Early Inhibition of Fatty Acid Synthesis Reduces Generation of Memory Precursor Effector T Cells in Chronic Infection

Despite some progress in the control of malaria, the World Health Organization estimates that 3.2 billion people are still at high risk worldwide (1). The most recent subunit malaria vaccine candidate, RTS,S, has shown poor efficacy and a lack of long-lived protection (2). Immunity to blood-stage malaria infection, the stage of parasite that causes disease pathology, requires CD4 T cell– and B cell–mediated mechanisms (3). Yet, mechanisms of activation and differentiation for these protective cells are not well understood. Infection of mice with Plasmodium chabaudi is an accurate and well-defined model for the immunity and pathology of mild malaria, and it has a chronic phase lasting up to 3 mo (4). Plasmodium falciparum is also documented to become chronic, lasting up to a year, even in the absence of a rainy season and mosquitos (5). Chronic infection and chronically stimulated T cells protect animals from reinfection against Plasmodium and other chronic parasites (6). We have shown that CD4 T cells in the memory phase of P. chabaudi infection primarily have an effector memory T cell (Tem) phenotype and that these cells contain an increased proportion of IFN-γ⁺TNF⁺IL-2⁻ Th1 cells during chronic infection compared with infections cured after 30 d (7). However, the mechanisms of Tem differentiation are much less well understood than those of central memory T cells (Tcm) generated in acute stimulation.

Two models have been proposed for the differentiation of memory T cells (Tmem). In simplified terms, these models propose a bifurcating model in which all Tmem, including Tem, are generated early in activation (8) or a linear model in which Tem are derived from effector T cells (Teff) and are predicted to have a short half-life (9). It should be noted that early differentiation steps do not exclude the role of later inflammatory effects that promote terminal differentiation in CD8 T cells or the role of regulatory T cells (Treg) in controlling the quiescence of Tmem (9). Furthermore, there are differences between CD4 and CD8 Teff differentiation. For example, Blimp-1 drives terminal differentiation in CD8 Teff, but not in CD4 Teff (10, 11), which indicates that there are likely differences in CD8 and CD4 Tmem differentiation as well. Our work on CD4 T cell differentiation in P. chabaudi infection definitively support the early divergence of CD4 Teff and Tem in chronic infection (12). We have recently identified CD4⁺IL-7Rα⁻CD62L^hi early Teff (Teff^Early), which can be detected as early as day 5 postinfection (p.i.), as precursors of Tcm and Tem (12). We also showed that the Tem phenotype of the CD4⁺ Tmem developed in P. chabaudi infection is determined within the first 5 d of infection, because it can be blunted by treatment of infection on day 3 p.i. but not on day 5 p.i. (12).

Upon activation, T cells undergo metabolic reprogramming to meet their energy and biosynthetic demands (13, 14). CD4 and CD8 Teff rely on aerobic glycolysis during proliferation and effector function (15). Quiescent CD8 Tcm have enhanced mitochondrial fatty acid oxidation (FAO), which they use to generate energy and molecular building blocks during the Tmem survival phase, which includes homeostatic proliferation (14, 16). A recent study showed that CD8 Tcm mobilize endogenous fatty acids, potentially synthesized within the cell, to fuel FAO (17). Furthermore, FAO promoted by drug treatment can drive Tcm differentiation after Teff contraction (18). In contrast, Tem do not do homeostatic proliferation and may use a combination of mitochondrial metabolism and glycolysis in the survival phase (19). However, it is not known what metabolic changes occur during early differentiation of activated T cells into either type of Tmem. Interestingly, inhibition of glycolysis during an active immune response enhances Tmem formation, suggesting that the metabolic switch controlling Teff differentiation concomitantly regulates bifurcation into the Tmem differentiation pathway (20).

In this study, we identified fatty acid synthesis (FAS) as a metabolic switch that appears to drive early CD4 bifurcation from memory precursor effector T cells (MPEC) (Teff^Early) into Tmem during differentiation in a mouse model of chronic infection. Starting from the previously reported observation that mice deficient in the first enzyme in the FAS pathway, acetyl coenzyme carboxylase (ACC)1, have fewer Tmem, we observed a small decrease in MPEC. We analyzed metabolic gene profiles and found that the expression of genes in the FAS pathway was greater in CD4 Tmem than Teff from P. chabaudi malaria infection. Tmem were confirmed to have increased neutral fatty acid content, but not uptake, whereas Teff showed a similar rate of FAS as Tmem. However, in this chronic infection, CD4 cells did not exhibit an increase in mitochondrial function to use the fatty acids generated, as previously shown for Tcm generated in acute stimulation (21). Deletion of ACC1 has previously been shown to have an effect on Tmem numbers, but the timing of the crucial effect was not explored (22). By inhibiting ACC1 at various times during P. chabaudi infection using a specific drug in vivo, we tested its role for long-term survival of Tmem, as well as during T cell priming and contraction. Strikingly, only early blockage of FAS (days 1–3 p.i.) had an effect on the number of malaria-specific Tmem remaining by day 60 p.i. This early blockade of FAS also appeared to drive increased Teff expansion by ∼40-fold, while reducing the fraction of responding Teff with a memory precursor (MPEC/Teff^Early) phenotype. Blockade of FAS also significantly reduced the parasite load, demonstrating the relevance of these findings. We demonstrate that early blockade of FAS on T cells impairs MPEC generation in a T cell–intrinsic manner and that MPEC have particularly high levels of FAS, which are very well inhibited, accounting for the MPEC-specific effect. Together, these data suggest that an early shift to FAS is important for the generation of CD4 memory and support an early bifurcation in the differentiation of Tmem from Teff.

