Peroxisome Proliferator-Activated Receptor–δ Supports the Metabolic Requirements of Cell Growth in TCRβ-Selected Thymocytes and Peripheral CD4⁺ T Cells

The development of αβ T cells occurs in the thymus through a highly regulated process that is controlled by interactions between thymocyte progenitors and thymic stromal cells (1, 2). T cell progenitors that arrive from the bone marrow are double negative (DN) for CD4 and CD8. The least mature of these cells, termed early T cell progenitors (ETP), can be identified within the DN population by CD25⁻CD44⁺CD117⁺ expression (3). Once in the thymus, ETP differentiate into DN2 (CD44⁺CD25⁺) and then DN3 (CD44⁻CD25⁺) cells and at the same time undergo T cell lineage commitment and initiate rearrangements at the TCRβ locus (2). TCRβ rearrangement ceases at the DN3 stage, at which time thymocytes undergo β-selection, a process whereby thymocytes are selected based on the productive rearrangement of the TCR β-chain (2, 4). During successful rearrangement, the nascent TCR β-chain pairs with the pre-T α-chain and CD3 signaling molecules to form a functional pre-TCR. Signaling via the pre-TCR rescues thymocytes from death by neglect, initiates allelic exclusion at the TCRβ locus, and induces thymocytes to proliferate and differentiate into DN4 and then CD4⁺CD8⁺ double-positive (DP) thymocytes (4). At the DP stage, thymocytes initiate TCRα rearrangements and undergo positive and negative selection based on the strength of interactions of the TCR with self-peptide/MHC class I and II to form single-positive (SP) CD4⁺ and CD8⁺ T cells (2).

The greatest expansion of thymocytes during their development occurs immediately following β-selection (4, 5). This amplification step serves to increase the diversity of the TCR repertoire because it permits a greater number of pairings of different TCR α-chains with successful TCR β-chains (6). This proliferative burst of TCRβ-selected DN3 (also termed DN3b) and DN4 thymocytes is signaled through the pre-TCR and stromal-derived factors (delta-like 4, IL-7, and CXCL12) acting through receptors on thymocytes (7–13). In response to these signals, TCRβ-selected DN3 thymocytes switch from a state of quiescence to that of active cycling and at the same time engage pathways to increase nutrient uptake to accumulate the necessary biomass to support repeated cell divisions (4). The major signaling pathway known to spur anabolic processes in thymocytes is the PI3K/3-phosphoinositide–dependent protein kinase 1/Akt signaling cascade (14, 15). This pathway is activated downstream of the pre-TCR, Notch1, and IL-7 receptor signals and promotes thymocyte growth by increasing the expression of nutrient receptors (glucose transporter 1, CD71, and CD98), glycolytic enzyme activity, and protein translation through phosphorylation of 90 kDa ribosomal S6 kinase (RSK) (7, 14–16). The PI3K/Akt pathway is similarly engaged to increase metabolism in peripheral CD4⁺ T cells downstream of stimulation of CD28 or by common γ-chain cytokines (17, 18). The downstream consequences of this signaling are that aerobic glycolysis increases, which not only enhances ATP levels but also helps to provide metabolite building blocks such as nucleotides and amino acids to support cell growth (7, 17, 18). More recently, it has been shown that liver kinase B1 (LKB1), which is activated in response to energy stress within cells, is also critical to thymocyte and peripheral T cell survival and proliferation (19, 20). Similar to PI3K, LKB1 activity promotes T cell growth by promoting the expression of the amino acid transporter CD98 and phosphorylation of RSK (20). Together, these studies have thus demonstrated a critical role for increased glucose and amino acid uptake, glycolytic metabolism, and protein translation in supporting the proliferation of TCRβ-selected thymocytes and activated peripheral CD4⁺ T cells.

Peroxisome proliferator-activated receptor–δ (PPAR-δ, also known as PPAR-β) is a ligand-activated transcription factor that belongs to the PPAR nuclear hormone receptor superfamily that also includes PPAR-α and PPAR-γ (reviewed in Ref. 21). PPAR-δ is broadly expressed in tissues, including the thymus and spleen (22), and is considered to be a sensor of fatty acid intermediates, including dietary fatty acids (both unsaturated and saturated medium- to long-chain fatty acids) as well as certain arachidonic acid derivatives (e.g., prostacyclin I) (21, 23). Upon ligand binding, PPAR-δ regulates the cellular expression of hundreds of genes involved in diverse processes, including cell proliferation, cell differentiation, fatty acid and glucose metabolism, cell signaling, and detoxification (24, 25). In the most commonly described mechanism of transcriptional activation by PPAR-δ, ligand binding leads to a conformational change in PPAR-δ that favors its heterodimerization with the retinoid X receptor, dissociation with corepressor complexes, and recruitment of coactivating complexes (21). This transcriptional activation occurs through the direct binding of PPAR-δ to PPAR-responsive elements in control regions of target genes (21). In addition, there is evidence that PPAR-δ represses some genes indirectly through interactions with other transcription factors (NF-κB, STAT-3) or transcriptional repressors such as B cell lymphoma 6 (BCL-6) (21). These mechanisms of transcriptional regulation by PPAR-δ have largely been defined in studies that administered synthetic ligands. How PPAR-δ functions in immune cells in the mammalian host with only endogenous fatty acid ligands available is not as well understood. Examining the phenotype of PPAR-δ−deficient mice is, therefore, important to understanding the activities of endogenous ligands on PPAR-δ in the regulation of physiological processes.

In this article, we report that mice with whole body deficiency of PPAR-δ (PPARδ^mut mice) exhibit peripheral lymphopenia and reduced thymic cellularity. This reduction in thymic cellularity correlated with compromised thymocyte growth at the DN3 and DN4 stage. Although in vivo DNA synthesis and expression of nutrient receptors was unaffected in PPAR-δ–deficient DN3b and DN4 thymocytes, further studies using the OP9–delta-like 4 (OP9-DL4) in vitro system of thymocyte differentiation revealed that DN3 and DN4 thymocytes divided a fewer number of times in culture. Characterization of the phenotype of peripheral CD4⁺ T cells in PPARδ^mut mice revealed even greater deficits in the ability of these cells to proliferate and survive under conditions of anti-CD3 and anti-CD28 stimulation. These growth defects in PPARδ^mut thymocytes and peripheral CD4⁺ T cells correlated with reduced extracellular acidification rate (ECAR), mitochondrial reserve, and decreased expressions of genes encoding enzymes involved in glycolysis, the tricarboxylic acid cycle (TCA), the electron transport chain (ETC), and lipid biosynthesis. By contrast, mice that had T cell–restricted deficiency in PPAR-δ starting at the DP stage of thymocyte development exhibited a normal immune compartment despite having defects in CD4⁺ T cell proliferation, pointing to developmental defects as the driver of the peripheral lymphopenia in PPARδ^mut mice. Taken together, our results implicate PPAR-δ as an additional regulator of the metabolic program that supports the growth of thymocytes and mature CD4⁺ T cells.

Mice that have targeted neomycin disruption of the last exon of PPAR-δ gene have been described previously (26). These mice express an unstable PPAR-δ transcript that is missing the C terminus ligand-binding domain, resulting in low to undetectable PPAR-δ protein expression (26). Homozygous mice on a mixed SV.129/C57BL/6 background (estimated two generations to C57BL/6J) were acquired from Dr. A. Chawla (27) and were crossed an additional six generations to C57BL/6J. Heterozygotes from the last cross were used to generate the homozygous PPAR-δ mutant (PPARδ^mut) and wild-type (WT) mice used for these studies. Mice were not bred further onto C57BL/6 background because PPARδ^mut mice were not generated at a high enough frequency in the ninth generation to support experimental studies.

