Reprogramming lipid metabolism prevents effector T cell senescence and enhances tumor immunotherapy

Abstract

The functional state of T cells is a key determinant for effective antitumor immunity and immunotherapy. Cellular metabolism, including lipid metabolism, controls T cell differentiation, survival, and effector functions. Here, we report that development of T cell senescence driven by both malignant tumor cells and regulatory T cells is a general feature in cancers. Senescent T cells have active glucose metabolism but exhibit unbalanced lipid metabolism. This unbalanced lipid metabolism results in changes of expression of lipid metabolic enzymes, which, in turn, alters lipid species and accumulation of lipid droplets in T cells. Tumor cells and T_reg cells drove elevated expression of group IVA phospholipase A₂, which, in turn, was responsible for the altered lipid metabolism and senescence induction observed in T cells. Mitogen-activated protein kinase signaling and signal transducer and activator of transcription signaling coordinately control lipid metabolism and group IVA phospholipase A₂ activity in responder T cells during T cell senescence. Inhibition of group IVA phospholipase A₂ reprogrammed effector T cell lipid metabolism, prevented T cell senescence in vitro, and enhanced antitumor immunity and immunotherapy efficacy in mouse models of melanoma and breast cancer in vivo. Together, these findings identify mechanistic links between T cell senescence and regulation of lipid metabolism in the tumor microenvironment and provide a new target for tumor immunotherapy.

Fig. 1.. Regulatory T cells and tumor cells induce T cell senescence.

(A) Naïve CD4⁺ T cells were incubated alone or cocultured with nT_reg cells or control effector CD4⁺CD25⁻ T cells at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for 3 days and then stained for SA-β-gal. Arrows indicate SA-β-gal⁺ T cells. Scale bars, 20 μm. Data are presented as means ± SD from three independent experiments. n = 6 different representative CD4⁺ T cells. **P < 0.01. (B) Anti-CD3–activated naïve CD4⁺ T cells were cocultured with MCF-7 or M586 cells at a ratio of 1:1 for 1 day, then purified, and stained for SA-β-gal after culture for an additional 3 days. Arrows indicate SA-β-gal⁺ T cells. Scale bars, 20 μm. Data presented as means ± SD from three independent experiments. n = 3 different representative CD4⁺ T cells. *P < 0.05 and **P < 0.01. (C and D) CD4⁺ T cells were cocultured with nT_reg cells (C) or MCF-7 cells (D) as in (A) and (B), and changes in cytokine mRNA expression were measured relative to untreated CD4⁺ T cells 24 hours later. Expression was normalized to β-actin expression as a control. Data are presented as means ± SD from three independent experiments. n = 3 to 6 different representative CD4⁺ T cells. *P < 0.05 and **P < 0.01. (E to H) CD4⁺ and CD8⁺ T cells were isolated from blood, lymph nodes (LNs), spleens (SPs), and tumor tissues of B16F0 subcutaneous tumor (E and F)– or E0771 subcutaneous tumor (G and H)–bearing mice and tumor-free littermates when primary tumors reached 10 to 15 mm in diameters. SA-β-gal staining (E and G) and cytokine mRNA expression (F and H) were measured. Data shown are means ± SD from seven mice in each group. *P < 0.05 and **P < 0.01. (I) CD3⁺ TILs were isolated from freshly digested human melanoma (MTIL) and breast cancer (BTIL) tissues and stained for SA-β-gal⁺. Arrows indicate SA-β-gal⁺ T cells. Scale bars, 20 μm. Naïve CD4⁺ and CD8⁺ T cells from healthy donor peripheral blood were included as controls. Data presented as means ± SD. Each dot represents an individual donor. **P < 0.01. (J) Proinflammatory cytokine expression was measured in purified CD3⁺ TILs isolated from human melanoma or breast cancer tumors. Purified naïve CD4⁺ or CD8⁺ T cells from healthy donor peripheral blood were used as controls. Data shown are means ± SD. *P < 0.05. One-way analysis of variance (ANOVA) was performed in (A), (B), (C), (F), (H), (I), and (J). Paired Student’s t test was performed in (D), and unpaired Student’s t test was performed in (E) and (G).

