Bioengineering the metabolic network of CAR T cells with GLP-1 and Urolithin A increases persistence and long-term anti-tumor activity

Abstract

Constant tumor antigen exposure disrupts chimeric antigen receptor (CAR) T cell metabolism, limiting their persistence and anti-tumor efficacy. To address this, we develop metabolically reprogrammed CAR (MCAR) T cells with enhanced autophagy and mitophagy. A compound screening identifies a synergy between GLP-1R agonist (semaglutide [SG]) and Urolithin A (UrA), which activate autophagy through mTOR (mechanistic target of rapamycin) inhibition and mitophagy via Atg4b activation, maintaining mitochondrial metabolism in CAR T cells (MCAR T-1). These changes increase CD8⁺ T memory cells (Tm), enhancing persistence and anti-tumor activity in vitro and in xenograft models. GLP-1R knockdown in CAR T cells diminishes autophagy/mitophagy induction, confirming its critical role. We further engineer GLP-1-secreting cells (MCAR T-2), which exhibited sustained memory, stemness, and long-term persistence, even under tumor re-challenge. MCAR T-2 cells also reduce cytokine release syndrome (CRS) risks while demonstrating potent anti-tumor effects. This strategy highlights the potential of metabolic reprogramming via targeting autophagy/mitophagy pathways to improve CAR T cell therapy outcomes, ensuring durability and efficacy.

Keywords: CAR T cells; GLP-1 peptide; T cell persistence; Urolithin A; anti-tumor activity; autophagy; metabolism; mitochondrial health; mitophagy.

Graphical abstract

Figure 1

Functional and metabolic characterization of CAR T cells after repeated tumor re-challenge (A) Schematic of the lymphoma model. Mice were injected with Raji cells (0.5 × 10⁶) on day 0, followed by CAR T cell (5 × 10⁶) administration on day 5. TR was done on days 12, 19, and 26 with 0.2 × 10⁶ Raji cells. Blood was collected on days 14, 21, and 28. (B) Sorting and flow cytometry gating of human CD3⁺ CAR T cells from blood samples across time points (3 mice were pooled for a total of 24 in each group to obtain n = 8 per condition). (C–F) Flow cytometric analysis showing expression of Granzyme B (C), Perforin (D), PD-1 (E), and LAG-3 (F) in CON and TR groups over time (n = 8; mean ± SD). The data are represented as mean fluorescence intensity (MFI) calculated from the flow cytometry histograms. (G–I) Percentage of CAR T cells in blood tracked over time (G). IL-2 (H) and IFN-γ (I) secretion quantified from co-culture of CAR T cells with Raji cells for 24 h (n = 8). (J and K) Human CD8⁺ T cell subsets (T_n, T_scm, T_cm, T_em, and T_eff) in CD3⁺ T cells measured by flow cytometry in TR vs. CON groups at each time point. The pie chart shows the percentage of T_ex in the T_eff population. (L and M) Sorting of human CD8⁺ cells from pooled blood samples (3 mice were pooled for a total of 24 in each group to obtain n = 8 per condition). Representative panels showing the purity of CD3⁺ and CD8⁺ cells. (N and O) Live-cell Ca²⁺ imaging showing representative images (after 300 s) of cells transduced with the mitochondrial Ca²⁺ indicator (GCaMP6f) and cytosolic (RCaMP), upon histamine stimulation, which was confirmed by quantitative analysis over time. (P and Q) OCR and ECAR measured using Seahorse XF in CON and TR CAR T cells with respective bar graphs showing the basal respiration (n = 6). (R1) Representative image of mtDNA, indicated by staining the cells anti-TFAM (Mitochondrial transcription factor A). (R2) Quantitative analysis of images (n = 11 images) was done and represented as integrated density of TFAM signal. (S) Fluorescent assessment of mitophagy by tracking LTDR/MTG ratios in CON and TR CAR T cells following FCCP treatment for 60 min. (T1) Representative images of CAR T cells, which were transduced with LC3-mCherry and mito-GFP showing co-localization of LC3 (red) with mitochondria (green). Blue arrowheads show co-localization, and white arrowheads indicate lack of co-localization. Cells were treated with FCCP for 30 min. (T2) Quantitative analysis of the LC3 images (n = 10 images). (U) p62, Beclin-1, and Atg14 expression measured by flow cytometry in CON and TR CAR T cells (n = 6). Data represents mean ± SEM. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. A non-parametric t test and one-way ANOVA were used for statistical analysis. Scale bars: 10 μm.

