Memory CD8⁺ T cells upregulate glycolysis and effector functions under limiting oxygen conditions

Abstract

Memory CD8⁺ T cells are indispensable for maintaining long‐term immunity against intracellular pathogens and tumors. Despite their presence at oxygen‐deprived infected tissue sites or in tumors, the impact of local oxygen pressure on memory CD8⁺ T cells remains largely unclear. We sought to elucidate how oxygen pressure impacts memory CD8⁺ T cells arising after infection with Listeria monocytogenes‐OVA. Our data revealed that reduced oxygen pressure during i n vitro culture switched CD8⁺ T cell metabolism from oxidative phosphorylation to a glycolytic phenotype. Quantitative proteomic analysis showed that limiting oxygen conditions increased the expression of glucose transporters and components of the glycolytic pathway, while decreasing TCA cycle and mitochondrial respiratory chain proteins. The altered CD8⁺ T cell metabolism did not affect the expansion potential, but enhanced the granzyme B and IFN‐γ production capacity. In vivo, memory CD8⁺ T cells cultured under low oxygen pressure provided protection against bacterial rechallenge. Taken together, our study indicates that strategies of cellular immune therapy may benefit from reducing oxygen during culture to develop memory CD8⁺ T cells with superior effector functions.

Memory CD8⁺ T cells are rewired compared to naïve CD8⁺ T cells to provide improved protection against reinfection. We have analyzed the impact of oxygen availability on the effector function, metabolism, protein profile, and protective capacity of memory CD8⁺ T cells. Our results provide insight into oxygen‐dependent aspects of memory CD8⁺ T cell responses.

Introduction

Memory CD8⁺ T cells are critical components of the adaptive immune system, providing immediate and long‐term protection against infections with intracellular pathogens. Upon recognition of virus‐derived peptides presented in major histocompatibility complexes, antigen‐specific naïve CD8⁺ T cells are primed, undergo extensive clonal expansion, and acquire effector functions, such as the production of the pro‐inflammatory IFN‐γ and cytotoxic molecules, including the serine protease granzyme B and perforin [1, 2, 3]. After resolving the infection, the vast majority of effector cells become redundant and are eliminated via apoptosis, leaving behind a minor fraction of memory precursors, approximately 5–10% of the effector pool, that will develop into long‐lived memory CD8⁺ T cells [4, 5]. Upon reinfection, memory CD8⁺ T cells proliferate more rapidly than naïve CD8⁺ T cells [6, 7, 8, 9], resulting in more robust secondary responses to counter invading pathogens upon reinfection. Similar to activated naïve CD8⁺ T cells, activated memory CD8⁺ T cells accomplish viral clearance through granzyme B‐driven elimination of target cells [10, 11] and activation of other immune cells by the release of pro‐inflammatory cytokines, including IFN‐γ, TNF‐α, and IL‐2 [8, 12]. These effector properties are considered to be cardinal functions of reactivated memory T cells to mediate protection against reinfection with pathogens. The CD8⁺ T cell memory pool includes central memory T cells (T_CM), effector memory T cells (T_EM), and tissue‐resident memory T cells (T_RM) that are marked by distinct trafficking patterns and distinct functions. T_CM recirculate in the blood and lymph, and are able to enter secondary lymphoid organs using lymph node homing molecules, such as CCR7 and CD62L (L‐selectin). Similar to T_CM, T_EM are circulating memory T cells, but lack lymph node homing molecules, denying these cells access to the lymph nodes. T_EM may circulate to non‐lymphoid peripheral tissues [13], but it appears that the majority is confined to the spleen and the intravascular compartment [14]. In contrast to circulating memory T cells, T_RM are permanently localized in barrier tissues such as the skin, small intestine, and lungs, where these cells are poised for rapid recognition of recurring pathogens and persist for a prolonged period of time [6, 15–20].

T cell activation triggers substantial alterations in cellular metabolism to support the energetically demanding processes of proliferation, effector differentiation and effector function [21, 22]. Quiescent naïve and memory T cells mainly rely on mitochondrial oxidative phosphorylation (OXPHOS) for maintenance under steady‐state conditions, utilizing derivatives of glucose, lipids or glutamine [23, 24]. Upon activation, a rapid metabolic switch occurs, resulting in upregulation of aerobic glycolysis to support effector functions, such as the synthesis and secretion of pro‐inflammatory cytokines (e.g., IFN‐γ, TNF‐α, and IL‐2) and proliferation [25, 26, 27, 28]. Early T cell receptor (TCR)‐mediated T cell activation also leads to an increased mitochondrial oxygen consumption. T cell activation enhances tricarboxylic acid (TCA) cycle activity and electron transport chain flux, which is important for the development of effector cells [26, 29]. Memory T cells retain the capacity of naïve T cells to upregulate their metabolic activity to support secondary responses upon re‐encounter of pathogens. In fact, the metabolic wiring of memory T cells enables more efficient expansion and differentiation into effector T cells, compared with naïve T cells. This is largely supported by the spare respiratory capacity (SRC), defined as the difference between basal and maximal respiration, and the increased mitochondrial mass that memory T cells possess, compared with naïve T cells [28, 30]. In addition, memory T cells appear to have a tightened cristae structure in their mitochondria, which facilitates a more efficient influx of pyruvate to boost OXPHOS [31]. Together, these metabolic advantages allow memory T cells to more effectively expand and produce effector cytokines, resulting in more robust recall responses upon antigen re‐encounter [17, 21, 23, 32, 33].

Memory T cells are exposed to a wide range of partial oxygen (pO₂) pressures across tissues, which may impact their metabolism and differentiation. In vivo pO₂ pressure varies from roughly 8 mmHg (1%) in the superficial region of the skin to 40–90 mmHg (5‐12%) in blood, and up to 105 mmHg (15%) in the lung alveoli [34, 35]. The local oxygen pressure impacts the ability of memory T cells to utilize OXPHOS to fuel their expansion and differentiation into effector T cells at the site of reactivation. In contrast to naïve T cells, which likely encounter antigens in the spleen or lymph nodes, memory T cells are dispersed throughout the body and may require to upregulate their metabolic activity in response to pathogens at peripheral sites with low oxygen pressure. Reducing oxygen pressure during the activation of naïve T cells in culture resulted in decreased expansion and proliferation of activated T cells, whilst enhancing differentiation into effector T cells with higher granzyme B production, stronger cytolytic activity, and an improved capacity to eliminate tumor cells [34, 36, 37].

Despite the considerable interest in the role of oxygen pressure during the activation of naïve T cells, its impact on the reactivation of memory CD8⁺ T cells has remained understudied. This information is relevant, given that memory T cells have distinct metabolic wiring compared with naïve T cells. Moreover, memory T cells can also be reactivated at peripheral sites with low oxygen pressure in contrast to naïve T cells. Our study focused on the impact of oxygen pressure during in vitro restimulation of pathogen‐specific memory CD8⁺ T cells using a Listeria monocytogenes‐OVA (Lm‐OVA) infection model. We observed that oxygen pressure impaired the ability of memory T cells to upregulate OXPHOS after antigenic stimulation. We used quantitative proteomics to unravel the metabolic pathways that are utilized by Lm‐OVA specific memory CD8⁺ T cells. Our data revealed an enhanced glycolytic signature of memory T cells that were maintained under limiting oxygen pressure during ex vivo restimulation. The oxygen‐induced metabolic alterations did not substantially impact T cell expansion, but rather resulted in upregulation of effector functions of the reactivated memory CD8⁺ T cells.

