|Home | About | Journals | Submit | Contact Us | Français|
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact email@example.com
Homologous recombination (HR) deficient cells are sensitive to methyl methanesulfonate (MMS). HR is usually involved in the repair of DNA double-strand breaks (DSBs) in Saccharomyces cerevisiae implying that MMS somehow induces DSBs in vivo. Indeed there is evidence, based on pulsed-field gel electrophoresis (PFGE), that MMS causes DNA fragmentation. However, the mechanism through which MMS induces DSBs has not been demonstrated. Here, we show that DNA fragmentation following MMS treatment, and detected by PFGE is not the consequence of production of cellular DSBs. Instead, DSBs seen following MMS treatment are produced during sample preparation where heat-labile methylated DNA is converted into DSBs. Furthermore, we show that the repair of MMS-induced heat-labile damage requires the base excision repair protein XRCC1, and is independent of HR in both S.cerevisiae and mammalian cells. We speculate that the reason for recombination-deficient cells being sensitive to MMS is due to the role of HR in repair of MMS-induced stalled replication forks, rather than for repair of cellular DSBs or heat-labile damage.
The DNA alkylating agent methyl methanesulfonate (MMS) has been used for many years as a DNA damaging agent to induce mutagenesis and in recombination experiments. MMS modifies both guanine (to 7-methylguanine) and adenine (to 3-methlyladenine) to cause base mispairing and replication blocks, respectively (1). DNA damage caused by alkylating agents is predominantly repaired by the base excision repair (BER) pathway and DNA alkyltransferases (2). The sensitivity of cells to MMS also increases significantly when other DNA repair pathways are compromised. For example, in Saccharomyces cerevisiae disrupting the homologous recombination (HR) pathway by mutating genes in the RAD52 epistasis group significantly increases sensitivity to MMS (3). The sensitivity of HR mutant cells to MMS has lead to this agent being called an ionizing radiation mimetic or a DNA double-strand break (DSB) agent, for many years including in some recent publications. More direct evidence that MMS causes DSBs comes from analyses of yeast chromosomes by pulsed-field gel electrophoresis (PFGE), in which treatment with low levels of MMS leads to fragmentation of yeast chromosomes (4,5). Furthermore, MMS induces both inter-chromosomal and intra-chromosomal recombination (6). Other S.cerevisiae genes with a wide range of functions are also known to protect the cell from MMS-induced damage. Interestingly, MMS1 to MMS5 and MMS22 were originally identified in a screen for genes affecting MMS sensitivity but not sensitivity to ionizing radiation (7). Thus, the MMS genes are unlikely to be required for DSB repair in general. RAD52 is epistatic to both MMS1 and MMS4 for MMS sensitivity, implying that these genes might act in the same pathway (8,9). One possibility is that the MMS2 and MMS4 are involved in recognizing a specific subset of DSBs leading to the recruitment of RAD52. It has been suggested that DSBs arise after MMS treatment when, during BER, single-strand breaks (SSBs) are encountered by a replication fork. These replication-associated DSBs would serve as a substrate for HR (10,11). To check if the role of the HR system in MMS-resistant cells is to repair S-phase-specific DSBs, we analysed DNA samples from mammalian and yeast cells treated with MMS or N-methyl-N′-nitro-N-nitrosoguanidine (MNNG). We find that recently replicated DNA from MMS- or MNNG-treated mammalian cells does not contain DSBs. Further to this, experiments with both mammalian and yeast systems indicate that neither MMS nor MNNG lead to DSBs in vivo but, when such DSBs are detected by PFGE, they are an experimentally induced in vitro artefact.
The EM9, irs1SF and V3-3 cell lines all originate from AA8 Chinese hamster ovarian (CHO) cells. EM9 has a mutation in the XRCC1 gene and has a defect in BER (12,13). The V3-3 is mutated in the XRCC7 gene resulting in a deficiency in DNA-PKcs and impaired non-homologous end-joining (NHEJ) (14). The irs1SF is defective in the XRCC3 gene resulting in deficiency in HR (15,16). CXR3 is an XRCC3 corrected irs1SF cell line (17). The HCT116 cell line was obtained from American Type Culture Collection (Manassas, VA). All cell lines were cultured in DMEM, with the addition of 9% fetal calf serum and penicillin–streptomycin (90 U/ml) at 37°C and 5% CO2 atmosphere.
