ABSTRACT
The triphosphate of the nucleoside deoxyribosyl dihydropyrimido[4,5-c][1,2]oxazin-7-one (dP) is known to be incorporated into DNA efficiently by Taq polymerase and is a useful tool for polymerase-mediated in vitro mutagenesis. It is shown here that dP is a potent mutagen in Escherichia coli and Salmonella typhimurium. In E.coli, this deoxycytidine analog induces both GC -> AT and AT -> GC transitions. No induced transversions are observed. It is highly mutagenic in wild-type E.coli, but this is much reduced in a strain lacking thymidine kinase. Mutagenesis induced by dP is efficiently inhibited by the addition of thymidine. Partially purified thymidine kinase from E.coli catalyzes phosphorylation of dP to its 5'-monophosphate. When E.coli was grown in the presence of dP, the nucleoside analog was incorporated into its DNA. The content of dP in DNA was dependent on the concentration of dP added to the medium. The incorporation characteristics of the 5'-triphosphate of dP (dPTP) were also studied using E.coli DNA polymerase I large fragment. The results confirm that this triphosphate can be incorporated opposite A and G in the template with similar efficiencies. This indicates that dP is metabolized as a thymidine analog and that the resulting triphosphate induces a high rate of mutagenesis through replicational errors.
The deoxyribosyl derivative of dihydropyrimido[4,5-c][1,2]oxazin- 7-one (1, dP) was earlier synthesized as a model compound for one of the conformers of N4-methoxydeoxycytidine (2), which results from the action of methoxyamine on cytosine residues in DNA. The formation of N4-methoxycytosine is believed to be a major cause of methoxyamine-induced mutagenesis. In dihydropyrimido[4,5-c][1,2]oxazin-7-one (P), the exocyclic amino substituent is constrained to an anti conformation (with respect to N1), whereas for methoxycytosine, both the syn and anti orientations can exist. Consequently there is no hindrance in P to hydrogen bonding (1 ,2 ). This property, together with the favorable tautomeric state, contributes to the strong pairing of P with both of the natural purines; for this reason it was considered to be a potential mutagen provided it could be metabolized and incorporated into DNA. The triphosphate dPTP causes mutations in vitro (3 ). We have now found that dP has a very potent mutagenic activity on bacteria. The mutagenicity of dP is higher than that of N4-aminocytidine (3), a strongly mutagenic cytidine analog which we have reported previously (4 ). In the present work we have investigated the mechanism of dP-induced mutagenesis. The structures of these nucleosides are shown in Figure 1 . A conference abstract reported briefly on dP mutagenicity (5 ).
Escherichia coli strain CSH50 [[Delta](pro-lac), ara-, thi-/F'(traD36, proAB, lacIQZ[Delta]M15)] was a gift of Dr T.A.Kunkel of NIEHS (Research Triangle Park, NC). CC101-106 were kindly provided by Dr J.H.Miller of the University of California (Los Angeles, CA). A thymidine kinase-deficient strain, SH10 (ilv, trpE, tonB, tdk-1, tsx), was obtained from Dr Akiko Nishimura of the Genetic Stock Center (National Institute of Genetics, Japan).
Mutagenic activities of dP on E.coli CC101-CC106 were assayed as reported (6 ). Briefly, an overnight culture of E.coli cells was centrifuged and the bacteria obtained were washed twice. The bacteria were plated with dP and top agar containing 0.5% NaCl and 0.2 mg/ml Nutrient Broth (Difco) onto minimal lactose plates containing 5 g lactose, 2 g citric acid monohydrate, 10 g potassium dihyrogen phosphate, 3.5 g sodium ammonium hydrogen phosphate tetrahydrate, 0.2 g magnesium sulfate heptahydrate and 15 g agar per liter. After incubation for 2 days, the numbers of revertant colonies were scored. The numbers of living cells were scored on glucose minimal plates in which the lactose in the lactose minimal plates was replaced with the same concentration of glucose. For the spot test, bacteria were plated onto minimal lactose plates without dP and then paper discs (10 mm in diameter) containing dP were placed on the plates. Salmonella reversion tests and mutations to rifampicin resistance were also performed as reported (4 ,7 ). For the rifampicin resistance assay, E.coli was grown for 4 h at 37oC with shaking in liquid LB medium containing 10 g Bacto Tryptone (Difco), 5 g Bacto Yeast Extract (Difco), 10 g NaCl per liter, adjusted to pH 7 with NaOH, supplemented with dP. The bacteria were then plated onto solid LB medium with rifampicin for mutants and onto medium without the antibiotic for survivors after appropriate dilution. The plates were incubated for 2 days at 37oC and the numbers of colonies scored.
