Patients included in the study suffered all from relapsing-remitting MS as defined by the revised MacDonald criteria published in 2005, and the demographic and clinical characteristics are shown in Table 1. Twelve out of 15 MS patients were naïve of any treatment at the time of teriflunomide start. The three other patients were previously treated with an immunomodulatory treatment (Interferon beta, dimethyl-fumarate) that was stopped at least three months before Teriflunomide initiation. Finally, Teriflunomide was started at least 6 months after a relapse for all the patients. The University Hospital Committee and the Committee for the Protection of Patients from Biological Risks approved the study (ABM PFS13-003 “Collection sclérose en plaques”), and all patients were recruited after obtaining informed signed consent.

PBMCs were separated from blood samples collected in EDTA tubes on a Ficoll gradient layer according to the manufacturer’s recommendations and were frozen in autologous serum with 10% DMSO. PBMCs were then thawed in CTL wash supplemented with benzonase (10 U/mL).

The antibodies used for flow cytometry analyses and key reagents are listed in Supplementary Tables 1 and 2.

PBMCs were cultured overnight in complete RPMI medium supplemented with 10% FCS, 100 U/mL penicillin and 100 U/mL streptomycin. CD8 T cells were purified using a human REAlease CD8 MicroBead kit. Naïve (CD45RA⁺CCR7⁺), EM (CD45RA^-CCR7^-) and TEMRA (CD45RA⁺CCR7^-) CD3⁺CD8⁺ cells were sorted from PBMCs by FACS (FACS Aria, BD Biosciences; the purity was greater than 95%).

Two million PBMCs were either directly stained with the “T cell panel” or stimulated with PMA (50 ng/mL) and calcium ionophore (500 ng/mL) for 3.5 h in the presence of brefeldin A (5 µg/mL) before being stained with the “functional panel” (i.e., IL-10, IL-13, IL-17, IFNγ, TNFα). PBMCs were analyzed using an Aurora 5-laser spectral flow cytometer (Cytek). After individual titration of each antibody, a deconvolution matrix was set up according to Cytek recommendations. Quality control procedure was performed prior each data acquisition according to Cytek recommendations.

Supervised and unsupervised (optSNE and FlowSOM) analyses were performed on OMIQ (https://www.omiq.ai/). FCS files were first cleaned using FlowCut (16) and by gating on CD45⁺Singlet⁺LiveDead^-. Non-naïve CD4 T cells and non-naïve CD8 T cells were analyzed separately. optSNE analysis was performed using equal amounts of samples comprising 1x10e5 non-naïve CD4 T cells or 5x10e4 non-naïve CD8 T cells obtained from each FCS pool, with 1000 iterations, a perplexity of 30, and a theta value of 0.5. For the non-naïve CD4 T cells, the expression of the following markers was measured CCR6, CXCR3, CCR4, CD161, CXCR5 and CD28. For the non-naïve CD8 T cells, the expression of the following markers was measured: CCR6, CXCR3, CCR4, CD161, CXCR5 and CD28. For non-naïve CD8 T cells, the expression of the following markers was measured: CD161, CD69, CXCR3, CD57, CD45RA, CD152, CXCR5, CD137, PD-1, TIGIT, CD39, CD194, CCR7, CD49a, TIM3, CD27, LAG3, CD28, KI-67, CD25, CD127, CD73 and CCR6. The resulting optSNE maps were fed into the FlowSOM clustering algorithm (17). For each cell subset, a new self-organizing map (SOM) was generated using hierarchical consensus clustering on the tSNE axes. For each SOM, 12 and 20 metaclusters were identified for non-naïve CD4 T cells and for non-naïve CD8 T cells, respectively.

After FACS sorting, purified CD8 T cell subsets were cultured overnight in TexMACS buffer at 37°C and 5% CO2. The oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured using 4×10⁵ or 2.5×10⁵ purified CD8 T cell subsets and Seahorse XF24 or XF96 analyzers, respectively. The assay was performed in Seahorse XF-base medium supplemented with 10 mM glucose, 2 mM glutamine, and 1 mM pyruvate. A mitochondrial stress assay was performed by successively adding oligomycin (1.5 μM), CCCP (1 μM), and anti-mycin A + rotenone (1 μM each). To assess the OCR and ECAR upon polyclonal stimulation, PMA (50 ng/mL) and ionomycin (500 ng/mL) were added 75′ after the start of the experiment. 2-DG (250 mM) was added before stimulation with PMA-Iono.

