Abstract

To investigate the possible role of glycosylase action in causing tumor resistance, a full-length, histidine-tagged human alkyladenine glycosylase has been purified from the cloned human gene contained in a pTrc99A vector propagated in a tag alkA mutant Escherichia coli . This human enzyme releases both 3-methyladenine and 7-methylguanine from methylated DNA but in contrast to previous studies of the bacterial AlkA glycosylase, it does not release any adducts from [ 3 H]chloroethylnitrosourea-modified DNA. This finding suggests that the alkyladenine DNA glycosylase-dependent resistance to the toxic effects of the chloroethylnitrosoureas reported previously in the literature may occur by a mechanism other than through direct glycosylase action.

Introduction

The chloroethylnitrosoureas (CNUs) were introduced into clinical practice as antitumor agents in the 1970s. Although they have achieved an important place in the treatment of brain tumors and certain lymphomas, other tumors are usually resistant to their cytotoxic action. Because of the importance of the resistance phenomenon in the treatment of cancer generally, the mechanism of resistance to the CNUs has been investigated in detail. Cytotoxicity is caused by DNA modification, in particular the formation of a cross-link between guanine and cytosine ( 1 ). Formation of this cross-link can be prevented by the DNA repair enzyme, O6 -alkylguanine-DNA alkyltransferase (AGT); high cellular levels of AGT are a major cause of resistance ( 2,3 ). This mechanism of cytotoxicity and cellular resistance has now become a model for understanding the role of DNA repair in causing tumor resistance. Indeed, protocols designed to decrease cellular levels of AGT and thereby restore tumor sensitivity have reached clinical trial and have been reviewed recently ( 4 ).

This mechanism of tumor resistance is analogous to the adaptive response in Escherichia coli ( 5 ). In these bacteria, Ada removes alkyl groups attached to the O6 position of guanine whereas a second enzyme, alkyladenine DNA glycosylase (AAG), releases 3- and 7-substituted purines from DNA. That a glycosylase might also be involved in repairing mammalian DNA and in causing tumor resistance was suggested by HPLC analysis of the DNA modifications in resistant tumor cells treated with a CNU ( 6 ). These studies showed that other DNA adducts besides the GC cross-link were absent from resistant cells in comparison with sensitive cells, indicating possible glycosylase action ( 6 ). Subsequent in vitro studies showed that many CNU-modified bases are substrates for the bacterial AlkA protein, 3-methyladenine DNA glycosylase II ( 7 ); further supporting the hypothesis that human AAG might be involved in the resistance of tumors to CNUs.

Investigations of the role of mammalian AAGs in reducing the toxicity of CNU to cultured cells have not, however, shown a simple relationship between glycosylase content and cellular resistance. Although a glial cell line resistant to the chloroethylnitrosoureas was found to have higher levels of alkyladenine glycosylase than a sensitive cell line ( 8 ), over-expression of human AAG in Chinese hamster ovary cells did not lead to increased resistance to CNUs ( 9 ). Comparison of CNU toxicity in wild-type and AAG knock out mouse embryonic stem cells showed a protective effect of AAG ( 10 ), but there was no evidence of protection when wild-type and AAG knock out mouse fibroblasts were compared ( 11 ).

In contrast to the enzyme alkyltransferase that completes DNA repair in a single step, the action of AAG is only the first step in the repair process; other enzymes are required to repair the abasic site left in DNA by AAG action. Thus, the varying results quoted above could depend on whether or not the glycosylase step is rate limiting in the repair process in these different cells.

As a first step in understanding these results, we have purified human AAG and tested its activity towards CNU-modified DNA. The human AAG gene has been cloned independently by several research groups, but the separate isolates differ in their N-terminal amino acid sequences ( 12 ). As these differences could affect substrate specificity, especially in view of the data of Roy et al . ( 13 ) showing that the N-terminal region of AAG is positioned close to the active site, we felt it was important to study the specificity of a full-length enzyme. The plasmid constructed for this study allows purification of full-length, histidine-tagged human AAG of the isoform described by Samson et al . ( 14 ). We report that the human enzyme differs from the bacterial enzyme in its activity towards CNU-modified DNA. Both human and bacterial glycosylases are able to remove modified bases from methylated DNA, but the human enzyme has no demonstrable activity towards CNU-modified DNA. As discussed below, this result could indicate either that hAAG requires an additional factor to be active on CNU lesions in vivo or that this protein provides cellular protection by a mechanism other than its glycosylase action.

