Differential Immune Susceptibility of Malignant Versus Normal Tissue Leads to Tumor Rejection without GVH after Adoptive Immunotherapy.

Adoptive immunotherapy based on the injection of allogenic cytotoxic T-lymphocytes (CTL) during or after bone marrow transplant (BMT) has established itself as a potent anti-neoplastic treatment for several malignancies. However this approach is limited by the occurence of graft versus host disease (GVH). Using a previously described murine adoptive immunotherapy model where donor and recipient are mismatched for a single dominant minor histocompatibility antigen (H7^a), and where a powerful anti-neoplastic effect is seen without GVH, we sought to determine what rendered cancer cells more vulnerable than normal cells to immune attack. B10.H7^a mice were lethally irradiated and reconstituted with B10.H7^b (H7^a negative) T-depleted bone marrow. On the day of transplant, these mice received a B16.F10 (H7^a positive) melanoma challenge. Adoptive transfer of splenocytes obtained from B10.H7^b mice previously immunized with B10.H7^a splenocytes was performed on day 7 post BMT. Following transfer of those splenocytes containing primed anti-H7^a CTL, neither GVH nor vitiligo was noted in recipients despite the fact that H7^a is expressed in all tissues and organs. 50% of treated mice, versus 0% of controls rejected the tumor and survived 100 days. Overall survival was increased to 80% when adoptive transfer was carried on day 3 post BMT. The injection of anti-H7^b CTL had no effect on melanoma growth. Thus, the anti-tumor activity was T-cell receptor recognition dependent and not a mere bystander effect. In treated mice but not in controls, tumor histology and flow cytometry revealed important CTL infiltration (80% of those CTL being MHC-H7^a tetramer positive), increased expression of MHC class I molecules (MHC I) at melanoma cell surface, expression of Rae-1 (an NKG2D ligand), tumor necrosis and decreased angiogenesis. Importantly, normal skin in treated or control animals showed no increased expression of MHC I or Rae-1. B16.F10 melanoma cells express almost no MHC I, and no Rae-1 when cultured in vitro. Co-incubation of B16.F10 cells with INFγ leads to increased MHC I expression but no induction of Rae-1 expression, implying that at least a second factor is present in vivo to account for the expression of this stress ligand. An additional role for INFγ was evidenced when anti-H7^a CTL were injected in INFγ receptor knock-out recipients. The angiostatic effect noted after anti-H7^a CTL injection was abrogated and no mice were cured. Thus, INFγ-mediated angiostasis on the tumor stroma was crucial for the inhibition of cancer progression.

Conclusion: in our model, the differential immune susceptibility of tumor versus normal cells appears to stem from the fact that neoplastic cells are induced to express more MHC I and stress ligands such as Rae-1. Those can respectively increase target antigen density at the cell surface and mediate CTL co-stimulation or cytotoxicity, through NKG2D receptors. Strategies exploiting stress ligand induction as well as the effect of INFγ on MHC expression and angiostasis may contribute to successfully separate anti-tumor from GVH effects.