The disclosure additionally sets forth methods of dominant selection of a second transduced gene in hematopoietic cells wherein a first transduced gene expressing an AGT mutant resistant to an 06-benzylated guanine derivative is co-cultured in the presence of the 06-benzylated guanine derivative and an alkylating or methylating agent. A marked enhancement of clonogenic survival of CD34+ cells transduced with the transgene expressing G156A AGT in the presence of BG and BCNU shows that any AGT mutant showing similar activity will be useful as a drug resistance gene.

Alkylating and methylating agents are important

groups of compound for use in cancer chemotherapy.

Chloroethylating agents, including but not limited to chloroethylnitrosoureas, form DNA adducts within the cell nucleus that promote alterations in DNA structure and/or function. These changes at the DNA level lead to cytotoxicity within the targeted cell. For example, the class of chemotherapeutic agents known as nitrosoureas alkylate guanine residues at the 06 position, forming lesions which become cytotoxic if left unrepaired (Pegg <BR> <BR> <BR> AE. Mammalian 06 Alkylguanine-DNA Alkyltransferase: Regulation and Importance in Response to Alkylating <BR> <BR> <BR> Carcinogenic and Therapeutic Agents, Cancer Res., 50: 6119-6129,1990; Gerson, S. L., and Willson, J. K., o6- alkylguanine-DNA alkyltransferase. A target for the <BR> <BR> <BR> modulation of drug resistance, Hematol. Oncol. Clin.

North. Am., 9: 1995).431-50, Methylation damage from agents such as MNU are believed to result in a process called abortive mismatch repair, forming multiple double strand breaks which cannot be repaire properly, leading to cytotoxicity. (Karran, P., Macpherson, P., Ceccotti, S., Dogliotti, E., Griffin, S., and Bignami, M., o6- methylguanine residues elicit DNA repair synthesis by human cell extracts, J. Biol. Chem., 286: 15878-15886, 1 9 9 3). T h e n i t r o s o u r e a, B C N U (N, N'-bis (2-cholorethyl)-N-nitrosourea), donates a chloroethyl lesion on 06 of guanine, which rapidly rearranges its structure to form interstrand DNA crosslinks between the modifie guanine and the antiparallel strand cytosine. (Tong, W. P., Kirk, M. C., and Ludlum, D. B., Formation of the cross link 1- (N3- deoxycyctidyl), 2- (N1-deoxyguanosinyl)-ethane in DNA treated with N, N'-bis (2-chloroethyl)-N-nitrosourea, Cancer Res., 42: 1982).3102-3105, Chloroethylnitrosoureas such as BCNU (N, N'-bis (2-cholorethyl)-N-nitrosourea) and CCNU <BR> <BR> <BR> (N- (2-cholorethyl)-3-cyclohexyl-N-nitrosourea) are lipid soluble compound that have been shown to possess

clinical utility against some neoplasms but with limited success in clinical trials. These chloroethylnitrosoureas promote DNA alkylation at the 06 position of guanine, leading to DNA interstrand crosslinking and altered fidelity of DNA replication and transcription. This induced interstrand crosslinking involves formation of a chloroethyl adduct at the guanine residue that undergoes an intramolecular rearrangement to produce an unstable intermediate that reacts with the cross strand cytosine residue. The result is a N'-guanine, N 3-cytosine-ethanol crosslink.

This N'-guanine, N 3-cytosine-ethanol crosslink can be prevented by the DNA repair protein, 06-alkylguanine-DNA alkyltransferase (AGT). This repair protein is active in mammalian tumor cells and is able to directly remove the lesion from the DNA through an irreversible covalent interaction, a"suicide"process which inactivates the transfer activity of the protein.

(Pegg AE. Mammalian 06Alkylguanine-DNA Alkyltransferase: Regulation and Importance in Response to Alkylating Carcinogenic and Therapeutic Agents, Cancer Res., 50: 6119-6129,1990). This repair protein is also responsible for protecting cells from the antitumor effects normally associated with chloroethylating agents such as BCNU and CCNU. AGT has a unique mechanism of action in that it brings about the transfer of alkyl groups present on the 06 position of guanine in DNA to a cystine residue located within the AGT amino acid sequence (Lindahl, et al., 1988, Annu. Rev. <BR> <BR> <BR> <P>Biochem. :57 133-157; Pegg, 1990, Cancer Res. :50 6119-6119). The endogenous expression of AGT is tissue specific and we found that AGT is expressed at the lowest levels in human and murine bone marrow. (Gerson, S. L., Miller, K., and Berger, N. A. 06 alkylguanine-DNA alkyltransferase activity in human myeloid cells, J.

Clin. Invest., 76: 1985).2106-14, The low expression of AGT in bone marrow explains the observation that

myelosuppression is the doselimiting toxicity of nitrosourea chemotherapy. Furthermore, we have demonstrated that over expression of AGT in transgenic mice (Dumenco, L. L., Allay, E., Norton, K., and Gerson, S. L. The prevention of thymic lymphomas in transgenic mice by human 06-alkylguanine-DNA alkyltransferase, <BR> <BR> <BR> Science, :259 219-22,1993), results in increased protection from nitrosourea mediated carcinogenesis.

These data suggest the therapeutic potential of MGMT gene therapy to mediate protection from the myelosuppressive effects of alkylating agent chemotherapy.

U. S. Patent No. 5,091,430. issued to Moschel, et al. on February 24,1992 discloses 06-substituted guanine compound which inhibit AGT. An exemplified compound which inhibits AGT is 06-benzylguanine (BG). It has been shown that BG is a strong time and concentration dependent inactivator of human AGT (Dolan, et al,, 1990, Proc. Natl. Acad. Sci. :87 686-690). This mechanism has been confirme by the identification of S-benzylcysteine in AGT and the formation of stoichiometric amounts of guanine following incubation with BG.

U. S. Patent No. 5,352,669 issued to Moschel, et al. on October 4,1994 discloses 06-substituted guanosine and deoxyguanosine compound which are shown to inhibit AGT.

U. S. Patent No. 5,358,952 issued to Moschel, et al. on October 22,1994 disclose pharmaceutical combinations for chemotherapy comprising 06-substituted guanine compound in tandem with anti-neoplastic alkylating agents.

A major limitation in the use of alkylating and methylating agents in the treatment of neoplastic disease is the profond myelosuppression produced by these drugs. This problem peaks at about 4-6 weeks after treatment and thus prevents the repetition of cyclic therapy at preferred intervals. This myelosuppression is due to low concentrations of AGT in hematopoietic cells.

Crone and Pegg (1993, Cancer Res. 53: 4750-4753) disclose a mutant human AGT protein with a single amino acid change of proline to alanine at aa #140 (P140A).

This mutant human AGT shows a decrease in sensitivity to BG in vitro. Crone and Pegg do not address in vivo activity of this mutant in regard to BG or BG/BCNU combinations.

Crone, et al. (1994, Cancer Res. 53: 4750-4753) disclose an additional mutant human AGT protein with a single amino acid change of glycine to alanine at aa #156 (G156A). As with human AGT P140A, G156A shows a decrease in sensitivity to BG in vitro. Again, Crone, et al. do not address in vivo activity of this mutant in regard to BG or BG/BCNU combinations.

Gerson, et al. (1994, Mutation Res. 307: 541-555) disclose transgenic mice expressing either the wild type human 06-methylguanine-DNA methyltransferase (MGMT) cDNA or the bacterial ada gene (which expresses a prolcaryotic version of AGT resistant to BG). The authors show a level of protection against addition of BCNU in transformed cell types shown to express the respective gene.

Retroviral transduction and expression of MGMT has been shown to provide cellular resistance to the DNA-crosslinking agent BCNU in murine hematopoietic progenitors (Allay, J. A., Dumenco, L. L., Koc, O. N., Liu, L., and Gerson, S. L., Retroviral transduction and expression of the human alkyltransferase cDNA provides nitrosourea resistance to hematopoietic cells, Flood, 85: 3342-3351,1995; Moritz, T., Mackay, W., Glassner, B. J., Williams, D. A., and Samson L., Retrovirus-mediated expression of a DNA repair protein in bone marrow protects hematopoietic cells from nitrosourea-induced <BR> <BR> <BR> toxicity in vitro and in vivo, Cancer Res., 55: 2608- 2614,1995), and human CD34'cells (Allay, J. A., Koc, O. N., Davis, B. M., and Gerson, S. L., Retroviral-mediated gene transduction of the human alkyltransferase cDNA

confers nitrosourea resistance to human hematopoietic <BR> <BR> <BR> <BR> progenitors, Clinical Cancer Res., 2: 1996),1353-1359, and reduce myelosuppression (Moritz, T., Mackay, W., Glassner, B. J., Williams, D. A., and Samson L., Retrovirus-mediated expression of a DNA repair protein in bone marrow protects hematopoietic cells from nitrosourea-induced toxicity in vitro and in vivo, <BR> <BR> <BR> <BR> Cancer Res. , :55 2608-2614,1995). However, the survival avantage to BCNU treatment in MGMT transduced cells was only two-fold that of mock-transduced controls (Allay, J. A., Dumenco, L. L., Koc, O. N., Liu, L., and Gerson, S. L., Retroviral transduction and expression of the human alkyltransferase cDNA provides nitrosourea resistance to hematopoietic cells, Blood, :85 3342-3351, 1995), severely limiting its therapeutic potential.

This fact has led to the study of a mutant AGT, which is resistant to inactivation by the AGT inhibitor 06-benzylguanine (BG). This protein, which contains a glycine to alanine substitution at amino acid 156 (G156A), was designed to mimic the region near the active site of the bacterial AGT homolog, Ada, which has been shown to be resistant to BG inactivation (Crone, T. M., Goodtzova, K., Edara, S., and Pegg, A. E., Mutations in human 06-alkylguanine-DNA alkyltransferase <BR> <BR> <BR> <BR> imparting resistance to 06-benzylguanine, Cancer Rets., 54: 6221-7,1994). This mutation alone was sufficient to render the protein 240-fold more resistant to inactivation than wild type AGT in cell free extracts (Crone, T. M., Goodtzova, K., Edara, S., and Pegg, A. E.

Mutations in human 06-alkylguanine-DNA alkyltransferase imparting resistance to 06-benzylguanine, Cancer Rets. 54: 6221-7,1994).

Although AGT P140A and G156A show resistance to BG and no apparent defect in the ability to repair 06-methylguanine in DNA in vitro, the rate of repair of methylated DNA is very rapid and is difficult to measure

accurately under physiological conditions. Also, it is not known whether the ability to act on the larger 2-chloroethyl group is affecte by these mutations.

Furthermore, some point mutations in AGT have a pronounced destabilizing effect and may reduce the steady state level of the AGT protein. These factors couid limit the ability of the mutant AGTs to protect cells from chloroethylation. Therefore, it is possible that some or all of these mutations do also affect the ability to repair 0'- (2-chloroethyl) guanine in cellular DNA and would therefore not produce resistance to sensitization by BG.

We have previously shown that retroviral transduction of G156A MGMT (MGMT) into human CD34+ cells provides increased resistance to combination BG and BCNU treatment (Reese, J. S., Kob, 0. N., Lee, K., Liu, L., Allay, J. A., Phillips, W. P., and Gerson, S. L., Retroviral transduction of a mutant MGMT into human CD34 cells confers resistance to 06-benzylguanine plus BCNU, Proc. Natl. Acad. Sci. USA, 93: 1996).14088-14093, In the present application, we have designed a murine model system to demonstrate in vivo resistance to BG and BCNU, as well as to demonstrate the ability to enrich for cells with high AAGT expression using in vivo BG and BCNU drug administration. Based upon the greater degree of drug resistance observe in AMGMT transduced human CD34+ cells compare to untransduced cells, a greater degree of in vivo selection for transduced cells is possible using aMGMT with BG and BCNU.

A major limitation of previous drug resistance gene transfer studies using MDR1, wtMGMT and DHFR is the inability to enrich for transduced progenitors while selecting against tumor cells. In the case of wtMGMT, there is a lack of a differential survival avantage for wtMGMT transduced cells over tumor cells expressing high levels of AGT. In fact, the expected result after transfer of these genes would simply be to render the

bone marrow as drug resistant as the tumor cells. Our <BR> <BR> <BR> current strategy using AMGMT should increase the differential survival avantage of transduced bone marrow cells versus tumor, since treatment with BG will deplete the wtAGT activity of the tumor, making the tumor sensitive to BCNU, while bone marrow cells expressing NAGE would be resistant to BG mediated AGT inactivation and would retain the ability to repair DNA damaged by BCNU.

Enrichment for transduced cells may be necessary since current protocols for retroviral mediated gene transfer into hematopoietic cells result in inefficient transgene expression in vivo, thereby limiting the potential therapeutic benefit of the procedure. Low level enrichment for cells with high proviral expression <BR> <BR> <BR> of the multiple drug resistance gene (MDR1) has been<BR> <BR> <BR> <BR> <BR> reporte by Sorrentino et. al. (Sorrentino, B. P., Brandt, S. J., Bodine, D., Gottesman, M., Pastan, I., Cline, A., and Nienhuis, A. W., Selection of drug- resistant bone marrow cells in vivo after retroviral transfer of human MDR1, Science, 257: 1992),99-103, and <BR> <BR> <BR> Podda et, a1. (Podda, S., Ward, M., Himelstein, A., Richardson, C., de la Flor-Weiss, E., Smith, L., Gottesman, M., Pastan, I., and Bank, A., Transfer and expression of the human multiple drug resistance gene into live mice. , Proc. Natl. Acad. Sci. USA, 89: 9676, 1992), by in vivo paclitaxel administration. <BR> <BR> <BR> <P>Additionally, Allay et al has demonstrated enrichment of cells with high wtMGMT expression in mice transplanted with wtMGMT transduced bone marrow progenitors after in vivo BCNU administration (Allay, J. A., Davis, B. M., and Gerson, S. L. In-vivo enrichment of MGMT transduced murine hematopoietic progenitors by BCNU treatment of MGMT transplanted mice, in press, Exp. Hem. , 1997).

SUMMARY OF THE INVENTION The present invention overcomes a major limitation in the use of methylating and alkylating agents in the treatment of neoplastic disease by addressing the problem of extreme myelosuppression produced by these drugs. Therefore, this specificatiori discloses gene therapy treatments to improve chemotherapeutic treatments of a neoplastic disease.

The present invention relates to methods of optimizing a chemotherapeutic regime whereby a compound of the regime selectively inhibits a tumor cell-localized wild type protein while a mutant version of this protein, resistant to the compound, is simultaneously provided in hematopoietic cells. This mutated version of the protein also exhibits DNA repair activity. These methods require transducing a population of hematopoietic cells with a transgenic construction expressing the mutant protein and reintroducing the transduced cells into the mammalian host so as to promote expression of the mutant protein in bone marrow cells during the respective chemotherapy regime.

To this end, a portion of the invention relates to use of these gene therapy treatments to improve a chemotherapeutic regime wherein O6-benzylguanine (BG) is added to render tumor cells containing high levels of O6-alkylguanine-DNA alkyltransferase (AGT) sensitive to chloroethylating agents. In this application a mutated nucleic acid sequence is generated which expresses an AGT DNA repair protein resistant to BG while retaining the ability to repair DNA adducts formed after exposure to various chloroethylating agents. The transduction and expression of such an AGT mutant gene in hematopoietic cells followed by reintroduction of the transduced cells into the mammalian host will result in a relevant cell population which both shows an elevated level of the mutant AGT and in turn is not sensitive to the presence

of BG as a chemotherapeutic agent. In other words, expression of the mutant AGT protein overcomes the effect of a chloroethylating agent not only through elevated concentrations of the protein but also due to resistance to BG. In contrast, BG inhibits tumor cell localized wild type AGT, preventing repair of DNA adducts generated in the tumor cell genome through the action of a respective chloroethylating agent.

