Proc. Natl. Acad. Sci. USAVol. 81, pp. 1494-1498, March 1984Genetics

Mutational specificity of depurination(bacteriophage M13mp2/transversion mutagenesis/SOS repair/base selection/noncoding lesion)

THOMAS A. KUNKELLaboratory of Genetics, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709

Communicated by James A. Miller, November 16, 1983

ABSTRACT The mutagenic consequences of damage toDNA produced by low pH and high temperature have beendetermined in a forward mutational system capable of detect-ing all classes of mutagenic events. When damaged single-stranded DNA from bacteriophage M13mp2 is used to trans-fect competent Escherichia coli cells, a 15-fold increase in mu-tation frequency, measured as loss of a-complementation bythe lac DNA in the phage, is observed compared with an un-treated DNA control transfection. The enhanced mutagenicityis largely dependent on induction of the error-prone SOS re-sponse and is proportional to the number of lethal hits intro-duced into the DNA. The effect is abolished by treatment ofthe damaged DNA before transfection with either apurinic/apyrimidinic endonuclease or alkali. Based on these observa-tions and the rate constants for formation of the known heat/acid-produced lesions in DNA, it is concluded that the major-ity of the induced mutagenesis results from apurinic sites.DNA sequence analysis of 87 spontaneous and 124 inducedmutants indicates that the major effect is on single base-substi-tution mutagenesis with a small increase in (deletion) frame-shift frequency. Approximately 80% of the base-substitutionmutations occur at purine positions in the viral strand, consis-tent with depurination as the predominant premutagenic le-sion. The preference of guanine over adenine sites mutated isconsistent with the preference for depurination of guanineover adenine. Transversions are observed for 57 of 79 (72%)induced base substitutions, with a strong preference for inser-tion of adenine residues opposite the putative apurinic site.These data in a forward mutational system provide insight intothe mechanisms used by a cell to replicate DNA containingnoncoding lesions.

Insight into the effects of damage to genetic information re-quires analysis of the interactions of cellular replication andrepair processes with specific lesions in DNA. One such le-sion that has received considerable attention in recent yearsis the abasic site (1, 2) produced by hydrolysis of the N-gly-cosyl bond between the sugar and base (3-5). This base losscan occur spontaneously at high frequency (3, 6) and thisfrequency can be increased either by base modifications re-sulting from certain DNA damaging agents (7-11) or enzy-matically by the action ofDNA glycosylases (1, 12). Apurinic/apyrimidinic (AP) sites are quite stable (4, 13), and cells haveevolved mechanisms to repair these lesions (12, 14). Unre-paired AP sites have been shown to have two biological con-sequences, lethality (15-18) and base-substitution errors(18). These errors can be produced in vitro with purifiedDNA polymerases (19, 20) or can occur in vivo in Escherich-ia coli under conditions in which the SOS response has beeninduced (11, 18, 21). These results, obtained in an assay spe-cifically designed to detect base-substitution errors that re-vert 4X174 amber mutations, indicate that for a limited num-ber of sites in 4X174 DNA there is a strong preference for

insertion of dAMP at apurinic sites (21), leading to distinc-tive G-C -+ T-A and A-T -- T-A transversions.The use of a single-base-substitution assay involving re-

version of nonsense codons in an essential gene focuses on avery specific set of base changes at a limited number of sites.The inability of this system to detect all possible single-basechanges as well as other classes of mutagenic events has ledto the development of a forward mutational assay in a nones-sential gene. This assay selects for loss of f3-galactosidase"a-complementation," encoded in the DNA of bacteri-ophage M13mp2 developed by Messing et al. (22). Becauseselection is for loss of a nonessential gene function, a widespectrum of mutagenic events can be scored. The precisenature of the mutagenic events, whether base substitutions,frameshifts, deletions, additions, or rearrangements, can bedetermined by determining the sequences of the DNAs ofthe mutants. This system is used here to determine the muta-tional consequences of a noncoding lesion, the apurinic/apyrimidinic site. Of particular interest are two issues, theimportance of base-substitution mutations relative to othermutagenic events and the specificity of the induced basechanges in a system capable of monitoring a wide spectrumof base changes at many sites.