B5 TCR–transgenic (Tg) mice, a kind gift from Jean Langhorne (The Francis Crick Institute, London, U.K.), were generated as previously described (23) and backcrossed to BALB/cJ mice (N4-N10; The Jackson Laboratory, Bar Harbor, ME). The B5 TCR recognizes merozoite surface protein-1 (MSP-1; 1157-1171, ISVLKSRLLKRKKYI/I-E^d); B5 TCR–Tg mice were typed using primers Vα2, 5′-GAACGTTCCAGATTCCATGG-3′ and 5′-ATGGACAAGATCCTGACAGCATCG-3′ and Vβ8.1, 5′-CAGAGACCCTCAGGCGGCTGCTCAGG-3′ and 5′-ATGGGCTCCAGGCTGTTCTTTGTGGTTTTGATTC-3′. Thy1.1 BALB/cByJ recipients were backcrossed to BALB/cJ (N4) mice and maintained in our specific pathogen–free animal facility with access to food and water ad libitum. ACC1^fl/fl mice were generously provided by Dr. W.J. Salih (Baylor College of Medicine, Houston, TX). These mice were on the C57BL/6 background (24) and were crossed with CD4-Cre mice. ACC1^fl/flCD4-Cre⁺ were generated and used for experiments. ACC1^fl/+CD4-Cre⁺, ACC1^fl/flCD4-Cre⁻ littermates (wild-type [WT]) were used as a control. RAG2° mice (Taconic, Germantown, NY) were maintained in our animal facility. Mice (6–12 wk old) were infected with 10⁵ P. chabaudi chabaudi AS–infected erythrocytes i.p. Parasites were counted by light microscopy in thin blood smears stained with Giemsa (Sigma-Aldrich, St. Louis, MO). All experiments were carried out in accordance with the protocol approved by the University of Texas Medical Branch Institutional Animal Care and Use Committee.

CD4 T cells were purified from uninfected B5 TCR–Tg mice, and CD4⁺ cells (2 × 10⁶) were adoptively transferred into Thy1.1 congenic mice. In some experiments, recipient mice received the antimalarial drug mefloquine hydrochloride (MQ; 4 mg/kg; Sigma) by oral gavage for 5 d starting day 3 p.i. Some animals received 5-(tetradecyloxy)-2-furoic acid (TOFA; 50 mg/kg; Cayman Chemicals, Ann Arbor, MI) daily on days 1–3, 10–12, or 51–53 p.i., as described (25). For in vivo labeling with BrdU (Sigma), BrdU was given in drinking water (0.8 mg/ml) on days 1–7 p.i. Cells were stained using an FITC BrdU flow kit (B44; BD Biosciences, San Jose, CA) and analyzed by flow cytometry, according to the manufacturer’s protocol. For lipid-uptake experiments, mice received 50 μg/kg BODIPY FLC16 (Invitrogen, Thermo Fisher Scientific, Waltham, MA) i.p. 1 h before euthanasia.

Single-cell suspensions from spleens were prepared in HEPES-buffered HBSS (Life Technologies, Thermo Fisher Scientific) and incubated in RBC lysis buffer (eBioscience, San Diego, CA). Cells were then stained in PBS, 2% FBS (Sigma), and 0.1% sodium azide with Abs (anti-CD90.2 [30-H12], anti-CD44 [IM7], anti-CD25 [PC61.5], and anti-CD27 [LG-7F9]) and the following combinations of markers: PerCP/eFluor 710, allophycocyanin/eFluor 780, PE, and allophycocyanin–conjugated Abs (all from eBioscience), anti-CD127 (A7R34), anti–PD-1 (RMP1-30) PE/cyanine 7 (Cy7), anti-CD62L (MEL-14) Brilliant Violet 605, and CD4 (RM4-5) Brilliant Violet 650 (BioLegend, San Diego, CA). Cells were collected on an LSR II Fortessa using FACSDiva software (BD Biosciences) and analyzed with FlowJo (version 9.7; Tree Star, Ashland, OR). Foxp3 staining was performed using eBioscience Perm/Wash Buffer after fixation with 2% paraformaldehyde. Compensation was performed in FlowJo using single-stained splenocytes (with CD4 in all colors). Each mouse was analyzed, and averages and SEM were calculated for graphs and contour plots. Data from three or four mice were concatenated to achieve sufficient cell numbers for presentation. Intracellular cytokine and CellTrace Violet (CTV) staining was performed as described previously (26). To determine the neutral lipid content, cells were incubated with 0.5 μg/ml BODIPY 493/503 (Invitrogen, Thermo Fisher Scientific) for 20 min. Mitochondrial potential was assessed by incubating the cells with 500 nM MitoTracker Green (MTG) and 250 nM MitoTracker Red (MTR; both from Thermo Fisher Scientific) for 30 min at 37°C prior to staining.

Splenic CD4 T cells from day 8–infected (Teff) or day 60–infected (Tmem) B5 TCR–Tg donors were purified using an EasySep biotin selection kit (STEMCELL Technologies, Vancouver, BC, Canada) and biotinylated anti-CD8a, B220 (RA3-6B2), CD11b (MI/70), CD11c (N418), F4/80 (BM8), and Ter119 (eBioscience). Enriched T cells were then stained with anti-CD4–FITC, CD44–allophycocyanin–Cy7, and CD127-PE for Teff and Tmem sorts. Cells were sorted on a FACSAria with FACSDiva software (BD Biosciences). Effector (CD44^int-hiCD127⁻) and memory (CD44^hiCD127^hi) CD4⁺ T cells were sorted from 5–12-wk-old B5 TCR–Tg mice on days 8 and 60 p.i., respectively (Supplemental Fig. 1). For lipidomics and microarray analysis, Teff (CD44^int-hiCD127⁻) subsets, including Teff^Early (CD62L^hiCD27⁺), intermediate Teff (Teff^Int; CD62L^loCD27⁺), and late Teff (Teff^Late; CD62L^loCD27⁻), were sorted on day 8 p.i., as described (12). Tmem (CD44^hiCD127^hi) subsets, including Tcm (CD62L^hiCD27⁺), early Tem (CD62L^loCD27⁺), and late Tem (CD62L^loCD27⁻), were sorted on day 60 p.i. in a similar manner (12), but data from the three subsets were similar and, therefore, were averaged for display.

