Monocyte infiltration induces CNS arginine catabolism to fuel neuroinflammation

Human samples and neuropathological assessment

All procedures done on human postmortem tissues in this study were approved by the Ethics Committee of Medical University of Vienna (votes: 1636/19 and 1067/2024). Tissue samples were acquired from the tissue bank of the Division for Neuropathology and Neurochemistry at the Medical University of Vienna. MS was diagnosed either during the individual’s life or via autopsy and neuropathological examination. The mean age was 52 ± 15 years (n = 5 men and n = 13 women). Disease clinical course was classified as acute (n = 3), relapsing and remitting (n = 5), secondary progressive (n = 4) and primary progressive (n = 2). In some cases, the definite clinical course was not available; these cases were designated as unknown (n = 4). Control individuals showed an absence of major pathological alterations; however, minor vascular pathology in the form of small vessel disease was frequently detected. Only white matter sections without pronounced pathology were selected. The mean age for control individuals was 63 ± 15 years (n = 4 men and n = 3 women). Participant characteristics are summarized in Supplementary Table 7.

Experimental animals

Mice were bred and housed in pathogen-free facilities at the Medical University of Vienna and were maintained on a 12-h light/12-h dark cycle at 21–23 °C and 45–65% humidity. C57BL/6J mice were obtained from the Animal Core Facility of the Medical University of Vienna or were purchased from Janvier (SC-C57J-M). Csf2rb^fl/flCcr2-Cre^{ERT2(−mKate2)} mice⁴⁹ were kindly provided by B. Becher (Institute of Experimental Immunology, University of Zürich, Switzerland) and were further crossed to mice expressing a CAG-loxP-STOP-loxP-tdTomato cassette in the Rosa26 locus (B6;129S6-Gt(Rosa)26Sor^{tm14(CAG−tdTomato)Hze/J}, The Jackson Laboratory, stock 007908, R26^tdTomato), which were provided by C. Österreicher (Center for Physiology and Pharmacology, Medical University of Vienna, Austria)⁸⁰, to generate R26^tdTomatoCcr2-Cre^ERT2 fate-mapping animals. R26^tdTomatoMs4a3-Cre mice were generated by crossing R26^tdTomato mice with Ms4a3-Cre mice (C57BL/6J-Ms4a3^em2(cre)Fgnx/J, The Jackson Laboratory, 036382). Arg1-eYFP reporter mice (B6.129S4-Arg1^tm1.1Lky/J, The Jackson Laboratory, 0015857) were crossed with R26^tdTomatoCcr2-Cre^ERT2 mice to generate R26^tdTomatoCcr2-Cre^ERT2Arg1-eYFP animals. Arg1^fl/fl mice (C57BL/6-Arg1^tm1Pmu/J, The Jackson Laboratory, 008817) were a kind gift of P. J. Murray (Immunoregulation Research Group, Max Planck Institute of Biochemistry, Germany)⁸¹ and were crossed with Cx3cr1-Cre mice (B6J.B6N(Cg)-Cx3cr1^{tm1.1(cre)Jung}/J, The Jackson Laboratory, 025524) or Ccr2-Cre^ERT2 mice to generate Arg1^fl/flCx3cr1-Cre or Arg1^fl/flCcr2-Cre^ERT2 mice, respectively. The presence of respective transgenes was confirmed by PCR analysis of DNA from ear biopsies of genetically modified animals using the following primer pairs (Microsynth): Csf2rb: 5’-GAGAGAGGGTCCTTTTGGTC-3’ (forward) and 5’-CCTCCCCTCTTCTGTATCTTC-3’ (reverse); Ccr2-Cre^ERT2-mKATE: 5’-CTCTACTTCATCGCATTCCTTGC-3’ (forward) and 5’-GGTTGATGAAGGTTTTGCTGC-3’ (reverse) or 5’-AGAAAGTGAGCCCTCTGTATGG-3’ (forward) and 5’-TTGGCATTTCCTGGTGAGC-3’ (reverse); Cx3cr1-Cre: 5’-TCGCGATTATCTTCTATATCTTCAG-3’ (forward) and 5’-GCTCGACCAGTTTAGTTACCC-3’ (reverse); Ms4a3-Cre: 5’-AGAGAAATCATCAGGGCAGAAAT-3’ (common forward), 5’-GAAAGGGGAACAAGCGAAGAT-3’ (wild-type reverse) and 5’-TTGGCGAGAGGGGAAAGAC-3’ (mutant reverse); tdTomato-wild-type: 5’-AAGGGAGCTGCAGTGGAGTA-3’ (forward) and 5’-CCGAAAATCTGTGGGAAGTC-3’ (reverse); tdTomato-mutant: 5’-CTGTTCCTGTACGGCATGG-3’ (forward) and 5’-GGCATTAAAGCAGCGTATCC-3’ (reverse); Arg1: 5’-TGCGAGTTCATGACTAAGGTT-3’ (forward) and 5’-AAAGCTCAGGTGAATCGG-3’ (reverse); Arg1-YFP_wild-type: 5’-AGAGCAAGCACCCCGTTTCTTCTC-3’ (forward) and 5’-GCTGTGATGCCCCAGATGGTTTTC-3’ (reverse); Arg1-YFP_mutant: 5’-TGAGCAAAGACCCCAACGAGAAGC-3’ (forward) and 5’-GCTGTGATGCCCCAGATGGTTTTC-3’ (reverse). For all experiments, male animals (8–12 weeks old) were used with the exception of BMDM generation and experiments involving Csf2rb^fl/flCcr2-Cre^ERT2 mice, where both sexes were included. All animal experiments were performed in strict accordance with regulations of the relevant animal welfare acts and protocols approved by the respective regulatory bodies (Austrian Ministry of Sciences, 2022-0.474.463 and 2025-1.040.322).

