A Costimulatory CAR Improves TCR-based Cancer Immunotherapy

Abstract

T-cell receptors (TCR) recognize intracellular and extracellular cancer antigens, allowing T cells to target many tumor antigens. To sustain proliferation and persistence, T cells require not only signaling through the TCR (signal 1), but also costimulatory (signal 2) and cytokine (signal 3) signaling. Because most cancer cells lack costimulatory molecules, TCR engagement at the tumor site results in incomplete T-cell activation and transient antitumor effects. To overcome this lack of signal 2, we genetically modified tumor-specific T cells with a costimulatory chimeric antigen receptor (CoCAR). Like classical CARs, CoCARs combine the antigen-binding domain of an antibody with costimulatory endodomains to trigger T-cell proliferation, but CoCARs lack the cytotoxic CD3ζ chain to avoid toxicity to normal tissues. We first tested a CD19-targeting CoCAR in combination with an HLA-A*02:01-restricted, survivin-specific transgenic TCR (sTCR) in serial cocultures with leukemia cells coexpressing the cognate peptide-HLA complex (signal 1) and CD19 (signal 2). The CoCAR enabled sTCR+ T cells to kill tumors over a median of four additional tumor challenges. CoCAR activity depended on CD19 but was maintained in tumors with heterogeneous CD19 expression. In a murine tumor model, sTCR+CoCAR+ T cells improved tumor control and prolonged survival compared with sTCR+ T cells. We further evaluated the CoCAR in Epstein-Barr virus-specific T cells (EBVST). CoCAR-expressing EBVSTs expanded more rapidly than nontransduced EBVSTs and delayed tumor progression in an EBV+ murine lymphoma model. Overall, we demonstrated that the CoCAR can increase the activity of T cells expressing both native and transgenic TCRs and enhance antitumor responses.

Figure 1.

CoCAR3 enhances antigen sensitivity of sTCR⁺ T cells. A, Schematic of retroviral vectors encoding the sTCR and CoCAR1–3. CoCAR1–3 are fusion proteins consisting of an anti-CD19 scFv (FMC63), CD28 transmembrane domain, and costimulatory endodomains. For detection and selection, the vectors contained an IRES followed by a truncated NGFR (ΔCD271). LTR, long terminal repeat; VH, variable region of heavy chain; VL, variable region of light chain. B, Frequency of transgenic (CD271⁺) T cells in culture with weekly antigen-specific stimulations (see Materials and Methods; CoCAR1⁺ and sTCR⁺CoCAR1⁺ T cells, n = 3; CoCAR2⁺ and sTCR⁺CoCAR2⁺ T cells, n = 5; others, n = 7). C, Fold expansion after transduction and following weekly stimulations (see Materials and Methods; n = 7). D, Frequency of IFNγ spot-forming cells (SFC) in response to survivin-peptide pulsed T2 cells by ELISpot (n = 6) over time (left) and separately at day 30 (right). E, Total IFNγ secretion in response to survivin-peptide pulsed T2 cells by ELISA (n = 7 day 10, n = 6 day 20, and n = 4 day 30). F, Memory phenotype of T cells defined by CD45RO and CCR7 expression (day 30, n = 5). G, 4-hour ⁵¹chromium-release cytotoxicity assay with NT, single-transduced CoCAR⁺, and single-transduced CD19-CAR⁺ T cells against B cells at the E:T ratios shown (n = 5). C–E, NT: nontransduced T cells. B–G, mean ± SD; “n” denotes the number of individual healthy donors. Statistical significance was determined by Tukey test (D) and ANOVA (area under the curve; G), *, P < 0.05; ****, P < 0.0001.

Figure 2.

Costimulation through CoCAR3 enhances the sequential killing activity of sTCR⁺CoCAR3⁺ T cells in an HLA class I- and CD19-restricted manner. A, Schematic of T-cell and tumor cell interaction in cocultures. B, Timeline and schematic of serial coculture experiment (showing first three timepoints). T cells and tumor cells were cocultured every 3–4 days at an E:T ratio of 1:5 up to eight times. C, Tumor cell elimination in serial cocultures (n = 7). Marks indicate the time of coculture failure with tumor outgrowth. D, T-cell expansion in serial cocultures (n = 7). E, IFNγ ELISA of coculture supernatants collected 24 hours after each tumor challenge (n = 5). F, T-cell fractions over time during serial cocultures (n = 5). G, Schematic of tumor cell and T-cell interaction in coculture with BV173 CD19-KO cells (left). BV173 CD19-KO cell expansion during serial coculture (right). Each line represents one donor (n = 5). H, Schematic of tumor cell and T-cell interaction in coculture with BV173 B2M-KO cells (left). BV173 B2M-KO expansion during serial coculture (right). Each line represents one donor (n = 5). Statistical significance was determined by log-rank test (C, G), ANOVA (area under the curve; D), Welch t test (area under the curve; E), paired t test of day 0 versus day 14 (F), *, P < 0.05; **, P < 0.01. D–F, mean ± SD; “n” denotes the number of individual healthy donors.

