PDC deficiency in T cells: TPdh-/- mouse
To understand the effects of disruption of this critical metabolic node in T cells, we developed a model of T cell PDC deficiency by targeting Pdha using a CD4-cre recombinase. TPdh-/- mice display normal litter sizes, body weight and length and life span (data not shown). To confirm deletion of the Pdha locus, we performed qPCR on gDNA from splenic T cells. TPdh-/- T cells displayed an absence of Pdha (Figure 1A). Similar to humans, Pdha is encoded on the X-chromosome11. Therefore, to determine the efficacy of our cre-recombinase and the utility of male and female mice for experiments, we studied both sexes for the presence of PDHA by immunoblot (Figure 1B). PDHA was absent in both male and female mice, enabling us to use both sexes for subsequent experiments. Finally, we wanted to determine whether pyruvate could be oxidized by activated TPdh-/- cells. To answer this question, splenic T cells from WT and TPdh-/- mice were isolated and activated for 24 hours with CD3/CD28 stimulation. Extracellular flux analysis was performed where glucose was removed and replaced by pyruvate (Figure 1C). While WT cells readily oxidized pyruvate, TPdh-/- cells were impaired, consistent with a block at the level of PDC.
TPdh-/- cells are dependent upon glycolysis
PDC is an important gatekeeper in metabolism, linking glycolysis and the TCA cycle. Since inhibition of PDC activity by PDK promotes aerobic glycolysis in culture and in vivo12,13, we predicted a similar increase in T cells with PDC deficiency. To profile glycolysis, we performed extracellular flux analysis following glucose injection on activated WT and TPdh-/- splenic T cells (i.e., glycolytic stress test, Figure 2A). Following glucose injection, the extracellular acidification rate rose promptly with TPdh-/- cells peaking about 50 mpH/min higher than WT. The addition of oligomycin to quantify glycolytic reserve resulted in a minimal increase in TPdh-/- T cells, indicating that these cells were operating at their glycolytic maximum. To confirm increased utilization of glycolysis, we anticipated an accumulation of glycolytic intermediates. To identify these points of substrate accumulation, we conducted metabolomic analyses of glycolytic intermediates on activated T cells. While other glycolytic and pentose phosphate pathway intermediates were similar to WT (Figure S1A and S1B), TPdh-/- T cells displayed elevated levels of glucose-6-phosphate and fructose 1,6 bisphosphate, the products of two key regulatory enzymes of glycolysis, hexokinase and phosphofructokinase, respectively (Figure 2B). The accumulations observed at critical metabolic checkpoints are consistent with enhanced glucose metabolism14.
The transition from glycolysis to the TCA cycle occurs in the mitochondria via PDC to produce acetyl-CoA, a critical metabolite for the TCA cycle 15. To confirm the interruption of glycolytic carbon transfer into the TCA cycle, we next examined the incorporation of 13C carbon from [U-13C] glucose into TCA cycle intermediates (Figure 2C). In WT, approximately 40% of the citrate pool was labelled as M + 2, indicating a considerable glucose-derived contribution to citrate through PDC (Figure 2D). Consistent with the genetic ablation of Pdha, the M+ 2 isotopologues of TCA cycle intermediates citrate, fumarate, and malate were essentially absent, with unlabeled (M + 0) intermediates comprising the predominant isotopologue (Figure 2D and S1C). The same was true for M + 2 aspartate, an amino acid derived from oxaloacetate. Cycling of the TCA from glucose derived carbon was also significantly depressed as reflected by the M + 4 citrate isotopologue (Figure 2D, lower right). To define functional glucose dependence of activated T cells, we studied proliferation by incubating stimulated T cells with increasing concentrations of 2-deoxyglucose (Figure 2E). While WT displayed a dose dependent effect, TPdh-/- was found to have significant inhibition of proliferation at all doses of 2DG. Overall, our results not only confirm PDC deficiency, but also define a functional dependence of glycolysis on TPdh-/- cells.
