Lineage-specific metabolic properties and vulnerabilities of T cells in the demyelinating central nervous system

Abstract

Multiple Sclerosis (MS) is a disease characterized by immune-mediated destruction of central nervous system (CNS) myelin. Current MS therapies aim to block peripheral immune cells from entering the CNS. While these treatments limit new inflammatory activity in the CNS, no treatment effectively prevents long-term disease progression and disability accumulation in MS patients. One explanation for this paradox is that current therapies are ineffective at targeting immune responses already present in the CNS. To this end, we sought to understand the metabolic properties of T cells that mediate ongoing inflammation in the demyelinating CNS. Using experimental autoimmune encephalomyelitis (EAE) in C57BL/6 mice, a well-studied model of MS, we showed that the CD4⁺ and CD8⁺ T cells that invade the EAE CNS are highly glycolytic. This elevation of glycolysis is mediated by upregulated expression of the glycolytic machinery, and is essential for inflammatory responses to myelin. Surprisingly, we found that an inhibitor of glyceraldehyde-3 phosphate (GAPDH), 3-bromopyruvic acid (3-BrPa), blocks IFN-γ but not IL-17A production in immune cells isolated from the EAE CNS. Indeed, in vitro studies confirmed that the production of IFN-γ by differentiated Th1 cells is more sensitive to 3-BrPa than the production of IL-17A by Th17 cells. Finally, in transfer models of EAE, 3-BrPa robustly attenuates the encephalitogenic potential of EAE-driving immune cells. These data are one of the first to demonstrate the metabolic properties of T cells in the demyelinating CNS in vivo.

INTRODUCTION

Multiple Sclerosis (MS) is a devastating disease of the central nervous system (CNS) with high prevalence, early age of onset, and no cure (1, 2). MS is a disease with prominent autoimmune features that results in the debilitating destruction of CNS myelin, or demyelination (3).

There is a critical need for novel strategies to treat MS. Nearly all therapies that are FDA approved to treat MS target the immune system, with the most effective being those that prevent peripheral immune cell infiltration into the CNS. While this therapeutic strategy robustly diminishes relapse rates, it paradoxically fails to prevent MS disease progression (4, 5). A better understanding of the mechanisms regulating in situ inflammation in the MS CNS may reveal new targets for therapeutic intervention. There are no treatments specifically designed to target, self-propagating immune responses in the MS CNS, and the development of such may have profound therapeutic benefit as a standalone or combinatory treatment.

A great body of evidence suggests that T cells, agents of the immune system that normally direct inflammatory responses against invading pathogens, play a critical role in MS pathogenesis (6, 7). In vitro studies have revealed that activated T cells require upregulated glycolytic metabolism to fuel inflammatory programs, leading to speculation that glycolytic blockade may block a wide range of inflammatory T cell functions in autoimmune disease (8–10). Related studies showing that glycolysis is required for the in vitro differentiation of T cells further support this hypothesis (11). More recent studies have begun to show that glycolysis is also detected in vivo during experimental autoimmune encephalomyelitis (EAE) (11–14). It is becoming increasingly clear that in vitro metabolic status of immune cells is not always recapitulated in vivo, thus necessitating a closer look at the metabolic pathways fueling autoimmunity in vivo (15).

In this study, we use experimental autoimmune encephalomyelitis (EAE), a murine model of CNS demyelination, to address a major questions in MS research: are the glycolytic properties of T cells in the demyelinating CNS unique from peripheral T cells, and how does this glycolysis contribute to their proposed role as orchestrators of demyelination. Currently, the metabolic state of T cells in the EAE spinal cord is still unclear. Furthermore, it remains unclear if distinct metabolic pathways fuel different inflammatory programs, and likewise, if intervention along all points of a metabolic pathway will produce the same effects. Using two different models in which encephalitogenic T cells are generated in vivo, active immunization-based EAE and spontaneous EAE in the 2D2 strain of mice, we identified the metabolic properties of T cells in the EAE CNS. Our studies implicate glucose and glutamine as primary metabolic substrates for inflammatory T cell responses during CNS demyelination in vivo. Given the therapeutic potential of targeting metabolic pathways to attenuate inflammatory disease, we characterized the effects of 3-bromopyruvic acid (3-BrPa), an inhibitor of GAPDH and the glycolytic cycle, on inflammatory responses in the EAE spinal cord. We found that 3-BrPa potently blocked IFN-γ production by immune cells from the EAE spinal cord, but was not as efficient at blocking IL-17A production. Subsequent studies confirmed that in vitro differentiated IFN-γ-producing Th1 cells are indeed more susceptible than IL-17A-producing Th17 cells to 3-BrPa. We likewise found that 3-BrPa was effective in ameliorating disease onset and severity in EAE induced by the adoptive transfer of Th1 cells. These data provide novel and important insight into the therapeutic potential of targeting T cell glycolysis to treat MS.

