Methods

Patients with multiple sclerosis and healthy controls

Human samples and biographic data were available from an observational cross-sectional investigation including 102 multiple sclerosis (MS) patients (31 men, 71 women) as described in (Schmitz et al., 2017; Schmitz et al., 2019). They were consecutively recruited from outpatients and inpatients of the Department of Neurology of the Goethe University Hospital Frankfurt, Germany. Data and blood collection was part of the local bio-banking project (Neurological Department of the Goethe University, Frankfurt). The diagnosis of MS was based on ICD10 criteria. Additional 14 patients with serious disease courses were recruited and observed up to 3.5 years for time course analyses. The patients participated in clinical efficacy studies of fingolimod or natalizumab (NTZ).
To cover the whole period, control samples were analyzed from four consecutive cohorts of healthy subjects (HC). The first encompassed 117 men and 233 women with a mean age of 28 ± 8 years (range 18-57 years). The second were 118 men and 183 women, aged 25 ± 6 years (range 18 - 57 years), the third were 108 m, 217 f with a mean age of 35 ± 12.8 years (range 18-68 years), and the last cohort comprised each 25 men and women above 50 years of age (mean ± SD: 62.9 ± 8.4 years, range 50-79). HC cohorts 1 to 3 were recruited through the Occupational Health Service at the University Hospital of Frankfurt, Germany. HC cohort-4 was recruited from family, friends and colleagues.
For the whole blood assay, venous blood of healthy donors was sampled in K+-EDTA tubes (Tegeder et al., 2008), each split into two samples, one stimulated with 10 µg/ml LPS, the other unstimulated and kept in a 37°C water bath for the indicated times, and biopterin was analyzed in plasma.
The studies were approved by the Ethics Committee of the Medical Faculty of the Goethe University and adhered to the Declaration of Helsinki. Informed written consent was obtained from each participating subject. Venous blood samples were collected to K+ EDTA tubes or in serum tubes and centrifuged at 3000 rpm for 10 min. Plasma and serum were frozen at -80°C up to analysis.

Animals and drug treatments

Female 10-12 weeks old SJL/J mice (Charles River, Germany) were used for study of relapsing-remitting EAE, and C57Bl6/J mice (Charles River, Germany) were used for the study of primary progressive EAE. Mice were housed at 2-4 mice per cage at constant room temperature (21 ± 1 °C) under a regular light/dark schedule with light from 7:00 A.M. to 7:00 P.M. Food and water were available ad libitum .
For the treatment of SJL/J-EAE mice, DAHP (Sigma #D19206; 4 mg/day, ̵̴200 mg/kg/d) or BH4 (Sigma #T4425; 2 mg/day, ̵̴100 mg/kg/d) were dissolved in the drinking water with 2% DMSO. Control animals received the respective vehicle. In C57Bl6/J-EAE mice, BH4 was administered perorally once daily in cornflakes soaked with 10% sucrose/5% ethanol in water. Treatments started at the day of immunization. Control animals received the respective vehicle.
AGMO LacZ reporter mice were used to assess the localization of AGMO in the brain and were created according to EUCOMM gene targeting strategy (Coleman et al., 2015).
The experiments were approved by the local Ethics Committee for Animal Research (Darmstadt, Germany) and adhered to the European guidelines and to those of GV-SOLAS for animal welfare in science and agreed with the ARRIVE guidelines.