Mice with a T cell–specific deficiency in PPAR-δ were generated by crossing mice that were homozygotes for a floxed exon 4 of the PPAR-δ gene (Ppard^fl/fl) (stock no. 5897; The Jackson Laboratory) (28) with mice that were transgenic for expression of the Cre recombinase (Cre) gene driven by either the proximal lymphocyte protein tyrosine kinase promoter (pLck-Cre) (stock no. 3802; The Jackson Laboratory) (29) or the distal Lck-cre promoter (30) (dLck-Cre) (stock no. 12837; The Jackson Laboratory). For studies using the pLck-Cre transgenic mice, comparisons were initially made between littermate control Ppard^fl/fl mice that were also heterozygotes for pLck-cre allele (pLck-Cre⁺) and Ppard^fl/fl counterparts (see data in Supplemental Fig. 2). Based on reports that expression of the pLck-cre allele increases thymocyte apoptosis at the DP stage (31, 32), mice that were heterozygotes for the pLck-Cre transgene that did not express the floxed transgene were used as controls for experiments in the OP9-DL4 system. For studies using the dLck-Cre transgenic mice, comparisons were made between littermate control Ppard^fl/fl mice that were also heterozygotes for dLck-Cre gene (dLck-Cre⁺) and Ppard^fl/fl mice. Rag1 knockout mice were obtained from Jackson Laboratories (stock no. 2216; The Jackson Laboratory). All mice were maintained within a specific pathogen–free facility at the University Health Network, and work was done under an animal use protocol (AUP 2863) that was approved by the institutional animal care committee.

PCR genotyping of tail DNA from PPARδ^mut and WT mice was performed as described previously (27). Genotyping of Ppard^fl/fl pLck-Cre⁺ and Ppard^fl/fl dLck-Cre⁺ mice was performed by PCR amplification of tail DNA using primers that recognized a segment of the floxed PPAR-δ allele (WT 359 bp, PPARδ^mut 400 bp) and the proximal (100 bp) or distal Lck-Cre (300 bp) transgene. Primer sequences are listed in Table I. PCR cycling conditions for amplification of the Ppard^fl/fl allele were 94°C for 3 min, followed by 35 cycles of 94°C for 30 s, 65°C for 1 min, and 72°C for 1 min, followed by a final extension at 72°C for 2 min. PCR cycling conditions for amplification of the pLck-Cre allele were 94°C for 3 min, followed by 35 cycles of 94°C for 30 s, 52°C for 1 min, and 72°C for 1 min, followed by a final extension at 72°C for 2 min. PCR cycling conditions for amplification of the dLck-Cre allele were 94°C for 3 min, followed by 35 cycles of 94°C for 30 s, 62°C for 30 s, and 72°C for 30 s, followed by a final extension at 72°C for 2 min.

Thymi and spleens were dissociated through a 70-μM cell strainer to produce a mononuclear cell suspension. Cells were washed with 1× PBS and then centrifuged at 335 × g for 10 min. All centrifugations were conducted at 335 × g for 10 min unless otherwise indicated. For thymocytes, the cell pellet was resuspended in FACS buffer [2% FCS in 1× PBS (pH 7.4)]. For splenocytes, RBCs were lysed with ammonium-chloride-potassium lysis buffer (0.15 M NH₄Cl, 10 mM KHCO₃, 0.1 mM Na₂EDTA) for 1 min and 15 s and then washed with 1× PBS, prior to resuspension in FACS buffer.

All flow cytometry staining steps were performed at 4°C in the dark. Fc receptors were blocked by incubating cells with purified CD16/CD32 (Clone 93) (0.25 μg/50 μl/10⁶ cells) (eBioscience) for 15 min. Cells were washed twice with 1× PBS, centrifuging in between, after which they were stained with fixable viability dye eFluor506 (1:1000) (eBioscience) for 30 min in 1× PBS. Cells were washed an additional two times with 1× PBS, then stained for 30 min with Abs that detected cell surface Ags (diluted in FACS buffer). The following Ab clones from eBioscience were used: CD8 (53.67), CD4 (GK1.5), CD44 (IM7), B220 (RA3-6B2), CD3 (145-2C11), CD45 (30-F11), NK1.1 (PK136), CD71 (R17217), CD19 (ebio1D3), and CD25 (PC61.5). The following Ab clones from BioLegend were used: CD98 (RL388) and CD117 (2B8). The lineage mixture contained the following Ab clones from BioLegend: CD3 (145-2C11), Ter119 (TER-119), CD4 (GK1.5), CD8 (53.67), CD11b (M1/70), CD11c (N418), NKp46 (29A1.4), Gr-1 (RB6-8C5), and B220 (RA3-6B2). After cell surface staining, cells were washed twice with FACS buffer. For intracellular staining for TCRβ, cells were fixed in 4% paraformaldehyde (PFA) in 1× PBS for 10 min, washed twice in 1× PBS, and then permeabilized by incubating cells for 20 min in Perm/Wash buffer (BD Biosciences). After centrifugation, cells were stained with anti-TCRβ (H57-597) (eBioscience) diluted in 1× Perm/Wash buffer for 45 min. Cells were washed twice with 1× Perm/Wash buffer and after a final centrifugation were resuspended in FACS buffer. Staining for regulatory T cells (Treg) was performed using the Mouse Regulatory T Cell Staining Kit (eBioscience). Annexin V⁺ apoptotic cells were stained using the Annexin V Apoptosis Detection Set PE-Cyanine7 (eBioscience) according to kit directions. Mitochondria content was detected by staining these cells with 400 nM of MitoTracker Green FM (Thermo Fisher Scientific) according to product specifications. Flow cytometry acquisition was performed using an LSR II (BD Biosciences), and data were analyzed using FlowJo software.

For analysis of thymocyte proliferation in vivo, mice were injected twice with 1 mg of BrdU (i.p.), 1 h apart. Thymi were harvested 2 h after the first injection and were processed for cell surface staining as described above. Intranuclear staining for BrdU was performed using the BD Pharmingen BrdU Flow Kit, according to kit instructions.

Thymocytes were processed into a mononuclear cell suspension, and non-CD4⁺ thymocytes were enriched from total thymocytes using the MagniSort Mouse CD4 Positive Selection Kit (eBioscience). Negatively selected thymocytes were then stained with lineage mixture and anti-CD44 and anti-CD25 as described above and were resuspended in 1× HBSS containing 1% FCS and DAPI (1:3333) (BioLegend) for sorting. Live DN2 (Lin⁻CD44⁺CD25⁺) or DN3 (Lin⁻CD44⁻CD25⁺) cells were sorted using a MoFlo sorter (Beckman Coulter), washed, and resuspended in MEM that contained 5% FCS, 1% penicillin/streptomycin, 1 nM of IL-7, and 1 nM of FLT-3 ligand (R&D Systems). These cells were then seeded into 24-well plates that contained 80%-confluent monolayers of OP9-DL4 cells. OP9-DL4 cells were precultured in MEM that contained 5% FCS and 1% penicillin/streptomycin. After 72–144 h of coculture with OP9-DL4 cells, thymocytes were enriched by passage through 40-μm filter paper (BD Falcon) into FACS tubes. Cells were then processed for flow cytometry as described above. To assess cell division, sorted DN2 or DN3 thymocytes were first labeled with CFSE (Invitrogen) prior to plating. CFSE staining was carried out as follows: cells (1 × 10⁶/ml) were incubated with CFSE (0.25 μM diluted in 1× PBS) for 15 min at 37°C, centrifuged (335 × g for 10 min), resuspended in 1× PBS for 30 min at 37°C, and then centrifuged again, prior to being resuspended in prewarmed complete MEM.

Naive CD4⁺ T cells were negatively selected from spleen and lymph node mononuclear cell preparations by magnetic cell isolation (MagniSort Mouse CD4 Naive T Cell Enrichment Kit; eBioscience). The purity of naive (CD44^lo) T cells was checked by flow cytometry staining for CD4 and CD44 and was always >97%. T cells were resuspended in complete RPMI media (2 mM l-glutamine, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 100 U/ml, penicillin, 0.1 mg/ml streptomycin, and 10% FCS [all from Life Technologies, Carlsbad, CA] and 50 μM 2-ME [Sigma-Aldrich, Oakville, ON, Canada]) and were plated in 96-well plates (0.2 × 10⁶ cells per well) that were precoated with a range of 0–5 μg/ml anti-CD3 (145-2C11) and 0–2 μg/ml anti-CD28 (37.51) concentrations (eBioscience). Cytokine levels in culture supernatants were measured using Ready-SET-Go cytokine ELISA kits. DNA synthesis was measured using a [³H]thymidine incorporation assay as described previously (27). For CFSE dilution assay, cells were labeled with CFSE as described above, centrifuged, resuspended in complete RPMI 1640, and plated (2 × 10⁶ cells per well) in 24-well plates that were precoated with anti-CD3 (5 μg/ml) and anti-CD28 (0.5 μg/ml).