dP (8 ) and its 5'-triphosphate, 5'-dPTP (3 ), were synthesized as reported previously. The 5'-monophosphate (5'-dPMP) was synthesized by treating dP (27 mg, 0.1 mmol) with phosphoryl chloride (10 [mu]l, 0.1 mmol) in dry trimethylphosphate (0.25 ml) at 0oC. After 30 min, 5 ml 0.1 M triethylammonium bicarbonate, pH 7.5, were added and the 5'-dPMP was purified by reverse phase HPLC using a Waters C18 (7.8 * 300 mm) column eluted with a linear gradient of 0-10.5% acetonitrile in 0.1 M triethylammonium bicarbonate, pH 7.5, over 20 min at a rate of 7.5 ml/min. Samples were monitored at 260 nm. Approximate retention times were 5'-dPMP 9.4 min and dP 16.7 min. Pure 5'-dPMP (27.4 [mu]mol, 27%) was obtained: 31P NMR [delta](D2O) 2.25 p.p.m.
One milliliter of E.coli CSH50 overnight culture was inoculated into 50 ml LB medium containing 0-10 [mu]g/ml dP. After incubation for 4 h with shaking, E.coli cells were collected by centrifugation and washed twice with cold TEG buffer composed of 25 mM Tris-HCl, pH 7.5, 10 mM EDTA and 50 mM glucose. Cells were then suspended in 1.5 ml TEG buffer supplemented with 154 U RNase T1 and 770 [mu]g RNase A and applied to an automated DNA extractor (Genepure 341; ABI Japan, Tokyo) under a procedure recommended by the manufacturer for the preparation of RNA-free DNA. Purified DNA samples (150 [mu]g) in 25 mM Tris-HCl, pH 7.8, containing 25 mM MgCl2 and 12.5 [mu]M coformycin were hydrolyzed at 37oC for 1 h with 100 [mu]g DNase I and further incubated for 3 h after addition of 10 [mu]g snake venom phosphodiesterase, 10 [mu]g alkaline phosphatase and 5 mM Tris-HCl, pH 8.9 (Y.Goto, H.Hayatsu and T.Negishi, unpublished procedure). The resulting nucleosides were fractionated by HPLC using a Waters Novapak C18 column eluted with 10 mM sodium dihydrogen phosphate, 8% methanol. Retention times for dT and dP were 13 and 17 min respectively. The amount of dP was measured by absorption at 320 nm and of the four normal nucleosides at 254 nm. dP has a substantial absorbance at 320 nm ([epsilon] = 4200), unlike the normal major nucleosides. For identification of P in DNA, DNA samples were analyzed after digestion to nucleotides. For this purpose, DNA was digested as described above but without the alkaline phosphatase treatment. The nucleotides were fractionated by HPLC using a Waters Novapak C18 column eluted with a linear gradient of 0-28% methanol in 10 mM potassium phosphate buffer, pH 4.0, at a rate of 0.7 ml/min and subsequent isocratic elution with 28% methanol. dTMP and dPMP eluted at 11 and 17 min respectively.
Thymidine kinase was partially purified from E.coli as reported (9 ). The reagent mixture for phosphorylation contained 1.5 mM dP (or 1.2 mM thymidine), 10 mM ATP, 10 mM MgCl2, 0.3 mg/ml bovine serum albumin and 70 mM Tris-HCl, pH 7.8. Typically, 75 [mu]l reagent solution were mixed with 25 [mu]l fractionated extract containing 125 [mu]g/ml protein. After an incubation at 37oC, acid-insoluble material was removed from the reaction mixture by the trichloroacetic acid/Freon/trioctylamine procedure (10 ). dP and dPMP were fractionated with the same gradient as used for nucleotide analysis of the DNA digests. Samples were monitored at 254 nm for thymidine and 320 nm for dP. Retention times for dPMP and dP were 17 and 36 min respectively
An 18mer oligonucleotide labeled with Texas Red at the 5'-end, 5'-(Texas Red)-TGTAAAACGACGGCCAGT-3', was purchased from Genset K.K. (Tokyo, Japan) and used as a fluorescent dye primer. The template-primers were obtained by annealing this primer to oligonucleotide templates. Typically, 20 [mu]l of a reaction mixture containing 10 [mu]M dNTP, 1 U Klenow fragment (New England Biolabs), 5 mM MgCl2, 7.5 mM dithiothreitol and 10 mM Tris-HCl, pH 7.5, was incubated at 37oC for 30 min. The enzyme was inactivated by heating at 75oC for 10 min. The elongated products were analyzed in an automated DNA sequencer (Hitachi) after denaturation at 95oC for 3 min.
If dP causes mutations by ambiguous base pairing during replication, dP must be metabolized into dPTP and then incorporated into DNA. In order to confirm incorporation of dP into bacterial DNA, E.coli cells were grown in the presence of the nucleoside. The DNA extracted from the cells was analyzed by HPLC after digestion to nucleosides (Fig. 4 a) and to nucleotides (Fig. 4 b). In both systems, the DNA was shown to contain peaks co-migrating with authentic dP or 5'-dPMP. The peak corresponding to dP did not result from decomposition of free dPMP or dP contaminating the DNA fraction, because the peak was not observed in DNA samples treated only with phosphatase, without DNase and PDase. Figure 5 shows that the content of dP in DNA from bacteria grown in medium containing dP had an approximately linear relationship to the concentration of dP added to the growth medium.