A further aspect of the present invention demonstrates the ability of in vivo administration of BG <BR> <BR> <BR> <BR> and BCNU to select for high level ^MGMT expression in transduced murine hematopoietic progenitors which were transplanted into lethally irradiated mice. The present invention demonstrates a survival avantage following combination BG and BCNU treatment for mice transplanted <BR> <BR> <BR> <BR> with ßMGMT transduced bone marrow progenitors compare to control animals transplanted with mock transduced cells.

An additional aspect of the invention is directe <BR> <BR> <BR> <BR> to demonstrating the infusion of AMGMT transduced mouse bone marrow cells into a mouse which is then treated with a chemotherapy combination of BG plus BCNU, allows for selection in vivo of the transduced cells and their maintenance for many weeks with an increased selection in favor of the transduced cells with each round of chemotherapy thus resulting in replacement of the bone marrow by transduced cells.

A still further aspect of the present invention <BR> <BR> <BR> <BR> relates to the demonstration that infusion of AMGMT transduced cells into a nude mouse which is then inoculated with human tumor cells which are allowed to grow because of the immunodeficiency state of the mouse, followed by treatment of the mouse with repeated cycles of the chemotherapy agents BG plus BNCU, results in in vivo selection of the transduced bone marrow cells.

These and other objects of the invention will be more full understood from the following description of

the invention, the figures and the claims appende hereto.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1. BG and BCNU resistance in murine <BR> <BR> <BR> hematopoietic progenitors transduced with oMGMT (n),<BR> <BR> <BR> <BR> <BR> wtMGMT (a) or lacz (o). Cells were treated with 0 (A), 10 (B) or 25 (C) UM BG for 1 h at 37 C prior to 0-40 yM BCNU for 2 h at 37 C. Bars, SD. P-values determined by paired t-test of the mean BCNU Icso and comparisons at each dose of BCNU.

Figure 2. Design of BG and BCNU mediated enrichment expriment in mice transplanted with AMGMT or lacZ transduced bone marrow progenitors. Arrows represent passage of time; syringes represent dates of BG and BCNU treatment. Mice were sacrifice 13 or 23 weeks post transplantation and analyzed as shown.

Figure 3A-Figure 3E. FACS analysis of QAGT expression in bone marrow from mice transplanted with QMGMT or lacZ transduced progenitors 13 weeks post transplant. A, Comparison of expression in lacZ (dotted line), unselected QMGMT (solid line) and BG and BCNU selected QMGMT (bold line) bone marrow cells obtained from transplanted mice. B, Overlay of lacZ and unselected QMGMT cells. C, Resulting histogram following subtraction analysis demonstrating expression in 30k of the unselected QMGMT cell population. D, Overlay of lacZ and BG and BCNU selected AMGMT cells.

E, Resulting histogram following subtraction analysis demonstrating expression in 60t of the selected AMGMT cell population.

Figure 4. BG and BCNU resistance in bone marrow derived CFU (A) after transduction and before transplantation, (B) from transplanted mice sacrifice 13 weeks post transplant (lacez, n=5; unselected QMGMT, n=3; selected nMGMT, n=5) and (C) from transplanted mice sacrifice 23 weeks post transplant (selected aMGMT, n=4; untransduced n=2). Bone marrow cells were

treated with 20 UM BG for 1 h at 37 C followed by 0-80 UM BCNU for 2 h at 37 C, then plated in triplicate in methylcellulose plates and CFU growth was enumerated.

Curves represent survival of lacZ transduced or untransduced CFU (o), unselected QMGMT transduced CFU (Q) and selected AMGMT transduced CFU (2). Bars, SE.

Figure 5. Kaplan-Meyer survival of mice transplanted with QMGMT (n=22) or lacZ (n=13) transduced progenitors after i. p. BG and BCNU administration. Open symbols represent mice censored from study due to <BR> <BR> <BR> scheduled sacrifice and intermediate analysis; Q, QMGMT mice;, lacez mice. P<0.0001, Mann-Whitney U.

Figure 6 shows Western blot analysis of AGT protein expression in CHO cells. Extracts from CHO cells expressing human AGT (lanes 1,2) or its mutants P140A (lanes 3,4) and G156A (lanes 5,6) and extracts from HT29 cells (lanes 7,8) were resolved by SDS-PAGE, transferred to nitrocellulose and developed using antibodies to a peptide corresponding to amino acids 8-20 of the human AGT. Untransfected CHO cells (lane 9) showed no AGT protein of this size. (Lanes 1,3,5-25 yg of protein loaded; lanes 2,4,6,7-50 yg; lanes 8, 9-100 Ug).

Figure 7 shows the effect of different concentrations of BCNU on killing of CHO cells. Control CHO cells (A) and CHO cells expressing human AGT (o) or its mutants P140A (O) and G156A (z) were treated with BCNU at the concentration shown for 2 h. The medium was then replace. After 16-18 h, the cells were replated as described in Example Four and colonies were counted 7-8 days later.

Figure 8A and Figure 8B show a loss of AGT activity after exposure to different concentrations of BG (Figure 8A) and 5-nitroso-BP (Figure 8B). CHO cells expressing human AGT (o) and its mutants P140A (o) and G156A (D) were exposed to BG or 5-nitroso-BP for 4 h. The cells were then harvested and the AGT activity determined.

Figure 9 shows the effectof BG on cell growth.

Cells of a CHO clone expressing G156A AGT were seeded at 105 per 75 cm2 flask. After 48 h the medium was replace with fresh medium containing different concentrations of BG as shown. The cells were maintained in the presence ot BG for 24 h. After this time, the medium was replace every 24 h. Cells were harvested and counted at time points indicated.

Figure 10 shows the effect of different concentrations of BG and 5-nitroso-BP on colony formation. BG was added to CHO cells expressing mutant AGTs P140 (*) and G156A (D) and 5-nitroso-BP was added to cells expressing G156A () for 18-20 h. The cells were then replated as described in materials and methods and colonies were counted 7-8 days later.

Figure 11A and Figure 11B show the effect of different concentrations of BG (Figure 11A) and 5-nitroso-BP (Figure llB) on killing of cells by BCNU.

Cells were treated with BG or 5-nitroso-BP at the concentrations indicated for 2 h and BCNU at 80 yM was added for 2 h. Medium was replace with fresh medium also containing the AGTinhibitor but not BCNU for the period of 18 h. The cells were then replated as described in materials and methods and colonies were counted 7-8 days later.

Figure 12A and Figure 12B show the effect of differing concentrations of BG on AGT activity for wtMGMT (AGT) and aMGMT for transduced CHO cells (Figure 12A) and transduced CD34* cells (Figure 12B).

Figure 13A and Figure 13B show the effect of 10ym BG and differing concentrations of BCNU on activity of the G156A mutant (nAGT) (), wild type AGT (wtMGMT) (o) and untransduced control for K562 cells (Figure 13A) and CD34'cells (Figure 13B).

Definitions BCNU-N, N'-bis (2-chloroethyl)-N-nitrosourea <BR> <BR> <BR> CCNU-N- (2-chloroethyl)-3-cyclohexyl-N-nitrosourea AGT-O6-alkylguanine-DNA alkyltransferase BG-O6-benzylguanine MPSV-myeloproliferative sarcoma virus MoMLV-moloney murine leukemia virus SDS-PAGE-SDS polyacrylamide gel electrophoresis MGMT-O6-methylguanine-DNA methyltransferase As used herein, the term"MGMT"denotes a gene or cDNA construct which encodes 06-alkylguanine-DNA alkyltransferase (AGT). The term"MGMT"may be used interchangeably with"AGT"to denote such a nucleic acid sequence expressing a wild type or mutant version of the human AGT protein.

As used herein, the term"patient"includes members of the animal kingdom including but not limited to human beings.

As used herein, the term"mammalian host"includes members of the animal kingdom including but not limited to human beings.

DETAILED DESCRIPTION OF THE INVENTION The present invention overcomes a major limitation in the use of alkylating and methylating agents in the treatment of neoplastic disease by addressing the problem of extreme myelosuppression produced by these drugs. The present invention overcomes this limitation by providing for methods of treating neoplastic disease whereby gene therapy methodology is combine with known chemotherapeutic regimes to optimize host tumor responsiveness to the anti-neoplastic agents administered to the patient. The basis of the present invention is grounded in using gene therapy techniques in conjunction with chemotherapy regimes. This

combination provides for a host environment whereby tumor cells are optimally sensitized to anti-neoplastic agents while transduced myeloid cells remain substantially unaffected.

There is a preferred scenario whereby the methods of the present invention will be particularly useful. It will be advantageous in certain cancer treatments to inactivate a specific tumor cell protein which is also expressed in non-malignant cell types. This will be accomplished by adding a compound to the chemotherapy regime known to inhibit a particular function of the protein. It follows that inactivating this biological function in non-malignant cell types may be deleterious to successful treatment of the patient. Therefore, a mutant protein shown to possess substantially wild type activity as well as imparting resistance to the inhibiting compound will be subcloned into a recombinant vector, transduced into a population of non-malignant cell types, such as hematopoietic stem and progenitor cells, and introduced back into the patient. The end result will be expression of the sensitive, wild type form of the protein in tumor cells as compare to a mutant version being expressed in hematopoietic cells.

This mutant version will retain wild type function as well as being insensitive to the inhibitory compound.

This combination of gene therapy and chemotherapy applications will provide maximum sensitivity of the tumor cells to anti-neoplastic agents while substantially decreasing myelosuppression within the patient.

In the following examples section, both in vivo and in vitro showings are made with respect to the above identifie utilities of the present invention.

Example One demonstrates the ability of in vivo administration of BG and BCNU to select for high level AMGMT expression in transduced murine hematopoietic progenitors which were transplanted into lethally

irradiated mice. Thus Example One will demonstrate a survival avantage following BG and BCNU treatment for mice transplanted with aMGMT transduced bone marrow progenitors compare to control animals transplanted with mock transduced cells.

Example Two is directe to demonstrating the <BR> <BR> <BR> infusion of AMGMT transduced mouse bone marrow cells into a mouse which is then treated with a chemotherapy combination of BG plus BCNU, allows for selection in vivo of the transduced cells and their maintenance for many weeks with an increased selection in favor of the transduced cells with each round of chemotherapy thus resulting in replacement of the bone marrow by transduced cells.

Example Three relates to the demonstration that <BR> <BR> <BR> infusion of AMGMT transduced cells into a nude mouse which is then inoculated with human tumor cells which are allowed to grow because of the immunodeficiency state of the mouse, followed by treatment of the mouse with repeated cycles of the chemotherapy agents BG plus BNCU, results in in vivo selection of the transduced bone marrow cells.

Example Four exemplify the use of human AGT sequence mutated at aa#140 (proline to alanine) and #156 (glycine to alanine).

Example Five exemplifies targeting of hematopoietic cells, namely CD34+ cells enriched from peripheral blood. Gene therapy based targeting of hematopoietic cells will be effective in overcoming myelosuppression associated with treatment of these anti-neoplastic drugs. However, this specification teaches the usefulness of targeting any non-malignant cell type, by either ex vivo or in vivo based methods, which will support biologically active concentrations of a human AGT mutant of the present invention.

To this end, a portion of the invention relates to use of gene therapy treatments to improve a

chemotherapeutic regime wherein the tumor protein targeted for modification is 06-alkylguanine-DNA alkyltransferase (AGT). It is known that the presence of AGT, a DNA repair protein active in mammalian tumor cells, imparts resistance to alkylating agents. The mechanism of action of AGT involves rection with the 06-position of guanine residues in DNA, the target for DNA adduct formation of many alkylating agents. It has been advantageous to inactivate tumor localized AGT so as to sensitize tumor cells to the alkylating agents.

Tumor cells have been successfully sensitized by adding 06-benzylated guanine derivatives, such as BG, to the chemotherapy regime (see Pegg, et al., 1995, Prog. Nucl.

Acids :51 167-223). These compound have been shown to be effective inactivators of AGT and inclusion in a chemotherapy mix enhances cytotoxic properties of chloroethylating and methylating anti-tumor drugs (see, <BR> <BR> <BR> e. g., Chae, et al., 1994, J. Med. Chem. :35 4486-4491).

Regardless, addition of BG or a related O5-benzylated guanine derivative does not address the problem of cytotoxicity within various host non-malignant cell types. In fact, inhibiting AGT in non-malignant cell types will exacerbate cytotoxicity in non-malignant cells expressing useful concentrations of AGT. An overriding problem of using alkylating agents in chemotherapy regimes has been severe myelosuppression caused by low levels of AGT expression. This problem will persist with or without addition ot 06-benzylated guanine derivatives.

Therefore, one aspect of the present invention relates to the use of a mutant AGT protein, preferably a human AGT protein, resistant to inactivation by a compound shown to inactivate or at least substantially to inactivate the DNA repair function of wild type AGT.

By way of example, and certainly not of limitation, compound which may be utilized are 06-benzylated guanine derivatives such as 8-aza-06-benzylguanine

(8-aza-BG); 06-benzyl-8-bromoguanine (8-bromo-BG); 2-amino-4-benzyloxy-5-nitropyrimidine (4- desamino-5-nitro-BP) 06-benzylguanine (BG) ; o'- [p- (hydroxymethyl) benzyll guanine (HN-BG); 06-benzyl-8- methylguanine (8-methyl-BG); 06-benzyl-7,8-dihydro-8- o x o g u a n i n e (8-o x o-B G) ; 2,4,5,-triamino-6-benzyloxyprimidine (5-amino-BP) ; 06-benzyl-9- [ (3-oxo-5 a-androstan-l7ß- yloxycarbonyl) methyl] guanine (DHT-BG); 06-benzyl-9- (3- oxo-4-androsten-17i-yloxycarbonyl) methyl] guanine (AND-BG); and, 8-amino-06-benzylguanine (8-amino-BG). ln addition, pyrimidine compound, including but not limited to 2,4-diamino-6-benzyloxy-5-nitrosopyrimidine (5-nitroso-BP) and 2,4-diamino-6- benzyloxy-5-nitropyrimidine (5-nitro-BP) may be utilized in conjunction with a preferred AGT mutant.

Another aspect of the present invention relates to the use of a mutant AGT protein, preferably a human AGT protein, resistant to inactivation by a compound shown to inactivate or at least substantially inactivate the DNA repair function of wild type AGT which inclue, but again are in no way limited to the 06-substituted guanosine and 2'deoxyguanisine compound disclosed in U. S. Patent No. 5,352,669 issued to Moschel, et al. on October 4,1994. As an example, but again not as a limitation, A dBG, the 2'-deoxyribonucleoside of BG. In an in vitro screen of potential AGT-inhibiting BG analogs, dBG was found to be ten-fold less potent than BG and BGs modifie at the benzyl ring, but it was among the most active of the relatively soluble AGT inhibitors. The relatively high potency of dBG combine with its superior solubility in aqueous media compare <BR> <BR> <BR> to other BGs prompte us to test in an in vivo xenograft system. Despite a difference in in vitro potency between BG and dBG, the BCNU-potentiating effects of the two compound were quite comparable. Furthermore, escalation of the dBG dose was not restricted by solubility as with

BG. Therefore, this class of compound, as exemplified by dBG, will be useful in practicing the present invention whereby the skilled artisan identifies a related AGT mutant which imparts resistance to addition of dBG both in vitro and in vivo.

Example Four exemplifies the use of human AGT sequence mutated at aa#140 (proline to alanine) and #156 (glycine to alanine). However, the skilled artisan will be fully aware that any mutation, whether it be a substitution mutant, deletion mutant or addition mutant, will be useful in the present invention as long as any such mutant shows resistance to the compound which inhibits wild type AGT and also retains DNA repair activity. It will be within the purview of an artisan of ordinary skill to generate additional mutants, and test such mutants with the aid of this specification, for use in practicing the disclosed invention. The combine use of such a mutant in gene therapy/chemotherapy applications will allow practically unfettered use of the compound of interest in chemotherapy regimes to sensitize target tumor cells while imparting both resistance and wild type AGT activity to the transduced hematopoietic cells and subsequent hematopoietic cells introduced back to the patient.