Naive CD4⁺ T cells were isolated to >97% purity as described above, resuspended in 1× PBS at a concentration of 4 × 10⁶/ml, and allowed to rest on ice for 1 h. One million cells were then dispensed into each FACS tube (1 per time point) together with 10 μg/ml of anti-CD3 Ab, followed by a 15-min incubation on ice. Cells were then washed with ice-cold 1× PBS, centrifuged at 335 × g for 5 min at 4°C, and then resuspended in the original volume (250 μl) of 1× PBS and left on ice. Stimulations were initiated by adding anti-hamster IgG (10 μg/ml) and anti-CD28 (5 μg/ml) to each sample and transferring the tube to a prewarmed 37°C water bath. Signaling was stopped after 0, 1, 2, 5, or 15 min by the addition of 100 μl of 16% PFA to the tube. After a 15-min incubation with PFA, cells were centrifuged at 500 × g for 10 min, the supernatant was removed, and the cell pellet was resuspended in 1 ml of prechilled 100% methanol. Cells were incubated for 20 min on ice and were washed twice in FACS buffer (3 ml). After the second centrifuge, cells were resuspended in 100 μl of FACS buffer containing the phospho-Akt(Ser473) (Clone M89-61), phospho-Akt(Thr308) (Clone JI-223371), or phospho-Erk1/2(pT202/pY204) Abs (BD Biosciences) and then incubated at 4°C for 1 h. Cells were then washed twice with FACS buffer prior to flow cytometry acquisition.

Prior to the assay, the XF96 or Xfe96 sensor cartridge was calibrated in Seahorse XF Calibrant at 37°C overnight in the XF Prep Station. The next day, the XF 96-well cell culture microplate was coated with 10 μl per well of 22.4 μg/ml Cell-Tak (Corning), sealed, and then incubated for 20 min at room temperature. Excess liquid was aspirated, and wells were washed with 200 μl per well sterile Milli-Q distilled water and air dried for 10 min at room temperature. Cells were suspended at a concentration of 4 × 10⁶ cells/ml either in Seahorse Mito Stress media [XF Base Medium containing 1 mM sodium pyruvate, 2 mM l-glutamine, and 5.5 mM D-+-glucose (pH 7.4)] or Seahorse Glycolysis media [XF Base Medium containing 2 mM l-glutamine (pH 7.4)] (Agilent) for analysis of oxygen consumption rate (OCR) or ECAR, respectively. A total of 2 × 10⁵ cells were then seeded into wells of the Cell-Tak–coated plate. The plate was spun at 60 × g for 2 min with 0 acceleration and deceleration and was incubated at 37°C in the XF Prep Station for 30 min. Afterwards, each well was topped up with 100 μl of the appropriate prewarmed media. The cartridge was loaded with the appropriate drugs and was placed on top of the cell plate. OCR was measured using either the Seahorse XF96 or the Seahorse XFe96 Extracellular Flux Analyzer (Agilent) under basal conditions and in response to addition of 1 μM of oligomycin and 1 μM of fluorocarbonyl cyanide phenylhydrazone (FCCP), and 1 μM of rotenone + 1 μM of antimycin A. The latter two drugs were only added in the experiments that used the XFe96 machine because this machine had more drug ports. ECAR was measured under basal conditions (both in Seahorse Mito Stress media and Seahorse Glycolysis media) and after the addition of 10 mM of D-+-glucose, 1 μM of oligomycin, and 50 mM of 2-deoxy-d-glucose to glycolysis media. After OCR and ECAR measurements were completed, cell numbers were determined using the CyQuant NF Cell Proliferation Assay Kit (C35007; Invitrogen), and these numbers were used to normalize data.

Naive CD4⁺ T cells were isolated from spleens of WT and PPARδ^mut mice (n = 6 mice per group) using the mouse CD4⁺CD62L⁺ T Cell Isolation Kit (Miltenyi Biotec). Samples from individual mice were processed separately for generation of samples used to probe the array. Half of the cells from each sample were used for assessment of gene expression under quiescent conditions, and half of the cells were stimulated in vitro in complete RPMI media with plate-bound anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml) for 20 h. Cells were washed in 1× PBS, resuspended in lysis buffer (provided with RNA kit), and then frozen at −80°C. Total RNA was isolated using the RNeasy Mini Kit (Qiagen), RNA integrity was determined using an Agilent Bioanalyzer, and purity was checked using Nanodrop. Samples from WT mouse no. 1 and no. 2 were pooled together to provide one biological replicate because of low sample abundance. Samples from WT mouse no. 6 and from PPARδ^mut mouse no. 2 were excluded because of low RNA integrity. Fifty nanograms of RNA from each sample (n = 4 WT and n = 5 PPARδ^mut mice per group) was amplified and labeled using the Illumina TotalPrep-96 RNA Amplification Kit (lot no. 1002014; Ambion). Samples were randomized, and 1.5 μg of cDNA of each sample generated from the amplification kit was hybridized into three separate Mouse WG-6 v2.0 Expression BeadChips. The BeadChips were incubated at 58°C in a rotational hybridization oven for 18 h. The BeadChips were washed and stained according to the Illumina protocol and scanned using the iScan System (Illumina). The data files were quantified in GenomeStudio Version 2010.1 (Illumina) and analyzed using GeneSpring software (Version 12.1; Agilent). Data were normalized using a quantile followed by median-centered normalization. Logarithm (base 2)-transformed values were used for all data analyses and visualization. Data were first filtered to remove probes with no signal to avoid a potential confounding effect they may cause, and only the top 80th percentile of the distribution of intensities in any one group was used. The final set contained 39,175 probes. An unsupervised clustering using a Pearson-centered correlation as a distance metric with average linkage rules was performed to assess the overall degree of gene-expression similarity among samples. A two-way ANOVA analysis was performed on all samples (n = 8 WT and n = 10 PPARδ^mut samples per group with both naive and stimulated samples pooled) with a Benjamini–Hochberg corrected p value cut off of 0.05 applied. The test found 491 probes that differed significantly with PPAR-δ status regardless of activation state. A subsequent t test analysis was performed on stimulated PPARδ^mut (n = 5 samples) and WT samples (n = 4 samples) with a corrected p value cut off of 0.05 applied, and this revealed 1340 probes that differed significantly between PPARδ^mut and WT groups. Gene sets were then interrogated to identify differentially expressed probes that encoded genes with known metabolic, mitochondrial, or antioxidant functions. These gene microarray data have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus and are accessible through Gene Expression Omnibus series accession number GSE117461.

Thymocytes were stained for lineage markers, CD25, CD44, and DAPI as described above. Live DN4 cells (Lin⁻, CD25⁻CD44⁻) were sorted using a MoFlo sorter (Beckman Coulter). Total RNA was isolated using the RNeasy Mini Kit (Qiagen). Reverse transcription was performed using SuperScript III Reverse Transcriptase (Invitrogen), using 50 pmol of Oligo(dT)₁₈ Primer and 1 nmol each dNTP (both from Thermo Fisher Scientific) according to instructions provided in the technical data sheet. cDNAs were amplified using FastStart Universal SYBR Green Master Mix (Roche) and specific primer sets (10 pmol/12 μl reaction). Primers are listed in Table I. Cycling conditions for all genes were as follows: 95°C for 15 min, followed by 45 cycles of 94°C for 20 s, 53°C for 30 s, and 72°C for 30 s. All reactions were run in triplicate and were analyzed using a LightCycler 480 (Roche) real-time PCR machine and software. cDNAs prepared from soleus muscle were used to generate standard curves, and β-actin mRNA expression was used for normalization. In each case, melting curves were performed to confirm a single product, and PCR products were run on agarose gels to verify that the product was of the desired length.