EAE model

SJL/J female mice were immunized using the Hooke Kit™ 2110PLP139-151/CFA emulsion PTX (EK-2120, Hooke Labs, St Lawrence, MA), which contains 200 µg of peptide 139-151 of myelin proteolipid protein (PLP) emulsified in 200 µl Complete Freund’s Adjuvant (CFA). The emulsion was injected subcutaneously (s.c.) at two sites followed by two intraperitoneal (i.p.) injections of 200 ng pertussis toxin (PTX) in phosphate buffered saline (PBS), the first 1-2 h after and the second 24 h after PLP139-151.
C57Bl6/J female mice were immunized using Hooke KitTMMOG35-55/CFA emulsion PTX (EK-2110), which contains 200 µg of a peptide 35-55 (amino acids) of myelin oligodendrocyte glycoprotein (MOG) in 200 µl CFA (Hooke Labs, US). Injections of the emulsion and PTX were done as described above.
EAE scores were assessed daily to evaluate the severity and extent of motor function deficits. Score 0, normal motor functions; score 0.5, distal paralysis of the tail; score 1, complete tail paralysis; score 1.5, mild paresis of one or both hind legs; score 2, severe paresis of one or two hind legs; score 2.5, complete paralysis of one hind leg; score 3, complete paralysis of both hind legs; score 3.5, complete paralysis of hind legs and paresis of one front leg.
Blood and tissue samples were obtained at the end of the clinical observation time, 19-22 days after immunization. Blood was collected in K+ EDTA microtubes (EDTA K+Microvette Sarstedt), centrifuged at 3000 rpm for 10 min and stored in standard Eppendorf caps at -80°C until analysis. Tissue samples were snap frozen on dry ice and stored at -80°C until lipid analysis or they were freshly prepared for FACS.

FACS analysis of surface marker proteins

Single cell suspensions were prepared from the spleen, and the lumbar spinal cord. Tissues were rapidly dissected, treated with lysis buffer (DMEM/accutase (PAA) 1:1, collagenase (3 mg/ml, Sigma), DNAse I (1U/ml, Promega)) for 30 min at 37°C, followed by mechanical disruption, which was done by forcing the tissue through a nylon mesh with 70 μm pore size (Cell Strainer, BD). Cell suspensions were mixed with 1 ml erythrocyte lysis buffer for 10 min at room temperature and CD16/32 blocking antibody (Fcγ RII/III receptor blocker, BD) for 15 min on ice. For staining of cell surface antigens, cells were incubated for 20 min at room temperature in staining buffer with the respective fluorochrome labeled antibodies (Suppl. Table 1) and were then counted with a flow cytometer (BD FACS Canto II). FACS scans were analyzed with FlowJo 10.6. For all FACS assays, antibody concentrations followed the recommendations of the manufacturers and the controls were FITC, PE, or APC-conjugated rat IgG.

Immunofluorescence analyses and LacZ histology

A subset of mice were used for histology. Mice were terminally anaesthetized with isoflurane and cardially perfused with cold 0.9% saline, followed by 4% paraformaldehyde (PFA) in 1x PBS for fixation. The lumbar spinal cord was excised, post-fixed in 4% PFA for 2 h, cryoprotected overnight in 20% sucrose at 4 °C, embedded in tissue molds in cryomedium and cut on a cryotome (12 µm). Slides were air-dried and stored at -80°C. After thawing, slides were immersed and permeabilized in 1x PBS with 0.1% Triton-X-100 (PBST), then blocked with 3% BSA in PBST, subsequently incubated overnight with the first primary antibody in 1% BSA in PBST at 4°C, washed and incubated with the secondary fluorochrome-labelled antibody (Alexa488 or Cy3) for 2 h at room temperature. The procedure was repeated for further antibody pairs, followed by 10 min incubation with 1 µg/ml DAPI and embedding in Fluoromount (eBioscience).
For beta galactosidase (LacZ) visualization in tissue sections of AGMO-LacZ reporter mice, cryosections were postfixated for 5 min in 2% PFA, washed in 1x PBS with 2 mM MgCl2 and 3 times in washing buffer containing detergent (1x PBS/2 mM MgCl2 with 0.1 % sodium deoxycholate, 0.02% Nonidet P40, pH 7.5) for 5 min at room temperature. Slides were then incubated at 37°C with the staining solution consisting in 0.5 mg/ml nitrotetrazolium blue chloride (NTB), 5 µg/ml phenazine methosulfate (PMS) in detergent washing solution. The incubation time was adjusted to the tissue. The reaction was stopped by washing the slides 3x in 1x PBS for 5-10 min. Slides were counter-stained with eosin, dehydrated in increasing ethanol concentrations and xylene before embedding in Pertex mounting medium.
Tiled images were captured (10x objective lens) on an inverted fluorescence microscope (BZ-9000, KEYENCE, Germany), and were stitched using the Keyene’s software to cover the complete spinal cord. Filter and acquisition parameters were set to assure comparability. Subsequently, higher magnification images (20x objective lens) of the grey-to-white matter border were obtained of various regions. CD3+ T-cells were quantified using the particle counter plugin of FIJI ImageJ after background subtraction and threshold setting according to the IJ IsoData algorithm implemented in FIJI. Zoom-in images (5x) were created from regions of interest. The area covered by immunoreactive cells relative to the total area (which was identical in all images) was used to assess treatment effects. Three or more non-overlapping images of three mice were analyzed per group.

