Chronic lymphocytic leukemia cells impair mitochondrial fitness in CD8+ T cells and impede CAR T-cell efficacy

Abstract

In chronic lymphocytic leukemia (CLL), acquired T-cell dysfunction impedes development of effective immunotherapeutic strategies, through as-yet unresolved mechanisms. We have previously shown that CD8+ T cells in CLL exhibit impaired activation and reduced glucose uptake after stimulation. CD8+ T cells in CLL patients are chronically exposed to leukemic B cells, which potentially impacts metabolic homeostasis resulting in aberrant metabolic reprogramming upon stimulation. Here, we report that resting CD8+ T cells in CLL have reduced intracellular glucose transporter 1 (GLUT1) reserves, and have an altered mitochondrial metabolic profile as displayed by increased mitochondrial respiration, membrane potential, and levels of reactive oxygen species. This coincided with decreased levels of peroxisome proliferator-activated receptor γ coactivator 1-α, and in line with that, CLL-derived CD8+ T cells showed impaired mitochondrial biogenesis upon stimulation. In search of a therapeutic correlate of these findings, we analyzed mitochondrial biogenesis in CD19-directed chimeric antigen receptor (CAR) CD8+ T cells prior to infusion in CLL patients (who were enrolled in NCT01747486 and NCT01029366 [https://clinicaltrials.gov]). Interestingly, in cases with a subsequent complete response, the infused CD8+ CAR T cells had increased mitochondrial mass compared with nonresponders, which positively correlated with the expansion and persistence of CAR T cells. Our findings demonstrate that GLUT1 reserves and mitochondrial fitness of CD8+ T cells are impaired in CLL. Therefore, boosting mitochondrial biogenesis in CAR T cells might improve the efficacy of CAR T-cell therapy and other emerging cellular immunotherapies.

Conflict of interest statement

Conflict-of-interest disclosure: J.A.F., D.L.P., J.J.M., and C.H.J. have sponsored research grants from Novartis and are inventors on patents related to CAR T-cell therapy. The remaining authors declare no competing financial interests.

© 2019 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

Impaired activation and reduced glucose uptake by CLL-derived CD8+ T cells upon stimulation. PBMCs from CLL patients and age-matched HDs were thawed and cultured for 2 (A-B,D-F) or 5 (C) days in the presence or absence of anti-CD3/CD28 antibodies. Cells were analyzed for (A) activation markers (CLL, n = 15-19; HD, n = 16-22), (B) degranulation (CLL, n = 12; HD, n = 12), (C) PD-1 expression (CLL, n = 8; HD, n = 8), (D) surface glucose transporter GLUT1 (measured by GLUT1 Ras-binding domain [RBD] green fluorescent protein [GFP] construct) (CLL, n = 6; HD, n = 4; data are the same as in supplemental Figure 1G), and (E) glucose uptake (2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose [2-NBDG]; CLL, n = 23; HD, n = 20; data are the same as in supplemental Figure 1F), and expression of (F) HIF-1α (CLL, n = 6; HD, n = 4). Normality was determined by a D’Agostino and Pearson normality test. The P value was calculated by an unpaired Student t test (A-C,E), or a Mann-Whitney test (A,D,F). Data are presented as mean plus or minus standard error of the mean (SEM). *P < .05; **P < .005; ****P < .0001.

Figure 2.

Reduced glucose uptake in CLL-derived CD8+ T cells can be reversed and is not attributable to competition for glucose with CLL cells. CLL-derived CD8+ T cells and age-matched HDs were either positively sorted by fluorescence-activated cell sorting (FACS) or by using a negative T-cell selection kit from Stemcell or left in their PBMC fraction, supplemented with IL-2, and stimulated for 2 days using anti-CD3/CD28 antibodies. After stimulation, cells were analyzed for (A) surface GLUT1 expression (CLL, n = 4; HD, n = 4), (B) glucose uptake (2-NBDG; CLL, n = 5-6; HD, n = 6), and (C) activation (CD71; CLL, n = 5-6; HD, n = 6). (D) CLL cells from PBMCs from CLL patients were purified using CD19 macrophage-activated cell sorting and CD8+ T cells were cell sorted afterward. CD8+ T cells were either cultured alone, or with CLL cells separated by a transwell (0.2 µm pore size), or cultured together with CLL cells without a transwell in a 1:9 ratio, in the presence of IL-2. Two days after stimulation using anti-CD3/CD28 antibodies, CD8+ T cells were analyzed for glucose uptake (2-NBDG) by flow cytometry (CLL, n = 6). (E) Supernatants of PBMCs from CLL patients and age-matched HDs stimulated for 2 days with anti-CD3/CD28 antibodies were analyzed for glucose and lactate concentrations (CLL, n = 6; HD, n = 6). Normality was determined by a D’Agostino and Pearson normality test. The P value was calculated by a Mann-Whitney test (A-C), or a Wilcoxon signed rank test (D-E). Data are presented as mean plus or minus SEM. *P < .05. ns, not significant; TW, transwell.

Figure 3.

Expression of intracellular GLUT1 is diminished in CLL-derived CD8+ T cells. CD8+ T cells were sorted (A,E) from PBMCs from CLL patients and age-matched HDs or were left in their PBMC fraction (B-D), and analyzed for basal (A) extracellular acidification rate (ECAR; indicating glycolysis; CLL, n = 8; HD, n = 10), (B) glucose uptake, (C) total expression of GLUT1 (intracellular plus surface; CLL, n = 4; HD, n = 4), (D) HIF-1α expression (CLL, n = 4; HD, n = 4), and (E) GLUT-1 mRNA (CLL, n = 8; HD, n = 7). Normality was determined by a D’Agostino and Pearson normality test. The P value was calculated by a Welch test (A,C), Mann-Whitney test (B,D), or an unpaired Student t test (E). Data are presented as mean plus or minus SEM. *P < .05.