EAE induction

Active EAE was induced by subcutaneous immunization with 75 μg of MOG_35–55 (Charité) emulsified in complete Freund’s adjuvant (1:2 dilution, 150 μl per mouse), which refers to incomplete Freund’s adjuvant (Merck, F5506-10X10ML) enriched with 10 mg ml⁻¹ Mycobacterium tuberculosis (Difco/BD Pharmingen, H37Ra). Pertussis toxin from Bordetella pertussis (List/Quadratech, 181 or Hooke Laboratories, BT-0105) was administered intraperitoneally at days 0 and 2 after immunization (200 ng per mouse for pertussis toxin from List and 150 ng per mouse for pertussis toxin from Hooke Laboratories at days 0 and 2). For tamoxifen-inducible Cre strains, with the exception of Csf2rb^fl/flCcr2-Cre^ERT2 mice, animals were fed a tamoxifen diet (Tam400/CreER, tamoxifen citrate 400 ppm, ENVIGO, TD.55125IC) for 3 weeks, followed by 1 week of a regular chow diet (ssniff, V1536-000). Thereafter, EAE was induced, and animals were re-fed a tamoxifen diet throughout the course of EAE. For Csf2rb^fl/flCcr2-Cre^ERT2 and Csf2rb^+/+Ccr2-Cre^ERT2 littermate control animals, tamoxifen (5 mg per mouse) was administered as previously described⁴⁹. Briefly, tamoxifen was dissolved in corn oil and administered by oral gavage (250 μl per mouse at 20 mg ml⁻¹) at the first sign of clinical symptoms. Tamoxifen was administered two times (24–48 h apart), and animals were collected 3–4 days after the first dose (scores of 1.5–2). To assess the effects of arginine in EAE, animals were fed a control (Research Diets, A10021B) or arginine-free (Research Diets, A10036) diet 14 or 21 days before EAE, and animals were maintained on their respective dietary regimens throughout EAE. For arginine-free diet studies involving clinical outcome, only animals developing peak disease (score 3) were included.

Clinical signs of EAE were assessed using the following disease scores: no clinical signs (0), partial tail weakness (0.5), full paralysis of the tail (1), changes in gait or inability to climb (1.5), paralysis of one hind limb (2), paralysis of one hind limb and restraints in second hind limb (2.5), total paralysis of the hind limbs (3), hind limb paralysis and restraints in one fore limb (3.5), tetraparese (4) and moribund animal/death (5). EAE experiments were terminated at the following stages of disease development: the ‘onset’ stage is defined as the first time point at which immunized animals lose weight (7–10%) compared to the previous day, ‘peak disease’ is defined as the first day the animals display a clinical score of 3–4 (with peak in scores), and ‘recovery’ is defined as a distinct drop in score from a previous peak score of 3–4 to 1–2. The course of disease was monitored and evaluated via clinical scores. Additionally, the cumulative scores (summation of all daily clinical scores per animal over time), maximum scores (the highest clinical score reached per animal) and AUC are reported. Animals exhibiting extreme weight loss due to tamoxifen diet (>20% after 2 weeks) or those that did not develop disease (score 0) were excluded from the study.

