Research ArticleMultiple Sclerosis

Teriflunomide treatment for multiple sclerosis modulates T cell mitochondrial respiration with affinity-dependent effects

See allHide authors and affiliations

Science Translational Medicine  01 May 2019:
Vol. 11, Issue 490, eaao5563
DOI: 10.1126/scitranslmed.aao5563

Teriflunomide tampers with T cells

Activated T cells need de novo pyrimidine biosynthesis, which has been exploited for autoimmune therapy. Klotz et al. studied samples from patients with multiple sclerosis treated with the dihydroorotate dehydrogenase inhibitor teriflunomide. They found that teriflunomide did not affect all T cells equally and led to repertoire and subset distribution changes. They also used a mouse model of T cells with higher and lower affinity for the same antigen to explore the drug’s effects. Collectively, their data demonstrate that high-affinity T cells preferentially use mitochondrial respiration, which is then inhibited by teriflunomide.

Abstract

Interference with immune cell proliferation represents a successful treatment strategy in T cell–mediated autoimmune diseases such as rheumatoid arthritis and multiple sclerosis (MS). One prominent example is pharmacological inhibition of dihydroorotate dehydrogenase (DHODH), which mediates de novo pyrimidine synthesis in actively proliferating T and B lymphocytes. Within the TERIDYNAMIC clinical study, we observed that the DHODH inhibitor teriflunomide caused selective changes in T cell subset composition and T cell receptor repertoire diversity in patients with relapsing-remitting MS (RRMS). In a preclinical antigen-specific setup, DHODH inhibition preferentially suppressed the proliferation of high-affinity T cells. Mechanistically, DHODH inhibition interferes with oxidative phosphorylation (OXPHOS) and aerobic glycolysis in activated T cells via functional inhibition of complex III of the respiratory chain. The affinity-dependent effects of DHODH inhibition were closely linked to differences in T cell metabolism. High-affinity T cells preferentially use OXPHOS during early activation, which explains their increased susceptibility toward DHODH inhibition. In a mouse model of MS, DHODH inhibitory treatment resulted in preferential inhibition of high-affinity autoreactive T cell clones. Compared to T cells from healthy controls, T cells from patients with RRMS exhibited increased OXPHOS and glycolysis, which were reduced with teriflunomide treatment. Together, these data point to a mechanism of action where DHODH inhibition corrects metabolic disturbances in T cells, which primarily affects profoundly metabolically active high-affinity T cell clones. Hence, DHODH inhibition may promote recovery of an altered T cell receptor repertoire in autoimmunity.

INTRODUCTION

As documented in several successful clinical trials, inhibition of de novo pyrimidine synthesis, which is crucial for rapid expansion of activated lymphocytes, is a well-established strategy for treatment of T cell–mediated autoimmune diseases such as rheumatoid arthritis (RA) and multiple sclerosis (MS) (1, 2). Leflunomide, along with its active metabolite teriflunomide, is a well-known inhibitor of the mitochondrial enzyme dihydroorotate dehydrogenase (DHODH), the fourth enzyme in the de novo pyrimidine biosynthesis. Both preclinical data and datasets from clinical trials illustrate the impact of pharmacological DHODH inhibition on lymphocyte proliferation and expansion (1, 3). As illustrated by a recently published placebo-controlled trial (4), it is still poorly understood how these drugs exert a selective effect on autoreactive T cells while only slightly affecting immune responses against bacteria and viruses. Although antibody titers for de novo immune responses were slightly impaired in this study, they were sufficient for seroprotection, whereas cellular memory responses to recall antigens were not affected.

It has been demonstrated that development of autoimmunity can be driven by avidity maturation of prevailing autoantigen-specific T cell populations. Selective depletion of high-affinity T cell clones can prevent development of organ-specific autoimmunity, e.g., in an animal model of autoimmune diabetes. Likewise, an increase in antigen affinities has been implicated in disease progression in different models of T cell–mediated autoimmunity (5, 6). Also, in an animal model of MS, transgenic mice bearing a higher-affinity autoantigen-specific T cell receptor (TCR) exhibit a substantially higher disease incidence than mice with a low-affinity TCR for the same antigen, demonstrating that high-affinity T cells bear high pathogenicity also in central nervous system (CNS) autoimmunity. On the basis of the SKG mouse for spontaneous arthritis, it has also been proposed that, in RA, high-avidity autoreactive T cells might be central to disease pathogenesis, although this has not been formally proven (7).

The process of lymphocyte activation, expansion, and acquisition of effector functions is unique with regard to the specific bioenergetic and biosynthetic needs of these cells. In recent years, several studies have elucidated the metabolic properties and requirements of distinct lymphocyte populations under different conditions. Briefly, resting T cells primarily use oxidative phosphorylation (OXPHOS) and the breakdown of fatty acids via the tricarboxylic acid cycle to supply energy (8). Upon activation, they rapidly switch to aerobic glycolysis to ensure energy supply and generate macromolecules and “building blocks” to enable cell growth and expansion. However, despite these general principles, there are fundamental differences in the metabolic profile of distinct lymphocyte populations depending on their activation state. For example, naïve T cells depend on the combined up-regulation of OXPHOS and aerobic glycolysis for initiation of T cell proliferation, whereas effector T cells mainly depend on glycolysis for fulfillment of effector functions (9). Moreover, activated memory T cells display an increased capacity for OXPHOS in comparison to freshly activated T cells, which is the basis for their bioenergetic advantage over naïve T cells and explains their increased expansion kinetics (10). These insights into the distinct bioenergetic profiles of T cells gave rise to the concept of immune metabolism as a therapeutic target, allowing a more selective interference with distinct immune cell subsets or activation states.

Here, we demonstrate that teriflunomide treatment of patients with relapsing-remitting MS (RRMS) resulted in a reduction in TCR repertoire diversity due to depletion of individual T cell clones. Mechanistically, teriflunomide-mediated inhibitory effects on T cell proliferation depended on OXPHOS inhibition, and their extent was closely linked to antigen affinity, because high-affinity T cells exhibited a greater dependence on OXPHOS than low-affinity T cells. Hence, our study demonstrates that DHODH inhibition exerts specific effects on distinct T cell clones based on their individual metabolic profiles, thus providing insights into the mechanisms underlying the selective immune interference in the context of autoimmune diseases while preserving protective immunity against pathogens.

RESULTS

DHODH inhibition causes distinct alterations in T cell subsets and the TCR repertoire of patients with RRMS

So far, the in vivo effects of teriflunomide on T cell subsets in patients with RRMS have not been fully elucidated. To address this question, we investigated the CD4+ T cell subset composition in patients with RRMS before and during teriflunomide treatment by multicolor flow cytometry as part of the TERIDYNAMIC clinical trial (NCT01863888). Baseline data from 50 patients are depicted in table S1. Because the known antiproliferative effects of teriflunomide treatment include a mild but consistent lymphopenia (Table 1) (1), we assumed that teriflunomide treatment would result in a uniform reduction in T cell subset counts in the peripheral blood of patients. Unexpectedly, we observed distinct effects of teriflunomide treatment on different T cell subsets with an absolute reduction in T helper 1 (TH1) cells but not in TH2 or TH17 cells (Fig. 1A). Although absolute numbers of regulatory T cells (Tregs) and of the subset of inducible Tregs (iTregs) remained unaffected (Fig. 1A and fig. S1A), we observed a selective increase in the proportion of iTregs under teriflunomide (Fig. 1B and fig. S1B). This relative increase in iTregs was even more pronounced in a subgroup of previously untreated patients with RRMS receiving teriflunomide (fig. S1, C and D). This differential effect of teriflunomide resulted in an increased ratio of iTregs/TH1, whereas the ratio of iTregs/TH17 was not modified (fig. S1E). With regard to markers of Treg functionality, the expression of CD39 and CTLA-4 remained unchanged under teriflunomide treatment (fig. S1, F and G). Accordingly, the suppressive capacity of Tregs derived from teriflunomide-treated patients in an autologous suppression assay was not altered when compared to their respective baseline levels (fig. S1H). Likewise, the cytokine profile of Tregs from teriflunomide-treated patients was unaffected (fig. S1I).