MATERIALS AND METHODSBacteria and Bacteriophage. E. coli strains S90C [A(pro-

lac), ara-, thi-, strA], S90C recA56, NR8036 [A(pro-lac),ara-, thi-, trpE9777], and NR8037 [A(pro-lac), ara-, thi-,trpE9777, umu C36::tnS] were provided by R. Dunn and B.Glickman of this Institute. E. coli strain CSH50 [ara-, thi-,A(pro-lac)/F'traD36, proAB, lacJPZAM15] (the AM15 dele-tion spans 93 bases of the lacZ gene carried on the episome;see Fig. 1), and bacteriophage M13mp2 were obtained fromJ. E. LeClerc (Univ. of Rochester, Rochester, NY).DNA Preparation. Single-stranded M13mp2 viral DNA

was incubated at 70°C for various times in 30 mM potassiumchloride/10 mM sodium citrate, pH 5.0. These conditionsintroduced one AP site per molecule in 4 min, measured bysurvival (18). Hydrolysis of AP sites was carried out eitherby incubating 10 ,ug of depurinated DNA for 30 min at 37°Cwith 20 units of apurinic/apyrimidinic endonuclease fromHeLa cells [fraction VII (23)] (kindly provided by D. Mos-baugh and S. Linn, Univ. of California, Berkeley, CA) in 500,ul of 50 mM Tris-HCl, pH 7.5/10 mM MgCl2 or by incuba-tion in 0.1 M NaOH (final pH 12.8) for 30 min at 37°C.SOS Induction. The various strains of E. coli were grown

at 37°C to a cell density of 2-4 x 108 cells/ml in YT medium(8 g of tryptone, 5 g of yeast extract, 5 g of NaCl per liter, pH7.8), harvested by centrifugation, and then suspended in0.9% NaCl at 10% of original volume. These cells were irra-diated at 25-150 J/m2 in thin layers in plastic Petri disheswith constant gentle shaking. After removal of small aliquotsfor determination of cell survival, cells were diluted to 50%of their original volume in prewarmed YT medium and incu-bated in the dark for 45 min with vigorous aeration. These

Abbreviation: AP, apurinic/apyrimidinic.

1494

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

1, 2

021

Proc. NatL. Acad. Sci. USA 81 (1984) 1495

cells, or unirradiated cells grown to 2-4 x 108/ml in YT me-dium, were treated with CaCl2 to produce competent cells asdescribed (24).

Transfection and Plating. Transfections were carried outby diluting the DNA into 75 mM CaCl2 and adding 2 vol ofcompetent cells. The volume used was determined by theratio ofDNA molecules to cells, which was kept constant forall variables in a single experiment, typically at 20:1. Theefficiency of transfection was 2 x 10-7 (50 plaques per ng ofsingle-stranded DNA) and varied 2- to 5-fold over severalexperiments. After a 40 min incubation at 0WC, the mixturewas heat shocked at 420C for 3 min (or 5 min for volumeslarger than 5 ml), then placed at 0C. An amount of this mix-ture sufficient to yield 100-500 infective centers per platewas added to 3 ml of 0.8% agar in 0.9%o NaCl containing 2 mgof 5-bromo-4-chloro-3-indolyl ,8-D-thiogalactoside and 0.2 mlof a logarithmic phase culture of E. coli cells (CSH50F'lacZAM15). This mixture was poured onto plates containing30 ml of minimal medium solidified with 1.5% agar and con-taining 0.24 mg of isopropylthio-,3-D-galactoside per plate.These plating conditions were specifically chosen to give in-tense blue color for wild-type M13mp2 plaques and to allowoptimum visualization of mutant phenotypes.

Scoring Mutants and Determination of DNA Sequences. In-activation of a-complementation (25) resulting from a muta-tion in the lac DNA in M13mp2 will give rise to mutantsreadily distinguished as lighter blue or colorless plaques (26).Mutants were scored after 12-15 hr of incubation at 370Cfollowed by 24 hr of incubation at room temperature. Toeliminate false positives due to plating artifacts, mutantplaques were picked from the plate, diluted in 50 mM sodiumborate buffer (pH 9.0), and mixed with an equal dilution ofwild-type M13mp2 phage. Plating this mixture allows a visu-al comparison of phenotypes on the same plate and reducesuncertainty in identifying light blue plaques. Single-strandedDNA was prepared from the mutants scored in each of threeindependent experiments and sequences were determined bythe chain-terminator method (27). Two individual oligonu-cleotides were used to prime the reactions for sequence anal-yses, (i) the 15-base primer from Bethesda Research Labora-tories that is complementary to the coding sequence for ami-no acids 11-16 of the lacZ gene and (ii) a newly synthesized15-base primer (New England BioLabs) complementary tothe coding sequence for amino acids 46-50 of the lacZ gene.