Results

Listeria monocytogenes‐specific memory T cells expand similarly under different oxygen conditions

To study the impact of oxygen pressure on the reactivation of memory CD8⁺ T cells, we made use of Listeria monocytogenes‐ovalbumin (Lm‐OVA) specific memory T cells that developed from naïve SIINFEKL_257‐264 (OVA)‐specific CD8⁺ OT‐I T cells after oral infection with Lm‐OVA (Figure 1A). Memory OT‐I T cells were identified using congenic markers in spleens of Lm‐OVA infected mice and contained both T_CM and T_EM fractions (SI Figure 1A). We stimulated these cells with MEC.B7.SigOVA cells and cytokines under low (3%) and high (20%) oxygen conditions (Figure 1A). Binding of SIINFEKL_257‐264 (H‐2 Kb OVA) peptide‐loaded tetramers underlined the specificity and purity of the memory OT‐I T cells before and after culture (SI Figure 1B‐C). We observed that the Ly5.1⁺ OT‐I T cells increased relative to host CD8⁺ T cells after stimulation with cognate antigen in the presence of IL‐2, IL‐7, and IL‐15 and further maintenance in IL‐7 and IL‐15 for up to 9 days (Figure 1B). Memory OT‐I T cells cultured under low oxygen conditions were able to expand to a similar extent as their counterparts cultured under high oxygen conditions (Figure 1B‐D). Prior to culture, memory CD8⁺ T cells expressed CD44, but the majority did not express CD62L or CD69, which is in line with a T_EM phenotype (Figure 1E). Phenotypic analysis at the end of the culture showed that the majority of the restimulated memory cells still lacked expression of CD62L regardless of oxygen pressure during culture (Figure 1E). To analyze the impact of oxygen on the differentiation of T_CM and T_EM, we isolated these subsets and cultured them separately. We found that T_CM similar to T_EM lacked CD62L expression after culture in low and high oxygen conditions, indicating that T_CM downregulate CD62L expression (Figure 1F). The expression of CD69 was upregulated on memory CD8⁺ T cells after in vitro stimulation and this increase was more prominent under low oxygen conditions than under high oxygen conditions (Figure 1E). Combined analysis of CD62L and CD69 expression showed that reduction of oxygen pressure during culture decreased the number and percentage of CD8⁺ T cells with a CD69⁻CD62L⁻ phenotype, whereas the percentage and number of CD8⁺ T cells with a CD69⁺CD62L⁻ phenotype increased (Figure 1G‐I). These findings suggest that restimulation of CD8⁺ memory T cell under low oxygen conditions does not substantially reduce their expansion capacity, but rather impacts T cell differentiation. The increased expression of CD69 may indicate a higher activation status of the CD8⁺ T cells, or alternatively, the acquisition of characteristics of T_RM‐type memory CD8⁺ T cells.

Figure 1.

CD8⁺ memory T cells similarly expand after in vitro restimulation under high and low oxygen pressure. The expansion and phenotypes of Lm‐OVA specific memory CD8⁺ T cells (OT‐I) from infected mice were analyzed using flow cytometry. (A) Schematic representation of Ag‐specific memory T cell generation in vivo, subsequent in vitro restimulation with MEC.B7.SigOVA (20 hour co‐culture) and IL‐2, IL‐7, and IL‐15, and further culture in the presence of IL‐7 and IL‐15. (B) Flow cytometry plots demonstrate the expression of Ly5.1 and Ly5.2 to identify Ly5.1⁺Ly5.2⁻ OT‐I donor T cells prior to restimulation (grey) and after culture in 3% O₂ (blue) and 20% O₂ (red). Ly5.1⁺Ly5.2⁺ host CD8⁺ T cells are depicted in black. (C‐D) Scatter plots display the frequencies of memory OT‐I T cells within the total CD8⁺ T cell population (C) and the total numbers of memory OT‐I cells (D) at the indicated days of in vitro culture in 3% O₂ (blue) and 20% O₂ (red) culture conditions. (E) Histograms show the expression of CD44 (top panel), CD62L (middle panel), and CD69 (bottom panel) of OT‐I T cells from non‐infected and infected mice as indicated, prior to (grey) and after in vitro culture in 3% O₂ (blue) and 20% O₂ (red) conditions. (F) Flow cytometry plots display expression of CD69 and CD62L on FACS‐purified T_CM (CD69⁻CD62L⁺) and T_EM (CD69⁻CD62L⁻) (left panel) and CD62L expression before (grey) and after culture (right panel) at the indicated oxygen conditions. (G) Contour plots show the expression of CD69 and CD62L after in vitro culture of naïve OT‐I T cells from uninfected mice (left panel) and donor OT‐I T cells from infected mice (right panel) in 3% O₂ (blue) and 20% O₂ (red). (H,I) Bar graphs display percentage (H) and number (I) of memory OT‐I T cells with central memory (CD69⁻CD62L⁺), effector memory (CD69⁻CD62L⁻) or CD69⁺ effector memory (CD69⁺CD62L⁻) phenotypes, after culture in 3% O₂ (blue) and 20% O₂ (red) conditions. Data from B, C, E, G, H, and I are representative of at least 3 independent experiments, each including three biological replicates (n = 3). Data from D were combined from 3 independent experiments (n = 9). Data from F were derived from a single experiment including 3 biological replicates (n = 3). Symbols represent individual mice; bars represent the mean; error bars represent mean ± SD. (C) Two‐way ANOVA with Bonferroni's multiple comparisons tests, (D) two‐tailed paired t‐test, (H‐I) multiple t‐tests with Holm‐Šídák correction; *P < 0.05; ***P < 0.001; **** P < 0.0001.

In line with previous findings [34, 36, 37], we observed that the expansion of OT‐I T cells with a largely naïve phenotype from spleens of uninfected mice was larger under high oxygen pressure than under low oxygen pressure (SI Figure 1D). These OT‐I T cells largely acquired a CD62L⁺CD69⁻ phenotype, as observed in T_CM‐type memory T cells, after culture under high oxygen pressure, whereas similar to reactivated memory OT‐I T cells, they acquired a CD62L⁻CD69⁺ phenotype under low oxygen pressure (Figure 1G and SI Figure 1E‐F). The use of FACS‐purified naïve OT‐I T cells from uninfected mice recapitulated the findings of total OT‐I T cells from uninfected mice (SI Figure 1G‐I). Thus, naïve and memory CD8⁺ T cells similarly respond to antigenic challenge under limiting oxygen conditions by differentiating into CD69⁺ T_EM‐type T cells. In contrast, under high oxygen conditions, naïve CD8⁺ T cells exhibit a larger expansion potential and display superior formation of T_CM‐type T cells compared with memory CD8⁺ T cells.

