Abscisic acid (GoldBio) was dissolved in 100 mM NaOH to create a 75 mM stock solution and stored at -20°C. Further dilutions were made in RPMI1640 medium (Gibco). In the mouse model of microglial activation, lipopolysaccharide (LPS; L5886, Sigma-Aldrich) was administered at a dose of 3 mg/kg body weight. The following antibodies were used for immunofluorescence: rabbit anti-myelin basic protein (MBP; 1:250; MAB386, Sigma-Aldrich; brain cryosections), rat anti-MBP (1:250; MCA409S, Sigma-Aldrich; cerebellar brain slices), rabbit anti-myelin basic protein (dMBP; 1:2000; AB5864, Merck; cerebellar brain slices), rabbit anti-neurofilament (NF; 1:1000; ab8135, Abcam), rat anti-F4/80 (1:100; MCA497G, Bio-Rad), rabbit anti-iNOS (1:100; ab15323, Abcam), mouse anti-CC1 (1:100; ab16794, Abcam), and goat anti-Olig2 (1:100; AF2418, R&D Systems). Appropriate secondary antibodies were obtained from Invitrogen.

For mouse bone marrow derived macrophage (BMDM) experiments, C57BL/6J mice were bred at the Biomedical Research Institute of Hasselt University. For in vivo experiments, male and female wild-type C57BL/6J mice (8-12 weeks old) were obtained from Janvier (France) and maintained under specific pathogen-free conditions. Mice were acclimatized for 2 weeks before initiation of the experiments. The mice were maintained on a 12-hour light/dark cycle with free access to water. Mice were fed either a standard chow diet or a specially formulated diet (abscisic acid-supplemented diet; 400 mg/kg, Research diets Inc.) as described previously (49). All animal procedures were conducted in accordance with the institutional guidelines and approved by the Ethical Committee for Animal Experiments of the VIB site at Ghent University Faculty of Sciences (EC2022-087, EC2023-029, EC2024-005) or the Ethical Committee for Animal Experiments of Hasselt University (202408K; ex vivo use). The number of animals was determined by power analysis using G*power software.

Acute demyelination was induced as described previously (36). Briefly, 9-11-week-old male mice were fed ad libitum a diet containing 0.3% cuprizone (bis[cyclohexanone]oxaldihydrazone; Sigma-Aldrich) mixed into powdered chow, either with the standard formulation or supplemented with abscisic acid (400 mg/kg), for 5 weeks. Following withdrawal of the cuprizone diet, mice were fed powdered standard chow or ABA-supplemented chow for 1 week. Male mice were used because they show a more consistent demyelination response than female mice. Mice were euthanized at 5 weeks (5 wk) or after 1 week of recovery (5 + 1 wk) with intraperitoneal injection of ketamine (100 mg/mL) and xylazine (20 mg/mL) followed by transcardial perfusion. Blood and brain tissues were collected for ABA measurement, histological and biochemical analysis. In this experiment, groups were composed of 7-8 animals per condition.

Starting 1 week before immunization and throughout the experiment, female mice were fed ad libitum with either an ABA-supplemented diet (400 mg/kg) or a control diet. At the age of 11 weeks, experimental autoimmune encephalomyelitis (EAE) was induced. In this experiment, female mice were used because they show more reliable and uniform EAE development. Female C57BL/6J mice were sedated with isoflurane (4%) to minimize suffering and were immunized subcutaneously, according to manufacturer’s instructions (EK-2110; Hooke Laboratories), with 2 times 100 µL of recombinant human myelin oligodendrocyte glycoprotein peptide (MOG35–55) (~300ng) emulsified in complete Freund’s adjuvant containing Mycobacterium tuberculosis H37Ra. Directly after immunization and after 24 hours, mice were injected intraperitoneally with 100 µL of pertussis toxin (121 ng). A control group (no EAE) was included where mice were not immunized and received an injection with phosphate buffered saline (PBS) instead of pertussis toxin. This group allowed for better comparison of body weight and clinical scoring. Throughout the experiment, from the day of immunization until the day of sacrifice, mice were weighed daily and clinically scored for neurological signs of the disease according to the manufacturer’s mouse EAE scoring guide: 0: no clinical symptoms; 0.5: distal tail paralysis; 1: tail paralysis; 2: tail paralysis and partial hindlimb paralysis; 2.5: tail paralysis and dragging of hind legs; 3: complete hindlimb paralysis; 4: paralysis to the diaphragm; 5: death by EAE. Animal welfare and suffering of the mice was monitored daily. Mice were humanely euthanized when they had lost 20% of their initial body weight or at day 24. Euthanasia was performed using intraperitoneal injection of ketamine (100 mg/mL) and xylazine (20 mg/mL) followed by transcardial perfusion. In this experiment, groups were composed of 8 animals per condition.