RNA from sorted B5 TCR–Tg Teff, spun into Buffer RLT (QIAGEN, Valencia, CA) every hour, was analyzed by real-time PCR using a custom PrimePCR assay with iQ SYBR Green Supermix and iScript (all from Bio-Rad, Hercules, CA) for cDNA synthesis. Normalized relative expression was calculated using GAPDH primers run on each plate as an internal control. Threshold cycle (CT) of the gene of interest was calculated as 2^ΔCT of gene of interest and 2^ΔCT GAPDH.

Oxygen consumption rate (OCR) and extracellular respiratory capacity (ECAR) were measured in an XF24 Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA). Sorted Teff and Tmem were resuspended in Seahorse XF Media and plated (5 × 10⁶ cells per well) into Seahorse 24-well plates coated with poly-d-lysine (50 μg/ml; Sigma) for T cell attachment. The metabolic profile was assessed by adding oligomycin (1 μM; EMD, San Diego, CA), and carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (2 μM), 2-Deoxy-d-glucose (50 mM), and antimycin A/rotenone (1 μM; all from Sigma) at the indicated time points.

P. chabaudi–infected mice were primed with deuterium water (²H₂O) at day 4 p.i. for Teff and day 57 p.i. for Tmem, by i.p. injection of 99% ²H₂O in 0.9% saline at 35 mg/g body weight, followed by free access to drinking water enriched with 4% ²H₂O for 3 d. Some mice were treated with TOFA (days 1–3) prior to the start of ²H₂O labeling. Plasma enrichment of ²H₂O was approximately 3% after 3 d of labeling. Teff (CD4⁺CD127⁻CD25⁻) and Tmem (CD4⁺CD44^hiCD25⁻) subsets, as above, were sorted on a FACSAria, with fluidics upgrade (BD Biosciences) for lipid extraction.

²H₂O enrichment in body water (i.e., plasma) was measured as described previously (27). Briefly, 100 μl of plasma or standards were placed in the inner well of an O-ring cap, the tube was screwed tightly, and the inverted tube was incubated at 80°C overnight (10–12 h). The water droplets were recovered from the sides of the tubes and incubated with 10 M NaOH (1 μl) and HPLC grade acetone (5 μl) overnight at room temperature. Acetone was extracted by addition of hexane (300 μl, HPLC grade) and NaSO₄ (0.5 g). Two hundred microliters of the mixture was tested for acetone using a DB-225 GC column (6890 Quadrupole GC system; Agilent, Santa Clara, CA). The Quad MS was operated in electron impact ionization mode, and m/z 58 and 59 were detected. Enrichment was estimated using a 10-point standard curve.

Total lipids were extracted from sorted cells using the chloroform/methanol (2:1) Folch extraction method, as described previously (28). The lipids were reacted with 14% boron trifluoride (Sigma-Aldrich) to produce palmitate methyl esters. Enrichment of ²H₂O in palmitate was measured using a gas chromatograph–mass spectrometer (GC-MS; MSD System; Agilent), monitoring m/z 270 and 271 for m+0 and m+1, respectively. Fatty acid composition was measured using a gas chromatograph system with flame ionization detection (model 6890; Agilent), and internal standards were used to calculate the total lipid content, as previously described (28).

The fractional de novo lipogenesis (fDNL) over the labeling period (3 d) was calculated from body water and palmitate deuterium enrichment using precursor–product relationships, as previously described, and assuming that the number of exchangeable hydrogens (n) is 22 (28). The absolute synthesis rate of palmitate was calculated by multiplying the fDNL by total lipid content and was normalized for the total number of cells.

RNA was isolated and processed according to the Agilent Array protocol (Phalanx Biotech, San Diego, CA) from sorted effector (Teff^Early, Teff^Int, Teff^Late, as defined above in cell sorting) and memory (Tcm, early Tem, late Tem) subsets. Briefly, 0.1 μg of total RNA was amplified and labeled using a Low Input Quick Amp Labeling Kit (Agilent). Dye incorporation and amplified RNA yield were tested (NanoDrop ND-1000 Spectrophotometer). Cy3-labeled cDNA (0.6 μg) was hybridized onto an Agilent SurePrint G3 Mouse Gene Expression Microarray (8 × 60K) using Agilent’s Gene Expression Hybridization Kit. Hybridization was performed for 17 h, rotating at a speed of 10 rpm at 65°C in an Agilent Microarray Hybridization Oven. Microarrays were washed and scanned on an Agilent Microarray Scanner. Two biologic replicates were performed for each of the three Teff and three Tmem subsets. The Z-score shown is calculated as an average of normalized intensity of all replicates of all Teff or Tmem subsets minus the average of normalized intensity of all three subsets of Teff or Tmem, divided by the SD among the three subsets in each group. The heat map was generated using hierarchical clustering of all 223 metabolic genes selected (29), using Spotfire Desktop software 7.6.0 (TIBCO, Palo Alto, CA).

All data are presented as mean ± SEM. A two-tailed unpaired Student t test was used (Prism; GraphPad, La Jolla, CA). *p < 0.05, **p < 0.01, ***p < 0.001.

It is well documented that activated Teff increase glycolysis for proliferation, whereas quiescent Tmem rely on FAO and mitochondrial oxidative phosphorylation for energy generation during homeostatic proliferation (15, 16, 21). Recent data suggest that cell-intrinsic FAS is important for memory generation (17). The first and rate-limiting enzyme in the FAS pathway is ACC1, the isoenzyme regulating biosynthesis and breakdown of long chain fatty acids (30). Previous studies showed that deletion of Acaca, the gene encoding ACC1, specifically in T cells (Acc1^fl/fl CD4^Cre), impairs CD4 and CD8 T cell survival at homeostasis and Tmem formation in acute infection (22, 31). However, the timing of this effect on CD4 Tmem has not been investigated, and no mechanism was proposed. To dissect features intrinsic to Acaca-deficient naive T cells, without detecting the additive potential effects on thymic development and naive T cell homeostasis or starting numbers, we used an adoptive-transfer strategy. Acaca-deficient CD4 T cells were transferred into Ly5.1 congenic WT animals, which were then infected with P. chabaudi (Fig. 1A). Splenocytes were harvested 7 or 60 d p.i., and CD4⁺Ly5.2⁺ T cells were analyzed by flow cytometry. This experiment tests the cumulative effect of deletion of the rate-limiting enzyme of the FAS pathway on CD4 Tmem differentiation and survival in this chronic infection. An average of 4-fold (day 7, Fig. 1B) or 70-fold (day 60, Fig. 1C) more CD4⁺Ly5.2⁺ T cells were recovered from WT T cell recipients compared with Acaca^−/− T cell recipients, similar to previous reports (22, 31).