Fig. 2.. Senescent T cells have increased expression of enzymes associated with glucose metabolism.

(A) Human naïve CD8⁺ T cells and nT_reg cells were isolated from PBMCs of healthy donors and cocultured at a ratio of 5:1 for indicated time points. Transcriptional analysis of nT_reg-treated CD8⁺ T cells was performed, and differentially expressed genes involved in glycolysis were identified. (B) Naïve CD8⁺ T cells were cocultured with nT_reg or control CD4⁺CD25⁻ T cells at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for 3 days, and glucose metabolism–related gene expression was evaluated by real-time quantitative PCR (qPCR). Expression of each gene was normalized to β-actin expression and adjusted to the expression in CD8⁺ T cells alone. Data shown are means ± SD from eight to nine different independent donors. Each dot represents one individual donor. *P < 0.05 and **P < 0.01. (C) Anti-CD3–activated naïve CD8⁺ T cells were cocultured with MCF-7 cells at a ratio of 1:1 for 2 days, and glucose metabolism–related gene expression was evaluated by real-time qPCR as in (B). Data shown are means ± SD from six different independent donors. Each dot represents one individual donor. *P < 0.05 and **P < 0.01. (D and E) CD8⁺ T cells were purified from blood and tumor tissues in melanoma B16F0-bearing (D) and breast cancer E0771-bearing (E) mice when primary tumors reached 10 to 15 mm in diameters, and mRNA expression of glucose metabolism–related genes was evaluated by real-time qPCR. CD8⁺ T cells purified from blood in tumor-free mice served as controls. Data shown are means ± SD from seven mice in each group. *P < 0.05 and **P < 0.01. (F) CD3⁺ cells were purified from melanoma (MTIL) and breast cancer (BTIL) tissues from patients and gene expression of glucose transporters, and key enzymes involved in glycolysis was evaluated by real-time qPCR. Naïve CD4⁺ or CD8⁺ T cells were purified from healthy donor blood and activated with plate-coated anti-CD3 (2 μg/ml) for 3 days for controls. Data shown are means ± SD from three to eight different independent donors. Each dot represents an individual donor or patient. *P < 0.05 and **P < 0.01. (G) Naïve CD4⁺ or CD8⁺ T cells were cocultured with nT_reg or control CD4⁺CD25⁻ T cells in anti-CD3–coated (2 μg/ml) plates in the presence of different concentrations of glucose and then stained for SA-β-gal after culture for 3 days. Data shown are means ± SD from three independent experiments. **P < 0.01. (H) Anti-CD3–activated naïve CD4⁺ or CD8⁺ T cells were cocultured with MCF-7 cells at a ratio of 1:1 in the presence of different concentrations of glucose for 1 day. The treated T cells were then separated and stained for SA-β-gal after culture for an additional 3 days. Data shown are means ± SD from three independent experiments. **P < 0.01. (I) Senescent CD8⁺ T cells induced by nT_reg or MCF-7 cells were cultured in the presence of normal (11 mM) or high (25 mM) dosages of glucose for 2 days and then stained for SA-β-gal. Data shown are means ± SD from three independent experiments. (J and K) Senescent CD8⁺ T cells induced by nT_reg cells (J) or MCF-7 cells (K) were cultured in the presence of glycolysis inhibitors 3-BrPA (30 μM) or 2-DG (2 mM), and mRNA expression of each cytokine was determined relative to untreated senescent CD8⁺ T cells 24 hours later by real-time qPCR. Expression was normalized to β-actin as a control. Data are means ± SD from three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001. One-way ANOVA was performed in (B), (D), (E), and (F). Paired Student’s t test was performed in (C), and unpaired Student’s t test was performed in (G), (H), (J), and (K).

Fig. 3.. Senescent T cells have unbalanced lipid metabolism.

