Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

O⁶-Methylguanine DNA methyltransferase (MGMT) directly repairs DNA damage at the O⁶-position of guanine generated by chemotherapeutic agents such as the chloroethylnitrosoureas (CENUs; Pegg et al., 1995). Left unrepaired, these adducts rearrange and lead to the formation of interstrand DNA cross-links, which are cytotoxic because they disrupt DNA replication (Toorchen and Topel, 1983). MGMT repairs the O⁶-adduct before the formation of a DNA cross-link by transferring the alkyl group to a cysteine residue located within the acceptor site of the protein. Removal of the O⁶-alkyl lesion protects against CENU-induced cytotoxicity (Pegg et al., 1995). Because this reaction is stoichiometric, the cellular level of MGMT correlates with CENU resistance.

Several preclinical studies have demonstrated CENU-resistant blood cells after increasing the expression of MGMT in the bone marrow (BM) via gene transfer. Retroviral-mediated expression of MGMT protects both murine and human BM cells against CENU-induced cytotoxicity (Allay et al., 1995, 1997; Moritz et al., 1995; Jelinek et al., 1996; Maze et al., 1996, 1997; Reese et al., 1996; Davis et al., 1997). Our laboratory has demonstrated that transduction of murine hematopoietic stem cells with a retroviral vector encoding the human MGMT cDNA protects progenitor cells in vitro and long-term repopulating myeloid and lymphoid cells in vivo against 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU)-induced cytotoxicity (Moritz et al., 1995; Maze et al., 1996).

Many primary tumors have demonstrated an increase in MGMT and CENU resistance compared with corresponding normal tissue (Preuss et al., 1996; Silber et al., 1998). MGMT can be rapidly depleted by 6-BG, which irreversibly binds to MGMT via the formation of S-benzylcysteine (Dolan et al., 1990a,b). Once benzylated, MGMT is degraded (Pegg et al., 1991). Several preclinical studies have demonstrated that 6-BG can restore CENU tumor cell sensitivity in vitro and in a variety of human tumor xenografts in vivo by depleting tumor cell MGMT activity (Futscher et al., 1989; Dolan et al., 1990a,b, 1991; Mitchell et al., 1992; Gerson et al., 1993; Marathi et al., 1994; Phillips et al., 1997). As a result, 6-BG is currently being tested in clinical trials (Koc et al., 1996).

Because hematopoietic cells express low levels of endogenous MGMT protein, the addition of 6-BG could further potentiate the degree of CENU-induced myelosuppression observed in vivo (Gerson et al., 1985; Gerson et al., 1986; Moritz et al., 1995; Reese et al., 1996). Several MGMT mutants have previously been generated, by site-directed and random mutagenesis techniques, and have been shown to confer resistance to 6-BG (Crone et al., 1994; Loktionova and Pegg, 1996; Encell et al., 1998; Xu-Welliver et al., 1998). Crone et al. (1994) demonstrated that two mutants, one containing an amino acid substitution of alanine for proline at position 140 (P140A) and another containing an amino acid substitution of alanine for glycine at position 156 (G156A), were 20- and 240-fold more resistant to 6-BG inactivation of MGMT compared with wild-type MGMT (WTMGMT). In addition, a double mutant containing both the P140A and G156A mutations (P140A/G156A) conferred 1200-fold more resistance to 6-BG compared with WTMGMT (Crone et al., 1994). Loktionova and Pegg (1996) further demonstrated that expression of either P140A or G156A protected Chinese hamster ovary cells against 6-BG sensitization to BCNU in vitro. Similarly, Hickson et al. (1996) observed that expression of P140A/G156A protected hamster fibroblast cells against 6-BG and mitozolomide treatment in vitro, while noting that the P140A/G156A protein demonstrated a 10-fold decrease in repair activity in these cells.

