Genetic Analysis of Repair and Damage Tolerance Mechanisms for DNA-Protein Cross-Links in Escherichia coli^{▿ §}

Strains, plasmids, and media.

All strains used in this study are derivatives of E. coli K-12 (1, 6-8, 13, 17, 19, 27, 29, 30, 34-37, 48, 53, 60, 63, 67, 68, 71, 72, 76, 78, 79). The relevant genotypes of the strains and the properties of plasmids are listed in Table 1. A portion of the strains in Table 1 was constructed in this study. The strains deficient in recQ (NKJ1514) and recJ (NKJ1515) were made via P1 transduction of recQ::Tn3 from KD2250 (53) and recJ::Tn10 from BIK814 (71) into the AB1157 recipient strain, respectively. Transductants were selected for ampicillin resistance (Amp^r) with NKJ1514 and for tetracycline resistance (Tet^r) with NKJ1515. To construct NKJ1500 (uvrC), uvrC279::Tn10 from CAG12156 (68) was transduced into AB1157 using P1 phage, and UV-sensitive and Tet^r derivatives were selected. Strains YG2238 (polA) and YG6341 (polB) were obtained by P1 transduction using AB1157 as the recipient and HRS7052 (chloramphenicol resistant [Cam^r]) (35, 76) and HRS6700 (kanamycin resistant [Kan^r]) (67) as donors, respectively. The dinB gene was disrupted as an in-frame deletion without a selection marker as described previously (14). Briefly, using the primers dinB W-F (5′-CAAACCCTGAAATCACTGTATACTTTACCAGTGTTGAGAGGTGAGCAATGATTCCGGGGATCCGTCGACC-3′) and dinB W-R (5′-GCACACCAGAATATACATAATAGTATACATCATAATCCCAGCACCAGTTGTGTAGGCTGGAGCTGCTTCG-3′), the Kan^r cassette on pKD13 was amplified and the resultant fragment was flanked upstream and downstream of the dinB gene. The fragment was introduced into AB1157 harboring pKD46, and then recombination events generated Kan^r colonies, which are expected to carry ΔdinB::Kan^r. Cultivation of the colonies at 43°C cured the temperature-sensitive (Ts) plasmid pKD46, which had been introduced to allow AB1157 to keep the linear PCR fragments intact with its encoding genes derived from lambda phage. The ΔdinB::Kan^r strain without pKD46 was designated YG6158. After introducing another Ts plasmid (pCP20) into YG6158, the strain was incubated at 43°C again. The high temperature induced a recombinase FLP and also removed the Ts plasmid from the strain. The FLP specifically acts on the FLP recombination target (FRT) sequences which had been introduced between the dinB flanking regions and the Kan^r cassette in advance (Table 1). The recombination removed the Kan^r cassette and produced the ΔdinB strain (YG6162). To construct YG6168, Δ(umuDC)596::ermGT was transferred to AB1157 using the DE2302 and EC8 strains as described previously for two-step P1 transduction (19, 78). YG6171 (a dinB umuDC double mutant) was constructed by introduction of ΔdinB::Kan^r into YG6168, followed by removal of the Kan^r cassette as described above. To introduce ΔpolB::Kan^r into YG6171, P1 transduction was carried out with HRS6700 as the donor. The resultant polB dinB umuDC triple mutant was designated YG6342. Finally, YG6344, which has deletions of polA, polB, dinB, and umuDC, was constructed by P1 transduction using YG6342 as the recipient and HRS7052 as the donor. RFM445 [gyrB203(Ts)] and RFM475 [gyrB203(Ts) Δ(topA cysB)204)] are the Ts mutants of topoisomerases. In keeping with the reported Ts properties (17), RFM445 and RFM475 exhibited heat and cold sensitivities, respectively (see Fig. S1 in the supplemental material). pNTR-SD is a set of mobile plasmid clones of E. coli open reading frames, and the expression of the open reading frames is strictly controlled by P_tac and lacI^q (61). pNTR-SD containing the dcm gene was designated pNTR-SD-Dcm (Table 1).

