Deamination of cytosine-containing pyrimidine photodimers in UV-irradiated DNA. Significance for UV light mutagenesis.

The realization that cytosine in cyclobutyl pyrimidine dimers rapidly deaminates to uracil raised the possibility that this chemical transformation, rather than an enzymatic polymerase error, is the major mutagenic step in UV mutagenesis. We have established a sensitive bioassay system that enabled us to determine the rate of deamination of cytosine in cyclobutyl pyrimidine dimers in plasmid DNA. This was done by in vitro UV irradiation and deamination of a plasmid carrying the cro gene, followed by photoreactivation, and assaying uracils in DNA by their ability to cause Cro− mutations in an indicator strain that was deficient in uracil DNA N-glycosylase. DNA sequence analysis revealed that 27 out of 29 Cro− mutants carried GC → AT transitions, as expected from deamination of cytosine. Deamination of cytosines in the cro gene in UV-irradiated plasmid pOC2 proceeded at 37°C with first-order kinetics, at a rate of (3.9 ± 0.6) × 10−5 s−1, corresponding to a half-life of 5 h. Physiological salt conditions increased the half-life to 12 h, whereas decreasing the pH increased deamination. The temperature dependence of the rate constant yielded an activation energy of 13.6 ± 3.3 kcal/mol. These kinetics data suggest that deamination of cytosine-containing dimers is too slow to play an important role in UV mutagenesis in Escherichia coli. However, it is likely to play an important role in mammalian cells, where the mutagenic process is slower.

UV radiation produces in DNA a multiplicity of photoproducts, of which two are believed to be responsible for most of the killing and mutagenic effects of UV radiation (1). These are the cyclobutyl pyrimidine-pyrimidine dimers, and the pyrimidinepyrimidone 6-4 adducts (2, 3). Since GC 3 AT transitions constitute a major class of UV mutations both in prokaryotes and eukaryotes (4 -6), cytosine-containing dimers are likely to be a major pre-mutagenic class of UV lesions.
It is believed that the key step in UV mutagenesis involves a bypass synthesis reaction, whereby a DNA polymerase replicates through a UV lesion in DNA (reviewed in Refs. 6 and 7). According to this model, the polymerase incorporates with high frequency an incorrect nucleotide opposite the lesion, a potentially mutagenic event. The chemical nature of the DNA lesion is a major parameter in determining the efficiency and specificity of misincorporation opposite the lesion (1). Thus, if a UV lesion undergoes a secondary chemical transformation to yield a different product, this may have a significant effect on its mutagenicity. This is the case with cytosine-containing photodimers which undergo spontaneous deamination to yield uracil-containing photodimers (8). The special interest in this reaction stems from the fact that a C 3 U change is potentially mutagenic, since uracil has the coding properties of thymine (9). Thus, if the spontaneous deamination proceeds fast enough, it adds a component of a non-enzymatic mutagenic reaction to the process of UV mutagenesis (10). Clearly, the rate of deamination is a major factor needed in order to evaluate its biological importance.
The deamination of normal cytosines in DNA into uracils is a slow process under physiological conditions. At 37°C and pH 7.4 it proceeds with a half-life of approximately 200 and 30,000 years in single-stranded and double-stranded DNA, respectively (11). This means a deamination of approximately 100 cytosines/human genome/day. However, saturation of the 5,6bond of cytosine greatly facilitates the rate of deamination. For example, 5,6-dihydrocytosine deaminates at pH 7.0 and 37°C with a half-life of only about 2 h (12). Since in both types of photodimers the 5,6-double bond is saturated, the rate of deamination of cytosine-containing pyrimidine dimers is likely to be much higher than that of cytosine. Previous attempts to determine the rate of deamination of cytosine-containing cyclobutyl pyrimidine dimers (CPDs) 1 led to conflicting results. One set of results suggested that CPDs deaminate in Escherichia coli with a half-life of 5-6 h (13, 14), whereas a second set of results suggested that deamination of CPDs in DNA proceeds with a sharp step kinetics, and is completed within 55 min in phage (10) or in purified phage S13 dsDNA (15), and within 30 min in ssDNA purified from phage S13, or in the virion state (10,15). This time difference is critical for evaluating the significance of deamination, since in E. coli, for example, mutations are believed to be formed within 30 min after UV irradiation (6, 16 -18).

