Suicide gene transduction sensitizes murine embryonic and human mesenchymal stem cells to ablation on demand-- a fail-safe protection against cellular misbehavior.

Stem cells and their progeny constitute a potential resource for replacing damaged tissues or supplying missing functions, but also pose a threat of aberrant behavior, including neoplastic growth or immunopathology. Suicide genes introduced into these cells before transplantation might provide a means of addressing this threat by permitting the ablation of the cells if they subsequently misbehave. Retroviral transduction of the E. coli gpt and herpes thymidine kinase (HSVtk) suicide genes was used to determine the degree to which stem cells could be sensitized to the prodrugs 6-thioxanthine (6TX) and ganciclovir (GCV) respectively, and whether this sensitivity could persist over many cell generations. The ES-E14TG2a murine embryonic stem cell line was rendered sensitive to quantitative ablation at prodrug concentrations well tolerated by untransduced cells (50 microM 6TX, 1 microg/ml GCV). The HSVtk gene also conferred GCV sensitivity on human mesenchymal stem cells and hematopoietic precursors derived from the murine cells, although ablation was not complete. Because ES-E14TG2a cells are deficient in the cellular enzyme HPRT, they are sensitive to hypoxanthine/aminopterin/thymidine (HAT). This property enhanced the persistence of chemosensitivity in gpt-transduced cells by permitting cells that lost 6TX sensitivity to be ablated with HAT.

Transduction of the herpes thymidine kinase gene into premalignant murine mammary epithelial cells renders subsequent breast cancers responsive to ganciclovir therapy.

Drug sensitivity ("suicide") genes can sensitize cancer cells to chemotherapy, but therapeutic use of these genes is limited by difficulties in delivering them to all areas of established cancers. An alternative strategy entails preemptive introduction of suicide genes into tissues at risk for cancer, thereby imparting drug sensitivity as a clonal property to cancers arising from sensitized cells. To test the preemptive approach, a retroviral vector was used to transduce the herpes thymidine kinase gene into the TM4 line of preneoplastic murine mammary epithelial cells to yield a clonal subline sensitized to the guanosine analog ganciclovir. Ganciclovir therapy of tumors that arose from the transduced cells retarded tumor growth and induced durable regressions in 7/20 mice; ganciclovir was ineffective against control tumors. The results imply the possibility of reducing cancer lethality by actions taken before cancers arise.

Escherichia coli gpt gene sensitizes rat glioma cells to killing by 6-thioxanthine or 6-thioguanine.

Genes that encode enzymes that convert inactive "prodrugs" into anticancer metabolites may be therapeutically useful against brain tumors. Unlike other genes tested to date in brain tumor models, the Escherichia coli gpt gene is unique in that it not only sensitizes cells to the prodrug 6-thioxanthine (6TX) but also encodes resistance to a different regimen (mycophenolic acid, xanthine, and hypoxanthine), thus providing a means to select for gpt-positive cells. In the present study, rat C6 glioma cells were infected with a retrovirus vector that transduces this gene. A clonal line (C6GPT-7) was derived that exhibited significant 6TX susceptibility in vitro with an ID50 of 2.5 mumol/L, whereas 50% growth inhibition of parental C6 cells was not achieved at concentrations tested (up to 50 mumol/L). This line also exhibited significant sensitivity to 6-thioguanine (6TG), with an ID50 of 0.05 mumol/L, whereas 50% growth inhibition of parental C6 cells was achieved at 0.5 mumol/L. In a "bystander" assay, C6GPT-7 tumor cells efficiently transferred 6TX sensitivity to C6 cells at ratios as low as 1:9 (C6GPT-7:C6). This in vitro bystander effect was abrogated when C6GPT-7 and C6 cells were separated by a microporous membrane, suggesting that it was not mediated by highly diffusible metabolites. In vivo both 6TX and 6TG significantly inhibited the growth of subcutaneously transplanted C6GPT-7 cells but not that of C6 cells in athymic mice. In an intracerebral model, both 6TX and 6TG exhibited significant antiproliferative effects against tumors formed by C6GPT-7 cells. These findings provide a basis for exploring further gene therapy strategies based on in vivo transfer of the E coli gpt gene to provide chemosensitivity against 6TX and 6TG.

The "bystander effect": tumor regression when a fraction of the tumor mass is genetically modified.