Table 1 Evolution and change from baseline in the percentages and absolute counts of overall lymphocyte populations and CD4+T cell subsets in PBMCs.

Per-protocol predefined population, people with RRMS treated with 14 mg of teriflunomide. Peripheral blood mononuclear cells (PBMCs) isolated and analyzed by flow cytometry at baseline and 6 months. Data are represented as mean (SD) or median (IQR). Change from baseline represented as least squares means (LSM) change (SEM) or median (IQR). P values from linear mixed model analysis of month 6 with baseline. N/A, not applicable

View this table:
Fig. 1 CD4+and CD8+T cell changes in teriflunomide-treated patients with RRMS from the TERIDYNAMIC study.

(A and B) CD4+ T cell subpopulations in patients with RRMS after 3 months (3M) and 6 months (6M) of teriflunomide (TF) treatment from the TERIDYNAMIC clinical study (table S1). Box plots represent the interquartile range (IQR) with the horizontal line indicating the median and error bars showing maximum and minimum values. P values were calculated from linear mixed model on change from baseline. (A) Absolute cell numbers of TH1 (n = 32), TH2 (n = 32), TH17 (n = 32), and total Tregs (n = 37). (B) Frequencies of iTregs (n = 37). (C and D) Global TCR repertoire analysis of CD4+ and CD8+ T cells from HCs (n = 10) and treatment-naïve patients (n = 14) (table S3). Graphs display numbers of unique clones and sample overlap of CD4+ (C) and CD8+ (D) T cells. For changes at baseline between HCs and patients, P values were calculated using a Wilcoxon rank sum test. (E and F) Global TCR repertoire analysis of CD4+ and CD8+ T cells from patients at baseline and after 3 and 6 months of teriflunomide treatment (n = 15) from the TERIDYNAMIC clinical study. Graphs display numbers of unique clones and sample overlap of CD4+ (E) and CD8+ (F) T cells. P values were calculated from linear mixed model on change from baseline. (G and H) Global TCR repertoire properties of CD4+ T cells from patients at baseline and upon treatment with DMF for 6 months (n = 14), IFN-β for 12 months (12M) (n = 10), and GLAT for 12 months (n = 10) (table S10). Graphs display numbers of unique clones (G) and sample overlap (H). For changes from baseline and (C), P values were calculated using a paired Student’s t test. Horizontal lines indicate mean, and error bars show SEM. *P < 0.05, **P < 0.01, and ***P < 0.001.

On the basis of these results, we hypothesized that such differential effects on distinct T cell populations might, in turn, result in changes of the TCR repertoire, because some T cell clones might be more affected than others. First, we performed a detailed analysis of the TCR repertoire from 14 treatment-naïve patients with RRMS at an early disease stage in comparison with 10 age- and sex-matched healthy controls (HCs). Here, we observed alterations in the TCR repertoire between treatment-naïve patients and HCs (Fig. 1, C and D) characterized by an increase in TCR repertoire diversity reflected by higher numbers of unique clones in the CD4+ compartment and, to a lesser extent, in the CD8+ compartment. Moreover, patients exhibited a greater percentage of clones with shared amino acid sequences (termed sample overlap), which indicates that the frequency of common or shared clones is higher in patients than in HCs (Fig. 1, C and D). Analysis of samples from the TERIDYNAMIC clinical trial demonstrated that teriflunomide treatment reduced CD4+ TCR repertoire diversity within several weeks of treatment initiation, as numbers of unique clones were profoundly reduced upon treatment (Fig. 1E). In addition, teriflunomide treatment resulted in a substantial decrease in sample overlap in patients, suggesting a reduction of shared clones by teriflunomide (Fig. 1E). For CD8+ T cells, similar, albeit less pronounced, teriflunomide-induced changes were observed (Fig. 1F). To exclude potential confounding effects of previous disease-modifying treatments (DMTs) on the TCR repertoire, we performed a second analysis focusing on treatment-naïve patients with RRMS before and during teriflunomide treatment and could corroborate the effects of teriflunomide on the TCR repertoire (fig. S1, J and K). We did not observe alterations in CD4+ TCR repertoire diversity in patients after immunomodulatory treatment with dimethyl fumarate (DMF), interferon-β (IFN-β), or glatiramer acetate (GLAT) (Fig. 1, G and H). These results indicate that immunomodulation per se does not result in TCR repertoire changes.

These differential effects suggest that some T cells might be more susceptible toward DHODH-mediated changes than others. To investigate the DHODH-mediated impact on antigen-specific T cell responses in more detail, we used a murine system using T cells from transgenic mice that are specific for distinct model antigens.

DHODH inhibition exerts affinity-dependent effects on T cell proliferation

We compared the effects of DHODH inhibition on T cell proliferation by using transgenic T cells that recognize antigenic peptides with different antigen affinities. First, to evaluate the effect of teriflunomide on CD4+ T cells, we made use of myelin oligodendrocyte glycoprotein (MOG)–specific T cells from 2D2 mice, which are known to cross-react with a particular neurofilament (NFM15–35) peptide (11), with a higher affinity than their cognate MOG35–55 peptide (11). Teriflunomide was more effective in restricting proliferation of CD4+ T cells upon high-affinity stimulation compared to low-affinity stimulation in vitro (Fig. 2, A and B). This was accompanied by a differential effect on CD4+ T cell expansion with a 91% inhibition of high-affinity stimulated CD4+ T cells compared to 59% inhibition of low-affinity stimulated CD4+ T cells (Fig. 2C), whereas the viability of cells was not affected (fig. S2A).

Fig. 2 Antigen affinity–dependent effect of teriflunomide on CD4+and CD8+T cells.

(A) Proliferation of CD4+ T cells from 2D2 mice upon stimulation with NFM15–35 (high-affinity) or MOG35–55 (low-affinity) peptide-loaded dendritic cells in the presence (+TF) or absence (w/o) of teriflunomide at day 3. Data are representative of four independent experiments. (B) Percentages of proliferated CD4+ T cells from 2D2 mice upon stimulation with NFM15–35 or MOG35–55 peptide-loaded dendritic cells in the presence or absence of teriflunomide at day 3 (n = 4). ns, not significant. (C) Absolute cell numbers of CD4+ T cells from 2D2 mice upon stimulation with NFM15–35 or MOG35–55 peptide-loaded dendritic cells in the presence or absence of teriflunomide at day 3 (n = 5). Data were normalized to cells without treatment. (D) Proliferation of OT-I CD8+ T cells upon stimulation with altered peptide ligands of OVA257–264 with different affinities: SIINFEKL (N4) > SIIQFEKL (Q4) > SIITFEKL (T4) loaded on splenocytes in the presence or absence of teriflunomide at day 3. Generation analysis was performed with the FlowJo proliferation tool. Data are representative of three independent experiments. (E) Percentages of proliferated OT-I CD8+ T cells upon stimulation with N4, Q4, or T4 loaded on splenocytes in the presence or absence of teriflunomide at day 3 (n = 3). (F) Absolute cell numbers of OT-I CD8+ T cells upon stimulation with N4, Q4, or T4 loaded on splenocytes in the presence or absence of teriflunomide at day 3 (n = 3). Data were normalized to cells without treatment. (G) Proliferation of OT-I (high-affinity TCR) and OT-III (low-affinity TCR) CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes in the presence or absence of teriflunomide at day 3. Generation analysis was performed with the FlowJo proliferation tool. Data are representative of three independent experiments. (H) Percentages of proliferated OT-I and OT-III CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes in the presence or absence of teriflunomide at day 3 (n = 3). (I) Absolute numbers of OT-I and OT-III CD8+ upon stimulation with OVA257–264 peptide-loaded splenocytes in the presence or absence of teriflunomide at day 3 (n = 2). Data were normalized to cells without treatment. All data are displayed as means ± SEM. Statistical analysis was conducted by Student’s t test and was defined as *P < 0.05 and **P < 0.01.