MATERIALS AND METHODS

Mice

All studies were performed with C57BL/6 mice (8–16 weeks) obtained from Jackson (Cat#000664). 2D2 were have been described previously (7) and were obtained from Jackson (#006912). All animal experiments complied with regulations of the Institutional Animal Care and Use Committee at University of Virginia.

Reagents

3-Bromopyruvic acid (3-BrPa) was purchased from Acros Organics (cat# AC325690250). 6-Diazo-5-oxo-L-norleucine (DON) was purchased from Cayman Chemical (cat#17580).

Experimental Autoimmune Encephalomyelitis

Active EAE was performed as described previously (16). Briefly, male or female (sex-matched within experiments) C57BL/6 mice were immunized subcutaneously with 50 μg of myelin oligodendrocyte glycoprotein peptide 35–55 (MOG_35–55) (CSBIO #CS0681) emulsified at a 1:1 ratio in Complete Freud’s Adjuvant (Sigma Aldrich #F5881). On days 0 and 2, 250 ng of pertussis toxin (List Biologicals #180) was administered intraperitoneally. Transfer EAE was performed based on previously described protocols (17, 18). Immune cells were isolated from spleens of 8–12 week old 2D2 male mice, and differentiated into Th1 lineage promoting conditions for 3 days as previously described (14). Cells were treated with either 10–20 μM 3-bromopyruvic acid or saline vehicle on day 0 and day 2 of skew. On day 3, cells (5×10⁶) were injected intraperitoneally into Rag1^−/− immunodeficient hosts, and the pertussis toxin was administered as for the active EAE. Disease severity was scored using a standardized scale: 0- no disease; 1- complete tail paralysis; 2- loss of hindlimb coordination on grate walk; 3- hindlimb weakness; 4-complete bilateral hindlimb paralysis; 5- moribund or death due to EAE.

T cell purifications from the EAE spinal cord and spleen

To determine the glycolytic profile of T cells isolated from the demyelinating CNS, active EAE was induced and T cells were harvested from the spinal cords and spleens of mice at peak disease (d.14–18). Harvesting T cells from the spinal cord was achieved first by digesting the tissue in HBSS supplemented with 2 mg/mL Collagenase IV (Worthington #LS004188) and 50 U/mL DNAse (Sigma Aldrich #10104159001). Debris generated during the digestion was removed by Percoll density gradient (GE Healthcare #17-0891-01). The resulting cell suspension was then stained with a PE-conjugated antibody against TCRβ (eBioscience #12-5961-83) and T cells were positively sorted with Miltenyi MACS anti-PE Microbeads (#130-048-801). Splenic T cells were isolated similarly after tissue homogenization. MACS purifications yielded ~90% purity. Results obtained with magnetically sorted cells were confirmed by analysis of TCRβ⁺CD4⁺ and TCRβ⁺CD8⁺ T cells sorted on the BD Biosciences Influx Fluorescence Activated Cell Sorting machine at the University of Virginia Flow Cytometry Core Facilities.

Transcriptional analysis of T cells isolated from the EAE CNS and spleen

Transcriptional analyses were performed by quantitative real-time PCR on the BioRad CFX384 Touch™ Real-Time PCR Detection System. Purified T cells were snap frozen at −80C. RNA was extracted using the Bioline Isolate II RNA Mini Kit (#BIO-52073) according to manufacturers instructions. RNA was converted into cDNA using the Bioline SensiFAST cDNA synthesis kit (BIO-65053). qRT-PCR (See Supplementary Table 1 for primer list) was performed using Bioline SensiFAST kits (BIO-98005, BIO-86005).

Seahorse Metabolic Flux Analysis

To assess the glycolytic capacity of T cells isolated from the EAE spinal cord and spleen, sorted T cells were subjected to a glycolytic stress test and simultaneously assessed for glycolytic activity using the Seahorse XFp. T cells were seeded at 1×10⁵/well in XF Base Medium Minimal DMEM (#103193-100) supplemented with 2mM L-glutamine (Gibco #25030-081) and subjected to the glycolytic stress test (Seahorse Bioscience #103017-100). All Seahorse experiments were performed according to manufacturer instructions.

