Role of Nucleotide Excision Repair in Processing of 04-Alkylthymines in Human Cells*

04-Alkylthymines have been implicated as potential carcinogenic DNAlesions. We have studied the effects of 04-methylthymine, 04-ethylthymine, and 04-n-propyl- thymine in a model system in which a single lesion was located at a defined position on a SV40-based shuttle vector and have found large differences in the effects of these lesions in repair-proficient and nucleotide exci- sion repair-deficient cells. In repair-competent human HeLa cells, normal fibroblasts, and XP-A (205) revertant cells, all 3 residues were highly mutagenic; a mutation frequency of 40% was found for both 04-methylthymine and 04-ethylthymine, whereas that of 04-n-propylthym- ine was 4 2 % . These frequencies were independent of the activity of the 06-alkylguanine DNA alkyltrans- ferase. All three 04-alkylthymines induced T * C transitions

* This work was supported by Grant NKI 87-17 from the Dutch Cancer Society, NKB, and a grant from the Cancer Research Campaign. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduerthis fact. was obtained that the miscoding potential of these residues might be very high (up to 80-90%) (4-6). During DNA replication in HeLa cells, however, the frequency of mutations induced by 04-ethythymine (04-EtT) appears to be much lower (-20% overall or -40% per adducted strand) (7). Other studies showed that in mammalian cells the persistence of 04-AlkT lesions varies with the size of the alkyl group (8-10); the half-life of 04-methylthymine (04-MeT) in DNA of cultured cells or rat liver varies from 2 t o 20 h, whereas 04-EtT has a half-life of 2-20 days. The processing of these lesions thus appears to depend on the size of the alkyl group (11). These observations indicate that 04-AlkT residues are actively repaired in mammalian cells.
This repair could be mediated by 06-alkylguanine DNA alkyltransferases, nucleotide excision repair, or an as yet unknown mechanism (4, 9, 12, 13). In Escherichia coli 04-AlkT lesions are substrates for the ada-and ogt-encoded alkyltransferases that normally act on 06-alkyldeoxyguanosine residues in DNA (9, 14, 15). 04-EtT can also be removed by nucleotide excision repair if alkyltransferases are not induced and there is virtually no repair of these lesions in bacterial cells that lack both the alkyltransferases and the nucleotide excision repair system (16).
The mammalian alkyltransferase efficiently removes alkylgroups from the 0-6 of dG, and, while it was thought that this protein could not or could very poorly repair 04-AlkT (9, 17,18,19), one study reported the presence of low levels of a 04-MeTspecific transferase-like activity in human liver (12), whereas another suggested that the removal of 04-EtT residues occurs by a non-transferase mode of repair (13). Recently it was shown that the yeast and human alkyltransferase can bind to doublestranded oligodeoxynucleotides that contain 04-MeT, although with a low affinity (20). Evidence for the actual removal of 04-MeT by both of these alkyltransferases is now emerging (21,22). In mammalian cells, it might also be that alkylated nucleotides are repaired by nucleotide excision repair. Evidence for the removal of 06-alkyldeoxyguanosine by this repair mechanism in human cells has been published (23)(24)(25).
In the present study, we investigated whether 04-AlkTs in human cells are substrate for nucleotide excision repair by comparing the mutagenic effects of individual 04-MeT, 04-EtT, and 04-nPrT moieties during replication in HeLa cells, normal (SV40-transformed) human fibroblasts and excision repair-deficient xeroderma pigmentosum cells of complementation group A (XP-A). All residues were located at a single position of a SV40-based plasmid that was transfected into the various cell lines and allowed to replicate transiently. We also determined the mutagenicity of 06-MedG in relation to the activity of the alkyltransferase in the same set of cell lines. Our experiments 25521

Repair and
Mutagenicity of O4-A1kylT in Human Cells provide evidence that the mutagenicity of 04-AlkTs in human cells is not influenced by the alkyltransferase activity but is primarily caused by incomplete excision repair.

MATERIALS AND METHODS
Enzymes a n d Cells-All enzymes used were purchased from Boehringer Mannheim.