EXPERIMENTAL PROCEDURES
Materials-Plasmid pOC2 is a 5-kilobase pair pBR322 derivative constructed in our laboratory (19). It carries the cro gene of phage , and the bla and kan genes, which confer ampicillin and kanamycin resistance, respectively. DNA photolyase was a gift from A. Sancar (University of North Carolina, Chapel Hill). Radiolabeled [␣-32 P]dATP at 3000 Ci/mmol and Hybond-N nylon membranes were obtained from the Radiochemical Center, Amersham. Dithiothreitol was purchased from Boehringer Mannheim, MOPS was purchased from Sigma, and eosin yellowish and methylene blue were obtained from Riedel-de Haen.
Bacterial Strains-The bacterial strains used in this study are listed in Table I. All the strains constructed in this study originated from E. coli CSH26 which is ⌬(pro-lac). The desired mutations were transferred by P1 transduction (20) from strains RW82 (⌬umuDC::cat) (21), BD2314 (ung::Tn10) (22), and WBM535 (⌬recA::Tn10) (19). In order to construct a ⌬recA::Tn10 derivative of an ung::Tn10 strain, non-revertible Tet s mutants of the ung::Tn10 strain were selected on fusaric acid plates as described (23). This enabled using Tet selection for the subsequent ⌬recA::Tn10 transduction. The lacY gene was introduced into the constructed strains by mating (20) with DP90C/RЈ22 which contains an FЈ episome carrying LacI q Z Ϫ -Y ϩ -Pro ϩ . The cells were then lysogenized (20) with 200ind, in which lacZ is fused to the O R P R operatorpromoter of phage (24). The recA genotype was checked by its extreme sensitivity to UV irradiation at 254 nm. The Ung Ϫ phenotype was verified by a comparison of the transformation efficiency of a normal plasmid to that of a uracil-containing plasmid. An Ung Ϫ strain does not distinguish between these types of plasmids, but Ung ϩ strains attempt to repair the uracil-containing plasmid by releasing from it free uracils, thus producing AP sites (25). When those are present in small amounts they are rapidly repaired by excision repair initiated by AP endonucleases (26). However, if the plasmid is heavily substituted with uracil, it will be degraded and lost. The result is that the transformation efficiency of both plasmids is similar in Ung Ϫ strains, but in Ung ϩ strains the transformation efficiency of the normal plasmid is at least 20-fold higher than that of a uracil-containing plasmid. The uracil-substituted plasmid was prepared from a culture of CJ236ung dut cells. The Umu Ϫ phenotype was checked by assaying for loss of UV-mutability to rifampicin resistance as described (27).
Photoreactivation-The photoreactivation mixture (30 l) contained 100 mM NaCl, 50 mM Tris⅐HCl (pH 7.4), 1 mM EDTA, 100 g/ml bovine serum albumin, 10 mM dithiothreitol, UV-irradiated plasmid DNA (0.3 g), and DNA photolyase (10 ng). The mixture was spread on parafilm as droplets (3 l each) and UV irradiated at 365 nm using a longwavelength UV lamp (Black-ray model UVL-56) which was at a distance of 13 cm from the droplets. Irradiation was for 15 min at a dose rate of 5 Jm Ϫ2 s Ϫ1 , as determined by a UV products radiometer using a UVX-36 sensor. The enzyme was heat inactivated for 10 min at 65°C and the mixture stored at Ϫ20°C. This procedure eliminates cyclobutyl pyrimidine dimers by direct reversal to the original pyrimidine-pyrimidine sequence.