MitoTracker green and MitoTracker red were used to quantify mitochondria and to assess the MMP, respectively. PBMCs were first incubated with 100 nM MitoTracker green or MitoTracker red for 30′ at 37°C and 5% CO2, and then, the cells were stained using anti-CD3, anti-CD8, anti-CD45RA, and anti-CD28 antibodies. The samples were immediately analyzed with an LSRII flow cytometer.

PBMCs were stimulated overnight or not with aCD3 (2 µg/mL) and then stained using anti-CD3, anti-CD8, anti-CD45RA, anti-CD28 and anti-Glut1 antibodies according to the manufacturer’s instructions. GLUT1 expression in CD8 T cell subsets was assessed according to mean fluorescence intensity (MFI).

CD8 T cell isolation or FACS-sorted CD8 T cell subsets were plated in 96-U bottom plates coated with anti-CD3 mAb (2 μg/mL) and cultured for 1–5 days in RPMI complete medium with IL-15 (10 ng/mL) or anti-CD28 mAb (2 µg/mL). Proliferation was monitored using diluted cell proliferation dye eFluor450 (CPD) and assessed according to the frequency of CPD^low cells on days 4 and 6. After culturing overnight, activation (CD25 and CD69) and proinflammatory cytokine production (TNFα and IFNγ) of the CD8 T cell subset was determined by cytometry. Teriflunomide (100 µM), uridine (50 µM) and 2-DG (10 mM) were added as indicated.

TEMRA (CD45RA⁺CCR7^-) and EM (CD45RA^-CCR7^-) CD3⁺CD8⁺ cells obtained from MS patients were sorted by FACS and cultured overnight in complete RPMI medium supplemented with IL-15 (2 ng/ml). CD8 T cell subsets stained with the cytosolic Ca²⁺ indicator Fura-2 AM probe (1 µg/mL for 45 min at room temperature) were seeded on a confluent monolayer of primary Human Dermal Microvascular Endothelial [HDMECs that have been previously and widely used as a good model for the in vitro study of inflammatory responses (18)] activated for 24 h with TNFα (100 U/ml) in µ-Slide 8 wells. Teriflunomide (100 µM) or DMSO (negative control) was added after 20 min of cell migration. Cell migration was tracked by time-lapse microscopy through a 20× objective of a Leica DMI 6000B microscope equipped with MetaMorph® software (Molecular Devices), and 120 sequential images were captured in 20-second intervals. The migration paths of individual cells were determined with ImageJ software, TrackMate-extras plugin and in-house scripts (R Studio), and these data were used to calculate average migratory velocity according to the definition of Meyjering (19) during the observation period.