MMS and MNNG was dissolved in phosphate-buffered saline and dimethyl sulfoxide just before use (DMSO; the treatment dose of MNNG did not exceed 0.2% of DMSO), respectively. Treatments with MMS and MNNG were performed in Hank's balanced salt solution (HBSS, GIBCO), unless otherwise indicated. All solutions were prepared just before treatment. γ-irradiation was performed in a 137Cs chamber (10.6 Gy/min).
Fresh MMS (Sigma, 100% purity) was added directly to yeast cultures in standard YEPD to a final concentration of 0.05 or 0.1%. Cells were shaken at 30°C until sampling when they were washed twice in fresh YEPD before use in assays described below.
Recombination was measured between arg4-nsp and arg4-bgl alleles by measuring the frequency of ARG4 cells in treated (0.05% MMS for 30 min) and untreated populations of SK1 diploid cells [construct described fully in (18,19)]. Cells were sonicated briefly to break down clumps, and plated onto both rich medium and arginine dropout plates immediately after exposure to MMS. The recombination frequency is the fraction of colony forming units on arginine dropout plates compared to rich medium. The frequency of ARG4 is a conservative estimate of the number of gene conversion events as it does not account for gene conversion to the opposite mutant allele or the double mutant allele arg4-nsp,bgl.
Viability was estimated by plating serial dilutions wild-type, rad52::KanMX and apn1::KanMX haploid BY4741 cells from the Euroscarf deletion collection on YEPD solid medium following exposure to 0.05% MMS for 0–90 min.
In the toxicity assay, 500 cells were plated onto a Petri dish ( 100 mm) 24 h prior to a 0.5 h treatment with MNNG or MMS. Following treatment, plates were rinsed three times with 10 ml HBSS (GIBCO) and 10 ml medium was added. After 7–12 days, when colonies were observed, the plates were harvested and the colonies were fixed and stained using methylene blue in methanol (4 g/l). Colonies containing >50 cells were counted.
Flasks were inoculated with 4 × 106 cells for 4 h prior to a 24 h treatment with hydroxyurea or etoposide, or 28 h prior to a 0.5 h treatment with MMS or MNNG. After treatment, the cells were released from the flask by trypsinization and 1 × 106 cells were set into each agarose plug (75 μl, 1% InCert Agarose, BMA). γ-irradiation (137Cs, 10.6 Gy/min) was performed after cells had been set into the agarose plugs. Inserts were incubated in 0.5 M EDTA, 1% N-laurylsarcosyl and proteinase K (1 mg/ml) at 50 or 20°C for 48 h, and thereafter washed four times in TE-buffer prior to loading onto an agarose separation gel (1% Chromosomal grade agarose, Bio-Rad). Separation was performed on a CHEF DR III equipment (BioRad; 120° field angle, 240 s switch time, 4 V/cm, 14°C) for 18 h. The gel was stained with ethidium bromide overnight and subsequently analysed by a scanning fluorescence reader (FLA-3000, Fujifilm) using Image Gauge software.
In the 14C-thymidine labelling experiments, flasks were inoculated with 2 × 106 cells 27.5 h prior to a 30 min incubation with 14C-thymidine (4.39 μM, 9.25 kBq/ml) or with 2 × 106 cells 4 h prior to 14C-thymidine labelling (0.439 μM, 0.925 kBq/ml) for 24 h. Flasks were rinsed twice with 10 ml HBSS (GIBCO) before initiation of a 30 min treatment with MMS or MNNG or a 24 h treatment with hydroxyurea. After treatment, cells were released by trypsinization, melted into agarose inserts as described above and separated on PFGE for 24 h (BioRad; 120° field angle, 60–240 s switch time, 4 V/cm, 14°C). After separation, the DNA was transferred from the gel to a nylon membrane according to the manufacturer's protocol (Hybond-N, Amersham Pharmacia Biotech). The membrane was then dried for 2 h at 80°C and exposed onto a phosphoimager plate (FujiFilm) for 18 h before quantification employing Image Gauge software (FLA-3000, Fujifilm).
When treating isolated DNA with MMS, untreated AA8 cells were set in agarose plugs and incubated in 0.5 M EDTA, 1% N-laurylsarcosyl and proteinase K (1 mg/ml) at 50°C for 48 h and thereafter washed as described above. The agarose plugs with naked DNA was then treated with MMS 1 mM in HBSS for 30 min before incubation in 0.5 M EDTA, 1% N-laurylsarcosyl and proteinase K (1 mg/ml) at 50 or 20°C for 48 h. Samples were rinsed and separated by PFGE for 18 h as described above.
