IFN-γ-resistant CD28 CAR T cells demonstrate increased survival, efficacy, and durability in multiple murine tumor models

Abstract

Interferon-γ (IFN-γ) plays complex and, sometimes, contradictory roles in cancer, which can affect patient responses to treatments such as immunotherapies. We recently demonstrated that IFN-γ production by chimeric antigen receptor (CAR) T cells is not required for efficacy in hematologic tumor models, whereas IFN-γ receptor (IFN-γR) signaling in solid tumor cells facilitates CAR T cell adhesion and antigen-specific cytotoxicity. Here, we show that IFN-γ induces apoptosis of CAR T cells bearing a CD28 intracellular signaling domain, which can be reduced through targeting of IFN-γ or IFN-γR. In hematologic malignancies, knockout of IFN-γR (IFN-γRKO) in CAR T cells increased their persistence without compromising efficacy. In xenograft and syngeneic solid tumor models, IFN-γR knockout CAR T cells displayed more potent tumor control, prolonged survival, and improved T cell memory that conferred protection from tumor rechallenge. RNA sequencing of tumor-infiltrating IFN-γRKO CAR T cells derived from tumor-bearing mice revealed increased cell death in tumor cells. Collectively, these data show that inhibition of IFN-γ signaling can increase the expansion and antitumor activity of CD28-based CAR T cells in liquid and solid tumors.

Fig. 1.. Knockout of IFNγ or IFNγ receptor drives greater CD28 CAR-T cell expansion in vitro.

(A) Schematic diagram of genetic constructs used to generate human chimeric antigen receptor (CAR)-T cells with a CD19 single-chain fragment variable (scFv), CD28 costimulatory domain, and CRISPR/Cas9 editing of T cell receptor A constant region locus (TRAC; CAR_CD19), TRAC and IFNG (IFNγKO CAR_CD19), or TRAC and IFNGR (IFNγRKO CAR_CD19). (B) Expression of CD3 (Allophycocyanin, APC; left), IFNγR (Alexa Fluor 647, AF647; middle) and phosphoSTAT1 (pSTAT1) signaling (AF647; right) following exposure to recombinant human IFNγ on resting CAR-T cells and donor-matched untransduced (UTD) T cells) with representative histograms from 3 biological replicates. (C) IFNγ (Brilliant Violet 510, BV510) expression on CAR-T cells was assessed by flow cytometry following four-hour activation with phorbol myristate acetate (PMA) and ionomycin; (n=3 biological replicates). Y-axis is an empty channel. (D) CD4 and CD8 expression on CAR-T cells and UTD T cells were determined using flow cytometry; n=6 biological replicates. Data are shown as mean ± SEM with P values determined by two-way ANOVA with Tukey’s multiple comparisons test. (E) CAR-T cells were activated overnight with CD19 antigen and secreted cytokines were measured by Luminex; bubble plot represents average cytokine expression (ng/ml) from 3 biological replicates. (F) Expansion of CD28 CAR-T cells in response to Nalm6 cells was monitored using real-time Incucyte imaging. Average expansion (±SEM) was measured over 72 hours. Data represents 4 biological replicates which were analyzed by one-way ANOVA with Tukey’s multiple comparisons analysis at 24, 48, and 72 hours. Significance of IFNγKO CAR_CD19 vs. CAR_CD19 (pink *) or IFNγRKO CARCD19 vs. CAR_CD19 (turquoise *,**) is shown. (G) Schematic diagram of constructs used to generate control, IFNγKO, and IFNγRKO CAR-T cells with a CD19 scFv and 4–1BB costimulatory domain. (H) Representative histograms of CD3 (APC; left) and IFNγR (AF647; middle) expression on resting CAR-T cells and donor-matched UTD T cells measured by flow cytometry alongside IFNγ following a four-hour activation with PMA and Ionomycin (BV510; right). Data was analyzed from 3 biological replicates. (I) Expansion of 4–1BB CAR-T cells in response to Nalm6 cells was monitored using real-time Incucyte imaging. Average expansion (±SEM) was measured over 72 hours. Data represents 3 biological replicates which were analyzed by one-way ANOVA. *P<0.05, **P<0.01, ****P<0.0001; ns, not significant.

Fig. 2.. IFNγKO and IFNγRKO CAR_CD19 maintain anti-tumor activity against hematologic malignancies and display greater persistence in vivo.