The cro Bioassay System-Plasmid pOC2 was UV irradiated and then incubated to allow deamination. Deamination was usually performed at 0.1 g/l DNA in 10 mM Tris⅐HCl, 1 mM EDTA (pH 7.5). The pH dependence of deamination was assayed in Teorell and Stenhagen's citrate-phosphate-borate buffer (29). Following deamination, the plasmid was subjected to photoreactivation, which monomerized the cyclobutyl pyrimidine dimers. The treated plasmid was introduced in parallel into isogenic Ung ϩ and Ung Ϫ indicator strains by transformation, and Cro Ϫ mutations were assayed on lactose-EMB plates as described (19,30). Experiments were performed 3 times. A typical trans-formation plate contained 4 -6 ϫ 10 4 colonies, and each data point was obtained based on the colony count of 8 plates (a total of 3-5 ϫ 10 5 colonies). Mutation frequencies, their standard deviations, and kinetic data were calculated using the Statview statistics software. The maximal Cro Ϫ mutation frequency for a plasmid irradiated at 400 Jm Ϫ2 was determined by a series of experiments designed to cause maximal deamination. This was done by incubating the DNA at 37°C (pH 6), or at 42°C (pH 7) for 5, 6, or 8 h. Under each of these conditions the same mutation frequency was obtained. Based on these measurements, MF m for plasmid pOC2 irradiated at 400 Jm Ϫ2 was (70 Ϯ 14) ϫ 10 Ϫ4 .
DNA Sequence Analysis of Cro Ϫ Mutants-Dark red colonies of E. coli WBY13T(pOC2) harbor plasmids with mutant cro (24). Plasmids were isolated from dark red colonies using the Promega Wizard Miniprep DNA purification system. The sequence of cro in these mutants was performed by the DNA Sequencing Unit of The Biological Services Department at our institute, by automated DNA sequence analysis in a Applied Biosystems 373 DNA sequencer using Taq DyeDeoxy Terminator Cycle Sequencing.

An in Vivo Assay for Deamination of Cytosine-containing
Cyclobutyl Pyrimidine Dimers-Our approach was to UV irradiate and deaminate plasmid DNA under defined in vitro conditions, and then to measure the deamination of the dimers by assaying the formation of uracils in the DNA using a bioassay. In preparation for the bioassay the DNA was treated with purified DNA photolyase. This enzyme uses light as a cofactor for a direct reversal reaction that converts all types of cyclobutyl pyrimidine-pyrimidine dimers (including uracil-containing dimers) into the original pyrimidine-pyrimidine sequence (31). The net result of deamination and photoreactivation of UVirradiated DNA was the conversion of cytosine residues, that were originally part of cyclobutyl dimers, into uracils ( Fig. 1). Since the latter have the coding properties of thymine, this transformation of a GC base pair into a GU base pair is mutagenic if the uracil is not removed from DNA (32). When such a plasmid is introduced into an Ung ϩ strain the uracil will be eliminated and the DNA will be repaired to the original GC base pair, thus preventing mutations. In an Ung Ϫ strain, the GU mismatch will persist, and after replication, one of the daughter chromosomes will carry an AU base pair instead of the original GC base pair, resulting an elevated frequency of GC 3 AT transitions (Fig. 1).
To assay deamination-induced mutations we used the cro mutagenesis system developed in our laboratory (24). It is based on the repressor properties of the cro gene product of bacteriophage (33), carried on plasmid pOC2. The tester strains have a LacZ Ϫ LacY ϩ genetic background and contain a 200ind prophage, in which lacZ is fused to the O R P R operatorpromoter of phage (200 itself has the immunity region of phage 21). Upon transformation, the plasmid which overproduces the Cro repressor is introduced into the tester strain cells. Under normal conditions, the overproduced repressor binds to the O R operator and represses expression of lacZ, thus blocking the production of ␤-galactosidase. In this case, cells are unable to ferment lactose and thus give rise to white colonies on lactose-EMB indicator plates. A mutation that sufficiently reduces the binding of the repressor to the O R region will cause full or partial derepression of lacZ, enabling lactose fermentation. This will lead to a local reduction of pH on the indicator plates and production of dark red colonies. To avoid loss of the plasmid from the cells during growth, kanamycin is included in the plates (19,24,30). The indicator strains used were an isogenic pair of Ung ϩ and Ung Ϫ strains. In order to eliminate any contribution from the SOS system, we engineered the strain to carry ⌬recA and ⌬umuDC mutations. This inactivates the SOS response in general (34), and the SOS-mutagenic pathway in particular (35,36), thus not allowing the processing of UV lesions into mutations. The conversion of unrepaired GU mispairs to mutations is unaffected by the lack of the UmuDC and RecA proteins.