More recently, investigators have demonstrated that retroviral expression of G156A in hematopoietic cells was associated with a 2-fold shift in the BCNU IC₅₀ for human (Reese et al., 1996) and a 5-fold shift for mouse BM cells (Davis et al., 1997) in vitro after 10 µM 6-BG treatment. Expression of G156A was also noted to result in less protein and repair activity compared with WTMGMT (Reese et al., 1996). This study, along with the observations noted above, raises questions about the stability and activity of MGMT mutants expressed in hematopoietic cells. Therefore, we focused our analysis on the P140A and P140A/G156A mutants, because the stability of these mutants has never been reported in hematopoietic cells. Our results clearly demonstrate that retroviral-mediated expression of the P140A mutant form of MGMT via a retroviral vector leads to a stable protein that protects hematopoietic cells against 6-BG and BCNU treatment and demonstrates a potential approach to increase the therapeutic index for the treatment of CENU-resistant tumors.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of Retroviral Vectors and Producer Cells. Moloney stem cell virus (MSCV2.1) retroviral vectors were constructed to express the cDNA sequence for either WTMGMT, MGMT^P140A (P140A), or MGMT^P140A/G156A (P140A/G156A) (Fig. 1). DNA sequences were amplified from pINAGT (Crone et al., 1994) expression plasmids containing the respective cDNAs by polymerase chain reaction. Briefly, a 5'-oligonucleotide sequence (GGCCGCGAATTCATGGACAAGGATTGTGAAATG) containing an EcoRI restriction site and a 3'-oligonucleotide (CCGCTCGAGTCAGTTTCGGCCAGCAGGCGGGGA) containing an XhoI restriction site were used to amplify the 623-base pair human MGMT cDNAs. The amplified products were purified with a Qiaex II gel extraction kit (Qiagen Inc., Chatsworth, CA), cut with EcoRI and XhoI, and cloned into EcoRI-XhoI restriction sites of MSCV2.1. Positive clones were identified by diagnostic restriction analysis and sequenced to confirm both the presence of the desired mutation and the lack of other mutations that may have arisen during polymerase chain reaction. Retroviral producer lines were generated for each construct by transfecting plasmid DNA into GP+envAm12 (Markowitz et al., 1988a) with N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate, according to the manufacturer's recommendation (Boehringer Mannheim, Indianapolis, IN). Transient virus, harvested 24 to 36 h after transfection, was used to infect GP + E-86 cells (Markowitz et al., 1988b) in the presence of 8 µg/ml Polybrene (Aldrich Chemical Co., Milwaukee, WI). GP + E-86 cells, split 1:40 in a six-well plate, were infected once every 24 h for 5 days with fresh virus and replated in limiting dilution in the presence of 0.75 mg/ml G418 (dry powder; GIBCO-BRL, Gaithersburg, MD). After 10 to 12 days of culture, G418-resistant clones were individually picked and expanded. Clones were screened both for viral titer by standard methods (Hanenberg et al., 1996), and for MGMT repair activity as described below (performed on G418-resistant NIH/3T3 populations). Retroviral producers used to infect L1210 cells (WT; clone 2, P140A; clones 19, 21, 43, and 62, P140A/G156A; clones 24, 31, 42, 48) were identified and had titers 9 × 10² to 6 × 10⁴ G418^r colony-forming units (CFU)/ml. High titer producer clones (61.19 and 33.5) used to infect primary murine BM cells were generated by ping-pong with GP + envAm12 and had titers of >1 × 10⁵ (Moritz et al., 1995). All producer lines were maintained in Dulbecco's modified Eagle's medium (GIBCO-BRL) supplemented with 10% calf serum (Summit Biotechnology, Ft. Collins, CO), 2% (v/v) penicillin-streptomycin (GIBCO-BRL).

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Schematic diagram of WT, P140A, and P140A/G156A retroviral vectors constructed in MSCV2.1. Relevant restriction sites are noted. Proviral size is 3.4 kb. Expression of the MGMT cDNA is via the 5'-LTR (arrow). MSCV2.1 contains a neomycin (neo)-selectable marker cloned under the control of the phosphoglycerate kinase (Pgk) promoter.

