Sensitivity to Restimulation-Induced Cell Death Is Linked to Glycolytic Metabolism in Human T Cells

Abstract

Restimulation-induced cell death (RICD) regulates immune responses by restraining effector T cell expansion and limiting nonspecific damage to the host. RICD is triggered by re-engagement of the TCR on a cycling effector T cell, resulting in apoptosis. It remains unclear how RICD sensitivity is calibrated in T cells derived from different individuals or subsets. In this study we show that aerobic glycolysis strongly correlates with RICD sensitivity in human CD8⁺ effector T cells. Reducing glycolytic activity or glucose availability rendered effector T cells significantly less sensitive to RICD. We found that active glycolysis specifically facilitates the induction of proapoptotic Fas ligand upon TCR restimulation, accounting for enhanced RICD sensitivity in highly glycolytic T cells. Collectively, these data indicate that RICD susceptibility is linked to metabolic reprogramming, and that switching back to metabolic quiescence may help shield T cells from RICD as they transition into the memory pool.

Figure 1. Increased RICD sensitivity in glycolytic CD8⁺ T cells

(A) Activated T cells from 3 normal donors (1-3) were restimulated with OKT3 Ab. Percent cell loss was measured 24 hrs later by PI staining and flow cytometry. (B) L-Lactate was measured in T cell supernatants by ELISA after 4 hrs of OKT3 restimulation. (C) Linear regression analysis comparing maximum % cell loss versus L-lactate production for 12 independent donors, including 95% confidence interval (dashed line). Pearson correlation R=0.8722, R²=0.7695 and p=0.0002. (D) Seahorse analysis of extracellular acidification rate (ECAR) of 6 different donors (4-9). (E) RICD sensitivity of the same 6 donors (4-9) used for Seahorse analysis (F) Linear regression analysis comparing maximum % cell loss versus maximum ECAR for 6 independent donors, including 95% confidence interval (dashed line). Pearson correlation R=0.9142, R²=0.8358 and p=0.0107. (G) Activated T cells cultured in glucose (Glc) or galactose (Gal)-containing media for ~14 days were restimulated and analyzed as in (A). Data represent % cell loss (avg ± SEM) for 5 individual donors for a 24h RICD assay. Glc and Gal T cells were compared by two-way ANOVA: OKT3 [5] n.s, [50] p=0.0057, [500] p=0.0003. (H) Activated T cells as in (G) were stimulated with anti-FAS agonistic Ab APO1.3 for 24 hrs. Data represent % cell loss (avg ± SEM) for 3 individual donors. Two-way ANOVA analysis showed no significant differences. (I) Representative FACS surface staining of CD3 (upper panel) and CD95 (lower panel) between Glc (black) and Gal (green) T cells versus isotype control (grey).

Figure 2. Acute glucose availability governs RICD sensitivity

(A) Annexin V binding to Glc or Gal T cells at baseline, or after 4 hr OKT3 restimulation ± 2-DG (2 mM) analyzed by flow cytometry. Numbers denote % of AnnexinV⁺ T cells. (B) Glc or Gal T cells were restimulated with 500 ng/ml OKT3 for 24 hrs ± 2 mM 2-DG pre-treatment, and analyzed for RICD as in Fig 1D. Data represent % cell loss (avg ± SEM) for 6 individual donors. Treatments were compared by two-way ANOVA: Glc-Gal p<0.0001, Glc-Glc+2-DG p=0.0003, Glc+2DG-Gal+2-DG p<0.0001, Gal-Gal+2DG p<0.0001. (C) L-Lactate was measured in T cell supernatants after 4 hours of OKT3 restimulation ± 2-DG, data represents 4 individual donors. Lines connect data points for each single donor. (D) Glc T cells were maintained or switched into Gal media on day 9 or day 13 in culture, then assayed for RICD sensitivity on day 14 as in (B). Data (average ± SD of technical replicates) are representative of 3 independent experiments using different donors. (E) Glc or Gal T cells were placed into fresh media or swapped into 3 day conditioned media of the opposite sugar with added IL-2 and analyzed for RICD sensitivity. Data (average ± SD of technical replicates) are representative of 3 independent experiments using different donors. (F) Gal T cells were washed and resuspended in media containing titrating doses of glucose (0.1 – 10 mM) and tested for RICD sensitivity as in (B) + 2-DG treatment. Data (avg ± SD of technical replicates) are representative of 3 independent experiments using different donors.

