Fatty Acids Play a Critical Role in Mitochondrial Oxidative Phosphorylation in Effector T Cells in Graft-versus-Host Disease

Acute graft-versus-host disease (GVHD) is among the leading causes of nonrelapse mortality and morbidity after allogeneic hematopoietic stem cell transplant (SCT). To date, glucocorticoid is still the mainstay of treatment for GVHD, and no clinically proven second-line treatment has been established for glucocorticoid-resistant GVHD (1, 2). Dose escalation of glucocorticoid, the addition of other immunosuppressive drugs, and a T cell depletion strategy have been attempted for patients with refractory GVHD; however, they often lead to delayed immune reconstitution, thereby resulting in severe infection and relapse of the underlying diseases. The development of novel targets and therapeutic strategies is urgently needed.

Cellular metabolism determines T cell fate and function (3–6). Upon TCR stimulation, naive T cells undergo metabolic reprogramming to meet the increasing energy demands for an efficient immune response. Although naive and memory T cell metabolism mainly depend on fatty acid oxidation (FAO), activated T cells preferably use glucose as a primary energy source, which is referred to as the Warburg effect (7–9). Besides glycolysis, other metabolic pathways are also dramatically altered during T cell activation. Metabolomic analysis using mass spectrometry has demonstrated that TCR-stimulated T cells rapidly accumulate metabolic intermediates involved in several anabolic pathways, such as amino acids, carbohydrates, and nucleotide synthesis, which are necessary for cell proliferation and differentiation (9–12). A wide range of lipid metabolites are also highly accumulated in activated T cells; however, their fate and biological roles remain to be elucidated.

With the use of mouse models of GVHD, it has been shown that activated T cells increase lipid synthesis in conjunction with the downregulation of mitochondria-dependent FAO, suggesting that intracellular lipids are used as biomaterials for plasma membrane (phospholipid) formation rather than as a fuel source (9, 10, 12). Conversely, recent studies have shown that activated T cells increase FAO and subsequent oxidative phosphorylation during GVHD, thus providing potential therapeutic targets for T cell–mediated immune disorders, including GVHD (13, 14). To address this controversial issue, we comprehensively investigated donor T cell metabolism in vivo using xenogeneic GVHD models. Xenogeneic GVHD models have been proved to faithfully recapitulate T cell immune response in humans in terms of the interplay of donor T cells with host APCs and subsequent processes that lead to cytokine and chemokine production (15, 16). In the present study, we demonstrate that donor effector T cells increase fatty acid (FA) uptake and mitochondrial activity during GVHD and that incorporated lipids serve as a crucial source of mitochondrial respiration.

Female NOD/Shi-scid-IL2rγ^null (NOG) and MHC class I/II–deficient (MHC^−/−) NOG mice were purchased from the Central Institute for Experimental Animals and housed in the animal facility of our institute. Mice were allowed to acclimate to the facility for at least 1 wk and were used at age 7–9 wk. Xenogeneic GVHD models were established by i.v. injection of 5 × 10⁶ human pan–T cells into NOG mice that had received a sublethal dose (250 cGy) of total body irradiation 4 h prior to transplant, as described previously (15, 16). Human PBMCs were collected from healthy volunteers (n = 7) after obtaining informed consent according to our institutional regulations. Pan–T cells were negatively isolated using magnetic separation (Miltenyi Biotec) with more than 97% purity. Shortly before injection, T cells were stained with a cell-tracking dye, CellTrace Far Red (0.1 μM; Thermo Fisher) according to the manufacturer’s instructions.

