Fibulin-2 is an extracellular matrix inhibitor of oligodendrocytes relevant to multiple sclerosis

Abstract

Impairment of oligodendrocytes and myelin contributes to neurological disorders including multiple sclerosis (MS), stroke, and Alzheimer's disease. Regeneration of myelin (remyelination) decreases the vulnerability of demyelinated axons, but this repair process commonly fails with disease progression. A contributor to inefficient remyelination is the altered extracellular matrix (ECM) in lesions, which remains to be better defined. We have identified fibulin-2 (FBLN2) as a highly upregulated ECM component in lesions of MS and stroke and in proteome databases of Alzheimer's disease and traumatic brain injury. Focusing on MS, the inhibitory role of FBLN2 was suggested in the experimental autoimmune encephalomyelitis (EAE) model, in which genetic FBLN2 deficiency improved behavioral recovery by promoting the maturation of oligodendrocytes and enhancing remyelination. Mechanistically, when oligodendrocyte progenitors were cultured in differentiation medium, FBLN2 impeded their maturation into oligodendrocytes by engaging the Notch pathway, leading to cell death. Adeno-associated virus deletion of FBLN2 in astrocytes improved oligodendrocyte numbers and functional recovery in EAE and generated new myelin profiles after lysolecithin-induced demyelination. Collectively, our findings implicate FBLN2 as a hitherto unrecognized injury-elevated ECM, and a therapeutic target, that impairs oligodendrocyte maturation and myelin repair.

Keywords: Extracellular matrix; Multiple sclerosis; Neuroscience.

Figure 1. FBLN2 is upregulated in lesions of MS and its animal models.

(A) Top: Luxol fast blue (LFB) and H&E histologically stained tissues showing demyelinated lesions from 3 MS brain samples. White matter lesions were defined by loss of LFB staining (yellow rings). Bottom: Immunofluorescence images labeled with CD45 for immune cells and FBLN2. Scale bars: 2 mm. (B) Yellow dotted square specifies the area shown at higher magnification to the right. Scale bars: 1 mm (left), 200 μm (right). (C) Representative image of large area of longitudinal section of spinal cord from EAE mice. White dotted box indicates a region of interest (ROI) tracked by loss of myelin. (D) Representative images of spinal cord sections from EAE mice comparing the normal-appearing white matter (NAWM) and lesion area. Scale bars: 100 μm. (E) Colocalization of FBLN2 in GFAP⁺ astrocytes but rarely in CD45⁺ immune cells using Imaris 3D rendering. Scale bar: 10 μm. (F) Quantification comparing the percentage of FBLN2 in NAWM and EAE lesions at different time points. n = 6 mice total per group. Data were acquired from 2 separate experiments, and each dot represents mean of 5 lesions analyzed per mouse. (G) Representative images of coronal sections of NAWM and LPC-induced lesion (dotted line) 14 days after injury. Scale bars: 100 μm. (H) Bar graph comparing the percentage of FBLN2 area within the LPC lesion over the time after injury. n = 6 mice for days 3, 14, and 21; n = 9 mice for day 7, from 2 separate experiments. Images in A and C were obtained by slide scanner. Images in B, D, E, and G were acquired by immunofluorescence laser confocal microscope (Z-stack). Data in F and H are presented as mean ± SEM; 1-way ANOVA, Bonferroni post hoc.

Figure 2. The better clinical recovery from EAE in FBLN2-deficient mice is associated with a profound myelination profile.

(A) Average EAE clinical score (mean ± SEM) is shown. n = 23 mice for WT, 18 mice for Fbln2^+/–, 19 mice for Fbln2^–/–, pooled from 3 independent experiments; 2-way repeated-measures ANOVA (mixed-effects model), Dunnett’s multiple-comparison test; *P < 0.05, **P < 0.01, WT vs. Fbln2^–/–; ^#P < 0.05, ^###P < 0.001, WT vs. Fbln2^+/–. (B and C) Violin plots of cumulative EAE scores for the first 17 days (B) and from day 18 after induction (C) (median, solid black lines; interquartile range, dotted black lines). Kruskal-Wallis test with Dunn’s multiple comparisons. (D) Percentage of mice undergoing remission on each day from peak clinical severity. Curves were compared using the log-rank test. (E and F) Uniform manifold approximation and projection (UMAP) plot of 34,395 cells from spinal cords of 9 EAE mice (E) and across different experimental groups (F) depicting 17 clusters as determined by 30 principal components and 0.3 clustering resolution. (G and H) Heatmap (G) and violin plots (H) comparing the levels of select DEGs associated with remyelination in oligodendrocyte cluster. Midline in the plot is the median. (I) Select top activated (red) or inactivated (blue) pathways in oligodendrocytes from Het and Homo FBLN2-knockout spinal cords compared with WT spinal cords as predicted by IPA.

