induced DSBs and repair.

Haploid cells growing logarithmically in YPDA media were arrested at G2/M with nocodazole and γ-irradiated with 5 to 80 krads from a Cs source. Cells from the 80 krad dose were returned to YPDA containing nocadazole and incubated for up to 4 hours to allow repair of DSBs (A) DSBs and repair in WT and exo1Δ strains. DSBs are indicated by the progressive decrease in intensity of linear chromosomal bands and concomitant increase in intensity of a smear of randomly broken DNA with increasing dose (left panel). The linearized circular chromosome III (III-L) containing a single DSB can be seen as a band just above the chromosomes I and VI doublet band starting at 5 krads and increasing in intensity at higher doses. With time, the position of the III-L band appears to shift upward to higher molecular weight (see text). By 2 hrs the band is no longer seen on stained gels due to repair, which results in recircularization and immobilization in the agarose plug. Repair of full-length linear chromosomes is indicated by increasing intensities of chromosomal bands along with decreasing intensity of the smear of broken DNA fragments (center and right panel). The DNAs were run on TAFE gels (see Materials and Methods). (B) Survival of γ-irradiated G2-arrested WT and mutant strains. Cell suspensions in ice-cold water were irradiated and plated to YPDA plates. Results for WT, rad50Δ, and rad52Δ were from 7 independent experiments; rad51Δ from 3 experiments; and exo1Δ and rad50Δ exo1Δ were from single experiments.

Figure 2. Changes in broken chromosomes following irradiation of HR defective mutants.

(A) Absence of repair in HR defective mutants. Logarithmically growing rad51Δ, rad50Δ, and rad52Δ cells were arrested in G2/M with nocodazole, irradiated with 80 krad, and returned to growth medium for up to 4 hrs (as described in Figure 1A). A single break in Chr III resulted in a band above the bottom doublet (230 kb) at 0 hrs. While there was no evidence of repair, the rad51Δ strain, unlike rad50Δ and rad52Δ, exhibited limited repair of ChrXII, half of which contains ribosomal DNA (first band below the wells). Chromosomal material was displayed by PFGE (CHEF, see Materials and Methods) and the gels were stained with SybrGold. (B) PFGE-shift of broken fragments in rad52Δ but not in rad50Δ. Cells of rad50Δ and rad51Δ were arrested in G2/M, irradiated, and returned to growth medium as described in Figure 2A. Samples were run on a pulsed-field gel (TAFE; see Materials and Methods), and a southern transfer was hybridized to a probe (V16) specific for unique sequence near the left telomere of Chr V (16 kb from the end). The lower portion of the smear of broken fragments below the intact full-length Chr V shifts upward by 0.5 hours in the rad52Δ but not the rad50Δ strain.

Figure 3. DNA resection leads to reduced mobility (PFGE-shift) of chromosomal size molecules.