Total CD4⁺ T cells were negatively selected from spleens and lymph nodes of Ppard^fl/fl mice and dLck-cre Ppard^fl/fl mice (n = 3 mice per group) using a similar procedure as described above, except that the MagniSort Mouse CD4 T Cell Enrichment Kit was used (eBioscience). Cells were labeled with CFSE dye as described above and were resuspended in sterile 1× PBS and injected i.v. into the tail vein of syngeneic Rag1^−/− recipients (8 × 10⁶ cells per mouse, n = 4 per group) via tail vein injection. Five days later, spleens and lymph nodes (cervical, axillary, inguinal) were harvested, processed as described above, and stained with viability dye and anti–CD4-APC, and the fluorescence of the CFSE dye in the CFSE⁺ CD4⁺ live gate was examined by flow cytometry.

Data are presented as mean + or ± SEM. When data were parametric, a Student t test (two groups) or a one-way ANOVA with Tukey post hoc test (>2 groups) was used to detect differences between groups. When data were nonparametric, ranks were compared among groups using a Mann–Whitney U test (two groups) or a Kruskal–Wallis test followed by a nonparametric post hoc test (>2 groups). A p value ≤ 0.05 was considered significant.

Previously, we noted that PPARδ^mut mice exhibited a higher percentage of memory CD4⁺ T cells in the peripheral lymphoid compartment as compared with PPAR-δ^+/+ (WT) mice (27). Because upregulation of CD44 occurs during homeostatic proliferation of CD4⁺ T cells in a lymphopenic environment (33), we further analyzed the lymphocyte composition of the spleens of PPARδ^mut and WT mice (Fig. 1). We observed that both male and female PPARδ^mut mice exhibited lowered spleen weights and cell numbers compared with WT counterparts (Fig. 1A, 1B). Both T and B cell numbers were reduced in this organ in PPARδ^mut mice (Fig. 1H). Consistent with our previous findings, we observed that the frequency of CD44^hi cells within the CD4⁺ compartment was higher in PPARδ^mut compared with WT mice (Fig. 1E, 1I), as was the frequency of CD25⁺Foxp3⁺ cells (Fig. 1F, 1I), which can be another indicator of T cell lymphopenia (34).

To further understand the underlying basis for the T cell lymphopenia, we examined the weight and cellularity of the thymus of WT and PPARδ^mut mice between 4 and 5 wk of age. We observed that there was a modest reduction in the size of the thymus in both male and female PPARδ^mut mice as compared with WT mice (Fig. 2A, 2B), suggesting that this could be a cause of the peripheral T cell lymphopenia. Because this phenotype did not appear to be sex dependent, subsequent studies used male mice.

To determine the underlying basis for the decrease in thymus cellularity, we examined the frequency and number of each major thymocyte subset in PPARδ^mut and WT thymi. We did not observe a difference in the proportion of DN, DP, SP CD4, or SP CD8 cells between PPARδ^mut and WT thymi (Fig. 2C, 2D); however, the number of each subset was significantly reduced in PPARδ^mut mice (Fig. 2E). This result suggested that the defect in thymocyte development in PPARδ^mut mice occurred at the DN stage or earlier. γδ CD3⁺ T cells and Foxp3⁺ CD25⁺ CD4⁺ (Treg) cell frequencies and numbers were also examined, and a reduction in the number of Treg was observed in PPARδ^mut compared with WT mice (Fig. 2F–H). The finding that the percentage of Treg was not increased in the thymus of PPARδ^mut mice further reinforced the idea that enrichment of these cells in the periphery was related to T cell lymphopenia.

We then questioned whether the defect in PPARδ^mut mice occurred earlier in thymocyte development by examining the frequencies and numbers of DN cells (ETP, DN2, DN3, and DN4) in thymi of PPARδ^mut and WT mice (Fig. 3A, see root gating strategy in Supplemental Fig. 1A). This analysis revealed that PPARδ^mut mice exhibited an increase in the frequency of DN2 cells compared with WT mice (Fig. 3B). When DN frequencies were converted to cell numbers, it became evident that thymocyte numbers in PPARδ^mut mice were significantly reduced at the DN4 stage of development (Fig. 3C). In some experiments, we also stained thymocytes with intracellular TCRβ to help distinguish pre– (iTCRβ⁻, DN3a) from post– (iTCRβ⁺, DN3b) TCRβ-selected DN3 cells (Fig. 3D, see root gating strategy in Supplemental Fig. 1B). This analysis revealed that although there was no significant difference in the total number of DN3a or DN3b cells between PPARδ^mut and WT mice (Fig. 3E), the fraction of DN3a to DN3b cells was higher in PPARδ^mut mice (Fig. 3F). Consistent with these findings of a role for PPAR-δ at the DN4 stage, it has been reported that Ppard mRNA expression increases from the DN to DP stage (35). We further delineated that Ppard expression increases between the DN3 and DN4 stage of thymocyte development (Fig. 3G).

In parallel studies, we examined the DN T cell compartment in mice that exhibited PPAR-δ deficiency restricted to the T cell compartment. These mice were generated by crossing mice with a floxed exon 4 of Ppard with those expressing Cre recombinase under the control of the proximal Lck promoter, which is expressed starting at the DN3 stage of thymocyte development (31). Mice that were homozygotes for the floxed allele and heterozygotes for the proximal Lck-Cre allele (Ppard^fl/fl pLck-Cre⁺) were generated and used for experiments. We observed that thymus weight was reduced by 17% in Ppard^fl/fl pLck-Cre⁺ mice (4.6 + 0.2 mg/kg body weight) relative to pLck-Cre⁺ counterparts (5.5 + 0.4 mg/kg body weight) (n = 8 mice per group, p = 0.03 by two-tailed t test). Similar to the phenotype observed in the PPARδ^mut mice, thymocyte numbers in Ppard^fl/fl pLck-Cre⁺ mice were reduced starting at the DN4 stage, with a higher fraction of DN3a to DN3b cells evident in these mice (Supplemental Fig. 2). These data supported a T cell–intrinsic role for PPAR-δ in thymocyte development.

Our findings of a higher fraction of DN3a to DN3b cells in both PPARδ^mut and Ppard^fl/fl pLck-Cre^+/− mice suggested that β-selection did not proceed as efficiently in these mice or that the proliferation or survival of PPAR-δ–deficient DN3 and DN4 thymocytes was compromised after passing the β-selection checkpoint. To help distinguish between these possibilities, we examined the proliferation of DN3b and DN4 cells in vivo by pulsing mice with BrdU and then measuring BrdU incorporation in DN3b and DN4 thymocytes 2 h later by flow cytometry (Fig. 4A). In addition, we examined thymocyte size and the expression of the transferrin receptor, CD71, and the neutral amino acid transporter, CD98, which are nutrient receptors that have been identified to be important for supporting biomass accumulation in DN3 and DN4 cells downstream of PI3K and LKB1 signaling (15, 19). These studies revealed that the extent of BrdU incorporation, the expressions of CD98 or CD71, and thymocyte size did not differ between PPARδ^mut and WT mice in either DN3b or DN4 cells (Fig. 4), suggesting that pre-TCR signaling pathways leading to DNA synthesis were intact in these cells.

One disadvantage associated with studying thymocyte survival with ex vivo flow cytometry characterization is that dead cells cannot be accurately enumerated because they are rapidly engulfed as they die in vivo. It is also unfeasible to measure how efficiently cells are dividing in the natural setting. We therefore turned to the OP9-DL4 in vitro system of thymocyte differentiation to better understand the impact of PPAR-δ deficiency on thymocyte survival and proliferation. In this system, thymocytes are cultured with IL-7, FLT-3 ligand, and OP9-DL4 stromal cells, which provide physiological notch ligands and other signals that support thymocyte proliferation and differentiation (36). A further advantage of this system is that it is a closed environment, allowing one to simultaneously examine cell-intrinsic gene effects on thymocyte commitment, cell division, and survival.