Hypoxia impacts production of inflammatory cytokines and cytotoxic molecules by memory CD8⁺ T cells

We were interested whether low oxygen pressure impacts the capacity of reactivated memory CD8⁺ T cells to produce cytotoxic molecules. Therefore, we measured intracellular granzyme B production. First, the intracellular granzyme B protein expression of memory CD8⁺ T cells was quantified using flow cytometry before and after in vitro restimulation under low and high oxygen conditions. The granzyme B content was drastically upregulated after in vitro restimulation of memory OT‐I T cells. Importantly, granzyme B production was higher in low oxygen settings than under high oxygen conditions (Figure 2A). Next, we analyzed the production of key effector cytokines IFN‐γ, TNF‐α, and IL‐2 after a 4–6 hour restimulation of the cultured memory OT‐I T cells with SIINFEKL_257‐264 (OVA) peptide. A higher percentage of memory OT‐I T cells cultured under high oxygen pressure produced IFN‐γ and TNF‐α at low peptide concentrations, compared with memory OT‐I T cells that were cultured using low oxygen conditions (Figure 2B‐F). In addition, memory CD8⁺ T cells cultured under high oxygen pressure more often co‐produced IFN‐γ, TNF‐α, and/or IL‐2 than memory CD8⁺ T cells cultured under low oxygen pressure (Figure 2G). In contrast, memory CD8⁺ T cells cultured under low oxygen produced approximately two‐fold more IFN‐γ on a per cell basis, compared with memory CD8⁺ T cells cultured under high oxygen (Figure 2H). The production of TNF‐α and IL‐2 was not similarly elevated on a per cell basis between oxygen conditions (Figure 2I‐J). Our data demonstrate that the oxygen regimen in culture has a major impact on the ability of restimulated memory T cells to produce cytokines and granzyme B. Reduced oxygen availability in reactivated memory CD8⁺ T cells increases the capacity to produce IFN‐γ at the expense of co‐production of TNF‐α and IL‐2. Thus, reactivation of memory CD8⁺ T cells under limiting oxygen appears to compromise polyfunctional cytokine responses, while enhancing effector functions including IFN‐γ and granzyme B production.

Figure 2.

Oxygen pressure impacts the potential of memory CD8⁺ T cells to produce pro‐inflammatory cytokines and cytotoxic molecules. (A) Granzyme B production of memory OT‐I T cells cultured in 3% O₂ (blue) and 20% O₂ (red), was measured by intracellular flow cytometry. Histogram (top) displays granzyme B expression prior to (grey) and after in vitro culture under 3% O₂ (blue) and 20% O₂ (red) conditions. Graph (bottom) displays the geometric mean fluorescence intensities (gMFI) of granzyme B expression after culture at the indicated oxygen conditions. (B‐J) Memory OT‐I T cells, cultured in 3% O₂ (blue) and 20% O₂ (red), were left unstimulated or were restimulated for 4–6 hours with indicated SIINFEKL_257‐264 (OVA) peptide concentrations to assess cytokine production via intracellular flow cytometry. (B) Cytokine production of IFN‐γ (top panels), TNF‐α (center panels) and IL‐2 (bottom panels) is represented in histograms after peptide restimulation at the indicated concentrations. (C) Dot plots display co‐production of IFN‐γ and TNF‐α (top panels), IFN‐γ and IL‐2 (center panels), and TNF‐ α and IL‐2 (bottom panels) after stimulation with 100 nM peptide. (D‐F,H) Graphs display the frequencies (D‐G) and the gMFI (H‐I) of IFN‐γ (D‐H), TNF‐α (E,I), and IL‐2 (F,J) production of OT‐I T cells. (G) Bar graph shows the cytokine profiles of OT‐I T cells after peptide restimulation. Data from A were derived from two independent experiments (n = 6) and data from B‐J is representative of at least two independent experiments (n = 3). Biological replicates are plotted as individual data points (n = 3) with mean ± SD. (D‐G) Multiple unpaired t‐tests and (H‐J) two‐tailed unpaired t‐tests; * P < 0.05; ** P < 0.01; *** P < 0.001.

Hypoxia enhances glycolysis in CD8⁺ T cells, whereas normoxia supports OXPHOS during culture

It has been previously shown that the metabolic wiring of memory CD8⁺ T cells impacts their ability to produce IFN‐γ [28]. In particular, the glycolytic potential of CD8⁺ T cells appears relevant for these cells to maintain the production of IFN‐γ. Therefore, to address how oxygen pressure levels impacted IFN‐γ production in reactivated CD8⁺ memory T cells, we analyzed the metabolic phenotypes of cultured memory T cells using the Seahorse Mito Stress Test [38, 39]. As expected, we observed a higher oxygen consumption rate (OCR) in T cells cultured in 20% oxygen compared with those that were cultured in 3% oxygen, which indicates that these cells rely more on OXPHOS (Figure 3A). The oxygen consumption was not only elevated under basal conditions, but also after mitochondrial uncoupling, suggesting that the SRC was increased after culture under high oxygen pressure (Figure 3B‐C). In contrast, we found that memory CD8⁺ T cells cultured under low oxygen pressure displayed a higher basal extracellular acidification rate (ECAR) than those that were cultured under high oxygen pressure, suggesting that these cells rely more on glycolysis to meet their energy demands (Figure 3D‐E). In line with these findings, we detected higher lactate concentrations in the culture supernatant of memory CD8⁺ T cells cultured under low oxygen conditions (Figure 3F). Similar observations regarding SRC and ECAR were made after reactivation of naïve OT‐I T cells (Figure S2A‐E). These data demonstrate that low oxygen conditions during culture trigger glycolysis, while high oxygen promotes OXPHOS in naïve and memory CD8⁺ T cells.

Figure 3.

Limiting oxygen conditions induce glycolysis, whereas high oxygen conditions support OXPHOS in memory CD8⁺ T cells. The metabolic profiles of FACS‐purified Lm‐OVA specific memory CD8⁺ T cells (OT‐I) cultured in 3% O₂ (blue) and 20% O₂ (red), were assessed using a Seahorse XFe96 analyzer. (A) OCR are displayed under basal conditions and in response to the indicated mitochondrial inhibitors. (B‐C) Bar graphs represents (B) basal respiration after d‐(+)‐glucose injection and (C) SRC after injection of FCCP. (D) ECAR are displayed under basal conditions and after d‐(+)‐glucose injection and the indicated mitochondrial inhibitors. (E) Bar graph represents the glucose‐induced change in ECAR after D‐(+)‐glucose injection. (F) The concentration of lactate was determined in supernatant 3 days after medium refreshments at day 3 and day 6 of culture, respectively. Data from A‐E are representative of at least 3 independent experiments. Data from F is representative of at least 2 individual experiments. Data points from A and D represent the mean ± SEM (n = 3). Symbols represent individual mice with mean ± SD. (B, C and E) Two‐tailed unpaired t‐tests and (F) Multiple t‐tests with Holm‐Šídák correction; * P < 0.05; ** P < 0.01; *** P <0.001.

Memory T cells cultured under high and low oxygen conditions display distinct metabolic profiles

The distinct metabolic profiles of memory CD8⁺ T cells reactivated under low and high oxygen pressure levels urged us to perform proteomic analysis of these cells. Given that we observed CD69 upregulation after culture under low oxygen pressure compared with high oxygen conditions, we separated memory CD8⁺ T cells cultured in 3% and 20% oxygen into CD69⁺ and CD69⁻ populations for proteome analysis using mass spectrometry (Figure S3A‐B). Principal component analysis showed that memory CD8⁺ T cells cultured in low and high oxygen conditions formed separate populations based on their proteomic profiles (Figure 4A). In contrast, proteome analysis of CD69⁺ and CD69⁻ T cells cultured under 3% and 20% oxygen did not reveal considerable differences between these populations (Figure 4A‐B and Figure S3C‐E). Given these minimal differences, the CD69⁻ fractions that contain the majority of CD8⁺ memory T cells were solely used for further proteomic analyses. From a total of 4130 quantified proteins, we identified a set of 342 differentially expressed (DE) proteins between the CD69⁻ fractions of memory CD8⁺ T cells that were cultured under low and high oxygen pressures. Unsupervised hierarchical clustering using the DE proteins showed that CD69⁺ and CD69⁻ populations of memory CD8⁺ T cells cultured in 3% oxygen segregated from CD69⁺ and CD69⁻ populations of memory CD8⁺ T cells cultured in 20% oxygen (Figure 4B).