(A) Transcriptional analysis of nT_reg-treated CD8⁺ T cells was performed as in Fig. 2A. Alterations of genes involved in lipid metabolism were identified and normalized to log₂ expression at different time points. (B) Naïve CD8⁺ T cells were cocultured with nT_reg at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for indicated time points. Lipid metabolism–related enzyme gene expression was evaluated by real-time qPCR. Expression of each gene was normalized to β-actin expression and normalized relative to expression in CD8⁺ T cells at 8 hours. Data shown are means ± SD from four to six different independent donors. Paired Student’s t test was performed. *P < 0.05 and **P < 0.01. (C) Anti-CD3–activated CD8⁺ T cells were cocultured with MCF-7 cells at a ratio of 1:1 for different times. The treated CD8⁺ T cells were then separated, and lipid metabolism–related gene expression was evaluated by real-time qPCR. Expression of each gene was normalized to β-actin expression and adjusted to expression in CD8⁺ T cells at 8 hours. Data shown are means ± SD from six different independent donors. Paired Student’s t test was performed. *P < 0.05 and **P < 0.01. (D) CD3⁺ TILs were purified from melanoma (MTIL) and breast cancer (BTIL) tissues, and gene expression of key enzymes involved in lipid metabolism was evaluated by real-time qPCR. CD4⁺ or CD8⁺ T cells from healthy donors activated with plate-coated anti-CD3 (2 μg/ml) for 3 days served as controls. Data shown are means ± SD from three to eight different independent donors. Each dot represents an individual donor or patient. One-way ANOVA and unpaired Student’s t test were performed. *P < 0.05. (E) Naïve CD4⁺ or CD8⁺ T cells were cocultured with nT_reg or control CD4⁺CD25⁻ T cells at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for 3 days. Treated T cells were purified, stained with Fillipin III, and then analyzed for cholesterol content by fluorescence microscopy. Scale bars, 20 μm. (F) CD3⁺ TILs were purified from breast cancer and melanoma tissues isolated from patients and stained with Fillipin III as in (E). Anti-CD3 (2 μg/ml) preactivated CD4⁺ and CD8⁺ T cells purified from healthy donors served as controls. Scale bars, 20 μm.

Fig. 4.. Alterations of lipid species in senescent T cells induced by T_reg cells and tumor cells.

(A and B) Naïve CD8⁺ T cells were cocultured with nT_reg or control CD4⁺CD25⁻ T cells at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for 3 days. Treated CD8⁺ T cells were purified, and total and multiple molecular species in free fatty acid (FFA) (A) and cholesteryl ester (CE), phosphatidylcholine (PC), phosphatidylethanolamine (PE), and Cer (B) from different treatment groups were subjected to mass spectrometry. Data shown are means ± SD from T cells purified from two to three independent donors. *P < 0.05, **P < 0.01, and ***P < 0.001. (C) Anti-CD3–activated naïve CD4⁺ T cells were cocultured with MCF-7 cells at a ratio of 1:1 for 1 day. The treated CD4⁺ T cells were then separated, and lipid extracts of T cells from different treatment groups were analyzed by mass spectrometry after an additional 3-day culture. Data shown are means ± SD from T cells purified from four different independent donors. *P < 0.05 and **P < 0.01. (D) Summary of lipidomic analysis results in senescent T cells induced by T_reg and tumor cells as described in (A) to (C). (E) Naïve CD4⁺ T cells were cocultured with nT_reg cells at a ratio of 4:1 in anti-CD3–coated (2 μg/ml) plates in the presence or absence of the indicated concentrations of lipid fractions for 3 days and then stained for SA-β-gal. Data shown are means ± SD from three independent experiments. n = 4 to 6 different representative CD4⁺ T cells. **P < 0.01. Unpaired Student’s t test was performed in (A) and (C). One-way ANOVA was performed in (B) and (E).

Fig. 5.. Accumulated LDs contribute to the development of T cell senescence mediated by T_reg and tumor cells.

