Proc. Nati. Acad. Sci. USA Vol. 84, pp. 2891-2895, May 1987 Genetics Loss of an apurinic/apyrimidinic site increases the mutagenicity of N-methyl-N'-nitro-N-nitrosoguanidine to Escherichia coli (xthA-/exonuclease III/DNA repair/alkylation damage/SOS response) PATRICIA L. FOSTER AND ELAINE F. DAVIS Division of Environmental Health, Boston University School of Public Health, and the Hubert H. Humphrey Cancer Research Center, Boston University School of Medicine, Boston, MA 02118 Communicated by Fred Sherman, January 16, 1987 (received for review October 6, 1986)

ABSTRACT xthA- Escherichia coli, which are missing a in spontaneous or induced mutagenesis has been major cellular apurinic/apyrimidinic (AP) endonuclease, are difficult to establish. 5- to 10-fold more sensitive than xthA+ bacteria to mutagenesis Exposure of E. coli to low concentrations of alkylating by N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) under con- agents induces the "adaptive response" (20), which includes ditions that induce the "adaptive response." The xthA-- two known DNA repair -06-methylguanine-DNA dependent mutations are also dependent on SOS mutagenic methyltransferase (21) and 3-methyladenine-DNA glycosyl- processing and consist of both transversion and transition base ase II (TAGII) (22, 23). Two other genes of unknown substitutions. When MNNG-adapted xthA- bacteria are chal- function, alkB (24) and aidB (25), also are induced. O6_ lenged with a high dose of MNNG, more xthA-dependent methyltransferase accurately repairs 06-alkylguanine (21) SOS-dependent mutations are induced, and transversions are and 04-alkylthymine (26), thus virtually eliminating any enhanced relative to transitions. The mutations induced by mutations due to these lesions. TAGII, the product of the challenge are eliminated in xthA- alkA- bacteria, which are alkA gene, removes purines alkylated at the N-3 position (22, also deficient for 3-methyladenine glycosylase II activity. These 23) and pyrimidines alkylated at the 0-2 position (26), leaving data are consistent with the hypothesis that AP sites, at least AP sites. AP sites are also produced after alkylation damage some of which are produced by glycosylase activity, are by the spontaneous loss of alkylated bases, particularly mutagenic intermediates following cellular DNA alkylation. N7-alkylguanine, and by the activities of the constitutive glycosylases formamidopyrimidine-DNA glycosylase and 3- Apurinic and apyrimidinic (AP) sites cause errors in DNA methyladenine glycosylase I (3). Thus, induction of the replication in vitro and are mutagenic when present in adaptive response by alkylating agents leads to the produc- transforming DNA (1, 2). In addition to being a probable tion of AP sites by several different pathways. Based on the cause of spontaneous mutations (3, 4), AP sites have been distribution ofalkyl lesions in DNA (27) and the known repair implicated as a common mutagenic lesion created by a activities, the majority of these AP sites are expected to be number of DNA-damaging agents (reviewed in ref. 1). These due to missing purines. agents include those that make bulky adducts to the DNA, In vitro studies have indicated that AP sites give rise to such as aflatoxin B1 (5), benzo[a]pyrene (6, 7) and p- mutations because DNA polymerases preferentially insert propiolactone (6, 8), as well as simple alkylating agents (6, 9). purines, particularly adenines, opposite such sites during However, the mutagenic role ofAP sites produced in vivo has bypass synthesis (1). Hence, depurination is expected to not been well established. yield a predominance of transversion mutations while The xthA gene ofEscherichia coli encodes exonuclease III, depyrimidination should yield transition mutations. In E. a multifunctional enzyme that is the cell's major AP endo- coli, at least some part of this postulated pathway must be (10). xthA- bacteria retain only 10-20% of the dependent on SOS mutagenic processing, as mutations are normal cellular AP endonuclease activity (11, 12). xthA- not induced when SOS-deficient cells are transformed with mutants are deficient in the repair of the AP sites