Antigen recall assays in CNS tissues of mice with EAE

Cells were isolated from the EAE CNS as described above. 5×10⁵ cells were stimulated in 200 μL of complete RPMI with 50 μg/mL MOG_35–55 and 10 μM 3-BrPa or 1mM 6-Diazo-5-oxo-L-norleucine (DON) for 48 hours, and IFN-γ (BioLegend #517902 and #505704) and IL-17A (eBioscience #eBio17CK15A5 and #eBio17B7) secretion were measured by ELISA. Post-stimulation cell viability was determined by flow cytometry. In some cases, spinal cord meninges were separated from the spinal cord parenchyma prior to recall as described in the text.

Effects of 3-BrPa on T cell proliferation and cytokine production

To assess the effects of 3-BrPa on T cell proliferation, T cells were isolated from spleens and lymph nodes of C57BL/6 mice using the R&D CD3⁺ T cell enrichment Column (MTCC-10), stained with 20 μM Cell Proliferation Dye eFluor450 (eBioscience #65-0842-85), and stimulated to divide in complete RPMI with platebound α-CD3 (BioXcell #BE0001-1) (1 μg/mL) and soluble α-CD28 (BioXcell #BE0015-5) (10 μg/mL) antibodies for 72hrs. After 24 hours of stimulation, a time when T cells have engaged glycolysis to fuel activation (19) but have not yet divided, 10 μM 3-BrPa or saline vehicle was added to the cells. Division was monitored for an additional 48 hours and determined every 24 hours by flow cytometric analysis of efluor450 dye dilution. The same T cell isolation procedure used for the proliferation assay was used to determine how 3-BrPa affects T cell cytokine production. Enriched T cells were stimulated with α-CD3 and α-CD28 in the presence of 3 μg/ml Brefeldin A (eBioscience #00-4506-51) as well as 10 μM 3-BrPa or saline vehicle for 6hrs at 37°C, 5% CO₂ and assessed for viability and cytokine production by flow cytometry.

Flow Cytometry

Flow cytometric analyses were performed on a 10 color Beckman Coulter Gallios flow cytometer. Antibodies used were as follows: TCRβ (H57-597), CD4 (RM4-5), CD8 (53-6.7), CD19 (eBio1D3), CD45 (30-F11), CD11b (M1/70), MHC-II (M5/114.15.2), IL-2 (JES6-5H4), TNF-α (MP6-XT22), IFN-γ (XMG1.2), IL-17A (Biolegend, TC11-18H10.1), T-bet (eBio4B10), and Zombie Aqua Fixable Viability kit (BioLegend #423101). For intracellular cytokine staining, the eBioscience Intracellular Fixation & Permeabilization Buffer kit (#88-824-00) was used. For intranuclear staining, the eBioscience FoxP3/Transcription Factor Staining Buffer Set (#00-5523-00) was used. All antibodies for flow cytometry were purchased from eBioscience unless otherwise noted.

GAPDH and hexokinase assay

Single cell suspension were prepared from the lymph nodes of C57BL/6 mice and incubated with increasing concentrations of 3-BrPa for 30 min at 37°C. Cells were next lysed in the assay extraction buffer and the enzymatic activity of both enzymes was determined according to manufacturer’s instruction (Biovision, GAPDH activity kit # K680 and hexokinase activity kit # K789).

Immunohistochemistry and histology

Slides were deparaffinized using xylenes and an ethanol gradient. and stained with Luxol Fast Blue to assess demyelination. Adjacent sections were submitted to the UVA Biorepository and Tissue Research Facility for CD3 staining. An investigator blinded to the status of the groups performed histological analysis.

Statistics

All statistical analyses were performed with Prism 7 (GraphPad software). The results of the statistical tests are presented within each results section. Analyses involving two groups were performed using a two-tailed t-test.

RESULTS

T cell glycolytic activity and inflammation are potentiated in the EAE spinal cord

