Expression in mammalian cells of the Escherichia coli O6 alkylguanine-DNA-alkyltransferase gene ogt reduces the toxicity of alkylnitrosoureas.

V79 Chinese hamster cells expressing either the O6-alkylguanine-DNA-alkyltransferase (ATase) encoded by the E. coli ogt gene or a truncated version of the E. coli ada gene have been exposed to various alkylnitrosoureas to investigate the contribution of ATase repairable lesions to the toxicity of these compounds. Both ATases are able to repair O6-alkylguanine (O6-AlkG) and O4-alkylthymine (O4-AlkT) but the ogt ATase is more efficient in the repair of O4-methylthymine (O4-MeT) and higher alkyl derivatives of O6-AlkG than is the ada ATase. Expression of the ogt ATase provided greater protection against the toxic effects of the alkylating agents then the ada ATase particularly with N-ethyl-N-nitrosourea (ENU) and N-butyl-N-nitrosourea (BNU) to which the ada ATase expressing cells were as sensitive as parent vector transfected cells. Although ogt was expressed at slightly higher levels than the truncated ada in the transfected cells, this could not account for the differential protection observed. For-N-methyl-N-nitrosourea (MNU) the increased protection in ogt-transfected cells is consistent with O4-MeT acting as a toxic lesion. For the longer chain alkylating agents and chloroethylating agents, the protection afforded by the ogt protein may be a consequence of the more efficient repair of O6-AlkG, O4-AlkT or both of these lesions in comparison with the ada-encoded ATase. ImagesFigure 2Figure 3

Alkylation of DNA is thought to be the mechanism by which alkylnitrosoureas and related alkylating agents exert their biological effects. In particular, the presence of 06-AlkG has been correlated with the toxicity, mutagenicity, clastogenicity and carcinogenicity of these agents (reviewed by Margison & O'Connor 1990; Barbin & Bartsch, 1989). The cytotoxicity of chloroethylating agents appears predominantly to be due to DNA crosslinking (Erickson et al., 1978) whilst the mutagenic effects of simple alkylating agents are probably mediated by 06-AlkG (Abbott & SafThill, 1979) and 04-AlkT (Duran & Wani, 1987) miscoding during DNA replication. The mechanisms by which other biological effects are mediated, such as sister chromatid exchange is not clear. The carcinogenic potency of a variety of alkylating agents has been correlated with the formation, persistence or accumulation of 06-AlkG in DNA (Margison & O'Connor, 1990). However, there are examples where no such relationship exists (Rogers & Pegg, 1977;Silinskas et al., 1985) and it is becoming increasingly evident that the formation and repair of all adducts having a mutagenic potential should be considered in order to evaluate their contribution to the biological consequences of alkylating agent exposure.
The ogt gene encodes a 19 kDa protein with a repair function similar to that of the 18 kDa subfragment of the ada ATase. Kinetic experiments using synthetic oligonucleotides containing modified bases have shown that the rates of repair of 06-MeG by the two ATases are almost identical (Wilkinson et al., 1989). However, the ogt ATase is able to repair 06-ethylguanine 173 times faster and 04-methylthymine 84 times faster than the ada ATase (Wilkinson et al., 1989 Cultured mammalian cells expressing low levels of endogenous ATase activity are ideal model systems for exploring the biological effects of 06-AlkG and 04-AlkT in DNA. Thus vectors containing the entire protein coding region or truncated versions of the ada gene encoding the 06-AlkG/04-AlkT ATase have been transfected and expressed in a variety of cell lines which have then been used to examine the contribution of these lesions to the toxic, mutagenic and cytogenetic effects of alkylating agents (reviewed by Margison & O'Connor, 1990). Since the same subunit of the ada ATase repairs both 06-AlkG and O4-AlkT, it has not been possible to assess the relative importance of each of these adducts. However, as the ogt ATase repairs 04-MeT and longer alkyl chain derivaties of 06-AlkG more efficiently than the ada ATase, a comparison of mammalian cell lines expressing each of these enzymes provides an approach to establish the relative biological importance of these lesions.