Figure 4.

Altered mitochondrial homeostasis in CLL-derived CD8+ T cells. (A) KEGG pathway analysis showing top 10 of most significantly different pathways in CD8+ T cells from HDs vs CLL, and heatmap of differentially expressed genes in OXPHOS pathway (Gene Expression Omnibus [GEO] accession number GSE8835; P = 3.4 × 10−15 for OXPHOS pathway). PBMCs from CLL patients and age-matched HDs were analyzed by flow cytometry (C,E-H) or Seahorse EFA (B,D) either directly after thawing (B-D,H), 2 days of culturing (E-F), or 3 days of culturing (G). (B) Cells were analyzed for OCR (indicating OXPHOS; CLL, n = 7; HD, n = 6). (C) Mitochondrial mass was determined by flow cytometry (Mitotracker Green; CLL, n = 29; HD, n = 29) and by qPCR by calculating the mitochondrial DNA (mtDNA)-to-nuclear DNA (nDNA) ratio (CLL, n = 8; HD, n = 5). (D) SRC was calculated as the ratio of maximum OCR over basal OCR (CLL, n = 6; HD, n = 6). (E) ΔΨm was determined by flow cytometry (Mitotracker Orange; CLL, n = 11; HD, n = 12; data are the same as unstimulated control in Figure 5D). (F) Mitochondrial ROS (mitoSOX; CLL, n = 11; HD, n = 12; data are the same as unstimulated control in Figure 5E). (G) Mitochondrial ROS as well as mitochondrial potential were analyzed in HD CD8 T cells cocultured with HD-derived B cells or CLL (HD, n = 8). (H) PGC-1α, SOD1, and SOD2 (CLL, n = 8; HD, n = 8), HO-1, NRF-2, and ERRα (CLL, n = 8; HD, n = 4). Data were normalized to HD to compile multiple independent experiments (B,D). Normality was determined by a D’Agostino and Pearson normality test. The P value was calculated by a Welch test (E-F), Mann-Whitney test (B-D,G-H), or an unpaired Student t test (H). Data are presented as mean plus or minus SEM. *P < .05; **P < .005; ****P < .0001.

Figure 5.

Impaired induction of mitochondrial biogenesis in CLL-derived CD8+ T cells. PBMCs from CLL patients and age-matched HDs were thawed and subsets of CD8+ T cells were directly analyzed by flow cytometry. (A) Mitochondrial mass as defined by Mitotracker Green; naive T (Tn) cells: CD27+CD45RA+CCR7+CD95−; memory stem cells (Tscm): CD27+CD45RA+CCR7+CD95+; memory T cells (Tm): CD27+CD45RA−; and effector T cells (Te): CD27−CD45RA−/+ (CLL, n = 9; HD, n = 9). (B) PGC-1α and SOD2 expression in CD8+ T-cell subsets defined as naive (Tn: CD27+CD45RA+), memory (Tm: CD27+CD45RA−), and effector T cells (Te: CD27−CD45RA−/+; CLL, n = 8; HD, n = 8). (C-E) PBMCs were stimulated by using anti-CD3/CD28 antibodies for 2 days and being analyzed for (C) induction of mitochondrial mass, which was calculated by dividing the MFI of Mitotracker Green of stimulated cells by unstimulated cells (CLL, n = 4; HD, n = 4), (D) ΔΨm (CLL, n = 11-12; HD, n = 11-12; unstimulated control is the same as in Figure 4D), and (E) mitochondrial ROS (CLL, n = 11-12; HD, n = 11-12; unstimulated control is the same as in Figure 4E). Normality was determined by a D’Agostino and Pearson normality test. The P value was calculated by a paired Student t test (A), an unpaired Student t test (B, D), a Mann-Whitney test (C), or a Welch test (D-E). Data are presented as mean plus or minus SEM. *P < .05; **P < .005; ***P < .0005; ****P < .0001.

Figure 6.

Increased mitochondrial mass in CAR T cells from CLL patients showing complete response to CD19 CAR T-cell therapy. Patients were separated into 2 groups based on response to therapy: CRs and NRs. Viably frozen infusion products from CLL patients undergoing a CD19 CAR T-cell trial (NCT01747486, and NCT01029366; CR, n = 7; NR, n = 20) were thawed and CAR+CD8+ T cells were analyzed for (A) mitochondrial mass (Mitotracker Green), glucose uptake (2-NBDG), mitochondrial membrane potential (Mitotracker Orange), and mitochondrial ROS (MitoSOX). (B) Mitochondrial mass of CAR+CD8+ T cells plotted against determinants of clinical outcome including expansion of the CAR T-cell culture (fold expansion), expansion of the CAR T cells after infusion (qPCR CAR days 0-35), persistence of CAR T cells calculated at days 0 to 35 postinfusion (AUC days 0-35), and overall expansion of CD8+ T cells after infusion (CD3+CD8+ peak days 0-28) (CR, n = 6; NR, n = 18). (C) Mitochondrial mass of CAR+CD8+ T cells plotted against the percentage of CD27+ cells negative for PD-1, TIM-3, and LAG-3 (CR, n = 6; NR, n = 18). Normality was determined by a D’Agostino and Pearson normality test. The P value was calculated by a Mann-Whitney test (A). Correlations were determined by a Spearman Rho test (B-C). Data are presented as mean plus or minus SEM. *P < .05.