Therefore, in a specific example, hematotoxicity will be overcome by use of gene therapy to express an alkyltransferase gene in the hematopoietic cells and/or hematopoietic stem or progenitor cells. A mutant human AGT resistant to inactivation by 06-benzylated guanine derivatives and pyrimidine compound, for example, will have considerable avantages for this purpose. The expression from a strong promoter of the resistant form of AGT in the patient's marrow will improve the therapeutic index of the treatment both by increasing the alkyltransferase activity in the critical hematopoietic cells and in rendering this AGT insensitive to inactivation by the modifying agent

whereas the tumor AGT would still be sensitive to the respective compound.

It follows that another aspect of the present invention relates to utilizing the resistance of cells expressing a mutant AGT, as exemplified in Example Four, to select a population of hematopoietic progenitor cells expressing high levels of a stable AGT mutant for use in the disclosed gene therapy protocols. This selection procedure will be based on selecting cultured cells expressing high levels of at least one mutant form of AGT in the presence of (1) the compound or compound which inhibit wild type activity but are inactive against the mutant, and (2) the alkylating agent of choice which will be at least one of the agents used in the planned chemotherapeutic regime.

It may be incumbent upon the skilled artisan to generate and test an AGT mutant resistant to a particular compound. As noted above, any such additional mutant is within the scope of the present invention and may be tested in conjunction with the data presented in Example Four. Although a mutant AGT resistant to a compound, such as an 06-benzylated guanine derivative or a pyrimidine compound, may show no apparent defect in the ability to repair an 06-methylguanine in DNA in vitro, the rate of repair of methylated DNA is very rapid and is difficult to measure accurately under physiological conditions. It may not be evident as to whether the ability to act on the larger 2-chloroethyl group is affecte by the respective mutation.

Furthermore, some point mutations in AGT have a pronounced destabilizing effect and may reduce the steady state level of the AGT protein (see. e. g. Irone, et al., 1994, Cancer Res. 54: 6221-6227; Ling-Ling, et al.. 1992. Carcinogenesis 13: 837-843; Pieper, et al.

1994, Carcinogenesis 15: 1898-1902). Any or all of these factors could limit the ability of the respective mutant AGTs to protect cells from chloroethylation. It

is possible that any such mutation derived by in vitro means may still not affect the ability to repair 0'- (2-chloroethyl) guanine in cellular DNA and would therefore not produce resistance to sensitization by a 06-benzylated guanine derivative or a pyrimidine compound. Therefore, the assays exemplified in Example Four will be necessary to predict the fidelity of a mutant for future transduction into hematopoietic progenitor and stem cells and reintroduction into the patient.

The present invention additionally relates to the use of any available alkylating agent, such as chloroethylating agents, or other known compound which may form one or more types of DNA adducts from which may be reverse by wild type AGT or a respective mutant AGT protein which substantially retains the wild type DNA repair function. In other words, the present invention contemplates the use, alone or in any combination recognized by the skilled practitioner, of one or more alkylating agents which are alkylated DNA at the 06-position in guanine. Examples of such alkylating agents, provided for the purpose of example and by no means of limitation, are the chloronitrosoureas N, N'-bis (2-cholorethyl)-N-nitrosourea (BCNU) and N- (2-cholorethyl)-3-cyclohexyl-N-nitrosourea (CCNU).

These compound may be utilized in combination with any exemplified or disclosed inhibitor of AGT and with any exemplified or disclosed transgene in a combination so as to sensitize tumor cells to the these alkylating agents while rendering transduced cells resistant to the inhibitor.

The present invention further relates to the use of any available methylating agent or other similar compound which may form one or more types of DNA adducts from which may be reverse by wild type AGT or a respective mutant AGT protein which substantially retains the wild type DNA repair function. As noted in

the previous paragraph for use of alkylating agents, the present invention contemplates the use, alone or in any combination recognized by the skilled practitioner, of one or more methylating agents which form a DNA adduct at the 0'-position in guanine. Examples of such methylating agents, provided for the purpose of example and by no means of limitation, are temozolomide, decarbazine, procarbazine, and streptozotocin. These compound may be utilized in combination with any exemplified or disclosed inhibitor of AGT and with any exemplified or disclosed transgene in a combination so as to sensitize tumor cells to the alkylating agents while rendering transduced cells resistant to the inhibitor.

The 06-benzylguanine derivatives employed in the present invention may be made into pharmaceutical compositions by accommodation with appropriate pharmaceutically acceptable excipients or carriers or diluent, and may be formulated into preparations in solid, semi-solid, liquid or gaseous forms such as tables, capsules, powders, granules, ointments, solutions, suppositories, injections, inhalant, and aerosols in the usual ways for their respective route of administration. The following methods and excipients are merely exemplary and are in no way limiting.

In pnarmaceutical dosage forms, the 06-benzylguanine derivatives employed in the present invention may be used in the form of their pharmaceutically acceptable salts, and also may be used alone or in appropriate association, as well as in combination with other pharmaceutically active compound.

In the case of oral preparations, the 06-benzylguanine derivatives of the present invention may be used alone or in combination with appropriate additives to make tables, powders, granules, or capsules. e. g. with the conventional additives such as lactose, mannitol, corn starch or potato starch; with

binders such as crystalline cellulose, cellulose derivatives, acacia, corn starch or gelatins; with disintegrators such as corn starch, potato starch or sodium carboxymethylcellulose; with lubricants such as talc or magnesium stearate; and, if desired, with diluents, buffering agents, moistening agents, preservatives, and flavoring agents.

Furthermore, the 06-benzylguanine derivatives employed in the present invention may be made into suppositories by mixing with a variety of bases such as emulsifying bases or water-soluble bases.

The 06-benzylguanline derivatives employed in the present invention may be formulated into preparations for injections by dissolving, suspending or emulsifying them in an aqueous or non-aqueous solvents such as vegetable oil, synthetic aliphatic acid glycerides, esters of higher aliphatic acids or propylene glycol; and if desired, with conventional additives such as solubilizers, isotonic agents, suspending agents, emulsifying agents, stabilizers and preservatives.

In the cases of inhalations or aerosol preparations, the O6-benzylguanine derivatives employed in the present invention in the form of a liquid or minute powder may be filled up in an aerosol container with gas or liquid spraying agents, and, if desired, together with conventional adjuvants such as humidifying agents. They may also be formulated as pharmaceuticals for non-pressured preparations such as via a nebulizer or an atomizer.

The amount of 06-benzylguanine derivatives employed in the present invention to be used varies according to the degree of the effective amount required for treating tumor cells. A suitable dosage is that which will result in a concentration of the 06-benzylguanine derivatives in the tumor cells to be treated which results in the depletion of AGT activity, e. g. about 1-2000 mg./kg prior to chemotherapy and preferably 10-800 mg/kg prior to

chemotherapy. In fact, a basis for the present invention is the use of gene therapy applications in conjunction with the addition of BG such that higher doses of BG may be added without the documente deleterious effects on non-malignant cell types.

Unit dosage forms for oral administration such as syrups, elixirs, and suspensions wherein each dosage unit (e. g. teaspoonful or tablespoonful) contains a predetermined amount of the 06-benzylguanine derivative employed in the present invention can be combine with a pharmaceutically acceptable carrier, such as sterile water for injection, USP, or by normal saline.

The 06-benzylguanine derivatives employed in the present invention can be administered rectally via a suppository. The suppository can include vehicles such as cocoa butter, carbowaxes and polyethylene glycols, which melt at body temperature, yet are solidifie at room temperature.

The 06-benzylguanine derivatives employed in the present invention can be administered transdermally in an appropriate vehicle or salt or converted to a salt.

Adsorption may be aided by the use of an electric current or field.

The 06-benzylguanine derivatives employed in the present invention may be administered with an appropriate vehicle for buccal or sublingual administration.

The O6-benzylguanine derivatives employed in the present invention can be utilized in aerosol formulations to be administered via inhalation. The 06-benzylguanine derivatives can be formulated into pressurized acceptable propellants such as dichlorodifluoromethane, propane, nitrogen and the like.

The term"unit dosage formas used herein generally refers to physically discrete units suitable as unitary dosages for human and animal subjects, each unit containing a predetermined quantity of the

O6-benzylguanine derivatives calculated in an amount sufficient to produce the desired effect in association with a pharmaceutically acceptable diluent, carrier, excipient or vehicle. The specifications for the novel unit dosage forms of the present invention depend on the particular compound employed and the effect to be achieved, and the pharmacodynamics associated with each compound in the host.

The pharmaceutically acceptable excipients, for example, vehicles, adjuvants, carriers or diluents are readily available to the public.

Any necessary adjustments in dose can be readily made to meet the chemotherapeutic treatment requirements and adjusted accordingly by the skilled practitioner.

A preferred mode of BG delivery is via a PEG 400/saline solution. This delivery vehicle has been studied in detail, is practically non-toxic, has been given to humans orally and is present in commercial drug formulations such as lorazepam for injection. The use of a PEG 400/saline solution has been shown to be an appropriate vehicle for delivery of BG (Dolan., et al.

1994. Cancer Chemother.. Pharmacol. 35: 121-126).

This specification teaches combinational gene/chemotherapy. Combinations of dosage and timing and intervals introduction of transduced cells to optimize treatment of a patient will be required and are within the scope of the present invention.

These compound may be administered using conventional techniques such as those described in Wasserman, et al., 1975, Cancer 36: ;1258-1268 and Physician's Desk Reference, 45th ed., 1991, Edward R.

Barnhart (publisher). For example, BCNU may be administered intravenously at a dosage from about 150- 200 mg/m2 every six weeks. CCNU may be administered orally at a dosage of about 130 mg/m2 every six weeks.

Other additional alkylating or methylating agents may be administered in appropriate dosages via appropriate

routes of administration known to the skilled artisan.

A specific embodiment of the present invention relates to using a mutated nucleic acid sequence which expresses an AGT protein, preferably a human protein, which is resistant to BG while still retaining wild type or near wild type ability to repair DNA adducts formed after exposure to various chloroethylating agents. The transduction and expression of such an AGT mutant gene in hematopoietic cells followed by reintroduction of the transduced cells into the mammalian host will result in an in vivo cell population which shows both an elevated level of the mutant AGT and in turn is not sensitive to the presence of BG as a chemotherapeutic agent. In vivo expression of a BG-resistant AGT mutant protein will, as disclosed above, overcome the effect of a chloroethylating agent not only through elevated concentrations of the protein but also due to resistance to BG.

In a specific embodiment of the present invention, gene therapy treatments disclosed within this specification are utilized in tandem with a chemotherapeutic regime comprising BG and the c h 1 o r o e t h y 1 a t i n g a g e n t N, N'-bis (2-cholorethyl)-N-nitrosourea (BCNU).

It will be advantageous to select an AGT mutant exhibiting both optimal resistance to the 06-benzylated guanine derivatives or a pyrimidine compound of choice as well as retaining wild type or substantially wild type function. This selection can occur not only in culture CHO (Chinese Hamster Ovary) cells as shown in Example Four but will also continue during infection of hematopoietic cells. It will be advantageous and is forwarded as a portion of the present invention to select recombinant AGT retroviral clones expressing high levels of the mutant AGT of interest. This may be accomplished by selecting for positive clones during the initial c-culture with producer cells in the presence

of an alkylating agent and/or the 06-benzylated guanine derivatives or a pyrimidine compound of choice.

Additionally or in combination with the latter, selection may occur during c-culture of the retroviral supernatant with hematopoietic cells. Only recombinant retroviral AGT mutants expressed at adequate levels will impart resistance in culture to the 06-benzylated guanine derivatives or a pyrimidine compound while exhibiting wild type activity. It will be a recombinant AGT retroviral vector with such characteristics which will lead to a greater therapeutic index when reintroduced into the patient in conjunction with anti-neoplastic alkylating or methylating agents.

The following examples are meant for illustrative purposes and are in no way intended to limit the scope of the invention.

EXAMPLES EXAMPLE ONE-MATERIALS AND METHODS MFG vector and producer cells The zMGMT cDNA was cloned into the 3MFG retroviral vector (provided by Dr. J. Barranger, University of Pittsburgh, Pittsburgh, Pennsylvania) as described elsewhere (Reese, J. S., Kob, 0. N., Lee, K., Liu, L., Allay, J. A., Phillips, W. P., and Gerson, S. L., Retroviral transduction of a mutant MGMT into human CD34 cells confers resistance to 06-benzylguanine plus BCNU, Proc. Natl. Acad. Sci. USA, 93: 1996).14088-14093, Producer cell lines were created as described elsewhere (Reese, J. S., Kob, 0. N., Lee, K., Liu, L., Allay, J.

A., Phillips, W. P., and Gerson, S. L., Retroviral transduction of a mutant MGMT into human CD34 cells confers resistance to 06-benzylguanine plus BCNU. Proc. <BR> <BR> <BR> <P>Natl. Acad. Sci. USA, 93: 1996).14088-14093, PMFGAMGMT was transfected into GP + E86 and GP + envAml2 cell lines (provided by Dr. A. Bank, Columbia University, NY, NY), using a modified"ping-pong"to increase viral

titer. Retroviral producer clones were generated and the titer was determined as described elsewhere (Reese, J. S., Kob, 0. N., Lee, K., Liu, L., Allay, J. A., Phillips, W. P., and Gerson, S. L. Retroviral transduction of a mutant MGMT into human CD34 cells confers resistance to 06-benzylguanine plus BCNU. Froc.

Natl. Acad. Sci. USA, 93: 1996).14088-14093, The titer was determined by analysis of expression using AAGT specific immunohistochemical stains on cells exposed to limiting dilutions of supernatant collecte from the retroviral producer cell cultures. The titer of the producer line used in these studies was 5 x 105 infectious particles/mL of supernatant.

Replication Comptent Retrovirus Detection Assay The producer cell lines and bone marrow cells obtained from mice 23 weeks after transplantation were assayed for replication comptent retrovirus (RCR). The producers or bone marrow cells were cultured in medium containing heat inactivated serum for 48 hours. The media was collecte, passed through a 0.45yM filter and overlayed upon cultures of NIH3T3 cells that had been <BR> <BR> <BR> previously transduced with the lacZ gene (Allay, J. A., Dumenco, L. L., Koc, O. N., Liu, L., and Gerson, S. L., Retroviral transduction and expression of the human alkyltransferase cDNA provides nitrosourea resistance to <BR> <BR> <BR> <BR> hematopoietic cells, Blood, 85: 1995).3342-3351, These cells were grown for 24 hours in fresh medium containing heat inactivated serum, then the medium was filtered and overlayed upon cultures of NIH3T3 cells (repeated for 4 consecutive days). None of the target NIH3T3 cells were <BR> <BR> <BR> positive for lacZ expression after X-gal staining in the proviral reactivation assay, which has a limit of detection of 1 in 105 viral particles/mL supernatant.

Additionally, MFG-aMGMT specific PCR performed on <BR> <BR> <BR> the NIH3T3 lacZ cells cultured in both producer and bone marrow supernatant was negative for amplification (the

limit of detection is 1 in 104_ viral particles/mL).

Similarly, there was no amplification observe following <BR> <BR> <BR> lacz specific PCR performed on the NIH3T3 cells, coinciding with the results of the X-gal stain.

Transduction protocol Bone marrow progenitors were obtained from the femur and tibia of 8 week old male C3H/HeN mice (harles River, Wilmington, MA) 48 hours after treatment with 150 mg/kg 5-fluorouracil (Pharmacia, Kalamazoo, MI). The cells were prestimulated at 5 x 105 cells/mL for 24 hours in a-MEM plus 100 U/mL mIL-3 (enzyme, Cambridge, MA), 100 U/mL mIL-6 (enzyme), and 100 ng/mL rSCF (Amgen, Thousand Oaks, CA), then co-cultured for 48 hours with <BR> <BR> <BR> <BR> Aml2 MFG-MGMT producer cells rendered replication defective by treatment with 5yg/mL of Mitomycin C. The co-cultures were performed using the above cytokines plus 6 yg/mL protamine sulfate in a-MEM supplemented with 20t heat inactivated fetal calf serum, again at a <BR> <BR> <BR> <BR> density of 5 x 105 cells/mL. The non-adherent hematopoietic cells were then separated from the adherent producers, washed and resuspended in serum free a-MEM at 5 x 106 cells/mL, and used for transplantation.

