Targeted Mutation at Cytosine-Containing Pyrimidine Dimers

Proc. NatL Acad. Sci. USA Vol. 80, pp. 4446-4449, July 1983 Genetics

Targeted mutation at cytosine-containing pyrimidine dimers: Studies of Escherichia coli B/r with acetophenone and 313-nm light (suppressor mutation/sensitized irradiation/SOS induction/transdimer synthesis) DOUGLAS FIX AND RICHARD BOCKRATH Department of Microbiology and Immunology, Indiana University School of Medicine, Indianapolis, Indiana 46223 Communicated by Richard B. Setlow, April 7, 1983 ABSTRACT We have tested the two-event model for UV mu- dexing signal and a premutational photoproduct that targets the tagenesis producing class 2 suppressor mutations at glutamine tRNA mutation (12-14). While the first of these could be any lesion genes in Escherichia coli. In the model used, the induction/in- that disrupts DNA synthesis to stimulate SOS induction or indexing lesion is any type of pyrimidine dimer and the premuta- dex critical overlapping daughter-strand gaps (15), the targeting tional photoproduct at the target site is a cytosine-containing di- photoproduct could be unique. For example, the transitions mer. Specific mutation-frequency responses were analyzed under accounting for the three examples of class 2 suppressor mu- conditions in which the ratio of thymine-thymine dimers to cy- tation affect cytosine residues to the 3' side of thymine residues tosine-containing dimers was modified by using 313-nm light and in the DNA target sequences. Therefore, the possibility of mu- 0.0%, 0.1%, or 0.2% acetophenone. Changes observed in the production of class 2 suppressor mutations were consistent with the tation at a thymine-cytosine (TC) pyrimidine dimer as the pre- model and suggested that the G-C -- AT transitions responsible mutational photoproduct has been proposed (16-18). This was for class 2 suppressor mutations are targeted by cytosine-con- supported by results showing cytosine-containing dimers to be taining pyrimidine dimers at the mutational sites. a likely premutational photoproduct for class 2 suppressor mutations (19). Cells mutagenized with 254-nm light were held in Mutagenesis in bacteria is now known to have considerable buffer at an elevated temperature and irradiated with 405-nm specificity. This is seen in distinct nonrandom or contrasting light to cause photoenzymatic monomerization of pyrimidine patterns of induced DNA alterations by mutagenic agents and dimers. This irradiation eliminated revertants except in cells suggests targeted phenomena (1, 2). With UV mutagenesis of lacking uracil-DNA glycosylase activity. When repair of uracil the lad gene, a major portion of the data is consistent with tar- residues in DNA was absent, class 2 suppressor mutations spe- geting at cytosine-containing pyrimidine dimers or pyrimidine- cifically became refractory to the 405-nm irradiation. This im- pyrimidine (64) photoproducts (3), but additional features are plicated a TC premutational photoproduct that could degrade clearly important to explain the "hot spots" preferred for these to a thymine-uracil dimer and be repaired by photoenzymatic mutations among all possible sites (4). Reversion mutation also monomerization to establish the cytosine to thymine (uracil) may be used to study specific DNA alterations if nonsense-de- transition responsible for class 2 suppressor mutation. fective auxotrophs are mutagenized. When UV-induced re- We tested the two-event mutagenesis model by assuming vertants of a nonsense-defective strain are selected on semi-en- that there is a qualitative difference in the type of photoproduct riched medium, a pronounced preference for the production of required for the two separate events. Irradiation with 313-nm suppressor mutations at glutamine tRNA genes is observed (5, light alone produces approximately two cytosine-containing di- 6). These class 2 suppressor mutations (7) indicate specific base- mers for every one thymine-thymine (TT) pyrimidine dimer [in pair changes and large numbers are easily distinguished. E. coli (20) or animal cells (R. B. Setlow, personal communi- Therefore, it is possible to associate the frequency of these spe- cation)]. However, in the presence of the photosensitizer ace- cific alterations quantitatively with different mutagenizing ex- tophenone, which transfers triplet state energy to thymine suf- periences. ficient only for the formation of TT dimers (21), the distri- Reversion of nonsense-defective auxotrophic cells can occur bution of pyrimidine dimers can be sharply altered to favor the at several sites in the bacterial genome-e.g., suppressor mu- formation of TT dimers (22, 23). We used this procedure to tations at different tRNA loci and back mutation in the non- introduce various distributions of pyrimidine dimers and de- sense-defective gene (7). The two species of glutamine tRNA termined specific mutation-frequency responses. For the three in Escherichia coli recognize the codons CAA and CAG and are examples of class 2 suppressor mutation, the responses changed encoded by tandem duplicates of the genes glnl and gln2, re- in a manner consistent with the model that induction/indexing spectively (8). Class 2 suppressor mutations result from G'C is by any type of dimer but the premutational photoproduct at A-T transitions at the distal or proximal ends of the anticodon the target site is specifically a cytosine-containing pyrimidine sites in these genes producing de novo ochre suppressor mu- dimer. tations (glnl -- glnl°) and de novo amber suppressor mutations (gln2 -- gln2a) (9, 10) or converted amber-to-ochre suppressor MATERUILS AND METHODS mutations (gln2a gln20) (11). The process by which these suppressor mutations arise may Excision-repair-defective strains of E. coli B/r, WU-11 and WU, require two types of DNA damaging events, an induction/in- were used (24). Cells were grown and irradiated with 313-nm radiation (2.5 W/m2) from a 1,000-W Hg/Xe lamp (19) filtered The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertise- Abbreviations: TT, thymine-thymine pyrimidine dimer; TC, thymine- ment" in accordance with 18 U.S.C. §1734 solely to indicate this fact. cytosine pyrimidine dimer. 4446 Downloaded by guest on September 24, 2021 Genetics: Fix and Bockrath Proc. Natl. Acad. Sci. USA 80 (1983) 4447

