The chemotherapeutic activity of 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU or carmustine) may be improved by the addition of O6-benzylguanine (O6-BG). The reaction of O6-BG with O6-alkylguanine-DNA alkyltransferase (AGT) prevents the repair of O6-chloroethyl lesions caused by BCNU. In clinics, the combination of O6-BG and BCNU is now being tested for the treatment of brain tumors. However, the effectiveness of this drug regimen may be limited by drug resistance acquired during treatment. To understand the possible mechanisms of resistance of brain tumor cells to the O6-BG/BCNU combination, we generated medulloblastoma cell lines (D283 MED, D341 MED, and Daoy) resistant to the combination of O6-BG and BCNU [O6-BG/BCNU resistant (OBR)]. DNA sequencing showed that all of the parent cell lines express wild-type AGTs, whereas every OBR cell line exhibited mutations that potentially affected the binding of O6-BG to the protein as evidenced previously by in vitro mutagenesis and structural studies of AGT. The D283 MED (OBR), Daoy (OBR), and D341 MED (OBR) cell lines expressed G156C, Y114F, and K165T AGT mutations, respectively. We reported previously that rhabdomyosarcoma TE-671 (OBR) also expresses a G156C mutation. These data suggest that the clonal selection of AGT mutants during treatment with O6-BG plus an alkylator may produce resistance to this intervention in clinical settings.

Carmustine or 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) is one of the most widely used drugs in treating brain tumors because it can efficiently cross the blood-brain barrier (1). In aqueous environment, BCNU decomposes into chloroethyl isocyanate and chloroethyl diazohydroxide (2). The latter gives rise to chloroethyl carbonium intermediate, which reacts with the O6 position of guanine (3). This initial adduction eventually leads to the formation of the cytotoxic lesion, which is the ethanyl cross-link between the N1 of guanine and the N3 of cytosine in the opposite strand (4).

The foremost problem in cancer chemotherapy is the development of drug resistance during treatment despite a promising initial response. A major resistance factor to O6-guanine alkylating drugs such as BCNU is O6-alkylguanine-DNA alkyltransferase (AGT; ref. 5), which acts as a suicide enzyme by removing the methyl or chloroethyl damage at the O6 position of guanine. The repair of the O6-(2-chloroethyl)guanine formed by BCNU prevents the formation of the interstrand cross-links described above. The activity of BCNU can also be hindered by glutathiones (6) and base excision repair (7, 8). The possibility of an interstrand cross-link repair mechanism has also been described (9).

O6-benzylguanine (O6-BG) is designed to inactivate AGT by acting as an alternate substrate (10–12). The combination of O6-BG with either BCNU or temozolomide is currently being tested in clinical trials for the treatment of several types of cancer including adult and childhood brain tumors (13, 14). Through in vitro mutagenesis, it has been shown that the ability of O6-BG to inhibit AGT may be altered by certain mutations (15). Therefore, a legitimate concern for oncologists is the possibility that these AGT mutations are present before or may arise during treatment. We established sublines of the medulloblastoma cell lines D283 MED and Daoy that are resistant to the O6-BG/BCNU drug combination [O6-BG/BCNU resistant (OBR)]. The parent and resistant cell lines were characterized for their AGT activity and genotype. In an earlier study, we reported that an OBR rhabdomyosarcoma (TE-671) harbors a G156C substitution in AGT (16). In that publication, we also described the generation of D341 MED (OBR), another medulloblastoma subline, with its AGT genotype included in this report.

Drugs

BCNU was purchased from Sigma Chemical Co. (St. Louis, MO). O6-BG was synthesized and purified according to published reports (10).

Cell Lines

D283 MED, D341 MED, Daoy, and TE-671 were grown in Improved MEM Zinc Option (Richter's modification, Invitrogen, Carlsbad, CA) supplemented with 10% FCS (Invitrogen; ref. 17).

Generation of Drug-Resistant Cells

The generation of the D341 MED and TE-671 OBR sublines has been described previously (16). The drug-resistant cell lines derived from D283 MED and Daoy were generated in a similar fashion. The parental cell lines were exposed to 100 μmol/L O6-BG for 10 minutes prior to treatment with 10 μmol/L BCNU for 1 hour. The cells were washed free of drugs and maintained in 10 μmol/L O6-BG. The cells were treated with 10 μmol/L BCNU in combination with O6-BG after every three passages until they showed increased survival to drug treatment. The BCNU dosage was subsequently increased by 5 to 10 μmol/L after every three passages to generate OBR cell lines.

Limiting Dilution Assay

The cytotoxicity of BCNU against D283 MED, D283 MED (OBR), Daoy, and Daoy (OBR) with and without O6-BG was determined using limiting dilution assay as described (17).

AGT Assay

The AGT activity of the cells was measured as described (10). Briefly, the 3H-methylated DNA substrate (prepared by reaction of calf thymus DNA with [3H]methylnitrosourea) was incubated with cell-free extract for 30 minutes at 37°C. The substrate was precipitated by adding ice-cold perchloric acid at a final concentration of 0.25 mol/L and depurinated by incubating the precipitate with 500 μL of 0.1 mol/L HCl at 70°C for 30 minutes. The modified bases were separated by reverse-phase high-performance liquid chromatography and quantified by scintillation counting. The enzyme activity was defined as the femtomoles of O6-[3H]methylguanine removed from 3H-methylated DNA per milligram of extracted protein. Protein concentration was determined using Bio-Rad protein assay with bovine serum albumin as standard (Bio-Rad, Hercules, CA).