(A and B) Naïve CD4⁺ and CD8⁺ T cells were cocultured with nT_reg cells (A) or with MCF-7 or M586 tumor cells (B). The cocultured CD4⁺ and CD8⁺T cells were purified and stained for Oil Red O to measure lipid droplet (LD) accumulation. Oil Red O ⁺ T cells are indicated by arrows. Scale bars, 20 μm. Data shown are means ± SD from three independent experiments. n = 3 to 5 different representative T cells. ***P < 0.001. (C) T_reg-treated senescent T cells were stained with BODIPY 493/503, sorted on the basis of BODIPY 493/503 fluorescence by FACS, and measured for SA-β-gal expression. Scale bars, 20 μm. Data shown are means ± SD from three different independent donor T cells. **P < 0.01 and ***P < 0.001. (D) Naïve CD4⁺ T cells were cocultured with nT_reg cells at a ratio of 4:1 in anti-CD3–coated (2 μg/ml) plate in the presence or absence of the indicated concentrations of lipid species for 3 days and then purified and stained for Oil Red O. Data shown are means ± SD from three independent experiments. n = 3 to 6 different representative CD4⁺ T cells. *P < 0.05 and **P < 0.01. (E and F) Naïve CD4⁺ and CD8⁺ T cells were cocultured with nT_reg cells (E) or MCF-7 cells (F), and mRNA expression of ACAT1 and ACAT2 was determined by the real-time qPCR. Expression was normalized to β-actin expression and adjusted to expression in T cells cultured alone. Data are presented as means ± SD from three to four independent experiments. *P < 0.05 and **P < 0.01. (G and H) Naïve T cells were pretreated with ACAT inhibitor avasimibe (Ava) for 24 hours and then cocultured with nT_reg cells at a ratio of 4:1 in anti-CD3–coated (2 μg/ml) plates for 3 days. The treated CD4⁺ and CD8⁺ T cells were purified and stained for SA-β-gal (G) and Oil Red O (H), respectively. Data shown are means ± SD from three independent experiments. n = 5 to 10 different representative T cells. **P < 0.01. (I) Naïve CD8⁺ T cells were cocultured with nT_reg cells at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for indicated time points, and mRNA expression of lipase A (LIPA) was evaluated by real-time qPCR. Expression was normalized to β-actin expression and adjusted to expression in T cells at 8 hours. Data are means ± SD from four independent experiments. **P < 0.01. (J and K) Naïve CD8⁺ T cells were cultured in anti-CD3–coated (2 μg/ml) plates in the presence or absence of lipase inhibitor orlistat (5 μM) for 3 days. The treated CD8⁺ T cells were stained for SA-β-gal (J) and Oil Red O (K). Data shown are means ± SD from five different independent donor T cells. **P < 0.01. One-way ANOVA was performed in (A) to (E), (G), and (H). Paired Student’s t test was performed in (F) and (I) to (K).

Fig. 6.. Elevated cPLA₂α promotes LD accumulation and induction of senescence in T cells.