Figure 3. Cell cycle progression and differential effector function do not contribute to differential RICD sensitivity in Glc vs. Gal T cells

(A) Flow cytometric PI cell cycle analysis of Glc (black) and Gal T cells (green) at baseline and after 4 hours of OKT3 restimulation ± 2-DG. Numbers denote % cells in sub-G1/apoptotic gate (upper left) or S+G2/M (upper right). Data are representative of 3 independent experiments using different donors. (B) EdU incorporation by Glc and Gal T cells treated as in (A) was measured by flow cytometry. Data (average ±SD) represent 4 independent experiments using different donors. Conditions compared by T-test were all n.s. (C) Secreted IFNγ was measured in Glc (blue) vs. Gal T cell (red) supernatants by ELISA after 4 hours of OKT3 restimulation; n=2 donors. Lines connect data points for each single donor. (D) Glc and Gal T cells were pretreated with 10 ng/mL IFNγ for 30 minutes, then assayed for RICD as above. Data represent % cell loss (avg ±SEM) of 3 donors. Treatments were compared by two-way ANOVA: (D) Glc-Gal p=0.0176, Glc+IFN-γ-Gal+IFN-γ p= 0.0140, Glc-Glc+IFN-γ n.s, Gal-Gal+IFN-γ n.s. (E) Representative surface staining and MFI of CD107a between Glc (upper) and Gal (lower) T cells versus isotype control (grey) at baseline (blue) or after 4 hours of restimulation (red) by flow cytometry. (F) Glc and Gal T cells were pretreated with 5μg/mL concanamycin A (CMA) for 30 minutes, then assayed for RICD as above. Data represent % cell loss (avg ±SEM) of 3 donors. Treatments were compared by two-way ANOVA: Glc-Gal p=0.0366, Glc+CMA-Gal+CMA p=0.0223, Glc-Glc+CMA n.s, Gal-Gal+CMA n.s.

Figure 4. Differential RICD is independent of mTORC1 activity

(A) Glc or Gal T cells were restimulated ± 2.7 μM rapamycin pre-treatment, and analyzed for RICD as previously described. Data represent % cell loss (avg ± SEM) for 4 individual donors. Data were compared by T-test: Glc-Gal p=0.0185, Glc+Rapa-Gal+Rapa p=0.01864. (B) Intracellular staining for pS6 in Glc or Gal T cells at baseline, or after 4 hr OKT3 restimulation ± 2mM 2-DG or ± 2.7 μM rapamycin analyzed by flow cytometry. Numbers denote MFI. Data are representative of 3 independent experiments using different donors.

Figure 5. Glycolysis enhances RICD specifically by facilitating FASL induction after TCR restimulation

(A) Lysates of Glc and Gal T cells at baseline or after 4 hr OKT3 restimulation ± 2-DG treatment were separated by SDS-PAGE and immunoblotted for the indicated proteins. Asterisk denotes non-specific band; arrows denote specific bands. β-actin serves as a loading control. Data are representative of 3 independent experiments using different donors. (B) Soluble FASL (sFASL) was measured in Glc (blue) vs. Gal T cell (red) supernatants by ELISA after 4 hours of OKT3 restimulation ± 2-DG, n=4 donors. Lines connect data points for each single donor. (C) Relative expression of FASL message measured by qPCR of Glc and Gal T cells treated as in (A). FASL mRNA was standardized to RPL13 control for each sample; Glc baseline was normalized to 1. Data (average ± SD of technical replicates) are representative of 2 independent experiments using different donors. Values in figure represent fold change from unstimulated sample. (D) Glc and Gal T cells were transfected with a FASL 3’ UTR GFP reporter plasmid (green histograms) or control GFP plasmid (black histograms), and analyzed for GFP expression after 6 hours by flow cytometry. Representative histograms are shown at left, including MFI values for the GFP⁺ population (right of dotted line). Bar graph represents avg ±SD of % change in MFI (3’UTR-GFP/GFP alone) for 3 independent donors. (E) Linear regression analysis comparing sFASL and L-lactate production in restimulated Glc T cell supernatants for 9 independent donors, including 95% confidence interval (dashed line). Pearson correlation R= 0.8033, R²= 0.6453, p=0.0091. (F) Glc and Gal T cells were pretreated with anti-FAS blocking Ab SM1/23 for 30 min, then assayed for RICD as above. Data represent % cell loss (avg ±SEM) for 3 independent donors. Treatments were compared by two-way ANOVA: Glc-Gal p=0.0124, Glc-Glc+SM1/23 p=0.0005, Glc+SM1/23-Gal+SM1/23 n.s, Gal-Gal+SM1/23 n.s.