Communication Vol. 262, No. 34, Issue of December 5, pp. 16267-16270,1987 THE JOURNAL OF BIOLOGICAL CHEMISTRY

Printed in U. S. A.

Escherichia coli mutT Mutator Effect during in Vitro DNA Synthesis ENHANCED A. G REPLICATIONAL ERRORS*

(Received for publication, May 27, 1987)

Roe1 M. Schaaper and Ronnie L. Dunn From the Laboratory of Genetics, National Institute of Enuironmental Health Sciences, Research Triangle Park, North Carolina 27709

The mechanism of the Escherichia coli mutT mutator effect was investigated using single-stranded phage as a mutational target. In vivo experiments showed that two M13mp2 lacZa nonsense mutants reverted at a higher rate on a mutTl host than on the wild-type host. The specificity of this mutator effect was identical to that observed for E. coli genes: A*T + C*G transver- sions. The mutT effect was subsequently demonstrated in vitro during DNA replication of M13mp2 DNA in cell-free extracts of E. coli. Replication (the single- stranded + replicative form conversion) in mutTl ex- tracts proceeded with a higher error rate than in wild- type extracts, and DNA sequence analysis of the in vitro revertants revealed the specific induction of A*T + C-G transversions. On the basis of the template specificity of the mutTeffect in vitro, we conclude that the mutT effect involves the aberrant processing of A-G rather than T*C mispairs.

Escherichia coli mutator strains provide an opportunity to investigate the mechanisms by which organisms control their mutation rates (1). Several known mutators originate from defects in DNA replication proteins. For example, the mutator properties of mutD strains result from a defect in the 3’ + 5’ exonuclease (or proofreading) activity of the DNA polymerase 111 holoenzyme, the enzyme primarily responsible for the replication of the bacterial chromosome (2). Mutations in the dnaE gene, coding for the polymerase subunit of this enzyme, yield mutator phenotypes as well (3). Related to the process of DNA replication in a wider context are mutators mutH, mutL, and mutS, whose phenotypes stem from a defective DNA-mismatch-correction system. This system is thought to closely follow the DNA replication fork and correct mis- matches resulting from replication errors, thus providing an important contribution to the fidelity of the replication proc- ess (4).

One other mutator, mutT, also called the Treffers mutator (5), exerts its action through an as yet unknown mechanism. Among the mutators, mutT is unique in displaying a strict specificity: only A.T + C . G transversions are induced. For such changes, mutT generally is a strong mutator, 100- to 10,000-fold (6, 7).

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertkement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The mutT mutator gene is recessive to mutT+ (8) and all known alleles of the gene, including one generated by insertion of bacteriophage p, display the same mutator intensity and specificity (7,9). From this it was concluded that the mutator phenotype is caused by the absence of a gene product (7, 9). Experiments with bacteriophage X indicated that DNA repli- cation is required for expression of the mutator phenotype (7, 10).

The mutTl allele is capable of thermally stabilizing tem- perature-sensitive dnaE mutations (7, 11). Since dnuE en- codes E. coli DNA polymerase 111, this result suggested that the mutF gene product participates directly in DNA repli- cation (7,lO). However, how this involvement may specifically prevent A . T + C .G transversions is unknown. The cloning of the mutT gene was recently reported (12).

In this study we have investigated whether the mutT mu- tator effect, previously shown to operate on E. coli and bac- teriophage X genes, operates on single-stranded DNA phages such as M13. If so, this might greatly facilitate the uncovering of the mutT mutator mechanism. A previous report on single- stranded phage S13 indicated that this phage might not be subject to the mutT mutator (B), although the high sponta- neous background in those experiments might have obscured an effect (1). Our results show that the mutT mutator can indeed operate on M13 phage replicating in a mutT host. Furthermore, during in uitro DNA replication of M13 DNA in cell-free extracts, mutTl extracts are more error prone than wild-type extracts and are specifically defective in pre- venting A. G replicational errors.

