c-Kit signaling potentiates CAR T cell efficacy in solid tumors by CD28- and IL-2-independent co-stimulation

Abstract

The limited efficacy of chimeric antigen receptor (CAR) T cell therapy for solid tumors necessitates engineering strategies that promote functional persistence in an immunosuppressive environment. Herein, we use c-Kit signaling, a physiological pathway associated with stemness in hematopoietic progenitor cells (T cells lose expression of c-Kit during differentiation). CAR T cells with intracellular expression, but no cell-surface receptor expression, of the c-Kit D816V mutation (KITv) have upregulated STAT phosphorylation, antigen activation-dependent proliferation and CD28- and interleukin-2-independent and interferon-γ-mediated co-stimulation, augmenting the cytotoxicity of first-generation CAR T cells. This translates to enhanced survival, including in transforming growth factor-β-rich and low-antigen-expressing solid tumor models. KITv CAR T cells have equivalent or better in vivo efficacy than second-generation CAR T cells and are susceptible to tyrosine kinase inhibitors (safety switch). When combined with CD28 co-stimulation, KITv co-stimulation functions as a third signal, enhancing efficacy and providing a potent approach to treat solid tumors.

Extended Data Fig. 1. Mz-KITv and M28z-KITv CAR T cells show relatively lower levels of PD1 expression.

a, Proportion of PD1⁺ cells in CD4 and CD8 CAR T cells after stimulation with antigen-expressing (MSLN⁺) target cells (n=5 biological replicates; two-tailed paired t test **p=0.0018, M28z vs. Mz-KITv; ***p=0.0009, M28z vs. M28z-KITv in CD4 T cells; and **p=0,0013, M28z vs. M28z-KITv; **p=0.0075, M28z vs. Mz-KITv for CD8 T cells). b, Proportion and kinetics of PD1⁺ cells in CD4 and CD8 CAR T cells upon stimulation with MSLN⁺ target cells every four days, as indicated by arrows. All data shown are mean ± standard deviation and are representative of three independently repeated experiments with at least three independent donors.

Extended Data Fig. 2:. Long-term functional persistence of M28z-KITv CAR T cells.

Mice were initially established with A549-M lung cancer metastatic tumor and treated with a single dose (1 × 10⁵) of M28z-KITv CAR T cells. After tumor eradication for 122 days, confirmed by multiple tumor bioluminescence imaging (BLI), mice were rechallenged with intraperitoneal (i.p) injection of 1 × 10⁶ A549-M cells on day (d) 123 after initial CAR T-cell treatment, and tumor burden was monitored with BLI (n=4 mice).

Extended Data Fig. 3:. KITv CAR T cells have molecular expression profiles associated with potent cytotoxic activity.

Mesothelin (MSLN)–targeted M28z and M28z-KITv CD8⁺ CAR T cells alone (unstimulated) or cocultured (stimulated) with MSLN-expressing 3T3-M cells for 24 hours (M28z Sti and M28z-KITv Sti) were collected, and their gene expression profiles were analyzed by RNA-seq analysis. a, Heat maps of the top 50 differentially expressed genes between unstimulated M28z and M28z-KITv CAR T cells. b, Individual gene expression of AP-1-bZip and IRF gene family members from the bulk RNA-seq analysis in unstimulated M28z (blue) and M28z-KITv (red) CAR T cells. Data are mean ± standard deviation from three samples across three different donors (*p<0.05; **p<0.01; ***p<0.001) determined by multiple t tests. c, Gene-set enrichment analysis, comparing the gene expression of M28z-KITv and M28Z CAR T cells for JAK-STAT3 and IL2-STAT5 gene sets. d, Heat maps of the top 50 differentially expressed genes between M28z Sti and M28z-KITv Sti CAR T cells. e, Gene-set enrichment analysis for the expression profiles of stimulated M28z-KITv Sti compared with M28z Sti CAR T cells for IFN-α and IFN-γ gene sets. Nominal p values and false-discovery rate q values are indicated.

