AMPK drives both glycolytic and oxidative metabolism in murine and human T cells during graft-versus-host disease

Abstract

Allogeneic T cells reprogram their metabolism during acute graft-versus-host disease (GVHD) in a process involving the cellular energy sensor adenosine monophosphate (AMP)-activated protein kinase (AMPK). Deletion of AMPK in donor T cells limits GVHD but still preserves homeostatic reconstitution and graft-versus-leukemia effects. In the current studies, murine AMPK knock-out (KO) T cells decreased oxidative metabolism at early time points posttransplant and lacked a compensatory increase in glycolysis after inhibition of the electron transport chain. Immunoprecipitation using an antibody specific to phosphorylated targets of AMPK determined that AMPK modified interactions of several glycolytic enzymes including aldolase, enolase, pyruvate kinase M, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH), with enzyme assays confirming impaired aldolase and GAPDH activity in AMPK KO T cells. Importantly, these changes in glycolysis correlated with both an impaired ability of AMPK KO T cells to produce significant amounts of interferon gamma upon antigenic restimulation and a decrease in the total number of donor CD4 T cells recovered at later times posttransplant. Human T cells lacking AMPK gave similar results, with glycolytic compensation impaired both in vitro and after expansion in vivo. Xenogeneic GVHD results also mirrored those of the murine model, with reduced CD4/CD8 ratios and a significant improvement in disease severity. Together these data highlight a significant role for AMPK in controlling oxidative and glycolytic metabolism in both murine and human T cells and endorse further study of AMPK inhibition as a potential clinical target for future GVHD therapies.

Graphical abstract

Figure 1.

AMPK KO T cells reduce both oxidative and glycolytic metabolism. A total of 2 × 10⁶ CD45.1⁺ WT (fl/fl) or AMPK KO (blue) T cells and 5 × 10⁶ T-cell depleted (TCD) B6 bone marrow cells were transplanted into irradiated allogeneic (B6D2F1) recipients. On day 7 after transplant, donor T cells were purified by negative selection over a magnetic column, placed into the Seahorse metabolic analyzer, and metabolism interrogated using the mitochondrial stress kit. OCRs (A), including both maximal OCR (B) and SRC (C), were measured simultaneously with ECARs, as shown in panel D. Response to oligomycin (w/oligo) was calculated by subtracting individual values from the averaged baseline values prior to oligomycin administration (E). Maximal ECAR values were simply the highest ECAR values obtained over the course of the analysis (F). (G-L) Donor T cells were recovered from a separate cohort of B6D2F1 recipients on day 21 posttransplant and measured for OCR (G-H), SRC (I), ECAR (J), response to oligomycin (K), and maximal ECAR values (L). Panels A-F, n = 2 pooled samples per group (3-4 mice in each pool), with plots representative of 2 independent experiments. In panels G-L, n = 3 pooled samples, with 3 to 4 mice per pool. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. fl/fl, floxed/floxed.

Figure 2.

Decreased aldolase activity in AMPK KO T cells. WT vs AMPK KO T cells were transplanted into B6D2F1 recipients, recovered on day 7, and proteins immunoprecipitated from cell lysates using an antibody detecting the phosphorylated AMPK-specific motif LxRxx(pS/pT). Precipitated candidate proteins were subsequently identified via LC-MS and a heat map generated of those recovered at threefold or higher levels in WT vs AMPK KO T cells (A). Representative LC-MS data from the heat map in panel A is shown for the 4 glycolytic enzymes of interest (B). Control protein samples were recovered from day 7 samples before immunoprecipitation and blotted for total levels of candidate proteins aldolase and GAPDH (C). (D-E) WT or AMPK KO T cells were stimulated on CD3/CD28 coated plates for 72 hours, followed by T-cell recovery and measurement of aldolase activity in cell lysates. GAPDH was measured in a similar fashion from T cells recovered on day 7 posttransplant (F). For data in panels A-B, 12× WT and 12× AMPK KO T cells were recovered on day 7 after transplant and divided into 4 groups of 3 recipients each. These 4 groups were then processed as individual replicates through cell lysis, IP, and LC-MS analysis. In panels C,F, n = 3 replicates pooled from ≥9 individual recipients (eg, 3 groups of 3 recipients each). Graphs in panels D-E represent data from 3 separate biological donors. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. PKM, pyruvate kinase M.

Figure 3.

Decreased IFN-γ production in AMPK KO T cells is cell-intrinsic. (A-C) WT or AMPK KO T cells were transplanted individually into irradiated B6D2F1 recipients, recovered on day 7, stimulated for 6 hours with fresh B6D2F1 splenocytes in the presence of monensin, and analyzed for intracellular cytokine production. Representative flow plots are shown in panel A, with the percentage of IFN-γ⁺ cells (B) and the median fluorescence intensity (MFI) of cells falling within the IFN-γ⁺ gate shown in panel C, respectively. (D-E) WT (CD90.1/2) and AMPK KO (CD90.2) T cells were transplanted separately or in a 1:1 combination (mixed) into irradiated B6D2F1 recipients and intracellular IFN-γ detected as outlined in panels A-C. Representative flow plots for individual and mixed samples are shown in panel D, whereas panel E represents the percentage of IFN-γ⁺ cells in CD4 vs CD8 T cells from multiple samples. (F-H) T cells were recovered from a separate cohort of B6D2F1 recipients on day 21 and analyzed by flow cytometry for CD4 vs CD8 expression (F), with CD4 percentages plotted for individual mice (G). CD4 and CD8 percentages were then multiplied by the total number of lymphocytes recovered to calculate the total number of CD4⁺ (left) and CD8⁺ (right) T cells on day 21 (H). n = 7 to 8 recipients per group for panels A-E, and n = 12 to 14 recipients per group for panels F-H. ∗P < .05; ∗∗∗∗P < .0001.