AGT Sequencing

Total RNAs and genomic DNAs of the parent and OBR cells were prepared using TRIzol reagent (Invitrogen) and DNeasy tissue kit (Qiagen, Valencia, CA), respectively. The AGT cDNA was generated from total RNA by reverse transcription (using SuperScript II RNase H reverse transcriptase and random hexamer primers from Invitrogen) followed by PCR (using PCR Master Mix from Promega, Madison, WI). The PCR primers used were AGT-L (5′-GCGCGGATCCGGTACTTGGAAAAATGGACAAGGATTG-3′) and AGT-R (5′-GCGCGAATTCCAAACATCCATCCTACTGCACATAC-3′). For genomic DNA, 0.2 μg was used as template to amplify AGT exons 4 and 5 by PCR. For exon 4 amplification, the primer pair was AGTexon4-L (5′-GCGTTTCCTGTTTTGGGACT-3′) and AGTexon4-R (5′-CGCCATGAGAACTCACAGGA-3′). For exon 5, the primers AGTexon5-L (5′-CTTGACCCCAAAGACCTCGT-3′) and AGTexon5-R (5′-TGTCGCTCAAACATCCATCC-3′) were used. For every PCR reaction, the conditions were as follows: (a) initial denaturation at 95°C for 2 minutes; (b) 40 cycles of 95°C for 30 seconds, 62°C for 1 minute, and 72°C for 1 minute; and (c) final extension at 72°C for 7 minutes. The PCR products were ethanol precipitated, dissolved in 15 μL of 10 mmol/L Tris-HCl (pH 8.0) and 1 mmol/L EDTA, and electrophoresed on a 1.0% agarose gel (with 0.5 μg/mL ethidium bromide). The AGT DNA fragments (680 bp for cDNA, 246 bp for exon 4, and 306 bp for exon 5) were extracted from the gel using QIAquick Gel Extraction Kit (Qiagen). Sequencing of the DNA fragments (using BigDye Terminator Cycle Sequencing) was carried out at the DNA Sequencing Facility at Duke University Medical Center (Durham, NC). The AGT cDNA was sequenced using both AGT forward (5′-CTGGCTGAATGCCTATTTCC-3′) and AGT reverse (5′-ATTGCTCCTCCCACTGCTC-3′) sequencing primers. The PCR primers AGTexon4-L and AGTexon5-L were used to sequence exons 4 and 5, respectively.

Production of Recombinant Wild-type and Mutant AGT Proteins

Recombinant wild-type AGT was isolated and purified by immobilized affinity chromatography as described previously using the pQE vector, which adds a 12–amino acid sequence (MRGS [H]6GS-) to the NH2 terminus of the AGT protein (18). The generation and purification of the recombinant mutant Y114F protein has been described (19). The mutants G156C and K165T were isolated during the screening of a library of plasmids encoding AGT molecules in which a 24–amino acid region was subjected to random mutagenesis (15). These mutant proteins were also purified using the pQE vector. Mutant K165T, which was found previously to be very unstable in Escherichia coli (20), could be purified in this way albeit in very poor yield.

Determination of Effects of O6-BG on Activities of Wild-type and Mutant Recombinant AGTs

Inactivation of the AGT activity by O6-BG was measured by incubating aliquots of the protein with different concentrations of O6-BG at 37°C for 30 minutes in 0.5 mL of 50 mmol/L Tris-HCl (pH 7.6), 5 mmol/L DTT, and 0.1 mmol/L EDTA containing 50 μg hemocyanin and 10 μg calf thymus DNA. The residual AGT activity was determined by incubating the samples with 3H-methylated DNA substrate for 30 minutes at 37°C and measuring 3H-methylated protein formed after collection on nitrocellulose filters (21). The results were expressed as the percentage of the AGT activity remaining and the graphs of AGT activity remaining against inhibitor concentration were used to calculate an ED50 value representing the amount of inhibitor needed to produce a 50% loss of activity.

Kinetics of Repair by AGT

The rate constant for the AGT reaction was determined by measuring the appearance of the 3H-methylated AGT at various times using concentrations of AGT protein determined in preliminary experiments to yield readily measurable rates under the assay conditions. The reaction mixture (1 mL) for each time point contained wild-type or mutated AGT (1.5 × 10−9–15 × 10−9 mol/L depending on the activity), 3.6 × 10−10 mol/L O6-[3H]methylguanine in 3H-methylated calf thymus DNA substrate, and 50 μg cold calf thymus DNA in a buffer containing 50 mmol/L Tris-HCl (pH 7.6), 5 mmol/L DTT, and 0.5 mmol/L EDTA. The rate constant was determined as described (19).