Approximately 1 x 106 cells were left untransplanted and treated with 20yM BG and 0-40yM BCNU, then plated in methylcellulose to determine CFU survival.

Transplantation and druq administration Recipient 8-10 week old male C3H/HeN mice were <BR> <BR> <BR> lethally irradiated with 1040 cGy using a 61 Co source and transplanted with 1 x 106bone marrow cells via tail vein injection. At the time of drug injection, 30 mg/kg of BG was injecte intraperitoneally in a solution of 40t v/v PEG 400 vehicle and 60% v/v 0. 05M PBS. One hour after BG administration, 10,15, or 20 mg/kg BCNU was injecte intraperitoneally in a solution of 10k v/v ethanol and 90k v/v 0.05 M PBS. The BCNU was

solubilized in 100k ethanol before PBS dilution and was used within 10 minutes of reconstitution. BCNU and Sentry Grade Union Carbide PEG-400 were obtained from the Developmental Therapeutics Branch, National Cancer Institute (Bethesda, MD). BG was synthesized by Dr.

Robert Moschel at the Frederick Cancer Research Institut (Fredericksburg, MD). Mice were kept in microisolator cages and given water supplemented with bacitracin and neomycin from 4 days before lethal irradiation to 3 weeks after the last drug administration.

CFU assai Bone marrow cells were resuspended in serum free a-MEM supplemented with pokeweed mitogen spleen cell conditioned medium, and incubated with 0,10,20, or 25yM of BG for 1 hour followed by 0-80 yM BCNU for 2 hours at 37°C with gentle rocking. Cells were washed free of drug and resuspended in serum free a-MEM, plated in triplicate in methylcellulose plus 100 ng/mL rSCF (Amgen), 100 U/mL mIL-3 (enzyme), 40 U/mL hEPO (Amgen), pokeweed mitogen spleen cell conditioned medium, O. lmM Hemin and 0 or 10yM BG, and cultured for 7 days at 37°C and 5W CO2. Total counts of CFU colonies greater than 50 cells were enumerated and survival differences between samples were analyzed by paired t-test of the mean BCNU IC50 and comparisons at each dose of BCNU. The transduction efficiency was determined by PCR using proviral specific primers on genomic DNA isolated from individual CFU-C.

PCR for provirus detection To obtain genomic DNA for PCR analysis, individual colonies were resuspended in 30yL of water and boiled for 8 min. One yL of a 10mg/mL Proteinase K solution was added and the colonies were incubated at 55°C for 2 hrs then boiled again for 8 min. Ten yL of the

preparation was used for each PCR rection. Proviral specific primers (5'TGGTACCTCACCCTTACCGAGTC and 3 CGGGGAACTCTTCGATAGCCT) were used to amplify a 443 bp fragment and mouse ß-globin primers (5t GAAGTTGGGTGCTTGGAGAC and 3'GAGACTGCTCCCTAGAATCG) which amplify a 400 bp fragment were used as an interna control to verify the presence of DNA. Samples were run on a 1.8k agarose gel and visualized by Ethidium Bromide staining.

AGT activity sassa AGT activity was measured as previously described (Gerson, S. L., Miller, K., and Berger, N. A., Ol alkylguanine-DNA alkyltransferase activity in human myeloid cells, J. Clin. Invest.,76 : 2106-14,1985).

Briefly, enzyme activity was measured as [3H]-methyl groups removed from [3H] methylnitrosourea (specific activity of 0.047 fmol 06 MeG/g DNA). The alkylated [3H1_ O6-MeG and N7-methylguanine bases were separated by high performance liquid chromatography and quantitated by liquid scintillation. N7-methylguanine was used as the interna standard. The DNA content of each sample was determined by staining with Hoescht dye and measuring fluorescence at 365nm in a TK-100 Hoeffer fluorometer. AGT activity was expressed as fmol O6-MeG removed/yg DNA.

FACS analysis for AGT expression Red blood cells were lysed from whole bone marrow or peripheral blood using 60mM NH4Cl, 10mM KHCO3, and O. OlmM tetrasodium EDTA at 37°C. The cells were washed with 2t PBS and fixed in 1k Paraformaldehyde for 30 min on ice, then permeabilized in 1% Tween-20 for 30 min at 37°C. The samples were blocked with 10% Normal Goat Serum for 30 min at room temp, resuspended in 2% PBS and incubated overnight at 4°C with llig of the AGT monoclonal antibody mT3.1 (provided by Drs. D. Bigner

and T. Brent) or IgGk isotype. The samples were incubated with Goat anti-mouse antibody conjugated to PE (Pharmingen, San Diego, CA) for 1 hr on ice. Flow cytometry was performed using the Becton-Dickinson FACSorter and 10,000 events were analyzed for each sample.

Experimental desian for enrichment expriment As depicted in Figure 2,12 lethally irradiated <BR> <BR> <BR> C3H/HeN mice were transplanted with AMGMT transduced bone marrow progenitors and 9 were transplanted with <BR> <BR> <BR> lacs transduced bone marrow progenitors. The mice were treated with 30mg/kg BG and 10mg/kg BCU or left untreated at 3 and 6 weeks post-transplant. The mice were divided into 6 cohorts: <BR> <BR> <BR> <BR> 6 oMGMT mice were treated 3 and 6 weeks post-transplant<BR> <BR> <BR> <BR> <BR> 3 aMGMT mice were treated 6 weeks post-transplant<BR> <BR> <BR> <BR> <BR> 3 oMGMT mice were left untreated<BR> <BR> <BR> <BR> <BR> <BR> 4 lacez mice were treated 6 weeks post-transplant; only 2 survive to the completion of the expriment <BR> <BR> <BR> 3 lacez mice were left untreated.<BR> <BR> <BR> <BR> <BR> <P> All but four of the AMGMT animals were sacrificed 13 weeks post-transplant and the bone marrow, peripheral blood and spleen were obtained. The remainder of the animals were sacrifice at 23 weeks post transplant.

For each animal, the bone marrow was analyzed for NAGE expression by FACS, AGT activity, CFU survival after BG and BCNU, and the percentage of CFU containing the provirus. The peripheral blood was analyzed for AAGT expression by FACS.

EXAMPLE ONE-RESULTS <BR> <BR> <BR> <BR> AMGMT transduced CFU-C are selectively protected from BG and BCNU Bone marrow derived hematopoietic progenitors obtained from 5-FU treated mice were transduced with <BR> <BR> <BR> either aMGMT, wtMGMT or lacZ by c-culture with

retroviral producer cells for 48 hours, then collecte and exposed to 0-25yM BG for 1 hour followed by 0-40yM BCNU for 2 hours. Survival curves derived from CFU-GM and BFU-E colony growth are shown in Figure 1. BG treatment markedly potentiated BCNU toxicity in both <BR> <BR> <BR> lacZ and wtMGMT transduced cells, but AMGMT transduction conferred protection from combination drug treatment. <BR> <BR> <BR> <P>Following BG treatment, the BCNU Icso in AMGMT transduced<BR> <BR> <BR> <BR> <BR> CFU was 5-and 4-fold higher than in lacZ and wtMGMT transduced CFU, respectively. At the highest concentration of BG analyzed, the élective avantage <BR> <BR> <BR> for aMGMT transduced CFU was greatest; 50% CFU survival was observe after 25yM BG and 40UM BCNU, compare to <BR> <BR> <BR> only 1.5% of wtMGMT and 0.07% of lacZ transduced cells.

The transduction efficiency was 60% for wtMGMT <BR> <BR> <BR> retrovirus and 90t for ^MGMT as determined by PCR using primers specific for the provirus. aAGT expression in hematopoietic cells of mice transplanted with transduced bone marrow proqenitors Cellular expression of human NAGE in bone marrow and peripheral blood was tested by FACS analysis on permeabilized cells obtained from mice sacrifice 13 weeks after transplantation with retrovirally transduced bone marrow progenitor cells. The histograms in Figure 3A compare the average NAGE expression in bone marrow <BR> <BR> <BR> from mice transplanted with lacZ transduced progenitors (background, since there is no crossreactivity of the mT3.1 monoclonal antibody with the murine AGT; n=5) to <BR> <BR> <BR> aMGMT transplanted mice either treated with BG and BCNU (n=5) or left untreated (n=3). A proportion of the <BR> <BR> <BR> bone marrow of untreated AMGMT transplanted animals were observe to express aAGT above background levels and <BR> <BR> <BR> this population increased in the AMGMT mice treated with BG and BCNU. By subtracting the number of events at <BR> <BR> <BR> each value of fluorescence intensity in the lacZ<BR> <BR> <BR> <BR> <BR> histogram from the untreated or treated aMGMT<BR> <BR> <BR> <BR> <BR> histograms, we were able to determine the percent of

cells expressing oAGT (Figure 3B-E). Using this <BR> <BR> <BR> <BR> technique, 301 of bone marrow cells from untreated aMGMT mice expressed AAGT above background, and this percentage increased to 60% of bone marrow cells from BG <BR> <BR> <BR> <BR> and BCNU treated AMGMT animals when analyzed 13 weeks post-transplant. Fourteen percent of peripheral blood <BR> <BR> <BR> <BR> cells from untreated ^MGMT animals expressed AAGT compare to 30% in peripheral blood cells from BG and <BR> <BR> <BR> <BR> BCNU treated AMGMT mice, an increase identical to the two-fold enrichment seen in the bone marrow.

The mean fluorescence intensity of bone marrow <BR> <BR> <BR> <BR> cell AAGT expression in untreated AMGMT mice as detected by FACS analysis 13 weeks post transplantation increased <BR> <BR> <BR> <BR> 2.6 fold over lacZ mice and further increased 5.5 fold<BR> <BR> <BR> <BR> <BR> <BR> <BR> <BR> in the bone marrow from treated AMGMT mice. There was no increase in the maximal value of fluorescence intensity observe in bone marrow from BG and BCNU <BR> <BR> <BR> <BR> treated AMGMT mice compare to bone marrow from<BR> <BR> <BR> <BR> <BR> <BR> <BR> <BR> untreated aMGMT mice, suggesting that NAGE expression did not increase within individual cells. The mean AGT activity present in the bone marrow of individual mice <BR> <BR> <BR> <BR> was compare to the average activity from lacz transduced mice. The AGT activity increased by 1.3 to 4-fold (mean 2.6 + 1.4 fold, corresponding to 1.18 + 0.62 fmol CH3 removed/yg DNA) in the bone marrow of <BR> <BR> <BR> <BR> individual untreated AMGMT (n=3) transplanted mice compare to the average endogenous activity detected in <BR> <BR> <BR> <BR> the bone marrow of lacZ transplanted mice (0.45 + 0.33 fmol CH3removed lAg DNA ; n=5), and further increased 1.6 to 7-fold (mean 3.4 + 2.2 fold, corresponding to 1.52 + 0.99 fmol CH3 removed lAg DNA) in bone marrow obtained from individual mice treated with BG and BCNU (n=5).

The low activity of NAGE which conferred drug resistance to transduced cells demonstrates that the protective function is primarily due to BG resistance, and not due to increased alkyltransferase activity of the mutant protein. Therefore, BG and BCNU administration leads to

enrichment for drug resistant progenitors expressing high levels of QAGT.

BG and BCNU resistance of CFU from the pre-transplant cell copulation A portion of the transduced murine bone marrow progenitor population used for animal transplantation was treated with 20yM BG for 1 hour followed by 0-80 yM BCNU for 2 hours. In three separate experiments, the <BR> <BR> <BR> <BR> BCNLT ICSO in AMGMT transduced CFU pretreated for one hour with 20 yM BG was 26.6 yM + 0.1 UM compare to 4.4 yM + <BR> <BR> <BR> 1.5 UM in lacZ transduced CFU, representing a six-fold<BR> <BR> <BR> <BR> <BR> <BR> <BR> shift and significantly increased survival of the QMGMT transduced cells (p<0.0001; Figure 4). The transduction <BR> <BR> <BR> <BR> efficiency of QMGMT into CFU was determined to be 69k (11 of 16) by proviral specific PCR amplification of individual progenitor colonies.

BG and BCNU resistance of CFU from the bone marrow of transplanted mice after enrichment Bone marrow cells from mice sacrifice 13 weeks after transplant were analyzed for BG and BCNU resistance by in vitro drug treatment with 20yM BG and 0-80 yM BCNU. As shown in Figure 4, BG and BCNU <BR> <BR> <BR> <BR> treatment of QMGMT mice enriched for BG and BCNU resistant CFU. The BCNU IC50 after 20yM BG in CFU from <BR> <BR> <BR> <BR> BG and BCNU treated QMGMT animals was 68yM compare to<BR> <BR> <BR> <BR> <BR> <BR> <BR> 6.5yM in untreated QMGMT and 6.2yM in both treated and<BR> <BR> <BR> <BR> <BR> <BR> untreated lacZ animals, representing a greater than ten- fold increase in drug resistance (p<0.0001). Although <BR> <BR> <BR> <BR> CFU survival was similar for untreated QMGMT<BR> <BR> <BR> <BR> <BR> <BR> <BR> transplanted mice and lacZ transplanted mice at low doses of BCNU, the BCNU IC90 after 20yM BG in the <BR> <BR> <BR> <BR> untreated QMGMT transplanted mice was 62 HM compare to<BR> <BR> <BR> <BR> <BR> <BR> 19 UM in the lacZ mice, a 3.3 fold shift (p<0.0001).<BR> <BR> <BR> <BR> <BR> <BR> <BR> <P>Therefore, mice transplanted with QMGMT transduced bone marrow progenitors retained BG and BCNU resistance in a proportion of CFU without BG and BCNU mediated selective

pressure, but increased resistance over lacZ controls was only detectable after in vitro treatment of CFU with higher concentrations of BCNU.

Mice transplanted with QMGMT transduced hematopoietic progenitors and treated with BG and BCNU at 3 and 6 weeks post transplant were sacrifice 23 <BR> <BR> <BR> weeks after transplantation. The BCNU IC50 af ter BG in bone marrow derived CFU was 41 HM compare to 6.2 yM in normal mice (Figure 4). A modest decrease in survival was observe in CFU derived from 23 week mice compare to 13 week mice at 80 4M BCNU plus 20 UM BG. FACS analysis demonstrated increased expression of QAGT in <BR> <BR> <BR> bone marrow cells from AMGMT mice 23 weeks post transplant compare with bone marrow from untransplanted mice. The retention of BG and BCNU resistant CFU 13 and 23 weeks after transplantation implies that drug treatment enriches for transduced early progenitors. <BR> <BR> <BR> <P>Thus, transplantation of QMGMT transduced bone marrow progenitors into lethally irradiated mice and enrichment by in vivo BG and BCNU administration led to reconstitution of the bone marrow with BG and BCNU resistant cells due to long term expression of QAGT.

Frequency of proviral sequence detected in CFU from transplanted mice The bone marrow derived CFU from untreated and BG and BCNU treated mice sacrifice 13 weeks post transplant were analyzed for proviral integration by PCR using proviral specific primers. While 22 of 33 (67W) of CFU from untreated QMGMT animals were positive for presence of the provirus, 50 out of 50 (100%) CFU from the BG and BCNU treated animals were positive, demonstrating that in vivo BG and BCNU administration enriches for the retrovirally transduced cells in vivo (Table 1). 32 of 34 CFU (94W) were positive in the mice sacrifice 23 weeks post transplant.

Table 1. Comparison of murine bone marrow derived MGMT transduced CFU and wtMGMT transduced CFU.