Spinal cord cell isolation

Mice were killed, and spinal cords were extracted, cut into small pieces and digested using either a tissue dissociation kit (Miltenyi Biotec, tissue dissociation kit 1 for inflamed neural tissue, 130-110-201) according to the manufacturer’s instructions or 3 ml of digestion medium (DMEM supplemented with 1 mg ml⁻¹ papain (Sigma-Aldrich, P4762), 0.03 mg ml⁻¹ DNase I (Roche, 11284932001) and 0.5 mg ml⁻¹ collagenase/dispase (Sigma-Aldrich, 10269638001)). In experiments where digestion medium was used, mechanical and enzymatic dissociation was performed with a GentleMACS Octo Dissociator with Heaters (Miltenyi Biotec) and a spinal cord-specific program (37C_ABDK_01). Digestion was stopped by the addition of 5 ml of neutralization medium (DMEM supplemented with 10% fetal bovine serum). Tissue homogenate was filtered through a prewet, 70-μm cell strainer and postwashing and centrifuged at 250g for 5 min at 21 °C. The supernatant was discarded, and isolated cells were separated from debris/myelin using a 70%/37%/30% layered Percoll gradient (Sigma-Aldrich, P1644) with centrifugation at 300g for 40 min at 18 °C without braking. Three milliliters of the immune cell-containing 70%/37% interphase was collected, washed with 9 ml of 1× HBSS (Gibco, 14175-05) and pelleted by centrifugation at 500g for 7 min at 4 °C. Debris removal in experiments involving the Miltenyi Biotec tissue dissociation kit was performed with Debris Removal Solution (Miltenyi Biotec, 130-109-398). After cell isolation with either protocol, cell pellets were resuspended in 1 ml of flow buffer (PBS supplemented with 1% fetal calf serum (FCS)). Cells were counted and again pelleted by centrifugation at 500g for 7 min at 4 °C. Finally, cells were resuspended with an appropriate volume of flow buffer and further prepared for flow cytometry or sorting by FACS.

CSF isolation

CSF was isolated from the cisterna magna as previously described⁸². Briefly, animals were anesthetized with ketamine (20% Ketasol, 100 mg ml⁻¹, Livisto) and xylazine (10% Xylasol, 20 mg ml⁻¹, Livisto) diluted in PBS. An incision from the top of the skull to the dorsal thorax was made, and musculature from the skull base to the first vertebrae was removed, exposing the meninges overlying the cisterna magna. The tissue above the cisterna magna was removed, and the surrounding area was cleaned with a cotton swab to eliminate residual blood or other interstitial fluid. The arachnoid membrane above the cisterna magna was then punctured with a microneedle. This resulted in the release of CSF, which is under positive pressure. CSF was pulse centrifuged and stored at −80 °C until subsequent downstream analyses.

BMDM culture and stimulations

For preparation of bone marrow cell suspensions, cells were flushed from mouse femurs and tibiae with sterile PBS (Gibco, 14190-094). Cell suspensions were filtered through 70-μm cell strainers (Falcon, 352350), and two times the volume of erythrocyte lysis buffer (0.15 M NH₄Cl, 10 mM KHCO₃, 0.1 mM Na₂EDTA, pH 7.2–7.4) was added for 5 min. Cells were pelleted and resuspended in RPMI-1640 (Gibco, 61870-044) supplemented with 10% FCS (Sigma-Aldrich, F7524-500ML), 100 U ml⁻¹ penicillin/streptomycin (Lonza, LON17-745E and Capricorn, AAS-B) and 30 ng ml⁻¹ M-CSF (R&D Systems, 416-ML-050). Cells were then transferred to 10-cm culture dishes (Sarstedt, 82.1473.001) at a concentration of 0.5 million cells per ml. After 3 days of cultivation, fresh RPMI-1640 supplemented as described above was added (1/10 of the total volume). Three days later, cells were washed once with prewarmed PBS and detached using prewarmed CellStripper (Corning, 15313661). Cells were replated at a concentration of 1 million cells per ml in 12- or 24-well plates (Costar, 3737) and stimulated with 100 ng ml⁻¹ LPS (Invivogen, tlrl-3pelps), 20 ng ml⁻¹ IFNγ (Miltenyi Biotec, 130-105-774), 20 ng ml⁻¹ GM-CSF (R&D Systems, 415-ML-050), 20 ng ml⁻¹ IL-4 (Miltenyi Biotec, 130-097-761) and 20 ng ml⁻¹ IL-13 (Miltenyi Biotec, 130-094-070). Supernatants from treated cells were centrifuged for 5 min at 4 °C and 500g, immediately snap-frozen in liquid nitrogen and stored at −80 °C until subsequent downstream analyses. For experiments where extracellular arginine was depleted, 300 ng ml⁻¹ recARG1 (Bio Cancer, PEG-BCT-100, rhArgIpeg5000) or RPMI medium without lysine and arginine (Thermo Scientific, 88365, substituted with 0.21857923 mM L-lysine hydrochloride and for control medium also with 1.1494253 mM arginine) was used and was present throughout the entire stimulation. For experiments with BMDMs from Csf2rb^fl/flCcr2-Cre^ERT2 and Csf2rb^+/+Ccr2-Cre^ERT2 control mice, 4-hydroxytamoxifen was given at a final concentration of 2 μM in full medium for the duration of the differentiation.