Materials and methods

N -(2-chloro-1,2-[ 3 H]-ethyl)- N -nitrosourea ([ 3 H]-CNU), sp. act. 1.3 Ci/mmol, was custom synthesized by Moravek Biochemicals (La Brea, CA). N -( 3 H-methyl)- N -nitrosourea ([ 3 H]-MNU) with a specific activity of 17.0 Ci/mmol was purchased from Amersham (Arlington Heights, IL). Unlabeled CNU was provided by the Division of Cancer Treatment, National Cancer Institute, Bethesda, MD. Optical markers of 3-methyladenine (m 3 A) and 7-methylguanine (m 7 G) used in the HPLC separations were obtained from Cyclo Chemical (Los Angeles, CA). Other chemicals were reagent grade materials.

[ 3 H]MNU- and [ 3 H]CNU-modified DNA were prepared by alkylating calf thymus DNA with the radiolabeled MNU and CNU as described previously ( 15 ). Purification and assay of E.coli AlkA glycosylase and assay of human glycosylase also followed our published procedures ( 7 ) except that bacterial glycosylase was assayed in its optimum buffer (70 mM Tris, pH 7.6, containing 10 mM NaEDTA and 2 mM dithiothreitol) whereas human glycosylase was assayed in its slightly different optimum buffer (50 mM HEPES, pH 7.5, containing 1 mM NaEDTA, 5 mM mercaptoethanol and 100 mM KCl).

Bacterial strains

Bacterial strains used in this study and their relevant genotypes are listed in Table I. Plasmids are described in detail below.

Construction of plasmids expressing human glycosylase

Our constructs started with plasmid pBU16, a gift from Professor Leona Samson (Massachusetts Institute of Technology, MA). Plasmid pBU16 carries a Bsu 36I– Xba I fragment from pKT218 that encodes the human alkyladenine DNA glycosylase gene ( hAAG ) ( 14 ) in the pSL301 vector (Invitrogen, San Diego, CA). We subcloned the hAAG -bearing Eco RI– Bst EII fragment from pBU16 into pTrc99A (Pharmacia Biotech, Piscataway, NJ) to produce plasmid pMV503. The pTrc99A vector carries the lacI gene of E.coli and has the synthetic trc promoter under lac operator control positioned upstream of the Eco RI site. The presence of lacI on the plasmid results in tight repression of the promoter that can be induced by the addition of the lac inducer isopropyl β- d -thiogalactopyranoside (IPTG), thus allowing IPTG-inducible expression of hAAG .

Plasmid pBU16 produces a fusion protein of hAAG and vector-encoded sequences 5′ to hAAG . To remove the pBU16 vector coding sequences from the 5′ end of the hAAG gene, the Eco RI– Bst EII fragment of pMV503 was removed by first digesting to completion with Eco RI, then partially digesting with Bst EII to produce a 5152 bp fragment. This fragment was purified from a 0.8% agarose gel using the Gene Clean System (Bio101, La Jolla, CA). Oligonucleotides MV1 (AATTCTAAGGAGGTATCTAATG) and MV2 (GTGACCATTAGATACCTCCTTAG) were first annealed to one another by heating and slow cooling, then ligated to the purified 5152 bp Eco RI– Bst EII fragment of pMV503 to produce plasmid pMV509. Oligonucleotides MV1 and MV2 have four important features: (i) they are complementary to one another; (ii) when annealed, they produce single-stranded DNA ends complementary to the Eco RI and Bst EII sites of pMV503; (iii) they reconstruct the ATG initiation codon of hAAG that lies within the Bst EII site; and (iv) they contain a consensus ribosome binding site, AGGAGG, appropriately positioned to allow optimal translation initiation from the downstream ATG initiation codon. The resulting plasmid, pMV509, carries the wild-type hAAG under the control of the IPTG-inducible trc promoter of pTrc99A. Construction of pMV509 was tested by loss of the Sal I restriction site that lies between Eco RI and Bst EII of pMV503 and restoration of the Eco RI and Bst EII sites. The hAAG sequence was confirmed independently.