Cells were preincubated with anti-human CD16/CD32 mAbs to block nonspecific Fc binding. To measure extracellular glucose and FA uptake, cells were stained with a fluorescence-labeled deoxyglucose analog, 2-NBDG (Cayman), and a long-chain FA analog, BODIPY 500/510 C₁₂ (Invitrogen), respectively, and accumulation was measured by flow cytometry. Cells were then stained in FACS buffer (HBSS supplemented with 2% BSA, 1 mM EDTA, and 25 mM HEPES) with a combination of Abs listed in Supplemental Table I. Events were acquired on a BD LSRFortessa (BD Biosciences) and BD FACSAria Fusion (BD Biosciences) and analyzed using FlowJo software (FlowJo LLC).

Transmission electron microscopic observation was performed using samples prepared for block-face observation by scanning electron microscopy. Briefly, freshly isolated human T cells and human T cells retrieved from two MHC^+/+ NOG mice were fixed with 4% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4), and their pellets were embedded in 2% low melting point agarose in PBS. Thereafter, the pieces of pellets in agarose were sequentially treated with 2% OsO₄ in 1.5% K₄[Fe(CN)₆] at 4°C for 1 h, 1% thiocarbohydrazide at room temperature for 20 min, 2% OsO₄ at room temperature for 30 min, 2% uranyl acetate at 4°C overnight, and 0.67% lead aspartate solution at 70°C for 30 min. Each of these treatments was followed by washing with double-distilled water. The pieces were then dehydrated in a graded series of ethanol and embedded in Quetol 812. Ultrathin sectioning was performed at a thickness of 70–80 nm, and the sections were observed under a transmission electron microscope (HT7700; Hitachi High-Technologies, Tokyo, Japan). The handling and analyses of acquired images were performed using ImageJ with Fiji plugins (http://fiji.sc/wiki/index.php/Fiji) and Amira (FEI Visualization Science Group, Hillsboro, OR). Intact cell profiles were randomly selected, and their mitochondria with cristae structures in addition to the cellular profiles themselves were manually segmented.

RNA was harvested from human T cells using the RT² q-PCR-grade RNA isolation kit (SA Biosciences), and total RNA was reverse transcribed with an RT² First Strand Kit (SA Biosciences) according to the manufacturer’s instructions. Real-time quantitative PCR was performed using RT² SYBR Green/ROX PCR Master Mix (SA Biosciences), and PCR arrays were run on an ABI StepOnePlus. Lipid metabolism–related gene expression was measured using the Fatty Acid Metabolism RT² Profiler PCR Array (330231, SA Biosciences). Data were analyzed with PCR Array Data Analysis Software provided online by SA Biosciences. Gene expression was normalized to that of housekeeping genes, and fold regulation was used to analyze changes in gene expression.

Oxygen consumption rate (OCR), an indicator of mitochondrial oxidative phosphorylation (OXPHOS), was measured using the Seahorse XF Cell Mito Stress Test Kit (Seahorse Bioscience, Billerica, MA) according to the manufacturer’s instructions. Briefly, T cells that had been preincubated under each condition were resuspended in XF basal media (Seahorse Bioscience) supplemented with 1 mM pyruvate, 10 mM glucose, and 2 mM l-glutamine, and then plated onto Seahorse cell plates (0.3 × 10⁶ cells per well) coated with Cell-Tak (BD Biosciences). With the use of a Seahorse XF96 extracellular flux analyzer (Seahorse Bioscience), OCR was measured under basal conditions followed by sequential treatment with etomoxir (a carnitine palmitoyltransferase I [CPT1] inhibitor, 100 μM, Sigma-Aldrich), oligomycin (an ATP synthase inhibitor, 2 μM), carbonylcyanide P-trifluoromethoxyphenylhydrazone (FCCP) (an uncoupling agent, 0.5 μM), and a combination of rotenone (complex I inhibitor, 0.5 μM) and antimycin A (complex III inhibitor, 0.5 μM). Actual values of cellular metabolic activity varied among individuals. Therefore, the OCR of each sample was normalized with respect to baseline values and fixed at 100%.