Because our primary interest is in understanding the fuel sources driving in situ inflammation in the demyelinating CNS, we explored the metabolic properties of T cells in the EAE spinal cord. To this end, we flow sorted CD4⁺ and CD8⁺ T cells from the CNS of mice at peak EAE and assessed their metabolic characteristics. Using Seahorse Extracellular Flux Analyses we determined that CD4⁺ T cells from the EAE spinal cord had elevated basal and maximal glycolytic rates in comparison to those isolated from the spleens of the same animals (Figure 1A and B). Basal mitochondrial respiration of spinal cord CD4⁺ T cells was also elevated (Figure 1B). Results from our transcriptional studies indicated that the elevated glycolytic rates in these spinal cord CD4⁺ T cells might be, in part, fueled by increased expression of glycolytic enzyme transcripts (Figure 1C). Basal expression of interferon gamma (IFN-γ), a Th1 cell-derived cytokine associated with EAE and MS pathology (20, 21), and T-bet, the transcription factor that drives differentiation of Th1 cells, was also increased in CD4⁺ T cells in the EAE spinal cord (Figure 1D), and correlated with increased IFN-γ and IL-2 production upon restimulation (Figure 1E). Transcriptional and Seahorse studies were confirmed using TCRβ⁺ cells magnetically sorted from the diseased spinal cord (Supplementary Figure 1). These results indicate that, during EAE, pathogenic T cells at the site of inflammation are more glycolytic than peripheral T cells in more quiescent tissues. This metabolic observation is likely to be tightly linked to the activation state of T cells in the EAE CNS, which as we found, are robust expressers of the glycolytically regulated cytokines IFN-γ and IL-2.

Figure 1. T cell glycolytic activity and inflammation are potentiated in the EAE CNS.

(A) Glycolysis stress test and extracellular acidification rate (ECAR) of TCRβ⁺ CD4⁺ T cells isolated from the spleens and spinal cords of mice at peak EAE; ***p<0.001 at all time points noted by Two Way ANOVA with Sidak Post Test. n=3 biological replicates per group. (B) Average glycolytic rates at baseline (basal) and after glucose stimulation (glucose) and basal oxygen consumption rate (OCR) from Seahorse in (A); **p<0.001 by Two Way Anova; *p<0.05 by paired Student’s T test. (C) Glycolysis diagram and qRT-PCR for mRNA expression of glucose transporter 1 (GLUT1), hexokinase 1/2 (HK1/2), glyceraldehyde 3-phosphate dehydrogenase (GAPDH), pyruvate kinase isoform m1/2 (PKM1/2), and lactate dehydrogenase A (LDHA) in TCRβ⁺CD4⁺ and TCRβ⁺CD8⁺ T cells flow sorted from the spleens and spinal cords of mice at peak EAE. Significance determined by ratio-paired Student’s t-test. RQ = relative quantity. Actin is used as the housekeeping reference. Each dot represents one mouse. (A–C) from N=1 experiment. (D) IFN-γ and T-bet expression in cells from (A–C). (E) MOG-induced IFN-γ and IL-2 production in splenic and spinal cord homogenates isolated from mice at peak EAE; gated on live, singlet, CD45⁺CD11b⁻ TCRβ⁺ CD4⁺; **p<0.01; ***p<0.001 by paired Student’s t-test. Representative of N=2 independent experiments.

Regulation of non-glycolytic metabolism in CD4⁺ T cells in the EAE CNS

To gain a more comprehensive understanding of the metabolic properties of T cells in the EAE spinal cord, we also explored the transcriptional regulation of non-glycolytic metabolism in these T cells (Figure 2A). We observed significant increases in glutaminase 2 (Gls2), glutamate dehydrogenase 1 (Glud1), lysosomal acid lipase (Lipa), and carnitine palmitoyltransferase 1a (CPT1a) in CD4⁺ T cells but not in CD8⁺ T cells (Figure 2B). Our data suggest that unlike their CD4⁺ counterparts, spinal cord CD8⁺ T cells may preferentially regulate glycolytic metabolism during EAE. Increased metabolism of glutamine and/or lipids and fatty acids may be related to the significant elevation in basal oxygen consumption rate (OCR) in T cells isolated from the EAE spinal cord (Figure 1A, Supplementary Figure 1). Previous groups have shown that during EAE, antigen-specific T cells leave lymphoid organs following immunization and are retained in the CNS (22–24). Our observations are likely a reflection of this CNS-proximal T cell activation and, as such, illustrate a profound divergence from what would be predicted based on in vitro observations, where activated T cells, in comparison to resting state T cells, are known to uniformly increase amino acid transporter expression while shutting down lipolytic pathways. Taken together, our data provide important insight into the metabolic basis of the site-specific perpetuation of the T cell response in the EAE spinal cord in vivo.

Figure 2. Regulation of non-glycolytic metabolism in T cells in the EAE spinal cord.

qRT-PCR for mRNA expression of ASC Amino Acid Transporter 2 (ASCT2), Amino acid transporter light chain, L-system (Slc7a5), glutaminase 1 and 2 (Gls1/Gls2), glutamate dehydrogenase 1 (Glud1), carnitine palmitoyltransferase 1a and 2 (CPT1a/CPT2), and lysosomal acid lipase (Lipa) in CD4⁺ and CD8⁺ T cells (CD45⁺CD11b⁻ TCRβ⁺ CD4/8⁺) isolated from spleen and spinal cord from mice at peak EAE. *p<0.05 by paired Student’s T test RQ = relative quantity. Actin is used as the housekeeping gene. Each dot represents one mouse.