For FA and azaC treatment, cells were grown in LB, minimal A, or M9 medium. Minimal A medium was comprised of 60 mM K₂HPO₄, 33 mM KH₂PO₄, 7.5 mM (NH₄)₂SO₄, 1.7 mM sodium citrate dihydrate, 1 mM MgSO₄·7H₂O, and 0.2% glucose, and supplemented with 1 μg/ml thiamine and 0.2% Casamino Acids (5). The composition of M9 medium was 47 mM Na₂HPO₄·12H₂O, 22 mM KH₂PO₄, 8.6 mM NaCl, 19 mM NH₄Cl, 2 mM MgSO₄, 0.1 mM CaCl₂, and 0.4% glucose, supplemented with 2 μg/ml thiamine, 20 μg/ml thymine, and 100 μg/ml each of arginine, leucine, tryptophan, histidine, and proline (36).

Cell survival assays with FA and azaC.

The working solution of FA was prepared freshly from the 37% FA solution (Nacalai Tesque) for every experiment. azaC (Sigma) was dissolved in 50% acetic acid and stored at −20°C until use. Except for the priA and other related primosomal mutants (priB, priC, and rep), cells were grown to an optical density at 600 nm (OD₆₀₀) of 0.3 at 37°C in LB medium for FA treatment or in minimal A medium for azaC treatment. A total of 0.2 ml of cell culture was diluted with 4.8 ml of 66-mM phosphate buffer (pH 6.8) containing different concentrations of FA (indicated in the figures) or with 4.8 ml of minimal A medium containing different concentrations of azaC (indicated in the figures), incubated at 37°C for 30 min with shaking. The priA, priB, priC, and rep mutants were similarly grown and treated with FA and azaC in M9 medium, since the priA mutant is sensitive to rich media (36). After FA or azaC treatment, cells were diluted and plated on LB, minimal A, or M9 agar plates, and colony formation was typically analyzed after overnight incubation at 37°C. Some strains used in this study grew slowly on M9 plates {AQ10459 [wild type (wt)] and AQ10479 [priA]) or LB plates (RFM445 [gyrB(Ts)] and RFM475 [gyrB(Ts) topA]} (see Table S1 in the supplemental material) so that colony formation was analyzed after a few days of incubation. Cells transformed with pNTR-SD-Dcm (Table 1) were grown to an OD₆₀₀ of 0.2 and incubated with 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) for 10 min to induce Dcm. The subsequent FA and azaC treatments were performed as described above for cells without a plasmid. Unless otherwise noted, the survival data are based on three to five independent experiments. Statistical significance was determined by a two-sided unpaired Student t test. A P value of less than 0.05 was defined as significant.

Measurement of UV and MMC sensitivity.

For the measurement of UV sensitivity, cells were grown to an OD₆₀₀ of 0.3 at 37°C in LB medium, serially diluted and plated on LB agar plates, and exposed to the UV light at doses indicated in the figures. After overnight incubation at 37°C, the number of growing colonies was counted. The sensitivity to mitomycin C (MMC; Sigma) was measured as described above for FA.

RR following HR of DPCs depends on PriA but not Rep helicase.

In E. coli, the PriA, PriB, and PriC proteins play vital roles in restarting chromosome replication through two PriA-dependent mechanisms, i.e., the PriA-PriB and PriA-PriC pathways (Fig. 1A) (24, 36). The RR proteins recognize forked DNA structures, such as arrested replication forks and D-loops, and sequentially load the replicative helicase DnaB (41, 64). To determine the roles of PriA, PriB, and PriC in RR following the HR of DPCs, the sensitivities of priA, priB, and priC mutants to DPC-inducing agents were assayed. The wt and mutant cells were treated with various concentrations of FA and azaC for 30 min, and the cell survival was analyzed. The priA mutant was hypersensitive to both FA and azaC (Fig. 1B and E), indicating that RR following the HR of DPCs proceeds through PriA-dependent mechanisms. Compared to wt, the priB mutant exhibited a moderate sensitivity to azaC at high concentrations (Fig. 1F), although it was not sensitive to FA (Fig. 1C). The priC mutant also showed a slight sensitivity to azaC at high concentrations (Fig. 1F), but the sensitivity increase was not statistically significant. The differential azaC sensitivities of the priB and priC mutants suggest that the PriA-PriB pathway contributes more to RR than the PriA-PriC pathway does, but the two pathways can compensate for each other to a significant degree when one is compromised. In addition to the PriA-dependent RR pathways, there is an alternative Rep-PriC restart pathway to load DnaB onto the forked DNA structures (63). However, the rep mutant was sensitive to neither FA nor azaC (Fig. 1C and F), indicating that the Rep-PriC pathway is dispensable in RR following the HR of DPCs.