created by depurinated DNA (1). the enzymatic removal of misincorporated uracil from DNA Since AP endonuclease activity is the first step in the repair (13). They are also slightly sensitive to killing by methyl of AP sites, we reasoned that if endogenous AP sites are methanesulfonate (11, 12), an alkylating agent that produces mutagenic, the mutagenicity of alkylating agents should be a predominance of N-alkylated DNA bases that give rise to enhanced by the xthA- defect under conditions that (i) maxi- AP sites both spontaneously and enzymatically (3). Other mize the formation of AP sites to saturate compensating phenotypes ofxthA- mutants include increased sensitivity to activities; (ii) minimize the mutagenicity ofotherlesions such as the oxidizing agent H202 (14), slight sensitivity to near UV 06-alkylguanine; and (iii) allow for SOS mutagenic processing. light (15), and partially defective expression of heat shock These conditions were achieved by exposing E. coli to a low proteins (16). Since the xthA gene product is also a duplex- concentration of N-methyl-N'-nitro-N-nitrosoguanidine (MN- specific DNA 3'-5' exonuclease (10), can remove a variety of NG), thus inducing both the adaptive response (20) and the SOS blocking groups from 3' termini in duplex DNA (17, 18), has response (9). When both of these responses are active, xthA- RNase H activity (10), and can recognize and nick near urea bacteria are more mutable by MNNG than xthA+ bacteria. residues in DNA (19), it is not known what contribution, if any, its AP endonuclease activity may make to these other MATERIALS AND METHODS phenotypes. Despite the several phenotypes of xthA- bac- Bacterial Strains. All genetic manipulations were as de- teria and the many activities of exonuclease III, a role for this scribed (28). The E. coli K-12 strains used for most of the

The publication costs of this article were defrayed in part by page charge Abbreviations: AP, apurinic/apyrimidinic; MNNG, N-methyl-N'- payment. This article must therefore be hereby marked "advertisement" nitro-N-nitrosoguanidine; TAGII, 3-methyladenine-DNA glycosyl- in accordance with 18 U.S.C. §1734 solely to indicate this fact. ase II.

Downloaded by guest on September 25, 2021 2891 2892 Genetics: Foster and Davis Proc. Natl. Acad. Sci. USA 84 (1987) experiments reported here are derivatives of PF260 [F- ara with MNNG-exposed cells to give the number of MNNG- A(gpt-lac-pro) thi hisS supD60]. hisS is a deletion ofthe entire induced mutations per plate. This difference was then divided gene cluster (29). One set of experiments used by the number of cells plated. In all cases except xthA+ and PF352 and PF353 (9), which are hisG- hisD' derivatives of umuC::TnS/F' hisG428 strains (see Table 1), the induced AB1157 (F- argE3 his4 leuB6 proA2 thr-J ara-14 galK2 mutations per plate were greater than twice the spontaneous lacYl mtl-i xyl-S thi-J rpsL31 supE44 tsx-33; see ref. 29) and background; for all experiments with xthA- alkA+ MS23 (AB1157 his' alkA-; see ref. 30), respectively. To umuDC+/F' hisG46 strains, the induced mutations per plate ensure that the strains were isogenic, PF353 and an xthA- were 20-40 times the spontaneous background. derivative of it were transduced to alkA' and tetracycline Analysis of Mutations. Revertants of hisG46 were single resistance with a Plvir lysate of NK5526 (alkA' hisG::TnJO; colony purified and infected with M13hol67.18 (34). To select obtained from B. Bachmann, Yale University). Methyl meth- for recombinant phage, a His- recipient (PF665 = P9OC anesulfonate resistant transductants were then made hisG- recAl3 hisG- hisD'/F::Tn3) was infected with selection for hisD+ as described (9). Results with these strains did not the His' phenotype. Single-stranded DNA was prepared and differ from those with PF352 and PF353. sequenced by standard dideoxy methods with a synthetic xthA- derivatives were constructed by transducing cells oligonucleotide priming close to the hisG46 region. In a few with a Plvir lysate of BW9115 [A(xth-pncA) pheSl ; see ref. cases, M13 crosses were repeated selecting for histidinol 11; obtained from B. Weiss, The Johns Hopkins