As can be seen in Table II, when an indicator Ung ϩ strain was transformed with unirradiated plasmid pOC2, a background mutation frequency of (0.2 Ϯ 0.1) ϫ 10 Ϫ4 was observed. A plasmid irradiated with 400 Jm Ϫ2 gave only a marginal increase in mutation frequency (0.6 Ϯ 0.3) ϫ 10 Ϫ4 , as expected from the non-mutability phenotype of the ⌬recA ⌬umuDC indicator strain. When the UV-irradiated plasmid was heated at 37°C for 5 h, then photoreactivated, and assayed in an Ung Ϫ strain, a dramatic increase of up to 200-fold was observed (mutation frequency (40.3 Ϯ 8.4) ϫ 10 Ϫ4 ). These mutations were almost entirely dependent on the combined treatments of UV irradiation, deamination (heating), and photoreactivation of the plasmid, and on the use of an Ung Ϫ strain.
Mutations arising from deamination of cytosines should give rise to GC 3 AT transitions. Indeed, DNA sequence analysis of Cro Ϫ mutants revealed that 27 out of 29 mutants (93%) contained GC 3 AT transitions, as expected. The remaining two mutations were GC 3 CG transversions. We analyzed also 12 mutants obtained from unirradiated plasmid, and found 8 GC 3 AT transitions, 2 GC 3 CG transversions, one deletion, and one insertion. All GC 3 AT mutations (in both irradiated and unirradiated DNA) occurred at cytosines that have one or more adjacent pyrimidines. For UV-irradiated DNA this is expected for mutations initiated by a cyclobutyl photodimer. Taken together these results strongly suggest that our system assays the deamination of cytosine-containing cyclobutyl pyrimidine dimers, and that this reaction occurs at 37°C on a time scale of hours.
Rate of Deamination of Photodimers in cro-The rate of deamination of CPDs is expected to be dependent on their concentration, and thus on the UV dose for a given DNA con-centration. Indeed, as can be seen in Fig. 2, Cro Ϫ mutation frequency was dependent on UV dose, increasing nearly linearly up to 400 Jm Ϫ2 , a dose at which the plasmid contained an average of 20 pyrimidine dimers/molecule. In order to determine the rate of deamination of CPDs we used the plasmid irradiated at 400 Jm Ϫ2 , and incubated it for various periods of time at 37°C. The time-dependent accumulation of Cro Ϫ mutations is represented in Fig. 3. The mutation frequency in the Ung ϩ strain was very low, in the range of 0.4 ϫ 10 Ϫ4 , indicating rapid repair of uracils prior to plasmid replication. In contrast, when assayed in the Ung Ϫ strain, mutation frequencies were dramatically higher, and increased with deamination time (Fig. 3).
The deamination reaction is expected to follow first-order kinetics, described by the rate equation (C*) t ϭ (C*) 0 e -kt , where (C*) 0 and (C*) t are the concentrations of cytosines in pyrimidine dimers before incubation (time ϭ 0) and after t hours of incubation. (U*) t , the concentration of uracils arising from the deamination of the cytosines is (C*) 0 Ϫ (C*) t . After rearrangement the equation becomes, ln(1-(U*) t /(C*) 0 ) ϭ -kt. How can U* be calculated from the observed mutation frequencies? Since the assay detects essentially only mutations at CPD sites, (U*) t is proportional to the mutation frequency at time t (MF t ), and the initial concentration of CPDs, (C*) 0 , is proportional to the maximal mutation frequency (MF m ) observed in our system. Thus, one can assume that (U*) t /(C*) 0 ϭ MF t /MF m . In order to determine MF m we have conducted several experiments for prolonged incubation times at 42°C or at pH 6.0, conditions that enhance deamination (see below). From these experiments, for a plasmid irradiated at 400 Jm Ϫ2 , MF m is (70 Ϯ 14) ϫ 10 Ϫ4 (see "Experimental Procedures"). Thus, the first-order rate constant, k, can be obtained by plotting ln(1 Ϫ MF t /MF m ) as a function of time. Such a plot is shown in Fig. 3B. The rate constant is (3.9 Ϯ 0.6) ϫ 10 Ϫ5 s Ϫ1 , implying a half-life of 5 h for the deamination of CPD in the cro gene.