Immune cells in the EAE spinal cord require glycolysis for inflammatory functions

Our data suggest that numerous substrates could fuel CD4⁺ T cell-mediated inflammation in the CNS (Figure 2). As a result, we hypothesized that T cells in the EAE spinal cord may maintain metabolic flexibility or, in other words, the capacity to adapt when discrete energy resources are limiting.

We sought to determine if glycolysis was essential for T cell effector functions in the EAE spinal cord. We first examined the effects of a glycolytic inhibitor, 3-Bromopyruvic Acid (3-BrPa) on T cell functions (25, 26). 3-BrPa has been described has an inhibitor for both GAPDH and hexokinase (27–30). Using a single cell suspension prepared from lymph nodes, we found that 3-BrPa was inhibiting GAPDH activity at a dose that was not affecting hexokinase activity (Figure 3A). We discovered that 3-BrPa is a potent suppressor of T cell proliferation (Figure 3B) without impacting cell viability at the end of the assay (Figure 3C). We also observed that the production of the cytokines, TNF-α and IL-2, was affected by 3-BrPa in a dose response manner (Figure 3D, left panel), independently of cell survival (Figure 3D, right panel). To understand if 3-BrPa could impact TCR signaling event, we examined activation of CD69 and the phosphorylation of ERK. 3-BrPa did not block TCR-mediated upregulation of CD69 expression (Figure 3E) or impact viability at this time point (Figure 3F). Similarly, the phosphorylation of mitogen activated protein kinase ERK was not affected by 3-BrPa (Figure 3G). Using the same assay, cell death and impaired TCR signaling were observed at a higher dose of 3-BrPa (100 μM) (data not shown).

Figure 3. 3-bromopyruvic acid is an inhibitor of T cell effector functions.

(A) Dose dependent inhibition of GAPDH and hexokinase activities in a lymph nodes single cell suspension by 3-BrPa. (B) Effect of 3-BrPa on T cell proliferation. Lymph node cells were stimulated with α-CD3 (1 μg/mL) and α-CD28 (10 μg/mL). At 24 hours, a time when glycolysis is engaged, but no cell divisions have occurred, 3-BrPa was added to a final concentration of 10 μM, and subsequent divisions were assessed after 48 additional hours. Proliferation of CD8⁺ T cells is shown. Gated on live, singlet CD8⁺ cells. (C) Effect of increasing concentrations of 3-BrPa on TNFα and IL-2 production in T cells stimulated with α-CD3 (1 μg/mL) and α-CD28 (2 μg/mL). Gated on live, CD11b⁻ CD19⁻ TCRβ⁺ CD4⁺ or CD8⁺. (D) Effect of 3-BrPa on CD69 induction on T cells after 6 hours of activation with α-CD3 (1 μg/mL) α-CD28 (2 μg/mL). Gated on live, singlet, CD11b⁻ B220⁻ CD4⁺ (E) Effect of 3-BrPa on phosphorylation of ERK in TCRβ⁺ CD4⁺ T cells stimulated with α-CD3 (1 μg/mL) and α-CD28 (2 μg/mL) for 60min. Gated on singlet, TCRβ⁺ CD4⁺. The same results were observed in CD8⁺ T cells

(A, E) N=2 experiments with n=1-2 mice per experiment; (B-D) representative of N=2–3 independent experiments is shown.

We next induced EAE in C57BL/6 mice and performed antigen recall assays on harvested spinal cord homogenates in the presence or absence of 3-BrPa to analyze the effects of glycolysis inhibition on cytokine production. Stimulation of EAE spinal cord homogenates with MOG_35–55 potently elicited production of IFN-γ, and this induction was profoundly sensitive to treatment with 3-BrPa (Figure 4A). Strikingly, production of IL-17A, another cytokine associated with autoimmune demyelination (31, 32), was largely refractory to 3-BrPa treatment (Figure 4A). There was no effect of 3-BrPa on the viability of antigen presenting cells, CD4⁺ T cells, or CD8⁺ T cells in these assays (data not shown). Unlike 3-BrPa, the glutaminase inhibitor 6-Diazo-5-oxo-L-norleucine (DON) potently inhibited both IFN-γ and IL-17A production, indicating that IL-17A production in this assay may be preferentially facilitated by glutamine metabolism (Figure 4B). In these recall assays, no cytokine production was detected in the absence of MOG stimulation (data not shown).