Retroviral Transduction of Cell Lines and Primary BM. L1210 cells (American Type Culture Collection, Rockville, MD) were maintained in 1630 RPMI supplemented with 15% calf serum (Summit Biotechnology), 2% penicillin-streptomycin, and 1% glutamine (GIBCO-BRL). L1210 cells were infected by incubating 1.5 × 10⁶ cells with 2 ml of filtered virus supernatant overnight at 37°C on plates precoated with 8 µg/cm² of fibronectin (FN) fragment CH296 (RetroNectin; Takara Shuzo, Biotechnology Group, Otsu, Japan). G418-resistant L1210 clones were generated after infection by plating 1 × 10³ cells/ml in 0.6% agar (Difco Bacto-agar; Difco Laboratories, Detroit, MI) in the presence of 1 mg/ml G418 for 7 days at 37°C. Individual G418-resistant L1210 colonies were harvested with a drawn Pasteur pipette with an inverted microscope, transferred to 96-well plates, and amplified for additional analysis. No mock-transduced cells survived G418 concentrations greater than 0.75 mg/ml after 7 days. B16 melanoma and Lewis lung (LL) carcinoma cells (American Type Culture Collection), maintained in Dulbecco's modified Eagle's medium (GIBCO-BRL) supplemented with 10% calf serum (Summit Biotechnology) and 2% penicillin-streptomycin (GIBCO-BRL), were infected in the presence of 8 µg/ml Polybrene (Aldrich Chemical Co, Milwaukee, WI) without FN CH296. Twenty-four hours after infection, cells were washed in PBS (GIBCO-BRL), trypsinized at 37°C, and plated in media supplemented with 1 mg/ml G418 (dry powder; GIBCO-BRL) for 7 days at 37°C. Primary murine BM cells were transduced in the presence of FN CH296. Briefly, BM was harvested from the hind limbs of 8- to 10-week-old female C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) 48 h after i.p. injection with 5-fluorouracil (150 mg/kg b.wt.; SoloPak Laboratories, Franklin Park, IL). These cells were prestimulated for 48 h at 37°C in 5% CO₂ with 100 U/ml recombinant human interleukin-6 (Pepro Tech Inc., Rock Hill, NJ), and 100 ng/ml recombinant rat stem cell factor (Amgen, Thousand Oaks, CA) in -modified Eagle's medium (GIBCO-BRL), supplemented with 20% fetal calf serum (Summit Biotechnology). Prestimulated BM cells were transduced by incubating 5 × 10⁶ cells with 2 ml of filtered virus supernatant overnight at 37°C on plates precoated with 8 µg/cm² FN CH296, as previously described (Hanenberg et al., 1996).

Analysis of MGMT Expression and Protein. Expression of retroviral vector-derived MGMT was determined by Northern and Western blot analysis and an O⁶-guanine repair assay. RNA was isolated and purified from infected and G418-selected L1210 clones with TriPure isolation reagent (Boehringer Mannheim) and resuspended in diethylpyrocarbonate (Sigma, St. Louis, MO)-treated water. Six micrograms of total RNA was electrophoresed through a 0.9% agarose gel containing 6% formaldehyde and 1× 3-(N-morpholino)propanesulfonic acid. RNA was subsequently transferred to nitrocellulose filters, hybridized to a ³²P-end-labeled EcoRI-XhoI WTMGMT cDNA fragment derived from MSCV2.1WTMGMT, and exposed to X-ray film at 70°C. Filters were probed with a ³²P-end-labeled actin probe to ensure equivalent loading of RNA.