Reprogramming of mitochondrial metabolism in TPdh-/-
The TCA cycle generates intermediates for sugars, amino acids, nucleic acids, and lipids, and provides reducing equivalents for oxidative phosphorylation16-18. Since the commitment of glucose derived carbon to the TCA cycle and aerobic metabolism was disrupted, we hypothesized that this would lead to metabolic adaptations19,20. To define these metabolic adaptations, we first profiled TCA cycle intermediates via metabolomics in CD3/CD28 activated T cells. Most TCA cycle intermediates were similar between TPdh-/- and WT, suggesting a potential role for anaplerosis (Figure S2A and S2B). Indeed, multiple anaplerotic amino acids were depressed in TPdh-/-, with phenylalanine, tyrosine, isoleucine and valine being the most significantly affected (Figure S2C). Notably, one TCA cycle intermediate was markedly reduced in our metabolomics study. Succinyl CoA, the product of a-ketoglutarate dehydrogenase and substrate for succinyl CoA synthetase21, was decreased along with a significant increase in GTP (Figure 3A). These findings suggest generation of GTP by substrate level phosphorylation via succinyl CoA synthetase22,23. GTP can subsequently be converted to ATP by nucleoside-diphosphate kinase (NDPK)24, thus contributing to the overall cellular ATP pool. To profile the incorporation of alternate carbon sources into the TCA cycle in TPdh-/-, we employed [U-13C] glutamine (Figure 3B). Glutamine is converted to glutamate and subsequently a-ketoglutarate, the substrate that generates succinyl-CoA and downstream metabolites succinate, fumarate and malate. The M + 5 isotopomer of glutamate was increased in TPdh-/- consistent with increased incorporation of 13C carbon (Figure 3C and S2D). Monitoring M + 4 isotopologues downstream showed enrichment of glutamine carbon in fumarate, malate, and aspartate (Figure 3C, Figure S2D).
Anaplerosis not only helps regulate rates of biosynthesis by augmenting substrate availability, but may also contribute to cellular energy status21. To test whether increased incorporation of glutamine carbon translated to enhanced OXPHOS in TPdh-/-, we next performed extracellular flux analysis on T cells activated as above. Increased incorporation into the TCA by glutamine did not result in enhanced OXPHOS, but rather a depression (Figure 3D), suggesting that this amino acid did not contribute to cellular energy status via OXPHOS. Based on our observations, we next sought to define OXPHOS by extracellular flux analysis. In activated TPdh-/- T cells, basal respiration, ATP synthesis, maximal respiration and spare respiratory capacity were all depressed (Figure 3E). Based on our stable isotope and extracellular flux analyses, we suggest that in activated T cells, a portion of glucose is completely oxidized in the mitochondria. In addition, this oxidation of glucose may help set the pace for OXPHOS. In PDC, a depression in OCR was also seen when the long chain fat palmitate was used as a substrate, indicating that mitochondrial fatty acid oxidation was also reduced (Figure S2E). Despite significant depressions in FAO and OXPHOS in TPdh-/- T cells, total cellular ATP was similar to WT, suggesting that substrate level phosphorylation was sufficient to account for the deficit (Figure 3F).
Deficiencies in T cell differentiation in TPdh-/- in vitro
T cells play multiple roles in the adaptive immune system. In addition to killing infected host cells, T cells activate and coordinate multiple arms of the innate and adaptive immune response to pathogens, allergens and tumors25. To study T cell expansion prior to differentiation, we stimulated cells for 72 hours and measured proliferation by cell trace violet (CTV) dilution (Figure 4A). TPdh-/- T cells displayed slightly compromised proliferation, with CD8+ cells showing a greater lag, consistent with their need for a more robust metabotype.
CD8+ T cells play a critical role in immunity, particularly viral infections. Following stimulation, CD8+ T cells rapidly proliferate and can differentiate into Ag-specific effector T (TE) cells or long-lived memory T (TM) cells that help protect against re-infection26. These differentiation states are dependent upon IL-2 and IL-15 stimulation, respectively9. To determine the impact of PDC deficiency on these cell types, we performed in vitro differentiation of TE (IL-2) and TM (IL-15) T cells (Figure S3A). TPdh-/- TE cells showed a retention of CD62L, a marker of TM (Figure 4B). Conversely, TPdh-/- TM cells showed depressed expression of Ly6C, indicating a breakdown in memory differentiation. In adaptive immunity to viruses, CD8+ TM cells comprise the memory pool, while TE cells control viral proliferation by killing infected cells. To define the TE phenotype, we performed cell killing assays using EL-4 cells. Interestingly, while TPdh-/- TE displayed reduced killing activity, TM also displayed killing activity (Figure 4C), likely due to the retention of granzyme B activity (Figure S3B), suggesting abnormal differentiation.
We next asked whether metabolites provided by the extracellular environment could overcome the effects of PDC deficiency in TM differentiation. Acetate, a ketone body produced during infectious states, has been shown to be involved in the acetylation of metabolic enzymes (e.g., GAPDH) and histones6,27. To test whether replacement of acetyl-CoA by acetate supplementation (10 mM) could aid in the differentiation of TPdh-/- TM cells, we performed in vitro differentiation with IL-15 as above. Since acetate alone did not produce changes in Ly6C expression (data not shown), we also employed a lactate dehydrogenase inhibitor (LDHi, 25 mM) to suppress Warburg metabolism and aid in the adoption of a TM metabotype. With the addition of acetate and LDHi, we saw a slight improvement in Ly6C, suggesting skewing towards the TM phenotype (Figure 4D). Interestingly, this improvement in TM skewing in TPdh-/- was not due to changes in the spare respiratory capacity of OXPHOS (Figure 4E), suggesting that the acetyl-CoA derived from acetate was involved in mechanisms of differentiation outside of bioenergetics.