HeLa cells were obtained from the American Type Culture Collection and grown in Dulbecco's modified Eagle's medium supplemented with 8% fetal calf serum. The SV40-transformed excision repair-proficient human fibroblasts (SV40 wild type Amsterdam, referred to as WtA fibroblasts) and SV40-transformed XP-A cell lines (xP2OS-SV and XP12RO-SV) were a gift from Dr. J. H. J. Hoeijmakers (Erasmus University, Rotterdam). The XP2OS-SV revertant cells (clone A24) were a gift from Dr. P. B. G. M. Belt (Veterinary Institute Lelystad, The Netherlands). These cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum.
Hanahan-competent E. coli DH5a cells (transformation efficiency 108-109/pg) were obtained from Life Technologies, Inc. and used as recommended.
Plasmids-The plasmids used were described earlier (7, 26); both contained the complete SV40 early region and origin of replication. pSVsupF carried the alkylated nucleotide at the unique ClaI site and contained the p-lactamase (Amp') gene for selection in bacteria. Plasmid pSVctr served as internal unmodified control and carried the bacterial chloramphenicol resistance (Cm') gene as selection marker.
Construction of Site-specific Modified Plasmids-Per construct, 250 pg of plasmid DNA was digested with EcoRI and BglII in order to remove the endogenous target sequence and mixed with an equal amount of BamHI-linearized plasmid to a concentration of 20 pg of DNNml in a buffer containing 50% (v/v) deionized formamide, 50 mM Tris-HC1 (pH 7.5), 2 mM EDTA, 0.3 M sodium acetate. This sample was denatured by heating for 10 min at 85 "C and allowed to reanneal slowly by cooling from 60 "C to room temperature. The DNA was precipated with ethanol, and the gapped duplexes were separated from the linear starting material by electrophoresis on agarose gels and subsequent electroelution (7).
Purified single-stranded oligodeoxynucleotides were phosphorylated and ligated into the gapped plasmids as described (7). The resulting covalently closed constructs were separated from remaining nicked plasmids on 4 ml of CsCl density gradients. All constructs were analyzed for the presence of the various lesions at the appropriate site by restriction enzyme cleavage.
Mutagenicity Assay-Cells were seeded into 60-mm (diameter) Petri dishes and transfected 24 h later at approximately 50% confluence using the calcium phosphate coprecipitation technique of Chen and Okayama (29). Per dish, 6 pg of DNA was transfected consisting of 1 pg of unmodified or adducted pSVsupF, 1 pg of pSVctr, and 4 pg of PA-CYC184 as carrier. The precipitate was left on the cells for 12-14 h. The cells were washed twice, trypsinized after 8-24 h of additional growth and divided over two new dishes. Cells were lysed after 68 h, and the plasmid DNA was reisolated using a modification of the Hirt extraction procedure (7,30). The complete DNA isolate was digested with DpnI in order to remove unreplicated input material; DpnI specifically cuts DNA that carries the E. coli dam methylation pattern. This pattern is lost after replication in eukaryotic cells. One-tenth of this material was AAF-) dG (marked with *) with respect to the origin of replication on plasmid pSVsupF. These modified nucleotides were all located in a single-stranded oligodeoxynucleotide (5'-AATTCATC-GATATCTA-3', broken underline) which was inserted into gapped duplex forms of the plasmid. Within the oligomer sequence, the modified residues were placed at the unique ClaI site (solid underline). The ClaI cleavage sites are indicated by vertical arrows. The leading and lagging strand during replication are indicated with t and -+, respectively. directly transformed into Hanahan-competent E. coli DH5a cells, and a n equal portion was additionally cleaved by ClaI before transformation. The mutation frequency was calculated from the relative numbers of transformants obtained mutation frequency = number of colonies obtained from ClaI + DpnI digested matenallnumber of colonies obtained from DpnI-digested material. Mutant DNAs were reisolated from E. coli colonies and analyzed using restriction enzyme digestions and dideoxy sequencing on double-stranded plasmid DNA (7, 31). Per lesion and per cell type, 2 4 0 clones were analyzed by restriction fragment analysis, and, of these, 2-10 were sequenced.