Western Analysis

Cellular extracts were prepared by incubating cells with lysis buffer [50 mmol/L Tris-HCl (pH 7.8), 0.15 mol/L NaCl, 1% NP40, complete mini EDTA-free protease inhibitor cocktail (Roche, Indianapolis, IN)] at 37°C for 10 minutes and isolating the supernatant after centrifugation. Sixty to 100 μg of protein extracts were mixed with an equal volume of 2× SDS loading buffer [20% glycerol, 4% SDS, 0.2 mol/L DTT, 0.01% bromophenol blue, 0.125 mol/L Tris-HCl (pH 6.8)], electrophoresed in 12% Tris-glycine SDS-PAGE gel, and transferred onto Immobilon-P membranes (Millipore, Billerica, MA). The anti-AGT monoclonal antibody clone 3.1 was used as described (22). Monoclonal antibody to α-tubulin (Calbiochem, San Diego, CA) was used as a loading control. The primary antibodies were diluted to 10 μg/mL (AGT) and 2 μg/mL (α-tubulin). Antibody binding was detected with the Enhanced Chemiluminescence Plus system (Amersham, Piscataway, NJ). Images were captured with a Kodak ImageStation 440CF and quantitated using 1D Image Analysis software (Kodak, Rochester, NY). Results for AGT were normalized to the values obtained for α-tubulin and were determined in at least three experiments for each cell line.

BCNU Cytotoxicity

The surviving fraction of D283 MED (OBR) was almost 100 times higher than D283 MED at every BCNU dosage (Fig. 1A). The addition of 10 μmol/L O6-BG did not increase the toxicity of BCNU for either cell line. BCNU toxicity for Daoy and Daoy (OBR) were similar (Fig. 1B). When AGT was inactivated by the addition of O6-BG, enhanced killing of Daoy, but not Daoy (OBR), was observed. The survival curves for D341 MED/D341 MED (OBR) and TE-671/TE-671 (OBR) resembled those for Daoy/Daoy (OBR; ref. 16).

Figure 1.

Survival of (A) D283 MED and D283 MED (OBR) and (B) Daoy and Daoy (OBR) after exposure to BCNU with and without O6-BG. Points, average of at least three independent experiments; bars, SD.

Figure 1.

Survival of (A) D283 MED and D283 MED (OBR) and (B) Daoy and Daoy (OBR) after exposure to BCNU with and without O6-BG. Points, average of at least three independent experiments; bars, SD.

Close modal

AGT Expression

The parental cell line D283 MED exhibited minimal AGT activity (20 fmol/mg protein; Fig. 2A), whereas D283 MED (OBR) was >5 times higher. The AGT activity for D283 MED (OBR) was measured in cells that were treated with 10 μmol/L O6-BG. When the O6-BG concentration was increased further to 20 μmol/L, the AGT activity measured in D283 MED (OBR) cell extracts decreased 35% to 74 fmol/mg. The expression of AGT was detectable in D283 MED (OBR) but not in the parental cell line when analyzed by Western blotting (Fig. 3). The results for D283 MED correlate very well with the limiting dilution data (Fig. 1A). The ineffectiveness of O6-BG in increasing toxicity of BCNU in D283 MED is consistent with the very low level of AGT expression. The parental Daoy cells registered AGT activity of 687 fmol/mg (Fig. 2B). When these cells were exposed to O6-BG, very little AGT activity was observed. As observed in the limiting dilution experiments, the ablation of active AGT by O6-BG greatly enhanced the sensitivity of the Daoy cells to BCNU. In contrast, no AGT activity was measured in Daoy (OBR). However, there was AGT protein detected in proteins extracted from these cells on a Western blot (Fig. 3). A similar result was observed for D341 MED, wherein no AGT activity was measured (16) but AGT protein was observed in cell extracts (Fig. 3).

Figure 2.

AGT activities of (A) D283 MED and D283 MED (OBR) and (B) Daoy and Daoy (OBR). AGT activities of D283 MED and Daoy with and without exposure to O6-BG. D283 MED (OBR) and Daoy (OBR) were maintained in a medium with 10 μmol/L O6-BG. Residual AGT activity in D283 MED (OBR) is still evident even when the O6-BG concentration was increased to 20 μmol/L. Bars, SD.

Figure 2.

AGT activities of (A) D283 MED and D283 MED (OBR) and (B) Daoy and Daoy (OBR). AGT activities of D283 MED and Daoy with and without exposure to O6-BG. D283 MED (OBR) and Daoy (OBR) were maintained in a medium with 10 μmol/L O6-BG. Residual AGT activity in D283 MED (OBR) is still evident even when the O6-BG concentration was increased to 20 μmol/L. Bars, SD.

Close modal
Figure 3.

AGT expression of D283 MED and D283 MED (OBR), Daoy and Daoy (OBR), D341 MED and D341 MED (OBR), and TE-671 and TE-671 (OBR) by Western analysis. P and O, protein extracts from the parent and OBR cell lines, respectively. The monoclonal antibody for α-tubulin was used as a control.

Figure 3.

AGT expression of D283 MED and D283 MED (OBR), Daoy and Daoy (OBR), D341 MED and D341 MED (OBR), and TE-671 and TE-671 (OBR) by Western analysis. P and O, protein extracts from the parent and OBR cell lines, respectively. The monoclonal antibody for α-tubulin was used as a control.