Mitochondrial FAO activity was measured using an FAO activity detection/assay reagent, FAOBlue (Funakoshi, Tokyo), according to the manufacturer’s instructions with minor modifications. FAOBlue was dissolved in DMSO to obtain 1 mM stock solution. Cells were first washed with PBS and stained with FAOBlue (20 µM in phenol red–free, serum-free RPMI media supplemented with 2.0 mM glucose, 0.5 mM l-carnitine, and 1 mM glutamine) for 2 h at 37°C in a 5% CO₂ humidified incubator. Cells were then stained with cell surface Abs and analyzed by flow cytometry.

Two‐tailed unpaired t tests or one-way ANOVA followed by Bonferroni’s post hoc comparisons tests were performed using Prism (GraphPad Software, La Jolla, CA) or EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), the latter of which is a graphical user interface for R (R Foundation for Statistical Computing, Vienna, Austria) (17). More precisely, it is a modified version of R Commander designed to add statistical functions frequently used in biostatistics.

To assess lipid metabolism in donor T cells after SCT, we first investigated FA and glucose uptake in donor human T cells in vivo using a xenogeneic GVHD model. Sublethally irradiated NOG mice that received human T cells developed progressive GVHD and died within 10 to 14 d after transplant. Lungs are the most severely affected organs in xenogeneic GVHD models, as we and others have previously shown (15, 16, 18). Hence, mice were sacrificed at day 9, and the transplanted human T cells were retrieved from lungs.

The fluorescence intensity of BODIPY, representing the uptake of extracellular FA, was significantly higher in donor human T cells isolated from wild-type (MHC^+/+) NOG mice than in T cells freshly isolated from healthy donors (Fig. 1A, upper panel). The BODIPY intensity of T cells in MHC^+/+ NOG mice steeply increased when the intensity of CellTrace, which is inversely correlated with the cell division rate, fell below a certain level. Thus, the BODIPY intensity of cells with low CellTrace intensity was significantly higher than that of cells with high CellTrace intensity for both CD4⁺ and CD8⁺ T cells in MHC^+/+ NOG mice (Fig. 1A, lower panel). A similar trend was observed in the intensity of 2-NBDG, corresponding to glucose uptake, in human T cells in the MHC^+/+ NOG mice (Fig. 1B).

Next, we examined FA and glucose uptake in human T cells transplanted in MHC^−/− NOG mice, because we and others have previously demonstrated that highly immunodeficient MHC^−/− mice receiving human T cells did not develop GVHD (15, 19). Donor human T cells were still able to proliferate moderately in MHC^−/− NOG mice by spontaneous proliferation, although their proliferation was significantly slower than that in MHC^+/+ NOG mice (Supplemental Fig. 1A) (15). The BODIPY intensity appeared to be elevated in T cells in MHC^−/− NOG mice, as in those in MHC^−/− NOG mice (Fig. 1A); however, the BODIPY intensity of T cells in MHC^−/− NOG mice was significantly lower than that in MHC^+/+ NOG mice (Supplemental Fig. 1B). Even when we compared T cells with low CellTrace intensity to compensate for the difference in the cell proliferation rate, the BODIPY intensity of T cells in MHC^−/− NOG mice was still significantly lower than that in MHC^+/+ NOG mice (Fig. 1C). These results suggest that the interaction of TCR with host MHC Ags facilitates extracellular FA uptake in donor T cells under inflammatory conditions. In addition, when we compared T cells with low CellTrace intensity, glucose uptake in CD8⁺ T cells in MHC^+/+ NOG mice was also higher than that in MHC^−/− NOG mice, but this was not significant in CD4⁺ T cells (Fig. 1D).