Figure 4.

Proteomic analysis reveals distinct phenotypes of memory T cells cultured under limiting and high oxygen conditions. The proteome of Lm‐OVA specific memory CD8⁺ T cells (OT‐I) was determined after culture in 3% O₂ (blue) or 20% O₂ (red) using mass spectrometry. (A) Principle component (PC) analysis is shown for CD69⁻ and CD69⁺ memory OT‐I T cell populations cultured under 3% and 20% O₂. (B) Heat map displays the protein expression profiles for the CD69⁻ and CD69⁺ populations of differentially expressed proteins between memory OT‐I T cells cultured under 3 and 20% O₂. Color coding represent z‐scored (log₂) LFQ intensity values as indicated. Non‐quantified values are depicted in grey. (C‐D) GO term enrichment analysis is shown using DE proteins between CD69⁻ OT‐I T cells cultured in 3 and 20% O₂. Graphs display biological processes (top 10) associated with CD69⁻ OT‐I T cells cultured in (C) 3% O₂ and (D) 20% O₂. (E) Volcano plot displays DE proteins between CD69⁻ OT‐I T cells cultured in 3% and 20% O₂. Selected proteins of interest are highlighted as indicated in legend. Data are representative of 4–6 biological replicates (n = 4–6). FDR < 0.05 and S0 correction of 1.

Gene Ontology (GO) term enrichment analysis was applied to identify biological processes associated with memory CD8⁺ T cells cultured in either high or low oxygen. We identified glucose metabolism‐related processes, cytolysis, and protein hydroxylation among upregulated biological processes in memory CD8⁺ T cells cultured in 3% oxygen (Figure 4C). In contrast, we identified that processes associated with OXPHOS and type I IFN responses were upregulated in memory CD8⁺ T cells cultured in 20% oxygen (Figure 4D). These findings indicate that oxygen pressure instructs changes at the protein level to adapt the metabolic wiring of memory CD8⁺ T cells reactivated in low oxygen pressure to glycolysis and that of memory CD8⁺ T cells reactivated in high oxygen conditions to OXPHOS. Analysis of DE proteins between memory T cells restimulated in high and low oxygen conditions showed an increased abundance of integrins, cytotoxic molecules including perforin, granzyme A, B, and E, the inhibitory receptor LAG3 and the costimulatory receptor GITR in memory T cells cultured under low oxygen and an increased abundance of type I IFN responsive proteins and the transcription factors T‐BET and EOMES in memory T cells cultured under high oxygen (Figure 4E). We did not detect IFN‐γ, TNF‐α, and IL‐2 using mass spectrometry, consistent with the absence of these pro‐inflammatory cytokines prior to short‐term peptide stimulation.

In line with our findings showing enhanced glycolysis in memory CD8⁺ T cells cultured under low oxygen pressure, we found an increased abundance of proteins associated with D‐glucose transmembrane transporter activity, such as GLUT1 and GLUT3 in memory CD8⁺ T cells that were expanded in limiting oxygen conditions (Figure 5A‐B). Furthermore, we detected robust increased abundance of proteins involved in the glycolysis pathway including HK, GPI, PFK, ALDO, TPI, GAPDH, PGK, PGAM, and ENO in memory CD8⁺ T cells cultured under low oxygen conditions, compared with those that were cultured under high oxygen conditions (Figure 5A,C). In contrast, proteins involved in OXPHOS, including TCA cycle proteins and proteins from mitochondrial respiratory chain complexes such as COX4I1, COX5A, COX7C, NDUFA4, NDUFS5, NDUFS8, and NMES1 showed decreased abundance in memory CD8⁺ T cells cultured under low oxygen pressure (Figure 5A,D‐E). To assess the importance of glucose for memory CD8⁺ T cells during reactivation in high and low oxygen, the glycolysis pathway was inhibited using the non‐metabolizable glucose analog 2‐deoxyglucose (2‐DG). We observed that inhibition of glycolysis impaired the expansion of memory CD8⁺ T cells at high concentrations of 2‐DG independent of oxygen availability (Figure 5F). To address the impact of oxygen availability on glucose uptake, we monitored the internalization of fluorescent glucose analog 2‐(N‐(7‐Nitrobenz‐2‐oxa‐1,3‐diazol‐4‐yl)Amino)‐2‐Deoxyglucose (2‐NBDG). Memory CD8⁺ T cells that were cultured under low oxygen conditions were able to internalize glucose to a higher extent than their high oxygen cultured counterparts (Figure 5G‐H). This difference was completely diminished in the presence of the competitive glucose inhibitor 2‐DG in concentrations that did not abolish T cell expansion (Figure 5G‐H). Together, these data show that memory T cells restimulated under limiting oxygen conditions have an increased abundance of glycolytic pathway components, including glucose transporters, resulting in elevated glycolytic metabolism, compared with memory CD8⁺ T cells after restimulation under high oxygen pressure.

Figure 5.

Limiting oxygen conditions robustly upregulate the glycolytic pathway in memory T cells. (A) Volcano plot displays DE proteins between CD69⁻ OT‐I T cells cultured in 3% and 20% O₂ as identified using mass spectrometry. Selected proteins involved in glycolysis (pink) and OXPHOS (blue) are highlighted. (B‐E) Heat maps display protein expression measured using mass spectrometry in CD69⁻ OT‐I T cells cultured in 3% O₂ (top) and 20% (bottom) for (B) D‐glucose transmembrane transporters; GO:0055056, (C) selected glycolysis proteins, (D) proteins of the TCA cycle; GO:0006099 and (E) proteins of the mitochondrial respiratory chain complexes I‐V; GO:0005747; GO:0005749; GO:0005750; GO:0005751; and GO:0005753, respectively. GO terms were selected from The Gene Ontology Resource. Color coding represent z‐scored (log₂) LFQ intensity values as indicated. Non‐quantified values are depicted in grey. (F) Graph displays the number of OT‐I T cells prior to (grey), and after in vitro culture in 3% O₂ (blue) and 20% O₂ (red) conditions in the absence or presence of the indicated concentrations of 2‐DG. (G‐H) Bar graph (G) and histograms (H) display the gMFI of 2‐NBDG uptake by OT‐I T cells before (grey) and after culture in 3% O₂ (blue) and 20% O₂ (red) with and without 2‐DG treatment. Dotted line represents the baseline uptake of 2‐NBDG. Data from A‐E are representative of 4–6 biological replicates (n = 4–6). FDR < 0.05 and S0 correction of 1, asterisks indicate significant upregulated proteins, arrows indicate steps within the indicated pathway. Data from F‐H are representative of at least 3 independent experiments and data points represent individual mice with mean ± SD (n = 4). Multiple t‐test with Holm‐Šídák correction; **P < 0.01.