Tumor cells expressing the herpes simplex virus thymidine kinase (HSV-TK) gene are sensitive to the drug ganciclovir (GCV). We demonstrate here that HSV-TK-positive cells exposed to GCV were lethal to HSV-TK-negative cells as a result of a "bystander effect." HSV-TK-negative cells were killed in vitro when the population of cultured cells contained only 10% HSV-TK-positive cells. The mechanism of this "bystander effect" on HSV-TK-negative cells appeared to be related to the process of apoptotic cell death when HSV-TK-positive cells were exposed to GCV. Flow cytometric and electron microscopic analyses suggested that apoptotic vesicles generated from the dying gene-modified cells were phagocytized by nearby, unmodified tumor cells. Prevention of apoptotic vesicle transfer prevented the bystander effect. The toxic effect of HSV-TK-positive cells on HSV-TK-negative cells was reproduced in an in vivo model. A mixed population of tumor cells consisting of HSV-TK-positive and HSV-TK-negative cells was inoculated s.c. into mice. Regression of the tumor mass occurred when the inoculum consisted of as few as 10% HSV-TK-expressing tumor cells. The bystander effect was also demonstrated in i.p. tumor studies. Initial experiments demonstrated that prolonged survival (> 70 days) occurred when a mixture containing 50% HSV-TK-positive and 50% HSV-TK-negative cells was injected i.p. followed by GCV treatment. Further, survival was prolonged for mice with a preexisting HSV-TK-negative i.p. tumor burden by injecting HSV-TK-positive cells and GCV. These results suggest that genetic modification of tumor cells may be useful for developing cancer therapies.

Retrovirally transduced Escherichia coli gpt genes combine selectability with chemosensitivity capable of mediating tumor eradication.

"Suicide genes" encoding exceptional sensitivity to chemotherapeutic agents can potentially improve the selectivity of cancer therapy. As a component of a retroviral gene therapy vector, a suicide gene might also improve the safety of gene therapy by permitting the subsequent ablation of transduced cells exhibiting neoplastic or other aberrant behavior. An extra gene, however, cannot easily be added to vectors already carrying a therapeutic gene and a selectable drug resistance marker without compromising gene expression. To circumvent this obstacle, we have investigated a retrovirally transduced Escherichia coli gpt gene on the basis of evidence that it might subserve a dual sensitivity/resistance function. A gpt vector was used to transduce the gene into murine K3T3 sarcoma cells. In vitro, gpt-positive K3T3 clones could be selected on the basis of resistance to a regimen containing mycophenolic acid and xanthine; the same clones were 18 to 86 times as sensitive to 6-thioxanthine (6TX) as their gpt-negative counterparts. In mice, systemic 6TX therapy induced durable regressions in 19/20 gpt-positive K3T3 sarcomas without affecting gpt-negative tumors. These results indicate that selectability and in vivo chemosensitivity can be expressed in the same cell population from a single retrovirally transduced gene and imply the additional possibility of fusing the gpt gene with a therapeutic gene to create vectors expressing three important functions from a single gene.

An experimental model of retrovirus gene therapy for malignant brain tumors.

Recent research using rodent models of central nervous system gliomas indicates that a combination of gene transfer and drug treatment may be successful in killing tumor cells. In the present study, a mouse fibroblast-derived packaging cell line, psi 2, which releases a replication-defective retrovirus vector bearing the herpes simplex virus type 1 (HSV)-thymidine kinase (TK) gene, was grown with rat C6 tumor cells in the presence and absence of wild type Moloney murine leukemia virus (MoMLV). Consequently, tumor cells became sensitive to ganciclovir, which is selectively converted to a toxic nucleotide analog by HSV-TK. This killing effect was more effective in the presence than in the absence of wild type retrovirus both in culture and in subcutaneous tumors in nude mice. Tumors regressed in vivo and failed to regrow over a subsequent 10-day observation period after combined treatment with packaging cells, wild type MoMLV, and ganciclovir. This killing effect may be augmented by the ability of the helper retrovirus to package the vector in tumor cells and thus extend delivery of the HSV-TK gene to more tumor cells. This represents significant improvement in tumor therapy in this model system as compared with helper-free systems previously reported by the authors and others. Although additional improvements in the therapy can be envisioned, this approach may prove useful in combination with current modes of therapy for these insidious and lethal tumors.

Gene therapy of malignant brain tumors: a rat glioma line bearing the herpes simplex virus type 1-thymidine kinase gene and wild type retrovirus kills other tumor cells.