Extended Data Fig. 4:. Dasatinib inhibits CAR T-cell cytotoxicity without affecting viability in vitro.

a, M28z-KITv CAR T cells were cultured in the presence or absence of dasatinib for 72 h. Target-cell viability was monitored by flow cytometry. The experiment was performed twice, and a representative example is shown. b, Luciferase-based cytotoxicity assay assessing M28z KITv CAR T-cell killing at 18 h when cocultured with ffLuc-expressing, mesothelin-positive target cells in the presence or absence of dasatinib (n=3 technical replicates). The experiment was performed three times, and a representative example is shown.

Extended Data Fig. 5:. Representative gating strategy used for flow cytometry analysis.

a, Tumor cells gating strategy corresponding to data shown in Figures 1d, 2e, 4b, and 4h. b, CAR T cells in vitro gating strategy corresponding to data shown in Figure 1e, h, and j. c, IFN-γ KO CAR T cells in vitro gating strategy corresponding to data shown in Figure 3c. d, CAR T cells in vivo gating strategy corresponding to data shown in Figure 6d.

Fig. 1:. KITv CAR T cells demonstrate constitutive STAT phosphorylation and higher antigen-specific target-cell lysis in vitro.

a, Schematic of the retroviral vector (SFG) encoding CARs targeting mesothelin (MSLN) or prostate-specific membrane antigen (PSMA), linked with or without an intracellular fragment of c-Kit or c-Kit D816V (KITwt or KITv). A c-myc-tag or ΔLNGFR is included for the detection of CARs. b, Representative flow cytometry histogram plots of transduced T cells, showing percentage of CAR expression. c, Western blot of phosphorylated (p-KIT) and unphosphorylated (KIT) c-Kit level in KITv CAR T-cell lysates. GAPDH is used as loading control. M28z and M28z CAR T cells transduced with wild-type c-KIT (KITwt) served as controls. d, MSLN expression level of MSTO-M (mesothelioma) cells by flow cytometry. e, Specific lysis of ffLuc-expressing MSLN⁺ target cells by indicated CAR T cells from luciferase-based cytotoxicity assay performed at 4 and 18 h (n=3 technical replicates; data representative of six independent experiments performed with two independent donors). f, Cumulative cell counts of indicated CAR T cells upon stimulation with MSLN⁺ target cells every 4 days (n=3 technical replicates; data representative of three independent experiments). g, After 18 h of coculture of respective CAR T cells with antigen (MSLN)–expressing target cells, released IFN-γ, TNF-α, and IL-2 as measured by Luminex assay are shown (n=6; three biological replicates with two technical replicates each; two-tailed unpaired t test ****p<0.0001, *p=0.0178, *p=0.02, **p=0.008). h, Comparative flow cytometric plot, showing the expression of p-STAT1, p-STAT3, and p-STAT5 in CD4⁺ and CD8⁺ CAR T cells. Untransduced (UTD) and M28z CAR T cells serve as controls. i, Specific lysis of ffLuc-expressing MSLN⁺ target cells by M28z KITv CAR T cells in the presence or absence of STAT 3/5 inhibitor after 18 h of coculture (n=3 technical replicates; data representative of two independent experiments). j, Representative flow cytometry density plots showing expression of CD45RA and CD62L in indicated CAR T cells. k, Relative frequency of memory (CD45RA⁺CD62L⁺) in CD8⁺ CAR T cells (n=4 independent donors; two-tailed paired t test *p=0.0346, **p=0.0060). Data are shown as mean or mean ± standard deviation.

