3.1. Transient and severe BBB breakdown following CCH
Central nervous system (CNS) homeostasis is dependent on the integrity of the BBB. The BBB prevents dysregulated transit of molecules into the brain and very effectively blocks toxins and pathogens in order to preserve delicate neural functioning [25]. The dynamic changes in BBB integrity and the cascade reaction following CCH are still largely unknown. We began by examining the trend of BBB permeability following CCH. Increased brain water content indicated increased BBB permeability. We found that brain water content was increased 1 day post operation, and the brain edema was most severe 3 days post operation (wet/dry weight ratio: 4.45 ± 0.17, 4.93 ± 0.15 and 5.33 ± 0.19 in Sham, 1 d and 3 d group, respectively) (Fig. 1A, B). Recovery began at 7 days (wet/dry weight ratio: 4.63 ± 0.14 in 7 d group), and brain water content had nearly returned to normal level by 28 days post operation (wet/dry weight ratio: 4.63 ± 0.25 in 28 d group) (Fig. 1B). Measurement of dye leakage following injection of EB into the tail vein was another tool we used to assess BBB integrity. As expected, EB was accumulated in the CC following CCH. EB extravasation was apparent on gross examination of the brain 3 days post operation (Fig. 1A). The trend of accumulation mirrored brain water content (EB concentration: 2.08 ± 1.38, 4.09 ± 1.79, 7.68 ± 1.45, 3.02 ± 1.54 and 2.42 ± 1.89 µg/g in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Fig. 1D). These results suggest that severe barrier leakage defects appear as early as 1 day following CCH, and thereafter spontaneous recovery occurs.
After BBB breakdown, endogenous circulating macromolecules leak into the brain, which may be toxic to neuronal function. Using immunostaining, we found significant leakage of the serum protein ALB outside vessels in the CC following CCH. ALB leakage was most severe 3 days post operation (ALB density/Sham: 0.89 ± 0.45, 8.74 ± 4.34, 22.70 ± 5.66, 12.75 ± 4.41 and 2.64 ± 1.16 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Fig. 1C, E). We then examined plasma-derived immunoglobulin G (IgG) deposition in the CC. Using immunostaining, we observed some level of IgG within vessels and some IgG leakage outside vessels (Supplemental Fig. 1A). All IgG was increased at 3 days (all IgG density/Sham: 0.92 ± 0.41, 1.46 ± 0.63, 1.75 ± 0.53, 1.00 ± 0.34 and 1.04 ± 0.38 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 1B). The leaked IgG was significantly increased 1 day post operation, reaching the most severe level at 3 days post operation (outside IgG density/Sham: 2.16 ± 1.16, 7.56 ± 2.21, 29.17 ± 7.60, 5.89 ± 2.26 and 5.19 ± 2.17 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 1C).
3.2. Reduction of pericyte coverage leads to BBB dysfunction following CCH
BBB integrity depends on the totality of BBB structure. We therefore studied the effects of CCH on BBB constituents. Pericytes play an import role in BBB function [24, 26]. Using dual immunostaining for PDGFR-β- and collagen IV-positive brain capillary profiles, we observed the initial changes in pericytes following CCH (Fig. 2). Compared to the sham group, pericyte coverage was significantly decreased 1 day post operation (Fig. 2A). Pericyte coverage loss reached the most severe level 3 days post operation, with a decrease of approximately 65% compared to the sham group (Fig. 2A, C). However, by 7 days post operation, pericyte coverage showed a slight degree of recovery. Recovery reached approximately 84% of coverage of the sham group at 28 days post operation, when pericyte coverage between two groups showed no significant difference (pericyte coverage: 60.13 ± 10.68, 43.81 ± 13.50, 20.97 ± 11.32, 46.12 ± 12.70 and 50.47 ± 19.01 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Fig. 2A, C). By 3 days post operation, capillary length was significantly reduced compared to the sham group (capillary length: 35.41 ± 8.03, 33.09 ± 7.92, 25.93 ± 6.46, 34.70 ± 7.16 and 34.12 ± 8.13 mm in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Fig. 2A, B). Correlation analysis found that pericyte coverage was not correlated with capillary length (Fig. 2D), indicating that pericyte coverage loss is not due to capillary length reduction. Western blot analysis also showed that PDGFR-β protein levels decreased from day 1 to day 3 post operation, then increased from days 7 to 28 (Fig. 2E). PDGFR-β protein reduction was most severe 3 days post operation (PDGFR-β/β-actin: 1.24 ± 0.45, 0.67 ± 0.19, 0.17 ± 0.09, 0.57 ± 0.25 and 1.14 ± 0.26 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Fig. 2F). Desmin is another pericyte marker. We sought to further confirm pericyte loss using desmin immunostaining, and the loss pattern was indeed similar to that of PDGFR-β immunostaining (desmin length: 469.7 ± 100.3, 306.0 ± 112.1, 141.3 ± 64.2, 409.9 ± 91.0 and 414.4 ± 124.4 µm in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Fig. 3). Interestingly, we found significant negative correlation between pericyte coverage and ALB accumulation (Fig. 3D), indicating that pericyte loss is associated with BBB impairment.