Figure 2

Decline in mitophagy/autophagy promotes dysfunction in CD8⁺ CAR T cells (A) Illustration of 3^rd generation CAR and flow cytometric analysis of CAR expression in CD8⁺ T cells. The y axis shows the total cell count, and the x axis represents the CAR-PE-positive cells. (B) CAR T cells were co-cultured with Raji cells (5:1 ratio) and then re-challenged with Raji cells every alternate day from day 6 to day 20 for functional analysis. (C) Analysis included CAR T cell multi-functionality, anti-tumor activity, exhaustion markers, and proliferation. (D) Heatmap of the various parameters showing anti-tumor activity by BLI and indicated as % lysis of tumor cells; expression of IFN- γ and IL-2 by ELISA; analysis of mtROS and TMRE by flow cytometry; mtDNA analysis by image quantitation; PD-1, LAG-3, and TIM-3 expression by flow cytometry; and cell proliferation by carboxyfluorescein succinimidyl ester (CFSE) staining (n = 6). The data are presented as MFI. (E1–E3) Correlation plots showing association between CAR T cell exhaustion and mitochondrial dysfunction. (F) Representative immunofluorescence images of Tom20-stained cells showing mitochondrial shape change in CAR T cells. (G) Quantification of various mitochondrial shapes in CON and TR groups (n = 6). (H) Schematic of reporter used to track mitophagy in live CAR T cells. Ratio of mCherry/GFP fluorescence indicates mitophagy progression. (I and J) Flow cytometric analysis showing mitophagy flux (mCherry/GFP ratio) over 60 min FCCP treatment. (K) Flow cytometry analysis showing the ratio of RFP to GFP as an indicator of autophagy (n = 6). (L) Transcriptomic analysis of existing patient-derived CAR T single-cell RNA sequencing data highlighting differential expression of autophagy-related genes, including BECN1 and ATG14, between CON and TR groups. (M) Beclin-1 and Atg14 expression in CAR T cells after TR suggesting alteration in autophagosome formation (n = 8). (N) Schematic of shRNA-mediated knockdown of ATG14 and BECN1 in CAR T cells. (O and P) MFI of PD-1 (O) and LAG-3 (P) measured by flow cytometry in CAR T cells transduced with scrambled (Scram) or shRNA targeting ATG14/BECN1 (n = 8). (Q and R) Ratio of mCherry/GFP or RFP/GFP fluorescence in mitophagy- and autophagy-deficient CAR T cells, respectively, following 60 min FCCP treatment (n = 7). (S) T cell subset analysis of CD8⁺ T cell (T_n, T_cm, T_em, and T_eff) in CAR T cells transduced with Scram or shRNA targeting ATG14/BECN1 (n = 6). (T) Proliferation rate over time in scrambled and shRNA knockdown groups (n = 6). (U) Analysis of CAR T cell persistence in vitro, comparing Scram and shRNA groups in the 21-day TR model (n = 6). Data represents mean ± SEM. ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. A non-parametric t test was used for statistical analysis. Scale bar: 10 μm.

Figure 3

SG and UrA enhance autophagy and reduce CAR T cell dysfunction (A) Schematics of workflow of test compound screening for autophagy and mitophagy using reporter Jurkat stable cells. (B and C) Heatmaps showing the effects of single, combinations, and dose titrations on autophagy (B) and mitophagy (C), assessed using GFP-RFP reporter systems. The blue scale indicates the intensity of activity (n = 3, mean). (D) Schematic of T cell isolation, CAR T cell transduction, drug treatment (SG and UrA), and TR (every alternate day from day 6 to day 20). CAR T cell collection was done on day 21. (E and F) Flow cytometric analysis of RFP/GFP ratios for autophagy (E) and mCherry/GFP for mitophagy (F) in TR CAR T cells treated with VEH, SG, UrA, or SG + UrA (n = 5). (G) Transcriptomic analysis of autophagy-related gene expression in TR CAR T cells, including Beclin-1 and Atg14, across different treatment conditions (SG, UrA, SG + UrA, and nGLP-1). (H and I) MFI of Beclin-1(H) and Atg14 (I) in CAR T cells treated with VEH, SG, UrA, SG + UrA, or nGLP-1 (H: n = 7; I: n = 8). (J) GLP-1R expression in activated (Act) vs. unactivated (Un) CAR T cells measured by flow cytometry (n = 8). (K) Representative images showing GLP-1R (red) and DAPI (blue) staining in activated and unactivated CAR T cells. (L) Flow cytometric analysis of GLP-1R expression across different T cell subsets (T_n, T_eff, T_cm, and T_ex) in activated CAR T cells (n = 7). (M) MFI of MitoSOX staining in CAR T cells transduced with scrambled shRNA or GRshRNA and treated with SG + UrA (n = 8). (N) Measurement of mitophagy flux in CAR T cells transduced with scrambled control (Scram) or GRshRNA using LTDR/MTG ratios after FCCP treatment (n = 6). (O1) Similarly, flow cytometric analysis of CYTO-ID showing autophagy. (O2) MFI of CYTO-ID (n = 8). (P) The cAMP activity assay with relative light unit (RLU) depicted as the inverse relationship between cAMP concentration and RLU (n = 6), with varying concentrations of SG. Data represent mean ± SE; from three independent experiments. ∗p < 0.01; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. A non-parametric t test and two-way ANOVA were used for statistical analysis between groups. Scale bar: 10 μm.