RESULTSThe biological activity of M13mp2 viral DNA, when trans-fected into competent E. coli cells, decreases in proportionto the time of incubation at 70°C in pH 5.0 buffer (data notshown). This result is similar to that observed with kX174single-stranded DNA (18). It is consistent with the conclu-sion that, when using single-stranded DNA subject to few orno repair processes, a major biological consequence of de-purination is lethality (14-16). Concomitant with this inacti-vation is an increase in mutation frequency (Table 1). Theincrease is double that of an untreated DNA control whenthe transfection is carried out using normal competent cells.However, if the transfection is carried out using competentcells made from bacteria UV irradiated to induce the SOSresponse, the mutation frequency increases to 97.8 x 10-4,15 times that of the untreated DNA in unirradiated cells.Both the actual mutation frequency and the relative increaseabove background are a function of the number of lethal hitsin the DNA and the UV dose to the bacteria (data notshown). These phenomena are similar to previous observa-tions using heat/acid-damaged 4X174 DNA (18) and implythat mutants result primarily from SOS-dependent bypass ofabasic sites.

Mutagenesis Due to AP Sites. If apurinic (or apyrimidinic)sites are indeed responsible for the enhanced mutagenesis,

Table 1. Mutation frequency of heat/acid-treated M13mp2 DNAin normal and SOS-induced cells

Mutation RelativeUV dose, Lethal hits per frequency mutationJ/m2 DNA molecule (x 10-4) frequency

0 0.0 6.2 ± 1.4 1.00 3.2 13.6 ± 3.0 2.2

100 0.0 11.8 ± 4.0 1.9100 3.2 97.8 ± 25.3 15.8

The transfections, using E. coli CSH50 F' lacZAM15, used eitherundamaged DNA (0.5-2.0 ,ug) or heat/acid-treated DNA (5.0-20Ag). Values shown are mean ± SD of six independent experiments,all carried out with a single M13mp2 viral DNA preparation. Surviv-al of the UV-irradiated cells was 55 ± 12%.

then treatments that eliminate AP sites should reduce oreliminate the effect. As shown in Table 2, two independentmethods of hydrolyzing AP sites reduced the enhanced mu-tagenesis. Hydrolysis by the highly specific (23) HeLa cellAP endonuclease (experiment 1) eliminated 95% of the en-hanced mutagenicity while alkaline hydrolysis (4) reversedthe effect by 85% (experiment 2). The residual mutagenesisabove background may represent incomplete hydrolysis ofAP sites or the presence of other lesions (3, 6, 10, 12, 28, 29).These data clearly indicate that the majority of the enhancedmutagenesis results from apurinic (and possibly apyrimi-dinic) sites in the DNA.SOS Dependence. The mutagenic response was examined

in recMA and umuC- host cells, which are defective forSOS-dependent mutagenesis (30, 31). The results (data notshown) indicate no damage-dependent increase in mutationfrequency either for the recA or umuC- mutant strains, us-ing conditions that gave enhanced frequencies for the wild-type parent strains (32-fold and 11-fold, respectively). Thesedata confirm the SOS dependence of the enhanced mutagen-esis and further support the concept that apurinic sites,which are known to impede DNA synthesis (21, 32-34), arethe responsible lesion.DNA Sequence Analysis. A total of 87 spontaneous and 124

induced mutants were subjected to DNA sequence analy-sis. The classes of mutations observed are summarized inTable 3. The most frequent events under either condition are

Table 2. Reversibility of induced mutagenesis by previoustreatment of DNA with AP endonuclease or alkali

Mutation RelativeUV dose, Lethal hits Previous frequency mutationJ/m2 permolecule treatment (x 10-4) frequency

Exp. 1: AP endonuclease treatment0 0.0 - 5.4 1.00 3.0 - 11.8 2.20 3.0 + 14.2 2.6

100 0.0 - 6.0 1.1100 3.0 - 57.8 10.7100 3.0 + 8.8 1.6

Exp. 2: alkali treatment0 0.0 - 6.1 1.00 3.0 - 16.6 2.70 3.0 + 17.7 2.9

100 0.0 - 13.8 2.3100 3.0 - 101.0 16.6100 3.0 + 27.1 4.4

E. coli CSH50 F' lacZAM15 were used for transfections. Cell sur-vivals were 61% and 70%, respectively, for experiments 1 and 2.Phage survival was unaffected by AP endonuclease treatment andwas halved by alkali treatment, in both normal and irradiated cells.Percentage reversal was calculated as follows: experiment 1, 1 -[(8.8-6.0)/(57.8-6.0)] x 100% = 95%; experiment 2, 1 - [(27.0-13.8)/(101.0-13.8)] x 100% = 85%.