Pericytes are not the only cell type that regulates permeability of the BBB [27]. EC, astrocytes and continuous complexes of endothelial junctions are also integral components of the BBB [28]. To further elucidate the effect of pericyte loss on BBB impairment, we also observed other components of the BBB. Glut1 is a marker of endothelial cells. Using Glut1 immunostaining to observe microvascular changes in the CC, we found no significant reduction in microvascular density (number of capillary: 20.25 ± 4.37, 20.75 ± 3.77, 18.38 ± 4.81, 21.50 ± 5.76 and 22.00 ± 6.09/section in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) and an increase in microvascular diameter (diameter of capillary: 5.58 ± 1.58, 7.19 ± 2.22, 10.44 ± 3.20, 7.55 ± 1.67 and 4.57 ± 1.21 µm in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 2). We then examined the expression of complexes between EC according to markers (i.e. occludin and claudin-5) for TJ. The expression of occludin was downregulated 3 days post operation compared to the sham group (occludin/β-actin: 0.57 ± 0.15, 0.47 ± 0.08, 0.34 ± 0.14, 0.50 ± 0.07 and 0.52 ± 0.16 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively). However, there was no significant downregulation of claudin-5 expression following CCH (claudin-5/β-actin: 1.27 ± 0.24, 1.26 ± 0.41, 1.05 ± 0.11, 1.17 ± 0.24 and 1.22 ± 0.22 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 3).
Astrocytic coverage of blood vessels is also vital for BBB integrity [29]. Although the number of GFAP-positive astrocytes was significantly increased (9.25 ± 3.62, 10.13 ± 3.31, 29.25 ± 11.80, 19.75 ± 5.31 and 10.63 ± 3.02 cells/section in Sham, 1 d, 3 d, 7 d and 28 d group, respectively), astrocytic vessel coverage was not increased following CCH (astrocytic coverage(%): 33.21 ± 6.47, 30.87 ± 12.31, 32.79 ± 8.14, 42.60 ± 5.89 and 37.81 ± 6.05 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 4). This indicates that astrocyte activation is intended to increase neuroinflammation, not to promote astrocytic coverage of microvasculature following CCH.
3.3. Neurotoxic molecules across the BBB occur through endothelial transcytosis following CCH
We further used TEM to observe the ultrastructural changes of BBB. We found the microvascular was edema and BM thickness was increased 1 day post operation (BM thickness: 100.20 ± 22.39, 186.50 ± 31.60 and 101.00 ± 17.87 µm in Sham, 1 d and 3 d group, respectively) (Fig. 4A, B). The edema was decreased and BM thickness returned to the normal level, but vesicles density in EC was significantly increased 3 days post operation (number of vesicles: 3.0 ± 1.41, 3.0 ± 1.41 and 7.0 ± 2.37/µm2 in Sham, 1 d and 3 d group, respectively) (Fig. 4A, C). CCH did not alter the ultrastructure of endothelial TJ (Supplemental Fig. 5). These results indicate that large neurotoxic molecules enter the brain parenchyma through increased endothelial transcytosis.
3.4. BBB dysfunction precedes neuroinflammation and demyelination following CCH
Neuroinflammation is an important factor in the pathogenesis of CSVD [30, 31]. The number of microglia reached its peak 3 days post operation, and 86% were activated microglial cells. From day 7 to 28 post operation, the number of microglia decreased (7.60 ± 2.86, 11.84 ± 3.24, 41.12 ± 10.69, 14.20 ± 6.31 and 13.52 ± 6.38 cells/section in Sham, 1 d, 3 d, 7 d and 28 d group, respectively). But 58% of microglial cells were still activated at 28 days (percentage of activated microglial: 23.32 ± 6.63, 32.62 ± 12.35, 87.08 ± 9.05, 61.23 ± 17.56 and 59.52 ± 16.07 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively), indicating that neuroinflammation persisted until at least 28 days post operation (Supplemental Fig. 6).