Starting 1 week before injection and throughout the experiment, mice were fed ad libitum with either an ABA-supplemented diet (400 mg/kg) or a control diet. Mice were injected intraperitoneally with LPS (3.0 mg/kg body weight, i.p.) or PBS, as described previously (64). Body weight and temperature were checked 5 hours, 10 hours and 24 hours post-injection and mice were sacrificed at the 24 hours timepoint. Euthanasia was performed using intraperitoneal injection of ketamine (100 mg/mL) and xylazine (20 mg/mL) followed by transcardial perfusion. Blood and brain tissues were collected for ABA measurement, histological and biochemical analysis. In this experiment, groups were composed of 5 animals per condition.

BMDMs were isolated and differentiated as previously described (36). Briefly, tibial and femoral bone marrow suspensions from 9-12-week-old female wild-type C57BL/6J mice were cultured in 10 cm petri plates at a concentration of 10 x 10⁶ cells/plate. The cells were differentiated in bone marrow medium (RPMI1640 supplemented with 10% fetal calf serum (FCS, Gibco), 50 U/mL penicillin (Invitrogen), and 50 U/mL streptomycin (Invitrogen)) supplemented with 15% L929-conditioned medium (LCM). After differentiation, BMDMs were detached at 37°C with 10 mM EDTA in PBS (Gibco) and plated for experiments at 0.5 x 10⁶ cells/mL in bone marrow medium supplemented with 5% LCM (37°C, 5% CO₂). Cells were treated daily with mouse myelin (100 μg/mL) for 24 hours (mye^24h) or 72 hours (mye^72h), and ABA (10 µM).

Cerebellar brain slices were obtained from C57BL/6 mouse pups at postnatal day 9 or 10 (P9 or P10), as described previously (65, 66). The brain slices were cultured in MEM medium (Thermo Fisher Scientific) supplemented with 25% horse serum (Thermo Fisher Scientific), 25% Hank’s balanced salt solution (Sigma‐Aldrich), 50 U/mL penicillin, 50 U/mL streptomycin, 1% Glutamax (Thermo Fisher Scientific), 12.5 mM HEPES (Thermo Fisher Scientific), and 1.45 g/L glucose (Sigma‐Aldrich). To induce demyelination, brain slices were treated with lysolecithin (LPC; 0.5 mg/mL, Sigma-Aldrich) for 16 hours at 3 days post isolation. After demyelination, brain slices were allowed to recover in culture medium for 1 day, followed by daily treatment with vehicle (PBS), ABA (1-10 μM) or PPARγ inhibitor (GW9662; 10µM; Sigma-Aldrich) for 1 week. For microglia depletion experiments, slices were treated with clodronate liposomes or empty liposomes (0.5 mg/mL; LIPOSOMA) for 24 hours, immediately after isolation.

Cerebellar brain slices were fixed in 4% paraformaldehyde (PFA) for 15 minutes at room temperature, while frozen brain cryosections were air dried and fixed in ice-cold acetone for 10 minutes at -20°C. For low permeabilization staining with MBP, cryosections were fixed in 4% PFA for 4 hours at 4°C and dehydrated using a sucrose gradient. Cerebellar brain slices were blocked with a buffer containing 5% normal horse serum and 0.3% Triton X-100 in PBS for 1 hour, while brain cryosections were treated with Dako protein block (Agilent) for 30 minutes and cryosections in the low permeabilization experiment were blocked with a buffer containing 1% normal goat serum, 0.5% BSA and 0.1% Tween. After blocking, brain slices and cryosections were incubated overnight at 4°C with primary antibodies. Following washing steps, they were exposed to the suitable secondary antibodies for 2 hours and 1 hour, respectively, at room temperature. Images of cerebellar brain slices were made on an LSM880 confocal microscope (Zeiss). For every brain slice at least 3 pictures were taken in different cerebellar lobes. The level of remyelination was quantified using the myelination index, computed as the ratio of the colocalized area of MBP and NF by the total NF area. Colocalization was calculated using the colocalize threshold plugin in ImageJ. dMBP staining was analyzed using ImageJ thresholder and denoted as percentage of total area. Images of brain cryosections were taken using a Leica DMi 8 microscope and LAS X Office software (Leica) and analyzed using the thresholder in ImageJ software. Measurements are shown as percentage of corpus callosum area. The researchers were blinded during sample processing.