through redistilled pyridine (25% in toluene) and by broad and mutations. All responses displayed fluence-squared increases narrow band pass filters (Oriel 5180 and G-521-3130). Just be- and were clearly enhanced by the addition of 0.1% or 0.2% ace- fore irradiation, a 10% (vol/vol) solution of acetophenone (East- tophenone to the irradiation buffer. An example is shown in man Kodak) in dimethyl sulfoxide (22) was diluted into the cell Fig. 2 for the conversion of class 2 suppressor mutations in WU- suspensions to the final concentrations indicated. After irra- 11 in the absence or presence of 0.1% acetophenone. Again, a diation, survival and mutagenesis were assayed on semi-en- dose-adjustment factor could be used to superimpose the data riched media (Difco nutrient broth, 0.2 mg/ml) as described (filled symbols labeled 1.4 in Fig. 2) but the dose-adjustment (24). Specific types of revertants were distinguished by their factor for mutation was not the same as that for survival. ability to propagate nonsense mutants of bacteriophage T4 (NG A selective increase in particular types of dimers by aceto- 75, NG 273, and PS 205) (18, 19). Tyr' reversion mutations pro- phenone could be expected to enhance inactivation and mu- duced by UV in WU-11 are primarily either back mutations of tation differently if survival were a function of any type of di- the UAA codon in the tyrosine gene (tyr- tyr') or class 2 mer and mutation depended, at least in part, on a particular ochre suppressor mutations. Since this strain carries a class 2 type of dimer. All of the data suggested dissimilar dose-ad- amber suppressor mutation (supE or gln2a) that suppresses a leu justment factors for survival and for mutation and thus that in- (UAG) mutation, the induced class 2 ochre suppressor muta- activation and mutagenesis resulted from different sets of pho- tions are either de novo suppressor mutations (glnl -- glnlI) toproducts. To determine whether the differential responses or converted suppressor mutations (gln2a -4 gln20). Strain WU might be consistent with distinct models, simple mathematical does not carry the class 2 amber suppressor mutation and this functions for survival and mutation-frequency responses were allows Leu+ reversion to be studied in addition to Tyr' rever- considered and tested with the data. sion. Leu+ revertants are primarily either back mutations (leu- -* leu+) or de novo class 2 amber suppressor mutations (gln2 ANALYSIS gln2a). Lethality and mutagenesis produced by 313-nm light with or RESULTS without acetophenone are approximately 95% or greater photo- Colony-forming ability of E. coli WU and WU-11 was more ef- reversible (22, 23). This indicates that pyrimidine dimers are ficiently inactivated by 313-nm light in the presence of 0.1% the principal cause of lethality by this irradiation. For mutation or 0.2% acetophenone. The survival curves were superimpos- involving two types of damaging events, photoreversibility in- able if dose-adjustment factors were used (2.2 for 0.1%, 3.3 for dicates only that at least one type is a pyrimidine dimer (15). 0.2%). These factors were derived from the exponential por- Nevertheless, in the following models, survival and mutation tions of the inactivation curves and were the numerical values frequencies are described in terms of pyrimidine dimers. This by which the fluence scales (abscissae) could be adjusted so that is considered further in the Discussion. survival data without acetophenone became congruent with We assume survival after 313-nm irradiation follows some survival data in the presence of acetophenone. An example is function S of the occurrence of dimers (t + c), where t rep- shown in Fig. 1 for WU-11 in the absence or presence of 0.1% resents TT dimers and c represents cytosine-containing di- acetophenone. As a check, cells were treated with longer ex- mers. Total dimers formed by irradiation in the presence of posure to 313-nm light in the absence of acetophenone to pro- acetophenone are indicated as ta + Ca. To describe superim- duce more extensive killing. These survival curves confirmed posable survival curves, we define a dose-adjustment factor for the superpositions (data not shown). survival, R, to relate total dimers produced in the absence and The mutation-frequency responses for Tyr' and Leu+ revertants were also determined and each was partitioned by phage testing to distinguish back mutation and specific suppressor CD,