For further corroboration, we switched to transgenic CD8+ T cells from OT-I mice recognizing altered peptide ligands of the model antigen ovalbumin with distinct antigen affinities (SIINFEKL > SIIQFEKL > SIITFEKL) (Fig. 2, D to F) (12). Again, we observed a differential effect of DHODH inhibition depending on the affinity of the peptide used, which was furthermore illustrated by generation analysis of proliferating CD8+ T cells (Fig. 2D). This differential effect cannot be explained by mere quantitative differences in T cell activation, because it was observed among different antigen concentrations (fig. S2B).

Last, we used transgenic T cells recognizing an identical peptide, albeit with distinct antigen affinities. SIINFEKL can also stimulate OT-III T cells, but the affinity of the interaction is lower than that with OT-I T cells (13). Again, teriflunomide was much more effective in restricting proliferation of high-affinity CD8+ T cells compared to low-affinity CD8+ T cells (Fig. 2, G to I). Polyclonal stimulation of high-affinity versus low-affinity CD8+ T cells abrogated teriflunomide-mediated differences in T cell proliferation (fig. S2, C and D), illustrating that the observed differences are linked to antigen-specific stimulation.

In contrast to the strong antiproliferative effect, we did not observe a profound effect of DHODH inhibition on effector molecule production when evaluating IFN-γ and granzyme B production by both high-affinity and low-affinity CD4+ and CD8+ T cells on a per cell basis using intracellular flow cytometry (fig. S2, E to H). Hence, the well-known and robust decrease in proinflammatory cytokine secretion in the supernatants of teriflunomide-treated T cells is thus most likely due to inhibition of T cell expansion rather than a direct effect on cytokine production on a cellular level (fig. S2, I and J).

Further experiments revealed that a structurally distinct DHODH inhibitor, brequinar, exerted comparable affinity-dependent effects on T cell proliferation (fig. S3, A to C). Other antiproliferative drugs such as the purine synthesis inhibitor mercaptopurine and the DNA intercalating agent mitoxantrone did not exhibit any affinity-dependent effects over a range of concentrations despite robust interference with T cell proliferation (fig. S3, A to D). We observed that the affinity-dependent effects of teriflunomide and brequinar on T cell proliferation could be rescued upon addition of the pyrimidine base uridine but not of the purine base guanosine (fig. S3E), further illustrating that inhibition of DHODH-mediated de novo pyrimidine synthesis is pivotal for teriflunomide-mediated effects on T cell proliferation; therefore, these data do not support a profound DHODH-independent effect of teriflunomide.

DHODH inhibition interferes with energy generation via OXPHOS and aerobic glycolysis

In light of the intimate topological relationship of DHODH with components of the electron transport chain, we wondered whether pharmacological DHODH inhibition might affect OXPHOS in activated T cells. In the presence of teriflunomide, OXPHOS was profoundly impaired in murine CD3/CD28-activated CD4+ and CD8+ T cells (Fig. 3, A to D). In naïve T cells, where generally low OXPHOS activity is observed, this effect could not be seen (Fig. 3, B and D, and fig. S4, A and B). Brequinar, a structurally distinct DHODH inhibitor, reduced the ability for maximal respiration (fig. S4C). In accordance with the role of DHODH, uridine supplementation was able to rescue impaired OXPHOS by teriflunomide (fig. S4D). Aerobic glycolysis was also severely impaired in activated CD4+ and CD8+ T cells in the presence of teriflunomide (Fig. 3, E to H, and fig. S4, E and F). Detailed kinetic analysis revealed that teriflunomide-mediated inhibition of CD8+ T cell proliferation was only effective during the first 24 hours of T cell activation, whereas DHODH inhibition at later time points only marginally affected T cell proliferation (Fig. 3I). Mitochondrial respiration was equally suppressed upon teriflunomide initiation at different time points throughout the 48 hours of T cell activation (Fig. 3J). The same was observed with respect to aerobic glycolysis (Fig. 3J). Thus, it can be concluded that OXPHOS is required during the initial phase of T cell proliferation, i.e., during the first 24 hours, whereas at later time points, DHODH-mediated inhibition of OXPHOS does not critically influence T cell proliferation. We then analyzed T cell proliferation in the presence or absence of low doses of oligomycin, an adenosine 5′-triphosphate (ATP) synthase inhibitor, at different time points. As expected, initiation of proliferation was substantially impaired in the presence of even low doses of oligomycin (5 to 10 nM) (fig. S4G), again demonstrating that the initiation of proliferation requires mitochondrial ATP synthesis (fig. S4H). In contrast, oligomycin did not influence the proliferation of already actively proliferating T cells and, hence, did not abolish the DHODH-mediated inhibitory effects on T cell proliferation (fig. S4, I and J). These data corroborate our hypothesis that DHODH-mediated OXPHOS inhibition is critical during the early stages of T cell proliferation but dispensable later on. Last, we addressed whether selective interference with OXPHOS using specific inhibitors of either complex I or III of the electron transport chain might affect glycolysis. OXPHOS inhibition by the complex I inhibitor rotenone or the complex III inhibitor antimycin A did result in an impairment of aerobic glycolysis, therefore indicating that aerobic glycolysis at least to some extent depends on preserved OXPHOS activity, providing further evidence of the close relationship between both pathways of energy generation in activated T cells (Fig. 3K).

Fig. 3 Influence of DHODH interference on T cell metabolism.

(A) Oxygen consumption rate (OCR) of α-CD3/CD28–stimulated 2D2 CD4+ T cells in the presence or absence of teriflunomide at day 3. (B) Bar graph of basal and maximal (Max.) respiration from unstimulated naïve (Naive) and α-CD3/CD28–stimulated (Stim) 2D2 CD4+ T cells in the presence or absence of teriflunomide at day 3. (C) OCR of α-CD3/CD28–stimulated OT-I CD8+ T cells in the presence or absence of teriflunomide at day 2. (D) Bar graph of basal and maximal respiration from unstimulated naïve and α-CD3/CD28–stimulated OT-I CD8+ T cells in the presence or absence of teriflunomide at day 2. (E) Extracellular acidification rate (ECAR) of α-CD3/CD28–stimulated 2D2 CD4+ T cells in the presence or absence of teriflunomide at day 3. (F) Bar graph of glycolysis and glycolytic (Glyc.) capacity from unstimulated naïve and α-CD3/CD28–stimulated 2D2 CD4+ T cells in the presence or absence of teriflunomide at day 3. (G) ECAR of α-CD3/CD28–stimulated OT-I CD8+ T cells in the presence or absence of teriflunomide at day 2. (H) Bar graph of glycolysis and glycolytic capacity from unstimulated naïve and α-CD3/CD28–stimulated OT-I CD8+ T cells in the presence or absence of teriflunomide at day 2. (I) Proliferation of OT-I CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes in the presence of teriflunomide added at indicated time points and measured at day 3. FACS, fluorescence-activated cell sorting; h, hours. (J) Mitochondrial respiration and glycolytic capacity of OT-I CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes in the presence of teriflunomide added at indicated time points and measured at day 2. (K) ECAR of α-CD3/CD28 stimulated OT-I CD8+ T cells at day 3. Rotenone (Rot) and antimycin A (AA) were added as indicated. Oligo, oligomycin; FCCP, carbonyl cyanide-p-trifluoromethoxyphenylhydrazone; Gluc, glucose; 2-DG, 2-deoxyglucose; Ctrl, control. All data are representative of three independent experiments or at least n = 3 per group (I and J) and displayed as means ± SEM. Statistical analysis was conducted by Student’s t test and was defined as **P < 0.01 and ***P < 0.001.