Close modal

AGT Mutations in OBR Lines

The AGT cDNA of D283 MED, D283 MED (OBR), Daoy, Daoy (OBR), D341 MED, and D341 MED (OBR) was amplified and sequenced. Results indicated a GGC (glycine) to TGC (cysteine) mutation at codon 156 for D283 MED (OBR), the same mutation found in TE-671 (OBR; ref. 16); TAC (tyrosine) to TTC (phenylalanine) at codon 114 for Daoy (OBR); and AAG (lysine) to ACG (threonine) at codon 165 for D341 MED (OBR; Fig. 4). Only the single base substitutions and not the wild-type sequences were observed in the AGT cDNA sequence. However, the corresponding genomic DNA sequences revealed that, unlike Daoy (OBR) and TE-671 (OBR), the AGT alleles of D283 MED (OBR) and D341 MED (OBR) were heterozygous. These results suggest that the AGT expression in D341 MED (OBR) and D283 MED (OBR) is limited to the mutated allele, whereas its wild-type counterpart is not expressed. AGT is among the several DNA repair genes with expression repressed by CpG island hypermethylation within their promoters (see ref. 23 for review). On the other hand, the lack of heterozygosity in genomic AGT mutations in Daoy (OBR) and TE-671 (OBR) suggests that both these cells possess a single AGT allele. Genomic AGT sequencing also revealed that D283 MED and D283 MED (OBR) have polymorphisms at codon 143 (ATC to GTC; Ile to Val) and codon 178 (AAG to AGG; Lys to Arg; Fig. 5). Apparently, this tandem polymorphism (24) occurs in the unexpressed AGT allele because these sequence variations are absent in both D283 (MED) and D283 MED (OBR) cDNA sequences.

Figure 4.

AGT mutations in OBR cells. The relevant portions of the AGT genomic and cDNA sequences of D283 MED and D283 MED (OBR), D341 MED and D341 MED (OBR), Daoy and Daoy (OBR), and TE-671 and TE-671 (OBR).

Figure 4.

AGT mutations in OBR cells. The relevant portions of the AGT genomic and cDNA sequences of D283 MED and D283 MED (OBR), D341 MED and D341 MED (OBR), Daoy and Daoy (OBR), and TE-671 and TE-671 (OBR).

Close modal
Figure 5.

Tandem polymorphisms (heterozygous) at codon 143 (ATC to GTC; Ile to Val) and codon 178 (AAG to AGG; Lys to Arg) detected in the AGT gene of D283 MED and D283 MED (OBR).

Figure 5.

Tandem polymorphisms (heterozygous) at codon 143 (ATC to GTC; Ile to Val) and codon 178 (AAG to AGG; Lys to Arg) detected in the AGT gene of D283 MED and D283 MED (OBR).

Close modal

Kinetic and Inhibition Studies of Recombinant Wild-type and Mutant AGTs

Compared with wild-type protein (ED50 for O6-BG = 0.2 μmol/L), the mutants were not as easily inhibited by O6-BG (Table 1). The recombinant G156C, Y114F, and K165T AGTs had ED50 values of 50, 73, and 36 μmol/L O6-BG, respectively. The mutations also decreased the ability of AGT to repair methylated DNA. The rate constants of repair were 38 × 106, 2.9 × 106, 2.6 × 106, and 5.6 × 106 mol/L−1 min−1 for wild-type, G156C, Y114F, and K165T AGTs, respectively.

Table 1.

Comparison of the ED50 (for O6-BG) and rate constant for repair of methylated DNA in recombinant wild-type and mutant AGTs

Recombinant AGTED50 for O6-BG, μmol/LRate Constant for Repair of Methylated DNA, ×106 mol/L−1 min−1
Wild-type 0.2 38 
G156C 50* 2.9 
Y114F 73 2.6 
K165T 36 5.6 
Recombinant AGTED50 for O6-BG, μmol/LRate Constant for Repair of Methylated DNA, ×106 mol/L−1 min−1
Wild-type 0.2 38 
G156C 50* 2.9 
Y114F 73 2.6 
K165T 36 5.6 
*

Previously reported as 49 μmol/L based on assays with crude extracts of E. coli expressing the G156C mutant (47).

Previously reported (19).

According to the Goldie and Coldman (25, 26) model, the fraction of drug-resistant cancer cells will increase over time and is a function of the mutation rate and initial number of the cells. In this particular study, the selection process for the BCNU-resistant AGT mutant cell lines was driven by two key processes: (a) the mutational events that occurred at the AGT locus of the BCNU-treated parental cell line and (b) the selection and propagation of a clone expressing O6-BG-resistant AGT. What determines the rate of mutation in the drug-exposed cells is the state of different DNA damage checkpoint (e.g., p53) and repair processes [e.g., mismatch repair (MMR) system] and, more significantly, the mutagenicity of BCNU. The loss of a functioning p53 has been correlated to increased mutations in cells exposed to DNA-damaging chemical and physical agents (27). DNA damage can lead to up-regulation of p53, which in turn activates expression of proteins involved in cell cycle arrest (28), DNA repair (29, 30), and apoptosis (31). Another important factor in maintaining genetic stability of cells is the MMR system. It has been shown that cancer cells with defective MMR have higher incidence of spontaneous (32–34) and DNA damaging agent–induced (35–37) mutations. When Liu et al. (38) treated colon cancer lines with three cycles of the combination of BCNU plus O6-BG, the two MMR deficient cell lines became resistant and that resistance was also caused by mutations in the AGT protein. The MMR proficient cell line maintained sensitivity to the drug combination and expressed wild-type AGT protein. In contrast, we observed mutations in AGT in the MMR proficient (Daoy and TE-671) as well as MMR deficient (D341 MED and D283 MED) cell lines (summarized in Table 2). This difference probably reflects the fact that, in our experiments, the AGT was characterized after selection for resistance to the drug combination. Because D341 MED and D283 MED have MMR deficiencies, their corresponding OBR sublines were selected more rapidly than the MMR proficient cells. Generation and selection of the mutant AGTs in the OBR sublines may require that wild-type AGT is not present to repair the DNA damage produced by BCNU. Thus, of the four sublines, two have a single, mutated AGT allele and the other two, although heterozygous for AGT, express mRNA from only the mutated AGT. Silencing of the wild-type AGT could be a result of hypermethylation of the CpG islands present in the promoter of AGT (23) or of mutation of the regulatory regions of the wild-type AGT allele.