Flow cytometry and cell sorting

For analysis of spinal cord cell populations, stainings were performed in PBS using fixable viability dye (Invitrogen, 65-0865-14, 1:2,000, 10 min at room temperature) as well as the following antibodies: CD45.2-BV650 (clone 104, Biolegend, 109836, 1:100), Ly6G-BV605 (clone 1A8, Biolegend, 127639, 1:100), Ly6G-AF700 (clone 1A8, Biolegend, 127622, 1:100), CD3-BV605 (clone 145-2C11, Biolegend, 100351, 1:100), Ly6C-BV605 (clone HK14, Biolegend, 128035, 1:100), Ly6C-BV510 (clone HK14, Biolegend, 128033, 1:100), CD11b-BV510 (clone M1/70, Biolegend, 101245, 1:100), CD11b-PB (clone M1/70.15, Invitrogen, RM2828, 1:100), CD11b-BV605 (clone M1/70, Biolegend, 101237, 1:100), CX3CR1-PeCy7 (clone SA011F11, Biolegend, 149015, 1:100), CCR2-BV421 (clone SA203G11, Biolegend, 150605, 1:100), CCR2-FITC (clone SA203G11, Biolegend, 150607, 1:100), F4/80-BV510 (clone BM8, Biolegend, 123135, 1:100), F4/80-BV605 (clone BM8, Biolegend, 123133, 1:100), F4/80-FITC (clone BM8, Biolegend, 123108, 1:100), CD11c-PE-Cy5.5 (clone N418, Invitrogen, 35-0114-82, 1:100), P2RY12-APC (clone S16007D, Biolegend, 848006, 1:100), P2RY12-AF488 (clone S16007D, Biolegend, 848016, 1:100), MerTK-APC (clone 2B10C42, Biolegend, 151507, 1:100), CD44-FITC (clone IM7, Invitrogen, 11-0441-82, 1:100), CD44-eF506 (clone IM7, Invitrogen, 69-0441-82, 1:100), CD64-BV605 (clone X54-5/7.1, Biolegend, 139323, 1:100), CD3-AF700 (clone 17A2, Biolegend, 100216, 1:100), CD3-rF710 (clone 17A2, Tonbo, 80-0032-U100, 1:80), CD8-BV510 (clone 53-6.7, Biolegend, 100751, 1:80), CD4-APC (clone REA604, Miltenyi Biotec, 130-116-487, 1:80), CD4-PE (clone REA604, Miltenyi Biotec, 130-116-509, 1:80) and CD25-BV421 (clone PC61, Biolegend, 102033, 1:80). Surface staining was performed for 20 min at 21 °C. Intracellular staining was performed using a Fixation and Permeabilization Buffer Set (eBioscience, 88-8824-00), True-Phos Buffer Set (Biolegend, 425401) or Foxp3/Transcription Factor Staining Buffer Set (Invitrogen, 00-5523-00) according to the manufacturer’s instructions and included antibodies specific to ARG1-PE (clone A1exF5, Invitrogen, 12-3697-80, 1:50), ARG1-APC (clone A1exF5, Invitrogen, 17-3697-82, 1:50), iNOS-PE (clone CXNFT, Invitrogen, 12-5920-82, 1:100), iNOS-PE-Cy5.5 (clone CXNFT, Invitrogen, 12-5920-80, 1:50), IL-10-APC (clone JES5-16E3, Invitrogen, 17-7101-82, 1:50), TGFβ1-PE (clone TW7-16B4, Biolegend, 141403, 1:50), HO-1-AF647 (clone EPR18161-128, Abcam, ab237268, 1:50), IL-17-PE (clone REA660, Miltenyi Biotec, 130-112-009, 1:50), IFNγ-FITC (clone REA638, Miltenyi Biotec, 130-117-780, 1:50) and FoxP3-APC (clone 3G3, Tonbo, 20-5773, 1:100) or isotype control rat IgG2a κ-APC (clone eBR2a, Invitrogen, 17-4321-81, 1:50), rat IgG2a κ-PE (clone eBR2a, Invitrogen, 12-4321-80, 1:50), rat IgG2a κ-PE (clone eBR2a, Invitrogen, 12-4321-41, 1:100), rat IgG2a κ-PE-Cy5.5 (clone eBR2a, Invitrogen, 35-4321-82, 1:50), rat IgG2b κ-APC (clone eB149/10H5, Invitrogen, 17-4031-81, 1:50), mouse IgG1 κ-PE (clone P3.6.2.8.1, Invitrogen, 12-4714-81, 1:50), rabbit anti-goat IgG-AF647 (Invitrogen, A21446, 1:50), REA control antibody-APC (clone REA293, Miltenyi Biotec, 130-113-446, 1:80), REA control antibody-FITC (clone REA293, Miltenyi Biotec, 130-113-449, 1:50), REA control antibody-PE (clone REA293, Miltenyi Biotec, 130-118-347, 1:50) and rat IgG1 κ-APC (clone P3.6.2.8.1, eBioscience, 17-4714-41, 1:100). Intracellular staining was performed overnight at 4 °C. For cell sorting of ARG1⁺ and ARG1⁻ cells, isolated cells from the spinal cord were filtered through a 40-μm cell strainer and sorted as viable CD3⁻Ly6G⁻CD45⁺CX3CR1⁺ARG1⁺/CD3⁻Ly6G⁻CD45⁺CX3CR1⁺ARG1⁻ cells. For analysis of Csf2rb expression within Mdcs from Csf2rb^fl/flCcr2-Cre^ERT2 and Csf2rb^+/+Ccr2-Cre^ERT2 control mice by quantitative real-time PCR with reverse transcription, isolated cells from the spinal cord were filtered through a 40-μm cell strainer and sorted as viable, CD45⁺CD3⁻Ly6G⁻CD11b⁺Ly6C⁺CX3CR1⁺ cells directly into Eppendorf tubes containing RLT Plus buffer.