Mouse BMDMs, cultured on glass cover slides, were fixed in 4% PFA for 15 minutes at room temperature. To visualize intracellular myelin degradation products, fixed BMDMs, unfixed frozen brain cryosections, and cerebellar brain slices were stained with 0.3% Oil Red O (ORO; Sigma-Aldrich) for 10 minutes. Cell nuclei were counterstained with hematoxylin. Images were captured using an Axio Scan.Z1 microscope (Zeiss) or an LSM880 confocal microscope (Zeiss) and analyzed with the QuPath software or the ImageJ software. For BMDMs, ORO stain was determined as percentage of cell area. For brain cryosections, ORO stain was determined as percentage of corpus callosum. For cerebellar brain slices, ORO stain was determined as percentage of total brain slice. The researchers were blinded during sample processing.

For transmission electron microscopy (TEM) analysis, 1 mm-thick coronal brain sections were collected at approximately -2.7 mm from the bregma. From these sections, the corpus callosum was dissected, and the portion from the right hemisphere was used for TEM. These dissected samples were immersed in a fixative solution containing 2.5% glutaraldehyde and 4% PFA in 0.1 M sodium cacodylate buffer (pH 7.2). The samples were placed in a vacuum oven for 1 hour, then rotated for 3 hours at room temperature. The fixative solution was subsequently replaced with fresh fixative, and samples were left rotating overnight at 4°C. After washing, the samples were post-fixed overnight in 1% OsO₄ with 1.5% K₃Fe(CN)₆ in 0.1 M sodium cacodylate buffer at 4°C. The samples were rinsed in distilled water and left in 1% uranyl acetate for 1 hour for bulk staining. Samples were then dehydrated through a graded ethanol series, followed by embedding in Spurr’s resin. Ultrathin sections were cut perpendicular to the corpus callosum using an ultramicrotome (Leica EM UC6) and were transferred to 0.7% formvar-coated copper grids (Aurion). Grids were viewed with a JEM-1400plus transmission electron microscope (JEOL, Tokyo, Japan) operating at 80 kV. ImageJ was used to calculate the g-ratio (the ratio of the inner axonal diameter to the total outer diameter) using 8-10 images per animal. MyelTracer software (67) was used to count the number of myelinated axons. The researchers were blinded during sample processing.

For 3D reconstruction of microglia, dissected brains were fixed overnight in 4% PFA and embedded in 5% agarose. 50 μm thick sagittal sections were cut using a vibratome (Leica). Brain sections were blocked with a buffer containing 5% normal goat serum and 0.5% Triton X-100 in PBS for 1 hour. After blocking, brain sections were stained overnight with anti-IBA1 antibody at 4°C, followed by secondary antibodies for 2 hours at room temperature. Z-stack images were taken in the CA1 region of the hippocampus with an LSM 780 microscope (Zeiss). The 3D reconstructions and measurements were done by the filament tracing algorithm from Arivis Vision 4D software (Zeiss). The researchers were blinded during sample processing.

For BMDM experiments, cells were treated with ABA (10 μM) and stimulated with myelin (100 µg/mL) for 24 or 72 hours. Cell lysis was performed using Qiazol (Qiagen). For mouse experiments, corpus callosum tissue was lysed using Trizol (Invitrogen). RNA from cells and tissue was extracted using the RNeasy Mini Kit (Qiagen) and Total RNA Mini Kit (Aurum), respectively. RNA concentration and quality were determined with a Nanodrop spectrophotometer (Isogen Life Science). DNase I treatment was performed on RNA using the DNAse I recombinant kit (4716728001, Sigma-Aldrich). cDNA synthesis was conducted using the SensiFast cDNA Synthesis Kit (GC Biotech) according to the manufacturer’s instructions. qPCR was performed on a Real-Time PCR system (Lightcycler 480, Roche) using the SensiFast SYBR No-Rox Kit containing 1x SYBR green (GC Biotech), 0.3 μM primers (Integrated DNA Technologies), 2-8 ng cDNA and nuclease-free water. mRNA expression was analyzed using qbase+ software version 3.2 (Biogazelle). GeNorm was used to select stable housekeeping genes. A list of primer sequences is provided in Supplementary Table S1 .