We previously showed that Teff early in activation [Teff^Early, IL-7Rα (CD127)⁻ CD62L^hi CD27⁺ PD-1^lo] have downregulated IL-7Rα upon activation but have not yet proliferated in P. chabaudi infection. We also showed that this population contains long-lived CD4 MPEC (12). Upon investigation of the phenotypes of Teff (CD44^int-hi CD127⁻) generated in transferred ACC1 CD4^Cre T cells on day 7 p.i., we found that there was a significant reduction in the proportion of Teff^Early/MPEC (CD62L^hiCD27⁺, Fig. 1D) in Acaca-deficient recipients. The impact of Acaca deficiency on the fraction of MPEC within Teff is suggestive of an early requirement for FAS on Tmem differentiation.

To understand the context of the role of FAS in Tmem metabolism, we analyzed the metabolic transcriptome of CD4 Teff and Tmem generated in P. chabaudi infection. Expression of metabolism genes was analyzed by microarray using cDNA from MSP-1–specific B5 TCR–Tg T cells. Teff (day 8 p.i.) and Tmem (day 60 p.i.) were sorted from splenocytes of B5 TCR–Tg animals. We have validated the direct infection of B5 TCR–Tg mice in earlier work by showing that infection induces an equivalent fraction of activated Teff and Tmem in TCR–Tg or BALB/c WT T cells (26). Genes from the glycolysis, tricarboxylic acid cycle, electron transport chain (ETC), FAS, and FAO pathways were included in the metabolic transcriptome analysis, as listed in Supplemental Table I. Strikingly, the pattern of metabolism-associated gene expression was almost completely distinct in Tmem versus Teff (Fig. 2A). The hierarchical clustering of the three T cell subsets shown on top of the heat map indicates that naive T cell, and not Teff, metabolism is closest to that of Tmem, despite the predominance of Tem in this infection (26). This observation indicates that Tem metabolism is more similar to that of resting cells than to Teff metabolism. Based on the similarity of expression patterns using Z-score, we defined two clusters of differentially regulated genes. In cluster A, 118 genes are significantly upregulated in Teff compared with Tmem, whereas in cluster B, 82 genes were upregulated in Tmem compared with Teff. Enrichment of upregulated genes in each metabolic pathway within cluster A (higher in Teff) or cluster B (higher in Tmem) is shown in Fig. 2B. As expected, many glycolysis genes (68.3% or 28/41 genes included in this group) are upregulated in Teff but not enriched in Tmem (26.8%). Expression of the tricarboxylic acid genes was also enhanced more in Teff (55.9%). Although there were some upregulated ETC genes in Teff and Tmem subsets, they were not enriched in either cluster (46.3% in Teff versus 40.7% in Tmem). Increased FAO gene transcription in Teff suggests that Teff use glycolysis and FAO to support proliferation and effector function more than Tmem, as also shown in CD8 Teff (14, 16). Despite reports that FAO is required for CD8 Tmem (32), we saw an enrichment in Teff of the CD4 genes that regulate the transfer of fatty acids from the cytosol into mitochondria, carnitine palmitoyl transferase (Cpt) isoforms (Cpt-1a, Cpt-1c, Cpt-2), in Teff (57.1%) compared with Tmem (28.6%). We also observed a striking increase in the expression of FAS genes in Tmem (68.4% of the 19 genes included).

To confirm the enrichment of FAS in microarray data, we tested expression of the five genes that make the enzymes central to the FAS pathway by real-time quantitative PCR (Fig. 2C). Real-time PCR confirmed expression of these genes in Teff and Tmem and showed that three of five FAS genes were significantly upregulated in Tmem compared with Teff. The upregulation of FAS in the absence of increased FAO at the transcriptional level is remarkable, because FAS provides lipid fuel to the mitochondria via the catabolic FAO pathway in CD8 Tmem in acute infection (17, 33).

To test the functional consequences of the high level of expression of FAS genes in Tmem, we assessed the neutral lipid content of Tmem (day 60) and Teff (day 8) by staining with BODIPY 493/503. This reagent binds neutral lipids, such as diglycerides and triglycerides (TG), but not polar lipids, such as membrane phospholipids (PL). B5 TCR–Tg T cells were adoptively transferred into Thy1.1 recipients and analyzed in the spleen at day 8 or 60 p.i. Tmem from malaria-infected mice showed higher BODIPY staining than Teff, indicating greater neutral lipid content (Fig. 2D). We also tested the potential of Tmem and Teff to uptake fatty acids in vivo by assessing uptake of BODIPY-labeled palmitate (FL-C16). No difference was found in lipid uptake between Tmem and Teff from animals infected for 60 or 8 d, respectively (Fig. 2E). These data reveal that CD4 Tmem contain more fatty acids than Teff, but they do not take up more extracellular fatty acids from the environment. This is consistent with previous findings that CD8 Tmem synthesize, but do not take up, fatty acids in vitro and during Listeria infection (17).

As a more accurate method of assessing fatty acid content, we tested the total palmitate content in CD4⁺ naive T cells (CD44^loCD25⁻), Teff (CD127⁻), and Tmem (CD44^hiCD25⁻) using a GC-MS (Fig. 3A). These data show increased lipid content in Tmem compared with Teff and naive T cells. Interestingly, the lipid profile is different in Teff and Tmem (Fig. 3B), with a striking predominance of diacylglyceride (DAG) in Teff (Fig. 3B). To measure the rate of FAS, we performed metabolic labeling in vivo by administering ²H₂O for 3 d to infected animals and measuring deuterium enrichment into newly synthesized fatty acids in Teff and Tmem. Teff have a higher absolute rate of newly synthesized fatty acid than naive T cells but similar to the rate of Tmem (Fig. 3C). The fractional rate of FAS is higher in Teff for free fatty acid, PL, and TG compared with Tmem (Fig. 3D). These data support previous reports that Teff generate complex fatty acids as building blocks (34, 35).

