DNA Damage-Inducible and RAD52-Independent Repair of DNA Double-Strand Breaks in Saccharomyces cerevisiae

Corresponding author: Carol Wood Moore, Department of Microbiology and Immunology, City University of New York Medical School/SDSBE, Science Bldg., Rm. 919, Convent Ave. at 138th St., New York, NY 10031., moore{at}med.cuny.edu (E-mail)

Communicating editor: L. S. SYMINGTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Chromosomal repair was studied in stationary-phase Saccharomyces cerevisiae, including rad52/rad52 mutant strains deficient in repairing double-strand breaks (DSBs) by homologous recombination. Mutant strains suffered more chromosomal fragmentation than RAD52/RAD52 strains after treatments with cobalt-60 irradiation or radiomimetic bleomycin, except after high bleomycin doses when chromosomes from rad52/rad52 strains contained fewer DSBs than chromosomes from RAD52/RAD52 strains. DNAs from both genotypes exhibited quick rejoining following irradiation and sedimentation in isokinetic alkaline sucrose gradients, but only chromosomes from RAD52/RAD52 strains exhibited slower rejoining (10 min to 4 hr in growth medium). Chromosomal DSBs introduced by irradiation and bleomycin were analyzed after pulsed-field gel electrophoresis. After equitoxic damage by both DNA-damaging agents, chromosomes in rad52/rad52 cells were reconstructed under nongrowth conditions [liquid holding (LH)]. Up to 100% of DSBs were eliminated and survival increased in RAD52/RAD52 and rad52/rad52 strains. After low doses, chromosomes were sometimes degraded and reconstructed during LH. Chromosomal reconstruction in rad52/rad52 strains was dose dependent after irradiation, but greater after high, rather than low, bleomycin doses with or without LH. These results suggest that a threshold of DSBs is the requisite signal for DNA-damage-inducible repair, and that nonhomologous end-joining repair or another repair function is a dominant mechanism in S. cerevisiae when homologous recombination is impaired.

CHROMOSOMAL double-strand breaks (DSBs), if left unrepaired, are lethal. In eukaryotic cells, several pathways repair DSBs. The most common pathway in mammalian cells is end joining (CHU 1997 ; JEGGO 1998 ; ECKARDT-SCHUPP and KLAUS 1999 ), but this pathway is far less common in Saccharomyces cerevisiae cells where homologous recombination predominates (reviewed by PETES et al. 1991 ; AGUILERA 1996 ; MOORE and HABER 1996 ; SIEDE et al. 1996 ; FRIEDL et al. 1998 ; LEWIS et al. 1998 , LEWIS et al. 1999 ; PAQUES and HABER 1999 ). Unlike homologous recombination, end joining requires little or no DNA homology, and thus is also called nonhomologous or illegitimate recombination (e.g., ROTH and WILSON 1988 ; THODE et al. 1990 ; GOEDECKE et al. 1992 ; MEZARD et al. 1992 ; LEHMAN et al. 1993 ; MEZARD and NICOLAS 1994 ; SCHIESTL et al. 1994 ). The three main repair pathways for DSB repair identified in S. cerevisiae are RAD52-dependent homologous recombination repair, Ku-protein-dependent complementary end-joining or nonhomologous repair, and the single-strand annealing or nonconservative repair occurring in direct-repeat DNA (reviewed by FRIEDL et al. 1998 ; LEWIS et al. 1998 ). Homology-dependent repair or recombination and single-strand annealing require the RAD52 gene product (RESNICK and MARTIN 1976 ; RESNICK 1978 ; OZENBERGER and ROEDER 1991 ; PETES et al. 1991 ; SUGAWARA and HABER 1992 ), but nonhomologous end joining (NHEJ) does not (KRAMER et al. 1994 ; MEZARD and NICOLAS 1994 ; SIEDE et al. 1996 ; LEWIS et al. 1999 ). Very few DNA repeats are found in S. cerevisiae (DUJON 1996 ).

It is generally accepted that cell cycle arrest ensures time for DNA repair before cells reenter the cell cycle. Direct evidence in yeast from experiments on expression of the homothallic switching endonuclease shows that DSBs can initiate cell cycle arrest (MALKOVA et al. 1996 ; MOORE and HABER 1996 ). In mammalian cells, arrest is dependent on p53 (NELSON and KASTAN 1994 ). Several proteins in yeast are involved in checkpoint control (WEINERT and HARTWELL 1988 ; WEINERT 1998A , WEINERT 1998B ; SIEDE et al. 1993 , SIEDE et al. 1994 ; SIEDE 1995 ; ALLEN et al. 1994 ; HARTWELL and KASTAN 1994 ; ELLEDGE 1996 ; PAULOVICH et al. 1997 ; TOCZYSKI et al. 1997 ). The signal mediating damage-induced cell cycle arrest in yeast is suspected to involve DNA strand breaks (SIEDE 1995 ). In mammalian cells, the sensing of broken DNA molecules and the initiation of cascade kinase reactions resulting in cell cycle arrest involve DNA-activated protein kinase (DNA-PK) (LEE et al. 1997 ), which plays an important role in Ku-mediated DSB repair via nonhomologous recombination and in V(D)J recombination (JACKSON and JEGGO 1995 ; JEGGO 1997 ).

The major objective in the current study was to optimize by experimental design the potential for the repair of chromosomal DSBs in S. cerevisiae. The goal was to produce different types of DSBs (i.e., directly or during the processing of DNA lesions) and afterward delay further cell cycle progression to maximize the opportunity for cells to reconstruct and repair their chromosomes. Depending upon the extent and nature of DNA damage, such a delay could assist cells in monitoring, processing, and repairing the DNA lesions in a manner somewhat analogous to the checkpoints for detecting DNA damage at G₁-S and G₂-M (reviewed by HARTWELL and KASTAN 1994 ; TOCZYSKI et al. 1997 ; WEINERT 1998A , WEINERT 1998B ). Thus, we imposed a prolonged delay identical to liquid holding conditions [(LH) e.g., PATRICK and HAYNES 1968 ; FRANKENBERG-SCHWAGER et al. 1980 ], rather than leaving uncertain whether the cells would otherwise allow for an adequate delay. During LH, "repair" complexes would have longer to access the lesions, possibly in a sequential fashion, because more than one complex may need to operate at a particular site. To distinguish between homologous and nonhomologous recombinational repair events, we investigated if diploid RAD52/RAD52 and rad52/rad52 strains responded similarly during LH. Since mutant rad52/rad52 cells are deficient in homologous genetic recombination and the repair of DSBs (e.g., RESNICK and MARTIN 1976 ; RESNICK et al. 1984 ; FRIEDBERG et al. 1991 ; PRADO and AGUILERA 1995 ), they are exquisitely sensitive to killing by ionizing radiation and radiomimetic DNA-damaging agents such as bleomycin (MOORE 1978 , MOORE 1982A , MOORE 1991 ; KESZENMAN et al. 1992 ). Pathways in S. cerevisiae are shared for the repair of chromosomal damage by ionizing radiation and bleomycins (MOORE 1978 , MOORE 1982A , MOORE 1982C , MOORE 1991 ; KESZENMAN et al. 1992 ; HE et al. 1996 ), and include those controlled by the RAD6, RAD9, RAD18, RAD50–RAD58, BLM1, BLM3, and BLM5–BLM7 genes [rad (reviews by HAYNES and KUNZ 1981 ; FRIEDBERG 1988 ; FRIEDBERG et al. 1991 ; GAME 1993 ; PRAKASH et al. 1993 ; RAMOTAR and MASSON 1996 ); blm (MOORE 1980 , MOORE 1991 ; M. MARTINEZ, J. F. MCKOY and C. W. MOORE, unpublished results)]. In this article, we analyze the processing and reconstruction of broken chromosomes after treatments with rays and bleomycin in repair-competent wild-type diploid cells and rad52/rad52 mutant diploid cells defective in homologous recombination. In particular, we focus on the effects of LH. Here, we report new and unexpected results that impart new information on cellular controls of the repair of DNA damage in RAD52/RAD52 and rad52/rad52 cells and indicate that damaged chromosomes are reconstructed substantially in both types of cells.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains:
The strains used in this study are listed in Table 1.