CD8⁺ T cells cultured under low oxygen pressure protect against pathogen rechallenge

Since CD8⁺ memory T cells cultured under limiting oxygen conditions retain their capacity to produce inflammatory cytokines and granzymes, we next assessed the protective capacity of these cells against rechallenge with the same pathogen. In order to do so, we transferred FACS‐purified populations of memory CD8⁺ T cells cultured in both low and high oxygen into naïve recipients to assess their protective capacity against challenge with Lm‐OVA. Bacterial loads in the spleen, liver, and small intestine were analyzed at 3 days post infection. We observed substantial pathogen clearance in the spleen after transfer of memory CD8⁺ T cells cultured in both low and high oxygen conditions, compared with control mice that did not receive donor cells (Figure 6A). In contrast, we did not detect improved pathogen clearance in liver and small intestine after transfer of the cultured populations of memory CD8⁺ T cells (Figure 6B‐C). The negligible impact on bacterial clearance in the small intestine is likely associated with the minimal migration of donor CD8⁺ memory T cells that were cultured under either low or high oxygen pressures into this tissue (Figure 6C). Our findings suggest that memory CD8⁺ T cells cultured under low oxygen pressure establish equal protection against bacterial challenge in the spleen compared to memory CD8⁺ T cells cultured in high oxygen. Thus, culturing memory CD8⁺ T cells under low oxygen pressure may be a viable strategy for adoptive T cell therapy.

Figure 6.

CD8⁺ T cells cultured under high and low oxygen pressure possess protective capacity against pathogen rechallenge. Bacterial loads were determined in control mice (gray) or in recipient mice containing donor OT‐I T cells cultured in 3% O₂ (blue) or 20% O₂ (red) conditions at day 3 after oral reinfection with Lm‐OVA. Bacterial loads are shown for (A) spleen, (B) liver, and (C) small intestine. Combined data from 3 individual experiments (n = 12) is displayed. Individual values represent mice with mean ± SD. Ordinary one‐way ANOVA with Tukey's multiple comparisons tests; **P < 0.01; ***P < 0.001.

Discussion

Memory CD8⁺ T cell effector function is critical to mediate protection against pathogen reinfection [2, 12, 40]. Our study focused on the impact of oxygen pressures on the restimulation of pathogen‐specific memory CD8⁺ T cells using a Listeria monocytogenes‐OVA infection model. We observed that reduced oxygen pressure directed the metabolic wiring of memory CD8⁺ T cells from an OXPHOS to a glycolytic profile. These metabolic modifications were captured in the proteome of low oxygen versus high oxygen cultured memory CD8⁺ T cells. In particular, the abundance of glucose transporters and components of the glycolytic pathway were increased, whereas the abundance of proteins involved in the TCA cycle and mitochondrial respiratory chain were decreased under reduced oxygen conditions. Functional consequences of reducing the oxygen pressure for memory T cells included an increased production of IFN‐γ and granzyme B at the expense of co‐production of TNF‐α and IL‐2. Both memory T cell populations were able to provide protection against rechallenge with Lm‐OVA, suggesting that low oxygen cultures of memory T cells may provide a viable strategy for adoptive therapy similar to high oxygen cultures of memory T cells.

In this study, the impact of oxygen was analyzed using Lm‐OVA specific memory OT‐I T cells. We found that lowering the oxygen pressure did not have a substantial effect on the expansion or the differentiation of memory T cells. Previous studies have addressed how naïve CD8⁺ T cells respond to oxygen pressure. Naïve CD8⁺ T cells were shown to expand less efficiently and differentiate more strongly into effector or T_EM type cells under low oxygen pressure [34, 36, 37]. Underlining these findings, we observed that low oxygen impaired expansion and enhanced T_EM at the expense of T_CM differentiation of naïve OT‐I T cells in our model system. In contrast, memory OT‐I T cells expanded to a similar extent and did not maintain a T_CM phenotype under both high and low oxygen pressures. Naïve T cells have greater potential than memory T cells to expand and maintain stemness [41, 42]. Potentially, this superior ability of naïve T cells in comparison with memory T cells requires high oxygen pressure. Our findings indicate that the expansion and stemness potential of memory T cells, in contrast to that of naïve T cells, is not impacted by oxygen pressure. We did not directly compare the impact of hypoxia on the stimulation of naïve CD8⁺ T cells versus memory CD8⁺ T cells at the protein level. However, comparison of our data to that of another study that analyzed the impact of hypoxia on antigen triggering of naïve CD8⁺ T cells suggests substantial overlap between the hypoxia‐dependent proteomic profiles of naïve and memory CD8⁺ T cells [43]. Similar to memory CD8⁺ T cells, naïve CD8⁺ T cells that were cultured in low oxygen conditions showed upregulation of the glycolytic pathway including glucose transporters (e.g. GLUT1 and GLUT3) and glycolytic enzymes (e.g. HK1 and HK2), cytotoxic mediators (e.g. GZMG and PRF1), and costimulatory and coinhibitory molecules (e.g. GITR and LAG3). Hypoxia may also have divergent consequences on naïve and memory CD8⁺ T cells during antigen stimulation. In contrast to memory CD8⁺ T cells, naïve CD8⁺ T cells stimulated under low oxygen pressure do not appear to upregulate IFN‐γ production [43]. Direct comparisons under identical experimental conditions are required to validate these similarities and differences between the responses of naïve and memory CD8⁺ T cells under hypoxia.

We observed that memory T cells cultured under low oxygen conditions upregulated CD69 expression more robustly than those that were cultured under high oxygen conditions. Similarly, it has been found that low oxygen pressure triggers CD69 expression on the majority of naïve T cells after stimulation in culture [36, 44]. Given that CD69 is highly expressed by T_RM in contrast to circulating memory T cells, this observation could indicate that low oxygen levels induce a T_RM‐like T cell population. Indeed, it has been proposed that hypoxia is a driving cue for the differentiation of T_RM cells [45]. However, analysis of the T cell proteome did not reveal major differences related to tissue residency between CD69⁻ and CD69⁺ fractions of cultured memory T cells. In fact, both populations displayed remarkably similar protein expression profiles. These findings suggest that CD69 expression may be indicative of a heightened activation state rather than formation of T_RM cells. CD69 is one of the first molecules upregulated by T cells upon antigenic stimulation [46]. Thus, low oxygen pressure may drive memory T cells into an activated state, whereas high oxygen pressure may direct quiescence in memory T cells.

CD8⁺ memory T cells cultured under low oxygen pressure accumulated larger amounts of the serine protease granzyme B and produced more of the pro‐inflammatory cytokine IFN‐γ upon restimulation, compared with memory CD8⁺ T cells that were cultured under high oxygen pressure. These findings suggest that deprivation of oxygen upregulates effector functions in reactivated memory T cells. The elevated granzyme B production may facilitate cytotoxicity and the elevated production of IFN‐γ may enhance the potential for recruitment and activation of other immune cells by memory T cells reactivated under low oxygen conditions. These increased effector properties may have clear benefits at infection sites, where the oxygen pressure is likely reduced compared with healthy tissues. In contrast, memory T cells that were reactivated under high oxygen pressure exhibited higher co‐production of TNF‐α and IL‐2, indicating a higher polyfunctional capacity than memory T cells reactivated under low oxygen pressure. The ability to retain multiple effector functions has been shown to identify the most potent memory T cells that counter infection or tumors [47]. The superior quality of polyfunctional memory T cells resides in their retention of potential to generate robust secondary immune responses compared with other memory T cells, which are more terminally differentiated and only generate modest secondary responses. Thus, the relatively high oxygen levels in, for example, lymph nodes and spleen, may be beneficial to preserve polyfunctional potential in memory T cells for enhanced responses upon secondary encounter with pathogens.