Tmem did not have any mitochondrial pathways (tricarboxylic acid, FAO, or ETC) that were transcriptionally enriched to a greater extent than in Teff. This was surprising, given that CD8 Tmem generated in response to acute stimulation have high levels of mitochondrial FAO and store fatty acids that they synthesize in FAS for energy, after breakdown by liposomal acid lipase, and transport into the mitochondria via CPT-1 (21). To test oxidative metabolism at the functional level, we measured mitochondrial activity in CD4 Tmem from animals infected with P. chabaudi. Naive B5 T cells that were adoptively transferred into congenic recipients that were infected and recovered day 60 p.i. (Tmem) did not show an increase in mitochondrial volume, as measured by MTG staining, compared with Thy1.2⁺CD127⁻ B5 T cells (Teff) harvested on day 8 p.i. (Fig. 4A). In fact, Teff showed significantly more mitochondrial staining than Tmem. However, Tmem and Teff displayed similar mitochondrial polarization, as measured by MTR (Fig. 4B). This suggests that CD4 Tmem in this chronic infection do not have more polarized mitochondria than Teff, as CD8 Tmem do when generated by acute stimulation (36). Increased mitochondrial volume in the absence of increased polarity indicates some depolarization of the mitochondria in the CD4 Teff in this chronic infection, as reported for exhausted CD8 T cells in chronic lymphocytic choriomeningitis virus (37). Interestingly, when the infection was shortened from 60 to 3 d by administering the antimalarial drug MQ (+MQ), mitochondrial volume (MTG; Fig. 4C) and mitochondrial polarization (MTR; Fig. 4D) of B5 T cells recovered on day 60 p.i. appeared to increase slightly, although these changes did not reach statistical significance. These data suggest that the differences in metabolic phenotype between CD4 Tmem in this infection and CD8 T cells in acute stimulation are due to increased Ag load or persistence.

A more definitive test of mitochondrial functionality is the consumption of oxygen by cells, which is the final acceptor of electrons in oxidative phosphorylation. Therefore, we tested OCR using extracellular flux analysis (Fig. 4E). The basal OCR of sorted B5 TCR–Tg Tmem (day 60 p.i., CD44^hiCD127^hi) was significantly higher than Teff (day 8 p.i., CD127⁻) and naive T cells (uninfected, CD44^loCD127⁺), which were similar. The maximal OCR (Fig. 4F) and the spare respiratory capacity (SRC), which is the unused mitochondrial capacity (above the basal rate), were also similar between Tmem and Teff (Fig. 4G). These data support the conclusion from the MTR data above that CD4 Teff and Tmem have similar mitochondrial capacity in this chronic infection. As expected, Teff had a higher basal ECAR than Tmem or naive T cells, indicative of increased production of lactate from glycolysis (Fig. 4H). This is consistent with the upregulation of glycolysis genes that we observed in Teff. Together, these data show that, during chronic malaria infection, Tmem exhibit a preference for FAS without a concomitant increase in FAO, suggesting differences in T cell metabolism between T cells stimulated in response to chronic or acute stimuli (32).

To study the stage of activation affected by inhibition of FAS, we studied treatment of animals with TOFA, an inhibitor of ACC1. TOFA is unique in that it specifically inhibits ACC1, but not ACC2 (38). The other drugs reported to block ACC1, soraphen A and ND646, block ACC1 and ACC2 (31, 39). This is critical, because ACC2 actually does the opposite of ACC1, clouding interpretation of previous studies. To evaluate the specific role of the FAS pathway in CD4 Tmem differentiation in this chronic infection, we tested the effect of specifically blocking ACC1 pharmacologically for short periods during each phase of Tmem generation. Tmem differentiation is regulated at three major time points: in T cell priming (12, 40, 41), on survival during the contraction of Teff (42, 43), and later in the memory phase, when decay can occur (44). B5 TCR–Tg CD4 T cells were purified, labeled with CTV, and adoptively transferred into Thy1.1 congenic mice, which were then infected with P. chabaudi (Fig. 5A). Recipient mice were given the ACC1 inhibitor TOFA daily for 3 d (days 1–3, 10–12, or 51–53 p.i.) to test the effect of transient blockade of FAS during T cell activation, contraction, and survival, respectively. A control group was given PBS (vehicle, days 1–3) instead of TOFA as a control. T cells were detected as CD4⁺Thy1.2⁺ (Fig. 5B). Double-positive staining, using the same Ab labeled with two different fluorophores, was used to detect all transferred cells most accurately. The fraction of Tmem (CD44^hiCD127^hi) within B5 T cells dividing during infection (CTV⁻) was measured 2 mo p.i. Strikingly, the proportion and number of Tmem recovered were significantly reduced in TOFA-treated animals compared with no treatment (NRx) animals, but only in mice receiving TOFA at the earliest time point, during T cell activation (Fig. 5C, days 1–3). These data indicate that the inhibition of FAS during T cell priming negatively impacts CD4 Tmem development, and it has no effect when blocked transiently at later time points.

To quantify the effect of FAS blockade, using TOFA, on lipid content and synthesis in T cells, P. chabaudi–infected mice were treated with TOFA (days 1–3 p.i.) and then provided ²H₂O for 3 d (day 4–7 p.i.) to label newly synthesized fatty acids. CD4 T cells were sorted on day 7 p.i., and ²H₂O incorporation into newly synthesized lipid was measured using a GC-MS. As previously reported for other cell types (45), TOFA treatment significantly reduced the absolute synthesis rate of fatty acids in CD4 T cells compared with the untreated control group (Supplemental Fig. 2A). The neutral lipid content of T cells, as measured by BODIPY staining, was also reduced in adoptively transferred B5 T cells on day 7 p.i. after TOFA treatment (Supplemental Fig. 2B). Because expression of the fatty acid synthase (Fasn) gene, which encodes the next enzyme downstream of ACC1 in the FAS pathway, is also increased in Tmem, we also tested C75, a Fasn inhibitor. However, treatment with C75 was lethal to infected animals, potentially due to a previously reported toxic accumulation of malonyl-CoA in multiple cell types (38, 45).

