Mice

8–10 weeks old SJL and C57Bl/6 mice were purchased from Charles River Laboratories and used for all experiments. All animal experiments were supervised by the local Institutional Animal Care Committee (OPBA) of the University of Verona and were conducted following the principles of the European Union’s Directive 2010/63.

PTTH preparation

D-pantethine (PTTH) was purchased from Sigma-Aldrich. For in vitro experiments, PTTH was dissolved in phosphate-buffered saline (PBS) to a final concentration of 100 mM and stored at − 20 °C before use. For in vivo experiments, PTTH was dissolved in physiologic saline solution (0.9% NaCl in water solution) to a final concentration of 300 mg/ml and stored at − 20 °C before use.

Production of PLP139-151-specific T cell lines

SJL mice were immunized subcutaneously (s.c.) with 300 μg of proteolipid protein (PLP)139–151 peptide (Genscript, USA) in 200 μl emulsion consisting of equal volumes of PBS and complete Freund’s adjuvant (CFA, Difco Laboratories), supplemented with 1 mg/ml of Mycobacterium tuberculosis (strain H37Ra; Difco Laboratories). 10–12 days later, draining lymph nodes (LNs) were collected and total cells were cultured in the presence of 30 μg/ml of PLP139-151 peptide for 4 days in complete medium [RPMI 1640 (Corning) supplemented with 1 mM sodium pyruvate (Sigma Aldrich), 4 mM GlutaMAX-I® supplement (Thermo Fisher Scientific), 100 U/ml penicillin/streptomycin (Sigma Aldrich), 10% v/v fetal bovine serum (FBS; Thermo Fisher Scientific) and 5 μg/ml plasmocin (InvivoGen)]. After 4 days, cells were collected, washed, and re-plated in fresh complete medium. PLP139-151-specific T cell lines were obtained by re-stimulation of these cultures every 14 days for at least 3 times in the presence of irradiated splenocytes as antigen presenting cells (APCs - ratio 1:8) and 30 μg/ml of PLP139-151 peptide. Cells collected 2 days post stimulation (dps) were considered as “actively-proliferating”, while those collected 10–12 dps were considered as “non-proliferating”.

Metabolomics analysis

Resting total CD4+ T cells were enriched via negative selection from LNs and spleens of 33 SJL mice using the CD4+ T cell isolation kit, mouse (Miltenyi Biotech), following manufacture instructions, and divided in four batches. PLP139-151-specific T cells were obtained from four independent actively-proliferating encephalitogenic T cell lines. Resting CD4+ T cells and PLP139-151-specific T cells were collected, washed twice with PBS, and pelleted. Cell pellets were immediately frozen in liquid nitrogen and stored at − 80 °C. In some experiments, resting CD4+ T cells and actively-proliferating encephalitogenic T cells were washed with PBS, re-suspended in fresh medium and treated with PBS or PTTH 1.0 mM for 6 h or 12 h before freezing. Total CoA levels were measured using the CoA assay kit (Abcam), following manufacturer’s instructions. Metabolite identification and quantification in T cells were performed in outsourcing by Metabolon (http://www.metabolon.com/). Metabolomics data visualization and statistical analysis were performed with R v4.3.1 [42] (using the packages readxl, openxlsx, stringr, plyr, dplyr, tidyr, tibble, reshape2, car, missMDA, colorspace, RColorBrewer, ggplot2, ggpmisc, ggrepel, ggforce, grid, scales, mixOmics, memoise, igraph, metap, MetaboAnalystR 4.0.0, nlme, emmeans) and TIBCO Spotfire v14.0.0 (TIBCO, Palo Alto, CA). Metabolites with less than < 30% missing data in all groups were used for multivariate analysis and univariate analysis (MVA and UVA, 137 metabolites; Supplementary data 1). Metabolites with data not missing at random were used only for univariate analysis in specific groups (66 metabolites). Six metabolites with very low signal intensities or high percentage of missing data were excluded from analysis. Samples were normalized to the median intensity of all MVA metabolites (metabolites before scaled 0 to 1) within the sample to correct for cell number differences. Median normalized intensities were log10-transformed and all normalized, transformed metabolites were normally distributed (Kolmogorov Smirnov test p > 0.05) and homoscedastic (Brown-Forsythe Levene-type test p > 0.05). All samples were checked for analytical outlier behavior according to the sample median, protein content and position in principal component analysis (PCA) excluding the first sample (UNVE-00062) from further analysis.

