Phosphoinositide 3-kinase δ inhibition promotes antitumor responses but antagonizes checkpoint inhibitors

Abstract

Multiple modes of immunosuppression restrain immune function within tumors. We previously reported that phosphoinositide 3-kinase δ (PI3Kδ) inactivation in mice confers resistance to a range of tumor models by disrupting immunosuppression mediated by regulatory T cells (Tregs). The PI3Kδ inhibitor idelalisib has proven highly effective in the clinical treatment of chronic lymphocytic leukemia and the potential to extend the use of PI3Kδ inhibitors to nonhematological cancers is being evaluated. In this work, we demonstrate that the antitumor effect of PI3Kδ inactivation is primarily mediated through the disruption of Treg function, and correlates with tumor dependence on Treg immunosuppression. Compared with Treg-specific PI3Kδ deletion, systemic PI3Kδ inactivation is less effective at conferring resistance to tumors. We show that PI3Kδ deficiency impairs the maturation and reduces the capacity of CD8+ cytotoxic T lymphocytes (CTLs) to kill tumor cells in vitro, and to respond to tumor antigen-specific immunization in vivo. PI3Kδ inactivation antagonized the antitumor effects of tumor vaccines and checkpoint blockade therapies intended to boost the CD8+ T cell response. These findings provide insights into mechanisms by which PI3Kδ inhibition promotes antitumor immunity and demonstrate that the mechanism is distinct from that mediated by immune checkpoint blockade.

Keywords: Adaptive immunity; Cancer immunotherapy; Immunology; Oncology; Signal transduction.

Figure 1. Deletion of phosphoinositide 3-kinase δ (PI3Kδ) in regulatory T cells (Tregs) mimics the effects of Treg depletion, but systemic PI3Kδ inactivation is less effective.

(A) Diphtheria toxin (DTx) was administered i.p. on days 3, 7, and 10 after s.c. tumor injection into the flank on day 0 (n = 6). EL4-OVA, MC38-OVA, and LLC-OVA tumors were removed on days 14, 24, and 18 after implantation, respectively. Proportions of Tregs in the blood of non–tumor-bearing mice (n = 2) were measured 24 hours after administration, and again immediately before the subsequent dose. (B) Proportion of tumor-infiltrating Tregs in the EL4-OVA, MC38-OVA, and LLC-OVA tumors at the time of collection. (C) Masses of EL4-OVA, MC38-OVA, and LLC-OVA tumors removed from WT or Foxp3^DTR mice as described in A. (D) Masses of EL4-OVA, MC38-OVA, and LLC-OVA tumors removed from WT or Foxp3^cre-PI3Kδ^fl mice. (E) Masses of EL4-OVA (n = 10), MC38-OVA (n = 8), and LLC-OVA (n = 8) tumors in WT or PI3Kδ^D910A mice. (F) Proportion of tumor-infiltrating Foxp3⁺ Tregs in WT or PI3Kδ^D910A mice; representative FACS plots of tumor-infiltrating lymphocytes are shown. Statistical significance was determined by multiple t tests with Holm-Sidak correction (B and F) or Mann-Whitney test (C, D, and E). *P < 0.05; **P < 0.01; ***P < 0.001. n.s., not significant.

Figure 2. PI3Kδ^D910A CD8⁺ T cells show functional defects.

In vitro assays were conducted with 3 biological replicates. (A) CD8⁺ T cell cytotoxic capacity was measured in an in vitro killing assay, using tumor cells expressing ovalbumin (OVA) protein (stained with CMFDA CellTracker Green) as targets for in vitro–stimulated OT-I CD8⁺ T cells, and parental tumor cells (stained with CMRA CellTracker Orange) as controls. Representative FACS plots shows EL4-OVA and EL4 cells incubated with, from left to right, WT, PI3Kδ^D910A, or no CD8⁺ T cells. (B) Cytotoxic efficiency measured in WT versus PI3Kδ^D910A OT-I CD8⁺ cells using EL4-OVA, MC38-OVA, and LLC-OVA cells. (C) Cytotoxic efficiency measured in WT versus PI3Kδ^D910A OT-I CD8⁺ cells using EL4 cells pulsed with SIINFEKL peptide, and single-amino-acid variants of decreasing binding affinity for the OT-I T cell receptor (TCR). (D) CD62L, granzyme A, and granzyme B expression measured in in vitro–stimulated WT versus PI3Kδ^D910A OT-I CD8⁺ cells. MFI, median fluorescence intensity. (E) Granzyme A and B expression measured in tumor-infiltrating CD8⁺ T cells from WT vs PI3Kδ^D910A mice (n = 8). Statistical significance was determined by multiple t tests with Holm-Sidak correction (B–D) or the Mann-Whitney test (E). *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 3. Defects in PI3Kδ-deficient CD8⁺ T cell response abrogate antitumor effect of LM-OVA immunization.

(A) OVA-specific CD8⁺ T cell responses in the tumor immune infiltrate (n = 6). Representative plots show MHC class I:SIINFEKL dextramer staining on tumor-infiltrating CD8⁺ T cells in LLC-OVA tumors. Differences between WT and PI3Kδ^D910A mice were not significant in all tumor types. (B) LLC-OVA tumor–bearing mice were immunized with attenuated LM-OVA or LM-YFP (both 10⁶ CFU) 3 days after tumor implantation. CD8⁺ T cell responses against the OVA antigen were measured in the blood 7 days after immunization (n = 6). (C) Tumor growth measured in WT and PI3Kδ^D910A mice receiving PBS control, LM-YFP, or LM-OVA immunization (n = 12, combined 2 experiments). (D) OVA-specific CD8⁺ T cell response in the tumor in response to immunization measured in WT and PI3Kδ^D910A mice (n = 6). Statistical significance determined by multiple t tests with Holm-Sidak correction (A and C) or 2-way ANOVA with Tukey’s multiple comparison (B and D). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. n.s., not significant.

Figure 4. PI3Kδ-deficient CD8⁺ T cells fail to respond to checkpoint blockade.

(A) Mice bearing LLC-OVA or MC38-OVA tumors received intravenous injections of antibodies against cytotoxic T lymphocyte antigen 4 (CTLA-4) or programmed death ligand 1 (PD-L1) on days 3, 6, and 10 after tumor injection (treatment of LLC-OVA tumor with anti–PD-L1 was performed on days 2, 5, and 9 after tumor injection) (n = 6). (B and C) Anti–CTLA-4 antibody treatment of LLC-OVA (B) and MC38-OVA (C) tumors in WT and PI3Kδ^D910A mice; no significant change was observed in tumor-infiltrating CD8⁺ T cells. (D and E) Anti–PD-L1 antibody treatment of LLC-OVA (D) and MC38-OVA (E) tumors in WT and PI3Kδ^D910A mice. Only 2 WT tumors were available for analysis, and no significant changes were found in the CD8⁺ infiltrate of these tumors. Statistical significance determined by Mann-Whitney test between antibody-treated and control groups. *P < 0.05; **P < 0.01.