University], utilization in WB351 (34) to ensure that spontaneous muta- selecting for resistance to both 1-2 mM 6-aminonicotinamide tions were not being generated during the phage crosses; no and fluorophenylalanine (ref. 40; the effective concentrations differences in the sequences were found. True revertants of varied among strains), and screening for methyl methanesul- hisG428 were identified by their ability to transfer the His' fonate sensitivity. umuC: :TnS derivatives were constructed phenotype by mating. To determine the percentage of true by transduction with a Plvir lysate ofGW2100 (obtained from revertants of hisG428 (Table 1), 291 revertants from MNNG- G. Walker, Massachusetts Institute of Technology) and treated cultures and 100 revertants from nontreated cultures screening kanamycin-resistant transductants for lack of UV of the xthA- umuDC+ strain and 50 revertants from both mutability to rifampicin resistance. treated and untreated cultures of each of the other strains F'4 episomes with either the hisG46 or hisG428 mutations were analyzed. The different classes of hisG428 revertants in were mated into recipient E. coli strains using appropriate our strains could not be identified using thiazole selection. The F'400 episome carrying the his gene cluster of sensitivity as described for S. typhimurium (32). Salmonella typhimurium was obtained from R. Brent (Mas- To calculate the MNNG-induced frequency of mutational sachusetts General Hospital) as hisO1242, hisG46, events, the fraction of the total analyzed represented by a gnd: :TnJO, rfb+. A hisG428 construction was made by given class of mutation was multiplied by the total number of transducing a S. typhimurium strain carrying the episome as colonies on the plate from which the revertants were taken. hisG' hisD- (obtained from R. Brent) to histidinol utilization The number of spontaneous occurrences of that mutation, (hisD+) with a P22 lysate of S. typhimurium strain hisG428H, calculated in the same way, was then subtracted and the which carries the hisG428 mutation on the chromosome difference was divided by the number of cells plated. (obtained from D. Levin, University of California at San Francisco). Transductants were then screened for tetracy- cline resistance and histidine autotrophy (His-). RESULTS hisG46 is a missense mutation in the S. typhimurium hisG The xthA- Defect Increases the Mutagenicity of MNNG. As gene that replaces a CTC with a CCC proline codon shown in Table 1, exposure to a low concentration of MNNG (31). Any base change at the first base and G-C to AT or G-C (0.5 ,ug/ml) in minimal medium for 4 hr was only weakly to T-A changes at the second base of this codon give a His' mutagenic to wild-type E. coli. However, the mutation phenotype (31). hisG428 is a nonsense mutation in the same frequency of xthA- E. coli under these conditions was 5- to gene that changes a CAA glutamine codon to a TAA ochre 10-fold greater than that of xthA+ bacteria. This effect was (32). In amber suppressor strains (as here), 8 base changes as not due to differential survival although adaptation was well as a small deletion will yield true revertants (ref. 32; slightly toxic to xthA- bacteria. With eight repetitions of unpublished observations). Extragenic tRNA suppressors of these experiments, the absolute number of His' revertants hisG428 are also readily induced (9, 32). per plate after MNNG treatment varied from 2- to 10-fold Media. Media used were as described (9). To select for higher for xthA- umuDC+ cells than for wild-type cells. His' revertants, a limiting amount of histidine (100 nmol for Reversion of both hisG46 and hisG428 was enhanced by the hisG46 or 50 nmol for hisG428) plus 2 mg of other required xthA- defect, although the majority of hisG428 revertants amino acids were added to VB minimal medium (33) plates. were due to the creation of extragenic suppressors (Table 1). Mutagenesis Experiments. Adaptation was induced by Thus, some MNNG-induced premutagenic lesions were ev- exposing cells growing in complete minimal medium to idently not being repaired in xthA- bacteria. MNNG (0.5 ,ug/ml) at 32°C for 4 hr as described (9). Cultures In vitro and in vivo evidence indicates that AP sites are only were then centrifuged, resuspended in phosphate buffer (pH mutagenic if the bacterial "SOS response" is induced (1). As 7), plated for mutants, and titered on complete VB minimal the functions of the umuDC operon are necessary for SOS plates. Challenge doses of MNNG were administered by mutagenic processing (35), xthA--dependent mutations resuspending the cultures in citrate buffer (pH 5.5) and should also be umuDC+ dependent ifthey are due to AP sites. adding MNNG (10 ,g/ml) for 5 min at room temperature. As predicted, the enhanced mutagenicity of xthA- bacteria Cultures were then centrifuged, washed, and plated as was greatly reduced by a defect in umuC (Table 1). described above. Sixty-five to 99% of the mutations induced When bacteria were exposed to a high concentration of by a challenge dose in unadapted bacteria were eliminated by MNNG (10 jig/ml for 5 min), the xthA- defect appeared to adaptation. That these conditions also induced the SOS have little effect on the mutation frequency (Table 2). Thus, response was confirmed by measuring the induction of an the xthA- mutagenic phenotype is obscured under normal umuD-lacZ gene fusion (9). conditions of MNNG mutagenesis, presumably because most To select for His' revertants, 1-6 x 108 viable cells were mutations are due to the directly miscoding SOS-independent plated. To calculate induced mutation frequencies, the num- lesion, 06-methylguanine (3, 35). However, if the bacteria ber of spontaneous revertants on the plates with untreated were first adapted by exposure to a low concentration of cells was subtracted from the number of revertants on plates MNNG, as in the previous experiments, and then challenged Downloaded by guest on September 25, 2021 Genetics: Foster and Davis Proc. Natl. Acad. Sci. USA 84 (1987) 2893

Table 1. His' revertants induced during adaptation to MNNG His+ revertants His' revertants % true MNNG, % viable per plate per 108 cells revertants Strain Ag/ml cells* hisG46 hisG428 hisG46 hisG428 hisG428I xthA' umuC' 0 100 24 101 0.5 89 204 197 40 24 35 xthA- umuC' 0 100 25 98 0.5 60 553 704 282 275 12 xthA' umuC- 0 100 7 209t 0.5 125 100 303t 26 W3t ND xthA- umuC- 0 100 12 130t 0.5 105 175 221t 47 24t 32t Values are means of two experiments with each strain unless indicated otherwise. ND, not done. *Average of the experiments with both hisG46 and hisG428 in each genetic background. tOne experiment.

by a high concentration of MNNG, >90% of the mutations dependent mutations at the second base of the hisG46 codon induced in unadapted wild-type bacteria were eliminated. could reflect either the distribution of DNA lesions or a Under these conditions, when 06-methylguanine lesions are preference for accurate repair at the first base. repaired by high levels of 06-methyltransferase activity (21), The umuC: :Tn5 defect reduced the frequency of all xthA-- xthA- bacteria were again -10-fold more mutable than enhanced mutations but had a larger effect in reducing wild-type bacteria. This increased mutagenesis was eliminat- transversions than transitions. That the frequency of transi- ed by the umuC::TnS defect (Table 2). As above, the xthA- tions at the first base differed little among the four strains effect was not due to differential survival-although fewer suggests that these mutations, which were neither xthA- nor cells were plated, the absolute number of His' revertants per umuDC+ dependent, were probably due to 06-methylguanine plate for challenged xthA- umuDC+ bacteria was 1.5 to 5 lesions. In contrast, umuDC+-dependent transitions at the times that of the wild-type bacteria. second base, which were enhanced by the xthA- defect, were The Specificity of xthA-Dependent Mutations. The bias of unlikely to be due to a simple miscoding lesion. DNA polymerases for inserting purines opposite AP sites Fig. 2 presents the frequencies of each base change predicts that in SOS-proficient bacteria, depurination should induced by a high dose ofMNNG given to previously adapted yield transversion mutations, while depyrimidination should xthA- umuDC+ bacteria. To illustrate