Analysis of the metabolic profiles of CD8⁺ T memory cells shed more light on the underlying mechanisms of how oxygen levels impacts their differentiation [38, 39]. We observed that glycolysis was elevated under low oxygen conditions, whereas OXPHOS was supported under high oxygen conditions in reactivated memory T cells. In addition, we observed that glycolysis‐derived pyruvate was preferentially shunted to lactate under low oxygen conditions, rather than oxidized in the mitochondria, which is a distinctive feature of Warburg physiology [48, 49]. These metabolic profiles were evident in the proteome of memory CD8⁺ T cells cultured under high and low oxygen pressure. In particular, we identified proteins that are associated with glucose transport, such as GLUT1 and GLUT3, and proteins that are involved in the glycolysis pathway under limiting oxygen conditions. In line with our observations, a previous study has reported enhanced expression of glucose transporters, and glucose uptake to underlie enhanced glycolysis as a result from CD28 co‐stimulation during T cell activation [49]. The metabolic profiles of memory T cells cultured under low and high oxygen conditions may underlie their functional phenotypes. Previously, it has been shown that effector T cells rely on glycolysis, whereas memory T cells adopt to OXPHOS to sustain their metabolic requirements [23, 30]. The upregulation of glycolysis has also been shown to drive the production of IFN‐γ in memory T cells after antigenic activation [28]. These observations suggest that the elevated glycolysis in memory T cells cultured under limiting oxygen conditions drives the enhanced effector functions, in particular the elevated IFN‐γ production of these memory T cells. It has been shown that glycolysis not only drives IFN‐γ production, but also IL‐2 production in memory CD4⁺ T cells [50]. This observation contrasts with the specific increase in IFN‐γ production under limiting oxygen conditions in our experimental setup. Possibly, oxygen deprivation‐dependent, but glycolysis‐independent effects have resulted in decreased TNF‐α and IL‐2 production in our culture conditions. Currently, it remains unclear how oxygen deprivation triggers glycolysis in our cultures. Earlier studies have shown that limiting oxygen conditions induce the expression of hypoxia inducible factor (HIF)‐1α [51], which instructs upregulation of glycolysis resulting in an enhanced production of effector molecules in T cells [11]. In fact, the forced upregulation of HIF‐1α instructs the differentiation of T_EM‐type T cells that persist long‐term in mice [11]. Potentially, HIF‐1α plays a role in metabolic switching to glycolysis under limiting oxygen conditions in our system. We similarly observed that limiting oxygen conditions instruct the differentiation of T_EM‐type T cells. However, we were unable to find evidence for enhanced expression of HIF‐1α in memory T cells cultured in low oxygen pressure using mass spectrometry. Interestingly, deliberate upregulation of the activity of HIF‐1α in memory CD8⁺ T cells induced the expression of targets including glucose transporters, proteins of the glycolysis pathway, granzymes, GITR, LAG3, and suppressed expression of other targets including T‐BET and EOMES [11]. The differential gene expression overlaps with the differential proteomic profiles that we observed of memory CD8⁺ T cells cultured under high and low oxygen, suggesting that HIF‐1α drives the effects of oxygen deprivation in memory CD8⁺ T cells. Hypoxic conditions have also been shown to upregulate HIF‐1α, and importantly also HIF‐1α targets GLUT1 and LDHA and glycolysis in human CD8⁺ T cells, indicating that HIF‐1α‐driven glycolysis is a conserved mechanism between mice and men underlying oxygen deprivation [52]. Thus, limiting oxygen pressure appears to substantially impact metabolism and in turn the effector potential of memory T cells, but future studies are required to resolve the molecular network underlying these processes.

The elevated effector qualities of memory T cells cultured under low oxygen pressure suggest potential for exploitation in cellular therapy. Our data demonstrated that donor memory T cells cultured under both high and low oxygen pressures provided substantial protection against bacterial challenge with Lm‐OVA, compared with control mice. Protection was observed in the spleen, but not in other tissues including liver and small intestine. Our findings imply that memory CD8⁺ T cells cultured under low oxygen pressure can establish at least equal protection against bacterial challenge in the spleen compared with memory CD8⁺ T cells that were cultured in high oxygen conditions. It is important to note that our experimental setup did not take into account the longevity of CD8⁺ T cells cultured in low versus high oxygen conditions. Possibly, the enhanced effector characteristics compromise the longevity of memory T cells cultured under low oxygen pressure compared with those cultured under high oxygen pressure. As noted above, previous work has indicated that low oxygen conditions may be compatible with long‐term survival as T_EM‐type T cells [11], suggesting that enhanced effector function and persistence are compatible. Adoptive immunotherapies, such as tumor‐infiltrating lymphocyte, or TIL therapy, show promising results in the clearance of various tumors in patients [53, 54, 55, 56, 57]. Tumor cells often create a hypoxic microenvironment [58], suggesting that targeting tumors with tumor‐specific memory T cells adapted to low oxygen conditions may further bolster the anti‐tumor potential of cellular immunotherapies. Although we did not directly address their efficacy in countering tumor growth, results from another study suggest that CD8⁺ T cells cultured under hypoxic conditions outperform CD8⁺ T cells cultured under normoxic conditions in the killing of tumor cells [52]. A different strategy to simulate low oxygen exposure through the deletion of the HIF‐1α regulator VHL in T cells similarly shows the generation of CD8⁺ T cells with superior capacity to control tumor growth [11, 59]. These studies suggest promise for the development of immunotherapies that exploit CD8⁺ T cells cultured using limiting oxygen conditions.

Materials and methods

Mice

Wild‐type Ly5.1⁺ and Ly5.2⁺ C57BL/6JRj mice and OT‐I TCR transgenic mice (C57BL/6JRj) were purchased from Janvier or The Jackson Laboratory, and were maintained at the animal facility of the Netherlands Cancer Institute (NKI). Ly5.1⁺ OT‐I TCR transgenic mice were generated via crossing of wild‐type Ly5.1⁺ C57BL/6JRj mice to OT‐I TCR transgenic mice. Ly5.1⁺Ly5.2⁺ mice that were generated via crossing of wild‐type Ly5.1⁺ and Ly5.2⁺ C57BL/6JRj mice were used as recipients for OT‐I TCR transgenic T cells. All animals were housed and bred under specific pathogen‐free (SPF) conditions and in individually ventilated cages (IVC) at the animal facility of the Netherlands Cancer Institute (NKI). Donor and recipient mice were gender‐matched for adoptive transfer experiments and were between 6 and 22 weeks of age. All experiments were performed in accordance with institutional and national guidelines and were approved by the national animal ethics committee (CCD registration numbers AVD3010020172205 and AVD30100202010644).

In vivo procedures

OVA‐specific memory CD8⁺ T cells were generated using an ovalbumin‐expressing Listeria monocytogenes infection model. For this purpose, 5×10⁴ FACS‐purified CD8⁺ OT‐I T cells were adoptively transferred into gender‐matched wild‐type recipient mice via intravenous injection. Recipient mice were orally infected with 2×10⁹ colony forming units (CFU) Lm‐OVA (InlA^M strain 10403s) kindly provided by Dr. Brian Sheridan (Stony Brook University) [60] at 1 day post transfer of OT‐I T cells. Recipient mice were sacrificed for tissue isolation to harvest memory OT‐I T cells for in vitro cultures at 30+ days post infection.

