Metabolic Reprograming of Exhausted Intratumoral CD8+ T-Cell Underlies Anti-Tumor Activity of Sumoylation Inhibitors in Large B Cell Lymphoma

Immune checkpoint inhibitors (ICI) targeting PD1 have revolutionized our approach to solid tumors and Hodgkin lymphoma, but they have not been effective in treating Large B Cell Lymphoma (LBCL). In prior work, we used single cell spatial profiling of human tumors to show that exhausted CD8 tumor and TREG populations were associated with PD1 resistance in LBCL (Colombo et al. Blood Adv 2022). Here we show, using a syngeneic model of LBCL, that PD1 resistance is associated with lower CD8+ T-cell basal metabolic rate (BMR) in tumor infiltrating lymphocytes (TIL), and that treatment with the covalent SUMOylation inhibitor (SB-4826) increases the metabolic rate of TIL by metabolic reprograming. Single cell metabolic profiling of human LBCL is on-going and will be presented as well. In summary these studies demonstrate that metabolic dysfunction of T cells in LBCL underlie resistance to PD1 checkpoint inhibitors and that this can be overcome by treatment with SUMOlyation inhibitors.

Direct measurement of intra-tumoral metabolic state is difficult to perform on clinical samples, however large tumor size is hypothesized to be a surrogate for increased metabolic derangement. As tumors expand and become more metabolically active themselves, infiltrating immune cells experience decreased O2 tension as well as nutrient deprivation, ultimately leading to decreased anti-tumor immune responses. In support of this idea, it has been observed that high metabolic tumor volume at baseline predicts survival in LBCL patients, independent of their initial response to therapy. We have shown that mice with established (>125mm³) A20 tumors are resistant to anti-PD1 therapy. However, A20 bearing mice treated ~5 days post implantation are responsive to anti-PD1 therapy. To understand these findings, we examined sensitive (small) and resistant (large) A20 tumors by flow cytometry analysis and found that resistant tumors had lower relative numbers of T cells, lower absolute numbers of CD8+ T cells, increased absolute numbers of TREG and a strikingly decreased CD8/TREG ratio. Additionally, the resistant tumors had increased absolute numbers of pre-exhausted and exhausted CD8+ T cells as well as decreased BMR.

Small ubiquitin-like modifiers (SUMO) proteins regulate a variety of cellular processes in tumor and immune cells and inhibitors of SUMOylation have shown anti-tumor activity in a variety of preclinical tumor models with potential synergy with checkpoint inhibitors targeting PD-1 (Cannon et al. Cancer Res, 2023). We demonstrate that a dose of 100mg/kg of SB-4826 led to a complete response in all mice treated (Fig. 1A) and we also show that this effect is dependent on CD8+ T-cells and elicits a long term memory response, with 0 of 15 mice in remission developing tumor upon rechallenge. A low dose (25mg/kg) of SB-4826 failed to elicit a response in any of the mice treated while anti-PD1 therapy elicited a response in 1/6 mice (Fig. 1A). Combination of 25mg/kg SB-4826 with anti-PD1 therapy demonstrated synergism with a response in 3/6 mice (Fig. 1A).

We also demonstrate that global SUMOylation levels are higher in PD1+ TIM3+ CD8+ TILs compared to PD1- TIM3- CD8+ T-cells, suggesting that SUMOylation may be acting as a negative regulator of T-cell function. Using SCENITH (Arguello et al. Cell Metabolism 2020) for single cell metabolic studies, we demonstrate that SB-4826 increases the BMR (Fig. 1B), mitochondrial dependency and proliferative capacity of tumor infiltrating PD1+ TIM3+ CD8+ T-cells (TILs). While both SB-4826 and PD1 inhibition led to increased CD8+ TIL numbers, SB-4826 led to increased BMR in CD8+ TILs while PD1 inhibition did not affect CD8+ TILs BMR. Additionally, PD1 inhibition significantly decreased CD8+ TILs mitochondrial dependency for metabolism, while SB-4826 increased mitochondrial dependency (Fig. 1B). Additionally, we recapitulate these metabolic enhancements in an in vitro system of T-cell activation, further confirming a T-cell intrinsic effect following SB-4826. Mechanistically, we found that SB-4826 led to increased HIF-1a levels in both CD4+ and CD8+ cells following TCR activation, likely through stabilization of HIF-1a due to decreased SUMOylation-dependent turnover (Cheng J et al. Cell 2007). We hypothesize that metabolic reprograming of CD8 T-cells through increased HIF-1a is the basis for synergy between SUMOylation inhibition and immune checkpoint inhibitors in LBCL.