EXPERIMENTAL PROCEDURES

Strains-Strain KA796 (F-, ara, thi, Aproloc) has been described (13). Strain NR9082 (like KA796 but rnutTI) was constructed by P1 transduction using as donor a mutTl derivative of W3110 obtained from E. C. Cox (Princeton University, Princeton, NJ). Strains NR9044 and NR9084 (rnutTI) are identical to KA796 and NR9082 (mutTI), respectively, except that they also contain the F’(prolocZ-Z-AM15) from strain NR9099 (13) to permit growth of M13mp2. M13mp2 mutant strains A88, A89, T90, and T108 were obtained from T. A. Kunkel (this Institute). They carry single-base substitutions in the lac20 gene (G + A at position 88 or 89, G + T at position 90, or C + T at position 108, respectively) which result in loss of a-complementation (14).

In Vivo Phage Mutagenesis-To measure the reversion frequency of M13mp2 nonsense mutants in mut+ and rnutTl strains, we first obtained single plaques on lawns of either NR9044(rnut+) or NR9084 (mutTI). A large number (50-100) of independent 1-ml phage cultures were then started, each from a separate single plaque and each containing a small inoculum of the corresponding mut+ or mutTl host. After overnight growth, dilutions of the resulting phage were plated on X-gal-IPTG plates to yield plates with up to 20,000 plaques. At this density, blue-plaque revertants can still be adequately scored, as determined empirically. The reversion frequency was calculated by dividing the number of revertants by the total number of plaques. The size of the experiment could be varied by plating multiple aliquots for each tube in parallel.

In Vitro DNA Replication-Extracts of E. coli KA796 and NR9082 were prepared by the lysozyme method of Wickner et al. (15) but using Hepes’ . KOH (pH 7.7) instead of Tris. HCl buffer. DNA repli- cation was performed in 200 pl containing 400 ng of M13mp2 single-

’ The abbreviations used are: Hepes, 4-(2-hydroxyethyl)-l-pipera- zineethanesulfonic acid ss, single stranded; RF, replicative form; X- gal, 5-bromo-4-chloro-3-indolyl-/3-D-galactoside; IPTG, isopropyl-l- thio-P-D-galactopyranoside.

16267

16268 mutT Mutator in Vitro

stranded DNA, 10 mM MgC12, 500 p~ ATP, 50 p~ each of GTP, CTP, and UTP, 1000 p~ (unless otherwise indicated) of each of the four dNTPs, 50 mM Hepes.KOH (pH 7.7), 100 mM KCl, and 20%

by adding 200 pl (10 mM Tris.HC1 (pH 8.1), 20 mM EDTA, and 500 sucrose. Reactions were incubated at 30 “C for 10-30 min, terminated

mM NaCl), and extracted once with an equal volume of a 1:l mixture of water-saturated phenol and chloroform/isoamyl alcohol (241). After ethanol precipitation and resuspension in 100 pl (10 mM Tris. HC1 (pH 8.1) and 0.1 mM EDTA), 10 p1 of the solution were treated briefly with RNase A prior to loading onto an 0.8% agarose gel, the remaining 90 pl being used for transfection.

Transfections-The product DNA was transfected into 2.0 ml of competent cells prepared from strain MC1061 by the CaCl, technique (16). The transfection mixture was spread on 50 to 100 X-gal-IPTG plates with strain CSH50 (14) as indicator to obtain 10,000 to 20,000 infective centers/plate. Further dilutions were also plated to deter- mine the total phage titer. The plates were incubated at 37 ‘C for 48 h before scoring the revertants (blue plaques).

RESULTS AND DISCUSSION

MutT Action on M13 Phage-We measured the relative mutability of bacteriophage M13 propagating in either wild- type or mutT strains of E. coli. For this purpose, we used a series of nonsense mutations in the hcZa portion of M13mp2. These mutants are defective in the a-complementing ability of mp2 and hence produce colorless plaques in an u-comple- mentation assay (14). Reversion of these phage to wild type (or pseudo-wild type) can be detected visually by restoration of the blue-plaque phenotype. A variety of base-substitution pathways may be studied this way since several different amino acid substitutions are permissible at each of the non- sense codons (14, 17) (see below).