To construct the histidine-tagged hAAG gene, pMV509 was digested to completion with Cel II and Hin DIII to remove the 3′ end of the gene and the resulting 5064 bp fragment was purified from a 0.8% agarose gel as described above to remove the 3′ end of the gene. Oligonucleotides MV3 (TGAGCAGGACACACAGGCCCATCATCATCATCATCACTGA) and MV4 (AGCTTCAGTGATGATGATGATGATGGGCCTGTGTGTCCTGC) were first annealed to one another as described above, then ligated to the 5064 bp fragment of pMV509 to produce plasmid pMV513. Oligonucleotides MV3 and MV4 have four important features: (i) they are complementary to one another; (ii) when annealed, they produce single-stranded ends complementary to Cel II and Hin DIII; (iii) they restore the last seven amino acid codons of the hAAG gene in the appropriate reading frame; and (iv) they add six histidine codons and one stop codon to the 3′ end of the hAAG gene. Construction of pMV513 was tested by loss of the Mlu I restriction site that lies within the Cel II– Hin DIII fragment of pMV509 and restoration of the Cel II and Hin DIII sites, and DNA sequencing of the insert.

Effect of plasmids expressing hAAG on sensitivity to methyl methanesulfonate

As the hAAG gene was initially cloned by functional complementation of an E.coli mutant strain lacking the alkA and tag DNA glycosylase genes ( 14 ), the function of the wild-type and histidine-tagged hAAG genes was tested by assessing the ability of pMV509 and pMV513 to complement the alkylation sensitivity of the alkA1 tag-1 E.coli double mutant. Strain MV2157 ( alkA1 tag-1 ), and its isogenic, plasmid-bearing derivatives, MV4122 (MV2157/pMV509) and MV4126 (MV2157/pMV513), or their uvrA6 derivatives, MV4236 ( alkA1 tag-1 uvrA6 /pTrc99 vector), MV4237 ( alkA1 tag-1 uvrA6 /pMV509), MV4238 ( alkA1 tag-1 uvrA6 /pMV513), were grown to early-log phase in LB ampicillin media at 37°C. Cultures were then divided into two aliquots, one of which was induced with IPTG (2 mM) and incubated at 37°C for an additional 90 min to allow induction. Cultures were then treated with methyl methanesulfonate (MMS) (20 mM) for 30 min at 37°C. After treatment, cells were diluted 1:100 in buffer containing sodium thiosulfate (4%) to inactivate MMS ( 16 ). Cultures were immediately diluted further and plated on LB ampicillin agar plates to assess cell survival. The results in Table II demonstrate that both plasmids are able to restore MMS resistance to MV2157 after IPTG treatment as expected if both the wild-type and histidine-tagged hAAG genes were inducible and active. Moreover, the level of resistance attained is similar, indicating that the addition of a histidine tag has not reduced activity of the hAAG enzyme.

Purification of glycosylases

Bacterial glycosylase II, the alkA protein, was isolated from E.coli MS23, which harbors the pYN1000 plasmid containing the AlkA gene ( 17 ). Lysed cells were treated with DEAE cellulose to remove DNA and the enzyme was purified through the phosphocellulose and DNA cellulose chromatography steps as described ( 17 ). The final product migrated as a single band with an apparent molecular weight of 39 000 Da as visualized by silver staining on a 12.5% polyacrylamide gel with 0.1% sodium dodecyl sulfate. This enzyme and the human glycosylase were assayed with [ 3 H]-MNU-DNA as described previously ( 18 ). One unit of enzyme activity was defined as the amount of enzyme that released 1 pmol of bases from [ 3 H]-MNU-DNA in 10 min at 37°C.

The hAAG-(his) 6 enzyme was isolated from strain MV4211 ( alkA1 tag-1 ompT /pMV513). Cells were grown to a Klett reading of 70 (∼5×10 8 cells/ml) in LB broth containing ampicillin (100 μg/ml), then induced by the addition of IPTG (2 mM). Fresh ampicillin (100 μg/ml) was also added at this time to insure plasmid maintenance, and cells were incubated for an additional 5 h.

After incubation, cells (8 g) were centrifuged, washed in saline buffer (10 mM Tris–Cl, pH 7.4, 1 mM EDTA, 100 mM NaCl) and lysed using a Kraft homogenizer followed by a French press. Crude cell extracts were then centrifuged at 12 000 g and the supernatant recovered. The supernatant was mixed with 24 ml of a 50% slurry of Ni–agarose (Qiagen) and, after stirring on ice for 1 h, was poured into a 1.6×24 cm column equilibrated with column buffer (5 mM Na 2 HPO 4 , pH 8.0, 300 mM NaCl, 10% glycerol). The column was washed with 200 ml of column buffer followed by 300 ml of 30 mM imidazole in column buffer, and finally eluted with a 100 ml gradient of 30–500 mM imidazole in column buffer. Two milliliter fractions were collected and assayed for glycosylase activity with [ 3 H]MNU-modified DNA. A sharp peak of activity appeared around fraction 14; fractions containing high activity were analyzed by gel electrophoresis as described in the Results section .