(A and B) Naïve CD8⁺ and CD4⁺ T cells were cocultured with nT_reg cells (A) or MCF-7 cells (B) for indicated time points, and mRNA expression of cPLA₂α was determined by real-time qPCR. Expression was normalized to β-actin expression and adjusted to expression in T cells cultured alone at 4 hours. Data shown are means ± SD from three independent experiments. *P < 0.05 and **P < 0.01. (C) Naïve CD4⁺ and CD8⁺ T cells were cocultured with or without T_reg cells or control CD4⁺CD25⁻ T cells at a ratio of 4:1 in the presence of plate-bound anti-CD3 (2 μg/ml) for 3 days. cPLA₂α protein in treated T cells was analyzed by Western blot. nT_reg-1 and nT_reg-2 cells are nT_reg cells purified from two individual healthy donors. (D and E) T cells were purified from blood and tumors of melanoma B16F0 tumor–bearing (D) and breast cancer E0771 tumor–bearing (E) mice, as described in Fig. 1. mRNA expression of cPLA₂α was evaluated by real-time qPCR. Data shown are means ± SD from seven mice in each group. *P < 0.05 and **P < 0.01. (F) Immunofluorescence after staining with anti-cPLA₂α antibody and BODIPY 493/503 was evaluated in responder CD4⁺ T cells treated with nT_reg or control CD4⁺CD25⁻ effector T cells. Scale bars, 20 μm. DAPI was used to stain cell nuclei. (G) Naïve CD4⁺ and CD8⁺ T cells were pretreated with the cPLA₂α inhibitor MAFP for 24 hours and then cocultured with nT_reg cells at a ratio of 4:1 in anti-CD3–coated (2 μg/ml) plates for 3 days. The treated CD4⁺ and CD8⁺ T cells were purified and stained for Oil Red O. Data are means ± SD from three independent experiments. **P < 0.01. (H) Anti-CD3 (2 μg/ml)–preactivated naïve CD4⁺ and CD8⁺ T cells were transfected with siRNAs specific for PLA2G4A or control (Ctr) nontargeting siRNA for 2 days. mRNA expression of cPLA₂α was determined by real-time qPCR. Expression was normalized to β-actin expression and adjusted to expression in T cells cultured in medium. Data shown are means ± SD from three independent experiments. **P < 0.01 and ***P < 0.001. (I) Anti-CD3 (2 μg/ml)–preactivated naïve CD4⁺ and CD8⁺ T cells were transfected with siRNAs specific for PLA2G4A or Ctr-siRNA and cultured for 2 days. The transfected T cells were further cocultured with nT_reg cells at a ratio of 4:1 for 2 days and were subsequently purified and stained for Oil Red O. Data shown are means ± SD from three independent experiments. **P < 0.01. (J) Cells were treated and cultured as in (G). Protein concentrations of cPLA₂α, P53, and P21 in treated CD4⁺ T cells were analyzed by Western blot and further quantitatively analyzed and compared against GAPDH expression using densitometry. Results shown are means ± SD from three independent experiments. **P < 0.01. (K and L) CD4⁺ and CD8⁺ T cells were pretreated with cPLA₂α inhibitor MAFP (K) or transfected with siRNAs specific for PLA2G4A (L) and cocultured with nT_reg cells. The treated CD4⁺ and CD8⁺ T cells were purified and stained for SA-β-gal. Data shown are means ± SD from three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001. (M) Flow cytometry analysis showed that inhibition of cPLA₂α with MAFP restored costimulatory molecules CD27 and CD28 in responder T cells. The gray-shaded histogram indicates T cells with the isotype control antibody staining. Unpaired Student’s t test was performed in (A) and (B). One-way ANOVA was performed in (D), (E), (G), and (H) to (L).

Fig. 7.. MAPK and STAT signaling regulate cPLA₂α and lipid metabolism in senescent T cells.

(A and B) Anti-CD3–activated CD8+ T cells were pretreated with or without the ATM inhibitor KU55933 (10 μM) for 1 day and then cocultured with nT_reg cells at a ratio of 4:1 for 24 hours (A) or 3 days (B). The treated CD8⁺ T cells were purified and cPLA₂α mRNA and protein expression were evaluated using real-time qPCR (A) and flow cytometry (B), respectively. Data shown in (A) were normalized to β-actin expression, adjusted to expression in CD8⁺ T cells without KU55933 treatment, and are presented as means ± SD from four different independent donor T cells. **P < 0.01. The gray shaded histogram (B) indicates T cell control with the secondary antibody staining. (C and D) Naïve CD4⁺ and CD8⁺ T cells were pretreated with KU55933 (10 μM) for 24 hours and then cocultured with nT_reg cells at a ratio of 4:1 in anti-CD3 coated (2 μg/ml) plate for 3 days. The treated CD4⁺ and CD8⁺ T cells were purified and stained for SA-β-gal (C) and Oil Red O (D), respectively. Data shown are means ± SD from three independent experiments. *P < 0.05 and **P < 0.01. (E and F) Immunofluorescence staining showing colocalization of MAPK (E) or STAT (F) molecules with LDs in responder T cells treated with T_reg or control CD4⁺CD25⁻ effector T cells using BODIPY 493/503 and antibodies against p-ERK, p-P38, p-STAT1, or p-STAT3. Scale bars, 20 μm. DAPI was used to stain cell nuclei. (G) Naïve CD4⁺ T cells were pretreated with indicated inhibitors for 24 hours and then cocultured with nT_reg cells at a ratio of 4:1 in anti-CD3–coated (2 μg/ml) plates for 3 days. cPLA₂α protein concentration in treated CD4⁺ T cells were determined by Western blot, quantitatively analyzed, and compared against GAPDH expression with a densitometer. Data shown are means ± SD from three independent experiments. **P < 0.01. (H and I) Responder CD4⁺ and CD8⁺ T cells were treated with indicated inhibitors and nT_reg cells, subsequently purified, and stained for Oil Red O (H) and SA-β-gal (I), respectively. Data shown are means ± SD from three independent experiments. **P < 0.01. One-way ANOVA was performed in (A), (C), (D), and (G) to (I).