BCNU IC50 BCNU ICgO % #MGMT<BR> #MGMT experiments after 20µM BGª after 20µM BGb positive CfU' i acZ 6.2 SM 19 µM N/A9<BR> <BR> <BR> <BR> <BR> <BR> <BR> pre transplant 25 µM >70 µMd 69%<BR> (11 of 16)<BR> <BR> <BR> <BR> <BR> <BR> <BR> unselected, week 13 6.5 pM 62 pM 6OX<BR> <BR> (22 of 33)<BR> 100%<BR> <BR> <BR> <BR> <BR> <BR> selected, week 13 68 pm >80 µMe (50 of 50)<BR> <BR> <BR> <BR> <BR> <BR> <BR> selected, week 23 41 uM gp uM 94%<BR> <BR> <BR> (32 of 34) wt MGMT experiments* BCNU IC50d BCNU IC90ª % wtMGMT<BR> <BR> <BR> <BR> positiveCRU' (acZ 5 uM ib uM N/A9<BR> <BR> <BR> unselecteduM3uM37y<BR> week 13-17 (18 of 49)<BR> selected, 18 µM 43 µM 90%<BR> week 13-17 (54 of 60) ª Inhibitory concentration of BCNU which reduces cell survival to 50% after BG pretreatment b Inhibitory concentration of BCNU which reduces cell survival to 10k after BG pretreatment c Determined by MFG-#MGMT proviral specific PCR d 13. 3k survival at 70 yM (highest dose tested)<BR> <BR> e 35. 1o survival at 8Q UM (highest dose tested) f Determined by human MGMT cDNA specific PCR g Not applicable * data from Allay, et. al. Exp. Hem., in press<BR> <BR> <BR> <BR> <BR> <BR> Mice transplanted with AMGMT transduced hematopoietic progenitors survive in vivo BG and BCNU administration that is toxic to control mice In four independent experiments, lethally <BR> <BR> irradiated mice were transplanted with 1 X 106 lacZ (n=13) or #MGMT (n=22) transduced bone marrow progenitors and treated with 30 mg, lkg BG and 10-20 mg/kg BCNU 2-6 weeks post-transplant as shown in Table 2. The mice were observe daily for survival for 90-120 days after transplantation before being removed from the <BR> <BR> <BR> study. Impressively, a total of 21 out of 22 AMGMT transplanted animals survive combination BG and BCNU <BR> <BR> treatment compare to only 3 of 13 lacZ transplanted animals (P<0.0001, Mann-Whitney U; Figure 5).

Periodically, mice were sacrifice before 90 days post transplant and were censored in this study (identified by the open circles and triangles in the figure). The censoring should not impact the interpretation of BG and BCNU tolerance since the majority of the toxicity occurred up to two weeks following drug administration.

LacZ transplanted mouse mortality occurred at a median <BR> <BR> <BR> 7.5 days after drug treatment, whereas the one aMGMT mouse death occurred 16 days after drug administration. <BR> <BR> <BR> <P>Therefore, transplantation of AMGMT transduced cells into mice protects against lethal doses of combination BG and BCNU treatment.

Table 2. BG and BCNU treatment schedule of transduced mice. BG/ time of Day of BCNU treatme # of death # of # of Expt. dose nt lacs ouf after nMGMT AMGMT (weeks anima lacZ drug animal mice Day of death (mg/k post ls mice treatm s dead/tot after drug g) transpl treat dead/to ent treate al treatment ant) ed tal d 1 30/15 week 2 3 0/3 --- 6 0/6 --- 30/20 week 4 3 3/3 4,6,7 6 1/6 17 2 30/10 week 3 0 --- --- 2ª 0/2 --- 30/10 week 6 4 2/4 6,7 5ª 0/5 --- 3 30/10 week 3 2 2/2 8,8 4 0/4--- 4 30/10 week 3 4 2/4 14,17 7 0/7 --- 30/10 week 6 2 1/2 12 7 0/7--- Total 13 10/13 median 22 1/22 median: s (77%) 7.5 (4.5%) 17 days days a Two mice were treated weeks 3 and 6, three mice were treated week 6 only.<BR> <P>Therefore, a total of 5 mice were treated in expriment 2.

EXAMPLE TWO Mice were treated with a non-myeloablative dose of 30 mg/kg BG and 10 mg/kg BCNU (half of the Lolo) 48 hours before transplant, reducing the total bone marrow cellularity by 57k and pregenitors by 77%. After transplant, the mice received 0 to 4 cycles of BG and BCNU every 4 weeks. Bone marrow progenitors were analyzed for in vitro resistance to BG and BCNU, frequently of #MGMT using flow cytometry and western blotting. #MGMT cDNA was not detected in any CFU from untreated, transplanted mice. However, in mice given two post-transplant cycles of BG and BCNU and analyzed 24 weeks after transplant, the BCNU ICgo in bone marrow CFU pre-exposed to 20yM BG increased up to 5-fold over non-treated mice and AMGMT provirus was detected in up to 971 of CFU tested (both in proportion with the number of AMGMT + cells infused; see Table 3), suggesting transduction of long term repopulating progenitors.

Table 3. BG and BCNU cycles 0 2 AMGMT+ ceLLs up to 1 x 106 5 x 10'25 X 104 1 x io, infused BCNUIC., (+20PM BG) 8 pM 15 OM 27 IIM 40 JJM % #MGMT + CFU 0% (0 of 89) 15% (4 of 26) 42% (10 of 24) 97% (32 of 33) We estimate a 10-fold enrichment for transduced cells with every successive cycle of BG and BCNU. The ability to enrich for transduced long term repopulating progenitors in non-myeloablated mice suggests that AMGMT may be a useful dominant selectable marker in vitro even when a minimal number of cells are transduced, and may permit high-level co-expression of a therapeutic gene in hematopoietic cells.

EXAMPLE THREE In this example, the effect of AMGMT transduced syngeneic marrow infusion of BG and BCNU treatment of tumor bearing nude (nu/nu athymic) mice was evaluated.

Prior to subcutaneous implantation of BCNU resistant, AGT+, human colon cancer (SW480) cells, cohorts of mice received ip injections of 30mg/kg BG and 10mg/kg BCNU (one half of the Ldlo) and then were infused with 1 x 106 syngeneic aMGMT (n=l9), or lacZ transduced (n-10) marrow cells. Treatment of AMGMT mice with three therapeutic cycles of BG (30mg/kg) and BCNU (10-25 mg/kg BCNU) resulted in repopulation of the marrow with 93+13 <BR> <BR> <BR> <BR> W aMGMT transduced CFUs which were resistant to BG and BCNU (BCNU IC50= 30yM vs 5yM for the lacZ CFUs).

Fourteen days after the third cycle of therapy, bone marrow cell and CFU counts, AGT activity, and block RBC and platelet counts were significantly (P<0.05) higher in AMGMT compared to lacZ mice. The improved marrow tolerance resulted in 80W survival of AMGMT mice following the fourth cycle of BG and BCNU treatment compare to 42% of the lacZ mice. Xenograft growth was significantly delayed in mice receiving three cycles of BG and BCNU, compare to a single cycle of therapy.

Thus, AMGMT transduced marrow cells can selectively repopulate the marrow after non-ablative, therapeutic doses of BG and BCNU and increase the tolerance to this combination. In this preclinical mode, AMGMT transduced marrow cell infusion allowed effective therapy of a BCNU resistant human tumor xenograft, providing support for clinical use of AMGMT transduced hematopoietic progenitors during BG and BCNU chemotherapy.

EXAMPLE FOUR-MATERIALS AND METHODS Materials-BG and 5-nitroso-BP, were synthesized and supplie by Dr. R. C. Moschel (NCI-FCRDC, Frederick, MD). BCNU was obtained from the Drug Synthesis and Chemistry Branch, Division of Cancer Treatment, NCI,

Bethesda, MD. N- ['HI methyl-N-nitrosourea was purchased from Amersham Inc., Arlington Hts., IL. Other reagents for molecular biology, cell culture and AGT assays were obtained from: GIBCO BRL Life Technologies.

Gaithersburg, MD; Sigma, St Louis, MO; Atlanta 10 Biologicals, Norcross, GA: New England Biolabs, Beverly, MA; Perkin Elmer, Bonchburg, NJ; and Promega, Stadison, WI.

Plasmid Constructions In order to express AGT in Chinese hamster ovary (CHO) cells, the plasmid pCMV-Neo-Bam (Baker, et al., 1990, Science :249 912-915) which has a CMV promoter controlling the expression of the sequences inserted at a unique BamHI site and a neomycin resistance gene for selection by geneticin of clones which have taken up the plasmid. The cDNAs corresponding to wild type, P140A and G156A AGTs were obtained by PCR using pGEMAGTs containing these inserts (Crone and Pegg, 1993, Cancer Res. :53 4750-4753; Crone, et al., 1994, Cancer Res. <BR> <BR> <BR> <BR> <P> 54: 6221-6227, hereby incorporated by reference). The primer used for the 31 zend corresponded to the SP6 RNA polymerase promoter sequence and the primer for the 5' end was 5'-CTCACTATAGGATCCAAAATGGACAAGGAT-3' (mismatches underlined). This primer creates a BamHI restriction site and also restores the initiation codon of the AGT sequence. The PCR product was digeste with BamHI and inserted into pCMV-Neo-Bam at the BamHI site.

Plasmids were isolated and checked for the insertion of the AGT sequence in the correct direction by digestion with XbaI and DraIII which cut unique sites in the vector and the AGT cDNA respectively. The AGT sequence in a plasmid showing the correct orientation was then verified by sequencing.

Cell culture CHO cells were maintained by seeding at 2.5 x 101 cells per 75 cm2 flask and grown in a-MEM medium containing 36 mM NaHCO3, 10W fetal calf serum, 100 units/ml penicillin and 100 mg/ml streptomycin. HT29 cells were grown in Dulbecco's modifie Eaglels medium containing 36 mM NaHC03 supplemented with 10t fetal calf serum plus 3o glutamine and gentamycin (50 yg/ml).

Transfection and selection Drocedure CHO cells were transfected using Lipofectin (GIBCO BRL) according to manufacture's protocol for stable transfection of adherent cells. Cells were seeded at 105 cells in 60 mm tissue culture plates. For each transfection, 2 Ug DNA and 10 y1 Lipofectin diluted in serum-free antibiotic-free medium were used. Cells were incubated for 5 h. then the DNA containing medium was replace with growth medium. At 48 h later, geneticin (GIBCO BRL) was added at a final concentration of 1 mg/ml to select the cells expressing the neomycin resistance gene. In addition, the cells transfected with AGT and mutant P140A AGT were selected for their resistance to killing effect of BCNU (80 tiM)-Clones from individual cell colonies were isolated and used for further experiments.

Cell qrowth In order to determine the effect of BG on the rate of cell growth, the cells were seeded at 105 cells per 75 cm2 flask and allowed to grow for 48 h. The medium was then replace with fresh medium or fresh medium containing different concentrations of the drug. After 24 h, the medium was replace and cells were allowed to grow until they reach 90k of confluence in a control flask. The medium was changed every 24 h. Control cells and cells after exposure to BG were harvested and counted at different time points.

Cytotoxicity assaYS Cell killing was determined using a colony forming <BR> <BR> <BR> assay (Pegg, et al., 1995, Bíochem. Pharmacol. :50 1141-<BR> <BR> <BR> <BR> <BR> <BR> 1148; Dolan, et al., 1986, Cancer Res. :46 4500-4504).<BR> <BR> <BR> <BR> <BR> <BR> <P>The cells were plated using 106 cells per 25cm 2 flask and grown for 24 h. After 2 h of incubation with either BG or 5-nitroso-BP, 80 itM BCNU was added for 2 h. The medium was then replace with fresh medium also containing the AGT inactivator and the cells were left at 37°C for an additional 16-18 h. The cells were then replated at densities of 100 to 1000 cells per 25 cm2 flask and grown for 7-8 days until discrete colonies could be stained and counted. The colonies were washed with 0.9% saline and stained with crystal viole and counted.

AGT activity assays Extracts were prepared and AGT activity was determined as described by Dolan et al. (Dolan, et al., 1990, Proc. Natl. Acad. Sci. U. S. A. :87 5368-5372) by incubation for 30 min at 37°C with a [3H] methylated calf thymus DNA substrate prepared by rection of DNA with N- [3 Hlmethyl-N-nitrosourea (Amersham Inc., Arlington Hts., IL). The AGT activity of the cells was then expressed as fmol of 06-methylguanine removed per mg of protein present in the cell extracts. Protein was determined by the method of Bradford (Bradford, 1976, Anal. Biochem. 12: 248-254). Inactivation of cellular AGT was measured by adding different concentrations of BG or 5-nitroso-BP to cell cultures which had reached about 80% confluence. After 4 h exposure, the cells were harvested and washed with PBS. Cell pellets were stored at 80°C until assayed. Extracts were then prepared and AGT activity determined. The results were expressed as the percentage of the AGT activity present in cultures to which no drug was added.

Western blot analysais The expression of the wild type and mutant AGT proteins in transfected CHO cells was measured using immunoblots. After SDS-PAGE in 12.5W acrylamide gels, the protein immobilized on the nitrocellulose membrane was determined by Western-Light Chemiluminescent Detection System (TROPIX, Inc.) using antibody MAP-I (Pegg, et al., 1991, Carcinogenesis :12 1671-1677). The intensity of chemiluminescence was measured by densitometric scanning using a laser densitometer.

EXAMPLE FOUR-RESULTS AGT activity was measured in individual clones of CHO cells after their transfection with the pCMV plasmids expressing human cDNA for wild type and mutant AGTs. The activity in untransfected CHO cells was below the limit of detection but multiple clones expressing high levels of AGT were readily obtained from the pools of transfected cells. There was a considerable range in AGT activities among the various clones. Those clones selected on the basis of resistance to BCNU as well as to G418 had on average a higher AGT activity but the range of activities observe in the two groups overlapped. Three clones containing equivalent activities of wild type, P140A and G156A mutant AGT were taken and used for further experiments. As shown in Table 4, these clones had AGT activities which were very similar and were about 2.5 times greater than that of HT29 cells. AGT protein could readily be measured in extracts from these cells using immunoblots developed with an antibody to peptide corresponding to amino acids 8-20 of the human AGT sequence (Figure 6). A single protein was detected with a M. W. of about 22 kDa. This band was completely absent from the untransfected CHO cells. Compare to the HT29 cells there was about 5.5 times more protein reacting with antibodies in the CHO cells transfected with wild type AGT, 8.2 times more for

those with P140A and 11.5 times more in case of G156A according to densitometric measurement of these Western blots. These results suggest that some of the AGT in the transfected cells is inactive and that a larger fraction of the mutant AGT is non-functional since the measured AGT activities were the same and only 52. times more than the HT29 cells.

Table 4. Activity of AGT in transfected CHO cells and its inactivation by BG and 5-nitroso-BP.

Inactivation of AGT by BG or 5-nitroso-BP, EDso (yM) after 4 h exposure AGT activity Cells (fmol/mg) BG 5-nitroso-BP HT29 456 0.05a 0.02b CHO-AGT 1144 0.5 0.05 HO-PUA 1082 15 1.25 CHO-G156A 1152 30 5 a Dolan, et al., 1990, Proc. Natl. Acad. Sci. U. S. A.

87: 5368-5372. b Pegg, et al., 1995, Biochem. Pharmacol. :50 1141-1148.

Control CHO cells were effectively killed by BCNU with less than 0. 1k survival after 80 UM BCNU. In contrast, the transfected cells expressing control or mutant AGT were completely resistant to killing by up to 100 yM BCNU which was the highest concentration tested (Figure 7).

The AGT activity in the cells expressing control AGT was readily decreased by exposure of the cells to BG or 5-nitroso-BG (Figure 8A, 8B). Considerably more of these drugs had to be used to obtain the same inhibition of AGT in the cells expressing P140A AGT and those expressing G156A were even more resistant. The 5- nitroso-BP was more active than BG as an AGT inactivator in all three cells lines (Figure 8B). A summary of the

inactivation based on the concentration of drug needed to reduce AGT activity by 50% in the 4 h exposure period [EDso] is shown in Table 4. The ED50 for BG was increased by 30-fold to 15 yM with P140A AGT and by 60-fold to 30 yM with G156A. Similarly, the EDso for 5-nitroso-BP was increased to 25-fold to 1.25 HM with P140A AGT and by 100-fol to 5 yM with G156A AGT.