Media and culturing conditions:
Strains were routinely grown with aeration in nonsynthetic complete medium (YPAD; MOORE 1982A , MOORE 1982B ) at 30°, harvested by centrifugation at 4°, and washed twice with deionized water.

2-¹⁴C- and 6-³H-prelabeled DNAs:
Cells were grown ~14 hr from starting inocula of fresh cells (5 x 10⁶ cells per ml) in supplemented synthetic minimal medium (PETES and FANGMAN 1972 ; FORTE and FANGMAN 1976 ) as described previously (MOORE 1982B , MOORE 1990 ). The [6-³H]uracil (specific activity = 20–30 Ci/mmol; New England Nuclear, Boston) was added to 7 µCi/ml, or [2-¹⁴C]uracil (specific activity = 40–60 Ci/mmol; New England Nuclear) was added to 5 µCi/ml. Cells were harvested by centrifugation at 4° in an RC-5B Sorvall SS34 rotor at 3000 x g, washed once in deionized water, chased in supplemented synthetic minimal medium (containing 0.0017% uracil) without the radiochemical for 60–90 min, and washed twice with deionized water. The two radioisotopes were alternated between RAD52/RAD52 and rad52/rad52 cells in replicated experiments, with no detectable effect on sedimentation profiles or molecular weights of DNAs. In addition, no differences were observed in replicated experiments between profiles or molecular weights of 2-¹⁴C- and 6-³H-prelabeled DNAs. RNA was hydrolyzed completely after velocity sedimentation (MOORE 1982B , MOORE 1990 ).

irradiation:
Exposures to radiation were conducted according to our published procedures (e.g., MOORE 1982B ; DARDALHON et al. 1994 ). Cells were grown to the stationary phase of growth, washed and suspended at 10⁷ cells/ml (for velocity sedimentation) or 5 x 10⁶ cells/ml [for pulsed-field gel electrophoresis (PFGE)] in deionized water, and chilled until ice cold. Irradiations were carried out on ice in cobalt-60 irradiators. For velocity sedimentation, Dr. Christopher Lawrence (University of Rochester, Rochester, NY) made the source (J. L. Shepard and Associates, Glendale, CA) available. The dose rate was determined from the decay constant of ⁶⁰Co after periodic ferrous sulfate dosimetry. RAD52/RAD52 and rad52/rad52 cells were irradiated together. For PFGE, a cobalt-60 -irradiation source (Institut Curie-Section de Biologie, Paris) was used at a dose rate of 20 Gy/min as determined by ferrous sulfate dosimetry. Irradiated cells were immediately pelleted by centrifugation at 4°. For velocity sedimentation, unirradiated and irradiated cells were either immediately converted to spheroplasts or resuspended without LH for postirradiation incubation with aeration.

Bleomycin treatments:
Bleomycins are a family of low-molecular-weight glycometallopeptides [M_r ~1500–1600 (UMEZAWA et al. 1966 ; UMEZAWA 1976 )]. The anticancer formulation Blenoxane [~55–70% (usually 68–69%) bleomycin A₂ and ~25–32% bleomycin B₂ (CROOKE and BRADNER 1976 ; W. T. BRADNER, personal communication)] was a gift from Bristol Myers Squibb Laboratories through the courtesy of Dr. William T. Bradner, Ms. Linda Sanders, and Mr. Daniel T. Elliott. Bleomycins were dissolved and diluted in deionized water just prior to use. The final pH was 5. Absorbance was monitored at 292 nm. Cells were grown to the stationary phase of growth, washed and resuspended at 1 x 10⁷ washed cells/ml of deionized water (pH 5), and incubated with bleomycin at 4° for 30 min with aeration. Neither metal ions nor reducing agents were added to reaction mixtures. Throughout this incubation, the pH of the reaction was 5. Rigorous consistency was followed in culturing and reaction conditions, including pH, chemical lots, cell densities, and temperatures during reactions. For most experiments, EDTA was added to 0.025 M at the end of 30 min.

LH:
At the end of reactions, cells were immediately pelleted by centrifugation at 4° and washed twice at 4° with 0.05 M EDTA. Cells were either diluted and plated immediately or resuspended at 1 x 10⁷ cells/ml for LH. The routine microscopic examination and visual counting of cells before and after LH showed that the cell populations did not bud or grow during the period of LH, and cell lysis was not observed. Phosphate buffer (0.005 M) was used for LH, although comparable results were usually obtained with phosphate buffer, phosphate-buffered saline, and deionized water. Plating efficiencies were also measured before and after LH.

Spheroplast formation:
Spheroplasts were prepared as described previously (MOORE 1982B ). Washed cells were suspended at ~10⁸ cells/ml in 0.9 M sorbitol, 0.05 M EDTA, and 0.1 M sodium citrate (pH 5.8) containing 3 mg of zymolyase (60,000 units/g; Kirin Brewery Co., Ltd., Takasaki, Japan) per milliliter at 34°. Spheroplasts were sedimented in an RC2B SS34 rotor at 2000 x g and resuspended gently in cold 0.01 M sodium citrate and 0.1 M EDTA. Approximately 5 x 10⁶ to 1 x 10⁷ spheroplasts in 100 µl of this suspension were layered onto the lysis layer on alkaline sucrose gradients. Spheroplasts were layered carefully to minimize potential shearing of DNAs.