To investigate the association between FA uptake and the functional phenotype of donor T cells during GVHD, we divided T cells into four distinct subsets (naive, central memory [CM], effector memory [EM], and effector memory reexpressing CD45 [EMRA] cells) on the basis of CD45RA and CCR7 expression. All T cell subsets were almost equally distributed in MHC^−/− NOG mice, whereas the majority of T cell subsets were EM T cells followed by EMRA T cells in MHC^+/+ NOG mice (Fig. 2A); the proportion of EM T cells in MHC^+/+ NOG mice was significantly higher than that in MHC^−/− NOG mice (55.7 ± 0.8% versus 31.6 ± 1.6%, p = 0.018; Fig. 2B). Rapidly dividing T cells (low CellTrace intensity) included a higher population of EM T cells than slowly dividing T cells (Fig. 2C), suggesting that differences in cell proliferation led to the difference in the T cell subset composition between MHC^+/+ and MHC^−/− NOG mice. Of note, however, EM T cells were still significantly dominant in MHC^+/+ NOG mice when gating on only rapidly dividing T cells (Fig. 2D). This indicates that effector differentiation of T cells is facilitated by TCR-dependent immune response independently of their proliferation.

Following the process for effector differentiation, the BODIPY intensity of T cells showed a stepwise increase in MHC^+/+ NOG mice and reached a maximum at EMRA cells in both CD4⁺ and CD8⁺ T cells (Fig. 2E, 2F). Donor CM T cells in MHC^−/− NOG mice also showed a higher BODIPY intensity than naive T cells; however, the intensity was not significantly different among CM, EM, and EMRA T cells in either CD4⁺ or CD8⁺ T cells (Fig. 2G, 2H). Although the FA uptake in CM T cells was comparable between MHC^+/+ and MHC^−/− NOG mice, EM and EMRA T cells in MHC^+/+ NOG mice exhibited a significantly higher FA uptake than did those in MHC^−/− NOG mice, respectively (mean fluorescence intensity of BODIPY 500/510 C₁₂, EM 1577 ± 140 versus 879 ± 138; EMRA 1863 ± 193 versus 932 ± 171), suggesting that a higher amount of FAs may be required for effector function in donor T cells. CD4⁺CD25⁺FoxP3⁺ regulatory T cells, which are known to generate energy through lipid oxidation (20), constituted only 3.1 ± 0.8% of total CD4⁺ T cells. Taken together, these results suggest that, besides a numerical predominance of EM T cells, a greater capacity to incorporate extracellular FAs into EM and EMRA T cells contributes to the higher lipid accumulation of T cells in MHC^+/+ NOG mice during GVHD.

To determine if structural changes in mitochondria coincide with the changes in FA metabolism in T cells during GVHD, ultrastructural analyses of T cells in MHC^+/+ NOG mice (referred to as GVHD T cells hereafter) and freshly isolated T cells from healthy donors (control T cells) were performed with sample preparation enhancing cell membranes and transmission electron microscopy (Fig. 3A, 3B) (21). Although the profile sizes of GVHD T cells were significantly larger than those of control T cells (Fig. 3A–3D), the cellular areas occupied by mitochondria (Mito area/cell area) in individual cells were not significantly different in either group (Fig. 3A–3C, 3E), suggesting that the total mitochondrial volume in individual T cells was increased during GVHD. This was further confirmed by flow cytometry using MitoTracker green staining and forward scatter (area) (Supplemental Fig. 2). The mitochondrial ultrastructure was also different, and the mitochondrial matrix was lucent and obvious in GVHD T cells compared with control T cells (Fig. 3A2, 3B2). Close interaction of mitochondria with endoplasmic reticulum (ER) (a characteristic feature of memory T cells is that lipids synthesized in ER are directly incorporated into proximal mitochondria [22, 23]) was not significantly different between control and GVHD T cells. Despite the increased FA uptake in GVHD T cells, abundant lipid droplets were absent in the cytoplasm of GVHD T cells, suggesting that the incorporated FA can be immediately used for further metabolic reactions. These results demonstrate that GVHD T cells exhibit larger cellular and mitochondrial volume along with structural changes in mitochondria, which support changes in mitochondrial function, depending on uptake in extracellular FA.