To determine how early blockade of FAS influences Teff activation, we studied malaria Ag–specific adoptively transferred B5 T cells on day 7 p.i. after early treatment with TOFA. Naive CD4 B5 TCR–Tg T cells (Thy1.2⁺) were labeled with CTV and adoptively transferred into Thy1.1 mice, which were then infected with P. chabaudi. Recipient mice were given the FAS inhibitor TOFA daily on days 1–3 p.i., as well as BrdU in the drinking water throughout the study (days 0–7 p.i.). Early blockade of FAS with TOFA increased MSP-1–specific B5 TCR–Tg (CD4⁺Thy1.2⁺) cell numbers (Fig. 6A) and the frequency of Teff (CD127⁻CD44^int-hi) within B5 cells compared with the control group on day 7 p.i. (Fig. 6B). The percentage of proliferating B5 T cells, as determined by the dilution of CTV, was also increased by an average of 26% in TOFA-treated animals on day 7 p.i. (Fig. 6C). We also tested proliferation by incorporation of the thymidine analog BrdU into DNA on division. We found that BrdU⁺ B5 TCR–Tg T cells were also significantly increased in TOFA-treated animals (Fig. 6D), supporting the conclusion that there is an increase in Teff caused by TOFA treatment on days 1–3 p.i.

Because early blockade of FAS increased generation of Teff, we hypothesized that this could affect P. chabaudi parasite growth. Parasitemia was determined for the first 2 wk of infection in mice receiving TOFA on days 1–3 p.i. (Fig. 6E). At day 8 p.i., which is the peak of parasitemia and the known point of the effect of T cells in this infection (26), early blockade of FAS significantly reduced parasitemia in TOFA-treated mice compared with untreated infected mice. It will be interesting to determine the mechanisms by which FAS blockade increases malaria-specific Teff in future studies, especially in light of the strong effect on the peak of parasitemia.

Inhibition of ACC1 and ACC2 in vitro using soraphen A and the T cell–specific deletion of Acaca in naive mice promote the generation of Treg (31). However, soraphen A blocks FAS and FAO, and permanent deletion promotes metabolic compensation. To test for effects of transient early TOFA treatment on Treg, we tested for Treg (Foxp3⁺CD25^hi) in P. chabaudi infection. There was no effect on the frequency of Treg out of all CD4⁺ cells (data not shown) or B5 TCR–Tg P. chabaudi–induced Treg in adoptive transfer (Supplemental Fig. 2C). Although ACC1 is reported to inhibit IL-17, there is very little IL-17 produced by T cells in this infection (46). However, we did investigate the effect of TOFA on other cytokines and found that early inhibition of FAS by TOFA had no effect on the IFN-γ, TNF, or IL-2 profile of CD4⁺Thy1.2⁺ Teff in infected mice (Supplemental Fig. 2D). Together, these data suggest that blockade of FAS does not significantly change the Treg response, but it does promote Teff over Tmem generation when tested at an early time point in T cell activation.

To investigate the mechanisms behind the concomitant increase in T cell expansion and the lack of surviving Tmem observed in this study, we next investigated whether inhibition of FAS specifically during the priming phase determined the development of memory precursor cells in early phases of the response. We adoptively transferred B5 TCR–Tg T cells into congenic recipients, infected recipients with P. chabaudi, and treated the with TOFA days 1–3 p.i. Seven days p.i., we detected CD127⁻CD62L^hiPD-1^lo MPEC. Strikingly, early blockade of FAS reduced the frequency and numbers of Teff^Early MPEC compared with the PBS recipient control group (Fig. 7A). MPEC are included in the PD-1⁻ fraction of Teff, because PD-1 is associated with terminal differentiation. The frequency of CD62L⁺PD-1⁻ cells was also significantly decreased (Fig. 7B), suggesting that, although proliferative Teff^Late (CD62L⁻CD27⁻) subsets were increased by blocking FAS (79.3 ± 2.2%), nonproliferative MPEC were decreased, potentially explaining the reduction in Tmem at day 60 (Fig. 5C).

To elucidate the mechanism of early blockade of FAS on MPEC generation, we investigated total lipid content and synthesis rate in MPEC and short-lived effector T cells (SLEC) (Teff^Int, Teff^Late) using a GC-MS. The data show that MPEC (CD127⁻CD62L^hi) have a higher lipid content and synthesis rate than CD62L^lo SLEC (Fig. 7C, 7D). Interestingly, transient blockade of FAS reduces lipid content as a fraction of control in MPEC (34.05 ± 1.61%) significantly more than in SLEC (Fig. 7E). TOFA had a very high level of inhibition of the rate of FAS in all three Teff subsets (Fig. 7F). Together, these data suggest that FAS is important in Tmem differentiation because of the high levels of fatty acid in MPEC compared with more terminal Teff subsets, which is specifically affected by transient blockade of ACC1.

Although TOFA clearly affects T cell activation, this was shown in the context of the whole mouse. Therefore, it remained to be determined whether the effect of TOFA on Tmem differentiation is a direct activity on T cells. To determine whether the decrease in MPEC in TOFA-treated animals is driven by an effect on innate immune cells, including APCs, we used mice deficient in adaptive immune cells (RAG2^o) (Fig. 8A). RAG2° recipient mice were treated with TOFA for 3 d, and on day 3 p.i. CD4 T cells from uninfected B5 TCR–Tg mice were adoptively transferred. The recipient mice were then infected with P. chabaudi, and splenocytes were harvested at day 7 p.i. The total number of recovered T cells was similar between nontreated and TOFA-treated recipients (Fig. 8B). Additionally, no difference was found in the percentage or total number of Teff (Fig. 8C) or MPEC (Fig. 8D). These results indicate that preblockade of FAS in innate cells is not the cause of the increase in T cell activation or the qualitative change in T cell differentiation.