We used M9 as a common medium for the FA treatment of RR mutants (Fig. 1B and C), since one of the RR mutants used (priA) is sensitive to rich media (36). Conversely, rich LB medium was used for the FA treatment of the other sets of mutants (Fig. 2 to 5). The use of M9 medium significantly increased the FA sensitivity of cells. For instance, AB1157 (wt) was more sensitive in M9 medium than in LB medium, with the sensitivity in M9 being comparable to that of AQ10459 (wt) in M9 medium (Fig. 1D). Thus, FA concentrations used in the survival assays of RR mutants (0 to 2 mM) (Fig. 1B and C) were lower than those used for other sets of mutants (0 to 7 mM or 0 to 16 mM) (Fig. 2 to 5). In view of the high reactivity of FA to amino and sulfhydryl compounds, it is likely that the constituents of rich LB medium scavenged FA more effectively than those of M9 medium. The azaC sensitivity of RR mutants was also assayed in M9 medium. The azaC concentrations used for the priB, priC, and rep mutants were the same as those used for other sets of mutants (Fig. 2 to 5). However, azaC concentrations used for the priA mutant were much lower than those for other sets of mutants due to an extreme sensitivity of the mutant. The priA mutant was hypersensitive to both FA and azaC, but the sensitivity to azaC was by far greater than that to FA (Fig. 1B and E). Also, the priA mutant grew very slowly (see Table S1 in the supplemental material) and was more sensitive to azaC than the recA mutant (Fig. 1E). Taken together, it is likely that the extreme azaC sensitivity of the priA mutant exhibiting complex phenotypes is attributable to defects not only in RR but also in yet-unknown factors.

Overproduction of Dcm increases azaC sensitivity of cells.

We have previously proposed that DPCs containing large cross-linked proteins (>12 to 14 kDa) are not repaired by NER and exclusively processed by RecBCD-dependent HR that involves the genes recA, recBCD, recG, and ruvABC. We confirmed that these mutants were sensitive to FA, whereas the recF mutant was not (see Fig. S2A in the supplemental material). azaC incorporated into the chromosome likely traps the Dcm protein (53 kDa), giving rise to large DPCs (Dcm-DPCs) that are exclusively processed by RecBCD-dependent HR. Consistent with this mechanism, it was reported that overproduction of the Dcm protein from the dcm-carrying plasmid increased the azaC sensitivity of wt cells by 930-fold and increased that of the recA mutant by fivefold (5). However, interestingly, the same report showed that the dcm mutant was not particularly resistant to azaC compared to the corresponding wt cell that expressed a physiological level of Dcm (5). Our tentative interpretation of the apparently inconsistent results regarding the effects of overproduction and mutational inactivation of Dcm is that the HR capacity of the repair-proficient wt cells is sufficient enough to deal with Dcm-DPCs produced from endogenous Dcm. Using the fluorescein isothiocyanate-labeling method (51), we attempted to detect Dcm-DPCs in DNA isolated from azaC-treated cells without success (data not shown), suggesting very low levels of the endogenous Dcm protein and concomitant Dcm-DPCs in cells. To confirm the role of HR in the processing of Dcm-DPCs more clearly, we used cells harboring pNTR-SD-Dcm, which overproduces the Dcm protein upon treatment with IPTG. A similar approach has shown that overproduction of EcoRII Dcm results in the replication arrest of plasmid pBR322 at the canonical methylation site in E. coli cells grown in the presence of azaC (38). Overproduction of Dcm from a plasmid dramatically increased the azaC sensitivity of cells that were insensitive (wt and recF) or moderately sensitive (recG) to azaC without a plasmid (Fig. 2A, F, and D). Similarly, overproduction of Dcm slightly increased the azaC sensitivity of recA, recB, and ruvA mutants that were hypersensitive to azaC without a plasmid (Fig. 2B, C, and E). We infer that the HR of recA, recB, and ruvA mutants was severely impaired so that Dcm-DPCs produced from endogenous Dcm were sufficient to kill the mutants. Thus, overproduction of Dcm-DPCs in the mutants did not result in a dramatic increase in cell killing.