Replication Impairment of Adducted Plasmids-In all transfections, the unmodified pSVctr plasmid was cotransfected as internal control. Since both plasmids carried different selection markers, the replication inhibition of adduct-carrying plasmids could be measured from the relative numbers of Amp' versus Cm' E. coli colonies obtained after transformation of DpnI-cleaved material: relative replication = (number of Amp' colonies/pg of DNA transfected)/(number of Cm' coloniesipg of DNA transfected). We also determined the ratios ofAmp'E. coli colonies obtained from two parallel transfections with modified and unmanipulated pSVsupF. Average values were calculated from at least two independent sets of experiments.
Alkyltransferase Activity Assay-Determination of the alkyltransferase activity in the different cell lines was performed essentially as described elsewhere (32,33). Cells were trypsinized, washed with phosphate-buffered saline, and pelleted by Centrifugation for 5 min at 1500 rpm. Cells were resuspended (2 x lo6 cells/ml) in a buffer of 50 mM Tris-HC1 (pH 8.3), 1 mM EDTA, 3 mM dithiothreitol containing 5 pdml leupeptin and disrupted by sonication (10 s at 10 p followed by 10 s at 16 p) after which point phenylmethylsulfonyl fluoride (87 pg/ml) was added. Cellular debris was removed by centrifugation for 10 min at 13,000 rpm at 4 "C. Allquots of the supernatants were incubated for 1 h at 37 "C with calf thymus DNA(-14.5 Cilmmol) that had been methylated by reaction with [3H]MNU. Specific activities were calculated from the fmol of methyL3H transferred to normalized amounts of protein.
Values represent the mean of five independent determinations.

RESULTS
Analysis of Site-specific Modified Plasmids-Constructs with an 04-AlkT-, an 06-MedG, or both an 04-EtT plus dG-C8-AAF' residue at the unique ClaI site were obtained after ligation of chemically synthesized oligodeoxynucleotides carrying just one of the modified nucleotides into plasmid molecules that contain a gap at the desired region (see "Materials and Methods"). A schematic representation of the region of modification of the constructs is given in Fig. 1.
Before ligation, all oligomers were extensively purified using reversed phase fast protein liquid chromatography. Levels of contaminations were below detection limits (51%) with analytical fast protein liquid chromatography (Fig. 2a). The modification of the oligomers is also demonstrated by the altered retention times (Fig. 2a). The oligodeoxynucleotides were enzymatically hydrolyzed to nucleosides and analyzed with reversed phase HPLC. Examples of these analysis are shown in

Mutation frequencies found after replication of different 04-T alkylated pSVsupF plasmids in human cell lines with different repair capacities
The single 04-AlkT residue was located at the unique ClaI site of the plasmids. These damaged plasmids and controls were transfected into HeLa cells, normal fibroblasts, and nucleotide excision repair deficient XP-A cells, allowed to replicate transiently, and reisolated. Mutation frequencies were calculated from the fraction of ClaI resistant plasmids among the total number of progeny clones.  28). These are due to experimental variations in the hydrolysis and chromatography steps. The modified nucleosides had the appropriate W absorption characteristics (Fig. 2c) and comigrated with coinjected reference nucleosides. The 04-EtT containing oligomer sample was also analyzed by 32P postlabeling in previous studies (7) and appeared to contain less than 1% impurities. Purified adducted plasmids were subjected to restriction fragment analysis. Plasmids carrying a lesion at the CZaI site could not be cleaved by CZaI, whereas unmodified controls could. Enzymes recognizing adjacent sequences cleaved both unmodified and adducted plasmids.