Fig. 8.. Reversal of T cell senescence via reprogramming of T cell lipid metabolism enhances antitumor immunity in vivo.

Mouse B16F10 tumor cells (2 × 10⁵ per mouse) were subcutaneously injected into C57BL/6 mice. Activated Pmel-1 T cells (1.7 × 10⁶ per mouse) were adoptively transferred through intravenous (i.v.) injection into B16F10 tumor–bearing mice at day 6 after tumor inoculation. MAFP (7.5 mg/kg per mouse) was injected intraperitoneally into the mice at days 1, 4, 7, and 10 after T cell transfer to inhibit cPLA₂α. (A) Tumor volumes were measured and presented as means ± SD (n = 6 mice per group). (B) Survival was evaluated with Kaplan-Meier analysis (n = 11 to 12 mice per group). (C and D) Transferred Pmel-1 T cells were isolated from blood, LNs, SPs, and tumors (TILs) in B16F10 tumor–bearing mice at day 28 after tumor injection and then stained for Oil Red O (C) and SA-β-gal (D), respectively. Data shown are means ± SD from four to five mice each group. **P < 0.01. (E) Mouse E0771 tumor cells transduced with retroviral vector encoding melanoma tumor antigen gp100 were subcutaneously injected into NSG mice at day 0. Activated Pmel-1 T cells were adoptively transferred into tumor-bearing mice at day 6. MAFP were administered intraperitoneally at days 1, 4, 7, and 10 after adoptive transfer of Pmel-1 T cells. Tumor sizes were measured and presented as means ± SD (n = 5 mice per group). (F and G) The transferred Pmel-1 T cells were isolated from blood, SPs, and tumors (TIL) of E0771-bearing mice at day 24 after tumor injection and then stained for Oil Red O (F) and SA-β-gal (G), respectively. Data shown are means ± SD (n = 4 to 5 mice per group). *P < 0.05 and **P < 0.01. (H) Mouse B16F10 tumor cells (2 × 10⁵ per mouse) were subcutaneously injected into C57BL/6 mice. Activated Pmel-1 T cells (2 × 10⁶ per mouse) were adoptively transferred into B16F10 tumor–bearing mice at day 11 after tumor inoculation by intravenous injection. MAFP (15 mg/kg per mouse) was administered intraperitoneally at days 1, 4, 7, and 10 after T cell transfer. Tumor volumes were measured and presented as means ± SD (n = 5 to 6 mice per group). (I) Human 586mel tumor cells (5 × 10⁶ per mouse) were subcutaneously injected into NSG mice. Tumor-specific CD8⁺ TIL586 cells (5 × 10⁶ per mouse) were intravenously injected on day 5 after tumor injection. MAFP (7.5 mg/kg per mouse) was administered intraperitoneally at days 1, 4, 7, and 10 day after adoptive transfer of CD8⁺ TIL586 T cells. Tumor volumes were measured and presented as means ± SD (n = 5 to 6 mice per group). (J and K) The transferred human TIL586 T cells were isolated from blood, SPs, and tumors (TIL) of 586mel tumor-bearing mice at day 39 after tumor injection and then stained for Oil Red O (J) and SA-β-gal (K), respectively. **P < 0.01. One-way ANOVA and unpaired Student’s t test were performed in (A), (E), (H), and (I). Unpaired Student’s t test was performed in (C), (D), (F), (G), (J), and (K).