(A) CAR-T cells were cultured with Nalm6 cells (1:1) and tumor cytolysis was measured by Incucyte. At 3 and 6 days, cells were collected, counted, and CAR-T cells were re-plated with fresh Nalm6 cells (1:1) before being returned to the Incucyte. Data represents mean (±SEM) for 2 biological replicates. (B to F) NSG mice bearing intravenously administered CBGGFP⁺ JeKo-1 tumor cells (1e⁶ IV) were left untreated (tumor only; TO) or treated with 1e⁶ CAR_CD19, IFNγKO CAR_CD19, or IFNγRKO CAR_CD19 (1e⁶ IV). Data was obtained from 4–5 mice/group per donor (2 donors total). (B) Schematic timeline of in vivo CAR-T cell treatment. Tumor burden was assessed by bioluminescent imaging for each CAR-T cell group and tumor only control over 75 days post-CAR-T cell infusion (C) and graphed by flux (D). (E) Overall survival was measured for each CAR-T cell treatment cohort and represented by Kaplan-Meier survival curve. (F) Mice were bled 14 days post-CAR-T cell infusion and T cell persistence was determined using flow cytometry. Data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons analysis (D, F) or Mantel-Cox test (E). **P<0.01, ****P<0.0001; ns, not significant.

Fig. 3.. Deletion of IFNγ or IFNγR protects CD19 CAR-T cells from cell death.

(A) CAR-T cells were activated overnight with plate-bound CD19 antigen prior to RNA sequencing using NanoString and graphed as volcano plots. Proliferation (blue dots) and cell death (pink dots) genes are highlighted in CAR_CD19 (purple shading) compared to either IFNγKO CAR_CD19 (pink shading) or IFNγRKO CAR_CD19 (turquoise shading). Three adjusted P value cutoffs are shown as a dotted bottom line (P<0.05), dashed middle line (P<0.01), and a dot-dash top line (P<0.001). Data was derived from 3 biological replicates. (B) Proliferation in response to CD19⁺ Nalm6 cells was assessed by Cell Trace Violet at 6 days post-activation. Data are represented as histograms and pie charts showing the percentage of cells (white numbers) for each number of cell divisions (0=gray, 1=red, 2=turquoise, 3=purple, 4=orange) derived from 3 biological replicates. (C and D) Annexin V (C) and cleaved Caspase 3 (D) expression were determined by flow cytometry for CAR-T cell groups at rest (no activation; NA) and 48 hours post-activation (CD19); n=3 biological replicates. (E) Annexin V expression on CAR-T cells activated with plate-bound CD19 antigen was assessed using Incucyte real-time analysis over a 72-hour period in 3–6 biological replicates]. Significance at 72 hours was noted for IFNγRKO CAR_CD19 vs. CAR_CD19 (turquoise *). (F and G) CAR-T cells were activated overnight with CD19 antigen in the absence or presence of exogenous recombinant human IFNγ prior to RNA sequencing using NanoString. Data is presented as bar graphs (F) and volcano plots with adjusted P value cutoffs shown as a dotted bottom line (P<0.05) and dashed top line (P<0.01) (G); n=3 biological replicates. Interferon signaling (purple dots) and cell death (magenta dots)-related genes are highlighted. (H) Wildtype CD19-specific CAR-T cells with no CRISPR/Cas9 editing were activated with CD19 in the absence or presence of blocking antibodies to IFNγ (αIFNγ) or IFNγR (αIFNγR) and Annexin V expression was assessed by Incucyte; n=2 biological replicates. Data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons analysis (C, D, E, F, H). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001; ns, not significant.

Fig. 4.. Loss of IFNγ signaling drives greater survival in mesothelin-specific CD28 CAR-T cells.