Figure 4

Analysis of mTOR activity, autophagy, and CAR T cell function (A) mTOR activity measured by flow cytometry in CAR T cells treated with VEH, SG (Scram and GRshRNA transduced), and MHY1485 as positive control (n = 6). (B) MFI of CYTO-ID (n = 6). (C and D) IL-2 secretion (C) and Granzyme B expression (D) measured by ELISA and flow cytometry, respectively, in CAR T cells transduced with Scram and GRshRNA or treated with VEH or MHY1485 cultured in the presence of SG (n = 6). (E) Analysis of CAR T cell persistence under various conditions. (F) Similarly, percentage of cancer cell survival after co-culture with CAR T cells (n = 5). (G1 and G2) Representative images of LC3 staining (green) showing autophagosome formation in CAR T cells treated with VEH, SG, UrA, SG + UrA, and nGLP-1. Red arrows indicate LC3 puncta. Quantification of LC3 puncta per cell is shown in (G2) (n = 14 images). (H1 and H2) Western blot analysis of LC3-I and LC3-II expression in CAR T cells. Quantification of LC3-II/LC3-I ratio is shown in (H2) (n = 5). (I) Schematic representation of the FRET-LC3II plasmid construction. The relative fluorescence unit (RFU) ratio of 530–475 nm (YFP/CFP) was calculated to quantify the extent of LC3B cleavage, where a decrease in the ratio indicated increased substrate cleavage. (J) FRET analysis using a CFP-LC3-YFP reporter. In vitro assay conducted by incubating purified Atg4b protein with CFP-LC3-YFP in the presence of VEH, UrA, SG, or nGLP-1. FRET signals were recorded at 0, 5, and 15 min (n = 5). (K) Structural model of Atg4b interaction with its substrates, highlighting key amino acids involved in the interaction, based on in silico docking predictions. (L) Fluorescence intensity (RFU ratio 530 nm/475 nm) over time in CAR T cells expressing wild-type and mutated forms (T28A, H300A, and L11A) of Atg4b, treated with UrA (n = 5). (M) Relative expression of LC3 normalized to ACTB in CAR T cells transduced with Scram or LC3-targeted shRNA (n = 5). (N) Ratio of LTDR/MTG after FCCP treatment for 60 min (n = 8). (O) MFI of Atg14 in CAR T cells after treatment (n = 6). Data represent mean ± SEM.∗p < 0.01; ∗∗p < 0.01; ∗∗∗∗p < 0.0001; (ns, not significant). A non-parametric t test was used for statistical analysis. Scale bar: 50 μm.

Figure 5

Anti-tumor activity and CAR T cell persistence after treatment with SG, UrA, or their combination in Raji and NALM6 tumor models (A) Schematic of the Raji tumor model. CAR T cells were isolated, transduced, and treated with SG + UrA in vitro (MCAR T-1), maintained for 7 days. Raji cells were injected into mice, followed by CAR T cell infusion (1 × 10⁷ cells) on day 5. Weekly in vivo imaging was conducted from day 7 to day 56, and weekly blood collection till day 70. (B) Representative images of tumor burden in mice treated over time (days 7, 14, 21, 42, and 56) (n = 4). (C) Quantification of bioluminescence radiance (photons/sec/cm²/sr) in Raji tumor-bearing mice over time (n = 5; at day 21; n = 1 for Raji group). (D) CAR T cell persistence in the blood at various time points till day 70 (n = 5). (E) Quantification of CAR T cell DNA copies per μg in blood samples over time. (F) Kaplan-Meier survival curve showing the survival of mice (n = 5). (G) Schematic of the NALM6 tumor model, with similar isolation, transduction, and treatment of CAR T cells (VEH, SG, UrA, and SG + UrA [MCAR T-1]) or direct administration of SG, UrA, or SG + UrA to mice. NALM6 cells were injected into mice, followed by CAR T cell infusion (5 × 10⁶ cells) on day 5. In vivo imaging was performed weekly from day 7 to day 56. (H) Representative images of tumor burden in mice over time (days 7, 14, and 28; n = 3 at day 21; n = 1 for Raji group). (I) Quantification of bioluminescence radiance (photons/sec/cm²/sr) in NALM6 tumor-bearing mice (n = 5). (J) Percentage of CAR T cells in blood at different time points (n = 5). (K) Survival plot of the mice over 70-day time period (n = 5). Data represent mean ± SEM. ∗∗∗∗p < 0.0001. A non-parametric t test and the Mantel-Cox test were used for statistical analysis.