For analysis of tdTomato expression within Ly6C^hi blood monocytes by flow cytometry, EDTA-treated blood was stained with antibodies and fixed with 2% paraformaldehyde for 15 min. The cell suspension was then subjected to red blood cell lysis and resuspended in PBS + 1% FCS before proceeding to flow cytometric analysis. The following antibodies were used: Ly6G-AF-700 (clone 1A8, Biolegend, 127622, 1:200), Ly6C-BV605 (clone HK14, Biolegend, 128035, 1:80), CD11b-PE-Cy7 (clone M1/70, eBioscience, 25-0112-82, 1:400), B220-PE-Cy5 (clone RA3-6B2, Tonbo, 55-0452-U100, 1:100) and CD3-PerCP-Cy5.5 (clone 145-2C11, Tonbo, 65-0031-U025, 1:100).

For analysis of BMDMs by flow cytometry, cells were detached using CellStripper (Corning, 15313661). Staining of detached cells was performed in PBS using fixable viability dye (Invitrogen, 65-0865-14) as well as the following antibodies: F4/80-BV605 (clone BM8, Biolegend, 123133, 1:80), F4/80-BV421 (clone BM8, Biolegend, 123137, 1:80), CX3CR1-PeCy7 (clone SA011F11, Biolegend, 149015, 1:100) and CD11b-BV510 (clone M1/70, Biolegend, 101245, 1:80). Intracellular staining was performed using a Fixation and Permeabilization Buffer Set (eBioscience, 88-8824-00), according to the manufacturer’s instructions, and included antibodies specific to ARG1-APC (clone A1exF5, Invitrogen, 17-3697-82, 1:100) and iNOS-PE (clone CXNFT, Invitrogen, 12-5920-82, 1:100) or isotype control rat IgG2a κ-APC (clone eBR2a, Invitrogen, 17-4321-81, 1:100) and rat IgG2a κ-PE (clone eBR2a, Invitrogen, 12-4321-41, 1:100).