These results suggest that rnuch higher levels of these AGT inhibitors may be needed to render the cells expressing mutant AGT sensitive to BCNU. Such high concentrations might have an effect on cell growth invalidating the colony forming assays so a control expriment was carried out to determine the effect of BG on cell growth. Exposure to BG for 24 h led to dose- dependent inhibition of cell growth with concentrations higher than 200 yM having a profond effect on cell growth during the 24 h after the drug was removed from the medium (Figure 9). However, during the next 48 h these cells recovered and were growing at the same rate as control cells. Since the effect of BG on cell growth was clearly reversible, the effects of exposure to the AGT inhibitors on colony forming ability could be determined (Figure 10). Although the colonies observe were smaller in size (by a factor of up to 3 corresponding to a reduction of about 40% in the number of cells making up a colony) with the higher concentrations of BG due to the inhibition of growth in the first 24 h, the cells were grown for a further 7 days so that smaller colonies were detected as readily as larger ones. The results showed that exposure to BG for 20 h had no effect on colony forming activity up to 200 UM BG. However, 5-nitroso-BP was more toxic with concentrations above 30 yM reducing colony formation.

Addition of BG to CHO cells expressing control AGT rendered them highly sensitive to 80 UM BCNU with 50 HM BG being required to reduce colony formation to < 1 per 1000 cells plated (Figure 11). This level of BG is

considerably greater than that needed to sensitize a wide variety of human tumor cells to BCNU (see e. g., <BR> <BR> <BR> Pegg, et al., 1995, Progr. Nucleic Acid Res. Mol. Viol.<BR> <BR> <BR> <BR> <BR> <BR> <BR> <P> 51: 167-223; Dolan, et al., 1990, Proc. Natl. Acad. Sci.<BR> <BR> <BR> <BR> <BR> <BR> <BR> <P>U. S. A. , :87 5368-5372; Pegg, et al., 1995, Biochem.

Pharmacol. :50 1141-1148; Dolan, et al., 1986, Cancer Res. :46 4500-4504). This difference is probably due to the high level of AGT in the transfected CHO cells which was 2.5 times that of HT29 cells which are among the highest expressors of AGT in human cell lines. Another factor may be the fact that AGT in the CHO cells is expressed from the CMV promoter.

The CHO cells transfected with mutant AGT cDNAs were much more resistant to sensitization by BG (Figure 11A). The addition of 200 yM BG was needed to reduce colony formation in the BCNU treated cells to < 1 per 1000 cells plated in the cells expressing P140A AGT.

The cells expressing G156A AGT were even more refractory to the combine effects of BG and BCNU with substantial survival even in the presence of 350 yM (Figure 11A).

As shown in Figure 11B, 5-nitroso-BP was very effective in sensitizing the CHO cells containing control AGT with about 5 yM required for the reduction of colon formation to <1 per 1000 cells plated. More than 25 yM of 5-nitroso-BG was needed to sensitize the cells expressing P140A AGT and cells expressing G156A mutant AGT required concentrations of >40 yM which as shown in Figure 10 affect colon formation even in the absence of BCNU.

These data presented in this example show clearly that the mutant P140A and G156A AGTs are sufficiently stable in the cell and active on the repair of chloroethyl groups in DNA to provide complete protection from killing by BCNU. Although these experiments were carried out in Chinese hamster cells, it is extremely likely that this conclusion would also be valid in both human tumor cells and in non-malignant cells. Our

experiments to establish the ability of the mutant AGTs to provide protection were carried out in CHO cells for several reasons. First, the ease of transfection and cloning of these cells made it possible to obtain clones expressing similar levels of AGT activity for control and mutant AGT cDNAs. This is critical for accurate comparison of the effects of AGT expression on protection from BCNU. Second, the use of CHO cells allows us to be certain that the expressed AGT does come from the plasmid cDNA rather than from the activation of a hamster gene since the antibodies used react exclusively with the human protein (Pegg, et al., 1991, <BR> <BR> <BR> <BR> Carcinogenesis :12 1679-1683). Such activation has been shown to have occurred in other studies (Pegg, et al., 1991, Carcinogenesis :12 1679-1683; Tano, et al, 1991, <BR> <BR> <BR> <BR> Mutation Res. 255: 175-182; von Wronski and Brent, 1994,<BR> <BR> <BR> <BR> <BR> <BR> <BR> Carcinogenesis :15 577-582; Arita, et al., 1990, Carcinogensis :11 1733-1738). Third, other studies show that expression of CHO cells of human DNA proteins including AGT has been studied for its effects on cellular physiology. Thus, previous studies have shown that the expression of normal mammalian or bacterial AGTs in CHO renders the cells resistant to the toxic effects of alkylating agents. The results shown in this Example are similar to those found when AGT is expressed in human tumor cells or in transgenic mice (see review, e. g., Mitra and Kaina, 1993, Nucleic Acid Res. 44: 109- <BR> <BR> <BR> <BR> 142; Pegg, A. E., 1990, Cancer Res. 50: 6119-6129; Pegg,<BR> <BR> <BR> <BR> <BR> <BR> <BR> et al., 1995, Nucleic Acid Res. Mol. Biol. :51 167-223;<BR> <BR> <BR> <BR> <BR> <BR> <BR> Gerson, et al, 1994, Mutation Res. :307 541-555).

Despite the differences in the content of protein revealed by Western blotting, no obvious differences in the stability of the mutant and control AGT proteins in the CHO cells were observe. The protein was too stable for measurement of its half life of decay after treatment with a protein synthesis inhibitor. Although the mammalian AGT proteins are highly conserve, it is

more likely that a reduced stability (or incorrect folding) of the human AGTs protein would be observe in the foreign environment of a rodent cell than in the human cells. Therefore, it is highly probable that these mutant proteins will be sufficiently stable and active in both normal and neoplastic human cells to prevent lethal crosslinking by chloroethylating agents.

The fate of the alkylated (or benzylated) form of the AGT protein produced by its rection with alkylated DNA or BG respectively is still not well understood. In some cell lines, including both HT29 cells and CHO cells, this protein is degraded rapidly (Gerson, et al, <BR> <BR> <BR> 1994, Mutation Res. :307 541-555) and recent studies with HT29 and CEM human tumor lines suggest that ubiquitination may be involved in this process (Pegg., <BR> <BR> <BR> et al., 1991, Carcinogenesis :12 1679-1683). However, another study (Ayi, et al, 1994, Cancer Res. 3726-54, 3731) the alkylated form of the protein was reporte to be sufficiently stable in CEM and HeLa S3 cells to be detected by Western blotting although a conformational change to increase cleavage by protase V8 was detected.

Whether these results represent a significant difference in how the AGT protein is regulated in different cell types remains to be determined but, in any event, the CHO cells have similar properties to the human colon carcinoma HT29 cells with respect to the degradation of the alkylated AGT protein.

The results in Figures 8 and 11 show clearly that the point mutations at position Prol4o and G1y156 do render the human AGT resistant to BG and thereby reduce the ability of this drug to sensitize cells having such AGTs to BCNU. BG is currently entering clinical trials as a means to render tumor cells containing high levels of AGT sensitive to chloroethylating agents (Gerson, et al, <BR> <BR> <BR> <BR> 1994, Proc.-Am. Assn. Cancer Res. :35 699-700; Pegg, et<BR> <BR> <BR> <BR> <BR> <BR> al., 1995, Progr. Nucleic Acid Res. Mol. Biol. 51: 167- 223). The selection of tumor cells expressing forms of

AGT insensitive to this drug is a potentially serious problem since several single amino acid changes have been shown to impact such resistance. These data from this Example Section strengthen this concern since the results presented here show that at least two of these alterations (P140A and G156A) do lead to an AGT activity in cells which is able to protect from the toxicity of BCNU even in the presence of BG. Based on the solubility and pharmacokinetics of BG in animals (Dolan, et al., 1994, Cancer Chemotherapy :35 121-126) and preliminary studies in primates (Berg, et aï., 1995, Cancer Res. :55 4606-4610), it is unlikely that plasma levels of BG of greater than 20 UM can readily be achieved. These levels are capable of completely inactivating the control AGT in a variety of human tumor cells (see, e. g., Gerson, et al., 1994, Proc. Am. Assn. <BR> <BR> <BR> <BR> <P>Cancer Res. :35 699-700; Pegg, et al., 1995, Nucleic Acid Res. Mol. Viol. :51 167-223) and produce a large increase in killing by BCNU. In contrast, the levels of BG that are needed to produce sensitization of the transfected CHO cells when the P140A and G156A mutant AGTs were used are much greater than the predicted plasma levels.

This problem may be overcome by producing inhibitors that are active even against the resistant mutant forms of AGT. Such inhibitors fall into two classes: those which are able to inactivate the mutant forms as readily as they inactivate the control AGT; and those which are much more potent than BG so that a plasma level could be maintained that inactivates the mutant AGT proteins even though there is a difference in reactivity with the mutant compare to the control AGT.

5-nitroso-BP, which is the most potent AGT inhibitor so far reporte on the basis of its ability to inactivate GAT in HT29 cells and cell extracts and to sensitize a variety of tumors to BCNU is not sufficiently active for this purpose.

It is very likely that a major reason for the resistance of the mutant forms of human AGT and the AGTs from microorganisms to BG is due to a steric limitation at the active site (see, e. g., Pegg, et al., 1995, <BR> <BR> <BR> Progr. Nucleic Acid Res. Mol. Biol. 51: 157-223; Pegg,<BR> <BR> <BR> <BR> <BR> et al., Biochemistry :32 11998-12006). Although the structure of the AGT has not yet been determined, the mutations that render it resistant are in residues that may decrease the size of the active site pocket.

Furthermore, alterations of the E. coli Ada-C alkyltransferase, which are predicted on the basis of the crystal structure to increase the accessibility of the cystine acceptor site, do render it sensitive to BG <BR> <BR> <BR> (Crone, et al., 1995, Carcinogenesis :16 1687-1692). As more information becomes available on the size of the active site pocket and the means of binding of low molecular weight substrats such as BG, it may therefore be possible to design compound that are able to fit into the active site and inactivate even the mutant AGTs. The CHO cell lines expressing these mutant AGTs will be useful for the screening of such inhibitors for their ability to sensitize cells to CNU.

It should be noted that the control AGT and the mutant AGTs were expressed from the CMV promoter in our experiments. This is a very strong promoter in CHO cells and the amount of AGT formed was considerably greater than that found in most if not all human tumors.

It is possible that the high level of AGT expression contributes to the insensitivity to BG. It is conceivable that both a mutation to a resistant form of AGT and a change in the level of AGT expression will be necessary to impact a high level of resistance to the BCNU/BG combination reducing the likelihood that such changes will be a problem in clinical trials.

Nevertheless, it seems prudent to design strategies to overcome such resistance. The concentrations of BCNU used in the present experiments were considerably higher

than those likely to exist in patients treated with this drug and much lower levels of expression of AGT may be sufficient to provide protection from chloroethylating agents in clinical protocols.

Finally, it is noteworthy that with cells expressing either the control of the mutants AGT, the concentration of BG or 5-nitroso-BP needed to get maximal enhancement of killing by BCNU is much higher than that needed to give > 90k inactivation of the AGT.

Similar results have been observe with a variety of human tumor cells treated with BG (Doland, et al., 1990, Proc. Natl. Acad. Sci. U. S. A. 87: 5368-5372; Dolan, et al., 1991, Cancer Res. :51 3367-3372; Marathi, et al., 1993, Cancer Res. :53 4281-4286) or other derivatives <BR> <BR> <BR> <BR> (Pegg, et al., 1995, Biochem. Pharmacol, :50 1141-1148).

The reasons for this are not known with certainty but two possibilities are that: (a) a small mount of residual AGT may provide significant protection or (b) it is may be necessary to ensure that the GAT made de novo over the entire period in which lethal cross links can be formed from the 06-chloroethyl guanine adducts reacts with BG rather than repairs DNA. This period may extend past the end of the 24 h period used in these experimental protocols and thus could involve a contribution from the residual drug present in the cells after the medium is change. Furthermore, the competition would require more of the drug than the inactivation of AGT prior to treatment with the alkylating agent. It has been difficult to investigate this problem in detail with cells expressing the control AGT since the methods of assaying BG or derivatives currently available are not sufficiently sensitive to measure their intracellular levels in cultured cells and relate them to the biological effects and the sensitivity of the AGT in vitro. The use of the mutant AGTs in which much greater levels of BG are needed would move these levels into the mesurable range and may

provide a means to investigate this question.

EXAMPLE FIVE-METHODS AND MATERIALS Retroviral Vectors Retroviral vectors pMFGwtMGMT and PMFGAMGMT were constructed by inserting the wild type and G156A mutant human MGMT cDNA coding sequences respectively, into the unique Ncol and BamHI restriction sites of the Moloney murine leukemia virus derived-retroviral vector pMFG (Ohashi et al. 1992, Proc. Natl. Acad. Sci. :89 11332-11336). The Ncol and BamHI sites at the 5'and 3' termini of human MGMT cDNA sequence, respectively, were generated by PCR amplification with a pair of primers <BR> <BR> <BR> (5'-p: 5'CTTGGAACCATGGACAAGGATTGTGAAA3', 3'-p : 5' CTTAGGATCCCATCCGATG CAGTGTTACACG3' : corresponding restriction sites underlined). The ATG sequence in the created Ncol site serves as start codon of human MGMT cDNA. aMGMT cDNA was generated by two steps of PCR amplification with the 5'-p and 3'-p primers and another pair or primers (5'-mp: 5'AGCGGAGCCGTGGCCAACTACTCCGGA3', 3'-mp: 5'TCCGGAGTAGTTGGCCACGGCTCCGCTG3'). These primers were derived from the cDNA sequence with a single mismatch (bol) at the second base of codon Glycine 156 to Alanine. First, 5'and 3'fragments were amplifie with primers 5'-p/3'-mp and 5'-mp/3'-3, respectively, then both 5'and 3'fragments were mixed to amplify the full length mutant cDNA with 5'-p and 3'-p primers. Sequences were confirme by the dideoxynucleotide chain termination method (fmol DNA Sequencing System. Promega Biotec, Maison, Wl).

Transfection of the vector constructs Into CHO cells Six Ug of each plasmid (pMFGwtMGMT and pMFGaMGMT) were cotransfected into 1.8 x 106 cells with 0.6 yg of pSV2neo plasmid DNA by Lipofectamine (Gibco BRL, Gaithersburg, MD) following manufacturer's protocol.

Clones of transfected cells were selected in G418

(lg/L).

Virus producinq cells Virus producing cells lines were made by co-tansfecting pMFGMGMT or pMFGMGMT and pSV2neo DNA into the packaging of line GP + E86. After selection in G418, viral supernatant was collecte and used to infect the amphotropic cell line GP + envAml2 (provided by Arthur Bank, Columbia Univ). To increase titer, a supernatant"ping-pong"method was used as previously described (Bodine, 1990, Proc. Natl. Acad. Sci. :87 3738). The amphotropic QCRIP MFG-lacZ was provided by Paul Robbins (Univ of Pittsburgh). Titer was estimated from supernatants collecte after 6 daily media changes by infecting 1x105 K562 cells (as described below) with limiting dilutions of viral supernatant.

K562 Transduction The human chronic myelogenous leukemia cell line K562 was transduced as follows. At 80k confluence, amphotropic vMGMT and VAMGMT producer cells were treated with 10 yg/mL mitomycin C for 2 hours, washed 4 times in serum-containing medium. trypsinized, and replated in complete medium, Twenty-four hours later, 2.5 X 105/mL K562 cells were added along with human interleukin-3 (IL 3) (100 U/mL, provided by Genetics Institute, Cambridge, MA), GM-CSF (100 U/mL, provided by Sandoz Research Institut. Nutley, NJ) and 6 yg/mL polybrene.

Forty-eight hours later, the nonadherent K562 cells were procure and either analyzed for gene transfer or grown in medium containing 500 Ug/mL G418. Selected cells were analyzed after 4 weeks.