the magnitude of the yield transition mutations. Transitions could also result from effect, the mutations induced during adaptation have not miscoding by unrepaired 06-methylguanine lesions, but these been subtracted from the total but are given separately. mutations would also occur in SOS-deficient bacteria (3, 35). Relative to adaptation, the challenge dose of MNNG en- Thus, it was of interest to determine the specificity of the hanced the occurrence of all mutational events 2- to 4-fold xthA--dependent mutations. Presented in Table 3 are the except transitions at the second base. sequenced base changes of 126 hisG46 revertants from The Mutations Induced in xthA- Bacteria by a Chailenge MNNG-exposed cultures and 45 spontaneous hisG46 rever- Dose of MNNG Are Dependent on alkA+ Activity. One tants isolated from the experiments given in Tables 1 and 2. explanation for the specificity of the xthA--dependent mu- Fig. 1 shows the frequencies of each base change at the tations induced by MNNG is that it reflects the distribution hisG46 codon induced during adaptation to MNNG in the of AP sites left by the removal of 3-methylguanine and xthA+, xthA-, umuDC+, and umuC::Tn5 strains. The major 02-methylcytosine by the activity of TAGII, the product of effect of the xthA- defect was to enhance both G-C to TA transversions and G-C to A-T transitions at the second base Table 3. Sequenced hisG46 revertants of the occurrence hisG46 codon. The bias for the of xthA-- Number of mutants Table 2. His' revertants induced by a challenge dose of MNNG Strain ACC TCC CAC CTC Total His+ revertants Adaptation % viable per 108 cellst xthA+ umuC+ 2 9 5 6 22 (3) Strain cells* hisG46 hisG428 xthA- umuC+ 3 4 16 11 34 (2) xthA+ umuC- 0 15 0 2 17 (1) Nonadapted cells xthA- umuC- 0 6 2 9 17 (1) xthA' umuC+ 78 820t 602 Challenge of adapted cells xthA- umuC+ 72 989t 714 xthA- umuC+ 3 7 24 2 36 (1) xthA' umuC- 88 682 704 Spontaneous xthA- umuC- 65 691 585 xthA+ umuC+ 1 1 9 5 16 Adapted cells xthA- umuC+ 4 0 12 0 17* xthA+ umuC+ 74 15t <1 xthA+ umuC- 1 1 0 3 5 xthA- umuC+ 43 1640 125 xthA- umuC- 0 4 0 3 7 xthA+ umuC- 61 <1 2 xthA- umuC- 60 <1 <1 Revertants from MNNG-adapted cultures and most spontaneous revertants were generated in the two experiments given in Table 1. Adapted and nonadapted cells were exposed to MNNG at 10 Revertants from MNNG-challenged xthA- umuDC+ cultures were ,ug/ml for 5 min. generated in the two experiments given in Table 2. Half of the *Average of the experiments with both hisG46 and hisG428 in each spontaneous revertants of the umuDC+ strains were generated in a genetic background. separate experiment. Numbers in parentheses are the expected tFor adapted cells, the number ofHis' revertants has been corrected number of spontaneous revertants contaminating each collection of for the mutants induced during adaptation. MNNG-induced revertants. tAverage of two experiments. *One GCC mutant was found. Downloaded by guest on September 25, 2021 2894 Genetics: Foster and Davis Proc. Natl. Acad. Sci. USA 84 (1987)

140 Table 4. Effect of the alkA- defect on the induction of hisG46 revertants by MNNG in xthA- and xthA' bacteria 120- Number of % viable His' revertants Strain determinations cells per 108 cells* =, 100* V) Adaptation xthA' alkA' 6 93 ± 17 106 ± 39 .0 xthA- alkA' 7 82 ± 34 302 ± 77 oc xthA' alkA- 5 47 ± 13 174 ± 70 60 xthA- alkA- 5 22 ± 11 319 ± 123 Challenge of adapted cells LO)E 40 xthA' alkA' 4 91 ± 14 68 ± 45 xthA- alkA' 5 73 ± 26 452 ± 129 20 xthA' alkA- 3 51 ± 6 73 ± 48 xthA- aIkA- 4 27 ± 15 91 ± 35 J1 i . 1 I Values are averages ± SD. ACC TCC CAC CTC *For challenged cells, the values have been corrected for the mutations induced during adaptation. FIG. 1. The base changes induced at the hisG46 codon by MNNG during adaptation. Frequencies were computed from the data in Tables 1 and 3 and have been corrected for spontaneously occurring TAGII activity is responsible for transversion, but not mutations of each class. 