Mutagenicity of Individual 04-AZkT Lesions in Human
Cells-Unmanipulated (wild type) plasmids, controls obtained after insertion of an unmodified oligomer or site-specific modified plasmids were transfected into the different repair-proficient and repair-deficient human cells. Plasmids were allowed to replicate transiently and were reisolated; unreplicated material was excluded from further analysis. Replicated plasmids were analyzed for the presence of mutations at the original region of modification (CZaI restriction site) and mutation frequencies were established from the relative numbers of mutated plasmids among the total amount of progeny (see "Materials and Methods"). Table I shows the mutation frequencies found after replication of the different constructs in the chosen set of cell lines. Except for the clones derived from wild type plasmids, we verified the presence of a mutation on a large number of clones by means of restriction fragment analysis. The reliability of the CZaI selection was also verified by the fact that cotransfected control plasmids (also carrying a CZaI site) were completely linearized. Progeny of untreated plasmids isolated from all cell lines carried (less than) 0.2-1% mutations. These clones were not analyzed further.
Insertion of an unmodified oligomer slightly increased the mutation frequency to -2%. This frequency was found with all cell lines tested. Table I1 shows the spectrum and distribution of the mutations found after sequencing of the mutant clones. The mutations, single base pair substitutions or small deletions, appeared to be randomly distributed over the region of modification. In addition, multiple mutations were found with  The presence of an 04-MeT or an 04-EtT increased the mutation frequency to -20% among progeny plasmids derived from repair proficient HeLa cells and WtA fibroblasts. Almost all mutations were found at the position of the alkylated T and consisted of a substitution into dC. The remainder of the mutations were multiple mutations of the type described above (see Table 11).
The mutation frequency induced by 04-nPrT in the two repair proficient cell lines was approximately half of that induced Mutagenicity of O4-Alky1T in Human Cells 25525

Mutation frequencies among progeny of pSVsupF plasmids carrying individual 04-EtT residues after transfection into different XP-A and XP-A revertant cell lines
These frequencies were also determined for undamaged (Wt) plasmids. Values were calculated from the relative numbers of progeny plasmids harboring an sequence alteration at the region of modification

+04-EtT
XP-20s 1 f 1 6 2 XP-20s rev  (Table I). In all sets of parallel transfections, we reproducibly found lower mutation frequencies for 04-nPrT than for 04-MeT or 04-EtT, and for the mutation frequencies of 04-EtT and 04-nPrT in HeLa cells, the difference was statistically significant ( p < 0.02). 0 4 -n P f l also exclusively induced T "-f C transitions. The few other mutations found consisted of the multiple mutations that were also observed with the analysis of controls containing unmodified oligomers (Table 11). We could not determine the mutagenicity of 04-AlkT in XP-A cells, since all 3 residues extensively inhibited plasmid replication (see below). The few replicated clones that were isolated carried almost exclusively the wild type sequence. From 04-EtT-adducted plasmids, only two clones were obtained that carried the specific T --f C transition. However, since plasmid replication was so severely impaired, it is not clear whether these two mutants were specifically induced by 04-EtT.
Interference with Plasmid Replication in XP20S (XP-A) Cells-In order to measure selective loss of the plasmids due to replication inhibition by the modifications, we calculated the ratios for the (normalized) amounts of progeny derived from adducted pSVsupF plasmids and the corresponding unmanipulated pSVsupF controls in parallel transfections. These amounts were determined using E. coli transformations (as is described under "Materials and Methods"). Mean ratios from at least two independent sets of experiments are depicted in Fig.  3. In addition, adducted plasmids were always cotransfected with equal amounts of an unmodified control plasmid (pSVctr). We also determined the ratios relative to the amount of progeny derived from pSVctr control plasmids (Fig. 4).
In both of the excision repair-proficient cell lines, adducted constructs and controls carrying an unmodified oligonucleotide yielded approximately equal amounts of replicated plasmids (ratios were -0.9 for HeLa and -0.7 for WtA fibroblasts, Fig.   3). When ratios were established relative to the replication of cotransfected pSVctr plasmids in these cells, we also found similar values (-0.4 for HeLa and -0.5 for WtA fibroblasts). However, in XP20S cells, plasmids carrying either one of the 04-AlkT lesions were less well replicated relative to both of the controls (undamaged pSVsupF and cotransfected pSVctr); the amount of progeny obtained from adducted constructs was only 5-8% (ratios of 0.05-0.08) of that obtained from wild type plasmids transfected in parallel experiments, while controls canying an unmodified oligodeoxynucleotide replicated almost as well as the untreated plasmids (Fig. 3).