All statistical analyses were performed using GraphPad Prism. Mann-Whitney U tests, Kruskal-Wallis tests followed by Dunn’s post hoc tests, and paired Wilcoxon tests were used as appropriate; the test performed is noted in the figure legends. All p-values are presented as exact values with * indicated as follows: * p<0.05; ** p<0.01; *** p<0.001; and **** p<0.0001.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Using high-dimensional flow cytometry, we investigated the impact of DHODH inhibition in patients with RRMS before and 12 months posttreatment. The clinical characteristics of the patients are depicted in Table 1. We first focused on major lymphocyte populations (Figure 1A). The frequencies of leukocytes among B lymphocytes, NK CD56 cells, TCRγδ T cells and TCRαβ CD4 T cells were similar before and after teriflunomide treatment. Interestingly, a significant decrease in TCRαβ CD8 T cells was observed in patients with RRMS 12 months posttreatment (Figure 1A), resulting in an increase in the ratio of CD4/CD8 TCRαβ T cells (Figure 1A). We then applied high-dimensional flow cytometric analysis to investigate the CD4 T cell subsets. The CD4 T cell regulatory compartment was not impacted by DHODH inhibition, as shown by the similar frequency at baseline and 12 months posttreatment of CD4 regulatory T cells (T_REGS; CD25^highCD127^low), memory CD45RA^-CD4 T_REGS, naïve CD45RA⁺CD4 T_REGS and follicular CXCR5⁺CD4 T_REGS (Figures 1B, C). The expression of functional markers of CD4 T_REGS was also not impacted by DHODH inhibition, as the expression of regulatory associated molecules by CD4 T_REG, such as CTLA-4, LAG-3, and ectonucleotidases CD39 and CD73 CD4 T_REG (20), was similar before and 12 months post treatment (Figure 1C). To gain more insights into the T helper (T_H) and T follicular helper (T_FH) CD4 T cells, we applied global high-dimensional mapping of markers (CCR6, CXCR3, CCR4, CXCR5, CD161 and CD28) related to T_FH, T_H1, T_H2, T_H17, T_H17.1, T_H22 T_H9 and cytotoxic CD4 T cells [identified as described in a consensus paper (21)]. An optimized t-distributed stochastic neighbor embedding (opt-SNE) representation (22) of the data highlighted 12 major populations that were identified by FlowSOM clustering and the comparison of marker expression (Figure 1D). The opt-SNE representation of non-naïve CD4 T cells of patients at M00 was similar to that of patients 12 months posttreatment (Figure 1E) and was also evidenced by the analysis of the frequency of T_H and T_FH CD4 T cell clusters (Figure 1F). The cytokine profiles (T_H1, IFNγ, TNFα; T_H2, IL-13; T_H17, IL-17; and T_REGS, IL-10) produced by TCRαβ CD4 T cells (Figure 1G) and the proliferation of non-naïve TCRαβ CD4 T cells, T_FH cells and T_REGS (% of KI-67⁺; Figure 1H) confirmed the lack of impact of DHODH inhibition on the CD4 T compartment. Collectively, the longitudinal immune monitoring of patients with RRMS showed that DHODH inhibition selectively impacts the CD8 T cell compartment.

Seven major CD8 T cell populations were examined by using the combination of CD45RA, CD27, CCR7, and CD28 cell surface markers to define naïve (CD45RA⁺CD27⁺CCR7⁺CD28⁺), TEMRA (CD45RA⁺CD27^-CCR7^-CD28^-), early-like effector memory (CD45RA^-CD27^-CCR7^-CD28⁺), effector RA- [CD45RA^-CD27^-CCR7^-CD28^- (TE RA^-)], early memory [CD45RA^-CD27⁺CCR7^-CD28⁺ (early)], intermediate memory [CD45RA^-CD27⁺CCR7^-CD28^- (intermediate)], and central memory [CD45RA^-CD27⁺CCR7⁺CD28⁺ (CM)] (Figure 2A) CD8 T cells [as described in a consensus paper (21)]. A nonsignificant increase in naïve CD8 T cells and a nonsignificant decrease in early CD8 T cells were observed in patients with RRMS 12 months posttreatment (Figure 2A). We then applied global high-dimensional mapping of the 27-parameter flow cytometry data to gain more insights into non-naïve αβ CD8 T cells. The opt-SNE representation of the data highlighted 20 populations that were identified by FlowSOM clustering and the comparison of marker expression (Figure 2B). The phenotype of the 20 populations at baseline was highly similar to those at 12 months posttreatment, as shown by the proximity of the 20 clusters after unsupervised clustering (Figure 2B). The opt-SNE representation of non-naïve CD8 T cells of patients at M00 was similar to that of patients at 12 months posttreatment (Figure 2C) and was shown by the analysis of the frequency of the 20 clusters of CD8 T cells (Figure 2C). Interestingly, the secretion of the proinflammatory cytokines IFNγ and TNFα produced by CD8 T cells was reduced in patients 12 months posttreatment (Figure 2D). The impact of DHODH inhibition was restricted to the production of proinflammatory cytokines by CD8 T cells, as the frequencies of IL-10 and IL-13 were similar before and 12 months posttreatment (Figure 2D). The reduced production of proinflammatory cytokines was confirmed in an analysis focused on TEMRA and EM CD8 T cell subsets (Supplementary Figure 1). Finally, a lower frequency of KI-67⁺ CD8 T cells was observed 12 months posttreatment (Figure 2E). The susceptibility of CD8 T cells of patients with RRMS to DHODH inhibition was confirmed in vitro, as the proliferation of EM and TEMRA CD8 T cells in response to TCR and IL-15 stimulation was blunted by teriflunomide, and the inhibition of proliferation can be reversed by added uridine (Supplementary Figure 2). In summary, we demonstrate that DHODH inhibition in patients with RRMS was associated with a lower in vivo proliferation of CD8 T cells with reduced proinflammatory properties 12 months after treatment.