PCA was performed centered and scaled using the inbuilt default missing imputation with mixOmics::pca [44]. For partial least-squares discriminant analysis (PLS-DA), missings were imputed with missMDA::imputePCA [21] and data was centered and scaled for mixOmics::plsda [44]. Significance of group differences was tested by receiver operating characteristic (ROC) with mixOmics::auroc [44] using the first two components delivering area under the curve (AUC) of each group against all other groups and p-values from a Wilcoxon test between the predicted scores.

For univariate analysis, three factors had to be considered: CELL , TREATMENT and incubation TIME . The analysis was performed in a way that same CELL and TREATMENT conditions are considered as one group, irrespective of incubation TIME since the metabolomes from different incubation TIME were highly similar to each other in contrast to the two other biological factors in both PCA and PLS-DA. To investigate and correct for the potential additional variability contributed by the incubation TIME, linear mixed models with the random factor ~ 1|TIME and the fixed factors CELL and TREATMENT, as well as their interaction, were fitted with nlme::lme. To avoid potential overfitting, additionally generalized least squares models for only the fixed factors CELL and TREATMNT without random factor were fitted as well with nlme::gls [40]. Model performance was compared within each metabolite for lower Akaike information criterion, higher log-likelihood (goodness of fit), significance in log likelihood ratio test and direct comparison of fitted values. The model CELL*TREATMENT ~ 1|TIME showed the best fit for most metabolites and results are reported throughout. Results of pairwise comparisons of interest were extracted with emmeans::emmeans and p-values were False Discovery Rate (FDR) multiple test adjusted. For UVA graded metabolites with specific missing group, accordingly simplified models were used (see Supplementary Data 1).

For quantitative enrichment analysis (QEA) with MetaboAnalystR [35], all included metabolites were manually matched to their respective CAS, PUBCHEM ID, HMDB ID, KEGG ID and HMDB conform compound names using said databases and the online MetaboAnalyst Compound ID Conversion tool. Missing data was imputed with ImputeMissingVar(…, method = “knn_var”), data was autoscaled with Normalization(…, “AutoNorm”) and all measured metabolites were set as specific background calculating the global test result against SMPDB and RaMP pathways with CalculateGlobalTestScore [20]. For pathway topological analysis against KEGG pathways, the global test without-degree centrality was calculated with CalculateQeaScore [20]. All pathway results can be found in Supplementary Data 1.

Phosphoproteomics and network analysis

After 6 h of treatment with PBS or PTTH 1.0 mM, actively-proliferating PLP139-151-specific T cells were washed twice with PBS and pelleted. Pellets were immediately frozen in liquid nitrogen and stored at − 80 °C. Analysis of protein expression and phosphorylation levels were performed in duplicate in outsourcing by Kinexus (http://www.kinexus.ca), using a Kinex™ KAM-850 Antibody Microarray Kit. The Kinex™ Antibody Microarray Service monitored changes in the expression levels and phosphorylation states of signaling proteins with more than 850 antibodies, which includes approximately 517 pan-specific antibodies (for protein expression evaluation) and 337 phospho-site-specific antibodies (for phosphorylation states evaluation). For each antibody, the background-corrected raw intensity data were logarithmically transformed with base 2. Furthermore, Z scores were calculated by subtracting the overall average intensity of all spots within a sample from the raw intensity for each spot, and dividing it by the standard deviations (SD) of all of the measured intensities within each sample [13]. Z ratios were further calculated by taking the difference between the averages of the observed protein Z scores and dividing it by the SD of all of the differences for comparisons between actively-proliferating PLP139-151-specific T cell treated or not with PTTH. A Z ratio of ± 1.1 was inferred as significant and used for bioinformatics analysis.