Substrate specificity of the human glycosylase

To obtain an HPLC profile of [ 3 H]MNU-modified DNA, a sample of the [ 3 H]MNU-DNA substrate containing 136 000 c.p.m. (49 μg) of [ 3 H]methyl adducts was depurinated in 0.1 N HCl for 18 h at 37°C. The hydrolysate was adjusted to pH 4.5 with 1 N NaOH and passed through a 2 ml A25 ion exchange column to remove oligonucleotides. Optical markers of m 3 A and m 7 G were added and an aliquot was separated on a C 18 column eluted with a 50 mM KH 2 PO 4 , pH 4.5 acetonitrile buffer system at 1 ml/min. One minute fractions were collected, counted in a Beckman liquid scintillation counter, and plotted versus fraction number. Similarly, spontaneous release from a separate sample of [ 3 H]MNU-DNA substrate containing 136 000 c.p.m. was determined after incubation for 1 h at 37°C in buffer (50 mM HEPES, pH 7.5, containing 1 mM NaEDTA, 5 mM mercaptoethanol and 100 mM KCl). Enzymatic release from [ 3 H]MNU-DNA by 0.15 U of bacterial or human enzyme was determined after a 1 h incubation at 37°C in the same buffer. To investigate the activity of the bacterial and human enzymes for [ 3 H]CNU-DNA, 3×10 4 c.p.m. of [ 3 H]CNU-DNA were incubated with increasing amounts of the enzymes for 10 min at 37°C, conditions under which the bacterial enzyme release is 30% complete and spontaneous release is low ( 19 ).

Results

The design of plasmids pMV509 and pMV513 is shown in Figure 1 . Both plasmids contain the full-length sequence for the human glycosylase gene; pMV513 also contains a (his) 6 insert at the C-terminus of the protein to facilitate its purification. This insert was positioned at the C-terminus rather than at the N-terminus because the amino end of the protein has been reported to be located near the active site ( 13 ), and a (his) 6 insert at this position might affect the specificity of the glycosylase. As described in the Materials and methods, the pTrc99A vector (Pharmacia Biotech) carries the lacI gene of E.coli and has the synthetic trc promoter under lac operator control positioned upstream of the Eco RI site to control transcription. The presence of the wild-type hAAG gene in the plasmids was tested by restriction digestion with Bst EII, Cel II, Bgl I, Eco RI, Hin DIII and Sac I followed by gel electrophoresis and confirmed by DNA sequencing.

The protective activity of pMV509 and pMV513 when these plasmids were propagated in E.coli cells deficient in glycosylase activity is shown in Table II . The doses of MMS and CNU used produced nearly identical levels of survival in the repair-deficient control strain MV4236. The base excision-proficient, nucleotide excision-deficient E.coli strain MV1176 shows an 89% survival after exposure to 20 mM MMS and a much lower, but appreciable, survival after exposure to 0.5 mM CNU. Survival after exposure to MMS and CNU in the repair-deficient control cells, E.coli MV4236 that lack glycosylase and excision repair capabilities but that harbor the unmodified vector, was lower. The presence of plasmids pMV509 and pMV513, carrying hAAG and hAAG-(his) 6 , respectively, greatly increased survival after exposure to MMS in the repair-deficient strain, but did not seem to increase survival after exposure to CNU. In fact when hAAG is expressed in E.coli , it results in a small but reproducible decrease in survival, suggesting hAAG expression may either interfere with other DNA repair mechanisms active on CNU lesions, or convert CNU lesions to a more toxic form. This is similar to the glycosylase-mediated sensitivity to a variety of alkylators described by others [Matijasevic and Volkert, unpublished results; ( 20 )]. Similar levels of MMS resistance and CNU sensitization were obtained regardless of the presence or absence of the histidine tag, thereby eliminating the possibility that the presence of the tag affects hAAG activity towards MMS or CNU lesions. Note that in this regard the glycosylase was originally identified by its ability to protect against the toxic effects of methylating and not chloroethylating agents.