To characterize the gene expression profile associated with lipid metabolism in donor T cells during GVHD, we then performed quantitative real-time PCR analysis of 84 key genes involved in the regulation and enzymatic pathways of FA metabolism (Fig. 4A). Gene expression levels were compared in human T cells obtained from MHC^+/+ and MHC^−/− NOG mice and freshly isolated T cells from healthy donors. Nearly half (38 of 84 genes) of lipid metabolism–related genes were upregulated in human T cells of MHC^+/+ NOG mice compared with those of MHC^−/− NOG mice (Fig. 4B, Table I). We observed the upregulation (defined as more than 2-fold) of mRNAs encoding the enzymes involved in FA transport, including the rate-limiting enzyme for FA oxidation, carnitine palmitoyltransferase (CPT1B), FA binding protein (FABP1-4, FABP6, and FABP7), and FA pathway, including acyl-CoA synthase (ACSBG2), acyl-CoA dehydrogenase (ACAD9-11, ACADS, and ACADL), and FA synthase (FASN) in T cells in MHC^+/+ NOG mice compared with MHC^−/− NOG mice (Fig. 4C). Similarly, the expression of genes encoding the enzymes in triacylglycerol metabolism, such as glycerol kinase (GK, GK2) and lipoprotein lipase (LPL), was upregulated in GVHD T cells. Furthermore, the expression of genes associated with mevalonate pathways, such as HMG-CoA synthase (HMGCS2), was also upregulated in GVHD T cells. These observations suggest that a series of processes of lipid metabolism, including lipid hydrolysis, mitochondrial FA transport, and FAO can be accelerated in donor T cells during GVHD, and excessive lipids can be stored as cholesterol esters.

Given that donor T cells upregulate genes related to mitochondrial FA transport and oxidation during GVHD, it is likely that the incorporated FAs are further transferred into mitochondria and subsequently used for bioenergy production. To test this assumption, we compared mitochondrial OXPHOS in T cells in MHC^+/+ and MHC^−/− NOG mice by measuring the OCR using extracellular flux analysis (Fig. 5A). The mitochondrial maximal respiration rate and spare respiratory capacity (SRC) in donor T cells of MHC^+/+ NOG mice appeared to be higher than those of MHC^−/− NOG mice (Fig. 5B), although the differences were not statistically significant (maximal respiration rate, p = 0.25; SRC, p = 0.45) (Fig. 5C, 5D).

Next, to investigate the role of mitochondrial FA transport in energy production in donor T cells, we assessed the OCR in the presence or absence of etomoxir, a well-described CPT1 inhibitor, in GVHD mice. Notably, treatment with etomoxir significantly decreased both the mitochondrial maximal respiration rate (Fig. 5C) and SRC (Fig. 5D) in T cells in MHC^+/+ NOG mice, but not in those from MHC^−/− NOG mice. Soon after treatment with etomoxir, both the mitochondrial maximal respiration rate and SRC in T cells in MHC^+/+ NOG mice decreased to levels comparable to those in MHC^−/− NOG mice (Fig. 5C, 5D). Supporting this fact, T cell metabolism was substantially impaired under FA-free conditions but was restored significantly by lipid supplementation (Supplemental Fig. 3 and data not shown). To further confirm these observations, we measured mitochondrial FAO activity in T cells by FAO staining. With its ability to exhibit fluorescence only after degradation by FAO in mitochondria, the FAO activity mediated by exogenous lipids can be assessed by flow cytometry (24). As shown in Fig. 5E, human T cells activated with T cell TransAct showed higher fluorescence intensity of FAOBlue than those stimulated with cytokines. Furthermore, we observed that the fluorescence intensity was significantly increased in GVHD T cells compared with pretransplant T cells (Fig. 5F). These observations suggest that mitochondrial FA transport may be necessary for achieving the maximal capacity of mitochondrial respiration in effector T cells during GVHD, whereas it is dispensable or can otherwise be compensated by other nutrients for T cells undergoing spontaneous proliferation.