Chemokines and cytokines were measured in plasma using a Bio-Plex assay for tumor necrosis factor α (TNFα) (12002444; Bio-Rad), interferon γ (IFNγ) (12002438; Bio-Rad), interleukin 6 (IL6) (12002241; Bio-Rad), interleukin 10 (IL10) (12002242; Bio-Rad), C-C motif chemokine 5 (CCL5) (12002256; Bio-Rad) and monocyte chemoattractant protein 1 (MCP1) (12002441; Bio-Rad) according to the manufacturer’s instructions. Signals were measured on a Bio-Plex 200 system (Bio-Rad).

Myelin was purified from postmortem mouse brain tissue by means of density gradient centrifugation, as described previously (68). Myelin protein concentration was determined by using the BCA protein assay kit (Thermo Fisher Scientific), according to manufacturer’s guidelines. Cells were treated with 100 μg/mL myelin. To evaluate the ability and extent of myelin phagocytosis, cells were pre-treated with vehicle (PBS) or ABA (10 µM) for 24 hours and were subsequently exposed to 100 µg/mL myelin that was fluorescently labeled with pHrodo™ intracellular pH indicator dye (P35372; Invitrogen) in presence or absence of ABA for 1.5 hours. Fluorescence intensity was analyzed using an LSRFortessa (BD Biosciences). BMDMs were stained for intracellular lipid droplet load by 15-minute incubation with BODIPY (493/503) (D3922, Thermo Fisher Scientific) at 37°C. The LSRFortessa (BD Biosciences) was used to quantify cellular fluorescence.

Cholesterol levels of BMDMs were defined by using the Amplex Red Cholesterol Assay kit (A12216; Thermo Fisher) according to the manufacturer’s instructions. BMDMs were treated with myelin for 24 hours or 72 hours in presence or absence of ABA (10µM). Fluorescence was measured using the Infinite 200 pro microplate reader (Tecan).

ABA extraction for measurement in brain samples was performed as described before by Maixner et al. (69). Briefly, snap frozen brain tissue was homogenized with ice-cold 80% methanol and shaken for 24 hours at 4°C. The supernatant was collected, additional methanol was added to the pellet, and samples were shaken for 1 hour at 4°C. The total supernatant was collected and evaporated on a rotary evaporator. Petroleum ether was added to the liquid and mixed. After the liquid became layered, the top layer of petroleum ether was removed by pipetting and the bottom methyl alcohol layer was collected and evaporated again. After methanol had completely evaporated, the pellet was resuspended in 50 µL of 10% methanol.

ABA measurements were performed using a highly sensitive biosensor, engineered by our lab (70). Briefly, this biosensor is a HEK 293T cell line stably transfected with a plasmid containing the Arabidopsis PYL1^H87P mutant ABA receptor coupled to a VP16 activation domain, and the ABA coreceptor ABI1 coupled to a GAL4 binding domain and a plasmid containing a luciferase reporter gene. Presence of ABA brings the VP16 activation domain to a GAL4-dependent promotor, leading to luciferase gene expression. To measure ABA in serum or brain samples, reporter cells were seeded at 10.000 cells per well in 96-well plates. One day post seeding, samples were added at a dilution of 1:10 when the medium was refreshed. Positive controls of 80, 40, 20, 10, 5, 2.5 and 1.25 nM ABA, as well as a negative control were used to create a standard concentration curve. This standard curve was spiked with ABA-negative serum or brain lysate to correct for sample type induced effects. After 24 hours, cells were lysed in luciferase lysis buffer (25mM Tris-phosphate pH7.8, 2 mM DTT, 2mM CDTA, 10% glycerol, 1% Triton X-100). To cell lysates, D-luciferin (E1605, Promega) was added. Luminescence signals were measured in triplicate using the Glomax^® 96 Microplate Luminometer (Promega).

