CRISPR/Cas9-mediated PD-1 disruption enhances anti-tumor efficacy of human chimeric antigen receptor T cells

Abstract

Immunotherapies with chimeric antigen receptor (CAR) T cells and checkpoint inhibitors (including antibodies that antagonize programmed cell death protein 1 [PD-1]) have both opened new avenues for cancer treatment, but the clinical potential of combined disruption of inhibitory checkpoints and CAR T cell therapy remains incompletely explored. Here we show that programmed death ligand 1 (PD-L1) expression on tumor cells can render human CAR T cells (anti-CD19 4-1BBζ) hypo-functional, resulting in impaired tumor clearance in a sub-cutaneous xenograft model. To overcome this suppressed anti-tumor response, we developed a protocol for combined Cas9 ribonucleoprotein (Cas9 RNP)-mediated gene editing and lentiviral transduction to generate PD-1 deficient anti-CD19 CAR T cells. Pdcd1 (PD-1) disruption augmented CAR T cell mediated killing of tumor cells in vitro and enhanced clearance of PD-L1+ tumor xenografts in vivo. This study demonstrates improved therapeutic efficacy of Cas9-edited CAR T cells and highlights the potential of precision genome engineering to enhance next-generation cell therapies.

Figure 1

PD-L1 expression in human K562 myelogenous leukemia cells inhibits anti-CD19 CAR T cell function in vitro and tumor clearance in vivo. (a) Schematic representation of CAR T cell interaction with either CD19+ or CD19+ PD-L1 K562 tumor cells. (b) CD8+ anti-CD19 CAR T cells exhibit reduced degranulation (CD107a staining) upon re-stimulation with CD19+ PD-L1+ K562 cells. Representative CD107a staining is shown, gated on CD8+ CAR+ T cells. Normalized CD107a staining shows ~15% reduction in degranulation upon co-culture with CD19+ PD-L1+ targets over three independent experiments (**p = 0.009, Student’s t-test). (c) CD19+ PD-L1+ K562 cells are resistant to anti-CD19 CAR-mediated lysis in an in vitro killing assay. Left panel: a complete effector:target ratio titration is shown for a representative experiment. Mean ± S.D. for triplicate wells in a single experiment are plotted. Right panel (bar chart): CD19+ PD-L1+ K562 cells induce ~40% reduction in specific lysis relative to CD19+ K562 cells at effector:target ratio of 2:1 (*p = 0.042, Student’s t-test). The experiment was performed three times; error bars are S.D. (d) Experimental design for subcutaneous xenograft model. (e) CD19+ PD-L1+ subcutaneous xenografts impair anti-CD19 CAR mediated tumor clearance. NOD-scid-IL-2Rγ^−/− (NSG) mice were injected with 5 × 10⁶ CD19+ or CD19+ PD-L1+ K562 cells subcutaneously. Mice with established tumors (100–250 mm³) were injected intravenously with 2 × 10⁶ CD4+ and 2 × 10⁶ CD8+ anti-CD19 CAR T cells and tumor burden measured longitudinally by caliper. Shown are tumor burdens for individual mice (n = 5 per tumor type). (f) Kaplan-Meier curve for experiment described in Fig. 1E. A statistically significant difference in survival was observed between animals bearing CD19+ vs. CD19+ PD-L1+ tumors (*p = 0.016, Gehan-Breslow-Wilcoxon test).

Figure 2

Pdcd1 can be efficiently disrupted in CAR T cells using Cas9 ribonucleoproteins (Cas9 RNPs). (a) Schematic of protocol for combined CRISPR gene editing and lentiviral transduction of human primary T cells. (b) Efficient PD-1 deletion and CAR transduction in primary human T cells. PD-1 surface staining and CAR transduction were assessed 48 hours post editing. A >50% reduction in PD-1+ cells was routinely observed, with CAR transduction >70%. Right panel, individual dots represent independent editing experiments. (c) PD-1 edited CAR T cells are stable in culture. Resting PD-1 edited CD8+ anti-CD19 CAR T cells were re-stimulated with CD19+ K562 cells. Activation was measured by CD69 induction and percent reduction in PD-1+ cells measured by flow cytometry based on surface PD-1 expression; percent reduction of PD-1+ cells was similar to that observed 48 hours after editing (rightmost panel, pairwise plot).

Figure 3

CRISPR-mediated PD-1 editing rescues anti-CD19 CAR T cell function in vitro and enhances tumor clearance in vivo. (a) Diagram of PD-1 edited CAR T cell: K562 interactions (b) PD-1 edited CD8+ anti-CD19 CAR T cells (ΔPD-1) exhibit greater degranulation (CD107a staining) upon co-culture with CD19+ PD-L1+ K562 cells as compared to control CD8+ CAR T cells (*p = 0.018, Student’s t-test). (c) PD-1 edited CAR T cells are partially resistant to CD19+ PD-L1+ mediated inhibition of cytolysis. Left panel: percent lysis for control and PD-1 edited CD8+ anti-CD19 CAR T cells is shown across a range of effector:target ratios. Error bars are S.D. of triplicate wells in a single experiment. Right panel: normalized killing shows reduced PD-L1 dependent inhibition of killing in PD-1 edited CD8+ CAR T cells at effector:target ratio of 2:1 (*p = 0.03, paired t-test). The experiment was performed three independent times. (d) PD-1 deficient anti-CD19 CAR T cells exhibit enhanced anti-tumor efficacy and clear subcutaneous CD19+ PD-L1+ tumor xenografts. NSG mice were injected with 5 × 10⁶ CD19+ PD-L1+ K562 cells subcutaneously. Mice with established tumors (100–250 mm³) were injected intravenously with 4 × 10⁶ CD4+ CAR+ and 4 × 10⁶ CD8+ CAR+ control T cells or PD-1 edited cells, and tumor burden measured longitudinally by caliper. Tumor burdens are mean ± SEM for each group (n = 6 mice per group). A statistically significant decrease in tumor burden of mice receiving PD-1 edited CAR T cells was observed at multiple points (*p < 0.05, **p < 0.01, Student’s t-test).