We therefore sorted DN3 progenitors from either PPARδ^mut or WT thymi, seeded these cells on OP9-DL4 cells, and then examined the differentiation and yield of mature CD45⁺ cells between 3 and 5 d of culture. OP9-DL4 cocultures that were seeded with PPARδ^mut DN3 cells yielded 50% less CD45⁺ cells at day 4 postculture compared with those seeded with WT cells (Fig. 5A). Consistent with our in vivo phenotype, we observed that cell numbers tended to be lowered by the DN4 stage and onward (Fig. 5C). The fraction of DN3/DN4 cells also tended to be higher in the PPARδ^mut cocultures (Fig. 5D). We achieved similar results of reduced CD45⁺ cell yields and reduced DN4 cell numbers when we conducted experiments seeding DN2 cells from PPARδ^mut and WT mice (Fig. 5B) or DN3 progenitors from Ppard^fl/fl pLck-Cre⁺ and pLck-Cre⁺ mice (Fig. 6A, 6B). The only difference in the studies using the Ppard^fl/fl pLck-Cre⁺ progenitors was that the reduction in CD45⁺ cell yields with PPAR-δ deficiency was more modest (25%), which we expect related to less-efficient knockdown of Ppard gene expression in this system (Fig. 6C). We also examined thymocyte death in these cocultures by monitoring for the percentage of cells with viability dye staining in the various cell gates but did not observe any consistent trend of increased death of PPARδ^mut or Ppard^fl/fl pLck-Cre⁺ progenitors relative to controls in any thymocyte subset that we examined, at any time point in these experiments (data not shown).

We next examined the proliferation of thymocytes in the OP9-DL4 cocultures using a CFSE dilution assay. Although we observed no difference in the ability of PPARδ^mut and WT progenitors to divide by day 3 of culture, the frequency of DN3 cells that divided was lower in the PPARδ^mut cultures compared with WT cultures by day 4 of culture (Fig. 5E, 5F). A similar defect in the ability of DN3 thymocytes to undergo repeated cell division was observed when progenitors were sourced from Ppard^fl/fl pLck-Cre⁺ mice (Fig. 6D, 6E). Together, these data suggested that the defect in thymocyte development in mice with PPAR-δ deficiency related to a decreased ability of post–β-selected thymocytes to undergo repeated cell division.

The OP9-DL4 coculture system can also promote the development of γδ T cells, particularly when DN2 progenitors are cultured. We therefore examined the yield of γδ⁺ CD3⁺ cells at day 5 of coculture with OP9-DL4 cells after plating DN2 cells isolated from PPARδ^mut, WT, Ppard^fl/fl pLck-Cre⁺, or Ppard^fl/fl mice. We observed that although the frequency of γδ⁺CD3⁺ T cells was elevated, the number of these cells and the extent of CFSE dilution were reduced in the cultures seeded with the PPAR-δ–deficient progenitors (Figs. 5G, 6F–H), suggesting that γδ T cells were not spared from the growth defects imposed by PPAR-δ deficiency. Together, these data support a cell-intrinsic role for PPAR-δ in regulating the division of DN3/DN4 and γδ CD3⁺ thymocytes.

The finding of proliferation defects in thymocytes with PPAR-δ deficiency prompted us to explore whether similar growth defects were evident in peripheral T helper cells in PPARδ^mut mice. Because of the altered T cell composition in these mice, we limited our studies to an assessment of naive CD4⁺ T cell responses. We observed that PPARδ^mut naive CD4⁺ T cells exhibited a profound defect in the ability to proliferate compared with WT naive CD4⁺ cells in response to anti-CD3 and anti-CD28 as assessed by [³H]thymidine incorporation and CFSE dilution assay (Fig. 7A, 7B). IL-2 and IFN-γ levels were also reduced in culture supernatants of PPARδ^mut CD4⁺ T cells (Fig. 7A). Contrasting with the findings for thymocytes, PPARδ^mut CD4⁺ T cells were also more likely to die as compared with WT CD4⁺ T cells. PPARδ^mut CD4⁺ T cells also exhibited more profound defects in the ability to accumulate cell biomass after anti-CD3 and anti-CD28 stimulation (as estimated by forward scatter area [FSC-A]) (Fig. 7C).

Because of the well-known role of CD28 and PI3K signaling T cell blasting (17), we also examined the phosphorylation of Akt in PPARδ^mut and WT naive CD4⁺ T cells after a short in vitro time course of stimulation with anti-CD3 and anti-CD28. These experiments revealed equivalent Akt phosphorylation at Thr308 or Ser473 sites in PPARδ^mut and WT CD4⁺ T cells (Fig. 7D). The phosphorylation of ERK was also examined as an additional output of CD3/TCR signaling and was not different between WT and PPARδ^mut CD4⁺ T cells (Fig. 7D). These findings of normal Akt signaling in mature PPARδ^mut WT CD4⁺ T cells therefore mirror our observations of preserved expression of PI3K targets CD98 and CD71 in thymocytes and suggest that the growth defects associated with PPAR-δ deficiency are not due to compromised PI3K/Akt signaling.

The decreased ability of thymocytes and peripheral CD4⁺ T cells to accumulate biomass and undergo repeated cell divisions was suggestive of possible metabolic deficiencies in these cells. To investigate how PPAR-δ deficiency altered cellular bioenergetics, we measured the OCR and the ECAR by WT and PPARδ^mut thymocytes and CD4⁺ T cells using a Seahorse extracellular flux analyzer. Basal OCR is an indicator of oxidative phosphorylation (OXPHOS), whereas basal ECAR reflects lactate formed during glycolytic metabolism (37). In the studies of thymocytes, we sorted DN3 and DN4 cells from thymi and pooled these fractions together for the analysis. In the studies using peripheral CD4⁺ T cells, we isolated naive CD4⁺ T cells from spleens and lymph nodes of WT and PPARδ^mut mice and measured the extracellular flux of oxygen and lactate in cultures of freshly isolated CD4⁺ cells or CD4⁺ T cells that had been preactivated in vitro with anti-CD3 and anti-CD28.

We observed that the PPAR-δ–deficient thymocytes exhibited small reductions in basal OCR and maximum OCR after the addition of FCCP, which uncouples ATP synthesis from the ETC and spurs the respiratory chain to operate at maximum capacity (Fig. 8A–C). This difference between basal OCR and maximum OCR is called the spare respiratory capacity (SRC) and reflects the cell’s ability for OXPHOS under conditions in which there is a sudden increase in energy demand (38). In separate studies, we measured basal ECAR both before and after the addition of glucose to the media and found it to be also reduced in PPARδ^mut compared with WT thymocytes after glucose addition (Fig. 8D).

We observed similar but more profound metabolic deficiencies when experiments were conducted comparing WT and PPARδ^mut naive CD4⁺ T cells but only after these T cells were first preactivated in vitro with anti-CD3/anti-CD28. Although activated CD4⁺ T cells from PPARδ^mut mice exhibited a tendency for reduced basal OCR (Fig. 8E, 8F), more profound reductions in SRC (Fig. 8E, 8G) and glucose-induced ECAR were noted for these cells (Fig. 8H). In some experiments, we also measured ECAR by PPARδ^mut and WT CD4⁺ T cells in the presence of pyruvate with no added glucose and observed ECAR to also be reduced in cultures of PPARδ^mut CD4⁺ T cells; however, the extent of this decrease did not approach that seen in the presence of glucose (Fig. 8I). These results suggest that the growth defects in PPARδ^mut thymocytes and CD4⁺ T cells are associated with decreases in basal glycolytic activity and mitochondrial reserve.

To further address whether decreases in mitochondrial reserve in PPARδ^mut CD4⁺ T cells contributed to the growth defects in these cells, we also compared the proliferation of WT and PPARδ^mut CD4⁺ T cells after culture in pyruvate- and glutamine-containing media that was supplemented with either galactose or glucose as a substrate. Culturing cells in the presence of galactose forces the cells to engage mitochondrial metabolism for cell growth (39, 40). These experiments revealed severe proliferation (or survival) deficits in PPARδ^mut CD4⁺ T cells regardless of whether the cells were cultured in the presence of galactose or glucose (Fig. 8J), suggesting that the glycolytic and mitochondrial defects in PPARδ^mut CD4⁺ T cells had the potential to influence cell growth and survival.