Gene Onthology (GO) enrichment analysis was performed by using GOnet [41]. The analysis was done on significant up- or down-regulated proteins. Only terms with p-value < 0.001 and FDR < 0.05 were considered. A bioinformatic probe, only including the set of eight down-regulated proteins, was extrapolated from the previous analysis and used as input for the subsequent topological network analysis. Topological network analysis was performed by using Cytoscape network analysis package (https://cytoscape.org). A global human PPI interactome dataset consisting of 19,439 proteins and 766,811 undirected binary interactions was compiled by integrating four global PPI datasets (http://iid.ophid.utoronto.ca, https://thebiogrid.org; https://cbdm-01.zdv.uni-mainz.de/~mschaefer/hippie/; https://apid.dep.usal.es). The compiled dataset included only experimentally determined interactions. Protein and interaction duplicates, and self-loops were removed with Cytoscape core tools. All protein IDs were normalized to HGNC nomenclature (https://www.genenames.org). A first interrogation of the global human PPI interactome dataset with the set of eight down-regulated proteins led to a first neighbors (FN) network of 1764 signaling proteins and 82,221 binary interactions. A topological network analysis procedure was then applied to reduce the complexity of the FN interactome. Four topological indexes of node centrality (centroid, betweenness, bridge and degree) and heat diffusion were computed by using the Cytoscape app Centiscape. The selection of proteins having topological indexes equal to or higher than 50% of the average led to a subnetwork of 487 proteins and 1587 interactions. Further interrogation of this sub-network with the set of eight down-regulated proteins coupled to Signor edge direction enrichment performed by using the Cytoscape core tools, led to a restricted directed signal transduction sub-network consisting of 51 signaling proteins and 105 interactions. Final network modularization with MCODE led to the identification of a highly interconnected directed signaling module of 6 proteins (all mitogen-activated protein kinases—MAPKs) and 15 interactions. Reactome pathway enrichment was then performed on these two signaling networks.

Proliferation assays 1.

In vitro proliferation assays. In vitro proliferation assays were performed by using non-proliferating PLP139-151-specific T cells (see above). Cells were incubated for 16 h with PBS or two different concentrations of PTTH (0.1 mM and 0.5 mM). After treatment, cells were washed and restimulated in 96-well plates in the presence of irradiated splenocytes as APCs (ratio 1:8) and increasing concentrations of PLP139-151 peptide (0, 10, 20, or 30 μg/ml). After 48 h, cells were pulsed with 1 μCi/well of 3H-thymidine (dT) and were left in culture for further 18 h.

2.

Ex vivo proliferation assay. Draining LNs cells were collected from vehicle- or PTTH-treated SJL EAE mice at disease peak. 0.4 × 106 total LN cells were re-stimulated in a 96-well plate with PLP139-151 peptide 30 μg/ml. After 48 h, cultures were pulsed for 18 h with 1 μCi/well of 3H-dT.

For both 1. and 2., samples were harvested after 18 h and supplemented with 3 ml of scintillation fluid (Ultima Gold from Perkin-Elmer). 3H-dT incorporation by proliferating cells was measured with a β-counter (Perkin-Elmer) and quantified as counts per minute (CPM). For each experiment, a 96-well plate was seeded for 3H-dT incorporation assay as explained before, and another identical one, but without 3H-dT, for supernatant collection for cytokine quantification.

