Design and Evaluation of TIM-3-CD28 Checkpoint Fusion Proteins to Improve Anti-CD19 CAR T-Cell Function

Abstract

Therapeutic targeting of inhibitory checkpoint molecules in combination with chimeric antigen receptor (CAR) T cells is currently investigated in a variety of clinical studies for treatment of hematologic and solid malignancies. However, the impact of co-inhibitory axes and their therapeutic implication remains understudied for the majority of acute leukemias due to their low immunogenicity/mutational load. The inhibitory exhaustion molecule TIM-3 is an important marker for the interaction of T cells with leukemic cells. Moreover, inhibitory signals from malignant cells could be transformed into stimulatory signals by synthetic fusion molecules with extracellular inhibitory receptors fused to an intracellular stimulatory domain. Here, we designed a variety of different TIM-3-CD28 fusion proteins to turn inhibitory signals derived by TIM-3 engagement into T-cell activation through CD28. In the absence of anti-CD19 CAR, two TIM-3-CD28 fusion receptors with large parts of CD28 showed strongest responses in terms of cytokine secretion and proliferation upon stimulation with anti-CD3 antibodies compared to controls. We then combined these two novel TIM-3-CD28 fusion proteins with first- and second-generation anti-CD19 CAR T cells and found that the fusion receptor can increase proliferation, activation, and cytotoxic capacity of conventional anti-CD19 CAR T cells. These additionally armed CAR T cells showed excellent effector function. In terms of safety considerations, the fusion receptors showed exclusively increased cytokine release, when the CAR target CD19 was present. We conclude that combining checkpoint fusion proteins with anti-CD19 CARs has the potential to increase T-cell proliferation capacity with the intention to overcome inhibitory signals during the response against malignant cells.

Keywords: CAR T cells; CD19; CD28; TIM-3; acute lymphoblastic leukemia (ALL); checkpoint fusion proteins; pediatric leukemia.

Figure 1

TIM-3 and CEACAM1 expression on T cells/leukemic cells and design of TIM-3-CD28 fusion proteins. (A) T cells were retrovirally transduced with an anti-CD19 CAR and co-cultured with CD19⁺ target cells (K562 cells transduced with CD19) for 48 h. TIM-3 expression on CAR T cells was analyzed by flow cytometry. N = 4 individual donors; unpaired t-test was performed. Data are representative of four independent experiments ** < 0.01. (B) Leukemia and lymphoma cell lines were either left unstimulated or stimulated with 100 ng/ml IFN-γ and 10 ng/ml TNF-α. CEACAM1 expression was analyzed by flow cytometry. N ≥ 3; unpaired t-test was performed. Data are representative of three independent experiments. (C) Correlation of CEACAM1 and TIM-3 (HAVCR2) expression in publicly available datasets was evaluated using the online tool Correlation AnalyzeR (H.E. Miller, correlationAnalyzeR, (2021), GitHub repository, https://github.com/Bishop-Laboratory/correlationAnalyzeR). (D) Schematic illustration of a T cell with its endogenous TCR and the TIM-3-CD28 fusion protein that is intended to turn co-inhibition into co-stimulation. (E) Schematic illustrations of the six different fusion proteins designed for this study. (F) Exemplary flow plot showing transduction of TIM-3/28-2 into primary human T cells as analyzed by TIM-3 expression in flow cytometry. (G) Transduction rates as analyzed by flow cytometric staining of TIM-3. N ≥ 3 individual donors. Data are representative of three independent experiments. AA, amino acid; SSC, side scatter. Schematic illustrations created using biorender.com.

Figure 2

Choosing the best TIM-3-CD28 fusion receptor based on proliferation and cytokine release. (A) Fusion receptor-transduced T cells were cultured on anti-CD3-coated plates. Percent proliferating T cells was evaluated by CTV staining and the fold change with/without CD3 stim calculated for each construct. To evaluate the impact of ligand addition, the fold change of proliferating T cells was calculated on CD3 stimulation plus ligand vs minus ligand for TIM-3/28-5 (B) and TIM-3/28-6 (C). Fold change of IFN-γ (D) and TNF-α (E) positive T cells compared to untransduced T cells was analyzed by intracellular cytokine stain for IFN-γ or TNF-α with/without CD3 stimulation. FC, fold change; stim, stimulation. Dotted line represents fold change of untransduced T cells (A, D, E) or CD3 stimulation only (B, C). Experiments were performed in two individual donors and technical duplicates. Data are representative of two independent experiments. Unpaired t-test was performed to determine significance. Physiologic expression levels of TIM-3 ligands are shown in Supplementary Figure 3. * <0.05, ** <0.01, *** <0.001, **** <0.0001, ns, not significant.

Figure 3

Transduction of primary human T cells with anti-CD19 CARs in addition to TIM-3-CD28 fusion receptors. (A) Schematic illustration of transduced CAR T-cell constructs with/without fusion receptors and control 19t. (B) Exemplary flow plot showing CAR (myc)/TIM-3 stain in 19_3z CARs with and without fusion protein. (C) Transduction rates of CARs with/without fusion proteins as determined by flow cytometric stain for myc. N = 2 individual donors. Data are representative of two independent experiments.

Figure 4

Functionality of anti-CD19 CAR T cells with TIM-3-CD28 fusion proteins. (A) Killing of CD19⁺/CEACAM⁺ K562 target cells by first-generation anti-CD19 CAR T cells with/without fusion proteins was calculated after 48 h of co-culture. N = 1 individual donor in technical duplicates for 1:1 E:T ratio and n = 2 individual donors in technical duplicates for 0.1:1 E:T ratio. One-way ANOVA was performed to determine statistical significance. (B) First-generation anti-CD19 CAR T cells were co-cultured with target cells (CD19⁺/CEACAM⁺ K562) for 72 h, and percent proliferating cells were analyzed by flow cytometry (CTV). N = 2 individual donors in technical duplicates. One-way ANOVA was performed to determine statistical significance. (C) Fold expansion of different CAR T-cell constructs throughout the culture process. Cell count was normalized on the day of transduction. N = 2 individual donors. (D) First-generation anti-CD19 CAR T cells with/without fusion proteins were co-cultured with target cells (CD19⁺/CEACAM⁺ K562), and cytokine production was analyzed by intracellular cytokine stain for IFN-γ, TNF-α, and IL-2 6 h after the start of the co-culture. (E) Second-generation CAR T cells with/without fusion protein were co-cultured with target cells (CD19⁺/CEACAM⁺ K562), and proliferative potential both in terms of percent proliferating cells (E) and absolute CAR cell count (F) were analyzed after 72 h. N = 2 individual donors in technical duplicates. One-way ANOVA was performed to determine statistical significance. (G) CD25 surface expression was evaluated by flow cytometry 14 h after target cell contact (CD19⁺/CEACAM⁺ K562). N = 2 individual donors, each in technical duplicates. One-way ANOVA was performed to determine statistical significance. Data are representative of two independent experiments (B–G). E:T ratio, effector-to-target ratio; IFN-γ, interferon gamma; TNF-α, tumor necrosis factor alpha; IL-2, interleukin-2. * <0.05, ** <0.01, *** <0.001, **** <0.0001, ns, not significant.