A previous study reported that SRC correlates closely with T cell mitochondrial content in CD8⁺ T cells (38). We therefore measured mitochondrial content in both DN3 and DN4 thymocytes and CD4⁺ T cells from WT and PPARδ^mut mice by staining these cells with MitoTracker dye and measuring dye fluorescence using flow cytometry. Although there were no major differences in the intensity of staining in DN3 or DN4 cells with PPAR-δ deficiency (Fig. 8K), mitochondrial content did increase in PPARδ^mut CD4⁺ T cells relative to WT CD4⁺ T cells after extended activation with anti-CD3/anti-CD28 (Fig. 8L). Together, these findings suggested that mitochondrial defects were present in both PPARδ^mut thymocytes and peripheral CD4⁺ T cells but that these defects had a greater impact on metabolism in peripheral CD4⁺ T cells, which increased mitochondrial content to maintain OXPHOS.

To define the metabolic gene targets of PPAR-δ in CD4⁺ T cells, we conducted a microarray study that compared gene expression in PPARδ^mut and WT naive CD4⁺ T cells (n = 4–6 biological replicates per group), both under quiescent conditions after activation with anti-CD3 and anti-CD28. We reasoned that genes transcriptionally activated by PPAR-δ would be present at lower levels in PPARδ^mut compared with WT CD4⁺ T cells. First, a two-way ANOVA was performed to examine global differences between PPARδ^mut and WT CD4⁺ T cells (samples from both quiescent and anti-CD3– and anti-CD28–activated cells combined). This analysis identified 33 genes with known metabolic or mitochondrial-related functions that were present at lowered levels in PPARδ^mut CD4⁺ T cells (Supplemental Table I). A subsequent t test analysis of gene expression identified an additional 32 genes that were differentially expressed between PPARδ^mut and WT CD4⁺ T cells under conditions of TCR/CD28 costimulation (Supplemental Table I). Just over one-third of these genes were either predicted or bona fide targets of PPAR based on in silico analyses (http://www.ppargene.org/) (Supplemental Table I).

Compared with WT CD4⁺ T cells, PPARδ^mut CD4⁺ T cells exhibited reduced levels of mRNAs encoding key enzymes in glycolysis (Eno1, Pkm2), an alternative glucokinase (Adpgk), a gene (Brp44) that regulates pyruvate entry into the mitochondria, and Pdk1, which encodes an enzyme that phosphorylates and inhibits pyruvate dehydrogenase kinase 1, the gatekeeper for pyruvate entry into mitochondria. PPARδ^mut CD4⁺ T cells also exhibited reduced expressions of Idh3b and Idh3g, which encode subunits of NAD⁺-dependent isocitrate dehydrogenase, a rate-limiting enzyme of the TCA cycle, as well as genes that are involved in the assembly (Cmc1, Ndufaf1) or functioning of the ETC (Ndufb10, Ndufb6, Coq10b, Cox10, Atp5o, Atp5f1). In addition, genes involved in peroxisomal β-oxidation (Abdc3, Acbd4, Hsd17b4, Crls1, Acaa1a, Scp2) were expressed at lowered levels in PPARδ^mut compared with WT CD4⁺ T cells.

PPARδ^mut CD4⁺ T cells also exhibited reduced expression of genes that encode enzymes involved in providing the amino acid or lipid substrates to support cell growth. Specifically, PPARδ^mut CD4⁺ T cells exhibited reduced mRNA expression of the lipogenic enzyme, Acsl5; enzymes in ketone metabolism involved in producing and activating acetoacetate for entry into lipid biosynthesis pathways (Bdh1, Oxct1); and enzymes (Shmt1, Tha1) that are involved in the liberation of one-carbon units for biosynthetic pathways. Organelles also divide during cell proliferation. In this respect, we observed a number of genes involved in mitochondrial (Crtc3, Ppargc1b, Tomm40l) and peroxisomal biogenesis that were lowered in PPARδ^mut compared with WT CD4⁺ T cells, including components of the mitochondrial transcriptional (Polrmt) and translational machinery (Mrps17, Mrps34, Mtrf1l, Mrpl10). Among the mitochondrial genes that were reduced in PPARδ^mut CD4⁺ T cells were Nnt and Txn2, which play roles in detoxification of mitochondrial reactive oxygen species (Nnt and Txn2) (Supplemental Table I).

Notably, our analysis also revealed a number of metabolic genes to be present at higher levels in PPARδ^mut than WT CD4⁺ T cells(see Supplemental Table II for complete list of genes), including sugar and amino acid transporters (Slc2a6, Slc35b1, Slc15a3, Scl7a8), proteins involved in mitochondrial biogenesis (Creb1) or fission (Fis1, Pgs1, Creb), mitochondrial protein translation (Mrpl22), or the respiratory chain (Ndufa12, Cox6a1). There were also a number of genes that were upregulated that are involved in mediating cellular adaptations to hypoxia and oxidative stress (e.g., Hba-a1, Hbb-b1, Clic4, Gfer, Gsto1, Cpox, Glut6) (Supplemental Table II), the most notable being a 10–20-fold upregulation of the genes encoding α- and β-chains of the oxygen-sensing molecule hemoglobin. Higher expression of hemoglobin in PPARδ^mut CD4⁺ T cells did not relate to erythrocyte contamination because it was not accompanied by higher expression of other erythroid lineage genes (Spta, Sptb, Gypa, Alas2) on the microarray. Only a few of the upregulated genes in PPARδ^mut CD4⁺ T cells were predicted PPAR-δ targets (Gsto1, Clic4, Pgs1), suggesting that these changes in expression were likely adaptations to the metabolic deficiencies imposed by PPAR-δ deficiency. Taken together, these findings suggest that endogenous PPAR-δ activity promotes expression of key enzymes in CD4⁺ T cells that are involved in glycolysis and lactate formation, OXPHOS, mitochondrial and lipid biogenesis, and peroxisome functioning. Moreover, CD4⁺ T cells and thymocytes appeared to adapt to these metabolic deficiencies by upregulating genes that are involved in mitochondrial biogenesis or countering hypoxic stress.

To address whether PPAR-δ deficiency impacted metabolic gene expression in a similar way in TCRβ-selected thymocytes, we sorted DN4 cells from PPARδ^mut and WT thymi and measured the mRNA expressions of select PPAR-δ target genes in these cells using real-time PCR. We focused this analysis on DN4 cells because this cell population exhibited higher Ppard expression and proliferative activity compared with DN3 cells (Figs. 3G, 4A). Consistent with our findings in peripheral CD4⁺ T cells, key genes in glycolysis and lactate production (Adpgk, Pkm2, Eno1, Pdk1), the citric acid cycle (Idh3b), and the assembly (Cmc1) and functioning (Ndufb10) of the mitochondrial respiratory chain were present at reduced levels in PPARδ^mut compared with WT DN4 cells (Fig. 9). The expression of the antioxidant gene Txn2 and genes involved in ketone metabolism and the generation of lipogenic precursors (Bdh1, Oxct1, Ascl5) were also reduced in PPARδ^mut compared with WT DN4 cells (Fig. 9). In addition, the expression of hemoglobin subunit Hbb-b1 was upregulated in PPARδ^mut compared with WT DN4 thymocytes (Fig. 9). Taken together, these findings suggest that PPAR-δ regulates expression of a similar set of key metabolic genes in thymocytes.

The question still remained of whether thymocyte or peripheral T cell growth defects were the driver of the T cell lymphopenia in the PPARδ^mut mice. To directly address the involvement of PPAR-δ in regulating T cell homeostasis in the periphery, we generated mice in which excision of the Ppard allele was driven by Cre recombinase expressed from the distal Lck promoter, which turns on at the DP stage of thymocyte development (30). We observed that CD4⁺ T cells harvested from the spleens of dLck-Cre⁺ Ppard^fl/fl mice exhibited a 65% decrease in Ppard mRNA expression relative to levels in Ppard^fl/fl controls (Fig. 10A). This decrease in Ppard expression in dLck-Cre⁺ Ppard^fl/fl CD4⁺ T cells was associated with defects in the proliferation of CD4⁺ T cells as assessed by [³H]thymidine incorporation and CFSE dilution assay (Fig. 10B, 10C). Contrasting with our studies of PPARδ^mut CD4⁺ T cells, dLck-Cre⁺ Ppard^fl/fl CD4⁺ T cells did not exhibit defects in survival or blasting (Fig. 10D). Experiments were also performed to contrast the ability of dLck-Cre⁺ Ppard^fl/fl and Ppard^fl/fl CD4⁺ T cells to undergo homeostatic proliferation upon transfer into syngeneic Rag1^−/− recipients. These experiments revealed that dLck-Cre⁺ Ppard^fl/fl CD4⁺ T cells exhibited a mild defect in the ability to undergo >2 cell divisions (Fig. 10E, 10F). Despite these T cell growth defects, dLck-Cre⁺ Ppard^fl/fl mice exhibited no differences in the weight or cellularity of the thymus or spleen (Fig. 10G) or in the frequency or number of the major thymocyte or spleen lymphocyte subsets (Fig. 10H–K) as compared with Ppard^fl/fl mice. Taken together, these findings confirm a T cell–intrinsic role for PPAR-δ in regulating CD4⁺ T cell growth and point to developmental defects as the cause of peripheral T cell lymphopenia in PPARδ^mut mice.