Figure 3.

Immunostimulatory pathways are activated in sTCR⁺CoCAR3⁺ T cells upon sequential tumor challenge. A, Detection of IFNγ, perforin, and TNFα in coculture supernatants 24 hours after the second tumor challenge (n = 4) as described in Fig. 2B. B and C, RNA expression in engineered T cells 24 hours after second tumor challenge using NanoString (see Materials and Methods). B, Heatmap showing genes with significant log₂ fold changes (sTCR⁺ vs. sTCR⁺CoCAR3⁺ T cells, P < 0.05; n = 4). C, Pathway scores for sTCR⁺ and sTCR⁺CoCAR3⁺ T cells as determined by NanoString. A and C, Statistical significance was determined by paired t test, *, P < 0.05; **, P < 0.01; ***, P < 0.005; mean ± SD, “n” denotes the number of individual healthy donors.

Figure 4.

CoCAR3 maintains immune-stimulatory effects in tumors with heterogeneous CD19 expression. A, Schematic of mixed tumor cell population cocultures. BV173 WT cells were mixed at the indicated ratios with BV173 CD19-KO cells to mimic various CD19 expression in the overall population. The mixed tumor cells were used in a serial coculture as described in Fig. 2B. T cells were rechallenged with the same tumor mixture every 3–4 days for a total of eight challenges. B, Tumor cell expansion during serial cocultures (n = 4). C, Coculture supernatants were harvested 24 hours after the second tumor challenge and analyzed for IFNγ by ELISA, mean ± SD. Statistical significance was determined by ANOVA with Tukey multiple comparisons test, *, P < 0.05; **, P < 0.01; “n” denotes the number of individual healthy donors.

Figure 5.

CoCAR3 activation enhances the antitumor function of sTCR⁺ T cells in vivo. A, Schematic of mouse model. NSG mice were sublethally irradiated then injected with BV173-ffluc cells. Seven days later, T cells expressing CoCAR3, sTCR, or sTCR⁺CoCAR3 were injected. Mice treated with NT cells were used as controls. Mice were monitored for tumor progression weekly. BLI, bioluminescent Imaging. B, Survival of mice. C, Bioluminescent images showing tumor signal over time. Statistical significance was determined by log-rank (Mantel–Cox), *, P < 0.01. NT n = 5, CoCAR3 n = 4, sTCR n = 6, and sTCR+CoCAR3 n = 5.

Figure 6.

CoCAR3 accelerates clearance of EBV⁺ lymphoma in vivo. A, Transduction efficiency of CoCAR3⁺ EBVSTs at the end of S1 and S2. S1 = first stimulation. S2 = second stimulation. B, Fold EBVST expansion following transduction on day 3 (n = 6 S1; n = 5 S2). NT: nontransduced. C, Frequency of IFNγ spot-forming cells (SFC) in response to EBV-pepmix at S1 and S2 by ELISpot (n = 6). Connected points indicate paired samples. D, Schema of in vivo xenograft model. NSG mice were subcutaneously injected with EBV-LCL tumor cells. Two weeks later, ffluc+ EBVSTs (NT or CoCAR3+) were injected into the mice. Ffluc⁺ CMVSTs were used as controls. BLI, bioluminescent Imaging. E, BLI of virus-specific T cell (VST) expansion over time in individual mice with established subcutaneous EBV lymphomas (n = 4). F, Left: Summary of BLI (p/s) of VSTs (n = 4). Right: Day 9 BLI summary comparing EBVST and CoCAR3⁺ EBVST signal intensity (n = 4). G, Left: LCL tumor volume (cm³). Each line represents one mouse (n = 4/group). Right: Day 13 summary of tumor volume in mice treated with EBVSTs or CoCAR3⁺ EBVSTs (n = 4). (A, B, F, G) mean ± SD. F and G, Statistical significance was determined by Student t test, **, P < 0.01; ****, P < 0.0001; “n” denotes the number of individual healthy donors.