Figure 4.

AMPK-deficient human T cells decrease oxidative and glycolytic metabolism in vitro. CD3⁺ human T cells were purified from the peripheral blood of healthy human donors, electroporated with ribonucleoprotein (RNP) complexes containing the Cas9 protein and a guide RNA (gRNA) targeting human AMPKα1 locus and expanded until day 10 in recombinant human IL-2. (A) Genomic DNA purified from day 10 T cells was used to amplify a 700 bp fragment covering the gRNA target site, followed by Sanger sequencing and decomposition analysis. (B) Protein from 1 × 10⁵ day 10 T cells was precipitated with trichloroacetic acid (TCA) followed by immunoblot analysis for specific phosphorylation of ULK-1 on Ser555 by AMPK. (C-H) Day 10 T cells were stimulated overnight with CD3/CD28 dynabeads in physiologic glucose (5.5 mM) and placed on the Seahorse metabolic analyzer (C). Both maximal OCR (D) and SRC (E) were measured simultaneously with ECAR (F-H), including both the response to oligomycin (G) and the maximal ECAR values (H). Plots in panels C,F are representative of 2 independent donors, whereas data in panels D-G are the composite analysis from these 2 donors. ∗∗∗P < .001; ∗∗∗∗P < .0001.

Figure 5.

Decreased glycolytic compensation in human T cells lacking AMPK. A total of 6 × 10⁶ day 10 human T cells were transplanted with 1 × 10⁶ autologous non–T cells into irradiated NSG recipients. Human CD3⁺ T cells were recovered from spleens of NSG recipients on day 25 after transplant and purified via magnetic selection. (A) Genomic DNA was purified from day 25 T cells followed by PCR amplification, Sanger sequencing, and decomposition analysis. (B-E) T cells recovered on day 25 were placed into the Seahorse metabolic analyzer, where the OCR (B) was measured simultaneously with ECAR (C). ECAR response to oligomycin (D) and maximal ECAR values (E) were calculated as in Figure 1. (F-H) Similar to panels A-E, 6 × 10⁶ day-10 human T cells were transplanted with 1 × 10⁶ autologous non–T cells into irradiated NSG recipients, recovered on day 28, and CD4 vs CD8 percentages quantitated by flow cytometry (F). These values were then used to calculate the posttransplant CD4/CD8 ratio (G). Pretransplant (post-CRISPR) CD4 and CD8 percentages are shown for comparison (H). Plots in panels B-E represent data from 2 pooled samples per group (3-4 mice in each pool; 6-8 mice total). Eleven mice per group in panels F-G. Each experiment was performed at least twice. Data in panel G represent the mean ± standard error of the mean. ∗P < .05; ∗∗∗P < .001.

Figure 5.

Decreased glycolytic compensation in human T cells lacking AMPK. A total of 6 × 10⁶ day 10 human T cells were transplanted with 1 × 10⁶ autologous non–T cells into irradiated NSG recipients. Human CD3⁺ T cells were recovered from spleens of NSG recipients on day 25 after transplant and purified via magnetic selection. (A) Genomic DNA was purified from day 25 T cells followed by PCR amplification, Sanger sequencing, and decomposition analysis. (B-E) T cells recovered on day 25 were placed into the Seahorse metabolic analyzer, where the OCR (B) was measured simultaneously with ECAR (C). ECAR response to oligomycin (D) and maximal ECAR values (E) were calculated as in Figure 1. (F-H) Similar to panels A-E, 6 × 10⁶ day-10 human T cells were transplanted with 1 × 10⁶ autologous non–T cells into irradiated NSG recipients, recovered on day 28, and CD4 vs CD8 percentages quantitated by flow cytometry (F). These values were then used to calculate the posttransplant CD4/CD8 ratio (G). Pretransplant (post-CRISPR) CD4 and CD8 percentages are shown for comparison (H). Plots in panels B-E represent data from 2 pooled samples per group (3-4 mice in each pool; 6-8 mice total). Eleven mice per group in panels F-G. Each experiment was performed at least twice. Data in panel G represent the mean ± standard error of the mean. ∗P < .05; ∗∗∗P < .001.

Figure 6.

Deletion of AMPK decreases xenogeneic GVHD severity without affecting antileukemia potential. (A-C) A total of 6 × 10⁶ mock or CRISPR-treated human T cells were transplanted with 1 × 10⁶ autologous non–T cells into irradiated NSG recipients, and recipients followed for survival (A) and clinical score (B) to 12 weeks posttransplant. Increased clinical scores were driven by significant skin manifestations, exaggerated fur ruffling, and dramatically hunched posture (representative photographs shown in panel C). To assess antileukemic potential, mock vs CRISPR–treated human T cells were plated at an effector-to-target ratio of 2:1 with GFP⁺ Molm13 leukemia cells and cytotoxicity measured using an Incucyte analyzer and the loss of Green integrated (G.I.) intensity over time (D). A similar approach was used to measure cytotoxicity of mock vs CRISPR–treated, CD19–targeting CAR T cells placed into Incucyte incubator with GFP⁺ Nalm6 leukemia cells (E). Ten to 11 mice per group in panels A-B, with clinical manifestations shown for 2 representative animals in panel C. Experiments in panels D-E are representative of similar results using T cells from 2 independent donors.