In order to avoid contamination with bacterial glycosylases in purifying hAAG, plasmids containing the hAAG gene were propagated in a glycosylase-deficient strain of E.coli , MV2157, which lacks both the tag and alkA genes. Attempts to purify human glycosylase from pMV509-carrying cells using classical separation methods resulted in poor yields of the enzyme. Accordingly, further purification studies were performed using pMV513 that produces the hAAG enzyme with a (his) 6 tail.

An active glycosylase was obtained when human enzyme was isolated on a Ni–agarose column from E.coli MV4126, which harbors this plasmid. However, N-terminal amino acid analysis revealed that the enzyme had been truncated between two arginines near the amino acid end of the protein. This site of cleavage indicated that the enzyme that was responsible for this truncation was probably the OmpT protease found in the outer membrane of E.coli cells ( 21 ). Accordingly, the OmpT gene in MV4126 was disrupted by the introduction of an allele of OmpT containing an insertion element expressing kanamycin resistance. This host cell was designated MV4211.

The construct pMV513 that contains the human glycosylase was grown in these cells and purified through a Ni–agarose column step as described in the Materials and methods. A typical separation is shown in the upper panel of Figure 2 . Fractions containing enzyme activity were concentrated using 3K Microsep microconcentrators (Pall Filtron Corporation, Northboro, MA), and separated by SDS–polyacrylamide gel electrophoresis (SDS–PAGE) on a 10% gel as shown in the lower panel of Figure 2 . Protein bands from the SDS–PAGE gel were transferred to a Millipore Immobilon-P SQ transfer membrane (0.1 μm pore size) of polyvinylidene fluoride (PVDF) in preparation for N-terminal analysis. N-terminal sequencing by Edman degradation using an Applied Biosystems 494 Procise protein sequencer showed that the 39 kDa band from fraction 14 terminated in VTPALQMKKP in agreement with the sequence found by Samson et al . ( 14 ). The apparent molecular weight is also in agreement with the value obtained for this enzyme by O’Connor ( 22 ), who also reported purification of intact full-length human glycosylase. The contaminating protein at 31 kDa was identified by N-terminal analysis as the histidine-rich E.coli protein, SlyD.

The activity of this glycosylase towards [ 3 H]MNU-modified DNA substrate is shown in Figure 3 . It is clear from this figure that hAAG releases both m 3 A and m 7 G from [ 3 H]MNU-DNA; although the human enzyme shows similar activity towards m 3 A, it shows somewhat lower activity towards m 7 G than the bacterial glycosylase. This result also agrees with O’Connor’s data on purified human glycosylase ( 22 ).

With active full-length human glycosylase available, it was now possible to test the activity of hAAG towards chloroethylnitrosourea-modified DNA. Bacterial and human enzymes were incubated with [ 3 H]-CNU-modified DNA for 10 min at 37°C and the c.p.m. of adducted bases released into the supernatant were measured as shown in Figure 4 . While the bacterial enzyme released CNU-adducted bases in an enzyme-dependent way, there was absolutely no evidence for their release by the human enzyme even when much higher levels of enzyme were added to the incubation mixture. Thus, the hAAG has no in vitro activity towards [ 3 H]-CNU-modified DNA under conditions where the bacterial enzyme is highly active suggesting that other factors must be involved in the cellular activity of hAAG as discussed below.

Discussion

After the construction of pMV513 and the development of the E.coli host cells MV4211, human alkyladenine glycosylase was successfully isolated from a Ni–agarose column as described above. This has made it possible to study the specificity of a full-length human alkyladenine glycosylase on CNU-modified DNA in vitro . In view of our earlier results showing that bacterial 3-methyl-adenine DNA glycosylase II releases a wide range of CNU-modified bases ( 7 ), we were indeed surprised at the lack of activity of the human enzyme towards CNU-modified DNA.