Sedimentation of DNAs and molecular weight determinations:
The 2-¹⁴C- and 6-³H-prelabeled DNAs from RAD52/RAD52 and rad52/rad52 cells were sedimented together through precalibrated, isokinetic alkaline sucrose gradients according to published procedures (CARRIER and SETLOW 1971 ; EHMANN and LETT 1973 ; RESNICK and MARTIN 1976 ; REYNOLDS and FRIEDBERG 1981 ; MOORE 1982B , MOORE 1990 ). The size of native yeast DNA (M_r ~3 x 10⁸) is ideal for this assay system, particularly at low centrifugal speeds, and the assay is quite sensitive under limited reaction conditions. With T4 and T7 DNAs as standards, the gradient system was determined to be isokinetic. The single-strand breaks (SSBs) assayed in this system include closely and more distally opposed SSBs. Alkali-labile lesions left in DNA by the release of free bases are also converted to DNA breaks, and thus are included in the SSBs assayed in this system.

PFGE:
The methods used to prepare cells and agarose plugs containing the cells (GAME et al. 1989 ; DARDALHON et al. 1994 ) were modified from the method of SOR 1988 and SCHWARTZ and CANTOR 1984 . Untreated and treated cells were washed twice with 0.05 M EDTA (pH 7.5) at 0° by centrifugation and resuspended at a concentration of 1.2 x 10⁹ cells/ml in 0.175 ml of 0.05 M EDTA. They were rapidly mixed with 0.1 ml solution I [1 M sorbitol, 0.1 M sodium citrate (pH 5.8) 0.01 M EDTA, and 5% v/v 2-mercaptoethanol containing 3 mg/ml zymolyase (10⁵ units/g; ICN Immunobiologicals, Costa Mesa, CA or Seikagaku Corporation, Tokyo)]. Agarose plugs (85 ml) were cast in a plug former containing 4.5 x 10⁷ cells/plug. Plugs were incubated overnight in solution II [0.45 M EDTA (pH 8), 0.01 M Tris-HCl, and 7.5 v/v 2-mercaptoethanol] at 37°. This solution was replaced by solution III [0.45 M EDTA (pH 8), 0.01 M Tris-HCl, 1% v/v sodium-N-lauryl sarcosinate, and 1 mg/ml proteinase K (Boehringer Mannheim, Indianapolis)], and incubated overnight at 50°. After replacement of the solution twice with 0.5 M EDTA (pH 8), samples were stored at 4° until PFGE analysis. We used a CHEF Mapper R (Bio-Rad, Richmond, CA). The agarose gel was made with 1% molecular biology certified agarose (Bio-Rad) in 0.5x TBE buffer prepared from 10x TBE buffer [108 g Tris base (Sigma, St. Louis), 55 g boric acid, and 40 ml of 0.5 M EDTA, pH 8, in 1 liter of distilled water]. PFGE gels were run in 0.5x TBE buffer at 12° with an angle of ± 60° (alternately) with a voltage gradient of 6 V/cm and switch times of 60 s for 15 hr and 90 s for 7 hr. Thereafter, gels were stained with 0.8% ethidium bromide solution for 2 hr and destained for 0.5 hr.

After PFGE, the quantitation of DSBs was carried out as described previously (GEIGL and ECKARDT-SCHUPP 1990 ; DARDALHON et al. 1994 ) using a densitometric station (Biocom, Villebon sur Yvette, France) by computerized image analysis (Elphor program). We used the equation n = -ln q, where q is the ratio between the intensities of relative areas of the migration profiles of treated and untreated samples, and n is the number of breaks per molecule. The estimation is based on the 8 chromosomal bands corresponding to the 12 largest chromosomes (mean molecular weight = 1020 kbp; OLSON 1991 ) revealed on the photonegatives of gels stained with ethidium bromide. These estimates were comparable to the calculations obtained using the three largest chromosomal bands.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Quick rejoining of radiation-induced DNA breaks in RAD52/RAD52 and rad52/rad52 cells in the absence of LH:
In this comparative study of DNA repair in RAD52/RAD52 and rad52/rad52 strains, experiments were first carried out to determine the capacities of cells of both genotypes to carry out the quick and slower components of DNA rejoining. The proportion of DNA breaks that are rejoined in RAD52/RAD52 and rad52/rad52 cells without LH was investigated by velocity sedimentation of single-stranded DNAs through calibrated isokinetic denaturing gradients. DNA molecules whose ends are substrates for yeast ATP-dependent DNA ligase [poly (deoxyribonucleotide):poly(deoxyribonucleotide) ligase] are immediately ligatable and constitute a quick component of DNA rejoining (MOORE 1982B , MOORE 1988A , MOORE 1990 ).

Typical sedimentation profiles of DNAs from RAD52/RAD52 and rad52/rad52 cells are shown in Fig 1. The sedimentation rates of DNAs from RAD52/RAD52 (A) and rad52/rad52 (B) cells are nearly equivalent before irradiation, except that the profiles from the rad52/rad52 cells contain a somewhat lower proportion of the highest-molecular-weight DNAs (2–3 x 10⁸ D) than RAD52/RAD52 profiles. This suggested that rad52/rad52 cells accumulated a limited number of DNA single-strand breaks before irradiation. After 250 Gy, the profiles of DNAs from RAD52/RAD52 and rad52/rad52 cells are also similar. After 10 min post-treatment incubation, the profiles of both RAD52/RAD52 and rad52/rad52 cells exhibit DNAs of higher molecular weights, indicating that both genotypes rapidly rejoined chromosomal breaks in the absence of LH. The total radioactivity recovered from unirradiated and irradiated cells was equivalent, indicating that the mean molecular weights of DNAs did not increase at the expense of losing low-molecular-weight DNA fragments in RAD52/RAD52 or rad52/rad52 cells.

View larger version (25K): In this window In a new window Download PPT slide   Figure 1. Sedimentation profiles illustrating the consequences of 250 Gy of irradiation and a 10-min postirradiation incubation on the sedimentation rates of DNAs from stationary-phase cells. Equal numbers of RAD52/RAD52 and rad52/rad52 cells were mixed and suspended before irradiation, and all subsequent experimental procedures were carried out on the mixed suspensions. (A) RAD52/RAD52 (CM-1293) cells. (B) rad52/rad52 (CM-1288) cells. Solid triangles, 0 Gy; open triangles, 250 Gy; solid circles, 10-min postirradiation incubation in YPAD medium at 24°.