6-BG Inactivation of MGMT Activity and BCNU Treatment. Resistance of expressed MGMT protein to 6-BG was determined in L1210 clones by incubating 5 × 10⁶ cells with 0, 5, 10, 20, or 40 µM 6-BG (kindly provided by Dr. R Moschel, Frederick Cancer Research Center, Frederick, MD) for 1 h at 37°C and assaying each sample for residual MGMT O⁶-methylguanine DNA repair activity. Resistance to 6-BG depletion of MGMT was also determined by incubating 2 × 10⁵ L1210 cells or 1 × 10⁶ transduced primary murine BM cells with 20 µM 6-BG for 1 h at 37°C, followed by subsequent exposure of treated cells to 0, 25, 50, or 75 µM BCNU (Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment, National Cancer Institute, Bethesda, MD) for 1 h at 37°C. Treated L1210 cells were centrifuged, washed twice with media, resuspended, and plated in agar as described above. Treated primary murine BM cells were centrifuged, washed two times with media, and resuspended; and 5 × 10⁴ cells/ml were plated in a 0.6% agar supplemented with 100 ng/ml recombinant rat stem cell factor and 50 ng/ml recombinant murine granulocyte, monocyte-colony stimulating factor. Cultures were incubated at 37°C in a humidified environment at 5% O₂ and 10% CO₂ for 7 days. BCNU survival was determined by dividing the number of colonies surviving 6-BG and BCNU treatment at each dose by the number of colonies scored in untreated plates and multiplying by 100.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Retroviral-Mediated Expression of 6-BG-Resistant MGMT Mutants. Expression of 6-BG-resistant MGMT mutants was determined in individual clones of L1210 cells, a murine hematopoietic cell line, after transduction with either WTMGMT, P140A or P140A/G156A vectors (Fig. 1). L1210 cells were used as targets for retroviral transduction because they lack endogenous MGMT and are extremely sensitive to BCNU treatment. Northern blot analysis of total cellular RNA isolated from a WTMGMT-transduced clone (clone 14; lanes 4), four P140A-transduced clones (19.2, 19.7, 43.2, and 43.4; lanes 5-8), and four P140A/G156A-transduced clones (clones 24.42, 31.31, 42.7, and 48.11; lanes 9-12) demonstrated similar levels of viral MGMT RNA transcripts of the expected 3.4-kb size (Fig. 2). WTMGMT-transduced clone 3 also demonstrated similar levels of RNA compared with the other clones (clone 3; lane 3). However, a smaller transcript was observed. RNA transcripts will vary in size because the MSCV vector contains a splice donor and an acceptor site 3' of the 5'-long terminal repeat (LTR), which allows for alternatively spliced RNA species (Hawley et al. 1994). No MGMT-hybridizing RNA bands were noted in uninfected L1210 (Fig. 2, lane 1) or in G418-resistant L1210 cells infected with an empty MSCV2.1 virus (1 × 10⁶ G418^r CFU/ml titer on NIH/3T3 cells) used as a control (Fig. 2, lane 2). The same blot was reprobed with actin to demonstrate similar RNA loading (Fig. 2, lanes 1-12, lower panel).

View larger version (66K): [in this window] [in a new window]   Fig. 2.   Northern blot analysis of MGMT transgene expression in L1210 clones. Six micrograms of total RNA was loaded, electrophoresed, blotted, and probed as described in Materials and Methods. Labeling above each lane represents clone numbers. Lane 1, uninfected L1210 cells; lane 2, L1210 cells infected with an empty MSCV2.1 virus; lanes 3 and 4, WT clones 3 and 14, lanes 5-8, P140A clones 19.2, 19.7, 43.2, and 43.4; lanes 9-12, P140A/G156A clones 24.42, 31.31, 42.7, and 48.11. The lower panel shows the same blot probed with actin to demonstrate equivalent RNA levels in each lane.

Protein Content and Activity of 6-BG-Resistant MGMT Mutants. To examine the content of MGMT protein in infected and G418-resistant L1210 cells, we analyzed L1210 clones by immunoblotting with a monoclonal antibody that recognizes the 21-kDa human MGMT (MT3.1) (Fig. 3). Similar levels of human MGMT protein were detected in clones expressing either WTMGMT (Fig. 3, lane 1) or P140A (Fig. 3, lanes 2-5). However, the amount of P140A/G156A protein was reduced in P140A/G156A-transduced L1210 clones compared with WTMGMT and P140A (Fig. 3, lanes 6-9). Densitometric analysis of the ratio of MGMT protein with an endogenous protein, equally present in every lane, demonstrated a ratio of 1.9 for WTMGMT-3 and 1.3 ± 0.26 for P140A-transduced L1210 clones. In contrast, these same ratios were 0.55 ± 0.10 for L1210 clones expressing the P140A/G156A protein. In addition, MGMT protein activity was analyzed with an O⁶-methylguanine DNA repair activity assay as previously described (Wu et al., 1987). As shown in Fig. 4, MGMT protein activity was low or absent in P140A/G156A-transduced L1210 clones (Fig. 4, lanes 9-12) compared with cells infected with WTMGMT (lanes 3 and 4) or P140A (lanes 5-8). Little to no P140A/G156A O⁶-methylguanine DNA repair activity was observed in cell extracts from any of the P140A/G156A-transduced clones, even after assaying 20-fold more protein (data not shown). In contrast, L1210 clones expressing the P140A single mutant demonstrated levels of repair activity to similar to those of WTMGMT-transduced clones. As expected, neither L1210 nor L1210 cells infected with an empty MSCV2.1 vector demonstrated any detectable O⁶-methylguanine DNA repair activity (Fig. 3, lanes 1 and 2).