CD34+ Transduction Peripheral blood mononuclear cells were obtained by apheresis from patients treated with cyclophosphamide and G-CSF. CD34'progenitors were isolated using the

Ceprate SC Stem Cell Concentrator (Cell Pro, Bothell, WA) according to manufacturer's directions. Briefly, cells were washed and incubated with biotinylated anti-CD34 antibody, passed over an avidin column and the bound fraction eluted by gentle agitation. Recovered cells had an average purity of 57*-.. Freshly obtained CD34+ cells (5 x 105 cells/ml) were resuspended in IMDM containing 20k heat inactivated FCS and supplemented with human SCF (100 mg/ml, Amgen), Il-3 (100U/ml) and IL-6 (100 U/ml both from Sandoz) and protamine sulfate (4 mg/ml, Sigma) and co cultured with MFG-nMGMT and MFG-lac producers prepared as above. At 48 hours, half of the media was removed, cells were pelleted and resuspended in fresh, supplemented media, and non-adherent cells were collecte at 96 hours.

In vitro BCAJ/BG treatment Cells transduced with MFG-MGMT, MFG wtMGMT, MFG-lac or uninfected cells were resuspended in serum free media containing 100U/ml GM-CSF with or without 10yM BG and incubated at 37°C for 1 hour with continuous mixing. BCNU was dissolve in 100k ethanol and diluted to lomm with serum free media and immediately added to the cells. Following a two hour incubation at 37°C, cells were incubated in methylcellulose (Stem Cell Technologies, Inc.) containing either SCF, IL3, hemin, erythropoietin, GM-CSF (for CD34+ cells) or GM-CSF and IL-3 (for K562 cells) in triplicate. Cells treated with BG received an additional 5 yM dose in the methylcellulose medium.

After incubation for 7-10 days at 37°C colonies were enumerated and survival was analyzed at each BCNU dose.

Immunsassay Cytospin preparations were stained for AGT using the monoclonal antibody mT3.1 (provided by D. Bigner) and the blotin/avidin horseradish peroxidase system

(Vector Laboratories, Burllngame, CA). Western blots were done by sonicating and boiling a sample in a buffer containing 50 mmol/L TRIS-HCL, pH 6.3.2t sodium dodecylsulfate (SDS), ltß-mercaptoethanols 0.1 mol/L dithiothreitol (DTT), 5W sucrose, and 300 ßmol/L NaO2, thovanadate. Total protein was quantitated by a modifie Bradford assay (Bio-Rad Laboratories, Hercules, CA), and 30 to 50 yg of denatured protein was separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and electroblotted onto immobilon PC (SandS, Keene, NH).

Human alkyltransferase was detected using the monoclonal antibody (MoAb) MT3.1 (provided by D. Bigner and T.

Brent). Signal was detected using a goat-antimouse IgG conjugated to horseradish peroxidase and developed via chemiluminescence using the ECL kit from Amersham according to the manufacturer's instructions. The human AGT was detected by immunostaining with the monoclonal antibody mT3.1 and horseradish peroxidase conjugated secondary antibody. Chemiluminescent signal was detected usir. g the ECL) cit (Amersham). AGT activity was correlated to densitometric band intensity using a standard curve generated from transduced K562 cells expressing high levels of MGMT.

AGT Assay AGT activity was measured as [³H]-methyl groups removed from [3 HI o'-methylguanine present in [3H]-methylnitrosourea treated alkylated calf thymus DNA.

The alkylated [3H-methyl] o6 methylguanine and N7 methylguanine bases were separated by HPLC and quantifie by liquid scintillation. AGT activity was expressed as fmol 06 mG removed/yg DNA or fmol 06removed/mg protein.

PCR Provirus Analysis DNA was isolated from single colonies selected from methylcellulose plates with a standard Proteinase

K/Triton X-100 method. A 152 bp human MGMT fragment and a 290 bp human dystrophin fragment was amplifie and the fragments were separated on a 2% agarose gel and detected by either ethidium bromide stain or Southern blot.

RT PCR Analysis Purifie total RNA or RNA prepared using the standard Proteinase K/Triton S-100 method was digeste extensively with DNaseI according to manufacturer's instructions. Reverse transcription and PCR was performed using the RNA PCR kit (Perkin-Elmer-Cetus, Norwalk, CT). In the presence of reverse transcriptase, a 497 bp fragment was amplifie using the sense primer (5'TGGTACCTCACCCTTACCGAGTC3') containing sequences of MFG proviral backbone and the anti-sense primer (5'ACACCTGTCTGGTGAACGACTCT3') specific to human MGMT.

Helper Virus Assay Viral supernatants from amphotropic producer cells were used to transduce K562 and NIH 3T3 cells.

Supernatants from these cells were then co-cultured with fresh NIH 3T3 cells. The latter were then analyzed for the presence of proviral sequences by PCR and supernatants were used to infect the NIH lac cells.

Since the latter contain proviral sequences, the presence of virus in the supernatant could be detected by supernatant transmission of lac + virus to naive NIH-3T3 cells. NIH 3T3 cells were cultured for one month at which time supernatant was reassayed for helper virus.

EXAMPLE FIVE-RESULTS Expression of AGT and BG resistance In CHO cells AAGT activity was compare to wtAGT in transfected CHO cells byenzyme assay and western blot. AGT activity was 3.5 fmol/yg DNA in aMGMT transfected CHO cells

compare to 84 fmol/yg DNA inwtMGMT transfectants.

Differences in AGT activity were greater than the differences observe by western blot. However, there was comparatively more immunoreactive protein determined by western blot than suggested by AGT activity. In CHO cells, AAGT was much more resistant to BG than wtAGT.

AMGMT transfected cells treated with 25UM BG had an IC50 for inhibition of AGT of approximately 30yM compare to <0. 1yM for cells transfected with wtMGMT. (Figure 12A).

Titer MFG-MGMT titer from Am-12 supernatant was estimated by immunohistochemical detection of infecte K562 cells. A clone with a titer of 5x105 AGT positive cfu/ml was used for further experiments. Of note, AGT was nuclear, indicating that the mutant protein retained its nuclear localization.

Expression and Drug Resistance In K562 cells K562 cells were retrovirally infecte by c-culture with wtMGMT or aMGMT Am-12 producer cells. AGT activity and resistance to BG were compare in the two cell cultures without prior selection. Mean AGT levels were 25.7 fmol/pg DNA in wtMGMT transduced cells, 2.7 fmol/ygDNA in zMGMT transduced cells and 0 fmol/Ag DNA in uninfected cells. The IC50 for BG inactivation of AGT was < 0.1 UM compare to approximately 18yM In wt and AMGMT transduced cells respectively. At 5 HUM BG AAGT containing cells retainedo90k activity while wtAGT was undetectable (Figure 12B). To determine whether ^AGIT expression could increase tolerance to BG and BCNU, transduced K562 cells were exposed to 10yM BG and various concentrations of BCNU and plated in methylcellulose. Clonogenic ^MGMT transduced cells were significantly more resistant to the drug combination than wtMGMT transduced and untransduced cells (Figure

12A). The BCNU ICso was 11.3 vs 4 vs 1.3 UM and the ICgo was > 30 vs 16 vs 5 ßM, for aMGMT, wtMGMT and untransduced cells, respectively. Furthermore, aMGMT cells maintained 20k clonogenic survival at 30 yM BCNU and BG compare to < 1% of cells transduced with wtMGMT. To assess proviral integration, individual colonies were subjected to PCR analysis and a 152bp MGMT fragment was amplifie in 22 of 33 colonies (67%). AGT immunoreactive protein detected in pooled colonies from wtMGMT infecte cells was 10 fold higher than in colonies from AMGMT transduced cells.

Expression and Drus Resistance in human CD34 cells The non-adherent cell count in co-cultures of human CD34'cells and MFG-MGMT or MFG-lac producers increased 5-fold over 96 hours. After co-culture, cells were treated with BCNU alone or with BG and plated in methylcellulose. oMGMT transduced CD34'cells had only a small increase in resistance to BCNU alone compare to lac Z transduced cells. However, after pretreatment with 10yM BG to deplete wtAGT, a striking resistance to BCNU was observe in AMGMT transduced CD34'cells (Figure 12B). The divergence between the lac Z transduced and AMGMT transduced clonogenic progenitor cell survival increased as the dose of BCNU increased. Thus, relative to the survival of cells transduced with lac Z, the survival of QMGMT was 73.7_10.6 at the ICso, 49.0+24.2 at the IC90 and approximately 25% were resistant to lOgM BG and 10yM BCNU, a dose which killed greater than 99% of lac Z transduced hematopoietic progenitor cells.

Individual progenitor colonies (BFU-E and CFU-GM) were analyzed for proviral integration and efficiency of transduction was 76% as determined by PCR. aMGMT mRNA expression was detected in pooled colonies by RT-PCR.

AGT levels detected by western blot were five-fold higher in aMGMT pooled progenitor colonies than lac progenitors and enzyme activity was increased two-fold.

A drug resistance gene selectively expressed in hematopoietic cells may provide a distinct therapeutic avantage during chemotherapy exposure for the treatment of cancer. This shows the utility of AAGT as a mutant drug résistance protein, resistant to inactivation by BG and is thus capable of protecting cells from the combination of BG and BCNU. We have shown that the retroviral vector MFGMGMT transmit AAGT expression into K562 cell lines and into human CD34'cells. These cells become remarkably resistant to the ccmbination of BG and BCNU compare to cells expressing wtAGT at higher levels.

Other disclosures regarding studies with cell lines which lack endogenous expression of AGT and are normally quite sensitive to BCNU have shown that over expression of MGMT results in resistance to BCNU. CD34+ cells express low but detectable levels of AGT and are more resistant to BCNU than these cell lines. Transduction of CD34+ cells with wtMGMT had little effect on BCNU resistance. Furthermore, both CD34+ cells and tumor cell lines with very high levels of wtAGT activity can be sensitized to BCNU after inactivation of AGT by BG. The combination of BG treatment with AMGMT gene transfer shows the feasibility of selectively protecting hematopoietic cells during systemic treatment with the BG and BCNU combination to an even greater extent than over expression of MGMT would protect these cells from BCNU alone.

In contrast to gene transfer of wtMGMT into CD34+ cells, transduction of AMGMT into CD34+ cells resulted in marked enhancement of clonogenic survival after BG and BCNU. These results relied on efficient gene transfer into CD34+ derived colonies (over 70'-.). Over 300 of colonies appeared very resistant to BG and BCNU. These data show that the aMGMT cDNA is a better candidate drug resistance gene than wtMGMT because the relative protection seen with AMGMT and BG and BCNU is much

greater than we observe with wtMGMT and BCNU alone at similar levels of gene expression. It is possible that higher expression of MGMT in human CD34'cells would partially protect cells from BG and BCNU as was observe with K562 cells. However, this is unlikely because transduction of cells expressing endogenous AGT results in less enhancement of BCNU resistance as originally seen and because the protection noted after wtMGMT gene transfer gene into murine hematopoietic cells was modest as well. Allay, et al. (1995, Blood :85 3342-3351) and Harris, et al. (1995,. Amer. Assoc. Can. Res. :36 419) have found less than a two-fold increase in BCNU resistance after retroviral mediated gene transfer of wtMGMT. In contrast, these data results indicate remarkable protection to cells expressing AMGMT and treated with BG and BCNU.

Another approach described by Harris (1995, Proc.

Amer. Assoc. Car.. Res. :36 419) is transfer of the bacterial ada gene into murine hematopoietic progenitors. These authors noted ada increased the resistance to BG and BCNU in vivo. The degree of resistance was not as high as disclosed in this example even though the bacterial protein is more resistant to BG than the AMGMT protein. Three caveats may explain the differences. First, there is evidence that the bacterial protein is not well nuclear localized and may not be an efficient DNA repair protein even if over expressed.

Second, mouse AGT has a higher ICso for BG than human AGT so that mouse cells are more resistant to the combination of BG and BCNU than human cells.

Consequently, the degree of protection noted can not be directly converted to that expected in human CD34+.

Third, regarding in vivo application, there is a significant possibility that an immune response would develop against the bacterial AGT that would not be expected following introduction of aAGT, with only a single amino acid change, into cells in vivo.

This example does notdirectly show gene transduction into cells more primitive than the BFU-E.

While targeting such primitive cells is within the scope of the present invention, transduction of very early progenitors may not be a prerequisite for successful introduction of this gene into humans for the purpose of protection from myelosuppression. Most chemotherapeutic regimens are administered for 2-6 months, suggesting that this would be the time window needed for persistent gene expression. While our earlier murine studies document gene expression out to 1 year, successful gene therapy may involve techniques that transduce CD34+ cells which, because of their state of differentiation repopulate the human marrow after every treatment cycle, contributing to hemtopoiesis for a period perhaps as short as 1-6 months. After this, if there is loss of the transduced cells due to entrance of more committed progeny into the cycle, it would not impact on treatment with other agents or patient outcome. AMGMT expressing cells might be significantly enriched with each cycle of drug treatment even if only a small proportion of cells are initially transduced, ameliorating the myelosuppression and preventing cumulative toxicity.

Furthermore, since use of nitrosoureas and other agents which attack at 0'ouf guanine may be associated with cumulative myelosuppression and eventual secondary leukemias, it may be postulated that after AMGMT gene therapy, these cells can be protected directly by gene transduction or indirectly by maintenance of a higher white count, decreasing the stimulus for their proliferation.

Use of MGMT as a drug resistance gene in CD34+ cells is fundamentally different than the use of other genes for this purpose. MDR is expressed at moderately high levels in early hematopoietic progenitors although over expression has been shown to increase drug resistance of CD34'cells in vitro. Likewise, aldehyde dehydrogenase

is expressed at high levels in early hematopoietic progenitors generating a relative resistance to cyclophosphamide. Over expression of DHFR protects cycling cells but there is no evidence that it protects early progenitors. In contrast, AGT is expressed at low levels in early hematopoietic progenitors and these cells are susceptible to cumulative cytotoxicity and secondary transformation into leukernia. The over expression of MGMT protects against both of these events.

The G156A AGT mutant appeared less active than the wt protein in CHO cells, K562 and normal CD34'cell cultures based on a comparison of the levels of immunoreactive protein to enzyme activity. While this difference was quite pronounced in the cell lines it was present to a much less extent in the r. ormal CD34'cells.

The mutant protein may be less stable than the wt protein. Perhaps it fails to maintain its active conformation or perhaps the inactive form is not degraded through the ubiquitin pathway as rapidly as the wt protein. In fact, finding the ^AGIT to be such potent protection from cytotoxicity from BG and BCNU is all the more remarkable because of the relatively low levels of enzyme activity in both the cell lines and the CD34+ cells.

Therefore, over expression of AMGMT transmit resistance to the chemotherapeutic combination BG and BCNU, providing a greater therapeutic avantage than does transduction of the wtMGMT protect against BCNU alone. This, plus the continue successful development of BG in clinical trials to potentiate the efficacy of BCNU in the treatment of cancer, show that gene therapy with AMGMT as disclosed throughout this specification will provide significant, selective protection of hematopoietic cells in the clinical setting.

SMARY A major limitation of previous drug resistance gene transfer studies is the inability to simultaneously enrich for transduced progenitors while selecting against tumor cells. The present invention relates to a novel gene therapy approach to overcome this hurdle, <BR> <BR> <BR> by utilizing the G156A mutant form of MGMT (MGMT).

Since this protein has been shown to be 240-fold resistant to inactivation by the AGT inhibitor BG which potentiates BCNU cytotoxicity by depleting endogenous <BR> <BR> <BR> AGT activity, we anticipate that AMGMT transduced hematopoietic progenitors will selectively survive BG and BCNU treatment while unmodified progenitors and tumor cells which express wtMGMT should not tolerate the combination chemotherapy.

It has been previously reporte that transfer of AMGMT into human CD34'cells provides increased resistance to combination BG and BCNU compare to untransduced cells (Reese, J. S., Kob, 0. N., Lee, K., Liu, L., Allay, J. A., Phillips, W. P., and Gerson, S.

