Fig 2: VEGF and MAG:PLP1 in brain tissue homogenates. VEGF and MAG:PLP1 in left middle (LM) and right middle (RM; infusion site) homogenates. VEGF (pg/mL/mg tissue/mL buffer) measured by sandwich ELISA in (a) RM and (b) LM samples, with values, mean and 95% CI shown. One-way ANOVA with Sidak’s multiple comparisons test compared VEGF in the groups. The variation in (a) RM between the 4 groups was not significant (F3,56 = 2.05, p = 0.117), and post-hoc analysis found no significant variation between the Aβ40 and saline groups. VEGF in the Aβ+zib group was, on average 961 (95% CI: 38.1–1884) pg/mL/mg tissue/mL buffer lower than in the Aβ group (p = 0.0398). We found significant variation between (b) LM homogenates (p = 0.0080), and although again there was no significant difference between the Aβ40 and saline groups (p = 0.983), VEGF in the Aβ+zib group was, on average, 1243 (410–2076) pg/mL/mg tissue/mL buffer lower than in the Aβ40 group (p = 0.0023). MAG:PLP1 in (c) RM and (d) LM samples, with the values, median and IQR shown, did not vary significantly between the 4 groups for (c) RM (H = 0.0898, p = 0.993) or (d) LM (H = 0.697, p = 0.874) samples on Kruskal-Wallis analysis.

Fig 3: Relationships of PDGFRβ, fibrinogen and endothelin-1 with amyloid-β42: influence of Braak stage and infection. (A and B) Fibrinogen correlated positively with amyloid-β42 in BSV–VI without and BS0–II with systemic infection. (C and D) PDGFRβ correlated negatively with amyloid-β42 (Aβ42) approaching significance (r = −0.0568; P = 0.0541) in BSIII–IV in the absence, and positively in BSV–VI in the presence, of terminal systemic infection in superior temporal cortex. (E and F) EDN1 correlated positively with MAG:PLP1 in BSV–VI in BSV–VI only without infection. The best-fit linear regression lines and 95% confidence intervals are shown.

Fig 4: Nox2 deficiency protects against BCAS-induced white matter damage. (a) Representative images of MAG staining in the corpus callosum, the internal capsule and the optic tract from the experimental groups (WT, WT-BCAS and KO-BCAS); scale = 50 µm. Arrows point to myelin debris in WT-BCAS images. (b) White matter integrity was measured by MAG grading in the corpus callosum, the internal capsule and the optic tract at 12 weeks after surgery. MAG grading increased significantly in WT BCAS compared to shams but not in Nox2 KO BCAS mice. Data was analysed using Kruskal-Wallis test followed by Dunn’s post-hoc test for multiple comparisons. **P < 0.01, ***P < 0.01 vs. WT-Sham; N = 7–12 per groups.

Fig 5: ANGPTL2 promotes oligodendrocyte differentiation and remyelination under pathological conditions. A Representative images of the corpus callosum region stained with oil red O in Angptl2+/+ and Angptl2−/− mice after the withdrawal of cuprizone for two additional weeks. Myelin was stained in red color. Meanwhile, more nonspecific oil red O positive deposits (a phenomenon that oil red O deposits in damaged myelin region) were observed in Angptl2−/− mice (arrows). B Quantitative data of the demyelinating areas in Panel A; six sections from each mouse were analyzed (n = 7). C Immunofluorescence staining for MBP, MOG, MAG and GFAP in the corpus callosum region from Angptl2+/+ and Angptl2−/− mice after the withdrawal of cuprizone for two additional weeks. D Quantification of the fluorescence intensity of MBP, MOG, MAG and GFAP in the corpus callosum region in Panel C. Four sections from each mouse were analyzed (n = 6–7). E Representative electron microscopy images of the corpus callosum from Angptl2+/+ and Angptl2−/− mice after the withdrawal of cuprizone for two additional weeks; red arrow indicates uncompacted myelin lamella; F Quantification of the G ratios of the remyelinated axons of the corpus callosum in Panel E. Approximately 70 axons were counted per mouse (n = 3). G The scatter plot for the individual G-ratio values and axonal size distribution of the corpus callosum in Panel E. H Motor coordination of Angptl2+/+ and Angptl2−/− mice after cuprizone (Cupz) treatment for five weeks and cuprizone withdrawal for two additional weeks in the rotarod test; 7–8 mice were used for each group. (*p < 0.05, **p < 0. 01, ***p < 0.001)