The growth of Plasmodium was recently shown to depend on type II FAS (47). Type II FAS is a unique FAS pathway using completely unique enzymes that would not be expected to be affected by TOFA (48). We tested for a direct effect of TOFA treatment on parasite growth, or via effects on innate immunity, by infecting RAG2° mice, which are immunodeficient because they lack B and T cells to control parasite growth. RAG2° mice were infected and treated with TOFA (days 1–3). Parasitemia was measured, and no effect was found in RAG2^o mice at the peak (Supplemental Fig. 3A). In fact, there was an increased rate of parasite growth in TOFA-treated RAG animals. To determine possible causes of this early effect, we evaluated dendritic cell (DC) number on day 3 p.i and found that monocytic DCs were decreased (Supplemental Fig. 3B). A reduction in phagocytes could explain the increased early parasitemia, although we did not explore functionality.

To test for a direct effect of TOFA on T cells alone, we cultured B5 TCR–Tg T cells for 1 h with TOFA in vitro and adoptively transferred them into Thy1.1 recipient mice, which were then infected (Fig. 9A). On day 7 p.i., the number (Fig. 9B) of B5 T cells recovered and the percentage and number of MPEC were measured and found to be decreased (Fig. 9C). Another way to test the effect of TOFA on T cells in isolation is pretreatment of B5 TCR–Tg donor mice in vivo with TOFA daily for 3 d before transfer, as shown in Fig. 9D. On day 3, CD4 T cells were purified and adoptively transferred into congenic (Thy1.1) mice. The recipient mice were then infected with P. chabaudi, and splenocytes were harvested at day 7 p.i. No difference was found in the total number of B5-specific CD4 T cells recovered (Fig. 9E). However, the percentage and number of MPEC (CD127⁻CD27⁺CD62L^hi) were reduced in TOFA-pretreated animals compared with untreated recipient mice (Fig. 9F). These striking findings demonstrate that blockade of FAS directly impairs MPEC differentiation in a T cell–intrinsic manner.

Given the positive effect of persistent infection on the mitochondrial activity of T cells suggested by the data in Fig. 4C and 4D, we next investigated whether the role of FAS on MPEC differentiation is dependent on Ag persistence. The antimalarial drug MQ was given to one group of recipient mice on days 3–7 p.i. to quickly reduce and stop the infection. Interestingly, when the infection was thus shortened to 3 d in TOFA-treated recipients, the fraction and number of MPEC were increased compared with the MQ-only control group (Fig. 9G), as might be expected from CD8 in acute infection. These data suggest that FAS is critical for early CD4 Tmem differentiation in persistent infection but may have a different role in CD4 T cell activation in acute infection.

In this study, we aimed to identify the metabolic pathways regulating CD4 Tmem development in chronic malaria infection. Our most striking finding is that blocking the FAS pathway, specifically at the time of T cell priming, reduces Tmem formation and reduces generation of MPEC. We show that FAS inhibition acts directly on T cells, and not APCs, leading to the decrease in MPEC. This specific result correlates with an especially high level of FAS in MPEC compared with terminally differentiated Teff.

Various roles for FAS have been described recently in T cell biology (17, 22, 31); however, the role of FAS in early Tmem development has not been previously shown. Pharmacological inhibition of FAS or T cell–specific deletion of Acaca has been shown to reduce peripheral CD4 and CD8 CD44^hi Tmem numbers at homeostasis, which is interpreted to indicate a requirement for FAS in the survival of CD8 T cells (22, 31). In addition to investigating CD4 T cells in more detail, and in the context of chronic infection, our data extend previous work by identifying the window of time during which FAS contributes to Tmem differentiation the most and highlight the high level of FAS and lipid content in CD4 MPEC, which were previously unknown. This observation has the potential to explain how blocking the FAS pathway impairs Tmem. This mechanistic work will inform efforts to increase Tmem production in chronic infection. In addition, it is worth noting that Treg, which can be increased in the T cell–specific ACC1 knockout, can also regulate Tmem generation (49, 50). In our short treatment, there was no increase in Treg, allowing us to conclude that this increase is not required for the effect on Tmem development seen in both studies. These results emphasize the role of metabolism in the early steps of Tmem differentiation.

A recent study demonstrated that CD8 Tmem show increased mitochondrial FAO compared with Teff (17). This study also showed that, in acute infection, CD8 Tmem, like most cells in the body, do not upregulate receptors to import fatty acids for FAO. Instead, they synthesize the fatty acids. Making fatty acids while simultaneously burning them in the mitochondria is referred to as a “futile cycle.” In the current study, we found that CD4 Tmem in chronic infection also exhibited an increase in FAS genes and total lipid content compared with Teff. However, Teff and Tmem showed similar absolute rates of FAS. As Teff proliferate, anabolism in general, and lipid synthesis in specific, are increased as part of the activation process; however, biosynthetic building blocks are used up with each division. The end product of FAS can be used as a precursor for more complex fatty acids, such as PL, which is required for plasma membrane formation and proliferation (33). The fractional rate of PL, free fatty acid, and TG synthesis was significantly increased in Teff compared with Tmem; however, content was not increased, suggesting equal usage of these building blocks. The increased DAG content and fractional rate of PL synthesis in Teff compared to Tmem probably reflects an increased need for lipids for membrane biosynthesis during cell proliferation. It is not known for what synthesized fatty acids are used in Tmem. Prior studies suggest that, in CD8 T cells, fatty acids are used in FAO, as an energy source. However, there was no increase in mitochondrial volume or oxidative metabolism, as seen in CD8 T cells, in chronically stimulated CD4 Tmem, suggesting a different usage of fatty acid.