During their proliferation, T cells and thymocytes switch from a “quiescent” metabolism that is characterized by oxidation of fats and glucose to a “proliferating” metabolism that is characterized by large increases in glucose and amino acid uptake, increased glycolytic flux and lactate production, and modest increases in glutamine oxidation (41–43). The increased flux of pyruvate and glutamine into the TCA cycle increases ATP production and, more importantly, provides the macromolecule substrates to promote biomass accumulation to support repeated cell division (41). To date, the PI3K signaling pathway, LKB1 activity, and transcription factors such as Myc have been shown to be critical in signaling some of these metabolic changes to support thymocyte and T cell growth (41, 42, 44). In this article, we report that the lipid sensor PPAR-δ is an additional regulator of this process. We observed in a variety of genetic, in vitro, and in vivo systems that PPAR-δ–deficient DN3/DN4 thymocytes and CD4⁺ T cells exhibited a reduced ability to undergo repeated cell division. The growth defects in these cell types associated most strongly with decreases in basal glycolytic activity and mitochondrial reserve, which correlated with decreased expression of a host of genes involved in aerobic glycolysis, OXPHOS, and mitochondrial and lipid biogenesis. These findings, therefore, implicate PPAR-δ as an additional regulator of the metabolic program that supports both the proliferative burst that occurs in thymocytes following the β-selection checkpoint and the growth of peripheral CD4⁺ T cells under conditions of activation.

Increases in glycolytic flux have been shown to be critical in supporting the growth and survival of rapidly dividing cells. This is because the diversion of glycolytic intermediates into the pentose phosphate shunt and into the synthesis of nonessential amino acids provides building blocks for nucleotide, tRNA, and ribosome synthesis (45). The process of lactate secretion itself is important for the growth of rapidly dividing cells because it creates an important driving stimulus for glutamine entry into the TCA cycle (and catabolization to lactate) and NADPH production, which are critical for supporting lipid and membrane biosynthesis (41, 46). Therefore, our observation that PPARδ^mut thymocytes and activated peripheral CD4⁺ T cells exhibited reductions in ECAR provided one explanation for the defective growth (and survival) of these cell populations.

Our gene profiling analysis identified a number of genes that were present at reduced levels in PPARδ^mut thymocytes and peripheral PPARδ^mut CD4⁺ T cells that could explain the decreased lactate flux by these cells. We observed decreases in the expression of the two most-distal enzymes in glycolysis (Pkm2, Eno1), an alternative glucokinase (Adpgk), and the gate keeper for pyruvate entry into the TCA cycle, Pdk1. Decreased Pkm2 and Eno1 expression have the potential to hinder cell growth by reducing the flux of carbon into the TCA cycle, which acts as a hub for macromolecule biosynthesis (41), or by slowing lactate extrusion (41). Evidence linking these two enzymes to cell growth is that Eno1 and Pkm2 are rapidly upregulated in CD4⁺ T cells upon activation (44), and inhibition of these enzymes is reported to slow tumor growth (47, 48). The finding that PPAR-δ regulated the expression of the distal-most enzymes in glycolysis may also explain why defects in DNA synthesis and survival were not observed in PPARδ^mut thymocytes, which exhibited milder decreases in ECAR. Decreasing the activity of Pkm2 and Eno1 can lead to the accumulation of glycolytic intermediates upstream, which can be diverted to nucleotide biosynthesis via the pentose phosphate shunt, thus acting to preserve nucleoside biosynthesis (47). In this way, the control of glycolysis by PPAR-δ contrasts with the PI3K pathway, which instead regulates proximal events in the glycolytic cascade, such as glucose transport and hexokinase activity, and exerts a comparatively more profound impact on DNA synthesis and cell survival (15, 49).

Beyond these core glycolytic enzymes, we also observed reduced mRNA levels of ADP-dependent glucokinase (Adpdk) in PPARδ^mut thymocytes and peripheral CD4⁺ T cells. Although the functioning of this enzyme is not completely understood, studies have resolved that it acts as an accessory glucokinase, phosphorylating glucose to generate glucose-1-phosphate, using ADP rather than ATP as a phoshoryl donor (50). Reports of human tumor cell lines of which the knockdown of Adpgk does not impact total glucose phosphorylation, glycolytic flux, or cell growth (50) have led some to conclude that this enzyme plays a subordinate role to hexokinases in glycolytic processes leading to macromolecule biosynthesis (50). However, another study in human T cells reported that ADP-dependent glucokinase activity is increased with anti-CD3 and PMA stimulation and that this activity results in increased glucose uptake and mitochondrial reactive oxygen species production in T cells (51). Further studies are necessary to elucidate whether this enzyme or other glycolytic enzymes played a role in the glycolytic or growth defects seen in PPARδ^mut cells.

In addition to key glycolytic enzymes, PPARδ^mut thymocytes and CD4⁺ T cells exhibited reduced mRNA expression of Pdk1, which would have the effect of enhancing the conversion of pyruvate into acetyl-CoA rather than lactate. At first glance this adaptation would appear to be counterintuitive to cell growth because it would enhance the TCA; however, decreased Pdk1 would also limit the availability of pyruvate for conversion into lactate and therefore limit associated growth pathways such as NADPH production. Evidence that reduced Pdk1 expression impacted lactate production was our observation that basal ECAR was lowered in PPARδ^mut CD4⁺ T cells relative to WT, even under conditions in which these cells were cultured in the presence of pyruvate but no added glucose (which isolates the influence of PPAR-δ deficiency on pyruvate to lactate conversion from the influence of PPAR-δ on glycolysis leading to lactate production). Proof that reduced Pdk1 activity can inhibit thymocyte and T cell growth is provided by the findings that treatment of rodents with the Pdk1 inhibitor, dichloroacetate, induces thymic atrophy in rats (52) and inhibits proinflammatory T cell processes in mice (reviewed in Ref. 53).

Beyond glycolytic deficiencies, we observed that PPARδ^mut thymocytes and activated CD4⁺ T cells exhibited subtle decreases in basal respiration and modest decreases in mitochondrial reserve. Our gene profiling studies identified that decreases in mitochondrial capacity correlated with decreases in the mRNA expression of the TCA cycle intermediate isocitrate dehydrogenase 3 (Idh3) and a host of ETC-associated genes, the most notable being Ndufb10, which plays a role in the biogenesis of ETC complex I (54). This finding that basal respiration was relatively preserved in both PPARδ^mut thymocytes and peripheral CD4⁺ T cells, despite decreases in mitochondrial gene expression, suggested that mitochondria in these cells were able sustain OXPHOS to meet the energetic demands of proliferation in the culture systems that we used. It remains unclear, however, whether the mitochondrial defects imposed by PPAR-δ deficiency contributed to the hindered growth or survival of thymocytes and CD4⁺ T cells. Our finding of more severe growth defects in PPARδ^mut CD4⁺ T cells when these cells were forced to grow on galactose as a substrate did at least provide evidence that the mitochondrial defects in the PPARδ^mut CD4⁺ T cells had the potential to impact T cell growth.