Plasmids
A 650 bp EcoRI fragment containing the ogt gene in pUC8 was provided by Dr P.M. Potter (of this Institute) and the vector pZipneoSV(X)l (Cepko et al., 1984) by Dr. R. Mulligan (Massachusetts Institute of Technology, USA).

Recombinant DNA techniques
All recombinant DNA manipulations such as routine subcloning employed standard procedures (Maniatis et al., 1982).
Site directed mutagenesis Site directed mutagenesis was achieved using the Bio Rad Mutagene in vitro mutagenesis kit (Kunkel et al., 1987). Briefly, single stranded M13 DNA containing the ogt gene was isolated from E. coti CJ236 (a strain deficient in dUTPase and uracil glycosylase which therefore allows uracil incorporation in the DNA). A mutagenic oligonucleotide was annealed and the second strand synthesised in vitro by the addition of dTTP, dATP, dCTP, dGTP and Klenow enzyme. Following transformation of this double stranded DNA into a wild type E. coli strain, the uracil containing strand was digested and the remaining mutant strand was replicated. Oligonucleotides were synthesised on a DuPont 3000 Coder using standard phosphoramidite chemistry and purified by urea polyacrylamide gel electrophoresis.
Mammalian cell culture and MTT survival assay The V79 Chinese hamster lung fibroblasts were maintained in modified Eagle's medium (MEM) containing 10% foetal calf serum at 37°C in a humidified atmosphere of 5% C02/95% air. Transfection of plasmid DNA employed Lipofectin (Gibco-BRL, Paisley, Scotland) under conditions described by the manufacturers. Cells were selected and maintained in medium containing 1.5 mg ml' G418 (Gibco-BRL). The MTT survival assay was based on the method of Carmichael et al. (1987). Cells (100/well) were plated into a 96 well plate (Costar) and following a three hour attachment period, varying concentrations of alkylating agents were added. After 5 days, 50 tl of a 3 mg ml-' solution of 3' (4,5,dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) in phosphate-buffered saline (PBS) was added to each well and incubated for 3 h. The media was removed and the formazan crystals formed in the viable cells were solubilised by the addition of 200y1 of dimethylsulphoxide (DMSO). The absorbance at 540 nm and 690 nm were determined using a Titertek Multiscan ELISA plate reader and surviving fractions were calculated as a percentage of the A690-A540 of untreated wells. Standard deviations were determined from values obtained from triplicate wells: the results of one representative experiment for each agent are presented.
Preparation of cell extracts Cells were rinsed with PBS, harvested by scraping and resuspended at I07 cells ml1' in 50 mM Tris-HCI, pH 8.3, 1 mM EDTA, 3 mM dithiothreitol. Following disruption by sonication on ice for two periods of 10 s each at 12 gm peak to peak distance, phenylmethylsulphonyl fluoride in ethanol was added to a final concentration of 0.5 mM and cellular debris was removed by centrifugation (10,000 g for O min at 4°C). Supernatants were used for protein estimation (Bio Rad) and for ATase assay.
Alkyltransferase assay This assay has previously been described . Briefly, cell extracts were incubated with calf-thymus DNA that had been methylated in vitro by reaction with [3H]MNU. Following incubation with the substrate DNA for 2 h at 37°C, the DNA was hydrolysed to acid solubility in 1 M perchloric acid at 75C for 40 min and labelled proteins were recovered by centrifugation and quantitated by liquid scintillation counting. The amount of protein in the cell extracts was determined and the ATase specific activity was calculated as fmole mg-' of total protein. The results shown are the means for at least three different amounts of protein which could be plotted as a straight line with standard deviations less than 10%.

Fluorography
Following incubation with [3H]-methylated substrate DNA, [3H]-labelled proteins were subjected to SDS-polyacrylamide gel electrophoresis and electroblotted to nitrocellulose. The membrane was air dried, wetted with Optiphase Highsafe II scintillation cocktail and exposed to X-ray film at -80°C for 2 weeks. This is based on the method originally described by Margison et al. (1985).