Genetics: Kunkel

Dow

nloa

ded

by g

uest

on

Aug

ust 2

1, 2

021

Proc. Natl. Acad. Sci. USA 81 (1984)

Table 3. Frequency of various classes of observed mutants

No treatment Treated DNA, induced cells(6.2 x 10-4) (97.80 x 10-4)

Events MF Events MF

Type of event No. % (X 10-4) No. t (X 10-4) Treated/normal

Base substitution 28/87 32.2 (42.4) 2.0 79/124 63.7 (85.9) 62.3 31.3Double mutation 0/87 <1.2 (<1.5) <0.1 3/124 2.4 (3.3) 2.3 >23.0(-) Frameshift 9/87 10.3 (13.6) 0.6 7/124 5.6 (7.6) 5.5 9.2(+) Frameshift 3/87 3.4 (4.5) 0.2 0/124 <0.8 (<1.1) <0.8 <4.0Deletion

93-base 21/87 24.1 (0) 1.5 32/124 25.8 (0) 25.2 16.8Other 26/87 29.9 (39.4) 1.9 3/124 2.4 (3.3) 2.3 1.2

When the value for single-base errors (60 x 10-4) is used, it is assumed that there are 3.2 AP sites in a molecule containing 3600 purines, andthe fact that (at least) 35 sites in the target are mutable (Fig. 1) is considered, the bypass frequency, calculated as described (18), is estimated tobe 20%o. Numbers in parentheses are percentages without considering the 93-base-deletion mutations. MF, mutation frequency.

single-base-substitution mutations. One-third of the sponta-neous and two-thirds of the induced mutations are of thistype. Depurination enhances base-substitution mutagenesis31-fold above background to a frequency of 62.3 x 10-4.A second mutation that is increased in frequency is the 93-

base-deletion mutation. This event is presumably the resultof recombination between the F' episome containing this ex-act deletion (i.e., lacZAM15; see Fig. 1), and M13mp2. Fur-ther evidence that recombination is responsible is the obser-vation that these deletion mutants have an associated ade-nine -* guanine base change at the EcoRI site (see Fig. 1)representing the original base altered in the construction ofM13mp phage. Since this deletion results from the use of F'lac-containing competent cells, these mutants need not beconsidered further. Base substitutions thus represent aneven greater percentage of the total induced mutations(79/92, 86%).Two other classes of mutations are increased in the in-

duced spectrum, double mutations and frameshifts. Threenontandem double mutations were observed, one of which(see Fig. 1, number 3, in the -35 promoter region) resemblesa similar mutation, at this same site, in the UV-induced spec-trum (26). Considering the rarity of double mutations, theexistence of two similar nontandem double mutants resultingfrom two different mutagenic treatments suggests a commonmechanism of production inherent to the structure of theDNA (35), the specificity of the replication apparatus, orboth. Double mutations have also been detected twiceamong 32 depurination-induced amber revertants in OX174DNA (21). They could reflect events at two different dam-aged sites in the same molecule or result from a targeted andan untargeted (semitargeted) event at some distance from thelesion. Frameshift mutations resulting from either the addi-

tion (+) or deletion (-) of bases were observed among thoseformed spontaneously. However, only deletion frameshifts(7/92) were observed among the induced mutations and,while these were enhanced 9.2-fold above spontaneous, theabsolute frequency was small compared with base-substitu-tion mutations (7.6% vs. 86%). No addition mutations wereobtained, and the frequency of large deletion mutations (oth-er than the 93-base deletion) was not increased.The spectrum of induced base substitution and frameshift

mutations is shown in Fig. 1. Of the seven frameshift muta-tions found, six represent the loss of a single base and one isa double-base loss. Four of the seven have lost a purine. The79 single-base changes occur at 38 different positions and ingeneral reflect sequences known to be important for expres-sion.The distribution of base-substitution mutants is not ran-

dom; 23% (18/79) are found in the 125-base operator/pro-moter region, while 77% (61/79) are found in the 129-baselacZ coding sequence. One "hotspot" is observed at the co-dons for amino acids 16 and 17 of the lacZ gene. Nineteenmutants (24% of the total) were observed at the three guano-sines, representing only 4.4% of the currently known muta-ble sites (ref. 26; this study) or just 1.2% of the 254-basetotal.An analysis of the specificity of the 79 induced base-sub-

stitution mutations is presented in Table 4. Of 38 sites mutat-ed, 30 were positions of purines in the template strand, witha preference of guanine over adenine sites mutated. Of the 79base substitutions, 64 (81%) were at these 30 purine posi-tions. Transversions (72%) were 3-fold more frequent thantransitions (28%), in contrast to the spontaneous base substi-tutions, which were 79% transitions (data not shown). As-suming that a misincorporation event occurred during the