High-affinity and low-affinity T cells display differential metabolic capacities

Following up on our initial observation of affinity-dependent differences in the extent of DHODH-related effects on T cell proliferation, we hypothesized that this phenomenon might be linked to differences in T cell energy metabolism. We therefore compared the metabolic profiles of high-affinity and low-affinity T cells and observed differences between OT-I and OT-III T cells both with regard to OXPHOS and aerobic glycolysis. High-affinity CD8+ T cells were shown to display greater capacities for OXPHOS and for aerobic glycolysis (Fig. 4, A and B). Kinetic analysis further revealed that both maximal respiration and glycolytic capacity were more rapidly up-regulated in high-affinity CD8+ T cells compared to low-affinity CD8+ T cells (Fig. 4C). The extent of DHODH-mediated inhibition of OXPHOS was comparable between high-affinity and low-affinity CD8+ T cells, suggesting that the differential susceptibility toward DHODH-mediated effects is rather due to differences in their metabolic capacity than affinity differences in DHODH function or relevance (fig. S5A). A direct comparison of the metabolic capacities of OT-I and OT-III T cells demonstrated that OT-I T cells up-regulate OXPHOS and glycolysis under stress, whereas OT-III T cells increase glycolysis more than OXPHOS (Fig. 4D). In general, OT-I T cells display a greater energetic capacity compared to OT-III T cells, as illustrated by the phenogram of OT-I and OT-III T cells 48 hours after antigen-specific stimulation. OT-III T cells display a strongly limited respiratory capacity compared to OT-I T cells, whereas the difference in glycolytic capacity was less pronounced (Fig. 4D). Also, for CD4+ T cells, high-affinity stimulation resulted in a more pronounced up-regulation of OXPHOS and glycolysis and ultimately a higher energetic capacity as compared to low-affinity stimulation (Fig. 4, E to G).

Fig. 4 Impact of teriflunomide on high-affinity versus low-affinity T cell metabolism.

(A and B) OCR (A) and ECAR (B) of OT-I (high-affinity TCR) and OT-III (low-affinity TCR) CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes at day 2. Bar graphs display OCR of basal respiration and maximal respiration (A) or ECAR of glycolysis and glycolytic capacity (B). (C) Kinetic analysis of maximal respiration and glycolytic capacity of OT-I and OT-III CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes at indicated time points. (D) XF phenogram of OT-I and OT-III CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes at day 2. (E and F) OCR (E) and ECAR (F) of 2D2 CD4+ T cells upon stimulation with NFM15–35 (high-affinity) or MOG35–55 (low-affinity) peptide-loaded dendritic cells at day 3. Bar graphs display OCR of basal respiration and maximal respiration (E) or ECAR of glycolysis and glycolytic capacity (F). (G) XF PhenoGram of 2D2 CD4+ T cells upon stimulation with NFM15–35 or MOG35–55 peptide-loaded dendritic cells at day 3. All data are representative of three independent experiments and displayed as means ± SEM. Statistical analysis was conducted by Student’s t test and was defined as *P < 0.05 and ***P < 0.001.

Last, when assessing the function of each respiratory complex separately, OT-I T cells showed a higher complex activity (I to IV) compared to OT-III T cells, again demonstrating that OT-I T cells have a higher OXPHOS ability (fig. S5B). In light of the strong dependence of initial T cell proliferation on OXPHOS, this might explain the distinct inhibitory effect of DHODH inhibition on high-affinity T cell proliferation in comparison to low-affinity T cell proliferation.

Given the affinity-dependent differences in kinetics of energy generation, we next addressed the kinetics of antigen-specific T cell proliferation in high-affinity and low-affinity T cells in more detail. High-affinity T cells display a small but crucial kinetic advantage in comparison to low-affinity T cells at early time points, i.e., 36 and 48 hours after T cell activation, thus mirroring the kinetic differences of energy generation (Fig. 5, A and B). TCR-derived signals are converted into metabolic activities via a range of key transcription factors serving as intracellular metabolic checkpoints such as IFN regulatory factor 4 (IRF4), mammalian target of rapamycin (mTOR), or c-Myc. We therefore analyzed the expression pattern of these key metabolic transcription factors in high-affinity versus low-affinity T cells early upon TCR-mediated activation. We observed a more pronounced nuclear expression of IRF4, c-Myc, and phospho-S6 ribosomal protein, a downstream target of mTOR, in OT-I versus OT-III T cells upon stimulation (Fig. 5, C to E). DHODH inhibition did not affect IRF4, c-Myc, and phospho-S6 ribosomal protein, neither in high-affinity nor in low-affinity T cells (Fig. 5, C to E).

Fig. 5 Affinity-dependent gene expression.

(A) Proliferation of OT-I (high-affinity TCR) and OT-III (low-affinity TCR) CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes at indicated time points. Data are representative of at least two independent experiments. (B) Statistical analysis of data depicted in (A). (C) Expression of transcription factor IRF4 in the nucleus of OT-I and OT-III CD8+ T cells after antigen-specific activation in the presence or absence of teriflunomide for 3 hours (at least n = 3). Graph displays mean fluorescence intensity (MFI) analyzed by flow cytometry. (D) MFI of c-Myc in the nucleus of OT-I and OT-III CD8+ T cells after antigen-specific activation in the presence or absence of teriflunomide for 1 hour (n = 4). (E) Expression of phosphorylated S6 (phospho-S6) protein, a downstream target of mTOR, in OT-I and OT-III CD8+ T cells after antigen-specific activation in the presence or absence of teriflunomide for 3 hours (at least n = 3). Graph displays MFI analyzed by flow cytometry. (F) Heat map of the expression of 84 glycolytic and 84 mitochondrial respiration genes assessed in OT-I and OT-III CD8+ T cells after antigen-specific activation for 12 hours in the presence or absence of teriflunomide (at least n = 3). (G) Principal components analysis (PCA) of data depicted in (F). (H) Venn diagram of all differentially expressed genes (OT-I versus OT-III). The orange area indicates the overlap between TCR affinity–regulated genes and stimulation dependent–regulated genes. All data are displayed as means ± SEM. Statistical analysis was conducted by Student’s t test or one-way analysis of variance (ANOVA) (C to E) and was defined as *P < 0.05, **P < 0.01, and ***P < 0.001. Unstim, unstimulated.

As these data suggest that affinity-dependent intracellular signals are routed into discrete transcriptional networks via modulation of several key transcription factors serving as checkpoints of metabolic T cell programming, we next compared the transcriptional profile of high-affinity and low-affinity T cells with regard to a broad range of metabolically relevant genes in more detail. We observed differences in the expression of numerous metabolically relevant genes in activated high-affinity versus low-affinity T cells as depicted by principal components analysis and scatterplot, respectively (Fig. 5, F and G, and fig. S5, C to E); these differences were only present upon T cell activation. Moreover, as expected, DHODH inhibition itself did not affect gene expression of metabolically relevant genes (Fig. 5, F and G). Last, we observed that, in total, 53 of 168 investigated metabolic genes were affinity dependent, and most of those (48 of 53) were also dependent on TCR stimulation (Fig. 5H and table S2). Together, these data indicate that affinity-dependent signals via the TCR are intracellularly translated into metabolic programs via distinct modulation of a set of key metabolic transcription factors, which, in turn, result in a differential metabolic capacity of high-affinity versus low-affinity T cells.

Pharmacological DHODH inhibition interferes with mitochondrial complex III function

Despite the notable differences in mitochondrial respiratory capacity, mitochondrial DNA content was not different between high-affinity and low-affinity T cells (Fig. 6A). Histochemical visualization and quantification of mitochondria confirmed equal presence of mitochondria in activated OT-I and OT-III T cells both in the presence and absence of teriflunomide (Fig. 6, B and C).

Fig. 6 Effect of DHODH inhibition on mitochondrial content and function.