(A) PFGE-shift of uncapped chromosomes. A temperature-sensitive cdc13-1 strain which is defective for telomere-capping (DAG760 described in [44]) was grown to stationary phase for 3 days in YPDA media at the 23°C permissive temperature, then diluted 20-fold into fresh YPDA at 23°C and 37°C. Presented are results following the 37°C incubation, a condition that results in 5′ to 3′ resection at telomeres (see text). Samples were collected at the indicated times and prepared for PFGE (TAFE). By 3 hrs, when 97% of cells in the 37°C culture were arrested in G2, novel bands appear in lanes from the 37°C time course at positions corresponding to molecular weights of approximately 25 to 40 kb above many of the known chromosomal bands (left image). The specific chromosomal shifts were confirmed by Southerns using the Chr I specific FLO1 probe (right image) and other probes specific for chromosomes II, III, V, and VIII (data not shown). These PFGE-shifts were not detected by either stained gels or Southerns for cells incubated at the permissive 23°C temperature (data not shown). (B) PFGE-shift associated with HO-induced DSB. Resection at a unique HO-endonuclease–induced site-specific DSB in Chr III leads to PFGE-shift in broken chromosome fragments. Following growth of WT cells in YEP-lactate media and G2/M arrest in nocodazole for 3 hrs, galactose was added (2% final concentration) to induce HO-endonuclease, resulting in a DSB at MATa (strain AM919). Chromosomes were separated by PFGE. [Because of differences in PFGE-CHEF running conditions for the HO-induced break system, the “apparent” MW shift is different than would be observed for the conditions used in the other PFGE experiments that are described in this paper.] A Southern transfer of the CHEF gel was hybridized with an ADE1 probe that identifies Chr III and also cross-hybridizes at its native position in Chr I (which also serves as internal standard) . HO cutting was approximately 90% complete by 0.5 hours and PFGE-shift is detected by 1 hour. Note that the unbroken chromosome III contains two copies of ADE1 inserted at HML and HMR, which is responsible for the higher intensity of its hybridization to the ADE1-specific probe as compared to the Chr I, and Cf A, bearing only one copy of ADE1. Also, to achieve the required separation between the resected and unresected fragments, the smaller fragment containing HMR::ADE1 (approximately 100 kb) does not remain in the gel, so that only the larger fragment containing HML::ADE1 is detected.

Figure 4. The PFGE-shift in rad52Δ and WT resulting from a random γ-induced single DSB is due to resection.

(A) Rapid PFGE-shift of broken ChrIII in rad52Δ mutant during postirradiation incubation. Nocodazole-arrested G2/M cells were irradiated with 80 krads and returned to YPDA media to allow processing of DSB ends. Presented is a Southern of a pulsed-field gel (TAFE; see Materials and Methods) using a Chr III specific CHA1 probe. Circular Chr III from unirradiated cells (“no γ”) remained trapped in the plug. Chr III molecules with only one DSB migrate at a position corresponding to 300 kb (“0” hr lane). Less than 10 min after returning cells to YPDA, most of the molecules in the band shifted to an apparent molecular weight that is greater than 300 kb. By 2 hrs the band reached a position of maximum shift corresponding to an apparent MW of approximately “430” kb. Smearing under the linear Chr III band is due to multiple DSBs in Chr III. (B) The PFGE-shift is due to resection based on digestion by mung bean nuclease. PFGE plugs from an experiment involving 80 krad to rad52Δ cells and postirradiation incubation (as in Figure 4A) were run on a CHEF gel under conditions to maximize resolution of the linearized 300 kb Chr III (see Materials and Methods). The probe used for this Southern labels the lambda DNA ladder as well as the LEU2 sequence on Chr III. As seen in the left group of lanes, there is a PFGE-shift of Chr III linearized with a single random DSB within 30 min postirradiation incubation. Duplicates of these plugs were treated with mung bean nuclease to remove 3′ single-strand tails. As shown in the group of lanes on the right, the nuclease treatment resulted in molecules from the ≥0.5 hrs incubation exhibiting faster PFGE mobility. By 6 hrs, the average size of the trimmed linearized Chr III was reduced by ∼15–20 kb, corresponding to an average resection rate of ∼1.5 kb per DSB end per hr. (C) Rapid PFGE-shift and appearance of novel bands in WT cells. WT cells were arrested at G2/M by nocodazole, irradiated with 80 krads and returned to YPDA media. Plugs were prepared at the indicated times and run on PFGE (TAFE; see Materials and Methods). The probe used for this Southern corresponded to the LEU2 sequence on chromosome III, which is circular at the beginning of experiment, but becomes linear following radiation-induced breakage (similar to Figure 4A). In addition to the non-resected (M) and resected (M**) bands seen in rad52, two additional bands referred to as D and D** appear within 30 min after irradiation. The D band corresponds to an MW that is twice that of the non-shifted linear monomer. Since the appearance of D and D** are RAD52-dependent, these are considered to be recombinant molecules (discussed in text). Band D** is proposed to be a linear dimer with both ends resected. [The “no γ” lane is from a different part of the same gel. The DNA in the well of the last lane did not transfer well. The “∼0.2 hr” time point is not precise because EDTA was not added although cells were put on ice. The amount of linear ChrIII detected in the “no γ” lane is <3% of that in the well.] (D) Resection in WT cells is the source of the PFGE-shift in ChrIII containing a random γ-induced single DSB. As in Figure 4B, mung bean nuclease treatment of plugs from the 80 krad+repair experiment (right half of image) abolished the PFGE-shift seen with untreated plugs (left half of image). The rapid and efficient repair of resected DSB ends in WT cells prevented accurate measurement of the resection rate.