Culture of splenocytes

Spleen tissue was rapidly dissected, treated with lysis buffer (DMEM/accutase (PAA) 1:1, collagenase (3 mg/ml, Sigma), DNAse I (1U/ml, Promega)) for 30 min at 37°C, followed by mechanical disruption, which was done by forcing the tissue through a nylon mesh with 70 μm pore size (Cell Strainer, BD). Cell suspensions were washed, resuspended in PBS and the cell number was counted with a Neubauer chamber. 5 x 105 cells were plated, cultured in RPMI1640-GlutaMax medium (GibcoTM, Life technologies) and restimulated with 25 ng/ml IFNγ for 24 h.

Griess assay of nitric oxide

The concentration of nitrite/nitrate was determined with the Saville-Griess assay adapted for microtitre plates. A standard curve was prepared with serial dilutions (0–50 µM) of a freshly prepared sodium nitrite (NaNO2) stock solution (100 mM). Cells were homogenized in 1x PBS and, after centrifugation, 200 µl of the supernatant were added to a well of a 96-well plate. 50 μl of sulfanilamide solution (4mg/ml in 1N HCl) were added to standards and samples. After 2 min incubation, 50 μl of N-(naphtyl)-ethylenediamine dihydrochloride solution (6 mg/ml in H2O) were added, followed by incubation for 5 min at room temperature and measuring absorbance at 540 nm with a Specta Fluor Plus® instrument and XFluor® software (Tecan, Crailsheim).

Analysis of lipid signaling molecules

Bioactive lipids including sphingolipids and ceramides, lysophosphatidic acids and endocannabinoids were analyzed by liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) as described in detail in the supplementary material of (Brunkhorst-Kanaan et al., 2019). All analytical methods were optimized based on previous methods (Brunkhorst-Kanaan et al., 2019; Schmitz et al., 2017; Zschiebsch et al., 2016).
In brief, the analytes were extracted using liquid-liquid-extraction. Sample volumes were 10 µl for sphingolipids, 50 µl each for LPA and 100 μl for endocannabinoids. The quantification of all analytes was performed using a hybrid triple quadrupole-ion trap mass spectrometer QTRAP 5500 or 6500+ (Sciex, Darmstadt, Germany) equipped with a Turbo-V-source operating in positive ESI mode for sphingolipids and endocannabinoids and in negative ESI mode for LPA.
Sphingolipids were separated using an Agilent 1200 HPLC system equipped with a Zorbax C18 Eclipse Plus UHPLC column (50 × 2.1 mm, 1.8 μm, Agilent technologies, Waldbronn, Germany) and the analysis of LPA was done on the same HPLC system using a Luna C18 column (50 × 2 mm, 5 μm, Phenomenex, Aschaffenburg, Germany). Analysis of the endocannabinoids was done using an Agilent 1290 Infinity I UHPLC system equipped with an Acquity UPLC BEH C18 UPLC column (100 × 2.1 mm, 1.7 μm, Waters, Eschborn, Germany).
Quality control samples of three different concentration levels (low, middle, high) were run as initial and final samples of each run. For all analytes, the concentrations of the calibration standards, quality controls and samples were evaluated by Analyst software 1.6.3 and MultiQuant software 3.0.2 (Sciex) using the internal standard method (isotope-dilution mass spectrometry) as described in (Gurke et al., 2019). Variations in accuracy were less than 15% for at least 67% of all QC samples. For the lower limit of quantification, a variation of 20% was accepted.