We had thought that hAAG might protect against the toxic effects of CNU in a manner similar to the bacterial glycosylase, which releases CNU-adducted bases ( 15,23 ). Although the data of Allan et al . ( 10 ) show that the presence of mouse AAG decreases the sensitivity of mouse cells to chloroethylnitrosoureas and the data of Matijasevic et al . suggest that hAAG may play a role in the resistance of human tumors to CNU ( 8 ) human AAG does not seem to release chloroethylnitrosourea-modified bases from CNU-modified DNA in vitro . Possible explanations for this could be that: (i) hAAG releases an extremely toxic base that is present in such small amounts that its release is not apparent in our experiments with [ 3 H]-CNU-modified DNA; (ii) hAAG requires the presence of an additional cellular factor to act on CNU-modified DNA; (iii) the activity of hAAG is affected by post-translational modification that differs in mammalian and E.coli host cells; or (iv) the hAAG protein serves some additional function besides base recognition and release in the cellular environment. The possibility of additional factors, post-translational modification or additional functions of hAAG must be specific for its role in repair of CNU lesions and do not apply to its interactions with methyl lesions, as the enzyme effectively protects against MMS exposure and methyl lesions are efficiently removed by the purified hAAG enzyme (Table II and Figure 3 ).

Unless there are other minor but very toxic DNA adducts that we have not detected, possibilities (ii)–(iv) seem more likely. However, for possibility (ii) to explain our results we would have to conclude that some factor in E.coli cells can selectively activate, or some factor from mammalian cells is not needed to activate the human enzyme towards repair of methyl lesions, because the human gene was identified by complementation of glycosylase-negative E.coli cells ( 14 ) and it is clearly active against this type of DNA damage (Table II and Figure 3 ). Post-translational modification of the enzyme has not been reported, but cannot be ruled out. The final possibility, number (iv), is intriguing but again there is no evidence for such a function. Because of the importance of the resistance problem in treating human tumors, further experimentation is needed to discriminate between these possibilities.

Table 1.

Bacterial strains

Strain name Relevant genotype Plasmid 
MV1176 uvrA6 None 
MV2157 alkA1 tagA1 None 
MV4122 alkA1 tagA1 pMV509 
MV4126 alkA1 tagA1 pMV513 
MV4211 alkA1 tagA1 ompT pMV513 
MV4236 alkA1 tagA1 uvrA6 pTrc99a 
MV4237 alkA1 tagA1 uvrA6 pMV509 
MV4238 alkA1 tagA1 uvrA6 pMV513 
Strain name Relevant genotype Plasmid 
MV1176 uvrA6 None 
MV2157 alkA1 tagA1 None 
MV4122 alkA1 tagA1 pMV509 
MV4126 alkA1 tagA1 pMV513 
MV4211 alkA1 tagA1 ompT pMV513 
MV4236 alkA1 tagA1 uvrA6 pTrc99a 
MV4237 alkA1 tagA1 uvrA6 pMV509 
MV4238 alkA1 tagA1 uvrA6 pMV513 
Table II.

Effect of genotype on cell survival a

Genotype MV1176 MV4236 MV4237 MV4238 
a As percent of growth of unexposed cultures.  
alkA – – – 
tagA – – – 
uvrA – – – – 
Plasmid – vector pMV509 (hAAG)  pMV513 [(hAAG-(his) 6 )]  
Survival after 20 mM MMS 89% 0.04% 42.4% 46.5% 
Survival after 0.5 mM CNU  0.12% 0.13%  0.023%  0.048% 
Genotype MV1176 MV4236 MV4237 MV4238 
a As percent of growth of unexposed cultures.  
alkA – – – 
tagA – – – 
uvrA – – – – 
Plasmid – vector pMV509 (hAAG)  pMV513 [(hAAG-(his) 6 )]  
Survival after 20 mM MMS 89% 0.04% 42.4% 46.5% 
Survival after 0.5 mM CNU  0.12% 0.13%  0.023%  0.048% 
Fig. 1.

Insert regions of the hAAG expression vector pMV509 (top) and the hAAG-(his) 6 expression vector pMV513 (bottom). Figures are drawn on the same scale. The P TRC promoter is indicated by the rightward arrow labeled P TRC , the hAAG coding sequences are indicated by the gray boxes, the histidine tag is indicated by the black box, and the ends of the pTrc vector sequences are indicated as heavy dotted lines.

Fig. 1.

Insert regions of the hAAG expression vector pMV509 (top) and the hAAG-(his) 6 expression vector pMV513 (bottom). Figures are drawn on the same scale. The P TRC promoter is indicated by the rightward arrow labeled P TRC , the hAAG coding sequences are indicated by the gray boxes, the histidine tag is indicated by the black box, and the ends of the pTrc vector sequences are indicated as heavy dotted lines.

Fig. 2.

(Top) Elution profile of the hAAG-(his) 6 enzyme from a Ni–agarose column. (Bottom) Gel separations of individual enzyme fractions 10–18. Molecular weight standards are in lanes marked S. N-terminal analysis was performed on the 39 kDa band from fraction 14.