Figure 5. The PFGE-shift associated with a random γ-induced single DSB requires Rad50 (MRX).

(A) Limited processing of Chr III DSBs in rad50Δ. Nocodazole arrested G2/M cells were irradiated with 80 krads and returned to YPDA media. Plugs were prepared from the indicated time points and run on PFGE (TAFE; see Materials and Methods). Presented is a Southern of the gel using a Chr III specific CHA1 probe. Circular Chr III from unirradiated cells (“no γ”) remained trapped in the plug. Chr III molecules with only one DSB migrate to a position corresponding to 300 kb (0 hr lane). Bands corresponding to M (linear non-resected monomer) and D (linear non-resected dimer) of Figure 4C (WT 80 krad+incubation) appear with the rad50Δ strain. However, bands corresponding to M** or D** of Figure 4C are not seen, indicating that resection is greatly reduced in MRX-deficient cells. The faint smear extending above the linear non-resected M band, detectable by ∼1 hr indicates limited resection and/or resection possibly of only one end of these molecules as discussed in the text. (B) The limited PFGE-shift in rad50Δ after 80 krad differs from rad51Δ and rad52Δ. The smear of broken Chr III (Figure 5A) was characterized by PFGE (CHEF; see Materials and Methods). Comparisons were made with the PFGE shift found with the rad51Δ and rad52Δ mutants. (The stained gel is presented in Figure 2A.) For rad50Δ (center lanes) no band is observed corresponding to the typical M** band seen with rad51Δ (left) or rad52Δ (right) mutants. Instead, there is a slow accumulation of a faint band (M*) with a PFGE-shift corresponding to ∼40 kb increase in apparent molecular weight. (C) The effect of mung bean nuclease treatment on the mobility of broken Chr III. PFGE plugs from an experiment involving 80 krad to rad50Δ and postirradiation incubation (as in Figure 5A) were run on a CHEF gel under conditions that maximize resolution of the linearized 300 kb Chr III (see Materials and Methods). A slight smear is observed above the linearized monomer band in the no MBN lanes (i.e., at 2 and 3 hrs after 80 krads). Mung bean nuclease (+MBN lanes) removes the smear and yields tighter ∼300 kb bands, indicating that the smear is due to resection. However, unlike for rad52Δ (Figure 4B), the products of the MBN treatment did not run at a detectably faster rate than the unresected monomer in the 0 hour lane, indicating that little resection had occurred. The “λ” lanes correspond to 48.5 kb lambda DNA ladder.

Figure 6. The PFGE-shift associated with a random γ-induced single DSB after low doses or HO-endonuclease produced DSBs.