Slow DNA rejoining of single-strand breaks in RAD52/RAD52 cells, but not in rad52/rad52 cells, without LH:
After the quick DNA rejoining, the processing and removal of additional lesions require more time before they become substrates for ligation (e.g., MOORE 1982B , MOORE 1988A , MOORE 1990 ; MOORE and LITTLE 1985 ). After 200 Gy irradiation (Fig 2A) and 5 min post-treatment incubation (Fig 2B), sedimentation profiles for RAD52/RAD52 and rad52/rad52 cells showed nearly equivalent shifts toward higher-molecular-weight DNAs, particularly at the leading edges. No significant additional shifts were detectable after 10 min (Fig 2C). By 1 hr, significantly more high-molecular-weight DNAs were apparent in RAD52/RAD52 cells than in rad52/rad52 cells (Fig 2D). In fact, the 1-hr (Fig 2D), 2-hr (Fig 2E), and 4-hr (Fig 2F) profiles for rad52/rad52 cells are quite similar to the 10-min profile. In contrast, the 1-, 2-, and 4-hr profiles of DNAs from RAD52/RAD52 cells showed conspicuous, time-dependent increases in high-molecular-weight DNAs. Thus, rad52/rad52 cells were clearly deficient in the slow rejoining of DNA breaks without LH. This deficiency is also manifested in the widely different sensitivities of RAD52/RAD52 and rad52/rad52 strains to lethal effects of ionizing radiation (Fig 3).

View larger version (28K): In this window In a new window Download PPT slide   Figure 2. Comparisons of sedimentation rates of DNAs from RAD52/RAD52 (—, CM-1293), and rad52/rad52 (— · — · — · —, CM-1288) cells as a function of postirradiation incubation from 5 min to 4 hr at 24° in supplemented synthetic minimal medium containing 0.05 M potassium phosphate buffer. The pH of the medium was 5.0. As for Fig 1, RAD52/RAD52 and rad52/rad52 cells were mixed and suspended prior to 200 Gy of irradiation, and all subsequent experimental procedures were carried out on the mixed suspensions.

View larger version (17K): In this window In a new window Download PPT slide   Figure 3. Survival of RAD52/RAD52 and rad52/rad52 cell populations. Strains CM-1293 and CM-1288 were grown to early stationary phase prior to irradiation. After irradiation, cells were diluted, plated, and allowed to grow on nonsynthetic complete growth medium (YPAD). The means of three independent experiments are plotted.

Double-strand breaks revealed by PFGE:
In PFGE, the spatial separation of individual chromosomes into distinct bands can be obtained according to molecular weight and electric field interaction. Radiation-induced DSBs cause the distinct chromosomal bands to lessen in intensity or disappear in a dose-dependent manner (CONTOPOULOU et al. 1987 ; GAME et al. 1989 ; GEIGL and ECKARDT-SCHUPP 1990 ; DARDALHON et al. 1994 ). Chromosomal bands reappear when the integrity of individual chromosomes is restored. The detection of DSBs by PFGE is quite sensitive, and doses as low as 20 Gy of rays or 0.01 µg/ml bleomycin can be used for mutant strains. DNA breaks increase approximately linearly with increasing doses of rays (MOORE 1982B ) or bleomycin (e.g., bleomycin A₂ or B₂; MOORE 1990 ). Bleomycin produces a less heterogeneous substrate than ionizing radiation for DNA repair.

Fig 4 illustrates the dose-dependent losses in the intensities of individual chromosomal bands that result from the production of bleomycin-induced DSBs. Intact DNA molecules were resolved by size, and chromosomes from untreated cells produced the strongest bands (Fig 4, lanes 1 and 8). After RAD52/RAD52 cells were treated with 1 and 2.5 µg/ml of bleomycin, a substantial fraction of all chromosomal bands was still detectable (Fig 4, lanes 2 and 3) despite the introduction of considerable numbers of DSBs. Above 2.5 µg/ml, individual chromosomes were severely fragmented (lanes 5–7).

View larger version (104K): In this window In a new window Download PPT slide   Figure 4. PFGE illustrating dose-dependent DSBs produced in chromosomes from RAD52/RAD52 cells (lanes 1–7, CM-1293) and rad52/rad52 cells (lanes 8–14, CM-1221). Lanes 1 and 8, 0 µg/ml; lanes 2 and 9, 1 µg/ml; lanes 3 and 10, 2.5 µg/ml; lanes 4 and 11, 2.5 µg/ml, followed by a 2-hr incubation in YPAD. Lanes 5 and 12, 5 µg/ml; lanes 6 and 13, 7.5 µg/ml; lanes 7 and 14, 10 µg/ml.

Comparisons of bleomycin-induced DSBs in RAD52/RAD52 and rad52/rad52 cells in the absence of LH:
Fig 4 Fig 5 Fig 6 compare, respectively, chromosomal bands, PFGE migration profiles, and estimates of the numbers of DSBs produced in chromosomes from RAD52/RAD52 and rad52/rad52 cells. The fragmentation of chromosomes from RAD52/RAD52 cells illustrated in Fig 4 (lanes 1–7) was characteristic and reproducibly obtained for several RAD52/RAD52 strains, but was very different from the fragmentation and apparent reconstruction of chromosomes from rad52/rad52 cells (Fig 4, lanes 8–14). The dose-dependent chromosomal fragmentation caused by 1 and 2.5 µg/ml of bleomycin is apparent in the PFGE migration profiles for the RAD52/RAD52 and rad52/rad52 strains, since the profiles for each strain were highest for untreated cells and lowest after 2.5-µg/ml treatments (Fig 5). Thus, the fragmentation observed after PFGE (Fig 4) corresponds well to the changes in PFGE profiles (Fig 5). Moreover, the profiles allow one to follow what happens to the individual chromosomes within the limits of resolution.

View larger version (28K): In this window In a new window Download PPT slide   Figure 5. Migration profiles for chromosomes from RAD52/RAD52 and rad52/rad52 strains. Profiles are for chromosomal bands in the pulsed-field gel shown in Fig 4. Each strain has a characteristic series of chromosomal bands that can differ from strain to strain because of chromosomal polymorphisms. However, the polymorphisms do not affect the overall results. Dose-dependent chromosomal fragmentation is shown for 0, 1, and 2.5 µg/ml of bleomycin. During the 2-hr incubation, chromosomes were further degraded in RAD52/RAD52 cells, but chromosomes were reconstructed in rad52/rad52 cells. The correspondence between the chromosomes and the profiles is shown for the RAD52/RAD52 strain.

View larger version (25K): In this window In a new window Download PPT slide   Figure 6. Dose-dependent induction of DSBs in chromosomes from RAD52/RAD52 and rad52/rad52 cells. Numbers of DSBs/107 bp were calculated from PFGE analyses (Fig 4 and Fig 5) as described in the text. DSBs could be determined at low concentrations of bleomycin, but reasonable estimates of DSBs at high doses are less reliable because chromosomes were extensively fragmented. Comparable results were obtained for additional RAD52/RAD52 and rad52/rad52 strains.