When studying repair of heat-labile sites, 1 × 106 cells were melted into each agarose plug (75 μl, 1% InCert Agarose, BMA) and treatment was performed by transferring the agarose plugs to HBSS++ containing 1 mM MMS for 30 min at 37°C. Following treatment, agarose plugs washed three times in ice-cold HBSS++ and then transferred to DMEM and kept on a turning table at 37°C. At given time points, agarose plugs were incubated in 0.5 M EDTA, 1% N-laurylsarcosyl and proteinase K (1 mg/ml) at 50°C for 48 h, treated as described above and separated on PFGE for 18 h as described above.
A single colony of haploid yeast was used to inoculate a 5 ml YEPD liquid culture grown overnight at 30°C. The whole culture was transferred to 250 ml of YEPD and grown overnight at 30°C. MMS (0.05 or 0.1%) was added for the times indicated in the figures, with shaking at 30°C followed by washing the cells twice in 250 ml of fresh YEPD. The cells assayed for repair of heat-labile damage were maintained at high density in stationary phase and sampled for PFGE at regular intervals. Plugs for PFGE were prepared as soon as the cells were sampled and stored in 50% glycerol/1× TE at −80°C until the final time point was sampled. All plugs were subsequently treated with Proteinase K (1 mg/ml) at 55 or 30°C for 24 h. Samples were rinsed and subjected to PFGE (BioRad; 120° field angle, 6 V/cm, initial switch time of 19 s, final switch time of 160 s for 26 h at 11°C).
Southern transfer from PFGE was undertaken in standard denaturing conditions (0.4 M NaOH) after soaking the gel in 0.25 M HCl for 15 min using ZetaProbe membrane (Biorad). The DNA probe specific to the SPO11 open reading frame on chromosome VIII was prepared from a plasmid (pAG191) by restriction enzyme digestion and gel purification. The probe was labelled with 32P using the HighPrime random primer labelling kit (Roche) as the manufacturer recommends.
Under the assumption of a Poisson process for the formation of alkylations, the positions of the alkylations were uniformly distributed along chromosomes. For each chromosome, a random number was generated for each alkylation event to determine its position. Our own in-house developed random number generator was used to undertake this. It was assumed that half of the total number of alkylation events occurred on each DNA strand. The number of times that the breaks on the two strands were separated by 14 bp or less was recorded using a purpose-written C programme. A total of 1000 replicates were simulated for each of four conditions. Under two conditions, the alkylation rate for MMS was used and it was assumed that either 100 or 10% of modified bases would be converted to an SSB in vitro. The other two conditions were set for the alkylation rate for MNNG with the same efficiencies of conversion to SSBs. The mean number of DSBs for each of the four cases is shown in Table 1.
It is well established that HR-deficient S.cerevisiae strains are hypersensitive to alkylating agents such as MMS (3). To test if HR-deficient mammalian cells are also hypersensitive to alkylating agents, we investigated cell survival in a wild-type (AA8) and a HR-deficient Chinese hamster cell line [irs1SF (16,17)] to MMS and MNNG. We also treated XRCC1-deficient EM9 cells (20), that are partially BER deficient (21), and DNA-PKcs-deficient V3-3 cells (14). We found that irs1SF cells were 7-fold more sensitive to MMS than wild-type cells and 9-fold more sensitive to MNNG (Figure 1). To ensure that this difference in sensitivity is due to deficiency in HR, we used an irs1SF cell line that is complemented with a functional hXRCC3 gene on a cosmid vector, CXR3 (17). The sensitivity to both MMS and MNNG in irs1SF cells is partially reverted by the human XRCC3 gene in Chinese hamster cells (Figure 1). The incomplete complementation by the human XRCC3 gene in hamster cells has previously been reported (17,22). We found that the NHEJ-deficient cell line V3-3 is insensitive to MNNG and only slightly sensitive to MMS (Figure 1), which is interesting given that NHEJ is recognized to repair the majority of DNA DSBs in mammalian cells (23–25), and yKu80 mutant cells are MMS sensitive (26). The EM9 cell line was found sensitive to MMS and MNNG consistent with what was previously reported (13).