Finally, to investigate whether donor T cells use glucose as a carbon source to fuel the tricarboxylic acid cycle, we assessed the OCR in GVHD T cells after incubation in normal media (10 mM) and media with low glucose (2 mM). There is evidence that, at such a low concentration of glucose, the OCR is significantly reduced in memory T cells, but not effector T cells, due to the different dependence on glucose availability in the process of OXPHOS (22, 25). Neither the maximal respiration rate nor SRC in GVHD T cells was decreased in low-glucose media compared with normal media (Fig. 6A, 6B, and data not shown). No additional inhibitory effect on OCR was observed when etomoxir was used under glucose-deprivation conditions (Fig. 6A, 6B). Furthermore, pharmacological blockade of mitochondrial pyruvate transport by UK5099 (mitochondrial pyruvate carrier; 5 μM) did not inhibit mitochondrial OXPHOS in GVHD T cells, whereas it did so at a high concentration (25 μM) (Fig. 6C, 6D). These observations indicate that although glycolytic substrates might also be involved in mitochondrial energy production, reduced glucose-driven OXPHOS can be compensated to a certain degree in effector T cells.

Using xenogeneic GVHD models, we have demonstrated that donor T cells increase extracellular uptake of FAs as well as glucose through the interaction of TCR with host MHC Ags. There was a clear association between FA uptake in donor T cells and their proliferation rate and effector differentiation. Furthermore, we found that lipid metabolism–related genes, including those associated with lipid hydrolysis, mitochondrial FA transport, and FAO, were upregulated in GVHD T cells. Pharmacological blockade of CPT1 reduced the maximal respiration rate and SRC in GVHD T cells, whereas substantial disruption of the glucose supply by either glucose deprivation or mitochondrial pyruvate transport blockade did not impair OXPHOS activity, demonstrating that exogenous FAs play a central role in mitochondrial biogenesis in T cells during GVHD. To the best of our knowledge, this is the first report to characterize FA metabolism of effector T cells in humans during GVHD.

Our results clearly demonstrate that GVHD T cells and, to a lesser extent, T cells undergoing spontaneous proliferation have a capacity to increase the uptake of FAs as well as glucose at the same time. Once stimulated with host MHC Ags, donor T cells continued to increase extracellular FA uptake during activation and even after they had entered a state of “exhaustion.” In contrast, pharmacological blockade of CPT-1 did not disrupt extra mitochondrial capacity in T cells in MHC^−/− NOG mice. These observations indicate that in T cells undergoing spontaneous proliferation, exogenous FAs can be used mainly for biosynthetic processes, but not for bioenergy production; thus, it is likely that disruption of the process of FA transport and/or FAO has a minimal effect on immune reconstitution after SCT. One limitation is that we were not able to fully compensate for the potential difference in donor memory/effector T cell composition between MHC^+/+ and MHC^−/− NOG mice, even though only actively proliferating T cells were used for flow cytometric analysis. Likewise, human T cells studied in our models include heterogeneous with distinct compositions, suggesting that a large number of differences observed in bulk assays can be associated with different cell types to some extent.

It has been shown that effector differentiation and full activation of T cells cannot be achieved if mitochondrial function is impaired (26, 27). Reactive oxygen species (ROS) generated in mitochondria appeared to be essential for the activation of NF of activated T cells and subsequent IL-2 induction during early T cell activation (27). Genetic reduction or pharmacological inhibition of complex III ROS disrupted Ag-specific CD8⁺ T cell expansion but did not affect spontaneous expansion. In fact, increased ROS levels in alloreactive T cells have been demonstrated in mouse bone marrow transplantation models (14, 28). These findings suggest that mitochondrial FA transport and FAO, followed by ROS generation, in donor T cells can be a potential target for the treatment of GVHD.