Fig. 2:. KITv CAR T cells show potent antitumor efficacy in mesothelioma and lung cancer models in vivo.

a-c, Orthotopic pleural mesothelioma–bearing mice were treated with CAR T cells. a, Schematic of the orthotopic pleural mesothelioma CAR T-cell therapy. Pleural mesothelioma was established by intrapleural (i.p) administration of mesothelin (MSLN)–expressing cancer cells (8 × 10⁵). At 10 days after establishment of tumor, cohorts of mice with equivalent tumor burden were treated with a single dose (5 × 10⁴) of respective CAR T cells (i.p). b, Kinetics of tumor growth tracked by bioluminescence imaging (BLI) after CAR T-cell treatment (P28z, n=6; M28z and M28z-KITv, n=10 each; Mz-KITv, n=9; Mz, n=8 mice per cohort). c, Kaplan-Meier survival analysis of data shown in b. Data analysis was performed using a one-sided log-rank Mantel-Cox test; *p=0.0144, Mz-KITv vs. Mz; *p=0.029, M28z-KITv vs. Mz-KITv; **p=0.0028, M28z-KITv vs. Mz; N.S. p=0.5947, M28z vs. M28z-KITv. d-g, Metastatic lung cancer xenograft–bearing mice were treated with MSLN-targeted CAR T cells. d, Schematic of the metastatic lung cancer xenograft model treated with CAR T cells intravenously (i.v). Antigen (MSLN)–expressing A549-M cancer cells (1×10⁶) were administered by tail-vein injection. At 20 days after establishment of tumor, cohorts of mice with equivalent tumor burden were treated with a single dose (1 × 10⁵) of respective CAR T cells (i.v). e, MSLN overexpression by A549-M cells demonstrated by flow cytometry. f, kinetics of tumor growth tracked by BLI after CAR T-cell treatment (n=6 mice for the P28z group, n=8 mice per group for the other groups). g, Kaplan-Meier survival analysis of data shown in f. Data analysis was performed using a one-sided log-rank Mantel-Cox test; **p=0.009, M28z vs. Mz-KITv; ***p=0.0003, M28z-KITv vs. Mz-KITv; ****p<0.0001, M28z vs. M28z-KITv. h, Plasma cytokine levels from mice at day 10 after CAR T-cell infusion (n=4 mice per group). Data displayed as mean ± standard deviation and statistical analysis determined by one-way ANOVA; ***p=0.0001, ****p<0.0001. Data shown in b and f are representative of three independently repeated experiments with at least three independent donors.

Fig. 3:. KITv costimulation enhances the antitumor activity of CAR T cells in a TGF-β–rich tumor microenvironment.

a-f, Mice with orthotopic pleural mesothelioma tumor overexpressing TGF-β treated with CAR T cells. a, TGF-β expression level as measured by flow cytometry in lung cancer (A549) and mesothelioma (MSTO-M and MSTOM-TGFβ) cells. b, TGF-β protein levels in culture medium (n=3 technical replicates), plasma (n=4 biological replicates), or tumor tissue (n=4 tumors for MSTO-M and n=5 tumors for MSTOM-TGFβ group) collected from mice with equivalent tumor burden. Data shown are mean ± standard deviation, and analysis was performed using a two-tailed unpaired t test; ****p<0.0001, ***p=0.0002, **p=0.0023. c, Representative flow cytometric plot showing the expression of IFN-γ in IFNγ wild-type (WT) and IFNγ knock-out (KO) M28z-KITv CAR T cells after 18 h of coculture with antigen-expressing MSTO-M target cells. d, Schematic of the TGF-β–overexpressing mesothelioma CAR T-cell therapy. At 10 days after establishment of orthotopic pleural mesothelioma with MSTOM-TGFβ cells, cohorts of mice with equivalent tumor burden were treated with a single dose (1 × 10⁵) of respective CAR T cells intrapleurally (i.p). e, Kinetics of tumor growth tracked by bioluminescence imaging (BLI) after CAR T-cell treatment (n=6 mice for the P28z group, n=8 mice per group for the other groups). f, Kaplan-Meier analysis of data shown in e. Data analysis was performed using a one-sided log-rank Mantel-Cox test; **p=0.0075, M28z vs. M28z-KITv; *p=0.039, M28z vs. Mz-KITv; ****p<0.0001, M28z-KITv vs M28z-KITv IFNγ KO. Data shown are representative of three independently repeated experiments with three independent donors.