RecQ, but not RecJ, is slightly sensitive to FA.

The RecQ helicase and RecJ exonuclease are the components of the RecFOR recombination machinery (39). The recJ mutant was sensitive to neither FA nor azaC (Fig. 2G and H). The recQ mutant displayed no sensitivity to azaC (Fig. 2H) but was slightly sensitive to FA (Fig. 2G). Although the FA sensitivity of the recQ mutant was significant, the phenotype to FA was by far milder than that of the recB mutant (see Fig. S2A in the supplemental material and reference 51). Accordingly, this is consistent with our previous observation that DPCs are processed exclusively by RecBCD-dependent HR and not RecFOR-dependent HR (51). In addition to its role in the RecFOR pathway, RecQ is suggested to be involved in SOS DNA damage signaling in response to replication fork stalling (26). However, it is not clear whether the weak FA sensitivity of the recQ mutant is related to SOS damage signaling.

TLS provides no alternative damage tolerance pathway to DPCs.

The presence of a lesion in template DNA often causes the replicative DNA polymerase to stall at the lesion. E. coli cells possess TLS DNA polymerases (Pol), including Pol II (polB), Pol IV (dinB), and Pol V (umuDC), that can transiently take over from the replicative polymerase to continue synthesis across the lesion, allowing replication to resume (55). With UV-induced lesions, there are conflicting reports on the role of Pol II in RR. Pol II was originally implicated in RR (57), but this possibility was ruled out by a subsequent study (11). Interestingly, Pol V becomes essential for RR in the absence of RecJ and RecQ, although the role of Pol V in RR is modest in wt cells (12). To ask whether TLS polymerases play any significant role in the resumption of replication, and hence contribute to damage tolerance to DPCs, the FA and azaC sensitivities of TLS polymerase mutants (polB, dinB, and umuDC) were assayed. The single mutants exhibited no sensitivity to FA or azaC (Fig. 3A and C). Moreover, the double (dinB umuDC) and triple (polB dinB umuDC) mutants were not sensitive to FA and azaC at all (Fig. 3B and D). These results clearly indicate that neither direct TLS nor indirect TLS involving template switching (23, 25) provides an alternative damage tolerance pathway to DPCs with respect to cell survival. In contrast, the polA mutants deficient in Pol I (YG2238 and YG6344) were hypersensitive to both FA and azaC (Fig. 3A to D). These results demonstrate the essential role of Pol I as a repair polymerase both in NER and HR.

NER mutants exhibit various degrees of FA and azaC sensitivity.

We have previously shown that the uvrA mutant is sensitive to FA but not azaC, demonstrating that NER partly contributes to the repair of FA-induced DPCs. Here we assessed the roles of a series of NER genes, including uvrA, uvrB, uvrC, uvrD, cho, and mfd (73), and those of repair-related transcription factors, including dksA and greA (72), in the processing of DPCs. We confirmed the moderate UV sensitivity of the mfd mutant (Fig. 4J) (66) and the lack thereof of the cho mutant (Fig. 4G) (48). The mutants were treated with FA and azaC, and their sensitivities were determined.

For analysis of NER genes, we used two sets of mutants derived from AB1157 and KMBL1001. With AB1157 derivatives, the mutants involved in damage recognition (uvrA and uvrB) and the dual incision of DNA (uvrC) were sensitive to FA, but their sensitivities differed significantly from each other (uvrB > uvrA > uvrC) (Fig. 4A). The mutants were also sensitive to UV and MMC (Fig. 4C and D), and shared a common order of sensitivity to FA, UV, and MMC (uvrB > uvrA > uvrC). However, the uvrC mutant exhibited a uniquely weak sensitivity to FA but not UV and MMC. The uniquely weak FA sensitivity characteristic of the uvrC mutant was confirmed using another set of strains derived from KMBL1001 (Fig. 4E). Conversely, the cho mutant exhibited a moderate sensitivity to FA (Fig. 4E), suggesting its in vivo role as an alternative nuclease in the NER of DPCs (see Discussion). To our knowledge, this is the first report of the phenotype of the cho single mutant to DNA-damaging agents. The cho uvrC double mutant exhibited an FA sensitivity comparable to that of the cho single mutant (Fig. 4E). Consistent with its role in NER, the uvrD mutant was sensitive to FA (Fig. 4E), indicating that the UvrD helicase can unwind the duplex containing DPC and dissociate the DPC-containing fragment from its complementary strand. Unlike uvrA, uvrB, uvrC, and cho mutants that were not sensitive to azaC (Fig. 4B and F), the uvrD mutant displayed a slight but statistically significant sensitivity to azaC as well (Fig. 4F), suggesting a role of UvrD outside NER (see Discussion). Mfd is a transcription-coupled repair (TCR) factor that translocates RNA polymerase stalled at the lesion, recruiting UvrA to the site (66, 73). However, the mfd mutant exhibited no sensitivity to FA and azaC (Fig. 4H and I), indicating that unlike the pyrimidine photodimers that are repaired through both global genome repair and TCR pathways, DPCs are eliminated from the genome by global genome repair-NER exclusively.