To determine OT‐I T cell protective capacity, in vitro restimulated memory OT‐I T cells (1.5‐2.5×10⁶) were adoptively transferred into recipient mice that were subsequently challenged with Lm‐OVA^ΔInlA (2×10¹⁰ CFU). Bacterial loads were determined in the spleen, liver and small intestine 3 days after challenge with Lm‐OVA^ΔInlA. For this purpose, infected recipient mice were sacrificed, and tissues were collected. All tissue preparations were treated as described in the section below and further disassociated via incubation with 1% saponin for 1 hour at 4˚C. Homogenates were plated on Brain Heart Infusion (BHI) agar (Thermo Scientific) containing 100 μg/mL streptomycin (Sigma). Bacterial colonies were counted manually after overnight incubation at 37˚C.

Tissue preparation

Lymphocytes were isolated from the spleen using mechanical disruption through a 70 μM strainer. Splenocytes were resuspended in phosphate‐buffered saline (PBS) containing 0.5% (v/v) fetal calf serum (FCS, Biowest). Erythrocytes were removed using red blood cell lysis buffer (155 mM NH₄Cl, 10 mM KHCO₃ and 0.1 mM EDTA). Untouched CD8⁺ T cells were obtained from splenocyte preparations through negative selection by depletion of CD4⁺ T cells, erythrocytes, NK cells and B cells using MACS sorting (Miltenyi Biotec). For this purpose, splenocytes were labeled with biotinylated antibodies (Biolegend) against murine CD4 (clone GK1.5), TER‐119 (clone TER‐119), NK‐1.1 (clone PK136) and CD45R/B220 (clone 30‐F11) to enable removal of non‐CD8⁺ T cells using streptavidin‐conjugated microbeads (Miltenyi Biotec) on MACS LS columns (Miltenyi Biotec).

Intraepithelial lymphocytes (IEL) were obtained from the small intestine. After removal of fatty tissue, feces and Peyer's patches and rinsing in Hanks’ balanced salt solution (HBSS, Gibco) with 2% (v/v) FCS, the remaining tissue was cut into 5–8 mm pieces and digested in HBSS with 10% (v/v) FCS, 5 mM EDTA and 1 mM DTT for 30 minutes at 37°C. After incubation, the IEL fraction was mechanically separated from the remaining tissue using a 70 μM strainer and isolated using 66%/44% Percoll density gradient centrifugation (GE Healthcare). The livers were directly mashed through a 70 μM strainer to obtain single cell suspensions. Liver lymphocytes were separated from hepatocytes using 66%/44% Percoll density gradient centrifugation and from erythrocytes using red blood cell lysis buffer.

In vitro (re)activation of CD8⁺ T cells

MACS‐purified OT‐I T cells (0.25‐1×10⁵/well) were stimulated under 3% O₂ or 20% O₂ and 5% CO₂ at 37°C with 1×10⁴ pre‐seeded MEC.B7.SigOVA cells in flat bottom 96 well plates in the presence of 10 ng/mL recombinant IL‐2 (Peprotech), 10 ng/mL recombinant IL‐7 (Peprotech) and 10 ng/mL recombinant IL‐15 (Peprotech) for 20 hours, as previously described [61, 62]. OT‐I cells were cultured in IMDM medium (Biochrome) with minor modifications as previously published [63]. Specifically, the culture medium was supplemented with 0.1% (v/v) human serum albumin (HSA), 10% (v/v) FCS, 100 U/mL penicillin (Sigma), 100 μg/mL streptomycin (Sigma), 2 mM L‐Glutamine (Sigma), 50 μM 2‐mercaptoethanol (Gibco), 300 μg/mL transferrin, 10 μg/mL insulin (I9278, Sigma) and 100 μM sodium pyruvate (Thermo Fischer Scientific). After overnight stimulation, OT‐I T cells were removed from the MEC.B7.SigOVA cell layer and subsequently cultured with IL‐2, IL‐7 and IL‐15 until day 3 of the culture. Further resting was done using only IL‐7 and IL‐15 for up to 4 days. In some cases, CD8⁺ T cells were stimulated using antibody‐mediated activation with 5 μg/mL plate‐bound anti‐CD3 (clone 145‐2C11, Bioceros) and 1 μg/mL soluble anti‐CD28 (clone PV‐1, Bioceros) for 72 hours. The glucose analogue 2‐Deoxy‐D‐glucose (2‐DG, Sigma) was added as indicated to study the impact of glycolysis on CD8⁺ T cell cultures. To analyze cytokine production using intracellular flow cytometry, cultured OT‐I T cells were restimulated with 1–100 nM SIINFEKL_257‐264 peptide (Genscript) in RPMI 1640 (Gibco), in the presence of 1 μg/mL Brefeldin A (Biolegend) for 4–6 hours. All T cell assays were performed in accordance with MIATA guidelines (http://miataproject.org/).

Flow cytometry and FACS‐sorting

Cells were labelled with fluorochrome‐conjugated monoclonal antibodies anti‐CD45.1 (clone A20), anti‐CD45.2 (clone 104), anti‐CD44 (clone IM‐7), anti‐CD62L (clone MEL‐14), anti‐CD103 (clone M290), anti‐IFN‐γ (clone 4.SB3), anti‐IL‐2 (MQ1‐17H12), anti‐TCRγδ (clone GL3, all Biolegend), anti‐CD4 (clone GK1.5, Biolegend and BD Biosciences), anti‐CD8α (clone 53–6.7, Biolegend and BD Biosciences), anti‐CD69 (clone H1.2F3, BD Biosciences and eBioscience), anti‐granzyme B (clone GB11, BD Biosciences), anti‐TNF‐α (Clone MAb11, BD Biosciences) and/or anti‐TCRβ (clone H57‐597, eBioscience). SIINFEKL_H‐2Kb (OVA) MHC tetramers were kindly provided by Dr. R. Arens (Leiden University Medical Center). All labelling procedures were carried out in PBS containing 0.5% (v/v) FCS at 4˚C for 25 minutes. The Foxp3/Transcription Factor Staining Buffer kit (eBioscience) was used for intracellular staining of transcription factors, the Fixation/Permeabilization kit (BD Biosciences) was used for intracellular staining of cytokines and granzyme B. Both staining procedures were performed according to the manufacturer's instructions. Unbound antibodies and tetramers were removed by washing cells in PBS supplemented with 0.5% (v/v) FCS. Dead cells were excluded from flow cytometric analyses using Live/dead Fixable Near‐IR Dead Cell Stain Kit (Thermo Fischer Scientific). The fluorescent glucose analogue 2‐NBDG (200 nM, Thermo Fischer Scientific) was used to measure glucose uptake after incubation at 37˚C for 15 minutes in serum‐free RPMI 1640 (Gibco). Samples were acquired using the BD FACSymphony or the BD LSRFortessa flow cytometer (BD Biosciences) and the collected data were analyzed using FlowJo software version 10 (FlowJo, LLC). FACS‐sorting was carried out for the purification of OT‐I cell populations from uninfected mice (CD44⁻CD69⁻CD62L⁺ T_Naïve, CD44⁺CD62L⁺T_CM‐like) and purification of OT‐I cell populations from memory mice (CD69⁻CD62L⁻ T_EM, CD69^dimCD62L⁻ T_EM and CD69⁻CD62L⁺ T_CM) using the FACSAria III cell sorter (BD Biosciences). All flow cytometric analyses and FACS‐sorting were performed in accordance with the Guidelines for the use of flow cytometry and cell sorting in immunological studies [64].