30 - 1.0 U, / / /

0.5

., 0

0 0.-

4,) 0.1 C00 lo .S 0e

0.05 0 750 1,500 2,250 3,000 3,750 4,500 0.02 I \ 1 Fluence (at 313 nm), j/M2 750 1,500 2,250 3,000 3,750 4,500 FIG. 2. Enhanced production of converted suppressor mutation in Fluence (at 313 nm), J/m2 the presence of acetophenone. Tyr' revertants ofWU-11 were selected on semi-enriched medium after irradiation (Fig. 1) and phage tested to FIG. 1. Enhanced inactivation of WU-11 in the presence of aceto- determine specific mutation examples. Conversion ofthe class 2 amber phenone. Colony-forming ability was determined on semi-enriched me- suppressor to an ochre suppressor mutation is shown-, symbols corre- dium as a function of exposure to 313-nm light. Inactivation in the ab- spond to those in Fig. 1. The three dose-adjustment factors for muta- sence of acetophenone (o) was similar to inactivation in the presence tion (3.3, 2.2, and 1.4) are from the three equations for mutation-fre- of 0.1% acetophenone (A) if adjusted by a dose-adjustment factor of 2.2 quencyresponse. Quadratic curves arefittedto the experimental points (A). ( and--) and the adjusted points (---). Downloaded by guest on September 24, 2021 4448 Genetics: Fix and Bockrath Proc. Natl. Acad. Sci. USA 80 (1983)