Fig. 2.

(Top) Elution profile of the hAAG-(his) 6 enzyme from a Ni–agarose column. (Bottom) Gel separations of individual enzyme fractions 10–18. Molecular weight standards are in lanes marked S. N-terminal analysis was performed on the 39 kDa band from fraction 14.

Fig. 3.

HPLC profiles of bases released from methylated DNA by acid treatment, by spontaneous release, by 0.15 U of bacterial enzyme, and by 0.15 U of the hAAG-(his) 6 enzyme. Incubation mixtures contained 49 μg of DNA with 1.36×10 5 c.p.m. of [ 3 H]-MNU adducts in 500 μl of buffer solution. Optical markers are shown for peaks corresponding to m 3 A and m 7 G.

Fig. 3.

HPLC profiles of bases released from methylated DNA by acid treatment, by spontaneous release, by 0.15 U of bacterial enzyme, and by 0.15 U of the hAAG-(his) 6 enzyme. Incubation mixtures contained 49 μg of DNA with 1.36×10 5 c.p.m. of [ 3 H]-MNU adducts in 500 μl of buffer solution. Optical markers are shown for peaks corresponding to m 3 A and m 7 G.

Fig. 4.

Activity of bacterial (open circle) and histidine-tagged human enzyme (shaded circle), towards 3 H-CNU-modified DNA. Incubation mixtures contained 27 μg of DNA with 2×10 4 c.p.m. of [ 3 H]-CNU adducts and the indicated number of units of bacterial or human glycosylase in 150 μl of buffer solution. Incubations were for 10 min at 37°C.

Fig. 4.

Activity of bacterial (open circle) and histidine-tagged human enzyme (shaded circle), towards 3 H-CNU-modified DNA. Incubation mixtures contained 27 μg of DNA with 2×10 4 c.p.m. of [ 3 H]-CNU adducts and the indicated number of units of bacterial or human glycosylase in 150 μl of buffer solution. Incubations were for 10 min at 37°C.

3
To whom correspondence should be addressed Email: michael.volkert@umassmed.edu

The authors would like to express their thanks to Professor Leona Samson for supplying the cloned human glycosylase and Kenneth Bonanno for technical assistance. Supported by NIH grant CA44499, by the Massachusetts Division of the American Cancer Society, and by the US Army Medical Research and Materiel Command under Contracts DAMD17-96-C-6073 and DAMD17-00-C-0012. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the views of the supporting organizations.