Bio-Plex assay for cytokines detection

Supernatants collected from in vitro and  ex-vivo proliferation assays were used for Milliplex cytokine assays (Merck Millipore), following manufacturer’s instructions. Briefly, anti-cytokine conjugated beads were plated in 96-well microtiter plates and then removed by vacuum filtration. Samples were then added, and the plate was incubated for 30 min by mixing at 300 rpm. Bio-Plex cytokine assays were sequentially incubated with the detection antibody and streptavidin-PE; samples were then analyzed immediately using the Bio-Plex array-system. Cytokine concentrations were automatically calculated by Bio-Plex software using a standard curve derived from recombinant cytokine standards.

In vitro adhesion assays on purified integrin ligands

12-well glass slides were coated for 18 h at 4 °C with purified mouse intercellular cell adhesion molecule (ICAM)-1 or vascular cell adhesion molecule (VCAM)-1 (R&D Systems), 1 μg/ml in PBS. Slides were then blocked with FBS for 10 min at 37 °C. Activated PLP139-151-specific T cells were pre-treated for 6 h with PTTH 0.5 mM or 1.0 mM. Cells were then re-suspended at 5 × 106/ml in standard adhesion buffer. 20 μl of cell suspension were added to each well, and cells were left spontaneously adhere on VCAM-1 or ICAM-1 for 20 min at 37 °C. After washing, adherent cells were fixed in glutaraldehyde 1.5% in PBS and counted by computer-assisted enumeration [9].

Intravital microscopy in brain pial venules

Intravital microscopy experiments were performed in inflamed brain microcirculation as previously described [39, 46]. Wild-type C57Bl/6 mice were injected intraperitoneally with 12 μg lipopolysaccharide (LPS) (Sigma-Aldrich) 5–6 h before starting the experiment. Animals were anesthetized and a heparinized PE-10 catheter was inserted into the right common carotid artery toward the brain. Blood vessels were visualized through the bone using fluorescent dextran as described [39, 46]. 2–3 × 106 activated PLP139-151-specific T cells were treated with vehicle or PTTH 1.0 mM for 6 h, labeled with CMFDA (5-chloromethylfluorescein diacetate) or CMTMR (5-(and-6)-(((chloromethyl)benzoyl)amino)tetramethylrhodamine) (Thermo Fisher Scientific), and injected into the carotid artery by a digital pump. Hemodynamic parameters were determined as described [39, 46]. Images were visualized with a silicon-intensified target video camera (VE-1000 SIT; Dage-MTI) and a Sony SSM-125CE monitor. Digitalized video images were analyzed frame by frame using the ImageJ software. Lymphocytes that remained stationary on the venular wall for ≥ 30 s were considered firmly adherent. Rolling and firm arrest fractions were determined as the percentage of cells that rolled or firmly arrested within a given venule on the total number of cells entering the venule.