For analysis of lipid content using LipidTox (Invitrogen, H34477, 1:150 in PBS), cells were stained for 25 min at 37 °C. Levels of lipid peroxidation were measured using C11-BODIPY (Invitrogen, D3861, 1:5,000 in PBS, 30 min at 37 °C) and were determined by dividing the mean fluorescence intensity of the FITC channel (oxidized lipids) by the sum of the mean fluorescence intensity of the FITC and PE channels (reduced lipids, together the total amount of lipids within a cell), which results in the ratio of oxidized/total lipids. For detection of cellular oxidative stress, CellRox (Invitrogen, C10422 or C10444, 1:2,000 in PBS, 30 min at 37 °C) was used.

Compensation was performed on single-stained samples of cells. All stained samples were acquired on a CytoFLEX S cytometer (Beckman Coulter), and results were analyzed using CytExpert (v2.5) and FlowJo software (v10.8.1). Cell sorting was performed on a FACSAria Fusion flow cytometer (BD Biosciences).

Lesion characterization and ARG1 immunohistochemistry of human samples

MS lesions were assessed on 2-μm cut sections using Luxol fast blue and CD68 immunohistochemistry staining as previously described^83,84. White matter MS lesions were identified by marked myelin loss as determined by Luxol fast blue staining and categorized following previously established criteria^36,37. Briefly, active lesions were identified by an accumulation of CD68⁺ macrophages and microglia, which frequently contain Luxol fast blue products. Chronic active lesions showed a demyelinated lesion center and an accumulation of CD68⁺ myeloid cells, occasionally containing myelin degradation products, at the lesion rim. Inactive lesions were identified by a hypocellular and demyelinated lesion center, showing no pronounced infiltration of CD68⁺ cells to the center or the rim. Remyelinated lesions were not examined in this study. Lesions were subdivided into separate zones. The lesion center is surrounded by the lesion rim, a distinct border, where myelin increases toward the level of the normal-appearing white matter. The periplaque white matter is a 1-cm-wide zone that transitions into the normal-appearing white matter. ARG1 levels within the aforementioned lesions were assessed by immunohistochemistry (polyclonal ARG1, Invitrogen, PA5-85267, sodium citrate buffer, 1:100) as previously described^83,84.

Immunofluorescence staining of human samples and image acquisition

For immunofluorescent co-staining of ARG1 and CD68 in human MS samples, a modified protocol for double labeling was used to enhance the ARG1 signal. After incubation, the first primary antibody was developed by a biotinylated tyramine precipitate and subsequently inactivated by a second heat steaming step with sodium citrate buffer^83,84. Briefly, after dewaxing, the slides were steamed at 80 °C using sodium citrate buffer at pH 6.0 for 40 min. Endogenous peroxidase was then blocked using Peroxidase Blocking Solution from Dako (SDS151) for 10 min. After 10 min of 10% FCS exposure, slides were incubated with primary anti-ARG1 (polyclonal, Invitrogen, PA5-85267, 1:100) overnight at 4 °C. The next day, a biotinylated secondary antibody (K675, Dako, SDS391) was applied for 30 min. Following this step, the slides were incubated with streptavidin-linked horseradish peroxidase (K675, Dako, SDS315) for 30 min, and the catalyzed signal amplification kit by Dako (K1500) was applied for 20 min. Afterward, the slides were steamed in sodium citrate buffer at 80 °C for 30 min and incubated with anti-CD68 (clone KP1, Dako, M0814, 1:500) overnight at 4 °C. Labeling was accomplished by exposure to streptavidin-linked Cy3 (Jackson ImmunoResearch, 016-160-084, 1:1,000) and secondary Alexa Fluor 488-linked anti-mouse IgG (polyclonal, Jackson ImmunoResearch, 115-545-166, 1:800) for 1 h at 21 °C. Nuclei were stained with DAPI (Thermo Fisher Scientific, D1306, 300 nM), and the slides were coverslipped using an aqueous mounting medium (Polysciences, 18606). Images were acquired with an Olympus BX63 microscope and exported as TIFF files. Figure composition and processing was performed with ImageJ⁸⁵ and GIMP (https://www.gimp.org).