At low doses, chromosomes were clearly more fragmented in rad52/rad52 cells than in the wild-type cells (Fig 4 Fig 5 Fig 6). At 1 µg/ml of bleomycin (Fig 4, lanes 2 and 9; Fig 5), more than three times more DSBs were produced in chromosomes from rad52/rad52 cells than in those from RAD52/RAD52 cells (Fig 6). An approximately linear increase in DSBs was calculated for 1, 2.5, and 5 µg/ml in RAD52/RAD52 cells, but chromosomes from rad52/rad52 cells were so fragmented in this dose range that reasonable dose-dependent estimates of the numbers of DSBs could not be obtained. For this reason, the actual numbers of DSBs introduced into rad52/rad52 cells are most likely underestimated in comparison to RAD52/RAD52 cells. Nevertheless, the fact that the numbers of DSBs were clearly less in RAD52/RAD52 strains than in rad52/rad52 strains suggests that the processing and repair of DSBs during the actual treatment period is different, and perhaps more efficient, in this dose range in RAD52/RAD52 strains than in rad52/rad52 strains.

In spite of the increased DSBs in rad52/rad52 cells at 1 and 2.5 µg/ml, weak chromosomal bands reappeared (Fig 4, lane 11; Fig 5) and DSBs decreased (Fig 6) in rad52/rad52 cells following a 2-hr post-treatment incubation in the absence of an LH. The reappearance of full-length chromosomes in rad52/rad52 cells suggests that they possess the capacity to reconstruct their chromosomes under these experimental conditions. This reappearance of the chromosomes and decrease in DSBs did not occur in RAD52/RAD52 cells (Fig 4, lane 4; Fig 5 and Fig 6). On the contrary, faint RAD52/RAD52 chromosomal bands were apparent after 2.5 µg/ml (Fig 4, lane 3; Fig 5), but the chromosomes became degraded (Fig 4, lane 4; Fig 5) and DSBs increased approximately twofold after a 2-hr post-treatment incubation without LH (Fig 6).

Furthermore, in contrast to the chromosomes from RAD52/RAD52 cells, the chromosomes from rad52/rad52 cells were far less degraded at the high doses (Fig 4, lanes 12–14) than at the two lowest doses (Fig 4, lanes 9 and 10). Faint bands were evident for chromosomes from rad52/rad52 cells treated at the high doses of 5, 7.5, and 10 µg/ml (Fig 4, lanes 12–14), but not for those from RAD52/RAD52 cells (Fig 4, lanes 5–7). These results further indicated that rad52/rad52 cells, after suffering extreme chromosomal fragmentation, actually reconstructed some of their chromosomes.

Survival of rad52/rad52 cells with and without short LH periods:
Since full-length chromosomes were reconstructed in rad52/rad52 cells during 2 hr without LH after bleomycin treatment (Fig 4 Fig 5 Fig 6), we determined if the 2 hr allowed for increased survival after irradiation or bleomycin treatments. We also determined if LH of the same duration allowed for increased survival. Separate aliquots of cells from the same culture were exposed to irradiation (100 or 200 Gy) or bleomycin (0.5 or 1 µg/ml) in parallel experiments. The 2.5-µg/ml treatment was not included because it kills such a high fraction of rad52/rad52 cells that it could not be compared to 100 or 200 Gy at equitoxic doses.

The results are shown in Fig 7. After both 100 and 200 Gy, survival was highest at 0 hr and decreased after 2-hr post-treatment incubation whether or not a short LH was imposed. In contrast, survival increased to some extent after bleomycin treatments (0.5 or 1 µg/ml bleomycin), except after the 2-hr LH following exposure to 0.5 µg/ml.

View larger version (38K): In this window In a new window Download PPT slide   Figure 7. Effects of 2-hr post-treatment incubation on the survival of rad52/rad52 cells (strain STX434) after irradiation and bleomycin treatment. 0 h, immediately following treatment. Comparable results were obtained with other rad52/rad52 strains.

RAD52/RAD52 cells after limited vs. extensive chromosomal fragmentation and LH:
For the remainder of the studies, chromosomal DSBs were studied after LH to permit processing of lesions and restoration of intact chromosomes. DNA repair was investigated following a bleomycin treatment dose that was so limited that numbers of DSBs and survival were the same as untreated cells, and following a bleomycin treatment dose that caused high numbers of DSBs and killing (Fig 8). In this approach, we took advantage of the fact that yeast cells can be incubated for long periods in water or buffer without acquiring DNA breaks and without significant effects on cell survival. As illustrated in Fig 8, chromosomes from untreated RAD52/RAD52 strains appeared intact after 24 hr (lane 1) or 48 hr (lane 5) of LH.

View larger version (65K): In this window In a new window Download PPT slide   Figure 8. Reconstruction of chromosomes after limited (lanes 1–4) and extensive (lanes 5–7) DNA damage to stationary-phase RAD52/RAD52 cells (strain CM-1293). DSBs were calculated as described in the text. For survival, cells were washed three times, diluted, and plated on YPAD.

Chromosomal bands were prominent after the limited treatment (no LH, Fig 8, lane 2). During the first 4 hr of the LH, chromosomal bands became significantly less intense, although all bands were still visible (data not shown). By 6 hr, chromosomes were extensively degraded, with only low-molecular-weight bands apparent (Fig 8, lane 3). Approximately three DSBs/10⁷ base pairs (bp) were introduced during the 6 hr (lane 3), but 50% of the DSBs were lost between 6 and 24 hr (Fig 8, lane 3 vs. lane 4). All chromosomal bands reappeared by 24 hr (Fig 8, lane 4), although the bands were less intense than before LH or after 4 hr of LH. These results indicated that substantial numbers of DSBs were introduced into chromosomes during the processing of DNA lesions in the first 4 hr of LH, that chromosomes were further fragmented between 4 and 6 hr of LH, and that some full-length chromosomes were reconstructed between 6 and 24 hr.

Table 2 compares numbers of DSBs introduced during 4 and 6 hr of LH after limited and moderate chromosomal fragmentation. Numbers of DSBs were dose dependent (0 hr LH) and increased substantially during the 4-hr LH. After the lower dose, DSBs nearly tripled between 4 and 6 hr. After the higher dose, however, DSBs doubled and reached their maximum after 4 hr, then decreased dramatically by 6 hr.

The repair steps appeared slower after the introduction of only limited DNA lesions (Fig 8, lanes 2–4; Table 2) than following extensive chromosomal breakage (Fig 8, lane 6). Numbers of DSBs were too high after the extreme breakage to reliably compare the estimates of DSBs at 0, 4, and 6 hr. Chromosomes were degraded further between 0 and 4 hr LH, but all chromosomal bands were prominent after 6 hr LH (data not shown). These results are analogous to those after moderate fragmentation (2.5 µg/ml; Table 2), and indicated that chromosomes were reconstituted by 6 hr. In fact, chromosomal bands were nearly as prominent after 6 hr as after 24 hr (Fig 8, lane 7) when ~85% of the DSBs were lost.

This clearly contrasts with the introduction of DSBs during the 6-hr LH after the limited treatment, and suggested repair steps were slower after only minimal DNA damage than after extensive DNA damage.