We examined the DNA isolated from cells treated with alkylating agents for evidence of DNA DSBs, a candidate lesion leading to increased HR. To do this, DNA was displayed by PFGE and compared to positive control samples from cells treated with etoposide (VP-16), hydroxyurea (HU) or ionizing radiation, all known to induce DSBs (27). Under the conditions used, chromosomal DNA remains in the wells and shorter DNA molecules, arising from the presence of DSBs, enters the gel creating a smear. Initial results using standard preparation for PFGE gave the impression that both MMS and MNNG produce high levels of DSBs (Figure 2A). However, because sample preparation for PFGE commonly involves treatment with proteinase-K at 50°C, and alkylated bases on DNA are known to be heat-labile (28,29), we investigated this issue further. Our concern was that the DSBs could be formed by spontaneous hydrolysis of alkylated bases to apurinic/apyrimidinic (AP) sites (28). These AP sites are, in turn, heat-labile and can be transformed into DNA SSBs (29,30). Subsequent alkylation on the opposite strand (and SSB conversion), and sufficiently close to the first alkylation, would result in a DSB.
After incubation with proteinase K at 20°C for 48 h (Figure 2A), positive control AA8 cells treated with etoposide (VP-16), hydroxyurea (HU) or IR, revealed smearing due to the presence of DSBs, as expected from previous findings (27,31). In contrast, no smearing was detectable for MMS or MNNG treated in samples prepared in 20°C (Figure 2B). This indicates that MMS and MNNG form lesions, in vivo, that are heat-labile but, at this level of resolution no in vivo DSBs are apparent.
To confirm that the sensitivity to MMS and MNNG in the EM9 and irs1SF cell lines are not due to the induction of DSBs, we compared induction of MMS- and MNNG-induced lesions in CHO cells deficient in different repair pathways. Also, chromosomal aberrations and sister chromatid exchanges (SCEs) formed after MMS and MNNG treatment have been suggested to arise from repair of DSBs produced by futile repair of O6-MeG by mismatch repair (MMR) proteins (32,33). To evaluate if release of DNA fragments after MMS and MNNG treatment is dependent on MMR, we also treated the MLH1-deficient human fibroblast cell line, HCT116, to the agents. All cell lines showed a similar induction of DNA fragments after both MMS and MNNG treatment as seen in Figure 2C (lanes 1–5; 11–15). DNA fragments were released in HCT116 cells as well, indicating that MMR is not sufficient for formation of these lesions. In samples incubated at 20°C (Figure 2C, lanes 6–10 and 16–20), no induction of DSBs were seen in either cell line.
To confirm that these results were not specific to mammalian cells, similar experiments were undertaken in yeast. Consistent with the results from mammalian cells, we found profound breakage of the DNA following preparation for PFGE at 55°C (Figure 3A), with significant repair of heat-labile sites being evident 16 h after MMS exposure. No smearing was detected for DNA from MMS-treated S.cerevisiae when preparing PFGE plugs at 30°C (Figure 3B). To enhance the sensitivity of the assay, the DNA was transferred onto a membrane for Southern hybridization using a probe specific to chromosome VIII. Still, we found no evidence of breakage of chromosome VIII following MMS treatment, supporting the view that DSBs are not formed in vivo following MMS treatment at 30°C.
One possible explanation for the dearth of DSBs after treatment at 30°C was because insufficient numbers of DSBs had been created by the level of MMS used (0.05%). To maximize our chances of capturing MMS-induced DSBs, rad52 mutant cells (which are unable to repair DSBs by HR) were exposed for up to 90 min to either 0.05% MMS or 0.1% MMS (Figure 4). The chromosomes of cells treated with 0.05% MMS for only 30 min were severely degraded when prepared for PFGE using a 55°C incubation step (Figure 4A). In contrast, no low molecular weight smearing was obtained for cells treated with either 0.05% MMS or 0.1% MMS for up to 90 min with sample preparation at 30°C (Figure 4B). To assess the impact of MMS on the genome, we measured HR as the appearance of Arg4+ prototrophs in diploids containing inserts of arg4–nsp and arg–bgl alleles (18,19). Exposure to 0.05% MMS for 30 min increased gene conversion to ARG4 about 6.5-fold compared to untreated cells (Figure 5A). Lastly, we checked that, as expected, rad52 cells were sensitive to MMS in our hands (Figure 5B). With this confirmed, we are confident that the MMS treatment is causing DNA damage, and the failure to detect severely damaged chromosomes by PFGE for samples treated at 30°C is because in vivo DSBs do not arise from exposure to MMS.