For immunofluorescent co-stainings of ARG1 and iNOS with markers for myeloid cells (IBA1), astrocytes (GFAP) and neurons (SMI32), formaldehyde-fixed paraffin-embedded 2-μm-thick tissue slides were deparaffinized in xylene for 20 min. Slides were rehydrated, and endogenous peroxidase activity was blocked using 0.9% hydrogen peroxide in methanol. Heat-induced epitope retrieval was performed in citrate buffer (GV80511, Agilent Dako) using a heater (PT200, Agilent Dako). Aldehyde-induced autofluorescence was blocked by 1% (wt/vol) sodium borohydride in TBS twice for 2 min each. Nonspecific protein binding was blocked by applying 10% normal goat serum (16210-064, Gibco) in PBS for 30 min. Primary antibodies from distinct host species (anti-NOS2, 1:5,000, rabbit, PA3-030A, Invitrogen; anti-ARG1, 1:50, mouse, PA5-85267, Invitrogen) were applied overnight in a commercially available antibody diluent (S2022, Agilent Dako) at 4 °C. Fluorophore-labeled secondary antibodies (Cy3 anti-rabbit, 1:1,000, goat, 711-485-152, Jackson ImmunoResearch; AF488 anti-mouse, 1:800, goat, 115-545-166, Jackson ImmunoResearch) were applied for 90 min in Dako antibody diluent at 21 °C. The third primary antibody (anti-IBA1, 1:300, rabbit, 019-19741, FujiFilm Wako; anti-GFAP, 1:1,000, rabbit, Z0334, Agilent Dako; anti-neurofilament H, clone SMI32, 1:400, mouse, 801701, Sternberger Monoclonals) was then linked to a fluorophore using species-specific CoraLite Plus 647 labeled FlexAble kits (KFA503 and KFA523, Proteintech), according to the producer’s recommendations, and applied overnight at 4 °C. Counterstaining of nuclei was performed using 1 μg ml⁻¹ DAPI (D1306, Invitrogen) in TBS for 5 min. Slides were coverslipped using a water-based mounting medium (18606, Polysciences) and stored at 4 °C.

Images of fluorescence multilabelings were acquired on an Olympus BX63 microscope with a ×100 objective and Olympus Cellsense software. Rolling ball background subtraction and bleaching correction were performed in Fiji and ImageJ2 v2.16.0 (refs. ^85,86,87). Horizontal reconstruction of the images was done on z stacks with 0.1-μm steps in ImageJ2. For quantification of IBA1, whole slides were imaged at ×10 and stitched together. Consecutively, 1 × 1 mm regions of interest were selected and exported. Cell detection was performed using the cyto3 model⁸⁸ in the cellpose GUI⁸⁹ on the DAPI and IBA1 channels. Cell labels were transformed to regions of interest in Fiji and imported to QuPath v0.5.1 (refs. ^90,91). The individual channels were thresholded for each slide, and a built-in combination classifier was applied.

scRNA-seq analysis

scRNA-seq data from CNS cells during EAE are available under accession number GSE130119, and the expression matrix, metadata and co-ordinates were kept as previously described⁴⁵. Clustered data were visualized using t-SNE plots, and, of the 22 clusters originally identified⁴⁵, myeloid cell-specific clustering (clusters 0–4 and 18) was performed with Arg1 expression identified in cluster 2. Cluster 2 was further subclustered into Slc7a2^hi and Slc7a2^lo, and differentially expressed genes between Slc7a2^hi and all other myeloid cells were determined using the Seurat R package v4.3.0. Subsequently, GSEA of the differentially expressed genes ranked according to their average log₂(fold change) values was performed, and normalized enrichment scores were determined for the most significant pathways (FDR < 0.05) related to GO biological process using the fgsea R package⁹².