Temperature Dependence of Deamination of CPD-Deamination of cytosines is temperature dependent. In order to examine the effect of temperature on deamination of UV lesions plasmid pOC2 was irradiated at 400 Jm Ϫ2 and incubated for an hour at different temperatures. The temperature range examined was 25-47°C, a range which includes physiological growth temperatures. The plasmids were then photoreactivated, and assayed in strains WBY11T Ung ϩ and WBY13T Ung Ϫ (Fig. 4). Consistent with previous experiments, the mutation frequency of  pOC2 in the Ung ϩ strain was constant, about 0.4 ϫ 10 Ϫ4 , at all temperatures, indicating rapid repair of uracils prior to replication. In contrast, a large increase in the mutation frequency of pOC2 in the Ung Ϫ strain was observed, from 7 ϫ 10 Ϫ4 at 25°C to 42 ϫ 10 Ϫ4 at 47°C. The increase was exponential, typical of an Arrhenius temperature dependence of reaction rate, and represents most likely the temperature dependence of deamination of cytosine-containing pyrimidine dimers. The first-order reaction rate constants are shown in Table III. They indicate that within the range of physiological temperatures (25-42°C), the half-life of deamination of CPDs varies 9-fold: from 7.4 to 0.8 h. The Arrhenius plot of the rate constants (Fig.  4B) gave an activation energy of 13.6 Ϯ 3.3 kcal/mol.

Effects of Salt and pH Conditions on the Deamination Rate of CPDs-
The intracellular environment contains salts at an estimated concentration of 0.1-0.2 M, and in both prokaryotes and eukaryotes K ϩ is the major intracellular cation (37,38). In E. coli the major anion is believed to be glutamate (37), whereas in eukaryotes a major free anion is chloride (38). In order to evaluate the effect of the intracellular salt environment on deamination, we measured reaction rates of deamination of CPDs in plasmid pOC2 in the presence of 0.15 M KCl, or 0.15 M potassium glutamate. The presence of 0.15 M KCl reduced the reaction rate 2.4-fold (k ϭ (1.6 Ϯ 0.2) ϫ 10 Ϫ5 ), representing a half-life of 12 h (Fig. 5). The rate was the same in 0.15 M potassium glutamate (not shown).
The effect of pH on the deamination of CPDs in UV-irradiated plasmid pOC2 was determined at 37°C. As can be seen (Fig. 6) deamination increased as the pH decreased, with the FIG. 2. UV dose dependence of deamination-induced Cro ؊ mutations. Plasmid pOC2 was UV irradiated at the indicated UV doses, after which it was subjected to deamination by incubation in 10 mM Tris⅐HCl, 1 mM EDTA (pH 7.5) at 37°C for 2 h. The plasmids were then photoreactivated, and assayed for Cro Ϫ mutations in E. coli WBY13TUng Ϫ as described under "Experimental Procedures." Each irradiated plasmid was assayed twice, each time in four independent transformation tubes. The total colony counts for the transformations with plasmids irradiated at 0, 50, 100, 150, 200, and 400 Jm Ϫ2 were 4 ϫ 10 5 , 4.5 ϫ 10 5 , 5.5 ϫ 10 5 , 4.5 ϫ 10 5 , 3.9 ϫ 10 5 , and 3.8 ϫ 10 5 , respectively.
FIG. 3. Kinetics of deamination of cyclobutyl dimers assayed by the production of Cro ؊ mutations. UV-irradiated plasmid pOC2 (400 Jm Ϫ2 ) was subjected to deamination by incubation in 10 mM Tris⅐HCl, 1 mM EDTA (pH 7.5) at 37°C for the indicated periods of time. The plasmids were then photoreactivated, and assayed for Cro Ϫ mutations in parallel in E. coli WBY13TUng Ϫ (closed circles) and in E. coli WBY11TUng ϩ (open circles) as described under "Experimental Procedures." A, kinetics of mutation formation. B, first-order reaction plot for the deamination of dimers, based on the data presented in A. MF t is the mutation frequency at time t, and MF m is the maximal mutation frequency observed in our system. See text for details. The 5-and 6-h data points were not included since they deviated from the earlier points. This is most likely due to the very high extents of reaction under these conditions (60% and more of MF m ). The resultant rate constant is (3.9 Ϯ 0.6) ϫ 10 Ϫ5 s Ϫ1 .