Tumor cells infected with a retrovirus vector (VIK) containing the herpes simplex virus thymidine kinase (HSV-TK) gene can be selectively killed by treatment with nucleoside analogues, such as ganciclovir. To mediate delivery of the HSV-TK gene to "recipient" tumor cells, "donor" C6 rat glioma cells infected with the VIK vector (C6VIK) were superinfected with wild type Moloney murine leukemia virus (WT Mo-MLV). These modified donor cells (C6VIKWT) produced both wild type retrovirus and the VIK vector. In culture, C6VIKWT cells were 300-fold more sensitive to the toxicity of ganciclovir than were C6VIK cells, suggesting that the presence of wild type retrovirus contributed to the toxicity. Co-culture of C6VIKWT cells with the C6 subline, C6BAG, sensitized the latter to ganciclovir treatment. Nude mice inoculated subcutaneously with a mixture of C6VIKWT and C6BAG cells showed regression of subsequent tumors when treated with ganciclovir. The observations show that tumor cells modified in culture by infection with a retrovirus bearing the HSV-TK gene and wild type retrovirus are not only sensitive to ganciclovir, but can transfer this sensitivity to neighboring "naive" tumor cells in culture and in vivo.

A model for predicting the risk of cancer consequent to retroviral gene therapy.

Theoretical estimates of the risk of cancer resulting from accidental insertion of retroviral gene therapy vectors into oncogenically vulnerable genomic sites may prove an important supplement to experimentally derived data in estimating risk/benefit ratios for future gene therapy trials. We have approached risk assessment by considering either a single vector insertion event or a single natural mutation to be potentially oncogenic, should either occur in a cell that would otherwise end with one less than the total number of mutations required for frank neoplasia. Estimates of the relative probabilities of these two phenomena yield a relative risk assessment, which in conjunction with epidemiologic data on natural cancer frequencies can be converted into an assessment of absolute risk. This approach yields an estimated range of relative risk over 10 years of about 1.00000026 to 25 for cells bearing single copies of inserted vectors; the upper limit is higher for multiple copies. These estimates, if accurate, imply that small experimental human or animal gene therapy cohorts will rarely, if ever, manifest vector-related cancers and that more precise future risk assessments will require additional data on natural and vector-induced mutation rates.

Multiple transduction as a means of preserving ganciclovir chemosensitivity in sarcoma cells carrying retrovirally transduced herpes thymidine kinase genes.

The potential value of retroviral gene transfer as a means of targeting therapeutic genes to neoplastic cells is threatened by the tendency of occasional cells to lose transduced genes or their expression. To determine whether this threat could be reduced by transducing multiple copies of a therapeutic gene, we compared 8 murine sarcoma sublines carrying from 1 to 6 copies of a retrovirally transduced herpes simplex virus thymidine kinase gene, which sensitizes cells to ganciclovir (GCV). When variability consequent to differences in vector integration site was excluded, increased copy number was associated with an increase in GCV sensitivity and a major reduction in the frequency of GCV-resistant mutants. The results suggest a potential means of preserving the efficacy of future antineoplastic gene therapy strategies.

Curability of tumors bearing herpes thymidine kinase genes transferred by retroviral vectors.

Retroviral vectors constructed to contain the herpes simplex virus thymidine kinase (HSV-TK) gene were used for transduction of this gene into murine sarcoma and lymphoma cells to yield sublines susceptible in vitro to the cytotoxicity of ganciclovir, a drug specifically activated by HSV-TK. In vivo, ganciclovir induced complete, durable regressions in most mice bearing transplanted HSV-TK-positive sarcomas; its efficacy against lymphomas was only marginal, possibly because of their greater instability of gene expression. The results imply the potential value of an anticancer strategy entailing the prophylactic use of retroviral vectors to create tissue mosaicism for drug sensitivity.

Lymphoma regression induced by ganciclovir in mice bearing a herpes thymidine kinase transgene.