Results and discussion
Ligation of the ogt protein coding sequence into pZipneoSV(X) 1 Site directed mutagenesis of the ogt gene was carried out in order to create an NdeI restriction endonuclease recognition site 5' to the ogt translation initiation codon.  Figure 1 Construction of pZipogtKL involving insertion of the ogt protein coding region into pZipneoSV(X)l following ligation of a double stranded oligonucleotide (linker) containing the eukaryotic translation initiation consensus sequence (Kosak, 1987 (Kosak, 1987) was ligated 5' to the coding sequence ( Figure  1). Subsequent insertion into the mammalian cell expression vector pZipneoSV(X)l (Cepko et al., 1984) created pZipog-tKL. These vectors are selectable in mammalian cells since they contain the neo gene encoding aminoglycoside phosphotransferase which confers resistance to G418. The neo gene and the inserted ogt gene were expressed under the control of Moloney Murine Leukemia Virus (MoMuLV) LTR promoter and polyadenylation signals. The resulting ATase was a fusion protein containing two additional amino acids at the amino terminal end.
Transfection of Chinese hamster V79 cells and isolation of clones expressing the ogt gene The plasmid pZipogtKL was transfected into Chinese hamster V79 lung fibroblasts which express very low levels of endogenous ATase (2-4 fmole mg-'). The parent plasmid, pZipneoSV(X)l was used in parallel to generate G418 resistant control cells. Ten G418 resistant clones were isolated, expanded and assayed for their ATase activity (Table I).
Various levels of ATase were detected in the pZipogtKL transfected cells one clone; LH2, expressed 300 fmole mg-' of total protein and was used in survival experiments. For comparison we used the previously described clone SB (Brennand & Margison, 1986a) which had been transfected with a truncated form of the ada gene and expressed a similar level (300fmolemg'1) of the ada 06-AlkG ATase.
This level of ATase is biologically relevant since human tissues have been shown to contain between 140 and 460fmmg-' of ATase enzyme (Myrnes et al., 1984). The pZipneoSV(X)l-transfected G418 resistant clone 6E was used as the negative control cell line.
Analysis of E. coli ATase expressing clones Southern analysis of DNA isolated from three pZipogtKLtransfected ogt ATase expressing clones ( Figure 2) and clone SB (data previously published by Brennand & Margison, 1986a) confirmed that the respective gene sequences has been incorporated into the cellular DNA. The [32P]-labelled ogt probe (650 bp) hybridised to DNA fragments of the expected sizes following digestion with BamHI, and or HindlII of DNA extracted from clones LH2 and LH7 indicating that there had been no detectable rearrangement of vector sequences essential for expression ( Figure 1). been generated by digestion of a site in the genomic DNA. The reduced size of EcoRI fragments produced may be explained similarly. Other clones (Table I) were not analysed.
To demonstrate that the ATase activities in extracts of LH2 and SB were not due to an increase in endogenous ATase gene expression, the molecular weight of the ATase protein was determined by SDS-PAGE and fluorography. Figure 3 shows that LH2 cell extract contains a single 19 kDa ATase which is the correct size of the ogt proten. SB cell extract contains four [3H]-labelled fusion proteins, the most abundant of which had the expected molecular weight of about 25 kDa (Brennand & Margison, 1986a). The additional bands may have been generated by either functionally active ATase fragments expressed from additional ATG codons in frame with the alkyl-acceptor cysteins residue or by proteolytic cleavage of the 25 kDa ada 06-AlkG ATase. Multiple protein bands following fluorographic analysis have been previously identified when the truncated ada gene was overexpressed in E. coli (Potter et al., 1987). The endogenous Chinese hamster ATase protein is also about 25 kDa (Morten et al., 1992) but the expression in clone SB was unlikely to have been upregulation of the hamster ATase gene since no G418 resistant clones produced from the transfection of the pZipneoSV(X)l parent plasmid had elevated levels of ATase activity. Furthermore, the only situation in which the endogenous hamster ATase has been seen by fluorography was in extracts of cells in which upregulation of ATase had occurred following selection with increasing doses of the chloroethylating agent, mitozolomide (Morten et al., 1992).
