vivo abrogated the cellular antitumor response, implicating a pivotal role for 
Class I MHC restricted CD8-positive cells. 
To approximate more closely a physiologic mode of cytokine secretion generated 
in the course of an immune response, the IL-2 gene was directly introduced into 
a murine fibrosarcoma cell line (CMS5). In this way a localized secretion of low 
levels of cytokine was achieved that led to the regression of injected tumor cells 
(14,16). When control mice were challenged by tumor, they died within 45 days. 
Mice challenged with IL-2 secreting tumor cells rejected implanted tumor cells 
and were cured. These mice developed specific immunological memory. 
Rechallenge with potentially lethal doses of CMS5 parental tumor cells after 
several months led to tumor rejection, while an unrelated tumor grew and killed 
immune mice, demonstrating the specificity of the response. Lymphocytes taken 
from spleens of immune animals were capable of killing the tumor in vitro 
throughout the life span of mice. Of note, only weak cytolytic responses were 
generated against CMS5 tumor cells after challenge with parental tumor cells; 
these responses were transient and disappeared within three weeks after tumor 
implantation. However, mice injected with IL-2 secreting tumor cells generated 
tumor specific cytolytic activity that lasted more than 75 days. 
Rosenberg et al (unpublished data) have reproduced these results in the poorly 
immunogenic tumor MCA- 102. Despite many attempts, they were never able to 
generate a cellular antitumor response using unmodified tumor cells. Introduction 
of the IL-2 gene into these cells increased their immunogenicity to a level which 
generated tumor rejection and memory. In addition, co-injection of IL-2 
producing tumor cells with unmodified, parental tumor cells produced an immune 
response strong enough to reject not only the lymphokine secreting tumor cells 
but also unmodified tumor cells. 
The mastocytoma cell line P815 was transfected with the murine IL-2 gene. 
Injection into mice led to tumor rejection. Analysis of the cellular immune 
response demonstrated that the frequency of CTL precursors and CTLs specific 
for P815 increased 2-4 fold as a result of IL-2 secretion by the tumor cells (58). 
At MSKCC the IL-2 effect was seen in different mouse tumor types, including 
B16 melanoma, 38C13 B cell lymphoma, and bladder carcinoma MBT2. Toxicity 
related to IL-2 has not been observed in any of the animals challenged with IL-2 
secreting tumor cells. Additional experiments using irradiated, IL-2 secreting 
CMS5 cells have demonstrated that there was no difference in their ability to 
induce immunological memory in vivo as compared to non-irradiated IL-2 
secreting tumor cells (see Appendix F). 
Recombinant DNA Research, Volume 15 