(A) Relative mitochondrial DNA (mtDNA)/nuclear DNA (nDNA) ratio of OT-I (high-affinity TCR) and OT-III (low-affinity TCR) CD8+ T cells upon stimulation with OVA257–264 peptide-loaded splenocytes in the presence or absence of teriflunomide at day 2 (n = 3). Data were normalized to OT-I without. (B) OT-I and OT-III CD8+ T cells were activated with α-CD3/2CD8 in the presence or absence of teriflunomide for 2 days. Cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue) and MitoTracker green (MT green, green). Scale bars, 20 μM. (C) Quantification of MT green intensity with ImageJ software. Dots represent the intensity of single images with the same size and exposure time. (D) Histochemistry of complex IV activity of OT-I CD8+ T cells unstimulated or activated with α-CD3/CD28 in presence or absence of teriflunomide at day 2. Scale bars, 10 μM. Graph data display quantification of densitometric mean of individual complex IV puncta. (E) Activity of complexes I, II/III, and IV of the mitochondrial respiration chain in OT-I CD8+ T cells activated with α-CD3/CD28 in the presence or absence of teriflunomide for 2 days (n = 5). Cells were permeabilized before measurement, and substrates and inhibitors of mitochondrial respiration chain complexes were successively added. Scheme shows simplified illustration of the mitochondrial respiration chain (complexes I to IV) with all substrates (green) and inhibitors (red). Succ, succinate; TMPD, tetramethylphenylendiamin; Asc, ascorbate; Cyt c, cytochrome c. All data are displayed as means ± SEM. Statistical analysis was conducted by Student’s t test (A and C), one-way ANOVA (D), or two-way ANOVA (E) and was defined as *P < 0.05, **P < 0.01, and ***P < 0.001.

We next investigated whether DHODH inhibition directly interferes with the function of individual complexes of the mitochondrial respiratory chain, as DHODH is in close proximity to the ubiquinone pool located between complexes II and III (14). First, teriflunomide diminished the increase in complex IV activity upon T cell stimulation (Fig. 6D). Furthermore, we measured the activity of mitochondrial complexes I, II/III, and IV of the respiratory chain, as described in Materials and Methods. Upon inhibition of complex III, the teriflunomide effect on OXPHOS was abrogated (Fig. 6E). Together, these data illustrate that DHODH inhibition functionally interferes with the complex III activity of the electron transport chain in activated T cells but does not affect mitochondrial content or structure.

DHODH inhibition exerts affinity-dependent effects on effector T cells in vivo

Given the affinity-dependent effects of DHODH inhibition in our in vitro model systems, we then aimed to evaluate whether such affinity-dependent effects might also occur in vivo in the context of T cell–mediated autoimmunity. To this end, we used an established animal model of MS, experimental autoimmune encephalomyelitis (EAE), where mice were immunized with the major histocompatibility complex (MHC) class II MOG35–55 peptide and received oral treatment with either the precursor of teriflunomide (leflunomide) or vehicle from 3 days before immunization throughout the disease course. In accordance with the literature (15), EAE disease course was significantly ameliorated under leflunomide treatment (P < 0.001) (Fig. 7A), accompanied by reduced infiltrating CD4+ T cell numbers—including numbers of cytokine-producing cells—in the CNS (Fig. 7, B and C). Matching our observations from the TERIDYNAMIC study, absolute numbers of IFN-γ–producing CD4+ T cells in the periphery were significantly reduced under leflunomide (P < 0.05), whereas numbers of interleukin-17A (IL-17A)–producing CD4+ T cells were not affected (Fig. 7D). In this line, teriflunomide interfered with de novo differentiation and proliferation of murine TH1 cells in vitro (fig. S5, F and G). When studying different Treg subpopulations in EAE mice under leflunomide, type 1 regulatory (Tr1) cells in the circulation increased, whereas other Treg subpopulations were not affected by leflunomide treatment (fig. S5H).

Fig. 7 Relevance of DHODH inhibition in vivo.

(A) Clinical EAE score of C57BL/6 mice treated orally with leflunomide (+LF, precursor of teriflunomide) or vehicle from 3 days before immunization throughout the disease course. Active MOG-EAE was induced in C57BL/6 mice (at least n = 13 per group), and the clinical disease score was assessed daily. (B and C) Absolute numbers of CD4+ T cells (B), CD4+ IFN-γ+, and CD4+ IL-17A+ T cells (C) from the CNS of immunized mice from (A) at day 14 analyzed by flow cytometry. (D) Absolute numbers of CD4+ IFN-γ+ and CD4+ IL-17A+ T cells from the spleen of immunized mice from (A) at day 14 analyzed by flow cytometry. (E) Tetramer staining and summary graph of MOG35–55-IAb+ CD4+ T cells isolated from the CNS of immunized mice treated with leflunomide or vehicle at day 14 (n = 7 to 8 per group). (F and G) Adhesion frequency of MOG35–55-specific CD4+ T cells isolated from the CNS of mice treated with leflunomide or vehicle at day 10 (n = 12 per group). T cells were tested for adhesion to MOG38–49-IAb or negative control hCLIP103–117-IAb. Receptor (R) density was assessed by flow cytometry (CD4+ T cells, 18 R/μm2; pMHC MOG38–49 -IAb, 770 R/μm2; and pMHC hCLIP103–117 -IAb, 393 R/μm2). (H and I) Treatment-naïve HLA-DR4+ patients with RRMS at baseline and during at least 6 months of teriflunomide treatment (n = 3) were analyzed for frequencies of myelin-specific T cells by using DRB1*0401/MOG97–109 and DRB1*0401/PLP180–199 tetramers and the DRB1*0401/CLIP87–101 control tetramer (table S8). FACS plots show one representative example. Bar graphs display the relative ratio of MOG97–109+ (H) or PLP180–199+ (I) CD4+ T cells to CLIP87–101+ CD4+ T cells. Data are displayed as means ± SEM. Statistical analysis was conducted by two-way ANOVA (A) or Student’s t test and was defined as *P < 0.05, **P < 0.01, and ***P < 0.001.

To specifically follow antigen-specific T cells during EAE, we performed tetramer staining of myelin-specific CD4+ T cells by using the MOG35–55-IAb tetramer, revealing a significant reduction in the frequency of MOG-specific CD4+ T cells in the CNS from leflunomide-treated mice (P < 0.01) (Fig. 7E). We used an elegant method to quantify affinities of antigen-specific T cell responses using two-dimensional (2D) microscopy, as described previously (16). At day 10 after EAE induction, immune cells were isolated from the CNS, and single-isolated CD4+ T cells were subjected to 2D microscopy; at least 26 cells per group (without, n = 26; +LF, n = 28; negative control, response to unspecific hCLIP103–117-IAb, n = 4) were analyzed in a blinded fashion. The adhesion frequency of CD4+ T cells to MOG35–55-loaded MHC class II molecules as a direct correlate to antigen affinities was significantly reduced in leflunomide-treated EAE mice compared to vehicle-treated EAE mice (P < 0.05) (Fig. 7F). Grouping of individual affinities in quartiles further revealed that the decrease in mean antigen affinities in leflunomide-treated mice was mainly driven by the ablation of high-affinity MOG-specific CD4+ T cells (Fig. 7G).

As these data indicate that DHODH inhibition has an impact on the affinity spectrum of autoreactive T cells in vivo, we aimed to translate this finding to the human situation. Here, we made use of a published setup (17), allowing detection of myelin-specific CD4+ T cells by DRB1*0401/MOG97–109 and DRB1*0401/PLP180–199 tetramer staining in HLA-DR4+ patients before and during teriflunomide treatment. Teriflunomide treatment reduced the frequency of MOG- and PLP-specific T cells in the subset of treatment-naïve HLA-DR4+ patients available to us (Fig. 7, H and I). Together, these data demonstrate that pharmacological DHODH inhibition preferentially affects high-affinity T cells during an antigen-specific autoimmune response in vivo.

Patients with RRMS are characterized by an altered metabolic profile of activated T cells

Last, we aimed to address the potential relevance of DHODH-mediated metabolic alterations in patients with RRMS. When comparing the metabolic profile of in vitro–activated CD4+ T cells from patients and HCs, we observed that T cells from patients during relapse (n = 24), but not in remission (n = 25), exhibit an enhanced OXPHOS and glycolytic activity as compared to HCs (n = 24) (Fig. 8, A and B), suggesting that, during disease activity, T cells exhibit a disturbed metabolic profile. Activation-induced increase in mitochondrial respiratory activity and glycolysis further illustrates the enhanced metabolic potential of these T cells (Fig. 8, C and D). Again, these metabolic changes were exclusively observed in T cells from patients with an ongoing relapse. We did not observe a difference in OXPHOS and aerobic glycolysis of CD4+ T cells in teriflunomide-treated patients (Fig. 8, E and F) or CD8+ T cells (fig. S6A). However, addition of teriflunomide in vitro inhibited both OXPHOS and aerobic glycolysis in activated human T cells both from HCs and patients (n = 10 per group) (Fig. 8, G and H, and fig. S6, B to E). Together, these data suggest that T cells from active patients with RRMS exhibit an augmented cellular metabolism, possibly contributing to the known immune dysregulation in MS and which is amenable to pharmacological DHODH inhibition.