View larger version (47K): [in this window] [in a new window]   Fig. 3.   Western blot analysis of MGMT protein in L1210 clones. Twenty micrograms of total protein was loaded in each lane, electrophoresed, and subjected to Western blot analysis. Arrow, the 21-kDa MGMT protein. Labeling above each lane represents clone numbers. Lane 1, WT clone 3; lanes 2-5, P140A clones 19.2, 19.7, 43.2, and 43.4l lanes 6-9, P140A/G156A clones 24.42, 31.31, 42.7, and 48.11.

View larger version (57K): [in this window] [in a new window]   Fig. 4.   O6-methylguanine DNA repair activity in L1210 clones. Repair activity was assayed by the conversion of the 18-bp band to the 8-bp band. The level of activity correlates with the ratio of these bands. Labeling above each lane represents clone numbers. Lanes 3 and 4, WT clones 3 and 14; lanes 58, P140A clones 19.2, 19.7, 43.2, and 43.4; lanes 9-12, P140A/G156A clones 24.42, 31.31, 42.7, and 48.11. Uninfected L1210 (lane 1) and L1210 cells infected with an empty MSCV2.1 virus (lane 2) were used as negative controls.

Activity of the P140A/G156A Mutant Form of MGMT in Other Mammalian Cells. To determine whether the reduced amount of P140A/G156A protein and O⁶-methylguanine DNA repair activity in L1210 clones was specific for hematopoietic cells, B16 melanoma and LL carcinoma cells, which also lack endogenous MGMT activity, were infected with retroviral supernatant, selected in G418, and analyzed for MGMT repair activity. Consistent with the results observed in L1210 cells, no O⁶-methylguanine DNA repair activity was detectable in either P140A/G156A-transduced B16 or LL cells (Fig. 5, lanes 4 and 8). In contrast, P140A-transduced B16 and LL cells (Fig. 5, lanes 3 and 7) demonstrated levels of repair activity similar to those of WTMGMT-transduced cells (Fig. 5, lanes 2 and 6).

View larger version (52K): [in this window] [in a new window]   Fig. 5.   O6-Methylguanine DNA repair activity in cell lines. B16 melanoma (B16) (lanes 1-4) and Lewis Lung (LL) cells (lanes 5 and 6) were infected and selected in G418 as described in Material and Methods, and analyzed for repair activity. Lane 1, uninfected B16 cells; lane 2, WT-transduced B16 cells; lane 3, P140A-transduced B16 cells; lane 4, P140A/G156A-transduced B16 cells; lane 5, uninfected LL cells; lane 6, MSCV2.1 WTMGMT-transduced LL cells; lane 7, P140A-transduced LL cells; lane 8, P140A/G156A-transduced LL cells.