Effects of 04-EtT in Other XP-A cells and a XP-A Revertant
Cell Line-To ascertain that the replication inhibition in XP20S cells was indeed caused by the nonfunctional XP-A factor, we introduced constructs with an 04-EtT into another XP-A cell line (XP12RO-SV) and revertant cells (clone A24) from the XP2OS-SV line.
X P 2 0 S cells are homozygous for a splicing mutation in intron 3, which results in two abnormally spliced XPAC mRNAs;  4 (34, 35). Both mRNAs yield largely truncated XP-A factors. The XP20S revertant cells have not been characterized at a molecular level, but their UV survival capacity is nearly equal to wild type levels (36). XPl2RO cells contain severely reduced amounts of normal sized D A C mRNA. However, this RNA contains a nonsense mutation in the fifth exon, giving rise to a truncated (207-amino acid) XP-A protein that lacks the 67 Cterminal amino acids (34, 35).
The replication of wild type and 04-EtT-adducted pSVsupF plasmids relative to the replication of cotransfected pSVctr controls in all cell lines is shown in Fig. 4. The relative amounts of the two unmodified plasmids undergoing replication appear to be cell line-specific. Similar ratios were found for adducted plasmids replicating in excision repair-proficient HeLa cells, fibroblasts, and XP20S revertant cells. Adducted plasmids failed to replicate only in the XPBOS-SV cell line (ratio 0.01) and were very inefficiently replicated in the XP12RO-SV cells (ratio 0.06) in comparison to undamaged plasmids, which produced ratios of 0.25 and 0.98, respectively. This implies that in XP12RO cells a relatively larger fraction of adducted plasmids has replicated.
The mutation frequency of 04-EtT in XPl2RO was -25% (Table 111). Mutations were again T + C transitions. In x P 2 0 S revertant cells, the mutation frequency of 04-EtT was 24%, which is comparable with the frequency found for the other repair proficient cell types. The majority (-21%) of the mutations were T "-f C transitions.
Comparison with the Mutagenicity of 06-MedG-To exclude a possible influence of the alkyltransferase on the mutagenicity of 04-AlkT, we transfected plasmids carrying a single 06-MedG at the ClaI site into the same cell lines. As is shown in Table IV, the mutation frequency of 06-MedG varied considerably among the different cell types. The mutagenicity in HeLa cells was comparable with that of 04-MeT and 04-EtT (mutation frequencies of 14 and 17-21%, respectively), whereas in WtA fibroblasts and XPPOS cells, the mutation frequency (2-5%) hardly exceeded background levels (2%). The majority of the mutations consisted of a G + A transition at the former position of the adducted dG (Table V), which is in good agreement with the mutational specificity of 06-MedG reported in literature (17).
The specific activity of the alkyltransferase varied from 18 fmol/mg of protein for the XP20S cells t o 768 fmoVmg of protein for the WtA fibroblasts (see Table IV). The mutation frequencies were in agreement with the alkyltransferase activity in the two excision repair proficient cell lines; the high mutation frequency of 06-MedG in HeLa cells corresponds with the relatively lower alkyltransferase activity in these cells and the much lower mutation frequency in WtA fibroblasts with the higher alkyltransferase activity.
The low mutation frequency of 06-MedG in x P 2 0 S cells does  not correlate with the very low alkyltransferase activity in these cells, since adducted plasmids again almost completely failed to replicate. Only two mutant clones were obtained both of which carried the G + A transition at the position of the modified dG, but again it is not clear whether these two mutants were spontaneously or specifically induced.