Targeted metabolomics of CSF and spinal cord tissue

Metabolite profiling was conducted at the Van Andel Institute Mass Spectrometry Core. Polar metabolites were extracted by the Bligh and Dyer method⁹³ from CSF (40 μl of CSF per 1 ml final extraction volume) and spinal cord tissue (80 mg of tissue per 1 ml final extraction volume). Spinal cord tissue was mechanically dissociated in a bead mill homogenizer. The top aqueous layer was collected, dried in a rotary vacuum and resuspended in water (40 μl for CSF and 100 μl for the spinal cord). Metabolites were profiled using two orthogonal liquid chromatography methods on an Agilent 6470 triple-quadrupole mass spectrometer. First, central carbon metabolites, including amino acids, tricarboxylic acid cycle intermediates and other organic acids, nucleotides/nucleosides, sugar phosphates and others, were profiled using ion-paired chromatography in ESI-negative mode, as described previously⁹⁴. Note, this ion-paired chromatography method is incompatible with positive-mode analysis and requires a dedicated liquid chromatography system. If ion-paired and standard methods are to be used on the same mass spectrometer, extensive source cleaning is required between methods. Next, additional metabolites of interest, including acyl-carnitines, kynurenine and related intermediates and S-adenosylmethionine/S-adenosylhomocysteine amenable to ESI-positive mode, were analyzed by reversed-phase chromatography (Cortecs T3, 2.1 × 150 mm, 1.6 μM, 186008500, Waters). For this method, mobile phase A was 0.1% formic acid (vol/vol), and mobile phase B was 90% acetonitrile with 0.1% formic acid (vol/vol). The gradient was as follows: 0–2.0 min, 100% A at 0.4 ml min⁻¹; 2.0–7.1, ramp to 100% B at 0.4 ml min⁻¹; 7.1–8.0 min, hold at 99% B. The column was then re-equilibrated in 100% A at 0.6 ml min⁻¹ for 2 min. Full liquid chromatography–mass spectrometry parameters for both methods and raw peak-area data are available in Supplementary Table 1.

Targeted metabolomics data analysis

Metabolomics data were log₂ transformed and scaled by using the following z-score formula: z = (x − μ) / σ, where x is metabolite abundance for one sample, μ is mean abundance for all samples, and σ is the population standard deviation. Metabolite profiles across experimental groups (Fig. 1c) were identified by soft clustering using the Mfuzz R package v2.56.0 (ref. ⁹⁵) under default parameters except that the fuzzifier was set to 1.3137 for spinal cord and 1.3099 for CSF. Significant metabolites (FDR < 0.15) between experimental groups in both the spinal cord and CSF were determined using two two-tailed t-tests or one-way ANOVA, and P values were adjusted for multiple hypothesis testing using the Benjamini and Hochberg method. Metabolite set enrichment analysis was performed using Metaboanalyst v5.0 (ref. ⁹⁶) and an FDR of <0.10. Metabolites were classified as strong or weak metabolites based on P value results from the different experimental groups (healthy, peak and recovery) using a one-way ANOVA followed by a Tukey’s honestly significant difference post-test (95% confidence level). Those metabolites passing a double threshold (P < 0.05 in two types of comparisons between the experimental groups) were grouped as strong metabolites whereas those passing only one threshold (one comparison between the experimental groups) were grouped as weak metabolites. Strong metabolites were subsequently subcategorized into DLMs, RLMs or HLMs, with DLMs significant between health and peak disease and between health and recovery, RLMs significant between health and recovery and between peak disease and recovery and HLMs significant between health and peak disease and between peak disease and recovery. Clusters of significant data from comparisons were depicted as heat maps using hierarchical clustering with Euclidean distance. For the average effect analysis examining impacts of ARG1 deficiency in CX3CR1⁺ cells at peak disease, mean z scores from Arg1^fl/flCx3cr1-Cre mice were subtracted from mean z scores from Arg1^fl/fl animals and plotted against the mean z scores of healthy animals subtracted from mean z scores of animals with peak disease.

Statistical analysis

Data collection and analysis were not performed blind to the conditions of the experiments, and mice were grouped randomly per cage. Data distribution was assumed to be normal, but this was not formally tested. No statistical methods were used to predetermine sample size, but our sample sizes were similar to those generally used in EAE studies.

Statistical analyses were performed by using two-tailed t-tests, an ordinary one-way ANOVA followed by a Tukey’s or Dunnett’s post-test, simple linear regression or two-way ANOVA followed by a Sidak’s post-test with Prism 9.4.1 software (GraphPad), unless otherwise stated. P values of clinical course AUC were determined by two-tailed t-test. Statistical significance is indicated by *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.001. All error bars indicate ±s.e.m.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.