Table 2.

Summary of characterization of parental and OBR cell lines

Cell LineParent Line AGT Activity, fmol/mg proteinParent Line AGT Activity (10 μmol/L O6-BG), fmol/mg proteinOBR Subline AGT Activity (10 μmol/L O6-BG), fmol/mg proteinRelative AGT Expression by Western (Parent)*Relative AGT Expression by Western (OBR)*AGT MutationMMR Phenotype
D283 MED 20 ± 25 6 ± 10 115 ± 43 0.09 ± 0.06 G156C − 
D341 MED 2,232 ± 329 49 ± 31 0.83 ± 0.16 0.71 ± 0.10 K165T − 
Daoy 686 ± 191 20 ± 8 3.0 ± 6.0 0.61 ± 0.23 Y114F 
TE-671 3,967 ± 341 18 ± 1 223 ± 36 1.68 ± 0.43 0.98 ± 0.52 G156C 
Cell LineParent Line AGT Activity, fmol/mg proteinParent Line AGT Activity (10 μmol/L O6-BG), fmol/mg proteinOBR Subline AGT Activity (10 μmol/L O6-BG), fmol/mg proteinRelative AGT Expression by Western (Parent)*Relative AGT Expression by Western (OBR)*AGT MutationMMR Phenotype
D283 MED 20 ± 25 6 ± 10 115 ± 43 0.09 ± 0.06 G156C − 
D341 MED 2,232 ± 329 49 ± 31 0.83 ± 0.16 0.71 ± 0.10 K165T − 
Daoy 686 ± 191 20 ± 8 3.0 ± 6.0 0.61 ± 0.23 Y114F 
TE-671 3,967 ± 341 18 ± 1 223 ± 36 1.68 ± 0.43 0.98 ± 0.52 G156C 
*

Relative to expression of D341 MED.

Determined by in vitro MMR reaction using a G:T mismatch (48).

AGT activity data and the TE-671 AGT mutation are from Bacolod et al. (16).

Obviously, the most significant source of AGT mutations would be the DNA damages caused by BCNU. This would be enhanced by the addition of O6-BG, which eliminates the enzyme that repairs the mutagenic lesion at the O6 position of guanine. We have made sublines resistant to BCNU in the absence of O6-BG. These sublines are characterized by high levels of AGT expression (16). When the AGT cDNAs from these sublines were sequenced, each retained the wild-type sequence.10

10

Unpublished results.

In Chinese hamster ovary cells that lack AGT, treatment with BCNU resulted in a mutation frequency in the adenine phosphoribosyl transferase (aprt) gene 6-fold above background (39). All possible transversions and transitions were seen, although the predominant mutations were G:C to T:A (51%) and G:C to A:T (20%). Overexpression of AGT in these cells reduced the frequency of both mutations, suggesting that they originate from initial O6-chloroethyl lesions and T:A to G:C transversions became the predominant mutation (40). Two hotspots for mutation in the Chinese hamster ovary cells lacking AGT were identified: 5′-ACTGgAGGT-3′ and 5′-ACTAgAAAT-3′, which have similar helical properties on application of Calladine's rules (41). Two of the four OBR cell lines we isolated have a G:C to T:A transversion at 5′-TGgGCAA-3′, which, like the first hotspot, has the 5′-Gg-3′ sequence. The other two mutations resulted in A:T to C:G and A:T to T:A transversions, both of which have been identified as minor mutations in BCNU-treated cells.

The crystal structure of wild-type human AGT has been elucidated by two independent laboratories—the most significant difference being the absence (42) or the presence (43) of zinc in the NH2-terminal region of the protein. Both studies provide a model on how O6-BG binds prior to a SN2 reaction with Cys145 and can be used as a basis for reasonable speculations on the reasons that the mutants described here have altered activity and resistance to O6-BG. It is highly likely that alteration of Gly156 modifies the shape of the binding pocket to exclude O6-BG due to the more restricted torsion angles allowed when other residues are placed at this site. Many different substitutions at Gly156 impart resistance (15). Similarly, interactions of Lys165, which is a highly conserved residue in AGT sequences, play an important role in stabilizing the active site loop and all 19 changes at this site impart resistance to O6-BG presumably by distorting the protein backbone (20). In the proposed reaction mechanism by Daniels et al. (43), the final step in removal of the benzyl group from guanine is the protonation of N3 (of guanine) by the hydroxyl group of Tyr114. This suggests that the loss of a hydroxyl group, which occurs in the Y114F mutant, would reduce the rate of reaction with O6-BG. Our in vitro studies showed that these three mutations adversely affected both the inhibitory effect of O6-BG (as evidenced in dramatic increase in ED50 values) and the ability of the protein to repair methylated guanine (the rate constants for repair all decreased significantly). However, these mutant proteins retained sufficient AGT activity to provide O6-BG-treated E. coli protection from the methylating agent N-methyl-N′-nitro-N-nitrosoguanidine during the selection of drug-resistant colonies (15).