Figure 3. The frequency of distinct oligodendrocyte populations in EAE WT versus FBLN2-deficient mice.

(A) UMAP plots of 2,592 oligodendrocyte-lineage cells reclustered into 7 distinct cell populations (principal components, 10; clustering resolution, 0.4). (B) Dot plot of representative marker genes enriched in oligodendrocytes. The size of the dot depicts percentage of cells expressing the gene in each cluster. The color represents the average gene expression level. (C and D) Bar graphs depicting percentage (C) and number (D) of cells in different subclusters of oligodendrocytes across groups (2-way repeated-measures ANOVA with Holm-Šídák post hoc test). (E) Heatmap showing the z scores of predicted pathways by IPA in oligodendrocyte subcluster. High and low z scores depict predicted activation and inhibition of pathways, respectively.

Figure 4. FBLN2 deficiency increases number of mature oligodendrocytes in demyelinating lesions.

(A and B) Representative images of coronal (A) and longitudinal (B) sections of spinal cord from EAE mice comparing WT and Fbln2^–/– mice. Tissues were stained for MBP and CD68. (C and D) Bar graphs comparing the percentage of ROI that is MBP⁺ (C) or CD68⁺ (D) (n = 10–12 mice; 2-tailed unpaired Student’s t test). (E–G) Representative images of longitudinal sections of spinal cord from EAE mice (E) and coronal sections of LPC lesions 14 days post injection (dpi) (F and G) stained for DAPI, Olig2, PDGFRα, and CC1. (H–M) Bar graphs comparing the number of Olig2⁺ lineage cells, Olig2⁺PDGFRα⁺ OPCs, and Olig2⁺CC1⁺ mature oligodendrocytes per square millimeter within EAE (H–J) and LPC (K–M) lesions. n = 19 mice for WT, 12 mice for Fbln2^+/–, and 15 mice for Fbln2^–/– over 3 separate EAE experiments; n = 11 mice for WT, 7 mice for Fbln2^+/–, and 10 mice for Fbln2^–/– over 2 separate LPC experiments; 1-way ANOVA, Bonferroni post hoc. All images were acquired by immunofluorescence laser confocal microscope (Z-stack). Scale bars: 200 μm (A) 100 μm (B, E, F, and G). Data are presented as mean ± SEM.

Figure 5. FBLN2 impairs maturation of oligodendrocytes and induces cell death.

(A) Mouse OPCs stained for O4 24 hours after plating onto PBS (control) or FBLN2. (B and C) Fold change in mean process outgrowth (B) and number of O4⁺ cells (C) of mouse OPCs cultured on control and different concentrations of FBLN2 for 24 hours. n = 5 independent experiments for FBLN2 (10 μg/mL), n = 3 for FBLN2 (1 and 5 μg/mL); 1-way ANOVA, Bonferroni post hoc. (D and E) Representative images (D) and number of MAG⁺ cells (E) from human OPCs cultured on control and FBLN2 (10 μg/mL). n = 3 independent experiments; 2-tailed, unpaired Student’s t test. (F and G) Mean process outgrowth (F) and number of O4⁺ cells (G) of mouse OPCs cultured on coated wells with different members of FBLN family (10 μg/mL). n = 3 independent experiments; 1-way ANOVA, Bonferroni post hoc. (H–K) Mouse OPCs (H) and fold change in mean process outgrowth (I), number of total cells (J), and number of O4⁺ cells (K) at 6, 12, and 24 hours after plating onto control or FBLN2 (10 μg/mL). n = 3 independent experiments; 2-way ANOVA, Bonferroni post hoc. (L) Representative images of mature mouse oligodendrocytes stained for O4 and MBP at 72 hours. (M) Quantification comparing number of O4⁺MBP⁺ cells. n = 3 independent experiments; 1-way ANOVA, Bonferroni post hoc. (N) Live imaging of mouse OPCs plated on control and FBLN2 (10 μg/mL) in the presence of propidium iodide (PI) at different time points. (O) Proportion of PI⁺ OPCs after 24 hours. n = 3 independent experiments; 2-tailed, unpaired Student’s t test. Each experiment (dot) included 3–4 replicates. (P) Ratio of Bax to Bcl2 mRNA expression in mouse OPCs using real-time PCR. n = 6 replicates over 2 separate experiments; 2-tailed, unpaired Student’s t test. Scale bars: 100 μm. Data are presented as mean ± SEM.

Figure 6. Blocking Notch pathway rescues the inhibitory effect of FBLN2 on oligodendrogenesis.