The small increase in mitochondrial volume detected when infected mice were treated with the antimalarial drug MQ to shorten the infection indicates a unique metabolism in Tmem in chronic infection. This is consistent with the known differences in the Tmem population in chronic infection, including constant generation of Teff, and surface markers indicating that the Tmem population remains somewhat activated in chronic infection. Early treatment of parasitemia with MQ also results in a higher ratio of Tcm than Tem at the memory time point, suggesting that the increased MPEC with MQ treatment survive to memory time points (12). Because Tcm have more mitochondria than Tem, the increase in mitochondria observed in this study is suggestive. Tcm are the cell type studied in the CD8 studies showing more mitochondria in Tmem compared with Teff. Therefore, we suggest that curing the infection early, making it “acute,” and thus making more Tcm, causes the metabolism of the Tmem generated in this shorter response to be more similar to the CD8 studies in acute infection. It should also be noted that although TOFA and MQ reduce parasitemia, they have opposing activities with regard to Tmem generation. Although TOFA reduces MPEC and Tmem generation, MQ actually increases MPEC. These opposite outcomes suggest that the effect of TOFA on MPEC, which we have shown is T cell intrinsic, is separate from the effect on parasitemia.

With regard to the intrinsic effect on T cells, a previous report found that inhibition of FAS impairs activation of GM-CSF–derived (CD11c⁺CD11b⁺) DCs in vitro (51); indeed, we saw a decrease in CD11c^loCD11b⁺ cells in RAG2° mice pretreated with TOFA. However, there was no change in T cell differentiation in these experiments, consistent with the observation that CD11c^loCD11b⁺ cells are not the dominant APC early in P. chabaudi infection (52). These data strongly support the conclusion that the effect of blocking FAS in vivo with TOFA early in P. chabaudi infection is due to the intrinsic effect that we can reproduce by pretreating T cells.

Our data suggest that the fatty acid synthesized in CD4 Tmem generated in P. chabaudi infection do not primarily feed increased mitochondrial respiration in FAO in the same futile cycle used by CD8 Tmem (36). However, the precise pathways downstream of FAS in chronically stimulated T cells remain to be investigated in our system. Increases in FAS could potentially induce acetylation of proteins, which regulate many cellular functions posttranslationally (53, 54). Acetylation regulates metabolic enzymes and histones, which control DNA accessibility to transcriptional enzymes. In addition, a recent study showed that increased systemic acetate concentration increased memory CD8 T cell function via increased lipid biosynthesis in a Listeria monocytogenes model (55, 56).

Several studies have shown that pharmacological modulators of metabolism can promote the formation of CD8 Tmem, at least later in the course of the immune response. For example, targeting the PI3K/mTOR signaling pathway that supports cell growth, proliferation, and survival (57) has been shown to regulate Tmem. Targeting mTOR directly, using rapamycin, or indirectly, with metformin, which promotes FAO (32), enhances CD8 Tmem numbers (58). However, these drugs work during the contraction phase of the T cell response to enhance survival of Tmem, suggesting little chance of using them for vaccination of healthy people to prevent infectious diseases. Metabolic drugs have also been shown to act on host T cells to reverse chronic autoimmune diseases (20, 59, 60). Studies targeting host metabolism using pharmacologic inhibitors have also shown protection against intracellular pathogens by targeting pathogen metabolism (61–64). To achieve a maximal level of dissemination or transmission, intracellular pathogens often alter host cellular metabolism to obtain sufficient amounts of energy and nutrients. For example, in Plasmodium yoelii, metabolic transcriptomic analysis of the host liver cells showed differential expression patterns of hepatocyte metabolism in infected cells compared with uninfected cells, which was shown to benefit sporozoites (65). Thus, targeting host cell metabolism for protection against pathogens represents a potential therapeutic strategy that must continue to be explored.

CD8⁺ MPEC were originally defined as Teff that maintained expression of IL-7Rα. However, this observation did not hold in CD4 T cells (66). Therefore, we defined the pathway of differentiation of CD4 memory in malaria and found that IL-7Rα⁻CD62L^hi Teff^Early contain the precursors of Tmem (12). Other markers have also been proposed for CD4 Teff^Early (such as CXCR5), and CD62L^hi CD8 Teff^Early have also been described (40, 41, 67–69). The data indicate that early blockade of FAS reduces Teff^Early/MPEC in a T cell–intrinsic manner (Figs. 7, 9) and concomitantly increases proliferating Teff (Fig. 6). This observation supports a model of Tmem differentiation, with an early bifurcation in which the ratio of Teff/Tmem is initiated during T cell priming (8). Increased Teff expansion was not apparent in every case in which MPEC were decreased; however, increased Teff progression to SLEC was seen. There are also several clear lines of evidence that some memory traits, such as the ability to expand in a second response, are acquired later on in the immune response (70). We show that MPEC have increased lipid content and synthesis compared with terminally differentiated Teff, making them highly sensitive to blocking of FAS. This is demonstrated by the strong effect of blocking FAS on MPEC lipid content. These data support a functional mechanism of Tmem differentiation from MPEC.

In summary, we have identified the FAS pathway as playing a key role in CD4 Tmem development in chronic malaria infection. From another perspective, we found that an inhibitor of the FAS pathway has a protective effect against malaria via the adaptive immune system. These findings are relevant not only to understand the metabolic pathways regulating early differentiation of Tmem, they also could represent a powerful target for vaccine development against infectious diseases. Further studies will determine whether reagents that promote FAS could be used as adjuvants to improve Tmem through this intriguing pathway.

We thank Michael DeSalvo, Cheryl F. Lichti, Larry Sowers, and Thomas Green for helping with microarray analysis bioinformatics. We also thank Mark C. Griffin of the Microbiology and Immunology Flow Cytometry core facility for assistance with cell sorting and Margarita L. Ramirez for animal husbandry. This project benefitted from feedback from the Joint Immunology Group (Drs. Rajsbaum, Soong, Sun, Cong, Endsley), Dr. Nisha Garg, and Dr. Peter C. Melby at the University of Texas Medical Branch. We thank Aaron L. Miller and Richard B. Pyles for help with the PCR analysis.

The authors have no financial conflicts of interest.