We next tested the impact of DHODH inhibition on the patrolling properties of EM and TEMRA CD8 T cells of with RRMS, two CD8 T cell subsets in which predominant T cell clones were notable among CNS lesions, CSF, and blood CD8⁺ T cells were found (15). The dynamic behavior of TEMRA and EM CD8 T cells on activated endothelial cells was tracked using live-cell imaging (Figure 3). This dynamic model of T cell interaction with the activated endothelium showed that the EM and TEMRA CD8 T cells of patients with RRMS exhibited similar patrolling behavior (Figure 3). Interestingly, the inhibition of DHODH resulted in the immediate and prolonged arrest of the EM and TEMRA CD8 T cells of patients with RRMS, as shown by the significant decrease in their average migratory velocity (Figure 3B). Finally, we assessed whether prolonged in vivo exposure to DHODH inhibition impacts the migratory properties of EM and TEMRA CD8 T cells of patients with RRMS. A nonsignificant decrease in the migratory velocity of the EM CD8 T cells was observed in the longitudinal follow-up of patients with RRMS 12 months after treatment (Figure 3C), whereas the velocity of the TEMRA CD8 T cells was not different before and 12 months posttreatment. Taken together, these results highlight that DHODH inhibition not only impacted the immune function of CD8 T cell subsets but also limited their migratory features.

As DHODH has been shown to interfere with components of the electron transport chain, we wondered whether the high and selective susceptibility of CD8 T cells of patients with RRMS might be linked to impaired CD8 T cell energy metabolism. In the steady state, all CD8 T cell subsets (naïve, EM and TEMRA T cells) in patients with RRMS exhibited a higher mitochondrial mass than those of HVs (Figure 4A) with a higher energy activation in mitochondria, as shown by the mitochondrial membrane potential (MMP) measured using a voltage-dependent MitoTracker probe (Figure 4B). The absence of mitochondrial defects in CD8 T cell subsets in patients with RRMS was confirmed using a Seahorse assay. Basal respiration was similar across the different CD8 T cell subsets (Figure 4C), and as expected, the injection of oligomycin, an inhibitor of ATP synthase, induced a rapid decrease in the oxygen consumption rate (Figure 4C), demonstrating the mitochondrial coupling efficiency. Finally, the protonophore CCCP resulted in an increase in OCR in all CD8 T cell subsets (Figure 4C). Collectively, these results showed that CD8 T cell subsets of patients with RRMS exhibited well-functioning mitochondria under steady state conditions.

The ability to rapidly switch to glycolysis-based metabolism upon stimulation is a hallmark of memory CD8 T cells (23). We showed that, upon stimulation, EM and TEMRA CD8 T cell subsets of patients with RRMS exhibited modified nutrient uptake and metabolism to satisfy their bioenergetic needs, as shown by the increase in their ability to take up glucose (increased expression of glucose transporter 1 GLUT1; Figure 4D) and in the extracellular acidification rate (ECAR; an indicator of glycolysis; Figure 4E) and OCR within minutes following polyclonal stimulation (Figure 4F). As expected, the addition of 2-deoxyglucose (2-DG), which blocks the first step of the glycolytic pathway, prevented both the immediate and sustained increases in the ECAR while not increasing the OCR (Figures 4E, F). The impact of 2DG on glycolysis inhibition resulted in reduced activation of the EM and TEMRA CD8 T cell subsets of patients with RRMS, which showed lower expression of the early activation marker CD69 and the high-affinity IL-2 receptor CD25 and lower TNF α production (Figure 4G). Collectively, these data showed the ability of EM and TEMRA CD8 T cell subsets of patients with RRMS to adapt their metabolism to engage their effector function, and therefore, the selective susceptibility of CD8 T cells of patients with RRMS to DHODH inhibition was not related to impaired CD8 T cell energy metabolism.