Fig. 8 Metabolic profiles of T cells from patients with RRMS and HCs.

(A and B) Basal and maximal respiration (A) or glycolysis and glycolytic capacity (B) of human CD4+ T cells from HCs (n = 24), treatment-naïve patients without (n = 25) and with relapse (n = 24) upon short-term stimulation with phorbol 12-myristate 13-acetate (PMA) and ionomycin (Iono) for 2.5 hours (table S11). (C and D) Activation-induced increase in mitochondrial respiration (C) and glycolysis (D) was calculated from human CD4+ T cells from the cohort depicted in (A) and (B) either left unstimulated or upon short-term stimulation for 2.5 hours with PMA/ionomycin before measurement of OCR or ECAR (stimulated OCR or ECAR values/unstimulated OCR or ECAR values). (E) CD4+ T cells from treatment-naïve patients at baseline and after at least 6 months of teriflunomide treatment (n = 14) were analyzed for maximal respiration and glycolytic capacity upon short-term stimulation with PMA/ionomycin for 2.5 hours (table S9). (F) OCR and ECAR values of CD4+ T cells from one representative treatment-naïve patients with RRMS at baseline and during teriflunomide treatment from (E). (G) Maximal respiration and glycolytic capacity of human CD4+ T cells from HCs (n = 12) and patients (n = 11) activated with α-CD3/CD28 in the presence or absence of teriflunomide for 3 days (table S12). (H) OCR and ECAR values of CD4+ T cells from one representative patient with RRMS from (G). All data are displayed as means ± SEM. Statistical analysis was conducted by one-way ANOVA (A to D) or Student’s t test (E to G) and was defined as *P < 0.05, **P < 0.01, and ***P < 0.001.

DISCUSSION

Here, we demonstrate that pharmacological interference with DHODH, a mitochondrial enzyme in the pyrimidine synthesis pathway and the target of therapeutic agents used for treatment of RA and MS, does not uniformly suppress proliferation of activated T cells. DHODH inhibition instead exerts differential effects on T cell clones, which is mainly determined by the affinities of antigen-specific T cell responses and its associated distinct metabolic profiles.

The data from our TERIDYNAMIC trial point toward a rather selective effect of the antiproliferative drug teriflunomide on different T cell populations, which was somewhat unexpected, as it is generally believed that teriflunomide and leflunomide interfere with the proliferation of any activated T cell due to the increased pyrimidine demand during activation-induced T cell proliferation (3). The preferential reduction in TH1 effector cells by DHODH inhibition not only in human patients but also in the EAE model is intriguing, especially in light of their acknowledged role in autoimmunity (18, 19). In support of these findings, we observed an inhibition of murine TH1 cell polarization and proliferation in vitro by teriflunomide. On the other hand, absolute iTreg numbers were unaffected, resulting in a relative increase in the proportions of iTregs within the CD4+ compartment and hence an increased iTreg/TH1 ratio, potentially indicating that DHODH inhibition in patients with RRMS might restore a disturbed balance between proinflammatory versus anti-inflammatory CD4+ T cell populations. A relative increase in Tregs upon DHODH inhibition has also been described in other mouse models of autoimmunity (20, 21). On a functional level, teriflunomide treatment neither affected the suppressive capacity of Tregs nor modulated the expression of key regulatory molecules and cytokines, indicating that DHODH is not implicated in the regulation of Treg function itself.

On the basis of the hypothesis that the observed selective effects of DHODH inhibition on distinct T cell populations might result in TCR repertoire changes, we assessed the effect of teriflunomide treatment on the TCR repertoire in patients with RRMS. Analysis of our patient cohort revealed that patients display an enhanced TCR repertoire diversity, as illustrated by increased numbers of unique clones and an increased sample overlap. Our findings support data from the cerebrospinal fluid (CSF) in a small cohort of five patients with RRMS (22) and, together, suggest that in MS there is a perturbation of clonal elimination, potentially caused by an impaired deletion of autoreactive clones (23, 24). We were unable to obtain more information about the TCR repertoire in the target organ, i.e., the CNS or at least the CSF. It is difficult to obtain CSF in a longitudinal fashion as part of a clinical study and was not included in our study protocol. Treatment of patients with teriflunomide resulted in a reduction of TCR repertoire diversity back to the levels of HCs. Although this remains speculative, the preferential sparing of Tregs by teriflunomide might help to restore their capacity to restrict TCR repertoire diversity by elimination of potentially autoreactive T cell clones (25). TCR sequencing does not provide any information about the antigen specificity or antigen affinity of the T cell clones investigated; however, the reduction in clonal diversity points to a deletion of distinct clones as a consequence of teriflunomide treatment. Our analysis of myelin-specific CD4+ T cell responses in HLA-DR4+ patients further supports this concept as we observed a reduction in frequencies of myelin-specific T cells in the course of teriflunomide treatment, albeit this analysis could only be performed in a small subset of patients.

Experiments with transgenic T cells revealed that DHODH inhibition differentially inhibited T cell proliferation depending on the antigen affinity of the T cell, as high-affinity T cells were more affected than low-affinity T cells. Recently, Man et al. (26) demonstrated that affinity-dependent signals from the TCR in the context of infections are translated into distinct transcriptional programs, which, in turn, determine the metabolic function of effector T cells. Mitochondrial respiration is essential for initial activation of naïve T cells (9, 10), and our data revealed a so far unappreciated role of DHODH in control of mitochondrial respiration during T cell activation. While the suppressive effect of DHODH inhibition on OXPHOS is plausible because of the intimate spatial relationship of DHODH with complex III of the respiratory chain, the inhibition of glycolysis was somewhat unexpected. However, strong inhibition of mitochondrial respiration using complex I and III inhibitors likewise impaired aerobic glycolysis, hence suggesting that the observed drop in glycolytic rate upon DHODH inhibition is most likely indirectly due to its impairment of mitochondrial respiration.

Comparison of metabolic profiles of activated high-affinity and low-affinity T cells revealed substantial differences both with regard to transcriptional regulators such as IRF4, c-Myc, and the mTOR pathway and transcriptional regulation of metabolic genes. Furthermore, a functional comparison of the metabolic capacities of high-affinity versus low-affinity T cells displayed a clear metabolic advantage of high-affinity T cells both with respect to OXPHOS and aerobic glycolysis. Together, these data do not only provide mechanistic insight into the molecular mechanisms of how TCR signals that result from antigen-specific interactions of certain affinities can be translated into distinct intracellular programs but also explain the differential effect of DHODH inhibition on high-affinity and low-affinity T cells. Irrespective of antigen affinities, DHODH inhibition interferes with mitochondrial respiration; however, as high-affinity T cells depend more on mitochondrial respiration for optimal energy supply during the initial steps of T cell activation, they are more susceptible to signals interfering with mitochondrial respiration such as DHODH inhibition.

With regard to the in vivo relevance of the observed affinity-dependent effects of DHODH inhibition, analysis of the antigen affinity spectrum of MOG-reactive T cell clones in the MOG-induced EAE model provides further evidence that DHODH-mediated affinity-dependent effects are also present in the context of a polyclonal immune response. They illustrate that DHODH inhibition shapes the affinity spectrum of an autoantigen response, as we observed a preferential ablation of high-affinity MOG-specific T cells in the CNS of leflunomide-treated EAE mice. It has been described that self-reactive T cells with relatively high avidity can escape thymic negative selection and contribute to autoimmune disease manifestation (5, 6). Development of organ-specific autoimmunity in a model of diabetes was driven by affinity maturation of the prevailing autoantigen-specific T cell population, and selective depletion of high-affinity T cells prevented disease development (5). These studies underline the relevance of high-affinity T cell clones for disease initiation and progression in different organ-specific autoimmune diseases. On the basis of these considerations, we propose that, during chronic autoimmune responses characterized by repetitive (re)activation of autoreactive T cells, DHODH inhibition might prevent affinity maturation and hence reduce disease propagation due to its preferential inhibition of high-affinity T cell clones.