(A) Schematic diagram of experimental design. Created with BioRender. (B) Volcano plots showing upregulated or downregulated genes in mouse OPCs cultured on FBLN2-coated (10 μg/mL) wells for 6 hours, identified by RNA sequencing. (C) Top upregulated pathways predicted by IPA from DEGs. RNA-sequencing data were acquired from 3 replicates per group (PBS or FBLN2). (D) Representative images of mouse OPCs cultured on control and FBLN2-coated (10 μg/mL) wells with or without Notch inhibitor (SAHM1, 10 μM) after 24 hours (top) and 72 hours (bottom). (E and F) Quantification for mean process outgrowth (E) and number of O4⁺ cells after 24 hours (F) (n = 4 independent experiments). (G) Number of O4⁺MBP⁺ cells at 72 hours (n = 3 independent experiments). (H) Number of O4⁺ cells 24 hours after transfection with 2 Notch1 siRNAs (200 nM each; n = 3 independent experiments). Each experiment (dot) in E–H included 3–4 replicates; 1-way ANOVA, Tukey’s post hoc. (I) Schematic depicting the Notch pathway reporter assay using the constitutively expressing Renilla luciferase vector (positive control) and CSL (CBF1/RBP-Jκ) firefly luciferase reporter vector. (J) Bar graph comparing relative luciferase activity (firefly to Renilla) in HEK239 cells transfected with luciferase reporter or negative control vectors in the presence or absence of 10 μg/mL FBLN2 (n = 8–10 replicates over 2 separate experiments; 1-way ANOVA, Bonferroni post hoc). (K and L) Ridge plots comparing expression of select genes involved in Notch pathway across oligodendrocyte subclusters (0, DA-MOL; 1, MOL; 2, IFN-MOL; 3 and 4, Stressed-OL; 5, COP; 6, NFOL) (K) and experimental groups (L). (M) Immunofluorescence images of MS brain sample labeled for Olig2 and NICD. White arrows show the presence of Notch signaling in oligodendrocytes. Scale bars: 100 μm, insets, original magnification, 2×. Data are presented as mean ± SEM.

Figure 7. Astrocytic deletion of FBLN2 improves functional recovery in EAE, which is associated with more robust oligodendrogenesis.

(A) Schematic showing the genome of AAVs used in this study. AAVs were injected 2 weeks before EAE induction or LPC surgery for targeted disruption of FBLN2 (KD) or nontarget guide RNA. (B) Average EAE clinical score. (C and D) Violin plots comparing the cumulative EAE scores for the first 17 days (C) and from day 18 (D). Solid black lines represent medians; quartiles are shown by dotted black lines. n = 17 mice for Ctrl, 15 mice for KD from 3 independent experiments; Mann-Whitney test; *P < 0.05. (E–H) Representative images of longitudinal spinal cord sections from EAE mice (E) and quantifications for FBLN2 (F), MBP (G), and CD45 (H) percentage area of the lesion. (I and J) Graphs comparing number of Olig2⁺PDGFRα⁺ OPCs (I) and Olig2⁺CC1⁺ mature oligodendrocytes (J) per square millimeter of EAE lesions. n = 15 mice per group over 3 separate experiments; 2-tailed unpaired Student’s t test. (K–R) Representative images of LPC lesions 14 dpi (K–M) and quantifications comparing percentage area of FBLN2 (N), MBP (O), Iba1 (P), and number of OPCs (Q) or mature oligodendrocytes (R) per square millimeter of lesion ROI. n = 9 mice for Ctrl, 11 mice for KD over 3 separate experiments; 2-tailed unpaired Student’s t test. Images were acquired by immunofluorescence laser confocal microscope (Z-stack). Scale bars: 100 μm. The insets are magnified ×1.7. Data are presented as mean ± SEM.

Figure 8. FBLN2 deficiency enhances remyelination.

(A) NG2^CreER MAPT^mGFP mice were used to identify newly formed oligodendrocytes and myelin as GFP⁺. (B) Representative images of LPC lesions 14 dpi from NG2^CreER MAPT^mGFP mice that had received AAVs 2 weeks before surgery. Images were acquired by immunofluorescence laser confocal microscope (Z-stack). Scale bars: 100 μm. The insets are magnified ×1.7. (C) Bar graph showing the extent of GFP in lesions (n = 6 mice over 2 separate experiments; 2-tailed, unpaired Student’s t test). (D) Electron micrographs of LPC-induced lesions from WT and Fbln2^–/– mice at 14 dpi. Blue pseudocolored regions indicate examples of axons from myelinated axons. Scale bars: 1.5 μm. (E) Dot plot of percentage remyelinated axons. (F and G) Mean g-ratio (F) and scatterplot of g-ratio (y axis) in relation to axon diameter (x axis) (G) of individual fiber (n = 150 axons per mouse, 6 mice per group; simple linear regression of slopes). E–G: n = 6 mice per group; 2-tailed, unpaired Student’s t test; mean ± SEM.