Figure 4. 3-BrPa is a potent inhibitor of IFN-γ but not IL-17A production in EAE spinal cords.

(A) Effect of 3-BrPa on MOG-induced IFN-γ and IL-17A production in spinal cord homogenates from mice at peak EAE; a representative of N=3 experiments with 3–5 mice is shown. Each dot is a mouse. (B) Effect of DON (1 mM) on IFN-γ and IL-17A production in spinal cord homogenates from mice at peak EAE; N=1 experiment with n=3 mice per group. (C) Flow cytometry of IFN-γ and IL-2 production by splenic CD4⁺ T cells from 2D2 mice with spontaneous EAE restimulated with MOG peptide. *p<0.05, **p<0.01 by One Way ANOVA with Tukey Post Test. N=1 experiment with n=2–3 mice per group. Each dot is a mouse. Gated on live, singlet, TCRβ⁺ CD4⁺ (D) Effects of MOG-induced IFN-γ and IL-17A production in meningeal and spinal cord parenchymal homogenates from 2D2 mice that developed spontaneous EAE; *p<0.05; ***p<0.001 by One Way ANOVA with Tukey Post Test; N=1 experiment with n=1 mouse

T cells that drive spontaneous EAE in 2D2 mice require glycolysis

Our results show that T cells require glycolysis for specific inflammatory functions in active, immunization-based EAE. To confirm these findings in a model of adjuvant-free EAE, we repeated these studies in 2D2 mice, a strain on the C57BL/6 background that is predisposed to developing spontaneous EAE. CD4⁺ T cells from 2D2 transgenic mice express a T cell receptor that recognizes MOG_35–55 (7). MOG_35–55 restimulation of splenocytes from 2D2 mice during EAE potently induced IL-2 and IFN-γ production by CD4⁺ T cells (Figure 4C). This pattern of cytokine secretion is not observed prior to the development of the spontaneous disease (data not shown). We further found that MOG-induced IFN-γ and IL-2 production can be potently inhibited with 3-BrPa (Figure 4C). This is in line with literature showing that production of both IFN-γ and IL-2 are regulated by glucose availability (33). In subsequent studies of spinal cord tissues isolated from 2D2 mice during EAE, we found that IFN-γ was profoundly sensitive to treatment with 3-BrPa, whereas IL-17A was only moderately affected (Figure 4D). These data suggest that, in two different mouse models of MS, IFN-γ production by T cells during active disease is inhibited by 3-BrPa, but 3-BrPa is a less effective modulator of IL-17A production.

In vitro-derived Th1 and Th17 cells are differentially responsive to 3-BrPa

Defining the properties of glycolysis inhibitors toward Th1 and Th17 cells is critical, especially in the case of MS, where patient stratification based on whether their disease is driven by Th1 or Th17 responses has been proposed as an important diagnostic criterion influencing subsequent treatment options (34). Our work with 3-BrPa suggests that glycolysis inhibition may manifest uniquely in different T cell types, a phenomenon that has been observed with other glycolytic inhibitors (14). To determine if 3-BrPa differentially affects Th1 versus Th17 responses, we generated Th1 and Th17 cells in vitro using established protocols (17) and, after terminal differentiation was achieved, restimulated these T cells with α-CD3 and α-CD28 in the presence or absence of 3-BrPa. We found that 3-BrPa blocked Th1 and Th17 production of IFN-γ and IL-17A as determined by ELISA and flow cytometry (Figure 5A and C). Furthermore, this inhibition correlated with reduced expression of the transcripts for these cytokines (Figure 5B). Interestingly, 3-BrPa treatment did not result in altered regulation of T-bet or RORγt transcript or protein levels over the course of our study (data not shown). Importantly, in these experiments, we also observed that 3-BrPa mediated inhibition of IL-17A production by Th17 cells was less effective than the inhibition of IFN-γ production by Th1 cells. These data support our earlier observations, and additionally provide further evidence for a direct and differential effect of 3-BrPa on CD4⁺ T cells of different lineages.

Figure 5. 3-BrPa is a more selective inhibitor of Th1 compared to Th17 cytokine production.