(A) PFGE-shift associated with a single radiation-induced DSB in circular Chr V in WT but not rad50Δ cells. Nocodazole arrested G2/M cells containing a circular chromosome V were irradiated with 40 krads and returned to YPDA media. Plugs were prepared at the indicated times and run on PFGE (TAFE). Southern transfers were hybridized with a MET6 probe specific for the 530 kb Chr V. Results are similar to those obtained using circular Chr III, except the resected form of the putative linear dimer band (D**) seen at ∼1080 krads in WT (left image, 1 hr) is not well-resolved from the unresected form, twice the molecular weight of the non-resected monomer band (M). PFGE-shift of the linearized monomer band in WT reaches a maximum at an apparent molecular weight of ∼“600” kb at 1 hr, after which the M** band mostly disappears due to repair and recircularization. In the rad50Δ strain the non-resected monomer band (M) persists for 4 hrs with only faint smearing above it as also seen in rad50Δ using the circular Chr III construct (Figure 5A). The putative linear dimer band (D) was also detected with rad50Δ cells. (B) Resection events at low dose suggest one- and two-ended events. Arrested rad50Δ and rad52Δ G2/M cells containing circular Chr III were irradiated with 20 krads and returned to YPDA media. Plugs were prepared at the indicated time points and run on PFGE using CHEF analysis (see Materials and Methods). Two PFGE-shift bands, M* (at 0.5 to 4 hrs) and M** (at 4 hrs) were detected. As suggested in the text, the M* band is consistent with linearized molecules that were resected at only one end of the DSB. The M** seen in rad50Δ at 4 hrs is at the same maximum shift position as the M** band typically seen in rad52Δ by 2 hrs (see right group of lanes) and is proposed to be due to molecules that were resected at both ends with 3′ single-strand tails long enough to cause maximum shift. The persistence of significant portions of non-resected and of partially resected molecules in rad50Δ but not in rad52Δ demonstrates that RAD50 is required for the rapid and efficient initiation of resection at damaged ends in WT and rad52Δ cells. (C) Lack of resection at HO-induced DSB in rad50Δ cells is reflected by absence of PFGE-shift. Resection was analyzed in the nocodazole-arrested rad50Δ cells (MN108) using procedures similar to those described in Figure 3B. Unlike for WT (Figure 3B), there was very little PFGE-shift, even at 5 hr.

Figure 7. Relationship between dose and dimer formation in rad50Δ.

Nocodazole arrested rad50Δ G2/M cells containing the circular Chr III were irradiated with 20, 30, and 40 krads and returned to YPDA media. Samples for PFGE were run on a CHEF (see Materials and Methods). The Southern transfer was hybridized to the Chr III-specific CHA1 probe. The bands correspond to unresected monomer (M), putative one-end resected monomer (M*) and two-end resected monomer (M**), as well as putative dimer (D) (see Figure 5 B). Bands were quantitated using Kodak MI software and the ratio of total dimer DNA (D) to monomer DNA (M+M*+M**) for each time point is shown below each lane.

Figure 8. Resection at DSBs in logarithmically growing cells and role of RAD50.

(A) Resection at a random DSB is rapid in asynchronous WT cells. Logarithmically growing WT cells containing the circular Chr V were irradiated (40 krads) and returned to growth conditions as described in Figure 1A and Materials and Methods. Samples were processed for PFGE and run on TAFE. The Chr V-specific MET6 probe was used to hybridize the Southern blot. The kinetics of PFGE shift of nearly all molecules is similar to that observed for nocodazole-arrested cells (see Figure 6A). (B) Limited resection in G2/M and asynchronous rad50Δ cells. Logarithmically growing rad50Δ cells containing the circular Chr V were either arrested at G2/M with nocodazole (the G2/M image is reproduced from Figure 6A) or not arrested (right group of lanes) before irradiation (40 krads) and returned to growth in YPDA. Samples were processed for PFGE (TAFE) and a Southern transfer was hybridized to the Chr V specific MET6 probe. Although more molecules experience Rad50-independent resection in asynchronous cells than in G2/M cells, the persistence of unshifted/unresected molecules as compared to WT (Figure 8A) indicates an important role for Rad50/MRX in initiation of resection even in asynchronous cells. (C) Resection of HO-cut chromosome fragments in asynchronous WT and rad50Δ cells as evidenced by PFGE-shift. The resection was analyzed in the asynchronous AM919 and MN108 cells (rad50Δ) using methods similar to those described in Figure 3B. The persistence of unresected “Cf A” molecules in rad50Δ indicates an important role for Rad50/MRX for initiation of resection. However, in asynchronous cells the requirement for Rad50 is not as great as in G2/M cells (Figure 6C).