0, xthA+ umuDC+; n, xthA- umuDC+; o, transition, mutations. xthA+ umuC::TnS; w, xthA- umuC::TnS). DISCUSSION The results presented here provide strong evidence that AP the alkA+ gene. Insertion of adenines opposite these missing sites produced in vivo are mutagenic. Furthermore, these bases would yield both G-C to T-A transversions and GC to lesions are intermediates in the induction of mutations by A-T transitions. We tested this hypothesis with isogenic MNNG and, by extension, other mutagens that produce AP xthA-, alkA-, and xthA- alkA- strains carrying the hisG46 sites. These conclusions are from the following observations: episome. It was not possible to make these constructions in (i) xthA- bacteria, deficient in the AP endonuclease, the genetic background used in the previous experiments. In exonuclease III, were more mutable by MNNG under con- the strains used, derived from AB1157 (29) and MS23 ditions that maximize the production of AP sites and mini- (AB1157 alkA-; ref. 30), induced and spontaneous mutation mize the occurrence ofmutations due to other lesions; (ii) the rates were more variable and the relative mutagenic effect of majority of the xthA--dependent mutations induced by xthA- was less. In addition, the alkA- defect increases both MNNG were also dependent on SOS mutagenic processing, the toxicity and mutagenicity of MNNG, although the mu- as expected for mutations due to AP sites; and (iii) the xthA tagenic effect at the hisG46 locus is slight (9). mutagenic effect was decreased by a deficiency in TAGII, As shown in Table 4, the frequencies of hisG46 revertants thus demonstrating that production of AP sites by glycosyl- induced during adaptation to MNNG were approximately the ase activity is mutagenic in the absence of exonuclease III. The simplest explanation for these results is that the AP same for the xthA- alkA- double mutant as for the xthA- endonuclease activity of exonuclease III participates with alkA+ strain. However, the enhanced mutagenicity of a TAGII in an error-free DNA repair pathway active on challenge dose of MNNG given to previously adapted xthA- alkylation-damaged bases. The activities of other glycosyl- bacteria was reduced by a factor of 5 by the alkA- defect. ases and the spontaneous loss of alkylated bases also may Since the mutations induced under these conditions are produce substrates for the AP endonuclease activity of mainly transversions (Fig. 2), these results suggest that exonuclease III. The occurrence of SOS-dependent mutations requires two conditions-the presence of a lesion in the DNA and the 160 induction of the SOS response (36). Hence, it is possible that lesions persisting in the DNA as the result of the loss of a DNA repair activity, such as exonuclease III, may serve as inducers for SOS mutagenic processing that then may act u120- elsewhere in the DNA (e.g., see ref. 9). However, there are several reasons why this explanation does not pertain to our results. As measured by the induction of a umuD-lacZ gene fusion, the xthA- defect enhanced the induction of the SOS 80 response by MNNG only 2-fold (data not shown), an effect unlikely to produce the differences we observed. In contrast, induction of the SOS response by MNNG is enhanced in alkA- bacteria =8-fold, but the mutagenic effect ofthe alkA- 40- defect differs from that ofxthA- (ref. 9; Table 4). In addition, induction of the SOS response by thermal shift of isogenic recA441 (tif-1) strains did not increase the mutagenicity of 0 MNNG under our experimental conditions (data not shown). ACC TCC CAC CTC Finally, GC to A-T transitions at the second base of the hisG46 were not induced FIG. 2. Base changes induced at the hisG46 codon of xthA- codon, although SOS dependent, by umuDC+ bacteria during adaptation and by a challenge dose of a challenge dose of MNNG (Figs. 1 and 2). Thus, simple MNNG. Frequencies were computed from the data in Tables 2 and induction of the SOS response cannot account for the 3 and have been corrected for spontaneously occurring mutations of mutations we observed. each class. o, Mutations induced during adaptation; u, mutations The mutagenic phenotype of xthA mutants has been induced by challenge of adapted cells. difficult to demonstrate (11), presumably because of the Downloaded by guest on September 25, 2021 Genetics: Foster and Davis Proc. Natl. Acad. Sci. USA 84 (1987) 2895