Differentiation and analysis of murine Th0, Th1, Th17 and Th2 cells

Resting CD4+CD62L+ T cells were isolated in a two-step separation from the spleens of 6–8 weeks old C57Bl/6 J mice, using the CD4 + T Cell Isolation Kit and the CD62L (L-selectin) MicroBeads (both from Miltenyi Biotech), following manufacturer’s instructions. T cells were incubated for 8 h in complete medium with vehicle (PBS) or PTTH 0.5 mM. Cells were then washed and seeded on plate-bound anti-CD3/anti-CD28 antibodies (1 µg/ml and 2 µg/ml, respectively—BioLegend) in complete medium. The following cytokines and blocking antibodies were also added to induce Th0, Th1, Th2 or Th17 cells, respectively: no cytokines for Th0 cells; interleukin-12 (IL-12) 20 ng/ml + anti-IL-4 antibody 5 µg/ml (BD Pharmingen) for Th1 cells; IL-4 40 ng/ml + anti-interferon gamma (IFN-γ) antibody 5 µg/ml (BD Pharmingen) for Th2 cells; IL-1β 10 ng/ml + IL-6 50 ng/ml + IL-23 10 ng/ml + transforming growth factor-beta (TGF-β) 0.5 ng/ml + anti-IL-4 and anti-IFN-γ antibodies 5 µg/ml for Th17 cells. Cytokines were purchased from ImmunoTools (IL-1β, IL-6 and TGF-β), Miltenyi Biotech (IL-12 and IL-23) or BioLegend (IL-4). For cell proliferation analysis, cells were stained with the fluorescent dye CellTraceViolet (CTV, Thermo Fisher Scientific) before activation, following manufacturer’s instructions. After 3 days, CTV-stained cells were collected, washed, and acquired. To analyze cytokine production, cells were re-stimulated for 5 h with brefeldin A (BFA) 10 µg/ml, ionomycin 1 µg/ml and phorbol-12-myristate-13-acetate (PMA) 50 ng/ml. Cells were then fixed and permeabilized with fixation buffer and intracellular staining permeabilization wash buffer (BioLegend) and stained with the following antibodies: APC anti-mouse IFN-γ, APC/Cy7 anti-mouse tumor necrosis factor alpha (TNF-α) and PE/Dazzle 594 anti-mouse granulocyte–macrophage colony-stimulating factor (GM-CSF) antibodies (Th0 cells); APC anti-mouse IFN-γ antibody (Th1 cells); APC anti-mouse IL-13 antibody (Th2 cells); PE anti-mouse IL-17A and Alexa647 anti-mouse IL-17F antibodies (Th17 cells) (all antibodies from BioLegend). All samples were acquired on a BD LSRFortessa and analyzed using FlowJo (BD Biosciences).

EAE induction and in vivo treatment with PTTH

For relapsing–remitting (RR)-EAE, SJL female mice were immunized s.c. with 300 μg of PLP139-151 peptide in 200 μl emulsion as described above. For chronic EAE, C57Bl/6 female mice were immunized s.c. with 300 μg myelin oligodendrocyte glycoprotein (MOG)35–55 peptide in 200 μl emulsion. In both models, 20 ng of pertussis toxin were injected intraperitoneally (i.p.) at the day of immunization and after 48 h. To evaluate the clinical benefit of PTTH, mice were treated i.p. with 30 mg/mouse/day of PTTH or saline (0.9% NaCl; control animals). In the pre-clinical setting, mice received a daily injection of PTTH or corresponding volume of saline starting 5 days post-immunization, before disease onset. In the therapeutic protocol, PTTH or saline were administered i.p. after the initial peak of the disease (day + 18 post-immunization) in the RR-EAE model, whereas treatment started at the beginning of the chronic phase (day + 23 post-immunization) in the chronic EAE model. In both pre-clinical and therapeutic settings, mice were treated for 20 consecutive days. Clinical score was recorded daily as previously described [2].

Neuropathology

SJL mice from pre-clinical treatment experiments were sacrificed at disease peak and perfused with PBS and 4% paraformaldehyde (PFA, Vetrotecnica). Lumbar spinal cords (SCs) were collected, put overnight in PFA 4%, and for a further overnight in a 30% saccharose water solution. SCs were finally embedded in tissue freezing medium (Leica Biosystems), frozen in dry ice, and stored at − 80 °C. 10 µm sections were histologically processed using standard hematoxylin/eosin (H/E) staining for detection of inflammatory infiltrates. Images were acquired using the Axio Imager Z2 (Zeiss, Germany) and quantification of infiltrated areas was performed on every fifth section of the SCs using the ImageJ software.