(A) Schematic diagram of CAR_MESO, IFNγKO CAR_MESO, and IFNγRKO CAR_MESO constructs with the anti-mesothelin SS1 scFv and CD28 costimulatory domain generated from healthy donors. (B) Representative histograms depicting expression of CD3 (APC), IFNγR (AF647), and IFNγ (BV510) for mesothelin-specific CAR-T cell constructs. (C) CD4 and CD8 frequencies were determined by flow cytometry for 6 biological replicates. Data are shown as mean ± SEM with P values analyzed by one-way ANOVA with Tukey’s multiple comparisons test. (D) Fold change expansion of CAR-T cells in response to mesothelin was monitored using real-time Incucyte imaging over a 72-hour period (left) and quantitative flow cytometry (right) from 3–4 biological replicates. Data are shown as mean ± SEM with P values analyzed by one-way ANOVA with Tukey’s multiple comparison test. Significance noted for IFNγRKO CAR_MESO vs. CAR_MESO (turquoise **). (E) Memory markers CD45RA (Peridinin chlorophyll protein-Cyanine5.5, PerCP/Cy5.5) and CD62L (Fluorescin isothiocyanate, FITC) were assessed on resting CAR-T cells and shown by representative flow cytometry plots and pie charts for overall averages for 6 biological replicates. Data is shown as naïve (CD45RA⁺CD62L⁺), central memory (CM; CD45RA^-CD62L⁺), effector memory (EM; CD45RA^-CD62L^-), and effector (EFF; CD45RA⁺CD62L^-). (F) CAR-T cells were activated with mesothelin, and exhaustion markers were measured by flow cytometry with pre- (PRE) and post-activation (POST) graphed. Data are shown as mean ± SEM from 3 biological replicates. (G) Proliferation of CAR-T cells in response to mesothelin was assessed using Cell Trace Violet and displayed as a representative donor and average pie chart (top) and collective averages (bottom) showing mean ± SEMt; n=3 biological replicates. Pie charts show the percentage of cells (white numbers) for each number of cell divisions (0=gray, 1=red, 2=turquoise, 3=purple, 4=orange, 5=blue, 6=green). (H) Annexin V on CAR-T cells exposed to mesothelin was assessed using Incucyte real-time analysis and displayed as fold change of average green area compared to timepoint zero; n=4 biological replicates. Data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons analysis (IFNγRKO CAR_MESO vs. CAR_MESO P=.0856). *P<0.05, **P<0.01, ***P<0.001.

Fig. 5.. IFNγRKO CAR_MESO exhibit greater anti-tumor activity and survival against pancreatic cancer.

(A) CAR-T cells and donor-matched UTD T cells were activated with mesothelin antigen overnight. Granzyme B and perforin expression were examined using flow cytometry with representative histograms shown for 3 biological replicates. (B) CAR_MESO, IFNγKO CAR_MESO, and IFNγRKO CAR_MESO were mixed with mesothelin⁺ AsPC-1, BxPC-3, or Capan2 cells at a 1:1 ratio and tumor cytolysis was assessed by Incucyte. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparisons test for 2–3 biological replicates and is shown as ± SEM. P values shown are CAR_MESO (purple *) or IFNγRKO CAR_MESO (turquoise *) compared to IFNγKO CAR_MESO. (C) NSG mice bearing subcutaneous AsPC-1 tumor cells (1.5e⁶ SC) were left untreated (tumor only; TO) or treated with 3e⁶ CAR_MESO, IFNγKO CAR_MESO, or IFNγRKO CAR_MESO (IV). (D) Mice were bled weekly through day 35 post-CAR-T cell infusion and CAR-T cell persistence was determined using flow cytometry. Data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons test. (E) Tumor burden was assessed by weekly caliper measurements. Data are shown as mean ± SEM and analyzed by one-way ANOVA with Tukey’s multiple comparison test. Statistical significance was assessed on days 56 (IFNγRKO CAR_MESO (turquoise *) or CAR_MESO (purple *) vs. TO), 77 (IFNγRKO CAR_MESO (turquoise *) vs. IFNγKO CAR_MESO), and 97 (IFNγRKO CAR_MESO (turquoise *) vs. CAR_MESO). (F) Overall survival of mice treated with CAR-T cells was assessed by Mantel-Cox test. For C-F, two biological replicates were used at 4–5 mice/group for each donor. *P<0.05, **P<0.01.

Fig. 6.. Pancreatic tumors from IFNγRKO CAR_MESO-treated mice exhibit greater interferon signaling and apoptosis.