P140A-Transduced L1210 Clones and Primary Murine BM Cells Are Resistant to 6-BG Depletion and BCNU Treatment. To examine the resistance of P140A-transduced L1210 cells to 6-BG inactivation of MGMT, we analyzed the level of MGMT repair activity with and without pretreatment with 6-BG. In these experiments, L1210 cells transduced with either WTMGMT or P140A were exposed to 5 to 40 µM 6-BG for 1 h. Figure 6 shows that two separate clones of L1210 cells transduced with WTMGMT (clone 3, Fig. 6A, and clone 14, Fig. 6B) could be depleted of MGMT repair activity in a dose-dependent fashion. A demonstratable reduction in repair activity (i.e., the lack of appearance of the 8-base pair band) was observed after exposure to 5 µM 6-BG. No repair activity was apparent after exposure to 10 µM or higher concentrations of 6-BG (Fig. 6, A and B). In contrast, all four L1210 clones (19.2, 43.2, 19.7, and 43.4) transduced with P140A that were analyzed maintained MGMT repair activity even at 20 to 40 µM 6-BG (Fig. 6, A and B). We next analyzed the ability of retroviral-mediated P140A MGMT expression in these cells to protect against 6-BG sensitization to BCNU. After treatment with 20 µM 6-BG, cells were exposed to increasing concentrations of BCNU, and survival was determined by using a clonogenic assay. All four of the P140A-transduced L1210 clones tested (Fig. 7B) were significantly more resistant to the combination of 6-BG and BCNU than clones expressing either WTMGMT (Fig. 7A) or P140A/G156A (data not shown). The IC₅₀ for BCNU in P140A-transduced L1210 clones (Fig. 7B) averaged 21.2 µM after 6-BG treatment compared with 7.5 µM for WTMGMT-transduced L1210 clones (Fig. 7A), and these mutants continued to show resistance even after treatment with 50 µM BCNU. Similar to transduced L1210 cells, primary murine hematopoietic cells expressing the P140A mutant form of MGMT were significantly more resistant (p < .05) to 6-BG inactivation of MGMT and sensitization to BCNU than mock- or WTMGMT-transduced cells (Table 1). Table 1 shows both the percentage survival for murine CFU-granulocyte macrophage (CFU-GM) and the IC₅₀ for BCNU derived from mock-, WTMGMT-, or P140A-transduced BM with and without 6-BG pretreatment. WTMGMT-transduced CFU-GM demonstrated only 20% survival and an IC₅₀ for BCNU of 27.5 µM after treatment with 20 µM 6-BG and 40 µM BCNU. In contrast, P140A-transduced CFU-GM demonstrated 77.0% survival and an IC₅₀ for BCNU of 50.0 µM.

View larger version (69K): [in this window] [in a new window]   Fig. 6.   6-BG depletion of O6-methylguanine DNA repair activity in L1210 clones. L1210 cells infected with either WT (clone 3) or P140A (clones 19.2 and 43.2) shown in A, were treated with 5, 10, 20, or 40 µM 6-BG as described in Material and Methods and analyzed for MGMT repair activity. B shows L1210 cells infected with either WT (clone 14) or P140A (clones 19.7 and 43.4) which were treated in a similar manner. Labeling above each lane represents clone number and 6-BG concentration.

View larger version (27K): [in this window] [in a new window]   Fig. 7.   Retroviral-mediated expression of P140A protects L1210 clones in vitro against 6-BG and BCNU treatment. L1210 cells were infected with either WT (clones 3 and 4) or P140A (clones 19.2, 19.7, 43.2, and 43.4) or empty retrovirus (MSCV) and analyzed for 6-BG inactivation of MGMT and BCNU resistance in vitro. The data are plotted as percentage survival versus BCNU dose and represent the averages from three independent experiments.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

CENUs are commonly used alkylating agents with moderate activity in brain tumor therapies. CENUs are cytotoxic to both hematopoietic stem and progenitor cells, and intensive use of these agents in humans leads to cumulative bone marrow toxicity, delayed myelosuppression, pancytopenia, and immune suppression (Schabel, 1976; Botnick et al., 1978; Neben et al., 1993; Kay et al., 1995; Maze et al., 1997). Our laboratory and other investigators have previously demonstrated the ability to generate blood cells resistant to CENUs by using retroviral-mediated gene therapy. Transduction of both murine and human hematopoietic cells with a retroviral vector encoding the human MGMT cDNA protects BM cells in vitro and in vivo against BCNU-induced cytotoxicity (Moritz et al., 1995; Allay et al., 1996, 1997; Jelinek et al., 1996; Maze et al., 1996, 1997; Davis et al., 1997). We previously reported that mice reconstituted with WTMGMT-transduced stem cells demonstrated significantly high survival after a myelosuppressive regimen of five weekly doses of 40 mg/kg BCNU as compared with control transplanted animals.