The next question was which enzyme converts dP to 5'-dPMP. This should be a key enzyme for dP metabolism in E.coli. Thymidine kinase is most likely to phosphorylate dP, because E.coli lacks deoxycytidine kinase (13 ). To confirm the involvement of thymidine kinase in dP mutagenesis, we grew cells of E.coli strains proficient (tdk+, CSH50) and non-proficient in thymidine kinase (tdk-, SH10) in medium containing dP and then applied them to plates containing rifampicin to score mutants resistant to this antibiotic. As shown in Figure 6 , dP had potent mutagenicity towards the tdk+ strain, but, in contrast, had very little mutagenic activity in the tdk- strain. The mutagenicity of N4-aminocytidine, which may be phosphorylated by uridine-cytidine kinase (7 ), was no different in the tdk- strain from that in the tdk+ strain (data not shown). The effect of thymidine, a possible competitor for thymidine kinase, was studied next. When 100 [mu]g/ml thymidine were present in a culture of wild-type E.coli with 6 [mu]g/ml dP, the mutant frequency was reduced to only 4% of that with dP alone. These results indicate that phosphorylation of dP by thymidine kinase is the major first step of dP metabolism.
Base pairing specificity of dPTP was studied using a DNA chain elongation reaction catalyzed by Klenow fragment. To analyze the incorporation of dPTP opposite adenine, a fluorescently labeled primer was annealed to an oligonucleotide with adenine at position 7 relative to the 5'-end to form primed oligo-`A', as shown in Figure 8 a. For incorporation of dPTP opposite guanine, we used primed oligo-`G', which is the same primer annealed to an oligonucleotide with guanine at position 7 relative to the 5'-end. These template-primers were incubated with the Klenow fragment and requisite triphosphates. Resulting elongation products were analyzed by automated DNA sequencing (Fig. 8 b). If only dATP was added as a triphosphate substrate, the primer was elongated by 1 nt (lanes 1 and 5). As shown in lanes 2 and 6, dCTP did not pair with adenine, but with guanine. As shown in lanes 3 and 7, dTTP paired with adenine, but not with guanine. In contrast, dPTP was incorporated opposite both A and G with similar efficiency (lanes 4 and 8). The bands in lanes 3, 4, 6 and 8 correspond to molecules of primer + AT, primer + AP, primer + AC and primer + AP respectively. The mobilities of these 20mers were slightly different from each other because they had different nucleotides at the 3'-end.
Previous Tm and NMR studies of oligonucleotide duplexes containing dP proved that dP can form stable base pairs with both adenine and guanine (1 ,2 ). The ambivalent pairing nature together with efficient conversion to dPTP in E.coli cells must be the basis for dP showing potent mutagenicity in E.coli and this must be so too for S.typhimurium. A probable pathway for the mechanism of this mutagenesis is summarized in Figure 9 . The nucleoside is first phosphorylated in bacterial cells by thymidine kinase. All the present data support this view. However, there might be a partial contribution of the free base liberated from dP by the action of cellular thymidine phosphorylase, which catalyzes conversion of thymidine to thymine and 2-deoxyribose-1- phosphate, and a reversal of the reaction. Thus we treated dP with thymidine phosphorylase (Sigma Chemical Co.) and degradation of dP was analyzed by HPLC. Although the velocity of degradation into the putative P base was over three orders of magnitude smaller than the thymidine to thymine reaction catalyzed by the same enzyme, dP can be metabolized to a state of apparent equilibrium. The putative P base purified by HPLC was found to be non-mutagenic in E.coli CC102 and CC106 (E.Mito and K.Negishi, unpublished results). Therefore, a contribution of the product of phosphorylase action to mutagenesis seems to be very unlikely.
This work was supported by Grants-in-Aid for Scientific Research (no. 07839010) and for International Scientific Research (Joint Research, no. 08044291) from The Ministry of Education, Science, Sports and Culture, Japan. Y.I. was a visiting student from the Department of Biotechnology, Okayama University. We thank her supervisor, Prof. Hiroshi Kanazawa of the Department of Biotechnology, for his constant encouragement to her. We also thank Dr David Loakes of the Laboratory of Molecular Biology, the Medical Research Council, for his advice on this research.
*To whom correspondence should be addressed. Tel: +81 86 251 7262; Fax: +81 86 254 0299; Email: knegishi@grrsews1.okayama-u.ac.jp
+Present address: Department of Chemistry, University of Sheffield, Brook Hill, Sheffield S3 7HF, UK
REFERENCES
Return