presence of acetophenone such that a function of any dimer (t + c) in Eq. 1, of T T dimers~t) in Eq. 2, and of cytosine-containing dimers (c) in Eq. 3: S[R(t + c)] = S[ta + Ca] MF1 (x (t + c)(t + c) [1] or R = (ta + Ca)/(t + c). MF2 x (t + c)(t) [2] The parameter c can be expressed in terms-of t: with 313-nm light, c = 2t (as noted in the Introduction). When acetophe- MF3 xC (t + c)(c). [3] average is because of ab- none is present, the fluence reduced Dose-adjustment factors for mutation, M1, M2, and M3, were sorption by the photosensitizer. The fractionfdenotes the cor- then derived from each equation by algebraic rearrangement in DNA by responding reduction in dimers produced directly and substitution of the terms defined above. For example, one di- 313-nm light. Thus, the production of cytosine-containing can use Eq. 1 and introduce the dose-adjustment factor for mu- mers will be decreased (they are assumed to result only from tation M1 into the function MF1: the direct effect of 313-nm light) to Ca = fc = 2ft. The production of T T dimers by the direct effect will also be reduced MF,[Ml(t + c)Ml(t + c)] = MFi[(ta + Ca)(ta + Ca)] but, since acetophenone photosensitization yields dimers ef- or ficiently, net T T dimer production increases such that ta = nt. MA1 = (ta + Ca)/(t + c) = R. The dose adjustment factor for survival then can be written Similarly, one can determine M2 = (Rn)05 and M3 = (Rf)05. R = (n + 2f)/3. For a comparison of data with irradiation in the absence or From the absorbance of 0.1% and 0.2% acetophenone at 313- presence of 0.1% acetophenone, the relevant values would be = = = mutation- nm, we calculated values for f of 0.83 and 0.70, respectively, R 2.2, n 5, andf 0.83. The three models for for sample irradiation in a spherical watch glass (see ref. 25). frequency responses (Eqs. 1-3) then predict dose-adjustment = = = 1.4. These values were confirmed to within approximately 10% by factors for mutation of M1 2.2, M2 3.3, and M3 These separate studies of survival inactivation at 254 nm using uridine are used in Fig. 2. Clearly, the data are superimposed best by in the irradiation buffer to absorb 254-nm light as acetophenone the dose-adjustment factor 1.4. This selects Eq. 3, the model absorbed the 313-nm light (data not shown). Thus, from the for targeting by cytosine-containing dimers. observed dose-adjustment factors for survival and the above equation for R, one may calculate n = 5 for 0.1% acetophenone DISCUSSION- (R = 2.2) and n = 9 for 0.2% acetophenone (R = 3.3). Comparisons of the mutation data are summarized in Table 1, In a similar way, equations for mutation-frequency responses which includes all types of reversion mutation. Examples of can be formulated to derive dose-adjustment factors for mu- class 2 suppressor mutations all show a satisfactory fit if the dose- tation. Three are considered here. All account for mutation fre- adjustment factor for mutation determined by Eq. 3 is used and quency (MF) increasing as the second power of fluence and as- this is in sharp contrast with the other models (line 4 corre- sume that one event for induction/indexing is a function of any sponds with the data of Fig. 2). Alternative models not con- type of dimer (t + c). A second event for targeting is given as sidered here-for example, assuming that the inducing event Table 1. Z values for comparison of model fitness z 0.1% acetophenone 0.2% acetophenone Strain Mutation M, = 2.2 M2 = 3.3 M3 = 1.4 M1 = 3.3 M2 = 5.4 M3 = 1.5 Class 2 suppressors WU gln2 gln2a 21 67 1.6 38 123 0.63 WU gn1- gln10 15 46 0.13 41 129 0.61 WU-11 gini glnl0 22 65 0.22 39 125 0.14 WU-11 gln2a -- g1n2' 22 67 0.18 47 152 0.66 Back mutations WU leu- leu4 4.3 18 2.5 2.6 15 3.0 Wu tyra-n tyr' 5.2 24 3.8 9.4 39 3.9 WU-11 tyra-n tyr' 1.6 17 5.8 5.2 29 5.5 Other suppressors WU ser sera 6.0 26 4.0 13 40 0.50 WU tyr tyre 1.7 1.4 3.2 1.4 8.2 1.7 WU-11 tyr tyr0 0.44 4.7 3.0 2.6 2.2 4.6 Ten individual examples ofmutation-frequency responses were determined using 313-nm radiation and 0.0%, 0.1%, or 0.2% acetophenone, with data sets from two or three independent experiments. To test the closeness offit provided by any particular dose-adjustment factor, the quadratic curves (as in Fig. 2) were transformed into straight lines by plotting mutation frequency against the square of fluence. The slopes of these lines were then obtained by using least-squares regression methods. The effect of acetophenone on a specific mutation-frequency response was expressed as the ratio ofthe slope under the sensitized condition to the slope under the control condition. The difference between this ratio and the square of each dose-adjustment factor was tested statistically with approximate normal deviate (Z value) tests, where the standard error of the difference was constructed using the approximate formula for the standard error of the ratio of two random variables (26). A Z value larger than 1.96 indicates a significantly poor fit at the 5% level. A fit is considered satisfactory if Z < 1.96. An Apple II Plus computer was used to visualize superimposed data and to determine Z values. Downloaded by guest on September 24, 2021 Genetics: Fix and Bockrath Proc. Natl. Acad. Sci. USA 80 (1983) 4449