We reasoned that the lack of DSBs found in samples treated at 20°C could reflect the fact that they are specifically associated in vivo with replicating DNA, which forms a small proportion of the total DNA examined. To enhance our ability to detect replication-associated DSBs, we labelled nascent DNA in vivo by incorporation of 14C-thymidine for 0.5 h or genome-wide DNA for 24 h labelling in wild-type Chinese hamster AA8 cells. Following labelling, cells were treated with MMS, MNNG, HU or ionizing radiation. A high level of labelled DNA fragments entered the gel following incubation of the plugs at 50°C (Figure 6A). For the cells treated with a 0.5 h pulse of 14C-thymidine, labelling is constrained to recently replicated, nascent DNA. Nascent DNA associated with replication forks is expected to form bubble structures preventing it from entering the gel (34). DSBs forming within these replicons, or close to replication forks, releases them into the gel (27,31). Only a small amount of nascent DNA fragments (0.5 h 14C-thymidine labelling) were released following MMS or MNNG treatments. Thus, most of these DSBs do not release the labelled DNA within replication bubbles, presumably because the heat-labile damage is spread throughout the genome and not restricted to replication forks. The same small release of nascent DNA fragments are also found following γ-ray treatments, consistent with random distribution of IR-induced DSBs. In contrast, HU releases large amounts of nascent DNA fragments into the gel, consistent with suggestions that HU produces DSBs close to replication forks (31).
For samples treated at 20°C, HU and IR treatments lead to labelled DNA being released into the gel (Figure 6B). It is particularly important to note that no nascent DNA was released into the gels by MMS or MNNG when cells were pulse labelled and treated at 20°C. This result suggests that alkylating agents do not induce replication fork associated DSBs in vivo.
Taken together, the data indicate that DSBs produced following MMS treatment are an in vitro artefact following a 50°C or 55°C treatment of the heat-labile alkylated DNA. To confirm that appearance of DSBs is a purely chemical process without biological relevance at physiological temperatures, we embedded control cells in agarose and stripped these with Proteinase K for 48 h. The naked DNA was treated with 1 mM MMS for 30 min (i.e. 0.5 mM h) and samples were incubated at 50°C or 20°C for 48 h, prior to analysis with PFGE. DSBs were induced by MMS in DNA exposed to 50°C, while no DSBs could be detected in DNA exposed to 20°C (Figure 7).
MMS or MNNG-induced heat-labile methylated bases may convert to a SSB following incubation at 50°C (29,30). One possibility is that DNA fragments visualized by PFGE are only released when two heat-labile methylated bases are in close proximity, and on opposite strands (so that two SSBs, are converted into a single DSB). By this argument, the amount of DSBs induced by methylations will depend on how the damage is distributed. Clustered alkylations would convert to high levels of DSBs, whereas randomly distributed damage would convert lower levels of DSBs.
We calculated the DNA alkylation grade following MMS or MNNG treatments using DNA reactivity data for MMS and MNNG (35–38). We found that 0.044% and 0.014% bases will be alkylated with MMS and MNNG, respectively, under the conditions used here for mammalian cells (Table 1).
Based on a random distribution of methylations, we calculated the expected number of DSBs that would be produced by MMS or MNNG. In the simulation, we hypothesized that 100 or 10% of the alkylations would give rise to a heat-labile site (Table 1). We also assumed that there is equal methylation on opposite strands and that heat-labile sites must be within 14 bases of each other on opposite strands to be converted into a DSB. According to the simulations, after in vitro heat treatment, MMS (1.5 mM h) would lead to 17364 DSBs per cell if all alkylated sites are converted by heat into SSBs or 173 DSBs if only 10% of alkylated sites are converted by heat into SSBs. Similarly, MNNG (5 μM h) would lead to, respectively, 1653 DSBs or 17.3 DSBs at 100% and 10% efficiency of damage conversion to SSBs (Table 1). The DNA fragments released by PFGE following MMS and MNNG treatments (and 50°C incubation) are equivalent to a treatment with ~200 Gy, producing ~8000 DSBs (Figure 2) (39,40). The predicted number of DSBs from conversion of heat-labile sites close on opposite strands is in 2-fold excess for MMS at 100% efficiency, but 5-fold lower for MNNG. From this different implications for the distribution of damage induced by MMS and MNNG can be considered. The inference is that damage caused by MMS could be randomly distributed to generate the quantity of heat-labile sites, demonstrated from the amount of low molecular weight DNA released. In contrast, MNNG damage sites appear to be clustered, otherwise based on our calculations MNNG heat-labile sites would not be sufficiently close to each other to liberate the amount of DNA fragments detected after heat treatment.