References

1
Tong,W.P., Kirk,M.C. and Ludlum,D.B. (
1982
) 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.
  ,
42
,
3102
–3105.
2
Erickson,L.C., Laurent,G., Sharkey,N.A. and Kohn,K.W. (
1980
) DNA cross-linking and monoadduct repair in nitrosourea-treated human tumour cells.
Nature
  ,
288
,
727
–729.
3
Pegg,A.E., Dolan,M.E. and Moschel,R.C. (
1995
) Structure, function and inhibition of O6-alkylguanine-DNA alkyltransferase.
Prog. Nucleic Acid Res. Mol. Biol.
  ,
51
,
167
–223.
4
Gerson,S.L. (
2002
) Clinical relevance of MGMT in the treatment of cancer.
J. Clin. Oncol.
  ,
20
,
2388
–2399.
5
Samson,L. (
1992
) The suicidal DNA repair methyltransferases of microbes.
Mol. Microbiol.
  ,
6
,
825
–831.
6
Bodell,W.J., Tokuda,K. and Ludlum,D.B. (
1988
) Differences in DNA alkylation products formed in sensitive and resistant human glioma cells treated with N-(2-chloroethyl)-N-nitrosourea.
Cancer Res.
  ,
48
,
4489
–4492.
7
Ludlum,D.B., Li,Q. and Matijasevic,Z. (1999) Role of Base Excision Repair in Protecting Cells from the Toxicity of Chloroethylnitrosoureas . IARC Scientific Publications, IARC, Lyon, Vol. 150, pp. 271–277.
8
Matijasevic,Z., Bodell,W.J. and Ludlum,D.B. (
1991
) 3-Methyladenine DNA glycosylase activity in a glial cell line sensitive to the haloethylnitrosoureas in comparison with a resistant cell line.
Cancer Res.
  ,
51
,
1568
–1570.
9
Bramson,J., O’Connor,T. and Panasci,L. (
1995
) Effect of alkyl-N-purine DNA glycosylase overexpression on cellular resistance to bifunctional alkylating agents.
Biochem. Pharmacol.
  ,
50
,
39
–44.
10
Allan,J.M., Engelward,B.P., Dreslin,A.J., Wyatt,M.D., Tomasz,M. and Samson,L.D. (
1998
) Mammalian 3-methyladenine DNA glycosylase protects against the toxicity and clastogenicity of certain chemotherapeutic DNA cross-linking agents.
Cancer Res.
  ,
58
,
3965
–3973.
11
Elder,R.H., Jansen,J.G., Weeks,R.J. et al. (
1998
) Alkylpurine-DNA-N-glycosylase knockout mice show increased susceptibility to induction of mutations by methyl methanesulfonate.
Mol. Cell. Biol.
  ,
18
,
5828
–5837.
12
Vickers,M.A., Vyas,P., Harris,P.C., Simmons,D.L. and Higgs,D.R. (
1993
) Structure of the human 3-methyladenine DNA glycosylase gene and localization close to the 16p telomere.
Proc. Natl Acad. Sci. USA
  ,
90
,
3437
–3441.
13
Roy,R., Kumar,A., Lee,J.C. and Mitra,S. (
1996
) The domains of mammalian base excision repair enzyme N-methylpurine-DNA glycosylase. Interaction, conformational change and role in DNA binding and damage recognition.
J. Biol. Chem.
  ,
271
,
23690
–23697.
14
Samson,L., Derfler,B., Boosalis,M. and Call,K. (
1991
) Cloning and characterization of a 3-methyladenine DNA glycosylase cDNA from human cells whose gene maps to chromosome 16.
Proc. Natl Acad. Sci. USA
  ,
83
,
5607
–5610.
15
Habraken,Y., Carter,C.A., Sekiguchi,M. and Ludlum,D.B. (
1991
) Release of N2,3-ethanoguanine from haloethylnitrosourea-treated DNA by Escherichia coli 3-methyladenine DNA glycosylase II.
Carcinogenesis
  ,
12
,
1971
–1973.
16
Volkert,M.R. and Nguyen,D.C. (
1984
) Induction of specific Escherichia coli genes by sublethal treatments with alkylating agents.
Proc. Natl Acad. Sci. USA
  ,
81
,
4110
–4114.
17
Nakabeppu,Y., Miyata,T., Kondo,H., Iwanaga,S. and Sekiguchi,M. (
1984
) Structure and expression of the alkA gene of Escherichia coli involved in adaptive response to alkylating agents.
J. Biol. Chem.
  ,
259
,
13730
–13736.
18
Habraken,Y., Carter,C.A., Kirk,M.C., Riordan,J.M. and Ludlum,D.B. (
1990
) Formation of N2,3-ethanoguanine in DNA after in vitro treatment with the therapeutic agent, N-(2-chloroethyl)-N′-cyclohexyl-N-nitrosourea.
Carcinogenesis
  ,
11
,
223
–228.
19
Carter,C.A., Habraken,Y. and Ludlum,D.B. (
1988
) Release of 7-alkylguanines from haloethylnitrosourea-treated DNA by E.coli 3-methyladenine-DNA glycosylase II.
Biochem. Biophys. Res. Commun.
  ,
155
,
1261
–1265.
20
Roth,R.B. and Samson,L.D. (
2002
) 3-Methyladenine DNA glycosylase-deficient Aag null mice display unexpected bone marrow alkylation resistance.
Cancer Res.
  ,
62
,
656
–660.
21
Sugimura,K. and Nishihara,T. (
1988
) Purification, characterization and primary structure of Escherichia coli protease VII with specificity for paired basic residues: identity of protease VII and OmpT.
J. Bacteriol.
  ,
170
,
5625
–5632.
22
O’Connor,T.R. (
1993
) Purification and characterization of human 3-methyladenine-DNA glycosylase.
Nucleic Acids Res.
  ,
21
,
5561
–5569.
23
Habraken,Y., Carter,C.A., Kirk,M.C. and Ludlum,D.B. (
1991
) Release of 7-alkylguanines from N-(2-chloroethyl)-N′-cyclohexyl-N-nitrosourea-modified DNA by 3-methyladenine DNA glycosylase II.
Cancer Res.
  ,
51
,
499
–503