The tumorigenic properties of L33 and B16F10 y-IFN-expressing tumor cell lines were also tested 
in immunocompromised athymic nude mice. The L33 tumors grew almost as well as the 
unmodified tumor (Fig 9), indicating that the decreased tumorigenicity observed in normal animals 
was at least partly due to an immune response (most likely a T cell). B16F10, however, also failed 
to grow as a tumor even in nude mice (Fig 9), indicating that y-IFN has additional effects on the 
tumor cells besides those that are T cell mediated. 
These data suggest that there are at least two mechanisms for the decreased tumorigenicity 
associated with y-IFN-transduced tumor cells: 1) T cell mediated immunity and 2) a non-T cell 
effect, due either to other immune cells (e.g., macrophages or NK cells) or to a direct inhibition of 
tumor growth. 
3. 3. 2. 3. GENERATION OF TUMOR-SPECIFIC CTL. The immunological mechanisms for the 
decreased tumorigenicity of tumor cells expressing y-IFN were of interest. There were at least 
three possible explanations for immune protection observed for y-IFN-expressing, but not 
unmodified, tumor cells. Either 1) the unmodified cells were not capable of eliciting cytotoxic 
lymphocyte responses, or 2) tumor-specific cytotoxic lymphocytes exist but the cells were resistant 
to lysis, or 3) both. The exceptionally low levels of Class I MHC on most tumor cells (see Figs 
6,7) and the need for Class I MHC for antigen presentation (3, 4) suggested that immune 
responses against the y-IFN-expressing tumor cells might not have been effective against the 
unmodified tumor. 
To test these possibilities, the CTL induction characteristics of unmodified and y-IFN-expressing 
tumor cells were compared, using the unmodified tumor cell as a target. The data in Fig 10A 
indicate that injection of B16F10 irradiated tumor cells resulted in no substantial CTL response. 
However, when mice were injected only once with y-IFN-expressing B16F10, substantial CTL 
responses were observed. Additionally, the CTL response was greater when the mice were 
immunized twice (Fig 10B). Furthermore, CTL induction was not substantially different whether 
the y-IFN-expressing B16F10 cells were injected intraperitoneally or intramuscularly (Fig IOC). 
The level of class I MHC observed for B16F10 therefore appears to present antigen sufficiently 
well to act as a target for, but not as an inducer of, CTL. 
Similarly, the data in Fig 1 1 indicate that injection of unmodified CT26 irradiated tumor cells 
resulted in no substantial CTL response. Injection with y-IFN-expressing CT26, however, 
resulted in measurable CTL responses (Fig 11 A). CTL induction was also tested in animals 
receiving unirradiated tumor cells. The data in Fig 1 IB indicate that y-IFN-transduced CT26 elicit 
more potent anti-tumor cytolytic activity than unmodified CT26. Control experiments with the 
Recombinant DNA Research, Volume 17 
[629] 