The AGT activity of D283 MED (OBR; which is G156C AGT) increased noticeably compared with that of the parental cells, although the former was maintained in O6-BG. The increase in AGT activity was consistent with the increased AGT expression (as confirmed by Western blotting). In contrast, Daoy (OBR; which is Y114F AGT) and D341 MED (OBR; which is K165T AGT) did not exhibit measurable AGT activity (16), although Western analysis indicated the presence of the protein. One possible explanation is the instability of K165T and Y114F AGTs and that those detected by Western analysis were actually a mixture of native and denatured forms. K165T AGT was observed to be very unstable in vitro, although it conferred resistance against O6-BG in E. coli (20). Liu et al. (38) identified two OBR colon cancer lines with K165N and K165E mutations. Both Western and enzyme activity assays indicated that the mutant cell lines had lower levels of AGT expression compared with parental lines. However, the cells had equal amounts of AGT mRNAs, thus prompting the speculation that the mutant lines had unstable AGTs. When the mutant AGTs were transfected to non-AGT-expressing, BCNU-sensitive Chinese hamster ovary cells, the O6-BG-treated cells became resistant to both BCNU and temozolomide. There is a possibility that the instability may lead to a new steady state at a lower amount of protein if degradation is enhanced and synthesis is not; however, if synthesis is also increased, the protein level would stay the same. These interpretations regarding the role of AGT in OBR are strongly supported by the limiting dilution results for Daoy (OBR; Fig. 1B) and D341 MED (OBR; ref. 16). The addition of O6-BG increased BCNU cytotoxicity in D341 MED and Daoy but not their corresponding OBR sublines, indicating that these had sufficient levels of active, mutant AGTs to repair the O6-chloroethyl lesions produced by BCNU. In TE-671 (OBR) and D283 MED (OBR), the role of the mutant AGT (both G156C) is more apparent. TE-671 (OBR) had measurable AGT activity even when O6-BG dosage was increased to 60 μmol/L (16). D283 MED (OBR) had a higher AGT activity than its parent and this activity persisted in the presence of 20 μmol/L O6-BG.

It is entirely possible that other mechanisms contribute to the drug-resistant phenotype of these cells. Initially, the selection of mutant AGTs may have been the most crucial factor for the survival of the O6-BG/BCNU–treated cells. However, as the cells were passed and treated with increasing amounts of BCNU, other resistance factors may have started to play equal or more important roles than AGT. For example, further investigations are under way, exploring the increased expression of metallothioneins that were observed in D341 MED (OBR; ref. 16).

Considering the frequency (4 of 4) of AGT mutations in OBR lines, we believe that a potential hindrance to O6-BG/BCNU chemotherapy in brain tumors may be the selection of clones with mutated O6-BG-resistant AGT (regardless of the MMR ability of the tumor). This conclusion agrees with previous detailed studies using site-directed mutagenesis and the selection from libraries of plasmids expressing human AGT in E. coli in which critical regions of the coding sequence were replaced by random inserts to identify mutants resistant to O6-BG. These studies have indicated that there are at least 28 individual amino acid residues in human AGT that can be altered to produce such resistance (44, 45). The current studies and those of Liu et al. (38) have identified mutations imparting resistance to O6-BG enhancement of killing by BCNU at only three of these sites. Although this may be related to hotspots for the induction of mutations by this alkylating agent, it is more likely a much wider range of resistant sites will emerge with more extensive use of this drug combination. In addition, at least one of the alterations identified in the screens of AGT described above, namely, G160R, is a naturally occurring variant. Although this variant is rare (46), the presence of this allele would generate resistance to therapy with O6-BG without need for mutagenesis. Our results therefore provide strong support for the development of second-generation inactivators of AGT that are able to inactivate those forms that are poorly affected by O6-BG.

Grant support: NIH grants NS-30245, CA-71976, and NS-20023.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Drs. Meng Xu-Welliver and Sreenivas Kanugula for originally generating the mutants and constructs used for recombinant protein expression and Dr. B.K. Ahmed Rasheed for technical advice.