Of the four mutants that we tested, two (A88, an amber mutant, and A89, an opal mutant) reverted at similar fre- quencies in a wild-type and a mutTl strain (Table I). How- ever, two ochre mutants displayed a significantly higher re- version frequency in the mutT1 strain than in the wild-type strain. Mutant T90, in two experiments, reverted 6.7- and 4.8-fold more frequently in the mutT strain than in the wild- type strain. Mutant T108 reverted 5.3-fold more frequently. We conclude that single-stranded phage are indeed subject to the action of the mutT mutator and may thus be used to investigate the mechanism of mutT action.

The mutT effect on single-stranded phage seems relatively modest compared with the effect on the E. coli chromosome (1). This probably results, at least in part, from the high background frequencies in phage stocks. In addition, simple

TABLE I Reversion frequency of mp2 h 2 - a after growth on wild-type or

mutTl strains M13mp2 derivatives, carrying nonsense mutations in the lacZa

gene, were grown on wild-type and mutTl strains as described under “Experimental Procedures” and revertants scored by restoration of a-comDlementation (blue-Dlaaue Dhenotme). wt. wild tvpe.

Strain Total uhaze Revertants Freauencv rnutT/rnut+

Experiment 1, phage rnpz(A88) NR9044(wt) 5.31 X lo6 NR9084(mutT) 3.76 X lo6

NR9044(wt) 10.8 X lo6 NR9084(mutT) 8.10 X lo6

NR9044(wt) 2.65 X lo6 NR9084(mutT) 1.88 X lo6 NR9044(wt) 32.2 X lo6 NR9084(mutT) 17.8 X lo6

Experiment 5, phage mp2(T108) NR9044(wt) 14.8 X lo6 NR9084(mutT) 11.3 X lo6

Experiment 2, phage mp2(A89)

Experiment 3, phage mp2(T90)

Experiment 4, phage mp2(T90)

12 9

14 10

3 15

16 42

7 28

2.26 X 2.39 X 1.1

1.30 X 1.23 X 0.95

1.1 x 7.4 X 6.7

0.50 X 2.4 X 10” 4.8

0.47 X 2.48 X 5.3

revertant frequencies represent the sum of several possible base-substitution pathways. Thus, for individual base-substi- tution pathways more substantial effects may occur. We therefore determined the DNA sequence changes in the re- vertants (see below).

Specificity of mutT-The sequencing results for mp2T90 (Table 11) indicate that the increase in mutation frequency in mutT cells is accompanied by a shift in the pattern of base substitutions. In the wild-type strain, the TAA (ochre) codon usually reverts to CAA, an A. T + G . C transition. A more disperse spectrum of changes is observed in the mutT strain. Most notably, the A . T * C. G transversion, represented by conversions of the TAA codon to GAA, TCA, or TAC, ac- counts for 13 of 16 changes (81%). In the wild-type strain, only 2 of 29 (7%) were of this category. Assuming an average increase in overall reversion frequency of 5.5-fold (Table I), the specific increase for the A. T + C. G transversions amounts to about 65-fold. The A. T * C. G specificity ob- served here for bacteriophage M13 is identical with the spec- ificity of mutT for the E. coli or bacteriophage X chromosome (1, 6).

M13DNA Replication in Vitro-The mutagenic action of the mutTl allele coincides with or follows DNA replication, and the m u t P gene product may be directly involved in the process of DNA replication (7, 10). In view of the extensive knowledge available about the replication of single-stranded phages (18), it seemed worthwhile to investigate the accuracy of in vitro M13 DNA replication using extracts derived from wild-type and mutTl strains. The first stage in the replication cycle, the conversion of the viral single strand to the double- stranded form (ss * RF), does not require any known phage- encoded gene products and proceeds readily in E. coli extracts (15) through the combined actions of E. coli RNA polymerase, E. coli single-stranded binding protein, and E. coli DNA polymerase I11 holoenzyme (18, 19).