Metabolic assays

Seahorse metabolic assays were performed on FACS‐purified OT‐I cells after 8 days of in vitro culture under 3% O₂ or 20% O₂ and 5% CO₂ at 37°C. Cells were resuspended in XF medium consisting of non‐buffered RPMI 1640 (Sigma) supplemented with 2 mM L‐glutamine and 1 mM sodium pyruvate (Sigma) and 2×10⁵ cells/microwell were plated onto a poly‐D‐lysine coated XF cell culture microplate. Cells were measured at 37°C under basal conditions and in response to sequential injections of 5 mM D‐(+)‐glucose (Sigma), the mitochondrial inhibitors 1 μM oligomycin, 1.5 μM carbonyl cyanide‐4 (trifluoromethoxy) phenylhydrazone (FCCP) and a mixture of 200 nM rotenone/1 μM antimycin A. The OCR and ECAR were simultaneously recorded using a XFe96 Extracellular Flux Analyzer (Agilent Technologies). The data was analyzed and the OCR was corrected for non‐mitochondrial respiration using Seahorse Wave Desktop Software (Agilent Technologies). The accumulation of the glycolysis end product lactate was measured in supernatants, collected at the indicated time points after culture, using a RapidPoint 500 system (Siemens Healthcare Diagnostics).

Mass spectrometry sample preparation

Cell pellets of FACS‐purified CD8⁺ T cells were washed in ice cold PBS, snap frozen in liquid nitrogen and stored at ‐80°C until further processing. Tryptic peptides were prepared as previously reported with minor adjustments (Kulak et al., 2014). In brief, cells were lysed in 1% (v/v) Sodium Deoxy Cholate (Sigma), 10 mM TCEP (Thermo Scientific), 40 mM ChloroAcetamide (Sigma Aldrich), 100 mM TRIS‐HCl pH 8.0 (Life Technologies) and HALT protease/phopsphatase inhibitor cocktail (Thermo Scientific). Samples were incubated at 95°C for 5 minutes and sequentially sonicated for 10 minutes in a Bioruptor Pico (Diagenode). After the addition of 50 mM ammonium bicarbonate (Sigma) containing 6.25 μg/mL Trypsin Gold (Promega), the samples were digested overnight at room temperature. After digestion, the samples were acidified by the addition of trifluoroacetic acid (Thermo Scientific) and loaded on an in‐house prepared SDB‐RPS STAGEtips (Empore). The peptides were desalted and eluted in 5% (v/v) ammonium hydroxide (Sigma) and 80% (v/v) acetonitrile (BioSolve). The sample volume was reduced by SpeedVac and supplemented up to 10 μl with 2% (v/v) acetonitrile and 0.1% (v/v) TFA. Peptide concentrations were determined using a colorimetric peptide kit (Thermo Scientific) and 3 μl of each sample (containing 0.75 μg peptides) was injected for MS data acquisition.

Mass spectrometry data acquisition

Tryptic peptides were separated by nanoscale C18 reverse phase chromatography coupled on line to an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific) via a nanoelectrospray ion source (Nanospray Flex Ion Source, Thermo Scientific). The peptides were loaded on a 20 cm 75–360 μm inner‐outer diameter fused silica emitter (New Objective) packed in‐house with ReproSil‐Pur C18‐AQ, 1.9 μm resin (Dr Maisch, GmbH). The column was installed on a Dionex Ultimate3000 RSLC nanoSystem (Thermo Scientific) using a MicroTee union formatted for 360 μm outer diameter columns (IDEX) and a liquid junction. The spray voltage was set to 2.15 kV. Buffer A was composed of 0.5 % (v/v) acetic acid and buffer B was composed of 0.5 % (v/v) acetic acid and 80% (v/v) acetonitrile. Peptides were loaded for 17 min at 300 nl/min at 5% buffer B, equilibrated for 5 minutes at 5% buffer B (17‐22 min) and eluted by increasing buffer B from 5–15% (22‐87 min) and 15–38% (87‐147 min), followed by a 10 minute wash to 90% and a 5 min regeneration to 5%. Survey scans of peptide precursors from 300 to 1600 m/z were performed at 120K resolution with a 4 × 10⁵ ion count target. Tandem mass spectrometry was performed by isolation with the quadrupole with isolation window 1.6, HCD fragmentation with normalized collision energy of 30, and rapid scan mass spectrometry analysis in the ion trap. The MS² ion count target was set to 1×10⁴ and the max injection time was 35 ms. Only those precursors with charge state 2–7 were sampled for MS². The dynamic exclusion duration was set to 30 s with a 10 ppm tolerance around the selected precursor and its isotopes. Monoisotopic precursor selection was turned on. The instrument was run in top speed mode with 3 s cycles. Data was collected using SII for Xcalibur software.

Mass spectrometry data analyses

The RAW files were processed with the MaxQuant computational platform version 1.6.2.10. Proteins and peptides were identified using the Andromeda search engine by querying the murine Uniprot database (downloaded March 2019). Standard settings were selected with additional matches between runs, Label‐Free Quantification (LFQ), and unique peptides for quantification. The generated ‘proteingroups.txt’ table was filtered for potential contaminants, reverse hits, and ‘only identified by site’ using Perseus software version 1.6.5.0. The LFQ values were transformed to log₂ scale, the replicates per experimental condition grouped and averaged based on the median, and only proteins found in all replicates of at least one of the experimental groups were further analyzed. Missing values were imputed by normal distribution (width = 0.3, shift = 1.8), assuming these proteins were close to the detection limit. Quantitative analyses (principle component analysis and Pearson correlation) were performed by Perseus software version 1.6.5.0 [65]. Statistical significance was determined using a volcano plot with an FDR < 0.05 and S0 correction of 1. From a total of 4130 quantified proteins, we identified a set of 342 differentially abundant proteins after imputation. Gene ontology analysis was performed using the web‐based platform GOrilla [66] with a P‐value threshold of 0.01. The data was subsequently reduced using the web tool REVIGO [67] and the top 10 GO terms were selected. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD032236 [68].

Statistical analyses

Values are expressed as mean ± SD or mean ± SEM as indicated. Statistical analyses between groups were assessed as indicated in figure legends using GraphPad software version 8.0.2 (Prism). Paired or unpaired two‐tailed Student's t tests were applied to assess differences between two groups as indicated in figure legends. Similarly, differences between more than two groups were assessed using multiple t tests with Holm‐Šídák correction, one‐way or two‐way ANOVA with Bonferroni's or Tukey's multiple comparisons test as indicated in the figure legends. P‐values of < 0.05 were considered to be statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001; **** P < 0.0001).

Conflicts of interest

The authors declare no financial or commercial conflict of interest

Author contributions

A.B.C., K.P.J.M.v.G., and M.v.d.B designed the experiments. A.B.C., F.P.J.v.A., J.J.F.v.H., N.A.M.K., M.R.G., and A.J.V. performed the experiments. A.B.C., F.P.J.v.A., N.A.M.K., and A.J.V. analyzed data, and A.B.C. and F.P.J.v.A. performed statistical analyses. A.B.C. drafted and edited the figures. A.B.C. and K.P.J.M.v.G. wrote the manuscript. All authors revised and approved the manuscript for publication.

Peer review

The peer review history for this article is available at https://publons.com/publon/10.1002/eji.202249918

Supporting information

Supplementary information

Data availability statement

The data that support the findings of this study are available from the corresponding authors upon reasonable request.