The dose limitations imposed on cancer chemotherapeutic agents by their lack of selectivity can, in theory, be circumvented by a strategy entailing the prophylactic insertion into hosts of drug-sensitivity genes that are acquired or expressed in some but not all cells. This strategy predicts that neoplasms arising from drug-sensitive cells might be safely treatable with tumor-eradicative drug doses because the presence of a modicum of drug-insensitive stem cells will protect vital tissues from lethal depopulation. To test this prediction, lymphomas were induced with Abelson leukemia virus in mice bearing a herpes simplex virus thymidine kinase (HSV-TK) transgene selectively expressed in lymphoid cells. Of 12 transgenic mice treated with the HSV-TK-specific substrate ganciclovir (GCV), 11 exhibited complete tumor regressions; 5 of these mice remained tumor-free over observation periods that exceeded 100 days. Among the lymphomas that recurred, most appeared to represent mutant subpopulations that were GCV-insensitive because they had lost HSV-TK, implying that independent insertion of multiple HSV-TK gene copies might provide a means of preventing recurrences. The results of this study demonstrate that chemosensitivity genes can enhance the efficacy of treatment in hosts who subsequently develop a neoplasm. While the use of a germ-line gene insertion model precludes direct human application, the results also imply the merits of exploring an alternative version of the strategy in which somatic insertion of chemosensitivity genes in mosaic fashion is used prophylactically to enhance the prospect that a subsequent tumor will respond to therapy.

An alternative to the magic bullet paradigm for specific cancer therapy.

To improve on current cancer therapies, which attack cells on the basis of their proliferative tendencies, much effort has been devoted to a search for properties of tumor cells that are tumor-specific rather than proliferation specific. Evidence from molecular genetic studies suggests, however, that most tumors may lack such properties. An alternative approach to therapy is described that is based on a property known to characterize the majority of human tumors; viz., a monoclonal origin. The strategy requires the prophylactic induction in tissues of mosaicism for genes dictating sensitivity or resistance to chemotherapeutic agents, and exploits the observation that any clone of cells arising in a mosaic tissue must inevitably differ from some other cells in the mosaic. Recent advances in genetic technology imply that the strategy is likely to be testable soon in animals, and that it may significantly improve the results of cancer therapy when a technology safe and efficient enough for its human implementation becomes available.

Tumor chemosensitivity conferred by inserted herpes thymidine kinase genes: paradigm for a prospective cancer control strategy.

The lack of highly exploitable biochemical differences between normal tissues and some tumors can theoretically be circumvented by a strategy utilizing gene insertion prophylactically to create tissue mosaicism for drug sensitivity, thereby ensuring that any tumor arising clonally will differ from part of the normal cell population. Elements of the strategy were tested with neoplastic BALB/c murine cell lines bearing the herpes thymidine kinase gene. Exposure to the herpes thymidine kinase-specific substrate 9-([2-hydroxy-1-(hydroxymethyl)ethoxy]methyl)guanine ablated the clonogenic potential of the cells in vitro, and administration of this drug to BALB/c mice bearing tumors produced by the cell lines uniformly induced complete regression of the tumors. The observed responses to therapy imply that the strategy may prove valuable when the genetic technology needed for its human implementation becomes available.

In vivo efficacy of HAT therapy against transplantable murine myelomas resistant to purine analogs.

The therapeutic efficacy of antineoplastic purine analogs can be jeopardized by the emergence of drug-resistant mutant subpopulations of tumor cells. To determine whether such mutant populations might be eradicable in vivo with the type of HAT combination (hypoxanthine + an antifolate + thymidine) known to be selectively cytotoxic to them in vitro, 2 thioguanine-resistant BALB/c murine myeloma lines were transplanted into BALB/c mice to produce tumors capable of progressive growth in the absence of therapy. Treatment of these mice with a modified HAT regimen induced permanent tumor regressions in 37/44 mice; the same treatment was ineffective against tumors produced by a non-mutant myeloma line from which one of the mutant sublines had been derived.

Nonspecific immunosuppressive factors in malignant ascites: further characterization and possible relationship to erythrocyte receptors of human peripheral T cells.

Nonspecific suppressor factor has been purified from malignant human ascites through partial delipidation, ammonium sulfate precipitation, ACA-22 gel filtration followed by goat anti-human IgG immunoadsorbent chromatography. Molecular weight estimated by SDS-PAGE under mild reducing conditions is 50,000 and 25,000 under vigorous reducing conditions. It inhibits PHA, Con A, and PWM responses of peripheral lymphocytes by 50% at microgram concentrations per milliliter in vitro and inhibits plaque-forming cell response to SRBC at 100 microgram/mouse in vivo. It also inhibits 2-way mixed lymphocyte reactions and natural killer cell activities. It inhibits SRBC rosette formation of peripheral T cells and appears to be an erythrocyte receptor of peripheral T cells.