In E. coli, it has been shown that the backed-up arrays of stalled transcription complexes which are impediments to replication are kept under surveillance of the stringent response regulators ppGpp and DksA or the GreA and Mfd proteins, which revive or dislodge stalled transcription complexes (72). Thus, it would be interesting to determine whether transcription complexes stalled by DPCs are under such a surveillance system. To clarify this, the sensitivities of the dksA and greA mutants to FA and azaC were analyzed. Unlike the mfd mutant, the dksA and greA mutants exhibited no UV sensitivity (Fig. 4J). The greA mutant was not sensitive to FA (Fig. 4H). However, the dksA mutant exhibited a slight but significant sensitivity to FA (Fig. 4H), implying a certain effect of DksA on the stability of transcription elongation complexes trapped by DPCs. The greA and dksA mutants were not sensitive to azaC (Fig. 4I).

BER mutants are slightly sensitive to FA, and DNA glycosylases alleviate azaC toxicity to cells.

In BER, aberrant bases with minor modifications are removed by DNA glycosylases, and the resulting abasic sites (or nicked abasic sites) are processed by apurinic/apyrimidinic (AP) endonucleases. In E. coli, endonucleases III (nth) and VIII (nei) and formamidopyrimidine glycosylase (fpg) account for the major DNA glycosylase activity for oxidized or fragmented pyrimidines (nth and nei) and purines (fpg) (77). Exonuclease III (xth) and endonuclease IV (nfo) account for the major AP endonuclease activity (16). To clarify whether BER contributes to the repair of DPCs, the sensitivities of glycosylase and AP endonuclease mutants to FA and azaC were assayed. With FA treatment, the nei nth fpg triple mutant and the xth nfo double mutant exhibited a slight but significant FA sensitivity (Fig. 5A and C). The fpg and nei nth mutants were virtually insensitive to FA (Fig. 5A). Thus, elimination of three major DNA glycosylases (or AP lyase activity associated with DNA glycosylases) or two major AP endonucleases confers some FA sensitivity on cells. It remains to be seen whether these BER enzymes are involved in the repair of the cryptic base damage induced by FA (i.e., minor base modifications) or FA-induced DPCs per se, although the latter seems unlikely. However, the relative weak sensitivities of the nei nth fpg triple mutant and the xth nfo double mutant indicate that BER plays at most a minor role in cell survival following FA treatment. Surprisingly, the nei nth fpg triple mutant exhibited unexpected strong sensitivity to azaC (Fig. 5B). However, the azaC sensitivity of the triple mutant was not as high as that of the hypersensitive recA mutant. The fpg single and nei nth double mutants were virtually insensitive to azaC. Also, the xth nfo double mutant was not sensitive to azaC (Fig. 5D). These results imply that Nei/Nth and Fpg glycosylases complement each other in vivo and remove lethal or potentially lethal DNA damage and that Xth and Nfo that generally act following glycosylases in BER are dispensable in mitigating azaC toxicity to cells (see Discussion).

topA mutants are slightly sensitive to FA.