1
Fine HA. The basis for current treatment recommendations for malignant gliomas.
J Neurooncol
1994
;
20
:
111
–20.
2
Lemoine A, Lucas C, Ings RM. Metabolism of the chloroethylnitrosoureas.
Xenobiotica
1991
;
21
:
775
–91.
3
Tong WP, Kirk MC, Ludlum DB. Molecular pharmacology of the haloethyl nitrosoureas: formation of 6-hydroxyethylguanine in DNA treated with BCNU (N,N1-bis[2-chloroethyl]-N-nitrosourea).
Biochem Biophys Res Commun
1981
;
100
:
351
–7.
4
Tong WP, Kirk MC, Ludlum DB. Formation of the cross-link 1-[N3-deoxycytidyl),2-[N1-deoxyguanosinyl]ethane in DNA treated with N,N′-bis(2-chloroethyl)-N-nitrosourea.
Cancer Res
1982
;
42
:
3102
–5.
5
Pegg AE. Properties of mammalian O6-alkylguanine-DNA transferases.
Mutat Res
1990
;
233
:
165
–75.
6
Ali-Osman F, Caughlan J, Gray GS. Decreased DNA interstrand cross-linking and cytotoxicity induced in human brain tumor cells by 1,3-bis(2-chloroethyl)-1-nitrosourea after in vitro reaction with glutathione.
Cancer Res
1989
;
49
:
5954
–8.
7
Allan JM, Engelward BP, Dreslin AJ, Wyatt MD, Tomasz M, Samson LD. Mammalian 3-methyladenine DNA glycosylase protects against the toxicity and clastogenicity of certain chemotherapeutic DNA cross-linking agents.
Cancer Res
1998
;
58
:
3965
–73.
8
Xu Y, Hansen WK, Rosenquist TA, Williams DA, Limp-Foster M, Kelley MR. Protection of mammalian cells against chemotherapeutic agents thiotepa, 1,3-N,N′-bis(2-chloroethyl)-N-nitrosourea, and mafosfamide using the DNA base excision repair genes Fpg and α-hOgg1: implications for protective gene therapy applications.
J Pharmacol Exp Ther
2001
;
296
:
825
–31.
9
Ali-Osman F, Rairkar A, Young P. Formation and repair of 1,3-bis-(2-chloroethyl)-1-nitrosourea and cisplatin induced total genomic DNA interstrand crosslinks in human glioma cells.
Cancer Biochem Biophys
1995
;
14
:
231
–41.
10
Dolan ME, Moschel RC, Pegg AE. Depletion of mammalian O6-alkylguanine-DNA alkyltransferase activity by O6-benzylguanine provides a means to evaluate the role of this protein in protection against carcinogenic and therapeutic alkylating agents.
Proc Natl Acad Sci U S A
1990
;
87
:
5368
–72.
11
Dolan ME, Stine L, Mitchell RB, Moschel RC, Pegg AE. Modulation of mammalian O6-alkylguanine-DNA alkyltransferase in vivo by O6-benzylguanine and its effect on the sensitivity of a human glioma tumor to 1-(2-chloroethyl)-3-(4-methylcyclohexyl)-1-nitrosourea.
Cancer Commun
1990
;
2
:
371
–7.
12
Dolan ME, Pegg AE. O6-benzylguanine and its role in chemotherapy.
Clin Cancer Res
1997
;
3
:
837
–47.
13
Friedman HS, Pluda J, Quinn JA, et al. Phase I trial of carmustine plus O6-benzylguanine for patients with recurrent or progressive malignant glioma.
J Clin Oncol
2000
;
18
:
3522
–8.
14
Quinn JA, Pluda J, Dolan ME, et al. Phase II trial of carmustine plus O(6)-benzylguanine for patients with nitrosourea-resistant recurrent or progressive malignant glioma.
J Clin Oncol
2002
;
20
:
2277
–83.
15
Xu-Welliver M, Pegg AE. Point mutations at multiple sites including highly conserved amino acids maintain activity, but render O6-alkylguanine-DNA alkyltransferase insensitive to O6-benzylguanine.
Biochem J
2000
;
347
:
519
–26.
16
Bacolod MD, Johnson SP, Ali-Osman F, et al. Mechanisms of resistance to 1,3-bis(2-chloroethyl)-1-nitrosourea in human medulloblastoma and rhabdomyosarcoma.
Mol Cancer Ther
2002
;
1
:
727
–36.
17
Friedman HS, Colvin OM, Kaufmann SH, et al. Cyclophosphamide resistance in medulloblastoma.
Cancer Res
1992
;
52
:
5373
–8.
18
Edara S, Kanugula S, Goodtzova K, Pegg AE. Resistance of the human O6-alkylguanine-DNA alkyltransferase containing arginine at codon 160 to inactivation by O6-benzylguanine.
Cancer Res
1996
;
56
:
5571
–5.
19
Goodtzova K, Kanugula S, Edara S, Pegg AE. Investigation of the role of tyrosine-114 in the activity of human O6-alkylguanine-DNA alkyltransferase.
Biochemistry
1998
;
37
:
12489
–95.
20
Xu-Welliver M, Kanugula S, Loktionova NA, Crone TM, Pegg AE. Conserved residue lysine165 is essential for the ability of O6-alkylguanine-DNA alkyltransferase to react with O6-benzylguanine.
Biochem J
2000
;
347
:
527
–34.
21
Xu-Welliver M, Kanugula S, Pegg AE. Isolation of human O6-alkylguanine-DNA alkyltransferase mutants highly resistant to inactivation by O6-benzylguanine.
Cancer Res
1998
;
58
:
1936
–45.
22
Brent TP, von Wronski M, Pegram CN, Bigner DD. Immunoaffinity purification of human O6-alkylguanine-DNA alkyltransferase using newly developed monoclonal antibodies.
Cancer Res
1990
;
50
:
58
–61.
23
Esteller M. Epigenetic lesions causing genetic lesions in human cancer: promoter hypermethylation of DNA repair genes.
Eur J Cancer