is a function of only t or c, fit poorly and therefore are unac- infrequent (2). A bias toward specific error could result because ceptable. Thus the systematic changes in the mutation-fre- the chemical nature of pyrimidine dimers in template DNA is quency responses for all three of the class 2 suppressor mu- significant. The tautomeric configurations of cytosine bonded tations are consistent with induction/indexing by any pyrimidine in a dimer may frequently have a hydrogen atom at the 3-po- dimer and targeting by a cytosine-containing dimer specifically sition of the ring and therefore pair with adenine (16, 17). Al- rather than any dimer or a TT dimer. ternatively, insertion of adenine residues at a noninstructive Table 1 also shows results for back mutation and for class 1 dimer may be an inherent characteristic of DNA polymerase in UAG suppressor mutation (ser -* sera) or class 3 UAA sup- transdimer synthesis (28). This would yield a de facto bias for pressor mutation (tyr -> tyre). The effect of acetophenone on GC to AT transitions at cytosine-containing pyrimidine di- the back-mutation frequency responses suggests a dose-ad- mers. justment factor of magnitude between those given by M1 and M3 (approximately 0.8 times M1). Since several possible base We thank L. M. Wolff for technical assistance; Dr. J. A. Norton for changes could account for back mutations, we have not tried to advice on statistical analysis; and Drs. H. J. Edenberg, R. B. Setlow, rationalize these data. The other suppressor mutations are 5- and D. E. Brash and the reviewers for comments helpful in drafting 10% as frequent as class 2 suppressor mutations. The absence the paper. This work was supported by National Institutes of Health of a consistent satisfactory fit probably reflects the large scatter Grant GM 21788. fac- in these data rather than an intermediate dose-adjustment 1. Miller, J. H. (1982) Cell 29, 11-22. tor. 2. Le Clerc, J. E. & Istock, N. L. (1982) Nature (London) 297, 584- One relatively efficient mechanism may be responsible for 598. the preferential production of the class 2 suppressor mutations 3. Brash, D. E. & Haseltine, W. A. (1982) Nature (London) 298, 189- gin1 -> gin)0, gin2 -- gin2a, and gln2a -* g1n2'. All three re- 192. quire a point mutation that is a GC to AT transition. The se- 4. Todd, P. A. & Glickman, B. W. (1982) Proc. Natl. Acad. Sci. USA at sites 5' 79, 4123-4127. quences the of the alterations could accommodate 5. Cheung, M. & Bockrath, R. (1970) Mutat. Res. 10, 521-523. TC 3' pyrimidine dimers (8-10) (confirmed by sequence anal- 6. Kato, T., Shinoura, Y., Templin, A. & Clark, A. (1980) Mol. Gen. ysis of the E. coli B/r genes in our laboratory; J. Engstrom, Genet. 180, 283-291. personal communication) and therefore all may be targeted pre- 7. Osborn, M., Person, S., Phillips, S. & Funk, F. (1967)J. Mol Biol cisely at a cytosine-containing dimer (a TC dimer). The param- 26, 437-447. eter used as a measure of dimers in the 8. Nakajima, N., Ozeki, H. & Shimura, Y. (1981) Cell 23, 239-249. c, cytosine-containing 9. Inokuchi, H., Kodaira, M., Yamao, F. & Ozeki, H. (1979)J. Mol. mathematical analysis, could measure any photoproduct pro- Biol 132, 663-677. duced directly by 313-nm light but not by acetophenone pho- 10. Inokuchi, H., Yamao, F., Sakano, H. & Ozeki, H. (1979)J. Mol. tosensitization. Therefore, a pyrimidine-pyrimidine (6-4) pho- Biol 132, 649-662. toproduct (3) could be the targeting photoproduct indicated by 11. Person, S. & Osborn, M. (1968) Proc. Natl Acad. Sci. USA 60, 1030- Eq. 3 if it is produced by 313-nm light but not by photosen- 1037. 