Differentiation and analysis of human Th1, Th17 and induced regulatory T (iTregs) cells

Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation with Lymphoprep (Proteogenix), from buffy coats obtained from the Transfusion Medicine Department of the Medical University of Graz. Naive CD4+ T cells were purified from PBMCs using the MagniSort human CD4 naive enrichment kit (Thermo Fisher Scientific), following manufacturer’s instructions. T cells were activated in vitro for 5 days in X-VIVO 15 medium (Lonza) with plate-bound anti-CD3/anti-CD28 antibodies (1 µg/ml and 2 µg/ml, respectively—BioLegend), in the presence of vehicle (PBS) or PTTH 1.0 mM. The following cytokines were also added to induce Th1, Th17, or iTreg cells, respectively: IL-12 10 ng/ml for Th1 cells; IL-1β 10 ng/ml + IL-6 20 ng/ml + IL-23 100 ng/ml + TGF-β 1 ng/ml for Th17 cells; IL-2 10 ng/ml + TGF-β 5 ng/ml for iTregs cells. Cytokines were purchased from ImmunoTools (IL-6, IL-23, TGF-β and IL-2), Peprotech (IL-1β) or R&D Systems (IL-12). For cell proliferation analysis, cells were stained with CTV, collected after 5 days of culture, washed, and acquired. To analyze cytokine production, cells were re-stimulated with PMA/BFA/ionomycin, fixed and permeabilized as described above for mouse T cells. Cells were then stained with the following antibodies: Alexa488 anti-human IFN-γ and PE/Dazzle 594 anti-human GM-CSF antibodies (Th1 cells); APC/Cy7 anti-human IL-17A and PE/Dazzle 594 anti-human GM-CSF antibodies (Th17 cells) (all antibodies from BioLegend). To analyze Treg induction, cells were collected after 5 days of culture and stained with the following antibodies: PE anti-human CD25 and Alexa647 anti-human-CD127 (both antibodies from BioLegend). Cells were then fixed and permeabilized with the eBioscience™ Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) and stained with the Alexa488 anti-mouse/rat/human Forkhead box P3 (FOXP3) antibody (BioLegend). All samples were acquired on a BD LSRFortessa and analyzed using FlowJo.

MS patient recruitment and analysis

All participants included in this study underwent clinical examination at the MS outpatient clinic of the Department of Neurology, Medical University of Graz. Participants (n = 15; Table 1) met the following criteria: (a) age > 18 years at baseline; (b) diagnosed with clinically isolated syndrome (CIS) suggestive of MS or with MS according to diagnostic criteria applicable at the time of presentation [12], (c) did not receive disease-modifying treatment and/or cortisone for at least 6 months prior to blood collection. PBMCs were isolated from 24 ml blood by density centrifugation, washed, counted, and resuspended in RPMI medium + FBS:DMSO (15 ml:3.7 ml) mixture. Cells were then divided in cryogenic vial (5–8 × 106 cells/vial) and gradually frozen at − 80 °C for at least 24 h. Cell vials were finally transferred into a cryo tank (liquid nitrogen vapor phase) and stored until use. At the time of the experiments, PBMCs were thawed, washed, counted, and recovered in X-VIVO medium for 1 h at 37 °C. Cells were resuspended in fresh X-VIVO medium and activated with plate-bound anti-CD3/anti-CD28 antibodies as above, in the presence of PBS or PTTH 1.0 mM. After 5 days, cells were collected, washed, re-stimulated as above with BFA/PMA/ionomycin, fixed, permeabilized, and stained with Alexa488 anti-human IFN-γ + PE/Dazzle 594 anti-human GM-CSF + APC/Cy7 anti-human IL-17A + BV421 anti-human IL-10 (BioLegend) + APC anti-human TNF-α (BioLegend) antibodies. Samples were acquired and analyzed as above.

Table 1 Demographic and clinical data of MS patients recruited in the studyStatistics

Statistical analyses for metabolomics and phosphoproteomics studies were performed as described above. For in vitro, ex-vivo, and in vivo experiments, statistical analyses were performed using GraphPad Prism 10 (GraphPad Software Inc.). Data are presented as mean values ± standard deviation (SD) or standard error of the mean (SEM). The statistical tests performed for each experiment are reported in the figure legend, and a p-value < 0.05 was considered significant.