Untargeted lipidomic analyses

Twenty microliter plasma or 40 µl lymph nodes homogenates (homogenated in 0.025 µg/ml water:ethanol 1:3 (v/v)) were extracted using methyl-tert-butyl-ether (Matyash et al., 2008). The organic phase was split into two aliquots, one for analysis in negative ion mode and the other in positive ion mode. After drying under a nitrogen stream at 45°C, the aliquots were reconstituted in 120 µl methanol or stored at -40°C until analysis. LC-MS analysis was performed on a Nexera X2 system (Shimadzu Corporation, Kyoto, Japan) coupled to a TripleTOF 6600 (Sciex). The chromatographic separation was done on a Zorbax RRHD Eclipse Plus C8 1.8 µm 50x2.1 mm ID column (Agilent, Waldbronn, Germany) with a SecurityGuard Ultra C8 pre-column (Phenomenex, Aschaffenburg, Germany), using a binary gradient with 40°C column temperature and a flow rate of 0.3 ml/min. For the positive mode, the mobile phase A consisted of 10 mM ammonium formate and 0.1% formic acid in water and mobile phase B of 0.1% formic acid in acetonitrile: isopropanol 2:3 (v/v). For measurement in negative mode 1 mM ammonium formate and 0.1% formic acid in water was used as for mobile phase A. The MS analysis encompasses a TOF MS Scan from 100-1000 m/z with six data dependent acquisitions per cycle and a mass range of 50 – 1000 m/z. The identification of the lipid species was based on the exact mass (+/- 5 ppm), the isotope ratio and the comparison of the MS/MS spectra with the reference spectra according to LIPID MAPS (http://www.lipidmaps.org), METLIN (http://metlin.scripps.edu) or the Human Metabolome Database (HMDB, version 4.0).
To reduce the impact of small variations in instrument sensitivity during the measurements all samples were randomized prior to analysis. Quality control samples were injected at the start and at the end of a run and after every 10th sample to verify system stability. Data evaluation was done with Analyst TF 1.7 and MultiQuant software 3.0, and peak areas were normalized to the quality control samples using median peak ratios by MarkerView software 1.2 (all Sciex).

Microarray data analysis

Microarray data of GEO dataset GSE60847 (own previous data) were reanalyzed and searched for genes involved in lipid metabolisms, regulation or function. Normalized data were analyzed with ArrayStar, which uses general linear models to assess differential expression. Data were log2 transformed, scored according to ”fold-regulation”, P-value and abundance and top scored genes were then clustered using Euclidean distance metrics. Valid genes are displayed as Volcano plots, showing the log2 difference i.e. fold change (positive for upregulated genes and negative for downregulated genes) versus the –log10 of the t-test P value. The P value was set at 0.05 and adjusted according to Benjamini Hochberg. Genes were text-filtered based on gene descriptions and GO ontology terms to find lipid regulating and metabolizing genes and genes involved in BH4 pathways (synthesis, recycling and coenzyme functions).

Statistics

Group data are presented as mean ± SD or median ± IQR for non-parametric data as specified in the respective figure legends. Data were analyzed with SPSS 24 and Graphpad Prism 8.0 and Origin Pro 2020. Data were mostly normally distributed, or log-normally distributed. For testing the null-hypothesis that groups were identical, two groups were compared with 2-sided, unpaired Student’s t-tests. The Mann Whitney U test (2 groups) or Kruskal Wallis (> 2 groups) were used as non-parametric alternatives in case of violations of t-test requirements. Time course data or multifactorial data were submitted to 2-way analysis of variance (ANOVA) using e.g. the factors ’time’ and ’genotype’. In case of significant differences, groups were mutually compared at individual time points using post hoc t-tests according to Dunnett, i.e. versus the control group, or according to Šidák. For time courses of non-parametric scores, the Friedmann test was used. Asterisks in figures show multiplicity-adjusted P-values.
Multivariate analyses of multiple lipid classes were used to reduce the dimensionality. Because raw lipid concentrations of different classes differ by several orders of magnitude, lipids were normalized and are expressed as percentage of the 90%-quantile. Canonical discriminant analysis (CanDisc) was employed to separate treatment groups and to assess the predictability of group membership. Partial least square (PLS) analysis was used if analytes exceeded the number of samples per group. Score plots and 95% confidence ellipses were created in OriginPro. Untargeted lipidomic data (normalized peak areas) were log2 transformed. Volcano plots were used to show the log2 difference (fold difference) versus the –log10 of the t-test P-value. Lipids of interest were further analyzed using 2-way ANOVAs for ”lipid X treatment”, and subsequent t-test for ”treatment”.