Mutation Induction in Repair-competent Cells-The observed loss of adducted plasmids in XP-A cells indicates that 04-AlkT and 06-MedG residues are recognized by nucleotide excision repair. Since the lesions are still highly mutagenic in repair competent cells, we hypothesized that in these cells excision of the damaged nucleotides is incomplete, possibly due to the relatively small size of 04-AlkT residues. To test this hypothesis, we constructed plasmids with an 04-EtT and a dG-C8-AAF next to each other in the same strand. The dG-CS-AAF was located 2 base pairs 3' to the 04-EtT (see Fig. 1). When this construct was transfected into the fibroblasts, the mutation frequency did not exceed the background frequency (-2%) and was comparable with that of dG-C8-AAF alone (Table VI; Ref. 26). The mutations were of the same types as found with control constructs.

DISCUSSION
We have determined the mutagenicity of 04-MeT, 04-EtT, and 04-nPrT residues in human cells with different repair capacities using site-specific modified plasmids. All three lesions were highly mutagenic in excision repair-proficient cells (-20% for methyl and ethyl adducts and -12% for n-propyl adducts) and specifically induced T + C transitions.
The mutational specificity of the 04-AlkT residues is as expected and in accordance with our previous studies on the mutagenicity of 04-EtT in HeLa cells (7). The same transitions were also found by other groups investigating the effects of 04-MeT and 04-EtT during replication in E. coli and in in vitro assays (5, 37-39). The mutation frequencies induced by these lesions in our system, however, were lower than those calculated from in vitro experiments (4, 6). This suggests that there is repair of 04-AlkTs in human cells, although this may be inefficient.
In order to investigate what type(s) of repair could be involved, we used different cell lines that varied largely with respect to the repair activities of the 06-alkylguanine DNA alkyltransferase or nucleotide excision repair system (AT+/ NER', AT-/NER+, AT-/NER-). In E. coli, 04-AlkTs are primarily removed by the 06-alkylguanine DNA alkyltransferase (9, 15, 331, but it appears unlikely that this is the case in human cells. In our studies, alkyltransferase activities varied widely in the different cell lines and correlated well with the mutation frequency of 06-MedG, but not at all with that of 04-AlkT. The alkyltransferase activity in HeLa cells appeared relatively low as compared with that in WtA fibroblasts, and this is reflected in the much higher mutation frequency of 06-MedG in HeLa cells. However, mutation frequencies of each of the 04-AlkT lesions were similar in both HeLa cells and WtA fibroblasts and approximately equal to those reported for 06-MedG (04-Me-, 04-EtT) and 06-EtdG (04-nPrT) in alkyltransferase-deficient CHO cells (17).
It has been reported recently that the human alkyltransferase is able to bind with very low affinity to 04-MeT in vitro but not to 04-EtT (20). Hence, it was suggested that the human alkyltransferase may also repair 04-MeT to some extent in vivo (20), and this may explain the differences in biological halflives of 04-MeT and 04-EtT in chromosomal DNA as observed with other experimental systems (tl,2 of 2-20 h and 2-20 days, respectively) (8-10). Such repair might have influenced the mutation frequency of 04-MeT in our system as well, but this frequency appeared to be similar to the mutation frequency of 04-EtT (-20%). Therefore, alkyltransferases do not seem to play a prominent role in the removal of any of the 04-AlkTs in human cells.  The multiple mutations have been described elsewhere ( 7 , 2 6 ) .
Normal WtA fibroblasts. Our data indicate that 04-AlkT moieties are recognized by nucleotide excision repair. Remarkably, we found these lesions to be inhibitory to plasmid replication and, consequently, not to be mutagenic in nucleotide excision repair-deficient XP-A cells. The replication inhibition of 04-EtT adducted plasmids was observed with two different XP-A cell lines (XP2OS-SV and XP12RO-SV), both of which are highly deficient in nucleotide excision repair but carry different mutations in the XP-A gene (34,35,40,41). This makes it unlikely that the replication inhibition can be ascribed t o mutations other than those affecting the excision repair capacity. Moreover, the replication block was relieved in a revertant of the XPZOS-SV cell line, and the mutation frequency of 04-EtT in these cells was equal to that found with the other repair proficient cells. 04-EtT also appeared t o inhibit plasmid replication in XP-D (HD2) cells. These cells carry a mutation in a different protein involved in nucleotide excision repair (421, and also, in this case, the amount of progeny derived from ethylated plasmids was largely reduced (-65% as compared with controls). This confirms our notion that the alkylated plasmids indeed fail to replicate because the 04-AlkT lesions are not removed and, therefore, impede the progress of the replication fork.