12. Bridges, B. A. (1966) Mutat. Res. 3, 273-279. sitization. We believe this is unlikely, however, because the (6- 13. Doudney, C. 0. (1975) in Molecular Mechanisms for the Repair 4) photoproduct is not known to be sensitive to photoenzymatic of DNA, eds. Hanawalt, P. C. & Setlow, R. B. (Plenum, New York), monomerization and would have to be produced by 313-nm light Part A, pp. 389-392. at the relevant sites about as frequently as TC. Assuming 9.8 14. Witkin, E. M. & Wermundsen, I. E. (1978) Cold Spring Harbor x 10-2 TC dimers per genome (20), an efficiency of about Symp. Quant. Biol. 43, 881-886. J/m2 15. Witkin, E. M. (1976) Bacteriol. Rev. 40, 869-907. 0.2 for mutation at a T^C photoproduct (e, see ref. 19) can be 16. Bockrath, R. & Cheung, M. (1973) Mutat. Res. 19, 23-32. calculated from the data of Fig. 2. Hence, any photoproduct 17. Person, S., McCloskey, J. A., Snipes, W. & Bockrath, R. C. (1974) less frequent than TC by a factor greater than about 5 could not Genetics 78, 1035-1049. account for the observed mutation frequencies. Furthermore, 18. Bockrath, R. & Palmer, J. (1977) Mol. Gen. Genet. 156, 133-140. 19. Fix, D. & Bockrath, R. (1981) Mol. Gen. Genet. 182, 7-11. studies with 254-nm mutagenesis have implicated a pyrimidine 20. Ellison, M. J. & Childs, J. D. (1981) Photochem. Photobiol. 34, dimer as the premutational photoproduct for class 2 suppressor 465-469. mutation (19, 27) that is specifically a cytosine-containing py- 21. Rahn, R. 0. & Patrick, M. H. (1976) in Photochemistry and Pho- rimidine dimer (19). tobiology of Nucleic Acids, ed. Wang, S. Y. (Academic, New York), One reasonable hypothesis is the following. An activity fa- Vol. II, pp. 123-129. cilitating DNA synthesis on a damaged template is induced or 22. Hodges, N. D. M., Moss, S. H. & Davies, D. J. G. (1980) Pho- indexed tochem. Photobiol. 31, 571-577. by any type of pyrimidine dimer and the cytosine of 23. Mennigmann, H. D. (1972) Mol. Gen. Genet. 117, 167-186. a TC dimer frequently interprets as thymine. Nonrandom in- 24. Bockrath, R., Harper, D. & Kristoff, S. (1980) Mutat. Res. 73, 43- sertion of bases opposite the dimer seems likely, since the fre- 58. quency of class 1 UAG suppressor mutations suggesting cyto- 25. Morowitz, H. J. (1950) Science 111, 229-230. sine to adenine transversion at potential TC sites 17) is small, 26. Colquhoun, D. (1971) Lectures on Biostatistics (Clarendon, Ox- ford), p. 297. the probability of the specific error estimated at TC dimers is 27. Kristoff, S. & Bockrath, R. (1983) Mutat. Res. 109, 143-153. larger than the 0.0625 predicted on a purely random basis (19), 28. Strauss, B., Rabkin, S., Sagher, D. & Moore, P. (1982) Biochimie and tandem base-pair changes at adjacent pyrimidines are too 64, 829-839. Downloaded by guest on September 24, 2021