Enforced Co-Stimulation and Co-Inhibitory Blockade Synergize to Enhance the Functions of Exhausted CTL

Abstract 1911

A major limitation of adoptive T cell therapies for cancer is the failure to maintain durable anti-tumor immunity. Graft-versus-tumor responses following bone marrow transplantation (BMT) may only be short-lived due to 1) defects in memory precursor generation and 2) exhaustion of surviving CTL that results from direct recognition of alloantigen upon non-hematopoietic cells {Flutter et al. JCI 2010}. In this study, we have explored the potential for enhancing co-stimulatory signals either alone, or in combination with co-inhibitory PD-1-PD-L1 blockade to improve the long term CTL response. Signalling through OX40, a TNF-receptor family member, has been shown to have an important role in long-term immunity, including an enhancement in the generation of CD8 T cell memory precursors. The mechanisms of action are complex and may include both direct effects on CD8 cells and indirect effects on CD4 helper cells or via inhibition of Treg.

In initial experiments, we evaluated the effects of early enforced OX40 co-stimulation following delayed transfer of donor T cells to haplo MHC-mismatched chimeras, 10 weeks following nonmyeloablative BMT. OX40 expression peaked on transferred CD4 and CD8 T cells in the first 1–2 weeks following transfer and was sustained thereafter, especially in the CD4 subset. 48 hours after T cell transfer, recipient mice were treated with agonistic anti-OX40 antibody (OX86) or isotype control. OX86 treatment led to a 9-fold increase in the expansion of CTL in comparison to isotype control treated mice, enhanced production of Granzyme B and IFNγ and led to more rapid eradication of host hematopoietic targets or host tumor cells. Moreover, OX86 antibody acted directly on CD8 T cells and bypassed the requirement for help from donor CD4 cells. However, although enforced OX40 co-stimulation boosted the primary effector response, it did not increase numbers of memory precursor cells, as assessed by survival and recall responses following transfer to antigen free hosts, and was unable to prevent eventual exhaustion of surviving donor CTL as tested at 60 days following transfer. Similarly, OX86 was unable to prevent exhaustion of CD8 cells transgenic for the male antigen-specific Matahari (Mh) TCR following adoptive transfer to male BMT recipients reconstituted with female BM.

We have shown previously that the functions of exhausted donor CD8 cells are partially restored by blockade of the co-inhibitory PD-1 pathway in both haplo mismatched and MHC-matched mHAg mismatch models. We hypothesized that provision of co-stimulatory signals when exhaustion had become established would increase the effectiveness of co-inhibitory blockade. Therefore, 6 weeks after Mh CD8 T cell transfer to male BMT recipients, we examined the effect of OX86, with or without additional blockade of the PD-1 pathway. Only a minority of Mh CD8 cells from animals receiving isotype control antibody were proliferating in vivo as measured by BrdU incorporation over a 7 day pulse (20 +/−3% BrdU+) and few cells were able to produce IFNγ following antigen stimulation in vitro (3.5+/−1.4 x10⁴ IFNγ+ cells/spleen). OX86 alone offered no restoration of function (15 +/− 2% BrdU+; 3.3+/−0.4 x10⁴ IFNγ+ cells; p=ns). Blockade of PD-L1 modestly increased turnover of cells (37 +/− 6 % BrdU+; p<0.01 vs isotype), but in the absence of CD4 cells, did not significantly increase production of IFNγ (4.4+/−0.9 x10⁴ IFNγ+ cells; p=ns). However, in vivo administration of OX86 combined with anti-PD-L1 blockade dramatically increased turnover of Mh CD8s (77 +/− 8% BrdU+; p<0.001 vs anti-PD-L1 alone, OX86 alone or Isotype) and enhanced their effector function ∼ 9-fold (27.4 +/− 6.8 x10⁴ IFNγ+ cells/spleen; p<0.01 vs all others).

In conclusion, forced co-stimulation via OX40 alone is unable either to prevent CTL exhaustion or restore CD8 T cell function when exhaustion has become established. In contrast, the marked synergy observed when agonistic OX40 signals are combined with co-inhibitory blockade, is consistent with a model in which the PD-1 pathway acts at a critical checkpoint that regulates the response to co-stimulation. Thus, these data suggest a novel approach to restoring the functions of exhausted anti-tumor CTL by modulating co-stimulatory and co-inhibitory pathways simultaneously.