A T T T TCA T T CT T T TCG T

GCGCA ACGCAATTAA TGTGAGTTAG CTCACTCATT AGGCACCCCA GGCTTC TTTATGCTTC CGGCTCG7AT GlIjGTGTGGA ATTGTGAGCG GATAACAATT TCACACG AACAGCTATG* A3~SA3Ly A4o) A 0

ATCTTT T A

A A TT T T A TA C T TT T AT A AT T T CA C T T ATT T TT T A A T TT A TA TA T TT T TA

ACC ATG ATT ACG AAT TCA CTG GCC GTC GTT TTA CAA CGT CGT GAC TGG GAA AAC CCT GGC GTT ACC CAA CTT AAT CGC CTT GCA GCA CAT CCC CCT TTC GCC AGC TGG CGT AAT AGC GAA GAG GCC CGC

t t c CS A2 A2 A lA1 A A A +

FIG. 1. Induced mutational spectrum for base substitutions and frameshifts in M13mp2 lac DNA. The 5' -. 3' DNA sequence of the viralstrand of the mutagenic target in M13mp2 is shown, from the first nucleotide after the laci termination codon through the coding sequence foramino acid 43 of the lacZ gene. Frameshifts are shown below the sequence, indicated by A. When a frameshift occurs in a run of two or more ofthe same base, the exact nucleotide lost is unknown and the A is centered under the run. Double mutations, linked by common subscripts, havethe indicated change. For example, the first double mutation, indicated with the subscript 1, contains both a C-G -* AT and a G'C -+ CGtransversion in the same molecule. Single-base-substitution mutations are shown above the wild-type sequence and indicate the nucleotidepresent in the viral strand. 0, CAP site; 0, -35 promoter; 0, -10 promoter; ®, transcription start; ®, operator site; 0, ribosome-bindingsite; (0, translation start; 0), position mutagenized to make the EcoRI site in the original mp2 construction; 0, first and last nucleotides of the93-base deletion; @0, known mutable purine sites at which no base substitutions were observed in this study.

1496 Genetics: Kunkel

Dow

nloa

ded

by g

uest

on

Aug

ust 2

1, 2

021

Proc. Natl. Acad. Sci. USA 81 (1984) 1497

Table 4. Specificity of induced base-substitution mutations

Sites Total Template Incorpo-Template mutated, mutations, mutation ration

base no. no. Base No. event*

Guanine 20 48 A 14 TT 27 AC 7 G

Adenine 10 16 G 1 CT 13 AC 2 G

Cytosine 6 13 T 7 AA 6 TG 0 C

Thymine 2 2 C 0 GA 2 TG 0 C

*Assuming that misincorporation occurs during the first round ofDNA replication in vivo.

first round of replication in vivo, there is a preference forincorporation of adenine residues (47/79 or 59%). The re-maining incorporation specificity is in the order thymine(28%), guanine (11%), then cytosine (1%).

DISCUSSIONThe major goal of this work was to establish the spectrum ofmutational events resulting from AP sites in DNA, to betterunderstand the available alternatives for processing a non-coding lesion. The data reported here establish both the fre-quency and specificity of mutagenesis resulting from the inviva SOS-dependent replication of unrepaired AP sites inM13mp2 DNA. Mutagenesis results primarily from an in-crease in single-base-substitution errors. The absence of mu-tants arising from the addition of a large number of bases andthe negligible effect on deletion mutations indicate that, atleast in this system, such events are not induced by AP sites.The spectrum of induced mutations yielded no additionframeshift mutants and only 8% deletion frameshifts. Thislow yield does not reflect a difference in target size forframeshifts versus other events, since frameshift mutants aredetectable at most sites in the lacZ coding sequence and atmany sites in the operator/promoter sequences. Thus, fail-ure to detect larger numbers of these events indicates thatthey are not a major component of depurination-induced mu-tagenesis. The fact that only deletion frameshifts were ob-served suggests that looping out of the template strand con-taining the AP site may occur more frequently than loopingout of the newly synthesized strand.The large absolute and relative increases in base-substitu-

