No abstract

Many tumor cells do not express co-stimulatory molecules, and this may account, in part, for their poor ability to stimulate T cells directly. One strategy to enhance immune recognition would be to express such molecules on the tumor cell. Here, we show that expression of a member of the B7 family of co-stimulatory molecules by CMT93 murine colorectal tumor or 1735 murine melanoma cells resulted in a local antitumor response in immunocompetent mice. The antitumor effect was diminished in athymic nude mice, indicating that T cells played an important part in this response. The ability of the B7-expressing tumor cells to generate systemic protective immunity was investigated by excision of tumors that developed from the initial inoculation followed by rechallenge with parental tumor cells. CMT93 is a poorly immunogenic tumor and no significant systemic immunity was elicited by the expression of B7-1 in these cells. 1735 melanoma is a mildly immunogenic tumor. Unexpectedly, the systemic immunity obtained with 1735 tumors expressing B7-1 or B7-2 was weaker than that generated by parental 1735 cells (p < 0.001, stratified logrank test), even when coexpression of interferon-gamma in the B7-1 cells produced high levels of surface MHC class I expression. These results suggest that some caution is appropriate when considering the use of these molecules in the gene therapy of cancer.

The herpes simplex virus-thymidine kinase/ganciclovir (HSVtk/GCV) system produces both direct and immune-mediated tumor cell killing. Here, we compare the efficacy of HSVtk/GCV with cytokines, alone and in combination, on the tumorigenicity and immunogenicity of B16 cells. With respect to single gene modifications, only HSVtk/GCV, or high-level interleukin-2 (IL-2) secretion, completely prevented tumor growth, whereas granulocyte-macrophage colony-stimulating factor (GM-CSF) generated the best levels of long-term systemic protection. To augment both local killing and immune activation, we constructed bicistronic constructs that express HSVtk and a cytokine within the same cell. Co-expression of HSVtk with IL-2 or GM-CSF enhanced the local antitumor activity of any gene alone. In a tumor-prevention model, HSVtk killing, in an environment preprimed with GM-CSF, generated the best long-term immune protection. However, in a short-term therapy model, continued IL-2 expression was most effective against 3-day established tumors. This probably reflects differences in the activities of IL-2 and GM-CSF in generating short-term, nonspecific immune stimulation compared to long-term immunological memory, respectively. As a prelude to in vivo delivery experiments, we also demonstrated that these bicistronic cassettes can be packaged normally into retroviral (5 x 10(5) virus/ml from pooled populations) and adenoviral vectors (5 x 10(9) virus/ml) and function as predicted within virally infected cells. This family of bicistronic vectors can be used to stimulate synergy between suicide and cytokine genes, overcomes the problems of delivering two genes on separate vectors, and should allow easier preparation of vectors for the delivery of multiple genes to patients' tumor cells.

Summary.-Peripheral blood lymphocytes (PBL) were obtained from 13 patients and tumour-intrinsic lymphocytes (TIL) from 20 patients with colorectal cancer. The PBL were separated on a Ficoll-Isopaque gradient and the TIL by digestion of the tumour with collagenase-DNase. Both PBL and TIL were passed through nylonwool columns and the eluted cells were co-cultured for 2 h with 51Cr-labelled tumour cells from the same patient. If patients in whom spontaneous 51Cr release from the tumour cells was greater than 3300 were excluded, PBL showed cytotoxicity for the autoplastic tumour cells in 5/10 cases and TIL in 3/10 cases (NS). In 12 cases the cytotoxicity of the TIL was compared with that for TIL from the same tumour after the lymphocytes had been washed a further 6 times in Medium 199. Three effector: target (E/T) ratios, 5:1, 10:1 and 20:1, were used. The proportion of effector populations showing cytotoxicity was 2/12 for unwashed TIL and 9/12 for washed TIL (P<0.006). At the 5:1 E/T ratio the level of cytotoxicity was not significantly greater for washed TIL, but at the 10:1 ratio washed TIL showed significantly more cytotoxicity (P <0-025). At the 20:1 E/T ratio, a comparison was possible in 15 cases and the washed TIL again showed greater cytotoxicity (P<0-001).

Summary In an attempt to enhance the anti-tumour immune response, the co-stimulatory molecules B7-1 or B7-2 were expressed on the surface of B16 melanoma cells. B7-expressing tumours grew more slowly in both syngeneic immunocompetent mice and athymic T cellimmunodeficient nude mice. The delay in growth of B7-expressing tumours was dependent on natural killer (NK) cells, as reductions in tumour growth rates were minimized in mice depleted of NK cells. Systemic immunity to B16 melanoma was examined by vaccination with irradiated tumour cells. Inoculation with irradiated B16 B7-1 cells failed to protect against a subsequent challenge with live parental B16 cells. but conferred partial protection against challenge with live B16 B7-1 cells. In contrast to the local anti-tumour reaction. this protective response was dependent on T cells. The results presented here reveal some of the mechanisms involved in the in vivo response to a poorty immunogenic tumour modified to express co-stimulatory molecules.