Statistical analyses and data visualization were performed using Prism (Graphpad, La Jolla, CA). Data are presented as mean ± standard error of the mean (SEM). The number of biological replicates is indicated by dots in the figure and denoted as “n” in the legend. Data distribution and variance characteristics were considered for statistical testing. For normally distributed datasets, an ordinary one-way ANOVA with correction for multiple testing or a two-tailed unpaired Student’s t-test was used. Statistical significance was defined as p<0.05, with significance levels indicated as *p<0.05, **p<0.01, ***p<0.001, and ****p<0.00001. No randomization was done, and the investigator was not blinded to the mouse group allocation. The sample size was determined by power analysis using G*power software. Outlier data points were removed on the basis of the robust regression and outlier removal (ROUT) method.

To determine the impact of ABA on remyelination, ex vivo and in vivo mouse models were used. In the ex vivo model, cerebellar brain slices were demyelinated with lysolecithin and subsequently exposed to ABA (experimental design in Figure 1A ). Fluorescent staining showed increased colocalization of myelin (MBP) and axons (NF) in brain slices treated with ABA, which was quantified by the myelination index (p<0.0001). This increased colocalization suggests more efficient axonal remyelination in ABA-exposed brain slices ( Figures 1B, C ). To evaluate the in vivo significance of ABA’s reparative capacity, a cuprizone-induced de- and remyelination model was employed in mice. The mice were given either a control diet or an ABA-supplemented diet, resulting in increased levels of ABA detectable in their serum (p<0.0002 and p<0.003 respectively) ( Supplementary Figure S1 ). Cuprizone feeding leads to reproducible, prominent demyelination in various CNS regions, particularly in the corpus callosum (71). Mice were fed a control or ABA diet supplemented with 0.3% cuprizone for 5 weeks, followed by 1 week on a diet without cuprizone, during which spontaneous remyelination occurred. The mice were pathologically characterized after demyelination (5 weeks) and during remyelination (5 + 1 weeks) (experimental design in Figure 2A ). One week after cuprizone withdrawal, ABA-fed mice displayed increased myelination efficiency in the CC, indicated by the ratio of MBP at 5 + 1 weeks compared to 5 weeks (p<0.004) ( Figures 2B, C ). Accordingly, TEM demonstrated that during the remyelination phase, ABA-fed mice had a significantly higher number of myelinated axons in the corpus callosum compared to the demyelination phase (p<0.03) ( Figures 2B–D ). However, no overall differences were observed in the g-ratio (ratio of the inner axonal diameter to the total outer diameter) during remyelination ( Figures 2B, E ). Interestingly, when axon size was considered, a higher g-ratio was observed in large-diameter axons of ABA-fed mice during remyelination ( Figure 2F ). During demyelination, a similar effect was noted in small-diameter axons ( Figures 2E, F ). These measurements suggest the presence of more myelinated axons but with a thinner myelin sheath. In support of these findings, ABA-fed mice exhibited a significantly increased mRNA expression of Mbp after demyelination (5 weeks) (p<0.02) and proteolipid protein (Plp) expression was significantly elevated during remyelination (p<0.03) ( Figure 2G ). Consistent with these findings, the corpus callosum of ABA-fed mice showed an increased abundance of Olig2+ CC1+ mature oligodendrocytes during remyelination compared to control-fed mice (p<0.04) ( Figure 3 ). These results indicate that ABA promotes remyelination in vivo by enhancing differentiation of oligodendrocyte precursor cells.

To investigate the impact of ABA on myelin clearance and processing by microglia, we initially examined foam cell formation. Counterintuitively, enhanced remyelination in ABA-fed mice was associated with increased intracellular lipid load, as evidenced by a higher ORO staining load (p<0.05) ( Figures 4A, B ). This observation prompted us to evaluate whether ABA facilitates remyelination by promoting the clearance of repair-inhibitory myelin debris (8). To this end, we quantified the presence of non-cell-associated myelin debris (31) by staining for MBP on PFA-fixed tissues without extensive permeabilization (72). Our results revealed a low MBP signal in both healthy mice and ABA-fed mice during remyelination, in contrast to a strong MBP signal in control-fed mice during remyelination (p<0.0002 and p<0.05 respectively) ( Figures 4C, D ). These findings indicate a higher presence of degraded myelin in control-fed mice compared to ABA-fed mice during remyelination, suggesting that the reparative impact of ABA is associated with increased myelin debris clearance and elevated intracellular lipid load.