TPdh-/- TM display altered epigenetic signatures
Based upon our TM studies, we hypothesized that TPdh-/- TM would display perturbations in gene expression due to aberrant epigenetic signatures, specifically, histone modification. Global histone acetylation levels are determined by glycolytic flux and PDC deficiency represents a major impediment to acetyl-CoA production28. To test this hypothesis, we began by characterizing expression signatures by RNAseq (Figure 5A). Compared to WT, TPdh-/- TM cells were found to have 16 downregulated genes and 414 upregulated genes (Log2 fold change >2, -Log10 P value > 1.3). Consistent with our functional assays, TPdh-/- TM cells displayed a divergent phenotype consistent with an inflammatory effector T cell phenotype. The top upregulated genes for TPdh-/- TM cells also included numerous granzymes and members of the killer-like receptor family, markers consistent with a TE phenotype (Figure 5A and 5B, top) 29,30. Similarly, upstream regulators lipopolysaccharide and interferon gamma were also consistent with an inflammatory effector T cell phenotype. We interpreted this gene expression profile as abnormal differentiation of TPdh-/- TM, with the cells retaining effector functions, an assertion which was supported by our cell killing assays (Figure 4D).
Since acetyl-CoA production is impacted by PDC deficiency, and is an essential component of gene regulation by histone modification, we next performed chromatin immunoprecipitation and sequencing (ChIPseq) studies by targeting acetylated histones. To visualize our genomic data, we constructed an MA plot (Figure S4A). In the figure, we found that WT TM have nearly a log fold difference (purple) in bound sites when compared to TPdh-/- TM. This translated into a generally lower number of genomic reads for the top 50 genes and at all loci in general (Figures 5C and 5D) for TPdh-/-. Although the number of acetylated sites was lower in TPdh-/-, the overall distribution of acetylated sites was similar between both groups (Figure S4B). To confirm our findings, we used ChIP PCR to probe several targets important for TM differentiation that were identified by our ChIPseq (Figure 5E). Consistent with our ChIPseq results, TPdh-/- TM displayed a decreased ratio of acetylated target genes by ChIP PCR. These results were also consistent with our ATACseq results that showed limitations in chromatin accessibility for the aforementioned genes (Figure 5F). Based on these findings, we hypothesized that histone acetylation would be altered. To answer this question, we conducted a proteomic analysis for histones lysine acetylation. In general, TPdh-/- TM displayed decreased amounts of histone acetylation (Figure 5G). Overall, our results indicate perturbations in the epigenetic signature of TPdh-/- TM, which impacts gene expression and by extension differentiation.
PDC deficiency alters the broader epigenetic landscape of histone modifications
Metabolic rewiring and epigenetic remodeling via histone modification are interconnected and reciprocally regulate each other, thereby impacting cellular phenotypes. In addition to acetylation, cellular metabolism also contributes to other histone modifications including lactylation, butyrylation, crotonylation, glutarylation, malonylation, succinylation, methylation, demethylation, and trimethylation; many of which play a role in modifying gene expression profiles 10,31,32. Based on the metabolic rewiring seen in TPdh-/- cells, we hypothesized that PDC deficiency would lead to broader perturbations in the histone code, beyond acetylation. To answer this question, we employed a histone proteomics approach. In general, TPdh-/- memory T cells were not only deficient in acetylation, as expected, but were also deficient in nearly every type of histone modification studied. The overall fold change, and histone subtypes (Figure 6), were markedly different from WT. To investigate the response to cellular and environmental changes, we employed our acetate and LDHi treatment (Rx) as earlier. In general, there was a decrease in histone modifications seen in both WT and TPdh-/- (Figures 6). In WT, the acetylation profiles remained mostly similar, with some replacements (e.g., acetylation for hydroxybutyrylation and lactylation at H2BK117 and HBK86, respectively). Following treatment, TPdh-/- showed attrition of multiple histone modifications, especially on histones H2A and H2B, including a loss of acetylated sites. Overall, these results indicate that PDC deficiency leads to an alteration of the histone code during differentiation in response to intracellular and extracellular sources, explaining the incongruent phenotype seen in TPdh-/- TM.