Next, we set out to test whether or not the heat-labile damage caused by MMS is responsible for HR sensitivity to alkylating agents. To test this, we examined the DNA of wild-type and HR compromised cells treated with MMS. For wild-type mammalian cells, repair of the heat-labile sites was apparent by the gradual disappearance of DSB DNA band (>2 Mbp) from cells incubated at 50°C (Figure 8A). For stationary phase wild-type yeast cells, similar repair was detected by the gradual increasing molecular weight of the smear and reappearance of chromosome-specific bands (Figure 3A).
Although the XRCC3 mutant cell line is severely compromised for HR (15,41) and sensitive to MMS (Figure 1A), this mutation had very little impact on the loss of heat-labile moieties, such that the DSB band from these cells disappeared with similar kinetics as wild-type cells (Figure 8). Similarly, Δrad52 yeast cells can regain whole chromosomes at virtually the same rate as wild-type cells (Figure 9). These data indicate that the hypersensitivity of HR-deficient cells is unlikely to be due to the presence of either DNA DSBs or loss of repair of heat-labile damage caused by MMS.
NHEJ is the quickest and most commonly used DSBs repair pathway in mammalian cells (23,24). If MMS produced an excessive amount of DSBs, one would imagine that the repair of these DSBs would be slow in an NHEJ-deficient cell line. To test this, we determined the repair of MMS-induced heat-labile sites in the DNA-PKcs deficient V3-3 cell line. We found that V3-3 cells repair MMS-induced heat-labile sites as efficiently as the wild-type HR deficient cells (Figure 8).
Proteins in the BER pathway play an important role in processing alkylated bases as well as clustered damage sites in vitro (42). We used the Chinese hamster cell line EM9, deficient in XRCC1 (12,20) to test if the repair of MMS-induced heat-labile sites, possibly multiple damage sites, is dependent on this enzyme. We found that ~40% of MMS-induced heat-labile sites remained in EM9 cells following 24 h, while most were repaired in wild-type cells (Figure 7B). This result implicates the XRCC1 protein in repair of MMS-induced heat-labile, possibly multiple damage sites. We also determined the repair of MMS-induced heat-labile sites in the S.cerevisiae apn1Δ mutant, deficient in the major yeast apurinic endonuclease (43). In contrast to EM9 cells, we found no defect in the repair of MMS-induced heat-labile sites in apn1Δ mutants (Figures 6B and and99).
Methylating agents methylate DNA mainly on oxygen or nitrogen residues present in either the DNA backbone or in the DNA bases. Methylated bases are efficiently repaired by BER or DNA methyltransferases (2). Further to this, alkylating agents induce both SCE and HR (6,44–47). The significance of this response to DNA alkylation is revealed by the well-established fact that budding yeast HR mutants are hypersensitive to alkylating damage, which we confirm to be the case also with mammalian cells deficient in HR. The HR-deficient irs1SF cell line is more sensitive to MNNG than the BER-deficient EM9 cell line, whereas the EM9 and irs1SF cell lines show similar sensitivity to MMS. One explanation for this difference could be that MNNG methylates the O6-guanine position much more efficiently than MMS, and O6-methyl guanine is linked with induction of HR (35,48,49).
Since it is well established that HR is involved in repair of alkylation-induced damage in yeast, such agents are regularly used in experimental systems to induce DNA damage responses for studying HR. Early reports using sucrose density sedimentation suggested that high MMS doses were required to induce DSBs in yeast, thought due to the proximity of multiple SSBs on opposite strands (50,51). More recently, low doses of MMS have been shown by PFGE to induce DSBs (5) and various authors continue to refer to MMS as a radiomimetic or DSB-inducing agent. A possible explanation for published PFGE data (5) was that DSBs arise during replication, when a replication fork encounters an MMS-induced SSB, causing the replication fork to collapse into a DSB (10,11). Replication fork collapse is known to trigger break-induced replication by HR in both yeast and mammalian cells (52,53). Our experiments were designed to search for such DSBs following exposure of mammalian and yeast cells to alkylating agents. Here, we present evidence that no such DSBs form in vivo following treatment with alkylating agents. First, chromosomes were not fragmented when analysed by PFGE after preparation at low temperatures in either wild-type or rad52 cells. Second, a higher resolution analysis of nascent mammalian DNA revealed that replicons remain trapped in the wells of agarose gels as long as heat-labile moieties remained stable, implying that they are not associated with DSBs. On the other hand, converting heat-labile MMS damage to DSBs by treating samples at 50°C permitted nascent DNA to travel into the gel. Similarly, DSBs induced in vivo by IR or HU also released replicons into the gel.