Survival of RAD52/RAD52 cells during LH:
The extensive reconstruction of chromosomes was surprising, and it was anticipated that the cells would lose viability due to lethal mutations introduced during inaccurate DNA repair. Nevertheless, we examined the role of LH in increasing the probability that RAD52/RAD52 cells survive the extensive DNA damage. Cell viability was measured in parallel with the PFGE analyses and is tabulated in Fig 8 for each of the experimental populations. The degradation of chromosomes during LH was accompanied by decreased survival (Fig 8, lane 2 compared to lanes 3 and 4). Strikingly, however, the loss of 85% of DSBs during the 24 hr was accompanied by a nearly 100-fold increase in survival (Fig 8, lane 6 compared to lane 7).

Dependency of chromosomal reconstruction after 24 and 48 hr LH on the amount of chromosomal fragmentation: a signal for inducing DNA repair after bleomycin treatment:
The fact that chromosomes became very extensively broken during 6 hr LH (Fig 8, lane 3) but weak chromosomal bands reappeared after 24 hr (Fig 8, lane 4) led to the hypothesis that the massive numbers of breaks somehow became a signal or otherwise were a requisite for processing DNA breaks and reconstructing the chromosomes. This hypothesis is supported by finding chromosomal degradation followed by reconstruction during 6 hr LH (Table 2, 2.5 µg/ml) and by the massive chromosomal degradation followed by extensive reconstruction during 24 hr (Fig 8, lanes 6 and 7).

Thus, the next experiments examined chromosomal reconstruction after 24 and 48 hr LH as a function of the amount of chromosomal fragmentation before LH. Low, medium, and high concentrations of bleomycin were employed. Results typical of RAD52/RAD52 cells are shown in Fig 9 and Fig 10. Chromosomes from untreated cells were intact after LH (Fig 9, lanes 1–3 and 13), and the migration profiles of individual chromosomes before and after 48 hr LH were practically the same (Fig 10).

View larger version (139K): In this window In a new window Download PPT slide   Figure 9. Induction of chromosomal repair during 24 and 48 hr LH in RAD52/RAD52 cells. A polymorphism in the region of the largest chromosomes is apparent only among the chromosomes from untreated cells from this diploid strain (CM-1461). EDTA was added to 0.025 M at the end of treatment periods, and cells were washed in 0.05 M EDTA. Lane 1, 0 µg/ml, no LH; lane 2, 0 µg/ml + 24 hr; lane 3, 0 µg/ml + 48 hr; lane 4 (last lane in figure), 1 µg/ml, no LH; lane 5, 1 µg/ml + 24 hr; lane 6, 1 µg/ml + 48 hr; lane 7, 2.5 µg/ml, no LH; lane 8, 2.5 µg/ml + 24 hr; lane 9, 2.5 µg/ml + 48 hr; lane 10, 5 µg/ml, no LH; lane 11, 5 µg/ml + 24 hr; lane 12, 5 µg/ml + 48 hr; lane 13, 0 µg/ml + 24 hr (duplicate of lane 2).

View larger version (31K): In this window In a new window Download PPT slide   Figure 10. Migration profiles for chromosomal bands in the pulsed-field gel shown in Fig 9 (RAD52/RAD52 cells). Profiles are shown for 0 and 48 hr LH after 0, 1, 2.5, and 5 µg/ml of bleomycin. The correspondence between the chromosomes and the profiles is shown.

Chromosomes were fragmented in a dose-dependent manner (Fig 9) after 1-µg/ml (lane 4), 2.5-µg/ml (lane 7), and 5-µg/ml (lane 10) treatments, and the profiles in Fig 10 clearly show this dose dependency. After the lowest treatment concentration, chromosomes were less fragmented before LH (Fig 9, lane 4; Fig 10) than after 24 hr (lane 5) and 48 hr (Fig 10, lane 6). On the other hand, chromosomes were extensively degraded after intermediate treatments (Fig 10, lane 7), but chromosomal bands reappeared after 24 hr (lane 8) and more intensely after 48 hr (Fig 10, lane 9). Finally, after the extreme degradation caused by the highest dose (Fig 10, lane 10), chromosomes also appeared to be reconstructed in a time-dependent manner (Fig 10, lanes 11 and 12). The most intense chromosomal bands appeared after the highest dose and 48 hr LH (Fig 10, lane 12). Thus, chromosomes were further degraded after 1 µg/ml and 48 hr LH, whereas 2.5 and 5 µg/ml stimulated the reformation of chromosomal bands. In fact, the 48-hr migration profiles for 1 and 2.5 µg/ml are nearly indistinguishable (Fig 10). In brief, these results are consistent with the hypothesis that extensive DNA damage provides the signal for inducing DNA repair and chromosomal reconstruction.

"Induced" repair accompanies "induced" recovery in RAD52/RAD52 cells:
Survival of RAD52/RAD52 cells was examined in parallel with the PFGE analyses illustrated in Fig 9 and Fig 10 and is plotted in Fig 11 as a function of dose and length of LH. Chromosomal and survival analyses were carried out on the same cell populations. Killing was dose dependent and highest when cells were plated without an LH. After 24 hr LH, cell viability increased and was highest after 48 hr LH. At the lowest dose, where chromosomal repair appears not to be induced or at least reconstruction does not occur (Fig 9, lanes 5 and 6; Fig 10), survival increased 50% after 24 hr LH and doubled after 48 hr (Fig 11). After the intermediate dose, chromosomes reconstructed during LH (Fig 9, lanes 8 and 9; Fig 10) and survival increased 3.5 times following the 24-hr LH and increased 30-fold (from 2–60%) during 48 hr. After the highest dose, survival increased over threefold after 24 hr and 50-fold during 48 hr LH, during which chromosomes were maximally restored (Fig 9, lane 12; Fig 10). Thus, cells recovered remarkable viability in spite of rearrangements of chromosomes at DSBs and a variety of other types of chromosomal mutations that could have arisen during chromosomal reconstruction.

View larger version (16K): In this window In a new window Download PPT slide   Figure 11. Survival of RAD52/RAD52 cells whose chromosomal bands and migration profiles are shown in Fig 9 and Fig 10.

DNA repair in rad52/rad52 mutant cells exposed to irradiation or bleomycin:
We next investigated if the processing and reconstruction of chromosomes observed in RAD52/RAD52 strains of different genetic backgrounds were exhibited in rad52/rad52 mutant strains. The experiments were conducted by growing rad52/rad52 strains to stationary phase as for RAD52/RAD52 diploids. Cells from the same culture were prepared for irradiation and exposure to bleomycin in parallel experiments. Nearly equitoxic doses of irradiation and the chemical were examined.