There was no statistically significant difference (one-way analysis of variance test) between either the growth rates or the fraction of the cells in the GI, S or G2/M phases of the cell cycle of the three clones LH2, SB and 6E (data not shown). This is important since certain stages of the cell cycle may be more susceptable to damage by alkylating agents (Madox-Jones & Mauro, 1975) and because cells with different growth rates vary in their sensitivities to cytotoxic effects (Wilkinson & Nias, 1971). It can therefore be concluded that any differences in resistance to the toxicity of the alkylnitrosoureas displayed by the clones can not be attributed to differential growth rates or cell cycle parameters.

Functional activity of the ogt A Tase in mammalian cells
To determine the ability of the ogt enzyme to act on alkylation damage in LH2 cell DNA, ATase depletion experiments were performed. Exponentially growing cells (105) were treated with increasing doses of MNU, ENU, or BNU in PBS for 1 h at 37°C and extracts were prepared and assayed for ATase activity. Complete ATase depletion was evident following 1 h treatment with 40 fg ml-' MNU and 50% depletion with 1O jig ml-' (Figure 4a). In contrast, total depletion of ATase activity could not be achieved using ENU or BNU: maximal depletion in cells treated was only 55% with ENU and 40% with BNU ( Figure 4a). The possibility that longer time periods were necessary for the ogt ATase to act on damage in ethylated and butylated DNA was assessed by exposing LH2 cells to 100 g ml-' BNU for 1 h and assaying the ATase activity at time intervals up to 19 h later. The maximum depletion was only 50% and this was after 5 h ( Figure 4b) suggesting that the level of adducts present was inadequate to completely deplete the ATase. Activity had returned to 70% of its original value by 19 h after treatment but the extent to which this incomplete recovery is a consequence of; (1) the slow resynthesis of ATase; (2) the continuing slow rate of action of the ATase on substrate lesions; (3) inhibition of protein synthesis by BNU; or (4) the repair of the adducts by non-ATase mechanisms remains to be established.
The dose dependent ATase depletion in LH2 cells is an indication that the bacterial ogt ATase produced by these cells is capable of acting on alkylation damage in mammalian DNA. Similarly, the ada ATase has previously been shown to [3H]-MNU (Brennand & Margison, 1986b) consistent with the protection of the cells against the toxic and other effects of such alkylating agents. Incomplete depletion of the ATase such as observed after treatment with ENU and BNU (Figure 4a) has also been observed in 06-AlkG ada expressing TG15SB7 Chinese hamster cells following incubation with the former agent (Fox & Margison, 1988). This also indicates a slower rate of removal from the 06 position of guanine of the larger lesions as compared to methyl adducts.

Effects of bacterial A Tases on cell survival
The survival of LH2 (ogt expressing), SB (truncated ada expressing) and 6E (control) cells was determined by MTT survival assays following exposure to increasing doses of three monofunctional alkynitrosoureas; MNU, ENU and BNU and three chloroethylating agents; mitozolomide, BCNU and chlorozotocin ( Figure 5). Statistical analysis of the results by two-way analysis of variance demonstrated that survival curves obtained for LH2, SB and 6E were statistically different (P<0.05) from each other for most of  I  I  I  1 2 3 4 the alkylating agents. The exceptions were that clones SB and 6E had similar sensitivities to the toxic effects of ENU and BNU. The differences in cytotoxicity are unlikely to be related to the clonality of the cells since there no significant differences between their growth rates and cell cycles as discussed above. The D30 values (the dose at which 30% of the cells survived) together with the initial ATase activities of the cells used for the survival assay are shown in Table II. Several other groups have now reported increased resistance to alkylating agents following overexpression of the ada (Samson et al., 1986;Kataoka et al., 1986;Ishizaki et al.,  Dose of agent (,ug ml -1) 1 1987) and human ATase genes (Hayakawa et al., 1990;Kaina et al., 1991;Wu et al., 1992) in mammalian cells. The relative increases in resistance of LH2 and SB cells compared to 6E (based on the D30 values) were not proportional to ATase expression in the cells for any of the drugs tested. Previously, expression in these Chinese hamster cells of the complete ada ATase to a level five fold higher than that of the truncated ada ATase resulted in identical survival curves following exposure to MNU (Brennand & Margison, 1986a). Indian muntjac cells expressing varying levels of ada ATase also displayed similar levels of resistance of MNU (Musk et al., 1989) and it would thus appear that the mammalian cells are unable to exploit the maximum potential of very high levels of exogenous ATase or that some other lesion is responsible for killing the cells.