Another indication for the removal of 04-AlkTs by nucleotide excision repair is the fact that mutation frequencies decreased with increasing size of the adduct (mutation frequency 04- to repair larger 0-alkyl groups less efficiently (43,44,45). We observed the opposite. The decrease in mutation frequency we found can only be consistent with removal by nucleotide excision repair, since this type of repair has been shown to recognize distortions of the DNA double helix caused by the adducts rather than the adducts themselves (46). With increasing adduct size, the distortion of the helix might become more extensive, allowing a more efficient repair. The effect is most pronounced with dG-C8-AAF, which has been shown to cause large alterations in the DNA conformation and which is indeed e%ciently removed by nucleotide excision repair (26,47). Finally, it is known from literature that different cell lines of the XP-A complementation group fail to replicate plasmids carrying other carcinogen-DNA adducts, which have been shown to be substrates for nucleotide excision repair (41,48,49). Previous experiments in our laboratory demonstrated that lesions that are substrates for glycosylase repair (i.e. 8-oxo-dG) do not inhibit DNA replication in XP-A cells (26). From the present studies, it appears that even the smallest 04-AlkT adducts completely prevent bypass of replication, indicating that nucleotide excision repair is the predominant repair mode operating on these adducts.
The fact that the observed inhibition of plasmid replication in XP-A cells is (almost) 100% indicates that the alkylated thymines are very efficiently recognized by the nucleotide excision repair system. This is unexpected because from the high mutation frequencies in repair-proficient cells, one would have anticipated a rather poor recognition of 04-AlkTs. Apparently, recognition and incisiodexcision are somehow uncoupled processes, since all lesions (small and large) are recognized effectively but not removed with equal efficiency. It may be that the smaller 04-AlkT lesions do not permit the specific identification and incision of the damaged strand but that, instead, the undamaged strand is incised, and subsequent repair replication using the damaged strand as the template introduces the mutation opposite 04-AlkT. Removal of the lesion by the proper operation of the excision repair system might then allow normal plasmid replication to proceed. Alternatively, the nucleotide excision repair factors in proficient cells might interact with the damage in a way that does not completely suppress replication, whereas in repair-deficient cells this interaction is more extensive and completely blocks replication.
The lower mutation frequencies of the larger (04-nPrT and dG-C8-AAF') adducts suggest that the incisiodexcision process is more reliable with larger adducts. When the strand carrying an 04-EtT was marked with an additional C8-AAF adduct on a 3' neighboring dG, it might have been predicted that this would reduce the mutagenicity of the adjacent 04-EtT and indeed, with the excision of dG-C8-AAF the 04-EtT was simultaneously

Repair and Mutagenicity of O4-A1kylT in Human Cells
removed and the mutation frequency concomitantly decreased to background levels. The recognition of 04-AlkTs by nucleotide excision repair has also been shown to occur in E. coli (16). However, Bronstein et al. (25) reported no differences between excision repair deficient and proficient human cells with respect to the t,, values for the removal of 04-AlkT residues from DNA, and they concluded that these lesions were not substrates for nucleotide excision repair. We have not measured the actual removal (t1J of these lesions in our system. Our experiments also differ with respect to the position of the lesions (plasmid uersus chromosomal DNA), the cell lines used, and their rates of division. Therefore, the apparent discrepancy between their conclusions and ours probably relates to gross differences in the experimental systems that were utilized. Altogether, our experiments demonstrate that 04-AlkT residues can be highly mutagenic in human cells, in spite of the fact that they appear to be efficiently recognized by the nucleotide repair system. Surprisingly, the nucleotide excision repair system seems instrumental in the induction of mutations at the position of 04-AlkT lesions. This clearly designates these lesions as potential inducers of neoplastic transformation.
with the HPLC analysis and Jos Domen, Hein te Riele, Chris Saris, and