ELISA (A) and qRT-PCR (B) for IFN-γ and IL-17A production by in vitro derived Th1 and Th17 cells stimulated with α-CD3 (2 μg/mL) and α-CD28 (2 μg/mL) in the presence of 3-BrPa (10 μM) for 20 hours; (C) flow cytometry for IFN-γ and IL-17A production by in vitro derived Th1 and Th17 cells stimulated as above for 6 hours in the presence of Brefeldin A. Gated on live, singlet, TCRβ⁺, CD4⁺ (A–C) are N=1 experiment with n=1–2 mice per group

3-BrPa limits the pathogenicity of encephalitogenic Th1 cells

Having determined that T cells in the demyelinating CNS have an elevated glycolytic profile and that inflammatory responses are sensitive to 3-BrPa treatment, we wanted to determine how 3-BrPa would affect the engagement of encephalitogenic programs in immune cells. To do this, we used the adoptive transfer model of EAE in immunodeficient hosts (Rag1^−/−). We prepared immune cells from the spleen of 2D2 mice and skewed them to promote generation of IFN-γ producing Th1 cells. Cells were treated at day 0 and day 2 with 10 μM 3-BrPa or vehicle (saline) and transferred at day 3 in Rag1^−/− mice. This treatment paradigm with 3-BrPa did not affect T cell viability (Figure 6A), but was sufficient to significantly inhibit IFN-γ production (Figure 6B). Importantly, we did not observe a difference in T-Bet expression suggesting that Th1 differentiation was not affected (Figure 6C). Mice receiving saline-treated cells were uniformly moribund by day 20 post transfer (Figure 6D). Treatment with 3-BrPa reduced disease incidence, and those mice that did develop EAE experienced a delayed disease course (Figure 6D). We performed luxol fast blue staining at day 20 and found a decrease in demyelination in mice from the 3-BrPa group (Figure 6E). We noticed a significant spread in the 3-BrPa group that is probably linked to the decreased in disease incidence observed within this group (Figure 6D). Logically, we found a decrease in CD3 positive cells in the 3-BrPa group at all the spinal cords levels examined (Figure 6F). To test if 3-BrPa could be used to reduce EAE severity we administered 3-BrPa at disease onset. We discovered that systemic administration of 3-BrPa (10 mg/kg) at the onset of the EAE was associated with severe mortality and hypothermia (data not shown). These data collectively show that pharmacologic inhibition of encephalitogenic immune cells could ameliorate the onset and severity of multifocal demyelinating events, highlighting the potential for developing a means of continuously targeting immune cell glycolysis as a therapy for autoimmune demyelination, especially in the context of IFN-γ driven disease.

Figure 6. 3-BrPa attenuates the encephalitogenic programming of immune cells.

Splenocytes were isolated from 2D2 mice and maintained in Th1 promoting conditions in the presence of MOG_35–55 peptide (50 μg/ml) for 3 days. 3-BrPa or saline was added at day 0 and day 2. (A) Cell viability, (B) IFN-γ production and (C) T-bet expression were measured prior adaptive transfer at day 3. (D) Disease severity and incidence of EAE in Rag1^−/− mice (n=5 mice group, 2 independent experiments) ***p<0.001 by Two Way ANOVA with Sidak Post Test. Spinal cord sections of the saline and 3-BrPa group were stained with (E) Luxol Fast Blue and (F) CD3 (n=5 for each group). Percentage of demyelination and cells number were manually quantified as presented on the right. **p<0.01, ****p<0.0001 by One Way ANOVA with Sidak Post Test.

DISCUSSION

Blocking immune infiltration of the CNS is a common and effective strategy to lower relapse rate in MS. However, this approach fails to block ongoing responses in the CNS. Perhaps for this reason, all FDA-approved MS treatments lower relapse rates yet none prevent disease progression (4, 5). Understanding what fuels the immune response in the demyelinating CNS is, therefore, of critical importance in revealing the potential of metabolic based therapeutics that could prevent disease progression.

While many studies have shown energy derivation pathways are important regulators of immune responses, reliance on individual nutrient requirements in disease-specific contexts remains unclear. To date, publications on the metabolic status of T cells in the demyelinating CNS have established that in vitro-derived T cells require glycolysis to differentiate into a pro-inflammatory phenotype that can drive EAE (11–14), and that these T cells maintain some aspects of glycolytic character after re-isolation from the spinal cord in transfer models of EAE and ex vivo culture (14). These seminal works have led to the speculation that pharmacologically or genetically manipulating T cell glycolytic metabolism could lead to improved patient outcomes in MS. However, no study has actually determined the metabolic state of T cells isolated from sites of ongoing CNS demyelination. Given that the metabolic characteristics of T cells generated in vitro can be strikingly different from those generated in vivo (15), this knowledge-gap is significant.