Fig. 4:. KITv potentiates the antitumor efficacy of CAR T cells against low-antigen (MSLN+) lung cancer targets.

a, M28z and M28z-KITv CAR T cells were stimulated by coculture with mesothelin (MSLN)–expressing target cells for 5 min, and p-ERK and p-CD3ζ levels were measured by flow cytometry. b, MSLN-expression level of A549 cells measured by flow cytometry. c, ⁵¹Cr-release cytotoxicity assay of CAR T cells. Low-antigen-expressing A549 target cells were labelled with ⁵¹Cr and cocultured with respective CAR T cells at multiple E:T ratios. Cytotoxicity at 18 h was measured by ⁵¹Cr-release (n=3 technical replicates; data representative of six independent experiments performed with three independent donors). d-g, Low-antigen-expressing A549 tumor–bearing mice were treated with CAR T cells. d, Schematic of the metastatic lung cancer xenograft model treated with CAR T cells. At 20 days after establishment of tumor, mice were treated intravenously (i.v) with a single dose (5 × 10⁵) of CAR T cells. e, Kinetics of tumor growth tracked by bioluminescence imaging (BLI) after CAR T-cell treatment (P28z and Mz, n=5; M28z and M28z-KITv, n=8 each; Mz-KITv, n=9; mice were selected and sorted into cohorts with equivalent tumor burden as determined by BLI). f, Kaplan-Meier survival analysis of data shown in e. Data analysis was performed using a one-sided log-rank Mantel-Cox test; **p=0.0024, M28z-KITv vs. Mz-KITv; **p=0.0074, M28z vs. Mz-KITv; ****p<0.0001, M28z vs. M28z-KITv. g, Plasma cytokine levels of mice at day 10 after CAR T-cell infusion (n=4 mice per group). Data shown are mean ± standard deviation and analyzed by one-way ANOVA (**p=0.002, M28z vs. M28z-KITv; **p=0.009, M28z-KITv vs. Mz-KITv). Data shown in e and g are representative of three independently repeated experiments with at least two independent donors.

Fig. 5:. KITv potentiates the antitumor efficacy of CAR T cells against low-antigen (MSLN⁺) mesothelioma cancer targets, without cytotoxicity against mesothelial cells.

a, MSLN-expression level of MSTO-M cells (high antigen), MSTO-M^low cells (low antigen), or human pleural mesothelial Met5A cells (very low antigen) as measured by flow cytometry. b, Control nonantigen-targeted P28z, antigen-targeted M28z, and M28z-KITv CAR T cells were cocultured with MSTO-M, MSTO-M^low, or Met5A cells for 48 h. Target cell viability was monitored by flow cytometry (n=3 technical replicates). Data shown are mean ± standard deviation, and analysis was performed using one-way ANOVA; ****p<0.0001, MSTO-M; **p=0.0031, MSTO-M^low P28z vs. M28z-KITv; **p=0.0062, MSTO-M^low M28z vs. M28z-KITv. c, Schematic of the orthotopic pleural mesothelioma model. Mice were established with intrapleural (i.p) administration of MSTO-M^low (8 × 10⁵) cells. At 10 days after establishment of tumor, mice were treated with a single dose (5 × 10⁵) of CAR T cells administered intrapleurally (i.p). d, Kaplan-Meier analysis of survival of mice. Data analysis was performed using a one-sided log-rank Mantel-Cox test (****p<0.0001). Data shown in b and d are representative of two independently repeated experiments.