High levels of MGMT activity have been demonstrated in multiple human tumors and cell lines (Pegg et al., 1995; Dolan and Pegg, 1997). 6-BG, which binds to and inactivates MGMT, has been shown to deplete tumor cell MGMT repair activity and to sensitize human tumors to CENU treatment both in vitro and in vivo (Futscher et al., 1989; Dolan et al., 1990a,b, 1991; Mitchell et al., 1992; Gerson et al., 1993; Marathi et al., 1994; Phillips et al., 1997). In this regard, we have recently demonstrated that continuous exposure to 6-BG results in a more prolonged inactivation of xenograft tumor MGMT repair activity in vivo compared with bolus 6-BG infusions (C. Kurpad, unpublished observations). Such prolonged inactivation of MGMT may be a critical component of tumor cell kill in vivo, because regeneration of repair activity would likely lead to reacquisition of the resistance phenotype.

Genetic approaches to expressing a 6-BG-resistant form of MGMT in hematopoietic cells provide a unique way to protect blood cells in vivo from CENU toxicity in the setting of pharmacological manipulation of primary resistant tumors. 6-BG will potentiate CENU-induced hematopoietic cytotoxicity in vivo (Gerson et al., 1996; Davis et al., 1997). Harris et al. (1995) have analyzed retroviral-mediated expression of ada, the bacterial homolog of MGMT, in murine hematopoietic cells. Compared with mammalian MGMT, this particular homolog is naturally resistant to 6-BG due to differences in the primary and secondary structure surrounding the adduct-binding site (Goodtzova et al., 1997). Retroviral expression of ada in murine hematopoietic cells after gene transfer was associated with only modest protection against 6-BG and BCNU treatment in vivo (Harris et al., 1995). Concerns regarding the expression of ada in mammalian cells, such as nuclear targeting and its ability to possibly evoke an immune response in vivo, have prompted our laboratory and others to pursue the use of 6-BG-resistant human MGMT mutants.

Crone et al. (1994) previously studied the ability to generate 6-BG-resistant human MGMT protein in O⁶-guanine repair-deficient bacterial cells by introducing specific amino acid substitutions around the active site that mimic ada. Several mutations provided significant resistance against 6-BG inactivation of MGMT. Cells expressing single mutations P140A or G156A were shown to be >20- and 240-fold more resistant to inactivation by 6-BG compared with WTMGMT. In addition, the expression of a double mutant form of MGMT containing both the P140A and G156A mutations (P140A/G156A) demonstrated 1200-fold more resistance to 6-BG.

In this report, we demonstrated that retroviral-mediated expression of P140A/G156A in transduced L1210 clones did not protect against the combination of 6-BG and BCNU. Molecular analysis of these clones demonstrated that, despite having similar levels of MGMT RNA compared with transduced L1210 cells expressing WTMGMT or the P140A mutant, P140A/G156A-transduced L1210 clones expressed less protein and little to no O⁶-methylguanine DNA repair activity. Hickson et al. (1996) have previously demonstrated that P140A/G156A protected cells against 6-BG inactivation of MGMT and sensitization to mitozolomide. Interestingly, they also observed a greater than 10-fold decrease in P140A/G156A repair activity compared with WTMGMT protein in both bacterial and Chinese hamster lung fibroblast cells. The ability of P140A/G156A to protect bacterial and Chinese hamster lung fibroblast cells, despite the reduction in repair activity and stability, may be due in part to the level of expression, because in these studies P140A/G156A was expressed via the human cytomegalovirus promoter, which has very strong transcriptional activity in cell lines compared with an MSCV2.1 5'-LTR promoter. Our results extend these observations to hematopoietic cells, where expression via a variety of viral promoters has been problematic. Taken together with the data previously reported, our data suggest that the P140A/G156A double mutant may be unstable in mammalian cells. Thus, this instability may be a major impediment to its use in modulating resistance to 6-BG in hematopoietic cells in vivo.