We did find extensive DNA fragmentation following preparation of the samples for PFGE at 50°C. Since isolated DNA can also be fragmented by exposure to MMS in vitro and heating to 50°C, we suggest that no enzymatic activity is required for MMS-associated DSB formation, but such DSBs are an experimental artefact. These DSBs may form in vitro following spontaneous hydrolysis of alkylated bases to AP sites (28), and heat-induced transformation into SSBs (29). Such SSBs could then be converted to DSBs through two routes. Either the SSBs are at sufficient density to be very close to each other on both strands (30) or SSBs can be formed on ssDNA of the lagging strand of replication forks. Significantly fewer nascent DNA fragments (after treatment at 50°C) are released by MMS and MNNG compared to HU. Therefore, the data support the view that the distribution of MMS and MNNG-induced heat-labile lesions is not concentrated in single-stranded DNA regions of replication forks. Computer modelling lead to the conclusion that, assuming 100% efficiency conversion of heat-labile sites to SSBs, SSBs separated by 6 bp (for MMS) or 70 bp (for MNNG) would be required to yield DSBs if the alkylated damage is randomly distributed. These calculations support the view that damage caused by MMS may be randomly distributed. For MNNG, the distance between randomized sites is almost certainly too great to explain the amount of DSB DNA generated by heat treatment, thus it is more likely that damage caused by MNNG is clustered.
While our data does not support that MMS or MNNG directly induce in vivo DSBs, HR repair may still be involved in the repair of heat-labile sites. To test this, we investigated the repair of MMS-induced heat-labile sites in wild-type and repair-deficient cell lines. For both mammalian and yeast cells, these sites could be repaired as quickly in HR-deficient cells as wild-type cells. The data suggest that, in mammalian cells, BER is involved in repair of MMS-induced heat-labile damage sites, since these are persisting for longer time in the XRCC1-deficient EM9 cell line. This is consistent with data showing that proteins involved in short-patch repair are important in repair of multi-damaged sites in vitro (42). It is not clear why we found no heat-labile repair defect in yeast apn1 mutants, though it could reflect the known partial redundancy between Apn1 and Apn2 (54). Another possibility is that heat-labile alkylated DNA bases are primarily repaired by bifunctional DNA glycolsylases in S.cerevisiae. Interestingly, expression of a human DNA glycolsylase in S.cerevisiae increase resistance to MMS (55).
Why then are HR-deficient cells hypersensitive to alkylating agents? One possibility is that at physiological temperatures, the number of alkylated sites converted to SSBs and then DSBs are too few to detect, but are sufficient to elicit a requirement for HR. If this is true, it is still important to note that studies based on analysis of chromosome fragmentation using PFGE are examining repair of heat-labile sites and not in vivo DSBs. Others have previously suggested that the HR may be required to correct various forms of DNA damage other than DSBs (56,57). One possibility is that stalling replication forks at sites of alkylated bases reverse to form a chicken foot including a Holiday junction (HJ), as has been suggested to occur at stalled replication forks (58–61). Alternatively, other structures resembling HJs may form at stalled replication forks that could mediate HR repair of MMS-induced damage (62,63). This notion is supported by the known sensitivity of mus81 mutant yeast to MMS and the need for a helicase, such as Srs2 or Sgs1, for both RAD52-mediated MMS resistance and removal of MMS-induced S-phase cruciform structures (62,64,65).
We wish to thank Dr Larry Thompson for providing the AA8, EM9, V3-3 and irs1SF cell lines. We thank Profs Mark Meuth and Chris Cannings for discussions. This investigation was supported by the Swedish Cancer Society, the Swedish Research Council, the Swedish Agency for Animal Welfare, the Biological and Biotechnological Sciences Research Council and Yorkshire Cancer Research. Publication charges were supported by JISC.
Conflict of interest statement. None declared.