Fig. 6:. KITv enhances the antitumor efficacy of prostate-specific membrane antigen (PSMA)–targeting CAR T cells.

a, Schematic of the retroviral vector (SFG) encoding CARs targeting PSMA, with CD28 costimulatory domain linked with LNGFR (P28z) or KITv (P28z-KITv) and a first-generation CAR targeting PSMA linked with KITv (Pz-KITv). A c-myc-tag is included for the detection of CARs. b, Representative flow cytometry histogram plots of transduced T cells, showing percentage of CAR expression. c, PSMA-expression levels of A549-P and PC3-P cells were measured by flow cytometry. d-g, A549-P tumor–bearing mice were treated with CAR T cells. d, Schematic of the metastatic cancer xenograft model treated with CAR T cells intravenously (i.v). e, Kinetics of tumor growth tracked by bioluminescence imaging (BLI) after CAR T-cell treatment (n=6 mice treated with untransduced [UTD] T cells, n=8 mice treated for the other groups; mice were selected and sorted into cohorts with equivalent tumor burden as determined by BLI). f, Kaplan-Meier survival analysis of data shown in e. Data analysis was performed using a one-sided log-rank Mantel-Cox test; *p=0.0406, P28z-KITv vs. Pz-KITv; ***p=0.0004, P28z vs. P28z-KITv; ***p=0.0005, P28z vs. Pz-KITv. g, Plasma cytokine levels of mice at day 10 after T-cell infusion (n=4 mice per group). Data shown are mean ± standard deviation and analysis was performed using one-way ANOVA; *p=0.0279, IFN-γ P28z vs. Mz-KITv; ***p=0.0003, IFN-γ P28z vs. P28z-KITv; ****p<0.0001, IL-10 P28z vs. P28z-KITv; ***p=0.0004, IL-2 P28z vs. Pz-KITv; *p=0.0277, IL-2 P28z vs. P28z-KITv. h-I, PC3-P tumor–bearing mice were treated with untransduced or CAR T cells. h, Schematic of the metastatic prostate cancer xenograft model treated with CAR T cells intravenously (i.v). i, Kinetics of tumor BLI (n=5 mice for the P28z group, n=6 mice per group for the other groups; mice with equivalent tumor burden as determined by BLI before treatment were selected and sorted into cohorts). j, Kaplan-Meier survival analysis of data shown in i. Data analysis was performed using a one-sided log-rank Mantel-Cox test; *p=0.0169, P28z vs. P28z-KITv; *p=0.0372, P28z vs. Pz-KITv. UTD, untransduced. All data shown are representative of at least two independently repeated experiments.

Fig. 7:. M28z-KITv CAR T cells are susceptible to tyrosine kinase inhibitors (TKIs) in vivo.

a, Schematic of mice with orthotopic pleural mesothelioma treated with M28z-KITv CAR T cells regionally (d-14; intrapleurally [i.p]) or systemically (d-10; intravenously [i.v]), followed by daily oral administration of TKI after receiving CAR T cells. b, Kinetics of tumor growth tracked by bioluminescence imaging (BLI) after treatment with M28z KITv CAR T cells administered intrapleurally or intravenously. After administration of M28z-KITv CAR T cells, mice received no drug (n=3 mice) or either dasatinib or midostaurin (n=5 mice per cohort). c, Kaplan-Meier survival analysis of data shown in b. Data analysis was performed using a one-sided log-rank Mantel-Cox test; **p=0.0017, intravenous no drug vs. dasatinib; **p=0.002, intravenous no drug vs. midostaurin; **p=0.0042, intravenous no drug vs. P28z KITv; *p=0.0123, intrapleural no drug vs. dasatinib; *p=0.0169, intrapleural no drug vs. midostaurin. Data shown in b and c are representative of three independently repeated experiments with at least two donors. d, Flow cytometric analysis quantifying the number of CAR T cells in the harvested tumors at day 3 (n=3 tumors for no-drug group, n=4 tumors for dasatinib group, n=5 tumors for midostaurin group). Data shown are mean ± standard deviation, and analysis was performed using one-way ANOVA; ****p<0.0001, no drug vs. dasatinib; **p=0.0035, no drug vs. midostaurin.