From a more general perspective, the relevance of DHODH-dependent changes in T cell metabolism highlights the attractiveness of the concept of immune metabolism as a therapeutic target in human autoimmune diseases. This concept has been fueled by several experimental studies demonstrating that modulation of immune metabolism can shape immune responses and thus ameliorate autoimmunity in different animal models (8, 27, 28). One intriguing example is the illustration of differential metabolic demands of effector T cell versus Treg subsets (28, 29). The specific metabolic profiles of distinct T cell subsets might contribute to our findings of a preferential effect of DHODH inhibition on TH1 cells as opposed to TH17 and Treg populations in vivo.

In the context of approved immunomodulatory MS drugs, it has recently been demonstrated that DMF downmodulates aerobic glycolysis in activated myeloid and lymphoid cells via inactivation of the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (30), which further supports the idea that therapeutic targeting of immune metabolism represents an attractive treatment concept also in MS. From another angle, these findings raised the idea that human autoimmune diseases might feature distinct perturbations in immune metabolism, which might be amenable to specific pharmacological modulation once fully understood. So far, a few studies have characterized the immune metabolism of T cells in the context of autoimmune diseases, and disturbances have been observed in RA and systemic lupus erythematosus, both in which CD4+ T cells are critical drivers of disease pathogenesis (27, 31). These studies revealed disease-specific differences in T cell immune metabolism, suggesting that there is no common “autoimmune signature” of immune-metabolic disturbance, but rather distinct alterations, which will require tailored strategies for each disease.

In MS, the focus has been mainly on metabolic disturbances within the CNS, especially in neurons and axons, and mitochondrial injury, and changes in glucose-metabolizing enzymes have been described in active MS lesions (32, 33). In peripheral immune cells from patients with RRMS, a study suggested that those cells might exhibit altered activities in several complexes of the electron transport chain and key enzymes of glycolysis such as hexokinase I pointing toward an impaired mitochondrial respiration and concomitantly decreased glycolytic activity; however, the patient cohort investigated was rather small and active, and stable patients were not separately investigated (34). Another study described impaired OXPHOS and glycolysis in activated PBMCs from patients with RRMS (35); however, using whole PBMCs for metabolic measurements does not provide information about the metabolic capacities of individual immune cell subsets. Our study showed that isolated CD4+ T cells from patients with active RRMS suffering from an acute relapse exhibit increased mitochondrial respiratory and glycolytic activity compared to those of clinically stable patients and HCs. This intriguing finding has two important implications. First, it supports the notion that T cell–mediated autoimmune diseases are characterized by distinct metabolic alterations, and the metabolic profile seems to be unique to the individual disease pathogenesis. Second, it appears that metabolic alterations in T cells from patients with RRMS correlate with disease activity. Together, therapeutic targeting of metabolic alterations might represent an attractive concept in MS and might represent an as yet unrecognized key mechanism of teriflunomide-mediated immune modulation in this disease.

MATERIALS AND METHODS

Study design

The goal of the study was to evaluate the effects of teriflunomide on immune cells especially focusing on human and murine T cells. TERIDYNAMIC (NCT01863888) was an open-label, phase 3b, clinical trial including patients with RRMS before/during teriflunomide treatment and age- and sex-matched HCs (see the “TERIDYNAMIC study design” section for more details). Further human studies included patients with RRMS and age- and sex-matched HCs and patients before/during teriflunomide treatment (see the “Study subjects independent from the TERIDYNAMIC trial” section for further details). For all human studies, blood sampling was approved by the local ethics committee, and all subjects signed informed consent. For mouse studies, all experiments were performed according to the guidelines of the Animal Ethics Committee and were approved by the government authorities of Nordrhein-Westfalen, Germany. Detailed study design, sample sizes, replicates, and inclusion/exclusion criteria are provided in the figure legends or in Materials and Methods. Primary data are reported in data file S1.

TERIDYNAMIC study design

TERIDYNAMIC was an exploratory, open-label, multicenter, phase 3b, clinical trial that recruited patients with RRMS (n = 50) from nine sites in Belgium, Germany, and The Netherlands. All age- and sex-matched HCs (n = 20) were recruited from one site in Germany. Eligible subjects were 18 to 56 years of age and met the McDonald 2010 criteria for RRMS. Patients were naïve to DMTs, received no DMT for ≥2 years, or received IFN-β or GLAT therapy (with ≤3 months of interruption) with a period of ≥2 weeks without IFN-β or GLAT (table S1). Patients were excluded if they experienced a relapse within 30 days before screening, had other relevant diseases, were pregnant, were breastfeeding, or were of childbearing potential and not using effective birth control. Patients were also excluded if they had previous or concomitant use of cytokine therapy or intravenous immunoglobulins within 3 months of screening, fingolimod within 1 year of screening, and natalizumab or other immunosuppressive agents within 2 years of screening or had ever used alemtuzumab or cladribine. After up to 4 weeks of screening, patients received 14 mg of teriflunomide orally once a day for 6 months. All patients who discontinued treatment underwent an accelerated elimination procedure as per local labeling. Follow-ups occurred 4 weeks after treatment discontinuation. Age- and sex-matched HCs (reference group) remained untreated during the screening (up to 1 week) and study period (6 months). For data analysis, all patients that had been treated per protocol were included. The protocol and its amendments were reviewed and approved by an independent ethics committee (2016-053-f-S). This trial was conducted in accordance with the Declaration of Helsinki and was registered on ClinicalTrials.gov (NCT01863888). All participants provided written informed consent before entering the study.

Study subjects independent from the TERIDYNAMIC trial

An independent cohort of 14 treatment-naïve patients with RRMS and 10 matched HCs were analyzed for TCR deep sequencing (table S3). Furthermore, a separate cohort of 56 patients with RRMS before and during teriflunomide treatment was characterized (tables S4 to S9). In addition, treatment-naïve patients with RRMS before and during treatment with DMF for 6 months (n = 14), IFN-β for 1 year (n = 10), or GLAT for 1 year (n = 10) were analyzed (table S10). For human T cell metabolism, a total of 60 patients with RRMS (stable, n = 36; relapse, n = 24) and 36 matched HCs were investigated (tables S11 and S12). Stable disease was defined as the absence of novel clinical symptoms and no magnetic resonance imaging activity within at least 4 weeks before PBMC collection. Relapse was defined according to acknowledged clinical criteria, e.g., new symptoms or deterioration of neurological symptoms, which lasts for at least 24 hours in the absence of infection. Moreover, the characteristics of freshly isolated versus frozen PBMCs were investigated in a cohort of 23 HCs (tables S13 and S14). Blood sampling of patients with clinically definite RRMS according to the McDonald criteria, as well as of age- and sex-matched HCs, was approved by the local ethics committee (2010-262-f-S), and all subjects signed informed consent.

Statistical analysis

The baseline value was compared between patients with RRMS and untreated HCs using a Student’s t test or a Wilcoxon rank sum test if strong violations from the Gaussian distribution occurred (conducted by Shapiro-Wilk test). The change from baseline to months 3 and 6 was analyzed using a linear mixed model including the respective baseline value and time as fixed effects and a random intercept for the patient. Adjusted least squares means (SE) at month 6 of changes from baseline are presented. To compare values obtained from two groups, two-tailed Student’s t test was performed. To compare values acquired from more than two groups, one-way or two-way ANOVA was performed as indicated.

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/11/490/eaao5563/DC1

Materials and Methods

Fig. S1. Changes in T cell subsets in patients with RRMS on teriflunomide treatment.

Fig. S2. Influence of teriflunomide on proliferation and cytokine production.