FIG. 4. Temperature dependence of deamination-induced
Cro ؊ mutations. UV-irradiated plasmid pOC2 (400 Jm Ϫ2 ) was subjected to deamination by incubation in 10 mM Tris⅐HCl, 1 mM EDTA (pH 7.5) for 1 h at the indicated temperatures. The plasmids were then photoreactivated, and assayed for Cro Ϫ mutations in parallel in E. coli WBY13TUng Ϫ (closed circles) and in E. coli WBY11TUng ϩ (open circles) as described under "Experimental Procedures." A, temperature dependence of mutation formation. B, Arrhenius plot for determination of the activation energy for deamination of dimers, based on the data presented in A and in Table III. The activation energy calculated from the slope is 13.6 Ϯ 3.3 kcal/mol. first-order rate constants varying nearly 50-fold in the tested pH range of 5-9. This implies half-lives of deamination varying from 0.7 h at pH 5 to 34.4 h at pH 9.

DISCUSSION
The degree of fidelity by which DNA polymerases copy damaged nucleotides is critical to the mutagenic effect of DNA lesions. Thus, any secondary chemical reaction that affects the primary DNA lesion might be a major factor in its final mutagenic outcome. This was suggested to be the case for UV light that produces two major types of DNA lesions: the cyclobutyl pyrimidine dimers, and the 6-4 pyrimidine-pyrimidone adducts. The TT cyclobutyl dimers are stable in DNA, and are not known to undergo any further chemical transformation. However, it was noticed in the early days of nucleic acid photochemistry that cytosines in cyclobutyl dimers were rapidly deaminated to uracils (8). This occurs generally in cytosine derivatives in which the 5Ј-6Ј bond is saturated. Since uracil has the coding properties of thymine, this secondary chemical reaction is potentially mutagenic.
Tessman and co-workers (10,15) have proposed that the major mutagenic step in UV mutagenesis is not due to a polymerase error, but rather due to deamination of CPDs, followed by accurate replication of the uracil-containing dimers. This suggestion is consistent with the observation that sitespecific TT and UT cyclobutyl photodimers in M13 ssDNA were copied with relatively high fidelity in E. coli in vivo (39,40), however, it is not clear whether it provides a general explanation for chromosomal UV mutagenesis in E. coli (41). Clearly, a critical parameter in assessing the importance of deamination in UV mutagenesis is the rate of deamination of pyrimidine dimers in DNA.
The assay system that we have developed measures in vitro deamination of CPDs in UV-irradiated plasmid DNA utilizing a mutagenesis bioassay. The presence of uracils produced by the deamination reaction was assayed by their ability to cause GC 3 AT transitions in the cro reporter gene. This is a sensitive bioassay that enabled us to determine rates of deamination in a biologically active DNA molecule, under physiological temperature and salt conditions. In our assay system deamination of CPDs follows first-order kinetics, as expected, with a rate constant of (3.9 Ϯ 0.6) ϫ 10 Ϫ5 s Ϫ1 at 37°C in 10 mM Tris⅐HCl, 1 mM EDTA (pH 7.5), corresponding to a half-life of 5 h. This result is in agreement with the deamination rate in E. coli in vivo, evaluated to be 2.2 ϫ 10 Ϫ3 min Ϫ1 (half-life 5.3 h (14)). Our results differ from the results of Tessman and co-workers, who reported a sharp step kinetics for the deamination of both ssDNA and dsDNA, with deamination completed within 55 min in dsDNAs from phages (10) or S13 (15), and within 29 min in ssDNA from phage S13 (10, 15). We do not know the reason for this difference, especially since some of the experiments were performed under similar buffer and temperature conditions.