We have also demonstrated that retroviral-mediated expression of the P140A mutant form of MGMT in L1210 clones leads to similar RNA, protein, and O⁶-methylguanine DNA repair activity levels compared with WTMGMT. Expression of P140A resulted in a 4-fold increase in resistance to 6-BG depletion of O⁶-methylguanine DNA repair activity, which correlated with protection against the combination of 6-BG and BCNU in vitro. As shown in Fig. 7, L1210 clones demonstrated a 2.7-fold increase in the IC₅₀ value for BCNU after 6-BG treatment, compared with WTMGMT. The expression and activity of P140A in L1210 clones were consistent with its activity in other mammalian cell types, such as B16 melanoma, LL cells (Fig. 5), and Chinese hamster ovary cells (Loktionova and Pegg, 1996). In addition, we report that gene transfer of P140A into primary murine hematopoietic progenitor cells protected against 6-BG depletion of MGMT and BCNU treatment. Expression of P140A correlated with a BCNU IC₅₀ of 48 µM, after treatment with 20 µM 6-BG. As shown in Table 1, we observed greater than 70% survival of murine hematopoietic progenitor cells after treatment with 20 µM 6-BG and 40 µM BCNU. These results are similar to results obtained with the G156A MGMT mutant by Davis et al. (1997), who reported a 50% clonogenic survival with 25 µM 6-BG and 40 µM BCNU after gene transfer into murine hematopoietic cells. Although, the increase in 6-BG resistance (compared with WTMGMT) is statistically significant, the level of this resistance will need to be examined in vivo to determine whether it is sufficient to prevent depletion of activity after pharmacological manipulation with 6-BG. These experiments are currently under way in our laboratory in a mouse model.

These results demonstrate that retroviral-mediated expression of the 6-BG-resistant P140A mutant form of MGMT in hematopoietic cells protects against in vitro 6-BG depletion of MGMT and sensitization to BCNU. We are currently testing the ability of P140A to protect murine hematopoietic cells against 6-BG and BCNU in vivo after gene transfer. In addition, Pegg and coworkers (Xu-Welliver et al., 1998) have recently demonstrated that other amino acid substitutions at position 140, such as P140K, confer significantly greater resistance to 6-BG compared with mutations P140A, G156A, or P140A/G156A. The median inhibitory dose (ID₅₀) for 6-BG against these 6-BG-resistant mutant forms of MGMT varies from 60 to 300 µM, as opposed to 0.25 µM for WTMGMT (Dolan et al., 1990a,b). These doses are well above the concentrations required to achieve in vivo biochemical modulation of MGMT expressed in human tumor cell xenografts. If more resistance to 6-BG is necessary, mutants such as P140K, which is more than 100 times more resistant than P140A, may be more logical to pursue than double mutants, which are likely to give problems with expression level and poor activity due to lack of stability or mis-folding.

These mutants may also be of importance with respect to the possible use of MGMT as a dominant selectable gene. We have previously reported that mice transplanted with MGMT-transduced hematopoietic stem cells demonstrated an enrichment of transduced cells in the lymphoid compartment after multiple doses of 40 mg/kg BCNU in vivo (Maze et al., 1997). Likewise, Allay et al. (1997), using WTMGMT, and Davis et al. (1997), using G156A, have observed an enrichment of transduced cells in the myeloid compartment in vivo after gene transfer of MGMT and BCNU treatment. Thus, MGMT may potentially be used to enrich for hematopoietic stem cells transduced with a retroviral vector containing a second nonselectable therapeutic gene, such as -globin, in a nonablative setting, as suggested by Davis et al. (1997). Expression of a significantly greater 6-BG-resistant human MGMT mutant, such as P140K, may prove to be more advantageous in minimizing the BCNU dose necessary for enrichment after 6-BG treatment, thus reducing any possible BCNU-related systemic toxicity.

In summary, this report demonstrates that retroviral-mediated expression of P140A protects hematopoietic cells against cytotoxic doses of BCNU after 6-BG depletion and provides an approach to protect blood cells whereas treating CENU-resistant tumors.