(A) Schematic diagram representing mouse NSG mice bearing subcutaneous AsPC-1 tumor cells (1.5e⁶ administered SC) which were left untreated (tumor only; TO) or treated with 3e⁶ CAR_MESO, IFNγKO CAR_MESO, or IFNγRKO CAR_MESO by IV. Tumors were collected 14 days post-CAR-T cell infusion, processed, and sorted into CAR-T cells and AsPC-1 tumor cells. (B) Representative flow cytometry plots of the GFP⁺ tumor and PE-TexasRed⁺ CAR-T cells isolated from the tumors of AsPC-1-bearing mice. (C and D) CAR_MESO and IFNγRKO CAR_MESO from before infusion (PRE) and post-infusion from the tumor (POST) (C) and tumor cells from mice receiving no treatment (TO) or treated with CAR_MESO or IFNγRKO CAR_MESO (D) were isolated and analyzed by NanoString. Z-scores for all genes in each group are displayed as a heatmap. (E) Interferon gene signatures are shown for mesothelin-specific CAR-T cells. Gene counts were normalized. Purple arrows indicate gene signatures which were expressed more in CAR_MESO, turquoise arrows indicate gene signatures expressed higher in the IFNγRKO CAR_MESO. (F) Cell death gene signatures were assessed by NanoString in CAR_MESO and IFNγRKO CAR_MESO T cells. Gene counts for anti-apoptotic (green arrow) and pro-apoptotic (red arrow) were normalized. (G and H) Interferon (G) and pro-apoptotic (H) gene signatures are shown by the scatterplot for CAR-T cell constructs pre and post infusion. Gene counts were normalized, and data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons test. (I) Interferon gene counts from tumors receiving no treatment (TO) or treated with CAR_MESO or IFNγRKO CAR_MESO were normalized with purple arrows indicating gene signatures which were expressed more in CAR_MESO-treated tumors and turquoise arrows indicating gene signatures expressed higher in IFNγRKO CAR_MESO-treated mice. (J) Cell death gene signatures were assessed by NanoString in tumors from untreated or CAR_MESO or IFNγRKO CAR_MESO-treated mice. Gene counts for anti-apoptotic (green arrow) and pro-apoptotic (red arrow) were normalized. (K and L) Interferon (K) and pro-apoptotic (L) gene signatures in tumors are shown by scatterplot for non-treated or CAR-T cell-treated mice. Gene counts were normalized, and data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons test. For this experiment, data was collected from 2 mice per group for each donor, with two donors total. Data shown in C to L are from 1 representative donor. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001; ns, not significant.

Fig. 7.. EGFR-targeting IFNγRKO CAR-T cells confer protection from tumor rechallenge.

(A) Schematic diagram of control and IFNγRKO CAR-T cells targeting EGFR generated from healthy donors. (B) Schematic of in vivo treatment protocol. NSG mice bearing subcutaneous AsPC-1 tumor cells (1.5e⁶ SC) were left untreated (tumor only; TO) or treated with 3e⁶ CAR_EGFR or IFNγRKO CAR_EGFR (IV) and tumor burden was measured by caliper weekly. Seventy days post-CAR-T cell infusion, mice displaying curative responses (3 mice/group) were re-challenged with 1.5e⁶ AsPC-1 tumor cells in the flank opposite of the initial tumor injection. Tumor burden was assessed using weekly caliper measurements. (C and D) Tumor size (C) and overall survival (D) were assessed in tumor only and EGFR-specific CAR-T cell-treated mice over a 200-day period. Mice were rechallenged on day 70. Significance between tumor only and both treated groups is shown at day 56, with IFNγRKO CAR_MESO vs CAR_MESO shown at day 143. Data are shown as mean ± SEM with P values determined by one-way ANOVA with Tukey’s multiple comparisons test (C; day 56), unpaired student’s t-test (C; day 143), or Mantel-Cox (D). 5 mice/group with 3 mice/group for re-challenge, taken from 2 donors. *P<0.05, **P<0.01,***P<0.001.

Fig. 8.. IFNγRKO CAR-T cells display increased efficacy and survival in a syngeneic model of triple-negative breast cancer.

(A) Schematic diagram of treatment protocol. Briefly, T cells were isolated from CD45.2⁺ C57BL/6 wildtype (WT) or Ifngr KO (IFNγRKO) mice and retrovirally transduced to express control (CAR_FITC) or B7H3-targeted (CAR_B7H3) CAR constructs containing a CD28 costimulatory domain prior to in vitro and in vivo assessment. (B and C) Transduction efficiency (B) and memory (C) were assessed by flow cytometry. Naïve (CD44^-CD62L⁺), central memory (CM; CD44⁺CD62L⁺), and effector memory (EM; CD44⁺CD62L^-) frequencies were determined for WT CAR and IFNγRKO CAR. Naïve cells were <2% for each group, making it not clearly seen on the graph. Data are shown as mean ± SEM with P values determined by unpaired t-test for 3 biological replicates. (D to G) 3e⁵ triple-negative breast cancer cells overexpressing B7H3 (E0771 B7H3) were injected into the mammary fat pad of CD45.1⁺ C57BL/6 mice. Tumor-bearing mice were conditioned with 100µg/g cyclophosphamide prior to intravenous (IV) administration of 5e⁶ CD8⁺ CAR-T cells and tumor size was monitored by caliper throughout the experiment and graphed as averages (D) and individual mice (E). On day 23, tumors were measured by caliper (F), collected, and frequencies of CAR-T cells were determined by flow cytometry (D). For in vivo experiment, data are derived from 5 mice per group. Data are shown as mean ± SEM with P values determined by two-way ANOVA (D), one-way ANOVA with Tukey’s multiple comparison test (F), and unpaired t-test (G). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001; ns, not significant.