2000
;
36
:
2294
–300.
24
Deng C, Xie D, Capasso H, Zhao Y, Wang LD, Hong JY. Genetic polymorphism of human O6-alkylguanine-DNA alkyltransferase: identification of a missense variation in the active site region.
Pharmacogenetics
1999
;
9
:
81
–7.
25
Goldie JH, Coldman AJ. A mathematic model for relating the drug sensitivity of tumors to their spontaneous mutation rate.
Cancer Treat Rep
1979
;
63
:
1727
–33.
26
Goldie JH, Coldman AJ. A model for tumor response to chemotherapy: an integration of the stem cell and somatic mutation hypotheses.
Cancer Invest
1985
;
3
:
553
–64.
27
Morris SM. A role for p53 in the frequency and mechanism of mutation.
Mutat Res
2002
;
511
:
45
–62.
28
el-Deiry WS, Tokino T, Velculescu VE, et al. WAF1, a potential mediator of p53 tumor suppression.
Cell
1993
;
75
:
817
–25.
29
Offer H, Wolkowicz R, Matas D, Blumenstein S, Livneh Z, Rotter V. Direct involvement of p53 in the base excision repair pathway of the DNA repair machinery.
FEBS Lett
1999
;
450
:
197
–204.
30
Grombacher T, Eichhorn U, Kaina B. p53 is involved in regulation of the DNA repair gene O6-methylguanine-DNA methyltransferase (MGMT) by DNA damaging agents.
Oncogene
1998
;
17
:
845
–51.
31
Selvakumaran M, Lin HK, Miyashita T, et al. Immediate early up-regulation of bax expression by p53 but not TGFβ1: a paradigm for distinct apoptotic pathways.
Oncogene
1994
;
9
:
1791
–8.
32
Kat A, Thilly WG, Fang WH, Longley MJ, Li GM, Modrich P. An alkylation-tolerant, mutator human cell line is deficient in strand-specific mismatch repair.
Proc Natl Acad Sci U S A
1993
;
90
:
6424
–8.
33
Glaab WE, Tindall KR. Mutation rate at the hprt locus in human cancer cell lines with specific mismatch repair-gene defects.
Carcinogenesis
1997
;
18
:
1
–8.
34
Ohzeki S, Tachibana A, Tatsumi K, Kato T. Spectra of spontaneous mutations at the hprt locus in colorectal carcinoma cell lines defective in mismatch repair.
Carcinogenesis
1997
;
18
:
1127
–33.
35
Glaab WE, Risinger JI, Umar A, Barrett JC, Kunkel TA, Tindall KR. Cellular resistance and hypermutability in mismatch repair-deficient human cancer cell lines following treatment with methyl methanesulfonate.
Mutat Res
1998
;
398
:
197
–207.
36
de las Alas MM, Aebi S, Fink D, Howell SB, Los G. Loss of DNA mismatch repair: effects on the rate of mutation to drug resistance.
J Natl Cancer Inst
1997
;
89
:
1537
–41.
37
Chen WD, Eshleman JR, Aminoshariae MR, et al. Cytotoxicity and mutagenicity of frameshift-inducing agent ICR191 in mismatch repair-deficient colon cancer cells.
J Natl Cancer Inst
2000
;
92
:
480
–5.
38
Liu L, Schwartz S, Davis BM, Gerson SL. Chemotherapy-induced O(6)-benzylguanine-resistant alkyltransferase mutations in mismatch-deficient colon cancer.
Cancer Res
2002
;
62
:
3070
–6.
39
Minnick DT, Veigl ML, Sedwick WD. Mutational specificity of 1,3-bis-(2-chloroethyl)-1-nitrosourea in a Chinese hamster ovary cell line.
Cancer Res
1992
;
52
:
4688
–95.
40
Minnick DT, Gerson SL, Dumenco LL, Veigl ML, Sedwick WD. Specificity of bischloroethylnitrosourea-induced mutation in a Chinese hamster ovary cell line transformed to express human O6-alkylguanine-DNA alkyltransferase.
Cancer Res
1993
;
53
:
997
–1003.
41
Dickerson RE. Base sequence and helix structure variation in B and A DNA.
J Mol Biol
1983
;
166
:
419
–41.
42
Wibley JE, Pegg AE, Moody PC. Crystal structure of the human O(6)-alkylguanine-DNA alkyltransferase.
Nucleic Acids Res
2000
;
28
:
393
–401.
43
Daniels DS, Mol CD, Arvai AS, Kanugula S, Pegg AE, Tainer JA. Active and alkylated human AGT structures: a novel zinc site, inhibitor and extrahelical base binding.
EMBO J
2000
;
19
:
1719
–30.
44
Encell LP, Coates MM, Loeb LA. Engineering human DNA alkyltransferases for gene therapy using random sequence mutagenesis.
Cancer Res
1998
;
58
:
1013
–20.
45
Loktionova NA, Pegg AE. Interaction of mammalian O(6)-alkylguanine-DNA alkyltransferases with O(6)-benzylguanine.
Biochem Pharmacol
2002
;
63
:
1431
–42.
46
Kaur TB, Travaline JM, Gaughan JP, Richie JP Jr, Stellman SD, Lazarus P. Role of polymorphisms in codons 143 and 160 of the O6-alkylguanine DNA alkyltransferase gene in lung cancer risk.
Cancer Epidemiol Biomarkers Prev
2000
;
9
:
339
–42.
47
Xu-Welliver M, Pegg AE. Point mutations at multiple sites including highly conserved amino acids maintain activity but render O6-alkylguanine-DNA alkyltransferase insensitive to O6-benzylguanine.
Biochem J
2000
;
347
:
519
–26.
48
Holmes J Jr, Clark S, Modrich P. Strand-specific mismatch correction in nuclear extracts of human and Drosophila melanogaster cell lines.
Proc Natl Acad Sci U S A
1990
;
87
:
5837
–41.