Extensive chemical studies were performed on the deamination of defined synthetic dinucleotide cyclobutyl dimers. As can be seen in Table IV, in all cases studied, a first-order deamination kinetics was observed, and the rate constants were close to the one calculated based on our bioassay. For example, the half-life of the cis-syn cyclobutyl dimer of pdCpdT in 10 mM phosphate buffer at pH 7.0 was 6.8 h at room temperature (42) (Table IV). Since the reaction conditions were not identical in the different studies, the results are not fully comparable. However, the emerging consensus from the data presented in  5. Kinetics of deamination of cyclobutyl dimers in the presence of salt assayed by the production of Cro ؊ mutations. UVirradiated plasmid pOC2 (400 Jm Ϫ2 ) was subjected to deamination by incubation in 10 mM Tris⅐HCl, 1 mM EDTA (pH 7.5), and 0.15 M KCl at 37°C for the indicated periods of time. The plasmids were then photoreactivated, and assayed for Cro Ϫ mutations in E. coli WBY13TUng Ϫ as described under "Experimental Procedures." A, kinetics of mutation formation. B, first-order reaction plot for the deamination of dimers, based on the data presented in A. It yields a rate constant of (1.6 Ϯ 0.2) ϫ 10 Ϫ5 s Ϫ1 .
FIG. 6. pH dependence of deamination of cyclobutyl dimers. UVirradiated plasmid pOC2 (400 Jm Ϫ2 ) was subjected to deamination by incubation at 37°C for various time periods in Teorell and Stenhagen's citrate-phosphate-borate buffer adjusted to the indicated pH values. The plasmids were then photoreactivated, and assayed for Cro Ϫ mutations in E. coli WBY13TUng Ϫ as described under "Experimental Procedures." The first-order rate constant, k, was calculated from the mutation frequency data and plotted as a function of pH. Table IV is that deamination of CPDs in DNA proceeds by first-order kinetics, and with a half-life of several hours, depending on conditions. It is noteworthy that one set of conditions (10 mM Tris⅐HCl, 1 mM EDTA, pH 7.5, 37°C) gave a half-life of 5 h, similar to that evaluated for E. coli in vivo (14). The activation energy for the deamination of CPDs in DNA calculated based on the bioassay gave a value of 13.6 Ϯ 3.3 kcal/mol, remarkably similar to the activation energy of the cis-syn isomer of the cyclobutyl dimer of the dinucleotide dpT-dpC, which was determined to be 13.7 kcal/mol (43). These two values are within the error margin of the activation energy of 17 Ϯ 3 kcal/mol estimated based on analysis of thermal dependence of UV mutations in glutamine tRNA in E. coli in vivo (44).
What is the significance of our data to in vivo UV mutagenesis? The rates of deamination are affected primarily by salt conditions and pH. Under physiological salt conditions (0.15 M KCl or 0.15 M potassium glutamate), deamination was slowed down to a half-life of 12 h. We have not determined the rate of deamination in ssDNA. However, if the dinucleotides are taken as a model system for ssDNA (Table IV), one can anticipate that deamination rates of dimers in ssDNA will not be much different from dsDNA. Another factor that may influence deamination rate is the pH (Fig. 6). In the cell this may be caused by local interactions with proteins or other molecules, and thus alter locally deamination rates.
The 6-4 adducts could not be assayed in our system since they are not photoreactivated by DNA photolyase. However, they do not interfere with our assay since they are most likely repaired by the UvrABC nucleotide excision repair in the indicator strain. The concomitant removal of uracils originating from CPD's by the repair of 6-4 adducts is not likely to be significant since the average interlesion distance in the UVirradiated plasmid is longer than 500 nucleotides (20 or less lesions in a 5-kilobase pair plasmid), whereas the repair patches are of the order of 10 -30 nucleotides (1). Any remaining 6-4 adducts will not give rise to mutations due to the ⌬recA⌬umuDC genetic background of the indicator strain. Studies on the deamination of the 6-4 adduct in the dinucleotide dCpT have shown that it deaminates with a half-life of 152-413 h (depending on conditions), nearly 2 orders of magnitude slower than CPDs (42). Extrapolating to DNA, this suggests that the deamination of the 6-4 adducts is unlikely to be a significant process under physiological conditions.
In E. coli UV mutations are believed to be fixed within 30 min (16 -18). Thus, our results lead us to conclude that deamination of pyrimidine dimers does not play a significant role in UV mutagenesis in E. coli. On the other hand, it is likely to be a significant factor in mammalian cells, where mutagenesis, repair, and replication are much slower (1).  (10) and phage S13 dsDNA (15).