Furthermore, improved remyelination in ABA-fed mice was associated with a significantly higher abundance of F4/80+ phagocytes during the demyelination phase (p<0.002) ( Figures 5A, B ). However, ABA treatment did not affect the percentage of iNOS+ phagocytes ( Supplementary Figure S2 ), nor did it result in significant changes in the mRNA expression of inflammatory mediators such as C-C motif chemokine 4 (Ccl4) and C-C motif chemokine 5 (Ccl5) ( Figures 5C, D ). Interestingly, ABA did affect the mRNA expression of neurotrophic factors. Specifically, ciliary neurotrophic factor (Cntf) was significantly increased during the demyelination phase (p<0.02), and there was a trend towards increased insulin growth factor 1 (Igf1) expression (p=0.0874). No significant differences were found for nerve growth factor (Ngf) or tumor growth factor β (Tgfβ) ( Figures 5C, D ). Subsequently, we assessed whether microglia depletion using clodronate liposomes in lysolecithin-demyelinated cerebellar brain slices would counteract the protective effects of ABA. Our results showed that the absence of microglia abrogated the protective effects of ABA on remyelination ( Supplementary Figure S3 ), indicating that ABA promotes remyelination in a microglia-dependent manner.

Given the critical role of F4/80+ cells in promoting remyelination, we were prompted to further investigate this cell population. For our in vitro experiments, we used primary BMDMs, as these cells are known to exhibit similar responses to myelin as microglia (11–13). BMDMs were cultured in the presence or absence of ABA and treated with myelin for 24 or 72 hours (experimental set-up in Figure 6A ). Consistent with in vivo observations, there was a trend towards increased intracellular lipid levels in ABA-treated cells after 24 hours of myelin exposure (p=0.1549), as shown by ORO staining ( Figures 6B, C ). However, no differences were detected at the 72-hour time-point. Further analysis revealed an increase in lipid droplet load in ABA-treated cells at steady state (p=0.0505) and after 72 hours of myelin exposure (p<0.03) ( Figure 6D ). To determine whether increased lipid uptake underlies the observed increase in lipid droplets following ABA treatment, we assessed the capacity of BMDMs to phagocytose myelin. Phagocytosis experiments demonstrated an increased uptake of pHrodo™-labelled myelin by BMDMs exposed to ABA (p<0.03) ( Figure 6E ). Supporting these findings, total cholesterol (TC) and free cholesterol (FC) levels were significantly increased at steady state in the ABA group (p<0.04 and p<0.02 respectively), with a trend towards increased TC after 72 hours of myelin exposure (p=0.1321) ( Figure 6F ). These findings indicate that the changes in the cellular lipid droplet pool upon ABA exposure are likely due to an enhanced capacity to internalize extracellular lipid-containing ligands.

Given that ABA has been reported to reduce inflammation in models of inflammatory bowel disease (47), Alzheimer’s disease (57) and HFD-induced neuroinflammation (56), we hypothesized that ABA could promote an anti-inflammatory and reparative phenotype in phagocytes. To investigate this, we used in vitro cultures of myelin-exposed phagocytes. We observed a significant increase in mRNA expression of neurotrophic factors Igf1 and Tgfβ at steady state (no myelin exposure) (p<0.008 and p<0.05 respectively), with a trend towards increased Igf1 expression after 24 hours of myelin exposure (p=0.1305) ( Figures 6G, H ). This was consistent with the increased expression of Cntf and Igf1 during the demyelination phase of the cuprizone experiment ( Figure 5C ). Although no differences were observed in the expression of inflammatory markers, including Ccl4 and Ccl5, in vivo and in vitro ( Figure 5C ), there was a significant decrease in Tnfα expression after 72 hours of myelin exposure (p<0.002) ( Figure 6I ), suggesting a potential anti-inflammatory effect of ABA. To further explore this possibility, an in vivo LPS model was employed. Mice were fed either a control diet or an ABA-supplemented diet for one week, followed by a single LPS injection. The mice were evaluated 24 hours post-injection (experimental set-up in Figure 7A ). To verify the effectiveness of diet and its ability to cross the blood-brain barrier, ABA levels were determined in both serum and whole brain samples, showing significant increases in ABA concentration (p<0.05) ( Supplementary Figure S4 ). Upon LPS injection, microglia typically shift from a resting phenotype to an activated phenotype. 3D- modeling analysis was used to compare microglial phenotypes in the CA1 region of the hippocampus across experimental groups. LPS treatment resulted in a significant decrease in microglial volume (p<0.03) ( Figures 7B, C ). However, in ABA-fed mice, volume of microglia was significantly higher compared to control-fed mice (p<0.0005), indicating that microglia in the ABA group were less activated. Additionally, dendrite count and -length were significantly increased in the ABA-treated group (p<0.03 and p<0.04 respectively) ( Figures 7D, E ). No significant differences were detected in the number of branch points, terminal points, or sections in LPS-injected mice treated with vehicle or ABA ( Figures 7F–H ). These findings suggest a more resting, less inflammatory phenotype in ABA-treated mice. To further investigate this, inflammatory cytokines and chemokines were measured in serum, revealing a significant decrease in MCP 1 (p<0.006) ( Supplementary Figure S5 ). These data collectively suggest that ABA may exert an anti-inflammatory effect by promoting a less activated state in microglia and by reducing pro-inflammatory chemokine levels. Despite the demonstrated protective effects of ABA in both the cuprizone-induced demyelination model ( Figure 2 ) and the LPS-induced microglial activation model ( Figure 7 ), we did not observe a protective effect on clinical score in the EAE model. Moreover, ABA-fed mice even exhibited a greater decrease in body weight during the acute phase of EAE ( Supplementary Figure S6 ).