Not much is known about whether DNA topoisomerases participate in DNA repair in E. coli (39, 69). E. coli has two type I topoisomerases (Topo I and III) and two type II topoisomerases (gyrase and Topo IV), and only Topo III is dispensable for cell growth (9). To obtain insight into the role of topoisomerases in the repair/tolerance of DPCs, we used two Topo mutants (Table 1). RFM445 contains mutations (gyrB221 and gyrB203) in the gyrB gene encoding the gyrase subunit B. The mutations confer coumermycin resistance (Cou^r) (gyrB221) and cause temperature sensitivity (gyrB203) (17, 45). The gyrase comprising GyrA (wt) and mutant GyrB (gyrB203) has an activity less than 1% of the normal enzyme at 42°C (45). Thus, the gyrB203 mutation causes partial inactivation of gyrase at 37°C, which was substantiated by the slow-growing phenotype of RFM445 relative to the wt at 37°C (see Fig. S1A in the supplemental material). RFM475 contains the deletion of the topA gene encoding Topo I, together with the mutations gyrB221 and gyrB203. The gyrB203 mutation compensates for the lack of DNA relaxing activity associated with the topA mutation (15, 56). Although the gyrB203 allele retains minimum activity to allow cell growth at 37°C, it regains gyrase activity at 30°C and is no longer sufficient as a compensatory mutation, rendering the gyrB203 topA mutant cold sensitive (17). This phenotype was confirmed in the present study (see Fig. S1B in the supplemental material). Keeping in mind the phenotypes described above, we assessed the sensitivity of RFM445 [gyrB(Ts)] and RFM475 [gyrB(Ts) topA] to FA and azaC at 37°C. The gyrB(Ts) mutant was sensitive to neither FA nor azaC (Fig. 5E and F). However, the gyrB(Ts) topA mutant was slightly sensitive to FA (Fig. 5E), suggesting a role of Topo I in the processing of FA-induced DPCs. The gyrB(Ts) topA mutant was not sensitive to azaC (Fig. 5F).

General characteristics of genes that alleviate the detrimental effect of DPCs.

Figure 6 shows the summary of the sensitivities to the DPC-inducing agents (FA and azaC) displayed by the mutants, including those deficient in damage tolerance (HR, RR, and TLS), excision repair (NER and BER), and miscellaneous aspects of DNA transactions. The data were derived from those shown in Fig. 1 to 5 and Fig. S2 in the supplemental material. The fold increases in the sensitivities of mutants relative to the corresponding wt were calculated at the FA and azaC concentrations indicated in Fig. 6. With hypersensitive mutants, the survival data at lower concentrations were used for calculation (Fig. 6). Although the data in Fig. 6 allow only semiquantitative analysis, they reveal general aspects of genes that alleviate the detrimental effect of DPCs (tentative threshold for significance set as a twofold increase in sensitivity). First of all, the damage tolerance mechanism involving HR and subsequent RR provides the most effective means for cell survival against DPCs. TLS does not serve as an alternative damage tolerance mechanism for DPCs so far as cell survival is concerned. Elimination of DPCs from the genome relies primarily on NER, which provides a second and moderately effective means for cell survival against DPCs. Interestingly, Cho rather than UvrC seems to be an effective nuclease for the NER of DPCs. The role of DNA polymerase I (polA) in both HR and NER has been confirmed. Together with the genes responsible for HR, RR, and NER, the mutation of genes involved in several aspects of DNA repair or transactions (i.e., recQ, nei nth fpg, xth nfo, dksA, and topA) rendered cell sensitivities to FA to increase by twofold or slightly more. Except for the nei nth fpg triple mutant, this was characteristic of FA but not azaC, probably reflecting the unique complexity of DPCs induced by FA or other FA-induced cryptic base modifications. UvrD may have a role outside NER, since the uvrD mutation conferred a threefold increase in azaC sensitivity on cells. The triple mutant of DNA glycosylases (nei nth fpg) exhibited weak and moderate sensitivities to FA and azaC, respectively. The moderate azaC sensitivity of the mutant may be related to the removal of azaC incorporated into DNA or related degradation products.

We measured the doubling time of strains used in this study to see whether their growth properties had something to do with FA and azaC sensitivities. Among the strains used, the following strains grew poorly and had more than 1.2-fold-greater doubling time than the wt: priA, recB, nei nth, nei nth fpg, polA, polA dinB polB umuDC, gyrB(Ts), and gyrB(Ts) topA (see Table S1 in the supplemental material). The poorly growing strains exhibited various degrees of FA and azaC sensitivities (see Table S1 in the supplemental material). The mutants such as priA, recB, and polA mutants that grow poorly and exhibit low viability under normal conditions may be sensitive to DNA damage for indirect as well as direct reasons.