existence of redundant activities in the cell. Certain muta- The results presented here provide direct evidence for the tional targets also appear to be insensitive to the effect. For mutagenic role of AP sites produced in vivo by spontaneous example, the induction of true revertants of hisG428 was not loss of damaged bases or as intermediates in the repair of enhanced by xthA- (Table 1), although this site provides ART other DNA lesions. The high rate of spontaneous depurina- base pairs as potential substrates for both 3-methyladenine tion (104 per mammalian cell per day; ref. 38), plus the glycosylase I and TAGII activity. Evidently, alkyl lesions do existence of multiple pathways for the repair of AP sites, not form, or repair enzymes do not act, at this locus. In argue for the potential importance of these lesions. When the addition, the multiple activities of exonuclease III compli- AP site repair pathways are saturated by exposure to a DNA cates the interpretation of experiments with xthA- strains. damaging agent, even a relatively low rate of base loss could The DNA lesions that are substrates for the other activities have severe mutagenic and carcinogenic consequences. of exonuclease III are not known to be produced directly by alkylation damage. However, spontaneous or enzymatic Note. It has recently been demonstrated that a strain defective in both removal of the base-free sugars left after the loss of alkylated xthA and nfo, the gene that encodes the AP endonuclease, endonu- bases could generate substrates for the 3' termini unblocking clease IV, is more mutable by methyl methanesulfonate than either activity or the exonuclease activity of exonuclease III. The single mutant strain (39). accumulation of unrepairable strand breaks apparently ac- We are grateful to Eric Eisenstadt and Bruce Demple for many counts for the extreme sensitivity of xthA- bacteria to the helpful discussions and for critically reading this manuscript. We lethal effects of oxidative damage (17), but no mutagenic thank Graham Walker and Jeffrey Miller for advice and discussions; consequence has been reported. Furthermore, it is doubtful the people mentioned in Materials and Methods, especially B. that DNA gaps or strand breaks would lead to base-substi- Weiss, for supplying bacterial strains; and Connie Caron for tech- tution mutations. Thus, the loss of activities other than the nical assistance. This work was supported by Public Health Service AP endonuclease activity of exonuclease III may contribute Grant CA37880 awarded to P.L.F. by the National Cancer Institute. to the lethality of alkylation damage, but it is unlikely to account the increase in 1. Loeb, L. A. (1985) Cell 40, 483-484. for base-substitution mutations that 2. Granger-Schnarr, M. (1986) Mol. Gen. Genet. 202, 90-95. we observe in xthA- strains. 3. Lindahl, T. (1982) Annu. Rev. Biochem. 51, 61-88. That the xthA--dependent mutations induced by MNNG 4. Miller, J. H. & Low, K. B. (1984) Cell 37, 675-682. included both G-C to ART transitions and G-C to TEA transver- 5. Foster, P. L., Eisenstadt, E. & Miller, J. H. (1983) Proc. Natl. Acad. sions was The cumulative effects ofall Sci. USA 80, 2695-2698. unexpected. pathways 6. Drinkwater, N. R., Miller, E. C. & Miller, J. A. (1980) Biochemistry 19, leading to the loss of alkylated bases and the distribution of 5087-5092. alkyl lesions in the DNA predict that depurination should be 7. Eisenstadt, E., Warren, A. J., Porter, J., Atkins, D. & Miller, J. H. much more frequent than depyrimidination after exposure to (1982) Proc. Natl. Acad. Sci. USA 79, 1945-1949. an if are 8. Schaaper, R. M., Glickman, B. W. & Loeb, L. A. (1982) Cancer Res. alkylating agent. Thus, purines preferentially insert- 42, 3480-3485. ed opposite AP sites in vivo as in vitro, transversions should 9. Foster, P. L. & Eisenstadt, E. (1985) J. Bacteriol. 163, 213-220. dominate over transitions in xthA- bacteria. However, this 10. Rogers, S. & Weiss, B. (1980) Methods Enzymol. 