The results for one of the rad52/rad52 strains are presented in Fig 12 and Table 3. Clearly, chromosomes in rad52/rad52 strains were more extensively degraded after low doses of bleomycin [ Fig 12 (lanes 1 and 4, STX434); Fig 4 (lanes 9 and 10, CM-1221)] than chromosomes in RAD52/RAD52 cells [e.g., Fig 4 (lanes 2 and 3); Fig 8 (lane 2, CM-1293); Fig 9 (lane 4, CM-1461)]. In spite of the extensive DSBs caused by 0.5 µg/ml in rad52/rad52 cells (Fig 12, lane 1; Table 3), chromosomes were reconstructed to some extent during 24 and 48 hr LH (Fig 12, lanes 2 and 3). Following the higher concentration of 1 µg/ml (Fig 12, lane 4), the chromosomal bands were also reconstructed and were quite prominent after 24 and 48 hr LH (Fig 12, lanes 5 and 6). In fact, the chromosomal bands after 24 hr (Fig 12, lane 5) appeared nearly identical to those from untreated rad52/rad52 cells (Fig 12, lane 7). The DSBs estimated in the rad52/rad52 cells before and after 24 and 48 hr LH are presented in Table 3. Losses of DSBs during LH could not be calculated because DSBs were so high and thus undoubtedly underestimated (Table 3). Following the higher, 1-µg/ml treatment clearly fewer DSBs were present at 24 and 48 hr LH than at 0 hr, indicating that many of the DSBs were removed during the 24 and 48 hr. Thus, in rad52/rad52 cells, as in RAD52/RAD52 cells, more complete (sometimes nearly complete) reconstruction of chromosomes took place after higher, rather than lower, treatment concentrations.

View larger version (147K): In this window In a new window Download PPT slide   Figure 12. PFGE analyses of chromosomes from rad52/rad52 cells. Lane 1, 0.5 µg/ml, no LH; lane 2, 0.5 µg/ml + 24 hr; lane 3, 0.5 µg/ml + 48 hr; lane 4, 1 µg/ml, no LH; lane 5, 1 µg/ml + 24hr; lane 6, 1 µg/ml + 48 hr; lane 7, no treatment; lane 8, 100 Gy, no LH; lane 9, 100 Gy + 24 hr; lane 10, 100 Gy + 48 hr; lane 11, 200 Gy, no LH; lane 12, 200 Gy + 24 hr. DSBs were calculated and are presented in Table 3. The higher dose of irradiation was nearly equitoxic with the higher concentration of the drug (Fig 13).

After irradiation (Fig 12, lanes 8–12; Table 3), chromosomes from rad52/rad52 cells were also efficiently repaired during 24 and 48 hr. Following 100 Gy, chromosomes were extensively fragmented (lane 8), but reappeared prominently after LH (lanes 9 and 10). Chromosomes from rad52/rad52 cells were clearly reconstructed during 24 hr and even more efficiently during 48 hr. After 200 Gy (Fig 12, lane 11), chromosomes were more extensively degraded than after 100 Gy, but reappeared after 24 hr (lane 12). This reappearance was less prominent than after 100 Gy. Lanes 4 and 11 in Fig 12 represent nearly equitoxic treatments (Fig 13), although bleomycin treatments resulted in more DSBs than irradiation (Fig 12). Thus, in contrast to the results obtained for bleomycin, where repair was more efficient after higher than after lower doses, repair of radiation-induced DSBs was higher after exposures to 100 Gy than after 200 Gy of rays.

View larger version (15K): In this window In a new window Download PPT slide   Figure 13. Dose-dependent killing of rad52/rad52 cells (strain STX434) after irradiation and bleomycin treatment. PFGE analyses and DSBs are presented in Fig 12 and Table 3.

Survival of rad52/rad52 cells:
Survival curves for the rad52/rad52 strains (Fig 3 and Fig 13) clearly illustrate the high lethality that rad52/rad52 cells suffer after exposure to bleomycin or irradiation, consistent with increased numbers of DSBs in the PFGE analyses of rad52/rad52 strains compared to RAD52/RAD52 strains. The high killing and numbers of DSBs measured for the rad52/rad52 cells and shown in Fig 12 and Fig 13 and Table 3, respectively, undoubtedly reflect the loss of the capacity for repairing DSBs by homologous recombination. Moreover, while repair of radiation-induced DSBs was higher after exposures to 100 Gy than after 200 Gy, the fact that DNA repair after bleomycin treatments was more efficient after higher than after lower doses quite likely accounts at least in part for the differing shapes of the survival curves after and bleomycin treatments (Fig 13). The reduced rate of killing at high bleomycin doses could be due to the induction and efficiency of DNA repair. The generally higher survival of irradiated cells than bleomycin-treated cells in these studies is consistent with the fact that irradiation produced fewer DSBs than bleomycin.

The magnitude of the increase in survival during LH among RAD52/RAD52 and rad52/rad52 cells was studied and compared and found to depend on the severity of the drug treatment. Recovery was generally less in rad52/rad52 cells than in RAD52/RAD52 cells when compared at the same phase of growth after equitoxic high concentrations, and generally higher in rad52/rad52 cells than in RAD52/RAD52 cells after equitoxic low concentrations (data not shown). The increases in survival after bleomycin doses that killed 90% (D₁₀) and 98% (D₂) of rad52/rad52 cells are presented in Fig 14. The 24- and 48-hr LH permitted recovery, respectively, of 5-fold and >9-fold at D₁₀ and 7- and 10-fold at D₂.

View larger version (42K): In this window In a new window Download PPT slide   Figure 14. Comparisons of survival of rad52/rad52 cells (strain CM-1221) at D10 and D2 (no LH) and after 24 and 48 hr LH.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Reconstruction of intact chromosomes is RAD52 independent:
In this study, cells executed repair steps that were induced or activated after sustaining high numbers of chromosomal DSBs. Chromosomes were reconstructed in diploid RAD52/RAD52 and rad52/rad52 strains. Since the mechanism of restructuring intact chromosomes was not dependent on the RAD52 gene product, it appears not to require homologous recombination. Moreover, the exquisite sensitivity of rad52/rad52 strains to low doses of radiomimetic damage appears to be caused, at least in part, by their failure to adequately inhibit cell cycle progression in response to DSBs in chromosomes. By analogy, mutations in DNA damage checkpoint genes cause cells to inadequately arrest cell cycle progression in response to DNA damage (HARTWELL and KASTAN 1994 ; TOCZYSKI et al. 1997 ; WEINERT 1998A , WEINERT 1998B ). Even chromosomes in RAD52/RAD52 strains greatly benefited from LH. In view of the highly efficient system of homologous recombination in RAD52/RAD52 S. cerevisiae cells, these findings were totally unexpected and indicate the presence of alternative ways of repairing DSBs in mitotic S. cerevisiae cells. Nonrecombinational endjoining—either precise or error-prone—appears to be an important candidate for the overall repair of chromosomal DSBs, particularly when homologous recombination is blocked (MEZARD and NICOLAS 1994 ; SCHIESTL et al. 1994 ; MILNE et al. 1996 ; MOORE and HABER 1996 ; SIEDE et al. 1996 ; TSUKAMOTO et al. 1996 ).