WML is another core pathological change of CSVD [32]. The myelin sheath is formed by mature myelin-producing oligodendrocytes, and WM damage is caused by the loss of mature myelin-producing oligodendrocytes [33]. Using immunohistochemistry staining for MBP, a marker for myelin sheath in neuronal axons, we found the MBP density in CC was progressive decreased from 3 days to 28 days after operation (MBP density/Sham: 0.96 ± 0.10, 0.82 ± 0.14, 0.80 ± 0.12, 0.55 ± 0.12 and 0.58 ± 0.15 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 7A, B). Until 28 days after operation, this downregulation had significantly difference compared with sham group (Supplemental Fig. 7B). WB analysis also showed lower MBP protein levels 28 days post operation (MBP/β-actin: 1.64 ± 0.41, 1.50 ± 0.31, 1.37 ± 0.30, 0.54 ± 0.20 and 0.65 ± 0.22 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) (Supplemental Fig. 7C, D). Although there was a slight increase of MBP protein level 56 days post operation, it was no significantly difference compared with 28 days. These results indicate that BBB breakdown precedes neuroinflammation and WML, and therefore BBB breakdown may be a key pathological event following CCH.
3.5. BBB dysfunction activates TGF-β signaling following CCH
After traumatic brain injury (TBI), serum protein leakage cause robust injury response by activating the transforming growth factor-β (TGF-β) signaling pathway [34, 35]. Hence, we next investigated whether TGF-β signaling pathway also as a candidate mechanism induced brain injury following CCH (Fig. 5). Immunostaining pSmad2, the downstream of the TGF-β receptors, we found the number of pSmad2-positive cells (102.30 ± 17.55, 120.60 ± 20.55, 145.40 ± 25.78, 123.70 ± 14.21 and 110.70 ± 21.65 cells/section in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) and the pSmad2 density (pSmad2 density/Sham: 1.01 ± 0.20, 1.62 ± 0.56, 5.22 ± 1.64, 3.84 ± 1.10 and 1.88 ± 0.61 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) were significantly increased 3 days post operation compared with sham group (Fig. 5A-C). Using western blotting, we further found increased concentration of pSmad2 (pSmad2/β-actin(%): 14.03 ± 2.97, 23.50 ± 2.51, 28.26 ± 2.48, 20.85 ± 2.34 and 15.06 ± 2.59 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) and TGF-β1 (TGF-β1/β-actin(%): 50.13 ± 4.34, 45.40 ± 4.13, 67.15 ± 4.56, 62.10 ± 3.93 and 45.40 ± 7.46 in Sham, 1 d, 3 d, 7 d and 28 d group, respectively) following CCH (Fig. 5D-F). The result indicates that the trend of TGF-β signaling activation is consistent with BBB breakdown. The consequences of BBB breakdown is regulated by TGF-β signaling following CCH.
3.6. Protection BBB integrity ameliorates brain damage following CCH
Neurotoxic molecules entered the brain parenchyma via increased endothelial transcytosis following CCH, we next investigated whether inhibition of endothelial transcytosis could ameliorate brain damage. Imatinib can effectively decreased BBB permeability occurred by endothelial transcytosis [24]. We found that imatinib treatment significantly reduced brain IgG accumulation 3 days post operation (100.00 ± 37.41 and 16.04 ± 7.81 in Saline and Imatinib group) (Fig. 6A, B). This indicates Imatinib shows effectively BBB integrity maintenance following CCH. We further assessed pathological outcomes after Imatinib treatment. We found the number of proliferative OPCs (16.67 ± 6.11 and 9.27 ± 3.63 cells/section in Saline and Imatinib group) and activated microglia (42.96 ± 8.72 and 21.52 ± 7.07 cells/section in Saline and Imatinib group) were decreased after imatinib treatment 3 days post operation (Fig. 6C-E, G). Further, TGF-β signaling was also decreased (126.60 ± 24.03 and 98.11 ± 22.27 cells/section in Saline and Imatinib group) (Fig. 6F, H, I). The results indicate that BBB dysfunction directly involves in the regulation of neuroinflammation responses and OPCs proliferation by regulating TGF-β signaling.