Fig. S3. Impact of antigen affinities and DHODH interference on proliferation.

Fig. S4. Metabolic assessment of T cells under teriflunomide treatment.

Fig. S5. Influence of DHODH inhibition on T cell metabolism, on gene expression of OT-I and OT-III T cells, on TH1 differentiation and proliferation, and on Treg subpopulations in EAE.

Fig. S6. Metabolism of T cells from patients with RRMS and HCs.

Fig. S7. Comparison of freshly isolated versus frozen PBMCs from HCs regarding immune cell subset composition and metabolism.

Fig. S8. Additional information supporting methodology.

Table S1. Demographics and baseline disease characteristics TERIDYNAMIC trial.

Table S2. Differentially regulated genes depicted in Fig. 5I.

Table S3. Demographics and baseline disease characteristics of HCs and treatment-naïve patients analyzed for TCR repertoire changes.

Table S4. Demographics and baseline disease characteristics of treatment-naïve patients before and during teriflunomide treatment analyzed for TCR repertoire changes.

Table S5. Demographics and baseline disease characteristics of treatment-naïve patients before and during teriflunomide treatment analyzed for immune cell phenotyping.

Table S6. Demographics and baseline disease characteristics of patients before and during teriflunomide treatment for at least 6 months analyzed for suppressive capacity of Tregs.

Table S7. Demographics and baseline disease characteristics of patients before and during teriflunomide treatment for at least 6 months analyzed for cytokine expression of Tregs.

Table S8. Demographics and baseline disease characteristics of treatment-naïve patients before and during teriflunomide treatment for at least 6 months analyzed for myelin-specific T cell frequencies.

Table S9. Demographics and baseline disease characteristics of treatment-naïve patients before and during teriflunomide treatment for at least 6 months analyzed for metabolic activity.

Table S10. Demographics and baseline disease characteristics of treatment-naïve patients before and during treatment with DMF, IFN-β, or GLAT analyzed for TCR repertoire changes.

Table S11. Demographics and baseline disease characteristics of HCs and treatment-naïve patients analyzed for metabolic activity.

Table S12. Demographics and baseline disease characteristics of HCs and patients analyzed for metabolic activity after 72 hours of in vitro stimulation.

Table S13. Demographics of HCs analyzed for immune cell phenotyping of freshly isolated versus frozen PBMCs.

Table S14. Demographics of HCs analyzed for metabolic differences of T cells isolated from fresh blood or frozen PBMCs.

Data file S1. Primary data.

References (3638)

REFERENCES AND NOTES

Acknowledgments: We thank B. Van Wijmeersch, R. Hupperts, M. Mäurer, M. Stangel, M. Lang, and B. Tackenberg for contribution of patient data and biomaterial within the TERIDYNAMIC trial and A. Posevitz-Fejfar, A. Lysandropoulos, D. Decoo, and S. Brette for assisting in the design and analysis of the TERIDYNAMIC clinical trial. We thank D. Zehn (Technical University Munich, Munich, Germany) for providing the OT-III transgenic mice. Teriflunomide was provided by Sanofi Genzyme as a part of a research project. Moreover, we thank A. Engbers and A. Pabst (University Hospital Münster, Münster, Germany) for excellent technical assistance. Funding: This study was funded by the Interdisziplinäres Zentrum für Klinische Forschung (IZKF), Münster (LK2/015/14 to L.K.) and by the Deutsche Forschungsgemeinschaft (DFG) SFB TR128 [projects A08 (to L.K.), A09 (to C.C.G. and H.W.), and Z2 (to H.W.)] and SFB 1009 [projects A03 (to L.K. and H.W.) and WI 1722/12-1 (to H.W.)]. The TERIDYNAMIC study was funded by Sanofi Genzyme. Author contributions: M. Eschborn, M. Lindner, M. Liebmann, B.T.G., V.P., J.B., M.H., A.S.-M., N.F., J.A., C.J., T.W., K.L., G.R.C., D.J.M., P.H., and G.N. performed experiments and analyzed data. M. Eschborn and M. Lindner wrote the paper. N.S., T.S.-H., C.C.G., B.P., D.B., S.G., K.B.B., J.R., M.S., and M. Eveslage analyzed data. L.K., S.G.M., V.P., T.T., and A.B.-O. designed the study and analyzed and interpreted data. L.K. wrote the paper. H.W. designed the overall study, analyzed and interpreted experiments, and edited the paper. Competing interests: L.K.: compensation for serving on scientific advisory boards (Janssen, Novartis, Merck Serono, and Sanofi Genzyme), speaker honoraria and travel support (Biogen, Novartis, Merck Serono, Sanofi Genzyme, and Teva); research support (Biogen, Merck Serono, and Novartis), and research funding [Deutsche Forschungsgemeinschaft (DFG), German Ministry for Education and Research, Interdisciplinary Center for Clinical Studies (IZKF) Muenster, and Innovative Medical research Muenster]; M. Eschborn: speaker honoraria and travel support (Sanofi Genzyme); M. Lindner, M. Liebmann, B.T.G., V.P., J.B., M.H., A.S.-M., N.F., J.A., G.N., B.P., J.R., S.G., T.W., K.B.B., G.R.C., D.J.M., K.L., M.S., and M. Eveslage: nothing to disclose; C.C.G.: speaker honoraria and travel support (Bayer HealthCare and Sanofi Genzyme); C.J.: travel support (Novartis); T.S.-H.: travel support (Biogen and Novartis); P.H.: speaker honoraria, travel support, and financial research support (Novartis and Merck Serono); D.B.: lectures, congress invitation, and board participation (Bayer HealthCare, Biogen, MedDay Pharmaceuticals, Merck, Novartis, Roche, Sanofi Genzyme, and Teva); N.S.: travel support (Sanofi Genzyme and Novartis); S.G.M.: lecture honoraria and travel support (Almirall, Amicus Therapeutics Germany, Bayer HealthCare, Biogen, Celgene, Diamed, Sanofi Genzyme, MedDay Pharmaceuticals, Merck Serono, Novartis, Novo Nordisk, ONO Pharmaceutical, Roche, Sanofi-Aventis, Chugai Pharma, QuintilesIMS, and Teva) and research funding [German Ministry for Education and Research (BMBF), Deutsche Forschungsgesellschaft (DFG), Else Kröner Fresenius Foundation, German Academic Exchange Service, Hertie Foundation, Interdisciplinary Center for Clinical Studies (IZKF) Muenster, German Foundation Neurology and Almirall, Amicus Therapeutics Germany, Biogen, Diamed, Fresenius Medical Care, Genzyme, Merck Serono, Novartis, ONO Pharmaceutical, Roche, and Teva]; T.T.: employee of Sanofi Genzyme; A.B.-O.: speaking, consultancy fees, and/or grant support (Amplimmune, Biogen, DioGenix, Genentech, Sanofi Genzyme, GSK, Merck/EMD Serono, Novartis, ONO Pharmaceutical, Receptos, Roche, and Teva Neuroscience); H.W.: honoraria for acting as a member of Scientific Advisory Boards and as consultant (Biogen, Evgen Pharma, MedDay Pharmaceuticals, Merck Serono, Novartis, Roche, and Sanofi Genzyme), speaker honoraria and travel support (Alexion, Biogen, COGNOMED, F. Hoffmann-La Roche Ltd., Gemeinnützige Hertie-Stiftung, Merck Serono, Novartis, Roche, Sanofi Genzyme, Teva, and WebMD Global), acting as a paid consultant (AbbVie, Actelion, Biogen, IGES, Johnson & Johnson, Novartis, Roche, Sanofi Genzyme, and Swiss Multiple Sclerosis Society), and research funding [German Ministry for Education and Research (BMBF), Deutsche Forschungsgemeinschaft (DFG), Else Kröner Fresenius Foundation, Fresenius Foundation, Hertie Foundation, NRW Ministry of Education and Research, Interdisciplinary Center for Clinical Studies (IZKF) Muenster, RE Children’s Foundation, Biogen, GlaxoSmithKline GmbH, Roche, and Sanofi Genzyme]. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article