tion mutations (Table 3) provide a large number of mutantsfor analysis of specificity under conditions in which most ofthe mutants are induced. The small increase not dependenton SOS induction or damage to the DNA (Table 1) couldresult from rare bypass ofAP sites under noninduced condi-tions, untargeted mutagenesis, or lesions other than APsites. However, under the conditions used here to damagethe DNA, other lesions (17, 28, 29, 36) are expected to berare relative to AP sites. The results in Table 2 and the obser-vation that 79% of the sites mutated and 82% of the totalsingle-base substitutions are at template purines are alsoconsistent with AP sites being primarily responsible for theobserved increase. This purine site specificity is not imposedby the DNA target; a recent study of UV mutagenesis usingthis same target gave opposite results; eight of nine sites mu-tated and 32 of 35 single-base changes were at template py-rimidines (26).

Previous studies on the reversion of three amber muta-tions in 4X174 DNA have indicated a strong preference for

insertion of adenine residues opposite apurinic sites (21, 32).The base-substitution specificity observed here for 38 sites issimilar; 59% of the mutants result from incorporation ofdAMP (Table 4). This specificity could result from severalfactors. The possibility that the base immediately precedingor following ain AP site can be used as a template to deter-mine the specificity of mutagenesis (by looping out the tem-plate or daughter strand) can explain only 25% of the ob-served specificity. Furthermore, if looping out is a generalmechanism for bypass, one would expect to see a greaterincrease in frameshift mutations than actually observed.This mechanism thus seems unlikely.Because an AP site is a noncoding lesion, incorporation

specificity could reflect nucleotide substrate concentrationsin vivo. Treating cells with DNA-damaging agents results inalterations in intracellular dNTP pools (37), the greatest ef-fect being a 3- to 8-fold increase in the dATP pool. This in-crease could be reflected in the specificity observed here.Alternatively, purified DNA polymerases are known to in-corporate dAMP opposite AP sites in vitro when equimolardNTP concentrations are used (32, 38). Moreover, using sin-gle nucleoside triphosphates, Strauss et al. (33), found a sim-ilar DNA polymerase incorporation specificity oppositeapyrimidinic sites. Thus dAMP incorporation at AP sitesmay not simply be a consequence of pool imbalances butmay reflect inherent properties of the AP site, the replicationapparatus, or both.The transversions characteristic of adenine incorporation

opposite positions of template purines are thought to be thepredominant mutagenic event induced by benzo[a]pyrene(39), and aflatoxin B1 (40), which produce primarily bulkyadducts in DNA. Such lesions could enhance mutagenesisthrough a common AP site intermediate, as suggested by thedata of Schaaper et al. (11) and Drinkwater et al. (9). Alter-natively, dATP may be the preferred substrate wheneverbase hydrogen bonding is not possible, regardless of the ac-tual lesion. This implies that specificity is imposed by thereplication apparatus, a suggestion supported by the recentobservation that, in the absence of mutagen treatment, theprimary base-substitution event in the lacI gene of E. coliobserved on constitutive expression of error-prone replica-tion is a G-C T-A transversion (J. H. Miller and K. B.Low, personal communication).Adenine incorporation specificity has been suggested to

result from preferential binding of adenine residues to theDNA polymerase in the absence of template instruction (33).Measurements with E. coli DNA polymerase I (ref. 41; bind-ing preference, guanine > adenine > thymine > cytosine)and terminal deoxynucleotidyl transferase (ref. 42; bindingpreference, guanine = adenine > thymine > cytosine) do notreflect the specificity observed in this study. However, simi-lar data are not available for other DNA polymerases, andother factors, such as exonucleolytic proofreading (32), areexpected to modulate any potential correlation. This conceptdeserves further study.Although incorporation of dAMP is the most frequent

event, both dTMP and dGMP are incorporated at substantialfrequencies. Incorporation of dTMP occurred at potentialapurinic sites in 14 mutants. However, this nucleotide is in-corporated opposite positions of template pyrimidines eighttimes, resulting in transversions not seen even once in thespontaneous spectrum. These unique mutants could repre-sent incorporation of dTMP opposite apyrimidinic sites, al-though this would imply that apyrimidinic sites are producedat frequencies somewhat greater than predicted from a previ-ous study (6). Alternatively, these mutants could result fromuntargeted events due to the specificity of error-prone repli-cation.Eleven transversions resulted from incorporation of