DNA-damage-inducible repair:
We propose that cells execute repair steps that are induced or activated after suffering sufficiently high numbers of chromosomal DSBs. Actually, chromosomal reconstruction occurred in rad52/rad52 cells during the 5-, 7.5-, and 10-µg/ml bleomycin treatments for 30 min, indicating that 24 or 48 hr LH was not required. The quite different responses of the cells to ionizing irradiation and bleomycin suggests that the DNA lesions produced by the two DNA-damaging agents do not all provide the DNA-damage-inducible signal. In fact, the different strengths of DNA lesions for producing the inducing signal are not well understood (SIEDE 1995 ). Chromosomes were less fragmented after ionizing radiation than after equitoxic bleomycin treatments in the current study. This may be due to the differing efficiencies of the two agents in inducing DSBs (reviewed by POVIRK 1996 ). In fact, in mammalian cells bleomycin causes both SSBs and DSBs at a ratio of 5:1 whereas rays are reported to induce 1 DSB for every 20 SSBs (POVIRK 1996 ). X rays were also less potent than bleomycin in inducing whole-chromosome loss in lymphocytes (ODAGIRI et al. 1990 ).

The actual reconstitution of full-length chromosomes may have been facilitated by the association of proteins with chromatin and the fundamental unit of organization of DNA in eukaryotic chromatin, the nucleosome. While genomic DNA prepared for PFGE is deproteinized, broken chromosomes in cells are held together or stabilized by proteins, thereby facilitating recombination by strand annealing and the assembly of complexes to carry out other types of enzymatic repair. Bleomycin preferentially cleaves chromosomes between nucleosomes (KUO and HSU 1978A , KUO and HSU 1978B ; KUO 1981 ; MOORE 1988B , MOORE 1989 ; MOORE et al. 1989 ; SIDIK and SMERDON 1990 ). This contrasts with the more random chromosomal breaks produced by ionizing radiation (WARD 1985 , WARD 1988 ). Thus, although nucleosomes are not released in significant numbers in the dose range employed in the current investigation, a preference for unprotected chromatin together with the DNase-like activity of bleomycin at low doses undoubtedly leave the damaged chromosomes and the ends where DSBs occur in a structure different from that in chromosomes damaged by ionizing radiation.

A prolonged LH period appears to increase the window for DNA repair complexes to access and process the chromosomal lesions. Without LH, diploid rad52/rad52 cells were blocked in the repair of DNA DSBs produced after ionizing radiation in previous studies by HO 1975 , RESNICK 1975 , and RESNICK and MARTIN 1976 using neutral sucrose gradient procedures, and by CONTOPOULOU et al. 1987 using PFGE. The optimum LH would depend upon several factors, including the nature and extent of the damage, chemistry and configuration of the DSBs, genotype of the cells, and proficiency of the strain in carrying out the necessary DNA repair steps. In this respect, RAD52/RAD52 strains and rad52/rad52 strains responded similarly to LH (Fig 8 Fig 9 Fig 10 and Fig 12; Table 2 and Table 3), indicating that the activation of the pathway(s) for reconstructing the chromosomes is RAD52 independent. This activation was observed in a variety of strains of different genetic backgrounds and was consistent from strain to strain.

Some of the chromosomal lesions produced after ionizing irradiation were immediately ligatable and led to the quick component of DNA rejoining in both RAD52/RAD52 and rad52/rad52 cells (Fig 1 and Fig 2). These lesions most likely were the direct SSBs or those resulting from rapid processing of the DNA damage and were handled equally well by RAD52/RAD52 and rad52/rad52 cells. We further suggest that the unrepaired breaks in rad52/rad52 cells in the absence of LH are those that would be repaired by the dominant homologous recombination pathways in yeast. Following LH, chromosomal bands were more intense after 100 Gy than after 200 Gy (Fig 12, lanes 9 and 10 vs. lane 12), raising the question of whether chromosomal fragmentation was caused by SSBs that can be repaired without LH. However, this was clearly shown not to be the case when SSBs were assayed without LH since the slower "components" of DNA repair were handled efficiently after irradiation by RAD52/RAD52 cells, but rad52/rad52 cells failed to carry out the slow repair of DNA breaks (Fig 2).

A threshold of DSBs appears to provide the critical signal for activating the repair steps:
The actual signalling event for activating the steps required for processing chromosomal damage and rebuilding the chromosomes may be a particular threshold of unrepaired DSBs. In the absence of a critical threshold of DSBs, the pathway or signal is likely to be repressed. The cell thereby conserves energy in the absence of extensive DNA damage. The signal for the activation or induction of the repair pathway appears to occur after lower treatment doses in rad52/rad52 cells than in RAD52/RAD52 cells, supporting the model that chromosomal degradation provides the signal for inducing the repair steps. The signal probably depends on DNA breaks that can be induced either directly or indirectly during processing of lesions within or at ends of broken DNA molecules. The signal may actually be activated (or derepressed) during exposures to DNA-damaging agents since chromosomal bands were prominent after the highest treatment doses in rad52/rad52 cells, but not in RAD52/RAD52 cells (Fig 4 and Fig 5). The time required for reconstructing the chromosomes was remarkably consistent for a particular strain.

Why would cells have such a mechanism? Certainly, repairing chromosomes increases the likelihood, but does not ensure, that the functionality of damaged genes will be restored. In turn, the likelihood that cells will survive would also increase. Thus, the driving force for the cell is survival, and time is undoubtedly limited for rescuing cells from death when their chromosomes are severely damaged. In the absence of damaged chromosomes, the proteins required for processing and repairing the damaged chromosomes need not be functional.

Relationship of DSB repair to survival:
The relationship of the repair of DSBs to survival is complex (MOORE and LITTLE 1985 ; MOORE et al. 1985 ; RADFORD 1985 , RADFORD 1986 ; FRANKENBERG-SCHWAGER 1990 ; MOORE 1990 ; ILIAKIS 1991 ). For example, it is possible that some cells put together faulty chromosomes and survive even though they do not restore normal DNA sequences. Or, cells may die because they reconstruct chromosomes at the expense of producing chromosomal rearrangements, deletions, aneuploidy, and other types of mutations. At a given dose of DNA-damaging agent in the current study, DSBs and killing were higher in rad52/rad52 cells than in RAD52/RAD52 cells. To our surprise, however, the reconstruction of chromosomes during LH led to increases in survival of both genotypes (e.g., Fig 7, Fig 8, Fig 11, and Fig 14). Actually, the recovery of cells was associated with the reconstruction of chromosomal DNAs during LH, and fragmentation of chromosomes during LH was accompanied by decreased survival. This correlation was not expected at the outset of these studies because we