LH2 cells are more resistant to the cytotoxic effects of MNU than SB cells ( Figure 5, Table II). It has been demonstrated ( Figure 4a) that 40 jig ml' is the minimum dose at which the ogt ATase is saturated in LH2 cells. However, the continued transcription and translation of the ATase results in minimal cytotoxicity at this dose. Since both ogt and ada ATases are able to repair 06-MeG at the same rate (Wilkinson et al., 1989), whereas the ogt ATase repairs 04-MeT more efficiently than the ada enzyme (Wilkinson et al., 1989), the difference between LH2 and SB cell survival may be attributed to the more efficient repair of 04-MeT by the ogt enzyme, suggesting that even though 04-MeT constitutes only 0.7% of the total alkylated bases following MNU exposure (Saffhill et al., 1985) it could be a toxic lesion. However, unequivocal determination of 04-MeT-induced toxicity cannot be established without direct measurement of the adducts. Because longer carbon chain alkylating agents produce a higher percentage of their total base DNA damage at the 04-position of thymine than MNU (e.g. ENU, 1-4.3% and BNU 0.8%, (Saffhill et al., 1985)), we expected a larger difference between the LH2 and SB cell lines with these agents. Indeed, LH2 cells were slightly but significantly more resistant than SB or 6E to ENU and BNU whilst there was no difference between the SB and 6E survival curves ( Figure  5). The human alkyltransferase has also been shown not to provide Chinese Hamster cells with any additional protection against the cytotoxicity of ENU (Wu et al., 1992). As the ogt encoded ATase repairs longer-chain 06-alkylguanine and presumably 04-alkylthymine adducts more efficiently than the ada ATase, it is not possible to assess the relative contribu-tion of these products to the overall toxicity of the longer chain alkylating agents. The minimal protection observed in LH2 cells may indicate that the longer chain alkyl lesions are less cytotoxic than the equivlent methylated lesions or possibly that non-ATase repair is able to deal with the damage more effectively.
The survival data for each of the three chloroethylnitrosoureas generated curves with different shoulder lengths. Since the molecular structures of the chloroethylating agents were different, their rate of uptake by and decomposition in the cells may vary and these effects could give rise to the different shapes of the observed survival curves ( Figure 5). All three clones were resistant to low doses of BCNU but the magnitude of the threshold was increased in cells with an increased ability to repair the damage i.e. the repair mechanisms were saturated at a higher dose.
LH2 and SB both exhibited greater resistance to the chloroethylating agents than to ENU and BNU ( Figure 5) when compared to 6E, suggesting that ATase repairable lesions (e.g. 06-chlorethylguanine and 06-hydroxyethylguanine (Robins et al., 1983) are more cytotoxic than 06-EtG or 06-BuG. As 06-chloroethylguanine forms interstrand crosslinks in the DNA (Kohn, 1977), that are known to be toxic (Erickson et al., 1978), any repair mechanisms which can act on crosslink precursors will prevent crosslink formation and increase the resistance of the cells to chloroethylnitrosoureas. The reaction of ATase protein with chlorethyl crosslink precursors may form a covalent complex with the DNA (Brent & Remack, 1988;Gonzaga et al., 1990) but the extent to which 06-chlorethyl adducts form such a complex in vivo, its toxicity and the mechanism of its repair are unknown. The greater resistance of LH2 cells to the toxic effects of the chloroethylnitrosoureas, may be due to the higher ATase activity of the LH2 cells at the time of exposure to the drugs (Table II) or due to a greater efficiency of the ogt than the ada ATase to repair chloroethyl or hydroxyethyl adducts. However, before any conclusions can be drawn on the relative toxic effects of 06-chloroethylguanine, 06-hydroxyethylguanine, 04-chloroethylthymine and 04-hydroxyethylthymine it will be necessary to determine their rates of repair by the ada and ogt ATases. This work was supported by the Cancer Research Campaign.