dGMP opposite potential apurinic sites. Two such transver-

Genetics: Kunkel

Dow

nloa

ded

by g

uest

on

Aug

ust 2

1, 2

021

Proc. NatL. Acad. Sci. USA 81 (1984)

sions were observed among the depurination-induced rever-tants of the 4X174 amber 18 mutation (21). Although lessfrequent than mutants arising from adenine incorporation,these mutants are not rare and should not be ignored in at-tempting to understand what information is used to deter-mine the mutagenic outcome of a lesion.The distribution of mutants is not random. As in measure-

ments of depurination-induced reversion of kX174 ambermutants (21, 32), site-specific differences are apparent here.For example, the middle guanine residue in the hotspot (Fig.1) gave nine mutants. In contrast, no mutants were observedat the adenine residue four nucleotides away (5' direction,Fig. 1), which is known (J. E, LeClerc and N. L. Istock,personal communication) to be mutable by the most frequentincorporation event associated with depurination (incorpo-ration of adenine). Based on current information, five purinesites (Fig. 1) known to produce a mutant phenotype are notfound in this spectrum, three of which could have resultedfrom the most frequently expected transversion. This appar-ent nonrandom distribution of mutants may reflect the rela-tively small number of mutants analyzed or could reflectmore important factors. Of eight purine sites yielding threeor more mutations, seven have purines immediately adja-cent, and the hotspot is in a run of seven purines. Perhapsdepurination may occur more readily in a purine-rich se-quence. Differences in the rate of depurination of double-and single-stranded DNA are known (3) and guanine resi-dues depurinate more readily than adenine residues (3). Thismay explain the preference for guanine versus adenine sitesmutated (Table 4). Alternatively the nonrandom distributionof mutants may reflect differences in the frequency of by-pass of apurinic sites (21).AP sites are highly mutagenic under SOS-induced condi-

tions, resulting in 0.2% mutants per lethal hit (Table 3) for atarget DNA sequence of only 250 bases. This is 5- to 10-foldgreater per lethal hit than the SOS-dependent mutagenesis ofthis same target resulting from UV irradiation of intactM13mp2 phage (26). This highly mutagenic response pre-

sumably results from two properties of AP sites. Unlikemodified bases, which may retain some correct base-codingpotential, AP sites are truly (base) noncoding lesions. In ad-dition, AP sites probably distort DNA to a lesser extent andsterically constrain polymerization less than certain bulkyadducts do (33, 43, 44), resulting in more frequent bypass.The estimates of AP site bypass frequency in vitro are in factquite high for several DNA polymerases (20, 21, 32, 38), incontrast to the results obtained with bulky base modifica-tions (33, 45). These properties of AP sites, in combinationwith the potentially high frequency of their production, makethese lesions interesting for future studies of repair.

I would like to express appreciation to J. Motto and J. Liu forassistance in DNA sequence analysis and to M. Volkert, R.Schaaper, B. Glickman, and J. Drake of this Institute for criticalevaluation of the manuscript. The gift of strains and the advice ofEugene LeClerc (University of Rochester) during the initial stagesof this work were most helpful and are gratefully acknowledged.

1. Lindahl, T. (1979) Prog. Nucleic Acid Res. Mol. Biol. 22, 135-192.

2. Kunkel, T. A., Schaaper, R. M., James, E. & Loeb, L. A.(1981) in Induced Mutagenesis: Molecular Mechanisms andTheir Implications for Environmental Protection, eds. Law-rence, C. W., Prakash, L, & Sherman, F. (Plenum, NewYork), Vol. 23, pp. 63-82.

3. Lindahl, T. & Nyberg, B. (1972) Biochemistry 11, 3610-3618.

4. Lindahl, T. & Andersson, A. (1972) Biochemistry 11, 3618-3623.

5. Grossman, L. & Grafstrom, R. (1982) Biochimie 64, 577-580.6. Lindahl, T. & Karlstrom, 0. (1973) Biochemistry 12, 5151-

5154.7. Margison, G. P., Capps, M. J., O'Connor, P. J. & Craig,

A. W. (1973) Chem. Biol. Interact. 6, 119-124.8. Lawley, P. D. & Brookes, P. (1963) Biochem. J. 89, 127-138.9. Drinkwater, N. R., Miller, E. C. & Miller, J. A. (1980) Bio-

chemistry 19, 5087-5092.10. Strauss, B., Scudiero, D. & Henderson, E. (1975) in Molecular

Mechanisms for Repair ofDNA, eds. Hanawalt, P. C. & Set-low, R. B. (Plenum, New York), Part A, pp. 13-24.