Figure 6

Engineering and evaluation of nGLP-1-secreting UrA-treated CAR T cells for autophagy, mitophagy, and anti-tumor activity (A) Various GLP-1 CAR constructs with nGLP-1 positioned at the N or C terminus of the CAR molecule, separated by P2A or T2A cleavage sites, including versions with a furin cleavage (FC) site and secretory signal peptide (SP). (B) Levels of GLP-1 secretion from CAR T cells with different constructs measured by ELISA at day 7 after transduction. (n = 6). (C1 and C2) Contour plots and analysis showing CAR expression and percentage positive CAR-expressing cells (n = 6). (D) cAMP levels in GLP-1R-expressing CHO cells exposed to supernatants (collected after 72 h of transduction) from CAR T cells transduced with various GLP-1 CAR constructs (n = 6). (E) Schematic of scGLP-1/CAR T cell treatment with UrA (referred as MCAR T-2). (F) Measurement of cAMP release in CAR T cells or scGLP-1/CAR T cells treated with the latter also transduced with Scram or GRshRNA (n = 6). (G and H) Flow cytometric analysis showing MFI of mtROS (G) and ΔΨm (TMRE, H) in CAR T or MCAR T-2 cells transduced with Scram or GRshRNA (n = 8). (I) OCR measured by Seahorse XF analyzer; the bar graph shows basal respiration (n = 6). (J) Similarly, ECAR measured by Seahorse XF analyzer; the bar graph shows basal respiration (n = 6). (K and L) Ratio of LTDR/MTG (K) and percentage of autophagy (L). (M and N) IL-2 secretion (M) and Granzyme B expression (N) measured by ELISA and flow cytometry (n = 6). (O–Q) Expression of PD-1 (O) and LAG-3 (P) measured by flow cytometry and represented as MFI. T cell subset analysis (Q) shows the percentage of different memory subsets (T_n, T_scm, T_cm, and T_em) after treatment (n = 6). (R and S) Persistence of CAR T cells (R) and percentage cancer cell survival (S) in co-culture assays (n = 5 or 6). Data represent mean ± SEM. ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001; ns, not significant. A non-parametric t test, one-way ANOVA, and the Mantel-Cox test were used for statistical analysis.

Figure 7

MCAR T-2 cells induce robust anti-tumor activity and improve long-term persistence in vivo (A) Schematic of the in vitro and in vivo experimental setup for Raji tumor-bearing mice. CAR T cells were transduced and cultured in UrA-containing media (MCAR T-2). Raji cells were injected into mice on day 0, followed by CAR T cell infusion on day 5. (B) Representative BLI images of mice showing tumor burden over time. Images were taken at days 7, 14, 21, and 42 (n = 4). (C) Quantification of tumor burden (radiance) over time (photons/sec/cm²/sr). (D) Kaplan-Meier survival curve. (E) Percentage of CAR T cells in blood over time. (F) CAR DNA copies measured by qPCR at different time points (n = 5 at day 21; n = 1 for Raji group). (G) GLP-1 levels in serum measured at days 7, 14, and 21 by ELISA. (H) Correlation of GLP-1 levels with CAR T cell percentage in blood at different time points (r² = 0.99) (n = 5). (I) Schematic of the tumor re-challenge model; mice were injected with Raji cells and treated with CAR T or MCAR T-2 cells on day 5. TR was induced on days 12, 19, and 26. (J) Percentage of CAR T cells in blood after TR. (K and L) (K) OCR and (L) ECAR measured by Seahorse XF analyzer, showing improved mitochondrial function in MCAR T-2-treated mice (n = 5). (M) Basal respiration plots of OCR and ECAR (n = 6). (N and O) Mitophagy and autophagy analysis. (N) Ratio of LTDR/MTG in CAR T or scGLP-1/CAR T^UrA-treated mice. (O) Percentage of autophagy-positive cells (n = 8). (P) Flow cytometric analysis of T cell subsets (Tn, Tscm, Tcm, and Tem) in CAR T or scGLP-1/CAR T^UrA-treated mice after TR (n = 5). Data represent mean ± SEM. ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. A non-parametric t test, one-way ANOVA, and the Mantel-Cox test were used for statistical analysis.