Despite the extensive FA uptake, we did not visualize lipid accumulation in the cytoplasm of GVHD T cells by electron microscopy. Similarly, lipid droplets are not observed in memory CD8⁺ T cells, although they synthesize large amount of lipids from extracellular glucose to support FAO (22, 29). Ultrastructure analysis reveals the accumulation of mitochondria in close association with ER in memory T cells, suggesting that lipids synthesized in ER can be immediately used for energy production in nearby mitochondria. In contrast, we observed that GVHD T cells had uniformly dispersed mitochondria within the cytoplasm, supporting the concept that the supply source of FAs primarily depends on extracellular uptake rather than on intracellular synthesis in effector T cells. Unlike adipocytes, nonadipose cells, including lymphocytes, have a limited capacity to store excess lipids as triglyceride (29). To avoid long-term exposure to free FAs, which can lead to cellular apoptosis referred to as “lipotoxicity,” intracellular FAs may need to be metabolized immediately for lipid synthesis or oxidative phosphorylation.

It has been widely accepted that T cells increase extracellular uptake of FAs as well as glucose in response to TCR stimulation (13, 14, 28, 30). To date, however, no consensus has been reached with regard to the subsequent fate of FAs after incorporation into T cells; it is controversial whether they are used as biomaterials for robust proliferation or as a fuel source for energy production. The discrepancy between studies is probably due to the differences in experimental conditions. The use of appropriate animal models is a prerequisite for studies on T cell metabolism during GVHD, because the cellular metabolism can be drastically affected by their surrounding environment during inflammation. Furthermore, the expression of CPT1α in alloreactive T cells appears to be different when different combinations of mice are used in transplantation (12, 14), which may lead to inappropriate conclusions.

Recently, TCR stimulation has been shown to upregulate mTOR proliferator-activated receptor-γ and sterol regulatory element-binding proteins, thereby promoting FA uptake and biosynthesis, respectively, in activated CD4⁺ T cells (10). Given the upregulation of genes related to lipid synthesis pathways, including glycerol kinase, lipoprotein lipase, HMG-CoA synthase, and FA synthase, in GVHD T cells (Fig. 4B, 4C), it is plausible that exogenous FAs are also used for biosynthesis processes to induce robust proliferation other than mitochondrial biogenesis. Decreased FA uptake due to reduced proliferator-activated receptor-γ expression has been shown to impair basal OCR and SRC (10), which is also consistent with our results. These observations suggest that effector T cells may need an extra supply of FAs from outside the cells to maximize their wide variety of functions. In addition, although glucose has been described as a carbon source of acetyl-CoA for histone acetylation, a recent study has shown that lipid-derived acetyl-CoA can be a major source of carbon for histone acetylation (31).

Given our novel findings that extracellular FA uptake and their mitochondrial transport are required to fully achieve the maximal respiration activity in effector T cells, CPT-1 can be considered an attractive target for the treatment of GVHD. Indeed, etomoxir treatment has been shown to reduce the clinical severity of GVHD in minor histocompatibility Ag-mismatched murine models (14). In a MLR, etomoxir inhibited proliferation of donor T cells obtained from GVHD mice, but not unmanipulated T cells, suggesting that inhibition of FAO selectively targets alloreactive T cells (14). One concern is that, when used at high concentrations, etomoxir may have off-target effects on the electron transport chain in T cells independent of CPT-1 inhibition (32, 33). In fact, treatment with etomoxir did not confer any survival benefit in our system; rather, long-term treatment with etomoxir seemed to have a toxic effect in mice (data not shown). This could be attributed to an innate vulnerability of NOG mice, off-target effect, or severe liver toxicity of etomoxir, which has been reported in clinical trials (34). Future studies should develop new therapies targeting extracellular FA uptake and subsequent oxidation other than CPT-1, using more specific inhibitors, that could regulate donor T cell bioactivity more efficiently.

The authors declare no conflicts of interest.

The authors thank Dr. Tom Koki (Jichi Medical University) for his excellent technical assistance.