11. Schaaper, R. M., Glickman, B. W. & Loeb, L. A. (1982) Can-cer Res. 42, 3480-3485.

12. Lindahl, T. (1982) Annu. Rev. Biochem. 51, 61-87.13. Duker, N. J., Hart, D. M. & Grant, C. L. (1982) Mutat. Res.

103, 101-106.14. Deutsch, W. A. & Linn, S. (1979) Proc. Natl. Acad. Sci. USA

76, 141-144.15. Brookes, P. & Lawley, P. D. (1963) Biochem. J. 89, 138-144.16. Lawley, P. D. & Martin, N. C. (1975) Biochem. J. 145, 85-91.17. Drake, J. W. & Baltz, R. H. (1976) Annu. Rev. Biochem. 45,

11-37.18. Schaaper, R. M. & Loeb, L. A. (1981) Proc. Natl. Acad. Sci.

USA 78, 1773-1777.19. Shearman, C. W. & Loeb, L. A. (1979) J. Mol. Biol. 128, 197-

218.20. Kunkel, T. A., Shearman, C. W. & Loeb, L. A. (1981) Nature

(London) 291, 349-351.21. Schaaper, R. M., Kunkel, T. A. & Loeb, L. A. (1983) Proc.

Natl. Acad. Sci. USA 80, 487-491.22. Messing, J., Gronenborn, B., Muller-Hill, B. & Hofschneider,

P. H. (1977) Proc. Natl. Acad. Sci. USA 74, 3542-3546.23. Kane, C. M. & Linn, S. (1981) J. Biol. Chem. 256, 3405-3414.24. Taketo, A. (1972) J. Biol. Chem. 72, 973-979.25. Langley, K. E., Villarejo, M. R., Fowler, A. V., Zamenhof,

P. J. & Zabin, I. (1975) Proc. Natl. Acad. Sci. USA 72, 1254-1257.

26. LeClerc, J. E. & Istock, N. L. (1982) Nature (London) 297,596-598.

27. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.Acad. Sci. USA 74, 5463-5467.

28. Karran, P. & Lindahl, T. (1978) J. Biol. Chem. 253, 5877-5879.29. Lindahl, T. & Nyberg, J. B. (1974) Biochemistry 13, 3405-

3410.30. Little, J. & Mount, D. (1982) Cell 29, 11-22.31. Schaaper, R. M., Glickman, B. W. & Loeb, L. A. (1982) Mu-

tat. Res. 106, 1-9.32. Kunkel, T. A., Schaaper, R. M. & Loeb, L. A. (1983) Bio-

chemistry 22, 2378-2384.33. Strauss, B., Rabkin, S., Sagher, D. & Moore, P. (1982) Biochi-

mie 64, 829-838.34. Lockhart, M. L., Deutch, J. F., Yamaura, I., Cavalieri, L. F.

& Rosenberg, B. H. (1982) Chem. Biol. Interact. 42, 85-95.35. Ripley, L. S. (1982) Proc. Natl. Acad. Sci. USA 79, 4128-

4132.36. Bingham, P. M., Baltz, R. H., Ripley, L. S. & Drake, J. W.

(1976) Proc. Natl. Acad. Sci. USA 73, 1273-1277.37. Das, S. & Loeb, L. A. (1984) Mutat. Res., in press.38. Boiteux, S. & Laval, J. (1982) Biochemistry 21, 6746-6751.39. Eisenstadt, E., Warren, A. J., Porter, J., Atkins, D. & Miller,

J. H. (1982) Proc. Natl. Acad. Sci. USA 79, 1945-1949.40. Foster, P. L., Eisenstadt, E. & Miller, J. H. (1983) Proc. Natl.

Acad. Sci. USA 80, 2695-2698.41. Englund, P. T., Huberman, J. A., Jovin, T. M. & Kornberg,

A. (1969) J. Biol. Chem. 244, 3038-3044.42. Kato, K., Goncalves, J., Houts, G. & Bollum, F. (1967) J.

Biol. Chem. 242, 2780-2789.43. Grunberger, D., Blobstein, S. & Weinstein, I. (1974) J. Mol.

Biol. 83, 459-468.44. Fuchs, R. P. P. (1975) Nature (London) 257, 151-152.45. Moore, P. D., Rabkin, S. D. & Strauss, B. S. (1980) Nucleic

Acids Res. 8, 4473-4484.

1498 Genetics: Kunkel

Dow

nloa

ded

by g

uest

on

Aug

ust 2

1, 2

021