Our data indicated that ABA enhances lipid droplet formation by promoting the uptake of lipid-containing complexes by macrophages ( Figures 6B–F ). To identify the pathway underlying this increased phagocytic capacity following ABA exposure, we investigated PPARγ, a receptor frequently implicated as an important downstream signaling component for ABA and reported to function in a LANCL2-dependent manner (59, 73). In the cuprizone model, we observed a significant upregulation of Lancl2 and Pparγ expression in the corpus callosum of ABA-fed mice during the remyelination phase (p<0.02 and p<0.04 respectively) ( Supplementary Figure S7A ). Additionally, the PPARγ response gene Plin2 was significantly upregulated in ABA-fed mice during the demyelination phase (p<0.05), while no significant differences were detected in Cd36 expression ( Supplementary Figure S7A ). Furthermore, we assessed the expression of Lancl2, Pparγ, and PPARγ-response genes in vitro using BMDMs exposed to myelin. Consistent with the in vivo results, ABA-treated BMDMs showed a significant increase in Pparγ expression at steady state (p<0.03) and a trend towards increased expression after 24 hours of myelin exposure (p=0.0861) ( Supplementary Figure S7B ). Additionally, expression of Cd36 was also significantly increased at steady state in ABA-treated BMDMs (p<0.05). However, no differences were observed in the expression of Lancl2 or Plin2 ( Supplementary Figure S7B ). The pronounced increase in Pparγ expression in both ABA-treated mice during remyelination and BMDMs suggests that the effects of ABA on enhancing remyelination, increasing lipid droplet formation, and promoting uptake of myelin debris might be dependent on PPARγ. To determine whether PPARγ is involved in these effects, we conducted an ex vivo experiment using cerebellar brain slices demyelinated with lysolecithin and subsequently treated with ABA in the presence of a PPARγ inhibitor (GW9662). In this setting, the previously observed increase in the myelination index seen in ABA-treated slices was abolished when PPARγ was inhibited ( Figures 8A, B ). To further explore the underlying mechanism, we examined the impact of PPARγ inhibition on the ABA-induced enhancement of myelin clearance by phagocytes. Consistent with previous in vivo findings, ABA treatment in the ex vivo brain slice model resulted in a significant reduction in non-cell-associated myelin debris (p<0.0001) ( Figures 8A, C ). However, when PPARγ was inhibited, ABA-treated samples still exhibited a decrease in myelin debris, but the reduction was less pronounced compared to samples without the inhibitor (p<0.01) ( Figures 8A, C ). To determine if this effect was related to microglia-mediated uptake, we measured the intracellular lipid load. In line with prior in vivo and in vitro data, ABA treatments significantly increased intracellular lipid accumulation in the ex vivo model (p<0.04) ( Figures 8A, D ). Interestingly, the PPARγ inhibitor alone also significantly elevated lipid load (p<0.002), with no additional increase observed in the presence of ABA ( Figures 8A, D ). These findings suggest that ABA might enhance remyelination through a PPARγ-dependent mechanism, primarily by facilitating the clearance of myelin debris.