Fig. 1 shows the products of such DNA replication obtained using extracts from mut+ and mutT cells. It can be seen that both extracts efficiently convert the starting single-stranded template into a mixture of double-stranded RF I (closed and supercoiled) and RF I1 (open) molecules. Once this stage is reached, no further replication takes place because the next step of replication requires the action of a phage encoded gene product (gene 2) which is not produced in these extracts.

Error Rates in Vitro-To determine the accuracy of the in vitro ss -+ RF conversion, we transfected the products of reactions using mp2(T90) as a template into competent cells and plated the resulting infective centers on X-gal plates. An increase in the frequency of blue plaques over the background frequency would indicate the production of revertants during in vitro synthesis and may serve as an estimate of the in vitro

TABLE I1 DNA sequence changes in mp2T90 revertants

Revertants were picked from experiments described in Table I and from related, similar experiments, selecting only one revertant per phage culture to ensure the independent origin of each revertant. Sequencing was performed by the chain-termination method of San- ger et al. (22). wt, wild type.

Strain Revertant codon”

CAA GAA AAA TCA TTA TAG TAT Total

NR9044(wt) 24 1 2 0 0 2 0 29 NR9084(mutT) 2 5 1 5 0 3 0 16

All seven revertant codons can be detected on the basis of the resulting blue phenotype, including TTA and TAT codons (R. M. Schaaper and R. L. Dunn, unpublished data). Changes that cannot be detected are A.T + G . C transitions at the second and third positions, which lead to TGA and TAG nonsense codons.

mutT Mutator in Vitro 16269 1 2 3 4 5 6 7 1 2 3 4 5 6 7

*- I RFII-1 R F I - 1 ss--l

I

"

-.lr

FIG. 1. DNA synthesis in cell-free extracts of E. coli (M13 ss + RF conversion). The reactions were performed as described under "Experimental Procedures" (1000 p~ for each dNTP) but including [cY-"~P]~TTP. Samples were run on an 0.8% agarose gel in the presence of 0.5 pg ml" ethidium bromide. Shown are an ethidium- stained gel (left) and the corresponding autoradiogram (right). Lune I, mp2 ss DNA, lane 2, mp2 RF DNA; lanes 3 and 4, DNA synthesis reaction with KA796 (mut') extract incubated for 0 or 30 min, respectively; lanes 5 and 6, DNA synthesis reaction with NR9082 (mutT1) extract incubated for 0 or 30 min, respectively; lane 7, extract KA796 but no DNA added, incubated for 0 min. The last reaction reveals the source of the band indicated with an asterisk. This band disappeared upon incubation. Identical results (like lanes 1-6) are obtained for reactions with imbalanced nucleotide pools (data not shown).

TABLE 111 Error rates of MI3 DNA replication in E. coli extracts

Single-stranded mp2T90 DNA was replicated by either wild-type or mutT1 extracts in the presence of balanced or unbalanced nucleo- tide pools. The data are combined from four independent samples. wt. wild tvoe.

dNTP 10" x 106 x Revertants total plaques frequency

Strain KA796(wt) GCTA = 1000 pM 26 17.7 1.5 f 0.3 GCT = lo00 pM, A = 10 pM 265 12.7 22 f 7 GCA = 1000 pM, T = 10 pM

GCTA = lo00 pM GCT = 1000 pM, A = 10 pM 100 5.42 19 f 3 GCA = 1000 p M , T = 10 pM

69 5.65 13 f 6

Strain NR9082(mutT) 20 9.90 1.7 f 1.0 21 11.1 1.9 f 0.7

f standard deviation.

TABLE IV DNA sequence of M13 (mp27'90) revertants resulting from in vitro

redieation errors

d N T P Revertant codon"

CAA GAA AAA T C A ' I T A T A C T A T Total

Strain KA796(wt) GCTA = 1000 8 9 6 0 0 0 0 2 3 GCT/A = 100 3 0 6 4 0 0 0 0 4 0 GCA/T = 100 8 2 0 2 2 C O 1 4

GCTA = 1000 1 0 3 2 1 0 2 0 1 8 GCT/A = 100 3 2 4 3 2 0 0 0 1 4 GCA/T = 100 1 0 0 2 3 0 6 0 3 0

Strain NR9082(mutT)

"All seven indicated revertant codons can be scored, including T T A and TAT codons (R. M. Schaaper and R. L. Dunn, unpublished data). Changes that cannot be detected are A . T .-, G .C transitions a t the second and third positions, which yield TGA and TAG non- sense codons.