L., Retroviral transduction of a mutant MGMT into human CD34 cells confers resistance to 06-benzylguanine plus BCNU., Proc. Natl. Acad. Sci. USA, 93: 14088-14093, 1996), and that this differential survival is significantly improved over that observe after transfer of wtMGMT. As shown in Figure 1, murine hematopoietic progenitors obtain a similar degree of BG and BCNU <BR> <BR> <BR> resistance after transfer of AMGMT as human CD34+ cells.<BR> <BR> <BR> <BR> <BR> <BR> <P>A 5-fold increase in the BCNU ICso was observe in aMGMT<BR> <BR> <BR> <BR> <BR> transduced murine progenitors compare to lacZ transduced cells after pretreatment with BG. The differential survival is much improved over that observe after wtMGMT gene transfer in murine progenitors which resulted in a less than two fold increase in the IC50 following BCNU treatment alone (Allay, J. A., Dumenco, L. L., Koc, O. N., Liu, L., and

Gerson, S. L., Retroviral transduction and expression of the human alkyltransferase cDNA provides nitrosourea resistance to hematopoietic cells, Blood, 85: 3342-3351, 1995).

Furthermore, we have demonstrated that <BR> <BR> <BR> <BR> transplantation of AMGMT transduced hematopoietic progenitors into lethally irradiated mice leads to protection from BG and BCNU mediated toxicity. Our results demonstrate significant differential survival between AMGMT and lacZ mice (95k vs. 23-*6) following multiple treatments of 30 mg/kg BG and a range of 10-20 mg/kg BCNU as shown in Figure 5 and Table 2. The differential animal survival following drug treatment observe in our studies is improved over the results reporte by Maze (Maze, R., Carney, J. P., Kelley, M. R., Glassner, B. J., Williams, D. A., and Samson, L., Increasing DNA repair methyltransferase levels via bone marrow stem cell transduction rescues mice from the toxic effects of 1, 3-bis (2-chloroethyl)-l- nitrosourea, a chemotherapeutic alkylating agent, Proc. Natl. Acad.

Sci. USA, 93: 1996),206-210, using wtMGMT and by Harris with ada (Harris, L. C., Marathi, U. K., Edwards, C. C., Houghton, P. J., Srivastava, D., Vanin, E. F., Sorentino, B. P., and Brent, T. P., Retroviral transfer of a bacterial alkyltransferase gene into murine bone marrow protects against chloroethylnitrosourea <BR> <BR> <BR> <BR> cytotoxicity, Clinical Cancer Res, l: 1995).1359-1365, Maze reporte 92% MGMT transplanted mice compare to 53% of mock transplanted mice survive weekly doses of 40 mg/kg for 5 weeks, a relatively high dose of drug considering the potential multiorgan cumulative toxicity observe with BCNU treatment. Harris noticed a narrow therapeutic window in mice transplanted with ada transduced hematopoietic progenitors. While little toxicity was observe between mock and ada mice when treated with 30 mg/kg BG plus 10 mg/kg BCNU, both ada and mock transplanted mice were similarly sensitive to

30 mg/kg BG and 15 mg/kg BCNU. Significant differential survival between ada and mock transplanted mice was observe only when treated with two doses of 30 mg/kg BG and 12.5 mg/kg BCNU. The avantage that is observed <BR> <BR> <BR> with the BG resistant AMGMT over the bacterially derived ada protein may be due to the difficulty of ada to translocate to the eukaryotic cell nucleus in the absence of a defined nuclear localization signal (Dumenco, L. L., Warman, B., Hatzoglou, M., Lim, I. K., Abboud S. L., and Gerson S. L., Increase in nitrosourea resistance in mammalian cells by retrovirally mediated gene transfer of bacterial 06alkylguanine-DNA alkyltransferase, Cancer Res., 49: 6044-6051, 1989.).

The present invention also demonstrates that BG and <BR> <BR> <BR> <BR> BCNU administration after AMGMT gene transfer is an effective means to enrich for transduced cells in a murine model. The degree of selection was much greater than was observe with wtMGMT gene transfer and BCNU administration alone. Specifically, the BCNU IC50 in CFU-C from mice transplanted with wtMGMT and given injections of BCNU was two-fold higher than that observe in untreated wtMGMT mice (19 vs. 10 yM) and four-fold higher than in mice transplanted with lacz transduced bone marrow progenitors (5 yM) (Allay, J. A., Davis, B. M., and Gerson, S. L., In-vivo enrichment of MGMT transduced murine hematopoietic progenitors by BCNU treatment of MGMT transplanted mice, in press, Exp. <BR> <BR> <BR> <P> Hem., 1997.). In contrast, CFU derived from the bone<BR> <BR> <BR> <BR> <BR> <BR> marrow of mice transplanted with AMGMT and given injections of BG and BCNU had a BCNU ICso of 68 yM when <BR> <BR> <BR> the cells were exposed to drug in vitro after 20yM BG, which was greater than 10 fold higher than untreated <BR> <BR> <BR> aMGMT mice (6.5 yM) and treated or untreated lacZ mice (6.2 yM).

The BCNU doses necessary to enrich for transduced cells in mice transplanted with wtMGMT transduced bone marrow progenitors ranged from 15-40 mg/kg (Allay, J. A.,

Davis, B. M., and Gerson, S. L., In-vivo enrichment of MGMT transduced murine hematopoietic progenitors by BCNU <BR> <BR> <BR> treatment of MGMT transplanted mice. in press, Exp.<BR> <BR> <BR> <BR> <BR> <P> Hem., 1997), compare to only 10 mg/kg of BCNU when<BR> <BR> <BR> <BR> <BR> given with 30 mg/kg of BG for aMGMT selection.

The use of lower doses of BCNU in combination with BG may be advantageous over conventional dose BCNU treatment alone, with the expectation of less systemic toxicity, especially if that toxicity, including pulmonary fibrosis, is due to total BCNU dose and not strictly the level of DNA adducts formed at 06 of guanine. Myelosuppression is still anticipated to be the dose limiting toxicity of BG and BCNU treatment, <BR> <BR> <BR> reiterating the importance of aIdGMT mediated protection.

In the experiments reporte here, blood was obtained 7 weeks after the last drug treatment at which time there was complete recovery of cellularity, and it is not surprising that mean blood and bone marrow cell counts <BR> <BR> <BR> in lacZ and AMGMT mice were similar. However, in subsequent experiments, there was assessed myelosuppression and there was an observe decrease in bone marrow and peripheral blood cellularity in normal mice 2-12 days after BG and BCNU administration, while <BR> <BR> <BR> mice infused with AMGMR bone marrow retained peripheral blood and bone marrow counts which were similar to non- drug treated control animals.

The ability to enrich for AMGMT expressing transduced progenitors may be advantageous over transduction of other drug resistance genes including <BR> <BR> <BR> MDR1 and DHFR. While expression of AAGT in 60% of the bone marrow nucleated cell population was detected after in vivo BG and BCNU administration, Podda et. al. <BR> <BR> <BR> <BR> reporte enrichment to only 71 of MDR1+ peripheral blood mononuclear cells following in vivo paclitaxel administration (Podda, S., Ward, bol., Himelstein, A., Richardson, C., de la Flor-Weiss, E., Smith, L., Gottesman, M., Pastan, I., and Bank, A., Transfer and

expression of the human multiple drug resistance gene into live mice, Proc. Natl. Acad. Sci. USA, 89: 9676, 1992). This may be due to the fact that there is no selection against cells expressing high endogenous levels of the non-specific drug efflux pump p-glycoprotein as is present in early hematopoietic progenitors (Chaudhary, P. M., and Roninson, I. B., Expression and activity of P-glycoprotein, a multidrug efflux pump, in human hematopoietic stem cells., <BR> <BR> <BR> <BR> Cell, 66: 1991;85-94, Drach, D., Zhao, S., Drach, J., Mahadevia, R., Gattringer, C., Huber, H., and Andreeff, M., Subpopulations of Normal Peripheral Blood and Bone Marrow Cells Express a Functional Multidrug Resistant Phenotype, Blood, 80: 2729-2734,1992). In cancer patients, this would correlate with enrichment for both transduced hematopoietic cells as well as non- <BR> <BR> <BR> <BR> transduced early hematopoietic progenitors and MDR1 positive tumor cells, leading to little therapeutic avantage after drug selection. Similarly, since methotrexate is cytotoxic to proliferating hematopoietic progenitors but not quiescent early progenitors (Gerson, S. L., Page, P. L., Hartwell, B. L., and Robinson S. H., Altered growth characteristics of murine hematopoietic cells induced cytotoxic drugs, Stem Cells, 2: 266-279, 1982), it is not clear that the bone marrow can be repopulated completely with DHFR positive cells after transplantation of DHFR transduced marrow cells. Recent experiments by Spencer et. al. have shown that transfer of the L22Y DHFR cDNA into murine hematopoietic cells protects against myelosuppression after trimetrexate treatment, suggesting that at least short term protection can be achieved using this gene (Spencer, H. T., Sleep, S. E., Rehg, J. E., Blakley, R. L., Sorrentino, B. P. A gene transfer strategy for making bone marrow cells resistant to trimetrexate, Blood, 87: 2579-87,1996). However, enrichment for stem cells containing the provirus may be limited when using MDR1

or DHFR because the drugs used with these genes are most effective against cells progressing through the cell cycle and are much less cytotoxic to quiescent stem cells. In contrast, BCNU mediated DNA crosslinking is permanent if the precrosslink lesion is not repaire and cytotoxicity will be observe when quiescent stem cells enter the cell cycle. aMGMT may have better potential to act as a dominant selectable marker for selection of transduced <BR> <BR> <BR> <BR> hematopoietic stem cells than MDR1 due to low endogenous expression of MGMT in primitive hematopoietic cells (Gerson, S. L., Phillips, W. P., Kastan, AMGMT M. B., Dumenco, L. L., and Donovan, C., Human CD34 hematopoietic progenitors have low, cytokine- <BR> <BR> <BR> unresponsive o'-alkylguanine-DNA alkyltransf erase and are<BR> <BR> <BR> <BR> <BR> <BR> <BR> sensitive to 06-benzylguanine plus BCNU, Blood, 88: 1649- 55,1996), predicting that most BG and BCNU resistant <BR> <BR> <BR> <BR> cells will be transduced with aMGMT. I'his cannot be<BR> <BR> <BR> <BR> <BR> <BR> <BR> predicted for MDR1 mediated drug resistance due to high endogenous expression of p-glycoprotein in the target primitive hematopoietic cells (23,24) (Chaudhary, P. M. and Roninson, I. B., Expression and activity of P- glycoprotein, a multidrug efflux pump, in human <BR> <BR> <BR> <BR> hematopoietic stem cells, Cell, 66: 1991;85-94, Drach, D., Zhao, S., Drach, J., Mahadevia, R., Gattringer, C., Huber, H., and Andreeff, M. Subpopulations of Normal Peripheral Blood and Bone Marrow Cells Express a Functional Multidrug Resistant Phenotype, Blood, 80: 2729-2734,1992). Unlike paclitaxel, BCNU acts as a stem cell toxin, which would lead to enrichment for <BR> <BR> <BR> <BR> AMGMT transduced hematopoietic stem cells after BG and BCNU administration. This would increase the proportion <BR> <BR> <BR> <BR> of early progenitor and stem cells expressing aMGMT, protecting both early and late progenitors from BCNU <BR> <BR> <BR> <BR> mediated toxicity. Further avantages of ^MGMT over MDR1 become apparent when considering construction of <BR> <BR> <BR> <BR> bicistronic vectors. The smaller size of AMGMT (625bp)

compare to MDRl (=3.9kb) allows for the inclusion of a second gene of larger size without significantly disrupting the transcription, stability or packaging ability into virions of the retroviral RNA.

The present applications demonstration of stable expression of AMGMT in bone marrow cells and continue enhanced resistance to BG and BCNU in CFU 23 weeks after transplant indicates that there is persistent transgene expression in early hematopoietic progenitors, and likely in stem cells. Furthermore, a recent report was made of efficient transduction and expression of AMGMT in human long term culture initiating cells (LTC-IC) demonstrate that early progenitors express proviral <BR> <BR> <BR> <BR> AMGMT and will survive after receiving normally cytotoxic levels of BG and BCNU (Koç O. N., Reese, J.

S., Szekely, E. M., Lee, K., and Gerson, S. L., FL dependent human peripheral blood (pb) long term culture initiating cell (LTCIC) expansion leads to retroviral transduction with mutant MGMT and resistance to o6- <BR> <BR> <BR> <BR> <BR> benzylguanine and BCNU, Blood Suppl, 88: 1996)1086, The modest decrease in the proportion of CFU with high level BG and BCNU resistance in mice sacrifice 23 weeks post transplant may be due to a combination of factors.

These include the absence of selective pressure for 17 weeks which may have allowed less resistant cells to proliferate and/or a reduction in proviral expression after progenitor differentiation along various hematopoietic lineages due to the induction of stage- dependent transcriptional controls. MFG is a MoMuLV derived virus which is thought to express less efficiently in myeloid cells than lymphoid cells (Baum, C., Hegewisch-Becker, S., Eckert, H. G., Stocking, C., and Ostertag, W., Novel retroviral vectors for efficient <BR> <BR> <BR> <BR> expression of the multidrug resistance (MDR-1) gene in early hematopoietic cells., J. Virol., 69: 7541-7, 1995), and it is possible that the lack of drug selection leads to decreased gene expression.

It is believed that the enhanced differential<BR> <BR> <BR> <BR> <BR> protective effect observe with AMGMT and BG and BCNU treatment would be particularly important when treating tumors that express high levels of AGT. In this scenario, it could be anticipated that treatment with BG would inactivate the wtAGT in the tumor and potentiate the cytotoxicity of low dose BCNU, whereas hematopoietic <BR> <BR> <BR> cells protected with AMGMT would be expected to exhibit no significant toxicity. Although there were initial concerns regarding solubility and biochemical efficacy of BG, this has not been born out in clinical trials which formulate BG in PEG 400, as was done in the present invention. These studies demonstrate AGT inhibitory levels of BG and its metabolite, 8-oxo-BG, for many hours after infusion of 80 mg/m2 of BG in PEG 400 without evidence of toxicity (Spire, T. P., Willson, J. K. V., Haaga, J., Hoppel, C. L., Liu, L., Majka, S., <BR> <BR> <BR> and Gerson, S. L., o6-benzylguanine and BCNU: Establishing the biochemical modulatory dose in tumor tissue for 06-alkylguanine-DNA-alkyltransferase directe DNA repair, Froc. ASCO, 15: 362,1996.). Therefore, the present invention is directe to establishing that <BR> <BR> <BR> the transfer of oMGMT into hematopoietic progenitors and subsequent BG and BCNU administration should enrich for <BR> <BR> <BR> BG resistant cells which express AMGMT while sensitizing tumor cells to BCNU.

Taken together, the data suggest that MFG-MGMT gene transfer into hematopoietic progenitors may provide therapeutic benefit for patients in clinical trials.

Combination BG and BCNU treatment reduces the BCNU dose necessary to achieve toxicity in tumor cells expressing wild type MGMT, due to BG mediated inactivation. Cells transduced with AMGMT are protected from BG and BCNU toxicity, since the mutant form of the protein is resistant to inactivation, and these cells maintain long term expression in vivo. In patients, this therapy has the potential to protect the marrow from the cumulative

effects of BCNU, primarily myelosuppression and perhaps the risk of secondary leukemic transformation. In summary, transplantation of aMGMT transduced<BR> hematopoietic progenitors followed by in vivo BG and BCNU administration leads to enrichment for transduced cells, long term expression of aAGT and a dramatic survival avantage in mice to combination BG and BCNU.

Whereas particular embodiments of this invention have been described above for the purposes of illustration, it will be evident to those persons of ordinary skill in the art that numerus variations of the details of the present invention may be made without departing from the invention as defined in the appende claims.