Here, using multiple models of EAE, we provide a comprehensive metabolic characterization of T cells orchestrating CNS demyelination. T cells isolated from the CNS of mice with actively induced experimental autoimmune encephalomyelitis (EAE) have a robustly elevated glycolytic transcriptional and metabolic profile in both the CD4⁺ and CD8⁺ T cell compartments. Interestingly, CD4⁺ T cells seem to more strongly regulate their ability to consume non-glucose metabolic substrates, indicating that CD8⁺ T cells may be more vulnerable to glycolytic inhibition. We have also made similar observations through analysis of T cells isolated from the spinal cords of 2D2 mice that develop spontaneous EAE (unpublished observations). It is important to note that the goal of our study is not simply to compare the metabolic state of active versus inactive T cells generated in vivo, but rather, to describe the unique metabolic properties and vulnerabilities of T cells that are specifically found in the EAE spinal cord.

Aerobic glycolysis or “Warburg effect” is also a hallmark of cancer cell metabolism, and the pursuit of novel cancer therapeutics has identified a plethora of non-classical metabolic inhibitors (35, 36). However, in many cases, experience with these drugs in immunological systems is lacking. One such drug is 3-BrPa, a robust GAPDH inhibitor, currently being tested in preclinical cancer research (29, 37). We found that 3-BrPa is a potent inhibitor of T cell GAPDH, and unexpectedly, that this inhibition had lineage specific effects on T cell populations. Production of TNF-α and IL-2 by antigen-inexperienced T cells is robustly inhibited by 3-BrPa, as is IFN-γ production by Th1 cells. In contrast, IL-17A production by Th17 cells is only mildly affected by 3-BrPa. Given that MS patients can be stratified into those with IFN-γ versus IL-17A driven disease, and that treatment regimens may soon be tailored according to this stratification (34), our experiences with 3-BrPa may have important implications. The thorough evaluation of diverse anti-glycolytics, like 3-BrPa, and how they impact immune cells, will be critical for developing a diverse array of metabolically-focused MS therapeutics that can be rationally deployed to accommodate patient-specific heterogeneity in disease. Further studies will be needed to understand the metabolic status of in vivo-derived pathogenic Th17 cells and their reliance on GAPDH activity and glycolysis for effector functions.

It is important to note that 3-BrPa has been also reported to inhibit hexokinase II, another enzyme of the glycolysis pathway (28). However, at the concentration that was used for our study, we were unable to detect inhibition of hexokinase activity by 3-BrPa (Figure 3A). Furthermore, 3-BrPa has also been documented as an inhibitor of other metabolic enzymes involved in glycolysis and the TCA cycle, including 3-phosphoglycerate kinase, isocitrate dehydrogenase and α-ketoglutarate dehydrogenase (38). Finally, due to the chemical nature of the inhibitor, 3-BrPa has been documented as an alkylating agent able to target free thiol group (29). Therefore, while we hypothesize that 3-BrPa mediate its biological effect on lymphocytes by inhibiting GAPDH, it is very likely that 3-BrPa function is not limited to this enzyme. Further studies are needed to fully understand the scope of 3-BrPa impact on T-cell biology.

Previous work has shown that sustained glycolysis inhibition during MOG-reactive T cell differentiation in vitro can prevent disease by Th17 cells (11), but left unanswered the question of whether glycolytic inhibition could derail a Th1 cell-mediated encephalitogenesis. In adoptive transfer models of EAE, we have determined that 3-BrPa significantly attenuated the pathogenicity of Th1-driven EAE. Systemic administration of 3-BrPa carries the risk of toxicity, as does the administration of many other broad-spectrum glycolytic inhibitors. Indeed, we did observe severe side effects when we administered 3-BrPa systemically in EAE (mortality and hypothermia), suggesting that 3-BrPa would not be a proper choice of drug to treat MS patients. This result is in contrast with recent work demonstrating that 3-BrPa was well tolerated and beneficial in an animal model of arthritis (39), perhaps highlighting some uniqueness about the CNS. The parallel development of well-tolerated drugs that can specifically target immune cells metabolic demands could result in a wave of novel biologics to treat inflammatory disease.

Understanding what fuels tissue-specific autoimmune responses is a critical next step in the treatment of auto-inflammatory diseases. This study definitively implicates glucose utilization by T cells in CNS tissues as a key contributor to pathology in autoimmune CNS demyelinating processes like MS. Studies of T cell metabolic properties in the target organs of autoimmune processes as well as those of in vitro-derived T cells were instrumental in helping to guide this work (40). The results of our studies build upon these by identifying the metabolic characteristics of T cells in a microenvironment long known specifically for its unique modulation of T cell responses, the central nervous system, specifically in the context of autoimmune demyelination.