65, 201-211. prediction was only upheld when previously adapted xthA- 11. Yajko, D. M. & Weiss, B. (1975) Proc. Natl. Acad. Sci. USA 72, bacteria were a dose of MNNG. Of the 688-692. challenged by high 12. Ljungquist, S., Lindahl, T. & Howard-Flanders, P. (1976) J. Bacteriol. xthA--dependent mutations induced during adaptation, 42% 126, 646-653. were transitions. These mutations were unlikely to be due to 13. Taylor, A. F. & Weiss, B. (1982) J. Bacteriol. 151, 351-357. miscoding by unrepaired 06-methylguanines as: (i) most 14. Demple, B., Halbrook, J. & Linn, S. (1982) J. Bacteriol. 153, 1079-1082. at 15. Sammartano, L. J. & Tuveson, R. W. (1983) J. Bacteriol. 156, 904-906. transitions, particularly those occurring the second base of 16. Paek, K.-H. & Walker, G. (1986) J. Bacteriol. 165, 763-770. the codon, were umuDC' dependent and (ii) no known 17. Demple, B., Johnson, A. & Fung, D. (1986) Proc. Natl. Acad. Sci. USA activity of exonuclease III would be likely to affect the 83, 1148-1153. mutagenicity of 06-methylguanine lesions. 18. Henner, W. D., Grunberg, S. M. & Haseltine, W. A. (1983) J. Biol. One for these results is the Chem. 258, 15198-15205. possible explanation that strong 19. Kow, W. K. & Wallace, S. (1985) Proc. Natl. Acad. Sci. USA 82, preference for inserting purines opposite AP sites during 8354-8358. bypass synthesis, demonstrated in vitro for a number of 20. Samson, L. & Cairns, J. (1977) Nature (London) 267, 281-283. polymerases, does not always obtain in vivo. in a study of 21. Karran, P., Lindahl, T. & Griffin, B. (1979) Nature (London) 280, 76-77. 22. Karran, P., Hjelmgren, T. & Lindahl, T. (1982) Nature (London) 296, mutations induced by depurination ofM13 DNA, Kunkel (37) 770-773. found transversions were only 3-fold more frequent than 23. Evensen, G. & Seeberg, E. (1982) Nature (London) 296, 773-775. transitions, and that after adenine, the base inserted opposite 24. Kataoka, H., Yamamoto, Y. & Sekiguchi, M. (1983) J. Bacteriol. 153, 1301-1307. AP sites most frequently was thymine. Incorporation of 25. Volkert, M. R. & Nguyen, D. C. (1984) Proc. Natl. Acad. Sci. USA 81, thymine opposite a missing guanine would yield a G-C to ART 4110-4114. transition, the mutation that we observed. Since G-C to COG 26. McCarthy, T. V., Karran, P. & Lindahl, T. (1984) EMBO J. 3, 545-550. 27. Singer, B. & Grunberger, D. (1983) Molecular Biology ofMutagens and transversions at the second base of the hisG46 codon give Carcinogens (Plenum, New York). only a partial His' phenotype (J. K. Miller and W. M. 28. Miller, J. H. (1972) Experiments in Molecular Genetics (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). Barnes, personal communication), our results may overesti- 29. Bachmann, B. J. (1972) Bacteriol. Rev. 36, 525-557. mate the true proportion of transition mutations and thus not 30. Yamamoto, Y., Katsuki, M., Sekiguchi, M. & Otsuji, N. (1978) J. be inconsistent with previous studies. Bacteriol. 135, 144-152. 31. Miller, J. K. & Barnes, W. M. (1986) Proc. Natl. Acad. Sci. USA 83, However, the fact that xthA--dependent transitions were 1026-1030. only induced by a low dose of MNNG suggests that these 32. Levin, D. E., Marnett, L. J. & Ames, B. N. (1984) Proc. Natl. Acad. mutations may be occurring via a different pathway or be due Sci. USA 81, 4457-4461. 33. Vogel, H. J. & Bonner, D. M. (1956) J. Biol. Chem. 218, 97-106. to a different DNA lesion than the transversions. For exam- 34. Barnes, W. & Tuley, E. (1983) J. Mol. Biol. 165, 443-459. ple, they may be due to enzymatic or spontaneous loss of 35. Walker, G. C. (1984) Microbiol. Rev. 48, 60-93. 36. Witkin, E. (1976) Bacteriol. Rev. 40, 869-907. alkylated cytosines or to some lesion other than an AP site 37. Kunkel, T. A. (1984) Proc. Natl. Acad. Sci. USA 81, 1494-1498. that is a substrate for exonuclease III activity. To be 38. Lindahl, T. & Nyberg, B. (1972) Biochemistry 11, 3610-3618. consistent with our data, this lesion must be not induced, or 39. Cunningham, R. P., Saporito, S. M., Spitzer, S. G. & Weiss, B. (1986) J. Bacteriol. 168, 1120-1127. be repaired by some other pathway, after exposure to a high 40. White, B. J., Hochhauser, S. J., Citrod, M. & Weiss, B. (1976) J. dose of MNNG. Bacteriol. 126, 1082-1088. Downloaded by guest on September 25, 2021