Our data did reveal that peripheral PPARδ^mut CD4⁺ T cells, which exhibited more profound decreases in mitochondrial reserve, were engaging compensatory mechanisms such as increasing mitochondrial content to maintain OXPHOS. This increase in mitochondrial content in PPARδ^mut CD4⁺ T cells correlated with increased mRNA expression of a number of mitochondrial biogenesis factors, a mitochondrial fission factor, and several ETC-related genes. Interestingly, this functional adaptation in PPARδ^mut CD4⁺ T cells mirrors the phenotype reported for T cells that are deficient in the ETC assembly factor, apoptosis-inducing factor (AIF). AIF^−/− T cells exhibit marked deficiencies in protein components of complex I, III, and IV of the ETC that lead to modest decreases in OXPHOS and a compensatory upregulation of mitochondrial content and TCA cycle activity (55). Interestingly, despite these T cell ETC deficits, mice with T cell–restricted AIF deficiency have normal thymocyte development, and CD4⁺ T cells from these mice exhibit growth deficits only when forced to grow in the presence of galactose (55). These observations of AIF^−/− CD4⁺ T cells, therefore, argue against ETC deficiencies being the major driver of thymocyte and T cell growth defects observed in PPARδ^mut mice.

Beyond the increase in mitochondrial content, PPARδ^mut CD4⁺ T cells also exhibited increased mRNA expression of genes that have known roles in mediating cell adaptations to hypoxia and oxidative stress, the most notable being a profound upregulation of mRNAs encoding the α- and β-chains of the oxygen-sensing molecule hemoglobin (by 19-fold in activated PPARδ^mut CD4⁺ T cells and 7-fold in PPARδ^mut thymocytes). Our finding of hemoglobin gene expression in T cells coincides with reports of a variety of other nonerythroid types (e.g., macrophages, dopaminergic neurons, vaginal epithelial cells, mesangial cells, etc.) (56–59). These past studies also provided genetic evidence that hemoglobin functions in nonerythroid cells to maximize OXPHOS by increasing the expression of genes involved in the ETC and in mediating protection from oxidative and nitrosative stress (56, 57, 59). Therefore, the upregulated expression of hemoglobin in PPARδ^mut thymocytes and CD4⁺ T cells was likely an adaptation by the cells to maintain OXPHOS or decrease mitochondrial stress.

A limitation of our study is that we did not explore how decreases in the expression of individual glycolytic or mitochondrial genes impacted thymocyte and peripheral CD4⁺ T cell metabolism or growth. This could have been further explored through the approach of gene knockdown or the use of specific enzyme inhibitors. We did not pursue these approaches because our microarray study identified that many metabolic genes were affected in PPARδ^mut CD4⁺ T cells, the majority of which with only modest changes in gene expression (by 20–35%). Such a result indicated that it was likely the concerted action of these gene deficiencies were contributing to metabolic phenotype observed in the PPARδ^mut thymocytes and CD4⁺ T cells.

Our gene profiling study also provided novel insights into the metabolic program of gene expression regulated by PPAR-δ in CD4⁺ T cells. Beyond the expression of key genes in glycolysis, the TCA, and ETC, endogenous PPAR-δ activity was important for maintaining the basal expression of genes involved in mitochondrial and peroxisomal biogenesis and in the generation of lipogenic precursors, which would also be supportive of cell growth and division. Our findings are consistent with the previous reported metabolic functions of PPAR-δ in skeletal muscle (60, 61), liver (62), heart (63), and macrophages (64). Interestingly, although PPAR-δ is most commonly described to promote the expression of genes involved in mitochondrial β-oxidation, we surprisingly found that only one gene in this pathway (Acadl) was modulated by PPAR-δ deficiency, and it was expressed at higher levels in PPARδ^mut compared with WT T cells. In this regard, our findings contrast with a report that showed the PPAR-δ agonist GW0742 to increase the expression of genes involved in fatty acid oxidation (Acca2, Acadvl, Cpt1a) in T cells (35). The fact that endogenous and synthetic PPAR-δ ligands have very different activities on the same metabolic genes in the same cell type is surprising, but it does have precedent. For example, the most comprehensive study to date of genes regulated by PPAR-δ in murine keratinocytes identified very little overlap between the genes regulated by endogenous PPAR-δ ligands (as read out by the comparison of WT and PPARδ^mut cells) and those induced by the high-affinity agonist GW0742 (24). It was speculated that the ability of the synthetic agonists to activate different sets of genes relates to the superior affinities of synthetic agonists for PPAR-δ, which may evoke conformational changes in the protein that lead to comparatively higher recruitment of coactivator proteins or enhanced dissociation of corepressors as compared with that elicited by endogenous ligands (24). Together, these findings highlight that the synthetic agonists designed to target this molecule may be having very different activities than endogenous fatty acid ligands present in the host or in cell culture media.

These discordant effects of endogenous ligands and synthetic agonists on PPAR-δ activity also reconcile the past-published observation that treatment of mice with the PPAR-δ agonist GW0742 induces thymus atrophy in mice (35). This study found that GW0742 activity hindered thymocyte growth by promoting fatty acid over glucose oxidation (65). Interestingly, this same study also reported that overexpression of PPAR-δ in T cells by using the pLck-Cre transgene to drive the expression of a CAG promoter-loxP-STOP-loxP-Ppard construct also compromised thymic cellularity in mice starting at the DN4 stage (35). Similar to findings with the agonist, the thymic atrophy associated with PPAR-δ overexpression was also accompanied by higher expressions of fatty acid oxidation genes and decreased DNA synthesis in DN4 thymocytes (35). It is unclear why higher expression of PPAR-δ would have the same effect as the synthetic ligand, but a different effect than endogenous PPAR-δ, because in the overexpression system, PPAR-δ activity would also be supported by endogenous ligands. One possible explanation is that this overexpression strategy used the very strong CAG promoter to drive Ppard expression, which is known to drive gene expression at very high levels (∼20-fold higher than normal) (66). Potentially, this higher expression of Ppard had the same effect as GW0742 of increasing promoter occupancy at genes that are more difficult to activate, such as those involved in fatty acid oxidation.

Our finding that spleen cellularity was preserved in distal Lck-cre⁺ Ppard^fl/fl mice, which also exhibited defects in CD4⁺ T cell proliferation, pointed to developmental defects as the cause of T cell lymphopenia in PPARδ^mut mice. However, our data do not rule out the possibility that T cell growth and survival deficits contributed to the peripheral lymphopenia observed in PPARδ^mut mice. This is because the extent of Ppard knockdown and resultant proliferation defects in distal Lck-cre⁺ Ppard^fl/fl mice were not as profound as seen in PPARδ^mut mice, which express an unstable null transcript. We experienced a similar difficulty interpreting the importance of PPAR-δ in thymocyte and peripheral lymphopenia using the proximal Lck-Cre⁺ Ppard^fl/fl system that produced only a 50% reduction in Ppard at the DN4 stage and only mild decreases in gene expression and no defects in T cell growth in the periphery (data not shown). Regardless of the limitations of these genetic systems, these experiments did provide support for a T cell–intrinsic role for PPAR-δ in thymocyte and peripheral CD4⁺ T cell growth. Interestingly, we observed PPARδ^mut mice to also exhibit a decrease in B cell numbers, which provided a clue that the developmental defects in this mouse extend beyond thymocyte development. In this regard, it has been reported that PPAR-δ plays a role in the maintenance of hematopoietic stem cells in the bone marrow (67). Our finding that PPARδ^mut mice showed normal numbers of ETP suggested that bone marrow defects did not impact the accumulation or survival of the earliest progenitor thymocytes. Therefore, more work will be required to further understand the basis of the peripheral lymphopenia in PPARδ^mut mice.

In conclusion, our study identifies PPAR-δ as an additional regulator of the metabolic program that supports the growth of TCRβ-selected thymocytes and peripheral CD4⁺ T cells and highlights that endogenous PPAR-δ activity is required for healthy thymic function and T cell output.

We thank Carl Virtanen (Princess Margaret Genomics Centre) for help with analysis of microarray data and staff at the SickKids–University Health Network Flow and Mass Cytometry Facility for help with cell sorting. We thank Dr. Pam Ohashi (University Health Network and University of Toronto) and Dr. Cynthia Guidos (SickKids and University of Toronto) for intellectual input in the project and Dr. Sylvie Lesage (Centre de Recherche Hôpital Maisonneuve–Rosemont) for critical review of the manuscript.

The authors have no financial conflicts of interest.