Concentrations of dNTPs as in Table 111.

error rate per round of replication. The results of a represent- ative experiment are presented in Table 111.

DNA synthesis reactions in which all four deoxynucleotide triphosphate precursors are present a t identical concentra- tions (1000 p ~ ) show that, for both wild-type and mutTl

extracts, no errors are produced above the background of about lo-'. Thus, DNA replication in these extracts must be considered highly accurate. I n vitro studies of DNA replica- tional fidelity have the advantage that imbalances can be introduced in the triphosphate precursor pools to increase error rates, the latter being generally proportional to the ratio of incorrect to correct nucleotide concentrations a t a given template position (20). The results for two such imbalanced conditions (low dATP to promote mutations at the first position of the TAA (ochre) codon or low dTTP to promote mutations at the second or third positions of the codon) are also presented in Table 111. A 100-fold lowered concentration of dATP results in a distinct increase (10- to 20-fold) in the reversion frequency, indicating that, under these conditions, DNA replication errors are made. However, the increase is identical for wild-type and mutT extracts.

A 100-fold lowered concentration of dTTP, in contrast, increases a reversion frequency specific for the mutT strain. The increase is about "-fold. Thus, the mutT mutator effect previously observed only in uiuo can be reproduced during in vitro DNA replication.

Specificity of DNA Synthesis in Vitro-We sequenced 166 of the mp2(T90) revertants from the experiments described in Table 111. The results are presented in Table IV. Low dATP is expected to promote mutations at the first position of the TAA codon and the majority of mutations do occur at that position: 40 out of 40 for the wild-type strain, and 39 out of 41 for the mutTI strain. There is no difference among the mutations at this position for the two strains: the majority are CAA with minor contributions of AAA and GAA.

As expected, lowering of dTTP in the copying reactions, which was specifically mutagenic with mutTl extracts, yielded mutations at the second and third position of the codon. The changes were almost exclusively (29 out of 30) TCA and TAC codons, examples of A ' T -+ C .G transversion mutagenesis. Thus, the specificity of the mutT mutator effect in vitro is identical to that in uiuo.

The factor of enhancement of the A.T + C.G transver- sions in mutT extracts can be calculated from the data in Tables 111 and IV (TCA and TAC codons): (13/1.7) X (291 30)/(2/14) = 52-fold. This represents a minimum estimate because no increase was observed above the background with the wild-type extract. This value agrees well with the 65-fold estimate obtained for the same phage in uiuo.

Nature of mutT-induced Mismatches-An important ques- tion regarding the mechanism of mutT-induced A. T + C . G transversions is whether they originate at template adenine or at template thymine residues. The use of single-stranded DNA in the present experiments allows us to distinguish between the two possibilities. As calculated above, the in- crease in error rate at template-A residues (production of TCA and TAC codons) was more than 50-fold. In contrast, no increase was observed at the template-?' residue (produc- tion of GAA codons): (19 x 10-'/22 x lo-') x (4/41)/(6/40) = 0.6-fold. Thus, the molecular basis of mutT action entails mutagenic events at adenine rather than thymine residues and, more specifically, must involve an A. G mismatch rather than the mutationally equivalent T. C mismatch.

mutT Function and DNA Replication-Our experiments demonstrate the mutT mutator effect during DNA replication in vitro. Previous experiments in uiuo had already indicated a requirement for DNA replication to express the mutator phenotype (7, 10). The coupling between DNA replication and generation of A. T + C . G transversions therefore seems firmly established. However, the exact role of the mutT gene product is still to be determined. The mutator phenotype

16270 mutT Mutator in Vitro

clearly results from the absence of functional mutT gene product (1, 7). Thus, in normal cells, the gene product specif- ically prevents A. T + C . G substitutions. One can envision two ways this could be achieved by preventing the DNA polymerase I11 holoenzyme complex from making A. G mis- incorporation errors during the polymerization step, or by reversing A.G mismatches in a separate repair step after DNA replication. Although the former possibility may be favored (see below), the mutT effect in vitro is formally consistent with both possibilities.

We favor a direct role for the mutF gene product in the polymerization step because the mutTl mutation thermosta- bilizes certain d n a P alleles (7, ll), implying a direct inter- action between the mutT gene product and the a subunit (=polymerase) of the holoenzyme. Of the seven subunits that have thus far been found to be unequivocal parts of the complex (21), only the 8 subunit (which with LY and t makes up the DNA polymerase core) has not been ascribed to a known gene. Identification of mutT with the 0 subunit, as already suggested (12), would provide a solid basis to explain the mutator effect through a direct involvement in the polym- erization process. However, holoenzyme preparations contain other, less defined subunits that may be part of the functional in vivo complex (e.g. Ref. 21), and these must also be consid- ered possible candidates for the mutT gene product. The present in uitro system for assaying mutT mutator activity may be helpful in characterizing the responsible factor (or factors) and in further defining the mechanisms by which the A.G mismatch (which is only one of the 12 possible mis- matches that can be formed during DNA replication) is spe- cifically prevented.

1. 2.

3.

4.

5.

6.

7. 8. 9.

10. 11. 12.

13.

14.

15.

16. 17.

18.

19.

20.

21. 22.

REFERENCES Cox, E. C. (1976) Annu. Reu. Genet. 10, 135-156 Scheuermann, R., Tam, S., Burgers, P. M. J., Lu, C., and Echols, H. (1983) Proc. Natl. Acad. Sci. U. S. A. 80. 7085-7089

Sevastopoulos, C. G., and Glaser, D. A. (1977) Proc. Natl. Acad.

Glickman, B. W., and Radman, M. (1980) Proc. Natl. Acad. Sci. Sci. U. S. A. 74, 3947-3950

U. S. A. 77, 1063-1067 Treffers, H. P., Spinelli, C., and Belser, N. 0. (1954) Proc. Natl.

Yanofsky, C., Cox, E. C., and Horn, V. (1966) Proc. Natl. Acad.

Cox, E. C. (1973) Genetics 7 3 , (suppl.) 67-80 Cox, E. C., and Yanofsky, C. (1969) J. Bocteriol. 100,390-397 Conrad, S. E., Dussik, K. T., and Siegel, E. C. (1976) J. Bacterid.

Cox, E. C. (1970) J. Mol. Bwl. 50, 129-135 Ray, U. (1978) Fed. Proc. 37 , 141 Akiyama, M., Horiuchi, T., and Sekiguchi, M. (1987) Mol. Gen.

Schaaper, R. M., Danforth, B. N., and Glickman, B. W. (1985)

Kunkel, T. A. (1984) Proc. Natl. Acad. Sci. U. S. A. 8 1 , 1494-

Wickner, W., Bmtlag, D., Schekman, R., and Kornberg, A. (1972)

Taketo, A. (1972) J. Biochern. (Tokyo) 72,973-979 Kunkel, T. A., and Alexander, P. S. (1986) J. Biol. Chen. 261,

Kornberg, A. (1980) DNA Replication, W. H. Freeman & Co., San

Geider, K., and Kornberg, A. (1974) J. Bwl. Chern. 249, 3999-

Kunkel, T. A., and Loeb, L. A. (1980) J. Biol. Chern. 255,9961-

McHenry, C. S. (1985) Mol. Cell. Biochern. 66, 71-85 Sanger, F., Nicklen, S., and Coulson, A. T. (1977) Proc. Natl.

Acad. Sci. U. S. A. 40, 1064-1071

Sci. U. S. A. 55, 274-281

125,1018-1025

Genet. 206,9-16

Gene (Amst.) 39, 181-189

1498

J. Bid. Chem. 69,965-969

160-166

Francisco, CA

4005

9966

Acad. Sci. U. S. A. 74,5463-5467